Interfacial Catalysisimage.sciencenet.cn/olddata/kexue.com.cn/blog/admin//images/upfiles/...Hydrogen...

10Phase Transfer Catalysis: Fundamentalsand Selected Systems

JING-JER JWO National Cheng Kung University, Tainan, Taiwan, Republic ofChina

I. INTRODUCTION

Heterogeneous chemical reactions between two reacting species located in immisciblephases are often inhibited due to the encounter problem. Conventional techniques tocircumvent this mutual insolubility problem rely on the use of rapid agitation and theuse of cosolvent, which exhibits both lipophilic and hydrophilic properties. If the reactiontakes place at the phase boundary, it is expected that the rapid agitation may have anaccelerating effect by increasing the interfacial contact. The addition of cosolvent mayeliminate the phase separation and provide a homogeneous mixing state for the reaction totake place. The cosolvents commonly used are the protic solvents such as methanol andethanol, and dipolar aprotic solvents such as acetonitrile, dimethyl formamide, anddimethyl sulfoxide. Although these cosolvents might resolve the mutual insolubility pro-blem, they render certain disadvantages such as the problem of promoting competinghydrolysis pathways and the difficulties in their purification and removal. A plausibletechnique now widely known as ‘‘phase transfer catalysis’’ (PTC) developed for overcom-ing the encounter problem due to the mutual insolubility of solvents appeared in the late1960s. In a PTC reaction, an added phase transfer catalyst is capable of transferring one ofthe reactants from its normal phase into a different phase where it can normally encounterand react under an activated state with the second reactant.

PTC, preceeded by some early reports [1–3], has emerged since 1971 as a versatiletechnique and become a very fascinating field of chemistry. Undoubtedly, it is worthy tocredit and compliment Starks [4,5], Makosza [6], and Brandstrom [7] for laying the foun-dations of PTC. The term ‘‘phase transfer catalysis’’ coined by Starks has been widelyaccepted and generally used. Other descriptive terms include ‘‘ion extraction,’’ ‘‘extractivealkylation,’’ and ‘‘catalytic two-phase reactions,’’ etc. PTC has attracted tremendousattention since 1965 and been applied to hundreds of reactions. The catalytic methodologyoffers many significant advantages over conventional methods, e.g., (1) acceleration of therate under mild reaction conditions, (2) use of inexpensive, recoverable, and nontoxicsolvents, (3) use of solvent-free reaction condition, (4) use of inexpensive and commerciallyavailable catalysts, (5) use of inexpensive inorganic bases for anion generation, (6)improvement of yield and enantioselectivity of products, and (7) use of continuous opera-tions for large-scale industrial applications. Based on the physical states of the phases,

Copyright © 2003 by Taylor & Francis Group, LLC

systems of PTC generally include liquid/liquid, liquid/solid, liquid/liquid/liquid, and gas/liquid. Although an overwhelming majority of publications and paterns in PTC deal withthe transfer of reactant anions from an aqueous or solid phase into an organic phase, theconcept of PTC is broader. It extends to the transfer of cations, neutral molecules, and freeradicals and includes reactions occurring exclusively or primarily at the interface. Ingeneral, PTC is an efficient methodology for the synthesis of a variety of compounds,such as haloalkanes, alkenes, aliphatic nitro compounds, nitriles, azides, sulfides, organo-metallic compounds, pharmaceuticals, amino acids, epoxides, peptides, pesticides, andpolmers. It has found widespread applications not only in research laboratories but alsoin numerous industrial processes.

In this chapter, an overview of the fundamentals, specific features, and selectedsystems of PTC is presented. An attempt is made to describe the basic concepts of PTCas clearly as possible and to confine its attention to those features of PTC that seem to beimportant for those who are interested in gaining a general knowledge of this attractivemethodology. It is hoped that those embarking on research in PTC may find this chapter auseful initial guide. Undoubtedly, they are required to read more comprehensive reviews,series chapters, and books [8–19] for advanced study in the field of PTC.

II. FUNDAMENTALS

A. Thermodynamic Aspects

1. Intermolecular Forces

The electrical properties of particles (molecules, atoms, or ions) play a key role in manyproperties of matter. The electrostatic attraction between opposite charges results in bond-ing (intramolecular) forces and intermolecular forces. Bonding forces (ionic, covalent, andmetallic bonds) are relatively strong because they involve larger charges that are closertogether. In contrast, intermolecular forces are generally weak because they typicallyinvolve partial charges that are farther apart. The types of intermolecular forces generallyconsidered in the molecular interactions are described briefly as follows.

(a) Ion–Dipole Forces. These forces arise from the attraction between an ion and apolar molecule (dipole) and are important in solutions of ionic compounds in polar sol-vents, e.g., the hydrated Me4N

þðaqÞ and Cl�ðaqÞ ions for Me4NþCl� in water. The

potential energy of an ion–dipole attraction is expressed as Eion-dipole ¼ �Z�=a"r2 whereZ is the absolute value of the charge on the ion, � is the dipole moment of the dipole, ris the distance between the ion and the dipole, a ¼ 4�"0, "0 is the vacuum permittivity,and " is the dielectric constant.

(b) Dipole-Dipole Forces. These forces arise from the interaction between the partialcharges of molecular dipoles. The interacted dipoles tend to orient themselves to maxi-mize the attraction between them. These forces are important in solutions of polar com-pounds in polar solvents, e.g., CH3Br in CH3CN. In a fluid of freely rotating polarmolecules, the interaction between dipoles average to zero. In fact, the molecules do notrotate freely even in a gas and there is a nonzero average interaction between polarmolecules. The average energy of interaction of two rotating dipoles is expressed asEdipole-dipole ¼ �2�2

1�22=ð3a2"2kTr6).

(c) Ion-Induced Dipole Forces. An uncharged nonpolar molecule can have a dipolemoment induced by the electric field of a nearby ion. The polarization of the nonpolarmolecule depends on its inherent polarizability (softness), �. The attractive interaction


between an ion and the induced dipole is important in solutions of ionic compounds innonpolar solvents, e.g., Bu4N

þBr� in benzene. The energy of interaction of an ion andan induced dipole is expressed as Eion-induced dipole ¼ �Z2�=ð2a"r4Þ:(d) Dipole-Induced Dipole Forces. A dipole can induce another dipole in a nearbynonpolar molecule, which results in an attractive interaction between them. These forcesare important in solutions of polar compounds in nonpolar solvents or vice versa, e.g.,CH3Br in toluene. The energy of interaction between a dipole and an induced dipolecan be expressed as Edipole-induced dipole ¼ �4�2�=ða2"2r6Þ:(e) Dispersion (London) Forces. Even in nonpolar molecules, instantaneous dipoleswill arise due to the momentary imbalance in electron distribution, which are capableof inducing dipoles in adjacent nonpolar molecules. The electrons in two or more ofthese nonpolar molecules tend to synchronize their movements at least partially to mini-mize electron–electron repulsion and to maximize electron–nucleus attraction. Theseinstantaneous dipole-induced dipole interaction are sometimes referred to as dispersion(London) forces and are responsible for the formation of condensed phases of nonpolarcompounds. These forces are important for solutions of nonpolar compounds in nonpo-lar solvents, e.g., benzene in toluene. The energy of such interactions may be expressedas Edispersion ¼ �ð2�1�2I1I2Þ=½3a2"2ðI1 þ I2Þ�, where I1 and I2 are the ionization energiesof the two nonpolar molecules.

(f) Hydrogen Bond. A hydrogen bond is an attractive interaction between moleculesthat have an H atom bound to a small, highly electronegative atom with lone electronpairs such as N, O, and F. If hydrogen bonding is present, it generally dominates theother intermolecular interactions with the exception of ion–dipole interactions.Hydrogen bonding is a primary factor in the ability of water to dissolve numerous O-and N-containing organic and biological compounds such as alcohols, amines, sugars,and amino acids.

2. Solubility

The intermolecular forces play an important role in determining the solubility of a solutedissolved in a solvent. The old rule of thumb ‘‘like dissolves like’’ usually provides a goodqualitative means to predict solubility. The energetics of solutions can be summarized asfollows. Keep in mind that there will usually be an entropy-driving force favoring theformation of solution. The solute–solute, solvent–solvent, and solute–solvent interactionsmust be considered in qualitative estimation of the enthalpy effect, i.e., the enthalpyof solution (�HsolutionÞ can be expressed as �Hsolution ¼ �Hsolute-solvent ��Hsolute-solute��Hsolvent-solvent, where the enthalpies may result from the various intermolecular forces.Solutions of nonpolar solutes in nonpolar solvents represent the simplest type of solution.The forces involved are all dispersion forces. If �Hsolution ¼ 0, the only driving force is theentropy of solution and an ideal solution is likely to form. At the other extreme fromthe ideal solutions of nonpolar compounds in nonpolar solvents are solutions of ioniccompounds in water. The enthalpy of solution may be expressed as�Hsolution ¼ �Hsolvation �U, where �Hsolvation is the total enthalpy of solvation and Uis the lattice energy of the ionic compound, respectively. The ion–ion bonding force in thelattice is inherently stronger than the ion–dipole forces between the ion and the polarsolvent molecules, but there are several of the latter interactions for each ion. As a result,�Hsolution may be either positive or negative or even close to zero. When �Hsolution isnegative, the free energy of solution ð�GsolutionÞ, �Gsolution ¼ �Hsolution � T�Ssolution, will


be especially favorable since both �Hsolution and the entropy of the solution (�SsolutionÞreinforce each other.

On the other hand, a compensation effect is exhibited when both �Hsolution and�Ssolution are positive. When �Hsolution has a small positive value, the mixing tendencyof entropy may force the solution to do work to pull the ions apart at the expense ofinternal energy, and the solution cools. If �Hsolution is sufficiently positive, such that theentropy factor is unable to overcome this, then the ionic compound will be insoluble. Thelattice energy of an ionic compound is inversely proportional to the sum of the ionic radii(i.e., rþ þ r�) whereas �Hsolution is the sum of enthalpies of solvation of the cation andanion, which are inversely proportional to the individual ionic radius (i.e., rþ or r� alone).For the dissolution of ionic compounds in water, the lattice energy is generally favoredrelative to the enthalpy of the solution when rþ ¼ r� and the reverse is true for r� � rþ orrþ � r�. For example, LiF is the least soluble lithium halide and the least soluble alkalifluoride in water, and CsI is the least soluble cesium halide and the least soluble alkaliiodide in water. In contrast, CsF and LiI are the most soluble salts in the series. A verypractical consequence of this argument is that the isolation of large complex ions likeR4N

þ is facilitated by isolating them as salts of equally large counterions. The solvationenergies of ionic compounds in nonpolar solvents are limited to those from the weak ion-induced dipole forces that are generally not large enough to overcome the very strong ion–ion forces of the lattice. Therefore, ionic compounds generally have limited solubility innonpolar solvents. The insolubility of nonpolar solutes in some polar solvents like watermight be rationalized by saying that the solute would willingly dissolve in water but thewater molecules would rather tie themselves together.

3. Surface Chemistry [20,21]

A molecule in the interior of a liquid interacts equally in all directions with its neighbors.Molecules at the surface of a liquid that is in contact with its vapor experience an unba-lanced intermolecular force normal to the surface, which results in a net inward attractionon the surface molecules. Subsequently, drops of liquids tend to minimize their surfacearea and to form an ideal spherical shape in the absence of other forces. Similarly, a liquidthat is suspended in another immiscible liquid so as to eliminate the effects of gravity alsotends to become spherical. Work must be done in creating a new surface. A fundamentalrelation of surface chemistry is shown in Eq. (1):

� ¼ �G=�Að ÞT;P;n¼ GS ð1Þ

where A is the surface area, � is the surface tension, and GS is the surface free energy perunit area with the unit of J=m2 or N/m. The surface tension of a liquid generally decreaseswith increased temperature due to the increased kinetic energy of molecules partially toovercome the attractions between molecules. The values of � at 293 K are (72.75, 21.69,28.88, 26.77, and 476Þ � 10�3 N/m for water, octane, benzene, CCl4, and mercury, respec-tively, and they are (51.68, 35.00, 45.0, and 37:5Þ � 10�3 N/m for the two-phaseH2O=C8H18, H2O=C6H6, H2O=CCl4, and H2O/Hg systems, respectively [22]. The workof cohesion is the reversible work per unit area required to separate a column of liquid andcreate two new equilibrium surfaces, which are beyond the range of their forces of inter-action. For a liquid X in contact with its vapor (V), the work of cohesion of X isWXX ¼ 2�XV. The work of adhesion per unit area between two different immiscible liquidsX and Y may then be expressed as WXY ¼ �XV þ �YV � �XY. A liquid will wet another


substance (liquid or solid), if its own work of cohesion is less than that of adhesionbetween it and the substrate.

A curious example is that of the distribution of benzene in water; benzene willinitially spread on water, then as the water becomes saturated with benzene, it willround up into lenses. Virtually all of the thermodynamics of a system will be affectedby the presence of the surface. A system containing a surface may be considered as beingmade up of three parts: two bulk phases and the interface separating them. Any extensivethermodynamic property will be apportioned among these parts. For example, in a two-phase multicomponent system, the extra amount of an ‘‘i’’ component that can be accom-mondated in the system due to the presence of the interface (ni) may be expressed asni ¼ ni � CIiVI � CIIiVII, where ni is the total number of molecules of i in the wholesystem, VI and VII are the volumes of phases I and II, respectively, and CIi and CIIi arethe concentrations of i in phases I and II, respectively. The surface (excess) concentrationof i is defined as �i ¼ ni=A, where A is the surface area. At equilibrium, the chemicalpotential of any component is the same in each bulk phase and at the surface. The Gibbsadsorption equation, which is one of the most widely used expression in surface andcolloid science is shown in Eq. (2):

�d� ¼Xi

ni=Að Þd�i ¼Xi

�id�i at constant T ð2Þ

where �i is the chemical potential of i component. Since the absolute value of �i isextremely dependent on the choice of dividing surface, the Gibbs dividing surface isnormally chosen so that ni and, hence, �i for the solvent are equal to zero so that allother components are measured with reference to that surface, giving the relative surfaceconcentrations. For example, �i;1 is the surface concentration of i relative to the solvent 1.Consider the simplest two-component system, containing solvent 1 and solute 2. For anideal solution, the surface concentration of solute 2 relative to solvent 1 may be expressedas follows:

�2;1 ¼ �ð1=RTÞd�=dðln c2Þ ð3Þ

where c2 is the molarity of solute 2. Surface-active substances that lower the surfacetension will have positive values of �, e.g., n-aliphatic (C6–C10) alcohols in water. Incontrast, electrolytes tend to raise the surface tension of water, indicating that they arenegatively adsorbed at the air–water interface, i.e., they tend to be repelled towards thebulk of water. In general, lyophobic solutes tend to accumulate at the surface in preferenceto remaining in the bulk solvent whereas lyophilic solutes tend to be repelled away fromthe air–solvent interface and thus raising the surface tension. For two liquid phases incontact at constant T and P, the Gibbs–Duhem equation requires that

Pnid�i ¼ 0 in

each phase [21]. Therefore, if two solvents are partially miscible, then the surface excess ofsolute 3 relative to solvents 1 and 2 may be expressed as follows:

�3;12 ¼ �3 � �1 n22n31 � n21n32ð Þ � �2 n12n31 � n11n32ð Þ� �= n22n11 � n21n12ð Þ ð4Þ

where nij’s are the moles of component i in solvent j; components 1, 2, and 3 are solvent 1,solvent 2, and solute 3, respectively. If solute 3 is distributed between the mutuallyinsoluble bulk solvents 1 and 2, then �3;12 reduces to Eq. (5) [23]:

�3;12 ¼ �3 � n31�1=n11ð Þ � n32�2=n22ð Þ ð5Þ


4. Mass Transfer [24]

Mass transfer, an important phenomenon in science and engineering, refers to the motionof molecules driven by some form of potential. In a majority of industrial applications, anactivity or concentration gradient serves to drive the mass transfer between two phasesacross an interface. This is of particular importance in most separation processes andphase transfer catalyzed reactions. The flux equations are analogous to Ohm’s law andthe ratio of the chemical potential to the flux represents a resistance. Based on the stag-nant-film model, Whitman and Lewis [25,26] first proposed the two-film theory, whichstated that the overall resistance was the sum of the two individual resistances on the twosides. It was assumed in this theory that there was no resistance to transport at the actualinterface, i.e., within the distance corresponding to molecular mean free paths in the twophases on either side of the interface. This argument was equivalent to assuming that twophases were in equilibrium at the actual points of contact at the interface. Two individualmass transfer coefficients (kcI and kcII) and an overall mass transfer coefficient (kc) couldbe defined by the steady-state flux equations:

JA ¼ kcI aIb � aIið Þ ¼ kcII aIIb � aIIið Þ ¼ kc aIb � aIIbð Þ ð6Þwhere JA was the flux of solute A, aIb and aIIb were the activities of A in the bulk phases Iand II, respectively, and aIi and aIIi were the activities of A at the place of contact forphases I and II, respectively. Under the assumption of equilibrium at the interface, theactivities aIi and aIIi were equal and then the following equation could be derived:

1=kc ¼ 1=kcI þ 1=kcII ð7ÞFor practical purposes, it was convenient to express transport rates in terms of the

bulk phase concentrations employed in the stoichiometry of the process. Furthermore, inthe simple two-film theory, it was assumed that the phases were in equilibrium at theinterface, i.e., there was no diffusional resistance at the phase boundary. However, sig-nificant interfacial resistances might be affected by the interfacial turbulence caused by thediffusion of solute or by the presence of surfactants that tended to concentrate at thesurface. For the system of solid particles suspended in a liquid in an agitated vessel, therewere many factors involved in the mass transfer, such as geometry of the vessel, the natureof the baffles, the type of impeller, the speed of agitation, the liquid viscosity, the mole-cular diffusivity of solute, and the size and porosity of particle, etc. It is not surprising thatthere is no reliable general correlation of mass transfer coefficients for such systems. Masstransfer between two liquids can be promoted by dispersing or suspending one liquid inthe second liquid as small drops, which provides a large surface of contact between the twophases. Applying the film theory to the system with simultaneous diffusion and chemicalreaction near an interface at constant temperature, the approximate rate of mass transferacross the interface for a first-order irreversible reaction could be expressed byNi ¼ ðkDÞ1=2Ci, where k was the rate constant and D was the diffusion coefficient of i [27].

5. Distribution Between Phases

The distribution of the phase transfer catalyst plays a crucial role in the success of PTCprocesses. For the distribution of a species A between the aqueous and organic phases,Aorg Ð Aaq, the distribution at equilibrium is determined by the standard free energychange (�G0

r ) of this process, which equals to �G0Aaq ��G0

Aorg. The distribution of Ain the organic phase is favored by positive �G0

r . The extraction of ionic compounds intoan organic phase from the aqueous phase or their solubilization in the organic phase


involves in many PTC reactions. A comprehensive review of the subject about the stateand properties of such solutions can be found in textbooks or monographs [28–30]. Theintermolecular forces are responsible for the stability and properties of an ion pair inorganic solvent. Polar protic solvents are expected to solvate both cations and anionsand lead to a high degree of dissociation of the ion pair into free solvated ions. Polaraprotic solvents such as dimethylsulfoxide (DMSO) and dimethylformamide (DMF) willsolvate cations easily. However, since the positive end of the solvent dipole cannot beapproached easily, anions are only poorly solvated. Salts are highly solvated in polaraprotic solvents. PTC reactions are usually carried out in low polar aprotic solvent withdielectric constants ranging from 8.9 (CH2Cl2) and 4.7 (CHCl3) to 2.3 (C6H6) and 1.9(C6H14). Although typical inorganic salts are negligibly soluble in aprotic solvents, organicquaternary onium salts are often quite soluble, especially in CH2Cl2 and CHCl3, with ionpairs being the dominant species.

Solvent extraction of ionic compounds from the aqueous to organic phase is wellknown to analytical and industrial chemists. For the extraction equilibrium of QþX� salt,Qþaq þX�aq Ð ðQþX�Þorg, the stoichiometric extraction constant EQX is defined by Schilland Modin [31,32] as

EQX ¼ QþX��

org= Qþ� �

aqX�½ �aq ð8Þ

To include the effects of competing side reactions such as association or dissociationequilibria of ion pairs in the organic phase, and pH-dependent equilibria in the aqueousphase, they also define a conditional extraction constant:

E QX ¼ EQX �QX=�QðXÞ�XðQÞ� � ð9Þ

where � coefficients serve as correction factors that deviate from unity. Extraction con-stants depend not only on the solvent system but also on the presence of foreign salts andare, therefore, determined generally at constant ionic strength of the aqueous phase.Quaternary ammonium (R4N

þ) ions have wide applications in PTC reactions. There isa relationship between the size of the R4N

þ ion and the extraction constant [9,33]. It isexpected that increasing the number of carbon atom in the R group will increase thelipophilicity (or organophilicity) of the R4N

þ ion and thus raise the extraction constantEQX. Gustavii [34] observed a linear relationship between logEQX and the total number ofcarbon atoms for the extraction of R4N

þ picrate� salt in CH2Cl2. The extraction of theR4N

þ ion is strongly influenced by the counterion. Combining the results of literature,Dehmlow and Dehmlow [18] arrived at the following order of lipophilicities of anions:picrate� �MnO

�4 > ClO�4 > SCN� > I�ðClO�3 , toluenesulfonate�Þ > NO�3 > Br� >

ðCN�;BrO�3 ; PhCOO�Þ > ðNO�2 ;Cl�Þ > HSO�4 > ðHCO�3 ; OAc�Þ > ðF�;OH�Þ > SO2�

4

> CO2�3 > PO3�

4 . A similar order of lipophilicities of anions is applicable to Ph4Pþ,

Ph4Asþ, and Ph3Sþ. In the practice of PTC, it is important to consider the competitive

extractions of two or more anions in the presence of a quaternary cation. For the compe-titive extraction reaction, ðQþX�Þorg þY�aq Ð ðQþY�Þorg þX�aq, the selectivity constant[KselðY=XÞ] is defined as:

KselðY=XÞ ¼ EQY=EQX ¼ QþY��

orgX�½ �aq= QþX�

� �org

Y�½ �aq ð10Þ

Selective constants, KselðCl=XÞ, are known for various anions and organic solvents[18,19,35–37]. It should be emphasized that the effects of side processes must not beneglected. Although the conditional extraction constants and the derived selectivity con-stants might not represent true constants, they are useful guides to the understanding of


anion exchange. For the competitive exchange of the simple anions, the percentageexchange in the organic phase is quite independent of the structure of the quaternaryammonium cation. In contrast, the percentage exchange increases with increasing stericavailability of the cationic nitrogen in halide–hydrogen sulfate and halide–sulfateexchange experiments [38]. Increasing the polarity and hydrogen-bonding ability of theorganic phase would exhibit a favorable effect on the extraction of small ions from theaqueous phase, but less effect on larger anions. Therefore, a leveling effect would beobserved [35]. Numerous PTC reactions are performed in the presence of the hydroxideion. In the absence of a residual amount of protic solvents, QþOH� ion pairs are veryinsolubile in nonpolar solvents. The amount of OH� ion extracted into the organic phasedepends on the structure of the Qþ cation [39]. In IPTC reactions involving OH� ions suchas alkylation, carbene additions and insertions, and isomerization, the OH� ion competeswith the other anions for the phase transfer cation. For the competitive extraction process,(QþOH�Þorg þX�aq Ð ðQþX�Þorg þOH�aq, the selectivity constant is defined as KselðX=OHÞ¼ ½QþX��org½OH��aq=½QþOH��org½X��aq. The values of KselðX=OHÞ in PhCl/NaOH(aq) med-ium are 30, 50, 120, 950, 2� 103, 3� 103, 1� 104, 5� 104, and > 1� 105 for X� ¼ SO2�

4 ,F�, OAc�, Cl�, PhCOO�, Br�, I�, SCN�, and MnO�4 , respectively [40]. In general, thehard monovalent anions compete much more favorably than the soft monovalent anionsand the divalent anions. In the presence of alcohols (ROH) (pKa � 18), the transfer ofRO� ions other than OH� can be important in promoting the base-initiated PTC reac-tions. Two processes are considered for this system. An acidity–selectivity constant isdefined as

ð1Þ ROHaq þOH�aq Ð RO�aq þH2O

ð2Þ ðQþOH�Þorg þRO�aq Ð ðQþRO�Þorg þOH�aq

ð3Þ K selðRO=OHÞ ¼ KaKselðRO=OHÞ ð11Þ

where Ka ¼ ½RO��aq=½ROH�aq½OH��aq and KselðRO=OHÞ ¼ ½QþRO��org½OH��aq=½QþOH��org½RO��aq.

Concluding remarks deduced from the experimental results [19,40,41] are:

1. The acidity–selectivity constant increases as the organophilicity of the alcoholincreases.

2. In general, the extraction of alkoxides of the diols is more favorable than thoseof monoalcohols, due in part to the intramolecular hydrogen bonding of themonoanion of the diol.

3. The extracted alkoxide ion may be solvated by unionized alcohol molecules viathe intermolecular hydrogen bonding.

It might be expected from simple pKa considerations that alkoxides would be more basicthan hydroxides. However, it turns out that hydroxide is a stronger base than alkoxideunder PTC conditions, similar to that observed in the gas phase.

6. Equilibria Involving Ion-Pair and Ion Aggregates

In principle, in the quaternary onium cation-catalyzed PTC reaction, the reactive speciescould be the free anion, the ion pair of the onium cation and anion, their complexaggregates, or a combination of all of these species. The behavior and structure of ionpairs and higher aggregates have been studied extensively using conductometric, spectro-photometric, spectroscopic, and magnetic resonance techniques [30]. In general, at low


concentrations, solvents with dielectric constant (") greater than 40 contain mainly dis-sociated ions whereas in those with " lower than 10–15 almost no free ions exist even athigh dilution. In aprotic solvents of low polarity, self-association between ion pairs leadsto the formation of aggregates [18,19,42,43] as shown below:

Qþ þX� Ð ðQþX�Þ Ð ðQ2XþÞX� Ð QþðQX2Þ� Ð ðQþX�Þ2 Ð etc.

In principle, the quaternary onium salt can exist as free ions, ion pairs, triple ions½ðQ2X

þÞX�, QþðQX2Þ��, quadrapole [(QþX�Þ2�; or higher aggregates. Electrochemicalconductance measurements of Bu4N

þNO�3 in benzene indicated that at a concentrationof less than 10�4:5 M, the salt existed mainly in the ion-pair form whereas within the range10�4:5–10�3 M it was probably in quadrapole form. In contrast, in solvents of higherpolarity like CH3CN and CH3OH, the salt was completely ionized within the range10�3–10�2 M.

B. Kinetic Aspects

1. Rates Involved in Phase Transfer Catalyzed Reactions

In general, PTC reactions involve processes occurring in series and/or parallel. A classicexample of PTC reaction is the two-phase reaction of 1-chloro-octane and aqueoussodium cyanide catalyzed by ðC6H13Þ4NþCl�ðQþCl�Þ [4,5]: ð1-C8H17Clorg þNaþCN�aq! 1-C8H17CNorg þNaþCl�aqÞ. In this reaction, the Qþ cation transfers CN� ion fromthe aqueous phase into the organic phase, activates the transferred CN� ion for reactionwith 1-C8H17Cl in the organic phase, and then transfers the product Cl� ion from theorganic phase back to the aqueous phase to start a new catalytic cycle. At least twoimportant steps are involved in this catalytic sequence, namely, the mass transfer stepand the intrinsic reaction in the organic phase. The kinetics of both steps are closely inter-related through the mediation of catalyst and reasonably high rates of both steps arenecessary to offer good PTC reactions. The overall rate of a PTC reaction will be deter-mined by the relative rates of both steps. If the transfer rate is faster than the intrinsicreaction rate, then the overall rate is limited by the rate of intrinsic organic phase reaction(e.g., the PTC reaction of 1-chlorooctane and aqueous sodium cyanide). On the otherhand, if the intrinsic reaction rate is faster than the transfer rate, then the overall rate islimited by the rate of mass transfer (e.g., the PTC reaction of benzyl chloride and aqueoussodium cyanide). Variables that may exhibit effects on the rates of mass transfer andintrinsic reaction include agitation, structure of catalyst, nature of reactant, organic sol-vent, and temperature, etc. These variables usually do not affect both rates equally, e.g.,the rate of agitation exhibits a strong effect on the transfer step whereas it shows littleeffect on the intrinsic reaction step. To increase the overall rate of an intrinsic reactionrate-limited PTC reaction, it is necessary to vary the factors such as catalyst, organicsolvent, and temperature to increase the rate of the intrinsic reaction in the organicphase. On the other hand, to increase the overall rate of a mass transfer rate limitedPTC reaction, it is helpful to vary the factors such as agitation, catalyst, and type ofinorganic anion to increase the rate of the mass transfer step. If both rates of transferand intrinsic reaction are very fast (e.g., in the PTC reactions of permanganate oxidation),it is easy to plan the reaction conditions to obtain satisfactory rates. In fact, the mainconcern is to think how to keep the reaction under control. If both transfer and intrinsicrates are very slow, it is then required to apply all possible skills to achieve a reasonableoverall reaction rate, e.g., use of dual catalysts, one to assist the mass transfer step and the


other to accelerate the intrinsic reaction step. It is reasonable to apply the rule, ‘‘likedissolves like,’’ to the kinetics of transfer as well as the thermodynamics of solubility, inwhich the nature of molecular interactions are similar.

For an intrinsic rate-limited reaction in the organic phase, if the rate equation can beexpressed as rate ¼ kirl½catalyst�org½substrate�org, then for a given substrate the rate ofreaction depends mainly on the concentration of the active catalytic species in the organicphase and on the intrinsic rate coefficient, kirl. The distribution of catalyst in the organicphase can be determined by the extraction constant for the two-phase organic/aqueoussystem. If the transferred catalyst is in the form of a catalyst–anion pair, then it is impor-tant to take the extent of aggregation into account to obtain the effective concentration ofthe active catalytic species.

The factors that affect the intrinsic rate coefficient include the nature of catalyst andsubstrate, the solvation of reactants, the transition state formed by the substrate and theactivated reactant anion, and the temperature. If the transferred catalyst is in the form ofcatalyst-anion pair, then it is important to understand the equilibria involving ion pair andion aggregates. The reactive ion in a different state of aggregation will exhibit differentvalues of kirl.

The thermodynamic formulation derived from the transition-state theory [44,45] isapplicable to the intrinsic reaction in the organic phase. The intrinsic rate constant may beexpressed as

kirl ¼ ðkT=hÞ exp ��G0 6¼=RT� � ¼ ðkT=hÞ exp �S0 6¼=R

� �exp ��H0 6¼=RT

� � ð12Þwhere, for kirl expressed in units of mol/dm3 (concentration) and seconds (time), theappropriate standard state for �G0 6¼ (standard free energy of activation), �H06¼ (standardenthalpy of activation), and �S0 6¼ (standard entropy of activation) is 1mol=dm3. It is alsopractical to apply the Arrhenius equation to understand the intrinsic reaction in theorganic phase, in which kirl can be expressed as

kirl ¼ A exp �Ea=RTð Þ ð13Þwhere A is the frequency factor and Ea is the activation energy, if A is temperatureindependent in the temperature range studied. A useful approach to the solvent effecton the reaction rate is in terms of the extent of solvation of the reactants and activatedcomplex (transition state) [46,47]. For example, consider the homogeneous displacementreaction of a tertiary amine (R3N) with an alkyl halide (R 0X) to form a quaternaryammonium halide:

R3NþR 0X! R3N�þ R 0 X�� ! R3NR

0þX�

Since the activated complex is partially ionized, it will be more solvated than thereactants in a polar solvent like nitrobenzene. The stabilization of the transition state leadsto a decrease in the free energy of activation and will accelerate the reaction rate. On theother hand, in the displacement reaction of an alkyl halide (RX) and a free anion (Y�):

RXþY� ! Y�� R X�� ! RYþX�

there is a decrease in polarity as the activated complex is formed. A polar solvent solvatesthe transition state less than the reactants and thus will decelerate the reaction rate. InPTC reactions, strong solvation of the reactant anion (including the hydration) will reduceits nucleophilicity.


2. Kinetic Order with Respect to Catalyst

In a PTC reaction catalyzed by quaternary onium salt involving the extraction of catalyst–anion ion pair, the kinetics is complicated by the reactive form of the reactant anion in theorganic phase. From both physical and kinetic points of view, two types of ion pairs canbe considered to exist, namely, the loose or solvent separated ion pairs and the tight orcontact ion pairs. Since any form of the anion (free ion, catalyst–anion ion pair, or ionaggregates) could be the reactive species in the PTC reactions, it is worthwhile exploringthe kinetics associated with the following two limiting cases of the reactive form of theanion.

Case 1. The quaternary salt is mainly in the monomeric ion pair form, which is inequilibrium with free ions, e.g., (QþX�Þorg Ð Qþorg þX�org. If the ionization equilibriumconstant (K) is very small and [QþX��org is approximately equal to the total concentra-tion of quaternary salt [QþX��t, then Eq. (14) can be easily derived:

½X��org ¼ ½Qþ�org ¼ K QþX��tÞ1=2�� ð14Þ

It then follows that the kinetic order with respect to the catalyst is expected to be 1/2,if the free anion (X�) is the reactive species. In contrast, if the ion pair is almost completelyionized (K � 1) or the ion pair is the reactive species (K � 1), then the kinetic order withrespect to the catalyst will be unity. The values of the ionization equilibrium constants ofquaternary ammonium salts are generally less than 0.01 in organic solvents (with " ¼ 2–20)most often used in PTC reactions [48]. It is clear that these onium salts exist in theseorganic solvents as ion pairs or perhaps some higher aggregates. For the PTC displace-ment reaction of 1-bromooctane and sodium cyanide catalyzed by quaternary phospho-nium salt [49] and the PTC halide exchange reactions catalyzed by quaternary ammoniumsalts [50], first-order kinetics with respect to the quaternary salt was observed, indicatingthat a monomeric ion pair was the reactive species of the anion. These results were con-sistent with the observations that ion pairs could react actively with alkyl halides with thereactivity of the anion correlating well with the cation–anion electrostatic interactionenergy [51].

Case 2. A quadrapole ion aggregate is the dominant species present, which is in equi-librium with the monomeric ion pair, e.g., ðQþX�Þ2org Ð 2ðQþX�Þorg. If the dissociationequilibrium constant K is very small and [(QþX�Þ2�org is approximately equal to half ofthe total concentration of quaternary salt [QþX��t, then Eq. (15) can be derived:

QþX��org ¼ K=2ð Þ1=2 QþX�� 1=2

t

hð15Þ

If the monomeric ion pair is the reactive species, then the kinetic order with respect to thequaternary salt is 1/2. If both the quadrapole and the monomeric ion pair are present incomparable amounts, then the kinetic order with respect to the quaternary salt depends onthe values of K and [QþX��t as indicated by

QþX��

org¼ �ðK=4Þ þ ðK=4Þ2 þ K QþX�

� �t=2

� �� 1=2 ð16ÞIn general, the kinetic order with respect to the quaternary salt is expected to be

between 1/2 and 1 [19]. It is important to point out that many anions and ion pairs areextracted into the organic phase along with hydrated water molecules. Typically, two tofive molecules of water are transferred along with an anion or ion pair from the diluteaqueous phase [5,49,52]. In general, the larger the charge/volume ratio of the anion thelarger the hydration number, e.g., the hydration numbers of Cl�, Br�, and I� ions in the


PhCl=H2O medium are 3.4, 2.1, and 1.0, respectively [5,49]. The presence of hydratedwater molecules tends to reduce the anion activation and vary the relative nucleophilicitiesof anions. For example, in the displacement reaction of n-octylmethanesulfonate andhalides under homogeneous conditions, the order of relative nucleophilicities isCl� > Br� > I� whereas it is Br� > I� > Cl� under PTC conditions [5,49,53]. The desic-cating ‘‘salting-out’’ effect provided by the presence of inorganic salt, especially concen-trated 50% aqueous NaOH solution, reduces substantially the hydration of anions and ionpairs [9].

3. Agitation

In principle, an ion pair is required to transfer physically from at least one bulk phase orinterface into another bulk phase in a PTC reaction. Without agitation, the interfacialarea is minimal and the PTC reaction tends to be mass transfer limited and is frequentlytoo slow to be useful. Agitation leads to an increase in the interfacial area as well as thesurface excess concentration of reactive species, and will thus accelerate the mass trans-fer rate. As the efficiency of agitation is increased, the reaction rate of a mass transferlimited PTC reaction becomes faster. As the transfer rate surpasses substantially theintrinsic reaction rate, then the rate of a PTC reaction will become independent ofagitation rate as reported in the classic example of the PTC reaction of alkyl halideand sodium cyanide [49]. Every PTC reaction tends to be dominated by either thetransfer limited or the intrinsic reaction limited, or both. Since the inherent maximumrate of an organic reaction is fixed under given reaction conditions, then the efficiency ofagitation will determine whether the overall reaction will be transfer rate limited orintrinsic reaction rate limited. Factors affecting the efficiency of agitation include thestirring speed, baffles, impeller shape, and positioning, etc. Since the transfer rate issolvent dependent, different rates of agitation may be required to maintain a constantlevel of transfer rate for a given PTC reaction performed in different solvents. A PTCreaction involving the transfer of anions having a high transfer rate such as I� andMnO�4 ions needs only minimal agitation. In PTC reactions involving transfer of anionshaving medium to slow transfer rate such as Cl�, CN�, OH�, HSO�4 , and SO2�

4 ions,more efficient agitation is required. The use of ultrasound may provide an extraordina-rily efficient means of agitation.

4. Temperature

The rates of most organic reactions increase with increasing temperature as expected fromthe transition-state theory. Therefore, increased temperature is likely to be considered forPTC systems that have slow organic phase reactions. However, in PTC reactions the effectof temperature is complicated by the thermal stability of the catalyst. Quaternary ammo-nium and other onium salts usually decompose at high temperatures (120–150�C) underneutral conditions and at lower temperatures (50–70�C) in the presence of concentratedNaOH (aq). The stability of complex formation of polyether catalyst with salts decreaseswith increasing temperature and thus reduces the catalytic activity of polyethers [54].Microwave irradiation is a good method of choice for heating in a PTC reaction [55].Under microwave irradiation, in the PTC reaction of o- and p-chloronitronenzene withethanol in the presence of NaOH (aq), a 144- to 240-fold increase in the reaction rate wasobserved due to the enormous increase in the reactivity of ethoxide ion resulting from thedehydration effect of the irradiation [56].


III. METHODOLOGY

Variables in reaction design for PTC reactions are more than those for homogeneousreactions. Since any PTC reaction can be transfer rate limited, intrinsic reaction ratelimited, or a combination of both, it is conceivable that there is no simple guideline forthe design, evaluation, and optimization of PTC reaction conditions. Rates of intrinsicreaction rate limited PTC reactions can be estimated by examining similar homogeneousreactions in the literature and also taking into account the deactivation of reactant anionsby hydration. The relative mass transfer rate of most anions into the organic phase can beestimated by examining similar PTC reactions using the same or a similar anion. If pre-liminary experiments or literature data indicate that the objective PTC reaction is feasible,then one can perform this reaction further by varying the reaction variables to optimizethe reaction conditions. Halpern and Lysenko suggested a guideline for exploring a newPTC reaction, based on substrate acidity [57,58]. More comprehensive approaches existfor considering separately the optimization of reaction variables in PTC reactions under avariety of conditions [19]. Based on the physical states of phases, PTC reactions aregenerally performed in the following systems: liquid/liquid, liquid/solid, gas/liquid, andliquid/liquid/liquid systems. In this section, the choice of some reaction variables and theireffects on the main features of PTC reactions are briefly described.

A. Catalysts

Selection or development of a phase transfer catalyst often plays the most important rolein developing a new PTC system. Two main factors considered in selecting a PTC catalystare the ability to transfer one of the reactants into the normal phase of the other reactantand the ability to activate the transferred species to facilitate the chemical reaction. Inpractice, other features of PTC catalysts considered by chemists or engineers in developinga PTC process include the stability, cost and availability, toxicity, recovery, recycling, anddisposal of catalysts.

1. Types of Catalysts

(a) Organic Soluble Catalysts for Extracting Anions into Organic Phase.

Quaternary onium salts. Quaternary ammonium salts include trioctymethylammo-nium chloride (Starks’ catalyst), Aliquat 336, tricaprylmethylammonium chloride,tetrabutylammo-nium hydrogen sulfate (Brandstrom’s catalyst), and benzyltrimethylam-monium chloride (Makosza’s catalyst); quaternary ammonium salts can also be generatedin situ from trialkylamines, etc. Other quaternary onium salts include tetrabutylphospho-nium bromide, tetraphenylphosphonium bromide, triphenylbenzylphosphonium chloride,tetraphenylarsonium chloride, and triphenylsulfonium chloride, etc. Special quaternarysalts are 4-aminopyridinium salts, bis-(quaternary ammonium) ½R3N

þ-ðCH2Þn-NRþ3 ;R3N

þ-ðCH2Þn-O-ðCH2Þn-NRþ3 � salts, 4,4 0-dialkylbipyridinium salts, cluster quaternaryammonium [e.g., PðC6H4SO

�3 NRþ4 Þ3] salts, crown-quaternary salts [e.g. (18-crown-6)-

(CH2Þ9PBuþ3 Br�], and chiral N-(4-trifluoromethyl)benzylcinchonium bromide, etc.

Macrocyclic polyethers. Crown ethers and cryptands include 18-crown-6, 15-crown-5, dibenzo-18-crown-6, dicyclohexane-18-crown-6, and [2.2.2]-cryptand, etc.

Open-chain polyethers (podands). Examples are polyethylene glycols (PEGs)[R-O-ðCH2CH2OÞn-R, R ¼ alkyl group] and tris-(3,6-dioxaheptyl)amine[(CH3-O-CH2CH2-OCH2CH2Þ3N, TDA-1].


(b) Water-Soluble Catalysts for Extracting Cations into Organic Phase. Examples arealkali metal salts of a lipophilic anion such as iodide, sulfonate, long-chain carboxylate,or tetra-arylborate, especially tetrakis-[3,5-bis(trifluoromethyl)phenyl]borate.

(c) Water-Soluble Catalysts for Extracting Organic Reactants into AqueousPhase. Examples are cyclodextrins, pyridine-1-oxide, 4-(dimethylamino)pyridine, tetra-methyl ammonium chloride, the rhodium complex of the trisodium salt of triphenylpho-sphine trisulfonic acid, and cuprous chloride.

(d) Insoluble Catalysts. Examples are:

1. Resin-bound PTC catalysts include polymer-NR3+, -PPh3

+, -SR2+, -crown

ethers, and -cryptands, etc.2. Inorganic solid-bound PTC catalysts include:

a. adsorption quaternary salts on organophilic clays such as smectite clay(hectorite), e.g., (n-C8H17Þ3NMeþ-hectorite;

b. adsorption of PhCH2NEtþ3 on SiO2, Al2O3-KF, SiO2-KF, Al2O3, C, orsand, and PEG chemically bonded to silica gel.

3. Third-liquid phase catalysts:Examples are toluene/Bu4N

þBr�=NaBr, toluene/Bu4NþBr�=NaOHðaqÞ,

Bu4NþHSO�4 =NaBr=NaOClðaqÞ, and toluene/PEG/KOH(aq), etc.

2. Quaternary Onium Salt Catalysts

Many quaternary ammonium, phosphonium, and arsonium salts are used as catalysts fortransferring anions in PTC reactions. Quaternary ammonium salts are the most frequentlyused due to their cost and availability. The criteria for selecting a quaternary onium salt asa PTC catalyst include extraction of the catalyst and reaction species into the organicphase and anion-activating ability, accessibility, and stability of the quaternary salt. Thestructural aspects and characteristics of quaternary onium salts, especially the quaternaryammonium salts are summarized as follows.

(a) Stability of Quaternary Onium Salt. Quaternary ammonium salts tend toundergo the following two main types of decomposition reactions: (1) the internal dis-placement (or dequaterization) reaction at high temperatures (100–200�C) to yield atrialkylamine and a displacement product, i.e., R4N

þY� ! R3NþRY; and (2) theHoffmann elimination reaction to yield a trialkylamine and an olefin in the presence ofa strong base, i.e., R 0CH2CH2NR3 þOH� ! R 0CH ¼ CH2 þR3NþH2O. The inter-nal displacement is usually not a serious problem at low temperatures (< 100�C).However, the presence of highly active substituents such as methyl and benzyl groupsattached to the central nitrogen atom tends to facilitate the internal displacement reac-tion. For the decomposition reaction, R 0R3NþOH� ! R3NþR 0OH, the relativereactivity of the R 0 group is allyl > benzyl > ethyl > propyl > methyl > isobutyl >phenyl [59]. Quaternary ammonium cations containing a methyl group tends to undergothe SN2 nucleophilic substitution reaction in the presence of a good nucleophile likethiophenoxide [60], e.g., MeðC8H17Þ3Nþ þ PhS� ! ðC8H17Þ3Nþ PhSMe. For the dis-placement of benzyl group from benzylpyridinium chloride in the absence of addedorganic solvent, the experimental results are somewhat surprising, which give the orderof relative reactivity of the neutral nucleophiles as Bu3N > Bu2NH > BuNH2 >HOAc > RSH � NH3 [61]. In contrast, quaternary phosphonium salts are much moreinert to internal displacement than the corresponding quaternary ammonium salts andare, therefore, more thermally stable under nonalkaline conditions. However, in the pre-


sence of a strong basic solution (NaOH) they become unstable and tend to undergodecomposition to trialkylphosphine oxide and alkane, i.e., R4P

þOH� ! R3POþRH.A comprehensive study of the stability of quaternary onium cations in the presence ofNaOH (aq) has been made by Landini et al. [62,63].

Some concluding remarks are:

1. Symmetrical tetraalkylammonium cation with longer alkyl chains tend to bemore stable e.g., ðn-C6H13Þ4Nþ > ðn-C4H9Þ4Nþ.

2. The presence of benzyl groups in quaternary ammonium cations tends to reducethe stability, e.g., ðn-C6H13Þ4Nþ > PhCH2ðn-C6H13Þ3Nþ.

3. Decomposition rates of quaternary ammonium and phosphonium cations inPhCl/NaOH(aq) medium increase dramatically with increased concentration ofNaOH(aq), due to the desiccating effect of concentrated NaOH(aq).

4. For a given quaternary ammonium or phosphonium cation in PhCl/NaOH(aq)medium, the relative order of the effect of halide ions on the decomposition rate isCl� > Br� > I�, e.g., (n-C6H13Þ4NþCl� > ðn-C6H13Þ4NþBr� > ðn-C6H13Þ4NþI� andPh4P

þCl� > Ph4PþBr� > Ph4P

þI� [64]), due to the increasing reluctance of halide ionto exchange with hydroxide ion to produce quaternary onium hydroxide in the organicphase.

5. Decomposition of a quaternary salt increases with agitation rate up to a pointand then levels off, which is consistent with the slow hydroxide transfer rate limitedprocess at low agitation rates, but slow intrinsic reaction rate limited process at highagitation rates.

(b) Extraction Ability, Anion-Activation Ability, and Accessibility of QuaternaryAmmonium Salts.

Extraction. A quaternary ammonium cation can be a successful catalyst only if ittransfers a sufficient quantity of the reactant anion from the aqueous phase into theorganic phase. In general, the ‘‘hard and soft acids and bases (HSAB)’’ empiricism [65]is applicable in considering the quaternary ammonium cation–anion–solvent interac-tions. Based on the intermolecular forces, it is expected that hard anions prefer to pairwith hard cations, and soft anions prefer to pair with soft cations [19]. The partition ofa catalyst cation–anion pair in the organic phase depends strongly on the structure ofthe quaternary ammonium cation for small anions such as Cl�, Br�, and CN�, but lessstrongly for large anions having considerable organophilicity such as picrate, MnO�4 ,and PhO�. Salts of Me4N

þX� (X ¼ F, Cl, Br, and CN, etc.) are not easily extractedinto most organic phase and are usually not good catalysts for extracting anions intothe organic phase, but may be useful for extracting cationic reactant into the aqueousphase. Tetra-alkylammonium (R4N

þ) cations with R ¼ C2H5 or C3H7 are usually poorfor extracting small anions, but may be useful for extracting organic anions, whereasthose with R ¼ n-butyl to n-decyl groups extract anions quite easily into almost allorganic phases.

Anion activation. It is generally required in PTC reactions that the reactantanion not only be transferred from the aqueous phase into the organic phase, but alsothat it is sufficiently activated for reaction with the other reactants in the organic phase.Bulky quaternary ammonium cations activate anions by increasing the distance separat-ing the cation from anion in the ion pair [e.g., Naþ Br� (r ¼ 285 nm) versusBu4N

þ Br� (r ¼ 0:628 nm)], which in turn will lower the energy of activation. Alarge bulky (‘‘soft’’) quaternary ammonium cation generally provides the required anionactivation for PTC reactions that tend to have slow intrinsic organic phase reactions. In


contrast, highly polar solvents tend to enhance the anion activation by reducing thecation–anion binding and allow less bulky or hard quaternary ammonium cations to besuccessful.

Accessibility. In contrast to PTC performed in neutral conditions in which orga-nophilic ammonium salts usually exhibit the highest activity, the base-promoted PTCreactions are effectively catalyzed by hard and even hydrophilic ammonium salts. Inparticular, the hydroxide-promoted PTC reactions were reported to have optimum reac-tivity with alkyltriethylammonium cations [17,19]. Halpern et al. [66] suggested the term‘‘accessibility’’ to rationalize the structural factor that determines the reactivity of aquaternary ammonium cation. This accessibility is important for PTC reactions whererates are limited due to slow anion transfer, e.g., those normally encountered in reac-tions with OH�, F�, OCl�, HSO�4 , and divalent anions. The accessibility of quaternaryammonium salt is especially important since hydroxide-promoted reactions account forover half of the PTC applications [19]. Quaternary ammonium cations that are rela-tively open-faced or accessible such as the hexadecyltrimethyl- or benzyltriethyl-ammo-nium cation readily occupy the interfacial positions and increase the interfacial areabetween the organic and aqueous phases via reduction of interfacial tension.Consequently, they increase the transfer rate of the anion into the organic phase [67].Benzyltriethylammonium chloride is extensively used in hydroxide-promoted alkylationPTC reactions due to the very strong tendency to lower the interfacial tension whereasthe use of hexadecyltrimethylammonium salts often leads to undesirable formation ofemulsions. Halpern suggested a quantitative parameter (q) for characterizing the accessi-bility of quaternary ammonium cations, which can be expressed as

q ¼X

1=CRið Þ ði ¼ 1; 2; 3; and 4Þ ð17Þ

where CRi is the number of carbon atoms of the alkyl group Ri, e.g. for CH3ðC8H17Þ3Nþ,q ¼ 1þ 3ð1=8Þ ¼ 1:38, and for CH3ðC4H9Þ4Nþ, q ¼ 1þ 3ð1=4Þ ¼ 1:75. Quaternaryammonium cations with q > 1 are generally considered to be accessible. Good correlationbetween q and reactivity in the PTC reaction of methylation of deoxybenzoin was obtained[19] using literature data [6,14,66]. It should be emphasized that the accessibility of aquaternary ammonium cation is not the only structural factor for determining the out-come of the transfer rate limited PTC reactions. A threshold organophilicity of the qua-ternary cation is generally required in order to form an ion pair with an anion that will besoluble to some extent in a suitable organic phase. Quaternary ammonium cations withq ¼ 1–2 are usually applicable in transfer rate limited PTC reactions. From the considera-tion of organophilicity, anion activation, and accessibility, it is not surprising that the n-Bu4Nþ cation is the most cited ammonium cation in patents as well as general PTCliterature, although it is usually not the optimal catalyst [19].

3. Uncharged Chelating Catalysts

Chelating agents such as macrocyclic polyethers (crown ethers, cryptands), open-chainpolyethers (polyethylene glycols), and acyclic cryptands have important applications inPTC reactions, attributed to their unique properties such as specific complex formationwith metal ions, the ability to solubilize and transfer ionic reagents from the aqueous orsolid phase to the organic phase, and the ability to activate the transferred anion in theorganic phase. The organic masking of the alkali metal ion provides an onium ion-likespecies that can be extracted or solubilized with the counteranion into nonpolar organic


solvents. Chelating ligands that complex Naþ and Kþ ions are particularly interestingsince sodium and potassium salts are the most frequently used salts in organic syntheses.

(a) Crown Ethers and Cryptands. The simple ‘‘lock and key’’ approach is helpful forselecting the crown ether; e.g., 18-crown-6 (cavity diameter, 0.26–0.32 nm) is more spe-cific for the Kþ ion (diameter, 0.266 nm) and 15-crown-5 (cavity diameter, 0.17–0.22nm) is more specific for the Naþ ion (diameter, 0.194 nm). However, the exact corre-spondence between cavity size and ionic diameter is not always a critical factor inorganic reactions. The solubilities of potassium salts in CH3CN are dramaticallyincreased by the presence of 18-crown-6 [68]. It is reasonable to believe that an impor-tant driving force for the increased solubilization of these salts is the organophilicity ofthe complex ion, which has a hydrophobic exterior. The solubility of a particular potas-sium salt is expected to be a complex function of the lattice energy of the salt and theorganophilicity of the crown ether. Distribution coefficients of alkali metal complexesof 18-crown-6 paired with inorganic counterions in H2O=CH2Cl2 medium were known[69]. It is misleading to describe the crown ether-mediated ‘‘anion activation’’ PTC reac-tions as the ‘‘reactions of naked anions,’’ since solvent–solute interactions are strongeven in weakly solvating or nonsolvating medium [69]. A leveling effect in nucleophilereactivity was observed in the investigation of 18-crown-6-mediated anion activation inCH3CN [70]. A total variation of less than one order of magnitude was observed in therate constants for displacements of benzyl tosylate for F�, Cl�, Br�, I�, CN�, N�3 , andOAc� ions. It should be emphasized that anion activation is suppressed substantially bythe presence of traces of water in the medium [71]. Cryptands (macrobicyclic multiden-tate ligands) are usually much superior to their macrocyclic counterparts in their abilityto complex alkali metal ions and to activate anions [72–74].

(b) Polyethylene Glycols and Acyclic Cryptands. Polyethylene glycols (PEGs), beingreferred to as a poor chemist’s crown ether, are open-chain analogs of crown ethers andare able to complex cations, to transfer anions into the organic phase, and to activatethe transferred anions. The formation constants of the Naþ–PEG complexex in anhy-drous MeOH for PEGs in the molecular weight range 200–14,000 range from 44 to12,000 [75]. It was concluded that the binding strength of complexation is a function ofthe total number of binding sites present and not the number of polymer chains, imply-ing that a long PEG chain may bind more than one cation. To obtain good partition ofa PEG into an organic phase may require the use of its mono- or di-ether derivative,since PEGs are themselves soluble in a dilute aqueous phase [76,77]. Based on the studyof the transfer of various potassium and sodium salts from the solid phase to PEG 400,Sasson and coworkers [78,79] made the following conclusions: (1) potassium salts aremore easily transferred than sodium salts; and (2) anions capable of hydrogen bondingwith the hydroxyl groups of PEG 400, such as OH�, F�, HSO�4 , and HCO�3 ions, aretransferred relatively easily from the solid phase to PEG 400. It is believed that theseanions are relatively free of hydration (‘‘naked’’) and that other anions such as Cl�,Br�, I�, and SCN� ions exist with significant hydration shell. Tris-(3,6-dioxaheptyl)-amine (TDA-1), an acyclic cryptand, is a highly effective catalyst for liquid/solid PTCreactions due to its hydrophilicity [80]. TDA-1 is an especially effective catalyst for thetransfer of sodium and potassium salts from the solid phase into the organic phase andis capable of dissolving sodium and potassium metals alloy in tetrahydrofuran (THF)to produce a deep blue solution which is useful for deoxygenation of acetates anddehalogenation reactions [81].


4. Insoluble Catalysts

Insoluble catalysts offer an important advantage of simple catalyst removal by filtration orcentrifugation after the completion of a PTC reaction. Regen [82] demonstrated thatquaternary onium cations chemically bound to insoluble resins could act as PTC catalystsand suggested the term ‘‘triphase catalysis’’ to describe the related PTC reactions.Insoluble PTC catalysts can be grouped into three categories, namely, the resin bound,the inorganic solid bound, and the third-liquid-phase catalysts as described in SectionIII.A.4(c).

(a) Resin-Bound Phase Transfer Catalysts. Tomoi and Ford [83] suggested the term‘‘polymer-supported’’ PTC to describe the PTC reactions occurring within the polymerphase. Resin-bound PTC catalysts include polymer-NRþ3 , -PPh

þ3 , -SR

þ2 , -crown ether,

-cryptand, -azacrown, -PEG, etc. In contrast to ordinary PTC reactions using solublecatalysts, PTC reactions using resin-bound catalysts require that both reactants diffuseto active PTC sites or the resin surface or to active sites inside the resin bulk phase forthe intrinsic reaction rate limited reactions. These also imply that both reactants arerequired to diffuse and penetrate the stagnant outer layer of the liquid(s) (i.e., theNernst layer) coating the resin particle as demonstrated in the reaction of 1-bromo-octane with NaCN(aq), known to have a slow intrinsic reaction rate, catalyzed by thestyrene–divinylbenzene resin-bound tributylphosphonium catalyst [84]. The resin-boundPTC catalysts generally consist of three elements, namely, the insoluble supportingcross-linked resin backbone, a spacer chain (optional), and the PTC functional group.Taking advantage of the huge amount of available ion-exchange resins, most publishedstudies on resin-bound PTC reactions use styrene–divinylbenzene resins and relatedresins.

Important factors affecting the efficiency of a resin-bound catalyst include levels ofcross-linking, ratios of chloromethylated rings to nonsubstituted ring (‘‘percent ring sub-stitution’’), and degrees of macroporosity. If percent ring substitution (RS, or PT-groupdensity) is too high, the resulting catalyst may tend to be too highly hydrophilic aroundthe active site, which inhibits the diffusion of hydrophobic organic reactants to the activesite. Resins having higher degrees of cross-linking tend to have smaller pores and are lesseasily swollen by liquids. Thus, catalyst activity decreases with increased cross-linking dueto increased resistance to reactant diffusion caused by increased tortuosity and rigidity ofthe resin. It is usually observed that catalysts with about 2% cross-linking exhibit thehighest catalyst activities whereas those with about 8–10% cross-linking exhibit bettermechanical stability.

Macroporous resins have greater internal porosity and surface area (up to 588 m2/g)than microporous or gel-type resins (0.06 m2=g), consequently allowing faster diffusionrates of reactants to active sites. For example, in the oxidation of benzyl alcohol tobenzaldehyde by NaOCl(aq) catalyzed by resin-bound catalysts, the observed effectivesurface diffusivity for the macroporous resin was 1.7-fold faster than that of the gel-type resin [85]. However, in the triphase-catalyzed reaction of 1-bromo-octane withNaCN(aq), the macroporous resin showed slightly slower rates than resin catalyst withlower porosity, which was explained by invoking that the pores in the macroporous resinwere completely filled with organic phase reactant, inhibiting the diffusion of anionicreactant to active sites, and consequently retarding the displacement rate [86]. The pre-sence of spacer chain (typically 8–20 carbon atoms) serves to separate the active sites fromthe resin backbone and from other active sites, especially the quaternary onium ions, so asto avoid the formation of ion aggregates, and to provide a reaction environment close to


that provided by soluble PTC catalysts [87]. Thus, the presence of space chains raises therates of reactions having a slow organic phase reaction, such as nucleophilic displacementreactions, by two- to four-fold [88,89]. The choice of functional group (such as quaternaryammonium and phosphonium groups, PEG chain, crown ether groupings, and cryptands,etc.) is usually very important for the feasibility of resin-bound catalysts, just as in thechoice of soluble catalysts, and is required to match the requirements of the reaction.

(b) Inorganic Solid-Bound Phase Transfer Catalysts. Two types of inorganic solid-bound PTC catalysts employed are the adsorption-type catalysts made by simpleadsorption of quaternary salts on organophilic clays, and the chemically bonded-typecatalysts made by chemical attachment of PTC functional groups to solid inorganicsupports. Adsorption of long-chain quaternary ammonium cations on particular formsof smectite clay (hectorite) is generally successful and commercially useful [90]. Using½ðn-C8H17Þ3N�þ–hectorite as the catalyst, the nucleophilic displacement of alkyl bro-mides with NaCN(aq), NaSCN(aq), Na2SðaqÞ, and NaOH(alcohols) in toluene/watermedium yields the expected nitriles, thiocyanates, sulfides, and ethers, respectively [91].Catalysts made by adsorbing PhCH2Et3N

þCl� on inorganic solid supports such asSiO2, SiO2-KF, Al2O3, Al2O3-KF, carbon, or sand were used as PTC catalysts for theN-alkylation of 2-oxazolidone [92]. Chemically bonded-type catalysts having -O-(CH2CH2OÞnR groups made by reacting a porous refractory oxide such as silica gel,containing surface hydroxyl groups, with a polyoxyalkylene oxide or monoalkyl etherof a PEG were patented [93] and shown to match the reactivity of resin-bound cata-lysts. In contrast to polymer-bound quaternary groups, the organic cations bonded tohigh surface-area silca and alumina exhibited a high affinity for hydrophilic anions suchas F�, HCO�3 , SO

2�4 , and PO3�

4 [94].

(c) Insoluble Third-Liquid-Phase Catalysts. Although insoluble solid-bound PTC cat-alysts have versatile industrial applications in PTC reactions, insoluble liquid-phase cat-alysts can be even more attractive. In an immiscible organic/aqueous two-phasemedium, it is expected that an increase in the difference between the cohesive forces(surface tension) of both phases due to changes in the compositions of solutes willdecrease their mutual miscibility and in turn change the solubilities of solutes in bothphases. If this medium effect causes a phase transfer catalyst to have limited solubilityin both the organic and aqueous phases, then this catalyst would rather exist in a third-liquid phase of its own. A well-known phenomenon called ‘‘coascervation’’ is used todescribe the formation of an additional phase (rich in surfactant) in a system when elec-trolyte is added to an aqueous solution of surfactant in large quantities. In the PTCreaction of the isomerization of allylanisole to anethol in toluene/KOH(aq) medium cat-alyzed by PEGs, Neumann and Sasson [95] observed a third liquid (PEG-KOH com-plex) phase formed, which increased the reaction rate dramatically. Nouguier andMchich [96] reported the formation of a third-liquid phase in the alkylation of pentaery-thritol in n-C7H15Br=NaOHðaqÞ medium to produce tri- and tetra-ethers.

Tetrabutylammonium salts frequently form third-liquid phases (or catalyst layers)when used in conjunction with organic solvents with low polarity such as toluene, hexane,and 1-chloro-octane, and with a concentrated aqueous solution of inorganic salts. Anexcellent example of tri-liquid-phase catalysis was demonstrated by Wang and Weng[97] in the displacement of benzyl chloride and sodium bromide in toluene/water mediumcatalyzed by Bu4N

þBr�, in which a third-liquid phase appeared under certain criticalconditions with a concomitant sharp increase in the reaction rate. Mason et al. [98]reported that Bu4N

þBr� uniquely formed a third-liquid phase in a toluene/Bu4NþBr�/


NaOH(aq) system, whereas Et4Nþ, ðC3H7Þ4Nþ, and (C6H13Þ4Nþ salts formed only two-

phase systems. Under static condition, the three liquid phases separate according to theirrelative densities, with the catalyst-rich phase being at the interface. For a stirred system,photomicroscopic observations revealed that dispersed drops of one phase were coated bya thick layer of the catalyst-rich phase, this being suspended in the continuous third phase.Through selective staining of the catalyst and organic phases, it was shown that the innerdroplets were the aqueous phase, which was coated by a catalyst-rich phase, and the wholewas dispersed in the toluene phase [98]. Correia [99] reported that the third-liquid phaseformed by Bu4N

þHSO�4 and NaOCl(aq) in the presence of NaBr consisted largely ofBu4N

þBr�3 , but also containing H2O, OCl�, Br�, Cl�, and possibly Br2, and that theaddition of cyclohexene to this system produced trans-1,2-dibromo- and (1-bromo-2-chloro)cyclohexane.

Weng and coworkers [100–103] investigated the PTC reactions of organic bromides(such as n-butyl bromide and ethyl 2-bromoisobutyrate) with sodium phenolate catalyzedby Bu4N

þBr�, focusing on the effects of solvents (such as toluene, hexane, and chloro-benzene) and inorganic salts (such as NaBr, NaOH, and Bu4N

þBr�) on the formation ofthe third-liquid phase and also focusing on the kinetics and mechanisms of these tri-liquid-phase catalyzed reactions. It was found that no third-liquid phase was formed when usingPhCl as the solvent of the organic phase. Although the tri-liquid-phase catalyzed reactionsare somewhat different from those of simple PTC reactions, the principles involved how-ever, are generally the same.

Wang and Weng [97,100] proposed that in tri-liquid-phase catalysis, both organicand inorganic reactants are transferred to the third-liquid (catalyst-rich) phase where mostof the intrinsic reactions take place. In commercial application, tri-liquid-phase catalysiswill allow organic reactions to proceed rapidly, with easy separation of the organic andaqueous phases, and the reuse of the catalyst-rich phase, as demonstrated by the tri-liquid-phase catalyzed reaction of n-butyl bromide and sodium phenolate catalyzed byBu4N

þBr� performed in a continuous-flow stirred vessel reactor [104]. Studies on model-ing mass transfer and interfacial reactions in tri-liquid-phase catalysis rationalize the mainfeatures of these systems, especially the jump in conversion on the formation of the third-liquid phase [105,106].

5. Comparisons of Catalysts

In addition to consideration of the structure–activity relationships, the criteria for select-ing a PTC catalyst usually include the following features: (1) stability, (2) cost and avail-ability, (3) removal, recovery, and recycling, (4) toxicity, and (5) waste treatment, etc. Thetetrabutylammonium cation is the most widely used quaternary ammonium cation. It iscommercially available in a wide variety of anions at moderate cost and has a uniqueapplication in tri-liquid-phase catalysis. It can also be easily separated and recovered byextraction, then recycled. Methyltributylammonium cation will become a popular catalystdue to its high reactivity in transfer rate limited PTC reactions, its lower toxicity than mostquaternary ammonium cations, and its low price. Methyltrioctylammonium cation is alsocommercially popular and is organophilic and anion activating enough to catalyze mostintrinsic reaction rate limited PTC reactions and is accessible enough to catalyze mosttransfer rate limited PTC reactions. In comparison with quaternary ammonium salts,quaternary phosphonium salts are generally more thermally stable and more activeunder neutral or acidic conditions, but less stable under alkaline conditions. Althoughthey have been used in a variety of PTC reactions, the greater cost compared to quaternary


ammonium salts limits their industrial applications. Toxic quaternary arsonium salts areused mainly for comparative processes. Tetraphenylarsonium salts are useful for PTCanalytical titration of highly organophilic unsaturated compounds [3]. The triphenylsul-fonium cation is stable under strong alkaline conditions and is effective for catalyzing thePTC displacement of 1-bromo-octane with NaCN, NaOPh, KSCN, and KI [107].

Catalyst separation or cost will usually be the main factor rather than the structure–activity relationshps for the choice of a specific PEG ether or crown ether. In contrast toquaternary onium salts, crown compounds are thermally and chemically more stable.Their application is limited mainly by the high cost and toxicity. 18-Crown-6 and itsderivatives have become available in ton quantities and in various grades of purity. Thecommercial application of crown ethers will be more feasible due to their reduced cost andtheir high reactivity. The PEGs and their capped ethers are more stable than quaternaryammonium salts and are attractive for processes using an excess of the PEGs due to theirlowest cost and least toxicity. PEG derivatives are generally included in standard screeningprograms for industrial processes, e.g., PEG 400 is always considered for hydroxide-pro-moted PTC reactions. PEGs and TDA-1 are inexpensive, thermally stable in the absenceof strong acids, usually easy to remove and recover, nontoxic, easily biodegradable, andcommercially available. Based on the yield and rate data of the displacement of chloridefrom benzyl chloride by acetate ion, so called the standard reaction for catalyst evaluation,several crown ethers, aminopolyethers, and cryptands, etc., were evaluated [108].

PTC reactions using insoluble catalysts offers the opportunities to separate easilyand recycle the catalyst, to prepare high-purity chemicals such as pharmaceuticals, and forcontinuous operations. Insoluble resin-bound catalysts are susceptible to stability pro-blems, mostly by thermal as well as mechanical degradation. Under sufficiently mildconditions, resin-bound catalysts with onium groups may be used for extended periodsor repeated cycles. Resin-bound PEGs, crown ethers, and cryptands are more chemicallystable than the corresponding onium salts. The disadvantages of resin-bound catalysts thatmust be overcome include the higher cost, the lower reactivity, and the lower capacities.Insoluble catalysts with quaternary onium cation adsorbed on hectorite are efficient,inexpensive, stable, and recyclable. The formidable task for tri-liquid-phase catalysts isto obtain third-liquid phase conditions that provide high catalytic reactivity yet do notcause significant loss of active catalyst by its being extracted into the organic and aqueousphases.

B. Counteranions and Anionic Reactants and Products

Most of the PTC reactions deal with the transfer and reactions of anions, especially thebase-promoted PTC reactions. It is apparent that factors such as the nature of the anion,the nature of the quaternary onium cation, and the effects of solvent are closely inter-related, and the combined effects of these factors should determine the outcome of a PTCreaction. Typical anionic reactants in PTC include nucleophiles, bases, and oxidants, andanionic products generally are leaving groups. In the order of lipophilicity of anions(Section II.A.5) [18], anions with higher lipophilicities will have a greater affinity forassociating with the quaternary ammonium cation, and those with lowest lipophilicitiesmay exhibit poison effects. The HSAB principle [65] is generally applicable for choosing asuitable quaternary ammonium cation to pair with an anion. The Bu4N

þ cation appears toexhibit sufficient lipophilicity and hydrophilicity and is able to perform reasonably wellwith the widest range of anions. Hydrogen sulfate ions are not only very good counter-anions for preparing many quaternary ammonium salts but also very useful for PTC


reactions, since in the strong basic solution they will be deprotonated to produce thehighly hydrophilic sulfate ions, which prefer to remain in the aqueous phase. Chlorideions are generally the second choice of counteranions to pair with quaternary ammoniumcations. If different anionic forms of quaternary ammonium salts are available, e.g.,Bu4N

þHSO�4 , Bu4NþCl�, Bu4N

þBr�, and Bu4NþI�, it is generally preferable to use

the HSO�4 and Cl� forms, and least preferable to use the I� form. Iodide anions andother anions that tend to pair strongly with the quaternary ammonium cation in theorganic phase tend to exhibit a poisoning effect, especially with anions that are difficultto transfer such as OH� and F�.

The effect of catalyst concentration on suppression of hydroxide ion extraction wasshown in the PTC isomerization of allylbenzene in toluene/40%NaOH(aq) catalyzed byBu4N

þ salt [109]. It was observed that the presence of 100-fold OH� ion relative to thecounteranion (X�) of Bu4N

þX� salts exhibited only a 45-fold increase in the reaction rateby varying X� from Br� to HSO�4 . It is clear that the reactant anion should be moreorganophilic than the leaving product anion, otherwise the latter would accumulate in theorganic phase and retard the reaction. Sometimes this catalyst poisoning is so severe that itis necessary to use stoichiometric amounts of catalyst and a hydrophilic counteranion, or acounteranion capable of conversion into a hydrophilic anion such as HSO�4 must be usedto pair with the quaternary ammonium cation. Deprotonation of organic substrates con-taining C–H, O–H, N–H, and S–H bonds, etc., with inorganic bases is perhaps one of themost plausible methods for forming a variety of organic anions used in the PTC reactions.The hard hydroxide ion is one of the most difficult anions to transfer from the aqueousphase to the organic phase.

However, it is one of the most valuable and frequently used anions in PTC reactions.It was shown that the quantity of OH� ion extracted into the organic phase decreased asits concentration in the aqueous phase increased, and the observed overall activity of OH�

ion actually increased due to the desiccating effect of the concentrated aqueous solution ofOH� ion [110]. Addition of a small amount of alcohol to a hydroxide-promoted PTCsystem usually causes a dramatic increase in reaction rate. One reason is that the alkoxideanions produced are more easily transferred into the organic phase than the highlyhydrated OH� ion and are at least as basic as OH� ions. The other reason is that thesolvation of the OH� ion with alcohol rather than with water increases its organophilicity[111]. In the PEG-catalyzed dehydrohalogenation of 2-bromo-octane in toluene/KOH(aq)medium, a maximum 126-fold increase in the reaction rate was observed in the presence ofmethanol [112]. It was observed that the decomposition of various quaternary ammoniumcations was retarded by the addition of methanol [113].

Clark and Macquarrie showed that PTC reactions with fluoride anion transfer couldbe considerably enhanced by use of Ph4P

þBr� as the catalyst [114]. Furthermore, theapplication of PTC methodology to the oxidation reaction of organic compounds byinorganic anionic oxidants such as MnO�4 ion [115–117] and OCl� ion [118] may improvethe yields and selectivity of products or even offer the possibility of performing the reac-tion that is impractical if the conventional methodology is employed because of thenarrow range of stable organic solvents that can be used.

Borohydride ion, an important inorganic anionic reductant, is sufficiently stable inan aqueous solution and can be transferred into a nonpolar organic solvent by typicalPTC catalysts. It was found that quaternary ammonium salts containing a �-hydroxylgroup are greatly superior PTC catalysts for borohydride reduction of aldehydes andketones [119].


C. Solvents

Some PTC reactions are conducted under ‘‘solvent-free’’ conditions [5,49,120].Nevertheless, it is more common to perform a PTC reaction in the presence of an organicsolvent or cosolvent, especially if the substrate is solid. In principle, an important factorfor choosing an organic solvent for a PTC system is that at least two phases are formed.Therefore, a very wide range of solvents may be considered, according to the nature ofphases and the nature of a given PTC reaction. The most commonly used solvents forliquid/solid PTC systems include benzene (and other hydrocarbons), dichloromethane,chloroform (and other chloro hydrocarbons), and acetonitrile. In liquid/liquid PTC sys-tems, the miscibility of the organic solvent with water is of particular importance. Formost of the applications, it appears that the chlorohydrocarbons such as dichloromethaneand chloroform are somewhat better solvents than the hydrocarbons such as benzene,toluene, and hexane. Dichloromethane and chloroform are commonly and successfullyused as organic solvents in PTC systems due to the high extraction capability for thestandard salts and to the low cost and the easiness of removal, although both may undergoside reactions. Chloroform is readily deprotonated to produce either trichloromethideanion or dichlorocarbene [121] and dichloromethane suffers the nucleophilic displacementreaction [122].

During the early stage (1970s) of developing PTC methodology, the key driving forcewas to optimize the PTC reaction to obtain high yield under mild and simple conditions.However, since the late 1980s, environmental issues have become increasingly dominant inevaluation of the industrial applications of PTC systems. Some of the major issues con-cerning the ‘‘green chemistry’’ are air emissions, occupational health and other industrialhygiene, wastewater treatment, etc. Therefore, although dichloromethane is one of themost common and useful solvents in the PTC literature, there is a tendency to eliminate itand other volatile chlorohydrocarbons due to stricter emission standards and to usemethyl isobutyl ketone instead. The criteria for choosing a solvent in a PTC system includethe nature of the chemical reaction, polarity, toxicity, volatility, flammability, cost, recycl-ability, and environmental considerations, etc.

In PTC reactions involving anionic reactants, the following solvent effects aregenerally considered: (1) the solubility and extraction of the catalyst–anion ion pair/complex in the organic phase, (2) the rate of transfer of the catalyst–anion ion pair/complex from the aqueous phase to the organic phase, (3) the activation of anion bysolvent separation of the catalyst–anion ion pair/complex, (4) the deactivation of anionby solvation of the anion (including hydration), (5) the extent of aggregation of thecatalyst–anion ion pair/complex in the organic phase, and (6) the stabilization of thetransition state formed by the active form of anion and the reactant in the organicphase, etc. For example, it was found that nonpolar solvents such as cyclohexane aremore effective than polar solvents such as chlorobenzene for the PTC reaction of thedisplacement of methanesulfonate by bromide ion catalyzed by (C16H33ÞBu3NþBr� salt[123]. Since this PTC reaction is intrinsic reaction rate limited, the nonpolar solvent canpromote the rate-determining step in the organic phase by reducing the extent of solva-tion (including hydration) of reactant anion (Br�) and increasing the concentration of(C16H33ÞBu3NþBr� ion pair in the organic phase, characterized by the extraction con-stant of the larger organophilic (C16H33ÞBu3Nþ cation.


IV. SELECTED SYSTEMS

Since in PTC, at least two immiscible phases and at least one interface separating thephases are present in the system, PTC reactions usually involve the transfer of reactantfrom its resident phase into the second (reaction) phase or the interfacial region forreaction with the second reactant, and the transfer of the product away from the reactionphase or the interfacial region. Thus, PTC reactions may involve several steps taking placeconcomitantly and/or in parallel and the detailed understanding of the relationshipsbetween steps and the factors that affect each step will be helpful for exploring andapplying PTC reactions. The overwhelming majority of PTC reactions involve the transferof one reactant, usually an anion, with a PTC catalyst from the aqueous or solid phaseinto the organic phase for reaction with the second reactant. In this chapter, this PTCmethodology is named as ‘‘normal phase transfer catalysis’’ (NPTC). In contrast, a com-plementary methodology named as ‘‘inverse phase transfer catalysis’’ (IPTC) [124]involves the transfer of one reactant with the assistance of a PTC catalyst from the organicphase into the aqueous phase for reaction with the second reactant. A special methodologynamed as ‘‘tri-liquid phase transfer catalysis’’ [98,100] involves the transfer of both organicand anionic reactants from the organic and aqueous phases, respectively, into the third-liquid (catalyst-rich) phase where the reaction takes place. In the following discussion,selected PTC systems will be presented and analyzed, focusing on the kinetic and mechan-istic aspects.

A. Normal-Phase Transfer Catalysis

1. Liquid–Liquid Phase Transfer Catalysis

Two limiting mechanistic models describing liquid–liquid PTC are the Starks extractionmechanism [4,5,49] and the Makosza interfacial mechanism [121,125]. However, theexperimental results of PTC reactions indicate that there is a spectrum of mechanismsthat fall within these two limiting mechanisms. Selected systems are discussed as follows.

(a) Starks Extraction Mechanism for Simple Displacement Reactions. The Starksextraction mechanism as illustrated in the classic example of the PTC displacement of1-chloro-octane (RY) with sodium cyanide (MþX�) catalyzed by quaternary onium salt(QþX�) [5,49] is depicted in Fig. 1. In this mechanism, PTC catalyst cation (Qþ) hasboth organophilic and hydrophilic properties and is distributed between the aqueousand organic phases, and the metal salts of reactant and product anions have limitedsolubility in the organic phase. The reactant anions are transferred across the interfacialregion into the organic phase as an intact catalyst cation–anion pair, so the productanions are transferred into the aqueous phase. However, if the metal salts of reactant

FIG. 1 Starks extraction mechanism for simple phase-transfer-catalyzed displacement reaction.


and product anions have sparing organophilicity and transfer themselves from the aqu-eous phase into the organic phase, then the exchange reaction, MþY� þQþX� !MþX� þQþY�, takes place in the organic phase. As a consequence, it is not strictly aPTC reaction. Nevertheless, the reactant anion is still activated by the catalyst cation. Ifthe catalyst cation is highly organophilic and distributed exclusively in the organicphase, then the Brandstrom–Montanari modification of Starks extraction mechanism(Fig. 2) is applicable [9,126], while the exchange reaction takes place at the interfacialregion. In these mechanisms, the catalyst cation–anion pair is considered as the reactivespecies in the organic phase. It is worthwhile noting that in the PTC cycle the ion pairtransferred into or generated in the organic phase does not need to be identical to theion pair added as PTC catalyst. It is only necessary that there is a lipophilic catalystcation or some equivalent cation solvator present in the solution, which dominates topair with the reactant anion (nucleophile) to be selectively extracted into the organicphase.

The Starks extraction mechanism and the Brandstrom–Montanari modification canbe described by the following reaction steps [19]:

ðQþY�Þorg þX�aq

k1

Ðk�1

ðQþX�Þorg þY�aq ð18Þ

ðQþX�Þorg þRYorg

k2!ðQ

þY�Þorg þRXorg ð19Þ

Step 1 [Eq. (18)] describes the competitive extraction reaction of reactant and productanions between the aqueous and organic phases in the presence of a catalyst cation. Therate constants k1 and k�1 include the effects of mass transfer across the interfacial regionand depend on the change in the interfacial area, i.e., on the agitation rate. Step 2 [Eq.(19)] describes an irreversible displacement reaction in the organic phase to produce theproduct RX and the product ion pair (QþY�Þorg, which subsequently exchanges with(QþX�Þaq by repeating Step 1. It should be emphasized that Step 2 need not be irrever-sible, the kinetics of the reaction will be more complicated, and the extent of reaction willdecrease. If Step 2 is irreversible, the rate equation can be expressed as

�d½RY�org=dt ¼ d½RX�org=dt ¼ k2½QþX��org½RY�org ð20Þ

FIG. 2 Brandstrom–Montanari modification of Starks extraction mechanism for simple phase-

transfer-catalyzed displacement reaction.


If there is no change in the individual [QþX��org and [QþY��org, then the steady-stateapproximation can be applied for [QþX��org, i.e., d[QþX��org dt ¼ 0, and the followingrate equation can be derived [19]:

�d½RY�org=dt ¼ k1k2½Qþ�org½X��aq½RY�orgÞ= k1½X��aq þ k�1½Y��aq þ k2½RY�org� ��

ð21Þwhere ½Qþ�org ¼ ½QþX��org þ ½QþY��org ¼ constant (the total concentration of Qþ salt inthe organic phase). The rate equation (21) allows a tractable analysis of a rather complexprocess in terms of some limiting cases that can be justified by experimental results. It isclear that simple integral-order kinetics would not be observed when rates of the masstransfer of the anions and the organic phase reaction contribute almost equally to theoverall rate. However, if large concentrations of X�aq and Y�aq ([X��aqi, ½Y��aqi) are presentinitially such that k1½X��aqi þ k�1½Y��aqi � k2½RY�org, then a pseudo-first-order kineticswill be observed:

�d½RY�org=dt ¼ k1k2½Qþ�org½X��aqi½RY�org� �

= k1½X��aqi þ k�1½Y��aqi� �

¼ kobs½Qþ�org½RY�org ¼ k 0obs½RY�orgð22Þ

Equation (22) is justified by the typical displacement of 1-chloro-octane with NaCN(aq)catalyzed by (C16H33ÞBu3PþBr�, performed in a saturated solution of NaCl and NaCN, inwhich the plot of k 0obs versus ½ðC16H33ÞBu3PþBr��org is linear [19,49]. Based on Eq. (21),some limiting cases are analyzed and discussed as follows.

1. If the rate of the organic phase reaction is slow compared to the mass transferrates such that k1½X��aq þ k�1½Y��aq � k2½RY�org, then Eq.(21) reduces to

�d½RY�org=dt ¼ k1k2½Qþ�org½X��aq½RY�org� �

= k1½X��aq þ k�1½Y��aq� � ð23Þ

Since ðk1½X��aq þ k�1½Y��aqÞ remains nearly constant during the reaction, the reactiongenerally follows reasonably good second-order kinetics, as observed in the reaction ofthiophenoxide ion with 1-bromo-octane in benzene/water medium using a variety of PTCcatalysts [122]. In the presence of excess initial concentrations of X�aq and Y�aq, a pseudo-first-order kinetics be observed, which is similar to Eq. (23) [18,48]. Equation (23) can berewritten as

�d½RY�org=dt ¼ k2K1½Qþ�org½RY�org� �

= K1 þ ½Y��aq� � ½X��aq� ð24Þ

where K1 ¼ k1=k�1. Under conditions of constant ½Y��aq and [X��aq and a well-stirredmixture, the ratio of [QþX��org=½QþY��org remains at a constant value of �, then Eq. (25)can be derived:

�d½RY�org=dt ¼ k2½QþX��org½RY�orgÞ ¼ k2 �=ð1þ �Þð Þ½Qþ�org½RY�org ð25ÞEquation (25) is supported by the reaction of 1-chloro-octane with NaCN(aq) catalyzedby various lipophilic quaternary ammonium cations [19,49].

2. If the product anions are also highly organophilic and predominate to pair withthe catalyst cation such that k�1½Y��aq �k1½X��aqþk2½RY�org, then Eq. (21) reduces to

�d½RY�org=dt ¼ k2K1½Qþ�org½RY�org½X��aq� �

=½Y��aq ð26ÞSince in Eq. (26), K1½X��aq=½Y��aq � 1, the reaction could stall after a sufficient amount ofthe product anion is formed, i.e., a catalyst poisoning phenomenon will be observed. Sucha poison effect can be avoided by using a stoichiometric quantity of catalyst. In the PTC


reaction of 1-iodo-octane with NaCN(aq), the reaction essentially came to a halt after 15–25% conversion due to the catalyst poisoning effect caused by iodide ion [19,49]. However,in the reaction of benzyl chloride with sodium benzoate in toluene/water medium, theiodide ion paired with quaternary ammonium cation acts as a cocatalyst rather than a‘‘catalyst poison’’ [127].

3. If both rates of the mass transfer of reactant anion into the organic phase andthe organic phase reaction are fast compared to the rate of the mass transfer of the productanion into the organic phase such that k1½X��aq þ k2½RY�org �k�1½Y��aq, then Eq. (21)reduces to

�d½RY�org=dt ¼ k1k2½Qþ�org½X��aq½RY�org� �

= k1½X��aq þ k2½RY�org� � ð27Þ

This limiting case represents a desirable process since both the rapid steps lead in thedirection of product. However, the reaction is expected to follow nonintegral orderkinetics, except under conditions such as k1½X��aq � k2½RY�org or k1½X��aq � k2½RY�org.The kinetics of the sequential substitution reaction of hexachlorocyclotriphosphazene with2,2,2-trifluoroethanol catalyzed by various quaternary ammonium salts in a chloroben-zene/NaOH(aq) medium were investigated. It was concluded that the reaction rate wascontrolled by both the organic phase reaction and the mass transfer of 2,2,2-trifluoroeth-oxide ion [128].

4. If the organic phase reaction is very fast compared to the mass transfer stepssuch that k2½RY�org �ðk1½X��aqþk�1½Y��aqÞ, then Eq. (21) reduces to

�d½RY�org=dt ¼ k1½Qþ�org½X��aq ð28ÞThe reaction is first order with respect to reactant anion and is zero order with respect toorganic reactant, RY. This limiting case is demonstrated by the hypochloride oxidation ofdi-n-butyl sulfide in CH2Cl2=H2O medium catalyzed by Aliquat 336 [129].

A graphical overview of extraction mechanism limiting cases is presented in aquantitative three-dimensional representation of the two mass transfer rates (k1[X

�]aqand k-1[Y

-]aq) and the rate of organic phase reaction (k2[RY]org) [19]. Gordon andKutina [130] discussed the implications of the interplay between extraction and chemicalreaction. The minimum ratio of [X�]aq/[RY]org that is sufficient for producing pseudo-first-order kinetics was calculated. If the concentration of the phase-transfer cation in theorganic phase ([Qþ]org) is changing during the reaction, the steady-state approximation of[QþX�]org is no longer valid and a more complex nonsteady-state treatment of the kineticsmust be considered [19], as shown in the displacement of 1-bromo-octane and NaCN(aq)catalyzed by Bu4P

þBr� [49]. In this system, no simple rate law is applicable since thereaction starts slowly and proceeds more rapidly with time. Initially, the Bu4P

þBr� saltwas distributed mainly in the aqueous phase of the water/bromo-octane medium.However, as the amount of the more polar product octyl cyanide increased, its distributionin the organic phase also increased, which led to acceleration of the reaction.

(b) Mechanisms of Hydroxide-Promoted Reactions of Organic Acids. Numerous two-phase catalytic reactions such as C-, O-, N-, and S-alkylations, generation of carbenes,isomerization, and H/D isotope exchange, etc., are carried out in the presence of strongalkali metal hydroxides. Nevertheless, much controversy exists in the mechanisticaspects. The transfer of the strong hydrophilic OH� ions into nonpolar medium is ener-getically highly unfavorable even in the presence of a great excess of metal hydroxides.Almost all of the monovalent anions dominate in the competitive extraction with OH�

ion for the quaternary ammonium cations into the organic phase [40]. Therefore, direct


extension of the typical extraction mechanism may not be suitable for all of the hydro-xide-promoted two-phase catalytic reactions. For example, in phenylacetomitrile/50%NaOH(aq) medium, greater than 99% of the benzyltriethylammonium cation formedion pairs with chloride ions. However, a 70% yield of alkylated product could beobtained in the alkylation of phenylacetonitrile with n-butyl iodide and 50% NaOH(aq)in the absence of PTC catalyst [131]. This observation prompted Makosza to suggestthe interfacial mechanism and a general term ‘‘catalytic two-phase systems’’ [15].

(c) Selected Mechanisms for Hydroxide-Promoted PTC Alkylation.

Starks–Brandstrom–Montanari extraction mechanism. The Starks–Brandstrom–Montanari extraction mechanism can be described by the following reaction steps [19]:

ðQþX�mH2OÞorg þOH�aq

k1

Ðk�1

ðQþOH�nH2OÞorg þX�aq ð29Þ

ðQþOH�nH2OÞorg þRHorg

k2

Ðk�2

ðQþR�ðnþ 1ÞH2OÞorg ð30Þ

ðQþR�ðnþ 1ÞH2OÞorg þR 0Xorg

k3!ðQ

þX�mH2OÞorg þRR 0org ð31ÞIn the first step (Eq. (29)], the phase-transfer catalyst cation hydroxide (QþOH�) ion pairis transferred from the aqueous phase through the interfacial region into the organic phase(Starks extraction mechanism) or the (QþOH�) ion pair formed by the exchange reaction(MþOH� þQþX� ! QþOH� þMþX�) at the interfacial region into the bulk organicphase (Brandstrom–Montanari modification). By applying the steady-state approximationto (QþOH�nH2OÞorg and (QþR�ðnþ 1ÞH2OÞorg, the following rate equation can bederived:

d½RR 0�org=dt ¼k1k2k3½QþX��org½OH��aq½RH�org½R 0X�org

k�1k�2½X��aq þ k�1k3½X��aq½R 0X�orgþk2k3½RH�org½R 0X�orgð32Þ

In principle, for this mechansim to be operative the PTC catalyst must have suffi-ciently high organophilicity in order to extract OH� ions into the organic phase. It isexpected that the reaction rate increases with increased organophilicity of catalyst cationand is independent of the agitation rate above a certain value. If the reaction is organicphase reaction limited, such that k�1k�2½X��aq � ðk�1k3½X��aq½R 0X�org þ k2k3½RH�org½R 0X�orgÞ, then Eq. (32) reduces to

d½RR 0�org=dt ¼ k3K1K2 ½QþX��org½OH��aq=½X��aq� �½RH�org½R 0X�org

¼ kobs½RH�org½R 0X�orgð33Þ

where K1 ¼ k1=k�1 and k2 ¼ k2=k�2. The reaction is expected to follow a second-orderkinetics or a pseuso-first-order kinetics (e.g., if initially, ½RH�orgi � ½R 0X�orgi). On the otherhand, if the reaction is mass transfer limited such thatk2k3½RH�org½R 0X�org �ðk�1ðk�2 þ k3Þ½R 0X�orgÞ½X��aqÞ, then Eq. (32) reduces to Eq. (34)and the reaction is expected to follow a zero-order kinetics.

d½RR 0�org=dt ¼ k1 ½QþX��org½OH��aq� ð34Þ


Makosza interfacial mechanism. The Makosza interfacial mechanism can bedescribed by the following steps [15,19]:

RHorg þOH�aq

k1

Ðk�1

ðR�H2OÞif ðif ¼ interfacial regionÞ ð35Þ

ðR�H2OÞif þ ðQþX�Þorgk2

Ðk�2

ðQþR�H2OÞorg þX�aq ð36Þ

ðQþR�H2OÞorg þR 0Xorg

k3!ðQ

þX�H2OÞorg þRR 0org ð37Þ

In this mechanism, the first step involves the deprotonation of organic acid substrate by(MþOH�) at the interfacial region to produce the carbanion (R�), which forms the‘‘anchored’’ ion pair (MþR�) and remains in the interfacial region. The second stepinvolves the detachment of R� ion from the interfacial region into the bulk organicphase under the assistance of catalyst (Qþ) cation. The third step is the reaction of R�

ion with the second reactant (R 0X) in the bulk organic phase. By applying the steady-stateapproximation to (QþR�Þorg and (QþR�Þif , the following rate equation can be derived:


k�1k�2½X��aq þ k�1k3½R 0X�org þ k2k3½RH�org½R 0X�orgð38Þ

It is expected that the reaction rate depends on the agitation rate. The accessibility ofthe Qþ cation plays a key role in the detachment of the carbanion (R�) from its anchoredposition at the interfacial region by forming an ion pair (QþR�). However, the (QþR�)ion pair should be sufficiently organophilic to dissolve itself in the bulk organic phase. Ingeneral, the kinetics of the reaction is complex. In the two-phase catalytic ethylation ofdeoxybenzoin with ethylbromide in (CH2Cl2, C6H6, or p-xylene)/50% NaOH(aq) mediumcatalyzed by various symmetric quaternary ammonium salts [132], in all cases it wasobserved that those catalysts reducing the interfacial tension most markedly were alsothe best catalysts and it was suggested that the Makosza interfacial mechanism bestaccounts for the experimental observations [11,12].

Modified interfacial mechanism. A modified interfacial mechanism is shown by thefollowing steps [19]:

ðQþX�mH2OÞorg þOH�aq

k1

Ðk�1

ðQþOH�nH2OÞif þX�aq ð39Þ

ðQþOH�nH2OÞif þRHorg

k2

Ðk�2

ðQþR�ðnþ 1ÞH2OÞorg ð40Þ

ðQþR�ðnþ 1ÞH2OÞorg þR 0Xorg

k3!ðQ

þX�mH2OÞorg þRR 0org ð41Þ


In the first step [Eq. (39)], the exchange reaction of OH�aq ion with (QþX�) ion pairtakes place at the interfacial region to produce the (QþOH�) ion pair. In the second step[Eq. (40)], the deprotonation of organic acid (RH) by (QþOH�) proceeds at the interfacialregion to produce the (QþR�) ion pair, which is transferred to the bulk organic phase toreact with the second organic substrate (R 0X) to yield the product RR 0 as shown in thethird step [Eq. (41)]. As in the Starks extraction mechanism, the PTC catalyst cation isinvolved in the formation of the carbanion. Based on this argument, it is clear that boththe Starks extraction mechanism and the modified interfacial mechanism are differentfrom the Makosza interfacial mechanism. By applying the steady-state approximationto (QþR�)org and (QþOH�Þif , the following rate equation can be derived:


k�1k�2½X��aq þ k�1k3½X��aq½R 0X�org þ k2k3½RH�org½R 0X�orgð42Þ

It is obvious that the Starks extraction mechanism and the modified interfacialmechanism are kinetically indistinguishable, Eq. (32) versus Eq. (42). The isotopeexchange reaction of fluorene in C6H6=16M NaODðD2O) in the presence ofBuEt3N

þCl� or BuEt3NþBr� was investigated [133]. It was observed that in the absence

of the catalyst no exchange took place, which was contradictory to the Makosza interfacialmechanism. The experimental results demanded the inclusion of the quaternary ammo-nium cation in the deprotonation of organic acid and provided evidence for the operationof the modified interfacial mechanism. In the tri-liquid phase catalysis of the dehydrobro-mination of �-phenylethylbromide to styrene in toluene/40% NaOH(aq) in the presence ofBuEt3N

þBr�, it was suggested that the modified interfacial mechanism best accounted forthe experimental results [134].

The acidity of organic acid may affect the mechanism of these hydroxide-promotedPTC reactions. Since relatively strong acids (e.g., acetylacetone, pKa � 9) can dissolve inNaOH(aq), the effect of the PTC catalyst cation is, therefore, to extract the conjugatebase anion in the form of an ion pair into the organic phase, where C- or O-alkylationoccurs. In other words, the classic Starks extraction mechanism is applicable. For ali-phatic alcohols (pKa � 18), both the uncharged alcohol (ROH) and its conjugate baseanion (alkoxide, RO�) can be extracted into the organic phase, which certainly willcomplicate the kinetics and mechanism of the reaction. The geminal diols (e.g., pinacol)can be extracted more efficiently than simple alcohols. Organic substrates(16 < pKa < 23), with somewhat activated methylene groups, are generally not watersoluble and do not readily dissociate in the absence of strong base, then the transferof OH� ion is usually required as mentioned above [19,57]. Organic substrates(23 < pKa < 38) are very difficult to deprotonate unless the base is very strong. It islikely that the rate of deprotonation is slow compared to the rate of OH� ion transferand predominates in the overall rate expression [19,57]. Organic substrates with pKa >38 are not likely to work in hydroxide-promoted PTC reactions [19,57].

(d) Reverse Transfer Mechanism for Dehydrohalogenation Reaction. An alternativemechanism called the ‘‘reverse transfer mechanism’’ is proposed for those PTC reactionsinvolving hydroxide ion, in which the active base in the organic phase is the quaternaryammonium halide (QþX�) rather than (QþOH�) or (QþR�). For example, in thekinetic study of the dehydrohalogenation of a series of substituted (1-haloethyl)benzeneswith 50% NaOH(aq) in the presence of Bu4N

þX� (X ¼ Cl, Br, I, and HSO4) [135], itwas found that the reactions followed a first-order kinetics and that the catalytic activ-ity of BuEt3N

þ salts followed the order Cl� > Br� > I� as expected. However,


Bu4NþHSO�4 salt showed an initial rapid reaction due to the initial presence of the

(Bu4NþOH�) ion pair extracted into the organic phase caused by the highly hydrophilic

sulfate ions. Nevertheless, as halide ion formed in the elimination reaction, it was pre-ferentially extracted into the organic phase due to its higher organophilicity, and actedas the base instead of hydroxide ion for the remainder of the reaction. Thus, when theextraction of hydroxide ion is inhibited, the reverse transfer mechanism may operate.The reverse transfer mechanism can be described by the steps shown below:

ðQþHX�2 Þorg þOH�aq

k1

Ðk�1

ðQþX�Þorg þX�aq þH2O ð43Þ

ðQþX�Þorg þRCH2CHXR 0orgk2!ðQþHX�2 Þorg þRCH ¼ CHR 0org ð44Þ

After the steady state is established, Eq. (45) can be derived by applying the steady-stateapproximation to (QþX�)org:

d½RCHCHR 0�org=dt ¼k1k2½QþHX�2 �org½OH��aq½RCH2CHXR 0�org

k�1½X��aq þ k2½RCH2CHXR 0�orgð45Þ

If reaction (44) is the rate-determining step such that k�1½X��aq � k2½RCH2CHXR 0�org,then Eq. (45) reduces to Eq. (46) and the reaction follows a first-order kinetics as observed[135]:

d½RCHCHR 0�org=dt ¼ k2K1 ½QþHX�2 �org½OH��aq=½X��aq� �½RCH2CHXR 0�org ð46Þ

(e) Mechanism of Dihalocarbene Generation and Addition Reactions. The method forgenerating dichlorocarbene in the two-phase CHCl3/50% NaOH(aq) medium in the pre-sence of quaternary ammonium salt [15,136] has gained widespread applications andopened a new chapter of carbene chemistry. Competitive reactions demonstrated thatthe dichlorocarbene generated in a PTC system is identical to that generated by othertraditional methods (e.g., by using potassium t-butoxide for the elimination) [5,49,137].In the absence of a reactant, the PTC mixture of CHCl3/concentrated NaOH(aq)/cata-lyst retains its ability to generate dichlorocarbene. In a 0.1 M solution of quaternaryammonium salt (QþX�) in CHCl3/50% NaOH(aq) medium, no evidence was found for(QþCCl�3 ) in the organic phase [137]. A mechanism based on the interfacial mechanismfor the dichlorocarbene addition reaction in a catalytic two-phase system is shownbelow:

HCCl3org þ ðNaþOH�Þaq Ð ðNaþCCl�3 Þif þH2O ð47ÞðNaþCCl�3 Þif þ ðQþX�Þorg Ð ðQþCCl�3 Þorg þ ðNaþX�Þaq ð48ÞðQþCCl�3 Þorg Ð ðQþCl�Þorg þ ð: CCl2Þorg ð49Þð: CCl2Þorg þ alkeneorg ! dichlorocyclopropane derivative ð50ÞStep 1 [Eq. (47)] involves the deprotonation of CHCl3 in the interfacial region to

form trichloromethylide (CCl�3 ) ions, which are anchored at the interfacial region [138].Step 2 [Eq. (48)] involves the detachment of the anchored CCl�3 ion from the interfacialregion into the bulk organic phase under the assistance of Qþ cation. Step 3 [Eq. (49)]


shows the reversible formation of : CCl2 in the organic phase with concomitant formationof a (QþCl�) ion pair, which stabilizes and increases the lifetime of : CCl2 in the organicphase. In contrast, when CHCl3 is treated with t-BuOK in a nonpolar solvent, theKþCCl�3 generated is insoluble and only : CCl2 can dissolve in the solution (i.e.,KþCCl�3 ! KCl # þ : CCl2). However, since the reaction of : CCl2 with alkene cannotcompete effectively with those with t-BuOH or t-BuOK, the irreversible consumption of: CCl2 takes place, which reduces its lifetime. In the presence of a reactant such as alkene,the irreversible addition of : CCl2 to alkene takes place to yield the dichlorocyclopropanederivatives as shown in Step 3 [Eq. (49)]. Since carbenes are usually strongly electrophilicspecies, they are capable of reacting with a variety of nucleophilic species such as alkenes,aromatic compounds, amines, alcohols, and S- and P-containing compounds [15].Reaction (49) competes with two reactions Eqs. (51) and (52), which are relatively slowdue to the phase boundary:

ð: CCl2Þorg þOH�aq! ðHOCCl�2 Þif ð51Þð: CCl2Þorg þH2O! ðHOCHCl2Þif ð52Þ

Reactions (51) and (52) lead to the production of HCOO� ion and CO(g) and in diluteNaOH(aq) they become more competitive and hydrolysis of CHCl3 is observed. In addi-tion, exchange reactions (53)–(55) may also take place, which lead to to the production ofXCCl�2 ion and HCXCl2:

ð: CCl2Þorg þ ðQþX�Þorg Ð ðQþCXCl�2 Þorg ð53Þð: CCl2Þorg þX�aq Ð ðCXCl�2 Þif þ ðQþX�Þorg ð54ÞðCXCl�2 Þif þH2O! ðHCXCl2Þif þOH�aq ð55ÞIn general, all other dihalocarbenes can be generated in a similar catalytic two-phase

system and subsequently add to alkenes. However, difluorocarbene is the only dihalocar-bene being excluded due to its very high rate of formation on deprotonation of theprecursor (CHF2Cl), which inhibits the transfer of CF2Cl

� ion by the Qþ cation [15]. Itwas generally observed that tertiary amines (R3N) catalyze the reactions involving dichlor-ocarbene. It was rationalized by invoking the reaction of R3N with (: CCl2) at the inter-facial region to form an ammonium ylide, which acted as a strong base and generated(: CCl2) in the bulk organic phase as shown in the following reactions [15,139]:

ðR3NÞif þ ðNaþClCCl�2 Þif ! ðR3Nþ � CCl�2 Þorg ð56Þ

ðR3Nþ � CCl�2 Þorg þ CHCl3org Ð ðR3N

þCHCl2CCl�3 Þorg ð57Þ

ðR3NþCHCl2CCl

�3 Þorg Ð ðR3N

þCHCl2Cl�Þorg þ ð: CCl2Þorg ð58Þ

2. Solid–Liquid Phase Transfer Catalysis

A major drawback in liquid–liquid PTC reactions involving transfer of anionic reactantfrom the aqueous phase into the organic phase is the coextraction of hydrated watermolecules. To cope with the problem of anion deactivation caused by the water of hydra-tion, it is reasonable to perform the PTC reactions with solid salts. This methodology iscalled ‘‘solid–liquid phase transfer catalysis,’’ which usually involves the reaction of ananionic reactant originally in a solid salt with a second reactant in the organic phase. Itwas reported that in some systems the liquid–liquid PTC reactions failed while the corre-


sponding reactions using the solid–liquid PTC method were successful [140,141]. It wasfound that the generation of (: CCl2) from sodium trichloroacetate (Naþ�O2CCCl3) undersolid–liquid PTC conditions was superior to that under liquid–liquid PTC conditions. Inanhydrous organic solvents such as absolute dimethoxyethane, (Naþ �O2CCCl3) decom-posed to generate ð: CCl2Þ, i.e. ðNaþ �O2CCCl3Þ ! NaþCl� þ CO2þ : CCl2. However, inthe presence of water, it decomposed to produce HCCl3, i.e.,(Naþ �O2CCCl3Þ þH2O! NaþHCO�3 þHCCl3. It was reported that very satisfactoryyields of dichlorocyclopropanes were obtained by the reactions of alkenes and : CCl2generated in the sodium trichloroacetate/quaternary onium salt/chloroform system[140,141].

The mechanistic description of the simple displacement reactions under solid–liquidPTC is quite similar to that of the liquid–liquid PTC counterpart [142]. However, incontrast to the liquid–liquid PTC system, the first step in the solid–liquid PTC systeminvolves the transport of a reactant anion from the solid phase to the organic phase by aPTC catalyst, which could be an organophilic quaternary onium cation, or an organo-philic complex cation formed by a metal ion such as Kþ with a polydentate ligand such ascrown ether, cryptand, PEGs and their derivatives, and TDA-1, etc. It is expected that thelattice energy plays an important role in the anion exchange of inorganic salt and qua-ternary onium salt. For example, it was observed that KCl salt did not exchange withBu4NBr in toluene, but KBr salt did exchange with Bu4NCl in toluene [143]. Similarly, thesolubilities of potassium salts in acetonitrile in the presence of 0.15 M 18-crown-6 followedthe order of KI > KBr > KCl > KF [68]. Under solid–liquid PTC conditions, the kineticsof displacement reactions of bromide and iodide ions and 1-bromo-octane and the mesy-late of 1-octanol catalyzed by Bu4NBr and 18-crown-6 ether were investigated [144]. Basedon the observed pseudo-first-order rate constants, it was concluded that 18-crown-6 cat-alyzed the reaction more effectively than Bu4NBr.

The effects of added water on the rates of displacement of benzyl bromide andbenzyl chloride with KCN salt in toluene catalyzed by 18-crown-6 were reported [145].It was observed that a small amount of water considerably increased the reaction ratescompared to the anhydrous conditions and that the rate increased sharply to a maximumvalue in the presence of an optimum amount of added water. An important observationwas that under anhydrous conditions, the reaction followed zero-order kinetics while inthe presence of added water it followed first-order kinetics. It was suggested that the initialsmall amounts of added water coated the surface of the salt particle, which extracted thecrown ether from the organic phase to form a new interfacial region called the ‘‘omega (!)phase.’’ It was believed that the catalytic reaction took place mainly in the omega phase,since the quantity of added water corresponding to the maximum quantity of crown etheron the surface of the salt particles correlated well with the optimum quantity of addedwater.

Furthermore, the results of a study on the distribution of 18-crown-6 between thetoluene phase and the omega phase indicated that the amount of crown ether in thetoluene phase remained low and relatively constant even though 3.50–11.45 mmol ofcrown ether were added [146]. It appeared that the omega phase acted like a spongethat was capable of adsorbing the added crown ether. For the 18-crown-6-catalyzed dis-placement of benzyl bromide with KCN salt in toluene [146], it was found that theobserved first-order rate constant was quite independent of the amounts (5.0–12.0mmol) of 18-crown-6, which implied that the displacement reaction probably took placemainly in the organic phase. Mechanistic rationalization of this crown ether-catalyzedtwo-phase reaction is described as follows:


1. In the absence of added water:

ðKþCN�Þs þ Lorg Ð ðKLþCN�Þorg ð59ÞðKþBr�Þs þ Lorg Ð ðKLþBr�Þorg ð60Þ

ðKLþBr�Þorg þ ðKþCN�Þsk1

Ðk�1

ðKLþCN�Þorg þ ðKþBr�Þs ð61Þ

ðKLþCN�Þorg þ PhCH2Brorgk2!PhCH2CNorg þ ðKLþBr�Þorg ð62Þ

Since it was required that KCN and KBr salts and crown ether (L) were stirred together intoluene for about 1 h prior to the addition of benzyl bromide in order to obtain repro-ducible kinetic data, the equilibria of Steps 1 [Eq. (59)] and 2 [Eq. (60)] were establishedand remained practically unchanged during the course of the reaction; in other words, theorganic phase solution was always saturated with (KLþCN�) and (KLþBr�) i.e.,½KLþCN�]org and [KLþBr�]org were constant. Therefore, as long as KCN and KBrsalts are present, the rate equation can be derived by considering mainly Steps 3 [Eq.(61)] and 4 [Eq. (62)]. By applying the steady-state approximation to (KLþCN�)org, thefollowing equation can be derived:

�d½PhCH2Br�org=dt ¼ k1k2½KLþBr��org½PhCH2Br�org=ðk�1 þ k2½PhCH2Br�orgÞ ð63ÞIf k2½PhCH2Br�org �k�1, then Eq. (63) reduces to Eq. (64) and the reaction is expected tofollow zero-order kinetics as observed [146]:

�d½PhCH2Br�org=dt ¼ k1½KLþBr��org ¼ constant ð64Þ2. In the presence of added water:

ðKþCN�Þs þ ðLxH2OÞ! Ð ðKLþCN�nH2OÞ! ð65ÞðKLþCN�nH2OÞ! Ð ðKLþCN�nH2OÞorg ð66ÞðKþBr�Þs þ ðLxH2OÞ! Ð ðKLþBr�mH2OÞ! ð67ÞðKLþBr�mH2OÞ! Ð ðKLþBr�mH2OÞorg ð68Þ

ðKLþBr�mH2OÞorg þ ðKþCN�Þsk1

Ðk�1

ðKLþCN�nH2OÞorg þ ðKþBr�Þs ð69Þ

ðKLþCN�nH2OÞorg þ PhCH2Brorg

k2!PhCH2CNorg þ ðKLþBr�Þorg ð70Þ

It was also required that KCN and KBr salts, crown ether (L), and water were stirredtogether in toluene for about 1 h to allow the equilibria [Eqs (65)–(68)] to be establishedbefore the addition of benzyl bromide. It is reasonable to assume that in the omega phase(!) the solution is always saturated with ðKLþCN�nH2O) and (KLþBr�mH2O), i.e.,


[KLþCN�nH2O�! and [KLþBr�mH2O�! are essentially constant so long as KCN andKBr salts are present. Thus, the rate equation can be derived based mainly on Steps 9 [Eqs.(69) and 10 (70)]. By applying the steady-state approximation to (KLþCN�nH2OÞorg, thefollowing equation can be derived:

�d½PhCH2Br�org=dt ¼ k1k2½KLþBr�mH2O�!½PhCH2Br�org=ðk�1 þ k2½PhCH2Br�orgÞð71Þ

If k�1 � k2½PhCH2Br�org, Eq. (71) reduces to Eq. (72) and the reaction is expected tofollow first-order kinetics as observed [146]:

�d½PhCH2Br�org=dt ¼ k2K1½KLþBr�mH2O�!½PhCH2Br�org ð72Þ

where ðK1 ¼ k1=k�1). This mechanistic interpretation is most applicable to the system inwhich an optimum amount of water is added. Since it was also observed that the quantityof water added was greater than the optimum value, the reaction rate decreased withincreased quantity of water and approached that of the reaction in toluene/H2O mediumand in the absence of crown ether [145]. Therefore, under these circumstances both thecrown-catalyzed solid–liquid PTC reaction and the uncatalyzed two-liquid-phase reactiontook place concomitantly.

B. Reversed Phase Transfer Catalysis

Besides the typical (normal) PTC reactions involving nucleophilic reactant anions andcationic catalyst, it is reasonable to believe that the PTC technique can be applied toreactions involving electrophilic reactant cations such as aryldiazonium and carboniumcations and anionic catalysts. In such ‘‘reversed phase transfer catalysis’’ (RPTC), acationic reactant in the aqueous phase is continuously transferred into the organicphase in the form of a lipophilic ion pair with a lipophilic, non-nucleophilic anioniccatalyst, and reacts with the second reactant in the organic phase.

Ellwood et al. [147] investigated the coupling reactions of 4-nitrobenzenediazoniumchloride with N-ethylcarbazole and N,N-dimethylaniline, etc., in H2O=CH2Cl2 mediumcatalyzed by sodium dodecylbenzenesulfornate and found that a 50-fold increase in thereaction rate was observed. Iwamoto et al. [148] reported coupling reactions of 4-nitro-benzenediazonium cation (generated in situ) with N-ethylcarbazole and 1-methoxy-naphthalene catalyzed by sodium tetrakis-[3,5-bis(trifluoromethyl)phenyl]borate(NaþTFPB�) in CH2Cl2=0:5M H2SO4ðaqÞ medium containing sodium nitrite [148].

They also investigated the Friedel–Crafts-type alkylation reactions of carboniumcations with m-methylanisole, o-cresol, and m-dimethoxybenzene catalyzed by(NaþTFPB�) in CH2Cl2=0:5M H2SO4ðaq) medium. The carbonium cations are generatedin situ by protonation of triphenylmethanol, diphenylmethanol, p-methoxybenzyl alcohol,and �-methylbenzyl alcohol [149]. Although the Friedel–Crafts-type reversed PTC appearsto be a promising methodology, its application is however, somewhat limited due to thegeneration and stability of the carbonium cation. Although relatively stable carboniumcations can be generated in dilute aqueous H2SO4, they can react only with active nucleo-philic aromatic compounds. On the other hand, less stabilized carbonium cations shouldbe generated in concentrated H2SO4 or other strong protic or Lewis acids. However, undersuch conditions, the reaction is interfered with by protonation of the aromatic substrateand the PTC catalyst.


C. Inverse Phase Transfer Catalysis

In contrast to the normal and reversed PTC methodologies, in which the chemical trans-formation takes place in the organic phase, it is reasonable to expect that PTC reactionscan also be performed by transferring the organic reactants from the organic phase intothe aqueous phase for reaction with a second reactant. Such a complementary methodol-ogy is named as ‘‘inverse phase transfer catalysis’’ (IPTC) by Mathias and Vaidya [124].Recently, the application of IPTC in organic synthesis has been reviewed by Li et al. [150].

1. Transition Metal Complexes

Cuprous chloride tends to form water-soluble complexes with lower olefins and acts as anIPTC catalyst, e.g., in the two-phase hydrolysis of alkyl chlorides to alcohols with sodiumcarboxylate solution [10,151] and in the Prins reactions between 1-alkenes and aqueousformaldehyde in the presence of HCl to form 1,3-glycols [10]. Similarly, water-solublerhodium-based catalysts (4-diphenylphosphinobenzoic acid and tri-C8-10-alkylmethylam-monium chlorides) were used as IPTC catalysts for the hydroformylation of hexene,dodecene, and hexadecene to produce aldehydes for the fine chemicals market [152].Palladium diphenyl(potassium sulfonatobenzyl)phosphine and its oxide complexes cata-lyzed the IPTC dehalogenation reactions of allyl and benzyl halides [153]. Allylic sub-strates such as cinnamyl ethyl carbonate and nucleophiles such as ethyl acetoactate andacetyl acetone catalyzed by a water-soluble bis(dibenzylideneacetone)palladium or palla-dium complex of sulfonated triphenylphosphine gave regio- and stereo-specific alkylationproducts in quantitative yields [154]. Ito et al. used a self-assembled nanocage as an IPTCcatalyst for the Wacker oxidation of styrene catalyzed by (en)Pd(NO3) [155].

2. Cyclodextrins

Cyclodextrins (cyclic oligosaccharide/polyalcohols of �-D-glucose with six to eight mono-meric units) form cyclindrical-like structures in aqueous solution having organophilicinteriors and hydrophilic exteriors and form complexes with a large number of compoundsand ions via the various intermolecular forces between host and guest molecules [156].Cyclodextrins can solubilize various organic compounds in aqueous phase solutions viaformation of host–guest complexes within the interior of the cyclodextrin structure; there-fore, they are expected to be good candidates of IPTC catalysts. Trifonov and Nikifornov[157] studied cyclodextrin-catalyzed IPTC nucleophilic substitution reactions of 1-bromo-octane and cyanide, iodide, and thiocyanate ions and found that both �- and �-cyclodex-trins could catalyze the reaction and that �-cyclodextrin was considerably more activethan �-cyclodextrin. �-Cyclodextrin has also been used as an IPTC catalyst in the follow-ing reactions:

1. The isomerization of 4-allylanisole with iridium(III) chloride to cis- and trans-anethol, in which a ternary allylanisole–cyclodextrin–iridium(III) chloride com-plex was proposed as an intermediate [158].

2. The selective oxidation of olefins, MeðCH2ÞnCH ¼ CH2 (n ¼ 5, 6, 7, 9, and 11),to ketones catalyzed by PdCl2=CuCl2 [159,160].

3. The epoxidation of alkenes such as styrene, cis-cyclo-octene, trans-hept-2-ene,and norbornene with iodosobenzene catalyzed by a water-soluble diaquo-N,N 0-ethylbenzenebis(salicyclideneaminato)chromium(III) complex [161].

4. The reductions of bromoanisoles with sodium formate in the presence of solid-phase (Pd/C) catalyst [162].


5. The hydrolysis of phthalic acid ester in 10% NaOH(aq) [163].6. The oxidation of methyl ketones by hypochlorite (the haloform reaction) [164].

An enhanced IPTC activity was observed for water-soluble �-cyclodextrin–epichlor-ohydrin copolymers in the nucleophilic substitution reactions of alkyl bromides andsodium iodide [165]. However, in the hypochlorite-induced oxidation of 1-phenyl-1-pro-panol or benzyl alcohol in the presence of �-cyclodextrin, the reactions were enhanced bylowering the pH of the aqueous phase rather than by the IPTC catalyst [166]. In contrast,the secondary alcohol was inert in aqueous hypochlorite solution maintained at high pH,even in the presence of the cyclodextrin.

3. Surfactants

Boyer and coworkers [167–170] have investigated the following IPTC reactions usingsurfactants as the phase transfer catalysts: (1) the reduction of ketones by sodium bor-ohydride in the presence of dodecenylsulfonate (monomer and polymer species) [167]; and(2) the epoxidation of �; �-unsaturated ketones (such as chalcone, carvone, citral, mesityloxide, and methyl cinnamate) by H2O2 in heptane/0.5 M NaOH(aq) medium and in thepresence of cationic surfactants (e.g., dedecylenyltrimethylammonium bromide) [168–170].The results indicated that the reaction was catalyzed by water-soluble micellar aggregatesof the surfactant and the catalytic effects depended strongly on the hydrophobicity of thesubstrate. Interesting results were observed in the study of the effect of surfactant con-centration on the epoxidation of chalcone by H2O2. It appeared that under slow agitation(100 rpm), the reaction occurred mainly via IPTC, while under vigorous agitation (1200rpm) it took place mainly at the interface due to the formation of an emulsion [170].

4. Tetramethylammonium Salts

Due to its low organophilicity (high hydrophilicity), the tetramethylammonium cation isnormally a poor PTC catalyst for transferring reactant anions into the organic phase.However, for certain systems where the reaction in the organic phase was not feasible,these salts could act as the IPTC catalysts as shown in the following reactions: (1) thefluorination of chlorobenzaldehydes [171] and the preparation of 1,2,2,2-tetrafluoroethyldifluoromethyl ether [172] with alkali metal fluorides, (2) the acetalization of sorbitol withbenzaldehyde to produce dibenzalsorbitol [173], and (3) the oxidation of benzyl alcohol byNaOCl(aq) to produce benzaldehyde, in which the catalyst was the trimethylammoniumgroups bound to resins [174].

5. Dimethylaminopyridine, Pyridine-1-Oxide, and Sulfide

An important class of IPTC reactions involves the conversion of a reactant and the IPTCcatalyst in the organic phase to an ionic intermediate that is transported into the aqueousphase for reaction with the second reactant to yield the organic product and to regeneratethe catalyst. This class of IPTC catalysts includes 4-(dimethylamino)pyridine (DMAP), 4-pyrrolidinopyridine, pyridine-1-oxide (PNO), tetrahydrothiophene, and diethylsulfide, etc.Mathias and Vaidya studied the first acylation reaction of alanine with decanoyl- or p-chlorobenzoyl chloride in H2O=CH2Cl2 medium catalyzed by DMAP [124]. In this system,the acid chloride reacted with DMAP in the organic phase to form the ionic intermediate,1-acyl-4-(dimethylamino)pyridinium ion, which was highly water soluble and sufficientlystable and was transported into the aqueous phase to react with the carboxylate ion ofalanine to yield the amide product. DMAP was also used as an IPTC catalyst to improve


the tosylation of alcohols and amines with tosyl chloride [175]. Similarly, Fife and Xin[176] reported the IPTC reaction of acid chloride with carboxylate ions catalyzed by PNOin H2O=CH2Cl2 medium to produce acid anhydride, in which the active ionic intermedi-ate, 1-(acyloxy)pyridinium ion, was formed by acid chloride and PNO in the organicphase. In the IPTC reaction of benzoyl chloride and phenols in H2O=CH2Cl2 mediumcatalyzed by PNO, it was observed that the IPTC reaction was more efficient than thenormal PTC reaction catalyzed by quaternary ammonium salts [177].

The pyridinyl- and 1-oxypyridinyl-substituted silanes and siloxanes were patented asIPTC catalysts in transacylation reactions [178]. In the IPTC nucleophilic substitutionreaction of benzoyl chloride with KSCN catalyzed by cyclic and acyclic sulfides such astetrahydrothiophene and diethyl sulfides, etc., the active ionic intermediate, benzylsulfo-nium ion, formed by benzyl chloride and sulfide in the organic phase, transferred into theaqueous phase to react with thiocyanate ion to produce benzylthiocyanate [179]. In thefollowing discussion, selected IPTC systems are presented, focusing on the kinetic andmechanistic aspects.

(a) DMAP-Catalyzed IPTC Reactions Involving �-Amino Acids. Asai et al. [180] stu-died the DMAP-catalyzed IPTC reaction of benzoyl chloride with glycine inH2O=CH2Cl2 medium in the absence of NaOH(aq), which produced high yields (up to94%) of hippuric acid, a precursor for the synthesis of aromatic amino acids such astryptophan and phenylalanine, etc., and the raw material of azlactone dyes. Themechanism of this reaction can be described as follows:

DMAPaq Ð DMAPorg ð73ÞDMAPorg þ PhCOClorg Ð DMAPCOPhþCl�org ð74ÞDMAPCOPhþCl�org Ð DMAPCOPhþCl�aq ð75ÞDMAPCOPhþCl�org þH2NCH2CO2Horg! PhCOONHCH2CO2Horg

þDMAPHþCl�org ð76ÞDMAPCOPhþCl�aq þH2NCH2CO2Haq! PhCOONHCH2CO2Haq

þDMAPHþCl�aq ð77ÞThe DMAPCOPhþCl� ion pair is the active ionic intermediate, formed by the reac-

tion of benzoyl chloride and DMAP in the organic phase. It was observed that the overallreaction rates were proportional to the interfacial concentration of DMAPCOPhþCl� inthe aqueous phase. In the absence of DMAP, the reaction was about three to four orders ofmagnitudes slower than that of the DMAP-catalyzed reaction. The yield of hippuric aciddecreased with increasing amounts of NaOH added, due to the hydrolysis of benzoylchloride. The overall rates could be rationalized by theoretical calculations based on theproposed model of this IPTC system including the consideration of the mass transferresistance of relevant reaction species. In contrast, Wang et al. [181] examined the feasibilityof the DMAP-catalyzed reaction of benzoyl chloride and the sodium salt of glycine in H2O(7 < pH < 10Þ=CH2Cl2 medium (Fig. 3). It was observed that the rates of both the unca-talyzed and DMAP-catalyzed reactions were fast and the yields of hippuric acid were veryhigh (up to 100%). These results were in contrast to those performed in the absence ofNaOH, in which both the reaction rate and the yield of hippuric acid were very low for theuncatalyzed reaction mentioned above [180]. It was observed that the reaction ratedepended on the agitation rate below 1200 rpm and on the shape of the reaction vessel


[181]. Both the uncatalyzed and DMAP-catalyzed reactions followed pseudo-first-orderkinetics in the initial presence of excess amount of the sodium salt of glycine (sodiumaminoacetate). Both the observed apparent pseudo-first-order rate constants increasedwith the initial concentrations of sodium aminoacetate and DMAP in the aqueousphase. The mechanism of the uncatalyzed reaction can be described by the followingreactions:

H2NCH2CO�2 aq þH2OÐ H2NCH2CO2Haq þOH�aq ð78Þ

H2NCH2CO2Haq Ð H2NCH2CO2Horg ð79ÞH2NCH2CO2Horg þ PhCOClorg! PhCONHCH2CO2Horg þHClorg ð80ÞH2NCH2CO

�2 ifPhCOClif ! PhCONHCH2CO2Hif þ Cl�if ð81Þ

The hippuric acid can be generated via Step 3 [Eq. (80)] in the organic phase and Step 4[Eq. (81)] in the interfacial region (if). The mechanism of the DMAP-catalyzed reactioncan be described by Eqs (73)–(83):

DMAPCOPhþClorg þH2NCH2CO2Horg! PhCOONHCH2CO2Horg

þDMAPHþCl�org ð82ÞDMAPCOPhþCl�aq þH2NCH2CO

�2 aq ! PhCOONHCH2CO2Haq

þDMAPaq þ Cl�aq ð83ÞIn the uncatalyzed reaction, the reaction rate was determined by Eqs. (80) and (81)

and in the DMAP-catalyzed reaction it was controlled by Eqs (74), (82), and (83). It wasdemonstrated that the reaction rates were similar in parallel experiments in which DMAPwas present initially in the organic and in the aqueous phase, respectively, which impliedthat the mass transfer of DMAP between the two phases was extremely rapid. Since thepKa values relative to water are 10–11, 4–5, and �1:74 for RNHþ3 , RCOOH, and H2O,respectively [182], the nucleophilicity of RNH2 is considerably higher than that of theRCCO� ion. Therefore, the reaction of PhCOCl with H2NCH2CO

�2 to yield

PhCOOCOCH2NH2 is negligible, as observed [181]. Since no benzoic acid was detected,the hydrolysis of PhCOCl was also negligible. Similar results were observed for othersodium salts of �-amino acids (RNHCHR 0COOH). These reactions proceeded rapidlyto produce PhCONRCHR 0COOH with high yields (85–100%). The order of reactivities ofamino acids was (N-methylglycine, l-prolineÞ � glycine� dl-alanine > 2-methylalanineð� acetic acid) [183]. The reactivities of these amino acids depended on their nucleopholi-cities and organophilicities (solubilities in CH2Cl2) and on the steric hindrance, e.g., thelow reactivity of 2-methylglycine was due to both the low solubility in CH2Cl2 and the

FIG. 3 Inverse phase transfer catalysis: the dimethylaminopyridine-catalyzed reaction of benzoyl

chloride and sodium salt of glycine.


steric hindrance of the 2-methyl group. An attractive application of the IPTC techniquewas demonstrated in the protection of the amino group of dl-serine with carbobenzoxychloride (benzyl chloroformate) in H2O=CH2ClCH2Cl medium catalyzed by DMAP [184].This method is useful for preparing the precursor for synthesizing the peptide containingthe serine moiety, since the protection of amino acids by the carbobenzoxy group isgenerally made in the alkaline solution, which is not applicable to dl-serine due to itsdecomposition in the alkaline solution to produce byproducts such as glycine.

(b) PNO-Catalyzed IPTC Reactions Involving Carboxylate Ions. Fife and coworkers[176,185] reported a similar IPTC process in which PNO was used instead of DMAP asthe IPTC catalyst in the two-phase reactions of acid chlorides and carboxylate ions tosynthesize the acid anhydrides (Fig. 4), which, being less reactive than acyl chlorides,are very important intermediates for the synthesis of esters, amides, and peptides. Jwoand coworkers [186–196] have carried out a systematic study on the kinetics andmechanism of the two-phase substitution reactions of benzoyl chlorides and carboxylateions using PNO as the IPTC catalyst, focusing on the substituent effects of benzoylchlorides, the structural effects of carboxylate ions, and the solvents, etc. Based on thekinetic results, a detailed mechanism was proposed for the PNO-catalyzed substitutionreaction of benzoyl chloride and benzoate ion in H2O=CH2Cl2 medium [186]. The mainreaction steps are shown as follows:

PNOaq Ð PNOorg ð84ÞPhCOClorg þ PNOorg ! PhCOONPþCl�org ð85ÞPhCOONPþCl�org Ð PhCOONPþCl�aq ð86ÞPhCOONPþaq þ PhCOO�aq! ðPhCOÞ2Oaq þ PNOaq ð87ÞPhCOONPþCl�org þH2O! PhCOOHaq þ PNOHþCl�aq ð88ÞPhCOClþH2O! PhCOOHþHCl ð89ÞReaction (89) can take place in both the organic and aqueous phases and in the

interfacial region. It was generally observed that without agitation the reaction rate wasslow and it increased with increasing agitation speed. However, the reaction rate wasindependent of the agitation speed beyond 1100 rpm in H2O=CH2Cl2 medium. ThePNO-catalyzed IPTC reactions of benzoyl chloride and benzoate ion produced a substitu-tion product (benzoic anhydride) and a hydrolysis product (benzoic acid). A high yield(> 95%) of benzoic anhydride could be obtained if a polar solvent like CH2Cl2 was used.Under suitable reaction conditions, the reaction followed pseudo-first-order kinetics asshown in Eq. (90):

FIG. 4 Inverse phase transfer catalysis: the pyridine 1-oxide-catalyzed reaction of benzoyl chloride

and sodium benzoate.


�d½PhCOCl�org=dt ¼ kobs½PhCOCl�org ð90Þ

The observed pseudo-first-order rate constant (kobs) depended linearly on the initial con-centration of PNO in the aqueous phase ð½PNO�iaq) and could be expressed as

kobs ¼ kh þ kc½PNO�iaq ð91Þ

In Eq. (91), kh and kc were the uncatalyzed (or hydrolysis) rate constant and cata-lyzed rate constant, respectively. Therefore, the reaction of PhCOCl and PNO in theorganic phase to yield the ionic intermediate, 1-(benzoyloxy)pyridinium chloride, [reaction(85)] was the rate-determining step in the PNO-catalyzed reaction path, which led mainlyto the production of benzoic anhydride. The value of kh obtained from the linear plot ofkobs versus ½PNO�iaq was generally consistent with that obtained in the uncatalyzed reac-tion. Therefore, reaction (89) was the main step in the uncatalyzed (hydrolysis) path,which led to the production of benzoic acid. In the following discussion, the main featuresof this IPTC system are described.

Solvent effects. In the PNO-catalyzed IPTC reaction of PhCOCl and benzoate ion,the order of the reaction rate in different two-phase media was H2O=CH2Cl2 >n-C6H14

=H2O > C6H6=H2O and the yield of benzoic anhydride in H2O=CH2Cl2 was considerablyhigher than those in the other two media. Similar results were generally observed for otherbenzoyl chlorides and carboxylate ions [191,192,194–196]. For example, for½PhCOCl�iorg ¼ 0:01M, ½PhCOONa�iaq ¼ 0:5M, and ½PNO�iaq ¼ 6� 10�4 M, the valuesof kobs at 22�C are (2.50, 2.25, and 0:417Þ � 10�3 s�1 for H2O=CH2Cl2, n-C6H14=H2O,and C6H6=H2O media, respectively. Although the reaction rate was fast in C6H6=H2Omedium, the reaction generated mainly the hydrolysis product, benzoic acid.Thermodynamically, the distribution of PNO in the organic phase is favored by thepolarity of the organic solvent. Kinetically, the reaction is also more favorable to takeplace in polar organic solvent as mentioned in Section II.B. Kinetic aspects, since thetransition state formed by PhCOCl and PNO (the rate-determining step [Eq. (85)] ismore ionic than both PhCOCl and PNO. These arguments were also strongly supportedby the PNO-catalyzed IPTC reaction of PhCOCl and acetate ion [187]. It was observedthat the order of relative reaction rates with respect to the effect of organic solvents wascyclohexanone > CH2Cl2 � CHCl3 > CCl4, which was consistent with the order of pola-rities. It was also found that in the H2O=CH2Cl2 medium (keeping the volume of organicphase constant), the reaction rate increased with the addition of an inert organic substancehaving larger polarity than CH2Cl2 such as nitrobenzene and benzonitrile whereas itdecreased with increasing amounts of added CCl4 [187]. Selected values of kobs areshown in Table 1.

Effects of carboxylate ions. The effects of carboxylate ions on the PNO-catalyzedIPTC reactions of PhCOCl and sodium carboxylates in H2O=CH2Cl2 medium were inves-tigated for selected carboxylate ions including formate, acetate, propionate, 2-methylpro-panoate, pentanoate, hexanoate, heptanoate, and octanoate ions [189]. It was found thatthe values of kobs depended somewhat on the type of the carboxylate ion under similarreaction conditions. For example, for ½PhCOCl�iorg ¼ 0:0100M, and½RCOONa�iaq ¼ 0:500M, the values of kc in kobs ¼ kh þ kc½PNO�iaq at 188C were (3.50,3.55, 3.52, 3.77, 3.83, 3.75, 3.83, and 3:35Þ �M�1 s�1 for RCOONa ¼ HCOONa,CH3COONa, C2H5COONa, ðCH3Þ2CHCOONa, n-C4H9COONa, n-C5H11COONa, n-C6H13COONa, and n-C7H15COONa, respectively. These results were rationalized bythe good correlations of the distribution of PNO in the CH2Cl2 phase and the carboxylate


ions in the aqueous phase [189,193]. The peculiar effect of the butanoate ion is described ina subsequent subsection.

The effects of dicarboxylate [RðCOONaÞ2] ions on the PNO-catalyzed IPTC reac-tions of PhCOCl and sodium dicarboxylates in H2O=CH2Cl2 medium were investigated onselected dicarboxylate ions including oxalate, malonate, maleate, fumarate, succinate,adipate, nonanedioate, phthalate, isophthalate, and terephthalate [190]. In general, theobserved products included mono- and bis-(benzoyloxycarbonyl) compounds, benzoicanhydride, and benzoic acid, which depended on the molecular structure of the dicarbox-ylate ion. Four types of dicarboxylate ions were classified according to the distribution ofproducts shown as follows:

1. Type I dicarboxyltes: the main product was PhCOOH (70–80%) and the minorproduct was ðPhCOÞ2O. Neither mono- nor bis-(mixed anhydride) productswere detected. They included oxalate, malonate, maleate, and succinate.

2. Type II dicarboxylates such as phthalate: the main product was the mono-(benzoyloxycarbonyl) compounds (PhCOOCORCOOH) and the minor pro-ducts were (PhCOÞ2O and PhCOOH.

3. Type III dicarboxylates: the main product was bis-(mixed anhydride)[RðCOOCOPhÞ2� with 70–88% yield. The minor products were (PhCOÞ2O,PhCOOH, or PhCOOCORCOOH. They included fumarate, isophthalate,and nonanedioate.

4. Type IV dicarboxylates such as adipate: the main products were PhCOOH andRðCOOCOPhÞ2; the minor products were (PhCOÞ2O and PhCOOCORCOOH.

Similar to the effects of monocarboxylates, the reaction rates depended significantlyon the type of dicarboxylates. For ½PhCOCl�iorg ¼ 0:0100M and½RðCOONaÞ2�iaq ¼ 0:500M, the values of the catalyzed rate constant (kc) at 18

�C were(4.10, 4.02, 3.83, 3.03, 4.27, 4.08, 3.80, 2.73, and 2:72Þ �M�1 s�1 for malonate, succinate,maleate, fumarate, adipate, nonanedioate, phthalate, isophthalate, and terephthalate,respectively. For ½PhCOCl�iorg ¼ 0:0100M, ½ðCOONaÞ2�iaq ¼ 0:200M, and½NaNO3�iaq ¼ 0:300M, the value of kc at 18

�C was 2:67M�1 s�1 for oxalate. These resultswere also rationalized by the good correlations of the distribution of PNO in CH2Cl2 andthe dicarboxylate ions, with the exception of the nonanedioate ion, due to interference by

TABLE 1 Effect of Composition of Organic Solvent on PNO-Catalyzed

Substitution Reaction of Benzoyl Chloride and Sodium Acetate in Two-Phase

H2O/Organic Solvent Medium

Organic solvent kobs ð10�4 s�1Þ Organic solvent kobs ð10�4 s�1Þ

Cyclohexanone 17.7 CH2CL2 (0.5M PhCN) 10.1

CHCl3 3.25 CH2Cl2 (0.5M PhNO2) 9.50

CCl4 2.70 CH2Cl2 (1.5M PhNO2) 10.1

CH2Cl2 8.08 CH2Cl2 (0.5M CCl4) 7.08

CH2Cl2 (0.5M PhCH2CN) 10.2 CH2Cl2 (1.5M CCl4) 4.75

CH2Cl2 (1.5M PhCH2CN) 12.0

½PhCOCl�iorg ¼ 0:0100M, ½PhCOONa�iaq ¼ 0:500M, and ½PNO�½iaq¼ 2:00� 10�4 M, at 18�C.Source: Ref. 187.


the emulsion phenomenon [189]. Reaction steps for the generation of RCOOCOPhCOOHand RðCOOCOPhÞ2 were proposed as follows:

Aqueous phase reaction:

PhCOONPþaq þRðCO�2 Þ2 aq ! PhCOOCORCO�2 aq þ PNOaq ð92ÞPhCOONPþaq þ PhCOOCORCO�2 aq ! RðCOOCOPhÞ2 aq þ PNOaq ð93ÞOrganic phase reaction:

PhCOONPþorg þRðCO2HÞ2org! PhCOOCORCO2Horg þ PNOorg ð94ÞPhCOONPþorg þ PhCOOCORCO2Horg ! RðCOOCOPhÞ2org þ PNOorg ð95ÞInterfacial reaction:

PhCOONPþif þRðCO�2 Þ2if ! PhCOOCORCO�2 if þ PNOif ð96ÞPhCOONPþif þ PhCOOCORCO�2 if ! RðCOOCOPhÞ2 if þ PNOif ð97ÞType I dicarboxylates tend to exist in the aqueous phase due to their low organo-

philicities. Reactions (92) and (93) or (96) and (97) were inhibited by the steric effect ofthe nearby second carboxylate group. As a consequence, the reaction was dominated bythe hydrolysis path [reactions (88) and (89)] to produce PhCOOH. Since the conjugateacids of phthalate (Type II dicarboxylate) ion and PhCOOCOC6H4CO

�2 ion had higher

organophlicities than those of the Type I dicarboxylates, the observed main product,PhCOOCOC6H4CO2H could be generated by reactions (92), (94), and (96). However,reactions (93), (95), and (97) were inhibited by the steric effect of the second carboxylatogroup at the ortho-position, since no C6H4ðCOOCOPhÞ2 was detected. In contrast, themain products were the bis(benzoyloxycarbonyl) compounds [RðCOOCOPhÞ2] for TypeIII dicarboxylates due mainly to the release of the steric hindrance of the second car-boxylato group. For isophthalate and terephthalate systems, reactions (92–97) wereinvolved in the generation of C6H4ðCOOCOPhÞ2. For the fumarate system, trans-C2H4ðCOOCOPhÞ2 was generated mainly via reactions (92) and (93) and (96) and(97), which was in contrast to its cis isomer, maleate (Type I). For the nonanedioatesystem, ðCH2Þ7ðCOOCOPhÞ2 was produced mainly by reactions (96) and (97) due to itssurfactant property. The propertities of succinate (Type IV dicarboxylates) ion seemedto occur at an intermediate position in these series and a wide distribution of products[(PhCOOH, ðCH2Þ4ðCOOCOPhÞCOOHÞ > ðPhCOÞ2O > ðCH2Þ4ðCOOCOPhÞ2] wasobserved.

Effects of substituents. In the PNO-catalyzed reaction of benzoyl chloride withbenzoate ion in H2O=CH2Cl2 medium, it was observed that the reaction of benzoylchloride and PNO in the CH2Cl2 phase to form the intermediate, 1-(benzoyloxy)pyridi-nium chloride, was the rate-determining step. Therefore, it was worthwhile investigatingthe effects of substituents on this system. The substituents included CH3, ðCH3Þ3C, CH3O,F, Cl, Br, and I groups. Similar to the PhCOCl/PhCOONa system, these reactions fol-lowed pseudo-first-order kinetics with the observed pseudo-first-order rate constant, kobs¼ kh þ kc ½PNO�iaq, Eq. (91). The values of the catalyzed rate constant (kc)[186,191,192,194,196] for XC6H4COCl are summarized in Table 2. The values of kc at22�C for the PNO-catalyzed reactions of Cl2C6H3COCl and the correspondingCl2C6H3COONa in H2O=CH2Cl2 medium to produce symmetric (Cl2C6H3COÞ2O were(15.6, 11.1, 15.4, and 57:3Þ �M�1 s�1 for, 2,3-, 2,4-, 3,4-, and 3,5-C6H3COCl, respectively[195]. The values of kc for the PNO-catalyzed reactions of Cl2C6H3COCl and PhCOONa


in H2O=CH2Cl2 medium to produce mixed PhCOOCOC6H3Cl2 were 11:6M�1 s�1 (22�C),

12:4M�1 s�1 (20�C), 7:98M�1 s�1 (20�C), and 10:5M�1 s�1 (22�C) for 2,3-, 2,4-, 3,4-, and3,5-C6H3COCl, respectively [195]. Good Hammett corelations were obtained for the meta-and para-substituents in the plot of logðkc=kcH) versus , where was the substituentconstant and kcH was the catalyzed rate constant of the parent compound (PhCOCl)(Fig. 5) [196]. The reaction constant (�) of the Hammett equation logðkc=kcHÞ ¼ �] soobtained for this reaction series was þ1:3, which implied that this reaction was a nucleo-philic substitution reaction and was expected to be accelerated by the electron-withdraw-ing substituent and retarded by the electron-donating substituent as observed in thesereactions. It is well known that the application of the Hammett equation to the ortho-substituent is usually poor mainly due to the steric effect. However, besides the inductiveand resonance effects, the electron-withdrawing ortho-substituent (F, Cl, Br, or I) alsofacilitates the reaction considerably by complexing with the positively charged nitrogenatom of the pyridinium moiety. In contrast, the electron-donating ortho-substituent (CH3

or CH3O) also retards the reaction via the steric effect.Reversible PNO-catalyzed benzoyl chloride / carboxylate systems. In contrast to

the other carboxylates (see earlier subsection), a peculiar phenomenon was observed in thePNO-catalyzed IPTC reaction of PhCOCl and butanoate (PrCOO�) ion in H2O=CH2Cl2medium, which led to an equilibrium with the PNO-catalyzed reaction of butanoylchloride (PrCOCl) and benzoate ion and vice versa [188]. It was observed that thePNO-catalyzed reaction of PrCOCl and PhCOO� ion reached equilibrium much morerapidly than that of PhCOCl and PrCOO� ion. For the PhCOCl/PrCOONa system, themain product was PrCOCl and the expected mixed benzoic butanoic anhydride

TABLE 2 Effects of Substituents on Catalyzed Rate Constant

(kc) for PNO-Catalyzed Reaction of Benzoyl Chloride

(XC6H4COCl) and Benzoate Ion (YC6H4COONa) in

H2O=CH2Cl2 Medium

X Y

kc(M�1 s�1) Ref. X Y

kc(M�1 s�1) Ref.

H H 3.60 186 2-Cl 2-Cl 8.10 191

2-CH3 2-CH3 1.49 196 2-Cl H 10.1 191

3-CH3 3-CH3 2.53 196 3-Cl 3-Cl 6.43 191

4-CH3 4-CH3 1.53 196 3-Cl H 6.37 191

4-ðCH3Þ3 H 1.87 194 4-Cl 4–Cl 5.37 191

3-CH3O 3-CH3O 3.40 196 4-Cl H 5.43 191

2-CH3O H nega 196 2-Br 2-Br 7.10 192

3-CH3O H 1.83 196 2-Br H 7.37 192

4-CH3O 4-CH3O 0.712 196 3-Br 3-Br 6.10 192

4-CH3O H 0.640 196 3-Br H 6.21 192

2-F 2-F 9.10 194 4-Br 4-Br 5.80 192

2-F H 10.4 194 4-Br H 5.61 192

3-F 3-F 6.10 194 2-I 2-I 17.5 196

3-F H 6.80 194 2-I H 10.9 196

4-F 4-F 3.40 194 4-I H 6.83 196

4-F H 3.93 194

a Negligible (mainly hydrolysis).


(PhCOOCOPr) was not observed. However, both PhCOOH and ðPhCOÞ2O (trace) wereobserved. The equilibrium conversion (Xeq) of PhCOCl depended on the concentrations ofPNO and PrCOO� ion for a given concentration of PhCOCl. For example, for½PhCOCl�iorg ¼ 0:0100M and ½PNO�iaq ¼ 6:00� 10�4 M, the values of Xeq at 18�C were0.536 and 0.552 for ½PrCOONa�iaq ¼ 0:500 and 0.100M, respectively. For½PhCOCl�iorg ¼ 0:0100M and ½PrCOONa�iaq ¼ 0:500M, the values of Xeq at 18�C were

0.494 and 0.519 for ½PNO�iaq ¼ 2:00� 10�4 and 4:00� 10�4 M, respectively. The value ofXeq depended insignificantly on the concentration of PhCOO� ion, the pH value(6:5 < pH < 10:7), the Cl� ion (by adding benzyltriethylammonium chloride), and thetemperature. The value of Xeq and the yield of ðPhCOÞ2O depended on the concentrationof PhCOO� ion, e.g., for ½PhCOCl�iorg ¼ 0:0100M, ½PNO�iaq ¼ 2:00� 10�4 M and½PrCOONa�iaq ¼ 0:500M, the values of [Xeq, ðPhCOÞ2O yield] at 18�C were 0.494 and2.78% and 0.554 and 25.7% for ½PhCOONa�iaq ¼ 0 and 0.2M, respectively. In contrast,in the PrCOCl/PhCOONa system, the reaction rapidly reached the equilibrium state withmuch smaller equilibrium conversion (< 0:1) to yield PhCOCl and no PrCOOCOPh beingdetected. PNO exhibited a great effect on the equilibrium yield of PhCOCl. Even in theabsence of PNO, a small amount of PhCOCl was observed. For example, for½PrCOCl�iorg ¼ 0:0100M and ½PhCOONa�iaq ¼ 0:500M, at 18�C, the values of the equili-brium concentration of PhCOCl were (1.16, 4.73, 7.09, and 10:3Þ � 10�4 M for½PNO�iaq ¼ ð0, 2.00, 4.00, and 8:00Þ � 10�4 M, respectively. In the uncatalyzed reaction,PhCOCl could be produced via the following reaction:

PrCOClorg þ PhCOOHorg Ð PhCOClorg þ PrCOOHorg ð98Þ

FIG. 5 Inverse phase transfer catalysis: Hammett plot for the pyridine 1-oxide-catalyzed reactions

of benzoyl chlorides and benzoate ions.


For ½PhCOCl�iorg ¼ 1:00� 10�3 M, the measured values of the equilibrium constantof reaction (98) at 18�C were (7.92 and 7:73Þ � 103 for ½PrCOOH�iorg ¼ 0:500 and 1.00M,respectively. The equilibrium concentration of PhCOCl was increased considerably by thepresence of a relatively small amount of Cl� ion (added as PhCH2Et3N

þCl�), in contrastto the insignificant effect of Cl� ion in the PhCOCl/PrCOONa system. The experimentalresults indicated that the mechanism of this system was very complicated. A simplifiedmechanistic description is shown in the following reaction steps:

PhCOClorg þ PNOorg Ð PhCOONPþCl�org ð85ÞPrCOClorg þ PNOorg Ð PrCOONPþCl�org ð99ÞPhCOONPþCl�aq þ PrCOO�aq Ð PhCðO�ÞðOCOPrÞðONPþÞCl�aq ð100ÞPrCOONPþCl�aq þ PhCOO�aq Ð PrCðO�ÞðOCOPhÞðONPþÞCl�aq ð101ÞPhCðO�ÞðOCOPrÞðONPþÞaq Ð PrCðO�ÞðOCOPhÞðONPþÞaq ð102ÞPhCðO�ÞðOCOPrÞðONPþÞaq Ð PhCOOCOPraq þ PNOaq ð103ÞPrCðO�ÞðOCOPhÞðONPþÞaq Ð PrCOOCOPhaq þ PNOaq ð104ÞPhCOONPþaq þ PhCOO�aq Ð ðPhCOÞ2Oaq þ PNOaq ð87ÞPrCOONPþaq þ PrCOO�aq Ð ðPrCOÞ2Oaq þ PNOaq ð105ÞReactions (100)–(104) play an important role in this reversible reaction. The beha-

vior of the reaction of mixed benzoic butanoic abhydride, PhCOOCOPr, and PNO issimilar to that of the acylation reaction of benzene catalyzed by AlCl3 [197].Furthermore, it is generally believed that the exchange reaction of acyl halide (RCOX)and carboxylic acid (R 0COOH) in a homogeneous organic medium takes place via a mixedacid anhydride intermediate:

RCOXþR 0COOHÐ ðRCOOCOR 0 þHXÞ Ð RCOOHþR 0COX ð106ÞThe reactivity of RCOCl is increased by an electron-withdrawing substituent and

decreased by an either electron-donating or a steric-hindered substituent. Similar argu-ments are applicable to acid anhydride (RCOOCOR 0). Acid anhydrides are generallymore stable than the related acyl chloride. As reported by Ugi and Beck [198], the relativereactivities of RCOCl toward hydrolysis in 89% aqueous acetone at �20�C were Cl3C(9200), ClCH2 (1.9), CH3 (1.0), C2H5 (0.69), n-C3H7 (0.54), (CH3Þ2CH (0.41), (CH3Þ3C(0.068), PhCH2 (0.33), and Ph (0.0038). As reported by Bunton et al. [199], the relativereactivities of RCOOCOR 0 toward hydrolysis in dioxane/water at 25�C were (CH3COÞ2ð1:0Þ > CH3COOCO Ph ð0:74Þ > ðPhCOÞ2O (0.033). In general, acid anhydrides withsmall alkyl groups are unstable and those with phenyl groups are considerably morestable. Thus, the order of relative reactivities is PrCOCl > PhCOCl > ðPrCOÞ2O >PhCOOCOPr > ðPhCOÞ2O, which is supported by experimental results [188]. Mixedacid anhydrides are unstable and prone to disproportionation and/or decomposition inthe presence of carboxylic acids and carboxylates. Wong and Jwo [200] studied theexchange reaction of symmetric benzoic and 2-chlorobenzoic anhydrides in CHCl3 toproduce mixed benzoic 2-chlorobenzoic anhydride. It was found that the reactionrate was slow and varied abnormally with the concentration ratio of½ðPhCOÞ2O�=½ð2-ClC6H4COÞ2O]. The reaction was promoted substantially by PNO andbenzoate salts. When 2-(ClC6H4COÞ2O was the limiting reactant, the order of reactivities


for promoting the exchange reaction was (PhCOOLi, PhCOONBu4Þ > PhCOONa> PhCOOH. Unexpectedly, the exchange reaction did not follow simple second-orderor pseudo-first-order kinetics. However, in the presence of PNO or benzoate salt, thereaction did follow pseudo-first-order kinetics under pseudo-order reaction conditions.Therefore, the mechanism of this exchange reaction was more complicated than expectedand the complex formation between two acid anhydrides could play a key role [200].

An interesting reversible phenomenon was also observed in the PNO-catalyzed reac-tion of nitrobenzoyl chlorides and the corresponding nitrobenzoate ions in H2O=CH2Cl2medium [196]. In the absence of PNO and carboxylate salt, a complete hydrolysis reactionwas observed for 2-, 3-, or 4-NO2C6H4COCl, whereas it reached an equilibrium in thepresence of PNO. The PNO-catalyzed reactions of 2-, 3-, and 4-NO2C6H4COCl andPhCOONa to synthesize mixed acid anhydride were unsuccessful. However, symmetric4-(NO2C6H4COÞ2O could be obtained with low yield, e.g., for [NO2C6H4COCl�iorg ¼0:010M, NO2C6H4COONa�iaq ¼ 0:50M, and ½PNO�iaq ¼ 4� 10�4 M. The reactionrapidly reached an equilibrium state with the yield of 4-(NO2C6H4COÞ2O being about40% [196].

D. Two-Phase Wittig Reactions

The Wittig reaction is one of the most important reactions in organic chemistry for thesynthesis of alkenes with unambiguous positioning of the C––C double bond. A compre-hensive review was made by Maryanoff and Reitz [201]. Maerkl and Merz [202] demon-strated that the Wittig reactions could be carried out in organic solvent/NaOH(aq)medium, in which ylides were generated by the reactions of quaternary alkyltriphenylpho-sphonium salts and NaOH in the aqueous phase, and then transferred into the organicphase to react with aldehydes to produce alkenes. One drawback of this two-phase Wittigreaction was the decomposition of quaternary phosphonium salts by NaOH(aq) to tri-phenylphosphine oxide, which depended on the solvent [CH2Cl2 � ðC6H6, n-C6H14Þ > nosolvent] and seemed to be catalyzed by Bu4N

þX� salts (X ¼ Cl > Br > I) [203]. Typicaltwo-phase Wittig reactions were performed in the following system: K2CO3ðsÞ=C6H6,Kþtert-BuO�(s)/C6H6, 50%NaOH(aq)/CH2Cl2, NaOH(s)/C6H6, and KF(s)/C6H6 orCH2Cl2 [204].

Based on the mechanistic aspect, two-liquid phase Wittig reactions cannot becounted as phase transfer catalyzed reactions, since it has been argued that phosphoniumsalts themselves are known to be PTC catalysts and the resulting ylenes are neutral speciesthat can diffuse into the organic phase without the assistance of a catalyst. However, theclosely related two-liquid phase Wittig–Horner and Horner–Emmons reactions are cata-lyzed by quaternary ammonium salts or crown ethers and are considered as PTC reac-tions. Therefore, it is useful to extend the concept of PTC to include two-phase Wittigreactions [10,18]. The two-phase Wittig reaction with NaOH(aq) is generally limited toaldehydes. Most of the studies on Wittig reactions were carried out homogeneously inorganic solvents such as THF, C6H6, CCl4, CHCl3, DMF, and CH3OH. In contrast, lesswork has been reported for the heterogeneous Wittig reactions. Jwo and coworkers[205,206] investigated the two-phase Wittig reactions of various benzyltriphenylphospho-nium salts (Ph3P

þCH2Ph0X�) and benzaldehydes (Ph 00CHO) in various organic solvent/

NaOH(aq) media (Fig. 6), focusing on the effects of substituents and organic solvents.This reaction system was chosen for study because of the convenience for two-liquid phasereactions and the versatility of varying the organic solvents and the substituents on the


ylide and benzaldehyde for studying their effects on the Z/E ratio of the product stilbene.These systems can be described by the following main reactions:

Ph3PþCH2Ph

0aq þOH�aq! ðPh3Pþ �CHPh 0 $ Ph3P ¼ CHPh 0Þaq ð107Þ

ðPh3Pþ �CHPh 0 $ Ph3P ¼ CHPh 0Þaq Ð ðPh3P ¼ CHPh 0Þorg ð108ÞðPh3P ¼ CHPh 0org þ Ph 00CHO! Ph 0CH ¼ CHPh 00org þ Ph3POorg ð109ÞThe mechanism of the homogeneous Wittig reaction [Eq. (109)] was generally

expressed in terms of two main steps: (1) the nucleophilic addition of the phosphorusylide to the carbonyl group to give intermediates (threo-betaine ¼ E form oxaphosphe-tane; erythro-betaine ¼ Z form oxaphosphetane), and (2) the irreversible decomposition ofthe intermediates to yield Z and E forms of alkene and phosphine oxide. The stereo-selectivity of the Wittig reaction is highly dependent on the substituents bonds to theylidic carbon and to the phosphorus atom, and on the reaction conditions, especiallythe organic solvent. In general, three categories of phosphonium ylides, namely, nonsta-bilized, semistabilized, and stabilized ylides, are classified. The Wittig reaction has beenshown to produce preferentially the thermodynamically stable E alkenes for stabilizedylides having strongly conjugating substituents such as COOMe or CN group; mixturesof the E and Z alkenes for semistabilized ylides bearing mildly conjugating substituentssuch as phenyl, vinyl, or allyl groups, and mainly contrathermodynamic Z alkenes for

FIG. 6 (a) Two-phase Wittig reaction mechanisms; (b) Organic phase reaction of

triphenylphosphonium ylide and benzaldehyde.


nonstabilized ylides lack such conjugating functionalities such as an alkyl group. In astudy of the two-phase Wittig reaction of Ph3P

þCH2Ph0X� salt and benzaldehydes

(Ph 00CHO) in organic solvent/NaOH(aq) medium [206], the substituents chosen forstudy included CH3, F, Cl, Br, CH3O, NO2, and CF3 and the organic solvents includedpolar solvents (CHCl3 and CH2Cl2) and nonpolar solvents (n-C6H14, C6H6, and CCl4).The main conclusions are summarized as follows:

1. The reaction rate was fast and independent of the agitation speed.2. The reaction of the intermediate, benzylidenetriphenylphosphorane

(PhP3 ¼ CHPh 0) and Ph 00CHO in the organic phase, was the decisive step responsiblefor the stereoselectivity.

3. In general, the yield and the Z/E ratios depended insignificantly on the concen-trations of phosphonium salt, Ph 00CHO, and NaOH, or agitation speed and temperature.

4. The substituents of Ph 00CHO exhibited considerably greater effects on the Z/Eratios of the product stilbene than those on the Ph3PCH2Ph

0þ ions. Therefore, the Z/Eratios of stilbene could change substantially by interchanging the substituents on thebenzyl group of the phosphorus atom and on the phenyl group of Ph 00CHO, even thoughthe product stilbene was the same. For example, the Z/E ratios were 1.4 and 3.3 for the (2-ClC6H4CH2PPh

þ3 -3-ClC6H4CHO) and (3-ClC6H4CH2PPh

þ3 -2-ClC6H4CHO) reactions in

NaOH(aq)/CH2Cl2 medium, respectively. The Z/E ratios were 4.5 and 1.8 for the (2-CH3C6H4CH2PPh

þ3 -2-BrC6H4CHO) and (2-BrC6H4CH2PPh

þ3 -2-CH3C6H4CHO) reac-

tions in NaOH(aq)/CH2Cl2 medium, respectively. The Z/E ratios were 6.8 and 1.5 forthe [2,5-(CH3Þ2C6H3CH2PPh

þ3 -2-ClC6H4CHO] and [2-ClC6H4CH2PPh

þ3 -2;5-ðCH3Þ2

C6H3CHO] reactions in C6H6/NaOH(aq) medium, respectively.5. In contrast to the meta- and para-substituted Ph 00CHO and the ortho-substituted

benzylidene ylide (Ph3P––CHPh 0), the ortho-substitued Ph 00CHO bearing heteroatom sub-stituent exhibited a pronounced enhancement for the Z selectivity with the order of effec-tiveness of substituents being CF3 > ðCl; BrÞ > CH3O > F > NO2. For example, for thereactions of 2,5-(CH3Þ2C6H3CH2PPh

þ3 ion and Ph 00CHO, the Z/E ratios were 0.93, 1.2,

2.3, 3.6, 0.71, 0.75, 4.0, 1.1, 3.6, 5.7, and 0.08 in NaOH(aq)/CH2Cl2 medium; and 0.63,0.84, 3.4, 6.8, 0.43, 0.41, 6.0, 2.0, 5.2, 9.6, and 0.04 in C6H6/NaOH(aq) medium forPhCHO, 2,5-(CH3Þ2C6H3CHO, 2-FC6H4CHO, 2-ClC6H4CHO, 3-ClC6H4CHO, 4-ClC6H4CHO, 2-BrC6H4CHO, 2-NO2C6H4CHO, 2,3,4-(CH3OÞ3C6H2CHO, 2-CF3C6H4CHO, and 2,6-Cl2C6H3CHO, respectively. This abnormal ortho effect was gen-erally rationalized by invoking the through space 2p–3d overlap effect. To avoid stericrepulsion, the phenyl group of the aldehyde should point away from the ylide and theC–P–O–C dihedral angle could then be varied freely. Assuming other things being equal,the complex would be expected to render a near 1 : 1 mixture of Z- and E-ozapho-sphetane, leading to a near 1 : 1 mixture of Z- and E-stilbenes. However, for an ortho-substituted Ph 00CHO bearing substituent such as F, Cl, Br, CH3O, or CF3, the Z selectivityof the oxaphosphetane could become more favorable due to chelate stabilization since thephosphorus atom (adopting a hypervalent octahedral structure) would co-ordinate to two-electron donating atoms, the carbonyl oxygen atom and the heteroatom of the ortho-substituent. However, in a 2,6-Cl2C6H3CHO system, the steric effect of two ortho-chlorosubstituents lowered the Z/E ratio substantially.

6. The effects of solvents depended somewhat on the substituent on the benzylgroup of the phosphonium ion. For XC6H4CH2PPh

þ3 (X ¼ H, CH3, F, Cl, Br,) ions,

the more polar solvents (CHCl3 and CH2Cl2) generally exhibited more favorable Z selec-tivity (CHCl3 > CH2Cl2). In contrast, for the ðCH3Þ2C6H3CH2PPh

þ3 ion, the nonpolar


solvents (CCl4, C6H6, and n-C6H14) could become more favorable for Z selectivity. Forexample, theZ/E ratioswere 1.2, 1.5, 1.2, 1.8, and 4.0 for the 2-ClC6H4CH2PPh

þ3 -2;5-ðCH3Þ2

C6H3CHO reaction system and 7.4, 6.8, 5.1, 3.6, and 5.8 for the 2;5-ðCH3Þ2C6H3CH2PPhþ3

-2-ClC6H4CHO reaction system with the organic phases being CCl4, C6H6, n-C6H14,CH2Cl2, and CHCl3, respectively.

7. The concerted asynchronous cycloaddition mechanism involving a four-centeredtransition state was suggested to be operating in these systems.

E. Asymmetric and Thermoregulated Phase Transfer Catalyses

Two novel methodologies termed ‘‘assymmetric phase transfer catalysis’’ and ‘‘thermo-regulated phase transfer catalysis’’ have been developed readily in the past decade andhave broadened greatly the scope of application of PTC. Therefore, it is worthwhile brieflydiscussing these two techniques.

1. Asymmetric Phase Transfer Catalysis

The use of optically resolved PTC catalysts for the synthesis of enantiomerically purecompounds is no doubt an attractive field. Asymmetric PTC has become an importanttool for both laboratory syntheses and industrial productions of enantiomerically enrichedcompounds. Recently, Lygo and coworkers [207–216] reported a new class of Cinchonaalkaloid-derived quaternary ammonium PTC catalysts, which have been applied success-fully in the enantioselective synthesis of �-amino acids, bis-�-amino acids, and bis-�-amino acid esters via alkylation [207–213] and in the asymmetric epoxidation of �=�-unsaturated ketones [214–216].

(a) Asymmetric �-Amino Acids. Chiral �; �-dialkyl-�-amino acids are an importantclass of noncoded amino acids in the design and the synthesis of modified peptides.Naturally occurring bis-�-amino acids such as dityrosines, isotyrosine, and meso-diami-nopimelic acid may act as cross-linking agents for stabilizing structural polymer ele-ments in plants and bacteria, and iodityrosine is a key structural subunit in a large classof bioactive peptides. O’Donnell et al. [217] first reported the use of chiral quaternaryammonium salts derived from Cinchona alkaloids as PTC catalysts for the asymmetricalkylation of amino acid imine esters to promote enantioselectivity. Lygo andWainwright [207] synthesized a class of Cinchona alkaloid-derived PTC catalysts viaquaterization of cinchonine, cinchonidine, dihydrocinchonine, dihydrocinchonidine, qui-nidine, and quinine using 9-chloromethylanthracene (Fig. 7). It was found that for theasymmetric phase transfer catalyzed alkylation of glycineimine in toluene/50%NaOH(aq), the quinidine- and quinine-derived catalysts were the least effective andthose derived from dihydrocinchonidine gave the best selectivity [94% enantiomericexcess (ee)]. N-Anthracenylmethylcinchonidinium chloride catalyzed (1) the PTC reac-tion of the enantio- and diastereo-selective synthesis of a series of bis-amino acids andesters with high enantioselectivity (> 95% ee) in toluene/50% NaOH(aq) medium [208–209]; (2) the PTC reactions of the alkylation of a series of alanine-derived imines withup to 87% ee in solid K2CO3=KOH medium [210]; and (3) the PTC alkylation of a ser-ies of benzophenone-derived glycineimines in toluene/50% KOH(aq) with up to 95% ee[213].

For environmental considerations (green chemistry), the organic solvent in thesesystems was generally toluene rather than CH2Cl2. However, the PTC alkylation of aseries of benzophenone-derived glycineimines catalyzed by N-anthracenylmethylcinchoni-


dinium chloride showed similar results using toluene of CH2Cl2 as the organic solvent[212]. In these systems, the alkylation reaction proceeded via ion-pair formation of thecarbanions and the chiral cinchona alkaloid-derived quaternary ammonium salts, whichrelied on the structure of the catalyst to promote the enantioselectivity. Lygo et al. [211]probed the role of key structural elements of a series of chiral cinchona alkaloid-derivedquaternary ammonium catalysts and concluded that the N-anthracenylmethyl substituentwas the key structural element that led to substantially enhanced enantioselectivity oversmaller N-alkyl substituents, which also suggested that the 1-quinonlyl group present inthe parent alkaloid played a key role in enantioselectivity.

Belokon and coworkers [218,219] reported a novel class of chiral metal complexesfor the asymmetric synthesis of �-amino acids under PTC conditions. The use of (4S,5S)-2,2-dimethyl-�; �; � 0; � 0-tetraphenyl-1,3-dioxane-4,5-dimethanol (TADDOL) promotedthe asymmetric PTC C-alkylation of Schiff’s bases of alanine esters with up to 82% ee[218], in which TADDOL functioned as a chelating agent for the alkali metal ions andmade the ion pair of the metal-ion complex and carbanion soluble in the organic solvent.Recently, Belokon et al. [219] have tested a series of chiral metal complexes of (1R,2R or1S,2S)-[N,N 0-bis-(2 0-hydroxylbenzylidene)-1,2-diaminocyclohexane (salen) as catalystsfor the C-alkylation of Schiff’s bases of alanine and glycine esters with alkyl bromidesunder PTC conditions in toluene/50% NaOH(aq) medium (Fig. 7) and found that theoptimal catalyst, (salen)Cu(II) complex gave �-amino and �-methyl-�-amino acids with eeof 70–96%.

(b) Asymmetric �; �-Epoxy Ketones. Recently, the enantioselective epoxidation of�; �-unsaturated ketones has received much attention. Lygo and coworkers [214–216]have investigated the enantioselective epoxidation of various �; �-unsaturated ketonesutilizing chiral Cinchona alkaloid-derived quaternary ammonium salts (e.g., N-anthrace-nylmethylcinchodinium salts) as PTC catalysts in conjunction with sodium hypochlorite,in which up to 90% ee could be obtained. A study on the factors affecting the rate ofthis reaction system suggested that ion exchange between the catalyst and sodium hypo-chlorite is the rate-determining step [216].

2. Thermoregulated Phase Transfer Catalysis

Recently, efforts to achieve facile catalyst/product separation in two-liquid phase transfercatalyzed systems have received considerable attention. Horvath and Rabai [220] devel-

FIG. 7 Asymmetric phase transfer catalysis: alkylation of glycine imine esters.


oped a fluorous two-phase system, based on the limited miscibility of partially or fullyfluorinated compounds with a nonfluorinated parent compound. Bianchini et al. [221]reported another method that is based on the solubility gap of metal–sulfur complexesin a two-phase methanol/hydrocarbon system. Both of these systems could become one-phase systems at appropriate higher reaction temperatures. However, after completion ofthe reaction, the reaction solution was cooled down to room temperature and then sepa-rated into two phases again, which facilitated the recovery of the catalyst.

Jin et al. [222] reported the synthesis of a series of novel polyether-substituted tri-phenylphosphines [4-HOðCH2CH2OÞn-C6H4�m-PPh3�m, m ¼ 1; 2 or 3] (PETPPs). PETPPsexhibited inverse temperature-dependent solubility in water and their Rh(III) complexescould act as PTC catalysts in thermoregulated PTC, which was successfully applied to thetwo-phase hydroformylation of higher alkenes (C6–C12) such as 1-dodecene in toluene/H2O medium (Fig. 8) with about 95% conversion and 85% aldehyde selectivity. Thisprocess can generally be described as follows. At room temperature, the Rh(III)–PETPP catalyst remains mainly in the aqueous phase. However, at a temperature higherthan the cloud point (or the critical solubility temperature), the catalyst precipitates fromwater and transfers into the organic phase, where it catalyzes the hydroformylation reac-tion of (CO=H2) and alkenes to produce aldehydes. After the reaction is complete, thesystem is cooled down to room temperature and the catalyst returns to the aqueous phase.Therefore, a simple phase separation allows continuous reuse of the catalyst. The inversetemperature-dependent solubility phenomenon of Rh(III)–PETPP catalyst in water isattributed mainly to the cleavage of the hydrogen bonds between the polyether chainsand water molecules on heating. Jin and coworkers [223–228] have also developed a seriesof nonionic water-soluble phosphine ligands bearing polyethylene moieties and appliedthem successfully in the hydroformylation of higher alkenes via the attractive thermore-gulated PTC.

V. POSTSCRIPT

In this chapter, an overview of fundamentals and selected systems of PTC is presented.The development of PTC has followed the scientific trend that a successful PTC applica-tion frequently stimulates further research that in turn leads to more applications andimproved processes. The growth of PTC is also accelerated by its applications in thechemical industry. Today, PTC has grown to become a very important and widely applied

FIG. 8 Thermoregulated phase transfer catalysis: hydroformylation of higher olefins (C6–C12).


methodology in organic syntheses. No doubt numerous novel catalysts, methodologies,and new applications based on PTC wait for discovery and exploration.

REFERENCES

1. G Wittig, M Heintzeler, MH Wetterling. Annalen 557:201–205, 1947.

2. MJ Jarrousse. Compt Rend Acad Sci, Ser C 232:1424–1426, 1951.

3. NA Gibson, JW Hosking. Aust J Chem 18:123–125, 1965.

4. CM Starks, DR Napier. Fr Pat 1573164, 1965.

5. CM Starks. J Am Chem Soc 93:195–199, 1971.

6. M Makosza, B Serafinowa. Rocz Chem 39:1223–1230, 1965.

7. A Brandstrom, K Gastavii. Acta Chem Scand 23:1215–1218, 1969.

8. WP Weber, GW Gokel. Phase Transfer Catalysis in Organic Synthesis. Berlin, Heidelberg,

New York: Springer-Verlag, 1977.

9. A Brandstrom. Advances in Physical and Organic Chemistry. vol. 15. London, New York:

Academic Press, 1977, pp 267–340.

10. CM Starks, C Liotta. Phase Transfer Catalysis, Principles and Techniques. London, New

York: Academic Press, 1978.

11. M Makosza. In: AT Scott, ed. Survey of Progress in Chemistry. vol. 9. New York: Academic

Press, 1980, p 1.

12. M Makosza, M Fedorynski. Polish J Chem 70:1093–1110, 1996.

13. F Montanari, D Landini, F Rolla. Top Curr Chem 101:147–200, 1982.

14. EV Dehmlow, SS Dehmlow. Phase Transfer Catalysis, 2nd ed. Weinheim: Verlag Chemie,

1983.

15. M Makosza, M Fedorynski. Advances in Catalysis. vol. 35. 1985, pp 375–422.

16. HH Fredman. Pure Appl Chem 58:857–868, 1986.

17. Y Goldberg. Phase Transfer Catalysis: Selected Problems and Applications, Gordon & Breach

Science Publishers (Switzerland, Australia, Belgium, France, Germany, Great Britain, India,

Japan, Malaysia, Netherlands, Russia, Singapore, USA), 1992.

18. EV Dehmlow. SS Dehmlow, Phase Transfer Catalysis. New York: VCH, 1993.

19. CM Starks, CL Liotta, M Halpern. Phase-Transfer Catalysis, Fundamentals, Applications,

and Industrial Perspectives. Chapman & Hall, London, Great Britain, (Weinheim, New York,

Tokyo, Melbourne, Madras), 1996.

20. LI Osipow. Surface Chemistry, Theory and Industrial Applications. Huntington, NY: RE

Krieger, 1972.

21. RJ Hunter, Foundations of Colloid Science. vol. I. Oxford: Oxford Science Publications,

1989, ch. 5.

22. R Aveyard, DA Haydon. Introduction to the Principles of Surface Chemistry. Cambridge:

Cambridge University Press, 1973.

23. S Ross, ES Chen. In: DE Guschee, ed. Chemistry and Physics of Interface. Washington, DC:

American Chemical Society, 1965, pp 44–56.

24. TK Sherwood, RL Pigford, CR Wilke. Mass Transfer. New York: McGraw-Hill, 1975,

chs 5, 8.

25. WG Whitman. Chem Met Eng 29:146–148, 1923.

26. WK Lewis, WG Whitman. Ind Eng Chem 16:1215–1220, 1924.

27. S Hatta. Tech. Rep. Tohoku Imperial U. 8:1–25, 1928; J Soc Chem Ind 31: 869–879, 1928.

28. LP Hammett. Physical Organic Chemistry: Reaction Rates, Equilibria, and Mechanisms. New

York: McGraw-Hill, 1970.

29. GA Olah, PVR Schleyer, eds. Carbonium Ions. vol. 1. New York: Interscience, 1968.

30. M Szwarc, ed. Ions and Ion Pairs in Organic Reactions. vols. 1, 2. New York: Wiley-

Interscience, 1972.


31. R Modin, G Schill. Acta Pharm Suec 4:301–326, 1967.

32. R. Modin. Acta Pharm Suec 9:157–166, 1972.

33. A Brandstrom. Preparative Ion Pair Extraction, An Introduction to Theory and Practice.

Stockholm: Apotekars ocieteten/Hassle Lakemedel, 1974.

34. K Gustavii. Acta Pharm Suec 4:233–246, 1967.

35. Y. Inoue, O Tochiyama. Bull Chem Soc Jpn 53:1618–1624, 1980.

36. CJ Coetzee, H Freiser. Anal Chem 41:1128–1130, 1969.

37. HK Bisiwas, BM Mandal. Anal Chem 44:1636–1640, 1972.

38. EV Dehmlow, B Vehre. J Chem Res (S) 350–351, 1987.

39. AW Herriott, D Picker. J Am Chem Soc 97:2345–2349, 1975.

40. J De la Zerda, Y Sasson. J Chem Soc Perkin Trans II, 1147–1151, 1987.

41. EV Dehmlow, R Thieser, Y Sasson, E Pross. Tetrahedron 41:2927–2932, 1985.

42. WF Luder, PB Kraus, RM Fuoss. J Am Chem Soc 58:225–258, 1936.

43. ED Hughes, CK Ingold, S Patai. Y Pocker. J Chem Soc 1206–1219, 1957.

44. H Eyring. J Chem Phys 3:107–115, 1935.

45. MG Evans, M Polanyi. Trans Faraday Soc 31:875–894, 1935.

46. ED Hughes, CK Ingold. J Chem Soc 244–255, 1935.

47. CK Ingold. Structure and Mechanism in Organic Chemistry. 2nd ed. London: Bell, 1969.

48. MJ McDowell, CA Draus. J Am Chem Soc 73:3293–3296, 1951.

49. CM Starks, RM Owens. J Am Chem Soc 95:3613–3617, 1973.

50. A Brandstrom, H. Kolind-Anderson. Acta Chem Scand Ser B 29:201–205, 1975.

51. J Ugelstad, T Ellingsen, A Berge. Acta Chem Scand 20:1593–1598, 1966.

52. D Landini, A Maia, A Rampoldi. Gazz Chim Ital 119:513–517, 1989.

53. D Landini, A Maia, G. Podda. J Org Chem 47:2264–2268, 1982.

54. N Kahana, A Deshe, A Warshawsky. J Polym Sci Chem Ed 23:231–253, 1985.

55. L Loupy, G Bram, J Sansoulet. New J Chem 16:233–242, 1992.

56. Y Yuan, D Gao, Y. Jiang. Synth Commun 22:2117–2119, 1992.

57. M Halpern. PhD thesis, Hebrew University of Jerusalem, 1983.

58. M Halpern, Z Lysenko. J Org Chem 54:1201–1203, 1989.

59. J von Braun. Justus Liebigs Ann Chem 382: 1–49, 1911; Chem Abstr: 5:3446.

60. HJM Dou, R Gallo, P. Hassanaly, J Metzger, J Org Chem 42:4275–4276, 1977

61. TH Lane, J Speier. J Org Chem 41:2714–2716, 1976.

62. D Landini, A Maia, A Rampoldi. J Org Chem 51:5475–5476, 1986.


64. EV Dehmlow, M Slopianka, J. Heider. Tetrahedron Lett 2361–2364, 1977.

65. T Ho. Hard and Soft Acids and Bases; Principles in Organic Chemistry. New York: Academic

Press, 1977.

66. M Halpern, Y Sasson, M Rabinovitz. Tetrahedron 38:3183–3187, 1983.

67. D Mason, S Magdassi, Y Sasson. J Org Chem 55, 2714–2717, 1990.

68. CL Liotta. In: RM Izatt, JJ Christensen, eds. Synthetic Multidentate Macrocyclic

Compounds. New York: Academic Press, 1978, pp 111.

69. VV Yakushkina, VM Abashkin, BN Laskorin. Dokl Akad Nauk Ser Khim 252:373, 1980.

70. CL Liotta, EE Grisdale, HP Hopkins Jr. Tetrahedron Lett 4205–4208, 1975.

71. F de Jong, DN Reinhoudt, CJ Smit. Tetrahedron Lett 1371–1378, 1976.

72. GW Gokel, SH Korzeniowski. Macrocyclic Polyethers Syntheses. Berlin, Heidelberg, New

York: Springer Verlag, 1982.

73. GW Gokel. In: JF Stoddart, ed. Crown Ethers and Cryptands. Monographs in

Supramolecular Chemistry, no. 3. London: Royal Society of Chemistry, 1991.

74. Y Inoue, GW Gokel. Cation Binding by Macrocycles. New York, Basel: Marcel Dekker,

1990.

75. GW Gokel, DM Goli, RA Schultz. J Org Chem 48:2837–2842, 1983.

76. JM Harris, NH Hundley, TG Shannon, EC Struck. J Org Chem 47:4789–4791, 1982.

77. JM Harris, MG Case. J Org Chem 48:5390–5392, 1983.


78. R Neumann, S Dermeik, Y Sasson. Israel J Chem 26:239–242, 1985.

79. R Neumann, Y Sasson: J Mol Catal 31:81–88, 1985.

80. P Lavelle. Spec Chem 6:16, 18, 20, 1986; Chem Abstr 105: 208179.

81. AK Bose, P Mangiaracina. Tetrahedron Lett 28:2503–2506, 1987.

82. SL Regen. Angew Chem 91:464–472, 1979; Chem Abstr: 91:44980c.

83. M Tomoi, WT Ford. In: DC Sherrington, P Hodge, eds. Syntheses and Separations Using

Functional Polymers. New York: John Wiley, 1988.

84. M Tomoi, WT Ford. J Am Chem Soc 103:3821–3828, 1981.

85. T Ido, S Hajime, K Sakurai, S Goto. Kagaku Kogaku Ronbunshu 10:287–293, 1984; Chem

Abstr 101:90171.

86. WT Ford, J Lee, M Tomoi. Macromolecules 15:1246–1251, 1982.

87. M. Tomoi, N. Kori, H Kakiuchi. Makromol Chem 187:2753–2761, 1986.

88. MS Chiles, DD Jackson, PC Reeves. J Org Chem 45:2915–2918, 1980.

89. F Montanari, S Quici, P Tundo. J Org Chem 48:199–202, 1983.

90. CL Lin, TJ Pinnavaia. Chem Mater 3:213–215, 1991.

91. TJ Pinnavaia, CL Lin. US Pat 5,099,054, 1992; Chem Abstr 117:51277.

92. HD Nguyen, NB Hiep, TNH Lee, PNS Chu. CR Acad Sci Ser 2 300:799–802, 1985.

93. RA Sawicki. US Pat 4,421,675, 1983; Chem Abstr 100: 74762.

94. P Tundo, P Venturello, E Angeletti. J Am Chem Soc 104:6547–6551, 1982.

95. R Neumann, Y Sasson. J Org Chem 49:3448–3451, 1984.

96. R Nouguier, M Mchich. Tetrahedron 44:2477–2481, 1988.

97. DH Wang, HS Weng. Chem Eng Sci 43:2019–2024, 1988.

98. D Mason, S Magdassi, Y Sasson. J Org Chem 56:7229–7232, 1991.

99. J Correia. J Org Chem 57:4555–4557, 1992.


101. DH Wang, HS Weng. J Chin Inst Chem Eng 27:419–426, 1996.

102. HC Hsiao, HS Weng. Ind Eng Chem Res 38:2911–2918, 1999.

103. HC Hsiao, HS Weng. J Chin Inst Chem Eng 31:573–583, 2000.

104. HS Weng, CM Wang, DH Wang. Ind Eng Chem Res 36:3613–3618, 1997.

105. HM Yang. J. Chin. Inst Eng 21: 399–408, 1998.

106. JJ Krueger, MD Amiridis, HJ Ploehn. Ind Eng Chem Res 40:3158–3163, 2001.

107. S Kondo, Y Takeda, K Tsuda. Synthesis 11:862–864, 1989.

108. A Knoechel, J Oehler, G Rudolph. Tetrahedron Lett 3167–3170, 1975.

109. M Halpern, Y Sasson, M Rabinovitz. J Org Chem 48:1022–1025, 1983.

110. D Landini, A Maia, G Podda. J Org Chem 47:2264–2268, 1982.

111. SS Shavanov, GA Tolstikov, TV Shutenkova, GA Viktorov. Zh Obshch Khim 57:1587–1594,

1987; Chem Abstr 108: 55217.

112. T Ido, M Saiki, S. Goto. Kagaku Kogaku Ronbunshu 14:539–542, 1988; Chem Abstr 110:

38372.

113. T Ido, Y Matsuura, S Goto. Kagaku Kogaku Ronbunshu 14:174–181, 1988; Chem Abstr

109:72972.

114. JH Clark, DJ Macquarrie. Tetrahedron Lett 28:111–114, 1987.

115. AW Herriott, D Picker. Tetrahedron Lett 1511–1514, 1974.

116. AP Krapcho, JR Larson, JM Eldridge. J Org Chem 42:3749–3753, 1977.

117. MS Newman, HM Dali, WM Hung. J Org Chem 40:262–264, 1975.

118. GA Lee, HH Freedman. Tetrahedron Lett 1641–1644, 1976.

119. S Colonna, R Fornasier. Synthesis 531–532, 1975.

120. HE Hennis, JP Easterly, LR Collins, LR Thompson. Ind Eng Chem Prod Res Dev 6:193–195,

1967; Chem Abstr 68:21662.

121. M Makosza. Pure Appl Chem 43:439–462, 1975.


123. D Landini, A Maia, F Montanari. J Am Chem Soc 100:2796–2801, 1978.

124. LJ Mathias, RA Vaidya. J Am Chem Soc 108:1093–1094, 1986.


125. M Makosza. In: R Scheffold, ed. Modern Synthetic Methods. Zurich: Schweizerischer

Chemiker Verband,1976, pp 7–100.

126. D Landini, A Maia, F Montanari. J Chem Soc Chem Commun 112–113, 1977.

127. JR Chang, MY Yeh, YP Shih. J Chin Chem Soc 31:185–190, 1984.

128. ML Wang, HS Wu. J Org Chem 55:2344–2350, 1990.

129. JH Ramsden, RS Drago, R Riley. J Am Chem Soc 111:3958–3961, 1989.

130. JE Gordon, RE Kutina. J Am Chem Soc 99:3903–3909, 1977.

131. M Makosza, E Bialecka. Tetrahedron Lett 2:183–186, 1977.


133. M Halpern, Y Cohen, Y Sasson, M Rabinovitz. Nouv J Chim 8:443–444, 1984.


135. M Halpern, HA Zahalka, Y Sasson, M Rabinovitz. J Org Chem 50:5088–5092, 1985.

136. M Makosza, M Wawrzyniewicz. Tetrahedron Lett 4659–4662, 1969.

137. EV Dehmlow, M Lissel. J Chem Res (S) 310, 1978.

138. M Makosza, I Krylowa. Tetrahedron 55:6395–6402, 1999.

139. M Makosza, A Kacprowicz, M Fedorynski. Tetrahedron Lett 2119–2122, 1975.

140. EV Dehmlow. Tetrahedron Lett 91–94, 1976.

141. EV Dehmlow. T Remmler. J Chem Res (S) 72–73, 1977.

142. KH Wong, APW Wai. J Chem Soc Perkin Trans 11:317–321, 1983.

143. OI Danilova, IA Esikova, SS Yufit. Izv Akad Nauk SSSR, Ser Khim 314–316, 1988; Chem

Abstr 109:169773.

144. HA Yee, HJ Palmer, SH Chen. Chem Eng Prog 33–39, 1987.

145. CL Liotta, EM Burgess, CC Ray, ED Black, BE Fair. In: CM Starks, ed. Phase Transfer

Catalysis: New Chemistry, Catalyst and Applications. ACS Symposium Series no. 326.

Washington, DC: American Chemical Society, 1987, ch. 3.

146. CL Liotta, EM Burgess, ED Black. Polym Preprints 31:65, 1990.

147. M Ellwood, J. Griffiths, P Gregory. J Chem Soc, Chem Commun 181–183, 1980.

148. H Iwamoto, T Sonoda, H Kobayashi. Tetrahedron Lett 24:4703–4706, 1983.

149. H Kobayashi, T Sonoda, H Iwamoto. Chem Lett 1185–1186, 1982.

150. HM Li, PJ Liu, CR Han. Chin J Org Chem 21:356–364, 2001.

151. A Suzuji, T Nakata, W Tanaka. Jap Pat 70/10/26; Chem Abstr 73:44885, 1970.

152. MJH Russell. Platinum Met Rev 32:179–186, 1988; Chem Abstr 110:117096.

153. E Paetzold, G Oehme, B Costisella. Z Chem 29: 447–448, 1989; Chem Abstr 112:244852.

154. M Safi, D Sinou. Tetrahedron Lett 32:2025–2028, 1991.

155. H Ito, T Kusukawa, M Fujita. Chem Lett 598–599, 2000.

156. ML Bender, M Komiyama. Cyclodextrin Chemistry. Berlin: Springer, 1978.

157. AZ Trifonov, TT Nikiforov. J Mol Catal 24:15–18, 1984.

158. G Barak, Y Sasson. Bull Soc Chim Fr 584, 1988; Chem Abstr 111:96429.

159. A Horada, Y Hu, S Takahashi. Chem Lett 2083–2084, 1986.

160. H Zahalka, K Januszkiewicz, H Alper. J Mol Catal 35:249–253, 1986.

161. PA Ganeshpure, S Satish. J Chem Soc Chem Commun 981–982, 1988.

162. S Shimizu, Y Sasaki, C Hirai. Bull Chem Soc Jpn 63:176–178, 1990.

163. F Trotta. J Mol Catal 85:L265–267, 1993.

164. F Trotta, D Cantamessa, M Zanetti. J Incl Phenom Macrocycl Chem 37:83–92, 2000; Chem

Abstr 133: 177001.

165. N Tanaka, A Yamaguchi, Y Araki, M Araki. Chem Lett 715–718, 1987.

166. RS Hays, GP Hobbs, M Kourouma, JJ Ries, TM Ste Marie, SKM Sherwood, RG Landolt.

J Mol Catal A: Chemical 129: L123–125, 1998.

167. B Boyer, JF Betzer, G Lamaty, A Leydet, JP Roque. New J Chem 19:807–811, 1995.

168. B Boyer, A Hambardzoumian, G Lamaty, A Leydet, JP Roque, P Bouchet. New J Chem

20:985–988, 1996.

169. B Boyer, A Hambardzoumian, JP Roque. Tetrahedron, 55:6147–6152, 1999.

170. B Boyer, A Hambardzoumian, JP Roque, N Beylerian. Tetrahedron 56:303–307, 2000.


171. J Wild, N Goetz. Ger Offen DE 3,820,979, 1989; Chem Abstr 113:40150.

172. T Kawai. Brit UK Pat Appl GB 2,219,292, 1989; Chem Abstr 112:234804.

173. G Fleche. Starch/Staerke 42:31–35, 1990; Chem Abstr 112:120970.

174. T Ido, N Ohyama, S Goto, H Teshima. Kagaku Kogaku Ronbunshu 9:58–63, 1983; Chem

Abstr 98: 125171.

175. J Du, Huaxue. Shiji 15:183–189, 1993; Chem Abstr 120:133964.

176. WK Fife, Y Xin. J Am Chem Soc 109:1278–1279, 1987.

177. YL Wen, MY Yeh, YS Lee, YP Shih. J Chin Inst Chem Eng 27:427–436, 1996.

178. M Zeldin. US Pat 4,855,433, 1988; Chem Abstr 112:099456.

179. M Takeishi, K Se, N Umeta, R Sato. Nippon Kagaku Kaishi 824–829, 1992; Chem Abstr 117:

131693.

180. S Asai, H Nakamura, W Okada, M Yamada. Chem Eng Sci 50: 943–949, 1995.

181. ML Wang, CC Ou, JJ Jwo. Chem. Eng Commun 179: 233–252, 2000.

182. J March. Advanced Organic Chemistry, 2nd ed. New York: McGraw-Hill 1977, pp 227.

183. ML Wang, CC Ou, JJ Jwo. J Chin Chem Inst Eng 23:171–178, 2000.

184. H Nakamura, S Asai, M Yamada. Chem Eng Sci 51:1343–1346, 1996.

185. WF Fife, Z Zhang. J Org Chem 51:3746–3748, 1986.

186. CS Kuo, JJ Jwo. J Org Chem 57:1991–1995, 1992.

187. ML Wang, CC Ou, JJ Jwo. Ind Eng Chem Res 33:2034–2039, 1994.

188. ML Wang, CC Ou, JJ Jwo. Bull Chem Soc Jpn 67:2949–2955, 1994.

189. ML Wang, CC Ou, JJ Jwo. J Mol Catal A: Chemical 99:153–160, 1995.

190. ML Wang, CC Ou, JJ Jwo. Bull Chem Soc Jpn 68: 2165–2174, 1995.

191. JS Ou, RH Chen, JJ Jwo. J Mol Catal A: Chemical 110:95–103, 1996.

192. YM Liou, JJ Jwo. J Chin Inst Chem Eng 27:405–410, 1996.

193. ML Wang, CC Ou, JJ Jwo. Chem Eng Commun 165:151–165, 1998.

194. ML Hung, JJ Jwo. J Mol Catal A: Chemical 154:55–63, 2000.

195. YS Chang, JJ Jwo. J Mol Catal A: Chemical 160:357–366, 2000.

196. YL Lu, JJ Jwo. J Mol Catal A: Chemical 170:57–65, 2001.

197. WR Edwards Jr., EC Sibille. J Org Chem 28:674–679, 1963.

198. I Ugi, F Beck. Chem Ber 94:1839–1850, 1961.

199. CA Bunton, NA Fuller, SG Perry, VJ Shiner. J Chem Soc 2918–2926, 1963.

200. YJ Wong, JJ Jwo. J Mol Catal A: Chemical 132:155–164, 1998.

201. BE Maryanoff, AB Reitz. Chem Rev 89:863–927, 1989.

202. G Maerkl, A Merz. Synthesis 295–297, 1973.

203. HJ Cristau, A Long, H Christol. Tetrahedron Lett 349–352, 1979.

204. EV Dehmlow, S Barahona-Naranjo. J Chem Res (S) 142, 1981.

205. ML Wang, SR Lay, JJ Jwo. J Chin Inst Eng 12:293–300, 1989.

206. JJ Hwang, RL Lin, RL Shieh, JJ Jwo. J Mol Catal A: Chemical. 142:125–139, 1999.

207. B Lygo, PG Wainwright. Tetrahedron Lett 38:8595–8598, 1997.

208. B Lygo, J Crosby, JA Peterson. Tetrahedron Lett 40:1385–1388, 1999.

209. B Lygo. Tetrahedron Lett 40:1389–1392, 1999.

210. B Lygo, J Crosby, JA Peterson. Tetrahedron Lett 40:8671–8674, 1999.

211. B Lygo, J Crosby, TR Lowdon, PG Wainwright. Tetrahedron 57:2391–2402, 2001.

212. B Lygo, J Crosby, TR Lowdon, JA Peterson, PG Wainwright. Tetrahedron 57:2403–2409,

2001.

213. B Lygo, J Crosby, JA Peterson. Tetrahedron 57:6447–6453, 2001.

214. B Lygo, PG Wainwright. Tetrahedron Lett 39:1599–1602, 1998.

215. B Lygo, PG Wainwright. Tetrahedron 55:6289–6300, 1999.

216. B Lygo, DCM To. Tetrahedron Lett 42:1343–1346, 2001.

217. MJ O’Donnell, WD Bennett, S Wu. J Am Chem Soc 111:2353–2355, 1989.

218. YN Belokon, KA Kochetkov, TD Churkina, NS Ikonnikov, AA Chesnokov, OV Larionov,

VS Parmar, R Kmar, HB Kagan. Tetrahedron Asymmetry 9:851–857, 1998.


219. YN Belokon, M North, TD Churkina, NS Ikonnikov, VI Maleev. Tetrahedron 57:2491–2498,

2001.

220. IT Horvath, J Rabai. Science 266:72–75, 1994.

221. C Bianchini, P Frediani, V Sernau. Organometallics 14:5458–5459, 1995.

222. ZL Jin, XL Zheng, B Fell. J Mol Catal A: Chemical 116:55–58, 1997.

223. XL Zheng, JY Jiang, XZ Liu, ZL Jin. Catal Today 44:175–182, 1998.

224. RF Chen, XZ Liu, ZL Jin. J Organomet Chem 571:201–204, 1998.

225. JY Jiang, YH Wang, C Liu, FS Han, ZL Jin. J Mol Catal A: Chemical. 147:131–136, 1999.

226. RF Chen, JY Jiang, YH Wang, ZL Jin. J Mol Catal A: Chemical 149:113–117, 1999.

227. YH Wang, JY Jiang, R Zhang, XL Liu, ZL Jin. J Mol Catal A: Chemical 157:111–115, 2000.

228. JY Jiang, YH Wang, C Liu, QM Xiao, ZL Jin. J Mol Catal A: Chemical 171:85–89, 2001.


11Interfacial Mechanism and Kinetics ofPhase-Transfer Catalysis

HUNG-MING YANG National Chung Hsing University, Taichung, Taiwan,Republic of China

HO-SHING WU Yuan-Ze University, Taoyuan, Taiwan, Republic of China

I. INTRODUCTION

A. General Considerations

As the chemical reactants reside in immiscible phases, phase transfer (PT) catalysts havethe ability to carry one of the reactants as a highly active species for penetrating theinterface, into the other phase where the reaction takes place, and to give a high conver-sion and selectivity for the desired product under mild reaction conditions. This type ofreaction was termed ‘‘phase-transfer catalysis’’ (PTC) by Starks in 1971 [1]. Since then,numerous efforts have been devoted to the investigation of the applications, reactionmechanisms, and kinetics of PTC. Nowadays, PTC becomes an important choice inorganic synthesis and is widely applied in the manufacturing processes of specialty che-micals, such as pharmaceuticals, dyes, perfumes, additives for lubricants, pesticides, andmonomers for polymer synthesis. The global usage of PT catalysts was estimated at overone million pounds in 1996, and PTC in industrial utilization is continuously growing atan annual rate of 10–20% [2]. PTC is a very effective tool in many types of reactions, e.g.,alkylation, oxidation, reduction, addition, hydrolysis, etherification, esterification, car-bene, and chiral reactions [2,3].

1. Reaction Cycle of PTC

The first reaction scheme addressed by Starks in 1971 was for the reaction of aqueoussodium cyanide and organic 1-chloro-octane. In contrast with the result of no apparentreaction occurring after more than 24 h in the absence of catalyst, the cyanide displace-ment reaction takes place rapidly with only 1% of the quaternary ammonium salt(C6H13Þ4NþCl� added, and achieving near 100% yield of 1-cyano-octane product in2–3 h [1]. The reaction scheme for the PT catalyzed cyanide displacement reaction inaqueous–organic phases is shown in the following:

NaCN ðaqÞ þ 1-C8H17Cl ðorgÞQCl

! 1-C8H17CN ðorgÞ þNaCl ðaqÞ ð1Þ


ð2Þ

The PT catalyst QCl should first react with the cyanide anion to form the activeintermediate QCN, which is then transferred into the organic phase to react with theorganic reactant 1-C8H17Cl and is then regenerated back to QCl to conduct the nextcycle of reactions.

2. Classification of PTC Reactions

PTC reactions can be classified into two types: soluble PTC and insoluble PTC. Each typecan be further divided into several categories. Figure 1 shows the classification of PTCreactions. Insoluble PTC consists of liquid–solid–liquid PTC (LSLPTC) and tri-liquidPTC (TLPTC), by which the catalyst can be recovered and reused, showing the greatpotential in large-scale production for industry. The catalyst used in LSLPTC is immo-bilized on an organic or inorganic support, while in TLPTC it is concentrated within aviscous layer located between the organic and aqueous phases. Soluble PTC includesliquid–liquid PTC (LLPTC), solid–liquid PTC (SLPTC) and gas–liquid PTC (GLPTC).There are also nontypical PTC reactions termed inverse PTC (IPTC) and reverse PTC(RPTC), and these are different in catalyst type and transfer route, compared to normalPTC [2,3].

PT catalysts commonly used are quaternary onium salts (ammonium and phospho-nium), crown ethers, cryptands, and polyethylene glycols. The essential characteristics of aPT catalyst are that the catalyst must have the ability to transfer the reactive anion into theorganic phase to conduct the nucleophilic attack on the organic substrate, and effect acation–anion bonding loose enough to allow a high reaction rate in the organic phase.

FIG. 1 Classification of PTC reactions.


Other factors in selecting a suitable catalyst that we should consider include the cost andstructure of the catalyst, the toxicity of the catalyst and solvent, the ease of separation ofthe catalyst from the products, the energy requirement for reaction, the stability of thecatalyst in process conditions, and the ease of treatment of the waste streams, in order tolead to an efficient and economic PTC process.

3. Interfacial Characteristics

Since PTC reactions are carried out between immiscible phases, the nature of the inter-face and the physical properties of the reacting compounds at the interface become veryimportant in promoting the desired reaction rate at a satisfactory level. In a liquid–liquidsystem under agitation, one phase should be dispersed as small droplets in the secondphase in a manner such that a large interfacial area between the two phases can beobtained. The nature of the interface includes interfacial tension, the presence of surfac-tants, and the degree of agitation rate [3]. These three factors determine the sharpness ofthe interface (or the thickness of interfacial film), the droplet size, and the interfacialarea available to transfer the reacting anion. The interfacial behaviors of the reactinganion include the surface equilibrium distribution of the active intermediate, the ease ofpenetration of the compounds into the other phase (the depth from the interface), andthe mass transfer rate across the interface. Adding extra salts may induce a change in theproperties of the interface. For example, by adding more inorganic salts or bases, thecatalyst is salted out of the aqueous phase and an organic solvent with low polarity, andthe interfacial film grows increasingly thick, finally becoming a separate observablephase. This situation alters the original reaction zone and the apparent reaction ratebecause the properties of the interface have been changed. Hence, the thickness of theinterfacial film (sharpness) is not only limited by the nature of the interface itself, butalso affected by the introduction of other ingredients. Figure 2 shows the scheme of aconcentration gradient of the reacting compound within a dispersed organic dropletunder a slow or fast diffusion rate, which indicates that the organic reaction is con-ducted at the interface or in the whole droplet.

4. Reaction at the Interface and in the Bulk Solution

In PT catalysis, the reaction mechanisms that have been proposed are the Starks’ extrac-tion mechanism and Makosza’s interfacial mechanism. These two mechanisms describethe zone where the organic reaction occurs or the phase where the rate-determining step islocated. However, in reality, it is realized that many PTC reactions are conducted both atthe interface and in the bulk solution, especially for a reaction controlled by the intrinsicorganic reaction [3]. The distinction between these two mechanisms is recognized as thedifference in the depth of the reaction zone penetrating the organic phase.

Under the conditions of no agitation with a flat interface or slow agitation with aslow mass transfer rate, as the solubility of the transferred species in the organic phase issufficiently large, the rate of diffusion within the organic phase would not influence theobserved reaction rate significantly. Fast diffusion rates may exhibit extraction mechanismbehavior, while with a slow diffusion rate the system is suitably described by an interfacialmechanism. In other words, for the case of strong agitation with extreme low solubility oftransferred species in the organic phase, the reaction should be mainly conducted near theinterface due to the short penetration depth of the transferred species, and so is describedby the interfacial mechanism. It is noted that, in general, increasing the agitation rateincreases the degree of dispersion of one phase and produces more tiny droplets, which in


turn generates a much larger interfacial surface area for transport. Hence, the mass trans-fer rate between the phases, the diffusion rate, and the solubility in the organic phase (orthe distribution equilibrium) incorporated with the intrinsic organic reaction play impor-tant roles in determining whether the PT reaction is dominated by an extraction mechan-ism or by an interfacial mechanism.

In the following sections of this chapter, the interfacial mechanism and the kineticsconcerning LLPTC, LSLPTC, SLPTC, and TLPTC will be reviewed.

B. Some Applications

The vast literature on PT catalysis has demonstrated in past years the very broad andeffective applications in organic synthesis [2,3]. Hundreds of articles are published per yearconcerning PTC. Hence, we do not intend to review the many uses of PTC that have beenreported, but just the typical later examples for illustration in this chapter.

FIG. 2 Concentration gradient in organic droplet: (a) slow diffusion rate (or low solubility); (b) fast

diffusion rate (or high solubility).


1. Applications in Biology

Orsini et al. [4] synthesized biologically polyphenolic glycosides via Wittig reactions fol-lowed by glucosylation under PT conditions. These compounds include (E)-3-(�-d-gluco-pyranosyloxy)-4 0,5-dihydroxystilbene (resveratrol 3-�-d-glucoside, piceid), (Z)-2 0,3 0-dihydroxy-3,4,4 05-tetramethoxystilbene (combretastatin A-1), �;�-dihydro-2 0,3 0-dihy-droxy-3,4,4 0,5-tetramethoxystilbene (combretastatin B-1), etc. Under PTC, the glucosyla-tion is stereoselective and gives the best results for yields with benzyltriethylammoniumchloride and aqueous sodium hydroxide. The use of nonaqueous bases in dry solventsleads to a sluggish reaction at room temperature, probably due to the poor solubility ofthe phenolate ion in the solvents. Carriere et al. [5] synthesized O-, S-, Se-, and C-glyco-sides by PTC. For the synthesis of O-glycosides under liquid–liquid conditions, usingdichloromethane as the organic solvent and aqueous NaOH as the base, the PT catalysttetrabutylammonium hydrogen sulfate is used to avoid the possibility of double halidedisplacement. PTC conditions are successfully applied in the synthesis of �- and �-naphthols to glycohydrolase substrate 7-hydroxy-4-methylcoumarin, to chromogenic sub-strate Fat Brown B1, and to estrone prodrug. In the preparation of thio- and seleno-glycosides, having saturated NAHCO3 and 1M Na2CO3 as the aqueous base is sufficientwith thiols and selenols, and together with tetrabutylammonium hydrogen sulfate as thecatalyst, and ethyl acetate as the solvent instead of dichloromethane, whereby the sideproducts are produced.

Albanese et al. [6] reported the synthesis of 2-substituted 3,4-dihydro-2H-1,4-ben-zoxazines by ring opening of glycidols under solid–liquid PTC. They used N-(2-fluoro-phenyl)toluene-p-sulfonamide as the nitrogen nucleophile by incorporating the aromaticmoiety of benzoxazine as the leaving group, and performed the ring opening by stirring at90�C a heterogeneous mixture of 1,2-epoxy-3-phenoxypropane, sulfonamide, anhydrousK2CO3, the catalyst BzEt3NCl, and dioxane to produce a 95% yield of N-(2-fluorophe-nyl)-N-(2-hydroxy-propyl)toluene-p-sulfonamide after 17 h of reaction. This method pro-vides a straightforward and new approach to the synthesis of chiral 2-substituted 3,4-dihydro-2H-1,4-benzoxazines.

Asymmetric PTC is an important method in the synthesis of �-alkyl and �-aminoacids. Belokon et al. [7] reported that the compound (4R,5R)-2,2-dimethyl-�;�;� 0;� 0-tetra-phenyl-1,3-dioxolane-4,5-dimethanol (TADDOL) was used to catalyze the C-alkylation ofC–H acids with alkyl halides to the asymmetric synthesis of �-methyl-substituted �-aminoacids under PTC conditions. The alkylations of the substrate C–H acids with benzylbromide or allyl bromide were conducted in dry toluene at ambient temperature withNaH or solid NaOH as base and TADDOL as a chiral promoter. The type of base isimportant in the asymmetric C-alkylation of C–H acids.

Lygo et al. [8] investigated the enantioselective synthesis of bis-�-amino acid estersvia asymmetric PTC. Under liquid–liquid conditions, the target amino acid esters wereobtained with high enantiometric excess (� 95% ee) from the alkylation reaction of ben-zophenone-derived glycineimine with an appropriate dibromide. They reported that eitherthe mono- or di-alkylated product could be obtained, depending on the reaction condi-tions; the monoalkylated product was obtained in good yield with excess dibromide,whereas with stoichiometric quantities of dibromide this led to the dialkylated product.By controlling the stoichiometry of the reaction, the selectivity of the desired product canbe accessed at a high level. Lygo et al. [9] also reported the asymmetric synthesis of bis-�-amino acids via alkylation of a benzophenone-derived glycineimine under PTC conditions.The target bisamino acids can be produced with high yields and high levels of stereo-selectivity by applying chiral quaternary ammonium salts. The core structure of the chiral


quaternary ammonium salts closely related to the cinchona alkaloid cinchonine can beused in the benzoylation of a glycineimine [10].

The indan-based �-amino acid derivatives can be synthesized by PTC. Kotha andBrahmachary [11] indicated that solid–liquid PTC is an attractive method that offered aneffective way of preparing optically active products by chiral PTC. They found that ethylisocyanoacetate can be easily bisalkylated in the presence of K2CO3 as the base andtetrabutylammonium hydrogen sulfate as the catalyst. The advantage of isolating waterfrom the reaction medium is to avoid the formation of unwanted hydroxy compounds inthe nucleophilic substitution reaction. If liquid–liquid PTC is applied in the system withthe strong base NaOH and dichloromethane as the organic solvent, the formation ofdihydroxy or cyclic ether can be observed.

2. Other Applications

PTC incorporated with other methods usually greatly enhances the reaction rate. Masstransfer of the catalyst or the complex between different phases is an important effect thatinfluences the reaction rate. If the mass transfer resistance cannot be neglected, animprovement in the mass transfer rate will benefit the overall reaction rate. The applica-tion of ultrasound to these types of reactions can be very effective. Entezari andKeshavarzi [12] presented the utilization of ultrasound to cause efficient mixing of theliquid–liquid phases for the saponification of castor oil. They used cetyltrimethylammo-nium bromide (CTAB), benzyltriethylammonium chloride (BTEAC), and tetrabutylam-monium bromide (TBAB) as the catalysts in aqueous alkaline solution. The more suitablePT catalyst CTAB can accumulate more at the liquid–liquid interface and produces anemulsion with smaller droplet size; this phenomenon makes the system have a high inter-facial surface area, but the degradation of CTAB is more severe than that of BTEAC orTBAB because of more accumulation at the interface of the cavity under ultrasound.

Recently, electron-transfer catalysis by viologen compounds has attracted muchattention. The compounds function as mediators of electron transfer and have beenapplied in the reduction of aldehydes, ketones, quinines, azobenzene, acrylonitrile,nitroalkenes, etc., with zinc or sodium dithionite in a monophase or a two-liquid phasesystem [13]. Noguchi et al. [13] found that a redox-active macrocyclic ionene oligomer,cyclobis(paraquat-p-phenylene), acted as an electron phase-transfer catalyst for the reduc-tion of quinines, as compared with acyclic benzyl viologen. The enhanced activity of thiscompound is due to the inclusion of the substrate into the catalyst cavity.

One of the important applications of PTC is in the field of pollution control. Anearly utilization was to apply the PTC method to recover phenolic substances from aqu-eous alkaline waste streams [14]. The methodology is based on the reaction of phenolicsubstances in the aqueous solution with materials such as benzoyl chloride, p-toluenesul-fonyl chloride, etc., dissolved in the organic solvent in the presence of PT catalysts:

ð3Þ


Tundo et al. [15] reported an efficient catalytic detoxification method for toxicpolychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans(PCDFs) under mild conditions (50�C and 1 atm of hydrogen) with a supported metalcatalyst modified by the PT agent Aliquat 336. Their results show that the methodologyproved successful for hydrodechlorinating the toxic samples to yield mixtures containingconcentration of contaminants lower than the experimentally detectable limit by gas chro-matography–high-resolution mass spectrometry. This method has the potential to bepractically applied in the detoxification of PCDDs and PCDFs.

PTC is also widely used in polymerization reactions. The main function of thequaternary ammonium salts is that they can transfer the diphenolate from the aqueousphase into the organic phase to react with the diacid chloride. Hodget et al. [16] presentedthe synthesis of polyesters by the reaction of dicarboxylic acid salts with bishalides ortosylates or by the self-condensation of salts of bromocarboxylic acids under liquid–liquidPTC. With benzyltrimethylammonium salts and halides in dry acetonitrile as solvent,using sodium or potassium salts, the yields of polyesters are, in degrees of polymerization(DP), in the range 17–47, and the rate of dissolution of salts is very slow and rate limiting;while in a liquid–liquid system, the DP is in the range 22–161. Liquid–liquid PTC is morefavorable in the synthesis of polyesters [16]:

RCOX þR 0OH! RCOOR 0 þHX ð4ÞRCOO�Mþ þR 0X! RCOOR 0 þMþX� ð5Þ

where X ¼ -Cl, -Br, -I, -OSO2CH3, or -OSO2C6H4CH3.The applications of PTC in polymerization are gradually increasing. Tagle and co-

workers [17,18] synthesized poly(amide ester)s from diphenols with the amide group in theside chain, using PT catalysts such as benzyltriethylammonium chloride, with good results.The use of anhydrous potassium carbonate as the base is to promote the organic reactionunder solid–liquid PTC. Albanese et al. [19] described some recent applications in thisarea, and the reactions of aza anions with 2-bromocarboxylic esters and expoxidesafforded protected �-amino acids and �-amido alcohols. Sirovski [20] described someexamples of PTC applications in organochlorine chemistry. Using a polymeric crownether the results of m-phenoxytoluene chlorination are also reported. Carboxylic acidsand picric acid act as inhibitors, while benzyl alcohol behaves as a strong promoter. Inthe absence of the promoter, the reaction is conducted either at the interface or in the thirdphase that is a border liquid film between the organic and aqueous phases.

The importance of triphase catalysis in industry grows continuously. The supportsfor immobilizing the triphase catalyst are mostly of organic type, i.e., copolymers ofpolystyrene. Yadav and Naik [21] reported that clay could be used as support for thePT catalyst; benzoic anhydride was prepared from benzoyl chloride and sodium benzo-ate using a clay-supported quaternary ammonium salt at 30�C. The polymer-supportedcatalysts are less active than the clay-supported catalyst for this reaction system.Desikan and Doraiswamy [22] investigated the enhanced activity of polymer-supportedPT catalysts for the esterification of benzyl chloride with aqueous sodium acetate. Theyfound that the reactivity using a triphase catalyst is higher than that using a solubleone. They hypothesized that the enhancement due to increased lipophilicity of thepolymer-supported catalyst was more than compensated by the decreased diffusionalresistance.

Jayachandran and Wang [23] prepared a new PT catalyst, 2-benzilidine-N,N,N,N 0,N 0,N-hexaethylpropane-1,3-diammonium dibromide (Dq-Br), to investigate


the cycloalkylation of phenylacetonitrile (PAN) with an excess of 1,4-dibromobutaneusing aqueous sodium hydroxide as the base, and the following pseudo-first-order kineticswas observed:

C6H5CH2CNþ BrðCH2Þ4Br�!Dq-Br ð0:75mol%Þ

C6H5CðCH2Þ4CN ð6ÞHwang et al. [24] studied the Wittig reaction of benzyltriphenylphosphonium

(BTPP) salts and benzaldehydes via ylide-mediated PTC. They concluded that the reactionof benzylidenetriphenyl phosphorane and the benzaldehyde in the organic phase is thedecisive step for stereoselectivity. The order of effectiveness of substituents isCF3 > ðCl;BrÞ >MeO > F > NO2. Satrio and Doraiswamy [25] proposed a case studyfor the production of benzaldehyde in a possible industrial application of PTC. Thereaction between benzyl chloride and hypochlorite anion is

C6H5CH2Cl ðorgÞ þOCl� ðaqÞ ! C6H5CHO ðorgÞ þHCl ðaqÞ þ Cl� ð7ÞThey show that the conventional route is the preferred one for a large-scale organicintermediate, and the improvements in merely one or two PTC steps can greatly enhancethe prospects of the PTC route.

II. LIQUID–LIQUID PHASE TRANSFER CATALYSIS

A necessary condition for a reaction is to cause the collision of two reactant molecules. Itis obvious that the reaction rate of two immiscible reactants is low due to their lowsolubilities. A general method for overcoming this difficulty was to employ a protic oran aprotic solvent in order to improve their mutual solubilities. Nevertheless, thisimprovement was not very significant. The problems of two-phase reactions were notsolved until Jarrouse [26] discovered the catalyzing effect of quaternary ammonium saltin the aqueous–organic phase reaction system. PTC is an effective tool for synthesizingorganic chemicals from two immiscible reactants [27–32]. It has been extensively applied tothe synthesis of special organic chemicals by displacement, alkylation, arylation, conden-sation, elimination, oxidation, reduction, and polymerization. The advantage of PTC inthe synthesis of organic chemicals are fast reaction rate, high selectivity of product, mod-erate operating temperature, and applicability to industrial-scale production.

A. Mechanism of Liquid–Liquid Phase Transfer Catalysis (LLPTC)

Quaternary salts, crown ethers, cryptands, and polyethylene glycol (PEG) are the mostcommon agents used for LLPTC. Over the last few decades, the two reaction mechanismsused to describe the phenomenon of a two-phase PTC reaction were the Starks extractionmechanism and Makosza interfacial mechanism.

1. Starks Extraction Mechanism

This reaction mechanism described by Starks [28,33] is widely accepted for a catalysttransferring between the two phases. Reactions occurring in such systems involve: (1) thereactant reacting with catalyst in the normal phase to form an intermediate catalyticreactant, (2) transfer of the intermediate catalytic reactant from its normal phase intothe reaction phase, (3) transferred intermediate catalytic reactant reacting with untrans-formed reactant in the reaction phase to produce the product and catalyst, and (4)


transfer of catalyst from the reaction phase to the normal phase. The reaction mechan-ism can be separated in three ways based on the reaction path, and can be described asfollows.

(a) Normal Liquid–Liquid Phase Transfer Catalysis (N-LLPTC). Traditionally, moreapplications of PTC have been reported in N-LLPTC. The reaction mechanism (8) ismostly applied to alkylation, esterification etherification, and simple displacement reac-tions in which a nucleophilic agent is transferred to the organic phase through the solu-ble catalyst therein:

ð8Þ

Mechanism (8) was first presented by Starks [33] for the reaction of 1-chloro-octane andaqueous sodium cyanide.

(b) Inverse Liquid–Liquid Phase Transfer Catalysis (I-LLPTC). The organic reactantis converted, by means of a reagent (e.g., pyridine 1-oxide, PNO) partially soluble inthe organic phase, into a reactive ionic intermediate and transferred into the aqueousphase where reaction takes place to produce the desired product. The processes havebeen termed inverse phase-transfer catalysis [34–36]. The reaction mechanism can beexpressed as follows:

ð9Þ

There are several examples where I-LLPTC has been used to synthesize acid anhy-drides, by means of a substitution reaction, and ketones from oxidation of alcohols [37–40]. The reaction of an acid chloride (RX) with the carboylate ions (MþR

0�) catalyzed byPNO is to proceed through an intermediate 1-(acyloxy)pyridinium chloride formed in theorganic phase. PNO and N,N-dimethylaminopyridine (DMAP) are widely used as inversePT catalysts. The formation of hippuric acid was conducted in the presence of 4-dimethy-laminopyridine as inverse PT catalyst [41].

(c) Reverse Liquid–Liquid Phase Transfer Catalysis (R-LLPTC). This reactionmechanism was expressed as follows:


ð10Þ

The dehydrohalogenation reactions of alkyl halides take place in the presence ofhydroxide ion and quaternary salts to form alkenes and alkynes [42–44]. The dehydroha-logenation is promoted by hydroxide ion. In general, two reaction conditions conducted inthis system were with highly lipophilic ammonium cation and 50% aqueous sodiumhydroxide. The reaction between 4-nitrobezenediazonium chloride and N-ethylcarbazolein aqueous media was accelerated by using a water–dichloromethane system containingsodium 4-dodecylbenzenesulfonate as a transfer catalyst for the diazonium ion [34].

2. Makosza Interfacial Mechanism

This reaction mechanism described by Makosza and Bialecka [45,46] is the acceptedcatalyst transport between the two phases. Reactions occurring in such systems involve:(1) transfer of ionic reactant from its normal phase and catalyst from the reaction phaseinto the interfacial region, (2) the ionic reactant reacting with catalyst in the interfacialregion to form intermediate catalytic reactant, (3) the intermediate catalytic reactanttransfer into the reaction phase to react with untransformed reactant to produce theproduct and catalyst. The reaction mechanism is expressed as follows:

ð11Þ

Usually, the aqueous salt could be too hydrophilic to allow the quaternary salt todissolve in the organic phase, and resided exclusively in the aqueous phase; anionexchange occured at or near the interface. The mechanism is applied to carbanion reac-tions, carbene reactions, condensation of polymerization, and C-alkylation of activemethylene compounds such as activated benzylic nitriles, activated hydrocarbons, andactivated ketones under PTC=OH�. In most cases, the reaction involves the QþOH�

complex because QOH is highly hydrophilic and has extremely low solubility in theorganic phase.

A mechanism can also be applied when the quaternary salt is too lipophilic todissolve in the aqueous phase, and resides exclusively in the organic phase, anion exchangeoccuring at or near the interface. This parallel mechanism is called the Brandstrom–Montanari mechanism. The ion-exchange reaction existing at the interface was verifiedby Landini et al. [47] and Brandstrom [48].


A summary of characteristic kinetic criteria to distinguish between the operation ofthe extraction and interfacial mechanisms has been suggested [28,49]. The extractionmechanism is characterized by: (1) increased rates with increased lipophilicity of catalyst,(2) reaction rates that are independent of stirring speed above a certain value, (3) first-order or fractional dependence of reaction rate on catalyst concentration, and (4) pseudo-first or second-order kinetics if the reaction in the organic phase reaction is rate controllingor zero-order kinetics if diffusion across the interface is rate controlling.

The interfacial mechanism is characterized by: (1) increased rates with increasedelectrostaticity of catalyst, (2) reaction rates are dependent on agitation rate, (3) fractionalkinetic order with respect to the catalyst concentration, and (4) the value of substrateacidity pKa is in the range 16–23.

B. Kinetics of a Liquid–Liquid Phase Transfer Catalysis

1. Starks Extraction Mechanism

A typical LLPTC cycle involves a nucleophilic substitution reaction, as shown in Eq. (8).A difficult problem in the kinetics of PT-catalyzed reactions is to sort out the rate effectsdue to equilibrium anion-transfer mechanism for transfer of anions from the aqueous tothe organic phase. The reactivity of the reaction by PTC is controlled by the rate ofreaction in the organic phase, the rate of reaction in the aqueous phase, and the masstransfer steps between the organic and aqueous phases [27–29]. In general, one assumesthat the resistances of mass transfer and of chemical reaction in the aqueous phase can beneglected for a slow reaction in the organic phase by LLPTC.

Although a large number of papers have been published on the synthetic applica-tions of PTC in the last three decades, little mathematical analysis of the phenomenon hasbeen done, and such an analysis is especially desirable in a large-scale application. Evansand Palmer [50] considered a process of interphase mass transfer and chemical reaction.Melville and Goddard [51] and Melville and Yortsos [52] presented an analysis of masstransfer in solid–liquid PTC. Chen et al. [53] derived algebraic expressions for the inter-phase flux of QY and QX. The reaction parameters were estimated from experimental datausing a two-stage method of optimal parameters. Wang and Chang [54–56] studied thekinetics of the allylation of phenoxide with allyl chloride in the presence of PEG asLLPTC. A simple mathematical model describing the liquid–liquid PT-catalyzed reactionwith the two-film theory was analyzed [57–59]. The results of the model’s prediction areconsistent with experimental data. Such mathematical analysis appears desirable andneeded in view of the widespread interest in PTC in the chemical industry in whichtwo-phase transfer and triphase catalysis are the most common industrial processes.

The reactivity in phase-transfer catalysis is controlled by: (1) the reaction rate in theorganic phase, (2) the mass transfer steps between the organic and aqueous phases, and (3)the distribution equilibrium of the quaternary salts between the two phases. The distribu-tion of quaternary salts between two phases directly affects the entire system reactivity[60–62]. On the basis of the experimental data and earlier literature [27,28,63], a general-ized approach describing a LLPTC reaction system uses a pseudo-first-order reaction. Therate expression is written as

� d½RX �dt¼ kint½QY �½RX � ð12Þ

¼ kapp½RX � ð13Þ


The fixed value of kapp is called the apparent first-order reaction-rate constant. The over-bar denotes the species in the organic phase. The reaction rate linearly increases withincreasing QY concentration. Equation (13) is established when the QY concentration isconstant. Most observed reaction rate would follow the pseudo-first-order kinetics for anexcess amount of aqueous reactant to that of organic reactant [37]. Wu [64] indicated thata pseudo-first-order hypothesis can be used to describe the PTC experiment data, eventhough the QY concentration is not kept constant. Wang and Wu [58] developed a com-prehensive model in a sequential phosphazene reaction. Their experimental results wereconsistent with a first-order reaction rate; the pseudo-first-order reaction-rate constantwas not linearly related to the concentration of the catalyst, because the mass transferof catalyst between the two phases influenced the reaction. Wang and Yang [57,65] andWu [63] indicated that the QY concentration is constant over time when the molar ratio ofnucleophile to catalyst is larger than unity. Therefore, in the general case, the QY con-centration cannot vary with time only when the ion-exchange rate in the aqueous phase ismore rapid than that in the organic phase [66], no mass transfer resistance of catalystbetween the two phases occurs, the molar ratio of nucleophile to catalyst is larger thanunity, and the ionic strength in the aqueous phase is high [67].

The complicated nature of the LLPTC reaction system is attributed to two masstransfer steps and two reaction steps in the organic and aqueous phases. The equilibriumpartition of the catalysts between the two phases also affects the reaction rate. On the basisof the above factors and the steady-state two-film theory [60,63,64,68], a phase-planemodel to describe the dynamics of a liquid–liquid PTC reaction has been derived. Thismodel offers physically meaningful parameters that demonstrate the complicated reactivecharacter of a liquid–liquid PT-catalyzed reaction. However, when the concentration ofaqueous solution is dilute or the reactivity of aqueous reactant is weak, the onium cationhas to exist in the aqueous phase. The mathematical model cannot describe this comple-tely. When the onium cation exists in the aqueous phase, several important phenomenainvolved in the liquid–liquid reaction need to be analyzed and discussed.

ð14Þ

On the basis of Eq. (12), and mechanism (14) [64,68], the species balance equationswere solved by eliminating the time variable (phase-plane model). The relevant rate equa-tions are

dyod�¼ y1oyo ð15Þ

dy1odyo¼ P1 �

�QY

yo�mQY

y1ay1o� 1

� �ð16Þ


dy1adyo¼ �QY

yo�mQY

y1ay1o� 1

� �� 1y3ay4aKd1y1oyo

þ �1y1ay1oyo

ð17Þ

dy2odyo¼ �QX

yo

y2oy1o� �mQX

y2ay1o

� �� P1 ð18Þ

dy2adyo¼ �2

y2ay1oyo

� �2y3ay5akd2y1oyo

��QX

yo

y2oy1o� �mQX

y2ay1o

� �ð19Þ

dy3adyo¼ �2y3ay5a

Kd2y1oyoþ �1y3ay4aKd1y1oyo

� �1y1ay1oyo

� �2y2ay1oyo

ð20Þ

The mass balances for Qi, Y�, and X� are given below:

1 ¼ y1o þ y1a þ y2o þ y2a þ y3a ð21Þy4a ¼ P2 � y1a � y1o þ ðyo � 1ÞP1 ð22Þy5a ¼ P3 þ 1� y2a � y2o þ ð1� yoÞP1 ð23Þ

in which the dimensionless variables and parameters are defined as

yo ¼½RX�o½RX�i

; y1o ¼Vo½QY�o

Qi

; y1a ¼Va½QY�a

Qi

; y2o ¼Vo½QX�o

Qi

; y2a ¼Va½QX�a

Qi

;

y3a ¼Va½Qþ�a

Qi

; y4a ¼Va½Y��a

Qi

; y5a ¼Va½X��a

Qi

; �QX ¼KQXA=Va

koQi=Vo

;

�QY ¼KQYA=Vo

koQi=Vo

; P1 ¼V0½RX�i

Qi

; P2 ¼Va½MY�i

Qi

; P3 ¼Va½MX�i

Qi

;

�1 ¼kd1ko; �2 ¼

kd2ko; � ¼ Vo

Va

; � ¼ tkoQi

Vo

ð24Þ

and ½MX�i, ½MY�i, and ½RX�i represent the initial concentrations of reactants MX, MY,and RX, respectively. By introducing the values of the parameters into Eqs (15)–(23), thedynamic phenomena of a liquid–liquid PT-catalytic reaction was obtained.

Wang and Yang [57] reported that the ion-exchange reaction-rate constant wascalculated with three differential equations as below for the dynamics of QY in boththe aqueous and organic phases in a two-phase reaction without adding the organicreactant by the numerical shooting method and correlating it with the experimental data.

d �CCQY

dt¼ KQYA CQY � �CCQY=mQY

� � ð25Þ

dCQY

dt¼ KdaCQYCQX � KQYA

�VV

VCQY � �CCQY=mQY

� � ð26ÞdCMY

dt¼ �KdaCQYCQX ð27Þ

The intrinsic reaction-rate constant in the organic phase is obtained by reacting QYwith RX in a homogeneous solvent and using Eq. (12). According to the literature, Wangand Yang [57] and Wu and Meng [69] have found the intrinsic reaction-rate constant fromtheir systems. The equilibrium constant and mass transfer constant of the catalyst betweentwo phases obtained are discussed in the next section.


Wu [64] characterized the transfer of QþX� from the organic phase to the aqueousphase and of QþY� from the aqueous to the organic phase by defining

QY ¼y1amQY

y1o; QX ¼

y2oy2amQX

ð28Þ

If the PT catalysts in the two phases are in extractive equilibrium and the mass transferresistance can be neglected completely, then QY and QX are each equal to 1.

The dynamics for a slow PT reaction and a mass transfer controlled instantaneousreaction were studied. Wu [63] and Wu and Meng [69] indicated that the pseudo-steady-state LLPTC model could describe the complicated nature of the LLPTC reaction. Therate equation from the report of Wu [63] is expressed as

d½RX�dt¼ k½RX�Q1= �VV

�mQY þ 1�mQY

þ DaQY�mQY

þDaQX

� �þ ð1þ �mQXÞ DaQY þ 1

�mQYþ �þ ��

� � ð29Þ

where DaQY ð¼ k½RX�=kQYA= �VVÞ and DaQXð¼ k½RXÞ=KQXA= �VVÞ are the Damkohlernumbers for QY and QX, respectively; � ð¼ k�2½MX�=k2½MY�Þ is the reaction ratio ofthe aqueous reverse reaction to the forward reaction for ion exchange; and � ð¼ k½RX�=k2½MY�Þ is the reaction ratio of the organic phase to the aqueous forward ion-exchange reaction.

Wu [63] also derived an expression for the catalyst effectiveness, which is defined asthe ratio of the actual reaction rate to that with all the catalyst present as QY, in terms ofseven physically meaningful dimensionless parameters:

� ¼ �mQY þ 1

�mQY

þ DaQY

�mQY

þDaQX

� �þ 1þ �mQX

� � DaQY þ 1

�mQY

�þ �� 1

ð30Þ

Before evaluating Eq. [30], the parameters of kinetics, mass transfer, and thermodynamicequilibrium must be established. The aim of this work is to evaluate the equilibrium andextraction of a quaternary salt in an organic solvent/aqueous solution. The studies ondistribution equilibrium of the quaternary salts enable one to clarify the true mechanismthrough which the reactant anion is transferred.

Models for LLPTC get even more complicated for special cases, e.g., reactions inboth aqueous and organic phases, systems involving a base reaction, or other complexseries–parallel multiple reactions. Wang and Wu [58] and Wu and Meng [69] studied thekinetics and mass transfer for a sequential reaction using LLPTC that involved a complexreaction with six sequential SN2 reactions in the organic phase along with interphase masstransfer and ion exchange in the aqueous phase.

Wang and Wu [70] analyzed the extraction equilibrium of the effects of catalyst,solvent, NaOH/organic substrate ratio, and temperature on the consecutive reactionbetween 2,2,2-trifluoroethanol with hexachlorocyclotriphosphazene in the presence ofaqueous NaOH. Wu and Meng [69] reported the reaction between phenol with hexachlor-ocyclotriphosphazene. They first obtained the intrinsic reaction-rate constant and overallmass transfer coefficient simultaneously, and reported that the mass transfer resistance ofQX from the organic to aqueous phase is larger than that of QY from the aqueous toorganic phase. The intrinsic reaction-rate constant and overall mass transfer coefficientswere obtained in three ways.


(a) Pseudo-Steady-State LLPTC model. The reaction relationship is given as

1

kapp¼

�VV

KQXA�þ

�VV

kQi

ð31Þ

where � denotes the reactivity of the phosphazene reaction. The plot of 1=kapp versus �, inwhich the data were measured at the initial time of different experimental runs, allows oneto obtain the mass transfer coefficient, KQXA, and the intrinsic reaction rate constant k,from the slope and intercept of the straight line.

(b) Extrapolation Method. If mass transfer resistance influences the reaction, the con-centration of the active catalyst QY cannot remain constant during the course of thereaction. Decreasing the concentration of organic reactant RX increases the apparentfirst-order reaction-rate constant. When the concentration of organic reactant decreases,both the reaction rate and the effect of mass transfer decrease. If the organic reactantconcentration extrapolates to zero ð½RX� ! 0Þ, the effect of mass transfer can beneglected. The intrinsic reaction-rate constant, k, is easily evaluated.

(c) Half-Reaction in the Organic Phase. The organic reactant reacted with an inter-mediate catalyst, tetra-n-butyl ammonium phenolate, in a homogeneous organic phase.The intrinsic reaction-rate constant was calculated from Eq. (12).

Another LLPTC is usually performed in an agitated system, in which the organicphase is mostly dispersed. Several efforts have been made in developing the theory for atwo-liquid phase with chemical reactions. For an organic phase being the dispersed phase,several phenomena take place: (1) formation of a single droplet in the continuous phase bystirring, (2) free rise or fall of a droplet through the continuous phase, and (3) coalescenceof a droplet at the end of the free-rise period. During the extraction of a catalytic inter-mediate, mass transfer from the bulk aqueous phase to the organic droplet surface influ-ences the rate of PT reaction. Yang [71,72] studied the general analysis of the dynamics ofa PT-catalytic reaction in a dispersed system of liquid–liquid phases, considering theirreversible and reversible reactions by solving the finite difference and Runge–Kuttafourth-order methods. The rates of change of RX, RY, QX, and QY in an organic dropletare described by the instantaneous equations of diffusion and reaction with the corre-sponding initial and boundary conditions as follows:

@ �CCi

@t¼ Di

r2@

@rr2@ �CCi

@r

� �þ �i �RR; i ¼ RX; RY; QX; and QY ð32Þ

where �i is the stoichiometric coefficient of the i component.The kinetics of inverse PT-catalytic extraction of species into the water phase was

carried out with partially water-soluble pyridines or derivatives [36,38,40,59,73], as shownin mechanism (9). These reactions can be described by a pseudo-first-order hypothesis[38,40]:

kapp ¼ kh þ kc½PNO�i ð33Þ

However, so far, the detailed kinetics of I-LLPTC are unclear.As mentioned above, the various approaches to LLPTC modeling have been taken,

and a comprehensive general model for N-LLPTC reactions is widely held. However, akinetic model for I-LLPTC and R-LLPTC reactions is yet to be developed.


2. Makosza Interfacial Mechanism

The interfacial mechanism is the most widely accepted mechanism for PTC reactions in thepresence of a base. However, although there are numerous industrially important applica-tions, very few kinetic studies or mathematical models for this mechanism are reported. Ingeneral, the mechanism is also described by a pseudo-first-order hypothesis.

Juang and Liu [74,75] proposed and discussed a possible mechanism based on amixed Makosza and modified interfacial mechanism. The reaction rate for the etherifica-tion of a substituted phenylacetic acid by PTC was measured using a constant interfacialarea cell, and expressed as

Rf ¼ k½R 0X�1=3½RH�½QX�½OH��5=21þ ka½QX�1=2½OH�� þ kb½RH�1=2½OH�� ð34Þ

C. Thermodynamic Equilibrium in LLPTC

Quaternary salts are generally used as normal liquid–liquid PT catalysts. In general, thefunctional groups of the quaternary cation will affect the dissolution of the catalyst in theorganic phase. Further, the phase transfer of the anion will also affect the reaction rate inthe two-phase reaction. Therefore, a proper choice of PT catalyst is very important inpromoting the reaction rate. Unfortunately, a universal guideline is unavailable for select-ing the proper PT catalyst to enhance the reaction. The reactivity in PTC is controlled by:(1) the reaction rate in the organic phase, (2) the mass transfer steps between the organicand aqueous phases, and (3) the distribution equilibrium of the quaternary salts betweenthe two phases. The distribution of quaternary salts between two phases directly affects theentire system reactivity [60–62].

In general, anion transfer and anion activation are the important steps involved intransferring anions from the aqueous phase to the organic phase where the reaction takesplace. Factors affecting the extraction ability of the anion from the aqueous to organicphase include cation–anion interaction energies, the ionic strength in the aqueous phase,ion-pair hydration, the lipophilicity of the catalyst, and the polarity of the organic phase.The extraction behavior and distribution coefficients of quaternary salts in various mediahave also been investigated [76–86].

Brandstrom [48] indicated that the distribution of quaternary salt between two(liquid–liquid) phases exists as complicated multiequilibrium constants, which dependon the structure of the anion, cation, and solvent, as well as on pH, ionic strength, andconcentrations in the aqueous solution. Such equilibrium properties have not yet beenevaluated completely. The relationship between quaternary salt and extraction constant isan important consideration for PTC work.

The distribution coefficient of quaternary cation DQ was obtained by measuring theconcentrations of quaternary cation (Q) in the organic and aqueous phases, respectively.The distribution coefficient is highly dependent on the nature and concentration of thequaternary salts:

DQ ¼½Q�obs½Q�obs

ð35Þ

The distribution coefficient of quaternary cations between both the phases not onlyprovides information on the phases to facilitate the modeling of the two-phase transfercatalysis system, but it can also give a criterion for evaluating the suitability of the catalyst.


The order of magnitude of DQ for quaternary salts is Aliquat 336 > TBA-TBPO >TBAI > TBPB > TBAB > TBAC. The sequence of DQ for solvents is CHCl3 >CH2Cl2 > 1;2-C2H2Cl2 > C6H5Cl. The order of influence on the extraction capabilityof quaternary salts is Br3C6H2O

� > I� > Br� < Cl� and Pþ > Nþ for the anion andcentral cation, respectively. Reasons for these behaviors have been discussed in previouswork [48,76,81,85,86]. The DQ value increased on increasing the temperature.

The true extraction constants of quaternary salts QX corresponding to their infi-nitely dilute solutions in a two-phase system were calculated using the following equation:

ETQX ¼

aQX

aQþ ; aX�¼ ½QX�½Qþ�½X��2�

ð36Þ

where a and �2� are the activity and the mean ionic activity coefficient of the quaternarysalts, respectively.

The distribution constant of quaternary salt at equilibrium between two phases is

m ¼ ½QþX��½QþX�� ð37Þ

The dissolved QþX� in the aqueous and organic phase may dissociate to

QþX� Ð Qþ þ X

� ð38ÞQþX� Ð Qþ þX� ð39Þ

Thus, the dissociation constants Kda and Kda of QX in the aqueous and organic phases arewritten as

Kda ¼½Qþ�½X��2�½QþX�� ð40Þ

Kdo ¼½Qþ�½X�� 2�½QþX�� ð41Þ

The dissociation constant in aprotic organic solvents can be derived from funda-mental principles based on Bjerrum’s theory for ion pairs. In most organic media, thedissociation constant of ion pairs is very low (of the order of around 10�5) [48].Brandstrom [87] reported that the ionic aggregation states of quaternary salts existingin the organic phase were of various types, i.e., dissociated ions (Qþ þX�), ion pairs(QþX�), quadruples ½ðQþX�Þ2�, etc. Hence, the total concentration of quaternary salt inan organic phase can be written as

�CCQ ¼ ½Qþ� þ ½QX� þ 2½Q2X2� þ ð42ÞSince the organic system is in electrical neutrality,

½Qþ� ¼ ½X�� ð43ÞEquation (42) can be transformed into

�CCQ ¼ ET1=2

Qþ��½Qþ�½X�� 1=2þET

QX�2� ½Qþ�½X�� þ 2ET

Q2X22�4� ½Qþ�½X��

� �2þ ð44Þ

where ETQþ , E

TQX, and ET

Q2X2are the concentration quotients represented as


ETQþ ¼

½Qþ�½X��2�½Qþ�½X��2�

ð45Þ

ETQX ¼

½QX�½Qþ�½X��2�

ð46Þ

ETQ2X2¼ ½Q2X2�½Qþ�½X��2�

ð47Þ

By using Eqs (42)–(47), the values of ETQþ , E

TQX, E

TQ2X2

, and the distribution constantm are evaluated. Corrections for the mean activity coefficient in the organic phase weremade using the Marshall and Grunwald expression, and the values of m, Kda, Kdo, and ��were calculated by a numerical iteration method. Beronius and Brandstrom [91] evenclarified the identical value of Kdo at ½QX� ¼ 0 within the limits of experimental errorand the conductance measurement. In view of past reports [87–92], most Kda valueswere located in the range between 1 and 10; Kdo values were located in the range between10�1 and 10�5. The dissociation ability of quaternary salt in the aqueous phase is greaterthan that in the organic phase.

The quaternary salts QX can be completely dissociated to free ions (Qþ and X�) inthe aqueous phase (, ½Qþ�=½QX� > 100Þ and partially dissociated in the organic phasewhen the concentration of the quaternary salt is 0.0125 kmol/m3. The quaternary salts QXcan be partially dissociated to free ions in the aqueous and the organic phases when theconcentration of quaternary salt is 0:1 kmol=m3. The incremental rules of the dissociationdegree of the quaternary salts were obtained as follows: (1) increasing the charge-to-volume ratio of the central cation or counteranion (e.g., Pþ > Nþ or I� > Br� > Cl�),(2) increasing the electron-releasing groups on the quaternary cation (e.g., Aliquat336 > TBAC), and (3) increasing the electron-withdrawing groups on the quaternaryanion (e.g., TBA-TBPO > TBA-BPO > TBAC). Electron-releasing (or electron-with-drawing) groups apparently make the transition state more stable on the quaternarycation (or anion) while the ion-pair type of quaternary salts transferring through theinterface between two phases is a transition state. Bockries and Reddy [93] reportedthat the association constant decreased when the effective ionic radius of the ion pairwas increased.

Quaternary salts in an organic phase must be determined experimentally to knowwhether the salts are dissociated or associated, and, if so, to what degree. The hydration ofthe anion plays an important role in dissociating the catalyst. Furthermore, the solvationof the anions increases the size of the ions, decreases their mobility and diffusion rate, andreduces the reactivity of the reactant. How many molecules of the coextracted water doeseach quaternary salt carry? Hence, the equation for the distribution of a tetralkylammo-nium halide into an organic phase can be written as [94,95]

Qþ þX�þjH2OÐ Qþ þX�:jH2O ð48Þ

Qþ þX�þjH2OÐ Qþ X�:jH2O ð49Þ

Depending on whether the species in the organic phase is dissociated as free ions [Eq. (48)]or associated as ion pairs [Eq. (49)], the corresponding equilibrium constants can bewritten as


ETQþ;H2O

¼ ½Qþ�½X�:jH2O��2�

½Qþ�½X��½H2O�j�2�ð50Þ

ETQX;H2O

¼ ½QX:jH2O�½Qþ�½X��½H2O�j�2�

ð51Þ

The j value can be calculated by dividing �ðH2OÞ by the amount of quaternary salts in theorganic phase. The water content difference in the organic phase ð�ðH2OÞÞ equals thedifference between the measured water content in the solvent and that in the solution atthe same temperature.

The order of magnitude of �H2O in the organic phase for quaternary salts is Aliquat336 > TBA-TBPO > TBAI > TBPB > TBAB > TBAC. The sequence of �½H2O� for sol-vents is 1,2-C2H4Cl2 > CH2Cl2 > CHCl3 > C6H5Cl. This tendency of the sequence of thecoextracted water is identical to that of the solubility of water in the organic phase of 1,2-C2H4Cl2 ð1:3Þ > CH2Cl2 ð0:81Þ > CHCl3 ð0:08Þ > C6H5Cl ð0:05Þ at 20�C. The orders ofinfluencing extraction capability of H2O are Cl� > Br3C6H2O

� > Br� > I� and Nþ > Pþ

for the anion and central cation, respectively. The trend for water content in the organicphase varied with increasing temperature. Landini et al. [96] indicated that the solvatingcapability between quaternary salt and water could reduce the quaternary salt’s reactivityin the organic phase in a PT-catalyzed reaction. This result was confirmed by previous work[61,76]. Hence, it is significant to study the liquid–liquid PT-catalyzed reaction and toevaluate how many molecules of the coextracted water are carried by each quaternarysalt. The water content in the organic phase increased with increasing temperature. The �½H2O� value increased when the charge-to-volume ratio of the anion increased and when thepolarity of the solvent increased, but decreased as the lipophilicity of the quaternary saltincreased. These tendencies correspond to those reported by Landini and coworkers[97,98]. Kenjo and Diamond [95] reported that the average water contents in a nitroben-zene/water system at 23�C were 3.3, 1.8, and 1 (mol/mol quaternary salt) for Cl�, Br�, andI�, respectively. Starks and Owens [99] reported that the hydration numbers ofC16H33Bu3P

þX� were 0.4, 4, and 5 for NO�3 , Cl�, and CN�, respectively. The average

water content in the organic phase ð�½H2O�Þwas about 1–3 mol/mol quaternary salt, exceptfor TBAC. Because the hydration numbers for different anions were different when thequaternary salt was TBAþ [(n-C4H9Þ4Nþ�, the results demonstrate that the water of hydra-tion is primarily associated with the anion, rather than with the quaternary cation.

Quaternary ammonium ions are used as PT catalysts because they are least likely tointerfere in chemical reactions. According to the experimental results of Brandstrom [48],Herriott and Picker [100], and Landini et al. [97], the organophilic quaternary cationsserved as more effective PT catalysts than quaternary cations with small alkyl chains.Thus, the incremental number of C atoms surrounding the central atom (e.g., N) of aquaternary salt will increase its lipophilicity, thus raising the extraction constant.However, these researchers did not give the relationship between the extraction constantand the structure of quaternary salts. According to the literature, four relationships forquaternary cations have been reported.

1. Gustavii [101] observed a linear relationship between log EQX and n, the numberof C atoms in an ammonium ion. He extracted picrates into methylene chloride usingprimary amines as well as symmetrical secondary and tertiary amines and symmetricalquaternary ammonium salts. The relationships for quaternary ammonium salts islogEQpicrate ¼ �2:0þ 0:54n.


2. A quantitative parameter for characterizing accessibility was suggested [28]based on the strong dependence of electrostatic interaction on the distance of closestapproach between the cation and anion (which is determined by steric factors). Thisparameter, termed q, is simply the sum of the reciprocals of the length of the linearalkyl chains attached to the central nitrogen of the quaternary cation;q ¼ 1=C#1 þ 1=C#2 þ 1=C#3 þ 1=C#4, where C# is the number of carbon atoms in eachof the four alkyl chains in the quaternary cation.

3. Fukunaga et al. [102] had presented a correlation function based on hydrophile–lipophile balance (HLB) ideas toassess the efficiencyofquaternary salts in thebenzene–watersystem in terms ofHildebrandparameters ½Dð�QXÞ ¼ ð�QX � �Þ2=ð�QX � �Þ2�where �QX, � and�� are, respectively, the solubility parameters of the catalyst, water, and organic solvent.

4. Sirovski [103] proposed that the structure–activity relationship for quaternarysalts can be described quantitatively using Hansch �-hydrophobicity constants. Theseconstants are defined analogously to Hammett and Taft constants [104]:�x ¼ logPx � logPH, where PH is the distribution coefficient for the standard compound,and Px is the same from its derivative with the X substituent in the standard 1-octanol–water system, which has low ion selectivity in relation to halide and hydroxide ions.

The former two relationships (paragraphs (1) and (2) above) were focused on to accessthe distribution of quaternary cations. The equilibrium property cannot reveal when thetotal carbon number for various quaternary salts is the same. In paragraph 3, theHildebrand parameter cannot be easily obtained for all quaternary salts. Hence, we tookthe results of paragraphs 1–3 and the concept of HLB for the surfactant to show that thedispersal efficiency of surfactant or emulsifier molecules is a function of the relative inter-actions of their polar, hydrophilic ‘‘heads’’ with the aqueous phase and of their nonpolar,lipophilic ‘‘tails’’ with the hydrocarbon phase [105,106]. We developed a new model as

HLB ¼ q0:475

ðMT �MHÞþ 9:4

MNX

� �MTBAB ð52Þ

in which 0.475 and 9.4 are hydrophilic group numbers of CH2 and N, respectively, whichwere defined by Davies [107]. The equation of the HLB was developed in respect of theextraction of quaternary salts between two phases based on molecular weights of hydro-philic and lipophilic groups. A linear relationship between extraction constant and HLBwas observed for ammonium cations. An average decrease in log EQX is about 10:5� 2unts per HLB value for various counteranions. The free energies of transfer for ion pairsand dissociated ions were determined and were shown to correspond to the experimentaldata in the literature.

It is of interest to determine the crude free energies of phase transfer between organicand aqueous phases for the quaternaries. This is combined with the free energies oftransfer for halide ions to give the free energies for the tetrabutylammonium and tetra-butylphosphonium ions, which are not well established. Do different salts give the samevalues? Tseng [92] reports the free energy of transfer of some anions from water to variouskinds of solvents based on the distribution data for quaternary salts, and evaluates theextraction behavior of quaternary onium salts in order to understand their performance ina PT-catalyzed reaction system.

An extensive and self-consistent set of data on free energies of transfer of someinorganic salts has been reported [89]. The free energy of the extraction constant, distribu-tion constant, and dissociation constant are expressed as

�Gi ¼ RT lnðiÞ; i ¼ EQX; ETQX; m; Kda; or Kdo ð53Þ


Gustavii [101] and Bockries and Reddy [93] indicated that the dissociation constantof quaternary salts increased when the dielectric constant of the solvents was increased.Nagata [108] reported that the logarithmic value of the association constant of a quatern-ary salt was proportional to the reciprocal of the dielectric constant of the mixed solvent.According to Eq. (53), the dissociation constant decreased slightly with increasing valuesof the reciprocal of the dielectric constant. Parker et al. [109] demonstrated that the freeenergies of transfer are very useful in correlation with the solvent effects on SN2 effects inPTC. The free energies of transfer for the quaternary salts of dissociated ions from waterto the solvent can be written as

�GtQþþX� ¼ �RT ln

�aaQþ

aQþ

�aaX�

aX�

!¼ �RT ln ET

QXKdo

� � ð54Þ

The free energies of transfer for free ions from water to the solvent can be written as

�Gti ¼ �RT ln

�aaiai

� �; i ¼ Qþ or X� ð55Þ

Abraham [88], Czapkiewicz et al. [110], and Taft et al. [111] have reported the freeenergies of transfer of (CnH2nþ1Þ4NX (n ¼ 1–3) for ion pairs and dissociated ions. The freeenergies of transfer for quaternary salts of ion pairs and dissociated ions from water tofour kinds of organic solvents were determined in these studies. The free energies oftransfer for ion pairs were less than those for dissociated ions, i.e., the transfer abilityof ion pairs was greater than that of dissociated ions. The result of the stronger cation–anion attraction in ion pairs is to reduce significantly the magnitudes of the endoergicsolvent cavity terms, as well as the exoergic anion–solvent attractive terms. The stability ofquaternary salts for ion pairs was greater than that for dissociated ions from water to theorganic phase. The result corresponds to that of Taft et al. [111]. The sequences of freeenergy of transfer for quaternary salts are of three sorts: (1) Pþ > Nþ, (2)TBPO� < I� < BPO� < Br� < Cl�, and (3) the long chain of an alkyl group is of lowvalue (Aliquat 336 < TBAC). The stability of ion pairs in dichloromethane (or dissociatedions in chloroform) was the highest among the four kinds of solvents. These results revealthat the incremental charge localization in the anion and decrement in the cation increasesthe stability of quaternary salt in the organic phase.

D. Mass Transport in LLPTC

Usually, it is recognized that the rate-determining step is controlled by the chemicalreaction in the organic phase under LLPTC conditions. For a fast mass transfer rate ofcatalyst between the two phases, the influence of mass transfer on the reaction can beneglected. In the past, the reaction rate was assumed to be independent of agitation andthe surface area of the interface beyond a minimum stirring rate (� 300 rpm). However,the reaction rates can increase with increased agitation in cases where the transfer rate ofanion between both phases is slower than the organic reaction. The phenomenon of masstransfer of quaternary salt between the two phases has received little attention. The reac-tivity of the reaction by PTC is controlled by the rates of the organic and aqueous reac-tions, the partition equilibrium, and the mass transfer steps of the quaternary saltsbetween the organic and aqueous phases [27,28]. The partition equilibrium of quaternaryammonium salts was obtained in our previous work [85,86,92].


The mass transfer rates of catalysts between two phases are difficultly realized due tothe difficult identification of the active catalyst during the reaction [57,112–115]. Masstransfer coupled rapid reactions subjected to LLPTC have been studied extensively[58,63,69,115,116]. Mass transfer rates of catalysts in the reaction of 2,4,6-tribromophenoland tetra-n-butylammonium bromide in a solution of KOH were determined [57,114].

Evans and Palmer [50] first consider theoretically the effect of diffusion and masstransfer in two well-mixed bulk phases of uniform composition separated by a uniformstagnant mass transfer layer at the interface. They studied the effect of the Damkohlernumber, organic reaction equilibrium rate constant, reactant feed-rate ratio, flow rate ofthe organic phase, and the organic reaction reactivity on conversion. Chen et al. [53]derived algebraic expressions for the interphase flux of QY and QX. The reaction para-meters were estimated from experimental data using a two-stage method of optimal para-meters. Naik and Doraiswamy [117] reported that future research should be directedtowards the use of a membrane module as a combination reactor and separator unitwith the membrane serving not merely to carry out the PT-catalyzed reaction, but alsosimultaneously and selectively to recover the organic product. Stanley and Quinn [118]reported the use of a membrane reactor for performing PT-catalytic reactions andincluded theoretical models and calculations to predict the kinetic behavior of the system.Matson [119] investigated the commercial feasibility of such membrane systems. However,the characterization of hydrodynamic phenomena in PT-catalyzed reactions has not beenattempted.

Rushton et al. [120] developed a method for measuring the mass transfer coefficient.However, their method can only be used in systems with unity distribution ratio. Asai etal. [121] measured the liquid–liquid mass transfer coefficients in an agitated vessel with aflat interface. In their later work [122,123] on the alkaline hydrolysis of n-butyl acetate andoxidation of benzyl alcohol in an agitated vessel, the overall reaction rate of PTC withmass transfer at a flat interface was analyzed. The observed overall reaction rate wasconcluded to be proportional to the interfacial concentration of the actual reactant.Wang and Yang [57] investigated the dynamic behavior of PT-catalyzed reactions bydetermining the parameters accounting for mass transfer and the kinetics in a two-phase system. The film theory was applied to interpret the behavior of PTC. The overallmass transfer coefficients of QX (or QY) from an agitated mixture of QX (or QY) werefirst calculated in known qualities of water and the organic solvent by using a simplecorrelation:

lnCQX

CQX;i

þ V

mQX�VV

CQX

CQX;i

� 1

� �" #�mQX

�VV

Vþ 1

!¼ �KQXAt ð56Þ

The overall mass transfer coefficient of QX was obtained by plotting the term on the left-hand side of Eq. (56) versus time. Yang et al. [124] developed a mathematical modelconcerning mass transfer in a single droplet to describe the dispersed phase system.They measured the distribution coefficient and the mass transfer coefficient of a PTcatalytic intermediate between two phases.

Also, the diffusion boundary layer resistances on either side of the membrane filter inmembrane transport processes have been extensively examined [125,126]. Most of thesestudies deal with cases wherein solute diffuses across a membrane filter separating twoaqueous phases with different concentrations. However, the individual film mass transfercoefficients in both liquid phases are unavailable.


The mass transfer resistances strongly depend on the nature of the hydrodynamics inthe contacting device and the mode of operation. Many devices have been used to studytwo-phase mass transfer at or near the liquid–liquid interface. Hence, the hydrodynamiccharacteristics of ion transport through a membrane were presented to evaluate the fea-sibility that this permeation system can be calibrated as a standardized liquid–liquidsystem for studying the membrane-moderated PT-catalyzed reaction. The individualmass transfer coefficients and diffusivities for the aqueous phase, organic phase, andmembrane phase were determined and then correlated in terms of the conventional Sh–Re–Sc relationship. The transfer time of quaternary salt across the membrane and thethickness of the hydrodynamic diffusion boundary layer are calculated and then the effectof environmental flow conditions on the rate of membrane permeation can be accuratelyinterpreted [127].

The mass transfer of quaternary salt from the organic phase into the aqueous phasethrough a lipophilic membrane is indicated in Fig. 3. Assume that the solute activity in thislipophilic membrane is identical to that in the bulk organic solution, then the mass fluxvalues for the individual species are described by

N ¼ ko ½QX� � ½QX�i� � ð57Þ

N ¼ ka ½QX�i � ½QX�� ð58ÞN ¼ km ½QX� i � ½QX� i

� � ð59Þ

The asterisk denotes the species in the organic phase and the membrane phase, respec-tively. The distribution coefficients of quaternary salts between membrane and aqueousphase or organic phase are defined as

m ¼ ½QX� i½QX�i

ð60Þ

and

�mm ¼ ½QX� i½QX�i

ð61Þ

FIG. 3 Mass transfer of the catalyst between two phases and membrane.


According to Eqs. (57)–(61), and for organic volume V and interfacial area S, the rate ofchange of solute concentration can be expressed by

�V

S

d½QX�dt¼ Ko �mm½QX� �m½QX�� ð62Þ

in which

Ko ¼m

kaþ 1

kmþ �mm

ko

� ��1ð63Þ

According to the initial extractive concept that the content of quaternary salt is restrictedto less than 10% in the aqueous solution, the quaternary salt is completely dissociated, i.e.,[QX] approaches zero, and the magnitude of the distribution coefficient �mm is less than 10.By plotting ðV=SÞd½QX�=dt against ½QX], the overall mass transfer coefficient Ko wasobtained by a least-squares regression. The regression factor � is more than 0.99.

If the extraction system is conducted in the absence of membrane, Eq. (63) is rewrit-ten as

Ko ¼m

kaþ 1

ko

� ��1ð64Þ

The values of diffusivities predicted for quaternary salts in the aqueous phase andthe organic phase are in the following descending order: TBPB > TBAB > TBAI �BTBAB and RBAB > TBAI > TBPB > BTBAB; respectively. The diffusivities of qua-ternary salts increased with increasing temperature. The effects of solvents on diffusivitiesare ranked in the following descending order: CH2Cl2 > C6H5CH3 > CHCl3 > C6H6 >C6H5Cl > 1; 2-C2H4Cl2 > H2O. The main influencing factor may be the viscosity of sol-vent. The overall mass transfer coefficients were determined by Lin [127]. The values of ko,ka, and km were calculated by a numerical method for four types of quaternary salts inseven kinds of solvents. Assuming that the hydrodynamic characteristics of the diffusionboundary layer in the aqueous phase and the organic phase were similar in the presence orabsence of the membrane system if the agitation rate was kept below 100 rpm, the indi-vidual mass transfer coefficient of the membrane could then be calculated by subtractingEq. (64) from Eq. (63). The individual mass transfer coefficients increased with increasingagitation rates and temperatures. The sequence of mass transfer coefficient iska � ko > km.

Kiani et al. [125] and Prasad et al. [126] reported the following equation for theintrinsic mass transfer coefficient in the membrane, km ¼ D"=�m�, where " and �m are theporosity and thickness of the membrane, respectively, � is the tortuosity factor of themembrane defined as the actual pore length divided by the membrane thickness, and Dis the diffusivity of species in the bulk liquid phase. The average tortuosities were calcu-lated and found to reduce from 4.3 to 2.7 when the agitation rates increased from 90 to 600rpm. Because the individual mass transfer coefficient of a membrane is not a constant andincreases with increasing agitation rate, the tortuosity decreases slightly with increasingagitation rate according to the equation of Kinai et al. [125].

If the mixing is so vigorous that the diffusion boundary layer can be eliminated, Eq.(62) can be reduced to

�V

S

d½QX�dt

��ka;ko!1

¼ km �mm½QX� �m½QX�� ð65Þ


The extractive effectiveness factor �, defined as the effect of the diffusion boundary layeron the extraction rate of quaternary salt can be characterized in terms of the ratio of Eq.(62) to Eq (65):

� ¼ Ko �mm½QX� �m½QX�� km m½QX� �m½QX�� ¼ ðBio þ Bia þ 1Þ�1

ð66Þ

in which Bio ð¼ �mmkm=koÞ and Biað¼ mkm=kaÞ are Biot numbers for the organic phase andaqueous phase, respectively. Equation (66) represents the mass transfer ratio of conduc-tion rate to convection rate of the quaternary salt at the interface. According to theexperimental data of Lin [127], the values of �, Bio, and Bia are calculated to be around0.96, 0.04, and 0.002, respectively, when the agitation rate is lower than 100 rpm. Hence, itclarified again that the membrane resistance at high agitation rates controls the masstransfer resistance of the membrane extraction.

Usually, mass transfer coefficients can be correlated from the classical equation:

Sh ¼ aRebScc ð67Þwhere Shð¼ kmd=DÞ is the Sherwood number; Re (¼ du�=�Þ is the Reynolds number, Sc(¼ �=�D) is the Schmidt number, D is the diffusivity in the bulk fluid, u is a characteristicvelocity of the fluid such as the mean fluid flow velocity, � is the density, � is the viscosity,and d is a characteristic dimension of the system.

In Eq. (67), a is an experimental constant and c usually has a value of 1/3 [128–130].The value of b depends on the type of equipment and system, and most of the theoriespredict a one-half power on the Reynolds number [131]. The mass transfer from bulksolution to the surface of the membrane is mainly controlled by the turbulence of the fluidmotion created by stirring. The characteristic velocity is defined in terms of the stirringspeed ðu ¼ ndÞ. The values of a and b were determined from the intercept and slope of theline of Sh=Sc1=3 against Re for the specified mass transfer coefficients of ka, ko, and Ko.These parameters are different and are dependent on the system geometry and flow pat-tern. However, it can be concluded that the exponent value on Re varied from 0.2 to 1.0,depending on the design of the membrane permeation system.

The correlating equation [67] established here can be used to evaluate the masstransfer coefficient and the thickness of the diffusion boundary layer, �ð¼ d=shÞ. Thethickness of this layer calculated for an organic solvent and aqueous solution were10�3–10�2 and 10�9–10�7 cm, respectively, for the four types of quaternary salts studied.For a solute crossing a mass transfer resistance film, the transfer time can be approxi-mately estimated by the following equation [131]:

Transfer time ¼ ðfilm thickness)2

diffusion coefficientð68Þ

Based on the data presented here, the estimated transfer times for a solute crossing theorganic and aqueous mass transfer resistance film are about 1–10 and 10�11–10�8 s, respec-tively.

E. Interfacial Phenomena in LLPTC

Starks [132] proposed that the transfer rate of an anion across the interface is largelygoverned by four factors: (1) interfacial area, (2) anion activity and hydration at the


interface, (3) bulkiness of the quaternary salt, and (4) sharpness of the interface. Starksindicated that three of the more important factors affecting the amount of interfacial areainclude interfacial tension, the presence of surfactants, and the degree of stirring or agita-tion. The interfacial area under steady-state stirring conditions will increase with decreas-ing interfacial tension. The chemical natures of the organic and the aqueous phasesdetermine the interfacial tension that exists between these two phases. The quaternarysalt present in the reaction mixture may lower interfacial tension because of its surfactantproperties. Elegant ESCA studies [132,133] suggest that if the anion is highly hydrated itwill not be tightly bound at the interface with the quaternary cation, but rather tend to bemore dispersed in solution, removed from the interface.

Reuben and Sjoberg [134] indicated that all boundaries are difficult to cross: poli-tical, legal, and geographical boundaries, and also phase boundaries in chemical systems.The interfacial mechanism is the most widely accepted mechanism for PTC reactions in thepresence of a base

Interfacial tension is an important property in the process design of liquid–liquidprocesses. The decrement of interfacial tension between both phases leads to an increasedinterfacial area [135]. Because the volumetric rate of extraction was found to be dependenton the interfacial area, interfacial tension data are useful in understanding the effect ofinterfacial area on the volumetric rate of extraction and overall reaction rates for a PT-catalyzed reaction. Dutta and Patil [136] reported that the effect on the interfacial tensionof the water/toluene system has been studied in the presence of four PT catalysts, i.e.,tricaprylmethyl ammonium chloride, hexadecyltrimethyl ammonium chloride, hexadecy-trimethyl ammonium bromide, and hexadecyltributyl phosphonium bromide. Thedecrease in interfacial tension by surfactants increases the interfacial contact area, enhan-cing the volumetric rate of extraction.

Juang and Liu [74,75] presented that the interfacial tensions between water/n-hexaneand water/toluene in the synthesis of ether–ester compounds by PTC could be measured.These two-phase systems contained PT catalyst, an aqueous phase reactant, and/or alkali.The interfacial data could be well described by the Gibbs adsorption equation coupledwith the Langmuir monolayer isotherm.

III. LIQUID–SOLID–LIQUID PHASE TRANSFER CATALYSIS

LLPTC is the most widely synthesized method for solving the problem of the mutualinsolubility of nonpolar and ionic compounds [27–31]. Two compounds in immisciblephases are able to react because of the PT catalyst. However, processes using a two-phase PT-catalytic reaction always encounter the separation problem of purifying thefinal product from the catalyst. Regen [137] first used a solid-phase catalyst [triphasecatalyst (TC) or polymer-support catalyst], in which a tertiary amine was immobilizedon a polymer support, in the reaction of an organic reactant and an aqueous reactant.From the industrial application point of view, the supported catalyst can be easily sepa-rated from the final product and the unreacted reactants simply by filtration or centrifuga-tion. In addition, either the plug flow reactor (PFR) or the continuous stirred tank reactor(CSTR) can be used to carry out the reaction. The most synthetic methods used for triphasecatalysis were studied by Regen and Beese [137–141] and Tomoi and coworkers [142–146].Another advantage of triphase catalysis is that it can be easily adapted to continuousprocesses [147–149]. Therefore, triphase catalysis possesses high potential in industrial-scale applications for synthesizing organic chemicals from two immiscible reactants.


Quaternary onium salts, crown ethers, cryptands, and polyethylene glycol have allbeen immobilized on various kinds of supports, including polymers (most commonlymethylstyrene-co-styrene resin cross-linked with divinylbenzene), alumina, silica gel,clays, and zeolites [137–156]. Because of diffusional limitations and high cost, the indus-trial applications of immobilized catalysis (triphase catalysis) are not fully utilized. Thisunfortunate lack of technology for industrial scale-up of triphase catalysis is mainly due toa lack of understanding of the complex interactions between the three phases involved insuch a system. In addition to the support macrostructure, the support microenvironment isalso crucial in triphase catalysis since it determines the interactions of the aqueous and theorganic phases with the PT catalyst immobilized on the support surface [117]. However, todate, few papers have discussed the microenvironment. The effect of the internal molecularstructure of the polymer support, which plays an important role in the imbibed composi-tion, on the reaction rate has seldom been discussed. In addition to the reactivity, for a TCin an organic and aqueous solution the volume swelling, imbibed different solvent ratio,amount of active site, and mechanical structure of the catalyst must be considered. Hence,these complex interactions in the microenvironment must be solved in order to obtain ahigh reactivity of TC.

A. Characterization and Mechanism of LSLPTC

1. Mechanism of LSLPTC

In general, the reaction mechanism of the fluid–solid reactions involves: (1) mass transferof reactants from the bulk solution to the surface of the catalyst pellet, (2) diffusion ofreactant to the interior of the catalyst pellet (active site) through pores, and (3) intrinsicreaction of reactant with active sites. Triphase catalysis is more complicated than tradi-tional heterogeneous catalysis, because it involves not merely diffusion of a single gaseousor liquid phase into the solid catalyst. Both organic reactant and aqueous reactant existwithin the pores of the polymer pellet. For step (3), a substitution reaction in the organicphase and an ion-exchange reaction in the aqueous phase occurred. Diffusion of both theaqueous and organic phases within the solid support is important and various mechanismshave been proposed for triphase catalysis. However, each mechanism can only explain asingle reaction system. Naik and Doraiswamy [117] discussed these mechanism in theirreview paper.

Tundo and Venturello [155,157] proposed a mechanism for a TC system using silicagel as support to account for the active participation of the gel by adsorption of reagents.Telford et al. [158] suggested an alternation shell model that requires periodical changes inthe liquid phase filling the pores of the catalyst. Schlunt and Chau [150] from the sameresearch group tried to validate this model using a novel cyclic slurry reactor, and indicatedthat only the catalyst in a thin shell near the particle surface was utilized. Tomoi and Ford[142] and Hradil et al. [159] reported that a realistic mechanism involves the collision ofdroplets of the organic phase with solid catalyst particles dispersed in a continuous aqueousphase. Svec’s model [160] for transport of the organic reagent from the bulk phase throughwater to the catalyst particle has been developed in terms of emulsion polymerization.

Because the triphase reaction involves not merely diffusion of a single phase into thesolid support, the organic reaction take places in the organic phase, and the ion-exchangereaction occurs in the aqueous phase. The catalyst support is usually lipophilic. Theorganic phase and aqueous phase fill the catalyst pores to form the continuous phaseand the disperse phase, respectively. The interaction between quaternary salts as well as


the organic phase and aqueous phase play a crucial role in promoting the triphase reactionrate. However, this information is unclear.

2. Characterization of LSLPTC

Poly(styrene-co-chloromethylstyrene) crosslinked with divinylbenzene, which is immobi-lized with quaternary ammonium salts, was investigated for the synthesis of the finechemicals in our previous studies [161–166]. The microenvironment of the polymer sup-port played a crucial role in enhancing the reaction rate. More information about char-acterization of the polymer structure, the interaction between organic solvent, resin, andaqueous solution, and the reuse of the catalyst is required to encourage application.

Wu and Lee [166] report that 24 kinds of ion-exchange resin were used to clarify thischaracter of the resin, including six kinds of commercial ion-exchange microresin, fivekinds of laboratory-produced macroresin, and 13 kinds of laboratory-produced microre-sin, using instrumental analysis by TGA, EA, and SEM-EDS, and the reaction method.The densities of active sites in the resin, titrated using the Volhard method for commercialanion exchangers, were higher than those for laboratory-produced resins.

ð69Þ

Scanning electron microscopy (SEM) analyzes electrons that are scattered from thesample’s surface, and monitors the morphological observation of the polymer resin. Theelemental analysis (EA) is effected by means of energy-dispersive X-ray spectrometer(EDS) methods. The chloride density was shown to be well distributed on the resin surfaceby X-ray images of chloride. It was also demonstrated that the active sites (-NR4Cl) in theresin were completely dispersed. Some other chemical compounds used for synthesizingthe polymer resin were also detected. Although the pretreatment of the resin was con-ducted by washing with water, NaOH solution, and acetone, the salts (Al, Si, and Ca) usedas reactants in the suspension method were slightly retained in the resin.

The immobilized content of tri-n-butylamine in the resin was determined by theTGA, EA, and Volhard methods. The polymer backbone formed in a one-stage processwhere the decomposing temperature range was 300�–450�C. The immobilized resin (mi4-20) was formed in a two-stage process, where the ranges of decomposing temperature forthe two stages were 160�–200�C and 350�–450�C. Although it is tempting to divide the twostages into two distinctive units, the correlation between quaternary salt content andweight loss in the first was qualitative. The weight loss in the first step is equal to theimmobilized amount of the functional group of -NðC4H9Þ3. The accuracy of the analyticaltechnique was within 10%. The commercial ion-exchange resins were revealed in a three-stage process. The decomposed compound and temperature for each decomposition stepare: imbibed water (� 100�C), functional group (160�–300�C), and polymer backbone(350�–450�C). The sequence of the imbibed capability of water is: IRA-900 ð20%Þ > A-


26 > Dowex 1� 2 > A-27 � IRA-410 > IRA-904 > mi4-20 (4%). Most commercial ion-exchange resins are of the hydrophilic functional group type.

In addition, the immobilized amount of the functional group of -NðC4H9Þ3 in theresin was determined from the mass fraction of nitrogen by EA for C, H, and N, and fromthe chloride ion density titrated by the Volhard method. The sequence of determiningmethod for the immobilized content of tri-n-butylamine in the resin wasTGA > EA > Volhard. The analyzed result of the TGA (or EA) method was based onthe elemental weight, and it revealed the real immobilized content. However, the analyzedresult of the Volhard method determined the free chloride ion in the solution by theAgNO3 titration method. The immobilized content of tri-n-butylamine in the resin bythe TGA (or EA) method was > 20% larger than that determined by the Volhard method.The immobilized content of tri-n-butylamine in the resin by the TGA (or EA) method wasindependent of the number of cross-linkages, and only dependent of the number of thering substitution.

These experimental results demonstrate that tri-n-butylamine could be immobilizedcompletely with the active site on the resin for an immobilizion duration of 6 days.However, the immobilized content of tri-n-butylamine by the Volhard method was depen-dent on both the number of cross-linkages and the number of ring substitutions. Theimmobilized contents for the Volhard method are about 50–70% that for TGA (orEA). Since the analyzed results of the Volhard method determined the free chlorideions in the solution by the AgNO3 titration method, the free chloride ion of the activesite were only measured at 50–70% of the amount of immobilized content. The trend ofthe varied content for microresin is larger than that for macroresin. This result indicatesthat the analysis by the Volhard method may be influenced by the diffusion problem, andmay be because the resin did not swell completely in the aqueous solution. On the otherhand, if the resin is used as a TC to react in an actual reaction system, and the resin couldnot swell completely to release all free chloride ions, then the reaction environment wouldbe influenced by the mass transfer of the reactant.

As indicated by Ohtani et al. [32] both organic reactant and aqueous reactant existwithin the pores of the polymer pellet. The HLB of the support structure determines thedistribution of the two phases within the catalyst support [167,168]. Therefore, the dis-tribution of the organic reactant and aqueous reactant within the pores of the polymerpellet will directly influence the reaction. The swollen capability of the resin is used toestimate the validity of the resin. The effect factor of the swollen capability of the resinincludes the cross-linkage, the number of ring substitutions (total exchange capability), theelectronic charge and diameter of the counterion, the polarity of the organic solvent, thecomposition of the functional group, the chemical bonding type between both exchangeions, and the electrolyte concentration in the aqueous solution.

Wu and Lee [166] and Tang [169] reported the amount of imbibed solvent, volumeratio, and porosity of 12 kinds of ion-exchange resin for seven kinds of solvents (dichlor-omethane, chloroform, 1,2-dichloroethane, benzene, toluene, chlorobenzene, and water)when 1 g of the resin was placed in 25mL of the pure solvent. The experimental results forthe commercial ion-exchange resin were as follows:

1. The amounts of the imbibed solvent for the aromatic solvents (benzene,toluene, and chlorobenzene) were larger than those for halide aliphatic solvents(dichloromethane, chloroform, and 1,2-dichloroethane) since the resin was ofthe styrene type; the sequence of the imbibed amount for the aromatic solventswas benzene > toluene > chlorobenzene.


2. The imbibed amounts for water and organic solvent were around 1 g and < 1 g,respectively.

3. The volume ratios were mostly located between 1 and 2.4. The porosities were located between 0.5 and 0; the porosity of most ion-

exchange resins was about 0.5 [170].

The experimental results for the laboratory-produced resins were as follows:

1. The amounts of the imbibed solvent were different, depending on the structureof the resins, and the amount for water was less than that for organic solvent.

2. The amounts of imbibed solvent were in the range 0–3g.3. The volume ratios were almost all located between 1 and 3, and decreased with

increasing cross-linkage of the resin.4. The porosities were located between 0.25 and 0.75.

The porosity and imbibed amount decreased for the solvents benzene, toluene, andchlorobenzene, and increased for the solvents chloroform, 1,2-dichloroethane, dichloro-methane, and water, with increasing number of ring substitutions. These results indicatethat the solubility of water in chloroform, 1,2-dichloroethane, and dichloromethane isgreater than the solubility in benzene, toluene, and chlorobenzene. The imbibed amountfor aromatic solvents was larger than that for halide aliphatic solvents when the number ofring substitutions was small, and the trend was opposite when the number of the ringsubstitutions was large.

Because the functional group of the laboratory-produced resin (tetrabutylammo-nium chloride) is more lipophilic than that of the commercial ion-exchange resin [tetra-methyl- (or ethyl-) ammonium chloride], the amount of imbibed water was larger than thatof the organic solvent for commercial resin; on the other hand, for laboratory-producedresin, the amount of water was less than that of organic solvent. The imbibed amount oforganic solvent for laboratory-produced resin was larger than that for commercial resin.Since the swollen A-27 and IRA-904 was high in the commercial resin in this study, theothers (IRA-900, A26, IRA410, Dowex IX2) were not good for swelling. Hence, they areimproperly used as PT catalysts in an organic phase/aqueous solution reaction system.

Tang [169] reported the amount of the imbibed solvent for commercial resin andlaboratory-produced resin in an organic solvent and in an aqueous solution in the presenceof KOH, NaOH, KCl, and NaCl. Four kinds of salts were used to investigate the swellingphenomenon since the KOH and NaOH were usually used as reactants and the chlorideion was a byproduct in the PTC reaction. Chlorobenzene was chosen as solvent because ofits high boiling point. The imbibed amounts of chlorobenzene and water increased for thecommercial resin, and decreased for the laboratory-produced resin when the salt wasadded. The imbibed amounts of chlorobenzene and water for NaOH were less than thatfor KOH, and that for NaCl was also less than that for KCl since the diameter of the aquaion for Na is larger than that for K. The aqua interaction between metal and waterincreased to increase the swelling capability of the resin when the diameter of the aquaion increased. Also, the imbibed amounts of chlorobenzene and water for KCl were lessthan that for KOH, and that for NaCl was also less than that for NaOH since the diameterof the aqua ion for Cl is larger than that for OH. The imbibed amounts of chlorobenzeneand water for microresin was larger than that for macroresin.

In general, the reaction rate increases with augmentation of the polarity of thesolvent. The apparent reaction-rate constant increased with a rise in temperature. Thesequence of the reactivity for macroresin was CH2Cl2 > CHCl3 > C6H5Cl > C6H6 >


C6H6 > C6H5CH3 and the trend for microresin was similar, except for chloroform. Theincrement of the reactivity of this triphase reaction corresponds to the polarity of thesolvent. Dichloromethane has the highest reactivity among the solvents. However, theboiling point of dichloromethane is 39�C, and it is unsuitable for reaction in the highertemperature system. The activity energy varied with the structure of the resin. Mostactivity energy levels for microresins were greater than those for macroresins, except forchloroform.

The four functions of a base in a liquid–solid–liquid triphase catalytic reaction werereported [165]: (1) reactant; (2) deprotonation of acidic organic compound to become thereactive form; (3) improving the reactive environment in the catalytic pellets, such asswelling volume, imbibed composition, solubility between two phases, etc; and (4) redu-cing the solvation of catalyst and water to upgrade the reactivity of active catalyst in theorganic phase. Wu and Lee [166] showed the effect of base concentration for the reactivityof 4-methoxyphenylacetic acid. The apparent reaction-rate constant was maintainedalmost constant when the NaOH (or KOH) concentration was greater than 1 kmol=m3.The increment of the deprotonation of 4-methoxyphenylacetic acid dramatically increasedthe reactivity of the reaction when the base concentration was below 1 kmol=m3. When thesalt concentration was increased to change the reactivity environment, the reactivity of thereaction was slightly increased with increasing base concentration. The reactivity for KOHwas greater than that for NaOH. The result corresponded with the imbibed compositionof the resin.

The advantage of using a triphase catalytic reaction is that it easily recovers thecatalyst and purifies the product and reactant. Hence, the reuse, stability, and degradationof the catalyst must always be considered. Resins with onium groups may be used forextended periods or repeated cycles only if the catalyzed reactions occur under sufficientlymild conditions to avoid degradation. The degradation of the triphase catalyst may havethree factors: high temperature, strong base, and mechanical degradation. In the pastliterature [143,147,148,166,171–173], the reactivity of the triphase reaction was slightlyinfluenced by the degradation of the catalyst (polymer-supported resin). The active sitewas seen to decrease slightly with increasing base concentration up to 9 kmol=m3. Thenumber of active sites remained constant up to 60�C and then decreased dramatically asthe temperature increased. The degradation of the catalyst with temperature is moresensible than that for base concentration.

In addition to the diffusion resistance of reactants affected by the particle size, it isalso influenced by the characterization of the polymer pellet, i.e., the degree of cross-linkage. In principle, the cross-linkage is related to the covalent bonds between two ormore linear polymer chains. For this reason, the degree of cross-linkage of the polymerwill affect the pore size and the amount of swelling [142,161]. The structure of the polymeris compact for a higher degree of cross-linkage. The pore size of the pellet is increasedwhen a polymer with a low degree of cross-linkage is swollen in an organic solvent. Thus, alower degree of swelling for a higher cross-linking polymer in an organic solvent is adisadvantage, i.e., a large diffusion resistance is obtained for using a higher degree ofcross-linking of the polymer. Hence, the application of a highly cross-linked polymer islimited because of low reactivity in the triphase catalytic reaction.

B. Kinetics and Modeling in LSLPTC

The reaction of triphase catalysis is carried out in a three-phase liquid (organic)–solid(catalyst)–liquid (aqueous) medium. In general, the reaction mechanism of the triphase


catalysis is: (1) mass transfer of reactants from the bulk solution to the surface of thecatalyst pellet, (2) diffusion of reactants to the interior of the catalyst pellet (active sites)through pores, and (3) surface or intrinsic reaction of reactants with active sites. For step(3), the substitution reaction in the organic phase and ion-exchange reaction in the aqu-eous phase occurrs.

�rRX0¼ RcV0

3VcalkIRX

þ V0

kMc�c

�¼�1 CRX0 ¼ kobsCRX0

ð70Þ

Wang and Yang [174–176] have proposed a general dynamic model for triphasecatalysis in a batch reactor. The mass transfer of reactants in the bulk aqueous and organicphases, diffusion of reactants within the pores of the solid catalyst particle, and intrinsicreactivities of the ion-exchange and organic reactions at the active sites within the solidcatalyst were investigated. Desikan and Doraiswamy [151] account for the effect of thereversibility of the ion-exchange reaction. The concentration of the catalytic active siteswithin the catalyst is given as

@qQX

@t¼ �k1CYqQX þ k�1CXqQY þ k2CRXqQY ð71Þ

Mass balances of organic substrate and inorganic species within the catalyst are written as

�""@ �CCRX

@t¼ DRX

r2@

@rr2@ �CCRX

@r

�� sk2 �CCRXqQY ð72Þ

and

"@CY

@t¼ DY

r2@

@rr2@CY

@r

�� s k1CYqRX � k�1qQYCX

� � ð73Þ

respectively.In a heterogeneous catalytic reaction, the intraparticle effectiveness, �c, for a first-

order reaction within a spherical catalyst at steady state is [177]

�c ¼3�cothð3�Þ � 1

3�2ð74Þ

where � is the Thiele modulus:

� ¼ Rc

3

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffik2MccVcatDe

sð75Þ

An apparent overall effectiveness factor of the catalyst is obtained by applying the pseudo-steady-state assumption to the mass balance equations within the catalyst, as

�0 ¼3

�2app

�app coth�app� �� 1

1þ �app coth�app� �� 1

� �=Bim

" #ð76Þ

where �app is the apparent Thiele modulus, and Bim is the Biot number.

�appRcðk2�cq0=DRXÞ0:5

1þ DQYk2DRXk1

� �� 2664

3775

1=2

ð77Þ


Many experimental studies on three-phase catalytic reactions indicated that the reactionrates for the organic phase and the aqueous phase follow pseudo-first-order kinetics[65,69,164,165]. The different type of the reaction expression can be written as

� d �CCRX

dt¼ �kkapp

Mresin

VT

�CCRX ¼ �kkobs �CCRX ¼ �kk 0aVC

�VV�CCRX ð78Þ

and

� dCMY

dt¼ kapp

Mresin

VT

CMY ¼ kobsCMY ¼ k 0aVC

VCMY ð79Þ

where �kkapp and kapp [cm3 ðmin g resinÞ�1] represent the apparent reaction rate constant inthe organic phase and aqueous phase, respectively.

C. Mass Transfer Problem in LSLPTC

The reactivity of a liquid–solid–liquid triphase reaction (i.e., polymer-supported catalyticreaction) is influenced by the structure of the active sites, particle size, degree of cross-linkage, degree of ring substitution, swollen volume, and spacer chain of a catalyst pellet.In the past, the characteristics of a triphase reaction, subjected to the mass transferlimitation of the reactants and ion-exchange rate in the aqueous phase, have been dis-cussed [146,158,162,178,179]. The ion-exchange rate in the aqueous phase affects thereactivity of the triphase reaction.

Past efforts have carried out this investigation macroscopically. The planar phaseboundary in a classical two-phase system cannot be described for the triphase system.Telford et al. [158] suggested an alternating shell model that requires periodical changes inthe liquid phase filling the pores of the catalyst. Schlunt and Chau [150] indicated that thereaction occurred in a thin shell near the particle surface. Tomoi and Ford [142] andHradil et al. [159] proposed that the droplet of organic (or aqueous) phase collided withthe solid catalyst. However, the mechanism and effects of the internal molecular structureof the polymer support with the reaction are seldom discussed. Although some rules werelisted in the text and clarified by the experimental results [27,28], the relationship betweenthe reaction mechanism and polymer resin in a liquid–solid–liquid triphase reaction hasnot been understood completely. Hence, this study aims to discuss the mechanism of apolymer-supported triphase reaction.

Among the vast scope of PTC application [27,28], approximately 40% of PTCpatents involve the hydroxide ion and it has been estimated that approximately 60% ofcommercial PTC applications involve the hydroxide ion [28]. Many papers[61,76,96,116,164,165] have proposed that the reactivity of a reactant in an organic reac-tion is influenced by the base concentration. The base concentration plays a crucial role ina PT-catalyzed reaction. However, the base effect for the reactivity of reactant in a tri-phase reaction was rarely paid attention to.

Most PTC reactions are carried out on an industrial scale in the batch mode inmixer–settler arrangements. In view of the reactor design in the liquid–solid–liquid PT-catalyzed reaction, Ragaini and coworkers [147–149] reported the use of fixed-bed reactorswith a recycling pump or with a recycling pump and an ultrasonic mixer, and emphasizedthe importance of effluent recycle concept. Schlunt and Chau [150] reported the use of acyclic slurry reactor, which allowed the immiscible reactants to contact the catalyst sites incontrolled sequential steps. However, for triphase reactions in liquid–liquid systems where


the catalyst is a solid phase, which reactor type should be properly used in this reactionsystem is not clear.

The substitution reaction of (NPCl2Þ3 with phenol is a sequential reaction [166,169].The reaction type is different from the common one-stage reaction. The experimentalresults can easily demonstrate the relationship between the reaction kinetic limitationand the particle diffusion limitation. In a triphase reaction, the overall kinetic cycle canbe broken up into two steps by virtue of the presence of two practically insoluble liquidphases: a chemical conversion step in which the active catalyst sites (Resinþ with phenolateions) react with the hexachlorocyclotriphosphazene in the organic solvent, and an ion-exchange step in which the attached catalyst sites are in contact with the aqueous phase:

ð80Þ

The function of base in a liquid–solid–liquid triphasic reaction has four roles asmentioned above. In previous studies, it was observed that the reactivity of organic reac-tant varied with the concentration of the base concentration in a liquid–liquid PT-cata-lyzed reaction [96,116,180]. In the liquid–solid–liquid triphase reaction, the effect of baseon the reactivity of reactant (or reactive environment) was rarely paid attention to the �kkappand kapp values dramatically increased with increasing concentration of NaOH when theratio of NaOH to C6H5OH was in the range 1–1.5. This trend corresponds to that ofliquid–liquid phase-transfer catalysis [165]. The apparent activity energies for an organicreaction decreased, and for an aqueous reaction increased when the NaOH concentrationwas increased. The reactive behavior of the organic reaction changed from reaction che-mical control to diffusion control Ea < 42 kJ=mol). The reports of Wang and Wu [161]and Wu and Lee [166] indicated that the imbibed solvent compositions of organic solvent,water, and salt, and the swollen volume of resin increased when the NaOH concentrationwas increased. In other words, the reactive environment of organic reactant near the activesite increased with increasing NaOH concentration. However, the activity energy of theion-exchange reaction increased when the NaOH concentration increased. The behavior ofthe ion-exchange reaction changed to diffusion control from reaction chemical control.Excess NaOH concentration caused the hydroxide ion to react with ResinþCl� to formResinþOH�, which was larger than that of the phenolate ion. Thus, the ion-exchange rateof C6H5O

� with ResinþOH� decreased.The concept of phase plane and superposition [69] was used to change the variable

from time to the consumed ratio of C6H5OH=ðNPCl2Þ3. Figure 4 displays the trajectoriesof the products and the reactant in terms of molar ratio of C6H5OH=ðNPCl2Þ3 consumed.In the organic phase, the reaction mechanism of the reaction of (NPCl2Þ3 withResinþ �OC6H5 can be expressed by Eq. (80). The expression cannot be directly solvedto yield compositions as a function of time because the amount of ResinþOC6H5 in theresin was unknown, and mass transfer resistance influenced the reaction. If the masstransfer resistance of the partially substituted (phenoxy) chlorocyclotriphosphazene inthe particle can be neglected, the maximum yield of monophenolated product can beobtained when approximately 1 mole of phenol per mole of (NPCl2Þ3 has been reacted,and the maximum yield of diphenolated product results when around 2 moles of phenolper mole of (NPCl2Þ3 has been reacted, etc., if the position of the maximum yield shifts tothe right, the higher shifting value means a larger mass transfer resistance of organic


reactant in the particle. In Fig. 4, the maximum yield of monophenolated product shifts tothe right by more than 0.2 unit, and the maximum yield of diphenolated product shifts tothe right by more than 0.1 unit. This reveals that the effect of intraparticle diffusion on theorganic reaction influences the reaction rate. This trend of shifting to the right of themaximum yield was increased with increasing concentration of NaOH.

The reactivity of a triphase reaction is influenced by the structure of the active sites,particle size, degree of cross-linkage, degree of ring substitution, swollen volume, andspacer chain of a catalyst pellet. All these make the triphase reaction a complicatedone. Past efforts have carried out this investigation macroscopically. However, themechanism and effects of the internal molecular structure of the polymer support haveseldom been discussed.

According to the steric effect of phenolate ion reacting with hexachlorocyclotripho-sphazene and the reports of Wu and Meng two-phase catalysis [69]; triphase catalysis[165]), the maximum yield of partially substituted phenolated product was increasedwith increasing degree of substitution reaction. Figure 4 shows that the maximum yieldof monophenolated product was larger than that of the diphenolated product, and themaximum yield of partially phenolated product decreased when the NaOH concentrationincreased (i.e., reactivity of the active site increased). This result reveals that the reactionrate of phenolate reacting with monophenolated (or diphenolated) product was greaterthan the diffusion rate of monophenolated (or diphenolated) product from active site tobulk solution and hexachlorocyclotriphosphazene from bulk solution to active site. Mostmonophenolated (or diphenolated) product reacted in situ with ResinþOC6H5 in the

FIG. 4 Yields of products and conversion of reactant as a function of reactant C6H5OH=ðNPCl2Þ3consumption ratio at different NaOH concentrations: (abcd) 0.5 kmol=m3, (efgh) 0.9 kmol=m3, (ijkl)

1.8 kmol=m3; (*) (NPCl2Þ3, (*) N3P3Cl5ðOC6H5Þ1, (!) N3P3Cl4ðOC6H5Þ2, (&) N3P3Cl3ðOC6H5Þ3,(^) N3P3Cl2ðOC6H5Þ4; ð�Þ N3P3CLðOC6H5Þ5.


neighborhood of the active site. Meanwhile, when the reactivity of ResinþOC6H5 increasedas the NaOH concentration increased, the diffusion resistance of reactants was obvious.

In the present reaction, the overall reaction includes organic substitution and anaqueous ion-exchange reaction, Eq. (80). Two rate-controlling steps influence the reactionrate simultaneously. The reaction is complicated. Hence, from the literature[142,146,158,164,165,181,182], four special relationships are established between the ion-exchange reaction and organic reaction with increasing concentration of organic reactant(NPCl2Þ3, according to Eqs (78) and (79), and these are listed below:

1. It is assumed that the ion-exchange rate in the aqueous phase is much higherthan the substitution reaction rate in the organic phase. The effect of the ion-exchange reaction could be eliminated from the reaction controlling steps.a. The intrinsic organic reaction is the rate-controlling step. Hence, the con-

centration of the active-site of triphase catalyst ResinþC6H5O� remains

constant. The value of kapp is constant. The value of kapp increases withhigher concentration of (NPCl2Þ3 due to an increase in the consumptionrate of phenolate ion. Similar results were obtained by Wu and Tang [164].

b. The organic reaction rate is limited by both reaction kinetics and particlediffusion. The values of kapp increase, and the values of �kkapp decrease withincreasing concentration of (NPCl2Þ3 [181].

c. The organic reaction rate is only limited by film diffusion of reactant fromthe bulk organic solution to the surface of the catalyst pellet; this is the rate-controlling step. The values of �kkapp are constant, and the values of kappdramatically increase with increasing concentration of (NPCl2Þ3. Similarresults were obtained by Tomoi and Ford [142].

2. It is assumed that the organic reaction rate in the organic phase is much higherthan the substitution reaction rate in the organic phase. Controlling the reac-tion could eliminate the effect of the organic reaction.a. If the film diffusion of ion from the bulk aqueous solution to the surface of

the catalyst pellet is the rate-controlling step, the value of kapp remainsconstant because the initial concentration of phenolate ion is kept constant,and the value of �kkapp dramatically decreases [181].

b. If the ion exchange rate is limited by both particle diffusion and filmdiffusion, the value of kapp decreases with increasing (NPCl2Þ3 concentra-tion, and the value of �kkapp dramatically decreases. With low-percentage ringsubstitution, the ion-exchange process is the rate-limiting step[146,153,158,182].

c. If the ion-exchange rate is limited by the intrinsic ion-exchange rate, thevalue of kapp remains constant with increasing (NPCl2Þ3 concentration,and the value of �kkapp dramatically decreases.

3. If the organic reaction rate is limited by both reaction kinetics and particlediffusion, and the ion-exchange rate is also limited by film (or particle) diffu-sion, the value of kapp decreases, and the value of �kkapp also decreases [165].

4. When the organic reaction rate competes with the ion-exchange rate, the valuesof kapp and �kkapp remain almost constant with increasing concentration of(NPCl2Þ3.

If mass transfer resistance influences the reaction, the concentration of the activecatalyst cannot remain constant during the course of the reaction. Also, when the con-centration of organic reactant decreases, both the reaction rate and the effect of mass


transfer of organic or aqueous reactant between solid and liquid phase decrease. However,the apparent first-order reaction-rate constant is increased by decreasing the concentrationof organic reactant [165,181].

Elemental analysis is studied by means of energy-dispersive X-ray spectrometer(EDS) methods. A high Cl peak was detected due to the active site. Some chemicalcompounds (Si, Ca) added in the procedure of synthesizing the polymer resin were alsodetected. Although the pretreatment of the resin was conducted by washing with water,NaOH solution, and acetone, the salts (Si, Ca) used as reaction agents by the suspensionmethod were slightly retained in the resin. A low Cl peak was detected due to the activesite. The peak height for the Cl atom was decreased and was increased for the O atombetween, before, and after the reaction. This finding demonstrates that the phenoxide ionexchanged the chloride ion as counterion on the polymer-supported catalyst during thecourse of the reaction, and did not, however, occupy all the active sites in the catalyst.Hence, the result reveals that the mass transfer resistance of the ion-exchange step influ-enced the concentration of anion on the active site.

The volume and wet porosity of catalyst was increased about three times when thecatalyst imbibed the organic solvent and water. Different catalyst interacts differentlywith the organic phase and aqueous phase. Wu and Lee [166] indicated that the imbibedamount of organic solvent was larger than that of water because the catalyst supportwas lipophilic. The imbibed amount of water was dependent of the amount of ammo-nium cation (i.e., active site). Hence, the imbibed amount of water increases withincreasing number of ring substitutions. If the structure of the resin is rigid (higherdegree of cross-linkage) or of larger particle size, the organic and aqueous phasesremains quiescent in the interior of the resin. The organic and aqueous reactants shouldnot diffuse simultaneously to the active site. The reaction occurs at a shell near thesurface of the resin. When the degree of cross-linkage of the resin is low, the structure ofthe resin is not solid. The flow rate of the organic and aqueous solutions in the interiorof the resin increases with increasing agitation rate. The number of the effective activesites in the resin is increased.

Wu and Lee [166] indicated that the free chloride ions on the active site (measured byVolhard analysis) were at only 50–70% of the amount of immobilized content (measuredby element analysis). The results of the Volhard analysis method determined the freechloride ions in the bulk solution measured by the AgNO3 titration method. Their resultsimplied that the active site in the resin could not react completely with the organic reactantin durating the triphase reaction. According to the experimental results, this reaction is atwo-zone model (or shell–core model). The reaction occurs in a shell zone, and does notoccur in a core zone. The triphasic reaction mechanism and the swollen type of resin areshown in Fig. 5. This mechanism can offer us an understanding of the reaction phenomenain triphase reactions.

IV. SOLID–LIQUID PHASE TRANSFER CATALYSIS

The function of solid–liquid phase transfer catalysis (SLPTC) is to conduct the reaction ofa solid salt and the organic reactant using a PT catalyst that is easily dissolved in theorganic phase in the absence of water. These catalysts can be tertiary amines, quaternaryammonium salts, diamines, crown ethers and cryptands, among which crown ethers, act asthe catalysts because of their specific molecular structures [183–186]. Starks et al. [183]indicated that 100% of the yield of product benzyl acetate was obtained at 258C in 2 h for


the substitution reaction of potassium acetate and benzyl bromide using 18-crown-6 asthe catalyst under solid–liquid PT conditions. This phenomenon of high conversion andproduct yield using SLPTC promotes more research work in investigating this type ofreaction.

The most important step in PT-catalyzed reactions is that the catalyst must have theability to transfer the reacting anion into the organic phase to react with the organicsubstrate. In an aqueous–organic two-phase system, the reacting nucleophile is locatedin the aqueous phase and is usually insoluble or slightly soluble in the organic phase underthe operating conditions. In the situation of the absence of water, the anion nucleophileshould be given by the solid salt reactant, such that the unfavorable side reaction isprobably inhibited. In addition, SLPTC can promote the weak nucleophiles, such assalts of acetate, to have much higher reactivity by eliminating the hydrolysis effect.Hence, for SLPTC, it has the advantages of easy separation of products from reactants,easy selection of organic solvents, easy recovery of catalysts, the inhibition or preventionof unfavorable side reactions, etc., and shows great potential for commercial applications

FIG. 5 Mechanism of the triphasic reaction (a) and the swollen type of resin (b).


[187–196]. Several reactions that cannot be performed in liquid–liquid phases can becarried out efficiently in solid–liquid systems.

Starks et al. [183] have addressed several questions regarding the mechanistic detailsof SLPTC, and those include what are the mechanisms of transport of anions from thesolid phase to the organic phase, the mechanisms of formation of reactive ion pairs, themechanisms of exchange of product anions located in the organic phase with reactantanions located in the solid phase, the effects of particle size on the rates of reaction, themechanistic differences between quaternary cation and crown ethers as PT catalysts, andthe mechanistic role of small quantities of water in SLPTC. Obviously, the behavior of theactive ion pairs or catalytic intermediates is important in realizing the mechanism ofSLPTC.

A. Interfacial Phenomena

1. The Omega Phase

For solid–liquid PT-catalyzed reactions using crown ethers as the catalyst, the correspond-ing cation of the solid reactant has some limitations, e.g., a potassium salt system can onlyuse 18-crown-6 as the catalyst, while 15-crown-5 can only catalyze the reaction of asodium salt. This is because metal salts carried by crown ethers depend on their molecularstructures with the cation size just fitting into the cage of the crown ether; the activecomplex is then transported into the organic phase. Moreover, the solubility of this activecomplex is related to its lipophilicity in the organic solvent [184,185].

In many solid–liquid systems using crown ethers as the catalyst, adding smallamounts of water enhances the reaction rate greatly. A trace amount of water inSLPTC obviously plays an important role. When small quantities of water are added,the solid particles are surrounded by water molecule to form a thin layer. This interfaciallayer between the solid and the organic phases is termed the omega phase, whereby thesolubility of solid reactant in the solution is enhanced to produce easily the active inter-mediate. Liotta et al. [186] indicated that, using 18-crown-6 as the catalyst for the solid–liquid reaction of benzyl halide and potassium cyanide, 92% of the 18-crown-6 (as asolution in toluene) and inorganic salts KCN and KCl resided in the toluene phase;however, about 1–2% of the crown ether was transferred on to the surface of the saltand coated the surface of the salt particles to form a third phase when adding smallamounts of water.

When the omega phase is formed, the overall reaction rate can be described bypseudo-first-order kinetics with respect to the organic reactant. While the reaction followspseudo-zero-order kinetics as the substitution reaction is conducted in the presence ofcrown ether and in the absence of water, it is independent of the benzyl halide concentra-tion. Crown ether directly dissociates the cation of the reacting salt. A reaction mechanismwas proposed for the esterification reaction of solid potassium 4-nitrobenzoate and benzylbromide by using crown ether [197]. The overall reaction is

O2N� C6H4 � COO�Kþ þ C6H5CH2Br��!CHCl3;25�C

O2N� C6H4 � COOCH2C6H5 þKBr ð81Þ


The reaction steps involve [197]:

1. Dissolution of solid potassium nitrobenzoate:

CEorg þKNBsolid Ð CE�KNBorg ð82Þ2. Intrinsic reaction in the organic phase:

CE�KNBorg þ PhCH2Brorg! PhCH2Brorg þ CE�KBrorg ð83Þ3. Release of crown ether:

CE�KBrorg Ð CEorg þKBrsolid ð84Þ

With further additions of water, the overall reaction rate does not inevitably increase, butreaches a maximum with an optimal amount of water added.

2. Solubilization of Solid Salt by Quaternary Ammonium Salts

SLPTC can also be conducted by using quaternary ammonium salt as the catalyst. Thisphenomenon is somewhat different from using crown ether. Vander Zwain and Hartner[198] concluded that, for the reaction of acetate and adeninyl anions in the solid–liquidPT-catalyzed reaction using tricaprylmethylammonium chloride, showed better efficiencythan crown ether. Yadav and Sharma [199] investigated the kinetics of the reaction forbenzyl chloride and sodium acetate/benzoate by SLPTC. They found that cetyldimethyl-benzylammonium chloride was the most efficient catalyst among those studied in thetemperature range 90–139�C, and the rate of reaction in the presence of water was lessthan that in the absence of water. The solubilities of NaOAc and NaCl in toluene assolvent at 101�C are 3:85� 10�5 and 3:24� 10�5 gmol/mL, respectively, while being 2:19�10�5 gmol=mL for the former in the presence of dimethylhexadecylbenzyl chloride. Theconcentrations of chlorides and acetates are 6:25� 10�5 and 5:6� 10�5 gmol=mL.Obviously, the solubilities of these two salts are affected by the reaction with the PTcatalyst. Yee et al. [200] showed that the slower reactions catalyzed by quaternary saltsin a well-mixed batch reactor were caused by the limited effectiveness of quaternary saltsin solubilizing the solid reactant.

Yang and Wu [201] investigated the esterification of dipotassium phthalate withbenzyl bromide in a solid–liquid system. We found that the catalytic intermediate, formedby the solid reactant with tetrabutylammonium bromide, was the key-reacting componentin SLPTC. Yang and Wu [202] explored the kinetics of the O-allylation of sodium phen-oxide with allyl bromide in the presence of quaternary ammonium salt catalyst in a solid–liquid system. The behaviors of the catalytic intermediate tetrabutylammonium phenox-ide, formed from the reaction of solid sodium phenoxide and tetrabutylammonium bro-mide in the solid–liquid phases, are important in conducting the etherification, andpseudo-first-order kinetics are observed.

The past efforts in SLPTC show that not only can the reactions be catalyzed byquaternary ammonium salt, but the interfacial reaction of the solid reactant with thequaternary ammonium salt is also important in this type of reaction. Moreover, thebehaviors of the active intermediate are also influenced by the addition of water in con-ducting the quaternary salts catalyzed reactions. A conceptual scheme describing thereaction mechanism for SLPTC was proposed by Melville and Goddard [203,204], i.e.,heterogeneous solubilization and homogeneous solubilization, by considering the solubi-lity of solid salts in the organic phase. For the heterogeneous solubilization mechanism,


the solid salt directly reacts with the quaternary catalyst at the solid–liquid interface toproduce the intermediate, which then transfers into the solvent and reacts with the organicsubstrate to form the product. For the homogeneous solubilization mechanism, the solidreactant can be dissolved in an organic solvent of generally higher polarity, and then reactswith the catalyst to form the intermediate. Melville and Yortsos [205] performed a theo-retical study regarding rapid homogeneous reactions based on a simple stagnant filmmodel in the system of SLPTC.

Naik and Doraiswamy [206] reported that the homogeneous solubilization could befurther subdivided into four types, models A to D, for the following reactions:

QXorg þMYs=aq Ð QYorg þMXs=aq ð85ÞQYorg þRXorg ! QXorg þRYorg ð86Þ

Model A assumes that the solid dissolution and mass transfer steps are very fast comparedwith the organic reaction and that the solid particles MY and MX are present at theirequilibrium solubility levels in the organic phase. The concentrations of QY and QX in theorganic phase are both constant, i.e.,

CQYo ¼Kq0K þ � ; CQXo ¼

�q0K þ � ð87Þ

Model B assumes that both the ion-exchange reaction in Eq. (85) and the organic reactionin Eq. (86) are under kinetic control with the solid dissolution and mass transfer steps stillfast, and a differential equation describing the variation of QY with reaction time in theorganic phase is required. Model C assumes that MY is no longer at saturation concen-tration in the organic phase, but is at some finite value. The rate of dissolution is governedby the interfacial area per unit volume of the organic phase, the dissolution rate constant,and the driving force between the saturation and the instant concentrations. Both the ion-exchange and the organic reactions take place in the bulk organic phase, and the transportof species from the solid surface to the bulk liquid is assumed to be fast; in addition, thevariation of the interfacial area according to the progress of the reaction should also beaccounted for. Model D accounts for the effect of transport of QY from the thin filmoutside the solid surface to the bulk liquid, and incorporates the rate of the organicreaction. The ion-exchange reaction is assumed to be fast and completed within the film.

In order to describe the solubilization of solid reactant in the organic phase, Yangand Wu [200] performed the ion-exchange reaction of sodium phenoxide with tetrabuty-lammonium bromide in a solid–liquid system. The interfacial reaction and mass transfersteps are shown as follows. The independent ion-exchange reaction is

PhONa ðsÞ þQBr ðorgÞ ! PhOQ ðorgÞ þNaBr ðsÞ ð88ÞThis reaction involves several steps:

(a) Dissolution of PhONa. Traces of water are present in the solid reactantPhONa:3H2O, and the omega phase around the solid particle is formed to enhance thesolubilization of PhONa in the solution. The rate of dissolution of PhONa is very fast,leading to the solid part of PhONa readily in equilibrium with its soluble part. The con-centration of PhONa at the interface is thus kept at its saturation state.

(b) Reaction of PhONa with QBr. PT catalyst QBr reacts with the soluble part ofPhONa to form PhOQ at the solid–liquid interface. The film reaction is assumed to bereversible with the equilibrium constant K1:


PhONa ðorgÞ þQBr ðorgÞk1�! PhOQ ðorgÞ þNaBr ðorgÞ ð89Þ

K1 ¼C PhOQC

NaBr

C PhONaC QBr

ð90Þ

In Eq. (90) the asterisk represents the component concentration in the layer adjacent to thesurface of the solid particle.

(c) Mass Transfer of PhOQ to the Bulk Organic Phase. The intermediate PhOQ trans-fers from the solid–liquid interface to the organic phase, wherein PhOQ has limitingsolubility. The equation for the rate of change is given as

Vorg

dCPhOQ

dt¼ KmAs C PhOQ � CPhOQ

� � ð91Þ

The term As denotes the surface area of the solid particle, which gradually reduces duringthe progress of the reaction, and is derived from the mass balance of PhONa used, i.e.,

As ¼ As 0

NPhONa;0 �NPhOQ

NPhONa;0

� �2=3

¼ As 0 1� q �CCPhOQ

� �2=3 ð92Þ

with

q ¼ NQBr;0

NPhONa;0

and �CCPhOQ ¼CPhOQ

CQBr;0

ð93Þ

The mass transfer coefficient km, which is also dependent on the particle size and theore-tically inversely proportional to the n power of the particle size where n is in the range0.25–1.0 (from high to low Reynolds number), can be expressed as a function of saltconversion:

km ¼ km0

NPhONa;0 �NPhOQ

NPhONa;0

� ��n=3¼ km0 1� q �CCPhOQ

� ��n=3 ð94Þ

By combining Eqs (91)–(93), the rate of change of PhOQ in the organic phase is deduced,

d �CCPhOQ

dt¼ � 1� � �CCPhOQ

� �1� q �CCPhOQ

� �2�nÞ=3 ð95Þ

where

� ¼ CQBr;0C NaBr

K1C PhONaC

QBr

and � ¼ km0As 0

�Vorg

ð96Þ

Yang and Wu [202] showed that near-saturated concentrations of PhOQ in chlor-obenzene as solvent at different temperatures were achieved after about 20 min of opera-tion. The difference in PhOQ concentraitons at various reaction temperatures was notsignificant in this case. This shows that the catalytic intermediate PhOQ can be formedfrom tetra-n-butylammonium salt reacted with PhONa in a solid-liquid system.

Polyethylene glycols (PEGs) can also be used as the catalyst in SLPTC. Chu [207]reported the kinetics for etherification of sodium phenoxide with benzyl bromide usingquaternary ammonium salts and PEG as the catalyst in SLPTC. When PEG is used as thecatalyst, formation of the complex PEG–NaþPhO� mainly occurs at the solid–liquidinterface. The phenoxide anion carried by PEG can dissolve much more than its original


solubility in the organic phase, thus enhancing the overall reaction. The reaction scheme isshown in Fig. 6. For the same reaction system, but using quaternary ammonium saltinstead PEGs, the active intermediate becomes PhOQ produced from solid NaOPh reactedwith catalyst QBr. The solubility of PhOQ varies in different kinds of solvent and leads todifferent reaction rates. The variations in the catalytic intermediate PhOQ with respect totime for chlorobenzene, dichlorobenzene, and heptane are shown in Fig. 7, from which theconcentration PhOQ in heptane is the least. However, the overall reaction rate in heptaneis still at a high level; this shows that the interfacial reaction is dominant in this case [207].

B. Adsorption Effect on the Solid Surface

1. Formation of the Active Complex

In contrast with the reaction mechanism of heterogeneous and homogeneous solubiliza-tion, Yufit et al. [208] proposed a new mechanism for SLPTC that included step-by-stepformation of a cyclic ternary complex [208]. This mechanism is based on the formation oftwo pairs of binary complexes (BCs) and ternary complexes (TCs) obtained from theorganic reactant RX, the solid reactant MY, and the PT catalyst QX adsorbed on asolid salt surface as follows [208]:

RXþMYþQXÐ TC1Ð TC2Ð RYþMXþQX ð97Þ

ð98Þ

They also analyzed the energetics of the substitutions with solid salts of different strengthof M—Y bonds and concluded that the rate-determining step was the rearrangement of

FIG. 6 Reaction scheme for benzyl bromide reacted with sodium phenoxide using PEG as the

catalyst in SLPTC.


the TCs. From the kinetic analysis of different reaction mechanisms, the observed reactionrate constant kobs was determined by the equation:

kobs ¼k½MY�0½QX�0

1þ K ½RX�0 ½MY�0 þ ½QX�0� � ð99Þ

In Eq. (99), k is the combined rate constant and K is the equilibrium constant for thereversible reaction of TC formation.

Generally, many experimental results can be described by applying pseudo-first-order or pseudo-second-order kinetics successfully. Sometimes, however, using confinedkinetic data to elucidate exactly the reaction mechanism is indeed difficult. Hence, severalsimplified reaction mechanisms are usually employed to describe the kinetic behaviors ofthe reaction systems successfully. The technique of topochemistry is an effective methodfor achieving an approximate and quite precise interpretation of the kinetic data. Sirovskiet al. [209] discussed the applicability of the models developed for the topochemical reac-tions in SLPTC. They considered that the simplest kinetic equation, called the Erofeevequation [210,211]:

x ¼ 1� exp �k�nð Þ ð100Þ

with the rate constant k, the conversion degree x, and the parameter n depending on thegeometry of the nuclei, is more appropriate for a description of SLPTC than more com-plicated ones recommended in the literature. Such a kinetic equation is widely used for thedescription of topochemical processes. Sirovski et al. [209] investigated an SN2 reaction ofbenzyl chloride with sodium acetate under SLPTC conditions:

PhCH2ClþAcON�!Aliquat 336PhCH2OAc ð101Þ

They observed that the reaction rate did not follow simple kinetic laws under their oper-ating conditions. A possible reaction scheme was thus proposed [209]:

FIG. 7 Effect of solvent on formation of PhOQ in SLPTC: benzyl bromide 0.005 mol, sodium

phenoxide 0.005 mol, TBAB 0.001 mol, solvent 50 mL, agitation 350 rpm, temperature 70�C; (*)

C6H5Cl, (~) C6H4Cl2, (&) n-C7H16.


QClþAcONa ðsÞ Ð QCl:AcONa ðsÞ ð102ÞQCl:AcONa ðsÞ þ PhCH2ClÐ QCl:AcONa:PhCH2Cl ð103ÞQCl:AcONa:PhCH2Cl! QCl:NaCl:PhCH2OAc ð104ÞQCl:NaCl:PhCH2OAcÐ QCl:NaCl ðsÞ þ PhCH2OAc ð105ÞQCl:NaCl ðsÞ Ð QClþNaCl ðsÞ ð106ÞThey also found that the Erofeev equation described the observed kinetics much better

than other simple kinetic equations. Yufit and Zinovyev [212] compared the kinetic study ofnucleophilic substitution under PTC conditions in liquid–liquid and solid–liquid systems.They observed the effect of initial exponential burst (IB) on the kinetic curve in the reactionwith solid salts for the SN2 reaction of 2-octylmesylate with potassium halides under PTCconditions. In their study, they assumed that the active sites on which the reaction occuredwere present on the solid surface through the formation of complexes of salts, catalysts, andsubstrate [212–215]. They also concluded that the phenomenon of IB was characterized bythe first-order dependence on the initial stage of conversion and by zero-order dependenceup to high conversion. Therefore, the kinetic equation for the reaction becomes a sum oflinear and exponential terms with correlated parameters A and B

½P� ¼ Atþ ½X�ð1� exp�BtÞ ð107Þwhere P represents the key product, [X] is the concentration of product formed by the first-order law, and t is the reaction time. They also proposed a reaction mechanism includingthe adsorption on the solid surface for the solid–liquid system [212]:

KBr ðsÞ þQCl ð1Þ ! KCl ðsÞ þQBr ð1Þ ð108ÞQBr ð1Þ þROMs ð1Þ ! QOMs ð1Þ þRBr ð1Þ ð109ÞQOMs ð1Þ þKBr ðsÞ Ð QOMs:KBr ðadsÞ ð110ÞQOMs:KBr ðadsÞ þROMs ð1Þ Ð QOMs:KBr:ROMs ðadsÞ ð111ÞQOMs:KBr:ROMs ðadsÞ ! RBr ð1Þ þQOMs:KOMs ðadsÞ ð112ÞQOMs:KOMs ðadsÞ þKBr ðsÞ Ð QOMs ð1Þ þKOMs:KBr ðadsÞ ð113ÞKOMs:KBr ðadsÞ ! KOMs ðsÞ þKBr ðsÞ ð114ÞAnother mechanism based on the concept of topochemical reaction, which means

that the reaction rate is dependent on the characteristics or properties of the interface, hasbeen proposed by Yang and Wu [216] in investigating the esterification of linear dicarbox-ylate using SLPTC for solid dipotassium sebacate (SeK2) reacted with benzyl bromide(RBr). The overall reaction is

KOOCC8H16COOK ðsÞ þ 2 C6H5C H2Br ðorgÞQBr

�!C6H5C H2OOCC8H16COOC H2C6H5 ðorgÞ þ 2 KBr ðsÞ

ð115ÞThe reaction steps involves [216]:

1. Dissolution of SeK2 from the solid surface to the organic film.

SeK2 ðsolidÞ ! SeK2 ðorgÞ ð116Þ


With the saturation solubility C SeK2in the organic solvent, the dissolution rate coefficient

kS is dependent on the surface property and the degree of agitation.2. Formation of a transition complex [R–Br–Q–Br] (TC) in the organic phase with

equilibrium constant K1.

RBr ðorgÞ þQBr ðorgÞ Ð TC ðorgÞ ð117ÞFormation of the active transition complex TC from the reaction of QBr and RBr isassumed to be fast and reversible.

3. The substitution reaction of TC and dissolved SeK2 in the organic film with thethird-order rate constant k.

SeK2 ðorgÞ þ 2 TC ðorgÞ ! SeR2 ðorgÞ þ 2 KBr ðorgÞ þ 2 QBr ðorgÞ ð118ÞThe adsorption of QBr on the solid surface of SeK2 plays an insignificant role in thekinetic description. The solid–liquid equilibrium of KBr between its soluble parts andsolid parts is still existed. Wu [217] reported that the kinetic data for S-shape curveswere found in this system, as shown in Fig. 8 for different amounts of potassium sebacateused. This revealed that the catalytic transition complex [R–Br–Q–Br] in the organic phasewould lead to a long induction period for the reaction of SeK2 with TC.

The concentration of TC in the organic phase and the rate of change of componentsare derived as follows [216]:

CTC;org ¼ K1CRBr;orgCQBr;org ð119ÞdCSeK2;org

dt¼ kS C SeK2

� CSeK2;org

� �� kCSeK2;orgC2TC;org ð120Þ

dCSeR2;org

dt¼ kCSeK2;orgC

2TC;org ð121Þ

Equation (120) is simplified by neglecting the term kCSeK2;orgC2TC;org to obtain the concen-

tration of SeK2 in the organic film, CSeK2;org ¼ C SeK2ð1� e�kStÞ, which can be further

transformed into the equation CSeK2;org ¼ C SeK2�ðkstÞn for ease of interpreting the kinetic

FIG. 8 Yields of product dibenzyl sebacate for different molar ratios (r) of dipotassium sebacate to

benzyl bromide: chlorobenzene 50 mL, benzyl bromide 0.01 mol, TBAB 0.0025 mol, agitation 350

rpm, temperature 80�C; r values: (*) 0.125, (^) 0.25, (&) 0.5, (�)1.0, (*) 2.0.


curve of the product SeK2. The parameters � and n are determined for a definite set ofdata and have the physical meaning of characterizing the surface properties of solidreactant SeK2. Different sets of data in ð1� e�kst) give different sets of � and n for bestfitting with a precision greater than 0.99. The apparent reaction rate constant is thendeduced from the equation:

dY

dt¼ 2kK2

1C2QBr;orgCRBr;0C

SeK2

�knS

1þ K1CQBr;org

� �2 tnð1� YÞ2 ¼ kapp;0CRBr;0tnð1� YÞ2 ð122Þ

where

kapp;0 ¼ 2�knSkC SeK2

K21C

2QBr;org= 1þ K1CQBr;org

� �2and Y ¼ 2CSeR2;org=CRBr;0

It it noted that kapp;0 is subjected to the effects of rate of dissolution, intrinsicreactivity, rate of formation of transition complex, catalyst amounts, the solubility ofsolid reactant in the organic phase, and the characteristics of the solid surface, and hasthe dimensions of [ðtimeÞ�1�nðconcentrationÞ�1]. The resultant equation from integratingEq. (122) is similar to the conversion equation deduced from topochemistry theory. Bytaking the natural logarithm on both sides, one can obtain a rather simplified equationused to correlate the kinetic behaviors, i.e.,

� lnY

1� Y

� �¼ � ln

kapp;0CRBr;0

nþ 1

� �þ ðnþ 1Þ½� lnðtÞ� ð123Þ

Moreover, further simultaneous generation of KBr during the reaction of SeK2 with TCwould make the organic solution much more slushy, which in turn would reduce the filmreaction rate due to the steric hindrance when a much higher catalyst amount was used.

2. Deactivation Behavior of Catalyst

In SLPTC, the effect of the side-product salt on the overall reaction rate is sometimesobserved to be severe after a specific reaction time. The kinetic curve shows that thepseudo-first-order reaction initially obeyed is no longer followed at later reaction times,behaving in a diminished kinetic order, and the whole reaction is finally stopped. This sideproduct is produced from the anion-exchange reaction at the solid–liquid interface.Depending on the polarity of the solvent, this generated metal salt is usually difficult todissolve in the organic phase and has a tendency to adsorb on to the surface of the solidparticle. The adsorbed salts thus strongly influence the subsequent anion-exchange reac-tion, from which the active intermediate is formed. Such an effect is more significant forthe fast anion-exchange reaction or at a higher reaction temperature.

Yang et al. [218] investigated the substitution reaction of sodium phenoxide withethyl 2-bromoisobutyrate in a solid–liquid PT-catalyzed system. The deactivation of cat-alyst activity on the apparent pseudo-first-order reaction rate was observed. Such a phe-nomenon results from the salts deposited on the surface of solid particles during theprogress of a reaction. The deposition of salts decreases the rate of formation of the activeintermediate, leading to the observed reaction order change [218]. We applied the pseudo-first-order reaction with catalytic deactivation kinetics to show that the initial reaction ratewas not influenced by the agitation rate when exceeding 350 rpm, but the deactivation rateincreased with increasing stirring speed and the amount of catalyst used. Adding a smallamount of water resulted in a reduction in the apparent reaction rate. A more lipophilic


quaternary cation solvates the solid reactant anion much more easily, and leads to a fasterinitial reaction rate. The order of reactivity for different PT catalysts is determined asTBPB > TBAB > TBAI � TBAHS � Aliquat 336 for the reaction.

Yang et al. [219] investigated the kinetics of the etherification of ethyl 2-bromoiso-butyrate (RX) with potassium 4-benzyloxyphenoxide in the presence of potassium iodidein SLPTC. In that work, we found that for various molar ratios of TBAB to RX (denotedas f ) the yield of product ArOR increased with increasing catalyst amounts up to f ¼ 0:60.Too much catalyst employed in the presence of KI results in the reduction of catalyticefficiency. This effect is due to two major reasons: first, the solubility of the catalyticintermediate in the organic solvent is limited; second, the formation of the catalytic inter-mediate in chlorobenzene is retarded because use of a higher amount of catalyst inducedrapid deposition of the generated potassium salts on the solid surface. Adding the extrasalt KI enhances the reactivity of PT catalyst, but the active intermediate in the organicphase is diminished when much KI is present. Small amounts of KI promote the conver-sion of RX into RI, which is more reactive in the organic reaction. The reaction stepsconcerning the deactivation of the catalyst are shown below [219].

The overall reaction is

ArOK ðsÞ þRX ðorgÞKI;QX

�! ArOR ðorgÞ þKX ðsÞ ð124Þ

The reaction mechanism for the overall reaction is as the following steps:

ArOKðsÞ! ArOK ðorgÞ ð125Þ

KI ðsÞ! KI ðorgÞ ð126Þ

KI ðorgÞ þQX ðorgÞK1

�! KX ðorgÞ þQI ðorgÞ ð127Þ

KI ðorgÞ þRX ðorgÞK2

�! KX ðorgÞ þRI ðorgÞ ð128Þ

KXðorgÞ! KX ðsÞ ð129Þ

ArOK ðorgÞ þQX ðorgÞK3

�! ArOQ ðorgÞ þKX ðorgÞ ð130Þ

ArOK ðorgÞ þQI ðorgÞK4

�!

ArOQ ðorgÞ þKI ðorgÞ ð131Þ

RX ðorgÞ þArOQ ðorgÞkb

�!ArOR ðorgÞ þQX ðorgÞ ð132Þ

RI ðorgÞ þArOQ ðorgÞkb

�!ArOR ðorgÞ þQI ðorgÞ ð133Þ

The rate of change of ArOR is then expressed as

dCorgArOR

dt¼ kaC

orgRX þ kbC

orgRI

� �C

orgArOQ ð134Þ


Taking the mass balance for the cation Q of the PT catalyst in the system gives

CorgQX;0 ¼ C

orgQX þ C

orgQI þ C

orgQrOQ ð135Þ

The expression for the active intermediate is

CorgArOQ ¼

CorgQX;0

1þ 1Corg

ArOK

CorgKX

K3

þ CorgKI

K4

� � ð136Þ

The concentration of ArOQ depends on the amounts of ArOK, KX, and KI, and theinitial usage of the catalyst QX. Combining the mass balance equation for initial RX in theorganic phase with Eq. (134), a deactivation function � can be introduced in the situationunder decline of catalytic efficiency, leading to the following equations:

dCorgArOR

dt¼ �kapp;0 C

orgRX;0 � C

orgArOR

� �ð137Þ

with

kapp;0 ¼ ðka þ kbK2CorgKI =C

orgKXÞCorg

QX;0

and

� ¼ 1

1þ K2CorgKI

CorgKX

� �1þ 1

CorgArOK

Corg

KX

K3

þ CorgKI

K4

� � � ð138Þ

In Eq. (137), kapp;0 is the initial apparent reaction rate constant and is dependent on theamounts of KI, KX, and QX in the organic phase. If the rate of change of ArOR follows apseudo-first-order reaction, then � would be approximately a constant. In such cases, nodeactivation effect appears. If the reaction rate behaves as a diminished first order, then �decreases with the progress of the overall reaction.

To evaluate the exact variation of � with time is to measure the rate of deposition ofKX and the other parameters directly or to apply an empirical correlation relating to theeffect of the decrease in C

orgArOK on the overall reaction. The expression of the deactivation

function and the kinetic data would determine the form of �, such as

� ¼ 1

ð1þ kdtÞ2or � ¼ 1

1þ kdtð139Þ

Lepertit and Che [220] discussed the definitions of interfacial co-ordination chem-istry (ICC) and surface organometallic chemistry (SOMC) and compared their maincharacteristics and applications. The concepts of ICC applied to catalyst preparation,adsorption, and relations with catalysis are also useful in the development of interfacialmechanisms.

C. Mass Transfer Effects

Sufficient kinetic information should be collected to proceed the process design for aspecific reaction system. The factors affecting the performance of SLPTC includes agita-tion rate, particle size of solid salt, reaction temperature, the amount of solid reactant, thekinds and amount of PT catalyst, the solubility and the dissolution rate of solid reactant inthe organic solvent, extra addition of other metal salts, the polarity, surface tension, and


viscosity of the solvent, the amount of organic reactant, and the presence or absence ofwater. The reactor employed in SLPTC is usually an agitated batch type with the solid saltsuspended in the solution. The mass transfer rate between the solid and the liquid phases isimportant in designing a SLPTC reactor.

In an agitated reactor, the effect of mass transfer resistance can be reduced to aminimum by adjusting the stirring speed. The mass transfer coefficient is also a function ofthe size of suspended particles. From the point of view of reactor design, to maintain theuniformity of the desired product from batch to batch the particle size distribution of thesolid reactant should be in a rather narrow range to render the mass transfer resistanceunimportant.

1. Mass Transfer Coefficient

Melville and Goddard [204] used rotating disk flow to measure the mass transfer coeffi-cient between the solid and liquid phases in SLPTC for the reaction of benzyl chloride andsolid potassium acetate using Aliquat 336 as the catalyst in acetonitrile as solvent. Theconcentration of quaternary ammonium acetate is expressed in the following equations:

CQOAc ¼ � 1� e��t� � ð140Þ

� ¼ K 01C KOAc; � ¼ kð1þ SK 01Þ

1þ K 01ð141Þ

K 01 ¼K1CQCl

CKC l

; k ¼ A

V

Dk

�; S ¼ DQ

DK

ð142Þ

where K1 is the equilibrium constant for the reaction, KOAc þQCl! KClþQOAc; Di isthe diffusivity coefficient of component i; � is the film thickness of mass transfer; A is thesurface area of solid potassium acetate; and V is the liquid volume. They concluded thatthe solid–liquid reaction was effected after the solid potassium acetate dissolved in thehigh-polarity solvent. Yee et al. [200] also applied rotating disk flow to carry out the masstransfer experiments for solid benzoate. The mass transfer coefficient K is obtained as

K ¼ 0:6205D1=3!1=2

�1=6f ðScÞ ð143Þ

f ðScÞ ¼ 1þ 0:2980Sc�1=3 þ 0:01451Sc�2=3 ð144Þwhere ! is the angular velocity (rad/s), � is the kinematic viscosity (cm2=s), D is thediffusivity (cm2=sÞ, and SC is the Schmidt number (�=DÞ:

2. Pseudo-First-Order Kinetics Neglecting Mass Transfer Effect

Pseudo-first-order kinetics are usually observed in many solid–liquid PT-catalyzed reac-tions when the mass transfer effect is insignificant. For the reaction between the organicsubstrate RX and the nucleophile MY, the equation is

RX ðorgÞ þMY ðsÞ þQþX�ðorgÞ ! RY ðorgÞ þMX ðsÞ þQþX�ðorgÞ ð145ÞIn the case of some solid MY dissolved in the organic phase, the equilibrium state isachieved in a short reaction time:

MY ðsÞ $MþY�ðorgÞ ð146Þ


The ion-exchange reaction occurs at the interfacial zone to form QY, then conducting theintrinsic reaction:

MþY�ðorgÞ þQþX�ðorgÞ $ QþY�ðorgÞ þMþX�ðorgÞ ð147ÞQþY�ðorgÞ þRX ðorgÞ ! RY ðorgÞ þQþX�ðorgÞ ð148Þ

The regenerated QþX�ðorgÞ continues to catalyze the formation of QþY�ðorgÞ, andM+X�(org) is always in equilibrium with MX(s):

MþX�ðorgÞ $MX ðsÞ ð149ÞThe rate of equation is derived as

� d½RX�dt¼ korg½RX�½QþY��

¼ korgKe

½MþY��½MþX�� ½RX�½QþX�� ð150Þ

Ke ¼½QþY��½MþX��½QþX��½MþY�� ð151Þ

If the value f ¼ ½MþY��=½MþX�� is approximately a constant, the rate of the equation canbe expressed in a pseudo-first-order form:

� d½RX�dt¼ kapp½RX� ð152Þ

kapp ¼ korgKef ½QþX�� ð153ÞDuring the course of reaction, when the value of f is gradually diminished, the initialreaction rate and deactivation rate should be applied. With or without adding waterinfluences the mass transfer rate from the interface into the organic phase.

D. Kinetic Modeling of Heterogeneous Solubilization

Naik and Doraiswamy [206] developed a mathematical model for the case of heteroge-neous solubilization that involves the steps of ion exchange in the solid phase, interphasetransport of the catalyst and the intermediate, and the organic reaction. In this model, ionexchange occurring within the solid phase is assumed due to the possible deposition of theproduct salt MX on the solid surface retarding formation of the catalytic intermediate.The controlling step can either be the liquid-phase transfer steps, the diffusion within thereactive solid, the adsorption–desorption steps, the surface ion-exchange reaction, or theliquid organic reaction. This treatment is similar to that in gas–solid catalytic reaction.The controlling step may shift to another step continuously with time. The reaction oforganic RX with solid MY in the presence of PT catalyst QX is considered, and a poroussolid wherein the ion-exchange reaction takes place throughout the whole pellet, ratherthan at a sharp interface due to the liquid penetrating into it is assumed. The governingequations are derived as follows [206]:

RX ðorgÞ þMY ðsÞ �!QXRY ðorgÞ MX ðsÞ ð154Þ

Within the solid phase:


@CSQX

@t¼ De

r2

@ r2 @CSQX

� �=@r

h i@r

� ks CsQX �

CsQY

K

� �ð155Þ

@CsQY

@t¼ De

r2@ r2 @Cs

QY

� �=@r

� �@r

þ ks CsQX �

CsQY

K

� �ð156Þ

In the organic bulk solution:

dCorgQX

dt¼ k2C

orgRXC

org

QY � kqa CoegQX � Cs

QX

h ið157Þ

dCorgRX

dt¼ �k2Corg

RXCorgQY ð158Þ

CorgQY ¼ q0 � C

orgQX � Cs

QX;a � CsQY;a ð159Þ

CsQX;a ¼ 3

ð10

�2CsQXd� ð160Þ

The initial and boundary conditions are:

IC: t ¼ 0; CorgRX ¼ C0

RX; CsQX ¼ 0; Cs

QY ¼ 0; CorgQX ¼ q0; C

orgQY ¼ 0 ð161Þ

BC: r ¼ 0;dCs

QX

dr¼ dCs

QY

dr¼ 0 ð162Þ

r ¼ R; Dq

dCsQX

dr¼ kq C

orgQX � Cs

QX;R

h ið163Þ

r ¼ R; Dq

dCsQY

dr¼ kq C

orgQY � Cs

QY;R

h ið164Þ

The above equations can be rendered dimensionless in terms of Thiele’s modulus, Biotnumber for mass transfer, and nondimensional time and distance, which are defined as

�2 ¼ ksR2

De

; Bim ¼kqR

Dq

; � ¼ Det

R2; � ¼ r

Rð165Þ

In the analysis of heterogeneous solubilization, the role of the solid-phase reaction ininfluencing the overall reaction is different from that for the usual gas–solid catalyticreaction. The most important situation is that the film and internal diffusion effects withinthe solid and at the solid–liquid interface are significant.

V. TRI-LIQUID PHASE TRANSFER CATALYSIS

Neumann and Sasson [221] investigated the isomerization of allylanisole using PEG as thecatalyst in a toluene and aqueous KOH solution. They observed that a third-liquid phasewas formed between the aqueous and the organic phases. This was the first report regard-ing tri-liquid PTC. In 1987, Wang and Weng [222] performed the reaction of benzylchloride and sodium bromide using tetra-n-butylammonium bromide as the PT catalystin liquid–liquid phases. They found that the overall reaction rate rapidly increased whenthe amount of catalyst used exceeded some critical value. In such reaction conditions, thePT catalyst was found to be concentrated within a viscous liquid phase that was insoluble


in both aqueous and organic phases [222]. This liquid phase enhanced the overall reactionrate as much as several fold that in two-liquid phase systems with PTC, and was called thethird-liquid phase. The third-liquid phase was found to contain little of the organic andaqueous reactants, but mainly the highly concentrated catalyst, which exhibited hydro-philic and lipophilic properties. In the bromide–chloride exchange reaction system ofWang and Weng [222], the third liquid phase was found to consist mainly of Bu4NBr,small amounts of toluene, water, and sodium bromide. Above about 70% of the tetra-butylammonium bromide was forced to form a separate liquid phase. The organic andaqueous reactants readily reacted with the concentrated catalyst to yield a high reactionrate. The PTC reaction in this situation was termed as tri-liquid PTC (TLPTC).

From the point of view of industrial practice, the formation of a third phaseprovides not only enhancement of the reaction rate, but also easier separation of thePT catalyst from the product stream than that in a two-liquid phase. However, in someparticular reaction systems, the catalyst could lose as much as approximately 25% ofthe initial amount used. Catalysis by TLPTC was briefly reviewed by Naik andDoraiswamy in 1998 [223]. The key results from the previous publications are discussedas follows.

A. Formation of the Third Liquid Phase

Tetrabutylammonium salts are found to be able to form a third liquid phase underappropriate conditions. In principle, the formation of a third catalyst phase can beobtained by using a PT catalyst having limiting solubility both in the aqueous phaseand organic phase under the interaction of other concentrated ingredients. Ido et al.[224] effected the elimination reaction of 2-bromo-octane with aqueous sodium hydro-xide using PEG as the catalyst [224]. By adding small quantities of methanol the solu-bility of PEG in the organic phase was greatly reduced, leading to the formation of athird liquid phase. Mason et al. [225] investigated the elimination of phenethyl bromideto styrene using tetrabutylammonium bromide under PT-catalytic conditions. Theyfound that the rate of reaction was accelerated rapidly due to the addition of morethan the critical amount of PT catalyst, and the third phase was rich in catalyst. Whenthe PT catalyst used was replaced by the tetrapropyl- or tetrapentyl-ammonium salts,the third liquid phase was not formed, and the precipitation of excess catalyst wassimply induced.

Wang and Weng [226] explored the effects of solvents and salts on the formation of athird liquid phase for the reaction between n-butyl bromide and sodium phenolate. Theyconcluded that the polarity of the solvent and the amount of NaOH are two importantfactors in influencing the formation of a third liquid phase, the distribution of catalyst, andthe reaction rate. The aqueous reactant NaOPh also exhibitis significant behavior incertain conditions. With the catalytic intermediate QOPh produced by the reaction ofNaOPh and the catalyst QBr, NaOH has the ability to extract QOPh from the organicphase or the third liquid phase into the aqueous phase. For example, when the amount ofNaOH added was 2 g, the amount of catalyst in chlorobenzene decreased to less than 10%of the original content, while the concentration of the catalyst in the aqueous phaseincreased with increasing NaOH added. In addition, when hexane was used as the solvent,adding a small amount of NaOH caused the disappearance of the third liquid phase, whichhad been formed before the addition of NaOH. This phenomenon is due to the dissolutionof QOPh in the aqueous phase.


Ido et al. (1997) reported a halogen substitution between benzyl chloride in theorganic phase and potassium bromide in the aqueous phase catalyzed under the thirdliquid phase that was formed by changing the concentration of KBr, types of PT catalysts,and the organic solvents [227]. Yadav and Reddy (1999) investigated the n-butoxylation ofp-chloronitrobenzene (PCNB) using the base NaOH under tri-liquid phase conditions.The typical composition of the third liquid phase was 55.12% of toluene, 22.52% oftetrabutylammonium bromide, 4.96% of p-chloronotrobenzene, 14.51% of water, and2.89% of n-butyl alcohol by weight. Distribution of the catalyst between the organicphase and the third liquid phase indicates about 89% of the total catalyst residing inthe third phase, and the overall reaction rate is attributed to the contribution of thereaction occurring in both the organic and the third liquid phases.

Jin et al. [229] further performed the dehydrohalogenation of 2-bromo-octane withdodecane as the organic solvent and potassium hydroxide in the aqueous solution toinvestigate the synergistic effect of two PT catalysts in the situation of a third liquidphase using a combination of tetrahexylammonium bromide and PEG. They concludedthat a molecule of tetrahexylammonium bromide surrounded by many molecules of waterand some PEG 200 led to the effect of water on the catalytic activity of tetrahexylammo-nium bromide becoming weaker when the amount of PEG was increased.

In summary, the operating conditions influencing the formation of the third liquidphase are: (1) type and quantity of the aqueous reactant, (2) type and quantity of PTcatalyst, (3) reactant and product in the organic phase, (4) the addition of otherinorganic salts, (5) lower polarity of the organic solvent, and (6) the reaction tempera-ture. Increasing the reaction temperature benefits the formation of the third liquidphase due to the breakage of hydrogen bonding between the PT catalyst and thewater molecule.

B. Interfacial Mechanism of TLPTC

1. Reaction Mechanism

The typical reaction mechanism for tri-liquid PTC in a batch reactor under agitation isillustrated in the schematic diagram of Fig. 9. Three types of reaction scheme consideringthe partition of the catalyst in the different phases and the place where the inherentreaction occurred have been proposed [226,227]. For the substitution reaction of alkylhalide (RX) and aqueous reactant metal salt (MY) using quaternary ammonium salt (QX)as the catalyst, the different types of reaction are addressed as follows [226].

FIG. 9 Reaction scheme for benzyl bromide reacted with sodium bromide in TLPTC.


Type I. The catalyst resides in the organic phase in a significant quantity:

ð166Þ

For type I, the partition of catalyst between the organic and aqueous phases is importantin determining the intrinsic reaction rate and the utilization of the catalyst in the organicphase.

Type II. The catalyst resides in the aqueous phase in a significant quantity:

ð167Þ

For type II, the catalyst mostly stays in the aqueous phase, and the transferred RBr fromthe organic phase to the aqueous–organic interface reacts with the catalytic intermediateQOPh that is transported from the aqueous to the interface. The intrinsic reaction ismainly conducted at the interface between aqueous and organic phases.

Type III. The third liquid phase appears with the catalyst and active intermediate allresiding in this viscous phase:

ð168Þ

By adding more NaOPh to the reaction system the catalyst is salted out to form the thirdliquid phase. The active intermediate is then formed at the interface of the aqueous and thethird liquid phases by the reaction of QBr and NaOPh, which is transferred from the


aqueous bulk phase. The main intrinsic reaction of QOPh and RBr is conducted in thethird liquid phase. The product ROPh is then transferred into the organic phase.

In reality, not all the catalyst would exist in the third liquid phase especially for thathaving high solubility in the aqueous phase. A distribution of the catalyst between theaqueous phase and the third liquid phase is probably retained. Hence, some parts of QOPhwould be produced in the aqueous phase, resulting in the distribution of QOPh betweenthe aqueous and the third liquid phases.

2. Factors Affecting Catalyst Activity in TLPTC

In TLPTC, the essential step is to form the third liquid phase by adjusting the contents ofinorganic salts and PT catalyst, and the interaction of the strong bases added. The overallreaction rates catalyzed by applying the third liquid phase are commonly enhanced tre-mendously, compared with the same reaction proceeding in liquid–liquid phases. Thevariables influencing the reaction rate can be summarized as follows:

(a) Agitation Speed. Agitation plays an important role in a multiphase reaction sys-tem. Increasing the agitation rate increases the mass transfer rate of the componentbetween the immiscible phases and reduces the droplet size of the dispersed phase.Under agitated conditions, the mass transfer resistance at the interface between thethird liquid layer and the organic phase is affected by the droplet size. When the agita-tion rate increases to a critical value, the limiting step is dominated by the reactionwithin the catalyst-rich phase [224–227,230–233]. Yadav and Reddy [228] reported thatwith a speed of agitation from 650 to 1400 rpm, the rate of reaction increased withincreasing stirring, and after 1400 rpm the rate was independent of the interfacial masstransfer resistance.

(b) Amount of Phase Transfer Catalyst. Different types of PT catalysts including qua-ternary ammonium salts and PEGs have been observed to enable the formation of thethird liquid phase, but under different conditions. Their common behaviors show thesharp discontinuity of the reaction rate before and after the formation of the thirdliquid phase. The observed reaction rate in the case of the tri-liquid phase increases line-arly with the total moles of quaternary ammonium bromide [228]. However, the qua-ternary ammonium salts with shorter alkyl chains show less tendency to form the thirdliquid phase, e.g., tetrapropylammonium bromide is ineffective for use as a catalyst in atri-liquid system. Mason et al. [225] indicated that, when a reaction mixture forming athree-liquid system was reconstructed by separating the middle third liquid phase, thereaction rate dropped by over half.

Ido et al. [227] investigated the kinetics of a halogen exchange reaction in a three-liquid phase system and applied first-order kinetics to describe the overall reaction rate.They observed that the reaction rate constant includes the contributions of reactions in thethird-liquid phase and in the organic phase, and is a first order proportional to the totalcatalyst moles mcat. The reaction rate kinter occur at the interface between the aqueousphase and the organic phase is also important. Their results are shown as the followingequations [227]:

� dmA

dt¼ korgxorg þ kinterxaq þ kthirdDAxthird� �

mcatCA;org ð169Þ

yA;org ¼Vorg

Vorg þDAVthird

exp �kobsmcat

Vorg

t

� �ð170Þ


kobs ¼Vorg

Vorg þDAVthird

korgxorg þ kinterxaq þ kthirdDAxthird� � ð171Þ

where xA denotes the mole fraction of the catalyst existing in the different phases, and KA

represents the distribution of A (benzyl chloride) between an organic phase and a thirdphase in equilibrium and is defined as

DA CA;third

CA;org

¼ mA;thirdVorg

mA;orgVthird

ð172Þ

Taking the logarithm for Eq. (170), one can determine the observed rate constant kobsfrom the experimental data by plotting ½� lnðyA;orgÞ� versus time t:

� lnðyA;orgÞ � lnVorg

Vorg þDAVthird

� �¼ kobs

mcat

Vorg

t ð173Þ

(c) Reactant and Alkali Salt in the Aqueous Phase. The overall reaction rate inTLPTC usually increases with the increase in amount of strong base reactant in theaqueous phase. In contrast with a base-catalyzed elimination reaction, the third liquidphase already formed will be precipitated under the excess base to dehydrate the cata-lyst phase. In the presence of 49% of NaOH, two liquids and one solid are observedinstead of three liquids at a somewhat lower base concentration.

Ido et al. [227] found that increasing the aqueous reactant KBr increases the reactionrate in TLPTC. The ionic strength in the aqueous phase also affects the ease of forming thethird liquid phase, since adding extra salts tends to salt out ion pairs produced from theaqueous reactant with the quaternary salt. In the system of n-butyl bromide reacted withsodium phenolate [225], the water molecules form hydrogen bonds with NaOPh as well aswith QOPh, leading to the amount of tetrabutylammonium salts in the organic phase andin the third liquid phase increasing with the amount of NaOPh added, which in turnenhances the overall reaction rate.

(d) Organic Solvent and the Reaction Temperature. In general, the more polar theorganic solvent the faster is the overall reaction rate in LLPTC due to the increasingsolubility of the catalytic intermediate in the organic phase, and leading to much easiertransport of ion pairs into the solvent to react with the organic substrate. In contrast,in TLPTC, the solubility of the catalytic ion pairs in the organic solvent should be lowenough to push the catalyst to form a separate phase. Thus, a solvent with low polarityor a nonpolar one is favorable. Under the same conditions of using KBr and a catalyst,the reaction rate in dodecane was observed to be much faster than in toluene [227].

Increasing the reaction temperature accelerates the reaction rate [221–226,230–233].However, the catalyst existing in the third liquid phase as well as in the organic phaseshould still be maintained. Under strong base conditions in TLPTC, the catalyst and theactive intermediate have the tendency to decompose at a high temperature, hence, a limit-ing reaction temperature should be kept in maintaining the third liquid phase.

C. Kinetic Modeling for TLPTC

Yang [234] has developed a theoretical model to investigate the effects of mass transfer anddistribution of the catalyst within the third liquid phase and organic or aqueous phase onthe overall reaction rate. The modeling considers the dispersed organic droplet surroundedby an interfacial catalyst layer under agitation conditions, as shown in Fig. 10. This type ofdroplet is similar to some oil/water emulsions in the presence of surfactants. The reactant


MY in the aqueous phase undergoes a substitution reaction with the organic reactant RX toform the product RY. Under some appropriate conditions, by introducing extra inorganicsalts or reactants into the system, a separate liquid phase appears and is composed of PTcatalyst (QX), active intermediate (QY), a little water, and organic solvent. The third liquidlayer at the aqueous/organic interface exists if the solubility of QY in both the organic andaqueous phases is limited. This reaction system is shown in the following scheme:

ð174Þ

In the third liquid phase system, the organic phase is considered as the dispersedphase with spherical and rigid droplets, and with a high distribution coefficient of catalystbetween aqueous and third phases. The rates of change of RX in the organic phase andQY/QX in the third phase are formulated as follows [234]:

FIG. 10 Conceptual scheme for dispersed droplet and third liquid layer in TLPTC.


@CorgRX

@t¼ DRX

r2@

@rr2@Corg

RX

@r

� �for r � rd ð175Þ

@CcatQY

@t¼ DQY

r2@

@rr2@Ccat

QY

@r

� �for rd � r � rc ð176Þ

@CcatQX

@t¼ DQX

r2@

@rr2@Ccat

QX

@r

� �for rD � r � rc ð177Þ

In which rd is the radius of the organic droplet and rc is the radius of the organic dropletplus the thickness of the catalyst layer. The relationship between rc and rd is

rc ¼ 1þ Vcat

Vorg

� �1=3

rd ð178Þ

The initial and boundary conditions are

at t ¼ 0; CorgRX ¼ CRX;0 for 0 � r � rd

CcatQX ¼ 0; Ccat

QY ¼ 0 for rD � r � rc ð179Þ

at r ¼ 0;@Corg

RX

@r¼ 0 ð180Þ

at r ¼ rd;3

rd

� �DRX

@CorgRX

@r¼ �k2Corg

RXCcatQY

3

rd

� �DQY

@CcatQY

@r¼ k2C

orgRXC

catQY

3

rd

� �DQX

@CcatQX

@r¼ �k2Corg

RXCcatQY ð181Þ

The distribution coefficients of QY and QX, mQY and mQX, are defined as

mQY ¼C

catðsÞQY

CaqðsÞQY

and mQX ¼C

catðsÞQX

CaqðsÞQX

ð182Þ

At r ¼ rc; DQY

@CcatQY

@r¼ KQY Caq

QY �1

mQY

CcatQY

� �

DQX

@CcatQX

@r¼ �KQXðCcat

QX �mQYCaqQXÞ ð183Þ

where KQY and KQX denote the mass transfer coefficients of QY and QX, respectively. Inthe aqueous phase, the rates of change of MY, QY, and QX are

dCaqMY

dt¼ k1C

aqMYC

aqQX ð184Þ

dCaqQY

dt¼ k1C

aqMYC

aqQX � KQY

3

rc

� �Vcat

Vaq

� �Caq

QY �1

mQY

CcatðsÞQY

� �ð185Þ

dCaqQX

dt¼ KQX

3

rc

� �Vcat

Vaq

� �C

catðsÞQX �mQXC

aqQX

� �� k1C

aqMYC

aqQX ð186Þ


The initial conditions are

At t ¼ 0; CaqMY ¼ CMY;0

CaqQX ¼ CQX;0; Caq

QY ¼ 0ð187Þ

The average concentration and the conservation of PT catalyst Q at any reaction time are

�CCorgRX ¼

3

r3d

ðrd0

r2CorgRXdr ð188Þ

VaqCQX;0 ¼ Vcat�CCcatQY þ �CCcat

QX

� �þ Vaq CaqQY þ Caq

QX

� �ð189Þ

Equations (175)–(189) constitute the system of PT-catalyzed reactions with the third liquidphase, and can be further tendered in dimensionless form and solved by finite differenceand Runge–Kutta methods.

Recently, Krueger et al. [235] developed a theoretical model, based on the dispersedorganic phase, for modeling the mass transfer and interfacial reactions of the brominationof benzyl chloride in three-liquid PTC. The reaction occurring at the interface between theinner organic droplet and outer shell (or layer) of the third phase is

r ¼ R: BzClþQBrk1�!BzBrþQCl ð190Þ

They assumed that the ion-exchange reaction occurs at the interface between the aqueousand the third liquid phase according to

r ¼ Rþ �: Br� þQþ Ð QBr ð191ÞQClÐ Qþ þ Cl� ð192Þ

The governing equations in the dimensionless form for the system are

@CBC

@�¼ 1

�2@

@��2@CBC

@�

� �ð0 < � < 1Þ ð193Þ

@CQB

@�¼ DT

QBr

D0BzCl

1

�2@

@��2@CQB

@�

� �ð1 < � < �=RÞ ð194Þ

The initial and boundary conditions become

� ¼ 0: CBC ¼ 1; CQB ¼ 1 ð195Þ

� ¼ 0:@CBC

@�¼ 0 ð196Þ

� ¼ 1:@CBC

@�¼ �NDCBCCQB

@CQB

@�¼ ND

D0BzCl

DTQBr

C0BzCl;i

KCBCCQB ð197Þ

� ¼ 1þ �=R: CQB ¼ 1 ð198Þwhere ND ¼ k1KR=D

oBzCl represents a Damkohler number. The value Do

BzCl=DTQBr denotes

the ratio of the QBr and BzCl diffusivities, and the dimensionless thickness of the thirdliquid phase and the initial concentration are �=R and K=Co

BzCl;i ¼ CTQBr;i ¼ CT

BzCl;i, respec-


tively. When the amount of the third liquid phase has little effect on the overall reactionrate, the contribution of greater amounts of third liquid phase is to increase the masstransfer resistance. This case is similar to the situation of a very thin film of the third phasewith �=R� 1. They also proposed an equation [235] for the concentration of benzylchloride using the analogous analytical solution of heat transfer in terms of a mass transferproblem as

CBC ¼X1n¼1

An exp � 2n�� sinð n Þ

n�ð199Þ

where the eigenvalues n and the coefficients An are given by

1� n cotð nÞ ¼ ND and An 4 sinð nÞ � n cosð nÞ½ �

2 n � sinð2 nÞð200Þ

VI. CONCLUSION

In recent years, researches in PTC have achieved great progress in the development of thebasic theory and applications; the number of publications and patents in PTC has grownsteadily during the past decade. Halpern [236] estimated that the 5-year average of TBABpatents issued per year was 38 in 1985, 46 in 1990, and 57 in 1995. The PTC publicationsfor a 5-year average issued per year are 389 in 1985, 467 in 1990, and 484 in 1995. Thisinformation indicates the potential applications of PTC in industries either revamping ordeveloping new processes, including the applications of PTC in biology.

The aspects of PTC publications were mostly concentrated in chemistry-based inves-tigations in past years. On viewing the nature of PTC, the presence of immiscible phasesand the transport of reacting species between the phases are the basic phenomena; there-fore, the mass transfer resistances at the interface and within the intraphase should betaken into account in accompanying the chemical reactions for the purpose of reactordesign. The engineering analysis in various types of PTC, together with other techniquesfor enhancing the overall reactivity, has the advantage of realizing the factors influencingthe observed reaction rate, which makes the process design closer to the inherent results. Itis hoped that the review in this chapter could serve to generate more applications of PTCin the future.


SYMBOLS

a activity of quaternary salt in a solutionA interfacial area between the organic and aqueous phase ðm2)As surface area of solid particles ðdm2)Aliquat 336 tricaprylmethylammonium chlorideC concentration of quaternary saltCj

i concentration of component i in phase j (j ¼ org:, aq., cat.)�CCPhOQ dimensionless concentration of PhOQC# number of carbon atoms in each of the four alkyl chains in the quaternary

cationDQ distribution coefficient of quaternary cation definedETQX true extraction constant ðkmol=m3Þ�1

j hydration number per quaternary salt in the organic phasekapp apparent first-order reaction-rate constant (s�1)kapp;0 initial apparent reaction rate constant (1/min)k2 forward reaction rate constant of the aqueous phase (kmol=m3 minÞ�1k�2 reverse reaction rate constant of the aqueous phase ðkmol=m3 minÞ�1ki reaction rate constantKi equilibrium constantKda dissociation constant of QX in the aqueous phase (kmol=m3)Kdo dissociation constant of QX in the organic phase (kmol=m3)KQY mass transfer coefficient of QY (m/min)KQX overall mass transfer coefficient of QX (kmol=m3 min�1)m distribution coefficient of onium saltMH molecular weight of hydrophilic group in quaternary salt (e.g., NþX�)MT molecular weight of quaternary saltMTBAB molecular weight of TBABMX side product in the aqueous solutionMY aqueous reactantQþ quaternary cationQX quaternary saltr spatial co-ordinate in radial directionrc mean radius of droplet containing organic phase and catalyst phase (m)rd mean radius of organic droplet (m)Rf reaction rate per unit area (mol=m2 sÞRH reactantRX organic reactantR 0X organic reactantt timeT absolute temperature (K)TBAB tetra-n-butylammonium bromideTBAC tetra-n-butylammonium chlorideTBAI tetra-n-butylammonium iodideTBA-TBPO tetra-n-butylammonium 2,4,6-tribromophenoxideTBPB tetra-n-butylphosphonium bromideV volume of aqueous phase (m3)Vaq volume of aqueous phase (L)Vcat volume of catalyst phase (L)


Vorg volume of organic phase (L)X� anionX conversion of RX in the organic phaseY product yield[ ] molar concentration of species in brackets (kmol=m3)

Greek� volume ratio (organic phase/aqueous phase)� Hildebrand parameters�G Gibbs free energy�� mean activity coefficient n eigenvalues� dimensionless spatial co-ordinate in radial direction� dimensionless time

Subscripts0, i initial valuei compound iapp apparent valueobs observed value

Superscriptsaq aqueous phasecat catalyst phaseI interface between catalyst and organic phasesj j phase (j ¼ org:, aq., cat).org organic phaseS droplet surface between catalyst and aqueous phasesðoverbarÞ species in the organic phase


REFERENCES

1. CM Starks. J Am Chem Soc 93:195–199, 1971.

2. CM Starks, CL Liotta, ME Halpern. Phase-Transfer Catalysis: Fundamental Application and

Industrial Perspectives. 1st ed. New York: Chapman & Hall, 1994, pp 1–22.

3. CM Starks. Modern Perspectives on the Mechanisms of Phase-Transfer Catalysis. ACS

Symposium Series 659. Washington, DC: American Chemical Society, 1997, pp 10–28.

4. F Orsini, F Pelizzoni, B Bellini, G Miglierini. Carbohydr Res 301:95–109, 1997.

5. D Carriere, SJ Meunier, FD Tropper, S Cao, R Roy. J Mol Catal A: Chem 154:9–22, 2000.

6. D Albanese, D Landini, M Penso. Chem Commun 2095–2096, 1999.

7. YN Belokon, KA Kochetkov, TD Churkina, NS Ikonnikov, AA Chesnokov, OV Larionov,

VS Parmar, R Kumar, HB Kagan. Tetrahedron: Asym 9:851–857, 1998.



10. B Lygo, J Crosby, TR Lowdon, PG Wainwright. Tetrahedron 57:2391–2402, 2001.

11. S Kotha, E Brahmachary. J Org Chem 65:1359–1365, 2000.

12. MH Entezari, A Keshavarzi. Ultrasoni Sonochem 8:213–216, 2001.

13. H Noguchi, H Tsutsumi, M Komiyama. Chem Commun 2455–2456, 2000.

14. VK Krishnakumar, MM Sharma. Ind Eng Chem Proc Des Dev 23:410–413, 1984.

15. P Tundo, A Perosa, M Selva, SS Zinovyev. Appl Catal B: Environ 32: L1–L7, 2001.

16. P Hodget, R O’Dell, MSK Lee. Polymer 37:1267–1271, 1996.

17. LH Tagle, FR Diaz, G Cerda, M Oyarzo, G Penafiel. Polym Bull 40:35–42, 1998.

18. LH Tagle, FR Diaz, C Cares, A Brito. Polym Bull 42:627–634, 1999.

19. D Albanese, D Landini, A Maia, M Penso. J Mol Catal A: Chem 150:113–131, 1999.

20. FS Sirovski. Org Proc Res Dev 3:437–441, 1999.

21. GD Yadav, SS Naik. Org Proc Res Dev 4:141–146, 2000.

22. S Desikan, LK Doraiswamy. Chem Eng Sci 55:6119–6127, 2000.

23. JP Jayachandran, ML Wang. Appl Catal A: Gen 198:127–137, 2000.

24. JJ Hwang, RL Lin, RL Shieh, JJ Jwo. J Mol Catal A: Chem 142:125–139, 1999.

25. JAB Satrio, LK Doraiswamy. Chem Eng J 82:43–56, 2001.

26. JC Jarrouse. CR Hebd Seances Acad Sci Ser C 1424–1434, 1951.

27. EV Dehmlow, SS Dehmlow. Phase Transfer Catalysis. Weinheim: Verlag Chemie, 1993.

28. CM Starks, CL Liotta, ME Halpern. Phase Transfer Catalysis, Fundamental, Application

and Industrial Perspectives. 1st ed. New York: Chapman & Hall, 1994, pp 1–17.

29. WP Weber, GW Gokel. Phase Transfer Catalysis: Organic Synthesis. New York: Springer-

Verlag, 1977.

30. ME Halpern. Phase-Transfer Catalysis: Mechanisms and Syntheses. ACS Symposium Series

659. Washington, DC: American Chemical Society, 1997.

31. G. Goldberg. Phase Transfer Catalysis: Selected Problems and Applications. Netherlands:

Gordon & Breach, 1992.

32. N Ohtani, CA Wilkio, A Nigam, SL Regen. Macromolecules 14:516–520, 1981.

33. CM Starks. J Am Chem Soc 93: 195–199, 1971.

34. M Ellwood, J Griffiths, P Gregory. J Chem Soc, Chem Commun 181, 1980.

35. H Iwamoto, T Sonoda, H Kobayash. Tetrahedron Lett 4703–4706, 1983.

36. LJ Mathias, RA Vaidya. J Am Chem Soc 108:1093, 1986.

37. ML Wang. In: Y Sasson, R Neumann, eds. Handbook of Phase Transfer Catalysis. New

York: Chapman & Hall, 1997, pp 36–109.

38. CS Kuo, JJ Jwo. J Org Chem 57:1991–1995, 1992.

39. ML Wang, CC Ou, JJ Jwo. Bull Chem Soc Jpn 67:2949, 1994.

40. ML Wang, CC Ou, JJ Jwo. Ind Eng Chem Res 33:2034, 1994.

41. S Asai, H Nakamura, W Okada, M Yamada. Chem Eng Sci 50:943–949, 1995.

42. J Dockx. Synthesis 441–456, 1973.

43. A Gorgoes, A Le Coq. Tetrahedron Lett 4521, 1976.


44. Halpern, HA Zahalka, Y Sasson, M Rabinovitz. J Org Chem 50:5088, 1985.

45. M Makosza. Pure Appl Chem 43:439–462, 1975.

46. M Makosza, E Bialecka. Tetrahedron Lett 2:183–186, 1977.

47. D Landini, A Maia, F Montanari. J Chem Soc, Chem Comm 112–113, 1977.

48. A Brandstrom. Adv Phys Org Chem 15:267–330, 1977.

49. M Rabinovitz, Y Cohen, M Halpern. Angew Chem, Int Ed Engl 25:960, 1980.

50. KJ Evan, HJ Palmer. AIChE Symp Ser 77:104–113, 1981.

51. JB Melville, JD Goddard. Chem Eng Sci 40:2207–2215, 1986.

52. JB Melville, YC Yortsos, Chem Eng Sci 41:2873–2882, 1986.

53. CT Chen, C Hwang, MY Yen. J Chem Eng Jpn 24:284–290, 1991.

54. ML Wang, SW Chang. Ind Eng Chem Res 33:1606–1611, 1994.

55. ML Wang, SW Chang. Can J Chem Eng 69:340–346, 1991

56. ML Wang, SW Chang. Ind Eng Chem Res 30:2378–2383, 1991

57. ML Wang, HM Yang. Chem Eng Sci 46:619–627, 1991.

58. ML Wang, HS Wu. Chem Eng Sci 46:509–517, 1991.

59. A Bhattacharya. Ind Eng Chem Res 35:645–652, 1996.

60. HS Wu. J Chin Inst Chem Eng 25:183–189, 1994.

61. HS Wu, SH Jou. J Chem Tech Biotechnol 64:325–330, 1995.

62. HS Wu, JJ Lai. J Chin Inst Chem Eng 26:277–283, 1995.

63. HS Wu. Chem Eng Sci 51:827–830, 1996.

64. HS Wu. Ind Eng Chem. Res 32:1323–1327, 1993.

65. ML Wang, HM Yang. Ind Eng Chem Res 30:631–635, 1991.

66. ML Wang and HM Yang. Ind Eng Chem Res 29:522–526, 1990.

67. CM Starks, CL Liotta. Phase Transfer Catalysis: Principles and Techniques. New York:

Academic Press, 1978, p 40.

68. HS Wu. J Chin Inst Chem Eng 27:185–193, 1996.

69. HS Wu, SS Meng. AIChE J 43:1309–1318, 1997.


71. HM Yang. Chem Eng Commun 179:117–132, 2000.

72. HM Yang. Ind Eng Chem Res 37:398–404, 1998.

73. WK Fife, Y Xin. J Am Chem Soc 109:1278, 1987.

74. RS Juang, SC Liu. Ind Eng Chem Res 36:5296–5301, 1997.

75. RS Juang, SC Liu. Ind Eng Chem Res 37:4625–4630, 1998.

76. BR Agarwal, RN Diamond. J Phys Chem 67:2785–2792, 1963.

77. S Asai, H Nakamura, Y Furuichi. J Chem Eng Jpn 24:653–658, 1991.

78. M Cerna, V Bizek, J Stastova, V Rov. Chem Eng Sci 48:99–103, 1993.

79. EV Dehmlow, B Vehre. J Chem Res Symp 10:350–351, 1987.

80. EV Dehmlow, B Vehre. New J Chem 13:117–119, 1989.

81. NA Gibson, DC Weatherburn. Anal Chim Acta 58:149–157, 1972.

82. NA Gibson, DC Weatherburn. Anal Chim Acta 58:159–165, 1972.

83. HMNH Irvine, A D Damodaran. Anal Chim Acta 53:267–275, 1971.

84. HMNH Irvine, A D Damodaran. Anal Chim Acta 53:277–285, 1971.

85. HS Wu, RR Fang, SS Feng, KH Hu. J Chin Int Chem Eng 29:99–108, 1998.

86. HS Wu, RR Fang, SS Feng, KH Hu. J Mol Catal A: Chem 136:135–146, 1998.

87. A Brandstrom. Pure Appl Chem 54:1769–1782, 1982.

88. MH Abraham. J Phys Org Chem 299–308, 1971.

89. MH Abraham, AF Danil de Namor. J Chem Soc, Faraday I 72:955–962, 1976.

90. AA Ansari, MR Islam. Can J Chem 66:1720–1727, 1988.

91. P Beronius, A Brandstrom. Acta Chem Scand A 30:687–691, 1976

92. MS Tseng. Thermodynamic Properties of Quaternary Salts in an Organic Solvent and

Alkaline Solution. MS thesis, Yuan-Ze University, Taoyuan, Taiwan, 2000.

93. JO Bockries, KN Reddy. Modern Electrochemistry. Plenum Press, 1980, pp 261–265.

94. T Kenjo, RM Diamond. J Phys Chem 76:2454–2459, 1972.


95. T. Kenjo, RM Diamond. J Inorg Nucl Chem 36:183–188, 1974.

96. D Landini, A Maia, G Podda. J Org Chem 47:2264–2268, 1982.

97. D Landini, A Maia, F Montanari. J Am Chem Soc 100:2796–2801, 1978.


99. CM Starks, RM Owens. J Am Chem Soc 95:3613–3617, 1973.


101. K Gustavii. Acta Pharm Suec 4:233, 1967.

102. K Fukunaga, H Shirai, S Ide, M Kimura. Nippon Kagaku Kaishi 1148–1153, 1980.

103. FS Sirovski. Phase-Transfer and Micellar Catalysis in Two-Phase System. ACS Symposium

Series 659. Washington, DC: American Chemical Society, 1997.

104. A Leo, C Hansch, D Elkins. Chem Rev 71:525–563, 1971.

105. WC Griffin. J Soc Cosmet Chem 1:311, 1949.

106. WC Griffin. Am Perfum Essent Oil Rev 65:26, 1952.

107. JT Davies. Proceedings of the 2nd International Congress on Surface Activity, 1957, vol. 1,

p 426.

108. S Nagata. Mixing Principles and Applications. New York: Halsted Press, 1975.

109. AJ Parker, U Mayer, R Schmid, V Gutmann. J Org Chem 43:1843–1854, 1978.

110. J Czapkiewicz, T Czapkiewicz, D Struck. Pol J Chem 52, 2203, 1978.

111. RW Taft, MH Abraham, RM Doherty, MJ Kamlet. J Am Chem Soc 107:3105–3110, 1985.

112. TB Lin, MY Yeh, YP Shih. Proc PAC Chem Eng Congr 3:178, 1983.

113. JB Melville, JD Goddard. Ind Eng Chem Res 27:551, 1988.

114. ML Wang, YM Hsieh. J Chem Eng Jpn 26:374–381, 1993.

115. SS Leie, RR Bhave, MM Sharma. Chem Eng Sci 38:765–773, 1983.


117. SD Naik, LK Doraiswamy. AIChE J 612–645, 1998.

118. T Stanley, J Quinn. Chem Eng Sci 42:2313–2324, 1987.

119. S Matson. Advances in Catalytic Technologies Seminars, Catalystica. Mountain View, CA,

USA, 1988.

120. JH Rushton, S Nagata, TB Rooney. AIChE J 10:298–302, 1964.

121. S Asai, J Hatanaka, Y Uekawa. J Chem Eng Jpn 16:463–469, 1983.

122. S Asai, H Nakamura, Y Furuichi. AIChE J 38:397–403, 1992.

123. S Asai, H Nakamura, T Sumita. AIChE J 40:2028, 1994.

124. HM Yang, CK Yu, TM Chen. J Chin Inst Chem Eng 29:161–169, 1998.

125. A Kiani, RR Bahave, KK Shirkar. J Membr Sci 20:125–145, 1984.

126. R Prasad, A Kiani, RR Bhave, KK Sirkar. J Membr Sci 26:79–97, 1986.

127. YH Lin. Mass–Transfer Behavior of Quaternary Salts in Membrane Extractor, MS thesis,

Yuan-Ze University, Taoyuan, Taiwan, 2000.

128. SV Save, SS Zanwar, VG Pangarkar. Chem Eng Sci 44:1591, 1989.

129. N Wakao, S Kaguei. Heat and Mass Transfer in Packed Beds. New York: Gordon & Breach,

1982, p 156.

130. WJ McManamey, DKS Multani, JT Davies. Chem Eng Sci 30:1536, 1975.

131. EL Cussler. Diffusion Mass Transfer in Fluid System. 2nd ed. New York: Cambridge

University Press, 1997, pp 48, 355.

132. CM Starks. Modern Perspective on the Mechanisms of Phase-Transfer Catalysis. ACS

Symposium Series 659. Washington, DC: American Chemical Society, 1997.

133. F Moberg, O Bokman, H Bokman, OG Siegbahn. J Am Chem Soc 113:3663–3667, 1991.

134. B Reuben, K Sjoberg. Chemtech 315–320, 1981.

135. LL Travlarides, M Stamatoudis. In: TB Drew, GR Cokelet, JW Hooper Jr, T Vermeulen, eds.

Advances in Chemical Engineering, vol. 11. New York: Academic Press, 1981, pp 119–273.

136. NN Dutta, GS Patil, Can J Chem Eng 71:802–804, 1993.

137. SL Regen. J Am Chem Soc 97:5956–5958, 1975.

138. SL Regen. J Am Chem Soc 98:6270–6274, 1976.

139. SL Regen. J Org Chem 42:875–879, 1977.


140. SL Regen. Angew Chem, Int Ed Engl 18:421–429, 1979.

141. SL Regen, JJ Besse. J Am Chem Soc 101:4059–4063, 1979.



144. M Tomoi, E Ogawa, Y Hosokawa, H Kakjuchi. J Polym Sci, Polym Chem Ed 20:3421, 1982.

145. M Tomoi, E Nakamura, Y Hosokawa, H Kakjuchi. J Polym Sci, Polym Chem Ed 23:49–61,

1985.

146. M Tomoi, Y Hosokawa, H Kakjuchi. Makromol Chem Rapid Commun 4:227–230, 1983.

147. B Ragaini, G Verzella, A Ghigone, G Colombo. In Eng Chem Proc Des Dev 25:878–885,

1986.

148. V Ragaini, G Colombo, P. Barzhagi. Ind Eng Chem Res 27:1382–1387, 1988.

149. V Ragaini, G Colombo, P Barzhagi, E Chiellini, S D’Antone. Ind Eng Chem Res 29:924–928,

1990.

150. P Schlunt, PC Chau. J Catal 102:348–356, 1986.

151. S Desikan, LK Doraiswany. Ind Eng Chem Res 34:3524–3537, 1995.

152. WT Ford, M Tomoi. Adv Polym Sci 55:49–104, 1984.

153. WT Ford, J Lee, M Tomoi. Macromolecules 15:1246–1251, 1982.

154. O Arrad, Y Sasson. J Org Chem 55:2252, 1990.

155. P Tundo, P Venturello. J Am Chem Soc 103:856–861, 1981.

156. P Tundo, P Venturello, E Angeletti. J Am Chem Soc 104:6551–6555, 1982.

157. P Tundo, P Venturello. J Am Chem Soc 101:6606–6613, 1979.

158. S Telford, P Schlunt, PC Chau. Macromolecules 19:2435–2439, 1986.

159. JF Hradil, C Konak, K Jurek. React Polym 9:81–89, 1988.

160. F Svec. Pure Appl Chem 60:377–386, 1988.

161. ML Wang, HS Wu. Ind Eng Chem Res 31:490–496, 1992.

162. ML Wang, HS Wu. J Polym Sci, Polym Chem 30:1393–1399, 1992.


164. HS Wu, JF Tang. J Mol Cata, A: Chem 145:95–105, 1999.

165. HS Wu, SS Meng S. Can J Chem Eng 77:1146–1153, 1999.

166. HS Wu, CS Lee. J Catal 199:217–223, 2001.

167. E Ruckenstein, L Hong. J Catal 136:378, 1992.

168. E Ruckenstein, JS Park. J Poly Sci C, Polym Lett 26:529, 1988.

169. JF Tang. Application and Kinetics of Liquid–Solid–Liquid Triphase Reaction Ether–Ester

Compound. MS thesis, Yuan-Ze University, Taoyan, Taiwan, 1998.

170. DH Freeman. In: JA Marinsky, ed. Ion Exchange. vol. I. New York: Marcel Dekker, 1966,

ch. 5.

171. JP Idoux, R Wysocki, S Young, J Turcot, C Ohlman, R Leonard. Synth Commun 13:139,

1983.

172. Y Okahata, K Ariga, T Seki. J Chem Soc, Chem Commun 920, 1985.

173. Y Shan, R Kang, W Li. Ind Eng Chem Res 28:1289, 1989.



176. ML Wang, HM Yang. J Chin Inst Chem Eng 23:200–208, 1992.

177. R Aris. The Mathematical Theory of Diffusion and Reaction in Permeable Catalysis. vol. I.

London: Oxford University Press, 1975, pp 119–126.

178. PF Marconi, WT Ford. J Catal 83:160–167, 1983.

179. DY Cha. React Polym 5:269–279, 1987.

180. HS Wu, SS Meng. Chem Eng Sci 53:4073–4084, 1998.


182. M Tomoi, Y Hosokawa, H Kakjuchi. J Polym Sci, Polym Chem Ed 22:1243–1250, 1984.

183. CM Starks, CL Liotta, ME Halpern. Phase-Transfer Catalysis: Fundamental Application and

Industrial Perspectives. 1st ed. New York: Chapman & Hall, 1994, pp 108–111.


184. Y Sasson. In: Y Sasson, R Neumann, ed. Handbook of Phase Transfer Catalysis. 1st ed.

London: Blackie Academic & Professional, 1997, pp 111–134.

185. A Esikova. In: Y Sasson, R Neumann, ed. Handbook of Phase Transfer Catalysis. 1st ed.

London: Blackie Academic & Professional, 1997, pp 1–35.

186. CL Liotta, J Berkner, J Wright, B Fair. Mechanisms and Applications of Solid–Liquid Phase-

Transfer Catalysis. ACS Symposium Series 659. Washington, DC: American Chemical

Society, 1997, pp 29–40.

187. A Lopez, MM Manas, R Pleixats, A Roglans, J Ezquerra, C Pedregal. Tetrahedron 52:8365–

8386, 1996.

188. A Lopez, R Pleixats. Tetrahedron: Asym 9:1967–1977, 1998.

189. FS Sirovski, ER Berlin, SA Mulyashov, EA Bobrova, ZI Batrakova, DF Dankovskaya. Org

Proc Res Dev 1:253–256, 1997.

190. T Perrad, JC Plaquevent, JR Desmurs, D Hebrault. Org Lett 2:2959–2962, 2000.

191. HM Yang, TM Chen. J Chin Inst Chem Eng 29:367–374, 1998.

192. S Cohen, A Zoran, Y Sasson. Tetrahedron Lett 39:9815–9818, 1998.

193. R Guilet, J Berlan, O Louisnard, J Schwartzentruber. Ultrason Sonochem 5:21–25, 1998.

194. A Knochel, G Rudolph. Tetrahedron Lett 3739–3740, 1974.

195. AW Herriot, D Picker. J Am Chem Soc 97:2345–2349, 1975.

196. BP Czech, MJ Pugia, RA Bartsch. Tetrahedron 41:5439–5444, 1985.

197. KH Wong, APW Wai. J Chem Soc, Perkin Trans II 317–321, 1983.

198. MC Vander Zwan, FW Hartner. J Org Chem 43:2655–2657, 1978.

199. GD Yadav, MM Sharma. Ind Eng Chem Proc Des Dev 20:385–390, 1981.

200. HA Yee, HJ Palmer, SH Chen. Chem Eng Progr 83:33–39, 1987.

201. HM Yang, HE Wu. Ind Eng Chem Res 37:4536–4541, 1998.

202. HM Yang, CM Wu. J Mol Catal A: Chem 153:83–91, 1999.

203. JB Melville, JD Goddard. Chem Eng Sci 40:2207–2215, 1985.

204. JB Melville, JD Goddard. Ind Eng Chem Res 27:551–555, 1988.

205. JB Melville, YC Yortsos. Chem Eng Sci 41:2873–2882, 1986.

206. SD Naik, LK Doraiswamy. Chem Eng Sci 52:4533–4546, 1997.

207. CS Chu. Kinetics for Synthesizing Benzyl Phenyl Ether via Solid–Liquid Phase–Transfer

Catalysis over Polyethylene Glycol and Quaternary Ammonium Salts. MS thesis, National

Chung Hsing University, Taichung, Taiwan, 2001.

208. SS Yufit, GV Kryshtal, IA Esikova. Adsorption on Interfaces and Solvation in Phase-

Transfer Catalysis. ACS Symposium Series 659. Washington, DC: American Chemical

Society, 1997, pp 52–67.

209. F Sirovski, C Reichardt, M Gorokhova, S Ruban, E Stoikova. Tetrahedron 55:6363–6374,

1999.

210. BV Erofeev. Dokl Akad Nauk SSSR 52:515–518, 1946.

211. AY Rozovskii. Heterogeneous Chemical Reactions: Kinetics and Macrokinetics. Moscow:

Nauka, 1980, pp. 180–300.

212. SS Yuift, SS Zinovyev. Tetrahedron 55:6319–6328, 1999.

213. SS Yufit, IA Esikova. J Phys Org Chem 4:336–340, 1991.

214. IA Esikova, SS Yufit. J Phys Org Chem 4:149–157, 1991.

215. IA Esikova, SS Yufit. J Phys Org Chem 4:341–345, 1991.

216. HM Yang, PI Wu. Appl Catal A: Gen 209:17–26, 2001.

217. PI Wu. Kinetics for Solid–Liquid and Solid–Liquid–Liquid Phase-Transfer Catalysis: I.

Etherification of Halo-ester II. Esterification of Aliphatic Dicarboxylate. MS thesis,

National Chung Hsing University, Taichung, Taiwan, 2000.

218. HM Yang, CM Wu, HE Wu. J Chem Technol Biotechnol 75:387–393, 2000.

219. HM Yang, PI Wu, CM Li. Appl Catal A: Gen 193:129–137, 2000.

220. C Lepetit, M Che. J Mol Catal A: Chem 100:147–160, 1995.

221. R Neumann, Y Sasson. J Org Chem 49:3448–3451, 1984.



223. SD Naik, LK Doraiswamy. AIChE J 44:612–646, 1998.

224. T Ido, M Saiki, S Goto. Kagaku Kogaku Ronbunshu 15:403–410, 1989.



227. T Ido, T Yamamoto, G Jin, S Goto. Chem Eng Sci 52:3511–3520, 1997.

228. GD Yadav, CA Reddy. Ind Eng Chem Res 38:2245–2253, 1999.

229. G Jin, T. Ido, S. Goto. J Chem Eng Jpn 32:417–423, 1999.

230. HC Hsiao, HS Weng. Ind Eng Chem Res 38:2911–2918, 1999.

231. HC Hsiao, SM Kao, HS Weng. Ind Eng Chem Res 39:2772–2778, 2000.

232. HS Weng, CM Wang, DH Wang. Ind Eng Chem Res 36:3613–3618, 1997.

233. T Ido, T Susaki, G Jin, S Goto. Appl Catal A:Gen 201:139–143, 2000.

234. HM Yang. J Chin Inst Eng 21:399–408, 1998.

235. JJ Krueger, MD Amiridis, HJ Ploehn. Ind Eng Chem Res 40:3158–3163, 2001.

236. ME Halpern. Recent Trends in Industrial and Academic Phase-Transfer Catalysis. ACS

Symposium Series 659. Washington, DC: American Chemical Society, 1997, pp 1–9.


12Enzymes in Reverse Micelles(Microemulsions): Theory and Practice

ANDREY V. LEVASHOV and NATALIA L. KLYACHKOMoscow State University, Moscow, Russia

I. INTRODUCTION

In ternary systems, various ‘‘surfactant – water – organic solvent’’ type surfactant aggre-gates called micelles are spontaneously formed. A ternary diagram of one of the widelyused systems Aerosol OT (AOT)–water–octane, is presented in Fig. 1. The initial descrip-tion of this diagram was first given by Tamamushi and Watanabe for iso-octane, asorganic solvent, and pure water [1]; it was corrected by us for n-octane and phosphate–acetate buffer solution [2]. Depending on the nature and the components’ ratio in thesystem, different types of micelles may be formed, varying by shape, by size, and byassembling of surfactant molecules (see the examples in Fig. 1). As seen, in diluted aqueoussolutions a surfactant forms normal spherical micelles (L1), the core of each of which isconstituted of apolar chains (tails), while the outer shell contacting with water consists ofpolar (and often charged) heads. In apolar organic solvents a surfactant also forms sphe-rical—but inverted (or reverse)—micelles; here, the core consists of polar (charged) headswhile nonpolar tails are oriented outside in the solvent. Normal micelles are capable ofsolubilizing apolar compounds while reverse micelles solubilize polar compounds, primar-ily water. Hydration of reverse micelles is accompanied by an increase in their sizes andformation of an independent aqueous phase inside the micelle. Therefore, hydrated reversemicelles are often—appropriately—called microemulsions. However, in this chapter wewill use the term ‘‘micelle’’ (originating from the Latin word ‘‘mica’’ meaning smallparticle), since in our opinion it is broader by sense and significance (unlike ‘‘microemul-sion’’ which is not applicable for, e.g., ‘‘dry’’ systems).

Apart from water, reverse micelles can solubilize other polar compounds, of bothlow-molecular and high-molecular mass, including proteins (enzymes). Enzymesentrapped in micelles reveal catalytic activity. Figure 2 represents a so-called enzogramreflecting catalytic activities of laccase in different phases of the ternary system ‘‘AerosolOT – water – octane’’ shown in Fig. 1. As seen from Fig. 2, a level of enzyme catalyticactivity depends both on the structural type of surfactant aggregates and on the micellesize (its inner polar cavity). At the same time, as seen from Fig. 2, the level of catalyticactivity of laccase achieved in aqueous solutions used in classical enzymology turns out tobe much lower than that in micellar systems. In other words, for enzyme catalysis, micellarmedia appear to be much more variable by regulatory parameters and more favorable forrevealing catalytic activity than water. In recent times, abundant data have been accumu-


FIG. 1 Phase diagram for Aerosol OT (AOT)–water–octane system. The boundaries of each

individual phase were determined with 50 mM phosphate+50 mM acetate buffer as an aqueous

component (—). (From Ref. 2.) L1, L2: normal and reverse micelles of surfactant, respectively; D, F:

liquid crystalline mesophases with lamellar and reverse hexagonal packing of surfactant molecules,

respectively. Concentrations of all components are expressed as %(w/w). Cross-section of � type

shows an example of the variation of water content at constant surfactant-to-organic solvent ratio;

cross-section of � type shows an example of the variation of organic solvent content at constant

water-to-surfactant molar ratio.

FIG. 2 Laccase catalytic activity in different phases (reverse micellar—L2, lamellar—D, and reverse

hexagonal—F) of the ternary diagram for the AOT–water–octane system presented in Fig. 1. (From

Ref. 2.) The ‘‘a’’ and ‘‘b’’ catalytic profiles were measured in the ternary mixtures corresponding to

some cross-sections of � and � type, respectively (Fig. 1). The dashed line shows the catalytic activity

of laccase in aqueous solution.


lated concerning enzyme catalysis in micellar (microemulsion) systems (see reviews [3–16]).In the present chapter we will try to concentrate on the most characteristic and importantaspects of obtaining enzyme-containing systems, to point out their structural peculiaritiesand regulation of catalytic parameters, and to give examples of their practical applica-tions.

II. METHODOLOGY FOR OBTAINING AND MANIPULATING PROTEIN-CONTAINING MICELLAR SYSTEMS

A. Methods of Protein (Enzyme) Incorporation Into Reverse Micelles

Protein (enzyme) incorporation into reverse micelles of surfactant in organic solvents maybe achieved by one of the following methods.

The first method was proposed by our group [17]; it is often referred to as an‘‘injection method’’ and is now most widely used. A small amount (of the order of a fewvolume per cent) of protein aqueous solution is introduced into the surfactant solutionin organic solvent (‘‘dry’’ or slightly hydrated). A ratio of water and organic solutions isdefined by the conditions of the experiment, primarily, by the value of the desired degreeof surfactant hydration (water-to-surfactant molar ratio, w0 ¼ ½H2O]/[surfactant]). Thus,the obtained mixture is shaken vigorously (for seconds or tens of seconds) until anoptically transparent solution is formed. This method is really simple and effective.Among the disadvantages of using this method there can be mentioned a question ofequilibrium in the system obtained (see, e.g., [18]) as well as processes and reactionsproceeding during the system preparation. The question of equilibrium is solved throughcomparison with the other ways of reaching this state (as the equilibrium state does notdepend on the way of achieving it). Some undesired side reactions (leading, for instance,to partial inactivation of the solubilized enzyme) may be excluded by choosing theoptimal order of adding and mixing of the components. For example, micellar solutionsof an enzyme and a substrate are prepared separately with the same degree of hydration,incubated (to equilibrate) for some time, and the reaction is initiated by mixing of theincubated solutions in the desired proportion. (It should be specially emphasized that inthis case at the final stage – enzymic reaction initiation – there is no change in thehydration degree of the surfactant, often a critical factor in revelation of nonequilibriumand side complicating effects.)

In the second method, proposed by Menger and Yamada [19], a desired amount ofwater (aqueous buffer solution) was first introduced into the solution of surfactant in anorganic solvent in order to achieve the desired value for the degree of hydration w0; afterthat, a dry (for instance, lyophilized) protein preparation was dissolved in the micellarsolution obtained and energetically shaken (mixed). The time needed for solubilization ofthe dry protein is normally much longer than in the case of solubilization of aqueoussolutions—it varies from a few minutes to a few hours. With this procedure a protein staysin contact with the surfactant and the organic solvent during a relatively long time and as aresult its denaturation often occurs. However, in the end (at the expense of the loss ofprotein), micellar solutions with much higher (often saturating) protein concentrationsmay be obtained compared with those in the first (injection) method.

In order to reduce the loss of protein it is better to carry out the procedure of proteinsolubilization in a few stages. First, a slight excess of a dry protein is used. After a fewhours of mixing of the suspension, thus obtained, the supernatant is withdrawn, a newportion of the protein is added to it, and the procedure of dissolution is repeated (a few


times, thereby controlling the contents of the protein dissolved in the supernatant). Itshould be kept in mind when using this method that a partial loss of surfactant andwater (which are going to precipitate) may occur in the process of system preparation.This requires an additional control, such as a chromatographic procedure [20]. In princi-ple, it is not difficult to define a surfactant concentration in micellar solution due to itshigh content. It may be done via gravimetric measurements, by estimating the weight ofthe dry residue after evaporation of the aliquot of the solution under analysis. A con-venient method of controlling water in micellar systems is by proton magnetic resonance,by defining the value of the chemical shift [21].

Concerning water control, there should also be mentioned the question of estimationof the hydration level of initial surfactant preparations (which should be taken intoaccount in order to calculate correctly the degree of hydration). For this purpose, weused infrared spectroscopy to control the water content at a frequency of 3420 cm�1

(the method is described in detail in Ref. 22).It should be noted that, generally, in the case of using the second method of solu-

bilization—solution of the dry protein preparation in micellar solution—the contents ofthe comixtures, both proteins and low-molecular mass compounds (salts), could be con-trolled in the final solution.

The third method, first described in the work of Hanahan [23], was intensivelyemployed and developed in the work of the group of Luisi [24,25] (see also the reviewin Ref. 5).

The principle of this method lies in the spontaneous transfer (distribution) of theprotein in a biphase system consisting of nearly equal volumes of aqueous protein solutionand surfactant containing organic solvent (actually, a micellar system with a fixed degreeof hydration). Protein transfer occurs without stirring or with slight stirring and lasts arelatively long time (from tens of minutes to one day). During such a procedure a protein(enzyme) interacts actively with molecules of surfactant and organic solvent in the aqueoussolution, concentrates on the interface, and undergoes denaturation. Because of thesecomplications this method is clearly disadvantageous compared to those described earlier.Yet, this method has a separate independent importance and is doubtlessly prospective forprotein separation and purification (see, e.g., [26]) since such parameters as pH and ionicstrength may regulate the processes of protein incorporation into the micelles and itswithdrawal back into the water solution. It is, therefore, possible to choose the optimalconditions for the selective entrapment of the protein by the micelles and thus to extractthe required protein from the mixture. After replacing the water phase by pure solutionwith the required pH and ionic strength values, re-extraction is carried out to obtain asolution of the target purified protein. This approach allows one to extract not onlyproteins (enzymes) from micellar systems, but also other water-soluble components,including reaction products. This will be discussed in the next section.

B. Separation of Components of Micellar Systems, EnzymeRegeneration, and Isolation of Reaction Products

Since micellar systems are microheterogeneous, isolation of this or that component fromthese systems is a quite difficult task. There is apparently no general solution for this task,every separate case requires an individual consideration. At the same time there exist alarge number of approaches to tackle this problem. We will try to consider theseapproaches consecutively.


The simplest and at the same time elegant trick is based on the reversible phaseseparation on physical action. Thus, for instance, a relatively slight temperature shift(increase in some cases and decrease in others) as described in Ref. 27], leads to separationof the phase that is rich in organic solvent and contains hydrophobic product and almostno surfactant. This method seems most prospective and technological; it is appropriate fororganic synthesis, when the product can be obtained in pure form by evaporation (dis-tillation) of the organic solvent while the initial system is easily restored to the workingcondition by adding pure organic solvent (stripper).

A general technique for separation of low-molecular mass components from micellarsystems without their destruction deals with semipermeable membranes. This principlemay lay the basis for construction of continuous reactors.

It is possible to extract water-soluble components from the micelles via conversion ofthe micellar system into a macrobiphasic system (corresponding to the third solubilizationmethod) by mixing with water or water–salt solutions.

It is easy to destroy micellar systems irreversibly by adding water-miscible organicsolvents such as, e.g., acetone and ethanol. This technique is exceptionally effective andconvenient when it is important to extract the solubilized protein (enzyme) delicately sothat it retains its structure and physiological (including catalytic) activity [28]. Here, itshould be emphasized that, first, the ethanol and especially acetone should be cold.Second, during precipitation with acetone in neutral and basic media, formation ofSchiff bases is possible; besides which, side reactions involving modification of the proteinsubjected to isolation (including the loss of the protein solubility in water) are also prob-able.

The optimal choice of organic solvent and the volume ratio of organic solvent andmicellar system allows phase separation and obtaining of solvents enriched in this or thatcomponent, such as enzymic reaction products [29].

Reversible transitions in ternary systems (surfactant–water–organic solvent), forinstance, from a micellar to a liquid-crystalline state (including the macroheterogeneousstate, when surfactant and enzyme are located in the condensed phase, while the reactionproduct is situated in the organic solvent), may also be considered as a general separationtechnique. This approach (see, e.g., [30]) seems extremely prospective for the purposes oforganic synthesis.

In conclusion, it should be mentioned that the described approaches may also beused in combinations, e.g., a temperature (pressure) action can be combined with addingof water (aqueous buffer solutions), salts, or organic solvents.

C. Limits of Protein (Enzyme) Solubilization in Systems of ReverseMicelles

Limiting amounts (concentrations) of proteins obtained in systems of reverse micellesdepend on many factors, including the nature of the protein (purity and quality of pre-paration), the nature of the surfactant, and water content. An example of �-chymotrypsinand regulation of its maximal concentration in reverse micelles by modification of theprotein surface is presented in Fig. 3. Noticeable protein solubilization occurs only inhydrated micelles; saturating protein concentration increases with increase in the degreeof surfactant hydration and then begins to decrease. The reason for such a decrease in theamount of solubilized protein at high degrees of hydration is probably connected with thereduction in the micelles’ concentration (their size is increasing while the amount ofmicelles is getting lower). As a matter of fact, most often simple proteins are entrapped


into the reverse micelles in a stoichiometric ratio of 1:1. Higher stoichiometry is achievedat extremely high concentrations of simple proteins and also in the case of oligomericproteins (see, e.g., [3,6,31]).

D. Structure of Protein-Containing Micelles

Hydration of empty micelles results in an increase of their size. In the case of micellescomposed of AOT in iso-octane their radius linearly depends on the degree of hydration(water-to-AOT molar ratio), w0 ¼ ½H2O�=½AOT� : rM ¼ 4þ 1:5w0 (in A). The size of theprotein may be larger or smaller than the size of the inner cavity of the micelle. A sche-matic representation of different examples of protein-containing micelles is shown in Fig.4. In the case of smaller micelles the protein creates a new micelle out of a few micelles sothat in a protein-containing micelle the protein molecule becomes surrounded by a sur-factant monolayer (induced fit model) [Fig. 4(a)] [32,33]. The size of the micelle thenconsists of the protein size plus the length of a surfactant molecule. In the case of largemicelles, hydrophilic proteins incorporate into the inner cavity without changing theexternal sizes of hydrated micelles (fixed size model) [Fig. 4(c)] [11,32–34]. This model(for large micelles) also works in the case when the radius of the inner cavity of the micelleis equal to the protein sizes [Fig. 4(b)] [11,32,33]. In this case, water from the inner cavityof the formed (due to protein entrapping) micelles shifts towards carbohydrate-containingsurfactant chains [34,35]. In the case of membrane-sensitive enzymes (actively interactingwith the micellar matrix), formation of protein-containing micelles proceeds predomi-nantly by the ‘‘induced fit’’ mechanism, when the protein itself forms an optimal micelle(micellar shell) [Fig. 4(d)].

FIG. 3 �-Chymotrypsin solubility (protein concentration) in reverse micelles of 0.1 M AOT in

octane, as a dependence on the degree of surfactant hydration (water-to-AOT molar ratio), for

protein molecules with chemically modified surface groups. (*) Acetyl-�-chymotrypsin; (*)

succinyl-�-chymotrypsin; (- - -) maximal concentration of nonmodified native �-chymotrypsin

solubilized in reverse micelles.


III. KINETIC CHARACTERISTICS OF ENZYMIC CATALYSIS IN SYSTEMSOF REVERSE MICELLES AND THEIR REGULATION

Micellar systems are microheterogeneous; hence, according to the distribution laws, in theequilibrium state local concentrations of the reagents are different in different phases.Enzymes are generally localized in the water–micellar phase (where their concentrationsare higher than the total enzyme concentration and depend on the volume ratio of thewhole system and the micellar phase). Hydrophobic substances concentrate in the bulkorganic solvent; therefore, in the vicinity of the enzyme their concentrations are lower. In

FIG. 4 Schematic representation of the formation of protein-containing micelles: (a) An initial

micelle is smaller than the protein molecule to be entrapped. In this case, a protein molecule

creates its own micelle with a size equal to that of the protein. (b) An initial micelle is equal in

size to the protein molecule. In this case, protein solubilization occurs without noticeable changes in

the dimensions and shape of the initial micelle. (c) An initial micelle is larger than the protein

molecule. In this case, no changes occur. (d) An initial micelle is larger (the same with smaller

and equal size micelles) than the protein molecule, but the latter is capable of interacting with the

surfactant matrix due to an anchoring group on its surface (membrane-sensitive protein). In this

case, the protein molecule creates its own micelle with a size equal to that of the protein.


other words, in order to define the reaction parameters the dimensions of which includesconcentration, it is necessary to consider distribution of the reagents and to define theirlocal concentrations. The kinetics of enzymic reactions, according to the Michaelis equa-tion, employ the values kcat and Km. The latter one is usually expressed in units of con-centration; therefore, an analysis of the observed Km values is rather problematic. Theobserved Km values depend on the distribution coefficient of the substrates and on thevolume ratio of the phases (therefore, Km may be regulated by the volume ratio). Differenttheoretical and experimental approaches considering two-, three-, or multi-phasic modelswere used for the interpretation of results for reverse micelles (see, e.g., kinetic analysis inRefs 3 and 36–39).

The catalytic constant kcat is quite another matter: it is a first-order rate constant thedimension ([time]�1) of which does not contain concentration. Therefore, this parameter inreverse micelles is not complicated by the distribution effects of the substances and may beregarded as an objective parameter, reflecting a true reactivity of the enzyme solubilized inthe system of reverse micelles.

A. Dependence of Catalytic Constant k cat on Degree of Hydration

1. Simple Enzymes

The most striking effects in micellar enzymology are seen in dependencies of enzymecatalytic activity, kcat, on surfactant degree of hydration. These dependencies are normallybell shaped (see examples in Fig. 5); the higher the molecular weight of the enzyme the

FIG. 5 Regulation of relative maximal reaction rates (V=Vopt) for different enzymes solubilized in

AOT reverse micelles by variation of the degree of surfactant hydration (water-to-AOT molar ratio).

The inset shows the correlation between the radius of an entrapped enzyme (rp) and corresponding

optimal aqueous micellar cavity (rm). (For details see Ref. 40.)


higher w0 at which the optimum is observed (detailed analysis of the phenomenon observedis given in Ref. 40). It deals with the fact that an optimal micelle is the micelle complemen-tary to the protein molecule. The principle of geometric correspondence of a protein and amicelle is well traced in Fig. 5 (inset), where a correlation between the radius of entrappedenzyme (rp) and corresponding optimal aqueous micellar cavity (rm) is demonstrated.

A reason for the high catalytic activity in the optimal micelle is, in our opinion, thestabilizing effect of the micellar matrix on the protein conformation. A complementary tothe enzyme micellar matrix fixes (‘‘freezes’’) the protein structure (damping unfavorablevibrations and protecting the protein from unfolding).

These assumptions are confirmed by experiments with the spin-labeled active centerof �-chymotrypsin [41] (Fig. 6). As seen from the data in Fig. 6, the enzyme entrapped inreverse micelles is located in the medium with decreased polarity. In similar polar media ofwater–organic mixtures, enzyme structures are normally disrupted, and protein denatura-tion and loss of catalytic enzymic activity occur. Yet, in systems of reverse micelles aprincipally different picture is observed. In optimal enzyme activity conditions the proteinbecomes tightly fixed by the micellar matrix and its conformational mobility is minimized.

It is possible to make a micellar matrix more effective and more rigid by the replace-ment of water by water–organic solvent mixtures [42] (Fig. 7). As seen from Fig. 7,replacement of water with glycerol in the inner cavity of reverse micelles leads to a shiftin dependency profiles, kcat (w0), to the area of lower values of hydration degree.

According to the principle of geometric correspondence an optimum is observedwhen the sizes of a protein molecule and the inner cavity of a micelle are equal. At thesame time, increasing the glycerol portion (and decreasing the water content) results in afurther decrease in protein structure mobility (‘‘freezing’’ of the structure). Figure 7 showsthat an increase in enzyme catalytic activity accompanies such structure ‘‘freezing.’’ As alimit, the enzyme catalytic activity in a ‘‘frozen’’ state is defined by the high reactivity ofthe constituent nucleophil of the active site of �-chymotrypsin [43].

It should be noticed that in the extreme case of replacing water by water-miscibleorganic solvent we deal with a practically ‘‘dry’’ system with a highly active enzyme. It

FIG. 6 Dependence of the first-order rate constant (kcat) for deacylation of N-trans-cinnamoyl-�-chymotrypsin (*), as well as rotational frequency, � (*), and hyperfine splitting constant, a (~) for

spin-labeled �-chymotrypsin on degree of hydration (water-to-AOT molar ratio) in the system AOT

(0.1 M)–water–octane. Spin label: 2,2,5,5-tetramethyl-4-iodoacetamidopyrrolidine-1-oxyl. Dashed

lines show kcat and a values in aqueous solution. (From Ref. 41.)


means that for revealing catalytic activity an enzyme does not need free, ‘‘solvent’’ waterbut just a trace amount of it, as a component of a protein structure.

Presented in Fig. 8 are the data on the determination of the minimal amount of waterrequired for �-chymotrypsin catalysis. The system CTAB–dimethyl sulfoxide/water–octane/chloroform was used as a reaction medium, and the volume ratio of water todimethyl sulfoxide was varied from 0 to 0.001 [12,44]. It can be seen that in the totallydry system the reaction does not take place. Introduction of water activates the enzyme. Itsfull activation (defined by titration) occurs in the presence of just a few molecules of water(around 5) per enzyme molecule. The plateau in Fig. 8 is explained by the fact that theenzyme used is a hydrolase; in low-water conditions acylation of the enzyme by thesubstrate occurs with quantitative formation of the acyl-enzyme. (Incidentally, it is a directdemonstration of acyl-enzyme formation in the reaction of chymotrypsin with an anilidesubstrate.) Hydrolysis of acyl-enzyme by water occurs at higher water concentrations,when, apparently, water appears as a reagent in a free state.

2. Complex and Oligomeric Enzymes

In the case of complex and oligomeric enzymes, a dependence of kcat on w0 in the system ofreverse micelles usually has few optima; a wave-like profile is generally observed [45–61](see examples in Fig. 9). A reason for such a characteristic of the dependence is the ability

FIG. 7 Dependence on the water/surfactant molar ratio of (a) the maximal reaction rate

normalized to the enzyme concentration, V=E0, of �-chymotrypsin-catalyzed hydrolysis of N-

benzoyl-L-tyrosine p-nitroanilide, and (b) the rotational frequency, �, of the spin label in the

active site of the enzyme in the system AOT–water/glycerol–octane. Water/glycerol volume ratios:

1—100:0; 2—80:20, 3—50:50, 4—20:80, 5—6:94. Dashed lines show V=E0 and � values in aqueous

solution. (From Ref. 42.)


of micelles to cause dissociation of a protein with a tertiary structure. Each optimumobserved functionally corresponds to one or another oligomeric form of the enzyme. Inthe example of glyceraldehyde-3-phosphate dehydrogenase shown in Fig. 9(a), threeoptima are observed, corresponding to the functioning of monomeric, dimeric, and tetra-meric forms of the enzyme. In the case of lactate dehydrogenase, a similar behavior isobserved, Fig. 9(b). Mixing of these enzymes results in formation of a heteroenzymecomplex [60]; this is clearly traced in the dependency of kcat (w0) in Fig. 9(c).

B. Dependency of Catalytic Activity of Solubilized Enzymes onSurfactant Concentration

Enzymes may be divided into two groups. Catalytic activity of the first group does notdepend on the surfactant concentration, while that of the second group does (see Fig. 10).

The second group normally consists of membrane enzymes that contain anchorgroups providing their interaction with membranes. The ‘‘micelle dependency’’ of thesecond group of enzymes arises from their membranotropic properties (membrane sensi-tivity): the enzyme interacting with micellar media creates its own micelle, i.e., micelleformation occurs by the ‘‘induced fit’’ model (see Fig. 4). In this case, two differentpopulations of micelles varying by sizes are found in the system: protein-containing andempty. Both populations are in dynamic equilibrium and interact via collisions. As a resultof such interactions new particles are formed which are not optimal by structure, and theactivity of enzymes located in such micelles decreases. Therefore, the activity of enzymes ofthe second group may be increased by dilution of micellar systems: decreasing the con-

FIG. 8 Determination of minimal amounts of water required for �-chymotryptic catalysis from the

dependence of the concentration of p-nitroaniline (P1), formed during �-chymotrypsin cleavage of

N-benzoyl-L-tyrosine p-nitroanilide in the system CTAB–dimethyl sulfoxide/water–octane/

chloroform, on the concentration of water added to the system, [H2O]add. The volume ratio of

water to dimethyl sulfoxide was varied from 0 to 0.001. (From Ref. 44.)


centration of micelles results in reduction of frequency of their collisions. At the limit, atinfinite dilution, we obtain an isolated optimal micelle in which an enzyme possesses thehighest catalytic activity (right column in Fig. 10).

In the case of nonmembrane enzymes of the first group it is possible to increase theircatalytic activity in micelles by choosing the right degree of hydration and by increasing

FIG. 9 Dependence of the catalytic activity (maximal rate) of (a) glyceraldehyde-3-phosphate

dehydrogenase, GAPDH; (b) lactate dehydrogenase, LDH; and (c) their heteroenzyme complexes,

GAPDH+LDH, on the degree of surfactant (water-to-AOT molar ratio) in reverse micelles of AOT

in octane. (From Ref. 60.)


the rigidity of the micellar matrix (as shown in examples in Fig. 6). In the case of enzymesof the second group all independent changes do not sum up but come to a single limitcharacterizing the realized potential catalytic efficiency of the enzyme created by thisenzyme optimal micelle (see Figs 10 and 11). As seen in Fig. 11, the activity of �-chymo-trypsin, a nonmembrane enzyme (of the first group), is controlled by a micellar matrix theproperties of which are defined ‘‘from the outside’’ without participation of the enzymeitself (replacement of water by a water/glycerol mixture inside reverse micelles enhancesthe level of �-chymotrypsin catalytic activity). In the case of peroxidase, a second-groupenzyme, the matrix is formed by the enzyme itself and a single limiting value of enzymicactivity is observed.

FIG. 10 Regulation of the catalytic activity of solubilized enzymes by variation of the surfactant

concentration at a constant degree of hydration in the systems: (*) AOT–water–octane; (~)

dodecylammonium propionate–water–diethyl ether/benzene; (*) Brij 96–water–cyclohexane; (&)

lecithin–water/methanol/pentanol–octane. Dashed lines show levels of corresponding catalytic

activities in aqueous solution. (From Ref. 10.)


IV. APPLIED ASPECTS OF MICELLAR ENZYMOLOGY

Micellar enzymology which employs microheterogeneous ‘‘surfactant–organic solvent–water’’ systems as reaction media prove to be exceptionally technological and prospectivein application. Therefore, it seems necessary once more to draw attention to the propertiesof reverse micelles.

First, these systems are diphilic, since both hydrophilic and hydrophobic (water-insoluble) substances may be dissolved (solubilized) therein. Second, surfactant aggre-gates, which are organized in such systems, are particles and may serve as matrix reactorsof molecular size by entrapping enzyme molecules and required reagents. This allows oneto carry out controlled local reactions in limited volume and to organize catalytic ensem-bles of controlled sizes (nanoparticles).

In the literature the possibilities of practical uses of enzymes entrapped in reversemicellar systems were repeatedly discussed (see, e.g., reviews [7,8] and also topical compi-lation [62–64]). We will further consider some examples from the most prospective areas:fine chemistry, chemical analysis, and supramolecular design.

FIG. 11 Combination of two factors regulating enzyme catalytic activity: variation in the surfactant

concentration and addition of water-miscible organic solvents. (a) Peroxidase in the system AOT–

water/glycerol–octane at water/glycerol volume ratios: (*) 100:0; (*) 20:80. (b) �-Chymotrypsin in

the system AOT–water/glycerol–octane at water/glycerol volume ratios: (*) 100:0; (*) 6:94. Dashed

lines show the catalytic activities of the enzymes in aqueous solution. (From Ref. 44.)


There are three major reasons why enzymically catalyzed micellar systems seemprospective for the purposes of fine chemistry:

1. First, they allow transformations of water-insoluble substances in pseudoho-mogeneous media (without diffusive restriction) to be carried out.

2. Second, they allow the equilibrium state of catalyzed reactions to be regulated,by, for instance, lowering water content;

3. Third, enzyme incorporation into micelles allows technological stabilizationeffects (both in the pure state and in combination with other approaches) tobe achieved.

In our opinion, the best example of transformation of water-insoluble substances inmicellar systems is the work of a Dutch group [65] on the reduction of steroid hormones.

The authors [65] employed a triple-enzyme system to carry out the reaction ofstereospecific reduction of progesterone by molecular hydrogen.

Their type of micellar system belongs to the group of biphasic water/organic systems.These systems are simple and convenient in regard to regulation of equilibrium of thereactions executed therein [66,67]. The presence of an additional micellar matrix (the thirdmicellar phase) in a biphasic system allows one to obtain extra effects in the equilibriumstate, often very significant [4,68].

We will give examples of stabilization techniques for organic synthesis in SectionIV.A, in particular for describing nanogranulated enzyme forms Section IV.A.1.

Enzyme-containing micellar systems may be successfully employed in analysis for:

. Enhancement of sensitivity of analyzed water-soluble substances

. Detection (analysis) of water-insoluble substances

. Creation of principal analytical systems

1. A good example of the possibility of sensitivity enhancement is a bioluminescentassay by firefly luciferase. The transfer of the process from water into the colloid system‘‘Brij 96–water–cyclohexane’’ leads to increased catalytic activity of the enzyme and sen-sitivity of the corresponding analysis by two decimal orders [69,70] (see Fig. 12).

2. The possibilities of analysis of water-insoluble biologically active compounds arewell illustrated by test systems described for water-insoluble riboflavin derivatives [71].

3. ‘‘Homogeneous immunoassay’’ may serve as an example of a principal analy-tical approach realized by employing micellar systems [72,73] (Fig. 13). The enzyme(peroxidase) modified by hapten (thyroxine) has the same dependency profile of kcat(w0) as the native enzyme (see Fig. 13). Formation of an immunocomplex (larger bysize) in compliance with the principle of geometric correspondence results in a shift ofthe profile kcat (w0) towards larger values of w0. At fixed w0 values, immunocomplexformation can be denoted by a decrease in the observed reaction rate. Destruction of theimmunocomplex, e.g., by introduction of free hapten, is detected by an increase in theobserved enzymic activity.

A. Modification of Proteins (Enzymes) and Their Supramolecular Designin Reverse Micelles (Nanoreactors)

Reverse micelles are an excellent and even unique tool for various chemical manipulationswith proteins (enzymes).


FIG. 12 Bioluminescence analysis using firefly luciferase-catalyzed reactions: (a) kinetic curves of

light emission (bioluminescence intensity) in aqueous solution and in Brij 96–water–cyclohexane

(membrane-like system); (b) calibration curves for ATP determination in aqueous solution (*)

and in Brij 96–water–cyclohexane (*). (From Ref. 70.)


1. Protein Hydrophobization

Micellar systems are exceptionally convenient for the introduction of hydrophobic andlow-water-soluble reagents into protein molecules in controlled quantities (first of all, ofone or two hydrophobic residues per protein molecule) [74]. This method is successfullyemployed for the introduction of long-chain fatty acid residues [74–76], phospholipids [77],

FIG. 13 Homogeneous immunoassay in reverse micelles of AOT in octane. Top—schematic

representation of immunocomplex formation in reverse micellar system. Middle—the shift in

catalytic activity – hydration degree profile due to an immunocomplex much larger than the

initial components formed. Bottom—shows, as an example, the dependence of the catalytic

activity of peroxidase–thyroxine conjugate on the concentration of antibodies against thyroxine

(IgG) in a system of reverse micelles of AOT in octane at fixed degree of hydration (maximum

for peroxidase–conjugate activity). The dashed line shows the level of conjugate activity in the system

in the absence of antibodies. Vertical arrows indicate changes in the conjugate catalytic activity on

addition of free thyroxine to the system. (From Ref. 73.)


hormones (thyroid and steroid) [73,78], and metallo-organic compounds (such as ferro-cene) into protein molecules [79].

2. Construction of Protein Ensembles of Assigned Stoichiometry

A protein molecule is entrapped in the inner cavity of a reverse micelle the size of whichmay vary with variation in the degree of surfactant hydration. Normally, simple proteins(enzymes) form protein-containing micelles in a stoichiometric ratio of 1:1. It is, in somecases, possible to entrap a larger amount of protein in one micelle when using extremelyhigh (>10 mM) protein concentrations [31].

Another case is oligomeric proteins (enzymes). Variation of micelle sizes in this caseallows a whole set of different oligomeric forms to be obtained: monomers, dimers, tetra-mers, etc., for many individual enzymes [45–59] and enzyme mixtures [60]. By employing amicellar matrix it is possible to construct nonconventional (from the point of view ofclassical ‘‘aqueous’’ enzymology) protein–protein complexes, such as a compact noncova-lent chymotrypsin dimer or a stable (dissociating in water only in the presence of 8 M urea)noncovalent complex of chymotrypsin with peroxidase [61].

3. Obtaining Protein Complexes with Polymers

In homogeneous solution it is rather problematic to link two protein molecules, consider-ing the vast number of intermolecular reactions leading finally to formation of a cross-linked polymer. In micelles we can restrict a sphere of reaction by just a space of onemicelle and to suppress (or to exclude when necessary) other reactions at the level ofintermicellar interactions. In other words, employing classical linking reagents in micellarsystems allows one to link (to fix chemically) complexes in an ‘‘intramicellar’’ mode. As anexample of the realization of this idea we can cite the studies [80,81] on formation ofcomplexes, protein conjugates, and synthetic polymers of different stoichiometries. At lowdegrees of hydration, when micelle sizes are small, after reaching a certain critical degree ofsurfactant hydration (this value depends on the size of the complex) a complex is formedwhich may be linked chemically with a practically quantitative yield. Further increase inmicelle sizes (degree of surfactant hydration) may lead to formation of protein–polymercomplexes of higher stoichiometry [80,81].

4. Nanogranulated (Nanocapsulated) Proteins (Enzymes)

A polymer shell surrounding the protein molecule may be formed not only from existingpolymer but also by polymerization of corresponding monomers in the reverse micellemedium. The first polymerization of monomers solubilized by reverse micelles was carriedout by Speiser [82–84] for the purposes of nanocapsulating and nanogranulating of bio-logically active compounds. We introduced proteins (enzymes) in macromeric forms,obtained by previous chemical modification with acryloyl chloride, into copolymerizationreactions in reverse micelles system [85–87]. As a result of copolymerization of monomersand macromonomers solubilized in reverse micelle systems, polymer particles are formed(nanogranules) which may be easily isolated by system destruction, e.g., by using preci-pitation with acetone. The particles thus obtained are close to monodisperse and their sizesare of the order of tens of nanometers (they are really nanoparticles); they contain acovalently entrapped (immobilized) and highly stable enzyme. These particles may bedissolved in water or suspended in organic solvent as biocatalyzers. In order to carryout biocatalytic reactions in organic solvents the solubility of such nanoparticles in


organic medium may be improved by incorporating hydrophobic fragments into theparticles, by, for instance, copolymerization with surfactants containing double bondsin their hydrophilic part [87,88]. Nanogranules may be also obtained in reverse micellesystems from existing polymers such as gelatin [89] or poly(vinyl alcohol) (by formation ofcryogel) [90]. Organogels based on lecithin were studied by the group of Luisi [91–93]. Anexample of a block copolymer with the structural elements of the reverse-micelle type isgiven in the work of an Indian group [94].

V. CONCLUSION

In conclusion we would like to draw the reader’s attention to the possibilities of regulationand practical applications of enzymes entrapped in micellar systems.

REFERENCES

1. N Tamamushi, B Watanabe. Colloid Polym Sci 258:174–178, 1980.

2. AV Pshezhetsky, NL Klyachko, AV Levashov, K Martinek. Biocatalysis 4:185–198, 1990.

3. AV Levashov, YuL Khmelnitsky, NL Klyachko, K Martinek. In: K Mittal, B Lindman, eds.

Surfactants in Solution. vol. 2. New York: Plenum Press, 1982, pp 1069–1091.

4. K Martinek, AV Levashov, NL Klyachko, YuL Khmelnitsky, IV Berezin. Eur J Biochem

155:453–468, 1986.

5. PL Luisi, LJ Magid. CRC Crit Rev Biochem 20:409–474, 1986.

6. AV Levashov. Itogi Nauki I Techniki VINITI AN SSSR (Russ) 4:112–158, 1987.

7. K Martinek, IV Berezin, YuL Khmelnitsky, NL Klyachko, AV Levashov. Collect Czech

Chem Commun 52:2589–2602, 1987.

8. KMartinek, IV Berezin, YuL Khmelnitsky, NL Klyachko, AV Levashov. Biocatalysis 1:9–15,

1987.

9. PL Luisi, M Giomini, MP Pileni, BH Robinson. Biochim Biophys Acta 947:209–246, 1988.

10. K Martinek, NL Klyachko, AV Kabanov, YuL Khmelnitsky, AV Levashov. Biochim

Biophys Acta 981:161–172, 1989.

11. YuL Khmelnitsky, AV Kabanov, NL Klyachko, AV Levashov, K Martinek. In: MP Pileni,

ed. Structure and Reactivity in Reverse Micelles. Amsterdam: Elsevier, 1989, pp 230–261.

12. NL Klyachko, AV Levashov, AV Kabanov, YuL Khmelnitsky, K Martinek. In: M Gratzel,

K Kalyanasundaram, eds. New York: Marcel Dekker, 1991, pp 135–181.

13. C Oldfield. Biotechnol Genet Eng Rev 12:255–327, 1994.

14. AV Levashov, NL Klyachko. Ann NY Acad Sci 750:80–84, 1995.

15. C Nicot, M Waks. Biotechnol Genet Eng Rev 13:268–314, 1995.

16. M Tuena De Gomez-Puyou, A Gomez-Puyou. Crit Rev Biochem Mol Biol 33:53–89, 1998.

17. K Martinek, AV Levashov, NL Klyachko, IV Berezin. Dokl AN SSSR (Russ) 236:920–923,

1977; Proc Acad Sci USSR (Doklady Phys Chem) 236:951–953, 1978 (Engl Transl).

18. AV Kabanov, SN Nametkin, EG Matveeva, NL Klyachko, K Martinek, AV Levashov. Mol

Biol (Russ) 22:473–483, 1988.

19. FM Menger, K Yamada. J Am Chem Soc 191:6731–6734, 1979.

20. KM Larsson, A Janssen, P Albercreutz, B Matiasson. Biocatalysis 4:163–175, 1990.

21. AV Levashov, NL Klyachko, K Martinek, VM Polyakov, GB Sergeev. Vestnik MSU (Russ),

28 (Ser 2, Chem):287–290, 1987.

22. AV Levashov, VI Pantin, K Martinek. Colloid Zh (Russ) 41:453–460, 1979.

23. DJ Hanahan. J Biol Chem 195:199–206, 1952.

24. PL Luisi, F Henningen, M Joppich. Biochem Biophys Res Commun 74:1384–1389, 1977.


25. ME Leser, G Wei, PL Luisi, M Maestro. Biochem Biophys Res Commun 135:629–635, 1986.

26. KE Goklen. Liquid–Liquid Extraction of Biopolymers: Selective Solubilization of Proteins in

Reverse Micelles. PhD dissertation, Massachusetts Institute of Technology, Cambridge, MA,

1986.

27. KM Larsson, P Albercreuts, B Matiasson. Biotechnol Bioeng 36:135–141, 1990.

28. AV Levashov, AV Kabanov, YuL Khmelnitsky, IV Berezin, K Martinek. Dokl AN SSSR

(Russ) 278:246–248, 1984.

29. NG Bogdanova, NL Klyachko, VE Kabakov, K Martinek, AV Levashov. Bioorg Khim

(Russ) 15:634–637, 1989.

30. P Miete, R Gruber, H Voss. Biotechnol Lett 11:449–454, 1989.

31. NL Klyachko. Kinetic Peculiarities of Enzymes Solubilized by Reverse Micelles. PhD disser-

tation, Moscow State University, Russia, 1983.

32. AV Levashov, YuL Khmelnitsky, NL Klyachko, VYa Chernyak, K Martinek. Anal Biochem

118:42–46, 1981.

33. AV Levashov, YuL Khmelnitsky, NL Klyachko, VYa Chernyak, K Martinek. J Colloid

Interface Sci 88:444–457, 1982.

34. YuE Shapiro, NA Budanov, AV Levashov, NL Klyachko, YuL Khmelnitsky, K Martinek.

Collect Czech Chem Commun 54:1126–1134, 1989.

35. YuE Shapiro, VYa Gorbatyuk, AV Levashov, NL Klyachko. Biol Membrany (Russ) 7:277–

290, 1994.

36. K Martinek, AV Levashov, NL Klyachko, VI Pantin, IV Berezin. Biochim Biophys Acta

657:277–294, 1981.

37. PDI Fletcher, BH Robinson, RB Freedman, Ch Oldfield. J Chem Soc, Faraday Trans I

81:2667–2679, 1985.

38. R Bru, A Sanches-Ferrer, F Garcia-Carmona. Biochem J 268:679–684, 1990.

39. LF Aguilar, E Abuin, E Lissi. Arch Biochem Biophys 388:231–236, 2001.

40. NL Klyachko, AV Pshezhetsky, AV Kabanov, SV Vakula, K Martinek, AV Levashov. Biol

Membrany (Russ) 7:467–472, 1990.

41. OV Belonogova, GI Likhtenshtein, AV Levashov, YuL Khmelnitsky, NL Klyachko, K

Martinek. Biochemistry (Russ) 48:379–386, 1983.

42. NL Klyachko, NG Bogdanova, VK Koltover, K Martinek, AV Levashov. Biochemistry

(Russ) 54:1224–1230, 1989.

43. AV Levashov, NL Klyachko, NG Bogdanova, K Martinek. FEBS Lett 268:238–240, 1990.

44. NG Bogdanova. Enzyme Catalysis in Surfactant Reverse Micellar Systems Solvated with

Water–Organic Solvent Mixtures. PhD dissertation, Moscow State University, Russia, 1989.

45. NL Klyachko, S Merker, K Martinek, AV Levashov. Dokl Akad Nauk SSSR (Russ)

298:1479–1481, 1988.

46. AV Kabanov, SN Nametkin, GN Evtushenko, NN Chernov, NL Klyachko, AV Levashov, K

Martinek. Biochim Biophys Acta 996:147–152, 1989.

47. S Merker, NL Klyachko, AV Levashov, K Martinek. Biocatalysis 4:65–66, 1990.

48. AV Kabanov, SN Nametkin, NN Chernov, NL Klyachko, AV Levashov. FEBS Lett 295:73–

76, 1991.

49. SN Nametkin, AV Kabanov, NL Klyachko, AV Levashov. Bioorg Chem (Russ) 17:606–609,

1991.

50. AV Kabanov, NL Klyachko, SN Nametkin, S Merker, AV Zaroza, VI Bunik, MV Ivanov,

AV Levashov. Protein Eng 4:1009–1017, 1991.

51. AV Pshezhetsky, AV Levashov, GYa Viderschain. Biochim Biophys Acta 1122:154–160,

1992.

52. VE Kabakov, S Merker, NL Klyachko, K Martinek, AV Levashov. FEBS Lett 311:209–212,

1992.

53. AV Pshezhetsky, EM Bayer, OA Buneeva, MV Vinogradova, AV Levashov, GYa

Viderschain. Bioorg Chem (Russ) 19:161–173, 1993.

54. OA Kost, TA Ort, II Nikolskaya, SN Nametkin, AV Levashov. Biochemistry (Russ) 59:1746–

1755, 1994.


55. GG Chang, TM Huang, SM Huang, WY Chou. Eur J Biochem 225:1021–1027, 1994.

56. OA Kost, TA Ort, II Nikolskaya, SN Nametkin, AV Levashov. Bioorg Chem (Russ) 21:403–

407, 1995.

57. NL Klyachko, AV Ugolnikova, MV Ivanov, AV Levashov. Biochemistry (Russ) 60:1048–

1054, 1995.

58. NL Klyachko, SV Vakula, VN Gladyshev, VI Tishkov, AV Levashov. Biochemistry (Russ)

62:1683–1687, 1997.

59. AM Azevedo, LP Fonseca, DMF Prazeres. Biocat Biotrans 17:401–415, 2000.

60. AV Levashov, AV Ugolnikova, MV Ivanov, NL Klyachko. Biochem Mol Biol Int 42:527–

534, 1997.

61. RV Rariy, NL Klyachko, EA Borisova, CJ Cortes-Penagos, AV Levashov. Biochem Mol Biol

Int 36:31–37, 1995.

62. PL Luisi, BE Straub, eds. Reverse Micelles. Biological and Technological Relevance of

Amphiphilic Structures in Apolar Media. New York: Plenum Press, 1984.

63. MP Pileni, ed. Structure and Reactivity in Reverse Micelles. Amsterdam: Elsevier, 1989.

64. M Gratzel, K Kalianasundaram, eds. Kinetics and Catalysis in Microheterogeneous Systems.

Surfactant Science Series, vol. 38. New York: Marcel Dekker, 1991.

65. R Hilhorst, C Laane, C Veeger. FEBS Lett 159:225–228, 1983.

66. AN Semenov, YuL Khmelnitsky, IV Berezin, K Martinek. Biocatalysis 1:3–8, 1987.

67. AN Semenov, V Cerovsky, MI Titov, K Martinek. Collect Czech Chem Commun 53:2963–

2985, 1988.

68. K Martinek, YuL Khmelnitsky, AV Levashov, NL Klyachko, AN Semenov, IV Berezin.

Dokl AN SSSR (Russ) 256:1423–1426, 1981.

69. EI Belyaeva, NN Ugarova, NL Klyachko, AV Levashov, K Martinek, IV Berezin. Dokl AN

SSSR (Russ) 273:494–497, 1983.

70. NL Klyachko, M Yu Rubtsova, AV Levashov, EM Gavrilova, AM Egorov, K Martinek, IV

Berezin. Ann NY Acad Sci 501:267–270, 1987.

71. BI Kurganov, LG Tsetlin, EA Malakhova, NA Chebotareva, VZ Lankin, GD Glebova, VM

Berezovsky, AV Levashov, K Martinek. J Biochem Biophys Methods 11:177–184, 1985.

72. AV Kabanov, MM Khrutskaya, MI Budavari, SA Eremin. Dokl AN SSSR (Russ) 305:1253–

1256, 1989.

73. AV Kabanov, MM Khrutskaya, SA Eremin, NL Klyachko, AV Levashov. Anal Biochem

181:145–148, 1989.

74. AV Levashov, AV Kabanov, IV Berezin, K Martinek. Dokl AN SSSR (Russ) 270:246–248,

1984.

75. AV Kabanov, AV Ovcharenko, NS Melik-Nubarov, AI Bannikov, VYu Alakhov, VI Kiselev,

AV Levashov, ES Severin. FEBS Lett 250:238–240, 1989.

76. AV Kabanov, AV Levashov, Vyu Alakhov. Protein Eng 3:39–42, 1989.

77. AN Lukyanov, AL Klibanov, AV Kabanov, VP Torchilin, AV Levashov, K Martinek.

Bioorg Chem (Russ) 14:670–673, 1988.

78. EA Yatsimirskaya, EM Gavrilova, AM Egorov, AV Levashov. Steroids 58:547–550, 1993.

79. AD Ryabov, AM Trushkin, LI Baksheeva, RK Gorbatova, IV Kubrakova, VV Mozhaev, BB

Gnedenko, AV Levashov. Angew Chem, Int Ed Engl 31:789–790, 1992.

80. AV Kabanov, VYu Alakhov, EYu Klinsky, MM Khrutskaya, AA Rakhnyanskaya, AS

Polinsky, AA Yaroslavov, ES Severin, VA Kabanov, AV Levashov. Dokl AN SSSR (Russ)

302:375–379, 1988.

81. AV Kabanov, AV Levashov, MM Khrutskaya, VA Kabanov. Macromol Chem 33:379–391,

1990.

82. P Speiser. Progr Colloid Polym Sci 59:48, 1976.

83. G Birrenbach, P Speiser. J Pharm Sci 65:1763–1766, 1976.

84. P Speiser. In: PL Luisi, BE Straub, eds. Reverse Micelles. New York: Plenum Press. 1984, pp

339–346.

85. EG Abakumova, AV Levashov, IV Berezin, K Martinek. Dokl AN SSSR 283:136–139, 1985.


86. EG Abakumova, AV Levashov, K Martinek. Vestnik Mosk Univ, Ser Khim (Russ) 26:509–

513, 1985.

87. EG Matveeva, AV Levashov, AA Shegolev, VV Ryzhova, VA Polyakov, K Martinek.

Biotechnology (Russ) 6:84–93, 1985.

88. YuL Khmelnitsky, IN Neverova, R Momcheva, AI Yaropolov, AB Belova, AV Levashov, K

Martinek. Biotechnol Tech 3:275–280, 1989.

89. C Haring, PL Luisi. J Phys Chem 90:5892, 1986.

90. YuE Shapiro, EG Pykhteeva, AV Levashov. J Colloid Interface Sci 206:168–175, 1998.

91. R Scartezzini, PL Luisi. J Phys Chem 92:829, 1988.

92. R Scartezzini, PL Luisi. Biocatalysis 3:377–380, 1990.

93. HL Willemann, PL Luisi. Biochem Biophys Res Commun 177:897–900, 1991.

94. A Gupte, R Nagarjan, A Kilara. Biotechnol Progr 7:348–354, 1991.


13Micellar Catalysis

VINCENT C. REINSBOROUGH Mount Allison University, Sackville, NewBrunswick, Canada

I. INTRODUCTION

In the early 1970s, the Russian school of Berezin et al. [1,2] attempted to rationalize theunique effect that surfactants and micelles in particular had on reaction rates in solution(Fig. 1). Sufficient kinetic data involving various types of chemical reactions in micellarmedia had accumulated to test a hypothesis [3,4]. This hypothesis stated that in thepresence of ionic surfactants some reaction rates were dramatically enhanced over anarrow concentration range in the vicinity of the critical micelle concentration (cmc)only to fall off more gradually with increasing surfactant concentration until seemingly itregained the original reaction rate observed when no surfactant was present [Fig. 1,curve (a)]. These lambda spikes often achieved heights corresponding to several thou-sand times the reaction rate in water. In other reactions, the observed rates rose moregently to plateaus in enzyme-like fashion [Fig. 1 curve (b)]. Undoubtedly, in moreconcentrated micellar solution, a drop off in rate would be observed as [Fig. 1, curve(a)] but, because micelles changed size and even shape in concentrated micellar solution,the upper limit of the range of investigation of catalysis effects was usually arranged tobe about five times the cmc. A further series of reactions in the presence of surfactantsdisplayed the opposite behavior beyond the cmc. For example, reaction rates decreasedseemingly exponentially to many times less than the original values [Fig 1, curve (c)].Usually, this rate of inhibition occurred, as with the enhancement effects, over a narrowsurfactant concentration range.

The interest that arose quite strongly in the 1980s and 1990s in this particularcatalysis phenomenon came from many areas. Industry was obviously tantalized by thetime, energy, and materials that might be saved [5], but what seemed to appeal most toresearchers were the biochemical implications. Nearly all chemical processes in the livingcell occur at interfaces, such as at the active site of an enzyme on a membrane. As a result,chemical reactivity in biochemical situations is critically dependent on the local microen-vironment, the local concentrations, and relative orientations of the bound reactants at cellinterfaces. Thus, the realization that micelles could be realistic cell mimicks has become amajor reason for this ongoing drive on the part of investigators in figuring out the specificsof micellar catalysis.

When micelles were present in solution, Berezin pictured bimolecular reactions asoccurring in two sites, the micellar region and in bulk solution. The overall reaction ratewas determined by the partitioning factor by which each of the reactants was assimilated


into the micelles from the bulk solution. A rate promotion occurred when both reactantswere preferentially located in the micelles. A rate decrease followed if only one of thereactants had been incorporated into the micelles. In this pseudo-phase treatment ofmicellar effects, the partitioning coefficients between bulk solution and micelles of thereactants and the product are critical data in the evaluation of the overall rate ofreaction:

AB þ BB , PB

KA m KB m KP mAM þ BM , PM

where A and B are reactants, P is the product, and the subscripts B and M refer to the bulksolution and micellar regions, respectively. KA, KB, and KP are the binding coefficients forthe reactants and products partitioning between the bulk solution and micelles. The kineticequations developed from this scheme can be used to obtain partitioning constants thatagree with the experimentally determined values from techniques such as solubilizationand gel filtration [1].

The immediate success of the Berezin model in accounting almost quantitatively forthe observed catalysis effect of micelles has an interesting implication. Is this truly a case ofcatalysis? In many instances, the micelles bring about considerable shifts in equilibriumpositions, which forced Berezin to admit that the term ‘‘micellar catalysis’’ was ‘‘somewhatincorrect’’ [2]. He justified its continued use on the basis that the surfactant is not con-sumed in the reaction and that for most surfactants the concentration required to bringabout marked effects is usually very low. Some workers in the field have opted for lesscontroversial terms such as ‘‘micellar rate enhancement’’ or ‘‘rate promotion.’’ The title ofa recent review, ‘‘Micellar Catalysis, a Useful Misnomer’’ [3], sums up the prevalentattitude of researchers today.

FIG. 1 Effect of surfactants on the rates of reaction in the region of the critical micelle

concentration (cmc). Curves (a) and (b) represent micellar catalysis while curve (c) is a typical

example of micellar inhibition.


II. DYNAMIC NATURE OF MICELLES

A major difference between micelles and many other catalysts is that micelles are dynamicand fluid reaction sites. They are continually in the process of forming and disintegratingand this must be taken into account when interpreting micellar rate enhancements. Tworelaxation times are found in micellar solutions. The faster time is usually in the micro-second range and the other in the millisecond range [6–10]. The faster time is associatedwith the exchange of surfactant monomer between the bulk solution and the micelle whilethe so-called slow relaxation pertains to the overall breakdown of the micelle. The latterprocess occurs as a result of the size fluctuation in the micelles that results from the fastexchange. When the micelle size dips below a certain critical value, the micelle is unstableand falls apart. When this happens, a replacement micelle will spontaneously form else-where in the solution to maintain the overall equilibrium. The slow relaxation is related tothe cmc because both are measures of the overall stability of the micelle. The micelle willbe more stable the lower the cmc and the longer the relaxation time. These findings meanthat reactants in relatively slow reactions occurring within micelles experience a continu-ally changing chaotic environment. Surfactant monomers slide into and out of solutionfrom the micelles, and the micelles themselves explode and reform elsewhere in the solu-tion while the reaction proceeds. Kinetic techniques that involve perturbing a system atequilibrium by means of an externally applied temperature, pressure, pH, or concentrationjump can promote micelle breakdown while leaving the faster exchange process unaffected[8]. Micelle break-up times can significantly affect the control of technological processessuch as foaming, wetting, emulsification, and oil solubilization [10]. The short lifetimes ofmicelles are in sharp contrast to those of vesicles, which are the bilayer aggregates formedby phospholipids in aqueous solution. The lifespan of vesicles can range from weeks ormonths [11]. This means that catalysis effects observed with these stable, cell-mimetic,micellar look-alikes are much more easily interpreted than the situation with the moreproblematic, continually changing micelles.

Another implication of the dynamic nature of micelles is that careful delineation ofthe micelle into regions having specific characteristics is of limited use in reaction kinetics.In aqueous solution, the headgroups are in the micelle surface and the hydrocarbon chainsoccupy mainly the micelle interior where significant wetting of the methylene groupsoccurs [8]. Each micellar region can only be vaguely defined given the prevalent generalstate of flux. Thus, the micelle is divided crudely into a polar exterior and a nonpolarinterior. Considerable leeway normally exists as to the location and positioning of solu-bilizates during the course of any interaction between the two roughly defined regions.Even practitioners of the art of micellar catalysis often fall into the trap of fruitlesslyattempting to situate the reactants carefully in specific regions of the micelle, hence,ignoring its fluid and short-lived structure. Therefore, micelles presenting moving andchanging targets to reactant molecules is not usually what is found when consideringother forms of catalysis.

III. PRE-CMC CATALYSIS

In many instances, it is found that rate enhancements in surfactant systems appear tobegin below the cmc (Fig. 1). This seemingly premature behavior is usually ascribed topremicellar aggregates of various types [12–18] although hard evidence for their existenceis too difficult to obtain. The fact that this is found with some experimental methods while


other techniques appear not to be sensitive to anything unusual in the pre-cmc region isdisconcerting. On top of this, the range of suspected candidates is wide ranging fromsurfactant dimers to substrate–surfactant complexes to oligomers of various forms.Premicelles, as the culprits, are a too facile explanation. One of the more convincingstudies is that of Brinchi et al. [19] who postulated premicelles as accelerating agents inan examination of the hydrolysis of dinitroalkoxyphenyl phosphate with cationic surfac-tants. On the other hand, Drennan et al. [20] found that, for the case of a particular metal-ion complexation occurring in anionic surfactant solutions, the ‘‘premicelles’’ were mostlikely mixed, true micelles formed early through the intervention of the metal-ion reactant.A similar explanation was given for the pre-cmc catalysis found for the chromium(VI)oxidation of dimethyl sulfoxide in picolinic acid solutions containing surfactant [21].Clearly, the pre-cmc region requires more probing to discern the mechanism by whichcatalysis is commonly effected.

IV. CATALYSIS WITH CATIONIC SURFACTANTS

Reactions between organic substrates and hydrophilic anions have received perhaps themost attention as suitable candidates to test and extend Berezin’s ideas on micellar cata-lysis because of their importance industrially. Cationic surfactants are the surfactants ofchoice for this class of reaction in order for both reactants to be induced to enter and toreside preferentially in the positively charged, hydrophobic micelle rather than in bulkaqueous solution. The solution behavior of the observed rate coefficient in these micellarsystems generally is represented by the enzymic-like curve (b) of Fig. 1. Romsted andcoworkers [3,22] pioneered the work in this field with a successful model that has come tobe known as the pseudophase ion-exchange (PIE) treatment. A similar approach hadreceived earlier airing by Menger and Portnoy [23]. The strengths and weaknesses of thePIE model have received extensive coverage [24,25] with the model being found applicablealso to some situations where anionic micelles were used as the catalysts [26].

The basic assumptions of the PIE model are as follows:

1. The micelles act as a separate phase; they are uniformly distributed throughoutthe solution and are invariant in composition throughout the micellar range.

2. The degree of counterion ionization remains constant irrespective of ion type orconcentration or of surfactant concentration.

3. The micellar surface region can be thought of as ion-exchange resin whereexchange processes can be handled mechanically in the same way as for a resin.

The emphasis placed on the last assumption is responsible for the name of the model. It isnow well known that these assumptions, especially the first two, are reliable with impunityonly over very narrow and dilute micellar concentration ranges. Nevertheless, the PIEmodel has provided invaluable insight over the past 25 years in elucidating micellarcatalysis. Its ‘‘failures’’ [27–31] are usually attributable to clear-cut violations of its simpleassumptions. Refinements or alternatives to these basic premises such as solving the non-linear Poisson–Boltzmann equation for the cell model have not proved to be particularlyenlightening nor more helpful [32]. The extension of the PIE model to complicated micel-lar systems where anomalous rate behavior is more often than not the rule rather than theexception is probably unwarranted [33]. Sudholter et al. [34] have critically reviewed theBerezin model and its Romsted variation, the PIE model, as matters stood 20 years ago. In


a recent joint publication, a worker from each camp compares the two approaches andapplies each to the particular case of amide exchange in micelles [35].

The kinetic equation at the heart of the PIE model is simply

kobs ¼ kW½AW�½BW�=½AT� þ kM½AM�½BM�=½AT�½VM� ð1Þwith the two terms respectively denoting the contributions from the reaction occurring ineach pseudophase, the bulk solution, and the micelles, where A is the organic substrateand B is the anionic nucleophile. The total concentration of A is [AT]. The bindingconstant of A to the micelle (KA) provides values of [AW] and [AM] and the independentlyobtained ion-exchange constant generates values of [BW] and [BM]. The term VM is definedas the reaction volume in the micelle and is usually assumed to be the molar volume of thesurfactant in the micelle. This term is required dimensionally because the reaction isperceived as occurring essentially in the necessarily nebulous surface region of the micelle.Its determination is problematic, but all evidence indicates that VM does not differ appre-ciably from the molar volume of the surfactant measured in water. Crucial to the modelare accurate assessments of substrate/micelle binding constants and the ion-exchangeparameters. Diverse techniques such as spectrofluorimetry [36], linear solvation freeenergy relationship analysis [37], and ion-selective electrode measurements [38] havebeen brought to bear on the determination of the requisite binding constants. The con-fidence level in the PIE model is now such that observed rate constants for many reactionscan be successfully predicted [39]. Despite its shortcomings, it is the theoretical approachmost often first invoked by workers in micellar catalysis [40–42].

In acting as catalysts, cationic micelles are usually found to invoke no change inmechanism from the reaction in water alone [43], but occasionally they provide an alter-native route to the end result. Xiang et al. [44] found that the hydrolysis of p-nitrophenylpicolinate by Cu(II) and Zn(II) tripeptide complexes involved a ternary complex intermedi-ate that was more stable in the micelle than in water. Broxton and Duddy [45] noted thatcetyltrimethylammonium bromide (CTAB) micelles induced a change in mechanism in thebasic hydrolysis of substituted N-methyl-p-toluanilides. Mixed cationic micellar systemswhere either the counterion [46] or the surfactant ion [47] was varied have also beeninvestigated. Kaneko et al. [46] discovered that the mechanism unexpectedly differedwhen bromide ion was substituted for chloride ion in the photolysis of a phenylhydrox-ylamine. Munoz et al. [48] introduced an interesting twist when they reversed the kinetics ofthe dehydrochlorination of 1,1,1-trichloro-2,2-bis(p-chlorophenyl)ethane (DDT) in tetra-decyltrimethylammonium bromide micelles in aqueous alcohol solution to discover thedissociation constants of the micelles themselves.

Tee and coworkers [49,50] have pushed the mechanistic analysis harder in seekingthe preferred reaction paths that the micelles must be providing in reactions such as thethiolysis of p-nitrophenyl alkanoates where the ‘‘concentration effect,’’ i.e., the partition-ing of both reactants into the micelle, which is at the heart of the PIE model, fails toexplain sufficiently the observed catalysis or inhibition brought about by CTAB micelles.For example, in a follow-up study the efficiency of CTAB-mediated aminolysis of p-nitrophenyl acetate increased systematically from five-fold retardation to 70-fold enhance-ment as the amine chain was lengthened to n-octyl [51]. Advantage was taken of Kirby’sdissection of the binding of the activated complex into passive and dynamic states in orderto gain insight into the binding sites of the reactants in the micelle [52]. Focusing on theinteraction of the micelles with the activated complex is being increasingly viewed as thenext step in elucidating the mechanism of micellar catalysis beyond what can reasonablybe expected of the PIE model [53]. Another direction in which improvement is required is


the proper assessment of the distribution of the reactants among the micelles, which hasbeen a vexing problem right from the beginnings of the theory [54].

V. CATALYSIS WITH ANIONIC SURFACTANTS

Another active area in micellar catalysis has been metal-ion complexations where anionicsurfactants are used to attract positive ions and hydrophobic solutes into the micellaraggregate. The rate coefficients in these cases often show curve (a) (Fig. 1) behavior inmicellar solution where the initial increase defies careful characterization with its suddenprecipitous climb occasioned principally by the strong coulombic forces between themicelles and multivalent cations. As opposed to the PIE approach, the focus instead inthe modeling has been on the gradually descending portion of the kobs curve that occurswith increasing micelle concentration. Following Berezin’s view, the basic equilibria oper-ating in this region are assumed to be those of Fig. 1 and the observed attenuation of thecatalysis effect is ascribed to the dilution of the micelle-bound reactants as the micelleconcentration is increased. The micelle concentration dictates the relative amounts ofreactant in each pseudophase through the partitioning constants. Thus, as with the PIEmodel, these binding parameters play an important role in analyzing the observed kineticbehavior even on the downslope of the curve.

One development that has proved fruitful and insightful in this area is due toRobinson and coworkers [55–58]. Inhibition as would occur when indicator anions arepresent in anionic micellar solutions is also well accommodated by this view [18,59–61].The kinetic relationship that emerges from the Robinson treatment is

kobs ¼ kW½AT�=fC VM½1þ ðC KAÞ�1�½1þ ðC KBÞ�1�g ð2Þwhere C is the concentration of the micellized surfactant, i.e., the total surfactant con-centration less the cmc, A is the metal-ion reactant, and B is the ligand [56,57,62]. It isassumed that kW ¼ kM and that the backward step in the metal-ion complexation is oflittle kinetic consequence especially when micelles are present. The binding parametersKA and KB are obtained independently as with the PIE approach through such studiesas ligand solubilization. The molar reaction volume VM is usually initially approximatedby the surfactant molar volume in water and then computer fitted to Eq. (2) to yield amore useful measure of the micelle structuring as envisioned by the reactants. It is oftenfound that best-fit KA values are more precise and just as accurate as independentlymeasured values. Insight into the structure of mixed or unusual micelles is often gainedthrough the VM and, to a lesser extent, KA estimates that emerge from this modeling[62,63]. Useful information has been gained about the micelles formed by two-tailed [64],two-headed [64], short-tailed [57], and fluorocarbon anionic surfactants [65], and in thepresence of additives such as urea [63], benzene [66], and inorganic salts [67] through thismethod. Confidence in the essential legitimacy of this modeling approach is now suchthat the use of kinetics to reveal subtleties in micelle structure is becoming more wide-spread, even with cationic surfactants [68]. Given the common source and the similarmethodology, it follows that the Robinson model has the same limitations as that for thePIE model.

A probe reaction that has seen much use in this context is the formation of thecomplex between Ni2þðaqÞ ion and the bidentate organic ligand, trans-pyridine-2-azo-p-dimethylaniline (PADA). The kinetics of this reaction is conveniently studied by thestopped-flow technique under pseudo-first-order conditions with nickel ions in excess.


Anionic micelles can provide a successful reaction site for positively charged reac-tants which otherwise do not react to any extent in aqueous solution. Dash andMohammed [69] used sodium dodecyl sulfate (SDS) micelles to enable oxalatopenta-aminecobalt(III) to undergo complexation with Ni(II) and Fe(III) metal ions. That thesituation is not always that simplistic is demonstrated by the metal-ion complexation ofPyrogallol Red in micellar dodecyltrimethylammonium bromide solution where a rateenhancement was observed despite the similar charges of the ions and the micelle [70].On the other hand, metal–ligand complex formation between Cu(II) and benzoylacetonein anionic and cationic micellar solutions occurred as expected, i.e., with rate increases anddecreases, respectively [71]. Stability constants for Cu(II) and Cd(II) ion complexes withseveral common ligands in micellar SDS are available [72].

VI. CATALYSIS WITH NONIONIC AND ZWITTERIONIC MICELLES

When neither reactant is charged, micelles formed from nonionic surfactants will not beexpected to have a pronounced catalytic effect on the reaction rate because the chargedspecies will have little incentive to be solubilized within the micelle. Thus, Harada et al.[73] found that the polyoxylethylene alkyl ether of hydrocarbon chain length 12 was only25% as effective as a catalyst as an anionic surfactant of the same chain length such asdodecyltrimethylammonium chloride in the hydroxide ion reaction with tetranitro-methane. The kinetics of complexation of some azophenol derivatives with Ni2+ andCu2+ ions was depressed in the presence of Triton X-100 micelles in keeping with theincorporation of the azophenols within and the exclusion of the metal ions from thenonionic micelles [74]. Carbone et al. [75] took advantage of this lack of attraction ofionic species to nonionic micelles to gain insight into the dominant hydrophobic attractiveforces between reactants and nonionic Triton X micelles. However, this exclusion of polarreactants is not always complete and may in fact be significant enough to account for thesometimes observed micellar inhibitions [76].

Like their nonionic counterparts, zwitterionic micelles have not received much atten-tion as catalysts although they present an interesting opportunity to explore the effect ofcharge separation in the surfactant monomer on the catalysis effect. Conceivably, thenegative and positive charges on the surfactant monomer might be sufficiently removedfrom each other in the micelle so that charged reactants could be preferentially solubilizedwithin the micelle with resulting marked rate enhancements. The above situation is notedin the Ni2+ complexation of PADA where Ni2+ ions apparently readily enter carboxy-betaine micelles, but are excluded from sulfobetaine micelles [77]. The more delocalizedsulfonate group does not afford the same degree of attraction for the Ni2+ ion as does thecarboxylate charge center. In the carboxybetaine micellar situation, the tetra-alkylammo-nium positively charged center is ineffective in preventing the small and hard Ni2+ ionfrom residing in the micelle. Ghosh et al. [78] have noted that the acid hydrolysis ofhydroxamic acids in two zwitterionic sulfobetaine surfactants was inhibited, but that alka-line hydrolysis was accelerated.

Another diagnostic kinetic application with zwitterionic micelles was that ofRodrıguez et al. [79], who investigated the reaction of DDT with hydroxide ions inmicellar sulfobetaine solutions to determine the role cations play in the process.Micellar charge plays an important role in micellar catalysis [80]. The use of nonionicmicelles affords a sensitive and refining means to measure the operative hydrophobic


forces by providing a situation where the overall coulombic forces are essentially elimi-nated [81].

VII. ELECTRON TRANSFER REACTIONS

The use of micelles as rate promoters or inhibitors in electron transfer reactions has beenrecently reviewed by Prado-Gotor et al. [82]. Possible applications lie in the fields of solarenergy conversion and storage, DNA modification, gated electron transfer reactions, andthe testing of the PIE and analogous models. Another purpose in using micelles in electrontransfer reactions is to ascertain how electrons might behave in biological systems forwhich micelles are fitting mimicks [83–85]. Micelles can also possess intense surface electricfields which would be expected to exert a powerful influence on electron transfer reactions.In fact, the surface potential of micelles is often revealed through electron transfer reac-tions as surface probes [86]. Micelles can also have a major influence on electron transferfluorescence reactions [87].

It must be noted that the dynamic nature of micelles must especially be borne inmind in dealing with electron transfers which invariably are fast processes. The seminalwork of Bruhn and Holzwarth [88], an examination of the kinetics of diffusion-controlledelectron transfer reactions in micellar sodium dodecyl sulfate solutions, disclosed thatsufficient heed must be paid to the continuous disintegration and reconstitution of themicelles in this time range.

An interesting twist in using micelles in electron transfer reactions is to slow rates sothat they might be measured more conveniently. Bunton and Cerichelli [89] found thatferrocene is too rapidly oxidized in water by ferricyanide but, in the presence of anionicmicelles into which ferrocene is preferentially partitioned, the reaction rate is easilyobtained. The rate attenuation results because the concentration of the negatively chargedferricyanide is lower in the vicinity of the anionic micelles than in bulk solution.

VIII. EXCITED STATE CHEMISTRY IN MICELLAR SOLUTION

The diffusion-controlled deactivation of excited states in micelles is essentially more com-plicated than for ordinary, slower reactions occurring in micellar media. As noted above,concentrative effects are chiefly responsible for the change in rate. First, quencher mole-cules must be present in each micelle and thus it is important to know the statisticaldistribution of the reactants among the micelles. Second, due to the very close proximityof the quencher with the excited species in the micelle, additional slow deactivation pro-cesses may occur as the contents of the micelles undergo mixing after the initial rapiddeactivation step [90]. Factors such as polydispersity, quencher or excimer migration andexchange, excimer reformation (a particular problem with pyrene which is the most com-mon excited probe used), and even the extent of counterion binding (for ionic quenchers)must be taken into account [90]. These additional concerns have meant that the elucida-tion of the kinetic mechanism whereby excited states are defused in micelles has notexperienced the same refinement as for ordinary reactions reacting in micelles. However,probing these problems is ongoing [11]. The potential rewards, however, of persevering inthe hunt for a satisfactory model for deactivation in micelles are so manifest that a livelyrecent literature in this field exists.


IX. MICELLAR CATALYSIS IN INDUSTRY

The earliest interest in micellar catalyis was in searching for quicker and more efficientroutes to synthesizing industrially important organic compounds. Fendler and Fendler [4]is a comprehensive compendium of the work performed in this area previous to 1975.Since then the chief emphasis in micellar catalysis has been on the theoretical or mechan-istic side even with potentially commercially significant products, e.g., Diels–Alder reac-tions [91]. In spite of the level of sophistication achieved in modeling and the wide range ofreactions investigated theoretically, industry has been slow in finding applications formicellar catalysis although a quick look at the patent literature shows that technologistsconsider many surfactant-mediated processes to have commercial promise. Emulsion poly-merization has been touted as a notable exception where the scaling up of the process hasbeen singularly successful [92,93]. Hydrophobic monomers concentrated within micellesmay undergo polymerization much more quickly than in solution and in the processhigher degrees of polymerization are often achieved [94]. Another advantage of usingmicelles is that the degree of solubilization of the monomers in aqueous solution is usuallymuch enhanced without the system becoming too viscous and less easily temperaturecontrolled. Ionic monomers are concentrated in the surface region of the micelles bycounterion binding and a similar catalysis effect is noted [95].

Other specific areas of micellar catalysis in which industry has expressed interest arein micellar phase-transfer catalysis and in the synthesis of mesoporous molecular sieves[92]. In the first example of the latter application, investigators at Mobil were able tocontrol pore size and properties by synthesizing the desired mesoporous material in thepresence of appropriately sized, structured, and charged micelles [96]. The burst ofresearch activity in this area that occurred in the next few years after this discovery hasbeen reviewed by Huo et al. [97].

X. MICELLAR CATALYSIS AND THE ENVIRONMENT

Concern for the well-being of the environment is expected to have a major impact onmicellar catalysis in two directions: the remediation of pollutants and the switch to lesstoxic detergents. The little that has been done to date on either score in this field has beenreviewed recently by Mackay [98]. In the first area, the dehydrochlorination of DDT insulfobetaine micelles [99] and the micellar destruction in soils of neurotoxic phosphorusesters such as fenitrothion, a commonly used pesticide for the control of spruce budworm[100], have been investigated. The latter involves cationic surfactants which can effectnucleophilic displacements on the esters with rates as high as 5400-fold. In the secondarea, surfactant workers are turning their attention to natural surfactants [101] and to thebenign, nonionic sugar surfactants, the alkyl glucopyranosides, and the maltosides in boththeir monomeric and polymeric forms [102], to determine their basic properties. As far asis known, no micellar catalysis studies have been reported with these ‘‘green’’ detergents.

XI. MICELLAR CATALYSIS IN REVERSE MICELLES

Increasing the nonpolar character of the medium causes micelles to do a flip-flop into areverse micelle with the hydrophobic alkyl tails of the surfactant pointing outwards andthe hydrophilic head groups inwards occupying ‘‘shore’’ positions on the enclosed water


pool. These reverse micelles or vesicles constitute along with the solvent a microemulsionof the self-explanatory water-in-oil (w/o) type. Oil-in-water (o/w) microemulsions havesimilarities to normal micelles in water. Clearly, the same conditions for catalysis basedon Berezin’s ideas of preferential partitioning of the reactants into regions of the vesiclefrom the solvent pertain as they would for a normal micelle. Thus, kineticists were notslow in exploring microemulsions for possible dramatic catalytic effects on test reactions.An added incentive was that reverse micelles were one step closer than normal micelles tothe liposomes or bilayer structures that are the form found in the natural membrane forthe two-tailed surfactants, the phospholipids. Another advantage often exploited was thegreater kinetic stability of the vesicle over the micelle. Friberg and Ahmad [103,104] areusually given credit for the first catalysis study in reverse micelles when they investigatedthe hydolysis of p-nitrophenyl dodecanoate in aqueous hexanol solution with CTAB pre-sent. An early kinetic foray into microemulsions revealed that a 7:0� 104-fold enhance-ment occurred in the thiolysis of p-nitrophenyl acetate by N-methylmercaptan indialkyldimethylammonium chloride vesicles [105]. Other studies were equally as promising[106–110].

Fletcher and Robinson [58] pointed out that the kinetic scheme was basically nodifferent for w/o microemulsions than for regular micellar solution. They investigatedkinetically the Ni2+/PADA complexation reaction in the sodium bis-(2-ethylhexyl)sulfo-succinate (AOT)/n-heptane system where a 10-fold rate enhancement was observed withno change in the rate coefficients. The AOT system is frequently the microemulsion ofchoice for w/o systems because its properties have been well characterized for differentratios of water to oil.

Despite this sanguine start, catalysis in microemulsions has not developed to thesame extent as catalysis in normal micellar solution with only about one paper in five inmicellar catalysis each year involving reverse micelles. Half the reason is the complexity ofthe system, which does not lend itself easily to modeling and characterization even in theAOT system where much is known [111]. An in-depth survey of catalysis with reversemicelles will not be attempted here because with the many opportunities the variousmicroemulsion systems offer for research possibilities the advancing front in knowledgeis thinly spread both horizontally and vertically compared with the state of the art withsimpler, normal micelles. A listing of representative papers in reverse micellar catalysispublished recently reveals the types of investigations that are currently of interest in thisfield: Azevedo et al. [112] carried out kinetic and stability studies with penicillin acylase inreversed micelles; Lee and Cho [113] used a microemulsion to synthesize nanocrystallinePbS particles; enzymes were trapped and reacted within reverse micelles [114]; and nucleo-philic aromatic substitutions were performed in reverse micelles [115]. Not surprisingly,biologically interesting reactions, especially involving enzymes, feature prominently in thelist. Despite this present lack of focus and of sophistication in modeling, it is safe to predictthat reverse micelles will continue to be actively explored as catalysts.

XII. MICELLAR CATALYSIS TOMORROW

Micelles in general as catalysts have a future not only in the practical side of chemistry,e.g., in effecting the rapid disintegration of environmental pollutants, but also in thetheoretical sphere of the science by revealing intimate details of reaction mechanisms,micelle structures, and the dynamics at work within micelles. Next-generation modelsare required to replace the PIE model and its analogs. The use of more sophisticated


techniques and the resorting to a more thorough analysis of the data in micellar catalysisshould see this development soon. Two interesting new developments are chemical trap-ping to elucidate the interfacial regions of micelles [116] and micellar autocatalysis [117].

A look at what can be done now and a hint at what soon will be routine is seen in thework of Woodward and Sakaguchi [118]. Hydrogen removal in benzophenones in micelleswas examined by pulsed microwave irradiation. By a combination of single pulse andpulse shift measurements, all the kinetic parameters were obtained for the reaction ineach micellar system, resulting in a global kinetic analysis. What emerged from this com-prehensive scrutiny was a clarification of the micelle interior and a better understanding ofthe role the ketyl radical was playing in the recombination kinetics. Pulsed microwave isjust one of the many new insightful approaches to micellar kinetics now possible. Anothermolecular information-rich technique that has seen little use to date in following micellarkinetics is NMR with all of its acronymic variants that in the hands of the experienced canreveal such intimate details of structure. As micelle kineticists become better versed in theinvestigative opportunities afforded by these and other new methods, it can be confidentlypredicted that great strides will be made in the understanding of the micelle itself as well asof the manner by which it affects reaction rates and mechanisms.

ACKNOWLEDGMENT

Financial support was provided by the Natural Sciences and Engineering ResearchCouncil of Canada.

REFERENCES

1. IV Berezin, K Martinek, AK Yatsimirski. Dokl Akad Nauk SSSR 194:840–842, 1970.

2. K Martinek, AK Yatsimirski, AV Levashov, IV Berezin. In: KL Mittal, ed. Micellization,

Solubilization, and Microemulsions. vol 2. New York: Plenum Press, 1977, pp 489–508.

3. LS Romsted, CA Bunton, J Yao. Curr Opin Colloid Interface Sci 2:622–628, 1997.

4. JH Fendler, EJ Fendler. Catalysis in Micellar and Macromolecular Systems. New York:

Academic Press, 1975.

5. D Myers. Surfactant Science and Technology. New York: VCH Publishers, 1988, p 182.

6. EAG Aniansson, SN Wall, M Almgren, H Hoffmann, I Kielman, W Ulbricht, R Zana, J

Lang, C Tondre. J Phys Chem 80:905–922, 1976.

7. EAG Aniansson. In: E Wyn-Jones, J Gormally, ed. Aggregation Processes in Solution.

Amsterdam: Elsevier, 1983, pp 70–99.

8. BH Robinson, S Bucak, A Fontana. Langmuir 16:8231–8237, 2000.

9. DG Hall. J Chem Soc, Faraday Trans 1 83:967–983, 1987.

10. A Patist, SG Oh, R Leung, DO Shah. Colloids Surfaces A:176, 3–16, 2000.

11. M Tachiya. In: GR Freeman, ed. Kinetics of Nonhomogeneous Processes. New York: John

Wiley, 1987, pp 575–650.

12. MJ Vold. J Colloid Interface Sci 116:129–133, 1987.

13. P Wangnerud, B Jonsson. Langmuir 10:3542–3549, 1994.

14. S Niu, KR Gopidas, NJ Turro. Langmuir 8: 1271–1277, 1992.

15. M Katsumata, K Katsatani, M Kawasaki, H Sato. Bull Chem Soc Jpn 55:717–720, 1982.

16. SJ Atherton, CMG Dymond. J Phys Chem 93:6809–6813, 1989.

17. G Cerichelli, G Mancini, L Luchetti, G Savelli, CA Bunton. Langmuir 10:3982–3987, 1994.

18. P Bilski, RN Holt, CF Chignell. J Photochem Photobiol A: Chem 110:67–74, 1997.

19. L Brinchi, P Di Profio, R Germani, S. Gianfranco, M Tugliana, CA Bunton. Langmuir

16:10101–10105, 2000.


20. CE Drennan, RJ Hughes, VC Reinsborough, OO Soriyan. Can J Chem 76:152–157, 1998.

21. AK Das, SK Mondal, D Kar, M Das. Int J Chem Kinet 33:173–181, 2001.

22. LS Romsted. In: KL Mittal, ed. Micellization, Solubilization, and Microemulsions. vol 2.

New York: Plenum Press, 1977, pp 509–530.

23. FM Menger, CE Portnoy. J Am Chem Soc 89:4698–4703, 1967.

24. CA Bunton, F Nome, LS Romsted, FH Quina. Acc Chem Res 24:357–364, 1991.

25. CA Bunton. J Mol Liq 72:231–249, 1997.

26. ZM He, PJ O’Connor, LS Romsted, D. Zanette. J Phys Chem 93:4219–4226, 1989.

27. CA Bunton, J Frankson, LS Romsted. J Phys Chem 84:2607–2611, 1980.

28. TJ Broxton, DB Sango. Aust J Chem 36:711–717, 1983.

29. DG Hall. J Phys Chem 91:4287–4297, 1987.

30. CA Bunton, CP Cowell, F Nome, LS Romsted. J Phys Org Chem 3:239–247, 1990.

31. N Ouarti, A Marques, I Blagoeva, M-F Ruasse. Langmuir 16:2157–2163, 2000.

32. CA Bunton. In: M Gratzel, K Kalyanasundaram. eds. Kinetics and Catalysis in

Microheterogeneous Systems. New York: Marcel Dekker, 1991, pp 34–36.

33. S Couderc, J Toullec. Langmuir 17:3819–3828, 2001.

34. EJR Sudholter, GB van de Langkruis, JBFN Engberts. Rec J Roy Neth Chem Soc 99:73–82,

1980.

35. CA Bunton, AK Yatsimirsky. Langmuir 16:5921–5931, 2000.

36. E Abuin, E Lissi. Bol Soc Chil Quim 42:113–134, 1997.

37. FH Quina, EO Alonso, JPS Farah. J Phys Chem 99:11708–11714, 1995.

38. MA Morini, PC Schulz, JE Puig. Colloid Polym Sci 274:662–668, 1996.

39. CJ O’Connor, A-L Tan. Aust J Chem 33:747–755, 1980.

40. MN Khan, E Ismail. Int J Chem Kinet 33:288–294, 2001.

41. G Savelli, R Germani, L Brinchi. In: J Texter, ed. Reactions and Synthesis in Surfactant

Systems. New York: Marcel Dekker, 2001, pp 175–246.

42. MN Khan, Z Arfin, IA Wahab, SFM Al, E Ismail. Colloids Surfaces A 163:271–281, 2000.

43. M Kabir-ud-Din, K Hartini, Z Khan. Int J Chem Kinet 33:377–386, 2001.

44. XC Xiang, XC Zeng, SQ Cheng, YT Li, JQ Xie. J Colloid Interface Sci 235:114–118, 2001.

45. TJ Broxton, NW Duddy. Aust J Chem 32:1717–1726, 1979.

46. T Kaneko, T Tokue, K Kubo, T Sakurai. Bull Chem Soc Jpn 72:2771–2780, 1999.

47. AB Mirgorodskaya, LA Kudryavtseva, YF Zuer, VP Archipov, ZS Idiyatullin. Mendeleev

Commun 196–198, 1999.

48. M. Munoz, A Rodrıguez, MM Graciani, ML Moya. Int J Chem Kinet 32:204–209, 2000.

49. OS Tee, AA Fedortchenko. Can J Chem 75:1434–1444, 1997.

50. OS Tee, OJ Yazbeck. Can J Chem 78:1100–1108, 2000.

51. OJ Yazbeck, OS Tee. Proceedings of the 84th CSC Conference, Montreal, 2001, Abstract 191.

52. AJ Kirby. Acta Chem Scand 50:203–210, 1996.

53. NJ Buurma, AM Herranz, JFBN Engberts. J Chem Soc, Perkin Trans 2 113–119, 1999.

54. Y Moroi. J Phys Chem 84:2186–2190, 1980.

55. AD James, BH Robinson. J Chem Soc, Faraday Trans 1 74:10–21, 1978.

56. VC Reinsborough, BH Robinson. J Chem Soc Faraday Trans 1 75:2395–2405, 1979.

57. PDI Fletcher, VC Reinsborough. Can J Chem 59:1361–1367, 1981.

58. PDI Fletcher, BH Robinson. J Chem Soc, Faraday Trans 1 80:2417–2437, 1984.

59. M Fischer, W Knoche, BH Robinson, J. Wedderburn. J Chem Soc, Faraday Trans 1 75:119–

131, 1979.

60. M Fischer, W Knoche, PDI Fletcher, BH Robinson, NC White. Colloid Polym Sci 258:733–

742, 1980.

61. JR Hicks, VC Reinsborough. Can J Chem 62:990–994, 1984.

62. YL Favaro, VC Reinsborough. Can J Chem 72:2443–2446, 1994.

63. KA Berberich, VC Reinsborough. Langmuir 15:966–969, 1999.

64. DJ Jobe, VC Reinsborough. Aust J Chem 37:1153–1159, 1984.

65. TJ Connolly, VC Reinsborough. Can J Chem 70:1581–1585, 1992.


66. JR Hicks, VC Reinsborough. In: KL Mittal and B Lindman, eds. Surfactants in Solution. vol

2. New York: Plenum Press, 1983, pp 1207–1215.

67. JR Hicks, VC Reinsborough. Can J Chem 62:280–284, 1984.

68. MN Khan, FBH Ahmad. Colloids Surfaces A 181:11–18, 2001.

69. AC Dash and SS Mohammed. Indian J Chem 31A:166–170, 1992.

70. D Silicia, S Rubio, D Perez-Bendito. Anal Chim Acta 297:453–464, 1994.

71. Y Miyake, M Shigeto, M Teramoto. J Chem Soc, Faraday Trans 1 82:1515–1526, 1986.

72. LP Loginova, OS Chernysheva, N Levin, NA Sukhomlinova. Visn Khark Univ 230–234,

1998.

73. T Harada, N Nishikido, Y Moroi, R Matuura. Bull Chem Soc Jpn 54:2592–2597, 1981.

74. H Watanabe, T Saitoh, T Kamidate, K Haraguchi. Mikrochim Acta 106:83–90, 1992.

75. AI Carbone, P Cavasino, C Sbriziolo. J Phys Chem 91:4062–4065, 1987.

76. CA Bunton, HJ Foroudian, ND Gillitt, CRWhiddon. J Colloid Interface Sci 215:64–71, 1999.

77. KA Berberich, VC Reinsborough, CN Shaw. J Solut Chem 29:1017–1026, 2000.

78. KK Ghosh, A Pandey, S Roy. Colloids Surfaces A 163:293–300, 2000.

79. A Rodrıguez, MM Graciani, M Munoz, ML Moya. Langmuir 17:1860–1863, 2001.

80. CA Bunton, ND Gillitt, MM Mhala, JR Moffatt, AK Yatsimirsky. Langmuir 16:8595–8603,

2000.

81. L Brinchi, R Di Profio, R Germani, G Savelli, N Spreti, CA Bunton. J Chem Soc, Perkin

Trans 2 361–366, 1998.

82. R Prado-Gotor, R Jimenez, P Perez-Tejeda, P Lopez-Cornejo, M. Lopez-Lopez, A Sanchez, F

Muriel-Delgado, F Sanchez. Progr React Kinet 25:371–407, 2000.

83. SA Alkaitis, G Beck, M. Gratzel. J Am Chem Soc 97:5723–5729, 1975.

84. LK Patterson, M Gratzel. J Phys Chem 79:956–960, 1975.

85. KV Ponganis, MA De Araujo, HL Hodges. Inorg Chem 19:2704–2709, 1980.

86. D Grand, S Hautecloque. J Phys Chem 94:837–841, 1990.

87. B Peterson, C. Reddy, CJ Marzzacco. Rec Res Dev Photochem Photobiol 3:77–78, 1999.

88. H Bruhn, J Holzwarth. Ber Bunsenges Phys Chem 82:1006–1011, 1978.

89. CA Bunton, G Cerichelli. Int J Chem Kinet 12:519–533, 1980.

90. M Almgren. In: M Gratzel, K Kalyanasundaram, eds. Kinetics and Catalysis in


91. S Otto, JBFN Engberts. In: J Texter, ed. Reactions and Synthesis in Surfactant Systems. New

York: Marcel Dekker, 2001, pp 247–263.

92. JF Rathman. Curr Opin Colloid Interface Sci 1:514–518, 1995.

93. K Tauer. In: J Texter, ed. Reactions and Synthesis in Surfactant Systems. New York: Marcel

Dekker, 2001, pp 429–453.

94. I Lacik, J Selb, F Candau. Polymer 36:3197–3211, 1995.

95. S Aoki, Y Morimoto. Colloid Polym Sci 273:733–739, 1995.

96. JS Beck, JC Vertuli, WJ Roth, ME Leonowicz, CT Kresge, KD Schmitt, CTU Chu, DH

Olson, EW Sheppard, SB McCullen, JB Higgins, JL Schlenker. J Am Chem Soc 114:10834–

10843, 1992.

97. O Huo, DI Margolese, U Ciesla, DG Demuth, P Feng, TE Gier, P Sieger, A Firouzi, BF

Chmelka, F Schuth, GD Stucky. Chem Mater 6:1176–1191, 1994.

98. RA Mackay. In: J Texter, ed. Reactions and Synthesis in Surfactant Systems. New York:

Marcel Dekker, 2001, pp 373–383.

99. A Rodrıguez, MM Graciani, A Guinda, M Munoz, ML Moya. Langmuir 16:3182–3186,

2000.

100. VK Balakrishnan, X Han, GW vanLoon, E Buncel. Proceedings of the 84th CSC Conference,

Montreal, 2001, Abstract 893.

101. K Holmberg. Curr Opin Colloid Interface Sci 6:148–159, 2001.

102. C. Stubenrauch. Curr Opin Colloid Interface Sci 6:160–170, 2001.

103. S Friberg, SI Ahmad. J Phys Chem 75:2001–2004, 1971.

104. SI Ahmad, S Friberg. J Am Chem Soc 95:5196–5199, 1972.


105. IM Cuccovia, RMV Aleixo, RA Mortara, PB Filho, JBS Bonilha, RH Quina, H Chaimovich.

Tetrahedron Lett 3065–3069, 1979.

106. JH Fendler. Acc Chem Res 13:7–13, 1980.

107. RA Mackay, K Letts, C Jones. In: KL Mittal, ed. Micellization, Solubilization, and

Microemulsions. vol. 2. New York: Plenum Press, 1977, pp 801–816.

108. T Kunitake, T Sakamoto. J Am Chem Soc 100:4615–4617, 1978.

109. CA Bunton, F de Buzzaccarini. J Phys Chem 85:3139–3141, 1981.

110. CA Bunton, F de Buzzaccarini. J Phys Chem 85:3142–3145, 1981.

111. PL Luisi. In: M Gratzel, K. Kalyanasundaram, eds. Kinetics and Catalysis in


112. AM Azevedo, LP Fonesca, DMF Prazeres. Biocatal Biotransform 17:401–415, 2000.

113. S Lee, SM Cho. Hwahah Konghak 38:43–46, 2000.

114. C Shah, S Sellappan, D Madamwar. Process Biochem 35:971–975, 2000.

115. NM Correa, EN Durantini, JJ Silber. J Org Chem. 65:6427–6433, 2000.

116. LS Romsted. In: J Texter, ed. Reactions and Synthesis in Surfactant Systems. New York:


117. J-C Micheau, R Nagarajan. In: J Texter, ed. Reactions and Synthesis in Surfactant Systems.

New York: Marcel Dekker, 2001, pp 413–428.

118. JR Woodward, Y Sakaguchi. J Phys Chem 105:4010–4018, 2001.


14Multiple Effects of Water Pools and TheirInterfaces Formed by Reversed Micelles onEnzymic Reactions and Photochemistry

AYAKO GOTO, YUKO IBUKI, and RENSUKE GOTO University of Shizuoka,Shizuoka, Japan

I. INTRODUCTION

Amphiphilic molecules, when dissolved in organic solvents, are capable of self-assembly toform reversed micelles. The reversed micelles are structurally the reverse of normalmicelles in that they have an external shell made up of the hydrocarbon chains of theamphiphilic molecules and the hydrophilic head-groups localized in the interior of theaggregate. Water molecules are readily solubilized in this polar core, forming a so-calledwater pool. This means that reversed micelles form microcompartments on a nanometerscale. The reversed micelles can host all kinds of substrate molecules whether hydrophilic,hydrophobic, or amphiphilic due to the dynamic structure of the water pool and theinterface formed by the surfactant layer, in contrast with a liposome system. The proper-ties of water molecules localized in the interior of reversed micelles are physicochemicallydifferent from those of bulk water, the difference becoming progressively smaller as thewater content in the micellar system increases [1,2]. The anomalous water at lowWo ¼[water]/[surfactant] obviously influences the chemical behavior of host moleculesin the water pools.

Ultimately, the water pool and the interfacial surfactant layer can exhibit multiplecatalytic effects, which result from the concentrations of reactants localized in the nano-compartmentalized region and the physicochemical properties of the micellar environ-ment. Accordingly, the reversed micellar systems have the possibility of controlling themultiple effects on the reactions by changing the physical factors of the reversed micellarsystems such as water mobility, micropolarity, and electrostatic force.

There are a variety of other types of nonbilayer lipid structures such as reversedmicelles sandwiched between monolayers of the lipid bilayers in vivo, while the mainstructural pattern of biological membranes is the flat bilayer of lipid molecules. Thesenonbilayer structures can explain many processes occurring in the living cell, such asfusion, and exo- and endo-cytosis. Because the water in the reversed micelle resemblesthat adjacent to biological membranes or biological reversed micelle-like microcompart-ments, reversed micelles may be an appropriate model for investigating biological catalysisat the molecular level [3–5].

In this study, we describe the multiple catalytic effects of the water pools and theirinterface in the reversed micelles, mainly based on our recent studies. One subject of our


studies is concerned with the catalysis by enzymes entrapped in water pools of reversedmicelles and photochemistry in the interface of water pools by UV irradiation from theviewpoint of a biomembrane model in vivo. Another subject is concerned with biopolymersynthesis in the reversed micellar system with a view to its application in bioengineering[6]. We focus on the enzymic polymerization of nucleotides in a reversed micellar systemutilizing the liquid/solid interface.

II. HOW CAN WATER POOLS AND THEIR INTERFACES IN REVERSEDMICELLES BE CONTROLLED AS A REACTION FIELD?

A. Advantages of Reversed Micellar System as a Reaction Field

We describe the physicochemical aspects of a reversed micellar system, and how to controlthe water pools and the interfacial surfactant layer as a reaction field by changing Wo andthe hydrophilic group of the surfactant.

The reversed micellar system has many advantages in common with aqueous/organicmedia, as follows [7,8]:

1. A reversed micellar system catalyzes enzymically a substrate that is insoluble inwater.

2. A reversed micellar system catalyzes the reverse reaction of hydrolysis.3. A reversed micellar system inhibits side reactions, which may easily proceed in

bulk water and can maintain the stability of a product that is unstable in a bulkaqueous solution [9–11].

4. A reversed micellar system inhibits self-digestion of the enzyme and contam-ination by micro-organisms.

5. A reversed micellar solution controls substrate specificity [12].

On the other hand, normal micelles scarcely have the advantages mentioned above(points 1–4), although they can control substrate specificity and self-catalysis on themicellar surface in an aqueous solution [13–15]. Furthermore, the reversed micelles havefavorable advantages as a microreactor, which are different from those of the aqueous/organic solvents. First, we should mention the macroscopic properties of the reversedmicelles. The reversed micellar systems are optically transparent and spontaneouslyformed and the equilibrium state is achieved fairly quickly. Also, large amounts of hostmolecules can be incorporated without disturbing these macroscopic properties. The bestpoint of the reversed micelles as a microreactor is the ability to control easily the sizes ofreversed micelles and water pools by changing Wo. The state of the interface between thewater pool and surfactant monolayer depends strongly on the hydrophilic group of thesurfactant, and accordingly one can control the electrostatic field by changing the hydro-philic group. Thus, the chemical structure of the surfactant andWo would appear to play akey role for the appearance of multiple catalytic effects on the reactions in the water pools.

B. Relationship Between Sizes of Reversed Micelles and Water Poolswith Wo

Because reversed micelles formed by sodium bis-(2-ethylhexyl)sulfosuccinate (AOT) in anapolar organic solvent have been extensively studied [1,2,8,9,16–18], and the propertiesand structure have been fairly well elucidated compared with other surfactants, we


describe mainly the properties of the AOT reversed micellar system as a type of micro-reactor.

Wo is a key parameter which significantly affects the physical properties of AOTreversed micelles. In the case of an AOT/oil solution, discontinuity of several physicalproperties of the solubilized water is observed at Wo ’ 10 [16]. Below Wo ’ 10, the wateris bound to the AOT polar head-groups and counterions, and further addition of waterleads to the appearance of free water in the core of the water pools. However, the state ofthe water in the AOT reversed micelles, especially below Wo ’ 2, appears unusual. Wefound that the solution enthalpy of the water in AOT/various organic solvents solutionsindicated a great change in the state of the solubilized water [17,18].

We have studied changes in the dynamic state of the water in AOT reversed micelles/toluene-d8 with Wo by 17O-NMR spectroscopy [19] because 17O relaxation rates of thewater reflect the motion of single water molecules [20–22]. Figure 1 shows line widths of17O-NMR spectra of the water as a function of Wo, where the spectra were too broad toobserve a signal below Wo ’ 2 at 278C (dotted line in Fig. 1). This indicates that themotion of water molecules below Wo ¼ 2 is significantly suppressed and the activationenergy of the motion increases with decreasing Wo based on the relation of the line widthswith temperature. The 1H and 13C relaxation times of AOT molecules showed that themobility of the AOT molecules constituting monolayers is restricted at smaller Wo inagreement with that of the water molecules.

Manabe et al. [23] found the formation of the nonionized hydrate, AOT(H2O)2,below Wo ¼ 2 by the measurement of its conductivity. Therefore, water below Wo ¼ 2should correspond to nonionized hydrated water different from bound water belowWo ¼ 10. (Hereafter called immobilized water.) When water in limited amounts isadded to the anhydrous AOT solution, immobilized water is formed. A computer simu-lation showed that neither a spherical micelle nor a water pool is formed below Wo ¼ 2[24]. It is unlikely that any water pool would be formed at Wo ¼ 2 at which the AOTaggregates are compact and rigid. When a small amount of water is added to the AOT

FIG. 1 17O line width of the H217O signal of AOT reversed micelles as a function of Wo at various

temperatures. Width of signal was measured at half-height at 278C(*), 358C(&), 558C(~) and

658C(^). (From Ref. 19.)


aggregates at Wo = 2, bound water appears and immobilized water disappears gradu-ally up to Wo ’ 4.

Interfacial phenomena in the AOT reversed micelle system are governed by thecompositions of various rotational isomers [25,26]. By measuring spin coupling constants,the AOT aggregates at small Wo was assumed to include the rotamer in which two alkylchains are in a transposition, causing the chains to extend in opposite directions. AboveWo ’ 4, the rotamer of which both alkyl groups are in a gauche position increases greatlyand comes to form the water pool. Above Wo = 10, free water appears in the center of thewater pool. These processes are shown in Scheme 1.

The radius of the water pool, rw, defined by the equation of rw ¼ 1:5WoA, wasderived from geometric considerations of the dimensions and packing of surfactant mole-cules [27–29]. Using synchrotron radiation small-angle x-ray scattering, Hirai et al. [30]found a linear relationship between the radius of the water pool and Wo in a high Wo

range (16 <Wo<50) of AOT, but not in a lowWo range (0 <Wo<12). The discrepancieswere understood to be a transient oligomerization of reversed micelles due to the forma-tion of cylindrical reversed micelles. A recent structural study of the water/cyclohexane/Co(AOT)2 system suggests a phase transition of reversed micellar shape from spheres tocylinders at Wo ¼ 0�10 and from cylinders to spheres at Wo ¼ 10�25 [31]. Bardez and Vy[32] reported the counterion-driven sphere to cylinder transition of reversed micelles usingthe Al3+ ion. Therefore, the linear relation of sizes of the water pools with Wo cannot beinferred conclusively.

SCHEME 1 Structural changes in AOT reversed micelles with Wo.


C. Dynamic Properties of Reversed Micellar System

The reversed micelles have very dynamic structures in which the micelles can exchangesurfactant molecules with each other at a high rate. Despite the high mobility of sur-factant molecules, the micellar interface is rather well kept. The fluidity and/or flexibilityof the surfactant monolayer are important for understanding the state of the interface ofthe water pool. In order to study the properties, the relaxations of the interface havebeen observed using the iodine-laser temperature jump (ILTJ) technique [33]. The ILTJcan measure the relaxation as a response of the system to perturbations in the micro-emulsion structure induced by rapid changes in the temperature of the microemulsion.The interfacial bending modulus corresponding to the flexibility of the monolayer isdefined as twice the amount of energy required to bend a unit area of surface by aunit of curvature. It corresponds to approximately 0.4 kT for AOT. The bending mod-ulus is reported to decrease with increasing temperature and by the addition of NaCl[33,34]. The AOT reversed micelle containing a polyoxyethylene surfactant has a lowerbending energy, but that containing an alcohol has a larger bending energy than thepure AOT reversed micelle [35,36]. Finally, almost all aspects of the dynamics andthermodynamics of reversed micellar systems are affected by the fluidity and/or flex-ibility of the interface.

We found by NMR measurement that the fluidity of the AOT monolayer is parallelto the mobility of the water in the interface of the monolayer [19]. The appearance oflonger relaxation times of water in the reversed micelles has been studied also by time-resolved fluorescence Stokes shift [37]. The mobility of water inside the AOT reversedmicelles was revealed to be substantially reduced regardless of the kinds of counterions[38,39]. Riter and coworkers [40,41] showed that Na+ ions as well as the AOT head-groupin the AOT reversed micelles are responsible for reducing the water motion in theseenvironments. According to Shirota and Horie [42], the solvation dynamics of methanolin reversed micelles becomes faster with a large ratio of polar solvent to surfactant (W ¼[polar solvent]/[AOT]), but the W dependence of the solvation dynamics of acetonitrile inreversed micelles is not observed, suggesting that the intermolecular hydrogen-bondingnetwork is responsible for the different W dependencies. Therefore, the source of theseslow relaxations is generally discussed in terms of water bound to the supramolecularassembly as well as the immobilization of the water in a confined environment [43]. Themobility of the water molecules in the reversed micelles is parallel with the fluidity of themonolayer. The water bound with the interface of the water pool is exchangeable with freewater in the core at largerWo and the mobility of the bound water is faster than that in thewater pool at low Wo.

D. Observation of AOT Reversed Micelles Using Freeze–FractureTransmission Electron Microscopy and Cryotransmission ElectronMicroscopy

Reversed micelles have very highly dynamic structures and are in rapid equilibrium withsurfactant monomers. Therefore, it is usually difficult to observe their real features bymicroscopy. A freeze–fracture transmission electron microscope (TEM) would probablyshow the real picture of a reversed micellar solution because a freeze–fracture film of thereversed micelles is made by rapid cooling to �1508C to stop instantly the dynamic natureof the structure. Figure 2(a) shows an electron micrograph of the AOT reversed micellarsolution (5% w/v AOT–iso-octane solution, Wo ¼ 1) [44]. The visual observation by a


freeze–fracture TEM indicates that the reversed micelles atWo ¼ 1 are spheroidal and thatthe particles exist individually. This is in agreement with the previous result of NMRspectroscopy [19] that the AOT aggregate below Wo ¼ 2 is rigid and compact.

The water pool atWo ¼ 20 contains free water. Figure 2(b) shows the image of a 5%w/v AOT/iso-octane solution with Wo ¼ 20. Small particles with sizes less than 20 nmwere observed in the overall picture. Besides these small particles, large particles withvarious sizes less than 200 nm were observed. The large particles were of nonsphericalshape and appeared to be a floc of the reversed micelles. The image at Wo ¼ 20 suggeststhat the small reversed micelles assembled together to form flocs in 5% w/v AOT solution.Flocculation, which means the formation of loose aggregates of the reversed micelles, isdifferent from percolation, which means the formation of more rigid and larger aggregatesor a network structure of the reversed micellar solution, often accompanied by phaseseparation. It is well known that percolation occurs in higher AOT and water concentra-tions [34,45–48]. Manabe et al. [23] have found by measurement of electrical conductivitythat the interaction of AOT reversed micelles themselves leads to the formation of clustersbefore the threshold of percolation. Hasegawa et al. [49] suggest this by excitation energytransfer. These results are supported using a cryo-TEM. It is confirmed that flocculationoccurs easily in higher AOT and water concentrations due to the decrease in bendingmodulus. We can illustrate the effects of Wo and the AOT concentration on the reversedmicellar system, as shown in Scheme 2. If Wo remains constant, the micelles do not changein the structure but increase in number when the concentration of AOT increases [50].However, the increase in fluidity and/or flexibility at the surfactants/water interface results

FIG. 2 Freeze–fracture transmission electron micrographs of AOT reversed micelles. (From Ref.

44.)


in increased attractive interactions to induce flocculation and further the percolationthreshold.

E. Effects of Electrostatic Charges of Surfactants and Cosurfactants onReversed Micellar Properties

The electrostatic charges of surfactants seriously affect the localization of host moleculesin the water pool. Monte Carlo simulation in which ionic reversed micelles are treated asspherical entities showed the presence of the electrical double layer in the interface of thewater pool, and the distribution of counterions followed the Poisson–Boltzmann approx-imation [51]. Mancini and Schiavo [52] assumed recently, by the yield of halogenation,that the specific interactions between bromide or chloride ions and an ammonium head-group in cationic reversed micelles keep the ions in a defined position on the interface.

The micropolarity of the reversed micelles influences the extent of ionization of ionicsurfactant [53]. Correa et al. [54] estimated the micropolarity using 1-methyl-8-oxyquino-linium betaine (QB) as an optical probe. The more structured water showed lower micro-polarity. At Wo > 10, once the polar heads of the anionic (AOT) as well as cationic(benzyl-n-hexadecyltrimethylammonium chloride, BHDC) surfactants were completelyhydrated, both reversed micelles showed the same micropolarity. The inner polarity ofthe aggregate of the nonionic surfactant with polyoxyethylene group at Wo ¼ 0 was higherthan that of AOT–hexane, due to the hydrogen bonding between QB and the free OHgroups of these surfactants. The polarity in the penta-ethyleneglycoldodecyl ether (C12E5)reversed micelles in heptane at Wo ¼ 10 is nearly the same as those in ionic surfactantreversed micelles. However, water does not form a pool in the core of the reversed micelles;rather, it seems to be dispersed between the polyoxyethylene chains [55]. It is of interestthat at the maximum possible Wo, the polarity measured by QB never reaches the value ofpure bulk water regardless of ionic and nonionic surfactants [54].

SCHEME 2 Effects of Wo and surfactant concentration on reversed micelle.


As Wo increases, the size of the water pool increases [56] and free water appears inthe center of the water pool, resulting in an increase in micropolarity [10,57–59]. It isconcluded that ionization of the ionic groups of the surfactant layer of reversed micelleswould be enhanced with increasing Wo due to increase in micropolarity of the water pool[53].

AOT which has two bulky alkyl chains can easily form reversed micelles, but asurfactant with a single alkyl chain cannot easily form reversed micelles in aliphaticsolvents such as iso-octane. Therefore, it needs a cosurfactant such as octanol for sin-gle-chain surfactants to form reversed micelles in iso-octane [53]. Distribution of alkanolsbetween the bulk organic solvent and the micellar pseudophase was measured based ontheir effect on the fluorescence intensity of indolacetate ions incorporated into AOTreversed micelles [60]. For all the alkanols, the extent of incorporation into the micellesdecreased with increasing Wo from 4 to 20. At a given Wo, the incorporation of the n-alkanol increased when the length of the alkyl chain increased. The ILTJ techniquerevealed that the aliphatic chain alcohol induced an increase in the interfacial rigidity.

F. Multiple Effects of Water Pools and Their Interface Region onReaction Field

The relationships between the physical properties of the water pools and the interfacialregion with Wo are summarized in Scheme 3. Multiple effects on the reaction in thereversed micelles are caused by the co-operation of some effects.

Because the nanosizes of reversed micelles as well as water pools change withWo, thevolume of the compartmentalized entity can be easily controlled. Host molecules arecompartmentalized and localized in the reversed micelles by the electrostatic and hydro-phobic interactions. The bound water exists overwhelmingly below Wo ¼ 10 regardless ofcationic and anionic surfactants. However, since in the case of nonionic ethylene oxidesurfactants, the oxygen atom of ethylene oxide is hydrated with two to four water mole-cules, the appearance of free water depends on the chain length of ethylene oxide [61–64].

SCHEME 3 The relationship between physical properties of water pools and theirinterfacial region with Wo.


With decreasing Wo, the micropolarity decreases and the microviscosity increases[65]. This means that the mobility of water molecules as well as host molecules comes tobe restricted with a decrease inWo. This might lead to the suppression of the side reactionsand the stability of the substrate or product. Furthermore, the electrostatic force can becontrolled by changing the hydrophilic groups of surfactants. This is important because thelocalization of the ionic host molecules depends on the electrostatic force of the interface.

We used the following three kinds of surfactants to study the multiple catalyticeffects of the water pools and their interface region: AOT, hexadecyltrimethylammoniumchloride (HTAC), and octa-ethyleneglycoldodecyl ether (C12E8) were used as anionic,cationic, and nonionic surfactants, respectively. The respective chemical structures areshown in Scheme 4. Unlike the ionic surfactant, the hydrophilic part of nonionic surfac-tants with polyoxyethylene may be longer than the hydrophobic part of the molecule.HTAC and C12E8 reversed micellar solutions were prepared by dissolving an appropriateamount of HTAC in iso-octane/octanol (85:15 v/v) and of C12E8 in iso-octane/octanol(90:10 v/v), respectively. The respective contents of octanol correspond nearly to theminimum contents in bulk organic solvent required to solubilize a large amount ofwater in reversed micelles [53].

The mean diameter of HTAC reversed micelles (Wo ¼ 15, Tris/HCl buffer, pH 8.0)in iso-octane/octanol measured by dynamic light scattering [53] was 280 nm, while that of0.2 M AOT in iso-octane was 20 nm, nearly corresponding to those estimated by othermethods [56,66]. The diameter of C12E8 reversed micelles (Wo ¼ 15, Tris/HCl buffer, pH8.0) was � 430 nm.

C12E8 has eight ethylene oxide (EO) residues and nine oxygen molecules. Whilepolyoxyethylene glycol (PEG) can acquire nearly two molecules of water per EO group,the EO group in an organized assembly, such as in reversed micelles, is associated withfour molecules of water [61–64]. Almost all the water molecules are hydrated to the EOchains of C12E8 up to Wo ’ 30 [67]. Thus, in the case of C12 E8 reversed micelles the freewater is assumed to appear above Wo ¼ 30.

SCHEME 4 Chemical structures of surfactants.


The respective models of the reversed micelles used in this study were shown inScheme 5. It was reported that the percolation was enhanced by the enzyme [47,68] andtended to be suppressed in the presence of octanol [54]. However, because the concentra-tion of respective surfactants used was relatively high (0.2 M), the flocculation occured to agreater or lesser degree.

III. STUDY OF WATER POOLS AND INTERFACES AS A MODEL OFBIOMIMETIC MICROENVIRONMENTS

A. Catalytic Activity of Hexokinase in Reversed Micelles

It is generally known that substrates are bound to enzymes by multiple weak attractions ofelectrostatic bonds, hydrogen bonds, van der Waals forces, and hydrophobicity. Theseattractions strongly depend on the microenvironment in the neighborhood of the enzyme.Most of the enzymes in a living system are working in an environment which is quitedifferent from an aqueous bulk solution. The biomembrane and the bound water existingnear the boundary surface must affect the catalytic activity of an enzyme in cells to agreater or lesser extent, resulting in the regulation of the enzyme function. It has beenreported that the effect of biomembranes on the catalytic activity of the enzymes has beenscarcely detected in an aqueous solution containing membrane fragments because theenzyme exists far from the membrane surface [69]. For example, concerning the activityof enzymes entrapped inside liposomes, it is assumed that the conformation of the enzymeis probably the same as in the bulk aqueous phase because the aqueous interior of lipo-somes is significantly larger than the size of the enzymes [70].

A reversed micelle is fairly far from a living cell model, but it has some advantages inexamining the effects of the hydration and the membrane shell on the activity of theenzyme. An enzyme molecule is located or entrapped in the water pool, which is a nano-compartmentalized system [44,70], and the water pool and surfactant shell grant protec-tion to the enzyme against inactivation caused by the bulk organic phase. Reversedmicelles are very dynamic in contrast with the liposome system and the enzyme kineticsare not affected by the exchange processes between reversed micelles. Various kinds ofenzyme reactions in the reversed micelles have been examined to clarify the effect of thebiomembrane circumstance on the enzyme activity by changing Wo [4,71–76]. We focused

SCHEME 5 Models of the reversed micelles used.


on hexokinase (HK), which is of importance to sugar metabolism in cells and is basicallynot a membrane enzyme [53]. HK catalyses the phosphorylation of glucose, using ATP asa phosphoryl donor in the presence of Mg2+. HK has a cleft shape [77]. A dramaticconformational change occurs when glucose binds with HK and, at the same time,water molecules are repelled [77,78]. Because this change in HK conformation is assumedto be related to its function, the activity of HK would depend greatly on the microenvir-onment. The HK derived from mammalian cells binds preferentially to outer mitochon-drial membrane pores by the insertion of the hydrophobic NH2-terminal tail into thehydrophobic pore of the membrane [79–82]. This binding is considered to involve bothelectrostatic and hydrophobic forces, but the yeast HK used in this experiment has beenreported to lack the hydrophobic NH2-terminal and seemingly not to bind with themitochondrial membrane [81,83]. Consequently, we studied the microenvironment inwhich yeast HK catalyzes by using the reversed micelles.

The HK catalytic reaction in a reversed micellar solution was carried out by applyinga micelles-mix technique. ADP and ATP contents were analyzed by HPLC to determinethe kinetic parameters [53]. The concentrations of HK, substrates, and Mg2+ ions werekept constant in water pools by changing Wo in order to avoid the concentration effects ofHK and substrates on the catalytic activities [84–86]. Figures 3(a) and 3(b) show the effectsof Wo on HK activity in the reversed micelles of AOT and HTAC, respectively. In thecases of AOT and HTAC, the profiles revealed an optimum at Wo ¼ 10 although theoverall activities in the HTAC solution were two to three times higher than those in theAOT solution. In the case of C12E8, it was observed that the activity increased with Wo,reaching a nearly constant value above Wo ¼ 20 and indicating a maximum value atWo ¼ 30. The ratio of the HK activity in the C12E8 reversed micelles of Wo ¼ 30 tothat in the aqueous bulk solution was 0.92. On the other hand, the ratios of the AOTreversed micelles (Wo ¼ 10) and of the HTAC reversed micelles (Wo ¼ 10) were 0.015 and0.045, respectively. These results mean that the HK activity is maintained in the C12E8

reversed micelles to the same extent as in the aqueous bulk solution, but the activities aregreatly inhibited in the AOT and HTAC reversed micelles.

Table 1 [53] shows the apparent kinetic parameters of HK in three kinds of reversedmicellar solutions based on concentrations of substrates in the water pools and in theaqueous bulk solution. The apparent Vmax in the AOT, HTAC, and C12E8 reversedmicelles, and in the aqueous bulk solution increased in that order. HK may become fairlydenatured in the AOT and HTAC reversed micelles because the ionic surfactants in theaqueous bulk solution are usually known as strong denaturants for protein.

In addition, since the micropolarity of the water pools increases with increases inWo, the ionization of ionic groups in the water pools is enhanced with increasing Wo andconsequently the denaturation of HK in the inner surface may be promoted. However, theinhibition of HK activity was confirmed to be reversible at least in the AOT reversedmicelles (Fig. 4). HK activity was inhibited at Wo ¼ 5 as well as at Wo ¼ 30, comparedwith that at Wo ¼ 10. However, even after HK was left in the water pools of the AOTreversed micelles of Wo ¼ 5 and 30 at 378C for 2 h, HK activity obviously recovered byadjusting to Wo ¼ 10. In addition, it was confirmed that HK specificity was maintained tosome extent in the AOT reversed micelles by examining various substrates [53].

It is often observed that the catalytic activity of a solubilized enzyme depends on Wo

and there is an optimal value of Wo at which the solubilized enzyme shows maximumactivity. The optimal Wo depends on the nature of the enzyme and surfactant. Someresearchers explain that the highest catalytic activity is observed when the entrappedenzyme molecules are forced into close contact with the inner micellar interface, helping


the enzymes to retain their most active conformation by making them more rigid [87].According to the model of Ruckenstein and Karpe [88], the factor that really controls theenzyme reaction rate is the concentration of the substrate at the surface of the enzyme,depending on the electrical potential distribution in the water pool.

The HK catalytic reaction should proceed favorably with increasing Wo because HKactivity is extremely high in the bulk aqueous solution. However, the activity indicated amaximum at Wo ¼ 10 in both of the AOT and HTAC reversed micelles, as shown in

FIG. 3 Effect ofWo on HK activity in reversed micelles. (a) 0.1M AOT/iso-octane reversed micelles

at 378C; [ATP]wp,o=10 mM, [DG]wp,o=30 mM, [HK]wp=1.74 mg/ml, [MgCl2]wp=10 mM, and 50

mM Tris/HCl buffer (pH 8.0). (b) 0.1M HTAC/iso-octane–octanol reversed micelles at 378C;[ATP]wp,o=10 mM, [DG]wp,o=30 mM, [HK]wp=1.74 mg/ml, [MgCl2]wp=10 mM, and 50 mM

Tris/HCl buffer (pH 8.0). (c) 0.1M C12E8/iso-octane–octanol reversed micelles at 378C;[ATP]wp,o=10 mM, [DG]wp,o=30 mM, [HK]wp=0.0174 mg/ml, [MgCl2]wp=10 mM, and 50 mM

Tris/HCl buffer (pH 8.0). (From Ref. 53.)


Figs 3(a) and 3(b). ATP favorably exists in the center zone of the water pool of the AOTreversed micelles due to the repulsive electrostatic force between the negative charges ofthe ATP and AOT surface layer. However, it is adsorbed on the inner surface of theHTAC reversed micelles due to the attractive electrostatic force. In a similar manner,the cationic and hydrophobic moieties of HK are assumed to be inserted into the negativecharged layers of the AOT reversed micelles due to the attractive electrostatic force andthe hydrophobic interaction, but in the case of HTAC, it is located to some extent far fromthe cationic surface layer due to the repulsive electrostatic force. In the case of AOT, thisassumption may be supported by the effect of KCl in high concentrations because a higherconcentration of KCl showed higher activity and the optimum Wo disappeared [53]. Sucha sufficient absorption of K+ ions on the inner surface layer of the AOT reversed micellesleads to a neutralization on the surface, and HK can be repelled from the AOT layer,resulting in a shortening of the distance between ATP and HK in the water pools.

If the enzyme is assumed to be spherical, the radius of HK consisting of four sub-units is estimated to be 6.5–7.0 nm from the atomic force microscopy (AFM) image. Thedifference of the sizes of the reversed micelles may contribute to the high activity of HK in

TABLE 1 Apparent Kinetic Parameters of HK in Three Kinds of

Reversed Micellar Solution

Bulk water AOT HTAC C12E8

Km (�mol/ml) 1:04� 0:07 3:74� 0:07 2:80� 0:18 1:04� 0:06

Vmax (�mol/min) 26:49� 0:72 5:11� 0:11 13:34� 0:31 18:87� 0:60

Source: Ref. 53.

FIG. 4 Recovery of HK activity by changing Wo in AOT reversed micelles. The respective HK

activities were at (1) Wo=5, (2)Wo=10 and (3) Wo=30; after HK was left in AOT reversed micelles

of (4) Wo=5 and (5) Wo=30 at 378C for 2 h, the Wo values were adjusted to Wo=10. The reaction

conditions were: [ATP]wp,o=10 mM, [DG]wp,o=30 mM, [HK]wp,=1.74 mg/ml, [MgCl2]wp=10 mM,

and 50 mM Tris/HCl buffer. (From Ref. 53.)


the HTAC reversed micelles compared with that in the AOT reversed micelles.Accordingly, such a maximum activity at Wo ¼ 10 seems to be determined by the locationof HK, a nonuniform distribution of substrates, bound water, and the sizes of the respec-tive water pools of AOT and HTAC. Figure 3(c) shows that the activity of HK in theC12E8 reversed micelles increases gradually with Wo up to Wo ¼ 20, reaching a nearlyconstant value above Wo ¼ 20. The enzyme reaction is expected to proceed favorably inthe sufficiently hydrated layers of the polyoxyethylene mantle.

The effects of the pH of the water pools in the AOT, HTAC, and C12E8 reversedmicellar solutions on the HK activities at optimal Wo are indicated in Fig. 5. In the case ofAOT [Fig. 5(a)], the optimum pH tends to shift in the alkaline direction with respect to theoptimum pH of 8.0 in the aqueous bulk solution. The alkaline shift is attributed to thelocation of the HK in the negatively charged AOT layer because H+ is absorbed by thenegatively charged AOT layer [72,88]. The optimum pH is not observed in the HTACreversed micelles, Fig. 5(b). If HK is located on the inner surface of the HTAC reversedmicelles, the optimal pH should shift to the acidic side because H+ should be repelled fromthe cationic HTAC layer. However, because such a phenomenon is not observed, HK isassumed to be relatively far from the interface of the cationic charged layer of the HTACreversed micelles. This profile does not conflict with the speculation described above thatHK is located to some extent far from the cationic surface layer due to a repulsiveelectrostatic force. Figure 5(c) shows the effect of pH on the activity in the C12E8 reversedmicelles of Wo ¼ 30, indicating that the optimum pH is the same as that in the aqueousbulk solution. This indicates that the proton distribution in the C12E8 reversed micelles isuniform, like that in the aqueous bulk solution.

Generally, the catalytic activity of the solubilized enzymes should not depend on thesurfactant concentration if Wo is kept constant, as demonstrated in the case of chymo-trypsin, etc. [87,89]. On the other hand, there is a group of enzymes such as peroxidase,etc., whose activity decreases with increases in surfactant concentration. The latterenzymes are characterized by the presence of anchoring groups capable of interactingwith micellar inner surfaces, which results in a dependence of catalytic activity on thesurfactant concentration [5,87]. HK activity increased with increasing concentrations ofC12E8 at Wo ¼ 30. This result supports the assumption that HK exists in the EO mantlewhich prevents HK from coming into contact with the alkyl layer or the organic bulksolvent. HK is assumed to be able to change easily its conformation in the EO mantlewhose thickness is much larger than that of the hydrophilic layer in the AOT or HTACreversed micelles, as shown in Fig. 5. Furthermore, the substrates, ATP and deoxyglucose,and Mg2+ may become uniformly distributed in the hydrated EO mantle due to thenoncharged field.

From the above results, it is assumed that the interaction of the hydrophobic andcationic moieties of HK with a negatively charged layer of AOT results in its insertion intothe AOT layer, which leads to the suppression of conformational change. On the otherhand, the HK conformation in the HTAC reversed micelles may be more easily changedbecause of the location of HK relatively far from the interface of the water pool due to theelectrostatic repulsion and the larger size of the micelles compared with that of the AOTreversed micelles. Consequently, the high electrostatic interface of AOT and HTAC is notfavorable for the appearance of HK catalytic activity.

The structure of C12E8 reversed micelles is greatly different from those of AOT andHTAC reversed micelles, as described in Section II. It is assumed that HK exists in thehydrated EO mantle, binds easily with the substrate, and changes its conformation toreveal catalytic activity. The HK activity in the C12E8 reversed micelles is of interest


because the micropolarity of the EO mantle is low compared with that of the aqueous bulkwater [89]. Based on these results, possible locations of HK, ADP, and Mg2+ ions inrespective reversed micelles are proposed (Scheme 6). The lipid biomembrane in vivo issurrounded by bridged adjacent polysaccharides with bound water molecules. This statemight to some extent be similar to the hydrated mantle of the C12E8 reversed micelles

FIG. 5 Effect of pH on HK activity in reversed micelles. (a) 0.1M AOT/iso-octane reversed micelles

(Wo=10) at 378C; [ATP]wp,o=10 mM, [DG]wp,o=30 mM, [HK]wp=1.74 mg/ml, [MgCl2]wp=10

mM, and 50 mM Tris/HCl buffer. (b) 0.1M HTAC/iso-octane–octanol reversed micelles

(Wo=10) at 378C; [ATP]wp,o=10 mM, [DG]wp,o=30 mM, [HK]wp=1.74 mg/ml, [MgCl2]wp=10

mM, and 50 mM Tris/HCl buffer (pH 8.0). (c) 0.1M C12E8/iso-octane–octanol reversed micelles

(Wo=30) at 378C; [ATP]wp,o=10 mM, [DG]wp=30 mM, [HK]wp=0.0174 mg/ml, [MgCl2]wp=10

mM, and 50 mM Tris/HCl buffer (pH 8.0). (From Ref. 53.)


because this mantle is formed by bridged adjacent EO chains with bound water. In con-clusion, the activity of the yeast HK can be also seen in the C12E8 reversed micelles withbound water and low polarity.

B. Photomerization of Cysteine by Ultraviolet Irradiation in ReversedMicellar System

Ozone depletion, which is probably related to chlorofluorocarbons emission, causes anincrease in ultraviolet (UV) radiation and significantly affects biological and ecologicalsystems [90–92]. The UV spectrum is divided into three groups: UVA (320–400 nm), UVB(280–320 nm), and UVC (200–280 nm). UVA is not absorbed by DNA, but induces theproduction of active oxygen to attack DNA and cell membranes, etc. [93,94]. UVB andUVC are directly absorbed by DNA to cause DNA damage. They also generate activeoxygens, which induce cell membrane damage as well as DNA damage [95,96]. One of theserious effects of UV on biological systems is induced through the biomembrane [97].

We focused on a system of UV-induced signal tranduction, which is considered tooriginate in the plasma membrane. Various receptor tyrosine kinases are reported to beactivated immediately following UV irradiation [98–101]. Knebel et al. [101] reported thatUVA, UVB, and UVC inhibited the receptor tyrosine dephosphorylation in a thiosensitiveand reversible manner. Furthermore, Aragane et al. [102] indicated that UV irradiationdirectly activates the apoptosis-related surface molecule CD95, independently of theligand CD95L, showing induction of clustering of CD95. It is well known that activeoxygens such as singlet oxygens [103–106], superoxide anions [107–109], and hydroxyradicals [104,105,110] are generated by UV irradiation. From these results, it was sug-gested that the S–S bond formed from SH groups by active oxygen, which is generated incell membranes by UV irradiation, participates in UV-induced signal transduction and theformation depends on the microenvironment near the cell membrane in living cells. It isdifficult to confirm this idea using living cells because they have many components in themembrane, which lead to more complex reactions. However, it is important to clarify thisphenomenon at the molecular level. The quenching of 1O2 by tryptophan residues inliposome systems was studied by Vilensky and Feitelson [111]. It was found that thelocation of the tryptophan residues in the liposome membrane or in the surrounding

SCHEME 6 Possible location of hexokinase, ATP, and Mg2+ ions in the reversedmicelles.


solution greatly affects the quenching rate constants of 1O2, but the mechanism has notbeen significantly well interpreted.

Therefore, we have investigated the effects of UV irradiation on the photomerizationof cysteine to cystine by encapsulation of cysteine in water pools of three kinds of reversedmicelles [112]. In this case, the reactant and the product, i.e., cysteine and cystine, are verysoluble in water, but not in organic solvents; the photomerization of cysteine proceeds inthe water pool or in the interface region. The photomerization was carried out in a waterpool of 50 mM Tris buffer of pH 9.0. The degree of the reaction was measured using afluorogenic reagent, N-(9-acridinyl)maleimide, which reacts with the SH group.

Figure 6 shows the amounts of cysteine in bulk water and in water pools of reversedmicelles after exposure to UV radiation. The degree of SH group reduction in water poolswas significantly different from that in aqueous bulk water. In aqueous bulk water, noreduction of cysteine was observed by UVA irradiation at 20 J/cm2, whereas UVB andUVC irradiation showed a reduction of about 20 and 60%, respectively. On the otherhand, cysteine in water pools of the three different kinds of reversed micellar solution(Wo ¼ 15) completely disappeared with UV irradiation at 20 J/cm2. It was confirmed bythin-layer chromatography that cysteine was mainly converted into cystine, indicatingoxidation by UV radiation.

The photomerization of cysteine was measured by changing Wo (Fig. 7). Even if Wo

increased, the concentration of cysteine was kept constant. Larger Wo showed a lowerreduction ratio of cysteine, regardless of the charges on surfactants and kinds of UVranges. However, the reaction fields, in terms of photomerization, are effective in theorder of HTAC A AOT > C12E8 in the overall Wo, showing the participation of theelectrostatic force.

Cysteine molecules exist in the water pools, but the molecular orientation ofcysteine in the interface of the water pool depends on the charge of the surfactants.The SH groups of cysteine are probably located on the cationic surface layer of HTACreversed micelles due to the electrostatic attraction. On the other hand, for AOTreversed micelles, the SH groups are oriented in the core of the water pool. In thecase of C12E8 reversed micelles, the cysteine molecules are distributed uniformly in the

FIG. 6 Photomerization of cysteine in reversed micellar solution and in aqueous bulk solution by

UV irradiation. Reversed micellar solutions and aqueous bulk solution containing 50 mM cysteine

was irradiated by UV in a sealed glass Petri dish. The amounts of cysteine were measured by the

reaction with N-(9-acridinyl)maleiimide.


water pool due to the noncharged surfactant. Photomerization occurs by the generationof active oxygen, such as singlet oxygen, due to irradiation with UV. The active oxygensare formed from dissolved oxygens. It is well known that the solubility of oxygen in anorganic solvent is one order of magnitude larger than in water [113], meaning thatoxygen is more readily dissolved in bulk organic solvent than in water pools. Thephotomerization might proceed favorably in the interface of the water pool becausemore active oxygen is located in the interface region in comparison with that in thecore of the water pool. Therefore, the photomerization would proceed favorably in theinterface region of the water pool formed by HTAC.

Next, we must consider the bound and free water in the water pool. The free waterappears at Wo ¼ 10 in the core of the water pool in both AOT and HTAC reversedmicelles. On the other hand, in the C12E8 reversed micelles, water molecules are hydratedto the oxygen of the EO chain and almost all the water molecules are bound up toWo ¼ 30. As Wo decreases, the photomerization proceeds favorably. It is suggested thatbound water is more important than free water for photomerization. As Wo decreases, thesize of the water pool also decreases to result in suppression of the mobility of the water.Therefore, the generated active oxygens come to be stable in the region of bound water,especially at low Wo.

In order to clarify these points, the effects of inhibitors of active oxygen, Trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid) (hydrophilic inhibitor) and �-tocopherol (hydrophobic inhibitor), were then examined (Fig. 8). The reduction ofcysteine was recovered to some extent by Trolox and significantly by �-tocopherol atWo ¼ 7. Since �-tocopherol is mainly present in the neighborhood of the interface ofreversed micelles and in organic solvents because of its hydrophobicity, this shows thatactive oxygens are more easily generated at the interface than in the core of the waterpools.

FIG. 7 Photomerization of cysteine in reversed micellar solutions with various Wo by UV

irradiation. Wo was adjusted by the addition of amounts of water to the surfactant solutions;

irradiation doses of 2 J/cm2 for UVA (&), UVB (B), and 0.5 J/cm2 for UVC (&) were used.


Furthermore, we examined using n-dodecanethiol, which has a longer hydrophobicalkyl chain than cysteine and may be distributed on the interface of the reversed micellesand in organic solvent (Fig. 9). Contrary to cysteine, larger Wo indicated a higher reduc-tion ratio of n-dodecanethiol. Because the n-dodecanethiol is minimally converted inmethanol or in iso-octane with UV irradiation, this photomerization needs a waterpool. In other words, as Wo increases, the thiol group comes into contact with thewater pool because the fluidity and/or flexibility of the AOT monolayer starts to increasewith increasingWo, resulting in dissociation of the thiol groups in Tris buffer of pH 9. Thisleads to the photomerization of n-dodecanethiol at the interface of the water pool. Theseresults demonstrated that photomerization of thiol compounds, regardless of differentchemical structures, proceeded in the interface region of the water pool.

From the above results, it is concluded that the bound water and electrostatic field inthe interfacial region are important for the active oxygens generated with UV irradiationand it is suggested that proteins, including the receptor at the interface of the cell mem-brane in vivo, are likely to be affected by UV irradiation.

IV. STUDY OF ENZYMIC POLYMERIZATION OF NUCLEOTIDES IN AREVERSED MICELLAR SYSTEM AS A DEVELOPMENT OFBIOPOLYMER SYNTHESIS UTILIZING THE LIQUID–SOLID INTERFACE

Reversed micellar enzymology has been extensively studied [114–116]. However, neither alarge-scale operation, nor the enzyme and product recoveries for its practical applicationhave been successfully carried out. The reversed micelle is a favorable reaction field for

FIG. 8 Effects of antioxidants on photomerization of cysteine in AOT reversed micellar solution by

UV irradiation: (a) Trolox; (b) �-tocopherol. (&) In the absence of antioxidant; (&) in the presence

of antioxidant.


enzymic polymerization when the solubility of substrates is poor in water. Furthermore,the reduced entropy of the substrate in the interface of the water pool is also important forpolymerization because it is generally considered that interface-bound molecules areentropically more favored to condense than those in the aqueous bulk solution owingto their reduced translational and rotational freedom [117]. Therefore, it is expected thatpolymerization in the bounded structures [118,119], such as the water pools of a reversedmicellar solution, proceeds successfully and selectively [117–121]. For example, a fluores-cent polymer of 2-naphthol was prepared using a peroxidase in AOT/iso-octane reversedmicelles [122]. Aniline and styrene were found to be easily polymerized in the reversedmicellar systems [123,124].

Oparin and coworkers [125,126] have studied the enzymic polymerization of ADP bypolynucleotide phosphorylase (PNPase) and Mg+2 ions in coacervates in an attempt toconstruct primitive forms of precellular structures. Walde et al. [127] have investigated thisenzymic ADP polymerization in AOT reversed micellar solutions instead of coacervates.The PNPase-catalyzed synthesis of poly(A) (polyadenylic acid) in the AOT reversedmicelles was carried out by mixing two reversed micellar solutions, one containing ADPand the other containing the enzyme.

This enzymic polymerization needed a high concentration of Mg2+ ions (10 mM inwater pools). The reaction usually proceeded at a low concentration of Mg2+ (1.56 mM)for the aqueous bulk solution. Polymerization occurred also in the HTAC reversedmicelles as well as in the C12E8 reversed micelles at a low concentration of the Mg2+

ions (1.56 mM) [85]. This can be interpreted in terms of the localization of Mg2+ and ADPin the water pool. Because the ADP and Mg2+ ions are located in the center of the waterpool in the AOT reversed micelles and in the interface of the water pool due to the

FIG. 9 Photomerization of n-dodecanethiol in AOT reversed micellar solution with various Wo by

UV irradiation; 500 mM 1-dodecanethiol was used in the place of cysteine. (&) UVA (2 J/cm2), (B)

UVB (2 J/cm2); and (&) UVC (0.5 J/cm2).


electrostatic force, respectively, this polymerization needs a higher concentration of theMg2+ ions compared with that in the HTAC and C12E8 reversed micellar solutions.

The poly(A) coprecipitated out of the AOT reversed micellar solution with thePNPase at Wo ¼ 20, while Banerjee et al. [121] have found a precipitation of phenolicpolymers formed by peroxidase in the AOT reversed micellar solution (we observed thatvery little peroxidase was included in the precipitate of the phenolic polymers).

It was found by HPLC that the poly(A) precipitated out at Wo ¼ 10; 20, and 30, ofwhich the highest polymerization proceeded at Wo = 20, but every supernatant includedhardly any poly(A). No precipitation occurred at Wo ¼ 40 [85]. The polymerization in theAOT reversed micelles seems to need an appropriate size of water pool, and the greater theWo and the closer it approaches an emulsion, the lesser the tendency to precipitate. Thismeans that the small amount of water in the reversed micellar system has an importantrole in the precipitation. No poly(A) precipitated out of the HTAC reversed micellarsolution, but a small amount of poly(A) did out of the C12E8 reversed micellar solution,[85]. The electrostatic repulsion between poly(A) and AOT must play an important role inthe precipitation.

Gel electrophoresis of poly(A) in the AOT reversed micellar system revealed that thesize of poly(A) in the precipitate (4.0 kb, 1.6 kb*–0.6 kb) was nearly the same as that inthe supernatant (2.5 kb–1.0 kb*, 0.4 kb), in which the asterisk denotes the highest dis-tribution of poly(A) [85]. A similar tendency was observed in the C12E8 reversed micelles.In the case of HTAC reversed micelles, the size of poly(A) (7.4 kb–5.3 kb*–1.0 kb) in thesupernatant was larger than that in the AOT reversed micelles. ADP molecules arestrongly adsorbed at the cationic charged interface of the water pool of HTAC, resultingin enhancement of the polymerization in the water pool due to the entropy effect.

Interestingly, as shown in Fig. 10, if the supernatant was removed from the test tubeand replaced by a fresh reversed micellar solution containing ADP but no PNPase, afterthe reaction equilibrium was nearly attained at 24 h, poly(A) continued to be produced

FIG. 10 Activities of PNPase on the interface of glass/0.2M AOT reversed micellar solution. After

the reaction (0.2 M AOT solution, [PNPase]w=1mg/ml, [ADP]w =10 mM, [MgCl2] w=10 mM,

Wo=20 [50 mM Tris–HCl, pH 9.5]) proceeded for a definite time, the supernatant was removed

and the precipitate was rinsed three times with AOT solutions containing buffer (Wo=20) without

ADP, and a fresh AOT solution containing ADP ([ADP]w=10 mM, [MgCl2] w=10 mM, Wo=20

[50 mM Tris–HCl, pH 9.5]) was added to the precipitate. The arrows show an addition of a fresh

AOT solution. (From Ref. 84.)


successively on the bottom of the glass test-tube [85,127]. Each repeated addition of a freshreversed micellar solution to the precipitates resulted in a new polymerization of ADP.The yield of �70% was maintained constantly for 4 weeks, and poly(A) was accumulatedsuccessively on the glass surface [85].

Figures 11(a) and 11(b) show the decrease in ADP concentration in the water poolsover time by the PNPases, which were included in the precipitate and in the supernatant,respectively, after the reaction proceeded at 258C. The polymerization occurred in both thePNPases in the supernatant and in the precipitate, but the amount of the precipitatedPNPase increased with the reaction time. It was confirmed that once the PNPase precipi-tates out of the micellar solution together with poly(A), it cannot be solubilized again in thewater pools. This shows that the precipitate of the PNPase and poly(A) on the glass surfacecorresponds to the functional aggregates self-organized on the interface of the oil/glass.

Ferris et al. [128] have observed the synthesis of the long prebiotic oligomers onmineral surfaces in an aqueous solution, and von Kiedrowski [117] has proposed that theearliest forms of life may have proliferated by spreading on solid surfaces. Our observa-tions also indicate that RNA-like molecules may be enzymically grown on an oil/solidinterface. A nanostructure of the functional aggregates was observed by AFM [86]. Figure12 shows the AFM images of the precipitated product after 2 h (a) and 12 h (b), respec-tively. Large particles with sizes of 200–500 nm and heights of 100 nm were observed.Such large particles must be covered thickly with AOT layers because they were observedafter rinsing off AOT. The AFM image of the PNPase molecules in the aqueous solutionshows several masses of which each unit corresponds to �20 nm, showing that thePNPases are much smaller than the precipitated particles. Figure 13 shows the AFMpictures of the supernatants. After 5 min, small particles were observed, and after 4 hthe size of the particles considerably increased, while after 12 h the number of the particlesdecreased. These observations suggest that the reaction progresses first in the water poolsof the reversed micelles, and the polymerized products precipitate after a relatively short

FIG. 11 PNPase activities in the precipitate (a) and in the supernatant (b) of 0.2 M AOT/isooctane

reversed micelles at 258C. After the polymerizations were performed in four different test-tubes for

2 h (&), 6 h (*), 12 h (~), and 24 h (^), the PNPase activities involved in the precipitate (a) and in

the supernatant (b) were measured. (From Ref. 85.)


time. The particles in Fig. 12 tended to be larger than those in the supernatant (Fig. 13).The results showed that the particles consist of not only the directly precipitated poly(A),but also the poly(A) further grown on the solid surface. These AFM images of the poly-merization process are in qualitative agreement with the results shown in Fig. 11.

Since the AOT micellar solution is very dynamic, the reactant monomers in the waterpools are successively and easily supplied to the enzymes in the aggregates on the glasssurface and the polymerization might be carried out in a manner akin to the solid-phase

FIG. 12 AFM images of poly(A)–PNPase complex precipitated from AOT reversed micelles after

2 h (a) and 12 h (b). (From Ref. 85.)

FIG. 13 AFM images of poly(A)–PNPase aggregates in the supernatants after 5 min (a), 4 h (b),

and 20 h (c). (From Ref. 85.)


synthesis of biopolymers [128–130]. When the precipitate was rinsed with aqueous ethanolsolution to remove the AOT layers wrapping the poly(A) and PNPase aggregates, almostall the activity of PNPase was lost. However, after the precipitate was rinsed with the AOTreversed micelle solution of Wo ¼ 20, the activity was maintained. This indicates that theAOT layers wrapping the products on the glass surface could play an important role inmaking the poly(A) particles grow on a solid surface. This model is schematically shown inScheme 7.

The polymerization in the AOT reversed micellar solution involves three processes.The first process is the polymerization in the water pools. The precipitation occurs as thesecond process and the solid polymerization proceeds as the third process. The precipita-tion of the product allows the simple isolation of the product in the reversed micellarsystem, leading to possibilities of large-scale applications. This precipitation in the AOTsolution must be related to the electrostatic repulsion between poly(A) and the AOTmonolayer at the interface of the water pool, resulting in the poly(A) and the enzymebeing repelled from the water pools because no precipitation occurrs in the HTAC solu-tion owing to the electrostatic attraction at the interface of the water pool. For solidpolymerization, the interaction of a surfactant with a glass surface is important. It wasobserved by AFM that the extent of the adsorption of AOT on the glass surface was small,in contrast with HTAC (unpublished data), because the charge on the glass surface isnegative. This is one of the factors for the strong adsorption of the poly(A)–PNPasecomplex on the glass surface.

Although a small amount of poly(A) precipitated out of the C12E8 reversed micellarsolution in the presence of Mg2+ ions, we have recently found precipitation of poly(A) inthe C12E8 reversed micelles above 378C when Fe3+ ions (0.05 mM) [131] were used instead

SCHEME 7 Schematic model of the nanostructures of functional aggregates self-organized at the interface between AOT reversed micellar solution and solid glass.(From Ref. 85.)


of Mg2+ ions (Table 2) [88]. This suggests that precipitation is related to changes in themicellar structure brought about by an increase in temperature. Furthermore, it wasconfirmed that the greater the glass area, the greater the precipitation. However, nosolid polymerization proceeded because the poly(A)–PNPase complex could not bestrongly bound to the glass surface. Accordingly, the extent of the interaction betweenthe surfactant and glass surface must play an important role in solid polymerization.

Surfactant–polymer interactions in an aqueous solution have been studied by manyresearchers [132], and the adsorption and surface-induced self-assembly of the surfactantat the solid–aqueous interface have been recently studied [133]. On the other hand, thesesubjects have been rarely studied for oil solutions. The surfactant–polymer interaction inoil and the surface-induced self-assembly of surfactants at the oil–solid interface areimportant for further research studies to enhance the polymerization at the interface ofthe liquid/solid in reversed micellar solutions.

V. CONCLUSION

The water pools surrounded by surfactant layers whose size is on a nanoscale are dispersedin organic solvents. Because their structures are very dynamic and the reversed micellarsolution is thermodynamically stable, it is easier to prepare their reaction mixtures thanthose of a liposome system.

We described that the water pool and the interface of the surfactant monolayerprovided a unique reaction field for enzymic reaction and photomerization. It is mostuseful for the reversed micellar system as a reaction field to be able to change the size andphysical properties of the water pool by changing Wo. At low Wo, the mobility of watermolecules is suppressed, resulting in a decrease in the fluidity and/or flexibility of thesurfactant monolayer and in a lowering of the micropolarity. At greater Wo, free waterappears in the core of the water pool, resulting in the opposite situation. In addition, theelectrostatic field at the interface of the water pool can be controlled by changing thehydrophilic group of the surfactant. Therefore, the water pools and the interfacial mono-

TABLE 2 Effects of Metal Ions and Temperature on ADP Polymerization in Reversed Micelles

Yield of Poly(A)(%)

378C 258C

Supernatant Precipitate Supernatant Precipitate

0 h 24 h 0 h 24 h 0 h 24 h 0 h 24 h

AOT Fe3þ 0:9� 0:0 0:6� 0:0 0:2� 0:0 0:3� 0:0 0:9� 0:2 1:4� 0:2 0:2� 0:1 17:4� 3:6Mg2þ 1:3� 0:2 5:7� 1:8 0:2� 0:1 6:3� 0:6 0:7� 0:1 1:9� 0:4 0:4� 0:1 55:8� 9:5

HTAC Fe3þ 14:9� 0:7 8:3� 0:6 0:6� 0:1 1:8� 0:3 10:9� 1:0 7:8� 0:8 0:4� 0:1 2:1� 0:5Mg2þ 14:5� 2:7 10:0� 1:9 0:9� 0:5 5:0� 1:3 12:2� 1:2 34:4� 1:4 0:9� 0:3 2:1� 0:7

C12E8 Fe3þ 1:9� 0:4 7:2� 0:8 0:6� 0:1 52:0� 12:5 1:1� 0:1 47:2� 7:7 0:2� 0:1 2:3� 0:6Mg2þ 1:0� 0:3 3:0� 0:9 0:3� 0:0 20:2� 0:4 1:5� 0:3 17:0� 1:1 0:6� 0:2 5:1� 0:1

The surfactant concentration was 0.2 mM and Wo was 20; [Fe2+]w=0.05 mM, [Mg3+]w=10 mM.


layers can exhibit multiple catalytic effects, which result from the localized concentrationsof the reactant and the physical properties of the water pools.

In this chapter, two subjects of our study were described. One was concerned withthe catalysis by enzymes entrapped in water pools and photomerization at the level of abiomembrane model in vivo. Based on the study of the activity of yeast HK in the waterpools, the activity of HK can be seen in noncharged polyoxyethylene mantles with rela-tively low micropolarity in which almost all the water molecules are bound up with EOchains. This suggests that yeast HK can work more actively in the vicinity of mitochon-drial membranes in vivo. The photomerization of cysteine in the water pool with UVirradiation shows that cysteine is easily converted into cystine with lower Wo. This sug-gests that active oxygen is generated at the interface of the biomembrane rather than inbulk aqueous solution in vivo and SH groups of proteins in the cell membrane are oxidizedsimilarly with UV irradiation.

Another subject is concerned with biopolymer synthesis utilizing the liquid/solidinterface in a reversed micellar system. The enzymic polymerization of ADP in AOTreversed micellar solution containing a Mg2+ ions resulted in the precipitation ofpoly(A) together with the PNPase. Further polymerization could proceed by the enzymein the precipitate by feeding ADP through the dynamic AOT monolayer on a glass sur-face. This is concluded to be a kind of solid polymerization in a reversed micellar solution.This process of polymerization provides: a simple isolation of both the product andenzyme; the maintenance of the enzyme activity for a long time; and a novel solid poly-merization on the oil/solid interface. This polymerization at the interfaces in the reversedmicellar solution could be applied to other biopolymer syntheses.

ACKNOWLEDGMENTS

A part of the study described in this chapter was supported by a grant from theCosmetology Research Foundation, and the authors are grateful to Mr. Y. Kuwahara,Mr. H. Hakamata, Mr. H. Suzumori, and Mr. H. Goto of this laboratory for theirassistance.

REFERENCES

1. PL Luisi, M Giomini, MP Pileni, BH Robinson. Biochim Biophys Acta 947:209–246, 1988.

2. MP Pileni, ed. Structure and Reactivity in Reverse Micelles. Amsterdam: Elsevier, 1989.

3. R Bru, A Sanchez-Ferrer, F Garcia-Cramona. Biochem J 310:721–739, 1995.

4. Y Miyake. Colloids Surfaces A: 109:255–262, 1996.

5. K Martinek, NL Klyachko, AV Kabanov, YL Khmelnitsky, AV Levashov. Biochim Biophys

Acta 981:161–172, 1989.

6. T Ono, M Goto. Curr Opin Colloid Interface Sci 2:397–401, 1997.

7. H Kise. J Jpn Oil Chem Soc 46:1447–1454, 1997.

8. JH Fendler. Membrane Mimetic Chemistry. New York: John Wiley, 1982.

9. J Sunamoto, K Iwamoto, M. Akutagawa, M Nagase, H Kondo. J Am Chem Soc 104:4904–

4907, 1982.

10. A Goto, H Kishimoto. Bull Chem Soc Jpn 62:2854–2861, 1989.

11. A Goto, H Kishimoto. J Chem Soc, Perkin Trans 1990:73–78.

12. A Goto, H Kishimoto. J Chem Soc, Perkin Trans 1990:891–896.

13. LS Romsted, CA Bunton, J Yao. Curr Opin Colloid Interface Sci 2:622–628, 1997.


14. JF Rathman. Curr Opin Colloid Interface Sci. 1:514–518, 1996.

15. PA Bachman, PL Luisi, J Lang. Nature 357:57–59, 1992.

16. H Hauser, G Hearing, A Pande, PL Luisi. J Phys Chem 93:7869–7876, 1989.

17. A Goto, H Yoshioka, H Kishimoto, T Fujita. Langmuir 8:441–445, 1992.

18. A Goto, S Harada, T Fijita, Y Miwa, H Yoshioka, H Kishimoto. Langmuir 9:86–89, 1993.

19. A Goto, H Yoshioka, M Manabe, R Goto. Langmuir 11:4873–4875, 1995.

20. A Llor, P Rigny. J Am Chem Soc 108:7533–7541, 1986.

21. A Yoshino, T Yoshida, K Takahashi, I Ueda. J. Colloid Interface Sci 133:390–394, 1989.

22. G Carstrom, B Halle. Langmuir 4:1346–1352, 1988.

23. MManabe, T Ito, H Kawamura, T Kinugasa, Y Sasaki. Bull Chem Soc Jpn 68:775–781, 1995.

24. D Brown, JHR Clarke, J Phys Chem 92:2881–2888, 1988.

25. F Heartley. J Chem, Faraday Trans I 83:517–526, 1987.

26. J Carnali, B Lindman, O Soderman, H Waldrhaug. Langmuir 2:51–56, 1986.

27. JN Israelachvili, DJ Mitchell, BW Ninham. J Chem Soc, Faraday Trans 2 72:1525–1568,

1976.

28. JN Israelachvili, DJ Mitchell, BW Ninham. Biochim Biophys Acta 470:185–201, 1977.

29. DJ Mitchell, BW Ninham. J Chem Soc, Faraday Trans 2 77:601–629, 1981.

30. M Hirai, R Kawai-Hirai, S Yabuki, T Takizawa, T. Hirai, K Kobayashi, Y Amemiya, M

Oya. J Phys Chem 99:6652–6660, 1995.

31. J Eastoe, DC Steytler, BH Robinson, RK Heenan, NA North, CJ Dore. J Chem Soc, Faraday

Trans 90:2497–2505, 1994.

32. E Bardez, NC Vy. Langmuir 11:3374–3381, 1995.

33. P Alexandridis, JF Holtzwarth, TA Hatton. Langmuir 9:2045–2052, 1993.

34. C Petit, JF Holzwarth, MP Pileni. Lagmuir 11:2405–2409, 1995.

35. LMM Nazario, JPSG Crespo, JF Holzwarth, TA Hatton. Langmuir 16:5892–5899, 2000.

36. LMM Nazario, TA Hatton, JPSG Crespo. Langmuir 12:6326–6335, 1996.

37. NE Levinger. Curr. Opin Interface Sci 5:118–124, 2000.

38. RE Riter, EP Undiks, NE Levinger. J Am Chem Soc 120:6062–6067, 1998.

39. D Pant, RE Riter, NE Levinger. J Chem Phys 109:9995–10003, 1998.

40. RE Riter, EP Undiks, NE Levinger. J Am Chem Soc 120:6062–6067, 1998.

41. RE Riter, DM Willand, NE Levingen. J Phys Chem B 102:2705–2714, 1998.

42. H Shirota, K Horie. J Phys Chem B 103:1437–1443, 1999.

43. J Faeder, BM Landanyi. J Phys Chem B 104:1033–1046, 2000.

44. A Goto, Y Kuwahara, A Suzuki, H Yoshioka, R Goto, T Iwamoto, T Imae. J Mol Liq

72:137–144, 1997.

45. C Petit, JF Holzwarth, MP Pileni. Langmuir 11:2405–2409, 1995.

46. S Ray, SR Bisal, SP Moulk. J Chem Soc, Faraday Trans 89:3277–3282, 1993.

47. JP Huruguen. M Authier, LJ Greffe, MP Pileni. Langmuir 7:243–249, 1991.

48. W Meier. Colloids Surfaces A: 94:111–114, 1995.

49. M Hasegawa, Y Yamazaki, N Sonta, Y Shindo, T Sugimura, A Kitahara. J Phys Chem

100:15575–15580, 1996.

50. J Bergenholtz, AA Romagnoli, NJ Wagner. Langmuir 11:1559–1570, 1995.

51. A Luzar, D Bratko. J Chem Phys 92:642–648, 1990.

52. G Mancini, C Schiavo. Langmuir 12:3567–3573, 1996.

53. Y Kuwahara, A Goto, Y Ibuki, Y Yamazaki, R Goto. J Colloid Interface Sci 233:190–196,

2001.

54. NM Correa, MA Bisutti, JJ Silber. J Colloid Interface Sci 184:570–578, 1996.

55. DM Zhu, ZA Schelly. Langmuir 8:48–50, 1992.

56. T Kawai, K Hamada, N Shinoda, K Kon-no. Bull Chem Soc Jpn 65:2715–2719, 1992.

57. M Wong, JK Thomas, T Gratzel. J Am Chem Soc 98: 2391–2397, 1976.

58. NM Correa, MA Biasutti, MA, JJ Silber. J Colloid Interface Sci 172:71–76, 1995.

59. BB Raju, SMB Costa. J Phys Chem B 103:4309–4317, 1997.

60. EA Lissi, D Engel. Langmuir 8:452–455, 1992.


61. U Kaatze, O Gottman, R Podbielski, R Pottel, U Terveer, J Phys Chem 82:112–120, 1978.

62. S Bisal, PK Bhattacharya, SP Moulik. J Phys Chem 94:4212–4216,1990.

63. DM Zhu, KI Feng, ZA Schelly, J Phys Chem 96:2382–2385, 1992.

64. SP Moulik, ML Das, PK Bhattacharya, AR Das. Langmuir 8:2135–2139, 1992.

65. H Yoshioka, S Kazama. J Colloid Interface Sci 95:240–246, 1983.

66. M Kotlarchk, JS Huang, SH Chen. J Phys Chem 89:4382–4386, 1985.

67. R Allen, S Banyopahyay, ML Klein. Langmuir 16:10547–10552, 2000.

68. V Papadimitriou, A Xenakis, P Lianos. Langmuir 9:912–915, 1993.

69. SN Nametkin, AV Kabanov, AV Levashov. Biochem Mol Biol Int 29:103–111, 1993.

70. P Walde. Curr Opin Colloid Interface Sci. 1:638–644, 1996.

71. CL Sodre, HM Scofano, H Barrabin. Biochim Biophys Acta 1419:55–63, 1999.

72. A Ohshima, H Narita, M Kito. J Biochem 93:1421–1425, 1983.

73. SS Tang, GG Chang. Biochem J 315:599–606, 1996.

74. H Barrabin, HM Scofano, MT Gomez-Puyou, A Gomez-Puyou. Eur J Biochem 213:757–763,

1993.

75. AA Burlakova, BI Kurgonov, NA Chebotareva, VG Debabov. Membr Cell Biol 10:543–551,

1997.

76. R Bru, A Sanchez-Ferrer, F Gacia-Carmona. Biochem J 310:721–739, 1995.

77. CM Anderson, FH Zucker, TA Steitz. Science 204:375–380, 1979.

78. E Alexander, AE Aleshin, HJ Fromm, RB Honzatko. FEBS Lett 434:42–46, 1998.

79. BD Gelb, V Adams, SN Johns, LD Griffin, GR MacGregor, ERB McCabe. Proc Natl Acad

Sci USA 89:202–206, 1992.

80. PG Polakis, JE Wilson. Arch Biochem Biophys 236:328–338, 1985.

81. JE Wilson. J Bioenerg Biomembr 29:97–102, 1997.

82. G Xie, JE Wilson. Arch Biochem Biophys 267:803–810, 1988.

83. K Nicholay, M Roio, T Wallimann, R Demel, R Hovius. Biochim Biophys Acta 1018:229–

233, 1990.

84. A Goto, K Hakamata, Y Kuwahara, R Goto, P Walde, PL Luisi, T Imae. Prog Colloid Polym

Sci 106:245–248, 1997.

85. A Goto, R Goto, Y Kuwahara, H Hakamata, Y Ibuki, P Walde, PL Luisi, T Imae. Langmuir

14 : 3454–3457, 1998.

86. A Goto, H Suzumori, Y Ibuki, K Yamazaki, R Goto: I Noda, E Kokufuta, eds. In

Polyelectrolytes. Osaka Yamada Science Foundation, 1999, pp 17–20.

87. YL Khmelnisky, AV Kabanov, NL Klyacho, AV Levashov, K Martinek. In: MP Pileni, eds.,

Structure and Reactivity in Reversed Micelles. Amsterdam: Elsevier, 1989, pp 230–261.

88. E Ruckenstein, EP Karpe. J Colloid Interface Sci 139:408–436, 1990.

89. D Mandal, A Datta, SK Pal, K Bhattacharyya. J Phys Chem B 102:9070–9073, 1998.

90. T Henriksen, A Dahlback, S Larsen, J Moan. Photochem Photobiol 51:579–582, 1990.

91. S Madronich, FR de Gruijl. Nature 366:23, 1993.

92. H Slaper, GJM Velders, JS Daniel, FR de Gruijl, JC van der Leun. Nature 384:256–258, 1996.

93. K Ito, S Kawanishi. Biol Chem 378:1307–1312, 1997.

94. C Kielbassa, L Roza, B Epe. Carcinogenesis 18:811–816, 1997.

95. HN Ananthawamy, WE Pierceall. Photochem Photobiol 52:1119–1136, 1990.

96. X Zhang, BS Rosenstein, Y Wang, M Lebwohl, H Wei. Free Radic Biol Med 23:980–985,

1997.

97. K Bender, C Blatter, A Knebel, M Iordanov, P Herrlich, HJ Rahmsdorf, J Photochem

Photobiol B 37:1–17, 1997.

98. CC Miller, P Hale, AP Pentland. J Biol Chem 269:3529–3533, 1994.

99. C Sachsenmaler, A Radler-Pohl, R Zinck, A Nordheim, P Herrlich, HJ Rahmsdorf. Cell

78:963–972, 1994.

100. I Warmuth, Y Harth, MS Matsui, N Wang, VA Deleo. Cancer Res 54:374–376, 1994.

101. A Knebel, HJ Rahmsdorf, A Ullrich, P Herrlich. EMBO J 15:5314–5325, 1996.


102. Y Aragane, D Kulms, D Metze, G Wilkes, B Poeppelmann, TA Luger, T Schwarz. J Cell Biol

140:171–182, 1998.

103. RM Tyrrell, M Pidoux. Photochem Photobiol 49:407–412, 1989.

104. S Basu-Modak, RM Tyrrell. Cancer Res 51:4505–4510, 1993.

105. GF Vile, RM Tyrrell. J Biol Chem 268:14678–14681, 1993.

106. M Berneburg, S Grether-Beck, V Kuetent, T Ruzicka, K Briviba, H Sies, J Krutmann. J Biol

Chem 274:15345–15349, 1999.

107. B Czochralska, W Kawczynski, G Bartosz, D Shugar. Biochim Biophys Acta 801:403–409,

1984.

108. ML Cunningham, JS Johnson, SM Giovanazzi, MJ Peak. Photochem Photobiol 42:125–128,

1985.

109. T Horio, H Okamoto. J Invest Dermatol 88:699–702, 1987.

110. H Masaki, T Atsumi, H Sakurai. Biochem Biophys Res Commun 206:474–479, 1995.

111. A Vilensky, J Feitelson. Photochem Photobiol. 70:841–846, 1999.

112. R Goto, H Goto, Y Ibuki, A Goto. Proceedings of 13th International Congress on

Photobiology, San-Francisco, CA, July 2000, Abstracts, p 28.

113. Landolt–Boernstein Tables. Berlin: Springer vol. II/2b. pp 1–121, 1960.

114. R Bru, P Walde. Biochem Mol Biol Int 31:685–692, 1993.

115. P Walde, D Han, PL Luisi. Biochemistry 32:4029–4034, 1993.

116. EP Melo, MR Alres-Barros, JM Cabral. Biotechnol Annu Rev 7:87–129, 2001.

117. G von Kiedrowski. Nature 381:20–21, 1996.

118. PA Bachman, P Walde, PL Luisi, J Lang. J Am Chem Soc 112:8200–8201, 1990.

119. PA Bachman, P Walde, PL Luisi, J Lang. J Am Chem Soc 113:8204–8209, 1991.

120. AM Rao, VT John, RD Gonzalez, JA Akkara, DL Kaplan. Biotech Bioeng 41:531–540, 1993.

121. S Banerjee, R Premchandran, M Tata, VT John, GL McPherson, J Akkara, D Kaplan. Ind

Eng Chem Res 35:3100–3107, 1996.

122. RS Premachandran, S Banerjee, XK Wu, VT John, GL McPherson. Macromolecules

29:6452–6460, 1996.

123. LM Gan, CH Chew, SO Hardy, L Ma. Polym Bull 31:347–350, 1993.

124. LM Gan, CH Chew, I Le, L Ma, G Li. Polymer 34:3860–3864, 1993.

125. AI Oparin, KB Serebroskaya, TL Auerman. Biokhimiya 26:499–547, 1961.

126. AI Oparin, KB Serebroskaya, SN Pantskhava, NV Vasil’eva. Biokhimiya 28:671–675, 1963.

127. P Walde, A Goto, PA Monnard, M Wessicken, PL Luisi. J Am Chem Soc 116:7541–7547,

1994.

128. JP Ferris, AR Hill, R Liu, LE Orgel. Nature 381:59–61, 1996.

129. JP Ferris, G Ertem. J Am Chem Soc 115:12270–12275, 1993.

130. K Kawamura, JP Ferris. J Am Chem Soc 116:7564–7572, 1994.

131. M Beljanski. J Mol Evol 42:493–499, 1996.

132. P Hansson, B Lindman. Curr Opin Colloid Interface Sci 1:604–613, 1996.

133. F Tiberg, J Brinck, L Grant. Curr Opin Colloid Interface Sci 4:411–419, 2000.


15Supported Planar BLMs (Lipid Bilayers):Formation, Methods of Study, andApplications

ANGELICA L. OTTOVA and H. TI TIEN Michigan State University, EastLansing, Michigan, U.S.A.

I. BACKGROUND

Those researchers whose work have no direct connection with biomembranes perhaps arenot acquainted with experimental lipid bilayers, commonly referred to as planar bilayerlipid membranes (BLMs) and spherical liposomes. These artificial systems have beenextensively used in the past four decades as experimental models of cell membranes.The work began with D. O. Rudin and his associates in 1959 [1]. They first investigatedlipid monolayers and multilayers of the Langmuir–Blodgett (L–B) type, and then theytoyed with soap bubbles and films. It was realized that the structure of a soap film in air, inits final stage of thinning, has a structure that may be depicted as two lipid monolayerssandwiching an aqueous solution. That is a system which may be represented as:

air | aqueous soap film | air

Once they recognized this structure together with its molecular organization, Rudin andco-workers simply proceeded to make an underwater ‘‘lipid film’’ separating two aqueoussolutions, i.e.,

aqueous solution | lipid bilayer | aqueous solution

Experimentally, it is far easier to form a BLM than spreading a monolayer at an air/water interface [2]. Table 1 presents a historical overview of topics under discussion. Toimpart relevant functions and specificity to BLMs, a variety of compounds such as iono-phores, enzymes, receptors, photoactive dyes and pigments, fullerenes, etc., have beenincorporated. Further, optical and electrical methods are readily applicable to such aplanar BLM system. For long-term studies and for practical applications, however, theconventional BLM system has one major limitation; it is not very durable. This extremefragility of the BLM has now been overcome. Simple methods are available for theformation of self-assembled, stable BLMs on either freshly cleaved metallic surfaces(s-BLMs) or freshly cut agar gel enclosed in Teflon tubing (i.e., a salt bridge; hence,sb-BLMs). Applications of these supported lipid bilayers can be envisaged, rangingfrom biosensors to molecular electronic devices. Finally, it should be stressed that such


applications will require an interdisciplinary approach involving inputs from biophysics,colloid and surface science, and microelectronics.

II. INTRODUCTION

A BLM is a two-layered arrangement of phosphate and lipid molecules that form a cellmembrane, in which the hydrophobic lipid portions face inward and the hydrophilicphosphate moieties face outward with one side extracellular fluid, and the other cyto-plasm. It is also called a planar lipid bilayer, or simply lipid bilayer. In brief, the lipidbilayer postulated as the basic structural matrix of biological membranes is widelyaccepted. At present, the BLM (or planar lipid bilayer) together with spherical lipidbilayers (liposomes), on suitable modification, serves as a most appropriate model forbiomembranes. In recent years, advances in microelectronics and interest in ultrathinorganic films, including BLMs and L–B films, have resulted in a unique fusion of ideastoward the development of biosensors and transducers. Furthermore, recent trends ininterdisciplinary studies in chemistry, electronics, and biology have led to a new field ofresearch: biomolecular electronics. This exciting new field of scientific–technologicalendeavor is part of a more general approach toward the development of a new,postsemiconductor electronic technology, namely, molecular electronics with a long-term goal of molecular computers.

It has been demonstrated recently [4] that BLMs, after suitable modification, canfunction as electrodes and exhibit nonlinear electronic properties. These and other experi-mental findings relevant to sensor development and to ‘‘biomolecular electronic devices’’will be described in more detail in the present chapter. Also, the potential use of the BLMsystem together with its modifications in the development of a new class of organic diodes,switches, biosensors, electrochemical photocells, and biofuel cells will be discussed.Additionally, this chapter reports also a novel technique for obtaining BLMs (or lipidbilayers) on solid supports. The presence of a solid support on one side of the BLM greatly

TABLE 1 Self-Assembling Systems Containing Amphiphilic Molecules [2]

System Literature source

1. Soap films R Hooke. The History of the Royal Society of London,

vol. 3, 1672, p 29

2. Monolayers I. Langmuir. J. Am. Chem. Soc., 39:1848, 1917

3. Langmuir-Blodgett (L-B) multilayers KB Blodgett, I. Langmuir. Phys Rev 51: 964, 1937

4. Planar lipid bilayers (BLMs) P Mueller, DO Rudin, HT Tien, WC Wescott. Nature

194:979, 1962

Liposomes (lipid microvesicles) AD Bangham. BioEssays 17:1081, 1995

5. Nucleopore-supported BLMs JD Mountz, HT Tien. Photochem Photobiol

28:395–400, 1978

6. Gold-supported monolayers RG Nuzzo, DL Allara. J Am Chem Soc 105:4481,

1983

7. Metal-supported BLMs (s-BLMs) [4,5]

8. Salt-bridge supported BLMs

(sb-BLMs)

(a) Yuan et al. [28] (b) Lu et al., [29]


enhances its mechanical stability, while retaining the dynamic properties of the lipidbilayer. Advantages of the new technique for self-assembling amphiphilic molecules onrigid substrates are discussed in terms of their possible uses. It is evident that the new BLMsystem (s-BLMs) is potentially useful for technological applications in the area of biosen-sors and enzyme electrodes as well as molecular electronics.

For decades, colloid and surface scientists have known that amphiphilic moleculessuch as phospholipids can self-assemble or self-organize themselves into supramolecularstructures of bilayer lipid membranes (planar BLMs and spherical liposomes), emul-sions, and micelles [2–4]. As a matter of fact, our current understanding of the structureand function of biomembranes can be traced to the studies of these experimental systemssuch as soap films and Langmuir monolayers, which have evolved as a direct con-sequence of applications of classical principles of colloid and interfacial chemistry. Asalready mentioned in Section I, the seminal work on the self-assembly of planar lipidbilayers and bilayer or ‘‘black’’ lipid membranes was carried out in 1959–1963. The ideastarted while one of the authors was reading a paperback edition of Soap Bubbles by C.V. Boys. These early researchers realized that a soap film in air in its final stage ofthinning has a structure, which may be depicted as two monolayers sandwiching anaqueous surfactant solution. The picture of the so-called ‘‘black’’ soap films had beensuggested many years ago by Gibbs, Overbeek, Mysels, Corkill, and others (see Ref. 2and references cited therein). Rudin and coworkers showed that an underwater ‘‘soapfilm’’ or a BLM formed from brain extracts was self-sealing to puncture with manyphysical and chemical properties similar to those of biomembranes [3,4]. On modifica-tion with a certain protein, this otherwise electrically ‘‘inert’’ structure of about 6 nmthick became excitable displaying characteristic features similar to those of actionpotentials of the nerve membrane. Thus, in the four plus decades since its inception,the conventional BLM along with the liposome has been extensively used as a model ofbiomembranes [3,4]. In particular, the BLMs have been adopted to elucidate themolecular mechanisms of biomembrane functions such as ion sensing, materialtransport, electrical excitability, gated channels, antigen–antibody binding, signal trans-duction, and energy conversion, to name a few. We will digress for a moment in thefollowing paragraphs to describe biomembranes of the cell.

The basic unit of all living organisms is the cell. Each cell is bounded on the outsideby a very thin, delicate membrane of about 5–7 nm thick. For example, the plasmamembrane is an integral part of the cytoplasm. Different kinds of membrane modelshave been proposed [3,4], and the most useful one is the bilayer leaflet model proposedby Gorter and Grendel. In this model the membrane is seen as a bilayer of phospholipidsin which protein molecules are embedded. A typical example of phospholipids is phos-phatidylcholine (PC or lecithin), making up most of the cell membrane. The PC moleculeconsists of two moieties. The phosphate + choline ‘‘head’’ is polarized, and, although themolecule and this polar region are both electrically neutral, the distribution on charges isasymmetrical. Such polarization leads to solubility in water. The other moiety of the PCmolecule is two fatty acid chains. Usually, one of these chains will be unsaturated, andthere may be multiple double bonds. There is considerable variation in the precise com-position of these chains, which are similar in size and form, and they are nonpolar. Thismeans they are attractive to each other, and the nonpolar region of the phospholipid issoluble in organic solvents of low dielectric constant, but not in water.

Many different phospholipids form plasma and organelle membranes in many speciesand many cell types. Thus, the fundamental structure of each cell is bound by a limitingplasma membrane, which is a protein-carbohydrate modified lipid bilayer. This is so


because of the unique property of the lipid molecule: The remarkable stability of a lipidbilayer is due to the combination of hydrophilic and hydrophobic forces, which makes thelipid bilayer a thermodynamically favored structure, thereby establishing the lipid bilayerprinciple of biomembranes [2,4]. This lipid bilayer is the basic structure of the unit mem-brane surrounding cells and composing the many intracellular organelles. The living cellsundertake intercellular communication, which take place across as well as between cellmembranes. This usually entails the sending, receiving, and decoding of signals—so-calledbiosensitivity. These signals are generally electrical and chemical in nature, predicated onthe presence of a membrane. Specifically, embedded in the lipid bilayer of cell membranesare ion channels (Naþ, Kþ, intracellular Ca2+, etc.) and receptors (growth factors, mgluR/serpintine, immune, cytokine, etc.); most of these are involved in signal transduction, and afew of which have been incorporated into the lipid bilayer [2,4].

From the viewpoint of membrane biophysics and physiology, biological membranesare essentially the basic structure of Nature’s sensors and devices, and the cell membraneplays a crucial role in signal transduction, energy conversion, and information processing.This is because most physiological activities involve some kind of lipid bilayer-basedligand–receptor contact interactions. Outstanding examples among these are ion sensing,antigen–antibody binding, light conversion and detection, and gated channels, to name afew. For example, the thylakoid membrane of green plants functions as an energy trans-ducer converting sunlight into electrical/chemical energy, the photoreceptor membrane ofa rod’s outer segment detects photons as the initial step in visual perception, and theplasma membrane of cells and organelles possesses the ability for ion sensing, for instance,differentiating between Naþ and Kþ with great specificity. Further, the plasma membraneprovides sites for a host of ligand–receptor contact interactions such as antigen–antibodyformation. As a result of these extensive studies, biomembranes have now been recognizedas the basic structure of Nature’s sensors and molecular devices. This is summarized inTable 2.

As already mentioned, the generally accepted bimolecular leaflet model of theplasma membrane is that first proposed by Gorter and Grendel in 1925, which has domi-nated our thinking ever since [2]. Until then, our knowledge or the properties of bimole-cular lipid layers was derived entirely from indirect experimental evidence. In 1961, thereconstitution of membranous structures from lipids of bovine brain was finally achieved[1–4]. These reconstituted membranes not only had a thickness ranging from 6 to 9 nm,

TABLE 2 Basic Structure of Nature’s Sensors and Devices [4]

Examples Transduction

Nerve membrane � ��! electric

Photosynthetic membranes h�! electric, chemical

Mitochondrial membranes Chemical (foodstuff) ! chemical (ATP)

Visual receptor membranes h�! electrical

Plasma membranes ��! mechanical

fire-fly Chemical ! light

electric eel ��! electrical

Note: �� and � �� denote, respectively, chemical potential gradient and

electrochemical potential gradient.


but also electrical properties resembling those of biomembranes (capacitance Cm ¼ 1 �F/cm2, resistance Rm >108 ohm cm2, and dielectric breakdown strength >250,000 V/cm). Ithas been 40 years since the first report of bimolecular membranes or BLMs. In retrospect,it is informative to mention in this section the crucial role played by the science of inter-faces in the reconstitution experiments [2].

In living cells, the tremendous interfacial areas that exist between the membraneand its surroundings not only provide ample loci for carrying out activities vital to theliving system, but also afford a clue for our understanding. Physically, its interfacial freeenergy or tension characterizes an interface uniquely, which is a result of the orientationof the constituent molecules. A thin film is a system whose interior is influenced by theproximity of its interfaces. It is of microscopic thickness in one dimension and is macro-scopic in its two other dimensions. In a sense, an interface can be thought of as a film soultrathin that it has no homogeneous interior. However, the kind of ultrathin films (e.g.,BLMs) under discussion here are heterogeneous from their contacting aqueous phases.This and other interfacial properties of membranes can be understood to a large extentin terms of the laws of interface chemistry and physics that govern them. In fact, thecurrent concept of the structure of biological membranes and their experimental modelshas been developed as a direct consequence of the applications of classical principlesadvanced by Langmuir, Adam, Harkins, McBain, Hartley, and others (for a recentreview, see Refs 2 and 3).

In the recent past, there have been a number of reports on self-assemblies of mole-cules as ‘advanced materials’ or ‘smart materials’. Without question, the inspiration forthis exciting work comes from the biological world, where, e.g., the lipid bilayer of cellmembranes plays a pivotal role. In this connection it should be stated that many otherresearchers have also described self-assembling systems such as the liposome. Liposomesare modeled after biomembranes, which have been extensively investigated since the late1960s (see Table 1 for references).

Since the mid-1960s, our work has been motivated by the desire to explain the livingsystem in physical, chemical, and physiological terms. The area chosen for investigation ismembrane biophysics, where the cell membrane plays a crucial role in signal transduction,energy conversion, and information processing [4]. Other functions of the plasma mem-brane include control of cell shape, cell movement and cell–cell interaction, diffusion,osmosis, and transport. Additionally, regulation of cell growth and the cell cycle, devel-opment, and differentiation—hormones and receptors, second messenger systems, theimmune system, synthesis and intracellular trafficking of proteins and glycoproteins andpost-translational processing of secretory and membrane proteins are involved.Furthermore, bioenergetics, evolution, and the role in diseases such as cancer, apoptosis,and aging are also crucial functions are implicated with the cell membrane [4]. Our presentapproach to the study of these interactions in vitro is facilitated by employing self-assembled BLMs of 5 nm in thickness. Our efforts have been focused on ion and/ormolecular selectivity and specificity using newly available BLMs on solid supports (i.e.,s-BLMs), whose enhanced stability greatly aids in research areas of membrane biophysics,biochemistry, and molecular cell biology as well as in biosensor designs and molecular-device development. It should be noted that the dimensions of a typical s-BLM underinvestigation is about 2 x 10-5 cm2 and 5 nm thick. As a result of the efforts of manyinvestigators, biologically relevant phenomena such as ion selectivity, excitability, anti-body–antigen reactions, active ion transport, and photoelectric effects have all beendemonstrated. This new type of s-BLM-based probe is intended for biotechnology usingmicroelectronic fabrication techniques [4–10].


In recent years, numerous attempts have been made to exploit the BLM system’spotential in practical applications [7–10]. Advances in microelectronics and interest inultrathin organic films, including BLMs, especially, and the newly developed self-assembled bilayer lipid membrane (s-BLM) on a nascent metallic surface, have resultedin a unique fusion of ideas to forward the development of intelligent biosensors andtransducers. Furthermore, recent trends in interdisciplinary studies in chemistry, electro-nics, and biology have led to a new field of scientific–technological endeavor that is part ofa more general approach toward the development of a new, postsemiconductor electronictechnology, namely, molecular electronics with a long-term goal of molecular computers.In this chapter, we will review some of the highlights of research since 1987 [4,11–15] onplanar lipid bilayers: from using BLMs as an experimental model of biological membranesto artificial photosynthesis, and to supported BLMs as biosensors. Before doing so, men-tion should be made concerning the closely related soap films, which, as already alluded to,provided the necessary insight for forming BLMs in 1960. Outstanding research on ultra-thin fluid films including soap films and foams, have been carried out by many investiga-tors (for citations, see Refs 2 and 4).

III. SELF-ASSEMBLED PLANAR LIPID BILAYERS

A. Techniques of BLMs and Supported BLMs

The history of the BLM system and its development as a model for biomembranes hasbeen recounted elsewhere [1–4]. It should be mentioned, however, that there are majordifferences between L–B layers on a rigid substrate and the BLMs. Evidence stronglysuggests that:

. A BLM is a nanostructure in one dimension with two junctions.

. A BLM is in a liquid-crystalline state without pinholes.

. A BLM is the site for embedding receptors, channels, photoabsorbers, etc.

Apart from its biomolecular thickness, a BLM is a liquid-like, dynamic structure in ametastable state from a self-assembling point of view; we conclude that it is difficult toenvision how a BLM separating two aqueous solutions can be made from two rigid, solid-like monolayers of lipids without a Plateau–Gibbs border. For biosensor development, itis our opinion that a fluid bilayer is of crucial importance. The aim of this section is todescribe in sufficient detail how to set up a simple BLM system using the self-assemblingtechniques [5–10].

1. Conventional BLMs

Since the pioneering work on bilayer (black) lipid membranes (BLMs) in the early 1960s,several generations of membrane biophysicists and bioscientists have exploited the con-ventional BLM system for biophysical and reconstitution studies [2–4]. A conventionalBLM separating two aqueous solutions is usually formed by spreading a lipid solutionover the hole, which is followed by spontaneously thinning to a BLM [3,5]. Formed in thisfashion, changes in trans-BLM properties such as electrical potential and current can bereadily measured by reference electrodes on either side of the membrane. Other methods ofBLM formation have been developed over the years. A host of techniques for BLM studiesare available in monographs and recent reviews [3,4,11–13,16]. Experimentally, all thetechniques used to form a conventional BLM needs a small hole connecting two aqueous


chambers. In this technique a drop of lipid solution (e.g., a 2% PC solution) in n-decane ora 2% glycerol dioleate solution in squalene) is spread over the hole. As the solvent drainsaway, a BLM is formed (Fig. 1). To date, the original method is still one of the simplesttechniques available. Figure 1 also shows a basic arrangement used for c-BLM and sup-ported BLM experiments.

A conventional BLM can also be formed by a modified L–B technique. It usesessentially a lipid monolayer spread from lipid in a volatile solvent (e.g., n-hexane) oneach aqueous chamber with the aqueous solution below the hole. Raising the aqueoussolution level on both sides above the hole results in the two monolayers combining into aBLM in the aperture, thereby forming a so-called ‘‘solvent-free’’ BLM.It should be men-tioned, however, that there is a major difference between the BLM and multilayers formedby the L–B technique. A BLM, formed either by the conventional ‘‘painting’’ method orself-assembling on substrates (e.g., freshly cleaved metallic wire and agar gel—see latersections) is a dynamic liquid-like structure which is capable of accommodating a host ofmodifiers such as polypeptides, proteins, oil-soluble compounds, etc. In contrast, a L–Bmultilayer of bimolecular thickness, albeit more stable than a BLM, usually contains pin-holes and is in a solid state [16,17].

Into these BLMs or planar lipid bilayers, many compounds have been embedded: (1)polypeptides and channel proteins, (2) pigments, electron acceptors, donors, and media-tors including many highly conjugated compounds such as meso-tetraphenylporphyrins(TPP), metallo-phthalocyanines (PLC), TCNQ (7,70,8,80-tetracyano-p-quinodimethane),TTF (tetrathiafulvalene), and fullerenes (C60, C70, etc. [16]), (3) redox proteins and metal-loproteins such as cytochrome c, and iron–sulphur proteins (ferrodoxins and thioredox-ins); and (4) substances partaking in ligand–receptor contact interactions. Specifically, theligand may be a substrate, an antigen, a hormone, an ion, or an electron acceptor ordonor, and the corresponding receptor embedded in the BLM may be an enzyme, anantibody, a protein complex, a carrier, a channel, or a redox species, and fine semicon-ductor particles (formed in situ) such as CdS, CdSe, and AgCl [5].

2. Conventional BLMs as Models of Biomembranes

To characterize a BLM, the techniques of electrochemistry have been used. Conventionalelectrochemistry is mainly concerned with metallic electrodes in electrolyte solutions,although classic membrane electrochemistry came into being with precipitated coppercyanide membranes. As such, membranes were treated as a physical barrier with or with-out fixed charges as in cellophanes, ion-exchange membranes, lipid-soaked filter papers,etc. The electrical properties of such membranes, often used as a model for biomembranes,had been studied by placing a pair of reference electrodes in the aqueous solutions bathingthe membrane. The above-mentioned membranes were quite thick (tens of micrometers ormore) in comparison with biomembranes whose thickness is of the order of 5–7 nm.Further, all biomembranes are composed of a lipid bilayer intercalated with other con-stituents such as proteins, carbohydrates, and their complexes of lipids. Insofar as can bedetermined, biomembranes are liquid-like and in a dynamic state. Thus, it is not surprisingthat, today, BLMs and liposomes are the most used model systems for biomembranestudies [5,17]. In this connection, Lee et al. [18] have reported recently the use of differ-ential confocal microscopy to detect the phase transition of BLMs of spherical configura-tion. From the deformation the authors have obtained the bending rigidity of membranesby simple geometric analysis. The bending modulus changes by an order of magnitude asthe temperature is changed across the transition temperature, Also, from the linear rela-


FIG. 1 Experimental arrangements for investigating the electrical properties of bilayer lipid

membranes (c-BLMs, s-BLMs, sb-BLMs, t-BLMs). (a) A setup for a conventional c-BLM

separating two aqueous solutions (note: the BLM may be illuminated by a focused light). (b)

Left: a setup for investigating salt bridge (sb) supported BLMs; Right: a setup for studying sb-

BLMs using a three-electrode system. (c) Illustrating the ligand (L) and receptor (R) interaction,

where receptor is shown to be embedded in a metal- or sb-supported BLM, whereas the ligand is in

the adjacent solution.


tion between the mean-square amplitudes of fluctuations and temperatures, the phase-transition temperatures of lipid bilayers can be determined. Concerning liposomalbilayers, an insightful paper by Seitz et al. should be consulted [17].

Electrically speaking, the ultrathin BLMs and biomembranes possess very largecapacitance (� 1 �F cm�2) and dielectric breakdown strength (> 200 kV/cm). These,along with other unique attributes, are listed in Table 3.

(a) Modified BLMs for the Olfactory System. Scents and odors evoke powerfulthoughts and emotions in human beings. However, how the sensing organ (nose) andbrain recognize molecules that affect our world view has been little understood, except-ing that odorant molecules first interact with the olfactory bulb from which neurons areconnected to the cortex of the brain. The average human being is said to be able torecognize some 10,000 different odors. Yet, it is almost impossible to describe howsomething smells to someone who has not experienced it. Up to about 1991, little wasknown about the biochemical process involved in detecting odors. Now it appears thatthe odorant receptors are quite similar to visual rhodopsin (see Section III.A.3(b))which consists of G-proteins that criss-cross seven times the lipid bilayer of the plasmamembrane. Upon odorant excitation, it seems likely that some sort of conformationtake place. This change in the odorant receptor molecule triggers a biochemical cascade,leading most likely to the generation of electrical signals to the brain. To find out howthe olfactory system works at the membrane level, we studied the response of the bull-frog’s olfactory receptors, reconstituted into BLMs, using three structurally relatedodorants: diethyl sulfide (DES), thiophene, and diethanol sulfide (thiogylcol or DOS).DOS is the dihydroxy derivative of DES. The details of this research are available [2,4].

(b) Model for the Nerve Membrane. The concept of electrical activity in biomem-branes dates back to more than two centuries to the time of Galvani and Volta andtheir debate on the so-called ‘‘animal electricity.’’ Today, it is widely accepted by bio-physicists and electrophysiologists that ionic channels are the crucial elements in themembranes of nerves and muscles of animals as well as in the other cells such as photo-receptors, plant cells, auditory hair cells, other sensory cells, epithelial cells, and in uni-cellular organisms. In fact, ionic channels are a unique structure present in all livingsystems and exhibit experimentally measured potential differences (p.d.) across the bio-

TABLE 3 Unique Aspects of BLMs and Biomembranes [3]

Property BLMs Biomembranes

Thickness (nm) 4–7 5-7

Resistance ( cm2) > 108 (unmodified) 103–105

Capacitance (�F cm�2 ) 0.3–1.3 � 1

Breakdown voltage (V cm�1) 2–10 � 105 105–106

Dielectric constant 2.1–5 2–3

Water permeability (�M/s�1) 8–24 35

Interfacial tension (dynes/cm�1) 0.2–6 < 1

Potential difference per 10-fold

concentration (e.g., KCl) (mV)

> 50 (modified) > 50

Electrical excitability Observed (modified) Observed

Photoelectric effects Observed (modified) Observed

Electronic (nonlinear) Observed (modified) Observed


membrane. This experimentally observed p.d. or transmembrane voltage is bestdescribed by the Nernst–Planck equation. Ionic channels are made of macromolecularproteins embedded in the lipid bilayer of the membrane. The availability of experimen-tal BLMs allows the membrane channel reconstitution to be carried out under well-defined and controlled conditions. Hence, direct characterization of the properties ofsingle ion channels at a molecular level has become possible [2,4 (Ch. 6)]. In the follow-ing subsections a few examples of recent papers on ion channels are described.

(c) Ion channels in BLMs. The first evidence of single-channel activities in BLMswas succinctly described more than 30 years ago. Nowadays, channels are found in allsorts of biomembranes including the plasma membrane of sperms, bacteria, and higherplants, the sarcoplasmic reticulum of skeletal muscle, synaptic vesicle membranes of ratcerebral cortex, and the skin mucus of carps, to name a few [2,4].

(d) Model for the Cristae Membrane of Mitochondria. Electron conduction in, BLMswas one of the most significant findings in recent years. It was demonstrated for thefirst time that experimental BLMs could be electron conducting. This significant findingconstituted a major science news story, reported in the August 6, 1984, issue ofChemical and Engineering News (pp. 19–20). The discovery of electron-conductingBLM, together with the application of cyclic voltammetry, was an event to be amongthe major breakthroughs in lipid bilayer research since the first BLM was reported in1962. The main finding was that after the saturation of an unmodified BLM withTCNQ, one of a class of so-called ‘‘organic metals,’’ the membrane became semicon-ducting in the presence of electron donors and acceptors, as evidenced by cyclic voltam-metry results [19]. Of immediate interest were our observations in the area of electrontransfer and redox reactions in biomembranes, of which the cristae membrane of mito-chondria and the thylakoid membrane of chloroplasts are examples.

Although the standard potentials (E0’) of these electron-transfer chain componentsin the aforementioned organelle membranes had been measured by a number of methods,none of the methods used up to that point were based on redox reactions taking place atthe membrane/solution interface, nor had they involved a lipid bilayer. Therefore, it wasconcluded that the values published in biochemistry texts and handbooks might be quitedifferent from the actual values in the membranes [4, ch. 8].

3. Light-Induced Phenomena in Pigmented BLMs [3,4,20,21]

Photoelectric effects in pigmented BLMs (p-BLMs) was reported in 1968 and subsequentlyused to elucidate the mechanisms of photosynthesis [20,21], in particular, the primaryquantum conversion step, in which light-induced charge separation into electrons andholes leads to redox reactions on the opposite side of the BLM (Fig. 2). This is so,principally owing to the ultrathin lipid bilayer (�5 nm) which can withstand an electricfield strength of more than 300 kV/cm. Further, the insulating nature of the lipid bilayer(an unmodified BLM’s typical resistance is greater than 109 cm2) prevents the thermo-dynamically favored back reaction (i.e., electron and hole recombination), which other-wise would not be possible in the production of oxygen and carbohydrate in naturalphotosynthesis. Thus, on the basis of pigmented BLM studies over the past decades,numerous investigators have been attempting some practical applications by mimickingthe natural visual receptor, as a model for the purple membrane of Halobacterium halo-bium, and photosynthetic membranes for solar energy utilization, as described briefly inthe following subsections.


(a) Model for the Thylakoid Membrane of Chloroplasts. Our knowledge of naturalphotosynthesis has emerged in physicochemical and molecular terms through in vitromembrane reconstitution and in vivo studies. In membrane reconstitution studies, reac-tion centers of PS-I (Photosystem I) and PS-II (Photosystem II) have been incorporatedinto artificial BLMs. The insights gained from such studies have led to the development

FIG. 2 Electron transfer across BLMs. Top: showing C60 (fullerene) doped s-BLM and sb-BLM.

Cyclic voltammograms and equivalent circuits. Bottom: redox mechanisms of C60 doped BLM under

load. See Ref. 67 for details.


of the so-called semiconductor septum electrochemical photovoltaic cell, a subject thathas been discussed elsewhere [22].

Returning now to the use of p-BLM as a model for the thylakoid membrane, thepartial reconstitution of the PS-I reaction center from chloroplasts in BLMs was accom-plished by two different methods. The first method was the association of liposomes thatcontained thylakoid membrane fragments with the BLM. The second method involved theincorporation of the reaction center into the BLM through fusion of liposomes with theBLM. In both cases, photoelectric effects were observed, which required the addition ofcarriers of reducing equivalents. The action spectrum of the photoelectric response fol-lowed the action spectrum of the PS-I reaction center. This set of experiments lend supportto the idea that charge separation in the reaction center results in the generation of apotential difference across the thylakoid membrane. Moreover, the experiments show theusefulness of the planar BLM for reconstitution studies. It should be mentioned thatContino et al. [23] reported the use of liposomes to characterize the assembly of supportedBLMs and found that assembly of the lipid bilayer occurs with a specific orientation andthat the protein plays no role in establishing the orientation of the supported BLM.Lamrabte et al. [24] reported photosensitization of covalently linked phthalocyanine com-plexes in BLMs and in SnO2 photovoltaic cells. The triad ZnTPP–ZnPLC–AQ was foundto be the best among other tested complexes (ZnTPP–AQ and ZnTPP–ZnPLC). Theresults were discussed in terms of fluorescence quantum yield and lifetime induced byquinines (see [4 (ch. 9)].

The mechanism of electron transfer at the membrane level has been investigated bymany groups using different BLM systems [20,21]. For example, Seta et al. reportedcertain carotenoporphyrin–quinone (CPQ) triads in BLMs and proposed an electrontransfer scheme. Briefly, an unmodified BLM is an excellent insulator (membrane resis-tance > 1010 cm2). Upon modification by molecules such as CPQ which spans theBLM, light-generated electrons are able to move from one side of the bathing solutionto the other; this is facilitated by the presence of an electron donor and acceptor onopposite sides of the lipid bilayer. Later, Lamrabte et al. [24] reported the light-inducedelectron transfer in BLMs containing a stacked metallotriporphyrin. The reaction schemedevised by them is supported by fluorescence and UV–visible light absorption data, whichfit the experimental time course of the photoelectric measurements.

(b) Model for the Visual Receptor Membrane of the Eye. Of our five common senses,vision, hearing, smell, taste, and touch, the eye perhaps, is the most crucial. In each ofthese senses, billions of nerve cells are involved. At almost the very instant that aphoton of light excites a cell in the retina of the eye, the photoreceptor cells of rods andcones convert this stimulus into an electrical signal, the language of the brain. Althoughthis transduction is fast, intricate, and precise, we do not yet fully understand all itsdetails. Nevertheless, the following is known or has been established. The main functionof the eye, via the retina, is conversion of a light signal to an oscillation of nerveimpulses. It has been known for some time that the primary light excitation process bythe retina takes place through a change in the structure of rhodopsin embedded in thelipid bilayer of the photoreceptor cell. When a photon of light hits a rod cell, for exam-ple, it is immediately absorbed by one of the 100 million rhodopsin molecules that areembedded in the lipid bilayer of the visual receptor membrane. There are sac (disk orflattened sac) membranes, G-proteins, enzymes, ion channels, and cyclic-GMP (cGMP).The rhodopsin molecules have been found to criss-cross the lipid bilayer seven times,and consist of retinal and opsin, in a Schiff base arrangement. In the dark, the retinal is


in a cis-conformation. Upon light excitation, the retinal straightens out into a trans con-formation. This cis-to-trans change in the retinal of the rhodopsin molecule triggers abiochemical cascade. The activated rhodopsin molecule in turn stimulates a membrane-bound protein, called transducin (a G-protein), then causes an enzyme to break downcGMP, thereby dramatically lowering its concentration. cGMP, a so-called ‘‘secondmessenger,’’ carries the signals from the sac membranes, where the photons of light areabsorbed, to the plasma membrane of the cell. The plasma membrane contains a largenumber of channels that control the ion flow into the cell. As ions move down theirelectrochemical gradients through the membrane, they alter its electrical potential.Before light excitation or in the dark, the channels embedded in the lipid bilayer of theplasma membrane are open because of a high level of cGMP. This allows Naþ andCa2+ to flow into the cell. Upon excitation by light, the channels close, thereby causingthe electrical potential inside the cell to depolarize or to become more negative, whichin turn reduces the amount of neurotransmitter that is released from the base of the cellto act on other cells, alerting neurons in the next layer of retinal cells that a photon oflight has arrived.

In summary, this complex cascade of transduction of events results in the generationof an electrical signal which is transmitted through a bundle of nerves and eventuallyperceived by the brain [2,4]. BLMs containing a photopigment extracted from bees’ eyeswere reported [3,4 (ch. 9)]. It was found that, on illumination of such pigmented BLMs, anincrease in membrane conductance and a change in membrane potential were observed.When the BLM was clamped at a constant voltage, random current fluctuations betweencertain levels appeared, suggesting the formation of dynamic channels whose conductancewas between 100 and 300 pS. The results appear to be consistent with those found inBLMs containing rhodopsin [4].

(c) Model for Purple Membrane of H. halobium. Bacteriorhodopsin (bR), a single-chain polypeptide bound to chromophore retinal from the purple membrane of H. halo-bium, has been intensely investigated in the last two decades. Structurally, it is relatedto the visual pigment rhodopsin. However, bR functions as a light-driven proton‘‘pump’’ converting the energy of light into chemical energy. The light-generated elec-trochemical potential gradient of protons is utilized in the synthesis of ATP. As a resultof proton translocation across the purple membrane, a photocurrent is observed fromthe intracellular side to the external medium. The study of light-induced electrical phe-nomena (photoelectric effects) in vivo in pigmented organelles is technically difficultowing to the minute size of the cells. Hence, resorting to experimental BLMs has beena favorite approach by many investigators interested in the study of photoelectric sig-nals in otherwise not accessible pigmented natural membrane systems. Reconstitution ofbR into BLMs was first reported in the mid-1970s. In the intervening years, a sizableliterature exists on bR-containing BLMs, liposomes, and related systems [4,20,21,25].

Owing to its primitive origin, H. Halobium is unique because of its role in photo-synthesis, energy transduction, and ion pumping, as well as in practical applications. Thefocal point of interest of H. Halobium is centered on bR, the chromophore of the purplemembrane as already mentioned above. Interest in bR is many fold; it is a highly stablepigment, it has a very fast (< 1 ps) photoresponse, it is sensitive to wavelengths of light,and it can be genetically engineered and produced in quantity. A bR molecule is composedof a retinal and an opsin which criss-cross seven times the lipid bilayer of the purplemembrane. Each of the opsins consists of an �-helix of 248 amino acids with a molecularweight of 26,000. Dancshazy and Karvaly reported in 1976 (see Refs. 3 and 4) and later


others [20,21] the photoelectric effects in bR-containing BLMs. The light-induced photo-electric effects have been analyzed in terms of photovoltage and photocurrent in connec-tion with the proton-pumping activity. Caplan and Fischer have measured and analyzedthe current-voltage (I=V) curves of bR membranes and concluded that proton-conduc-tance channels operate in the dark and in the light. The photocurrent of bR–BLMs dis-plays a polarity reversal at certain applied voltages, which has also been observed byothers [3,4]) In this connection, Herman and Rayfield proposed a voltage-independentcurrent generator shunted by light-dependent conductance on the basis of their data. Someof these interesting findings have been discussed [4,20–22].

Spectroscopically, the bR of the purple membrane absorbs at 570 nm, which is thepeak of the retinal chromophore. BR isomerizes between either an all-trans or 13-cisconformation, which has a peak at 554 nm. Flash excitation of the purple membranecauses a transient change in its absorption spectrum. The photocycle of the purple mem-brane of H. Halobium takes about 10 ms and consists of seven intermediates, of which thethree main ones are bR570, K610, and M410. It is of interest to note that the last step ofthe photocycle involves reprotonation of the Schiff base and reisomerization of the retinal.The photoelectric signal of the photocycle intermediates is important for practical appli-cations such as photodiodes. Ideally, a photodiode should have fast response, broadspectral range, high sensitivity, and fast recovery. The properties of BR immoblized inBLMs, L–B thin films, and on conducting surfaces appear to be a suitable candidate forphotodiode construction. Indeed, such construction has been reported as early as in 1980.A neural-network architecture based on a bR photodiode synapse has been proposed.Thus, the activity in this area termed ‘‘molecular and biomolecular electronics’’ lies at theinterface of surface chemistry, electrical engineering, membrane biophysics, and solid-statephysics. Bacteriorhodopsin- and retinal-based photoelectric devices, besides photodiodes,are being actively researched and developed [4,25]. In this regard, Sanger and Sigrist [26]reported light-dependent immobilization of biomolecules on BLMs and other materialsurfaces. The performance of sensing molecules on surfaces is improved, and molecularresponses are enhanced with covalently immobilized and oriented biomolecules. Light-dependent immobilization is exemplified by a procedure which leads to covalent binding ofimmunoreagents to BLM surfaces.

B. Supported s-BLMs, sb-BLMs, and t-BLMs: The New BLM Systems

The conventional BLM system has one major drawback in that it is not very stable, rarelylasting more than 8 h. As a result, many attempts have been made to stabilize thisextremely delicate lipid bilayer structure for fundamental studies and for practical appli-cations. The formation of a BLM on solid substrates has been extensively studied as afunction of the lipid composition, electrical properties, and BLM modifiers. Electricaltechniques (e.g., cyclic voltammetry, electrochemical impedance spectroscopy) have beenproven very useful for performing biophysical and biochemical studies related to theelasticity of the lipid bilayer and the transition from the gel to the liquid-crystallinephase. Supported lipid bilayers (s-BLMs) have been also used as a biorecognition systemfor monitoring protein binding and antigen–antibody interactions. Biotinylated IgG hasbeen anchored to a lipid bilayer via a streptavidin bridge in order to develop a stable,specific, and biocompatible biorecognition site. Natural receptors with clinical relevancecan be also embedded in the BLM for the formation of probes and subsequent study ofreceptor–ligand interactions [4–10,27–30].


One recent successful result has been the formation of a BLM on metallic andhydrogel supports [8,28,29]. The origin of supported BLMs dates back many yearswhen we were interested in developing a model system for the thylakoid membranewith sufficient strength and size for use as a solar energy conversion system and otherpurposes. Later, supported BLMs were formed on metallic wires, conducting SnO2 glass,gel substrates, and on microchips, as described in the following paragraphs. These self-assembled, supported BLMs, have not only overcome the long-term stability problem ofconventional planar lipid bilayers, but have also have opened up a range of possibilities inmanipulating interfacial films as well as in developing practical biosensors [11–13,16].

Mention should be made here that the development of electrochemical biosensors isgrowing at a rapid pace since the early 1980s [4,10,27]. Many approaches such as the L–Btechnique, monolayers on gold electrodes, a variety of modified electrodes (carbon paste,glass carbon, tin oxide, etc.), and artificial BLMs have been used. Of all these, the BLMsystem is by far the most biomembrane-like, fluid in nature, and is capable of the ligand–receptor interaction. With all other systems the compound of interest is immobilized in arigid, solid-like structure, whereas in the BLM it is embedded. By embedding is meant thatthe compound(s) (membrane modifiers such as polypeptides and proteins) of interest in thelipid bilayer is relatively free to adapt to its surroundings. The functions of biomembranesare mediated by specific modifiers, which assume their active conformations only in thelipid bilayer environment. Further, the presence of the lipid bilayer greatly reduces thebackgound noise (interferences) and effectively excludes hydrophilic electroactive com-pounds from reaching the detecting surface and causing undesired reactions. Hence, thes-BLM system offers a wider opportunity for biosensor development. From the specificity,selectivity, and design points of view, the lipid bilayers or BLMs are the natural environ-ment for embedding a host of materials of interest. A number of comprehensive assess-ments are available [7–10,16,17,23,24].

1. BLMs on Polycarbonate and Cellulose Filters

The first report was published in 1978 describing the formation of supported BLMs inpolycarbonate filters with much improved stability to both chemical and mechanical dis-turbances. For example, pigmented BLMs in these microporous filters could exhibitphotoeffects lasting a period of days. It was then concluded that an extension of the systemto other area of membrane biophysics was anticipated. Thus, notably, Yoshikawa andcolleagues (for citations, see Ref. 4) have reported related systems using this type of BLMwith interesting results.

2. BLMs on Metallic Substrates

Supported lipid bilayers or s-BLMs on a solid support can be formed by a number ofmethods including the consecutive two-step technique as follows: Step (1) placing aTeflon-coated metal wire (e.g., Pt, Ag, Cu, Ni, and stainless steel or other alloys) to becut in contact with a BLM-forming lipid solution and then cut with a sharp knife, and step(2) immersing the lipid layer that has adsorbed on to the cut end of the metal wire surfaceinto an aqueous solution. Depending on the diameter of the wire and its composition, anail clipper may be used for cutting purposes. For the best cutting of metal wires, aminiature guillotine has been used, where the sharp knife is moved vertically on to thewire placed on a flat surface and immersed in a lipid solution (Fig. 1(b) and (c) and Fig. 2).The compounds used and methodologies developed for conventional BLMs are beingapplied to the s-BLM systems [5–10, 27–31].


3. BLMs on Hydrogels (sb-BLMs)

Although s-BLMs on metallic substrates are attractive for certain purposes, as will bedescribed later in this chapter, the metallic substrate, however, precludes ion transloca-tion across the lipid bilayer. Therefore, the pursuit of a simple method for obtaininglong-lived, planar BLMs separating two aqueous media has been elusive until a fewyears ago [17, 28–30]. A brief description of forming a planar BLM on hydrogel sub-strates is given below.

The formation procedure consists of three steps. In the first step, a chlorided Ag wire(Ag/AgCl) is inserted into Teflon tubing, which has been previously filled with a solutionof agar (or agarose) in KCl saturated with AgCl. The AgCl electrode and the filled Teflontubing are glued together with wax at the point of insertion. In this way an Ag/AgCl–Teflon tubing salt bridge (sb) is constructed. In the second step, the tip of the other end ofthe Teflon sb is cut while immersed in a BLM-forming solution with a sharp knife, as isdone with the s-BLM technique. In the third step, the Ag/AgCl–Teflon sb with the tipfreshly coated with lipid solution is then immersed in 0.1 M KCl solution in the cellchamber. Alternatively, the second step described above may be carried out in air andthen the freshly cut end of the sb is immediately immersed in the lipid solution for a fewminutes. In either case, the cell chamber filled with an appropriate aqueous solution (e.g.,0.1 M KCl) contains an Ag/AgCl reference electrode and an Ag/AgCl–Teflon sb with aself-assembled BLM at its end. In contrast to s-BLMs on metallic substrates where ionconduction is excluded, a sb-based BLM allows ion translocation across the lipid bilayer.The lead wires of the two electrodes shown in Fig. 1 are connected to the measuringinstrumentation (see also Fig. 2). In this connection, it should be noted that the sbmight be identified, respectively, as items 1 and 2, thereby eliminating one of the chambers[4,8,28,29].

4. s-BLMs on Interdigitated Electrodes

As mentioned above, BLMs have been earlier deposited on a variety of substrates includingSnO2 (conducting glass), freshly cut metal, and hydrogel supports. Formed in this manner,it is envisioned that the first monolayer of the lipid is ‘‘sorbed’’ on to the support therebyforming a fixed half-BLM. The second half-BLM is then self-assembled on to the anchoredmonolayer. As a result of hydrocarbon-chain interactions and of being fluid, the secondhalf-BLM is relatively free to move with respect to the anchored half-BLM, a situationwhich is not unlike that of ‘‘oil’’ lubrication. More recently, we have extended the experi-ment described above to the interdigitated structures (IDS). IDS are finger-like electrodesmade by microelectronic technologies and used in microchip applications [5,6,30].

The precise arrangement and degree of ordering of the lipid molecules in the finalstructure is not known for certain. However, it seems highly probable that the bilayernature of the assembly is a consequence of the thermodynamics of free-energy changes atthe metal–lipid surface and at the lipid–aqueous solution interface [4,13]. Our measure-ments of the electrical properties of supported lipid bilayers described here are consistentwith those of conventional BLMs and closely related systems.

C. Equipment and Measurement

The electrical properties can be easily measured using instruments ranging from a goodpH meter to a computer-controlled workstation. In our laboratory we have used home-assembled components to commercially available units. In voltammetry the potential of


the cell is varied and the corresponding current is monitored. One type of voltammetry iscalled cyclic voltammetry (CV) and employs a triangular voltage waveform. As a means ofinvestigating redox reactions, CV has been widely used and has been termed the equivalentof spectroscopy. A graph with the current plotted on the vertical axis versus the potentialon the horizontal axis is called a voltammogram, which is characterized by several para-meters. The working electrode (WE) may be made of Pt, Au, carbon paste, glassy carbon,or semiconductor SnO2. An Ag/AgCl or a saturated calomel electrode (SCE) is used as thereference electrode. The potential scan or sweep is carried out between two potentialvalues of interest (e.g., from about 1.2 to �0:8 V versus the SCE). The scan rates canbe anywhere from 0.1 mV to 100 V or more per second but values between 10 and 400 mVs�1 are frequently used. The current response of processes at a metal electrode is indicativeof the nature of the redox reaction at the interface. Experimental results derived frommeasurements of this kind permit the elucidation of mechanism and the thermodynamicparameters of the process (e.g., charge-transfer reaction). Frequently, a ‘‘duck-shaped’’voltammogram is obtained for redox reactions. The underlying physical mechanismresponsible for the ‘duck-shaped’ profile is based on the interplay between the kineticsof the charge-transfer process and mass transport of the charge carriers (oxidants andreductants). These basics of CV and its elegance and simplicity are well known to electro-chemists. According to our knowledge, this powerful CV technique was applied for thefirst time to membrane studies in 1984 with interesting results [5,19].

For membrane CV, a Lucite block containing two adjacent 2 cm diameter cham-bers (8 mL), one of which holds a 10 mL Teflon cup, is used. The Teflon cup wasreferred to as the inside, and the other chamber as the outside. The voltammograms ofthe BLM are obtained using an X–Y recorder fed by a picoammeter and a voltagegenerator (e.g., Princeton Applied Research, Universal Programmer, Model 175). Thevoltage from the programmer is applied through the potentiometer to the SCE immersedin the inside solution. Another SCE immersed in the outside solution is connected to thepicoammeter. The important feature of the setup is a very weak dependence of its inputvoltage on the current being measured. This means that the current is measured under‘‘voltage clamp’’ with an accuracy of �1 mV. We have also used a three-electrodesystem for obtaining voltammograms in the following configuration: one SCE is placedin the Teflon cup and two other calomel electrodes are on the outside. An EC/225Voltammetric Analyzer (IBM) in the cyclic voltammetry mode, in conjunction with anX–Y recorder, is used throughout, in the second setup. In this connection, Kalinowskiand Figaszewski [32] have described a four-electrode capacitance-to-period converterdesigned for capacitance measurements of BLMs. The capacitance measurement consistsof cyclically charging and discharging the measured capacitance with a constant current,regarding its absolute value. The voltage of the studied capacitor is triangularly shaped.The cycle duration is proportional to the input capacitance. According to the authors,capacitance measurement with a four-electrode system makes it possible to reduce con-siderably the errors caused by electrode and electrolyte impedance. It is possible to usehigh-resistance microelectrodes. The system makes it possible to measure the capacitanceat an imposed polarization potential; the voltage oscillates about that value during themeasurement. This makes it possible to measure the membrane capacitance as a functionof polarization potential. Research in this field necessitates specialized equipment, whichis not available on the market, as well as expensive software for processing the analyticalinformation. Toward this end, Katrivanos et al. [33] have prepared a paper aimed tohelp researchers engaged in the field of biosensors, who, in many cases, lack the properelectronic support and integrated technology. Their paper provides some details of an


integrated system and all the subunits that are necessary for proper measurements forthese kinds of experiments.

For the investigation of light-induced effects in membranes, a new technique termedphotoelectrospectrometry has been developed in our laboratory, the details of which maybe found in many papers [4,5,17,22,23].

1. Monitoring Solid-Supported Lipid Bilayer Formation

Similar to time-resolved spectrometry through which the transient molecular rection canbe observed optically, the dynamics of modified or unmodified BLMs, or reconstitutedbioreaction on a BLM, can be studied electrically by CV technology where the capaci-tance, resistance, membrane potential, and current peak are the fundamental parametersin determining the static or the dynamic change in the BLM system. Among these studies,the formation of a s-BLM is valuable in analyzing BLM mechanics and further electro-chemical reactions thereon [34,35].

Here, the membrane capacitance and resistance are chosen as two principle para-meters in monitoring the formation of different concentrated unmodified s-BLMs. Thetime-resolved capatance and resistance are measured following the model described above,based on the recorded s-BLM voltammogram.

2. Dynamic of s-BLM Formation [35]

The major principle of the CV technique is the imposition of extra potential on the s-BLMin monitoring its dynamics. However, since about 100 mV is carried on the s-BLM, andthis value is even superior to the biological resting potential, so if the CV detectingpotential affects the dynamics of the s-BLM it becomes a significant problem. This influ-ence is detected by comparison of the ‘‘continuous’’ measurement and ‘‘point’’ measurement(also called nonimpact measurement). The point measurement is conducted through asingle cyclic voltammogram recording at several predetermined time points, while holdingthe s-BLM electrically isolated during other periods. In contrast, continuous measurementis the cyclic recording without interruption during the entire formation. The polar mole-cules of the lipid is supposed to be charged for this behavior. In the initial stage of BLMformation, the lecithin or PC polar molecules are in a random state. In the case of no CVpotential imposed on the BLM, the preferred self-organization (parallel arrangement ororientation) is completed only through the attraction between the highly hydrophilicnascent metal surface and the polar groups of the lipid molecules. However, the detectingcyclic potential wave gives rise to an additional alternative electrical field. The ‘‘infield’’molecules are forced by the sweeping voltage to accelerate orientation. This, in macro,shortens the bilayer formation on the nascent metal surface.

According to the capacitance monitor of the s-BLM formation, there are, in general,four characteristic stages that can be distinguished from the first ‘‘cutting operation’’ tothe final formation of s-BLMs (Fig. 3):

1. At the start, the capacitance fluctuates at random for a few seconds because thetip of the Teflon-coated metallic wire has just been cut off with a sharp bladewhile immersed in a lipid solution.

2. During the next few seconds, the capacitance is relatively low due to the drop ofBLM forming solution on the newly cut surface on the tip of the metallic wires.A monolayer of lipid molecules is irreversibly and rapidly adsorbed onto it.


3. The capacitance tends to increase after immersion of the wire in the aqueoussolution. The lipid drop on the tip becomes small and thin and tends sponta-neously to form a lipid bilayer. Moreover, the potential of CV also helps toquicken the s-BLM formation.

4. After the s-BLM adsorbed on the metal support has been formed, the capaci-tance becomes stable, though sometimes it fluctuates slightly due to the transferof the solvent and of the excess lipids to the aqueous phase.

3. Instrumentation

A number of methods have been developed over the years to study the properties of BLMssuch as optical, electrical, mechanical, transport, and permeability. Of these methods, weshall describe only the electrical methods. In the last decade, many new electrochemicalmethods have been developed and applied to membrane research. Among them, CVturned out to be a very powerful method. The basics of CV consist of cycling the potentialof a WE in an unstirred solution and measuring the resulting current. The potential of the

FIG. 3 s-BLM formation dynamics: (a) time-resolved membrane resistance during the formation

process (curves 1, 2, and 3 represent continuous monitoring for 0.5, 2, and 10% solutions,

respectively; ‘‘+’’, point monitoring without CV potential effect); (b) time-resolved membrane

capacitance during the formation process (curves 1, 2, and 3 represent continuous monitoring for

0.5, 2, and 10% solutions respectively; ‘‘+’’, point monitoring without CV potential effect); (c)

characteristic stages of the s-BLM formation monitored by the membrane capacitance.


WE is controlled relative to a reference electrode, which is provided by a triangularpotential waveform generator. The instrumentation used with BLMs can be much simplerthan that used in conventional CV, because the high resistance of BLMs can be studiedwith a two-electrode setup. Thus, a picometer together with a voltage waveform generatoris all that is required. If a computer or an X–Y displaying device is available, the current/voltage (I/V) curves may be obtained, which are known as voltammograms. From suchvoltammograms, information about the thermodynamic and kinetic parameters of theBLM system may be obtained, thereby providing insights into the mechanism of themembrane process under investigation [4,34,35].

A standard circuitry of the setup for obtaining cyclic voltammograms of the s-BLMsystem has been used. The supporting metallic wire serves as a WE in a one-cell chamber.The reference electrode, usually a chlorided Ag wire (Ag/AgCl) electrode, is dipped in the0.1 M KCl solution placed in another chamber, a sb spanning over the two chambers. Fora two-electrode system as is usually used in the measurement, the newly cut tip of themetallic wire, coated with absorbed lipids, acts as the WE. The current through the s-BLMis measured in the auxiliary electrode during the CV. The setup is housed in a Faradaycage to minimize interference by external noise and electrical transients. In spite of shield-ing, external noise may still be picked up by the switch box; therefore, for the criticalmeasurements the switch box should be incorporated within the same Faraday shield asthe cell. All cables used are shielded and the shields are grounded.

4. Parameters Determining BLMs from Voltammogram

(a) Planar BLMs. The typical equivalent circuit of a traditional planar BLM system(Fig. 4) is represented by a membrane resistance Rm in parallel with a membrane capa-citance Cm. The triangular sweep wave in the range of �V0 with a scan rate A(mV s�1) is the input from the circuit. The current in nanoamperes or picoamperes ismeasured. There are two components in the current through the membrane, namely,the charging current ic and resistance current ir. The former is determined by the capaci-tance as follows:

ic ¼ Cm

dV

dt¼ CmA ð1Þ

It can be shown that the capacitance current ic through the membrane capacitance is aconstant. From Ohm’s law, the latter component ir of the membrane current is caused bythe membrane resistance, i.e.,

ir ¼V

Rm

ð2Þ

So the net current passing through the membrane can be expressed as

i ¼ ir þ ic ¼V

Rm

þ CmA ð3Þ

Equation (3) shows that the current through the resistor increases with increasingscan voltage. In the case of the constant scan rate A, with fixed values of Cm and Rm, thecurrent i has a linear relationship with the sweeping potential V . Thus, the slope reflectsthe value of Rm, whereas Cm can be determined by measuring ic according to the graph ofthe I–V response; ic will jump to its negative values suddenly (�ic) only at such pointswhere the sweeping wave reaches its maximum and begins to reverse. The jump distance 2h


equals 2ic; thus, Cm will be calculated by merely measuring h. However, the typicalvoltammogram of an unmodified s-BLM obviously has a different shape when comparedwith parallelograms obtained for the planar BLMs (Fig. 5). .The difference between themindicates that the equivalent circuit proposed for the conventional BLM is no longer validfor the s-BLM system. Measurement errors will have a great impact on the accuracy ofparameter determination unless the circuit is improved [16,34,35].

IV. SELF-ASSEMBLED LIPID BILAYER-BASED BIOSENSORS

A. Biosensors and Their Development

The interaction of biophysics with cell physiology and molecular biology has given rise toan exciting area of research termed membrane biophysics, which integrates up-to-datefindings on molecules and processes involved in inter- and intra-cellular recognition andcommunication [7–10]. Knowledge of the ideas and findings resulting from such interdis-ciplinary research are now being used for practical applications in analytical chemistry,immunology, photobiology, chemical/biosensors and transducers, and in molecular elec-tronics.

A biosensor is an analytical device incorporating biological and chemical sensingelements either intimately connected to or integrated with a suitable transducer, enablingthe conversion of concentrations of specific chemical compound concentrations into digi-tal electronic signals. A key aspect is the interface between biological materials and elec-tronics. A majority of the biosensors developed thus far have incorporated an enzyme as abiological recognition component. All types of enzyme sensors are based on the classic

FIG. 4 Voltammograms on the BLM simulator: (a) voltammograms with different values of Cm

(C1m < C2

m < C3m); (b) voltammograms with different Rm (R1

m < R2m < R3

m).


idea that involves glucose by detecting the reduction in oxygen when the oxidation ofglucose is catalyzed by the enzyme. To date, there are many substrates that have beenmeasured by the use of oxygen oxidoreductases, and the majority of the enzyme biosensorshave been designed specifically for the determination of a large number of ‘‘cardiac’’enzymes in blood. Tissue materials from plant and mammalian sources have been success-fully employed for the construction of biosensors as well. In brief, receptors, enzymes,antibodies, nucleic acids, tissue, micro-organisms, organelles, cell, etc., which can beembedded in/on to the lipid bilayer have been used as physiochemical transducers. Thisclass of biocatalytic materials simply maintains the enzyme of interest in its natural envir-onment (e.g., lipid bilayer), which results in a considerable stabilization of the desiredenzymic activity. The method of detection can be based on a variety of systems that maybe electrochemical, optical, thermometric, piezoelectric, or magnetic. Biosensors usuallyyield a digital electronic signal, which is proportional to the concentration of a specificanalyte or group of analytes. While the signal may in principle be continuous, devices canbe configured to yield single measurements to meet specific market requirements. Forexample, the microbial sensors are composed of immobilized micro-organisms and anelectrochemical device and are suitable for the on-line control of biochemical processes.These sensors involve the assimilation of organic compounds by the micro-organisms,change in respiration activity, or production of electro-active metabolites. These changeshave been monitored directly by an electrochemical device.

The sensitivity of electrical measurements developed for the electrochemical biosen-sors coupled with the specificity of antigen–antibody reactions provides a useful tool forimmunology. However, more recently, optical sensors for immunoassays have been receiv-ing considerable attention in research laboratories and also for in vivo applications.Among the different types of optical biosensors, two appear to be especially promising.One is based on a surface plasmon resonance phenomenon and the second is a fluores-cence capillary fill device. Surface plasmon resonance in a thin metal film deposited on awaveguide can be induced by an electromagnetic wave generated when light is reflected

FIG. 5 Method for calculating membrane parameters from s-BLM voltammogram. (From Refs. 34

and 35.)


within the waveguide, and is highly sensitive to variations in the refractive index of theimmediate surrounding medium. This phenomenon is monitored by a reduction in theintensity of reflected light.

Generally, the biological components used in biosensor construction can be dividedinto two categories: (1) those where the primary sensing event results from catalysis (suchas whole micro-organisms, tissue slices, organelles, and enzymes) and (2) those whichdepend on an essentially irreversible binding of the target molecule (i.e., cell receptors,nucleic acids, and antibodies). The essential element, however, in making a successfulbiosensor is to provide a suitable immobilization procedure for biological compounds.Four main approaches to enzyme immobilization have been employed: (1) physicaladsorption at a solid surface, (2) entrapment in polymeric gel or within microcapsules,(3) cross-linking by means of bifunctional reagents, often in combination with (1) and (2),and (4) covalent binding to a reactive insoluble support. These resulted in numerousmatrices that have been used.

In practice, researchers continue to search for ideal combinations of biocatalysts–enzymes, antibodies/antigens, bacteria, whole cells, plant slices (even isolated receptors),and transducers. In this context, genetic engineering technology will have a role in improv-ing the biological component of enzyme-based and whole-cell biosensors. Material scienceand chemical engineering are helping to find solutions for such problems as suitableimmobilization procedures, transducers, and the effective coupling of the biological com-ponent to the transducer and the subsequent amplification system. The main types oftransducers used in biosensor construction are listed in Table 4. Although in recentyears a variety of different biological components have been used, it is by no means certainthat all possible combinations of sensing element and transducers have been explored.

In recent years, the development of biosensor configurations has been concentratedlargely around the design of the transducer used. Further researchers’ attention, however,should be focused on the mechanism of molecular recognition and catalysis. The funda-mental properties of the device must be better understood in order to optimize criticalfactors such as response time, selectivity, and stability. Immobilization technologies andnew membrane materials may basically change the present performance of biosensors.

TABLE 4 Types of Transducers and

Measurements Used in Biosensor Technology

Transducers Measurement

Oxygen electrode Electrochemical

Ion-selective electrode Potentiometry

Modified metal electrode Amperometry

Field-effect transistor Conductometry

Conductometry Impedometry

Spectrophotometry Photometric

Laser light scattering

Optical fibers combine with

absorption and fluorescence

Surface plasmon resonance

Thermistors Thermometric

Piezoelectric crystal Acoustic

Surface acoustic wave device


B. Fundamentals of the Lipid Bilayer-Based Sensors and Devices

It should be pointed out that, first of all, unmodified planar lipid bilayers (i.e., BLMsformed from common phospholipids or oxidized cholesterol dissolved in n-octane) in 0.1M KCl will typically have the following electrical properties: Rm greater than 108 cm2,Cm of about 0.4 �F/cm2, Em about 0, Vb about 200 50 mV, and current/voltage (I/V)curves obeying Ohm’s Law. However, incorporating a host of materials such as pigments,dyes, polypeptides, membrane proteins, organic metals, and semiconductor particles candrastically alter the electrical properties of BLMs.

Recent success in interdisciplinary research in biology combined with electronicshas led to exciting new developments based on enzymology and transducer techniques.They are known as enzyme electrodes, enzyme thermistors, CHEMFET/ENFET devices,and immunosensor or enzyme transistors. Collectively, they are called ‘‘biosensors orbiochips.’’ A common feature of all these devices is a close connection between theenzyme and a transducing system, which is used to follow the enzymic reaction. Theessential principle of the devices, broadly speaking, is predicated on the ligand–receptorcontact interactions. Application of such developments in the fields of medicine, phar-maceuticals, biochemistry, environmental settings, robotics, and the food industry areobvious. For example, enzyme thermistors make use of the heat that is liberated duringan enzymic reaction. Their usual sensitivity is around 10�28C. A recent modification ofthe enzyme thermistor is the ‘‘TELISA’’ electrode, which achieves a sensitivity of about10�13 M using an immunoabsorbent. It is expected that this measuring technique willfind broad application in continuous measurements of the release of hormones and/orantigens–antibodies in blood circulation. The rapid achievement of a new steady state inthe reaction occurring at an enzyme electrode after a random perturbation makes thelatter ideally suited for monitoring an industrial process, e.g., the production of anti-bodies. Classic calorimetric methods require much more time than an enzyme thermistorassay to perform a quantitative analysis. Two other interesting developments are ellip-sometry and piezoelectric crystals. In ellipsometry, a close connection between theenzyme and the transducing device is not required. The method relies on the changein the angle of polarization of incident light that is reflected by a layer of biomoleculesbound to a solid surface. A change in the thickness and conformation of this layer,under the influence of other macromolecules interaction with the layer, can be easilymonitored. This principle is now used in the fermentation industry. Piezoelectric crystalscan be used in the analysis of traces of certain compounds, mainly anesthetics. Thefrequency of the crystal depends strongly on the absence or presence of adsorbed mole-cules on the surface of the crystal. A coating process may increase the selectivity ofcrystals toward a given compound, e.g., with hydrophobic substances such as oils andfats [11,12,14,15].

Another exciting new research area is the combination of semiconductor technologywith enzymes and other biological macromolecules. Here, mostly field-effect transistors(FETs) are used. If the sensitivity of a FET toward certain chemicals or ions can beachieved, the prototype of an ‘‘ISFET’’ is born. A common feature of all these devicesis the use of a MOS (metal oxide semiconductor) structure. In combination with a thinlayer of palladium, a high sensitivity toward gaseous hydrogen can be achieved. In thiscase, the membrane separates the gaseous from the liquid phase. Addition of traces ofcertain metals (e.g., Ir) to the Pd–MOS device also increases its sensitivity toward ammo-nia. It has been shown that such a device is capable of monitoring reliably the productionof hydrogen by micro-organisms, e.g., Clostridium acetobutylicium [4,15].


C. Molecular Electronics and Planar Lipid Bilayer-Based Biosensors[4,7–10,14,15,27–30]

Molecular electronics uses molecular materials in which the molecules retain separateidentities. As a result, the properties of such materials depend on the molecular arrange-ment, properties, and interactions. Theory seeks to guide the design and synthesis ofeffective molecular materials. It does so by analysis, interpretation, and prediction, leadingto the development and evaluation of concepts, models, and techniques. The role of theoryin treating molecular properties (mainly by molecular orbital methods) and arrangement(by electromagnetic or quantum-mechanical approaches) is of importance. When thesefactors are combined, the material properties can be treated more successfully in caseswhere the interactions are not essential in, e.g., in nonlinear optics as opposed to electronictransport properties.

The major advantage of molecular electronics with a lower limit in micrometers isthe further development of lithographic techniques. The changed physical properties in thesubmicroscopic region are the major obstacles to further miniaturization in semiconductortechnologies. The physical border for the silicon technologies is about 100 nm, becauseone cannot overcome the characteristic lengths such as diffusion, Debye, and tunnellengths. With still smaller dimensions, we enter the realm of biological and molecularsystems. The human brain (as well as our sensor organs), is also, without silicon, enor-mously capable. Although biotransducers function much slower than silicon-based devicesand are not very reliable, they are extremely efficient. Also, despite their disadvantages,nature’s molecular device functions more generally and is superior to technical computersor sensors. In contrast to macromolecular biological systems, the main advantage ofmolecular devices, purportedly, is their relatively simple construction. In this sense, mole-cular devices may be readily synthesized and are always easily accessible experimentallyfrom a quantitative point of view.

The main elements of molecular electronics are the molecular wire, conductingmaterial, molecular-specific transducers of signals similar to the particles, and molecularswitches, memories, emitters, detectors, etc. The flux of information between the moleculescan be released in many ways. One of the most important is the transfer of individualcharges in terms of electrons, holes, or hydrogen ions, or of other shapes similar to theelements, like solitons, soliton waves, or excitons. Molecular switches may be optical,electrical, magnetic, or thermally reversible systems. Storage of information in a molecularsystem can be realized through a change in the electronic as well as geometric structures ofthe molecules in reversible thermal reactions, e.g., conformational or configurationalchanges on replacement of hydrogen or protons.

The key advantage of molecular and biomolecular computing is specificity. Thelarge number of variations that are possible with organic polymers allows for fine-tuning of electronic motions to a much greater extent than is possible in organicmaterials. In biological molecules, unclear configurational motions are comparable insignificance to electronic motions. This is certainly the case in all conformation-basedrecognition processes. Enzymic recognition is itself a basic form of information proces-sing. When proteins and other macromolecules are combined into highly integratedcomplexes, it becomes possible for conformational switching processes to propagateover significant distances. The cytoskeleton is a good candidate for such a signal-pro-cessing network.

One of the most delicate ‘‘molecular wires’’ is the so-called hydrocarbon chain,which is best represented through the chain of carbon atoms in polymers. Most organic


polymers are well-known insulators. However, polyacetylene [(CH)x], polydiacetylene,and polysulfonitride [(SN)x], with their conjugated double bonds, are semiconductors orsuperconductors. Such conjugate systems form the group of organic conductors andsemiconductors. The most important organic electroactive polymer is polyacetylene.The foundation of the electron–hole pairs and the positive and negative charges arequite well known. In the outer electric field, the electron and the hole are accelerated inopposite directions. These properties can be used in optical switches for switching on andoff the flux of information. Combining molecular ‘‘wires’’ and switchable molecules couldlead to the construction of electronic systems based on molecules. Present research isoriented toward discovering peptides/proteins that can transduce electrical current orexist in two-electrical stages. This would lead to future ‘‘biochip.’’ Research on biochipscould lead not only to a better understanding of higher nerve function, but also to thefoundation of qualitative computer systems that could provide many of the activitiescurrently performed only by the human brain.

Biochips can be considered as highly sophisticated biosensors. The unique proper-ties of biochips are their analog and digital computing potential, self-perpetuating andpotentially self-repairing. Biochips hold promise in a variety of applications such asbionic implants, memory-intensive systems, image processing and storage, artificial intel-ligence, language processing, and molecular computers. For instance, the analog cap-ability of biochips could enable the creation of ‘‘artificial intelligence.’’ As such, biochipsare at a very early stage of research and development. As with biosensors, the currentproblem is our inability to produce uniform, high-activity stable biomolecular layers andtheir associated transduction systems. These problems notwithstanding, it seems likelythat the initial application of biochips as advanced biosensors, based on ligand–receptorcontact interaction, may be in the clinical setting, where they could serve as automatedcontrol devices for drug delivery. It also appears probable that in order to extend thecapabilities of present silicon-only systems, hybrid biochips and silicon-chip deviceswould be first produced for computing and memory-intensive systems. The key to thesuccessful application of biochips will be to fill places that are not well served by currentsilicon-chip technology. Thus, the future development of biochips requires the successfultechnologies of stable biomolecule immobilization, biotransduction, and molecular litho-graphy. Urgent problems to be solved are biologically based amplification, molecularswitching, electron transport, and memory function. In the coming decade, the answersto some of these problems will undoubtedly be found. In this connection, the develop-ment of lipid bilayer-based sensors and biological electronic devices seems to be a logicalfirst step. With the BLM system, especially s-BLMs, we now have an experimentalapproach for testing new ideas in the development of sensors for practical applications[4,14,15].

V. ADVANCES IN SELF-ASSEMBLED LIPID BILAYER-BASEDBIOSENSORS

In the last decade or so, there have been a number of reports on self-assembled moleculesor structures described as advanced materials or smart materials. Without question, theinspiration for this exciting work comes from the biological world, where the lipid bilayerof cell membranes plays a pivotal role. Past and recent achievements in self-assembledlipid bilayers as biosensor will now be described below.


A. Incorporation of Ferrocene on s-BLMs

To test the versatility of s-BLMs as a ‘‘smart material,’’ an amperometric sensor wasconstructed for ferri-/ferro-cyanide ions. The results have shown that: (1) ferrocene canbe very easily immobilized in the lipid bilayer on the tip of a metallic wire (s-BLM) system,and (2) ferrocene in a s-BLM on a Pt support increases the sensitivity by about two ordersof magnitude to potassium ferri-/ferro-cyanide ion than that of a bare pt electrode.Recently, substituted ferrocenes were incorporated into s-BLMs on a Pt support andinvestigated using the CV. These demonstrate that the s-BLM system offers a novelapproach to electrode modification by the simple incorporation of compounds within alipid bilayer [11,12,16].

B. Hydrogen Peroxide-Sensitive s-BLMs

The embedding of appropriate active molecules (modifiers) into the matrix of the lipidbilayer should be able to impart the functional characteristics of s-BLMs. We choseTCNQ (tetracyanoquinodimethane) and DP-TTF (dipyridyltetrathiafulvalene) as modi-fiers because of their properties as typical electron acceptor and donor molecules, respec-tively. It was found that DP-TTF could not only improve the stability, but also increasedthe range of the s-BLM’s sensitivity to hydrogen peroxide. In contrast, TCNQ-containings-BLMs showed little response to H2O2. This was not entirely unexpected since TCNQshould behave as an electron acceptor [5,17]. If highly conjugated compounds such asTCNQ are incorporated in the s-BLM forming solution, the resulting s-BLM is able todetect the presence of ascorbic acid, which is consistent with the findings obtained withconventional BLMs [5,11,27]. Concerning electron mediators, Cheng et al. [36] havereported a potential dependence of transmembrane electron transfer across a phospholipidBLM mediated by ubiquinone 10, whereas Yamada et al. [37] have investigated electrontransfer with three different redox couples through a BLM-containing TCNQ using a.c.impedance spectroscopy (see also Ref. 38).

C. Modified s-BLMs as pH Sensors

Of all the ions crucial to the functioning of cellular processes is the hydrogen ion (Hþ),which plays the leading role in enzyme catalysis and membrane transport. Thus, it is notsurprising that the measurement of pH is of the utmost importance. Currently, the pHglass electrode is routinely used in chemical and clinical laboratories. However, the largesize and fragility of pH glass electrodes preclude their use in many situations such as invivo cell studies and in monitoring membrane boundary potentials. For example, thehydrolysis of membrane lipids by phospholipid enzymes (lipases A and C) changes theboundary potential of a BLM (or cell membrane) as a result of local pH change.Additionally, it has been known for many years that BLMs formed from chloroplastextracts exhibit Nernstian behavior as a function of pH [3,4]. These observations suggestthat s-BLMs can be used as pH probes in membrane biophysical research and in biome-dical fields where the conventional glass electrode presents many difficulties. To test ourconcept, we incorporated a number of quinonoid compounds (chloranils) into s-BLMs.We have found that, indeed, s-BLMs containing either TCOBQ (tetrachloro-o-benzoqui-none) or TCPBQ (tetrachloro-p-benzo-quinone) responded to pH changes with a nearlytheoretical slope (55 � 3 mV) [16,27,30]. This new pH-sensitive s-BLM offers prospects forligand-selective probe development using microelectronic technologies.


D. s-BLMs as Ion Sensors

1. Modified s-BLMs as Ion Sensors

s-BLMs containing six different kinds of crown ethers were synthesized and investigatedusing the CV [5,19,30]. In particular, s-BLMs formed from a liquid crystalline aza-18-crown-6 ether and cholesterol-saturated n-heptane solution were found to be sensitive toKþ in the concentration range 10�4 to 10�1 M with theoretical Nernstian slope. Thespecificity for three alkali metal cations and NHþ4 of five different kinds of bis-crownethers in BLMs were investigated. The order of specificity for most of these bis-crownethers was found to follow the hydrated radii of cations, i.e., NHþ4 >Kþ>Naþ>Liþ. Theresults obtained with these s-BLMs compare favorably with conventional BLMs contain-ing similar compounds such as valinomycin [5,16,17].

2. Ion Sensors

Sato et al. [37] reported c-BLMs embedded with ionophores that selectively induce mem-brane permeability changes on binding analyte ions at the membrane/aqueous solutioninterfaces. A variety of ionophores for various metal ions, such as valinomycin, a lipo-philic derivative of 2,20-bipyridyl, and dihexadecyl phosphate, were directly incorporatedinto planar BLMs formed by two techniques, i.e., the folding and tip-dip methods. Theselective changes in transmembrane ion permeability were generally observed on com-plexation of respective primary ions with membrane ionophores, which were monitoredas membrane conductance changes. It was found that the primary ion-induced membraneconductance changes were due to permeation across the ionophore-incorporated BLMs ofnot only the primary ions but also their counteranions such as Cl� and ClO�4 . Themembrane conductance thus observed was discussed in terms of selectivity and the extentof permeability change per unit ionophore concentration in the BLM.

3. Supported BLMs for Urea Detection

Several groups of researchers have proposed a planar conductometric urea sensor basedon a microfabricated interdigitated electrode array [16,30]. Their biosensors are based onan interdigitated structure from platinum, gold, or a silver–palladium paste. The support isusually a glass or ceramic plate. Urease is immobilized on the surface of these electrodes ina membrane, cross-linked with bovine serum albumin by glutaraldehyde. Once immobi-lized on the surface of this device, the enzyme splits urea into ammonia and carbondioxide. These compounds are ionized in an aqueous medium, thereby increasing thesensor’s conductivity. This way of measuring seems to be the most convenient and, dueto its simplicity and efficiency, it is a preferred method over other detection techniques forthe urease reaction (measuring of NH3, NH4

þ, CO2, or pH with special electrodes) [5,6,8].Changes in the ion concentration give rise to a combination of changes in the double-layerstructure and in bulk electrolyte resistance. This result is an apparently concentration-dependent cell constant; in fact, the cell constant cannot be calculated in this case. It isworthwhile, therefore, to repeat the measurements with a larger conductometric cell. Thismight lead to a more suitable disposable biosensor for urea determination, especially in thewhole blood obtained from the finger tip of patients with anticipated renal failure[6,16,30].

Using the interdigitated arrays (IDAs) described above, the conductivity cells consisteither of the IDA of microelectrode pairs (width/gap of 10 �m) or the compact twoelectrodes of area 0.36 mm2 (gap of 300 �m). Conductivity microelectrode cells were tested


in different KCl solutions, from 0.1 mM to 1 M KCl, at an applied external voltage of 100mV and frequencies from 1 to 50 (100) kHz. On the basis of the results obtained one canconclude that further research of the conductance or impedance sensor is necessary. Onlyin such a way can one improve the experimental setup for impedance measurements aswell as carry out a systematic impedance analysis of a series of IDA thin-film structures ofwhich both the finger width and the interelectrode gap are varied. An optimization of theIDA can lead to use of the conductivity-based sensor for urea monitoring in physiologicalmedia. It makes the concentration range narrower (10–100 mM KCl) and allows largercharacteristic dimensions of the IDA (10–100 �m) [6,16,30].

E. Modified s-BLMs as Molecular Sensors

Many authors have reported sensors for the detection of glucose using glucose oxidase[16,30]. Interestingly, using s-BLMs containing redox compounds and electron mediatorsbut without the enzyme, glucose was detected in buffered solution. The results are pre-liminary and further experiments are in progress. If highly conjugated compounds such asTCNQ are incorporated in the s-BLM forming solution, the resulting s-BLM was able todetect the presence of ascorbic acid, which is consistent with the findings obtained withconventional BLMs [36]. From a technical point of view, transducers for use in biosensorscan be divided into four categories: electrochemical, semiconductive, optical, and otherssuch as piezoelectric. We shall be concerned only with the electrochemical category here,which consists of potentiometric and amperometric approaches. Earlier, we have reportedthe embedding of glucose oxidase in a polypyrrole–lecithin BLM with good results [40]. Inconnection with the conducting polymer polypyrrole, mention should be made of the workreported recently by Albers et al. [41] who have prepared extended di-(4-pyridyl)thiopheneoligomers by different methods. The authors suggest that these compounds are useful as‘‘molecular wires’’ and should be of interest in amperometric biosensors in connectionwith redox enzymes.

In the case of the glucose sensor, the potentiometric approach has been less success-ful than the amperometric one. The steady-state current for amperometric glucose sensorsis largely determined by the effective membrane thickness and the concentration of theembedded enzyme. We have tested a glucose sensor by embedding glucose oxidase (GOD)in avidin on s-BLM formed from biotinylated phospholipids. Avidin conjugation withGOD was achieved with glutaraldehyde according to the procedure of Rivnay et al. [42].Essentially, after biotinylated s-BLM formation, the coated wire tip was immersed in a 2.7�M avidin–GOD solution to allow the coupling between avidin–GOD and s-BLM toestablish. This was evidenced by a current reduction of about one order of magnitude.When the glucose was added to the cell, an increase in redox current was observed, whichwas a function of the applied voltage having a maximum at +670 mV. This avidin–GODcomplexed biotinylated s-BLM sensor was capable of detecting glucose with a linearresponse up to 9 mM. Since the lipid bilayer is ‘‘liquid crystalline’’ with a self-sealingproperty, the presence of pinholes (defects) seems unlikely. Hence, a credible explanationfor this glucose sensor would be based on electron transfer from the enzyme GODembedded in the lipid bilayer to the metallic substrate, where the major barrier for theelectron pathway is most likely to exist. One way to test this hypothesis is by incorporatingelectron carrier (mediator) in the lipid bilayer phase. Recently, we have found that full-erenes C60, C70, etc., serve this purpose very well (see below) [43]. Experiments with C60-doped s-BLM glucose sensors will be tried in due course.


Many researchers have reported sensors for the detection of molecular speciesbesides glucose such as antigens and antibodies. For example, Eray et al. [44] havereported a highly stable BLM formed in a microfabricated polyimide aperture containingnicotinic acetylcholine receptor modified with antibody, which was used for the detectionof specific antigen. Recently, Knoll and associates [45] reported that the dimer species(MW 580,000) of the nicotinic acetylcholine receptor, isolated from the electric organ ofTorpedo californica, was incorporated into a thiopeptide supported lipid bilayer. Theincorporation was achieved by fusion of liposomes with reconstituted receptor on to agold-supported thiopeptide lipid monolayer. Surface plasmon resonance spectroscopy(SPS) was used to monitor in real time the fusion process as well as the specific bindingof the antagonist cu-bungarotoxin. A recently developed extension of SPS offeringenhanced sensitivity and specificity, surface plasmon fluorescence spectroscopy (SPFS),was then used to monitor subsequent binding of the monoclonal WF6 and polyclonalantibody, respectively. The latter was fluorescence labeled with Cy5. The different bindingassays indicated the successful incorporation of the receptor in the lipid bilayer.

In this connection, we have reported in a feasibility study of an antigen–antibodyreaction using s-BLMs as biosensors with electrical detection [43]. The antigen (HBs-Ag—hepatitis B surface antigen) was incorporated into a s-BLM, which was then interactedwith its corresponding antibody (HBs-Ab—monoclonal antibody) in the bathing solution.This Ag–Ab interaction resulted in some remarkable changes in the electrical parametersof s-BLMs. The magnitude of these changes was directly related to the concentrations ofthe antibody in the bathing solution . The linear response was very good, ranging from 1to 50 ng/ml of antibody, demonstrating the potential use of such an Ag–Ab interaction viathe s-BLM as a transducing device [43].

F Electron-Transfer Experiments in s-BLMs

The early experiments in the field of electron-transfer processes in BLMs were first con-ducted in the late 1960s to understand the primary step in natural photosynthesis [3,4]. Itwas discovered that a light-driven electron-transfer process between donor and acceptorspecies can occur across the thickness of a pigmented bilayer lipid membrane. This findinghas subsequently led to the view that the reaction center of natural photosynthesis func-tions in a similar way to that of a photovoltaic device of molecular dimensions. In the mid-1980s, electron transfer in the dark was seen in BLMs doped with either organic ‘‘metals’’or semiconducting nanoparticles formed in situ. These phenomena were explained in termsof light-induced charge separation, field-driven charge transport, and subsequent redoxreactions on opposite sides of the BLM. In the absence of light, the theory of electrontunneling was invoked (see above on TCNQ- or TTF -containing BLMs). When a s-BLMdoped with Zn–phthalocyanine was excited by light, a voltage and a current wererecorded, with the action spectra closely paralleling that of the absorption spectrum ofthe photoabsorber [4,5,20,21]. Thus, we have shown that a pigmented s-BLM can functionas a light transducer or photon-activated switch or detector. In this connection, Biancoand Haladjian [46] have reported electron-transfer reactions between c-types of cyto-chromes and a lipid-modified electrode.

G. s-BLMs Deposited on Piezoelectric Quartz Crystals

Smell and taste (olfaction and gustation) are among living organisms two most vitalsensing systems, the biophysics of which have been increasingly elucidated at the mole-


cular level [2,4]. Here again the crucial receptors are BLMs. In our preliminary experi-ments, several kinds of BLMs were successfully deposited on AT-cut quartz resonators.These were verified by observing frequency (fm), potential (Em), capacitance (Cm) and I=Vcurves. Frequency change (versus that in air) ranged from 9 to 16 kHz, and no redox peakscould be observed or the peak was largely damped in the presence of Fe(CN)6

3-. Em andCm also showed characteristic values, but the exact values of these parameters were foundto be related to the lipid solution, the pH of the bathing solution, and the scan time ofvoltammograms. If the BLM failed to form or broke, obvious changes in these parameterswere observed. In this case, fm increased several kHz (frequency decreased to about 6 kHz,which corresponds to that induced only by pure viscous loading); Cm and Em also largelyincreased and characteristic redox peaks were observed. Our findings show that BLMs canbe formed on piezoelectric quartz crystals and that piezoelectric techniques can be appliedas a powerful tool for characterizing the s-BLM system. Concerning gustation, Hayashi etal. [47] have reported the effect of a bitter-tasting chemical, picric acid, on a BLM formedon a silicon wafer containing a single pore.

1. Minisensor for Screening of Sucralose

Nikolelis et al. [48] reported that the interactions of sucralose with s-BLMs producedincreases in electrochemical ion current which appeared to be reproducible within a fewseconds after exposure of the membranes to the sweetener. The mechanism of signalgeneration was found to be associated with alteration of the electrostatic fields of thelipid film. These studies revealed that an increase in the molecular area of the lipids atthe membranes and stabilization of the gel phase structure occurred due to adsorption ofthe sweetener. Water molecules are adsorbed at the polar head-groups of the lipids, whichchanges the electrostatic field at the surface of the membranes. The current signal increaseswere related to the concentration of sucralose in bulk solution in the micromolar range. Itis claimed that the present BLM-based sensor provided a fast response (i.e., in the order ofa few seconds) to alterations in sucralose concentration (5–50 �M) in electrolyte solution.The electrochemical transduction of the interactions of this artificial sweetener with s-BLMs was applied in the determination of the compound in granulated sugar substituteproducts by using the present minisensor. In a related study, Nikolelis and Pantoulias [47]described a minisensor for the rapid and sensitive screening of acesulfame-K, cyclamate,and saccharin based on surface-stabilized s-BLMs.

2. s-BLMs for Detecting Hybridization of DNA Oligomers

Krull et al. [50] reported the ion currents through s-BLMs on Ag wire that were used tomonitor hybridization of 20 mer and 25 mer oligonucleotides (single-stranded DNA oli-gonucleotides that were modified by attachment of a 16-carbon aliphatic chain). Theresults indicated that hybridization could be detected for mixed-base sequences, and forpartially complementary sequences. Quantitative results are dependent on the degree ofsurface occupancy by DNA, on the degree of complementarity of sequences, and on thebase sequence within the oligonucleotides.

3. Supported BLMs and DNA

Zhang et al. [51] reported the interaction of DNA (calf thymus) with hemin using CV in asb-BLM system. The sb-BLM was modified with lauric acid (LA) dissolved in the mem-brane-forming solution, then hemin molecules in electrolyte could also be embedded into


the BLM by electrostatic interaction between hemin and LA. Hemin showed a well-defined CV behavior. The cathodic peak current (I-Pc,) of hemin decreased in the presenceof DNA, which is consistent with the results of decreasing hemin concentration andillustrates the existence of an interaction between hemin and DNA.

4. Peptide t-BLMs

Naumann et al. [52] reported a new class of solid supported membranes tethered to thesupport by a peptide spacer. They are referred to as peptide-tethered lipid membranes (t-BLMs), formed by the fusion of liposomes with a thiopeptide–lipid monolayer chemi-sorbed on a gold support. Peptide t-BLMs are designed as a biomimetic system forinvestigating integral membrane proteins. As an example, cytochrome c oxidase (COX)from bovine heart is incorporated into the preformed peptide t-BLM by dilution of thesolubilized protein below the critical micellar concentration. The formation of the lipidfilm as well as the incorporation of the protein were monitored by SPS and SPFS. COX isactivated by adding the reduced form of cytochrome c to the air-saturated buffer solution.Using electrochemical techniques, such as square-wave voltammetry (SWV) and chron-oamperometry (CA), direct electron transfer between COX and the gold electrode isobserved as well as proton transport from the inside to the outside across the lipid bilayer.Proton transport is then further investigated using impedance spectroscopy, although theelectrode is shown to be only partially (70%) covered with a bilayer while there are defectdomains where only a monolayer of peptide or peptide–lipid coexist (�30%). Protontransport carried out by COX is shown to be voltage dependent. This transport is indi-cated as a resistance in parallel to the resistance of the lipid film. As a consequence, thetotal resistance decreases as a function of the concentration of cytochrome c and increasesagain either by removal of the substrate or by addition of cyanide as an inhibitor of COX.The conductance in the presence of the activated enzyme correlates with the known turn-over rate of COX. These experiments demonstrate the possibility of assessing the activityof integral membrane proteins incorporated in peptide t-BLMs by using electrochemicaltechniques. According to the authors, the system could thus be promising for screening aswell as biosensor applications.

In this connection, Raguse et al. [53] reported the formation of t-BLMs with an ionicreservoir. Self-assembled monolayers of reservoir-forming lipids were first adsorbed on tothe gold surface by using gold–sulfur interactions, followed by formation of the t-BLM.The properties of the t-BLM were investigated by impedance spectroscopy. The capaci-tance of the t-BLM indicated the formation of bilayer membranes of comparable thicknessto a solvent-free c-BLM. Also, the ionic sealing ability was comparable to those of classi-cal BLMs. The function of the ionic reservoir was investigated using the potassium-specificionophore valinomycin. Increasing the size of the reservoir by increasing the length of thehydrophilic region of the reservoir lipid or laterally spacing the reservoir lipid resulted inan improved ionic reservoir. Imposition of a d.c. bias voltage during the measurement ofthe impedance spectrum affected the conductivity of the t-BLM.

In this connection, Mirsky et al. [54] earlier reported antigen–antibody reactions onalkylthiol films on gold electrodes. These lipid layers are stable at neutral pH and displaypure capacitive behavior at frequencies around 20 Hz. Various reagents, including succi-nimides, thionyl chloride, p-nitrophenol, and carbodi-imides, were used to activate thecarboxy groups of the adsorbed monolayer of !-mercaptohexadecanoic acid.Glutaraldehyde, cyanuric chloride, and phenylene di-isocyanate were used to activatethe amino groups of the monolayer of !-mercaptohexadecylamine. The immobilization


of albumin on the activated surface was studied by capacitive measurements. The N-hydroxysuccinimide and carbodi-imide methods were identified as most suitable for pro-tein immobilization in that they did not compromise the insulating properties of thealkylthiol layer and led to maximal increase in its dielectric thickness. These approacheswere used for a layer-by-layer preparation of a capacitive immunosensor. Specifically,antibodies to human serum albumin were immobilized on the alkylthiol monolayer.Binding of the antigen led to a decrease in the electrode capacitance. According toMirsky et al. [54] the detection limit of the immunosensor is as low as 15 nM (1 mg/L).

5. s-BLM Based on Nonionic Surfactant

Karyakin et al. [55] reported the self-assembling of the uncharged nonionic surfactant Brij-52 on gold, and the resulting amphiphilic layers being characterized by means of impe-dance spectroscopy. Self-assembled Brij-52 layers have uniform structure and contain onlya minor number of defects. The capacitance of an optimal Brij-52 membrane, calculatedfrom the whole impedance spectrum rather than at a certain frequency, is 1:0� 0:4 �Fcm�2, which is close to the capacitance of the lipid bilayer. The uniform bilayers of Brij-52on the gold support are stated to be stable, in contrast to lipids, which provide theopportunity of their application for analytical purposes. In this connection, Nassar etal. [56] reported salt and pH effects on the electrochemistry of myoglobin in thick filmsof a bilayer-forming surfactant (didodecyldimethylammonium bromide) (DDAB).Protonation of (aquo)metmyoglobin [MbFe(III)—H2O] in these films precedes electrontransfer from electrodes, causing formal potentials to shift to negative values as the pHincreases from 5 to 8. At pH > 8, MbFe(III)–H2O dissociates to MbFe(III)—OH, whichis reduced directly at the electrode at rates higher than those MbFe(III)—H2O.Correlations of voltammetric data with FT–IR spectra suggested that at pH < 4.6, anunfolded form of Mb resides in the films and is reduced directly. The concentration of saltin solution influences the electrochemical properties of Mb–DDAB films by its influenceon Mb conformation and by its effects on interfacial Donnan potentials. NMR indicatedstrong binding of anions to Mb within DDAB films. Bound anions may neutralize positivecharge on the Mb’s surface so that it can reside in a partly hydrophobic environment, aspostulated on the basis of previous ESR and linear dichroism studies.

6. Electrochemical Transduction by s-BLMs

Siontorou et al. [57] reported that the interactions of Aflatoxin M-1 with s-BLMs formedfrom egg phosphatidylcholine produced ion current increases which reproduciblyappeared within �8–10 s after exposure of the lipid membranes to the toxin when usinga stirred solution. The magnitudes of the current signals were related to the toxin con-centration, which could be determined within the range 1.9–20.9 nM. In another series ofexperiments, Aflatoxin M-1 was found to affect the kinetics and time of signal generationdue to DNA hybridization, which was electrochemically monitored by using s-BLMs.According to the authors, alterations of electrochemical signals due to DNA hybridizationcan be used for rapid detection of this toxin. The ‘‘receptor’’ oligomer was single-strandeddeoxyribonucleic acid (ssDNA)–thymidylic acid icosanucleotide terminated with a C-16alkyl chain to assist incorporation into s-BLMs [dT(20)–C-16]. The target oligomer wasdeoxyadenylic acid icosanucleotide [dA(20)]; dT(20)-C-16 was incorporated into s-BLMsand complementary dA(20) (cDNA) was injected into the stirred bulk electrolyte solution.The electrochemical ion current across s-BLMs was found to increase due to the presenceof ssDNA and decrease due to the formation of double-stranded DNA. The toxin reduced


the initial rate of signal change and increased the time to reach equilibrium. This provideda means for the rapid (<1 min) and sensitive (detection limit 0.5 nM) detection ofAflatoxin M-1 based on measurements of the initial rate of hybridization.

7. Charge Transport by Membrane Proteins Embedded in s-BLMs

Seifert et al. [58] reported a method for the investigation of ion-translocating membraneproteins across BLMs. Protein-containing membrane fragments or vesicles are adsorbedon a solid supported membrane. Specific conductance and capacitance of the s-BLM arecomparable to those of a c-BLM. However, the s-BLM has the advantage of a muchhigher mechanical stability. Seifert et al. have also studied the electrical activity of bacter-iorhodopsin, Na,K-ATPase, H,K-ATPase, and Ca-ATPase in these s-BLMs. Accordingto the authors, the solid-supported BLM therefore represents an alternative method forthe investigation of electrical properties of ion-translocating transmembrane proteins. Inthis connection, Naumann [59] reported as-BLM containing Hþ-ATPase separated fromthe gold support by a peptide spacer. The translocation of protons across the lipid bilayerto the inner side is coupled with the discharge of protons at the gold surface. The overallprocess is investigated by SWV and double potential-pulse CA. As a result, the formationof a proton gradient is monitored by SWV whereas currents measured by CA monitor thestationary state when the enzyme activity is directly coupled with the charge transfer at theelectrode. These currents markedly depend on the number of ATPases present in thebilayer.

8. Some Electrochemical Features of s-BLMs

Gao et al. [60] reported the CV and electrochemical impedance spectroscopy of self-assembled, stainless steel supported s-BLMs, using a three-electrode system. The mem-brane resistance and capacitance calculated from the whole impedance spectrum ratherthan at certain frequencies were close to the conventional BLMs. The time course ofimpedance under certain applied frequencies was investigated. The results showed thatthe membrane capacitance of s-BLM fluctuated under low frequency and graduallyreached a constant value with the increase in applied frequency. Fullerene-doped s-BLMs were demonstrated to intensify this fluctuation behavior of membrane capacitance.A possible mechanism of this peculiar property of s-BLMs under low frequencies is dis-cussed in Ref. 43.

9. s-BLMs for Detecting Pathogenic Bacteria

Ivnitski et al. [61] reporteda new ion-channel biosensor based on a s-BLM for direct andfast detection of Campylobacter species. The sensing element is composed of a stainless-steel WE, which is covered by a self-assembled BLM. Antibodies to bacteria embedded inthe BLM are used as channel-forming proteins. The biosensor has a strong signal ampli-fication effect, which is defined as the total number of ions transported across the BLM.The total number of (univalent) ions flowing through the channels is 1010 ions s�1. Thebiosensor showed a very good sensitivity and selectivity to Campylobacter species.

10. Receptor–Ligand Interactions in BLMs

Wang et al. reported a simple method to reconstitute membrane receptors into c-BLMs.After reconstitution, the receptor still retains its ligand activity. Furthermore, the relation-ship between receptor–ligand interactions and electrical properties of reconstituted BLMs


such as membrane capacitance (Cm) and membrane resistance (Rm) was studied. Whenglycophorin in erythrocytes and asialoglycoprotein in hepatocytes were taken as examples,it was found that the resistance of reconstituted BLM decreased when adding blood-typemonoclonal antibody or solutions of galactose, respectively, and the decrease was ligandconcentration dependent; however, the membrane capacitance was not influenced. Thisprovides a simple, practical approach to determining the interactions between the receptorand its ligand [2,4,62].

In this connection, Costello et al. [63] reported improved techniques for the fabrica-tion of improved gel-protected bilayers. The gel protects the membrane from mechanicalcontact, from low-energy solvents, and from drying out. The new methods allow thepreparation in high yield of sensor cells with a front gel thickness less than 100 lm.During fabrication, gel-protected membranes of DOPC (dioleoyl phosphatidylcholine)readily incorporate valinomycin in functional form. DOPC membranes respond asexpected and within a few minutes to aqueous solutions of gramicidin applied to thefront gel layer. In long-term studies of protected membranes of the synthetic lipidPSPC, no failures were observed over runs of up to 3 weeks and the membrane leakagedecreased, reaching within 7 days values comparable to those reported for unsupportedBLMs. Initial attempts to incorporate functional nicotinic acetylcholine receptors haveshown promise.

Earlier, we described the effect of glucose and trehalose on the stability of s-BLMsformed on the freshly cut tip of Teflon-coated Ag wire [64]. Addition of saccharides to theelectrolyte resulted in a decrease in the elasticity modulus of the s-BLM formed while thecapacitance increased. In addition, the trehalose had a considerable stabilizing effect onthe above parameters of the s-BLM. Treatment of the s-BLM in an electrolyte containing0.3 M of the trehalose allowed storage of the s-BLM under dry conditions and underrefrigeration, with the subsequent recovery of membrane parameters after the wire hadbeen dipped into the electrolyte.

11. Photoelectric Effects in s-BLMs

Feng et al. [65] reported a new method based on photoelectric measurement for analyzingapoptosis of cell-free MCF-7 nucleoli. The supported s-BLM was used to enclose nucleoliin a biological environment. The s-BLM was self-assembled on the wall of a super-thincell. During the apoptosis induced by Taxol, the photoelectric current of the self-assembled s-BLM/nucleoli was found to decreasie with time, suggesting the degradationof nucleus DNA. Electron transfer along the DNA double helix and along the nuclearskeleton is assumed in the interpretation. This novel photoelectric analytical method mayprovide a rapid and sensitive technique to evaluate apoptosis. Further, the same groupreported [66] the photoelectric conversion properties of two self-assembled bilayer lipidmembranes (a BLM and a C60 fullerene containing BLM) on indium tin oxide (ITO)conducting glass. The influences of applied voltage, donor and receptor on the transmem-brane photocurrent as well as the facilitating effect of C60-doped BLMs on photoinducedelectron transfer across the BLM have been demonstrated (see Fig. 2). According to theauthors, this novel self-assembled ITO/BLM electrode may provide a simple and mechani-cally stable model for studying the photoelectric properties of biomembranes.

Related to this study is work reported by Jiang et al. [67] who used gold-supported s-BLMs, made of a octadecanethiol/phosphatidylcholine hybrid bilayer. They found that s-BLMs containing C60 can act as both a photosensitizer for electron transfer from a donormolecule and a mediator for electron transport across a lipid bilayer. The steady-state


photocurrent behaviors in different concentrations of ascorbate, Co(bpy)2þ=3þ3 , orFe(CN)4�=3�6 solution have been studied and it was found that the rate-limiting step ofthe whole photoinduced electron transfer depends on the applied potential and the redoxconcentration in solution. The sigmoidal-shaped steady-state photocurrent versus appliedpotential curve observed is different from the linear-shaped i(ph)/E curve of C60 modifiedBLM reported by Bensasson et al. [68].

VI. CONCLUDING REMARKS AND FUTURE PROSPECTS

The development of BLMs and later s-BLMs has made it possible for the first time tostudy, directly, electrical properties and transport phenomena across a 5 nm ultrathinbiomembrane element separating two interfaces. As a result of these extensive studies,biomembranes have now been recognized as the basic structure of Nature’s sensors andmolecular devices. For example, the plasma membrane of cells provides sites for a host ofligand–receptor contact interactions. To impart relevant biofunctions in BLMs, a varietyof compounds such as ionophores, enzymes, receptors, etc., have been incorporated. Someof these incorporated compounds cause the BLMs to exhibit nonlinear phenomena. Amodified or reconstituted BLM (or s-BLM) is viewed as a dynamic system that changes inresponse to environmental stimuli and as a function of time. This is best described by thedynamic membrane hypothesis as a basis of the biomembrane function. The self-assembledlipid bilayer, the basic component of biomembranes, is in a liquid-crystalline and dynamicstate. A functional biomembrane system based on self-assembled lipid bilayers, proteins,carbohydrates, and their complexes should be considered in molecular and electronicterms; it can facilitate both ion and electron transport and is the site of cellular activitiesin that it functions as a ‘‘device’’ for either energy conversion or signal transduction. Sucha system, as we know it intuitively, must act as some sort of a transducer capable ofgathering information, processing it, and then delivering a response based on this receivedinformation.

It should be reiterated that the research area covered in this chapter is highly inter-disciplinary. Emphasis has been placed on fundamental research. Our past work has beenbenefited by cross-fertilization of ideas among various branches of science. It seems likelythat the devices based on ‘‘smart’’ materials may be constructed in the form of a hybridstructure, e.g., utilizing both inorganic semiconducting nanoparticles and synthetic lipidbilayers. The biomimetic approach to materials science is unique and full of excitingpossibilities. In this connection, the rationale of our work in membrane biophysics hasbeen to understand the living organisms in physical and chemical terms. Hence, in a sense,we have been mimicking Nature’s approach to ‘‘smart’’ materials science or life, as weunderstand it, which may be summarized by one phrase trial and error. This approach wasfine for Nature but not a viable one for us now, since we do not have unlimited time andresources at our disposal. Nevertheless, we can glean the design principles from Nature’ssuccessful products and apply them to our search for better materials upon whichadvanced devices ultimately depend. So, the approach of our research is a biomimeticone. The success of our past work is evidenced by self-assembled lipid bilayers, photo-electric effects in pigmented BLMs, TCNQ-based BLM rectifiers, and, most recently,supported BLMs on interdigitated structures as biosensors by microelectronics techniques.

We now know a great deal about the structure of cell membranes, ‘‘ion pumps,’’electroporation, and membrane channels. In membrane reconstitution experiments, theevidence is that intracellular signal transduction begins at membrane receptors. The work


described here offers new and exciting opportunities for the preparation of a variety ofsupported lipid bilayer (BLM) probes with applications in membrane biophysics andbiotechnology. For example, the membranes can function in such important processesas electron transfer, signal transduction, and cellular environmental sensing. We haveestablished that supported BLMs (s-BLMs) are ‘‘smart materials’’ that can function asprobes in membrane biophysics and biotechnology. The work is invaluable in providingthe fundamental insight necessary for the design of biosensors and biomolecular electronicdevices. The s-BLM as a smart material is classified as an ultrathin film, which cantransduce a ligand–receptor contact interaction into an electrical response (i.e., changesin capacitance, potential, conductance, or dielectric breakdown or in voltammograms).For example, the rate of electron transport across BLMs by quantum tunneling is one suchphenomenon. Work is in progress experimenting on size quantization using s-BLM dopedwith semiconducting nanoparticles formed in situ. Supported BLMs or lipid bilayers areultrathin films of nanometer thickness, as such the interior of a BLM is strongly influencedby the close proximity of its interfaces. At the nanometer dimension, the quantum sizeeffects come into play. Thus, the properties of BLMs depend strongly on their constituentmolecules and interfaces. Remarkable changes in the electrical, optical and electromecha-nical properties are observed. This biomimetic approach to materials science research andbiotechnology will be pursued in the years to come.

REFERENCES

1. RC Burton. J Am Oil Chem Soc 45:201–224, 1968.

2. HT Tien, AL Ottova. J Membr Sci 189:83–117, 2001.

3. HT Tien. Bilayer Lipid Membranes (BLM): Theory and Practice. New York: Marcel Dekker,

1974, 672 pp.

4. HT Tien, A Ottova-Leitmannova. Membrane Biophysics: As viewed from Experimental

Bilayer Lipid Membranes (Planar Lipid Bilayers and Spherical liposomes). Amsterdam:

Elsevier Science, 2000, 648 pp.

5. A Ottova-Leitmannova, HT Tien. Prog Surface Sci. 41:337–445, 1992.

6. VH Tvarozek, HT Tien, I Novotny, T Hianik, J Dlugopolsky, W Ziegler, AL Ottova, J

Jakabovic, V Rehacek, M Uhlar. Sensors Actuators B: Chemi 19:597–602, 1994.

7. JM Kauffmann. Bioelectrochem. Bioenerg 42 (special issue):1–104, 1997.

8. YS Umezawa, K Kihara, N Suzuki, N Teramae, H Watarai. Anal Sci 14 (special issue):1–245,

1998.

9. CMA Brett, AM Oliverira-Brett. Electrochim Acta, 43 (special issue):1–234, 1998.

10. P Krysinski, HT Tien, A Ottova. Biotechnol Prog 15:974–990, 1999.

11. KL Mittal, DO Shah, eds. Surfactants in Solution. vol. 11. New York: Plenum Press, 1992,

pp. 61–88.

12. RF Taylor, JS Schultz, eds. Handbook of Chemical and Biological Sensors. Philadelphia,

PA:Institute of Physics, 1996.

13. R Guidelli, G Aloisi, L Becucci, A Dolfi, MR Moncelli, FT Buoninsegni. J Electroanal. Chem

504:1–28, 2001.

14. HT Tien, Z Salamon, P Kutnik, J Krysinski, J Kotowski, D Lederman, T Janas. J Mol

Electron 4:1–31, 1988.

15. FL Carter, RE Siatkowski, H Wohltjen, eds. Molecular Electronics Devices. Amsterdam:

Elsevier, 1988.

16. HT Tien, AL Ottova. Electrochim Acta 43:3587–3610, 1998.

17. M Seitz, E Ter-Ovanesyan, M Hausch, CK Park, JA Zasadzinski, R Zentel, JN Israelachvili.

Langmuir 16:6067–6070, 2000.


18. CH Lee, WC Lin, J Wang. Opt. Eng 40:2007–2014, 2001.

19. HT Tien. Bioelectrochem. Bioenerg 5:318–334, 1978; 9:559–570, 1982; 13:299–316, 1984;

15:19–38, 1986.

20. GJ Kavarnos. Fundamentals of Photoinduced Electron Transfer. New York: VCH

Publishers, 1993.

21. AG Volkov, DW Deamer, DL Tanelian, VS Markin. Liquid Interfaces in Chemistry and

Biology. New York: John Wiley, 1998.

22. HT Tien, AL Ottova. Curr Topics Biophys 25:1–50, 2001.

23. PB Contino, CA Hasselbacher, JB Ross, Y Nemerson. Biophys J 67:1113–1122, 1994.

24. A Lamrabte, JM Janot, E Bienvenue, G Miquel, P Seta. Bioelectrochem. Bioenerg 27:449–

463, 1992.

25. FT Hong. Prog Surface Sci 62:1–237, 1999.

26. M Sanger, H Sigrist. Sensors Actuators A: Phys 51:83–88, 1995

27. A Ottova-Leitmannova, AT Martynski, A Wardak, HT Tien. In: R Birge, ed. Molecular

Electronics and Bioelectronics. Adv Chem Series no. 240. Washington, DC: American

Chemical Society, 1994, pp 439–454.

28. HP Yuan, A Leitmannova-Ottova, HT Tien. Mater Sci Eng C—Biomim 4:35–38, 1996.

29. XD Lu, AL Ottova, HT Tien. Bioelectrochem Bioenerg 39:285–289, 1996.

30. A Ottova, V Tvarozek, J Racek, J Sabo, W Ziegler, T Hianik, HT Tien. Supramol Sci

4:101–112, 1997.

31. H Haas, G Lamura, A Gliozzi. Bioelectrochemistry 54:1–10, 2001.

32. S Kalinowski, Z Figaszewski. Meas Sci Technol 6:1043–1049, 1995.

33. PI Katrivanos, AJ Purnell, A Aleksandridis, C Siontorou, C White. Lab Robot Autom

10:239–246, 1998.

34. LQ Gu, A Leitmannova-Ottova, HT Tien. Encyclop Electr Electron Eng 10:475–493, 1999.

35. LQ Gu, LG Wang, J Xun, AL Ottova, HT Tien. Bioelectrochem Bioenerg 39:275–288, 1996.

36. YF Cheng, DJ Schiffrin. J Chem Soc, Faraday Trans 90:2517–2522, 1994.

37. H Yamada, H Shiku, T Matsue, I Uchida. J Phys Chem 97:9547–9552, 1993.

38. C Steinem, A Janshoff, HJ Galla, M Sieber. Bioelectrochem Bioenerg 42:213–220, 1997.

39. H Sato, H Hakamada, Y Yamazaki, M Uto, M Sugawara, Y Umezawa. Biosensors

Bioelectron 13:1035–1046, 1998.

40. J Kotowski, T Janas, HT Tien. J Electroanal Chem 253:277–282, 1988.

41. WM Albers, JO Lekkala, L Jeuken, GW Canters, APF Turner. Bioelectrochem Bioenerg

42:25–33, 1997.

42. B Rivnay, EA Bayer, M Wilchek. Methods Enzymol 149:119–123, 1987.

43. LG Wang, YH Li, HT Tien. Bioelectrochem Bioenerg 36:145–147, 1995.

44. M Eray, NS Dogan, SR Reiken H Sutisna, BJ Van Wie, AR Koch, DF Moffett, M Silber,

WC Davis. BioSystem 35:183–188, 1995.

45. J Sabo, A Ottova, G Laputkova, HT Tien. Thin Solid Films 306:112–118, 1997

46. YL Cheng, RJ Bushby, SD Evans, PF Knowles, RE Miles, SD Ogier. Langmuir 17:1240–

1242, 2001.

47. K Hayashi, K Yama-Uji, K Toko, N Ozaki, T Yoshida. Sensors Actuators 16:25–33, 1989.

48. DP Nikolelis, S Pantoulias, UJ Krull. Electrochim Acta 46:1025–1031, 2001.

49. DP Nikolelis, S Pantoulias. Biosens Bioelectron 15:439–444, 2000.

50. UJ Krull, DP Nikolelis, SC Jantzi, J Zeng. Electroanalysis 12:921–925, 2000.

51. YL Zhang, HX Shen, CX Zhang, A Ottova, HT Tien. Electrochim Acta 46:1251–1257, 2001.

52. R Naumann, A Jonczyk, R Kopp, J van Esch, H Ringsdorf, W Knoll, P Graeber. Angew

Chem, Int Ed Eng 34:2056–2058, 1995.

53. B Raguse, V Braach-Maksvytis, BA Cornell, LG King, PDJ Osman, RJ Pace, L Wieczorek.

Langmuir 14:648–659, 1998.

54. VM Mirsky, M Riepl, OS Wolfbeis, Biosens Bioelectron 12:977–989, 1997.

55. AA Karyakin, MY Vagin, MA Choba, NI Klyachko. Electroanalysis 11:1094–1097, 1999.

56. AEF Nassar, JF Rusling, TF Kumosinski. Biophys Chem 67:107–116, 1997.


57. CG Siontorou, DP Nikolelis, A Miernik, UJ Krull. Electrochim Acta 43:3611–3617, 1998.

58. K Seifert, K Fendler, E Bamberg. Biophys J 64:384–389, 1993.

59. R Naumann. Bioelectrochem Bioenerg 42:241–247, 1997.

60. H Gao, GA Luo, J Feng, HT Tien, AL Ottova. J Electroanal Chem 496:158–161, 2001.

61. D Ivnitski, E Wilkins, HT Tien, A Ottova. Electrochem Commun 2:457–460, 2000.

62. J Strutwolf, JA Manzanares, DE Williams. Electrochem Commun 1:139–144, 1999.

63. RF Costello, IR Peterson, J Heptinstall, DJ Walton. Biosensors Bioelectron 14:265–271, 1999.

64. T Hianik, J. Dlugopolsky, M Gyepessova. Bioelectrochem Bioenerg 31:99–111 1993.

65. J Feng, YX Ci, CY Zhang, AL Ottova, HT Tien. Photoelectr Electrochem Commun

1:145–147, 1999.

66. JS Ye, A Ottova, HT Tien, FS Sheu. Electrochem Commun 3:580–584, 2001.

67. DL Jiang, P Diao, RT Tong. Bioelectrochem Bioenerg 44:285–288, 1998.

68. RV Bensasson, JL Garaud, S Leach, G Miquel, P Seta. Chem Phys Lett 210:141–146, 1993.


16Bioelectrocatalysis

KENJI KANO and TOKUJI IKEDA Kyoto University, Kyoto, Japan

I. INTRODUCTION

Enzymes are biocatalysts that are essential for the acceleration and regulation of the ratesof a huge number of reactions involved in the metabolism of living cells. Because of thephysiological role of enzymes, they have novel properties of substrate specificities and highcatalytic efficiencies, allowing each of them to function in a specific biological reactionunder such mild reaction conditions as atmospheric pressure, temperatures around 208 to408C, and pH values near neutrality. Oxidoreductases are a group of enzymes catalyzingbiological redox reactions and have received considerable attention in connecting theirreactions with electrochemical reactions, since the connection has been expected to open anew one for applying the enzymes in a variety of fields as well as for studying fundamen-tals of the oxidoreductase reactions [1,2]. The electrochemical connection has been realizedby coupling the enzymatic reactions with electrode reactions of redox compounds, calledelectron-transfer mediators, which shuttle electrons between the enzymes and the electro-des [3,4]. Bioelectrocatalysis is the term expressing the enzyme-catalyzed acceleration ofelectrochemical reactions of substrates [1,2], and more fully, the word mediated bioelec-trocatalysis is used to describe the current enhancement by the enzyme–electrochemicalreactions with mediators. In the last two decades, the area relevant to bioelectrocatalysishas rapidly grown. In the mid-1980s, a growing demand for developing second-generationamperometric biosensors promoted extensive studies of bioelectrocatalysis, because it wasthe working principle of the biosensors [3,5]. Since then, a great number of papers haveappeared dealing with mediated bioelectrocatalysis with emphasis on the applied aspect.In recent years, bioelectrocatalysis has proved to be very useful for studies in the fields offundamental and applied electrochemistry, biochemistry, and biotechnology.Bioelectrocatalysis is now recognized as a key reaction for developing not only novelbiosensors but also bioreactors and biofuel cells and also for understanding the kineticsand thermodynamics of oxidoreductase reactions.

A typical electron flow in mediated bioelectrocatalysis is illustrated in Fig. 1. Theprinciple is described by a combination of an enzyme-catalyzed reaction with an electrodereaction. The oxidoreductase reaction obeying the ‘‘ping-pong’’ mechanism is written by

SðredÞ þ Eox$k1

k�1ES �!k2 PðoxÞ þ Ered ð1aÞ

Mox þ Ered$k3

k�3EM �!k4 Mred þ Eox ð1bÞ


where S and P are the substrates and product, Eox and Ered are the oxidized and reducedforms of the enzyme, and Mox and Mred are an electron acceptor and its reduced form,respectively; k1,k�1, k2, k3, k-3, and k4 denote the rate constants of the respective steps. Thesteady-state kinetics of the enzyme reaction (�E) is expressed by

�E ¼kcat½E�

1þ KM=½Mox� þ KS=½S�ð2Þ

where [E] is the concentration of the enzyme, kcat ½¼ k2k4=ðk2 þ k4)] is the catalytic con-stant, and KM½¼ k2ðk�3 þ k4Þ=k3ðk2 þ k4)] and KS½¼ k4ðk�1 þ k2Þ=k1ðk2 þ k4)] are theMichaelis constants of Mox and S, respectively. On the other hand, Mred and Mox undergoan electrode reaction:

Mred

k!

$k Mox þ ne� ð3Þ

where n is the number of electrons and k!

and k

are the electron-transfer rate constants. Inoxidative mediated bioelectrocatalysis, the thus-generated Mox takes part in the enzymereaction of Eq. (1b). As a result, the oxidation current of Mred is amplified by the enzymicreaction. Theory, methods, and applications of mediated bioelectrocatalysis are describedin this chapter. Items included are bioelectrocatalysis at a solution/electrode interface(Section II), bioelectrocatalysis at enzyme-modified electrodes (Section III), microbialcell-based bioelectrocatalysis (Section IV), and an application of bioelectrocatalysis inbiofuel cells (Section V). The data given in these sections are mostly cited from our ownworks, as this is the simplest way of providing a systematic explanation of bioelectroca-talysis.

II. BIOELECTROCATALYSIS AT A SOLUTION/ELECTRODE INTERFACE

This section treats the theory of homogeneous mediated bioelectrocatalysis in a quietsolution. An empirical equation explaining the catalytic current is presented, which isconveniently used for the determination of kinetic parameters of the enzyme reaction. Anovel method of protein redox potential measurements is also described using a mediatedcontinuous-flow column electrolytic spectroelectrochemical technique.

FIG. 1 Kinetic scheme of mediated bioelectrocatalysis.


A. Theory of Homogeneous Mediated Bioelectrocatalysis in QuietSolution

It is noted that the enzyme reaction is nonlinear with respect to both [Mox] and [S], Eq. (2).When conditions [Mox] � KM and [S] � KS are fulfilled, Eq. (2) is reduced to�E ¼ ðkcat=KMÞ½E�½Eox�. The mediated bioelectrocatalysis under such conditions isdescribed by the established theory of the catalytic current [6]. Thus, the steady-statelimiting current Is,lim is written by

Is;lim ¼ nFA½Mred� ðDk0½E�Þ1=2 ð4Þwhere F and A are the Faraday constant and the electrode surface area, D and [Mred]

arethe diffusion coefficient and the bulk concentration of Mred, respectively, and k0 is the rateconstant given by kcat=KM [7,8]. The quantity kcat=KM is the bimolecular reaction rateconstant between the enzyme and Mox [kcat=KM ¼ k3k4=ðk3 þ k4)) and can be determinedfrom Eq. (4) by plotting Is,lim against [Mred]

. An experimental observation of a linear Is,limversus [Mred]

relationship confirms the required condition, [Mox] � KM.The Is,lim versus [Mred]

plot will deviate downward at higher [Mred] , reflecting the

nonlinear character of the enzyme kinetics given by

�E ¼kcat½E�½Mox�KM þ ½Mox�

ð20Þ

Is,lim under such conditions is approximately expressed by [9,10]

Is;lim ¼nFA½Mred�

1þ f

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2Dkcat½E�

2KM þ ½Mred� s

ð5Þ

where f is an empirical correction factor determined on the basis of a digital simulation:

f ¼ ð½Mred� =KMÞ0:85ð½Mred� =KMÞ2 þ 11:4ð½Mred� =KMÞ þ 17:9

ð6Þ

A nonlinear regression analysis can be applied to the relation between Is,lim and [Mred] to

evaluate kcat and KM separately. However, the parameter f may be assumed to be zero atthe first approximation; even with this assumption, the maximum error of Is,lim is onlyabout 5% around [Mred]

/KM � 4�5 [9]. Equation 5 can then be rewritten as a Michaelis–Menten type expression:

ðIs;limÞ2½Mred�

¼ 2ðnFAÞ2Dkcat½E�1þ 2KM=½Mred�

ð7Þ

Accordingly, several linear regression analyses such as Lineweaver–Burk type plots([Mred]

/Is,lim2 versus 1/[Mred]

) and Hanes–Woolf type plots fð½Mred� =Is;limÞ2 versus[Mred]

g are useful in obtaining kcat and KM values [9,10].Cyclic voltammetry is a method frequently used to measure Is,lim. Mediated bioelec-

trocatalysis yields cyclic voltammograms (CVs) of different shapes as illustrated in Fig. 2,depending on the measuring conditions [11]. Curve (a) is the wave for a reversible elec-trode reaction of an Mox/Mred redox couple. Bioelectrocatalysis mediated with the Mox/Mred redox couple produces a sigmoidal catalytic wave as curve (c) under the conditions[Mred]

� KM and [S]� KS. When [Mred] is increased to higher concentrations, an anodic

peak of the diffusion current of Mred may be overlapped on the catalytic current asdepicted by curve (d); the current, however, becomes steady state after appropriate periods


of the potential scan to give a plateau on the voltammogram. Thus, Is,lim values areobtained from the plateau regions on curves (c) and (d).

The above discussion assumes the condition [S] � KS. When the dependence of [S]on the catalytic current is to be studied, the bulk concentration of S, [S] , must be lowered.Under such conditions, however, no steady-state current is observed on CVs as shown bycurve (b). This is because the substrate depression occurs in the vicinity of the electrodesurface, and no steady state is attained. A digital simulation technique [12] would be themost straightforward way to analyze such nonsteady-state currents or [S] dependence ofthe catalytic current [13–15]. Substrate depression can be avoided when two enzymereactions are coupled in mediated bioelectrocatalysis in such a way that S in the firstenzyme reaction is regenerated from the product P by the second enzyme reaction tokeep the S/P ratio constant [16]. Under such conditions, the steady-state limiting currentcan be given by [16,17]

Is;lim ¼ nFA

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2Dkcat½E�½M�t1þ Ks=½S�t

sð8Þ

where [M]t and [S]t express the total analytical concentrations of the mediator and sub-strate, and it is assumed that the condition [Mox]� KM is fulfilled within almost the wholeregion of the diffusion layer of the mediator. Thus, Eq. (2) can be reduced to�E ¼ kcat½E�=ð1þ KS=½S�t), and the diffusion equation of Mox coupled with the [Mox]-inde-pendent enzyme reaction under steady-state conditions is given by DMðd2½Mox�=dx2Þ��E;ox ¼ 0, which can be solved to obtain Eq. (8). An example of two-enzyme-linked bioe-lectrocatalysis will be described below for the case when the cofactor, nicotinamide ade-nine dinucleotide (NAD+), is the substrate of the first enzyme reaction.

The Is versus E curve for a bioelectrocatalysis reaction exhibits a sigmoidal shape asgiven by Eq. (9) [e.g., see curve (c) in Fig. 2], which can be deduced on the basis of thereaction layer approximation [18]:

FIG. 2 Typical CVs representing homogeneous mediated bioelectrocatalysis. (a) Mred alone (n ¼ 1,

totally reversible case); (b)–(d) Mred + Enz + S; [Mred]* ¼ 0.1 mM (d: 2 mM); [S]* ¼ 100 mM (b: 0.7

mM), kcat[E] ¼ 1 mM s�1 (D: 0.1 mM s�1), KM = 1 mM, KS = 1 mM, A = 0.1 cm2 (d: 0.01 cm2).

Scan rate: 10 mV s�1. (From Ref. 11.)


Is ¼Is;lim

1þD=ð� ~kkÞ þ k

=k! ð9Þ

where � is the reaction layer thickness (� ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiDKM=kcat½E�

pat [Mred]

� KM and � ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiD½Mred� =2kcat½E�

pat [Mred]

� KM), and k!

and k

[Eq. (3)] are given, for instance, bythe Butler–Volmer equation:

k!

¼ k� expf�nFðE � E�MÞ=RTg ð10Þ

k!

¼ k� expf�ð1� �ÞnFðE � E�MÞ=RTg ð11Þwhere k8, �, and E8M are the standard rate constant, the anodic transfer coefficient, andthe standard redox potential of the Mox/Mred redox couple, respectively. When k

!

þ k

�ffiffiffiffiffiffiffiffiffiffiffiffiffiDM=�p

(this is the case where the mediator has high reversibility and/or the enzymticreaction is slow), Eq. (9) is reduced to:

Is ¼Is;lim

1þ expf�ðnF=RTÞðE � E�MÞgð12Þ

Equation (12) predicts that the sigmoidal curve is symmetrical with respect to E�M, as in thecase of the reversible polarogram or rotating-disk voltammogram, Fig. 2, curve (c). With adecrease in ðk

!

þ k

Þ= ffiffiffiffiffiffiffiffiffiffiffiffiffiDM=�p

, the half-wave potential (E1/2) of the sigmoidal curve shifts tothe direction of the positive potential, as in the case of the quasireversible polarogram orrotating-disk voltammogram, Eq. (9). Such phenomena are frequently encountered inmediated bioelectrocatalysis with increased catalytic activity.

B. Kinetic Analysis of Oxidoreductase Reactions

Typical examples of the bioelectrocatalytic current are shown in Fig. 3, where diaphorase(DI)-catalyzed oxidation of NADH and reduction of NAD+ are coupled with the elec-trode reaction of mediators, vitamin K3 (VK3) and alizarin red S (ARS), respectively [16].VK3 and ARS produce reversible waves as shown by curves (a) and (c), respectively.Additions of DI and NADH (to the VK3 solution) and DI and NAD+ (to the ARSsolution) cause large increases in the anodic [curve (b)] and cathodic [curve (d)] currents,respectively, producing sigmoidal waves with steady-state limiting currents. Thus, Is,limvalues are obtained from the limiting currents and Is,lim versus [VK3H] (or [ARS] ) plotsin the linear region allow the calculation of kcat/KM values by Eq. (4), where [VK3H] and[ARS] are the bulk concentrations of the reduced form of VK3 and of ARS, respectively.Here, the quantity kcat/KM represents the bimolecular reaction rate constant for oxidationof the enzyme with VK3, or the rate constant for reduction of the enzyme with the reducedform of ARS. In this manner, the kcat/KM values for a series of reactions between DI andredox compounds with different values of their standard redox potentials (E�M) are easilydetermined. Figure 4 reveals that logarithms of kcat/KM are in linear relations with the E8Mvalues when the difference between E8M and the standard redox potential of protein (hereDI) (E8P) is less than about 200 mV. The result indicates that the oxidation and reductionreactions of the enzyme with the mediators satisfy linear free-energy relationships, i.e. thatthe enzyme exhibits no specificity in the reactions with the mediators [16]. There are manyexamples to support a linear free energy relation in the oxidoreductase reactions withartificial electron acceptors [18,19]. Figure 4 also shows that the kcat/KM values for the


FIG. 3 CVs of (a) VK3 (0.097 mM), (b) (a) + DI (0.11 �M) + NADH (4.8 mM); (c) ARS (0.09

mM) and (d) (c) + DI (56 �M) + NAD+ (4.6 mM) at a bare glassy carbon electrode at pH 7.0.

Scan rate: 5 mV s�1. (From Ref. 16.)

FIG. 4 Logarithmic plots of the bimolecular rate constants for (*) the oxidation (kcat,ox/KM,ox)

and (&) the reduction (kcat,red/KM,red) of DI with mediators as a function of the formal potential of

the mediators (E�M0) at pH 8.5. MV: methyl viologen; ARS: alizarin red S; FMN: flavin

mononucleotide; FAD: flavin adenine dinucleotide; 2,6AQS: anthraquinone-2,6-sulfonate; VK3,:

2-methyl-1,4-naphthoquinone; NQ: 1,4-naphthoquinone, 2,6DMBQ: 2,6-dimethylbenzoquinone;

BQ: 1,4-benzoquinone; AP: p-aminophenol; FMA: ferrocenylmethanol. (From Ref. 16.)


oxidation reaction become constant when the difference between E�M and E�P is larger than200 mV, which is attributed to diffusion control limitation in the enzyme reactions [16].

In the DI-catalyzed electrochemical oxidation of NADH (or reduction of NAD+)mentioned above, relatively large amounts of NADH (or NAD+) were used. Under suchconditions, the catalytic waves show steady-state characteristics in most cases. However,when the total analytical concentration of NADH, [NADH]t, is decreased below theMichaelis constant, KNAD,ox, the catalytic voltammogram has a peak shape followed bya diffusion tail-like current decrease as shown by curve (f) in Fig. 5. This is the case forcurve (b) in Fig. 2 and is ascribed to the substrate depression occurring in the vicinity ofthe electrode surface. On the addition of L-lactate dehydrogenase (LDH) and an excess ofL-lactate, the catalytic voltammogram changes to steady-state characteristics, as shown bycurve (e) in Fig. 5. This is due to the coupling of the LDH reaction (L-lactate + NAD+,pyruvate + NADH + H+) with the DI-catalyzed NADH oxidation; the coupling allows[NADH] in the vicinity of the electrode surface to be kept at [NADH]t during the catalyticreaction. In the DI-LDH-linked system, the steady-state catalytic waves are well recog-nized at low [NADH]t down to at least 10 �M. The steady-state limiting current (Is,ox)increases parabolically with increasing [NADH]t as shown by curves (b)–(e) and the insetin Fig. 5. The [NADH]t dependence of Is,ox is well reproduced by Eq. (8) (where [E], KS,and [S]t should be read as [DI], KNAD,ox, and [NADH]t, respectively). The nonlinearregression analysis yields KNAD,ox ¼ 0.93 mM and kcat ¼ 740 s�1, which are close tothe values separately determined from conventional kinetic measurements in the bulkphase (KNAD,ox ¼ 0.54 mM and kcat ¼ 1370 s�1) [16].

To evaluate kcat andKM values separately, bioelectrocatalysis data at higher mediatorconcentrations are needed. The bioelectrocatalytic reaction of bilirubin oxidase (BOD)

FIG. 5 CVs of (a) VK3 (9.8 �M) in the presence of DI (0.26 �M), lactate dehydrogenase (64 unit

mL�1) and L-lactate (47 mM), (b) (a) + NADH (19 �M), (c) (a) + NADH (38 �M); (d) (a) +

NADH (56 �M), and (e) (a) + NADH (98 �M) at a bare glassy carbon electrode at pH 7.0; scan

rate: 5 mV s�1. Curve (f) was taken under conditions similar to those for curve (e), but in the absence

of L-lactate. The inset shows the steady-state catalytic limiting current as a function of [NADH]t in

the presence of VK3 (9.8 �M), DI (0.26 �M), lactate dehydrogenase (64 unit mL�1), and L-lactate

(47 mM). (From Ref. 16.)


with 2,20-azinobis- (3-ethylbenzothiazolin-6-sulfonate) (ABTS2�) as a mediator is taken asan example; the bioelectrocatalysis system allows the reduction of O2 to H2O. The combi-nation of BOD and ABTS2� is very effective for reducing O2 to H2O [20] as is evident fromFig. 6, where the bioelectrocatalytic current for O2 reduction starts to appear at around0.58 V versus Ag|AgCl and reaches a steady-state limiting current at about 0.4 V. Thesigmoidal catalytic wave has a half-wave potential (E1/2) of 0.49 V; this is only 0.13 V morenegative than the standard redox potential of the reaction 1/2O2 + 2e� + 2H+!H2O atpH 7.0. The bioelectrocatalytic reaction rate is large as revealed by analysis of the depen-dence of Is,lim on the bulk concentration of the oxidized form of ABTS2�, [ABTS

�] , which

is equal to the initial concentration of ABTS2�, [ABTS2�]. The Is,lim versus [ABTS2�] plots(Fig. 7) deviate downward with increasing [ABTS2�] as predicted by Eq. (5) and allow thecalculation of the kcat and KM values to be 8:2� 102 s�1 and 11 �M, respectively [20]. Thekcat value is considerably large, and the bimolecular reaction rate constant (given by kcat/KM) has value of 7:5� 107 M�1 s�1, which is large enough to be close to the rate constantof a diffusion controlled reaction in enzyme kinetics (cf. Fig. 4). The result of the BOD–ABTS2� system reveals that it is one of the most effective electrocatalytic systems for thereduction of O2 to H2O. Application of the BOD-ABTS2 system in biofuel cells will bedescribed in Section V. The analysis based on bioelectrocatalysis has proved to be veryuseful in evaluating the kinetic parameters of oxidoreductases, and the evaluation is helpfulfor proper use of the enzymes in biosensors, bioreactors, and biofuel cells.

Kinetic analysis based on bioelectrocatalysis also provides a novel means of studyingoxidoreductase reactions in biochemistry. Quinohemoprotein amine dehydrogenase (QH-AmDH) is a newly found oxidoreductase from Paracoccus denitrificans (IFO 12442) [21]; itproduces QH-AmDH and at least three kinds of hemoproteins in the cytoplasmic spacewhen grown on n-butylamine. Bioelectrocatalysis experiments with QH-AmDH as anenzyme and the hemoproteins as mediators revealed that the combination of QH-AmDH with cytochrome c-550, one of the hemoproteins, produces a large catalytic cur-rent, but not with the other two hemoproteins. This is clear evidence for the role ofcytochrome c-550 as a natural electron acceptor of QH-AmDH; cytochrome c-550 med-iates the electron transfer from QH-AmDH to the terminal oxidase and constitutes a new

FIG. 6 CVs of (a) pH 7.0 phosphate buffer, (b) (a) + 0.25 mM ABTS2�, and (c) (b) + 11 �MBOD. Scan rate: 10 mV s�1. (From Ref. 20.)


pathway for the amine oxidation respiratory chain of P. denitrificans (22). The kcat and KM

values of the reaction between QH-AmDH and cytochrome c-550 are determined to be18 s�1 and 1:0� 10�7 M�1, respectively [22]. As in the present case, when both an enzymeand an electron acceptor have heme c groups as the redox centers and KM is very small, itis difficult to evaluate the enzyme kinetic parameters by conventional spectrophotometricmethods, because of spectral overlapping of the enzyme and mediator.

C. Redox Potential Measurements of Proteins and Enzymes

When substrates are removed from mediated bioelectrocatalytic systems, the reactions areconsidered to be indirect electrolysis of enzymes (or more generally proteins):

Pox þMred$kf

kbPred þMox ð13Þ

where kf and kb are the rate constants of the forward and backward reactions, respectively.After complete electrolysis (i.e., under the equilibrated state) by bulk electrolysis, theNernst equation is given by

E ¼ EoM þ

RT

nMFln½Mox�eq½Mred�eq

!¼ Eo

P þRT

nPFln½Pox�eq½Pred�eq

!ð14Þ

The equilibrated concentration ratio of the oxidized and reduced forms of a redox proteinf½Pox�eq=½Pred�eqð �P) may be obtained, e.g., by spectroscopy, then the E�P can be evaluatedby Eq. (14). This concept is utilized in the mediated optically transparent thin-layerspectroelectrochemical technique for the redox potential measurements of proteins [23].In potentiometric-spectroscopic titration as a related method, the solution potential ischanged by addition of a suitable oxidant or reductant. However, in these indirect meth-ods, [Mox] and [Mred] (as well as [Pox] and [Pred]) are the function of the time after theperturbation of E. This means that it may take a long time for equilibration, and thebackground subtraction is complicated when spectral overlapping of a protein and med-iator(s) occurs. This problem can be overcome by using a continuous-flow column elec-

FIG. 7 Plot of steady-state limiting current Is against [ABTS2�]; Is was measured at 0.2 V in 0.11

�M BOD solution of pH 7.0. (From Ref. 20.)


trolytic method coupled with a flow-injection method and spectrophotometric detection.The system is depicted in Fig. 8 [24], where [Mox] and [Mred] can be maintained at constantvalues and the background absorbance becomes independent of time. Therefore, E�P valuescan be easily evaluated [21,24,25].

Figure 9 shows the spectra of horseradish peroxidase (HRP) obtained with thismeasuring system [25]. Although the difference between the background and peak-toptotal spectra are small due to a large concentration ratio of the mediator IrCl2�=3�6 againstHRP (=212), both of the absorption spectra are reproducible and then a simple subtrac-tion gives the background-corrected spectra of HRP. The spectra are independent of thetotal flow rate up to 1.0 mL min�1 as well as the direction and the width of the potentialstep, which support the redox equilibrium between HRP and IrCl2�=3�6 . In a similarmanner, background-corrected spectra of a number of proteins and enzymes are obtain-able. Figure 10 demonstrates an example of spectra thus obtained [24]; the E�P value of asample of methylamine dehydrogenase is easily determined as �0:068 V versus Ag/AgClfrom the absorbance at 331 nm, which is attributed to the redox potential of tryptophantryptophylquinone in the molecule.

Since the reaction in the column can be expressed by

�E ¼ kf ½Eox�½Mred� � kb½Ered�½Mox� ð15Þat least at low concentrations of mediators, and [Mox]t ¼ [Mox]eq and [Mred]t ¼ [Mred]eq([Mox]eq/[Mred]eq �M), the reaction rate with the initial condition of [Pox]t=0 ¼ [P]o([Pox] + [Ped]) can be analytically expressed by

½Pox�t½P�o¼ exp �kf 1þ �Pð Þ½M�0t=ð1þ �MÞ

� �þ �P1þ �P

ð16Þ

with the half-life time (�1=2) for the equilibration:

�1=2 ¼�M þ 1

kf ½M�oð�P þ 1Þ ln 2 ð17Þ

Such kinetic considerations are very useful for optimization of the system and allow thekinetic evaluation of E�P [24].

In general, there is a risk in thermodynamic analysis in indirect methods. Figure 11shows [Pox]t/[P]o at several reaction times (t), as an example of the case that E�P > E�M. All

FIG. 8 Schematic diagram of a mediated continuous-flow column electrolytic spectroelectro-

chemical system. A protein sample is injected into the buffer flow and mixed with the mediator

solution equilibrated at the precolumn electrode, and then re-electrolyzed at the main-column

electrode. A photodiode array detector monitors the redox state of the protein. (From Ref. 11.)


profiles look to be Nernstian, but the apparent redox potential (E at [Pox]t/[P]o ¼ 1/2)deviates from E�P to the direction of E�M with a decrease in time. Therefore, special careshould be taken in confirmation of the equilibration. In this sense, kinetic analysis in thecolumn electrolytic method would be helpful to ensure E�P evaluation. Although someother kinetic methods are proposed for protein redox potential measurements [26,27],those are based on several assumptions and it would be risky to adopt them, as discussedin the literature [24].

III. BIOELECTROCATALYSIS AT ENZYME-MODIFIED ELECTRODES

Electrodes on which a redox enzyme(s) and an electron-transfer mediator(s) are co-immo-bilized are called (mediated) biocatalyst electrodes [1,2]. Figure 12 illustrates the kinetic

FIG. 9 Background-corrected three-dimensional spectra of flow-injection analysis peaks of

horseradish peroxidase (HRP) at E ¼ 0.550, 0.715, and 0.850 V versus Ag/AgCl. HRP,

3:2� 10�4 M (10 �L); K2IrCl6 in reservoir, 6:8� 10�4 M; total flow rate, 0.5 mL min�1. Theinset shows (A) a peak-top total spectrum and (B) a background spectrum at E ¼ �0:850 V,

where the difference between the two spectra is very small. Even in such a case, reproducible

background spectra are obtained. (From Ref. 25.)


scheme of a biocatalyst electrode. A homogeneously immobilized-enzyme/mediator layerof thickness l is confined to the electrode surface, usually by covering with a semiperme-able membrane of thickness lm. Mediated bioelectrocatalysis at biocatalyst electrodes isthe working principle of second-generation amperometric biosensors and has been exten-sively studied. This section describes the theory and some applications of mediated bioe-lectrocatalysis at biocatalyst electrodes. The theory of the catalytic current at a biocatalystelectrode will be helpful in constructing second-generation amperometric biosensors ofsuitable properties. Direct bioelectrocatalysis, which is a catalytic reaction based on directelectron transfer between an electrode and an enzyme immobilized thereon, is also men-tioned in this section. Direct electron transfer between an active site of a redox enzyme andan electrode provides a simple way of electrochemical control of biocatalytic reactions; the

FIG. 10 Background-corrected absorption spectra of the flow-injection analysis peak of

methylamine dehydrogenase (50 �M� 10 �L) in the presence of phenazine methosulfate (10 �M)

and phenazine methosulfate-2-sulfonate (40 �M). E = (A) 0 V, (B) �0:035 V, (C) �0:050 V, (D)

�0:060 V, (E) �0:070 V, (F) �0:080 V, (G) �0:095 V, (H) �0:110 V, and (I) �0:140 V. (From Ref.

24.)

FIG. 11 [Pox]t/[P]o versus E profiles at several reaction times (t) of a protein and a mediator

equilibrated to E. Parameters for calculation are kf[M]o ¼ 1.07 s�1, E�M ¼ E�p � 64 mV, [Pox]t=0

= [P]o; t/s = (a) 8, (b) 16, (c) 32, (d) 64, (e) 128, and (f)1 (equilibrium condition). (From Ref. 11.)


potential applied to the electrode controls directly the rate and direction of the biocatalyticreactions. Direct bioelectrocatalysis is expected to be a key reaction of third generationamperometric biosensors.

A. Steady-State Currents in Coimmobilized Systems

In the kinetic scheme of a biocatalyst electrode given in Fig. 12, the concentration polar-ization of the mediator occurs only within a finite thickness of the immobilized layer. Sincethe solution is usually stirred, the concentration polarization of the substrate will beneglected at the outside of the membrane and then the substrate polarization is restrictedwithin the immobilized layer and the membrane. As a result, the catalytic current reaches asteady state after a certain period of time in any case. The enzymic reaction in the immo-bilized layer with excess amounts of Mox is expressed by

�E ¼kcat½E�½S�KS þ ½S�

ð200Þ

A steady-state limiting current Is,lim is observed under steady-state conditions at E � E�M(DSðd2½S�=dx2Þ � �E ¼ 0 and DMðd2½Mred�=dx2Þ þ �E ¼ 0; DS is the diffusion coefficient ofS and is related to the concentration gradient as

Is; limnFA

¼ DM

d½Mred�dx

� �x¼0¼ DS

d½S�dx

� �x¼l¼ PS ½S� �

½S�x¼l�S

� �ð18Þ

where PS and �S are the permeability of the semipermeable membrane to S and thedistribution coefficient of S between the immobilized layer and the membrane, respec-tively. The maximum value of Is,lim (Imax

s;lim) is given by

Imaxs;lim ¼ nFAkcat½E�l ð19Þ

from which Is,lim values can be numerically obtained for given values of [S]x=l. Numericalcalculation supports the fact that Is,lim versus [S]

profiles exhibit Michaelis–Menten type

curved characteristics [28]. The empirical equation is given by

Is;lim ¼Imaxs;lim;app½S�

KS;app þ ½S� ð20Þ

FIG. 12 Kinetic scheme of a mediated biocatalyst electrode.


where Imaxs;lim;app and KS,app are the apparent values of I

maxs;lim and KS, respectively, and involve

the effect of the concentration depression of S in the immobilized layer and the perme-ability of S in the membrane. Equation (20) is frequently utilized for the characterizationof biocatalyst electrodes. It should be noted, however, that only when the enzyme reactionrate is much less than the diffusion rate of S in the immobilized layer and semipermeablemembrane, Imax

s;lim;app and KS,app approach the corresponding true values.Although the physical meaning of Imax

s;lim;app and KS,app are complicated, KS,app

becomes larger than the corresponding true values with an increase in the enzyme reactionrate compared with the diffusion rate of S in the immobilized layer and the semipermeablemembrane [28]. This property is very useful for widening the linear response range ofbiocatalyst electrodes to higher [S] regions. When Mred permeates the membrane towardthe solution, Is,lim decreases. Details of this problem have been discussed in the literature[29].

When an enzyme reaction in the immobilized layer is expressed by Eq. (2), Is,lim canbe empirically expressed, in a manner similar to that described above, as a function of thetotal concentration of Mox ([Mox]

) in the immobilized layer and [S]

in the bulk solution

[28].

Is;lim ¼Imaxs;lim;app

1þ KM;app=½Mox� þ KS;app=½S� ð21Þ

where KM,app is the apparent value of KM and involves the effect of the concentrationdepression of Mox as well as of S in the immobilized layer.

The current–potential curve of the steady-state current (Is) is sigmoidal. However,E1/2 is a complicated function of the electrode kinetics, the enzyme kinetics, and the masstransfer property [30]. Generally speaking, E1/2 becomes more positive than E�M withdecreasing electrode reversibility, as is the case of the homogeneous system. In contrast,E1/2 shifts to more negative potentials than E�M by the increase in the reversibility of theelectrode reaction, the enzyme kinetics, [Mox]

, and/or the mass transfer rate of the med-iator [30]. This is a typical surface catalytic property [31,32].

An example of the catalytic current due to mediated bioelectrocatalysis is shown inFig. 13, where the biocatalyst electrode is composed of a benzoquinone (BQ)–mixedcarbon paste electrode with glucose oxidase (GOD) immobilized behind a dialysis mem-brane on the electrode surface [33]. Panel (a) shows the CV of the electrode; the anodic andcathodic waves are attributable to the redox reaction of BQ dissolved from the carbonpaste bulk into the GOD layer on the electrode. Addition of D-glucose to the solutioncauses a large increase in the anodic current, and the current reaches a steady-state limitingcurrent Is,lim [panel (b)], where BQ in the GOD layer serves as an electron-transfer med-iator between the electrode and GOD. Is,lim increases with increasing concentration of D-glucose, cs, as illustrated in Fig. 14. The downward deviation from a linear relationship ispredicted by Eq. (20), and the slope of the Is,lim versus cs plots becomes smaller withincrease in the thickness (decrease in permeability) of the membrane covering the GODlayer as predicted by the above theoretical consideration. Carbon paste electrodes providea convenient means for preparing enzyme-modified electrodes [34,35]; the bulk of thecarbon paste serves as a sink for mediator compounds, and enzymes may be immobilizedon, or embedded in, the carbon paste electrodes [36]. A huge number of reports onbiocatalyst electrodes have appeared since the reports on the ferrocene-mediated enzymeelectrode [3] and the GOD-immobilized BQ–carbon paste electrode [5] for the ampero-metric determination of glucose. Biosensors based on mediated bioelectrocatalysis are now


commercially available for the measurement of glucose in blood, and efforts to improvethe sensor performance have been continued. Many papers dealing with mediated amper-ometric biosensors have appeared and continue to appear in journals relevant to analyticalchemistry and electroanalytical chemistry. ‘‘Fundamental Reviews’’ and the ClinicalChemistry chapter of ‘‘Application Reviews’’ in Analytical Chemistry will be helpful forbiosensor application.

B. Direct Bioelectrocatalysis

Catalytic currents based on direct electron transfer were observed for the reduction of O2

at carbon-black electrodes with immobilized laccase [37], the oxidation of H2 and reduc-tion of H+ on organic metals with adsorbed hydrogenase [38], and the oxidation of lactateat cytochrome b2-modified organic metal electrodes [39]. Since these observations of directbioelectrocatalysis, a considerable number of papers have appeared reporting direct bioe-lectrocatalysis, which have been summarized in review articles [40–43]. Interestingly, manydirect bioelectrocatalytic reactions use redox enzymes containing more than two redoxcenters: they are cytochrome b2 (flavin mononucleotide–heme), p-cresolmethylhydrolase[flavin adeninedinucleotide (FAD)–heme], flavocytochrome c552 (FAD-2 heme), cellobiosedehydrogenase (FAD–heme), D-fructose dehydrogenase [pyrroloquinoline quinone

FIG. 13 (a) A CV obtained with a film (50 �M)-coated glucose oxidase (18 �g)–BQ (0.25%)–

carbon paste electrode in acetate buffer of pH 5.0. The broken line is the CV of a film (50 �M)-

coated glucose oxidase (18 �g)–carbon paste electrode (not containing BQ). The solution was stirred

with a stirring bar at 500 rpm; scan rate: 50 mV s�1. (b) Similar to (a) except that 41 mM D-glucose

was added to the solution. (From Ref. 33.)


(PQQ)–heme], alcohol dehydrogenase (PQQ-4 heme), succinate dehydrogenase (FAD–Fe–S cluster), fumarate reductase (FAD–Fe–S cluster), and D-gluconate dehydrogenase(FAD–heme–Fe–S cluster). Redox enzymes with a single redox center can also take part indirect bioelectrocatalytic reactions, but the reactions seem to be limited to those for smallsubstrates: laccase (4Cu) for O2, peroxidase (heme) for H2O2, and hydrogenase (Fe–Scluster) for H+ and H2 [42].

The active sites of many redox enzymes seem to be more or less buried within theenzyme molecules as evidenced in the case of glucose oxidase by the three-dimensionalstructure [44]; consequently, the direct electron transfer of the redox enzymes at an elec-trode seems to be difficult. It is interesting to see that a unidirectional electron flow occursin the respiratory reaction in biological cell membranes, in which iron–sulfur clusters,flavins, quinones, and heme groups serve as electron-relaying centers. There exist flavo-cytochrome enzymes and/or quinocytochrome enzymes in some bacterial cell membranes,functioning as channels of electron flow from substrates to respiratory chains in themembranes. The enzymes have more than two redox centers: one is FAD or quinonecofactors, the site to accept electrons from the substrates, and the others are heme groupsserving as electron-relaying centers to donate electrons to ubiquinone in the membranes.When such a membrane-bound enzyme is adsorbed on an electrode, the electrode mayaccept electrons from the adsorbed enzyme through the heme groups in the catalyticoxidation of the substrate.

Figure 15 illustrates a schematic model of an expected electron flow from the sub-strate to the electrode through the redox centers in the enzyme adsorbed on the electrodesurface; the heme group (denoted as B in Fig. 15) is functioning as a built-in mediator to

FIG. 14 Dependence of the steady-state limiting current Is on the concentration of D-glucose *cS; Iswas measured at 0.5 V versus Ag/AgCl with film-coated glucose oxidase (180 �g)–BQ (30%)–carbon

paste electrodes at a film thickness of (a) 50 �m (nitrocellulose film), (b) 50 �m (dialysis membrane),

and (c) 100 �m (dialysis membrane). (From Ref. 33.)


achieve the unidirectional electron flow. This type of direct bioelectrocatalytic reactionoccurs when membrane-bound redox enzymes are adsorbed with an appropriate orienta-tion. The direct bioelectrocatalytic behavior is described by taking alcohol dehydrogenase(ADH) as an example. ADH is a bacterial membrane-bound quinocytochrome enzyme,which functions as the primary dehydrogenase of the ethanol respiratory chain to transferelectrons from ethanol to ubiquinone. It consists of three subunits of molecular weights78,000, 48,000, and 14,000; the first subunit contains tightly bound PQQ and heme c, thesecond three hemes c, and the third no redox group. ADH can be isolated from thebacterial membranes by the use of a detergent, Triton X-100. The solubilized enzymeattaches spontaneously on the surfaces of carbon and metal electrodes by physical adsorp-tion. Figure 16 shows CVs recorded with a gold-plated platinum (Au–Pt) electrode withadsorbed ADH [45]. A clear anodic current appears when the solution contains ethanol[curve (b)], though the wave attributable to the surface redox reaction of the adsorbedADH is not observed. When a current is measured at a fixed potential, a steady state isattained seconds after the addition of ethanol to the solution, and the steady-state catalyticcurrent increases with increasing concentration of ethanol, cEtOH, to approach saturation.The saturation tendency supports the catalytic oxidation of ethanol by ADH adsorbed onthe Au–Pt electrode surface.

FIG. 15 Schematic representation of direct bioelectrocatalysis. Electrons are transferred from

substrate to an electrode through an intramolecular electron transfer from redox center A to

redox center B in the enzyme molecule adsorbed on the electrode surface.

FIG. 16 CVs recorded with an Au–Pt electrode with adsorbed alcohol dehydrogenase in a buffer

solution at pH 6.0 (a) in the absence and (b) in the presence of 10 mM ethanol. Scan rate: 5 mV s�1.(From Ref. 45.)


The kinetic scheme of direct bioelectrocatalysis may be written as

Sþ Eox;ad$k1

k�1S � Ead!

k2Ered;ad þ P ð22Þ

Ered;ad$kfe

kbeEox;ad ð23Þ

where S and P are the substrate and product, respectively; Eox,ad, Ered,ad, and S-Ead are theoxidized and reduced forms of the adsorbed enzyme, and the enzyme–substrate complex,respectively; k1, k�1, and k2 are the rate constants of the enzymic reactions; and kfe and kbeare the rate constants of the electrode reaction of the adsorbed enzyme. In this simplemodel, the reaction S + Eox,ad ! S-Ead involves the step of intramolecular electrontransfer within the enzyme molecule. The steady-state current for reactions (22) and(23) can be expressed by [45]

I ¼ Imax=ð1þ kbe=kfeÞKm þ ð1þ k2=kfeÞ½S� ð24Þwith

Imax ¼ nFAk2½Enz�ad ð25Þand

Km ¼ ðk�1 þ k2Þ=k1 ð26Þwhere [S] (here, cEtOH) and [Enz]ad (here, Enz ¼ ADH) are the concentrations of thesubstrate in solution at the surface of the enzyme-modified electrode and the total con-centration of the adsorbed enzyme, respectively. At potentials where a limiting current isobtained, Eq. (24) is simplified to

I ¼ Imax½S�=ðKm þ ½S�Þ ð27ÞThe two quantities Imax and Km are obtainable from the dependence of Imax on cEtOH. TheKm values thus obtained with the different ADH-modified electrodes agree with each otherand with the Michaelis constant, 1.6 mM, for the ADH reaction in solution. Consideringthat the Michaelis constant expresses the affinity of an enzyme for the substrate, one maysay that the affinity of ADH for ethanol is affected little by the adsorption on the electro-des. The k2 value in Eq. 25 is estimated to be �20 s�1 as detailed in the literature [45]. Thisis much smaller than the catalytic constant, 482 s�1 of the isolated ADH in a solubilizedstate.

Electroreflectance (ER) measurement is a powerful technique that provides informa-tion on the redox state of the substance adsorbed in a monomolecular layer on an elec-trode [46]. When the ER spectrum is measured with an ADH-modified gold electrode at0.0 V in buffer at pH 6.0, small signals due to heme c appear in the range 390–590 nm. Aspectrum of adsorbed heme c is obtained by subtracting from the spectrum at 0.0 V aspectrum at �0:2 V, at which potential such signals as those at 0.0 V are not observed. Thedifference spectrum is shown in Fig. 17 [45], which is very similar to the spectrum obtainedwith cytochrome c coadsorbed with 4,40-bipyridyl on an Au electrode [46]. Heme c ofADH has absorption peaks at 417, 522, and 553 nm in the reduced form and at 409 nm inthe oxidized form, which are very similar to those of mammalian cytochrome c: 415, 521,and 550 nm in the reduced form and 409 nm in the oxidized form. Thus, the ER measure-ments strongly support the fact that heme c of the ADH adsorbed on a gold electrodeexchanges electrons with the electrode.


Although details of the adsorbed state of ADH are not clear, an oriented adsorptionis postulated with the PQQ moiety facing toward the solution as schematically illustratedin Fig. 15. The oriented adsorption is anticipated from the indication that ADH in vivo ispartially buried in the bacterial cytoplasmic membranes with the PQQ moiety exposed tothe periplasmic space and with the heme c moieties within the membranes. The PQQmoiety is presumed to be hydrophilic, and is allowed to stay apart from the electrode ofhydrophobic nature, while the heme c moiety is allowed to be in close contact with theelectrode surface. Thus, electrons can flow from the substrate to the electrode through theenzyme during the bioelectrocatalytic reaction, in which heme c acts as a built-in mediator.The observations that the adsorption affects the Michaelis constant of the ADH reactiononly slightly, but significantly affects the catalytic constant, as mentioned above are con-sistent with the oriented adsorption; the small catalytic constant suggests a conformationalchange in the hydrophobic moiety of ADH.

Direct bioelectrocatalyis is expected to be a key reaction in realizing mediatorlessbiosensors [40–43].

IV. MICROBIAL CELL-BASED BIOELECTROCATALYSIS

This section deals with bioelectrocatalysis based on whole cells of micro-organisms.Microbial cells contain a number of oxidoreductases; thus, they may be regarded as bagsof enzymes. Section IV.A describes how the catalytic activity of whole bacterial cells isevaluated. The bioelectrocatalysis of whole cells is then detailed in Section IV.B, in whichapplications in biosensors and bioreactors are also described. In Section IV.B.2, a novelapplication of bioelectrocatalysis to the study of enzyme reactions in vivo is mentioned.

FIG. 17 Electroreflectance spectrum of heme c of alcohol dehydrogenase adsorbed on an Au

electrode in a buffer at pH 6.0 with a modulation of 70 mV at 8.0 Hz. (From Ref. 45.)


A. Evaluation of Catalytic Activity of Microbial Cells

Many kinds of bacterial cells contain integrated protein complexes that take part in therespiratory electron-transport reaction in the cytoplasmic membrane. It is located near thecell surface beneath the cell wall, which is permeable to substances of relatively lowmolecular weights. Thus, oxidoreductases existing in the periplasmic space or bound tothe cytoplasmic membrane may function as catalysts to oxidize or reduce substancesoutside the cells using externally added artificial electron acceptors or donors. A schemeof the bacterial cell catalysis is illustrated in Fig. 18. The rate (vcell) of the bacterial cell-catalyzed reaction can be written as (47)

�cell ¼kcat;cell½B�

1þ KM;cell=½M� þ KS;cell=½S�ð28Þ

where [B], [M], and [S] are the concentrations of the bacterial cell (it is expressed by theoptical density of the cell suspension, which can be converted to the molar concentration[48]), mediator, and substrate, respectively; kcat,cell is an apparent catalytic constant of asingle cell, and KS,cell, and KM,cell are apparent Michaelis constants for S and M, respec-tively. Here, ‘‘apparent’’ means that these constants involve the effects of cell-wallpermeability to the substrate and mediator, and their distribution between the solutionand the medium within the cells. Furthermore, it is noted that a single bacterial cellcontains the enzyme molecules catalyzing the redox reaction in large numbers.Therefore, the constants should be regarded as parameters characterizing the catalyticactivity of the bacterial cell.

Values of these parameters can be determined from measurements of the concentra-tion changes in electron acceptors (or donors) by bacterial cell catalysis [47]. Table 1 givesthe values of kcat,cell, KS,cell, and KM,cell obtained for several bacterial species. The kcat,cellvalues expressing the turnover numbers of single cells lie in the range 0:6� 42� 106 s�1,

FIG. 18 Schematic model of catalytic oxidation of substrate, S, using an electron acceptor, Mox, by

an intact enzyme, DHase, in the cytoplasmic membrane of a micro-organism. (From Ref. 47.)


which are much larger than the values usually obtained with enzyme catalysis; kcat,cell maybe regarded as the product (zkcat) of a catalytic constant of intact enzyme (kcat) and thenumber of enzymes in a cell (z) [49]. It is noted that not only KM,cell values but also KS,cell

values in a given bacterial cell catalysis depend on the kind of electron acceptors. This isindicative that these values reflect the Michaelis constants of enzymic reactions in bacterialcells. The reactions are given by Eqs (1a) and (1b), thus, the Michaelis constants, KM andKS, involve the rate constants in both reactions (1a) and (1b). Relatively large KM,cell

values for Fe(CN)3�6 compared with those for the other electron acceptors suggest thatthe permeability of the cells to Fe(CN)3�6 is low. In any event, the bacterial cells can betreated as bags of enzymes and their catalytic activity can be described by a Michaelis–Menten type equation.

B. Mediated Bioelectrocatalysis of Microbial Cells

Since bacterial cells behave like bags of enzymes in catalyzing redox reactions of sub-strates, using artificial redox compounds as electron acceptors or donors, they may workas catalysts to produce catalytic currents for the electrocatalytic oxidation or reduction ofsubstrates in the presence of appropriate electron-transfer mediators. In fact, whole bac-terial cells both in a suspension and in an immobilized state produce similar bioelectro-catalytic currents to those obtained with enzymes.

TABLE 1 Kinetic Parameters of the Catalytic Reactions of Micro-organisms

Micro-organisms Substrate

Electron

acceptor/donor

KS,cell

(mM)

KM,cell

(mM)

kcat,cell(106 s�1)

A. aceti (IFO3284) Ethanol Q0 1.8 0.59 0.74

Fe(CN)3�6 4.9 7.7 1.6

O2 1.8 <0.02 3.0

P. fluorescens (TN5) Nicotinic acid DCIP 0.45 0.89 17.3

Fe(CN)3�6 0.21 7.2 4.6

O2 0.20 <0.02 10.8

G. industrius (IFO3260) D-Glucose BQ 7.6 1.2 8.2

DCIP 0.41 0.20 0.58

O2 3.1 <0.06 3.1

E. coli K12 (IFO3301) D-glucose Q0 0.64 1.1 6.7

(in presence of PQQ) PMS 0.90 0.80 15.0

DCIP 4.2 0.80 7.1

Fe(CN)3�6 0.19 3.5 1.6

O2 0.14 <0.05 0.45

T. ferrooxidans FeSO4 O2 0.037 <0.03 3.0

Fe(CN)4�6 O2 0.001 <0.07 1.5

D. vulgaris H+ MV+�— 0.16 42.0

(Hildenborough) H2 DMBQ 0.027 0.14 35.5

NQ 0.025 0.31 42.5

VK3 0.015 0.62 35.2

Q0: 2,3-dimethoxy-5-methyl-1,4-benzoquinone; DCIP: dichlorophenol-indophenol; BQ: 1,4-

benzoquinone; PMS: phenazine methosulphate; DMBQ: 2,6-dimethyl-1,4-benzoquinone; NQ: 1,4-

naphthoquinone; VK3: 2-methyl-1,4-naphthoquinone (vitamin K3).


1. Mediated Bioelectrocatalysis of Bacterial Cell Suspension

Bacterial cells can be suspended in a solution at 109 cells mL�1, which is in the order of10�12 M. In spite of this low concentration, the bacterial cells produce observable catalyticcurrents because of their high catalytic activities as given in Table 1. Figure 19 shows theeffect of Acetobacter aceti cells on the CVs of Q0 [50]. Q0 produces a totally irreversible CVat a glassy carbon (GC) electrode [curve (a)]; a cathodic wave for the reduction of Q0 tothe reduced form (Q0H2) has a peak potential at �0:19 V and a small anodic waveattributable to the oxidation of Q0H2 at 0.34 V. In the presence of A. aceti cells, the heightof the cathodic wave is decreased, while that of the anodic wave is increased [curve (b)].The effect of A. aceti cells on the CV is attributable to the reducing activity of the cells.Addition of ethanol to the solution causes significant increase in the anodic current anddrastic decrease in the cathodic current [curve (c)]. This clearly shows that ethanol isoxidized by A. aceti cells with Q0 as an electron acceptor to produce Q0H2 and that theQ0H2 is electrochemically oxidized at the electrode to regenerate Q0. The resulting anodicwave becomes independent of the scan rate, and changes to a sigmoidal shape, which ischaracteristic of a catalytic current. The current magnitude increases, depending on theconcentration of ethanol, cEtOH, and approaches a maximum current at and abovecEtOH ¼ 6 mM. The current also increases with the increase in the concentration of Q0

(cQ); it is proportional to cQ at cQ < 50 �M, but begins to bend downward at a cQ higherthan 50 �M. The nonlinear dependence of the current on both cEtOH and cQ is character-istic of an enzyme-based mediated catalytic current.

The catalytic reactions of whole cells can be expressed by Eq. (28), which is of thesame form as the equation for the catalytic reactions of enzymes, Eq. (2). Accordingly, thebioelectrocatalytic currents produced by whole cells can be analyzed by equations of thesame form as those employed for the analyses of bioelectrocatalytic currents by enzymes.Under the conditions cEtOH > 6 mM and cQ < 50 �M, where the current is linearly pro-portional to cQ, the steady-state limiting current is written as

Is;lim ¼ nFAfDðkcat;cell=KM;cellÞ½B�g1=2cQH ð29Þ

FIG. 19 Effect of A. aceti cells on CVs of 3-dimethoxy-5-methyl-2-1,4-benzoquinone (Q0): (a) pH

6.0 buffer containing 1 mM Q0; (b) (a) + A. aceti cells; (c) (b) + 20 mM ethanol. Scan rate: 5 mV

s�1. (From Ref. 50.)


which is of the same form as Eq. (4); here, D and cQH are the diffusion coefficient and thebulk concentration of the reduced form of Q0 (Q0H2), respectively, and [B] is the concen-tration of A. aceti cells. The quantity kcat,cell/KM,cell corresponds to the second-order rateconstant of the reaction between an A. aceti cell and Q0, which is calculated to be kcat,cell/KM;cell ¼ 1:0� 109 M�1 s�1 using D ¼ 9:8� 10�6 cm2 s�1 estimated separately and ½B� ¼7:07� 10�12 M. This is in good agreement with the kcat,cell/KM,cell value of 1:25� 109

M�1 s�1 calculated from the kcat,cell and KM,cell values in Table 1.

2. Mediated Bioelectrocatalysis at Bacterial Cell-Modified Electrodes

Bacterial whole cells can be immobilized on electrodes in a similar manner to the immo-bilization of enzymes [5]; e.g., A. aceti cells can be immobilized behind a dialysis mem-brane on the surface of a carbon paste electrode [48]. A detailed study of theelectrocatalytic properties of the current produced by the A. aceti cell-modified Q0-mixed carbon paste electrode in the presence of ethanol has revealed that the electroca-talytic behavior is described by

Is;lim ¼Imaxs;lim½S�

KS;cell þ ½S� ð30Þ

Imaxs;lim ¼ nFAkcat;cell½B�l ð31Þ

These equations are of the same forms as the equations of the catalytic current derived forenzyme-based electrocatalysis, Eqs (19) and (20). Analysis of the data obtained with theelectrode at low concentrations of ethanol gives the bimolecular rate constant for thereaction between the immobilized A. aceti cell and ethanol, kcat,cell/KS,cell, as 2:0� 108

M�1 s�1 [48]. This value agrees fairly well with the kcat,cell/KS,cell value of 4:1� 108 M�1

s�1 calculated from the kcat,cell and KS,cell values given in Table 1.Whole cell-based biosensors have usually been considered to have poor selectivity

with a longer response time and require lengthy recovery times [51,52]. However, there area number of bacterial cells whose mediated bioelectrocatalytic behavior is very similar tothat observed with isolated enzymes. Figure 20 shows calibration graphs of ethanolobtained with the A. aceti-modified Q0-mixed carbon paste electrode [48]. A linear rela-tionship is obtained between the current and cEtOH at cEtOH lower than 1 mM [panel b].Oxygen influences the current magnitude; the current measured under air-saturated con-ditions is reduced to 69%. The response time required to attain 95% of the magnitude ofthe steady-state current is 2 min. The reproducibility of the current measured at cEtOH ¼ 1mM is 1.3% (coefficient of variation; n ¼ 5). The electrode has very low response tomethanol, and poor sensitivity to glucose; the relative magnitudes of the currents formethanol and glucose to the current for ethanol are 0.08 and 5.5%, respectively. Theelectrode has no response to glycerol, a substance contained in the culture for the growthof A. aceti. The properties of the A. aceti-modified electrode are similar to those obtainedwith an enzyme-modified electrode prepared from ADH isolated and purified from A.aceti cells except that the former is somewhat sensitive to the presence of oxygen andglucose. It should be mentioned that genetic manipulation provides a powerful means ofobtaining a bacterial strain containing one particular enzyme of much higher activity. Thisis interesting in view of the fact that bacterial cells of high activity in one particular enzymeare preferable for the construction of a whole cell-modified electrode with high currentdensity and high selectivity. It has been shown that an electrode modified with an ADH-deficient strain, Acetobacter pasteurianum (NP2503), had a very poor current response to


ethanol, while the electrode modified with the strain cloned with the gene of ADH, A.pasteurianum (NP2503c), produced a catalytic current of significant magnitude [48]. Suchgenetically manipulated bacteria may be promising biocatalysts in constructing whole-cellbased mediated amperometric biosensors.

Since bacterial cells serve as bags of enzymes to catalyze redox conversions of sub-strates to products, they may be used in bioelectrocatalytic conversion. Pseudomonasfluorescens TN5 catalyzes the hydroxylation of nicotinic acid (NA) into 6-hydroxynicoti-nic acid (6HNA), an important compound as a starting material for the synthesis of a newtype of pesticide. Under aerobic conditions, however, 6HNA is metabolized in the P.fluorescens cells. The use of Fe(CN)3�6 as an extracellular electron acceptor enhances thebiotransformation of NA into 6HNA and completely suppresses the subsequent oxidationof 6HNA. Thus, flow electrolysis of NA in the presence of Fe(CN)3�6 at a P. fluorescensimmobilized column electrode allows the accelerated and complete transformation of NAinto 6HNA without any byproduct [53].

Mediated bioelectrocatalysis at a whole cell-modified electrode provides a novelmeans of studying in vivo enzymic reactions. A method using an Escherichia coli-modifiedelectrode has been successfully applied to the kinetic and thermodynamic studies of the invivo activation of glucose dehydrogenase (GDH) apoenzyme in E. coli cells [54]. E. colihas a membrane-bound GDH in the form of an apoenzyme, since E. coli is unable tosynthesize its prosthetic group PQQ. Apo-GDH in E. coli cells is converted into holo-

FIG. 20 Calibration graphs for ethanol obtained with the A. aceti (87 �g)-modified Q0

(10%)–carbon paste electrode under (*) deaerated and (&) air-saturated conditions. (From Ref.

48.)


GDH, the active form, with exogenous PQQ and Mg2+. The method using an E. coli-modified electrode allows continuous measurement of the activation process as demon-strated in Fig. 21, panel (A). When PQQ is added to the solution containing Mg2+, Q0,and D-glucose, the E. coli-modified electrode immersed in the solution begins to producethe oxidation current attributable to the bioelectrocatalytic oxidation of D-glucose by theimmobilized E. coli cells. The increase in the current reflects the increase in holo-GDH inthe cells [the current is proportional to the concentration of holo-GDH, cf. Eq. (31)] by thereaction of apo-GDH with PQQ added to the solution. The slope of the current–timecurve becomes larger with increasing concentration of PQQ (cPQQ). Accordingly, the time-dependent increase in the current measures the rate of the holo-GDH formation withPQQ. The rate of the holo-GDH formation in the E. coli cells immobilized on the elec-trode can be written as

apo-GDHþ PQQ !kf;PQQ

kb;PQQ

holo-GDH ð32Þ

FIG. 21 Holo-GDH formation in E. coli cells immobilized on an electrode was monitored by

measurement of the current for the catalytic reaction of the immobilized cells. Panel (A): the E.

coli-modified electrode was incubated in buffer containing 5 mM Mg2+ for 15 min, then 0.3 mM Q0

and 10 mM D-glucose was added to the buffer solution and the current was recorded at 0.5 V to

measure the increase in current on the addition of PQQ at a final concentration of (a) 100, (b) 60, and

(c) 30 nM. Panel (B): the data used in these plots were taken from the current–time curves in panel

(A). (From Ref. 54.)


where kf,PQQ and kb,PQQ are the rate constants for the formation and dissociation of holo-GDH, respectively. The currents at the three PQQ concentrations attained the steady stateof the same magnitude, indicating that reaction (32) is almost completely shifted to theright-hand side under these experimental conditions. Accordingly, the rate equation can beexpressed as

� ¼ d�

dt¼ kf;PQQð1� �Þ cPQQ ð33Þ

with � ¼ [holo-GDH]/{[holo-GDH]+[apo-GDH]}, where [holo-GDH] and [apo-GDH]are the concentrations of holo-GDH and apo-GDH in the E. coli cells on the electrode.Since the current (i) is proportional to � as given by i ¼ �imax, the dependence of i on time(t) is written as

ln½ðimax � iÞ=imax� ¼ �kf;PQQ cPQQt ð34Þhere, imax is the current at the steady state, which corresponds to the full holo-GDHformation. Applying Eq. (34) to the current–time curves in panel (A) of Fig. 21, kf,PQQ

can be determined from the slope of the ln[imax/(imax � i)] versus t plots in panel (B). In thisway, the kf,PQQ value is determined to be 3:8� 0:4� 104 M�1 s�1.

The currents in panel (A) reach steady states, indicating attainment of the equili-brium state in the reaction, Eq. (32). The steady-state current becomes smaller withdecreasing cPQQ. Accordingly, the equilibrium constant of the binding of PQQ to apo-GDH can be determined from the dependence of the steady-state current on cPQQ. Thedissociation constant of the equilibrium is calculated to be 1:0� 0:1 nM, revealing that theaffinity of PQQ for apo-GDH is very high.

V. APPLICATION OF BIOELECTROCATALYIS IN FUEL CELLS

Fuel cells are devices for converting chemical energy into electrical energy and havereceived considerable attention as practical devices for energy transfer because of theirefficient and nonpolluting properties. They use H2 and O2 gas as the most efficient fuels,noble metals as catalysts, and are driven at moderate to high operating temperatures underacidic or alkaline conditions. The use of biocatalysts in place of the metal catalysts wouldallow fuel cells to operate at neutral pH and ambient temperatures, which are the condi-tions much more favorable for the handling of fuel cells. Studies of such fuel cells (biofuelcells) have recently become a field attracting many electrochemists because of increasingrequirements for developing implantable batteries and human-life friendly energy sources.Reactions in biofuel cells involve an electron-transfer reaction between a biocatalyst andan electrode, and the rapid electron-transfer reaction is crucial in the operation of biofuelcells. Mediated bioelectrocatalysis is expected to be a method appropriate for realizingrapid electron transfer, and in fact, a biofuel cell based on mediated bioelectrocatalysis hasrecently been developed, which uses H2 as fuel to produce H2O in the reaction with O2 atneutral pH and ambient temperature [55].

The anodic oxidation of H2 was accelerated by methyl viologen-mediated electroca-talysis with bacterial cells Desulfovibrio vulgaris (Hildenborough) [D. vulgaris (H)] ascatalysts, and the cathodic reduction of O2 was accelerated by 2,20-azinobis-(3-ethylbenzo-thiazoline-6-sulfonate)-mediated electrocatalysis with bilirubin oxidase (BOD) as a catalyst.The bioelectrocatalytic reduction of O2 to H2O has been described in Section II.B, and thebioelectrocatalytic oxidation of H2 has been fully investigated [49,56]. Curve (a) in Fig. 22


shows a CV for the redox reaction of ABTS��/ABTS2� at pH 7.0. In the presence of BOD,

the redox reaction is coupled with the BOD-catalyzed reduction of O2 to produce a largecathodic catalytic current [curve (b) in Fig. 22]. The half-wave potential (0.48 V) of thesigmoidal voltammogram is close to the standard redox potential of the O2/H2O redoxcouple (0.618 V at pH 7.0). As mentioned in Section II.B, kinetic analysis of the catalyticcurrent has proven that ABTS2� functions as a very efficient electron donor in the BOD-catalyzed reduction of O2 to H2O. Curve (c) in Fig. 22 shows a CV for the methyl viologen(MV)-mediatedD. vulgaris (H) cell-catalyzed redox reaction of 2H+/H2. It is noted that thesigmoidal CV has both cathodic and anodic limiting currents with the half-wave potential(�0:635 V) agreeing with the midpotential Em (�0:634 V, the potential at the midpointbetween the anodic and cathodic peaks) of the CV for the redox reaction of MV2+/MV

�+.This means that the methyl viologen-mediated electrocatalysis is voltammetrically reversi-ble [49], i.e., the electrocatalytic system allows both the oxidation of H2 and the reduction ofprotons, depending on the potential applied to the electrode. The anodic current for theoxidation of H2 starts to appear from �0:61 V, which agrees with the standard redoxpotential of the 2H+/H2 redox couple (�0:611 V at pH 7.0). Curve (d) in Fig. 22 is theCV obtainedwith anthraquinone 2-sulfonate (AQS). AQS has anEm of�0:42 V at pH 7.0, apotential more positive than �0:611 V in favor of the H2 oxidation reaction. Accordingly,only an anodic limiting current is observed, and the magnitude is larger than that observedwith methyl viologen [56]. The catalytic activity of D. vulragis (H) is due to hydrogenaseexisting in the periplasmic space behind the outer membrane of the bacterial cell and is ashigh as the isolated hydrogenases [49]. This is a fortunate result, since whole cells of D.vulgrais (H) are more stable and easier to handle than isolated hydrogenases.

A construction of a biofuel cell is schematically illustrated in Fig. 23. A carbon feltsheet was used for both anode E1 and cathode E2. An anion-exchange membrane 180 �mthick was used for a separator membrane S. The contact area of S with the electrolyte ineach compartment was 12.5 cm2. Each compartment had an electrolyte solution (adjustedto pH 7.0 with NaH2PO4 and Na2HPO4) of 5 mL. The cell was used as a prototype biofuelcell to evaluate the performance of the fuel cell composed of a biocathode (ABTS

��/ABTS2�–BOD–O2/H2O) and a bioanode (MV2+/MV

�+–D. vulgaris (H)–2H+/H2). Thebiofuel cell was operated with O2 and H2 gas bubbling in the cathode and anode compart-

FIG. 22 CVs recorded with a glassy carbon electrode in a phosphate buffer at pH 7.0: (a) 0.25 mM

ABTS2�; (b) (a) + 0.1 �M BOD saturated with O2 gas; (c) 0.5 mM MV2+ + D. vulgaris (H)

saturated with H2 gas; (d) 0.25 mM AQS + D. vulgaris (H) saturated with H2 gas. Scan rate: (a) and

(b) 10 mV s�1; (c) and (d) 2 mV s�1. (From Ref. 55.)


ments, respectively, at atmospheric pressure and 258C. Filled circles in panel (a) of Fig. 24plot Ecell against I, where Iwas calculated from the Ecell value at a given value ofR. The Ecell

value is 1.17 V at an open circuit, which is close to the standard electromotive force (1.23 V)for the reaction H2 þ 1=2O2! H2O, and remains at 1.0 V for the current flowing at I =0.9 mA. This is a significant result in view of the fact that it is even larger than the value ofthe cell voltage (1.0 V) realized in solid polymer electrolyte fuel cells at an open circuit at508C and 1 atm pressure [57]. The value of Ecell, however, begins to decrease rapidly downto zero at around I = 1 mA (this leads to 0.2 mA cm�2 for the current density at the feltelectrode calculated per projected surface area), which is much less than the currentattained in the solid polymer electrolyte fuel cells, 0.2 A cm�2 or more [57]. Open circlesin panel (a) plot the quantity Ec � Ea against I. Each open circle is located only slightlyabove the closed circle (Ecell) at the corresponding value of I. The small difference betweenEc and Ea and Ecell assures that the internal resistance of the biofuel cell is small.

Panel (b) in Fig. 24 shows the dependence of Ec (filled square) and Ea (filled circle) onI. Solid curves in the figure depict linear-sweep voltammograms recorded on a three-electrode system with the cathode or the anode in the biofuel cell system as the workingelectrode. The voltammograms are essentially the same in shape as the CVs obtained witha glassy carbon electrode [curves (b) and (c) in Fig. 22] and are superimposable on theplots of Ec (filled square) and Ea (filled circle) against I. The sigmoidal shape of thevoltammograms reflects the fact that the bioelectrocatalytic reactions at the felt electrodesare in a steady state. The voltammogram recorded with the anode exactly traces the Ea

versus I plot at I > 0, and the voltammogram with the cathode traces the Ec versus I plot,but increases further to reach a limiting current. The result reveals that the maximumcurrent obtained with the biofuel cell is limited by the anodic reaction. The agreement ofthe voltammograms with the Ec (and Ea) versus I plots is not surprising, since a givenelectrochemical system should give the same current–potential curve independent of themethod of electrochemical measurement in a steady state. Accordingly, conventional

FIG. 23 Schematic illustration of the construction of the 2H2/O2 biofuel cell. A: anode

compartment with pH 7.0 phosphate buffer containing D. vulgaris (H) cells and MV2+ bubbled

with H2 gas; C: cathode compartment with pH 7.0 phosphate buffer containing BOD and ABTS2�

bubbled with O2 gas; E1 and E2: carbon felt sheets as an anode and a cathode, respectively, S:

separator membrane, R: resistor. (From Ref. 55.)


voltammetry provides a convenient method of analyzing the dependence of the cell voltageon the output current of a biofuel cell.

The steady-state bioelectrocatalytic currents would become larger at higher concen-trations of biocatalysts, and the current would come to be controlled by the rate of masstransfer of O2 and/or H2. The magnitude of the mass transfer-controlled current for O2 isestimated as 40 mA in an air-bubbled buffer solution (0.25 mM O2) from the limitingcurrent for the direct reduction of O2 to H2O at potentials more negative than �1:3 V atthe felt electrode. Thus, a maximum current of 200 mA is expected for the solutionbubbled with O2 gas (1.26 mM O2). The same argument applies to the current controlledby the rate of the mass transfer of H2 in the anodic reaction. The maximum currentattainable in a solution saturated with H2 (0.78 mM) can be estimated to be about 120mA. It should be noted that the magnitudes of the mass transfer rate-controlled currentsare still less than the current magnitude realized in ordinary solid polymer electrolyte fuelcells [57]. A strategy for realizing such high current density would be the use of a porouselectrode with three-phase boundary regions as employed in the solid polymer electrolytefuel-cell system [58].

FIG. 24 (a) Plot of Ecell (*) and Ec � Ea (*) against I; (b) plots of Ec (&) and Ea (*) against I.

Measurements were made with a biofuel cell containing 0.1 �M BOD and 0.4 mM ABTS2� in the

cathode compartment bubbled with O2 gas, and D. vulgaris (H) and 1.5 mM MV2+ in the anode

compartment bubbled with H2 gas. Solid curves are voltammograms measured with the anode (or

cathode) as the working electrode at a scan rate of 5 mV s�1. (From Ref. 55.)


Bioelectrocatalysis-based fuel cells operate under mild conditions and are able to useas fuels many kinds of organic compounds such as carbohydrates and alcohols. They maybe constructed in a variety of sizes ranging from a conventional type of fuel cell down tocells of nanometer scale. Therefore, bioelectrocatalysis-based fuel cells are expected to finda variety of applied fields: molecular batteries, implantable batteries, disposable-typebatteries, conventional type of fuel cells, etc.

REFERENCES

1. MR Tarasevich. In: S Srinivasan, YA Chizmadzhev, JO’M Bockris, BE Conway, E Yeager,

eds. Comprehensive Treatise of Electrochemistry. vol. 10. New York: Plenum Press, 1985,

pp 231–295.

2. T Ikeda, M Senda. In: M Senda, M Aizawa, N Oyama, eds. Kobunshi Kino Denkyoku

(Polymer Functional Electrodes). Tokyo: Gakkai Shuppan, 1983, pp 131–159 (in Japanese).

3. AEG Cass, G Davis, GD Francis, HAO Hill, WJ Aston, IJ Higgins, EV Plotkin, LDL Scott,

APF Turner. Anal Chem 56:667–671, 1984.

4. T Ikeda, I Katasho, M Kamei, M Senda. Agric Biol Chem 48:1969–1976, 1984.

5. T Ikeda, H Hamada, K Miki, M Senda. Agric Biol Chem 49:541–543, 1985.

6. AJ Bard, LR Faulkner. Electrochemical Methods. New York: John Wiley, 1980, pp 455–461.

7. LA Coury Jr, BN Oliver, JO Egekeze, CS Sonsnoff, JC Brumfield, RP Buck, RWMurray. Anal

Chem 62:452–458,1990.

8. Y Ogino, K Takagi, K Kano, T Ikeda. J Electroanal Chem 396:517–524, 1995.

9. K Kano, T Ohgaru, H Nakase, T Ikeda. Chem Lett 439–440, 1996.

10. T Ohgaru, H Tatsumi, K Kano, T Ikeda. J Electroanal Chem 496:37–43, 2001.

11. K Kano, T Ikeda. Anal Sci 16:1013–1021, 2000.

12. M Rudolph, DP Reddy, SW Feldberg. Anal Chem 66:A589-A600, 1994.

13. T Matsue, H Yamada, H-C Chang, I Uchida. Bioelectrochem Bioenerg 24:347–354, 1990.

14. K Yokoyama, Y Kayanuma. Anal Chem 70:3368–3376, 1998.

15. H Tatsumi, K Takagi, M Fujita, K Kano, T Ikeda. Anal Chem 71:1753–1759, 1999.

16. K Takagi, K Kano, T Ikeda. J Electroanal Chem 445:211–219, 1998.

17. F Battaglini, EJ Calvo. Anal Chim Acta 258:151–160, 1992.

18. LA Coury Jr, RW Murray, LJ Johnson, KV Rajagopalan. J Phys Chem: 95:6034–6040, 1991.

19. J Kulys, T Buck-Rasmussen, K Bechgaard, V Razumas, J Kazlauskaite, J Marcinkeviciene,

JB Christensen, HE Hansen. J Mol Catal 91:407–420,1994.

20. S Tsujimura, H Tatsumi, J Ogawa, S. Shimizu, K Kano, T Ikeda. J Electroanal Chem

496:69–75, 2001.

21. K Takagi, M Torimura, K Kawaguchi, K Kanol, T Ikeda. Biochemistry 38:6935–6942, 1999.

22. K Takagi, K Yamamoto, K Kano, T Ikeda. Eur J Biochem 268:470–476, 2001.

23. WR Heineman, FM Hawkridge, HN Blount. In: AJ Bard, ed. Electroanalytical Chemistry.

vol. 13. New York: Marcel Dekker, 1984, pp 1–113.

24. A Sato, M Torimura, K Takagi, K Kano, T Ikeda. Anal Chem 72:150–155, 2000.

25. M Torimura, M Mochizuki, K Kano, T Ikeda, T Ueda. Anal Chem 70:4690–4695,1998.

26. VD Parker, LC Seefeldt. Anal Biochem 247:152–157, 1997.

27. AE Kasmi, R Brachmann, G Fuchs, SW Ragsdale. Biochemistry 34:11668–11677, 1995.

28. T Ikeda, K Miki, M Senda. Anal Sci 4:133–138, 1988.

29. M Senda, T Ikeda, K Miki, H Hiasa. Anal Sci 2:501–506, 1986.

30. T Ikeda, K Miki, F Fushimi, M Senda. Agric Biol Chem 51:747–754, 1987.

31. T Ikeda, R Schmehl, P Denisevich, K Willman, RW Murray. J Am Chem Soc 104:2683–2691,

1982.

32. K Kano, T Konse, B Uno, T Kubota. In: G Dryhurst, K Niki, eds. Redox Chemistry and

Interfacial Behavior of Biological Molecules. New York: Plenum Press, 1988, pp 267–280.


33. T Ikeda, H Hamada, M Senda. Agric Biol Chem 50:883–890, 1986.

34. M Senda, T Ikeda. In: DL Wise, ed. Bioinstrumentation. Boston: Butterworth, 1990,

pp 189–210.

35. T Ikeda. Bull Electrochem 8:145–159, 1992.

36. L Gorton. Electroanalysis 7:23–45, 1995.

37. MR Tarasevich, AI Yaropolov, VA Bogdanovskaya, SD Varfolomeev. Bioelectrochem

Bioenerg 6:393–404, 1979.

38. SD Varfolomeev, SO Bachurin, A Nagui. J Mol Cat 9:223–226, 1980.

39. JJ Kulys, G-SJ Svirmickas. Anal Chim Acta 117:115–120, 1980.

40. JJ Kulys. Biosensors 2:3–13, 1986.

41. AL Ghindilis, P Atanasov, E Wilkins. Electroanalysis 9:661–674, 1997.

42. L Gorton, A Lindgren, T. Larsson, FD Munteanu, T. Ruzgas, I. Gazaryan. Anal Chim Acta

400:91–108, 1999.

43. T Ikeda. In: FW Scheller, F Schubert, J Fedrowitz, eds. Biosensorics I. Basel: Birkhauser

Verlag, 1997, pp 244–266.

44. HJ Hecht, HM Kalisz, J Hendle, RD Schmid, D Schomburg. J Mol Biol 229:153–172, 1993.

45. T Ikeda, D Kobayashi, F Matsushita. J Electroanal Chem 361:221–228,1993.

46. T Sagara, H Murakami, S Igarashi, H Sato, K Niki. Langmuir 7:3190–3196, 1991.

47. T Ikeda, T Kurosaki, K Takayama, K Kano. Anal Chem 68:192–198, 1996.

48. T Ikeda, K Kato, M Maeda, H Tatsumi, K Kano, K Matsushita. J Electroanal Chem

430:197–204, 1997.

49. H Tatsumi, K Takagi, M Fujita, K Kano, T Ikeda. Anal Chem 71:1753–1759, 1999.

50. T Ikeda, K Kato, H Tatsumi, K Kano. J Electroanal Chem 440:265–269, 1997.

51. JE Frew, HAO Hill. Anal Chem 59:933A–944A, 1987.

52. J Racek. Cell Based Biosensors. Basel: Technomic, 1995, pp 3–12.

53. M Torimura, H Yoshida, K Kano, T Ikeda, T Yoshida, T Nagasawa. J Mol Catal B 8:265–273,

2000.

54. D Iswantini, K Kano, T Ikeda. Biochem J 350:917–923, 2000.

55. S Tsujimura, M Fujita, H Tatsumi, K Kano, T Ikeda. Phys Chem Chem Phys 3:1331–1335,

2001.

56. H Tatsumi, K Kano, T Ikeda. J Phys Chem B 104:12079–12083, 2000.

57. S Srinivasan, OA Velev, A Parthasarathy, DJ Manko, AJ Appleby. J Power Sour 36:299–320,

1991.

58. JO’M Bockris, SUM. Khan. Surface Electrochemistry. New York: Plenum Press, 1993,

pp 880–882.


17Energetics and Gating of Narrow IonicChannels: The Influence of ChannelArchitecture and Lipid–Channel Interactions

PETER C. JORDAN, GENNADY V. MILOSHEVSKY, and MICHAEL B.PARTENSKII Brandeis University, Waltham, Massachusetts, U.S.A.

I. INTRODUCTION

Physiological ion channels have evolved to accomplish one basic task, using controlled ionflow to transmit electrical signals from one region of an organism to another [1]. This hasmany manifestations. A few examples include: sodium and potassium ion counter flowsgoverning muscle motion (the action potential); calcium ion flow in regulating cardiacrhythms; chloride flow in regulating skeletal muscle resting potential. Structurally similarare channels that rigorously exclude ions, but transport particular neutral species (likeaquaporins and glycerol facilitators) [2,3].

All such channels are designed with three aims in mind: promotion of rapid trans-membrane flow of the preferred moiety (permeability); strong discrimination against theflow of any other major physiological solute (selectivity); and rigorous control of thesefunctions (gating) so that the organism does not fall prey to dehydration or severe elec-trolyte imbalances.

A channel, in its conducting (open) state, behaves like (and can be described as) anenzyme, with substrate turnover typically in the range 106–107 s�1 and a recovery time in therange of milliseconds. Unlike pumps, where turnover is in the range 102–103 s�1 [4], move-ment through a channel is diffusive in nature and requires no addition of energy. Channelsare extraordinarily effective catalysts for the transport of their chosen ion ormolecule. Somequantitative estimates of their amazing success in such endeavors are readily obtained. Theenergy required to transfer an ion the size of Kþ across a membrane would be in excess of200 kJ mol�1. The corresponding transit time would be �1022 s or � 1014 years. Quiteclearly, ion channels completely suppress the electrical image barrier posed by the mem-brane. However, they are functionally far more impressive than this simple estimate sug-gests. There is strong evidence that a typical potassium channel has a narrow constriction�1.5 A in radius and �15 A long [5]. Treating the water-filled pore as a conducting ‘‘wire’’of radius R and length L, and assuming that the ionic mobility through this pathway isdetermined by the aqueous equivalent conductance, , estimates of the pore’s conductancecan be given. Presuming a point ion and no wall friction, the conductance, �, is

� ¼ �R2=L ð1Þ


for physiological concentrations of 100mM this would imply a conductance of�50 pS, onlyslightly less than that of large-conductance potassium channels, �200 pS [1]. However, ifsmall pore friction and access limitations are taken into consideration [6,7], the maximal �estimate drops precipitously, to �0.05 pS, far less than that which nature has been able todesign. Not only do ion channels negate the electrostatic barrier, they do so in ways that, ineffect, imply that their interiors are frictionless and their entrances are nonreflecting.

Section II discusses the energetic determinants of catalysis: permeation and selectiv-ity. In narrow pores they are observed to have some puzzling and superficially contra-dictory properties. The two best studied systems are the gramicidin-A (GA) dimer and theKcsA potassium channel. Here, we limit consideration to the more physiologically impor-tant KcsA system. Our focus is on the interplay of various interactions governing ion entryinto this channel, and the ion’s subsequent passage through the pore. The general theore-tical problem has been reviewed recently [8]. Our discussion first provides a brief overviewof standard theoretical approaches to the problem, from both the microscopic and meso-scopic perspectives. We then outline the semimicroscopic Monte Carlo formalism that wehave developed [9–11], highlighting its differences from the more orthodox methods andemphasizing its particular advantages. We finally apply this method to KcsA, and focus onsome of the insights that it yields.

Section III describes the mechanical influence that a membrane can have on channelstabilization, focusing on simple physical models used to understand the mechanism ofmembrane–peptide interaction. We describe the elastic interaction between a membraneand an inserted peptide, focusing on GA, and first provide a brief overview of the basics ofthe elastic theory of membranes, demonstrating the formal equivalence of these issues to aclassical problem in mechanical engineering. We then consider different descriptions of theinteraction of a membrane with an inclusion and follow with a discussion of lipids’influences on channel lifetimes. Finally, we describe a new perspective for describing themembrane-inclusion interaction. It emphasizes the inclusion-induced perturbation ofmembrane elastic constants at the lipid–peptide interface.

In Section IV we extend the ideas of Section III and treat membrane-mediatedinteraction between peptides. First, we consider the interaction between two insertionsand study the dependence of the free energy on both the interinsertion distance d, and the‘‘contact slope’’ s, the normal component of the gradient of the membrane’s deformationalprofile at the lipid–peptide interface. When minimized as a function of s this free energy,FminðdÞ is repulsive for all d. However, when treated for fixed s, the profiles FsðdÞ undergoa van der Waals-like transition from attractive behavior for s � sð1Þcr to repulsive behaviorfor s � sð2Þcr ; in the intermediate region, the profiles exhibit a local minimum separated by abarrier from a repulsive region at larger d: Finally, we focus on multiparticle effects,particularly those involving five and seven insertions and find that many-body interactionssmooth out the repulsive behavior of FminðdÞ for a cluster of five and make it purelyattractive for a cluster of seven. Treated at fixed s, the profiles FsðdÞ are similar to thosefound for two insertions, with three distinct regions and two ‘‘critical’’ points.

Section V summarizes our results and outlines promising avenues for future work.Of particular interest are how increasing the insertion concentration may lead to phasetransitions and may affect the lifetime of ion channels.

II. SELECTIVITY AND PERMEABILITY: CATALYSIS IN NARROW PORES

Selectivity and permeability are essentially the mappings of differential free energy profilesfor the various transportable species. As such they are sensitive to the membrane, which


provides low dielectric surroundings for the channel protein. The energetic determinant ofpermeation is that the favored species’ permeation free energy profile is essentially flat;relative to the free energy of the aqueous ion there should be neither a deep well (leading toblock) nor a high barrier (leading to impermeability). All competing species must havesignificantly higher barriers to permeation in some region of the permeation pathway andmust not have any deeper wells where they might effectively block the channel. This canrequire a highly sophisticated discrimination mechanism. Potassium channels rigorouslyexclude sodium; clearly, this is not just molecular sieving. When no calcium is present,calcium channels are highly permeable to sodium; however, at physiological calcium con-centrations (a few millimolar) the predominant sodium (150 mM) can no longer pass.Aquaporins and glycerol channels permit neutral molecules to pass; much smaller ionsdo not enter the channel.

A. Free Energy Compensation and Molecular Structure

Both permeability and selectivity are determined by a moiety’s ‘‘permeation free energyprofile,’’ the difference between its solvation free energy within the channel (�Gsolvation)and its free energy in bulk water (�Ghydration). Discrimination between solutes in thechannel interior requires a relatively narrow pore; otherwise the most that can be expectedis sieving in response to ionic polarity. In traversing the pore there are major changes inthe solute’s local solvation environment. Physiologically significant ions are fully hydratedin water, with approximately six to eight surrounding solvent molecules in their firsthydration shell [12]. As an ion enters the pore most of this water is stripped away, anddepending on the radius of the channel constriction, it may be left with only a pair of watermolecules, one leading and one trailing. The remainder of its immediate solvation envir-onment is composed of functional groups from the channel peptide. Rapid turnoverrequires that this environment be essentially as ‘‘solute friendly’’ as bulk water. Howmight ions be stabilized within channels? Given the structure of the amino acid buildingblocks, there are some obvious candidates: cations could interact favorably with backbonecarbonyl oxygens, with negatively charged residues or, via a polarization mechanism, withthe �-electrons of aryl residues [13]; anions might couple with backbone amido hydrogensor with positively charged residues. How might charged groups be excluded from trans-membrane aqueous pathways? Possibly this might involve electrostatic rejection by apolarchannel surroundings. In the absence of structural clues, choosing among these alterna-tives is purely speculative.

Detailed structural data are only available for a few highly selective channels. Thebest characterized is the model system gramicidin, which has been elucidated using NMR-based techniques [14,15]. Among physiological systems, atomic level resolution structures(based on x-ray or cryoelectron microscopy) are available for a potassium channel from S.lividans (KcsA) [5], two prokargotic chloride channels [16], a stretch activated channelfrom M. tuberculosis [17], human red cell aquaporin-1 [18], and a glycerol facilitator fromE. coli [19]. All share a common feature, a narrow region where the transported speciesmust shed most of its surrounding water and become intimately associated with thechannel protein. In KcsA the constriction is associated with a single file region structurallymuch like that in gramicidin, but far shorter, only �10–15 A long; there are backbonecarbonyl oxygens with which the ions could co-ordinate. The chloride channel’s struc-ture [16] confirms the ‘‘double-barreled shotgun model’’ [25]; anion stabilization arises byinteraction with a number of partially positively charged groups. In both aquaporin-1 and


the glycerol facilitator, the nonpolar groups that line the long single file would appear toform inhospitable surroundings for ionic species.

Atomic level resolution structures are not yet available for sodium or calcium chan-nels or for nicotinic receptors. Electrophysiological inference suggests strongly that theselectivity domain of calcium channels involves four glutamates; in the sequentially similarsodium channels three of these have been replaced by an alanine, and aspartate and alysine, mutating a region with net charge of �4 to one with net charge of �1 [20,21]. Thereare structural data on the acetylcholine receptor (AChR, the most heavily studied memberof the nicotinic receptor family), but at just below the resolution required for detailedtheoretical analysis [22]; nonetheless, electrophysiology suggests that, in AChR itself, setsof negatively charged residues create annular regions of negative potential in the channel,and that these regions stabilize cations and exclude anions [23]. As yet there is no clearstructural basis for the mild univalent/divalent discrimination observed [24].

Gramicidin’s structural stability arises from hydrogen bonds between backbonecarbonyl oxygens and backbone amido hydrogens. Cation entry locally disrupts thehydrogen bonds with simultaneous local co-ordination to two or more oxygens. As theion becomes solvated by the groups forming the channel backbone, all but two of itswaters of solvation must be stripped off. Whether the origin of the channel’s cationselectivity is primarily due to the channel carbonyl’s ability to co-ordinate positive ions[26] or reflects, in roughly equal measure, the greater difficulty of partially dehydratinganions [9] is still open to dispute.

The KcsA structure exhibits numerous features consistent with electrophysiologicalinferences [5]. As had been deduced, the channel is effectively a cylindrically symmetricaltetramer. In further accord with biochemical evidence, there are negatively charged resi-dues near the channel mouths, augmenting and depleting the concentrations of cationsand anions, respectively. Kinetic models of the permeation pathway implied a multi-ionchannel [27]; this was corroborated by the x-ray structure, which displays three ions in thepore. The local filter architecture, a narrow constriction lined with carbonyl oxygens,provides clear rationalization for this multi-ion stability. As illustrated in the idealizationof Fig. 1, two ions, separated by a single water molecule, are each surrounded by eightcarbonyl oxygens; the third is in a small aqueous pool with the carboxyl termini of the four�-helices aimed directly towards its center. However, the ionic pathway in this filter is sonarrow (with a mean constriction radius �1.4 A) [28] that it is not obvious how an ion canpass through it essentially unimpeded. In fact, recent electrophysiological evidence sug-gests that the selectivity filter in KcsA may have undergone a conformational change in thetransition to the open state of the channel, since ions as large as tetramethylammonium(TMA, radius �2.9 A) appear to permeate effectively [29]. Even in this well-characterizedsystem, unresolved questions remain. How can the free energy profile for potassium beessentially barrierless; i.e., why is potassium conductance so high? If the conducting chan-nel is �5.8 A in diameter, what makes sodium permeation effectively impossible?; i.e., whyis the K/Na permeability ratio so high? How does the structure translate into a free energyprofile that accounts for the macroscopic observations?

B. Modeling Conductance and Permeation Free Energy

The basic goal in theoretical modeling of ion channels is relating molecular structure tophysiological function (for a recent comprehensive review, see Ref. 8). A wide range ofapproaches have been used in this quest, most often targeted at the valence-selectivesimple model channel, gramicidin [e.g., 9,30–34]. Here, our focus is a truly ion-selective


physiological system, the well-characterized KcsA potassium channel, which hasrecently become the object of much scrutiny, using both microscopic and mesoscopicmethods.

1. Conventional Approaches

Both molecular dynamics (MD) simulation and Monte Carlo (MC) computation of equi-librium behavior can be carried out at full atomic detail, but each makes large computa-tional demands. Were atomic motions adequately represented by Newtonian mechanics(reasonable unless some mechanistic steps depended heavily on proton transfers [33]),either could, in principle (with runs sufficiently long), provide an accurate, well-resolvedand reliable picture of ionic energetics along a permeation pathway (for reviews focusingon biophysical applications, see Refs 32,35,36). There are, however, well-known difficul-ties with these approaches. The results are only as reliable as the underlying force field,which must correctly describe interactions among the atoms of both the protein and itssurrounding solvents (cell membrane, external water, and cytoplasmic water). Since theion–atom terms in the typical force field have been parameterized to mimick behavior in

FIG. 1 Semimicroscopic model geometry for the KcsA selectivity filter. It includes solvating CO

groups (residues 75–78 of each tetramer strand), single file ions and waters, peptide dipoles, the80Asp carboxylates, and the aqueous cavity and its included ion. In its simplest form, bulk electrolyte

and the cavity are treated as dielectric continua. The Helmholtz layer (accounting for water

immobilized by interaction with polar surfaces) separating the explicit sources in the filter from

extracellular bulk water has a width of 2 A; that between the filter and the midchannel water

pool is 1.5 A. As illustrated, the pool radius is 5.0 A. The crystallographic occupancy sites (2 and

4) are � 18.5 and 11.0 A from the center of the water pool. In modifications including more explicit

water, the pool accommodates �20 waters. Two optional geometry modifications involving

additional explicit waters are indicated in italics. For modeling an open channel the pool is

deformed into a tubular connector (dotted lines). To describe better the influence of the outer

mouth on ion-filter energetics, the heavily cross-hashed region in the left Helmholtz layer can be

filled with explicit water.


aqueous solutions, it is uncertain if they adequately account for ion–solvent interaction inthe channel interior, where both water and ion–water structures differ sharply from that inbulk solution. Certainly, the effective water–water potentials used in most MD and MCstudies bear no resemblance to quantum mechanically determined water dimer potentials[37]; this strongly suggests that either polarization or explicit three-body interactions [38]must be included if water is to be accurately modeled under conditions sharply differentfrom those used in the parameterization (ambient water). As these microscopic approachesare computationally very time consuming, individual simulations are rarely if everrepeated using fundamentally different force fields. Thus, their reliability can only beestablished by whether the computational setup is satisfactory and whether the resultsare, by and large ‘‘reasonable,’’ i.e., in agreement with experimental observation andphysical preconception. If this is the case, then credence is given to unusual observations.

Recent applications of MD simulational have addressed a wide range of issues. Clearevidence has been adduced to show that the permeant ions play a significant role instabilizing the channel assembly [39]. A detailed model of the channel’s duty cycle, invol-ving concerted movements of a Kþ–H2O–Kþ cluster from the aqueous cavity through thefilter [40], was found to be an energetically favorable permeation pathway. Somewhatunexpectedly, a computation of the multi-ion free energy profile suggests that the channelmight possibly operate in a four-ion mode, with one ion in the cavity and three additionalions in the filter, each separated by a single water molecule [41]. There is evidence that theoriented �-helices, in addition to helping to stabilize the ion in the cavity, align the watermolecules of the cavity and thereby contribute indirectly to stabilizing the ions in the filter[42]. While ionic diffusion appears to be significantly slowed in the channel, permeation isnot hindered since it appears driven by coulombic repulsion between the ions in the filter[28,43]. In all cases where computations have addressed the issue of selectivity the resultsaccord qualitatively with the observed high K/Na permeability ratio. However, the struc-tural basis for this discrimination remains an open question. Possibly it reflects the rigidityof the filter, so that sodium cannot bind as effectively as potassium to the carbonyloxygens in the binding pockets [5,11,28,40,43]. Alternatively, it may be a reflection ofthe greater energetic cost of dehydrating Na, i.e., removing it from water [44]. Most ofthese simulational results are consistent with experimental observations on the conductingchannel. A few are surprising and none is contradictory. Their internal consistency pro-vides a strong indication that, by and large, they realistically describe significant aspects ofthe interaction between ions and the KcsA selectivity filter. However, given that the x-raystructure describes a constriction that is just large enough to accommodate Csþ [5], thelargest of the alkali cations, what is uncertain is whether any of these MD results is trulydescriptive of behavior in the open state of KcsA, which appears to be permeable to themuch larger TMA [29].

Mesoscopic studies begin by drastically simplifying the computational model or, incases where direct structural data are unavailable, building models based on electrophy-siological inference [45–48]. Instead of treating a fully hydrated KcsA channel embeddedin a model membrane or surrounded by a domain formed from aliphatic chains [28,39–44],the solvents (exterior water, cytoplasmic water, and the membrane) are replaced by dielec-tric continua [49,50]. The fraction of the peptide that is explicitly included in the modelsystem depends on the specific aspects of channel behavior that are under investigation. Incomputations designed to model conductance, the model’s geometry necessarily differsfrom that determined crystallographically, where the channel, at its intercellular end, is fartoo narrow to permit ion entry. In all mesoscopic modeling only a very few moieties areassigned mobility. The goal is not detailed simulation of behavior at the atomic level, but


development of a (more or less realistic) model that accords with a specified range of thechannel’s properties. The results of such modeling are crucially dependent on the simplify-ing assumptions employed. Their ultimate goals must be limited; there will always remainsome ambiguity in the correlations of molecular structure with physiological function.However, whenever the results of dissimilar theoretical approaches, based on rather dif-ferent (but plausible) physical assumptions, are in basic agreement, their overlap providesstrong evidence for the fundamental correctness of a particular correlation.

In a study of the electrostatic influence of the oriented �-helices and the aqueouscavity in stabilizing an ion in the cavity [49], stringent dielectric assumptions wereimposed. All aqueous regions were treated as dielectrically equivalent to bulk water,with an " of �80, an approximation that must break down in the narrow constriction,where the effective " is more likely to be in the range �2–4 (this choice was first suggestedand discussed in Refs 51 and 52) [53,54]. The solvating protein was treated as an immobileset of charges at their crystallographic locations and embedded in a continuum dielectricwith " ¼ 2; this approximation ignores the stabilizing influence of solute-induced solventreorganization, which would tend to raise the effective ". These constraints have importantconsequences: the first overestimates dielectric shielding in the constriction, leading tomuch reduced interionic repulsion; the second ignores electrically induced reorientationof the peptide’s charged and polar groups, sharply increasing the strength of the electricalinteraction (both attractive and repulsive) between the peptide and ions in the channel.Nonetheless, the conclusions are illuminating and clearly demonstrate a direct relationbetween the channel’s structure and its kinetic properties. Consistent with the x-ray struc-ture and the properties of K-channels generally, monovalent cations were found to bemore stable in the cavity than in bulk water. Further consistence with the general behaviorin K-channels, divalent cations were also more stable in the cavity than in bulk water, butsignificantly less stable than the monovalent cations.

A completely different mesoscopic study used crystallographic data as input forBrownian dynamics analysis, with the aim of modeling potassium channel conductanceusing the KcsA structure as a guide [50]. The only parts of the KcsA assembly specificallyincorporated were the charged residues near the channel mouths, the carbonyl oxygenslining the constriction, and the carboxyl termini of the helix dipoles; the positions of allthese moieties were fixed, and determined from the x-ray structure. Dielectric conditionssimilar to those of the electrostatic [49] analysis were imposed (" ¼ 2 in nonaqueousregions and a constant, but adjustable, " extending the length of the aqueous pore). Anextracrystallographic assumption, based on electrophysiological observations, fixed thechannel diameter of the intracellular mouth at �6.0 A, consistent with studies of theinterior block. The diameter of the channel constriction was fixed by the x-ray structure,and taken to be 3.0 A. With this model, the multi-ion kinetics observed experimentallycould be reproduced. Rectifying current–voltage relationships in good quantitative agree-ment with experiment were found when the pore was assigned an " of 60. As with theelectrostatic study [49], this too demonstrates that macroscopic observables can be directlydeduced from knowledge of the channel’s structure, augmented by physical assumptionsbased on electrophysiological inference.

2. Semimicroscopic Method

While they provide considerable insight into aspects of ion-channel behavior, and therelation between structure and function, the approaches outlined in Section II.B.1 havetheir drawbacks. The reliability of the force fields, the real time length of the simulations,


and possible complications in treating long-range interactions limit the range of applic-ability of microscopic studies. Also, the introduction of unconfirmable, and somewhatcounterintuitive, dielectric assumptions, the use of a continuum description of water in ahighly constricted environment, and the immobility of the channel’s electrical surround-ings limit the range of applicability of mesoscopic studies. A model treatment that bridgesthe gap between these methods might provide insights not readily obtained using either.

The semimicroscopic (SMC) formulation was designed for just this purpose [9]. Itsbasic philosophy is that a certain, limited number, of interactions are crucial determinantsof ion-channel permeation that must be accounted for microscopically. These wouldinclude the contents (ions and water) of the narrow part of the channel and the solvatinggroups forming the boundaries of this constriction. The more distant parts of the systemneed only be described approximately. In the case of KcsA, these could include theprotein’s charged residues and the termini of the oriented �-helices. Bulk water domains,the remainder of the protein and its surrounding membrane are described as dielectriccontinua. Figure 1, an abstraction of the KcsA selectivity filter, illustrates how these ideasmay be translated into practice. In its minimalist application, the only explicitly mobilefeatures are the single-file moieties: ion(s), single-file waters, and the backbone carbonyloxygens of residues 75–78 that form the lining of the selectivity filter constriction.Additional electrical features (the backbone carbonyl carbons of residues 75–78, the sepa-rated termini of the oriented �-helices, and charged amino acid residues) are immobilized.The aqueous cavity and both bulk water regions are assigned a high " � 1; the regionsbetween the aqueous phases, comprising the membrane and the remainder of the peptide,are assigned a low " � 2 [10,11]. In this way, a protein system comprising some 600residues with another 100 or so water molecules in the extracellular mouth, the constric-tion, and the aqueous cavity has been described in terms of �20 mobile units, with theremainder forming part of an electrical environment.

Unlike strictly mesoscopic approaches, the SMC method permits the inclusion ofsubstantially more mobile molecular detail, treating various structural features both indi-vidually and collectively. Unlike the microscopic approaches, the computations are rapidenough that a wide range of plausible perturbations of the basic structure can be essayed.Both qualitative and quantitative consequences of significant structural changes can bereadily determined. While the x-ray structure of KcsA depicts a very narrow filter [5], theelectrophysiology suggests a substantially wider filter when the channel is in its conductingstate [29]. Using the SMC protocol it is possible to carry out a range of thought experi-ments. A particularly interesting example would be to enlarge the filter gradually anddetermine how the filter radius affects channel selectivity.

This approach involves some major approximations, the most important being thedielectric constant used to describe the intramembrane region and the resulting force fieldsdevised to describe interactions between the groups explicitly modeled. Because "backgroundis 2, standard force fields such as CHARMM [55], AMBER [56], and GROMOS [57] forwhich "background 1 are not appropriate. What is the rationale for assigning an " of 2? Thebasic idea is to account for electrically induced solvent reorientation of the ion’s surround-ings in the channel–membrane–water ensemble, i.e., to treat the structural aspects ofdielectric reorganization. Dielectric response is frequency dependent. Consider water. Atlow frequency (� < 1 GHz), molecular motion is effectively instantaneous; the solventresponds rapidly and in phase with an electrical stimulus. The dielectric constant doesnot differ from its static value [58]. At higher frequency (from � 1 GHz to � 1 THz),rotational and vibrational responses are gradually frozen out and water’s dielectric con-stant drops from � 80 to its high-frequency value of 1.8, representing the electronic con-


tribution to water’s dielectric permittivity. The SMC method is designed to take fullaccount of electrically induced structural rearrangement, which suggests assigning thesystem a background " � 1:8. The actual choice of "background is not quite so straightfor-ward since the ensemble also includes lipid for which the high-frequency dielectric con-stant is somewhat larger, �2 [59].

One approach to a ‘‘best’’ value for " is through the development of appropriateforce field parameters. For the simplest SMC modeling, based on hard-core molecules andions [9–11], this requires establishing the effective radii for water and the solute ions. Usingas criteria the density of bulk water and the hydration free energies of monovalent alkalications and halides, the optimal choice of " is 2 with a water radius of 1.4 A [60]. Withthese parameters the hydration free energies of both alkali cations and halides can bereasonably well reproduced using the Pauling radii [61]. Again, it should be stressed thatSMC modeling is basically designed to establish the qualitative influence that differentstructural features have on permeation, and how permeation responds to structuralchange. High-accuracy quantitative agreement with experiment, while it would be a desir-able outcome, is not an overarching goal.

C. SMC Computation of Permeation Free Energy

In ion permeation through a transmembrane channel, the basic process is one in which afully hydrated ion is removed from water and exchanged for a water molecule in the filter.This is summarized by the reaction:

Ionaqueous þWaterchannel ! Ionchannel þWateraqueous ð2ÞIn applications of the SMC approach the permeation process is decomposed into threedistinct transfer steps. First, an ion in bulk water is dehydrated and exchanged for a watermolecule in the gas phase; this is followed by transfer of the ion to a cavity in an infinitebackground dielectric, exchanging it for a water in the cavity; finally, the ion in the infinitedielectric background is stabilized by exchange for water in the channel. The individualsteps are

Ionaqueous þWatervacuum ! Ionvacuum þWateraqueous : dehydration ð3Þ

Ionvacuum þWaterbackground ! Ionbackground þWatervacuum : cavity ð4Þ

Ionbackground þWaterchannel ! Ionchannel þWaterbackground : stabilization ð5ÞThe hydration energy is experimentally accessible (to within � 3–4 kT at 300 K); thecavity term is a Born energy, determined by the ion’s cavity radius in the channel; andthe stabilization energy is computed using standard perturbation methods [9].

It should be noted that the process of Eq. (3) is actually hypothetical; nevertheless,these energies can be estimated with some considerable certainty. What is needed is theequivalent of a measurement of the absolute reduction potential for the standard hydrogenelectrode (SHE), i.e., the process:

HþðaqueousÞ þ e�o !1

2H2ðgasÞ ð6Þ

Standard tabulations of ionic free energies assign this process a reference free energy of0.00 kJ mol�1. All ionic reduction potentials are based on this standard; to computeabsolute values from standard values requires shifting them by Z times the real reduction


free energy for the SHE, with Z being the ionic valence. The SHE potential cannot bedirectly measured. All estimates rely on reasonable, but ultimately unverifiable, assump-tions [62]. The most recent experimental and theoretical determinations differ by only�7 kT [63,64], suggesting an estimated experimental uncertainty in �Gdehydration of�3–4 kT.

The cavity step, Eq. (4), involves the transfer of the ion from vacuum, with its " of 1,to the dielectric background, where " �2, and the complementary process of transferringwater from the background to vacuum. This is equivalent to two Born charging processes,one for a point charge and the other for the water charge distribution; as long as thecavities are chosen to be spherical the associated energies are readily computed [65].Naturally, transfer energetics is dominated by the ion, but the water contribution cannotbe ignored. Using a water model of radius 1.4 A incorporating three charges located at theatomic sites that reproduce water’s intrinsic dipole moment, the free energy for transfer ofwater is �10 kT; for a monovalent ion of the same radius the transfer free energy is 10times larger, i.e., �100 kT.

The free energy defined by Eq. (4) requires establishing a ‘‘cavity radius’’ for theparticular solvation environment. This is done by noting that, in a Born model, the ioniccharge (or the water charge distribution) is in a cavity with dielectric constant of unity,surrounded by solvent domains with their associated dielectric constants. The cavityboundary is, therefore, the position where the solvent begins [66]. When viewed from aquantum mechanical perspective, this is where the ion’s (or water’s) charge distributionbegins to overlap significantly with the charge distribution of the molecules forming thecentral species’ first solvation shell; this is essentially the van der Waals radius of thesolvent molecules. In studying ion–water systems containing a fully charged solvatedion, with the ion at the center of a first solvation sphere, the mean ion–water distancenaturally exceeds the sum of the two moieties’ hard-core radii [60]. Using the quantummechanical identification, the cavity radius is simply the difference between the mean ion–water distance for molecules in this first solvation shell and water’s hard-core radius; it islarger than the ion’s hard-core radius [60].

The stabilization contribution, Eq. (5), in which a charge in the dielectric back-ground is exchanged for a multipolar charge distribution in the channel, accounts fordielectric reorganization of the solvent [9–11]. Because this perturbation involves processesoccurring in a uniform dielectric background, it would appear that there is no Borncontribution. This is correct as long as the exchanged species (ion and water) have thesame radii; however, if their radii differ there is an additional small term in the free energywhich can be accurately accounted for by decomposing stabilization into two more steps[60]. The ion–water exchange takes place without a radius change creating an intermediate‘‘water’’ of modified cavity size; the radius change is then dealt with by exchanging thisintermediate species for a normal water.

1. Applications: Parameterizing Ion–Water Interactions

The reliability of the SMC approach is crucially dependent on devising reasonable inter-action parameters. Here, we describe their determination by computing the free energy ofionic dehydration and contrasting the consequences of two values of "background, either 1.8or 2, and two values of water’s hard-core radius, either 1.4 or 1.5 A. The rationale for the "choices has already been given. The radius choices correspond to the mean water–waternearest neighbor distance in liquid water and the water–water separation in the dimer,respectively [67]. Instead of exchanging an ion in the dielectric background for a water in


the channel, the third step of the equivalent thermodynamic process of Eqs. (3)–(5) nowinvolves exchanging the ion for an explicit water in a sphere containing a limited numberof explicit waters (�40) surrounded by bulk water. Both energies, Eq. (4) and the analog toEq. (5), depend on the choice of "background and the water hard-core radius. Table 1presents dehydration energies for six hypothetical ions of various hard-core radii; theseare compared with ‘‘experimental’’ data, estimated from the absolute dehydration freeenergies for Kþ, Rbþ, Csþ, F�, Cl�, and Br�, assuming Pauling’s ionic radii [61]. An"background of 2 is clearly preferable to the choice of 1.8; both water hard-core radii yieldacceptable free energy values for cations, but the overall agreement is better using a waterradius of 1.4 A. Only for the Br� analog (R ¼ 1:95 A) are disagreements significantlylarger than the computational uncertainty.

It should be stressed that the SMC parameterization procedure is very different fromstandard Born model analysis where an ion of assigned radius is embedded in a uniformcontinuum dielectric with bulk permittivity. Here, the ion is fully solvated by the modelwater, and the dielectric constant reflects the solvent’s electronic properties. The structuralcomponent of dielectric reorganization is treated explicitly.

2. Applications: Permeation Energetics in KcsA

The great strength of SMC modeling is its flexibility. KcsA provides an excellent demon-stration system. Structural features can be added sequentially to determine which havesignificant co-operative influences on �Gpermeation. The effect that mobility of portions ofthe peptide distant from the channel ion transport domain have on �Gpermeation can beaddressed separately, in effect providing a measure of how such motions alter the channelprotein’s effective ". The extent to which water, in the channel mouth, in the central cavity,or in the conducting system’s intracellular channel connecting the protein’s inner mouth tothe central cavity, can be described in continuum terms is readily investigated. The pos-sible influence of filter radius on �Gpermeation, important for understanding how the widerfilter can be selective [29], is accessible to study. We can use Fig. 1 to illustrate how some ofthese questions can be approached by addition of layers of complexity.

The model, in its simplest form, has only 21 mobile features, the ion(s) and waterswithin the filter and the carbonyl oxygens that form the binding pockets [11]. The carbonylcarbons, the termini of the oriented �-helices, and the various charged groups explicitly

TABLE 1 Dehydration Free Energies (in kJ mol�1) for Ions of Various Size and Polarity

as Functions of Water’s Hard-Core Radius, R, and Background Dielectric Constant, "a

Ion propertiesb " ¼ 2 " ¼ 1:8

Radius (A) Valence R ¼ 1:4 A R ¼ 1:5 A R ¼ 1:4 A R ¼ 1:5 A ‘‘Experiment’’

1.3 þ1 356� 14 345 339 371 351

1.5 þ1 313� 12 320 309 314 317

1.7 þ1 292� 11 295 275 285 305

1.3 �1 436� 9 428 444 453 445

1.8 �1 304� 5 294 279 283 309

1.95 �1 265� 5 265 264 260 280

a See text.b Computational uncertainties are only listed in the first column; they are essentially independent of "

and R.


treated are immobilized. The model geometry [11] is derived from the x-ray structure. Theaqueous cavity is presumed to be a continuum, filled with permittive, high ", water. Thelarge diameter but thin extracellular mouth is assumed to be filled with essentially immo-bilized water, in effect forming a low " Helmholtz layer [68]. There is no water-filledconnector extending from the central cavity to the channel’s inner mouth. All these arefeatures that can be modified or turned on and off individually, or are restrictions that canbe lifted. In Fig. 1, sites 2–5 are the physiologically important ones, equivalent to thosefound fully or partially occupied in the x-ray structure [5]. Three of them have roughelectrophysiological equivalents, identified from single-channel conductance measure-ments: sites 2 and 4 represent the ‘‘outer lock-in’’ and ‘‘enhancement’’ sites, respectively[27] and site 5 approximates the Ba block site [69]. Site 3 corresponds to the location of thefilter’s explicit water [5].

Consider first the role of the central cavity. Its major purpose is to stabilize an ionnear the center of the membrane [5,49]. Might it have some secondary influence on per-meation? Does the cavity stabilize filter ions as well as single file water or as bulk water?Does it isolate the filter from the channel’s low " intracellular domain? How reasonable isthe continuum description of the cavity? Would filter energetics be very different in anopen channel? To address these questions we limit consideration to ion–water interactionand treat five variants from the default model geometry [11]:

1. Fill the cavity with explicit waters.2. Vary the cavity radius, keeping overall system width constant.3. Eliminate the cavity, replacing it by single file waters.4. Approximate the open state by creating contact between the cavity and the

intracellular region, either by shrinking the overall system width or by con-structing a water filled connecting tube.

Using the SMC energy decomposition, Eqs. (3)–(5), only the stabilization term is affectedby the changes outlined. Thus, we can answer these questions by focusing on that termonly, and base conclusions on a hypothetical monovalent ion with the same hard-coreradius as water, i.e., a cation intermediate between Kþ and Rbþ. Results are presented inTables 2–4, for which the computational uncertainties in the energies are � 1 kJ mol�1.

Table 2 treats various modifications of the cavity, either filling it with explicit wateror varying its radius. These modifications noticeably affect the stabilization free energyonly at site 5, adjacent to the cavity. Clearly the high ", continuum cavity approximationadequately describes the system except at site 5. Table 3 illustrates that, if the filter

TABLE 2 Effect of Cavity Size and Cavity Descriptiona on Monovalent

Cation Stabilization Free Energies, Eq. (5), for Single Occupancy of Sites

of Model Filter (Fig. 1)

Geometric propertiesExplicit

waters

Stabilization energies (kJ mol�1Þ

Width (A) Rcavity (A) Site 1 Site 2 Site 3 Site 4 Site 5

49 5 File-5, Cavity-0 �71 �65 �51 �51 �4849 5 File-5, Cavity-20 �70 �64 �51 �52 �5249 8.5 File-5, Cavity-0 �71 �65 �53 �54 �5549 0 File-5, Cavity-0 �70 �63 �50 �47 �37a See text.


geometry is unaltered, transition to an open state can have little effect on ion energetics inthe filter itself, again with the exception of site 5. Table 4 compares the stabilizing effect ofthe cavity with that of hypothetical single file waters and shows that just one single filewater is sufficient to compensate for the whole water-filled cavity, underscoring the excep-tional stabilizing influence of linear dipolar arrays [70,71].

The overall picture is clear. The cavity did not evolve in order to stabilize mono-valent filter ions; more single file waters would do this task better. It does have a real effecton ionic stability at site 5, roughly the Ba block site [69]. Since ion–cavity interaction arisesfrom image forces, quadratically dependent on ionic valence, the site 5 interactions wouldbe four times more effective in stabilizing a divalent Ba than a monovalent alkali cationand may play an important role in determining the site of the block. The cavity effectivelyisolates the filter from the low ", interior side of the channel assembly. Without a change infilter geometry, transition to the open state could only marginally alter filter energetics.

Just as we have separated out the influence that the cavity has on ions in the filter, wecan ask how the various electrical sources included in the model of the peptide affect theion’s permeation free energy. Assuming that joint effects do not alter the cavity size, thiscan also be answered by computing stabilization free energies. Again considering(hypothetical) single occupancy we can separately assess the influence of:

TABLE 3 Effect of Intracellular Channel Geometrya on Monovalent

Cation Stabilization Free Energies, Eq. (5), for Single Occupancy of Sites

of Model Filter (Fig. 1)


waters



49 5 File-5, Cavity-0 �71 �65 �51 �51 �4841 5 File-5, Cavity-0 �71 �65 �51 �52 �4933 5 File-5, Cavity-0 �72 �66 �53 �54 �5349 Tubeb File-5, Tube-72 �71 �64 �52 �54 �56a See text.b See text and Fig. 1.

TABLE 4 Effect of Replacing Cavity by Single File Watersa on

Monovalent Cation Stabilization Free Energies, Eq. (5), for Single

Occupancy of Sites of Model Filter (Fig. 1)


waters



49 5 File-5, Cavity-0 �71 �65 �51 �51 �4849 0 File-5, Cavity-0 �70 �63 �50 �47 �3749 0 File-6, Cavity-0 �70 �64 �51 �51 �5549 0 File-7, Cavity-0 �71 �65 �52 �54 �5949 0 File-8, Cavity-0 �71 �65 �53 �55 �61a See text.


1. Bulk and cavity water2. Single file water3. The binding pocket carbonyls4. The oriented �-helices5. The 80Asp (aspartate) residues near the extracellular mouth

Figure 2 describes each feature’s contribution to the stabilization free energy, Eq. (5), forthe channel occupied by the hypothetical Kþ/Rbþ hybrid ion described above.Additionally, it indicates the extent to which there is any co-operativity (defined as thedifference between the influence of all sources collectively and their contributions indivi-dually) in this model of the system, where only the ion, waters, and carbonyl oxygens aremobile. Here, the cavity is presumed to be ion free.

With this model, the total stabilization energy is �200 kJ mol�1, which compensatesfor �60% of the ionic dehydration energy. Attraction to the dielectric background (bulkand cavity water) is strongest at the sites near the two boundaries, with bulk water twice aseffective. Stabilization by single file waters is complementary, being strongest in the filterinterior where the ion has two single file neighbors. Net stabilization by waters of all kinds(bulk, cavity, and single file) varies from �20 to �30 kT. The rest of the dielectric stabi-lization is dominated by ion–helix and ion–carbonyl interactions. For filter sites nearestthe cavity, interaction with the �-helices dominates since these are very close to the fourhelices’ carboxyl termini each with an effective negative charge of �0:5e0 (with e0 thecharge of an electron). In contrast, the negatively charged carboxylate groups of the80Asp residues near the peptide–water interface are strongly shielded by nearby bulksolvent, greatly reducing their ability to stabilize filter cations. Co-operativity appears to

FIG. 2 Individual contributions of the various electrical features (continuum background ^, single

file waters &, binding pocket carbonyls ~, oriented �-helices *, negative residues *) and of co-

operativity (x) to monovalent ion-stabilization energy at various occupancy sites in the model KcsA

filter. The water pool does not contain an ion.


make an insignificant contribution to ionic stabilization. These conclusions are basicallyunaltered if the central cavity is ion occupied. However, there is one interesting change. Anion at site 5 interacts with the single file waters just as well as an ion at site 4; because thefields due to the cavity ion and the site 5 ion reinforce, the single file waters are morefavorably aligned than when the cavity is unoccupied.

Finally, consider an example that illustrates the importance of releasing constraintson mobility. The mobile moieties are the ion(s) and waters of the single file, the termini ofthe �-helices, and the explicit waters in the central pool, which may or may not contain anion. The carbonyls and the 80Asps are ignored. As long as joint effects do not alter the sizesof the ionic Born cavities, changes in the permeation free energy track changes in�Gstabilization. The results are presented in Table 5 [72]. For immobile helices, the restpositions of their charged termini are, by definition, the crystallographic positions.Including mobility complicates matters. The termini’s rest positions and the crystallo-graphic positions are not the same, as the structure was determined at �80–100 K [5].Rest positions are deduced by requiring that, for crystallographic occupancy (two filterions and one pool ion), their mean thermal locations at 100 K reproduce the crystallo-graphic data; the force constants are deduced from MDmodeling [41]. The data of Table 5may appear counterintuitive. Because rest positions change, including mobility need notimply increased ion–channel interaction. For immobile modeling, the rest positions of thecarboxyl termini are very close to the inner sites, 4 and 5. With mobility included, the restpositions move away from these sites. At all sites, incorporating mobility decreases thestabilization energy, the effect being more pronounced at the inner sites. While qualita-tively independent of pool occupancy, the effect is amplified for the occupied pool.Naturally, for a specific set of rest positions, energy always decreases when mobility ispermitted. Thus, we find that: co-operative effects may be greater than suggested by theprimitive model; their sign cannot be predicted with certainty; and treating mobility self-consistently may have surprising consequences.

How dependent these various observations are on model limitations is an openquestion. Assuming that only the filter contents and the carbonyl oxygens are mobile isquite restrictive; helix terminus mobility has a considerable influence on co-operativity.Since solvent reorganization is the physical basis for dielectric stabilization, we may expectthat the mobility of the carbonyl carbons also matters. This is especially true for KcsA,where the filter contents and their surrounding carbonyl oxygens are very tightly packed,so that the oxygens appear very immobile, with mean deflections from their equilibrium

TABLE 5 Effect of Helix Terminus Mobilitya on Monovalent Cation

Stabilization Free Energies, Eq. (5), for Single Occupancy of Sites of

Model Filter (Fig. 1)


waters



49 5 File-5, Cavity-0 �71 �65 �51 �51 �4849 0 File-5, Cavity-0 �70 �63 �50 �47 �3749 0 File-6, Cavity-0 �70 �64 �51 �51 �5549 0 File-7, Cavity-0 �71 �65 �52 �54 �5949 0 File-8, Cavity-0 �71 �65 �53 �55 �61a See text.


orientations of �3�. Permitting carbonyl carbon motion might markedly unfreeze theoxygens, with possibly significant energetic consequences. These are unpredictable since,to treat thermal motion self-consistently, the atomic rest positions must first be deduced.Of equal interest is explicit solvent in the outer mouth of the filter and its ability to shieldthe filter contents from the electrostatic influence of 80Asp residues.

III. MEMBRANE INFLUENCES ON GATING AND CHANNEL STABILITY

In their functional forms, channel-forming proteins (or peptides) are embedded in mem-branes, specifically in lipid bilayers. This implies that lipids affect the proteins’ ability toperform their functions by allowing them to form the channel. Changing the membranecomposition may affect the protein’s affinity to the membrane, and thereby influence itsability to act as an ion-transport catalyst. An illustrative example is the channel formed bydimerizing gramicidin-A (GA). The coupling between GA monomers and the membrane isdue primarily to a lipophilic interaction between the peptide’s apolar side chains and themembrane’s lipid core. In this system the gating process is especially simple. It occurswhen monomers on opposite sides of the membrane, each attached to water at the inter-face by means of its tryptophan residues, diffuse along the surfaces and, colliding, self-assemble and form the membrane-spanning dimer. This creates a highly conductive andcomparatively simple ion channel with well-established properties (see Ref. 73 for a com-prehensive review). Its stability depends on bilayer composition. Quite importantly, it hasbeen shown that chemical specificity is a relatively insignificant influence on protein–lipidinteraction [74–76]. Much more important are the membrane’s general physical character-istics, such as its hydrophobic thickness and elastic moduli.

More generally, channel gating requires a conformational change, where mechanicalcoupling to the membrane may affect either kinetics or equilibria. The clearest cases ofmembrane involvement arise in self-assembly of small polypeptides, when channels formfrom peptide monomers of a size comparable to phospholipid molecules [76,77]. In GA thechannel lifetime, �, depends on the membrane’s mechanical properties [76]. Alamethicinforms channels with a wide range of conductances, reflecting different oligomerizationstates; the population equilibrium correlates with changes in the membrane’s mechanicalproperties [78]. It is thus reasonable to expect lipid–peptide mechanical coupling to beimportant in gating processes that involve the self-assembly of small peptides [79] or as incolicin Ia or diphtheria toxin, the translocation of single peptide strands [80,81]. Wherelarger, already assembled structures are involved, the gating machinery is likely to beisolated from the surrounding membrane. Thus, there are only a few instances wherelipids are known to affect gating kinetics and/or equilibria and the mode of coupling isunclear. Equilibria in rhodopsin is sensitive to changes in acyl chain saturation [82].Doping membranes with phosphatidylino-sitol-y,5-bisphosphate alters activation in aninward rectifier potassium channel [83] and inactivation of a Na/Ca exchanger [84].

In this section we only discuss the ‘‘ultimate gating event,’’ i.e., membrane effects informing and breaking the conductive GA channel. Our focus here is the elastic influencesarising from perturbation of the membrane’s thickness in contact with the insertion.

A. Theoretical Background: Elastic Plates and Shells

The elastic problem of a membrane with an insertion is very similar to some classicalproblems of mechanical engineering in the theory of elastic plates and shells. We first give


a brief review of those problems to provide a necessary background for the discussion ofthe membranes.

1. Equilibrium Equations

The free energy of a bent plate is [85,86]

Fbendpl ¼

Zb ½ ð�uÞ2 þ 2ð1� ÞTðuÞ �df ð7Þ

the coefficient b is called the flexural rigidity or cylindrical rigidity of the plate, u ðx; yÞ is the(small) vertical displacement of its surface, is Poisson’s ratio, � @2=@x2 þ @2=@y2 is theLaplace operator, TðuÞ ð@2u=@x@yÞ2 � @2u=@x2 @2u=@y2, and df ¼ dxdy.

The natural extension of this model, more closely corresponding to the description ofmembranes, is the so-called ‘‘floating plate’’ [86]. In a sense this is like a pontoon bridge.Different parts of the floating plate experience deviations uðrÞ, from the flat arrangementopposed by a restoring ‘‘Archimedes force,’’ �auðrÞ. In this case the ‘‘elastic constant’’ adepends on the density of the supporting liquid. The total energy is now

Fpl ¼ F0pl þ

ZauðrÞ2df ð8Þ

(in the literature a is also called the modulus of the foundation). This equation contains both‘‘bending’’ and ‘‘compression’’ components which are also the most important contribu-tors to the membrane deformation energy. Although their origin in lipid bilayers is dif-ferent (e.g., compression is due to steric interaction between the hydrocarbon tails of lipidmolecules opposing the thickness fluctuations), the mathematical analogy provides thebridge between these distinctly different systems. The equation of equilibrium for aplate [the deformational profile uðrÞ] can be derived from the condition that its free energyis a minimum corresponding to the appropriate boundary conditions. To do so, we mustcalculate the variation of Eq. (8), �Fpl. It is important to notice that �Fpl can be expressedas a sum of two contributions [85,86]:

�Fpl ¼ �F surfpl þ �F edge

pl ð9ÞThe first term contains an integral over the surface of the plate:

�F surfpl ¼ 2

Z�uðb�2uþ auÞdf ð10Þ

which is the energy variation for the plate subject to the arbitrary surface deformation�uðrÞ. The second term is the variation in the energy of the edges of the plate. As such it isproportional to a contour integral along all the boundaries of the plate:

�F edgepl ¼

Idl �u pðrÞ þ @�u

@nmðrÞ

�ð11Þ

with

pðrÞ=2b ¼ @�u

@nþ ð1� Þ @

@l

1

2sinð2�Þ @

2u

@x2� @

2u

@y2

!þ cosð2�Þ @

2u

@x@y

" #ð12Þ

mðrÞ=2b ¼ �uþ ð1� Þ sinð2�Þ @2u

@x@y� sin2ð�Þ @

2u

@x2� cos2ð�Þ @

2u

@y2

" #ð13Þ


where n is the normal to the contour of a boundary and � is the angle between n and the x-axis.

In the theory of elastic plates and shells it is sometimes natural to introduce explicitlynot only the external forces gðrÞ acting on the surface, but also external shearing forces Qn

and mechanical moments Mn applied to the edges. The equilibrium conditions are then(see Ref. 86, Section 23)

�Fpl ¼Z�ugðrÞdf þ

Idl �uQnðrÞ þ

Idl@�u

@nMnðrÞ ð14Þ

Further on we limit consideration to situations where a membrane is perturbed on itsinternal boundaries (contacts with peptides) and no other external forces are present:gðrÞ ¼ 0. This leads to

�F surfpl ¼ 0 ð15Þ

and [given that �uðrÞ is arbitrary]b�2uþ au ¼ 0 ð16Þ

Similarly, identifying the contour variations in Eq. (14), we findIdl �u ½pðrÞ �Qn� ¼ 0 ð17Þ

and Idl@�u

@n½mðrÞ �Mn� ¼ 0 ð18Þ

2. Boundary Conditions

The application of Eqs. (17) and (18) depends on the choice of boundary conditions. If Qand M are fixed (denoted as Q;M conditions) on a boundary L, then from Eqs. (17) and(18) it follows that

½pðrÞ �Qn�L ¼ 0 ð19Þ

½mðrÞ �Mn�L ¼ 0 ð20ÞAlternatively, the constraints may be expressed in ‘‘geometrical’’ terms. For instance, theedge of the plate may be ‘‘supported’’ in which case ujL ¼ u0 ¼ const. and/or ‘‘clamped’’ inwhich case rujL ¼ s ¼ const. With these conditions, �ujL ¼ 0 and @�u=@njL ¼ 0, so thatEqs. (17) and (18) are automatically satisfied. In what follows we always assume that bothuðrÞ and ruðrÞ vanish at the external boundary of the membrane, which is associated withareas ‘‘infinitely distant’’ from the insertions:

uð1Þ ¼ 0; ruð1Þ ¼ 0 ð21ÞFor the problems we study, it is impossible to measure or fix the forces acting at the

interfaces, and the geometrical approach is the only one that can be used. For instance, asis discussed below, in the contact between a membrane and an inserted peptide, the‘‘hydrophobic matching’’ condition naturally defines a displacement u0 rather than theshearing force causing that displacement:

uðr0Þ ¼ u0 ð22Þ


Here, r0 is the radius of the inclusion (or the circular hole); implicit to Eq. (22) is cylind-rical symmetry, a simplification that we will generally use.

In some studies [e.g., 87–89] the fourth boundary condition required for solving Eq.(16) is chosen by assuming that the slope of the membrane surface at the edge of theinclusion (the contact slope) is fixed:

rujL ¼ s ð23Þwhich for cylindrical symmetry gives

@u

@�jr0 u0ðr0Þ ¼ s ð24Þ

The constraints of Eqs. (22) and (23) [or Eq. (24)] are denoted as (u; s) boundary condi-tions.

In other cases, the second boundary condition is chosen by assuming that themoments MjL ¼ 0 [90,91], which leads to

mðrÞjL ¼ 0 ð25ÞWith cylindrical symmetry, mð�Þ ¼ @2u=@�2 þ =� @u=@�, and the last equation becomes

u00ðr0Þ þ

r0u0ðr0Þ ¼ 0 ð26Þ

We denote these constraints, Eqs. (22) and (25) [or Eq. (26)], as (u;M ¼ 0) boundaryconditions. They imply that after u0 is fixed, the slope s adjusts to its optimal value.The corresponding value of free energy, Fminðu0Þ; is the minimal free energy correspondingto a given u0:

It is important to notice that imposing ‘‘geometrical’’ restrictions on the systemimplicitly suggests that there are external forces acting at the edges of the inclusion (orthe hole). For instance, the force acting at the edge of cylindrical hole and conjugate to u0is

Q ¼ �Fðu; sÞ�u

js;u¼u0 ¼ 2�r0pðr0Þ ð27Þ

with

pð�Þ ¼ @�u

@�¼ @

3u

@�3þ 1

�

@2u

@�2� 1

�2@u

@�ð28Þ

where pð�Þ is to be calculated from the solutions of Eq. (16) with fixed u0 and s. In the samemanner, the mechanical moment conjugate to s is

M ¼ �Fðu; sÞ�sjs;u0 ¼ 2�r0mðr0Þ

Therefore, for any choice of M and Q one can find corresponding geometrical parametersu0 and s, while the imposition of geometrical conditions determines the correspondingforces. The choice between two alternative descriptions is simply a matter of convenience.

3. Hertz’s and Related Problems

Hertz [92] solved a mechanical problem in which a circular plate bears a single load P at itscenter � ¼ 0. This is mathematically similar to our problem, a membrane containing aninserted peptide. Our discussion is based on that of Ref. 86 and covers the extension


treating circular plates (membranes) with a hole. The fundamental problem is to determinethe membrane’s surface deformation profile, uðrÞ, the solution to Eq. (16) subject to theappropriate boundary conditions; in cylindrical coordinates Eq. (16) becomes

uð4Þ þ 2

�uð3Þ � 1

�2u00ð�Þ þ 1

�3u0ð�Þ þ a

buð�Þ ¼ 0 ð29Þ

Its general solution can be expressed in terms of Kelvin functions:

u ¼ c1keiðxÞ þ c2kerðxÞ þ c3beiðxÞ þ c4berðxÞ ð30Þwith x ¼ �=, where

¼ ðb=aÞ1=4 ð31Þis the characteristic elastic decay length. Using the conditions Eq. (21), we should set bothc3 ¼ 0 and c4 ¼ 0: As a result,

u ¼ c1 keiðxÞ þ c2kerðxÞ ð32ÞThe remaining constants c1 and c2 are defined by the boundary conditions at the plate-insertion interface. Numerous solutions to similar problems with cylindrical symmetry arediscussed in Ref. 93. For instance, if (u; s) conditions are appropriate:

c1 ¼u0kerðx0Þ0 � s kerðx0Þ

ðx0Þð33Þ

c2 ¼s keiðx0Þ � u0 keiðx0Þ0

ðx0Þð34Þ

ðx0Þ ¼ keiðx0Þ kerðx0Þ0 � kerðx0Þ keiðx0Þ0

with x0 ¼ r0=. Using the (u;M ¼ 0) conditions, Eqs. (22) and (26), we find

c1 ¼ u0r0 keiðx0Þ þ ð1� Þkerðx0Þ0

�ð35Þ

c2 ¼ u0r0 kerðx0Þ � ð1� Þ keiðx0Þ0

�ð36Þ

� ¼ ð1� Þðx0Þ þ r0jK0ðffiffiip

x0Þj2

where K0ðzÞ is a modified Bessel function of the second kind. The free energy calculated forthe surface deformation profile, Eq. (32), is

F ¼ 2 � ½ x0 ðc21 þ c22Þffiffiffiffiffiabp

ðx0Þ þ ð1� Þ b u0ðr0Þ2� ð37ÞInstead of solving the Hertz problem directly, with the load P fixed in the center, we

fix uð0Þ ¼ u0, use the solution for the (u;M ¼ 0) conditions [Eqs. (35) and (36)], and onlythen express u0 in terms of P: In the limit r0 ! 0, c2! 0, and

c1 !u0

keið0Þ ¼4u0�

Noticing that in the same limit x0ðx0Þ ! �=4; we find for the energy:

FHertz ¼ 8u2offiffiffiffiffiabp

ð38Þwhere the second term in Eq. (37) drops out because as r0! 0, u0ðr0Þ ¼ u0ð0Þ ¼ 0.


The load P, conjugate to the displacement u0 and acting at the center of the plate,can be found as:

P ¼ @FHertz

@uo¼ 16uo

ffiffiffiffiffiabp

The same result could be derived using the definition of Eq. (28) for pð�Þ and the conditionof force balance [86]:

P ¼ Q ¼ 2�r0pðr0Þjr0!0 [see Eq. (27)]

It is interesting to notice, based on Eq. (38), that, in relation to lifting its central point, aninfinite membrane behaves like a spring with the effective force constant:

KHertzspr ¼ 16

ffiffiffiffiffiabp

ð39Þ

B. Membranes with a Single Insertion

Membranes are self-assembled bilayers of amphiphilic molecules. When a protein isinserted into a membrane, the thickness of the membrane adjusts to match the thicknessof the hydrophobic region of the inclusion in order to minimize the exposure of thenonpolar parts to the aqueous environment. Thus, the natural choice for a boundarycondition is to define a perturbation of the membrane thickness at the interface:

2u0 ¼ h0 � lh ð40Þwhere h0 and lh are the unperturbed hydrophobic membrane thickness and the hydro-phobic length of the channel, respectively [87,88] (see Fig. 3).

The possibility of an exact hydrophobic match, Eq. (40), has been analyzed recently[89]. The elastic energy of matching should be compared with the free energy of hydro-phobic mismatch, i.e., the increase in surface energy due to the contact of the nonpolarlipid tails with water. While the elastic energy is proportional to u20; the mismatch energy isa linear function of u0: This means that for a sufficiently large value of u0 the membrane

FIG. 3 Schematic diagram of the peptide–membrane interface.


would prefer to remain undeformed, gaining some mismatch energy instead. The estimateshows that the matching condition, Eq. (40) is likely to be fulfilled for u0 � 2:65 A, andthat for larger u0 some slippage or incomplete matching may occur. This means that uðr0Þfor larger mismatches may be smaller than the u0 defined by Eq. (40). This importantnotion should be kept in mind, although throughout our discussion we assume that thehydrophobic matching condition is fulfilled.

Being perturbed at the interface, the membrane profile uðrÞ adjusts itself gradually tominimize the elastic energy. The corresponding free energy of membrane deformation canstrongly affect both protein conformation and protein function. For the GA insertionconsidered below, these effects manifest themselves through the influence of membraneparameters (elastic constants, thickness of the bilayer) on the lifetime � of the ion channel,i.e., the dissociation of GA dimers into separate monomers.

1. Elastic Description of Membranes

In theoretical studies of these phenomena, membranes are traditionally described bymeans of the ‘‘smectic bilayer’’ model [76,87,88,94,95], although other approaches havealso been used [96,97]. The original model was designed for the description of smecticliquid crystals [98]. The elastic energy in such systems depends on both the positions andorientations of their constituent molecules. The corresponding degrees of freedom aredescribed through the displacements uðrÞ and a director nðrÞ. Under the assumption thatdeformation is smooth enough, and that the director adjusts to the local normal to thesurface, the deformation energy can be expressed in terms of the uðrÞ alone. Adapted to thedescription of bilayers, the energy in this model consists of three contributions.

1. Splay (‘‘bending’’ or ‘‘curvature’’ energy):

Fbending ¼Z

Kh02ð�uÞ2df ð41Þ

where K is the bending (splay) constant related to the curvature elastic modulus Kc

(Kc ¼ Kh0). Formally, Eq. (41) can be derived from the plate bending energy [Eq. (7)]by the substitutions b! Kh0=2; and T ! 0.

2. Compression–stretching energy:

F stretching ¼Z

2B

h0u2df ð42Þ

where B is the stretching modulus.3. Surface tension contribution:

F tension ¼Z�

2ðruÞ2df ð43Þ

where � is the surface tension coefficient.

The issue of surface tension is quite controversial [99–107]. Fortunately, existingestimates [76,88] and experimental data [108] show that, even when included, this con-tribution is not important. For this reason we omit the surface tension term throughoutthe remainder of this paper.

Considering only the first two terms, and comparing Eqs. (4) and (42) with Eqs. (7)and (8), we find that the problem of a membrane with an insertion is equivalent to that of afloating plate [Eq. (16)], assuming the correspondences:


a ¼ 2B

h0; b ¼ Kh0

2; and T ½u� ¼ 0 ð44Þ

In addition, the surface forces and moments, Eqs. (12) and (13), must be substituted by

pðrÞ ¼ @�u

@nð45Þ

mðrÞ ¼ �u ð46Þ

2. Elastic Problem for a Single Cylindrical Insertion

The elastic problem for a membrane with a cylindrical insertion, accounting for bothcompression and splay contributions, was first studied by Huang [87]. In this study, Eq.(16) was solved using the (u; s) boundary conditions, Eqs. (22) and (24). The generalsolution is given by Eq. (32) with the coefficients c1 and c2 determined from Eqs. (33)and (34) with the characteristic decay length:

¼ Kh204B

!1=4

ð47Þ

In a membrane with a comparatively small stretching modulus is large. Conversely, largestretching and relatively low bending constants lead to a small value of . This result iseasy to understand. If B is large, the most effective way to reduce the energetic cost of athickness perturbation is for uðrÞ to decay to 0 as rapidly as possible. If K were negligible, uwould have dropped abruptly (with infinite slope). If K is finite but small, the decay is stillsteep and is small. Obviously, if B is comparatively small, uðrÞ decays smoothly, so thatthe curvature contribution is small. For the typical solvent-free membrane, is close to 10A (see below). Interestingly, despite a significant variation in B, K , and h0, the values of are typically found in a narrow range, between � 8 and � 12 A. Calculating the membranedeformation energy:

F ¼ 2�

Z1r0

½bð�uÞ2 þ a uð�Þ2�� d �

using the general solution, Eq. (32), and the correspondences, Eq. (44), we find [87]:

F ¼ 2 �x0 ðc21 þ c22ÞffiffiffiffiffiffiffiKBp

ðx0Þ ð48Þan expression quite similar to that for the plate, Eq. (37).

The results for a GA insertion into a glyceryl mono-oleate (GMO) membrane areillustrated in Figs 4 and 5. The parameters are those used in the literature [87,88]:B ¼ 5� 10�8dyn A

�2; K ¼ 10�6dyn; h ¼ 28:5 A. The hydrophobic length of GA in

Eq. (40) is chosen to be lh ¼ 21:7 A, and its external radius is r0 ¼ 10 A. From Eq. (40)it follows that u0 ¼ 3:4 A. The asterisk indicates that these parameters define a ‘‘reference’’set [76].

Figure 4 demonstrates how the total elastic energy, Eq. (48), and its stretching andbending components, depend on the contact slope s. The coefficients c1 and c2 are deter-mined from Eqs. (33) and (34). It is important to notice that in the vicinity of the energyminimum, the stretching contribution strongly dominates bending, while the bending


FIG. 4 Elastic free energy (3), decomposed into its stretching (1) and bending (2) components as

functions of slope s.

FIG. 5 Membrane deformation profiles for s ¼ 0 (1), �0:2 (2), �0:4 (3), �0:6 (4) and �0:8 (5):


energy’s s dependence is much sharper. As has already been indicated, the surface tensioncontribution to the insertion energy is negligible [76,86].

Figure 5 illustrates deformation profiles for various s: An important feature of thesecurves is their nonmonotonic behavior, which can result (depending on the value of s) in aprominent peak at distances � 20–30 A from the insertion, so that in that region themembrane’s thickness exceeds its unperturbed value h0. In addition to their variation withs, the magnitude of these oscillations also depends on the elastic constants B and K[through the dependence of uðrÞ on the decay length ]. These and other properties ofthe deformation profiles have been discussed in detail in Ref. 76. In a sense, this ‘‘over-expansion,’’ in response to thinning of the membrane at the interface, is reminiscent of‘‘overscreening’’ of electrical charges in electron plasmas or electrolytes [109], when theelectrical field in a particular region reverses its direction. These oscillations can be impor-tant for membrane-mediated elastic interactions between insertions. They can becomemore prominent if the discreteness of the membrane (its molecular nature) is taken intoconsideration [110]. It is interesting to note that the possibility of ‘‘overexpansion’’ wasdiscussed in relation to the problems of elastic plates (the Hertz problem) a long time ago.It was noticed that raising of the loaded elastic plate above its unloaded level is neverobserved experimentally [86]. It has also been shown that the modified theory of plates,treating them as three-dimensional bodies, does not normally predict such an ‘‘anomaly.’’This does not mean, however, that such behavior is forbidden in soft materials likemembranes.

C. Experimental Results: Analysis Using Elastic Models

1. Lifetime of Gramicidin Channels and Linear Spring Model

A release of the elastic energy is considered a major factor in dissociation of gramicidinchannels into separate monomers with a corresponding loss of conductance [87,88,111].The dissociation rate constant kdiss can be described as [87,111]

lnfkdissg ¼ � lnf�g ¼ ��G=RT � lnf�0g ð49Þwhere

�G ¼ �Gint þ�Gdef ð50Þis the activation energy for channel dissociation, �Gint is the intrinsic activation energyindependent of u0, and �Gdef is the contribution due to deformation [111]. From thenature of �Gdef , the only part which depends on h0; one can expect that in a group ofsimilar membranes an increase in the difference between the length of the channel hydro-phobic exterior and the thickness of the membrane leads to a higher dissociation rate. Thisis supported experimentally.

Three groups of membranes have been analyzed: monoglyceride/squalene bilayers(thickness changes up to 0.7 nm) [112], monoglyceride/n-hexadecane bilayers (changesup to 2 nm), and monoglyceride/n-decane bilayers (changes up to 1.7 nm) [113]. It hasbeen shown that in each group, ln � is a linear function of the membrane thickness h0: Thisresult naturally leads to a phenomenological ‘‘linear spring’’ model [76,111] for the bilayerdeformation energy due to insertion:

Gdef ðu0Þ ¼ Hð2u0Þ2 ð51Þ


where H is the effective spring constant.* It is assumed that dissociation of GA channelsoccurs abruptly after a certain separation �z is reached between the two monomers con-stituting the channel. While Huang [87] suggested using �z � 0:1 nm, Lundbæk andAndersen [111] proposed that �z � 0:16 nm would be a better choice for a critical separa-tion for breaking of the hydrogen bonds holding the GA monomers intact. The mismatchdependence of the dissociation rate constant is then given by [111]

dðlnfkdissgÞ=du0 ¼ 4H�z=RT ð52ÞThis equation permits expression of the effective spring constant H in terms of the mea-sured slope of d lnðkdissÞ=dh0. The results depend on the choice of �z. For the monoglycer-ide/squalene system, d lnðkdissÞ=du0 � 17:8 nm�1 [111]. Assuming �z � 0:16 nm one finds

Hexp � 69 kJmol�1nm�2 ð53Þ

2. Interpretation of Experimental Data in Terms of Elastic Theory

This result, Eq. (53), posed a challenge to the elastic description of the insertion barrier.For illustration, let us consider the (u;M ¼ 0Þ conditions. As previously mentioned, theboundary condition corresponds to complete minimization of the elastic energy with u0fixed.

The coefficients c1 and c2 are defined by Eqs. (35) and (36) from which it follows that

c1 ¼ u0keiðx0ÞjK0ð

ffiffiip

x0Þj2ð54Þ

c2 ¼ u0kerðx0ÞjK0ð

ffiffiip

x0Þj2ð55Þ

Both c1 and c2 are proportional to u0, which leads directly to the linear spring model:

Fmin ¼ Hmin ð2u0Þ2

with the elastic constant for the minimized energy equal to

Hmin ¼ 4�x0ffiffiffiffiffiffiffiKBp

ðx0ÞjK0ð

ffiffiip

x0Þj2

For the GMO parameters this leads to

HminGMO � 19 kJmol�1nm�2 ð56Þ

The corresponding value of the elastic energy induced by the insertion is [88]

Fmin � 4 kT

with the contact slope smin � �0:45 (see Fig. 2). As we can see, HminGMO is far less than the

experimental prediction, Eq. (53).

* Note that defined in this way, the spring constant is half its conventional value.


Huang [87] was the first to address this challenge. Instead of minimizing F over s, theequivalent of the ðu;M ¼ 0Þ condition, he suggested treating s as a parameter. Its value,s ¼ sexp; was determined by comparison with experiment. As smin � �0:45 corresponds tothe minimum of FðsÞ, any shift of s from this value, either negative or positive, increasesH. Huang considered only small jsj and found that sexp � 0*. This choice has been com-monly accepted and apparently justified in a number of later publications [e.g.,76,89,91,111]).

One of the arguments against energy minimization was presented in Ref. 89. It wassuggested that total free energy consists of a membrane deformation energy term, FðsÞ,and a boundary energy term, EbdðsÞ. The correct solution for smin should then be obtainedfrom F 0ðsÞ þ E 0bdðsÞ ¼ 0: Although a sensible idea, this suggestion leaves some importantquestions unaddressed.

For instance, if the contribution Ebd were important, it would be unreasonable not toinclude it directly in the analysis of the membrane-mediated interaction between theinsertions, and in the energetics of channel formation. However, this has never beenattempted. Also, the interactions responsible for this ‘‘surface’’ contribution cannot belocalized at the geometrical surface, but must affect the membrane’s properties in thevicinity of the insertion. This poses a question of how valid it is to describe the membranewith Eq. (16), which ignores these interactions. It is well known that even well-definedsolid interfaces are actually transition regions where the properties of both phases arestrongly modified. This should be even more the case for the soft contacts involvingmembranes.

D. Nonuniform Membrane Model

Based on these and similar considerations, we decided to attempt to incorporate theinterfacial perturbation directly into the membrane model. Hydrophobic interactionwith the peptide imposes some limitations on the mobility of the molecules and, conse-quently, on their ability to readjust. We describe these limitations through the materialproperties and the membrane elastic constants, and assume that in the vicinity of aninsertion they must differ from their bulk values. The decay length of this perturbationmust be comparable to the length of lipid molecules. Equating these two characteristiclengths, we reduce the number of parameters to two ‘‘surface enhancement factors’’ forboth stretching and bending moduli. As will be seen from our analysis, only the perturba-tion of the stretching constant has a significant effect on the insertion energetics, whicheffectively reduces the number of parameters to one.

The idea that membrane elastic constants could be modified by the insertion is not anew one. It was suggested [76,89] that, on the molecular scale involved in the insertionenergetics, the elasticity coefficients might differ from their macroscopic values (see alsoRef. 114). Furthermore, membranes’ material constants are nonlocal [115–117] (see alsosimilar predictions for the interfacial tension [118]), which implies that their behavior onshort-length scales differs from the macroscopic limit even for uniform membranes, as hasbeen recently observed experimentally [119,120].

* It follows from Fig. 4 that the same increase in energy would also correspond to s � �0:8. There isno criterion for choosing between these two possible values for sexp. Probably his argument was

that a lage negative contact slope would be harder to justify by geometrical (or molecular packing)

arguments.


1. Euler–Lagrange Equation

Assuming that elastic moduli K and B depend on r:

aðrÞ ¼ 2BðrÞh0

; bðrÞ ¼ h0KðrÞ2

ð57Þ

we rewrite Eq. (8) for the nonuniform membrane model as

Fnu ¼Z½bðrÞð�uÞ2 þ aðrÞu2ðrÞ�df

The Euler–Lagrange equation for a membrane with a single cylindrical insertion thentakes the form:

c0uðrÞ þ c1u0ðrÞ þ c2u

00ðrÞ þ c3uð3ÞðrÞ þ c4u

ð4ÞðrÞ ¼ 0 ð58Þwith

c0 ¼ aðrÞ; c1 ¼bðrÞr3� b0ðrÞ

r2þ b00ðrÞ

r;

c2 ¼ �bðrÞr2þ 3

b0ðrÞrþ b00ðrÞ; c3 ¼ 2

bðrÞrþ b0ðrÞ

�; c4 ¼ bðrÞ

ð59Þ

For the nonuniform model, the MjL ¼ 0 condition, Eq. (25) is unaltered, while Eq. (26) ismodified to

@

@r½bðrÞ�u�jr¼r0 ¼ 0 ð60Þ

2. Possible Implications of Nonuniformity

The boundary problem for Eq. (58) was solved [117] using the Shooting Algorithm imple-mented with Mathematica 4 [121]. Comparison with analytical solutions in the uniformlimit demonstrated agreement to within 2% for deformation profiles and to within 5% forthe elastic energy.

We are suggesting that the perturbation of B and K , in contact with the insertion,propagates in the membrane for distances c, comparable to the length of lipid moleculesthemselves, � h0=2: For simplicity, we first presume that the relative distance dependenceof both stretching and bending moduli can be described by the same transition function:

TðrÞ ¼ 1þ ð� � 1Þ exp � r� r0c

� �ð61Þ

so that

aðrÞ ¼ 2B

h0TðrÞ; bðrÞ ¼ 2K

h0TðrÞ ð62Þ

The coefficient � accounts for the relative increase in elastic constants at the interfaceas compared to their bulk values. We shall then consider how independently perturbingthe bending and stretching constants affects the elastic energy.

It can be seen from Fig. 6 that a local increase in elastic constants significantlyincreases the elastic energy barrier. Thus, for GMO a three-fold increase in Emin, whichwould be consistent with the experimental values of the channel lifetimes, requires � � 4.Weobtained similar results for ditridecanoyl-sn-glycero-3-phosphocholine (DTPC) and dimyr-


istoyl lecithin (DMPC) using the literature values of their bulk properties [122] and follow-ing Ref. 76, assuming that a reasonable estimate for the energy barrier is given by settings � 0 in uniform model calculations. In both cases the enhancement parameter � is � 4.

We have assumed that both K and B are scaled by the same transition function TðrÞ.However, the surface perturbation of these moduli may and must be different. Localincrease in K is mainly due to the restricted orientational mobility of the lipid moleculesnear the inclusion, while variation of B is mainly due to reduced conformational mobilityof the ‘‘tails’’ of lipid molecules responsible for the resistance to compression (obviously,these effects are not completely independent). Therefore, a more accurate model mustinclude two surface constants, �K and �B. To estimate their relative influence we studythem independently. In other words, two limits are considered, �K ¼ �, �B ¼ 0 and �K ¼ 0,�B ¼ �. The results of this analysis are presented in Fig. 7.

It is clear that perturbation of the stretching constant (increase in �B) has a muchgreater effect on the elastic energy barrier, Emin, than the increase in �K . In fact, theperturbation of B alone is responsible for about 80% of the total increase in the elasticbarrier. Moreover, if only the bending constant were affected, than the threefold increasein Emin would require �K � 40(!). These tendencies are illustrated in Fig. 8.

In addition, one can see that �B and �K drive the optimal slope smin in oppositedirections. Increasing �B tends to make uðrÞ steeper, thus more effectively reducing thestretching energy, while increasing �K has the opposite effect. As we can see, the energybarrier is amazingly insensitive to the surface perturbation of the bending modulus. On theother hand, variation of �K greatly affects the shape of FðsÞ. Such behavior is similar tothat found in the uniform model, where bending mainly influences the s-dependence of theenergy, but not the equilibrium energy barrier (see the previous discussion, and Refs 76and 88).

It should be noted that with � � 4 (the value required to account for the energybarrier to GA dimerization in GMO), the contact slope is smin � �0:39: This reduction in saccords with suggestions made previously [76,87]. However, in our approach it is a directconsequence of complete optimization of the nonuniform energy functional.

FIG. 6 Elastic energy versus contact slope, s, as a function of the enhancement factor, yK yB y:(1) y ¼ 1, ——; (2) y ¼ 3, *; (3) yK ¼ 5, &.


Let us now consider how our treatment differs from the conventional approach ofusing s as an adjustable parameter. We have suggested that there is a nonuniform increasein the membrane elastic constants in the vicinity of the peptide inclusion. Based on thisassumption, we allowed the deformation profile to attain its equilibrium slope smin, andfound the corresponding energy barrier or, equivalently, the effective elastic constant H.Thus, instead of adjusting s, we have stressed the possibility of local perturbation of theelastic constants. We found that, setting � � 4 yields results equivalent to fixing the contactslope s � 0 in the uniform model. Even though the conclusions are similar, we think thenonuniform model has some distinct advantages:

FIG. 7 Elastic energy versus contact slope, s, as separate functions of bending (yK ) and stretching

(yB) enhancement factors: (1a,b) yB;K ¼ 1, —; (2a) yK = 3,*; (3a) yK = 5,&; (2b) yB = 3, *; (3b)

yB= 5, &.

FIG. 8 Energy barrier, Fmin, as a function of the enhancement factor y if only: (1) the stretching

factor yB, *; or (2) the bending factor yK , &, is perturbed by the insertion.


1. It is widely presumed that in the vicinity of insertion the elastic properties mustdiffer from the macroscopic limit [76,89]. However, this notion has not previously beenimplemented in a computational model.

2. The argument for fixing the slope requires introducing an additional ‘‘surfaceenergy’’ component [89] which is otherwise not present in the picture. There is a reason tobelieve that such a perturbation not only influences the contact slope, but also directlyaffects the barrier and modifies the membrane properties. That is why we think that thesurface perturbation should be explicitly incorporated in the membrane model.

3. It is worthwhile formulating a model in terms of material properties that aremeasurable (or at least, potentially so). For instance, the short-range membrane thicknessfluctuations at a point r depend on the averaged values of the elastic constants in itsvicinity:

< uðrÞ2 >� kT

Beff ðrÞwhere Beff ðrÞ is an effective elastic constant averaged in the neighborhood of the point r.

These fluctuations may be studied by neutron scattering, spin labeling, and x-rayscattering techniques. There is also evidence from molecular dynamics simulations thatlocal fluctuations near inclusions are smaller than those in the unperturbed bilayers[123,124].

IV. MEMBRANE-MEDIATED INTERACTION BETWEEN INCLUSIONS

Biological membranes contain large numbers of insertions, e.g., embedded proteins andcholesterols. Interaction between insertions includes direct forces, such as electrostatic andvan der Waals, and also interactions mediated by the membrane. Our focus here is on theinteractions caused by the insertion-induced elastic deformations in membranes.

A number of studies in this field [97,100,125,126] have considered two contributionsto the bilayer energy: molecular stretching/compression and the interfacial tension. Yetothers have focused on the membrane’s resistance to bending [127,128]. Since the pioneer-ing studies of Huang [87] and of Helfrich and Jakobsson [88] (see also Section III) it wasclear that membrane deformation is mostly governed by bending and stretching elasticity.The role these forces play in membrane-mediated interaction between insertions has beenconsidered for two flat insertions [129] and for a two-dimensional hexagonal lattice ofcylindrical insertions [90,94]. More recently, this problem has been studied for an ensembleof cylindrical insertions (GA channels) and their aggregation, simulated using a MonteCarlo approach [89].

Here, we describe the problem of two or more cylinders, trying to determine whencollective (non pairwise) interactions become important. To do this, we develop a high-precision algorithm for solving the Euler–Lagrange equation, Eq. [58], for several insertions.

A. Finite Difference Approach to Elastic Deformations Caused byInclusions

The technical problem that arises when several inclusions are present is that the system nolonger exhibits cylindrical symmetry. The Euler–Lagrange equation:

�½bðx; yÞ �uðx; yÞ� þ aðx; yÞ uðx; yÞ ¼ 0 ð63Þ


[the generalization of the cylindrically symmetric representation, Eq. (58)] can still besolved numerically using a finite difference representation [130] on a two-dimensionalmesh. We have developed a discretization scheme for solving this equation that can beused for any number of inclusions [131]. The computational domain for a single inclusionis shown in Fig. 9. Since both u and ru must be specified at the boundaries, in a finitedifference representation this implies that u is constrained at two layers of grid points tosatisfy boundary conditions at both the periphery of the computational domain and at theedge of each insertion. At the external edge of the computational domain and at theimmediately neighboring points we set u ¼ 0, which is equivalent to using the boundaryconditions, Eq. (21).

For the boundary conditions at the insertion edges (the circles that are the two-dimensional projections of the included cylinders) we presume that u is fixed: ujcircle ¼ u0:In addition, we constrain the slope s on these circles (in some cases s is not held constant,but varies along the boundary circle with a prescribed functional dependence). Theseassumptions are implemented through the condition:

uði; jÞ ¼ u0 þ sdði; jÞ

for the points ði; jÞ nearest the circumference both inside and outside the circle. Here,dði; jÞ ¼ rði; jÞ � r0 is the distance between the point ði; jÞ and the circumference of the circleand rði; jÞ is the distance from that point to the center of a circle. In most cases we choseu0 ¼ 3:4 A typical of GMO. (More commonly, peptide interaction is studied in thinnermembranes, such as dilauroylphosphatidylcholine (DLPC) or dimyristoylphosphatidyl-choline (DMPC), corresponding to u0�< 1:5 A [89,90,132]. Consequently, our choice ofu0 exaggerates the interactions. However, it does not affect the qualitative results).

We first tested the finite difference technique for a single inclusion, which has asimple analytical representation (in the uniform case) or a cylindrically symmetric, simplenumerical solution (the nonuniform case, Section III). We used the approximation of Eq.

FIG. 9 Schematic view of the computational domain: &, first layer of the boundary points; *,

second layer of the boundary points; & , interior points of the computational domain.


(62) for the elastic constants. The coefficient �, Eq. (61), was either equated to 1 (theuniform case) or to 4, which we showed in Section III accounted for experimental lifetimesin the GA channel.

Calculations were carried out for different mesh spacings h (0.5, 1, and 2 A).Regardless of the value of h, the deformation profiles uðrÞ were in excellent agreement(to within 2%) with the results described in Section III for both values of �. However, thetotal deformation free energy is very sensitive to the choice of h.

Two main sources of error affect the energy integral. First, the surface of distortionis calculated using a finite number of points. Therefore, uðx; yÞ defined on a discrete set ofpoints is an approximation the precision of which is very sensitive to the choice of h. Thebending energy, determined by the second derivatives of u, is especially dependent.Second, in our approach the circles are approximated by zigzag strips of grid points.Given that uðx; yÞ is both maximal in these regions and its variation is greatest, thealteration of the boundary’s shape may introduce another source of error.

Both these errors decrease with refinement of the grid. We found that the gridspacing h ¼ 0:5 A (which is close to the limit of our computational capability) reproducesprevious results with good accuracy. Comparison of the deformation free energy profilesas a function of the slope is shown in Fig. 10. The agreement between numerical andanalytical data is especially good in the physically most interesting range of slopes (�0:5 to0.0). The error becomes greater for large negative slopes; for s ¼ �1:0 the error in thedeformational energy is about 20%.

B. Two Inclusions

We have solved the uniform (� ¼ 1) elastic problem numerically for two inclusions atvarious separations, d. We chose u0 ¼ 3:4 A and varied s. The deformation profilesalong the line connecting the centers of two insertions are presented in Fig. 11 fors ¼ smin. As d increases, the well in uðxÞ becomes deeper. It reaches its minimal value,

FIG. 10 Deformation free energies versus slope in the uniform (� = 1) and nonuniform (� ¼ 4)

cases. The results of one-dimensional calculations explicitly utilizing the problem’s cylindrical

symmetry (CS) either analytically (the uniform case) or numerically (nonuniform model, see

Section III) are compared with finite difference (FD) results obtained with the two-dimensional

grid.


u � �0:2 A, for d � 60 A. For larger d it splits into separate minima with a maximum inthe middle. Only for d �> 80 A is the deformation profile reminiscent of the superposedprofiles for two independent insertions.

For the relaxed boundary conditions ½s ¼ sminðdÞ� the elastic interaction between twoinclusions is repulsive; minimized over s, the interaction free energy decreases with increas-ing d, Fig. 12. It is interesting to analyze the free energy, FðdÞ, as a function of the contactslope s: We consider the range 0:0 � s � �0:6. When s ¼ 0, FðdÞ increases as d increases,i.e., the inclusions attract one another. At large negative slopes the interaction is repulsive.In the intermediate region, �0:15�> s�> �0:35; a minimum exists at finite distancesbetween the insertions, which disappears at scr � �0:35. In addition, a barrier separatesit from the large d region. These results may be physically significant. If the slope s canreally be controlled by the insertion, and if it is different for different lipids, one canexpect, based on these results, a change in the peptides’ clustering behavior (phase transi-

FIG. 11 Displacement profiles between two inclusions along the X axis passing through the centers

of the insertions, for different distances between the inclusions: (1) 10 A; (2) 20 A; (3) 30 A; (4) 40 A;

(5) 60 A; (6) 80 A; (7) 120 A.

FIG. 12 Elastic energy per insertion for the interaction of two peptides as a function of the contact

slope. The distances d are: (1) 1 A; (2) 2 A; (3) 5 A; (4) 7.5 A; (5) 10 A; (6) 20 A; (7) 40 A.


tion) as the membrane’s lipid composition changes. We can expect that similar effects canbe formulated in terms of the nonuniform elastic constants, conditions under which �differs from 1.

We also tested the validity of the assumption that the contact slope remains cylin-drically symmetrical when there is more than one inclusion. Actually, along different partsof the insertion surface the membrane’s surface may exhibit different equilibrium slopes.One would expect that the perturbation of s should be more significant on the side facingthe neighboring insertion. To test this idea, we approximated the slope by a primitive trialfunction:

sð’Þ ¼ sþ s1 cos ’ ð64Þwhere s and s1 are parameters and ’ is the ‘‘azimuthal’’ angle. In Eq. (64) the quantity s1 isadded to the average slope s at the segment, ’ ¼ 0, and subtracted at the segment on theopposite side of the circle, ’ ¼ �. The free energy of membrane deformation per inclusionas a function of s is shown in Fig. 13 for various s1. Increasing s1 initially reduces the freeenergy. The optimal values of s and s1 depend on d. This example shows that angularvariation of the contact slope may play some role in the interaction between the insertions.However, due to the restricted trial function that we introduced, this only provides a crudeinitial estimate. Further analysis requires either using a more advanced set of trial func-tions, or the solution of the ðs;M ¼ 0Þ boundary problem.

C. Many-Body Effects in Membrane-Mediated Interaction BetweenInsertions

We have shown that, minimized with respect to the contact slope, the interaction betweentwo insertions is repulsive. On the other hand, the interaction involving a larger numbersof insertions may become attractive. This has been observed in a model system of twointeracting flat walls [129], which can approximate the interaction between two parallelarrays of insertions. Similarly, an attractive region is present in the interaction free energyfor a two-dimensional array of cylindrical inclusions [90,94]. Clustering was also seen in

FIG. 13 Deformation free energy per inclusion for two inclusions versus average slope s, with

asymmetric boundary conditions; distance d ¼ 10 A. The slope anisotropy parameter s1 are (1)

0.0; (2) 0.2; (3) 0.4; (4) 0.5.


the Monte Carlo modeling of an ensemble of inclusions [89]. These predictions raise theissue of multiparticle effects in membrane-mediated interaction. Here, we compare theresults for systems constituted of two, five, and seven cylindrical insertions.

First, we consider five inclusions, with one at the center and four at the vertices of asquare. Figure 14 demonstrates how the deformation free energy per inclusion depends onthe contact slope for several distances d between the central peptide and its neighbors (d isthe radial distance between the outer surfaces, so that d ¼ 0 corresponds to direct contact).For the minimized energy, there is no attraction. However, it differs quantitatively fromthat seen when only two inclusions are present (Fig. 12). The free energy for five insertionsis much less dependent on d. As d decreases from 10 to 1 A the energy increase perinclusion is only 0.4 kT compared to 2.5 kT for the case of two inclusions.

We have also considered seven inclusions, with one at the center and six at thevertices of a regular hexagon. A contour map of the distortion field is shown in Fig. 15.At the separation illustrated, d ¼ 15 A , the minimum in the free energy occurs for a slopeof s ¼ �0:35: The closely packed contours near the boundaries of the inclusions demon-strate steep distortion gradients. As is seen from our calculations, seven inclusions canform a stable aggregate. The free energy profiles demonstrating this effect are shown inFig. 16. They indicate that the attractive region is separated by a barrier from the repulsiveone. The shape of the curves is similar to some of the predictions of [90,94] for a two-dimensional hexagonal lattice, where a slightly different elastic membrane model has beenused. Our approach does not rely on a Wigner–Seitz type approximation reducing the two-dimensional problem to a one-dimensional (cylindrically symmetric) one. Thus, it has thepotential of being able to treat general configurations and arbitrary numbers of insertions,and to model their aggregation. In this sense it is similar to that described in Ref. 89, butour method is more accurate (the grid spacings can be roughly eight times smaller, whichcould be critical for the evaluation of energy integrals) and has the potential for treating anonuniform model.

Similar to the case with two insertions, the free energy behavior changes if the slope sis fixed. Again, at small jsj � 0, the interaction is attractive (see Fig. 17), while at largenegative slopes, s�< � 0:4, it is repulsive. The intermediate region corresponds to nonmo-notonic sigma-type curves. It is interesting that, in agreement with previous results [89], at

FIG. 14 Deformation free energy (per insertion) versus slope s for five insertions. The distances d

are (1) 1 A; (2) 3 A; (3) 5 A; (4) 10 A; (5) 20 A.


FIG. 15 Distortion field for seven insertions, one at the center and six at the vertices of a regular

hexagon. The distance between the surfaces of the neighboring insertions is d ¼ 15 A; the contact

slope s ¼ smin ¼ �0:35.

FIG. 16 Elastic energy per inclusion for seven insertions as functions of the contact slope s. The

distances d between the neighboring insertions are: (1) 1 A; (2) 3 A; (3) 5 A; (4) 10 A; (5) 15 A; (6) 20

A; (7) 30 A.


larger concentrations (seven insertions, small d) the equilibrium smin shifts towards s � 0:However, in no way can this serve as an argument for fixing s � 0 for a single (isolated)insertion.

V. SUMMARY

The semimicroscopic (SMC) method outlined in Section II provides a middle groundbetween microscopic and mesoscopic approaches to understanding free energy profilesof ions permeating transmembrane channels. By focusing on a limited set of ‘‘criticalinteractions,’’ determination of the free energy profile for a solvated biomolecule can bedetermined very efficiently. Ion-induced solvent reorientation is described rigorously,including all long-range electrostatic contributions. We presented a simple model forcefield for ionic hydrates with intuitively reasonable parameters; SMC ionic Born radii areroughly equal to their physical crystal radii. We then considered the physiologicallyimportant KcsA K-channel. Decomposition of the influence that individual structuralfeatures have on ion stabilization in the narrow filter showed that:

1. The central water cavity (basically designed to stabilize cations in the cavity)helps stabilize ions at the cavity–filter boundary, which may account for thelocation of the barium block site.

2. The oriented �-helices contribute importantly to ion stabilization at the interiorsites of the filter, again with implications for the stabilization of divalent ions.

Preliminary work incorporating flexibility indicates that movement of the �-helical terminisubstantially affects co-operativity, and suggests that electrostatically induced reorienta-tion of moieties quite distant from the solvated ion(s) may significantly influence stabiliza-tion of the filter’s ionic contents.

We presented a new approach allowing reconciliation of the requirement of self-adjustment of the membrane deformation profile (relaxed boundary conditions) with (1)the notion that membrane properties must be modified at short distances from the inser-

FIG. 17 Deformation energy per inclusion versus the distance d, for the optimized contact slope

smin: (a) two inclusions; (b) five inclusions; (c) seven inclusions and for the contact slope s ¼ 0: (d)

two inclusions; (e) five inclusions; (f) seven inclusions.


tion, and (2) the comparatively high experimental values of the elastic energy of insertion.This model suggests certain local stiffening of the membrane elastic constants. This shouldaffect lipid fluctuations in the vicinity of insertions and may be experimentally studiedusing neutron scattering, x-ray scattering, and spin labeling techniques, and computation-ally, using molecular dynamics simulation.

In our discussion of deformation phenomena, some important features were notconsidered (due to space limitations). One of them, which has received much attention,is the influence of the monolayer equilibrium curvature on the membrane deformation[91,133]. These features should be addressed in further development of the nonuniformmodel.

The elastic membrane model, formulated in terms of elastic moduli and uðrÞ, pro-vides a significantly reduced description of insertion phenomena. More detailed analysisshould account for the orientation and displacement of the lipid molecules as well as someof their internal degrees of freedom. A step in this direction has been made, for instance, inRef. 95. At short-length scales and near nonuniformities, lipid molecules cannot attain the‘‘normal’’ orientations typical of their mean behavior on a macroscopic scale, which mustinevitably affect their elastic properties. More detailed statistical mechanical analysis andsimulational studies might provide useful insight into such behavior.

Nonmonotonic behavior of membrane-mediated interactions and significant many-body effects could be responsible for the complex phase behavior of the peptides incor-porated in the membranes, and its dependence on the membrane lipid composition.Interestingly, a similar variation of the interaction potential for two insertions (of radii r0¼ 9 A) as a function of lipid composition has been found recently in a simulational study[132]. We are planning to study these transitions in the composition–protein concentrationphase plane. Another interesting issue is the possibility that peptide clustering can affectthe lifetime of ion channels, a study which is already under way.

ACKNOWLEDGMENTS

We greatly appreciate valuable and stimulating discussions with Dr. Yury Neustadt(Samara State University, Russia) about various problems in the theory of elastic platesand shells. This work was supported by the National Center for SupercomputingApplications and by the National Institutes of Health, Grant no. GM-28643.

REFERENCES

1. B Hille. Ionic Channels of Excitable Membranes. Sunderland, MA: Sinauer Associates, 1992.

2. P Agre, GM Preston, BL Smith, JS Jung, S Raina, C Moon, WB Guggino, S Nielsen. Am J

Physiol 265:F463–F476, 1993.

3. C Maurel, J Reizer, JI Schroeder, MJ Chrispeels, MH Saier. J Biol Chem 269:11869–11872,

1994.

4. P Lauger. Electrogenic Ion Pumps. Sunderland, MA: Sinauer Associates, 1991.

5. DW Doyle, JM Cabral, RA Pfuetzner, A Kuo, J Gulbis, SL Cohen, BT Chait, R MacKinnon.

Science 280:69–77, 1998.

6. A Solomon. J Gen Physiol 51:335s–364s, 1968.

7. DG Levitt. Biophys J 13:186–206, 1973.

8. DP Tieleman, PC Biggin, GR Smith, MSP Sansom. Q Rev Biophys 34:473–561, 2001.


9. VL Dorman, MB Partenskii, PC Jordan. Biophys J 70:121–134, 1996.

10. VL Dorman, S Garofoli, PC Jordan. In: DJ Chadwick, G Cardew, eds. Gramicidin and

Related Ion-Channel Forming Peptides. Chichester, UK: John Wiley, 1999, pp 153–169.

11. S Garofoli, G Miloshevsky, VL Dorman, PC Jordan. In: DJ Chadwick, J Goode, eds. Ion

Channels—from Atomic Resolution to Functional Genomics. Chichester, UK: John Wiley,

2002, pp 109–122.

12. S Koneshan, JC Rasaiah, RM Lynden-Bell, SH Lee. Phys Chem B 102:4193–4204, 1998.

13. DA Dougherty. Science 271:163–168, 1996.

14. AS Arseniev, AL Lomize, IL Barsukov, VF Bystrov. Biolog Membr (USSR) 3:1077–1104,

1986.

15. TA Cross. Methods Enzymol 289:672–696, 1997.

16. R Dutzler, EB Campbell, M Cadene, BT Chart, R MacKinnon. Nature 415:287–294, 2002.

17. G Chang, RH Spencer, ATL At, MT Barclay, DC Rees. Science 282:2220–2226, 1998.

18. K Murata, K Mitsuoka, T Hirai, T Walz, P Agre, JB Heymann, A Engel, Y Fujiyoshi Nature.

407:599–605, 2000.

19. D Fu, A Libson, LJW Miercke, C Weitzman, P Nollert, J Krucinski, RM Stroud. Science

290:481–486, 2000.

20. SH Heinemann, H Terlau, W Stuhmer, K Imoto, S Numa. Nature 356:441–443, 1992.

21. J Yang, PT Ellinor, WA Sather, JF Zhang, RW Tsien. Nature 366:158–161, 1993.

22. A Myazawa, Y Fujiyoshi, M Stowell, N Unwin. J Mol Biol 288:765–786, 1999.

23. K Imoto, C Busch, B Sakmann, M Mishina, T Konno, J Nakai, H Bojo, Y Mori, K Fukuda,

S Numa. Nature 335:645–648, 1988.

24. DJ Adams, TM Dwyer, B Hille. J Gen Physiol 75:493–510, 1980.

25. C Miller, Phil Trans Roy Soc London B 299:401–411, 1982.

26. B Roux. Biophys J 71:3177–3185, 1996.

27. J Neyton, C Miller. J Gen Physiol 92:569–586, 1988.

28. TW Allen, A Bliznyuk, AP Rendell, S Kuyucak, S-H Chung. J Chem Phys 112:8191–8204,

2000.

29. D Meuser, H Splitt, R Wagner, H Schrempf. FEBS Lett 462:447–452, 1999.

30. E Jakobsson, S-W Chiu. Biophys J 52:33–45, 1987.

31. SW Chiu, E Jakobsson, S Subramaniam, JA McCammon. Biophys J 70:121–134, 1991.

32. B Roux, M Karplus. Annu Rev Biophys Biomol Struct 23:731–761, 1994.

33. R Pomes, B Roux. Biophys J 71:19–39, 1996.

34. TB Woolf, B Roux. Biophys J 68:876–892, 1997.

35. B Jayaram, DL Beveridge. Annu Rev Biophys Biomol Struct 25:367–394, 1996.

36. WL van Gunsteren, FJ Luque, D Timms, AE Torda. Annu Rev Biophys Biomol Struct

23:847–863, 1994.

37. S Tanizaki, PC Jordan, unpublished results, 2000.

38. GC Groeneboom, E Mas, R Bukowski, K Szalewicz, PES Wormer, A van der Avoird. Phys

Rev Lett 84:4072–4075, 2000.

39. IH Shrivastava, MSP Sansom. Biophys J 78:557–570, 2000.

40. J Aqvist, V Luzhkov. Nature 404:881–884, 2000.

41. S Berneche, B Roux. Biophys J 78:2900–2917, 2000.

42. L Guidoni, V Torre, P Carloni. FEBS Lett 477:37–42, 2000.

43. TW Allen, S Kuyucak, S-H Chung. Biophys J 77:2502–2516, 1999.

44. PC Biggin, GR Smith, I Shrivastava, S Choe, MSP Sansom. Biochim Biophys Acta 1510:1–9,

2001.

45. B Corry, TW Allen, S Kuyucak, S-H Chung. Biochim Biophys Acta 1509:1–6, 2000.

46. D Boda, DD, Busath, D Henderson, S Sokolowski. J Phys Chem B 104:8903–8910, 2000.

47. W Nonner, L Catacuzzeno, B Eisenberg. Biophys J 79:1976–1992, 2000.

48. W Nonner, D Gillespie, D Henderson, B Eisenberg. J Phys Chem B 105:6427–6436, 2001.

49. B Roux, R MacKinnon. Science 285:100–102, 1999.

50. SH Chung, TW Allen, M Hoyles, S Kuyucak. Biophys J 77:2517–2533, 1999.


51. M Partenskii, M Cai, P Jordan. Chem Phys 153:125–131, 1991.

52. M Partenskii, M Cai, P Jordan. Electrochim Acta 36:1753–1756, 1991.

53. MB Partenskii, PC Jordan. Quart Rev Biophys 24:477–510, 1992.

54. MB Partenskii, VL Dorman, PC Jordan. Biophys J 67:1429–1438, 1994.

55. BR Brooks, RE Bruccoleri, BD Olafson, DJ States, S Swaminathan, M Karplus. J Comput

Chem 4:187–217, 1983.

56. SJ Weiner, PA Kollman, DA Case, UC Singh, C Ghio, G Alagona, S Profeta Jr, P Weiner. J

Am Chem Soc 106:765–784, 1984.

57. HJC Berendsen, JPM Postma, WF van Gunsteren, J Hermans. In: B. Pullman, eds.

Intermolecular Forces. Dordrecht: Reidel, 1981, pp 331–342.

58. JB Hasted. In: F Franks, ed. Water, a Comprehensive Treatise. vol. 1. New York: Plenum

Press, 1973, pp 405–458.

59. DR Lide, ed. Handbook of Chemistry and Physics. 78th ed. Boca Raton, FL: CRC Press,

1998.

60. VL Dorman, PC Jordan. J. Chem. Phys. (submitted).

61. L Pauling. The Nature of the Chemical Bond. 3rd ed. Ithaca, NY: Cornell University Press,

1960.

62. E Grunwald. Thermodynamics of Molecular Species. New York: Wiley-Interscience, 1997.

63. H Reiss, A Heller. J Phys Chem 89:4207–4213, 1985.

64. MD Tissandier, KA Cowan, WY Feng, E Gundlach, MH Cohen, AD Earhart, JV Coe, TR

Tuttle. J Phys Chem A 102:7787–7794, 1998.

65. DL Beveridge, GW Schnuelle. J Phys Chem 79:2562–2566, 1975.

66. PC Jordan. Ionic energetics in narrow channels. Proceedings of the IMA Workshop on

Membrane Transport and Renal Physiology. New York: Springer-Verlag, 2002.

67. JA Odutola, TR Dyke. J Chem Phys 72:5062–5070, 1980.

68. R Guidelli. J Chem Phys 92:6152–6160, 1990.

69. Y Jiang, R MacKinnon. J Gen Physiol 115:269–272, 2000.

70. PC Jordan. Mol Phys 25:961–973, 1973.

71. PC Jordan. Chemical Kinetics and Transport. New York: Plenum Press, 1979.

72. G Miloshevsky, PC Jordan, 2001.

73. OS Andersen, RE Koeppe II. Physiol Rev 72:S89–S158, 1992.

74. P Devaux, M Seigneuret. Biochim Biophys Acta 822:63–125, 1983.

75. A Bienvenue, J Marie. Curr Top Membr 40:319–354, 1994.

76. C Nielsen, M Goulian, OS Andersen. Biophys J 74:1966–1983, 1998.

77. SM Bezrukov, RP Rand, I Vodyanoy, VA Parsegian. Faraday Disc 111:173, 1998.

78. S Keller, S Bezrukov, S Gruner, W Tate, I Vodyanoy, VA Parsegian. Biophys J 65:23–27,

1993.

79. K Akerfeldt, J Lear, Z Wasserman, L Chung, W Degrado. 26:191–197, 1993.

80. K Jakes, P Kienker, A Finkelstein. Q Rev Biophys 32:189–205, 1999.

81. K Oh, L Senzel, R Collier, A Finkelstein. Proc Nat Acad Sci USA 96:8467–8470, 1999.

82. B Litman, D Mitchell. Lipids 31:S193, 1996.

83. C Huang, S Feng, DW Hilgemann. Nature 391:803–806, 1998.

84. Z He, S Feng, Q Tong, D Hilgemann, K Philipson. Am J Physiol 278:C661–C666, 2000.

85. L Landau, E Lifshitz. Theory of Elasticity. Oxford: Butterworth-Heinemann, 1970.

86. S Timoshenko, S Woinowsky-Krieger. Theory of Plates and Shells. New York: McGraw-Hill,

1959.

87. H Huang. Biophys J 50:1061–1070, 1986.

88. P Helfrich, E Jakobsson. Biophys J 57:1075–1084, 1990.

89. T Harroun, W Heller, T Weiss, L Yang, H Huang. Biophys J 76:3176–3185, 1999.

90. H Aranda-Espinoza, A Berman, N Dan, P Pincus, S Safran. Biophys J 71:648–656, 1996.

91. C Nielsen, OS Andersen. Biophys J 79:2583–2604, 2000.

92. H Hertz. Wiedemann’s Ann Phys Chem 22:449–461, 1884.

93. F Schleicher. Kreisplatten auf elasticher Unterlage. Berlin: Springer, 1926.


94. N Dan, A Berman, P Pincus, S Safran. J Phys II (France) 4:1713–1725, 1994.

95. A Ring. Biochim Biophys Acta 1278:147–159, 1996.

96. J Owicki, M Springgate, H McConnell. Proc Natl Acad Sci USA 75:1616–1619, 1978.

97. JC Owicki, HM McConnell. Proc Natl Acad Sci USA 76:4750–4754, 1979.

98. P-G de Gennes, J Prost. The Physics of Liquid Crystals. Oxford: Clarendon Press, 1995.

99. D Bensimon, F David, S Leibler, A Pumir. J Phys France 51:659–695, 1990.

100. M Bloom, E Evans, O Mouritsen. Q Rev Biophys 24:293–397, 1991.

101. D Sornette, N Ostrowsky. In: W Gelbart, A Ben-Shaul, D Roux, eds. Micelles, Membranes,

Microemulsions, and Monolayers. New York: Springer-Verlag, 1994, pp 251–302.

102. D Marsh. Biochim Biophys Acta 1286:183–223, 1996.

103. F Jahning. Biophys J 71:1348–1349, 1996.

104. SE Feller, R Pastor. Biophys J 71:1350–1355, 1996.

105. R Goetz, R Lipowsky. J Chem Phys 108:7397–7409, 1998.

106. H Tien, A Ottova-Leitmannova. Membrane Biophysics as viewed from Experiemental Bilayer

Lipid Membraness. Amsterdam, New York: Elsevier, 2000.

107. MB Partenskii, PC Jordan. Biophys J 78:325A, 2000.

108. M Goulian, O Mesquita, D Fygenson, C Nielsen, O Andersen, A Libchaber. Biophys J

74:328–337, 1998.

109. MB Partenskii, PC Jordan. Mol Phys 98:193–200, 2000.

110. MB Partenskii, PC Jordan. Biophys J 80:544A, 2001.

111. JA Lundbæk, O Andersen. Biophys J 76:889–895, 1999.

112. J Elliot, D Needham, J Dilger, D Haydon. Biochim Biophys Acta 735:95–103, 1983.

113. H Kolb, E Bamberg. Biochim Biophys Acta 464:127–141, 1977.

114. J Boon, S Yip. Molecular Hydrodynamics. New York: Dover Publications, 1980.

115. M Partenskii, V Dorman, P Jordan. In: R Lieberman, T Vo-Dinh. Progress in Biomedical

Optics. Proceedings of Biomedical Sensing and Image Technologies. vol. 3253. San-Jose, CA:

SPIE, 1998, pp 266–278.

116. M Partenskii, V Dorman, P Jordan. J Chem Phys 109:10361–10371, 1998.

117. M Partenskii, P Jordan. In: AG Volkov, ed. Liquid Interfaces in Chemical, Biological, and

Pharmaceutical Applications, vol. 95 of Surfactant Science Series. New York: Marcel Dekker

2001, pp 51–82.

118. K Mecke, S Dietrich. Phys Rev E 59:6766–6784, 1999.

119. C Fradin, A Braslau, D Luzet, D Smilgies, M Alba, N Boudet, K Mecke, J Daillant. Nature

403:871–874, 2000.

120. S Chen, CY Liao, H Huang, T Weiss, M Bellisent-Funel, F Sette. Phys Rev Lett 86:740–743,

2001.

121. S Wolfram. The Mathematica Book. 4th ed. New York: Cambridge University Press, 1999.

122. W Rawicz, K Olbrich, T McIntosh, D Needham, E Evans. Biophys J 79:328–339, 2000.

123. T Woolf, B Roux. Proteins Struct Funct Genet 24:92–114, 1996.

124. H Petrache, J Killian, RI Koeppe, T Woolf. Biophys J 78:342A, 2000.

125. D Fattal, A Ben-Shaul. Biophys J 65:1795–1809, 1993.

126. P Kralchevsky, V Paunov, N Denkov, K Nagayama. J Chem Soc, Faraday Trans

5:3415–3432, 1995.

127. M Goulian, R Bruinsma, P Pincus. Europhys Lett 22:145–150, 1993.

128. J Kim, KS Neu, G Oster. Biophys J 75:2274–2291, 1998.

129. N Dan, P Pincus, S Safran. Langmuir 9:2768–2774, 1993.

130. K Morton, D Mayers. Numerical Solution of Partial Differential Equations. Cambridge, UK:

Cambridge University Press, 1994.

131. G Miloshevsky, M Partenskii, P Jordan. (2002), in preparation.

132. P Lague, M Zuckermann, B Roux. Biophys J 81:276–284, 2001.

133. N Dan, S Safran. Biophys J 75:1410–1414, 1998.


18Biocatalysis: Electrochemical Mechanismsof Respiration and Photosynthesis

ALEXANDER G. VOLKOV Oakwood College, Huntsville, Alabama, U.S.A.

I. INTRODUCTION: KINETIC ASPECTS OF SYNCHRONOUSMULTIELECTRON REACTIONS

Vectorial charge transfer and a molecular recognition at the interface between two dielec-tric media are important stages in many bioelectrochemical processes such as thosemediated by energy-transducing membranes [1–4]. Many biochemical redox reactionstake place at aqueous medium/membrane interfaces and some of them are multielectronprocesses. About 90% of the oxygen consumed on Earth is reduced in a four-electronreaction catalyzed by cytochrome c oxidase. Multielectron reactions take place in photo-synthesis, which is the most important process on earth [5–10]. Life on Earth began asphotosynthesis.

Synchronous multielectron reactions may proceed without formation of intermedi-ate radicals, which are highly reactive and can readily enter a side reaction of hydroxyla-tion and destruction of the catalytic complex. Since multielectron reactions do not poisonthe environment with toxic intermediates and they are ecologically safe, they are used byNature for biochemical energy conversion in respiration and photosynthesis [11,12]. In themultielectron reaction that takes place in a series of consecutive one-electron stages, theGibbs energy necessary per single electron transfer obviously cannot be completely uni-formly distributed over the stages [7]. The energy needs for various stages will be differentand the excess energy in the easier stages will be converted into heat. In a synchronousmultielectron reaction the energy will be used very economically [11].

An important parameter in the quantum theory of charge transfer in polar media isthe medium reorganization energy Es that determines activation energy [1,12–17]. Theenergy of medium reorganization in systems with complicated charge distribution wascalculated by Kharkats [18]. Reagents and products can be represented by a set of Nspherical centers arbitrarily distributed in a polar medium. The charges of each of thereaction centers in the initial and final state are zi and zf, respectively. Taking Rk torepresent co-ordinates of the centers and "i for dielectric constants of the reagents itfollows that


Es ¼ 0:81

"opt� 1

"st

� �

�XNp¼1

ð�zkÞ22apþXNk¼1k 6¼p

ð�zpÞð�zkÞ2Rpk

þXNk¼1k6¼p

XNl¼1l 6¼p

ð�zpÞð�zlÞa3pð ~RRpk~RRplÞ

R3pkR

3pl

3"2st

ð2"st þ "iÞ2� 1

2

!264

375

8><>:

9>=>;ð1Þ

where (�zkÞ ¼ zfk � zik, Rpk ¼ Rp � Rk, zfk, z

ik are charge numbers of particle k in the initial

and final states, respectively, ap is the radius of particle p, Rk is the co-ordinate of the k -particle center, and "i is the dielectric constant of the reactant. Reactions with synchronoustransfer of several charges present a particular case of Eq. (1).

It follows from Eq. (1) that Es is proportional to the square of the number ofcharges transferred. This factor makes multielectron processes impossible in themajority of homogeneous redox reactions due to the high activation energy resultingfrom a sharp rise in the energy of solvent reorganization. For multielectron reactionsthe exchange currents of n-electron processes are small compared to those of one-electron multiple step processes, which makes the stage-by-stage reaction mechanismmore advantageous. Therefore, multielectron processes can proceed only if the for-mation of an intermediate is energetically disadvantageous. However, conditions canbe chosen which reduce Es during transfer of several charges to the level of thereorganization energy of ordinary one-electron reactions. These conditions requiresystems with a low dielectric constant and large reagent radii. Furthermore, thesubstrate must be included in the co-ordination sphere of the charge acceptor, withseveral charge donors or acceptors bound into a multicenter complex. Recent papers[14–17] have presented theoretical studies on the kinetics of heterogeneous multielec-tron reactions at water/oil interfaces, which proved to be capable of catalyzing multi-electron reactions and sharply reducing the activation energy.

The most effective coupling of ATP synthesis to electron transport can be obtained ifthe activation energy of the coupled process is lower than that of the charge transfer in theelectron-transport chain. It is obvious from Eq. (1) that this requires a simultaneoustransfer of opposite charges, so that the second and the third terms of the equation arenegative. An optimal geometry between the centers of charges of donors and acceptorsmust also be chosen [19].

To illustrate this point, we can consider two instances of multielectron reactions:simultaneous transfer of n charges from one donor to an acceptor and simultaneoustransfer of several charges (one from each of the centers) to m acceptors (m � n). Inthe former case Es is proportional to n2, while in the latter it may be significantlylower (depending on the sign of the charge being transferred and the reciprocalpositions of reagents). A multicenter process with Es ffi n is also possible. The con-certed multicenter mechanism of multielectron reactions markedly reduces Es, andhence the activation energy, compared to a two-center multielectron process.Electrostatic interactions between reactants also reduces the activation energy of mul-tielectron processes at the interface. Such electrostatic energies in a heterogeneousprocess can never be equal to zero due to interactions with image charges.However, by appropriate arrangement of the reactants the electron-transfer activationenergy in a heterogeneous multielectron reaction can be much lower than the energyof medium reorganization.


II. CYTOCHROME OXIDASE: ELECTROCHEMISTRY OF RESPIRATION

The function of the enzymes of the mitochondrial respiratory chain is to transform theenergy of redox reactions into an electrochemical proton gradient across the hydrophobicbarrier of a coupling membrane. Isolated oligoenzyme complexes of the respiratory chainof mitochondria, cytochrome c oxidase, succinate–cytochrome c reductase, and NADH–CoQ reductase, are able to catalyze charge transfer in model systems, e.g., at a water/octaneinterface, which can be followed by a change in the interfacial potential at this interface [20–22]. A necessary condition for this measurement is the presence of the enzymes and sub-strates in the aqueous phase and a charge acceptor in the octane phase [22].

Cytochrome oxidase (EC 1.9.3.1) is the terminal electron acceptor of the mitochon-drial respiratory chain. Its main function is to catalyze the reaction of dioxygen reductionto water using electrons from ferrocytochrome c:

4Hþ þO2 þ 4e�respiration! 2H2O

photosynthesisð2Þ

Reaction (2) is exothermic, and the energy can be used to transport protons acrossthe mitochondrial membrane (Fig. 1). Mitochondrial cytochrome c oxidase is a dimer,each monomer being composed of 13 subunits. The enzyme contains cytochromes a anda3, one binuclear copper complex Cua, one mononuclear copper site Cub and one boundMg2+ per monomer [23]. It has a molecular weight of about 180,000 – 200,000 kDa for themost active form. Cytochrome oxidases can transport up to eight protons across themembrane per four electrons. Four of the protons bind to the reaction complex duringdioxygen reduction to water and up to four other protons are transported across themembrane. The resulting chemiosmotic proton gradient is used in ATP synthesis.

Respiration is the reduction of O2 to H2O during the oxidation of carbohydrate toCO2. There are two types of respiration in photosynthetic organisms: a dark respirationand a photorespiration [3]. Dark respiration includes O2 reduction and the oxidation ofNADH and FADH2 in mitochondrial membranes, glycolysis, the Krebs cycle, and theoxidative pentose phosphate pathway. Respiration is commonly subdivided into two func-tional components: growth respiration, supplying energy for new biomass production, and

FIG. 1 Scheme of the structural organization of cytochrome c and cytochrome c oxidase monomer

in the inner mitochondrial membrane.


maintenance respiration providing energy to maintain the integrity of already existingstructures and their turnover. In the present chapter we will consider the thermodynamicsof oxygen reduction in plant mitochondria.

Plant mitochondria have two membranes: a smooth outer membrane that surroundsa invaginated inner membrane. The respiratory chain of mitochondria is an integral partof the inner mitochondrial membrane. It is composed of four electron-transporting proteincomplexes (NADH dehydrogenase complex I, succinate dehydrogenase complex II, cyto-chrome reductase complex III, and cytochrome c oxidase complex IV), ATP synthase(complex V), and the mobile electron carriers ubiquinone and cytochrome c. Plant mito-chondria have additional enzymes not found in the mitochondria of animals: the cyanide-insensitive alternative oxidase, an internal rotenone-insensitive NADPH dehydrogenase,and an externally located NADPH dehydrogenase, which do not conserve energy. Thealternative oxidase catalyzes the oxidation of ubiquinol to ubiquinone and the reductionof oxygen to water and is inhibited by salicylhydroxamic acid. In some photosyntheticcells the carbohydrates formed during photosynthesis can serve as the Gibbs free energysource for respiration, which leads to ATP synthesis and water and CO2 production.Oxygen reduction, catalyzed by cytochrome c oxidase accounts for a significant portionof the water eliminated from the mitochondria.

Kharkats and Volkov first presented proofs that cytochrome c oxidase reducesmolecular oxygen by synchronous multielectron mechanism without O�2 intermediateformation [15,17–19,24]. Our calculations predicted that the first step in dioxygen reduc-tion by cytochrome c oxidase should be a concerted multielectron process although oxygenintermediates at room temperature were not detected before our estimations. As the fieldprogresses after these pioneering observations, it became clear that the first step of dioxy-gen reduction is a two-electron concerted process. In the present chapter a possible mole-cular concerted 2:1:1-electron and 2:2 proton pump mechanisms for cytochrome c oxidasefunctioning are discussed.

The 1:1:1:1-electron mechanisms of oxygen reduction by cytochrome oxidase are themost frequently discussed in biochemistry. The reaction implies that the Gibbs free energyof the first electron transfer from cytochrome oxidase to O2 is positive (Fig. 2). As a result,

FIG. 2 Energy diagram for possible routes of the reaction O2 + 4H+, 2H2O; Gm is the reaction

Gibbs free energy at pH 7.


this route should either not be followed or the reaction rate should be extremely low. Sincethe Gibbs free energy of O2 binding in the catalytic site of cytochrome oxidase is �21 kJmol�1 [25], the cytochrome c redox potential is 0.25 V, and the Gibbs free energy of thefirst electron donation to oxygen at pH 7 is +33 kJ mol�1, the Gibbs free energy of thereaction:

O2 þ e� ! O�2 ð3Þin a cytochrome oxidase catalytic site is equal to +78 kJ mol�1. The activation energy forO2 reduction by fully reduced cytochrome oxidase amounts to 16 kJ mol�1 [26]. Since theGibbs free energy of the endothermic reaction (3) is five times the measured activationenergy for O2 reduction by cytochrome oxidase, the 1-electron mechanisms 1:1:1:1, 1:2:1,1:1:2, and 1:3 at room temperature are unlikely. They could only be feasible if the energyof the 1-electron intermediate binding were negative, but greater than 52 kJ mol�1 inmagnitude. Such significant energy of covalent binding allows this intermediate to beexperimentally detected. However, it has not been detected thus far.

The fact that the first electron addition to O2 is endothermic accounts for the relativechemical inertness of oxygen in Nature and permits the existence of life on Earth.

The high-energy binding of oxy-intermediate provides a strong restriction on theenergetics of the next step of second electron transfer and formation of a peroxide inter-mediate. The probability of such a pathway is too small due to kinetic and thermodynamiclimits. Therefore, evolution could reserve either ecologically clean sequential 2:1:1mechanism, with intermediates tightly bound at the catalytic site. A possible mechanismof dioxygen reduction by cytochrome c oxidase is outlined in Fig. 3 and will be consideredin detail after discussing the thermodynamic and kinetic aspects of the problem.

FIG. 3 Scheme of 2:1:1 electron oxygen reduction mechanism at the cytochrome c oxidase active

site. (From Refs. 15,17–19, and 24.)


A. Reaction Center Architectonics

Equation (1) lays down conditions for the structure of cytochrome c oxidase catalyticsites necessary for dioxygen reduction to occur by the concerted n-electron mechanism.To lower the medium reorganization energy and thus activation energy, it is necessarythat:

. The dielectric constant of the medium where oxygen reduction takes placeshould be low, i.e. the catalytic site should be immersed in the hydrophobicphase of the membrane (protein).

. There should be n spatially separated electron donors. For the mechanism pro-posed, these are hemes and protein–copper complexes.

. It is desirable that electron transfer via cytochrome c oxidase be accompanied bythe transport of cations (e.g., protons). It follows from Eq. (1) that, whencharges of opposite signs are transferred simultaneously in close directions, themedium reorganization energy could be lowered due to the third and fourthterms in Eq. (1). Equation (1) implies that the coupling of the electron andproton pumps in cytochrome oxidase can be attained if medium reorganizationis neutralized by the simultaneous transfer of unlike charges in close directions.If electron transfer via cytochrome oxidase is coupled with proton transportacross the mitochondrial membrane, the energy liberated in reaction (2) willbe consumed for useful work instead of being converted into heat.

. The radii of electron donors should be sufficiently large. This is achieved not byusing free metal ions, but their organic complexes (hemes and cysteines) withsystems of conjugated bonds and ligands capable of participating in redoxreactions.

B. Bridge Electron-Transfer Mechanism

Electrons generated in the oligoenzyme complexes of the mitochondrial respiratory chainare transferred to the cytochrome c oxidase active site by the 1-electron bridge mechanism.The reduction of the oxygen molecule to water requires the stepwise transfer of fourelectrons from cytochrome c to cytochrome a and a3 as well as to two Cu-containingproteins, Cua and Cub. A quantum mechanical calculation has been made of the prob-ability of electrons transfer via an intermediate virtual state as a possible model of anelectron mechanism with an activated outer sphere and a bridge ion.

The stepwise transfer of electrons from cytochrome c to cytochrome a3 via cyto-chrome a is kinetically favorable due to a substantial decrease in the medium reorganiza-tion energy for direct electron transfer from cytochrome c to cytochrome a3. The redoxpotential of Fec may not be smaller, but even greater than the redox potential of Fea. It isessential that only the minimum of the intermediate term on the reaction energy diagrambe below the cross-point of the initial and final terms.

C. Activation Energy and Mechanism of Dioxygen Reduction

Studies on the temperatures dependence of the oxygen reduction rate have revealed thatcytochrome oxidase exists in two conformations—‘‘hot’’ (h) and ‘‘cold’’ (c). The respectiveactivation energies Eh

a and Eca are equal to 16 kJ mol�1 (at 23–358C) and 60 kJ/mol (below

208C) [27]. A phase transition accompanied by changes in conformation and the absorp-tion spectrum takes place between 18o and 238C. The temperature Tc depends on the


composition of the surrounding lipid. The low Eha value means that the 1-electron mechan-

isms 1:1:1:1, 1:2:1, 1:3, and 1:1:2 are unlikely at temperatures higher than Tc since theenthalpy for the transfer of the first electron from reduced cytochrome oxidase to dioxygenis five times the measured activation energy. If the energy of binding of one electronintermediate a3–O

�2 is less than �52 kJ mol� it gives thermodynamic and kinetic possibi-

lities for a multistep one-electron mechanism of oxygen reduction. However, the energy ofreorganization of the medium for one-electron transfer in cytochrome oxidase in ‘‘hot’’and ‘‘cold’’ conformations cannot be different in four times. From Eq. (1) it follows that ifwe change the distance h12 of electron transfer from h12 = a1 + a2 to h12 > a1 + a2 theenergy of reorganization changes by less then 50%. Ea can be equal to 16 and 60 kJ/molfor the one-electron mechanism of oxygen reduction according to Eq. (1) only if Es > 770kJ/mol when Ecomp < �52 kJ/mol�1. Such a high value of Es also shows that 1:1:1:1-electron mechanism of oxygen reduction in vivo is unlikely.

For the multielectron reaction 2:1:1, according to Eq. (1), Es for 2-electron reactionsbetween O2, a3, and Cub strongly depends on the geometry and distances in a catalytic site.Only 2:1:1 mechanism of oxygen reduction by cytochrome oxidase can be realized in vivoin both ‘‘hot’’ and ‘‘cold’’ conformations.

Consider the molecular mechanism of dioxygen reduction outlined in Fig. 3 in moredetail.

The oxidized catalytic site of cytochrome oxidase composed of cytochrome a3 andCub is reduced via the bridge mechanism by two electrons supplied from the electronreservoir of the respiratory chain to form a reduced complex, which then binds an oxygenmolecule. The reaction center is oxidized to the initial state in a 2-electron reaction withthe formation of a peroxide bridge between a3 and Cub. The partially reduced (to per-oxide) oxygen molecule must be bound in the reaction center since cytochrome oxidase isknown to reduce dioxygen to water without the release of any intermediates from themembrane. After that, the catalytic complex accepts two electrons in turn from the elec-tron reservoir Fe(c)!a3. At the next step, the peroxide bridge undergoes 1:1-electronreduction and protonation to water.

The concerted electron-transfer step (10�15 � 10�13 s) is followed by enzyme relaxa-tion with a complex set of characteristic times and constants. Rate constants for suchprocesses can range from very low (1 s�1) to very high (109 s�1). Studying the relaxation offully reduced cytochrome oxidase on its interaction with dioxygen allowed the followingcharacteristic constants to be resolved: 7� 107 M�1s�1, 6:8� 104 M�1 s�1, 1:7� 104 M�1

s�1, and 1:1� 103 M�1 s�1. Such a complex relaxation pattern has led some authors tosuggest that electron transfer to O2 is stepwise and proceeds via the 1:1:1:1 mechanism.However, since no intermediate that should be formed on the first electron donation to O2

has been detected in the native enzyme at a temperature higher than 218C, this spectrum ofcharacteristic times can be attributed to the relaxation of the metalloenzyme during 2:1:1-electron oxygen reduction to water. The redox potentials of cytochromes a and a3 as wellas of Cua and Cub are about the same, which means that the energy states of all the fourmetal centers in the reaction complex of the native enzyme are similar. This also favorsconcerted reactions.

Sucheta et al. [28] published the experimental proofs of the theoretical mechanism ofcytochrome c oxidase functioning proposed originally be Kharkats and Volkov [15,17–19,24]. Using time-resolved optical absorption difference spectra and singular valuedecomposition analysis, Sucheta et al. [28] found the presence of peroxy and ferryl inter-mediates at room temperature during reduction of oxygen by cytochrome c oxidase andmeasured the rate constants.


D. Proton Pump

Water molecules released in the course of oxygen reduction are removed from the hydro-phobic catalytic site to the aqueous phase. The product being continuously removed fromthe reaction center will shift the equilibrium of reaction (2) to the right. Energy liberated inthe exothermic reaction (2) is sufficient for transporting eight H+ ions across the mem-brane. Four of them are used to form with O2 two H2O molecules. The remaining H+ ionscan be transported across the hydrophobic zone of the membrane and used for ATPsynthesis via the ATP-synthetase complex, with the cytochrome oxidase H+ pump servingonly to transform the energy of ferrocytochrome c oxidation. Proton translocation can bedirect if the ligands to redox centers provide the protons, or indirect if the redox reactionscause conformational changes transmitted to proton-donating groups remote from theredox centers. As follows from thermodynamics (Fig. 2), energy for the H+-pump func-tioning is liberated only at the last steps of water formation on the addition of third andfourth electrons independently of the reaction route. The functioning of the protons’pump after formation of ferryl intermediate is possible only if the difference betweenthe Gibbs energy of ferryl and peroxy intermediates’ binding is more negative than �35kJ/mol�1.

The energy of binding of ferryl intermediate is negative and sufficiently high, whichgives a possibility to the functioning of a proton pump not only at last stage of addition offourth electrons, but also after formation of a 3-electron oxygen intermediate. The stoi-chiometry of proton pumping by cytochrome oxidase can be 0:2:2 if the ferryl intermediatehas �35 kJ more negative energy of binding than the peroxy intermediate.

As follows from Eq. (1), media reorganization energy corresponding to simultaneoustransfer of electrons and protons will be minimal in the case when the directions of theirtransfers are close. In the case of charge transfer in cytochrome oxidase the donor ofelectrons is situated on side C and the protons come from side M. In this case the minimalactivation energy will be achieved at the maximally possible given geometry of the systemsangle between the directions of electrons and proton transfers.

It is to be noted that cytochrome oxidase can reduce O2 without concomitant protontransfer. In such a case the enzyme would work as a machine converting the energy ofelectron transfer to heat. It may be that evolution has reserved only the e�, H+-form ofcytochrome oxidase with minimum energy dissipation.

III. PHOTOSYNTHETIC SYSTEMS

The annual consumption of energy by mankind is currently about 4� 1017 kJ, risingrapidly and doubling every 20 years. The known reserves of fossil fuels are limited toan estimated energy equivalent of 5� 1019 kJ, so new sources of energy are of fundamentalimportance. One obvious possibility is solar energy. The amount of solar energy incidenton the Earth is about 5� 1021 kJ per year, of which 3� 1018 kJ is converted into chemicalenergy by photosynthesis in plants and micro-organisms. In water-oxidizing photosynth-esis two large membrane-integrated protein complexes photosystem II (PS II) and photo-system I (PS I) are operating in series [3]. The electron transfer starts in both photosystemsvectorially across the membrane. Light energy is harvested by photosynthetic pigmentsystems in which the electronic structure of excited-state chlorophyll donates an electronto a primary acceptor pheophytin, the first component of an electron-transport chain. Theelectron carries with it the energy of the original photon of light that was absorbed, and in


the process of electron transport the energy is captured in two ways. The first involvescoupling a proton pump mechanism with the sequential redox reactions in one part of theelectron-transport chain, so that a proton gradient is established across the thylakoidmembrane. The electrochemical energy of the proton gradient is then used to driveATP synthesis by the ATP synthase enzymes embedded in the membrane [1]. The secondenergy capture occurs when an acceptor molecule such as NADP is reduced to NADPH,which in turn is used to reduce carbon dioxide in the Calvin cycle. Systems modelingphotosynthesis should have the capability of carrying out relatively simple versions ofthese fundamental reactions.

The last part of this chapter focuses on electrochemical mechanisms of water oxida-tion in the PS II of green plants.

IV. STRUCTURE AND COMPOSITION OF THE OXYGEN-EVOLVINGCOMPLEX IN VIVO

The redox map of photosynthesis in green plants can be described in terms of the well-known Z-scheme proposed by Hill and Bendal [29]. The main advantage of the currentlyaccepted Z-scheme depicted in Fig. 4 lies in the specific mechanism of charge transfer atthe stage of water oxidation, which is a multielectron reaction mechanism involving nounknown intermediates [6,12].

The molecular organization of a thylakoid membrane is shown in Fig. 5. The spec-tral characteristics of PS II indicate that the primary electron donor is the dimer ofchlorophyll P680 with absorption maxima near 680 and 430 nm. Water can be oxidizedby an oxygen-evolving complex (OEC) composed of several chlorophyll molecules, twomolecules of pheophytin, plastoquinol, several plastoquinone molecules, and a manga-nese–protein complex containing four manganese ions. The OEC is a highly orderedstructure in which a number of polypeptides interact to provide the appropriate environ-ment for cofactors such as manganese, chloride, and calcium, as well as for electrontransfer within the complex. Figure 6 shows the electronic equivalent circuit of PS I andPS II.

Manganese-binding centers were first revealed in thylakoid membranes by EPRmethods, and it is now understood that four manganese ions are necessary for oxygenevolution during water photo-oxidation. Plastoquinone (PQ) acts as a transmembranecarrier of electrons and protons between reaction centers of two photosystems in thecase of noncyclic electron transfer and may also serve as a molecular ‘‘tumbler’’ thatswitches between one-electron reactions and two-electron reactions. Pheophytin is anintermediate acceptor in PS II. Direct formation of P680 pheophytin ion radical pairswas revealed by experiments on magnetic interactions between pheophytin and PQreflected in the EPR spectra.

V. THERMODYNAMICS OF WATER OXIDATION

The photocatalytic oxidation of two molecules of water to dioxygen cannot be a single-quantum process since the total energy expenditure of a catalytic cycle cannot be less than476 kJ mol�1. However, there is no fundamental reason why one quantum should notinduce the transfer of several electrons. For instance, a two-quantum process wouldrequire light with a wavelength of less then 504 nm while a four-quantum process


would involve a sequential mechanism in which each light quantum is used to transfer oneelectron from the photocatalyst to an electron acceptor. The threshold wavelength for theoxidation of water in this case is 1008 nm. The eight-quantum scheme which is actuallyused in photosynthesis can be explained by the need to compensate for energy losses in along electron-transfer chain of redox reactions.

Water oxidation to molecular oxygen is a multielectron process that proceeds with asurprisingly high quantum efficiency. The water oxidation reaction can proceed on illu-mination at 680 nm, a wavelength of light that excludes one-electron mechanisms usinghydroxyl and oxygen radicals (Fig. 2). For a three-electron reaction a stronger oxidantthan the cation radical P680+ is needed. A synchronous two-by-two 2:2-electron pathwayof the reaction is thermodynamically possible if the standard free energy of binding of thetwo-electron intermediate is about �40 kJ mol�1. This value corresponds to the energy oftwo hydrogen bonds forming between H2O2 and the catalytic center. For this case amolecular mechanism can be proposed (Fig. 7) and will be discussed below.Synchronous four-electron oxidation of water to molecular oxygen (Fig. 8) is also thermo-dynamically possible.

FIG. 4 Electron transfer during photosynthesis in higher plants. The abscissa shows the midpoint

redox potential at pH 7.0. Light quanta (h�) are absorbed in two sets of antenna chlorophyll

molecules, and the excitation energy is transferred to the reaction center chlorophyll a molecules

of PS II (P680) and PS I (P700) forming (P680)* and (P700)*. The latter two initiate electron

transport. Abbreviations: Z and D are tyrosine residues; Cyt b559 is cytochrome b559 of unknown

function; Pheo is pheophytin; QA, QB, and PQ are plastoquinone molecules; Fe2S2 represents the

Rieske iron–sulfur center, Cyt f stands for cytochrome f , PC is plastocyanin; Ao is suggested to be a

chlorophyll molecule and A1 is possibly vitamin K; FNR is ferredoxin NADP oxidoreductase.


One-electron mechanisms of water oxidation are likely to be operative in some modelsystems with a low quantum efficiency, but two- or four-electron reactions cannot occurdue to kinetic limitations. The intermediates formed in these systems would be highlyreactive and could enter into side reactions of hydroxylation, oxidation, and destructionof chlorophyll and other components of the reaction center.

FIG. 5 Schematic model of the electron-transport chain with most of the light-harvesting pigment–

protein complexes omitted.

FIG. 6 Electronic equivalent circuit of photosystems I and II.


VI. MOLECULAR MECHANISM OF OXYGEN EVOLUTION IN VIVO [6–12]

Membrane-bound P680 enters an excited state on illumination. In dimers and otheraggregated forms of chlorophyll the quantum efficiency of triplet states is low, and it isthe singlet excited states that undergo photochemical transformations. In several pico-seconds, an electron is first transferred to pheophytin, then to plastoquinone QA, and fromplastoquinone QA to another polypeptide-bound plastoquinone QB in the thylakoid mem-brane (Fig. 4), resulting in an oxidized pigment and a reduced acceptor. The cation radicalP680+ successively oxidizes four manganese ions, which in turn drives the production ofmolecular oxygen. Formation of a cation radical of chlorophyll or oxidation of manganeseions is accompanied by dissociation of water bound to the reaction center and ejection ofprotons. A synchronous multielectron process that describes all four oxidizing states of theOEC was proposed earlier. The transfer of electrons in a 1:1:1:1 series from a manganesecluster to the electron-transport chain is accompanied by the ejection of 1:0:1:2 protonsand the evolution of molecular oxygen.

Protons are released from reaction centers either by regulators of proton distributionor by hydrogen-bond transfer (analogous to a Grotthus mechanism) through the hydra-tion shell of manganese ions. The hydration sphere of manganese is known to containwater molecules that rapidly exchange protons with bulk water. The presence of divalentcations at the interface between two immiscible electrolyte solutions facilitates strong

FIG. 7 Proposed 2:2-electron mechanism of water photo-oxidation.

FIG. 8 Proposed four-electron mechanism of water photo-oxidation.


adsorption of water molecules belonging to the second hydration shell of ions. Thus, aportion of co-ordinatively bound water enters the compact part of the electrical doublelayer, which changes its differential capacity at the interface. In the case of multivalentions with small radii, the electric field near a cation is large. This can disturb the micro-structure of the adjacent intrathylakoid space and bring about dielectric saturation effects.

Manganese ions play a particularly important role in the evolution of dioxygenduring photosynthesis. Although there are several manganese pools in chloroplasts,only one is involved in water oxidation. The manganese ions associated with chloroplastOEC can perform a number of functions:

. The Mn–polypeptide complex is a redox intermediate that protects the reactioncenter from redox and radical destruction.

. Mn clusters are redox buffers facilitating accumulation of four holes in thereaction center of PS II, which are needed to ensure water photo-oxidation.

. Hydrated multivalent Mn cations bring water to the reaction center so that rapidproton exchange and transport through the hydration shell of Mn ions in thezone of water oxidation are affected.

. Multivalent Mn ions induce dielectric saturation effects in the polar region of thereaction center of PS II, which reduces the reorganization energy of the mediumduring charge transfer.

REFERENCES

1. AG Volkov, D Deamer, D Tanelian, VS Markin. Liquid Interfaces in Chemistry and Biology.

New York: John Wiley, 1998.

2. AG Volkov, DW Deamer, eds. Liquid–Liquid Interfaces: Theory and Methods. Boca Raton,

FL: CRC Press, 1996.

3. OS Ksenzhek, AG Volkov. Plant Energetics. New York: Academic Press, 1998.

4. AG Volkov, ed. Liquid Interfaces in Chemical, Biological, and Pharmaceutical Applications.

New York: Marcel Dekker, 2001.

5. AG Volkov. Biophysics 30:491–491, 1985.

6. AG Volkov. Mol Biol 20:728–736, 1986.

7. AG Volkov. J Electroanal Chem 205:245–257, 1986.

8. AG Volkov. Photobiochem Photobiophys 11:1–7, 1986.

9. AG Volkov. Sov Electrochem 21:91–98, 1985

10. AG Volkov. Biol Membr 4:984–993, 1987.

11. AG Volkov. Uspekhi Sovr Biol (Progr Modern Biol) 105:467–487, 1988.

12. AG Volkov. Bioelectrochem Bioenerg 21:3–24, 1989.

13. YuI Kharkats, AM Kuznetsov. In: AG Volkov, DW Deamer, eds. Liquid–Liquid Interfaces.

Theory and Methods. Boca Raton, FL: CRC Press, 1996, pp 139–154.

14. YuI Kharkats, AG Volkov. J Electroanal Chem 184:435–439, 1985.

15. YuI Kharkats, AG Volkov. Biochim Biophys Acta 891:56–67, 1987.

16. YuI Kharkats, AG Volkov. Bioelectrochem Bioenerg 22:91–103, 1989.

17. AG Volkov, YuI Kharkats. Biol Membr 5:920–931, 1988.

18. YuI Kharkats. Sov Electrochem 14:1721–1724, 1978.

19. YuI Kharkats, AG Volkov. In: MJ Allen, SF Cleary, AE Sowers, eds. Charge and Field

Effects in Biosystems—4. Singapore: World Scientific, 1994, pp 70–77.

20. YuI Kharkats, AG Volkov, LI Boguslavsky. Dokl Akad Nauk SSSR 220:1441–1444, 1975.

21. YuI Kharkats, AG Volkov, LI Boguslavsky. Biophysics 21:634–638, 1976.

22. YuI Kharkats, AG Volkov, LI Boguslavsky. J Theor Biol 65:379–391, 1977.


23. O Einarsdottir. Biochim Biophys Acta 1229:129–147, 1995.

24. YuI Kharkats, AG Volkov. Anal Sci 14:27–30, 1998.

25. B Chance, C Saronio, IS Leigh. J Biol Chem 250: 9226–9237, 1975.

26. M Erecinska, B Chance. Arch Biochem Biophys 151:304–315, 1972.

27. YM Orii, M. Manabe, M Yoneda. J Biochem 81:505–517, 1977.

28. A Sucheta, KE Georgiadis, O Einarsdottir. Biochemistry 36:554–565, 1997.

29. R Hill, P Bendal. Nature 186:136–137, 1960.


19New Types of Membrane ReactionsMimicking Biological Processes

SORIN KIHARA Kyoto Institute of Technology, Kyoto, Japan

I. INTRODUCTION

Generally speaking, the membrane transport of a charge (ion or electron) means thetransfer of a charge from one aqueous solution (W1) to another (W2) across a membrane(M) as shown in Fig. 1(a). This membrane transport is realized in the presence of apotential gradient between W1 and W2 perpendicular to the W1|M or W2|M interface,and hence will be called ‘‘perpendicular transport’’ hereafter. However, the reaction whena charge is incorporated from W1 into M at one site (site A) of the W1|M interface andreleased from M to W1 at another site (site B) of the same interface after transfer in M canalso be regarded as a membrane transport, Fig. 1(b). This transport is realized under apotential gradient between sites A and B in W1 parallel to the W1|M interface [1], andhence will be called ‘‘parallel transport of type I’’ hereafter. Another variety of ‘‘paralleltransport’’ is the reaction when a charge is released from M to W1 or W2 at one site (siteA) of the W1|M or W2|M interface and incorporated from W1 or W2 to M at another site(site B) of the W1|M or W2|M interface after transfer in W1 or W2, Fig. 1(c). This varietywill be called ‘‘parallel transport of type II’’ hereafter [2].

The ‘‘parallel transports’’ are considered to often occur at the interface between anaqueous solution and a heterogeneous biomembrane with various domains [3–5].Therefore, the electrochemical elucidation of the mechanisms of ‘‘parallel transports’’ isexpected to be very important for a better understanding of biomembrane phenomena aswell as for the design of novel analytical methods or other chemical techniques mimickingthe phenomena, though such investigations have been very few so far.

In this chapter, the fundamental feature of ‘‘parallel transports of types I and II’’elucidated with the aid of voltammetry for ion transfer at the interface of two immiscibleelectrolyte solutions is introduced, and compared with that of ‘‘perpendicular transport’’[6–9].

II. MEMBRANE SYSTEMS

A. Cell Used to Observe ‘‘Perpendicular Transport’’

The electrolytic cell with an M used for the observation of ‘‘perpendicular transport (W1 –M–W2 transport)’’ is shown in Fig. 2(a) [1,2,6,7]. An example of the cell configuration isas follows:


1 M MgSO4

½W1�

0:1 M TPenAþTPhB�

½M�

��

2 M MgSO4

½W2�ð1Þ

CEW1 REW1 � EW1jM �REM;1 REM;2 � EMjW2 �REW2 CEW2| | | || | | |___________________

EW1�W2______________

| || || |__________________________

IW1�W2_____________________

In the cell, W1 and W2 (5 mL each) containing MgSO4 as a supporting electrolytewere separated by a 1,2-dichloroethane solution (DCE), containing tetrapentylammoniumtetraphenylborate (TPenA+TPhB�) as a supporting electrolyte. The DCE worked as theliquid membrane (M) of thickness 1 cm. The concentrations of MgSO4 in W1 and W2were made to be 1 and 2 M (= mol L�1), respectively, in order to stabilize M between W1and W2 by means of the difference in specific gravities. Silver–silver chloride electrodeswere used as reference electrodes (REW1 and REW2) in W1 and W2, and platinum wireswere used as counter electrodes (CEW1 and CEW2) in W1 and W2. Two tetraphenylborateion-selective electrodes (TPhBE) were set in M near to the W1|M and W2|M interfaces asreference electrodes (REM,1 and REM,2).

B. Cell Used to Observe ‘‘Parallel Transport of Type I’’

The cell used for the investigation of ‘‘parallel transport of type I’’ is illustrated in Fig.2(b). The M was composed of DCE (10 mL, 0.7 cm thick) containing 0.05 MTPenA+TPhB� as a supporting electrolyte; W1 (10 mL) was distilled water withoutany electrolytes and W2 (10 mL) contained 2 M MgSO4. In connection with the cell,W2 was used for the investigation described in Section VI.B, though it was not necessaryfor those in Sections VI.A, VI.C, or VI.D. Two silver|silver chloride reference electrodes(REA,W1 and REB,W1), were set at sites A and B in W1 in the vicinity of the W1|M

FIG. 1 Three types of membrane transports: (a) ‘‘perpendicular transport’’ (W1 – M – W2

transport); (b) ‘‘parallel transport of type I’’ (W1 – M – W2 transport); (c) ‘‘parallel transport of

type II’’ (M – W1 – M or M – W2 – M transport).


FIG. 2 Membrane systems: (a) electrolytic cell for ‘‘perpendicular transport’’; (b) electrolytic cell

for ‘‘parallel transport of type I’’; (c) electrolytic cell for ‘‘parallel transport of type II.’’ W1 and W2:

aqueous solutions; M: membrane of organic solution; REW1, REW2, REA,W1, REB,W1, REA,W2, and

REB,W2: silver|silver chloride reference electrodes. REM,1, REM,2, REA,M, and REB,M: TPhB� ion-

selective reference electrodes; CEW1, CEW2, CEA,M, and CEB,M: platinum-wire counter electrodes;

(A) and (B): sites A and B.


interface. The distance between the two sites was 12 cm. Two platinum wires were set atsites A and B in W1 as counter electrodes (CEA,W1 and CEB,W1). In order to monitorpotential differences at the W1|M interfaces of sites A and B, two TPhBE (REA,M andREB,M) were set in M near to the interfaces.

C. Cell Used to Observe ‘‘Parallel Transport of Type II’’

The cell used for the investigation of ‘‘parallel transport of type II’’ is illustrated inFig. 2(c). The M was composed of DCE (10 mL, 0.7 cm thick) containing diluteTPenA+TPhB� (e.g., 10-4 M), and W1 (10 mL) and W2 (10 mL) contained 1 and 2 MMgSO4, respectively. Two TPhBE (REA,M and REB,M) were set at sites A and B in M. Thedistance between the two sites was 12 cm. Two platinum wires were set at sites A and B inM as counter electrodes (CEA,M and CEB,M). In order to monitor potential differences atthe W1|M or W2|M interfaces of sites A and B, four silver|silver chloride referenceelectrodes (REA,W1 and REB,W1 or REA,W2 and REB,W2) were set at sites A and B inW1 or W2 in the vicinity of the W1|M or W2|M interfaces.

III. CHARACTERISTICS OF VOLTAMMOGRAMS FOR ‘‘PERPENDICULARTRANSPORT’’ AND ‘‘PARALLEL TRANSPORTS OF TYPES I AND II’’

In order to elucidate the features of ‘‘parallel transports of types I and II,’’ based on thevoltammetric method and concept, voltammograms for ‘‘parallel transports of types I andII’’ were measured and compared with that for ‘‘Perpendicular transport.’’

A. Fundamental Feature of Voltammogram for ‘‘Perpendicular Transport’’

The voltammogram for ‘‘perpendicular transport’’ was recorded by scanning the potentialdifference, EW1�W2, between W1 and W2 and measuring the current, IW1�W2, between W1and W2, Fig. 2(a). During the recording of the voltammogram, variations in potentialdifferences, EW1|M and EM|W2, at the W1|M and W2|M interfaces were monitored as afunction of IW1�W2, and voltammograms for ion transfers at the W1|M and W2|M inter-faces were obtained.

Curve 1 in Fig. 3 realizes an example of the voltammogram for the ‘‘perpendiculartransport (W1�M�W2 transport),’’ which was recorded with the cell as Eq. (1). Curves2 and 3 are voltammograms for the ion transfer at the W1|M and M|W2 interfacesobserved during the recording of curve 1. The final rise and the final descent in curve 2are attributable to the transfer of TPhB� from M to W1 and that of TPenA+ from M toW1, respectively. Here, the final rise and the final descent mean the large positive andnegative currents, respectively, limiting the potential window. The final rise and the finaldescent in curve 3 correspond to the transfer of TPenA+ from M to W2 and that ofTPhB� from M to W2, respectively. Comparing curve 1 with curves 2 and 3, it is obviousthat (1) the potential window in curve 1 is about twice that in curve 2 or 3, and (2)the slopes of the final rise and final descent in curve 1 are much less than those in curves2 and 3.

The above results lead to a conclusion that the relationship shown in Eq. (2) is heldamong EW1�W2, EW1jM, and EMjW2 at any IW1�W2 [6,7], and the relationship can beapproximated by Eq. (3) when W1, W2, and M contain sufficient concentrations ofelectrolytes as in the present case.


EW1�W2 ¼ EW1jM þ EM jW2 þ IW1�W2 RW1�W2 ð2Þ

where RW1�W2 denotes the resistance between W1 and W2.

EW1�W2 ¼ EW1jM þ EMjW2 ð3Þ

Equation (3) indicates that the membrane potential in the presence of sufficientelectrolytes in W1, W2, and M is primarily determined by the potential differences attwo interfaces in the membrane system that depend on ion-transfer reactions at the inter-faces. Taking into account the relationship in Eq. (3) and the electroneutrality in themembrane phase, the final rise in curve 1 (Fig. 3) is attributable to the simultaneoustransfers of the same amounts of TPhB� from M to W1 and of TPenA+ from M toW2 [6,7]. Similarly, the final descent in curve 1 is attributable to the simultaneous transfersof the same amounts of TPenA+ from M to W1 and of TPhB� from M to W2.

B. Comparison of Voltammogram for ‘‘Parallel Transport of Type I’’ withThat for ‘‘Perpendicular Transport’’

The voltammogram for ‘‘parallel transport of type I’’ was recorded by scanning thepotential difference EW1ðA�BÞ between REA,W1 and REB,W1 and measuring the currentIW1ðA�BÞ between sites A and B in W1 and W2, Fig. 2(b). Here, EW1ðA�BÞ was applied asthe potential of REA,W1 versus REB,W1. During the recording of the voltammogram,variations in potential differences EW1|M,A and EM|W1,B at the W1|M interface of sitesA and B were monitored as a function of IW1ðA�BÞ, and voltammograms for the chargetransfer at the W1|M interface of sites A and B were obtained. Here, EW1|M,A and EM|W1,B

FIG. 3 Voltammograms for ‘‘perpendicular transport’’: voltammograms for ion transfer through

M (W1 – M – W2 transport, curve 1), at the W1|M interface (curve 2) and at the M|W2 interface

(curve 3). Compositions of W1, M, and W2: 1 M MgSO4 in W1, 0.1 M TPenA+TPhB� in M, and 2

M MgSO4 in W2.


were measured as the potential difference of REA,W1 versus REA,M and that of REB,M

versus REB,W1, respectively.The voltammogram shown as curve 1 in Fig. 4 was recorded with the cell of Fig.

2(b). The voltammogram was very similar to that for ‘‘perpendicular transport’’ (curve 1in Fig. 3), indicating that the voltammogram was realized mainly by the composite of twointerfacial ion-transfer reactions, i.e., reactions at the W1|M interface of sites A and B. Inother words, this result demonstrated that ‘‘parallel transport of type I (W1–M–W1transport)’’ could be realized when a potential difference was applied between two sitesin one aqueous phase of a membrane system.

It was confirmed by comparing the voltammogram of curve 1 in Fig. 4 with voltam-mograms for ion transfers at the W1|M interface of sites A and B (curves 2 and 3 in Fig. 4)that transfers of TPhB� from M to W1 at site A and of TPenA+ from M to W1 at site Bcaused the current at the potential of the final rise in the voltammogram of curve 1, andtransfers of TPenA+ from M to W1 at site A and of TPhB� from M to W1 at site Bcaused the current at the potential of the final descent.

C. Comparison of Voltammogram for ‘‘Parallel Transport of Type II’’ withThat for ‘‘Perpendicular Transport’’

The voltammogram for ‘‘parallel transport of type II’’ was recorded by scanning thepotential difference EMðA�BÞ, between REA,M and REB,M, applied as the potential ofREA,M versus REB,M, and measuring the current (IMðA�BÞ between sites A and B in M,Fig. 2(c). During the recording of the voltammogram, variations in potential differencesEMjW1;A and EW1|M,B at the W1|M interface of sites A and B or EM|W2,A and EW2|M,B atthe W2|M interface of sites A and B were monitored as a function of IMðA�BÞ, and

FIG. 4 Voltammograms for ‘‘parallel transport of type I’’: voltammograms for ion transfer

through M (W1 – M – W1 transport, curve 1), at the W1|M interface of site A (curve 2) and at

the W1|M interface of site B (curve 3). Compositions of W1, M, and W2: without electrolyte in W1,

0.05 M TPenA+TPhB� in M, and 2 M MgSO4 in W2.


voltammograms for the charge transfer at the W1|M or W2|M interface of sites A and Bwere obtained. Here, EM|W1,A and EW1|M,B or EM|W2,A and EW2|M,B were measured aspotential differences of REA,M versus REA,W1 and REB,W1 versus REB,M, or REA,M versusREA,W2 and REB,W2 versus REB,M, respectively.

The voltammogram shown as curve 1 in Fig. 5 was recorded with the cell of Fig. 2(c).Curves 2 and 3 are voltammograms for ion transfers at the W1|M interface of sites A andB, respectively, observed during the recording of curve 1. The final rise and the finaldescent in curve 2 are attributable to the transfer of TPenA+ from M to W1 and thatof TPhB� from M to W1, respectively. The final rise and the final descent in curve 3 areattributable to the transfer of TPhB� from M to W1 and that of TPenA+ from M to W1,respectively.

FIG. 5 Voltammograms for ‘‘parallel transport of type II’’: voltammograms for ion transfers

through W1 and W2 (M – W1 – M and M – W2 – M transports, curve 1), at the W1|M interface

of site A (curve 2), at the W1|M interface of site B (curve 3), and those at the W2|M interface of site

A (curve 20) and at the W2|M interface of site B (curve 30). Compositions of W1, M, and W2: 1 M

MgSO4 in W1, 10�4 M TPenA+TPhB� in M and 2 M MgSO4 in W2.


The voltammogram of curve 1 was very similar to that for ‘‘perpendicular transport’’(curve 1 in Fig. 3) or ‘‘parallel transport of type I’’ (curve 1 in Fig. 4), indicating that thevoltammogram was realized mainly by the composite of two interfacial ion-transfer reac-tions, i.e., reactions at W1|M and W2|M interfaces of sites A and B. In other words, thisresult demonstrated that M–W1–M transport (‘‘parallel transport of type II’’) could berealized when a potential difference was applied between two sites in the membrane of amembrane system.

Curves 2 0 and 3 0 in Fig. 5 are voltammograms recorded by the same procedure asthat for curves 2 and 3, but at the W2|M interface instead of the W1|M interface. Curves 20

and 3 0 are almost identical with curves 2 and 3, respectively, indicating that reactionsidentical with those at the W1|M interface, i.e., M – W2 – M transport, proceeded at theW2|M interface. Here, the current IM(A-B), between sites A and B in M, was found to bethe sum of currents flowing through W1 and W2.

Taking into account the result mentioned above together with the discussion inpapers reported by the group of the present author [1,6,7], the current at the poten-tial of the final rise or that of the final descent in the voltammogram of curve 1 wereattributed to transfers of TPenAþ from M to both W1 and W2 at site A and thoseof TPhB� from M to both W1 and W2 at site B or transfers of TPhB� from M toboth W1 and W2 at site A and those of TPenAþ from M to both W1 and W2 atsite B, respectively.

IV. EFFECT OF CONCENTRATIONS OF ELECTROLYTES IN AQUEOUSAND MEMBRANE PHASES ON ‘‘PARALLEL TRANSPORT OF TYPE I’’OR ‘‘PARALLEL TRANSPORT OF TYPE II’’

Curve 2 in Fig. 6 is the voltammogram for ‘‘parallel transport of type I’’ recorded by thesame procedure as that for curve 1 in Fig. 4 (reproduced as curve 1 in Fig. 6), but afteraddition of an electrolyte (0.01 M MgSO4) to W1. When the concentration of MgSO4 inW1 was increased from 0 to 0.01 M, the voltammograms became more distorted withincreasing concentration (cf. curves 1 and 2).

In order to interpret the distortion, the current – potential relation was observedby using only W1 containing MgSO4 (without M and W2), and by scanning EW1ðA�BÞand measuring IW1ðA�BÞ. Curve 3 in Fig. 6 is an example obtained with 0.01 MMgSO4. As a matter of course, a straight line which could be expressed by Ohm’sLaw [IW1ðA�BÞ ¼ EW1ðA�BÞ=RW1ðA�BÞ, where RW1ðA�BÞ denotes the resistance betweensites A and B in W1] was observed. When the relation among currents in curves1, 2, and 3 was investigated at a definite EW1ðA�BÞ, it was found that IW1ðA�BÞ in curve2 was the sum of IW1ðA�BÞ in curves 1 and 3, i.e., the current between sites A and Bin the membrane system of Fig. 2(b) was composed of the current due to ‘‘paralleltransport of type I W1–M–W1 transport)’’ and that flowing in W1. Therefore, thecurrent efficiency for ‘‘parallel transport of type I’’ is higher when the resistance ofW1 is higher.

The effect of concentration of the electrolyte (TPenA+TPhB�) in M on voltammo-grams for ‘‘parallel transport of type II’’ was also investigated. When the concentration ofTPenA+TPhB� was increased from 10�4 to 10�3 M, the voltammograms became moredistorted. Based on the consideration similar to that in the case of ‘‘parallel transport oftype I’’, it was found that the current between sites A and B in the membrane system of


Fig. 2(c) was composed of currents due to ‘‘parallel transports of type II (M – W1 – M andM – W2 – M transports)’’ and the current flowing in M. Therefore, the current efficiencyfor ‘‘parallel transport of type II’’ is higher when the resistance of M is larger.

V. DISTRIBUTION OF POTENTIAL DIFFERENCE AT THE W|MINTERFACE BETWEEN SITES A AND B

A definite potential difference, EW1ðA�BÞ or EMðA�BÞ, was applied between sites A and B inW1 of the cell of Fig. 2(b) for ‘‘parallel transport of type I’’ or in M of the cell of Fig. 2(c)for ‘‘parallel transport of type II’’, respectively, and the potential difference ðEW1jM;CÞ atthe W1|M interface at a site (site C) between sites A and B was measured by using tworeference electrodes (REC,W1 and REC,M), set at site C as the potential of REC,W1 versusREC,M. Where REC,W1 and REC,M were silver|silver chloride reference electrodes placed inW1 and TPhBE in M near to the interface, respectively.

The plot in Fig. 7 and plot 1 in Fig. 8 show relationships between EW1|M,C and thedistance of site C from site A observed when EW1ðA�BÞ ¼ 1 V and EMðA�BÞ ¼ 1 V, respec-tively. The EW1|M,C in both cases changed almost linearly with the distance from site Aexcept for parts in the vicinities of sites A and B where resistances in W1 or M were smallbecause of transfers of ions between W1 and M.

Plot 2 in Fig. 8 shows the potential difference (the potential of REC,W2 versus that ofREC,M) at the W2|M interface of a site (site C) between sites A and B, EW2|M,C, observedwith EMðA�BÞ ¼ 1 V. This plot is almost identical with plot 1 in Fig. 8.

FIG. 6 Effect of concentration of MgSO4 in W1 on voltammograms for ‘‘parallel transport of type

I.’’ Concentration of MgSO4 in W1 (M): 1; 0, 2, and 3; 10�2. Compositions of M and W2: 0.05 M

TPenA+TPhB� in M and 2 MMgSO4 in W2. Curve 3: current–potential relationship observed using

only W1.


VI. UNIQUE REACTIONS REALIZED UNDER THE CONDITION FOR‘‘PARALLEL TRANSPORT OF TYPE I’’ OR ‘‘PARALLEL TRANSPORTOF TYPE II’’

A. Ion Transfer at the W1|M Interface in a Region Between Sites A and BUnder Potential Difference Applied Between Sites A and B in W1

When EW1ðA�BÞ was applied between sites A and B in W of the cell of Fig. 2(b), not onlythe ion-transfer reactions in the vicinities of sites A and B, but also reactions in a regionbetween sites A and B, were found to occur.

An example of the ion transfer was investigated with a membrane system in which Mcontained 2� 10�4 M picrate (Pic�) in addition to the supporting electrolyte, and W1 did

FIG. 7 Relationship between the potential difference (EW1|M,C) at the W1|M interface of site C and

the distance of site C from site A observed when 1 V was applied between sites A and B in W1.

Compositions of W1 and M: without electrolyte in W1 and 0.05 M TPenA+TPhB� in M.

FIG. 8 Relationship between the potential difference EW1|M,C or EW2|M,C at the W1|M interface

(curve 1) or at the W2|M interface (curve 2) of site C and the distance of site C from site A observed

when 1 V was applied between sites A and B in M. Compositions of W1, M, and W2: 1 MMgSO4 in

W1, 10�4 M TPenA+TPhB� in M and 2 M MgSO4 in W2.


not contain any electrolytes. Plots 1 and 2 in Fig. 9 show the concentration profile of Pic�

found in W1 and M, respectively, after electrolysis for 3 h by applying 1 V of EW1ðA�BÞ.From this result, it is clear that the transfer of Pic� from M to W1 occurred in a regionbetween site A and about 6 cm from site A. Here, the standard potential for the transfer ofPic� at the WjDCE interface was reported to be �0:36 V versus TPhBE [10], whichcoincides with EW1|M,C at site C of 5.7 cm from site A (cf. Fig. 7).

B. Ion Transport from W2 to W1 Through a Domain Under PotentialDifference Applied Between Sites A and B in W1

Figure 10(a) shows the membrane system constructed with a view to simulating the iontransport through a membrane in the presence of a domain. In the membrane system, W1did not contain any electrolytes, the membrane contained supporting electrolyte and W2contained 10�3 M Mg(TPhB)2 and 2 M MgSO4. Here, the domain was formed in a region1 cm wide in M (called ‘‘domain region’’ hereafter) separated from the other part of M byusing two porous tetrafluoroethylene resin films (48 �m thick and 0.45 �m pore size) asphysical boundaries. When an EW1ðA�BÞ of 1 V was applied between sites A and B in W1,and 500 �L of 10�2 M bis(diphenylphosphinyl)methane (BDPPM), which complexes withMg2+, was added to the domain region in M and then allowed to stand for 3 h, Mg2+ wastransferred from W2 to W1 through the domain. Figure 10(b) shows an example of theresult obtained with a domain in M formed 9 cm from site A. The Mg2+ transport wasrealized only when the domain was formed in a region between 7 cm from site A and site Bwhere EW1|M,C was available for the transfer of Mg2+ from M to W1, as described later.

The Mg2+ transport can be explained as follows: Mg2+ in W2 is extracted sponta-neously with TPhB� into the domain, since BDPPM facilitates Mg2+ transfer owing to thestrong complex formation, and the counterion, TPhB�, is highly hydrophobic [10,11]. TheMg2+ moved into M can be transferred to W1 since the potential difference EW1|M,C at the

FIG. 9 Distribution of the concentration of picrate (Pic�) between sites A and B in W1 (curve 1)

and M (curve 2), after electrolysis by applying a potential difference EMðA�BÞ between sites A and B

in M of 1 V for 3 h. Initial compositions: 2� 10�4 M H+Pic� + 1 M MgSO4 in W1 and 10�4 M

TPenA+TPhB� in M.


M|W1 interface is appropriate. In this regard, E1/2, for the transfer of Mg2+ from DCEcontaining 5 �10�4 M Mg2+ and 10�2 M BDPPM to W was determined to be �0:52 Vversus TPhBE by current-scan polarography at theWjDCE interface. The E1/2 correspondsto EW1|M,C at site C of 7 cm from site A (cf. Fig. 7).

C. Ion Transfer Along the W1|M Interface with Adsorbed SurfactantUnder Potential Difference Applied Between Sites A and B in W1

Poly(oxyethylene)octylphenyl ether (Triton X) is a nonionic amphiphilic compound.Triton X adsorbs at the aqueous|organic (W|O) solutions interface when it is addedinto W or O in a W–O system. The oxygen atoms in the polyoxyethlene group ofTriton X attract the positive charge of certain metal ions (Mnþ) in W, resulting in a ratherhydrophobic Mnþ–Triton X complex. On account of this complex formation, the facili-tated transfer of Mnþ from W to O can be attained in the presence of Triton X at the W|Ointerface when a potential difference is applied at this interface [12]. Here, the appropriatepotential difference is that more positive than about �0:4 V versus TPhBE if O is DCE.The facilitated ion transfer is accompanied by the desorption of Triton X (Fig. 11).

FIG. 10 (a) Membrane system with a domain in M used to investigate the transport of Mg2+ from

W2 to W1 through the domain; 100 �l of 10�2 M BDPPM was injected into M in order to form the

domain. (b) Relationship between the concentration of Mg2+ at site C in W1 and the distance of site

C from site A after electrolysis at EW1ðA�BÞ ¼ 1 V for 3 h. Compositions of W1, M, and W2: without

electrolyte in W1, 0.05 M TPenA+TPhB� (+ 0.01 M BDPPM in the domain) in M, and 10�3 M

Mg(TPhB)2 + 2 M MgSO4 in W2.


When 2� 10�3 M Triton X with 9.5 oxyethylene units (Triton X-100) was added toM of the cell of Fig. 2(b) of which W1 and M contained 10�3 M Na2SO4 and 0.05 MTpenA+TPhB�, respectively, and EW1ðA�BÞ of +0.6 V was applied, a white precipitate wasformed at the W1|M interface of site A. Since EW1|M,A is about �0:15 V versus TPhBE(where it is more positive than �0:4 V) when EW1ðA�BÞ ¼ þ0:6 V, and hence the transfer ofNa+ facilitated by Triton X-100 occurs at the W1|M interface, the precipitate is consid-ered to be the transferred Na+–Triton X-100 complex. The Triton X-100 which had beenadsorbed at the interface is desorbed during the facilitated transfer. When EW1ðA�BÞ waschanged to �0:6 V, the white precipitate moved toward site B along the W1|M interface.Since EW1|M,A is about �0:75 V versus TPhBE (where it is more negative than �0:4 V)when EW1ðA�BÞ ¼ �0:6 V, and hence the adsorption of Triton X-100 takes place at theW1|M interface of site A, the adsorbed Triton X-100 pushes the white precipitate towardsite B, which is responsible for the movement of the precipitate. The result described in thissection might have some relation with the fluidity of membranes.

D. Ion Transport from W1 to a Special Region of W2 Under PotentialDifference Applied Between Sites A and B in M

When EMðA�BÞ was applied between sites A and B in the cell of Fig. 2(c), reactions in aregion between sites A and B were also found to occur. Curves 1, 2, and 3 in Fig. 12 showconcentration profiles of Pic� found in W1, M, and W2, respectively, after electrolysis for3 h by applying EMðA�BÞ of 1 V between sites A and B in M of the membrane system ofwhich W1 contained 2� 10�4 M Pic� in addition to the supporting electrolyte (1 MMgSO4); M contained 10�4 M TPenA+TPhB� and W2 contained the supporting electro-lyte (2 M MgSO4). The transfer of Pic� from W1 to W2 occurred in a narrow regionaround 3 to 6 cm from site A. The transfer can be explained as follows: the standardpotential for the transfer of Pic� at the W|DCE interface (�0:36 V versus TPhBE) coin-cides with EW1|M,C or EW2|M,C at site C of 4 to 5 cm from site A (see Fig. 8), which meansthat Pic� in W1 transfers into M in a region between site A and the region around site C.On the other hand, Pic� transfers from M to W2 in a region between around site C and

FIG. 11 Schematic depiction of transfer of Mnþ in the presence of Triton X.


site B (see Fig. 8), if Pic� presents in M. In this experiment, however, Pic� did not exist inthe region between sites C and B in M. Therefore, Pic� transfers only in a small regionaround site C.

The result described in this section suggests that the ion transport from W1 to W2 ata special region of a membrane that resembles the transport at a biomembrane with an ionchannel or a domain can be realized even in the absence of any channel proteins ordomain-forming reagents.

VII. ELECTRON TRANSFER AT THE W1|M INTERFACE IN A REGIONBETWEEN SITES A AND B

The electron-transfer reaction could also be realized in a region between sites A and Bwhen EW1ðA�BÞ was applied to the membrane system of Fig. 2(b), which contained anoxidant or a reductant in M and a reductant or an oxidant in W1, respectively.

An example of the electron transfer was investigated with a membrane system inwhich M contained 10�2 M 7,7,8,8-tetracyanoquinodimethane (TCNQ) in addition to thesupporting electrolyte, and W1 contained 2� 10�3 M K4[Fe(CN)6]. Figure 13 shows theconcentration profile of TCNQ� produced in M after electrolysis for 3 h by applying 1 Vof EW1ðA�BÞ. Here, the concentration of TCNQ� was determined by spectrophotometrybased on the absorption of TCNQ� in DCE at 852 nm. The plot indicates that the electrontransfer from W1 to M due to the interfacial redox reaction, Eq. (4) [7], proceeded in aregion between about 5 cm from site A and site B:

½FeðCNÞ6�4�ðW1Þ þ TCNQ ðMÞ> ½FeðCNÞ6�3� ðW1Þ þ TCNQ�ðMÞ ð4Þ

FIG. 12 Distribution of the concentration of picrate (Pic�) between sites A and B in W1 (curve 1),

M (curve 2), or W2 (curve 3) after electrolysis by applying a potential difference EMðA�BÞ betweensites A and B in M of 1 V for 3 h. Initial compositions: 2� 10�4 M H+Pic� + 1 M MgSO4 in W1,

10�4 M TPenA+TPhB� in M, and 2 M MgSO4 in W2.


where ( ) denotes the phase where the species exists. In this connection, the half-wavepotential, E1/2, for the electron transfer at the interface between W1 containing 10�3 MK4[Fe(CN)6] and DCE containing 10�2 M TCNQ, was determined to be �0:43 V versusTPhBE by current-scan polarography at the W|DCE interface [13]. The E1/2 correspondsto EW1|M,C at site C of about 6 cm from site A (cf. Fig. 7).

VIII. SPONTANEOUS ‘‘PARALLEL TRANSPORT OF TYPE II’’ REALIZEDBY CHEMICAL REACTIONS

The ‘‘parallel transport of type II’’ mentioned in previous sections was attained in thepresence of a potential gradient between sites A and B applied by using an external circuit.The similar ‘‘parallel transport of type II’’ could also be attained spontaneously by thepotential gradient produced by chemical reactions at W|M interfaces of sites A and B.

An example of the cell that enabled such spontaneous ‘‘parallel transport of type II’’is illustrated as Fig. 14(a). Here, the cell system was assembled by layering W2 containing0.1 M NaCl + 2 M MgSO4, M of DCE containing 10�4 M TPenA+TFPB� and W1containing 2� 10�4 M HPic + 0.1 M NaCl, and then injecting 0.1 ml of DCE solutionscontaining 0.05 M TPenA+Cl� and 0.05 M Na+TFPB� into M of sites A and B, respec-tively. Here, TFPB� was used in this experiment as an anion in M instead of TPhB�, sinceNa+ can be kept in M more stably when the counteranion is more bulky as TFPB� thanTPhB�.

Figure 14(b) shows concentration profiles of Pic� in M (curve 1) and W2 (curve 2)observed 2 h after establishment of the cell system of Fig. 14(a). It is clear that Pic� wastransferred from W1 to W2 of a definite region (between 3 and 7 cm from site A), whichindicates that ‘‘parallel transport of type II’’ similar to that in Fig. 12 could be realizedeven by the potential difference between sites A and B produced chemically. The trans-port of Pic� can be understood by consulting plots 1 and 2 in Fig. 15, which showdistributions of potential differences, EW1|M,C and EW2|M,C at the W1|M and W2|Minterfaces of site C between sites A and B observed 1.5 to 2 h after injection of

FIG. 13 Relationship between the concentration of TCNQ� in M and the distance of site C from

site A after electrolysis at EW1ðA�BÞ ¼ 1 V for 3 h. Compositions of W1 and M: 2� 10�3 M

K4[Fe(CN)6] in W1 and 0.01 M TCNQ + 0.05 M TPenA+TFPB� in M.


TPenA+Cl� and Na+TFPB� into M of the cell of Fig. 14(a). The EW1|M,C or EW2|M,C is�0:61 V versus TPhBE when site C is in a region near to site A. This potential isconsidered to be determined by the composite transfers of Cl� from W1 to M andfrom M to W1 or those from W2 to M and from M to W2. The EW1|M,C or EW2|M,C

is 0.06 V versus TPhBE when site C is in a region near to site B. This potential isconsidered to be determined by the composite transfers of Na+ from W1 to M and fromM to W1 or those from W2 to M and from M to W2.

Since EW1|M,C is more negative than �0:36 V versus TPhBE in the region betweensite A and site C of about 6 cm from site A, as seen in curve 1 of Fig. 15, Pic� transfersfrom W1 to M in this region [the transfer was confirmed as curve 1 in Fig. 14(b)]. On theother hand, if Pic� presents in M, Pic� transfers from M to W2 in a region between sites Cand B in M, since EW2|M,C is more positive than �0:36 V in this region as seen in plot 2 ofFig. 15. In the present experiment, however, Pic� did not exist in the region between sitesC and B. Therefore, Pic� transferred only in a small region around site C [see curve 2 inFig. 14(b)].

FIG. 14 (a) Example of cell system that realizes a spontaneous ion transport from W1 to a part of

W2 by using a potential gradient between sites A and B in M produced by chemical reactions. (b)

Distribution of the concentration of picrate (Pic�) between sites A and B in M (curve 1) or W2 (curve

2). Results obtained 2 h after injection of TPenA+Cl� and Na+TFPB� to sites A and B,

respectively, in M.


In connection with the results in Fig. 15, it is obvious that regions where sharpchanges of EW1|M,C and EW2|M,C are observed are narrower than those in Fig. 8, andpotentials of wide regions in the vicinities of sites A and B are almost constants, whenthe potential difference between sites A and B is produced chemically. This change ofpotential can be understood by considering the diffusion of TPenA+Cl� added at site Atoward site B and that of Na+TFPB� added at site B toward site A.

IX. CONCLUSIONS

The fundamental features of ‘‘parallel transport of type I or II’’ which were realized byapplying a potential difference between two sites (A and B) in W1 or M, respectively, wereintroduced in this chapter. The ion- or electron-transfer reaction at the W1|M interface ina region between sites A and B (site C), and the ion transport from W2 to W1 through adomain formed in a part of M were found to be possible under the condition of ‘‘paralleltransport of type I’’. The ion-transfer reactions at the W1|M and W2|M interfaces inspecial regions between sites A and B, and the ion transport from W1 to a narrow regionin W2 through M were also found to be possible under the condition of ‘‘parallel transportof type I.’’

The membrane phenomena introduced here might endow some views useful forbetter understanding of phenomena at membranes, including biomembranes.

REFERENCES

1. N Kurauchi, Y Yoshida, N Ichieda, H Ohde, O Shirai, K Maeda, S Kihara, J Electroanal

Chem, 496:118–123, 2001.

2. N Kurauchi, Y Yoshida, N Ichieda, M Kasuno, K Banu, K Maeda S Kihara, J Electroanal

Chem, 526:101–106, 2002.

3. P Mitchell, Nature, 191:144–148, 1961; P Mitchell, Eur J Biochem 95:1–20, 1979.

FIG. 15 Relationship between the potential difference EW1|M,C or EW2|M,C at the W1|M interface

(curve 1) or at the W2|M interface (curve 2) of site C and the distance of site C from site A. Curves 1

and 2 were measured 1.5 to 2 h after the injection of TPenA+Cl� and Na+TFPB� to sites A and B,

respectively, in M of the cell of Fig. 14(a).


4. LS Yaguzhinsky, LI Boguslavsky, AD Ismailov, Biochim Biophys Acta, 368:22–28, 1974.

5. RG Gennis, Biomembranes: Molecular Structure and Function. Tokyo: Springer-Verlag,

1990, chs 5 and 6. (Japanese edition.)

6. O Shirai, S Kihara, M Suzuki, K Ogura, M Matsui, Anal Sci, 7 (suppl): 607–610, 1991.

7. O Shirai, S Kihara, Y Yoshida, M Matsui, J Electroanal Chem, 389:61–70, 1995.

8. C Beriet, HH Girault, J Electroanal Chem, 444:219–229, 1998.

9. Z Samec, A Trojanek, J Langmaier, E Samcova, J Electroanal Chem, 481:1–6, 2000.

10. Y Yoshida, M Matsui, O Shirai, K Maeda, S Kihara, Anal Chim Acta, 373:213–225, 1998.

11. S Umetani, M Matsui, Anal Chem, 64:2288–2292, 1992.

12. Z Yoshida, S Kihara, J Electroanal Chem, 227:171–181, 1987.

13. S Kihara, Z Yoshida, M Suzuki, K Maeda, K Ogura, M Matsui, J Electroanal Chem,

271:107–125, 1989.


20Ion-Transport Processes ThroughMembranes of Various Types: LiquidMembrane, Thin Supported LiquidMembrane, and Bilayer Lipid Membrane

OSAMU SHIRAI Japan Atomic Energy Research Institute, Ibaraki, Japan

SORIN KIHARA Kyoto Institute of Technology, Kyoto, Japan

I. INTRODUCTION

The membrane potential, which determines the membrane transport of an ion from oneaqueous phase (W1) to another (W2) through a membrane (M), is composed of potentialsdue to transfers of the ion at the W1/M, and M/W2 interfaces and those due to masstransfers within W1, M, and W2. Here, the membrane potential is defined as the potentialdifference between W1 and W2. It has been widely supposed that the potential for themass transfer in M contributes significantly to the membrane potential, since most inves-tigations on the membrane transport so far carried out employed membranes of highelectrical resistance containing negligible concentrations of ions. However, the membranetransport in the presence of a high concentration of ions in the membrane is also impor-tant especially in connection with the ion transfer through biomembranes or bilayer lipidmembranes [1–12]. These membranes concentrate hydrophobic ions spontaneously intotheir lipid phases and even hydrophilic ions when the membranes contain hydrophobiccomplexing agents. In this case, the share of the potential for the interfacial ion transfer inthe membrane potential is considered to be more significant than that for the mass transferin the membrane.

A quantitative understanding of membrane transport is very important for elucidat-ing physiological reactions occurring at biomembranes such as nervous transmission,respiration, and metabolism as well as the application of membrane transports to thedevelopment of analytical methods such as liquid membrane (LM)-type ion sensors andmembrane separations [1,2,13].

The voltammetry for the ion transfer at the interface of two immiscible electrolytesolutions, VITIES, is expected to offer much information available for analyzing the iontransfer at the aqueous/membrane interface [14,15], if the organic solution is regarded asthe membrane. Transfer energies of ions at the aqueous/membrane interface and amountsof ions transferred can be evaluated precisely by VITIES. The kinetics of the ion transferand the interfacial adsorption can also be investigated by this method. The present authorsmeasured the relations between the membrane potential and the membrane current (the


current flowing between W1 and W2), and obtained a voltammogram for the ion transferthrough a membrane (VITTM). When the VITTM was compared with the voltammo-grams at the W1/LM and LM/W2 interfaces recorded simultaneously with the VITTM, itwas found that the ion transfer through a membrane was determined practically by thecomplementary ion-transfer reactions at two aqueous/membrane interfaces when the LMcontained sufficient electrolytes [16,17].

The ion-transport process through a bilayer lipid membrane (BLM), which is muchthinner than an LM, was also explained on the basis of ion-transfer reactions at theaqueous/BLM interface [17,18].

In this chapter, the interpretation of the membrane transport process through an LMor a BLM based on the voltammetric concept and method is introduced, and the impor-tant role of charge-transfer reactions at aqueous/membrane interfaces in the membranetransport is emphasized [16–18].

II. ION TRANSPORT THROUGH MEMBRANES IN PRESENCE OFSUFFICIENT ELECTROLYTES

A. Relations Between Voltammograms for Ion Transfers Through aLiquid Membrane and Those at Aqueous/Membrane Interfaces

An example of the voltammogram for ion transfer through an M is realized as curve 1 inFig. 1. The cell used for the measurement of the VITTM was as shown in Eq. (1) in whichW1 and W2 (5 mL each) containing a supporting electrolyte (MgSO4) were separated by anitrobenzene (NB) solution containing a supporting electrolyte (crystal violet tetraphenyl-borate, CVþTPhB�) and an ionophore (dibenzo-18-crown-6, DB18C6). The NB solutionworked as the LM of interfacial area 1 cm2 and thickness 1 cm.

2:5� 10�4 MK2SO4

1MMgSO4

ðW1Þ

��0:02MDB18C6

0:1MCVþTPhB�

ðLMÞ

�� 2MMgSO4

ðW2ÞCE1 RE1 � EW1=LM � RE3 RE4 � ELM=W2 � RE2 CE2

ð1Þ

EW1–W2

IW1-W

The VITTM was recorded by scanning the membrane potential, EW1�W2, and mea-suring the membrane current, IW1�W2. Here, EW1�W2 was applied using two silver–silverchloride reference electrodes, RE1 and RE2, as the potential of RE1 versus RE2. Duringthe recording of the VITTM, voltammograms for ion transfer at the W1/M and M/W2interfaces (curves 2 and 3) were also recorded by monitoring the potential differences(EW1=LM and ELM=W2) at the W1/LM and LM/W2 interfaces as a function of IW1�W2.Here, EW1=LM and ELM=W2 were measured as the potential of RE1 versus RE3 and that ofRE2 versus RE4, where RE3 and RE4 were TPhB� ion-selective electrodes placed in theLM near to the W1/LM and LM/W2 interfaces, respectively. Corresponding voltammo-grams in the absence of K2SO4 in W1 are shown using broken lines (curves 1 0, 2 0, and 3 0).Although voltammograms at the W1/LM interface (curves 3 and 3 0) are identical witheach other, positive and negative peaks exist in voltammograms 1 and 2 in the presence ofK2SO4. The positive peak, the final rise, and the final descent in curve 2 are attributable to


the transfer of Kþ from W1, to LM facilitated by DB18C6, TPhB� from LM to W1, andCVþ from LM to W1, respectively, which was confirmed by changing concentrations ofthe ions in each phase. Here, the final rise and the final descent mean large positive andnegative currents, respectively, limiting the potential window. The negative peak is due tothe transfer of Kþ, which has moved into the LM during the positive scan from LM toW1. In this regard, consulting the voltammetric work on the transfer of Kþ at an aqueous/NB interface [44], the positive and negative peaks in curve 2 are attributable to the transferof Kþ from W1 to LM and that of Kþ complexed with DB18C6 from LM to W1 con-trolled by diffusion of Kþ species. The final rise and the final descent in curve 3 correspondto the transfer of CVþ from LM to W2 and that of TPhB� from LM to W2, respectively.

Comparing curve 1 with curves 2 and 3, it is obvious that (1) the potential window incurve 1 is about twice that in curve 2 or 3, (2) the potential regions where the positive andthe negative peaks appear in curve 1 are different from those in curve 2, and (3) the slopesof the positive peak, negative peak, final rise, and final descent in curve 1 are much smallerthan those in curves 2 and 3.

FIG. 1 Voltammograms for ion transfer from one aqueous phase (W1) to another (W2) through a

liquid membrane (LM) (curves 1 and 1 0), at the W1/LM interface (curves 2 and 2 0) and at the LM/

W2 interface (curves 3 and 3 0). Compositions of W1, LM, and W2 for curves 1, 2, and 3: 2:5�10�4M K2SO4 þ 1M MgSO4 in W1, 0.02 M dibenzo-18-crown-6 þ0:1M CVþTPhB� in LM, 2 M

MgSO4 in W2; for curves 1 0, 2 0, and 3 0: the same as those for curves 1, 2, and 3, but in the absence of

K2SO4 in W1. Scan rate of EW1�W2; 0:01V s�1.


In order to elucidate the reaction involved in curve 1, the relations among EW1�W2 incurve 1, EW1=LM in curve 2, and ELM=W2 in curve 3 were investigated at a definite mem-brane current, I0W1�W2, taking into account that currents flowing across both the W1/LMand LM/W2 interfaces must be the same and equal to the membrane current. The relation:

EW1�W2 ¼ EW1=LM þ ELM=W2 þ I0W1�W2R ð2Þwas found to be held at any I0W1�W2, where R is the resistance between RE1 and RE2.

When W1, W2, and LM contain sufficient concentrations of ions, the contribution ofI0W1�W2R to EW1�W2 is not significant, and Eq. (2) can be approximated by

EW1�W2 ¼ EW1=LM þ ELM=W2 ð3ÞAs for the LM system of Eq. (1), the resistances of W1 and W2 containing MgSO4 werenegligibly small and that of NB solution containing 0.1 M CVþTPhB� was 1:05 k cm,which produces I0W1�W2R ¼ 13:4mV when I0W1�W2 is 10�A.

Equation (3) suggests that the membrane potential in the presence of sufficientelectrolytes in W1, W2, and LM is primarily determined by the potential differences attwo interfaces which depend on ion-transfer reactions at the interfaces, although thepotential differences at interfaces are not apparently taken into account in theoreticalequations such as those of Nernst–Planck, Henderson, and Goldman–Hodgkin–Katz,which have often been adopted in the discussion of the membrane potential [13,19–22].

The characteristics of VITTM in Fig. 1 can easily be understood by taking intoaccount the relations of Eq. (3) as follows. Since the membrane potential where thepositive wave appears (EW1�W2 indicated by A in curve 1) is the sum of EW1=LM indicatedby B in curve 2 and ELM=W2 indicated by C in curve 3, the positive wave in curve 1 isconsidered to be caused by the coupled reactions of both the transfer of Kþ from W1 toLM facilitated by DB18C6 (the positive wave in curve 2) and that of CVþ from LM to W2(the final rise in curve 3). Hence, the potential region for the positive wave in curve 1differs from that in curve 2.

On the basis of a similar analysis, the negative wave in curve 1 is considered toconsist of the transfer of Kþ from LM to W1 (the negative wave in curve 2) and thatof CVþ from W2 to LM (the negative wave in curve 3). The final rise in curve 1 involvesthe transfer of TPhB� from LM to W1 (the final rise in curve 2) and that of CVþ from LMto W2 (the final rise in curve 3), and the final descent in curve 1 involves the transfer ofCVþ from LM to W1 (the final descent in curve 2) and that of TPhB� from LM to W2 (thefinal descent in curve 3). These coupled reactions are responsible for the wide potentialwindow.

One of the reasons for the small slope of the positive wave in curve 1 is the mem-brane resistance [R in Eq. (2)]. However, the slope is still smaller than the slope of thepositive wave in curve 2 even after it has been corrected for the membrane resistanceemploying R ¼ 1:05 k cm. The small slope after the correction can be explained byconsidering that the slope of the positive wave in curve 1 is composed of the slope ofthe positive wave in curve 2 and that of the final rise in curve 3. The small slopes of thefinal rise and the final descent in curve 1 are also attributable to the membrane resistanceand the coupled reactions at two W/LM interfaces.

Horvath and Horvai [23] also investigated the VITTM using a poly(vinyl chloride)membrane, which contained hydrophobic electrolytes by changing the ionic compositionof aqueous phases, and found that the VITTM was characterized by the ion transfer attwo aqueous/membrane interfaces. Samec et al. [24] proposed the theory of the cyclic andconvolution potential sweep voltammetry of a reversible ion transfer through a liquid


membrane, and the theory was verified by voltammetric measurements of a semihydro-phobic ion transfer across a supported o-nitrophenyloctyl ether (NPOE) membrane. Thekinetics of charge transfer across a supported o-NPOE membrane was studied by Barkeret al. [25]. A theory of cyclic voltammetry for the ion transfer across a liquid membrane inthe absence of electrolytes using the Nernst–Planck equation and the electroneutralityassumption was described by Kakiuchi [26].

In the analysis of VITTM, Eq. (2) is an important relation. Figure 2 shows timecourses of EW1=LM and ELM=W2 obtained when EW1�W2 was scanned linearly under thesame condition as that for Fig. 1. Although the relations of Eq. (2) was kept at all timesamong EW1�W2, EW1=LM, and ELM=W2, EW1=LM and ELM=W2 did not vary linearly, asdemonstrated by Beriet and Girault [27].

B. Influence of Ion Transfer Reaction at One W/M Interface on That atAnother W/M Interface Under Applied Membrane Potential

Voltammograms shown as curves 1, 2, and 3 in Fig. 3(a) were recorded under the samecondition as that for curves 1, 2, and 3 in Fig. 1, but adding 0.01 M MgBr2 to W2. Thepositive wave in curve 1 appears at a potential about 0.19 V less positive than that in theabsence of MgBr2 [cf. curves 1 in Figs 1 and 3(a)]. The effect of MgBr2 on the positivewave in the VITTM can be understood by considering ion-transfer reactions at the W1/LM and LM/W2 interfaces. Although the voltammogram at the W1/LM interface isindifferent to the presence of MgBr2 in W2 [cf. curves 2 in Figs 1 and 3(a)], the finalrise in the voltammogram at the LM/W2 interface in the presence of MgBr2 in W2, whichis attributable to the transfer of Br� from W2 to LM, lies at a potential about 0.19 V lesspositive than that in the absence of MgBr2, which is attributable to the transfer of CVþ

from LM to W2 [cf. curves 3 in Figs. 1 and 3(a)], indicating that the Gibbs energy oftransfer of Br� is 18:3 kJ mol�1 less than that of CVþ at the LM/W2 interface.Consequently, the positive wave in the VITTM [curve 1 in Fig. 3(a)] composed of thetransfer of Kþ from W1 to LM and that of Br� from W2 to LM appears at a membranepotential about 0.19 V less positive than that (curve 1 in Fig. 1) composed of the transferof Kþ from W1 to LM and that of CVþ from LM to W2.

FIG. 2 Time courses of EW1=LM and ELM=W2 during the cyclic scan of EW1�W2 observed with the cell

of Eq. (1). Scan rate of EW1�W2; 0:01V s�1.


The voltammetric information given here suggests that the transfer of a given cationfrom W1 to LM can be achieved under a lower membrane potential when an anion, ofwhich the Gibbs transfer energy at the LM/W2 interface is lower, is added to W2. In thecase of the above-mentioned membrane system, the transfer of Kþ from W1 to LM in thepresence of 0.01 M MgBr2 in W2 is expected to be attained even at a membrane potentialof 0.19 V (which corresponds to a Gibbs energy of transfer of 18:3 kJ mol�1) less positivethan that required for the transfer in the absence of MgBr2. Actually, when the electrolysiswas carried out for 60 h, applying a membrane potential of þ0:20V to the cell of Eq. (1),72% of the Kþ was transferred from W1 to LM or W2 in the presence of 0.01 MMgBr2 inW2 although practically no transfer of Kþ was observed in the absence of MgBr2. In thisregard, Kþ once transferred to LM can be readily transferred to W2 at a membranepotential of þ0:20V, as mentioned later.

Curves 1, 2, and 3 in Fig. 3(b) are voltammograms recorded under the same condi-tion as that for curves 1, 2, and 3 in Fig. 1, respectively, but adding 0.1 M tetrabutylam-monium tetraphenylborate (TBAþTPhB�) as supporting electrolyte to the LM instead ofCVþTPhB�. The positive wave and the final rise in the VITTM with TBAþTPhB� [curve1 in Fig. 3(b)] were about 0.10 V less positive than those with CVþTPhB� (curve 1 in Fig.1), and the final descent in curve 1 in Fig. 3(b) was about 0.10 V more positive than that in

FIG. 3 Voltammograms are the same as those in Fig. 1, but (a) in the presence of 0.01 M MgBr2 in

W2, and (b) 0.1 M TBAþTPhB� in LM instead of CVþTPhB�.


curve 1 in Fig. 1. These characteristics of the VITTM can be understood by taking intoaccount ion-transfer reactions at the W1/LM and LM/W2 interfaces. For example, sincethe positive wave in curve 1 is composed of the positive wave in curve 2 due to the transferof Kþ from W1 to LM and the final rise in curve 3 due to the transfer of TBAþ from LMto W2, which lies at a potential 0.10 V less positive than that in curve 3 in Fig. 1 due to thetransfer of CVþ from LM to W2, the positive wave in curve 1 of Fig. 3(b) appears at apotential 0.10 V less positive than that in curve 1 in Fig. 1.

The results given here suggest that, although Eq. (3) is simple, the relations describedin this equation might be very important in elucidating the membrane transport phenom-ena under a membrane potential applied not only by an external electrical source but alsoby chemicals such as redox agents.

C. Voltammograms for Ion Transfer Through an LM in Presence of theObjective Ion in W1, LM, and/or W2

Voltammograms 1 shown in Fig. 4 are VITTMs recorded adding Kþ to W1, W2, and/orLM of the cell containing the same concentrations of supporting electrolyte and DB18C6as those in the cell of Eq. (1). Two positive and two negative waves were observed in theVITTM, when 5� 10�4 MKþ was added to both W1 and W2 or when 5� 10�4 MKþ wasadded to the LM as shown as curve 1 in Fig. 4(a) or 4(b), respectively.

Positive and negative waves symmetrical about the origin (the point of 0 V and 0 A)of each other were observed in the VITTM [curve 1 in Fig. 4(c)], when 0.1 M Kþ wasadded to both W1 and W2 and 10�3 MKþ to LM.

Although the assignment of the above-mentioned waves in VITTMs seems to becomplicated, it is easily attained if we take the relations of Eq. (3) into account togetherwith the voltammograms at the W1/LM and LM/W2 interfaces, which are shown ascurves 2 and 3, respectively, in Fig. 4. For example, the positive wave in curve 1 in Fig.4(c) is composed of the final rise in curve 2 due to the transfer of Kþ from W1 to LM andthe positive wave in curve 3 due to the transfer of Kþ from LM to W2, and the negativewave in curve 1 is composed of the final descent in curve 3 due to the transfer of Kþ fromW2 to LM and the negative wave in curve 2 due to the transfer of Kþ from LM to W1.Since both the positive and the negative waves in curve 1 in Fig. 4(c) involve the same ion-transfer reactions as mentioned above, i.e., the transfer of Kþ from an aqueous solution(W1 or W2) to LM and that from LM to an aqueous solution (W1 or W2), these wavesappear symmetrically about the origin.

D. Ion Separation by Electrolysis Under Applied Membrane Potential

In the following, electrolysis under an applied membrane potential will be discussed fromthe viewpoint of the separation of ions, employing the membrane system as shown in Eq.(4):

2:5� 10�4 MK2SO4

2:5� 10�4 MNa2SO4

1MMgSO4

ðW1Þ

��

0:02MDB18C6

0:1MCVþTPhB�

ðLMÞ

�� 2MMgSO4

ðW2Þ

ð4Þ

Curves 2 and 3 in Fig. 5(a) are imaginary steady-state voltammograms at the W1/LM and LM/W2 interfaces before the electrolysis, which are illustrated by reference to


FIG. 4 Voltammograms for ion transfer between W1 and W2 through an LM (curve 1), at the W1/

LM interface (curve 2) and at the LM/W2 interface (curve 3). Compositions of W1, LM, and W2: (a)

2:5� 10�4 M K2SO4 þ 1M MgSO4 in W1, 0.02 M dibenzo-18-crown-6 þ0:1 M CVþTPhB� in LM,

2:5� 10�4 M K2SO4 þ 2 M MgSO4 in W2; (b) 1 MMgSO4 in W1, 5� 10�4 M KþTPhB� þ0:02 M

dibenzo-18-crown-6 þ0:1 M CVþTPhB� in LM, 2 MMgSO4 in W2: (c) 0.1 M KClþ 1M MgSO4 in

W1, 10�3 M KþTPhB� þ 0:02 M dibenzo-18-crown-6 þ0:1 M CVþTPhB� in LM, 0.1 M

KClþ 2 M MgSO4 in W2. Scan rate of EW1�W2; 0:01V s�1.


voltammograms in Figs 1 and 4 and by knowledge of the transfer of Kþ and Naþ at theW/NB interface facilitated by DB18C6. The first and the second positive waves in curve 2correspond to the transfers of Kþ and Naþ, respectively, from W1 to LM. The final rise incurve 3 corresponds to the transfer of CVþ from LM to W2. Curve 1 can be obtained asthe VITTM before the electrolysis by combining curves 2 and 3 in the light of Eq. (3).Figure 5(a) indicates that only Kþ can be transferred from W1 to LM, if þ0:35V isapplied as the membrane potential, EW1�W2. The transfer of CV

þ from LM to W2 occursat the LM/W2 interface under this EW1�W2.

Considering the state that 10 and 40% of Kþ have been transferred from W1 to LMand W2, respectively, by the electrolysis, the voltammograms at the W1/LM and LM/W2interfaces are transformed to be curves 2 and 3, respectively, in Fig. 5(b) and hence theVITTM becomes curve 1 in Fig. 5(b). A wave composed of the transfer of Kþ from W1 toLM and that from LM to W1 and a positive wave for the transfer of Naþ from W1 to LM

FIG. 5 Imaginary steady-state voltammograms for ion transfer through LM (curves 1), at the W1/

LM interface (curves 2) and at the LM/W2 interface (curves 3). Compositions of W1, LM, and W2:

(a) 5� 10�4 M Kþ5� 10�4 M Naþ þ 1 M MgSO4 inW1, 0.02 M dibenzo-18-crown-6 þ0:1 M

CVþTPhB� in LM, 1 M MgSO4 in W2; (b) 2:5� 10�4 MKþ þ 5� 10�4 MNaþ þ 1M MgSO4 in

W1, 5:0� 10�5 MKþþ0:02M dibenzo-18-crown-6 þ0:1M CVþTPhB� in LM, 2:0� 10�4 M

Kþ þ 1M MgSO4 in W2.


are observed in curve 2. A positive wave for the transfer of Kþ from LM to W2 is observedin curve 3. A decrease in Kþ concentration in W1 brings about a decrease in the positivecurrent at the W1/LM interface, IW1=LM, due to the transfer of Kþ from W1 to LM. At thesame time, since IW1=LM and the current at the LM/W2 interface, ILM=W2, should be thesame (and equivalent to IW1�W2) in a membrane system, ILM=W2 due to the transfer of CVþ

from LM to W2 decreases along the final rise of the voltammogram at the LM/W2interface (curve 3) which has a slope, causing the negative shift of ELM=W2 (from A toA 0). The negative shift of ELM=W2 results in the positive shift of EW1=LM (from B to B 0)because EW1�W2 is kept constant [cf. Eq. (3)]. At the EW1=LM of B 0, not only Kþ but alsoNaþ transfer from W1 to LM.

The above discussion suggests that the quantitative separation of Kþ from Naþ

cannot be attained by electrolysis under the controlled membrane potential when themembrane system shown in Eq. (4) is employed. Here, it is noteworthy that the wavefor the transfer of Kþ from LM to W2 lies at potentials more negative than A or A 0 asseen in curve 3 in Fig. 5(b), indicating that Kþ once moved to LM can easily be transferredto W2 under the EW1�W2 of þ0:35V. The transfer of Naþ from LM to W2 is considered toproceed more easily than that of Kþ, since the wave for the transfer of Naþ, which is morehydrophilic than Kþ, lies at potentials more negative than that of Kþ in the voltammo-gram at the LM/W2 interface, although it is not shown in Fig. 5.

The imaginary steady-state voltammograms in Fig. 6 are illustrated under the sameconditions as those for voltammograms in Fig. 5, but assuming the presence of 2� 10�3 Mvalinomycin in the LM instead of DB18C6. Since valinomycin facilitates the transfer ofKþ from aqueous to NB solution more significantly and selectively than DB18C6, wavesfor the transfer of Kþ and Naþ from W1 to LM are separated more in the presence ofvalinomycin than in the presence of DB18C6 [cf. curves 2 in Figs. 5(a) and 6(a)]. Figure6(a) indicates that only Kþ transfers from W1 to LM when þ0:20V is applied as EW1�W2.

Figure 6(b) includes the voltammograms at the W1/LM and LM/W2 interfaces andthe VITTM at the stage when 10 and 40% of Kþ have been transferred from W1 to LMand W2, respectively, by the electrolysis at EW1�W2 ¼ þ0:20V. Although the decrease inthe current for the transfer of Kþ from W1 to LM causes shifts in ELM=W2 (from C to C 0)and EW1=LM (from D to D 0) similarly to the behavior with DB18C6, the transfer of Naþ

from W1 to LM hardly occurs in the case with valinomycin, because the difference inpotential between the wave for the transfer of Kþ and that of Naþ from W1 to LM in thepresence of valinomycin is much larger than the shift in EW1=LM. Therefore, effectiveseparation of Kþ from Naþ by electrolysis under the controlled membrane potential isexpected if valinomycin is used as the ionophore in the LM instead of DB18C6.

III. ION TRANSFER THROUGH A THIN SUPPORTED LIQUID MEMBRANE

Voltammograms 1 and 1 0 in Fig. 7 were recorded under the same conditions as those forcurves 1 and 1 0, respectively, in Fig. 1, but employing a thin supported liquid membrane(Teflon–LM impregnated with the NB solution) instead of LM and scanning EWI�W2 at arate of 0:1V s�1 instead of 0:01V s�1. The characteristics of these voltammograms withTeflon–LM such as potential windows, potential regions where waves appear and theslopes of the curves, and the final rise or the final descent resemble closely those withLM, which indicates that even ion transfer through a thin membrane such as Teflon–LMof 48�m thickness is determined by the ion-transfer reactions at two aqueous/membrane


interfaces similarly to that through a thick membrane such as an LM of 1 cm thicknesswhen the membrane contains sufficient electrolyte.

IV. ION TRANSPORTS THROUGH BILAYER LIPID MEMBRANES

The ion transport from one aqueous solution (W1) to another (W2) through a bilayer lipidmembrane (BLM) in the presence of hydrophobic ions has been investigated extensivelyfor the fundamental understanding of the feature of ion transfers through biomembranes[2–12]. Although a BLM represents a high-energy barrier for hydrophilic ions such as Kþ,Naþ, or Cl� [28], ion transport occurs easily when a hydrophobic ion is added to W1 and/or W2 in the presence of hydrophilic salts even if the additive is fairly dilute (e.g., 10�6 M)[2–12]. The mechanism and energetics of ion transport in the presence of a hydrophobicion have been discussed frequently [2–12,29–34]. Most authors assumed that condensation

FIG. 6 Imaginary steady-state voltammograms are the same as those in Fig. 5, but in the presence

of valinomycin in LM. Compositions of W1, LM, and W2: (a) 5� 10�4 MKþ þ 5� 10�4 MMgSO4

in W1, 2� 10�4 M valinomycinþ 0:1M CVþTPhB� in LM, 1 M MgSO4 in W2; (b) 2:5� 10�4 MKþ þ 5� 10�4 M Naþ þ 1M MgSO4 in W1, 5:0� 10�5 M Kþ þ 2� 10�3 M valinomycinþ 0:1MCVþTPhB� in LM, 2:0� 10�4 M Kþ þ 1M MgSO4 in W2.


of the hydrophobic ion into the BLM occurred, and attributed the rate-determining pro-cess to the diffusion-controlled mass transfer of the hydrophobic ion from the bulk of theaqueous solution, W, to the W/BLM interface when the hydrophobic ion was dilute.When the concentration of the hydrophobic ion was high, LeBlanc [5] considered thesaturation of the space charge in the BLM with the ion, Bruner [29] and Ketterer et al.[30] the saturation of the interface by the adsorption of the ion, and de Levie and Seidah[6] the variation in concentration of the ion in the stagnant (Nernst) and space charge(Gouy–Chapman) layers in W due to the partition equilibrium or the first-order phase-transfer kinetics at the W/BLM interface.

If we adopt the diffusion-controlled process in the case of the dilute hydrophobicion, however, it is difficult to explain not only the extraordinarily large current caused bythe ion transport (10–100 times larger than the ordinary diffusion-controlled current), butalso the cyclic voltammogram symmetrical about the origin (point of 0 V and 0 A), whichis observed by scanning the membrane potential and measuring the membrane currentwith the BLM system in the presence of hydrophobic ions in one of two aqueous phases[7,10,17]. In order to overcome these difficulties, Kutnik and Tien [9] considered thetransfer of hydrophobic ion, which had been concentrated in the BLM, from the BLMto both W1 and W2, and applied the thin-layer electrode theory to the analysis of thevoltammogram.

The present authors found that the role of the hydrophilic ions in aqueous phases,which must be distributed into the BLM together with the hydrophobic ion as the counter-ion in order to hold the electroneutrality in the membrane and aqueous phases, has notbeen taken into account in most papers on the transport process, and hence it seemsdifficult to understand the different features of the ion transport observed by varyingthe kind or concentration of hydrophilic salts based on the concepts so far proposed. Inprevious studies [17,18], the voltammogram for ion transport through a BLM was com-pared with that through a liquid membrane, and it was pointed out that the hydrophilicion moves into the BLM spontaneously as the counterion of the hydrophobic ion; it is thehydrophilic ion that transfers between W1 and W2, not the hydrophobic ion.

FIG. 7 Voltammograms are the same as those in Fig. 1, but with a thin supported liquid membrane

instead of an LM. Supported liquid membranes: Teflon films immersed in nitrobenzene solution

containing 0.02 M dibenzo-18-crown-6 þ 0:1M CVþTPhB�. Compositions of W1 and W2 for

curves 1 and 1 0; the same as those for curves 1 and 1 0 in Fig. 1. Scan rate of EW1�W2; 0:1V s�1.


Voltammograms for transports of various ions through BLMs of various compositionswere explained. Processes involved in the transports were also then elucidated on the basisof the results, taking into account solution chemical properties of the transferring ion, thecoexisting ion, and the constituents of the BLM.

A. Various Types of Voltammograms for Ion Transfer Through BLMs

The voltammograms were investigated employing a typical BLM made from n-decanecontaining a 1:1 mixture of phosphatidylcholine (PC) and cholesterol (Ch) by the brushingmethod and choosing ions in aqueous phases in the light of the standard Gibbs’ transferfree energy, �G0

tr;W!Org, from aqueous (W) to organic (Org) such as NB or 1,2-dichlor-oethane (DCE), which are summarized in Table 1 [14,15]. Here, the �G0

tr;W!org waspresumed to be a good measure of the hydrophilic or hydrophobic property of an ion,which might be deeply involved in the ion transport. In Table 1, ions are classified into fivecategories, i.e., hydrophilic ions with �G0

tr:W!NB > 10 kJmol�1 (category I), slightlyhydrophobic ions with 10 > �G0

tr;W!NB > 0 kJmol�1 (category II), rather hydrophobicions with 0 > �G0

tr:W!NB > �20 kJmol�1 (category III), very hydrophobic ions with�20 > �G0

tr;W!NB > �37 kJmol�1 (category IV), and extremely hydrophobic ions with �37 kJmol�1 > �G0

tr;W!NB (category V). In Table 1, Dipicrylamide, picrate, ethyl violet,tetraphenylarsonium, tetrapentylammonium, tetrabutylammonium, tetrapropylammo-nium, tetraethylammonium, and tetramethylammonium ions are abbreviated as DPA�,Pic�, EVþ, TPhAsþ, TPenAþ, TBAþ, TPrAþ, TEAþ, and TMAþ, respectively.

TABLE 1 Standard Gibbs Free Energies for

Transfers (�G0tr) of Various Ions from Aqueous (W)

to Nitrobenzene (NB) or 1,2-Dichloroethane (DCE)

Phases

Ion

W/NB

�G0tr (kJ mol�1)

W/DCE

�G0tr (kJ mol�1) Category

DPA� �39.4 V

TPhB� �35.9 �35.1 IV

Pic� �4.6 III

ClO�4 8.0 17.2 II

Br� 28.4 38.5 I

Cl� 31.4 46.4 I

SO2�4 >67.3 I

EVþ � �44a V

CVþ �39.5 V

TPhAsþ �35.9 �35.1 IV

TPenAþ �35.1 �34.7 IV

TBAþ �24.0 �21.8 IV

TPrAþ �10.0 �8.8 III

TEAþ �5.7 4.2 III

TMAþ 3.4 17.6 II

Kþ 23.4 I

Naþ 34.2 I

Mg2þ 69.6 I

aS. Kihara and O. Shirai, unpublished work.


When an ion was added to one of two aqueous phases in the presence of hydrophilicsalt, the voltammogram was transformed, depending on the property and concentration ofthe added ion. The voltammograms observed could be classified into four types, which willbe denoted as Types A–D hereafter.

Type A: Voltammograms observed with the cell in the absence of an extremelyhydrophobic ion, a very hydrophobic ion, or a rather hydrophobic anion in aqueousphases.

In the voltammogram that was obtained with the cell of Eq. (5) in which W1 and W2contained a salt composed of hydrophilic ions of category I (Table 1), there was no peakindicative of the ion transfer.

0:1 or 0:2M

Hydrophilic salt

ðe.g., KCl, NaCl, MgCl2

K2SO4; Na2SO4; MgSO4ÞðW1Þ

�� ðBLMÞ

��

0.1 or 0.2M

Hydrophilic salt

ðe.g., KCl, NaCl, MgCl2;

K2SO4; Na2SO4; MgSO4ÞðW2Þ

ð5Þ

As an example, curve 1 in Fig. 8 realizes the voltammogram recorded with W1 and W2containing 0.1 MMgSO4 by scanning EW1�W2 in the region between þ0:1 and �0:1V and,simultaneously, by measuring IW1�W2. Similar voltammograms without peaks wereobserved when the hydrophilic salt in W1 and W2 was NaCl, KCl, MgCl2, K2SO4, orNa2SO4 instead of MgSO4. The voltammograms of Type A were also observed when oneof the slightly hydrophobic ions of category II or rather hydrophobic cations of categoryIII was added in place of MgSO4.

Type B: Voltammograms observed with the cell in the presence of an extremelyhydrophobic anion, an extremely hydrophobic cation, or a very hydrophobic anion inW1 in addition to the hydrophilic salt.

In the voltammogram of Type B, which was obtained by the addition of one of theextremely hydrophobic ions of category V or very hydrophobic anions of category IV toW1 of the system of Eq. (5) in the presence of a hydrophilic salt in W1 and W2, well-

FIG. 8 Voltammograms for ion transfer through a BLM composed of PC and Ch. Curve 1: 0.1 M

MgSO4 in both W1 and W2; curves 2 and 3: as curve 1, but in the presence of 10�6 M DPA� in W1

(curve 2) and 10�6 M TPhB� in W1 (curve 3). Scan rate of EW1�W2: 0:01V s�1.


established positive and negative peaks symmetrical about the origin (the point of 0 V and0 A) appeared, even though the concentration of the added hydrophobic ion was verydilute (< 10�6 M) and the ion was added to one of two aqueous phases. The half-peakpotentials of both the positive and negative peaks were around 0 V. An example of thevoltammogram of Type B is shown as curve 2 in Fig. 8; this was observed when 10�6 MDPA� was added as the form of Mg2þ(DPA�)2 to W1 of the BLM system in the presenceof 0.1 M MgSO4 in W1 and W2. The peak current density was nearly proportional to theconcentration of DPA� in the range between 5� 10�8 and 10�5 M and to the square rootof the scan rate of EW1�W2 in the range between 0.01 and 0:1V s�1. It increased slightlywith an increase in concentration of the hydrophilic salt, MgSO4, in aqueous phases from0.1 to 1 M, and increased when the hydrophilic salt in the aqueous phases was changedfrom 0.1 M MgSO4 to 0.1 M K2SO4 or 0:1MNa2SO4. Here, the current density wascalculated from division of the current observed by the area of the BLM measured micro-scopically.

Voltammograms observed employing 0.1 M MgCl2, 0.2 M KCl, or 0.2 M NaCl inplace of 0.1 M MgSO4, 0:1MK2SO4, or 0:1MNa2SO4, respectively, were almost identicalwith those with sulfates.

Adding TPhB� instead of DPA� as the form of Mg2þ(TPhB�)2 to W1 of the systemin the presence of 0.1 M MgSO4 in W1 and W2, the voltammogram observed was that ofType B, shown as curve 3 in Fig. 8. The characteristics of the voltammogram were similarto those with DPA�, except that the peak current density with TPhB� was less than thatwith DPA� even though the concentration of TPhB� was the same as that of DPA�.Voltammograms with dilute TPhB� in W1 in the presence of 0.1–1 M MgSO4, K2SO4, orNa2SO4 were identical with those in the presence of 0.1–1 M MgCl2, KCl, or NaCl in W1and W2.

The voltammograms of Type B were also observed by the addition of dilute(5� 10�7–10�5 M) CVþ or EVþ to one of two aqueous phases containing 0.1 MMgSO4 or MgBr2, respectively. The peak current density depended on the species ofthe anion in the hydrophilic salt, but was practically independent from the cation in thesalt.

The characteristics of the peak currents with DPA�, TPhB�, or CVþ are summar-ized in Table 2.

TABLE 2 Peak Currents in Voltammograms of

Type B Observed with a BLM Composed of

PCþ Ch

Supporting

electrolyte

in W1 and W2

Peak current density (�Acm�2)/Ion added in W1 (10�6 M)

DPA� TPhB� CVþ

0.1 M MgSO4 0.20�0.03 0.10�0.02 0.04�0.011 M MgSO4 0.23�0.03 0.12�0.02 0.05�0.010.1 M K2SO4 0.60�0.10 0.25�0.03 0.04�0.010.1 M Na2SO4 0.22�0.03 0.12�0.02 0.04�0.010.1 M MgBr2 0.21�0.03 0.12�0.02 0.05�0.010.1 M MgCl2 0.20�0.03 0.11�0.02 �00.2 M KCl 0.62�0.10 0.23�0.03 �00.2 M NaCl 0.19�0.03 0.13�0.02 �0


When an extremely hydrophobic ion or very hydrophobic anion was added to bothaqueous phases to be the same concentration, the positive and negative peaks symmetricalabout the origin of each other were larger compared with those observed when the ion wasadded only to W1 (e.g., about 1.6 times when the added ion was 10�6 M DPA� and thehydrophilic salt was 0.1 M MgSO4).

Type C: Voltammograms observed with the cell in the presence of a very hydro-phobic cation in W1 in addition to the hydrophilic salt.

In the voltammogram of Type C, which was obtained by the addition of one of thevery hydrophobic cations (category IV) such as TPenAþ, TBAþ, or TPhAsþ (10�5–10�4 M) to W1 of the system in the presence of a 0.1–1 M hydrophilic salt such asMgSO4, K2SO4, or Na2SO4 in W1 and W2, positive and negative currents of differentmagnitudes appeared at around the origin.

Curve 1 in Fig. 9 is a typical example of Type C observed with 10�4 M TPenAþ inW1 and 0.1 M MgSO4 in both W1 and W2. The magnitudes of the positive and negativepeaks (or limiting currents) were proportional to the concentration of the very hydropho-bic ion added in the range between 10�5 and 10�4 M, while the ratio of the positive tonegative peak (or limiting current) was almost constant. The ratio depended on the kind ofthe added hydrophobic cation. Among cations that belong to category IV in Table 1, themore hydrophobic ion gave the smaller ratio.

The characteristics of peaks or currents of Type C are summarized in Table 3.When the very hydrophobic cation was added to both aqueous phases at the same

concentration, the positive and negative peaks or currents were the same magnitude andsymmetrical about the origin. Curve 2 in Fig. 9 is an example that was recorded by adding10�4 M TPenAþ and 0.1 M MgSO4 to both W1 and W2.

Type D: Voltammogram observed with the cell in the presence of a rather hydro-phobic anion in W1 in addition to the hydrophilic salt.

In the voltammogram of Type D, which was observed by the addition of a ratherhydrophobic anion of category III in fairly high concentrations (such as 10�4–10�3 M) toW1 of the BLM system with W1 and W2 containing a hydrophilic salt, a current decrease,which resembles to current limiting the potential window (the so-called final descent), and

FIG. 9 Voltammograms for ion transfer through a BLM composed of PC and Ch. Curve 1 10�4 MTPenAþ in W1 and 0.1 M MgSO4 in both W1 and W2; curve 2: as curve 1, but in the presence of

10�4 M TPenAþ in W1 and W2.


a small limiting current appear. Curve 1 in Fig. 10 is a typical example observed with10�4 M Pic� in W1 and 0.1 M MgSO4 in both W1 and W2. Current densities at EW1�W2 ¼�0:1 (the final descent) and þ0:1V (the limiting current) depended on both the concen-trations of Pic� and Mg2þ in W1 as summarized in Table 4.

When Pic� was added to both aqueous phases to be the same concentrations, thefinal rise and the final descent (which were symmetrical about the origin) were observed, asshown by curve 2 in Fig. 10.

B. Ion-Transfer Processes Involved in Voltammogram for Ion TransferThrough a BLM

The ion-transport processes involved in the voltammograms through BLMs of Types A toD were elucidated in comparison with those through LMs.

Curves 1 in Figs 11–14 are voltammograms for ion transfers through LMs (VITTM)observed with the LM systems shown in Eqs (6)–(9), respectively, by scanning EW1�W2 andmeasuring IW1�W2.

TABLE 3 Currents at EW1�W2 of þ0:1 and �0:1V, IðþÞ and Ið�Þ,Observed in Voltammograms of Type C with W1 and W2 Containing (1) 0.1

M MgSO4, (2) 0.1 M MgBr2, or (3) 0.1 M MgCl2 as Supporting Electrolyte

(SE)

SE

Current density (�Acm�2Þ=Ion added in W1 (10�5 M)

TPhAsþ TPenAþ TBAþ

IðþÞ Ið�Þ IðþÞ Ið�Þ IðþÞ Ið�Þ

(1) 0.09�0.02 �0.05�0.01 0.04�0.01 �0.02�0.01 0.03�0.01 �0.01�0.005(2) 0.15�0.03 �0.11�0.02 0.06�0.02 �0.03�0.01 0.04�0.01 �0.02�0.01(3) � 0 � 0 � 0 � 0 � 0 � 0

FIG. 10 Voltammograms for ion transfer through a BLM composed of PC and Ch. Curve 1: 10�3

M Pic� in W1 and 0.1 MMgSO4 in both W1 and W2; curve 2: as curve 1, but in the presence of 10�3

M Pic� in W1 and W2.


0:1MMgSO4

ðW1Þ

��10�3 MBDPPE;

MCVþTPhB�

ðLMÞ

��0:1MMgSO4

ðW2Þð6Þ

As Eq. ð6Þþ5� 10�7 M

Mg2þðDPA�Þ2ðW1Þ

��


Mg2þðDPA�Þ2ðLMÞ

��

As Eq. ð6Þ

ðW2Þ

ð7Þ


ðTPenAþÞ2SO2�4

ðW1Þ

��


ðTPenAþÞ2SO2�4

ðLMÞ

��

As Eq. ð6Þ

ðW2Þ

ð8Þ


Mg2þðPic�Þ2ðW1Þ

��


Mg2þðPic�Þ2ðLMÞ

��

As Eq. ð6Þ

ðW2Þ

ð9Þ

The LM consisted of NB to which CVþTPhB� and bis(diphenylphosphinyl)ethane(BDPPE) were added as a supporting electrolyte and a neutral ligand to stabilize Mg2þ

in the LM.Curves 2 and 3 in Figs. 11–14 are voltammograms for ion transfer at the W1/LM

and LM/W2 interfaces, respectively, recorded by monitoring variations of EW1=LM andELM=W2 as a function of IW1�W2 during the measurements of VITTMs. Ion-transfer reac-tions involved in voltammetric waves at the W1/LM and LM/W2 interfaces were assignedto those indicated in Figs. 11–14 by comparing with voltammetric results on the iontransfer at the aqueous/NB interface, which were obtained by changing the concentrationsof ions in W1, LM, or W2.

The shapes of curves 1 in Figs. 11–14 in the potential region between þ0:1 and�0:1V resemble those of voltammograms of Types A–D (curves 1 and 2 in Fig. 8, curves

TABLE 4 Currents at EW1�W2 ¼ þ0:1 and �0:1V Observed in

Voltammograms of Type D by Addition of Pic� to W1 in Presence

of 0.1 or 0.5 M MgSO4 in W1 and W2

Concentration

of Pic� in

W1

Concentration

of MgSO4 in

W1 and W2

Current density (�Acm�2)

EW1�W2 ¼ �0:1V(final descent)

EW1�W2 ¼ þ0:1V(limiting current)

10�4 M 0.1 M �0.25 0.03

10�3 M 0.1 M �1.90 0.40

10�4 M 0.5 M �0.29 0.09


1 in Figs. 9 and 10), respectively, observed with the BLM. Therefore, it may be natural toconsider that the processes of ion transfer through BLMs are analogous to those throughLMs. Ion-transfer reactions responsible for voltammograms of Types A–D were assignedas follows.

Type A: The VITTM of this type was observed when W1 and W2 contained hydro-philic ions. In this case, the hydrophilic ions cannot transfer easily from aqueous phases to

FIG. 11 Voltammograms for ion transfer through an LM (curve 1), at the W1/LM interface (curve

2) and at the LM/W2 interface (curve 3). Compositions of W1, LM, and W2: 0.1 M MgSO4 in W1,

10�3 M BDPPE þ0:1 M CVþTPhB� in LM, 0.1 M MgSO4 in W2.

FIG. 12 Voltammograms are the same as those in Fig. 11, but in the presence of 5� 10�7 M

Mg2þðDPA�Þ2 in W1 and 5� 10�5 MMg2ðDPA�Þ2 in LM.


the BLM. Therefore, the ionic current flow through the BLM is negligible when the appliedmembrane potential is not very large, such as between þ0:1 and �0:1V. In addition, thecontribution of the ohmic drop [IW1�W2R in Eq. (2)] on EW1�W2 should also be taken intoaccount, since the BLM contains little electrolyte and the resistance of the BLM is large.

Type B: The VITTM of this type was observed when one of the extremely hydro-phobic ions or very hydrophobic anions was added to W1. In this case, the hydrophobicion (e.g., DPA�) may easily be accumulated by spontaneous distribution to the hydro-

FIG. 13 Voltammograms are the same as those in Fig. 11, but in the presence of 5� 10�5 MðTPenþÞ2SO2�

4 in W1 and 5� 10�5 M ðTPenAþÞ2SO2�4 in LM.

FIG. 14 Voltammograms are the same as those in Fig.11, but in the presence of 5� 10�5 MðTPenAþÞ2SO2�

4 in W1 and 5� 10�5 M (TPenAþÞ2SO2�4 in LM.


phobic BLM. The distribution should be accompanied by the transfer of the hydrophilicion (e.g., Mg2þ) as the counterion of the hydrophobic ion in order to hold electroneutralityin both W1 and BLM. If we accept this assumption, the BLM system for Type B isanalogous to the LM system of Eq. (7), and hence the transfer reactions involved in theVITTM of this type can be considered to be identical with those for the VITTM in Fig. 12.The positive current peak is due to the transfer of Mg2þ from W1 to BLM and that ofMg2þ from BLM to W2, and the negative current peak is due to the transfer of Mg2þ fromW2 to BLM and that of Mg2þ from BLM to W1.

The ion transfer through a BLM of Type B has been investigated voltammetricallyby several authors [7,9–12]. All of these authors assumed the transfer of the hydrophobicion such as TPhB� or DPA� in their explanations of the current peaks in the VITTM. Onthe other hand, based on the above described analysis of the VITTM, the authors considerthat the hydrophilic ion which has been concentrated in the BLM with the aid of thehydrophobic ion transfers from the BLM to W2 (or W1), and, simultaneously, the sameamount of the hydrophilic ion in W1 (or W2) transfers to the BLM. During the transfer ofthe hydrophilic ion, the hydrophobic ion concentrated in the BLM remains there, behav-ing like a mobile site for the hydrophilic ion. Consequently, the concentration of thehydrophilic ion in the BLM is maintained constant. Adopting the mechanism presentedhere, the characteristics of the positive and negative current peaks in the VITTM such asEW1�W2 where peaks appear, the mutually symmetrical peaks, the extraordinary largecurrent peaks, and the dependence of magnitudes of peaks on the kind or concentrationof the hydrophilic ions in aqueous phases can be easily understood.

Type C: The hydrophobicity of the added ion in W1 for this type is a little weakerthan that for Type B. Hence, addition of the ion at a concentration higher than that in thecase of Type B is required to concentrate the added ion into the BLM to be at a concentra-tion identical with that of the ion in the BLM for Type B. When a proper concentration(e.g., 5� 10�4 M) of the ion (e.g., TPenAþ) is condensed spontaneously in the BLMtogether with the hydrophilic counterion (e.g., SO2�

4 ), the BLM system for Type C isanalogous to the LM system of Eq. (8), and hence the transfer reactions involved in theVITTM of this type are analogous to those in Fig. 13. Therefore, it is concluded that thepositive current peak of the VITTM is composed of the transfer of TPenAþ from W1 toBLM, that of SO2�

4 from BLM toW1, and that of SO2�4 fromW2 to BLM, and the negative

current peak of the VITTM the transfer of SO2�4 from W1 to BLM and that of SO2�

4 fromBLM toW2. Since the current due to the transfer of SO2�

4 from BLM toW1 is equal to thatof SO2�

4 from BLM to W2, the magnitude of the positive current density, I(þ), whichincludes the current due to the transfer of TPenAþ from W1 to BLM in addition to thecurrent due to the transfer of SO2�

4 from BLM to W1, is larger than that of the negativecurrent density, I(�), which is caused by the transfer of SO2�

4 from BLM to W2 alone.When Br�, which is more hydrophobic than SO2�

4 , is used instead of SO2�4 in the

BLM system of Eq. (8), the concentration of Br� distributed to the BLM is expected to belarger than that of SO2�

4 . Hence, the currents due to the transfer of Br� from BLM to W1and that from BLM to W2 may be larger than those observed with SO2�

4 . This considera-tion may explain the result in Table 3 that I(þ) caused by both the transfer of Br� fromBLM to W1 and that of TPenAþ from W1 to BLM and I(�) caused by the transfer of Br�

from BLM to W2 are larger, and the ratio of I(þ) to I(�) is smaller and closer to unitythan those observed with SO2�

4 . Here, the current due to the transfer of TPenAþ isconsidered to be unchanged by the use of Br� instead of SO2�

4 .Type D: The added anion in W1 for this type is not very hydrophobic. Hence, the ion

is not concentrated to the BLM to be at a concentration identical with that of the ion in


the BLM for Type B or C unless a rather high concentration of the ion is added to W1.When a proper concentration (e.g., 5� 10�4 M) of the ion is condensed spontaneously inthe BLM together with the hydrophilic counterion (e.g., Mg2þ), the BLM system for TypeD is analogous to the LM system of Eq. (9). Hence, the transfer reactions involved in theVITTM of this type are considered to be identical with those for the VITTM in Fig. 14.The positive limiting current is due to the transfer of Pic� from BLM to W1, that of Mg2þ

from W1 to BLM, that of Pic� from BLM to W2, and that of Mg2þ from BLM to W2,and the negative current which looks like the final descent is due to the transfer of Pic�

from W1 to BLM, that of Mg2þ from BLM to W1, that of Pic� from BLM to W2, andthat of Mg2þ from W2 to BLM. Here, the negative current is attributable to a part of anegative current peak such as that in curve 3 in Fig. 14, which might be observed ifEW1�W2 could be scanned to more negative potentials than �0:1V. This could not beattained here because of the breakdown of the BLM.

The above-mentioned explanations for the ion-transfer processes of voltammogramsof Types A–D suggest that not only the transfer of hydrophobic ions but also those ofhydrophilic ions must be considered in order to elucidate the ion-transfer processesthrough a BLM.

C. Ion Transfers Through BLMs Composed of Various Lipids

As described in Section IV.B, the peak height of the voltammogram of Type B is con-trolled by the concentration of the hydrophilic ion in a BLM, which has been distributedspontaneously in the BLM as the counterion of the hydrophobic ion. The distribution ofthe added ion with the counterion is determined by the �G0

tr of these ions from theaqueous to the BLM, which may depend on the interaction between the distributed ionsand lipid(s) composing the BLM.

Table 5 summarizes peak current densities observed in voltammograms recordedwith the cell system of Eq. (7) equipped with BLMs of different compositions. In thisinvestigation, a hydrophobic ion such as EVþ, CVþ, TPhAsþ, DPA�, or TPhB� wasadded in 10�6 M concentration to one of two aqueous phases containing 0.1 M MgSO4.The indication of C or D in the table means that the peak of Type B was not observed

TABLE 5 Peak Currents in Voltammograms Observed with

BLMs of Various Compositions. W1 Contained 0.1 M MgSO4 and

10�6 M of Additive, and W2 Contained 0.1 M MgSO4

Lipid

composing

BLM

Peak current density (�A cm�2)/Ion added in W1 (10�6 M)

DPA� TPhB� EVþ CVþ TPhAsþ

DOPC 0.40�0.05 0.10�0.01 0.12�0.03 0.07�0.02 (C)

PS 0.20�0.02 0.10�0.01 0.11�0.02 0.06�0.01 (C)

PE 0.09�0.01 (D) (C) (D) (D)

DOPCþ Ch 0.50�0.10 0.14�0.02 0.09�0.02 0.05�0.01 (C)

PCþ Ch 0.20�0.03 0.10�0.02 0.07�0.02 0.04�0.01 (C)

PSþ Ch 0.09�0.01 (D) 0.25�0.04 0.15�0.02 (C)

PEþ Ch 0.15�0.03 0.06�0.01 (C) (D) (D)

Sphþ Ch 0.18�0.03 0.07�0.01 (C) (D) (D)


when the hydrophobic ion was 10�6 M, but the current of Type C or D appeared when theconcentration of the hydrophobic ion was increased to 10�4 M or more.

The following can be deduced from Table 5:

1. In accordance with the results summarized by Flewelling and Hubbell [33], thepeak current density was larger irrespective of the composition of the BLM when the morehydrophobic one between two cations (CVþ or EVþ) or between two anions (TPhB� orDPA�) was adopted as the additive.

2. The peak current density with TPhB� or DPA� was larger than that withTPhAsþ or CVþ, respectively, although the hydrophobicity of TPhB� or DPA� arealmost identical with that of TPhAsþ or CVþ (cf. �G0

tr in Table 1), respectively, suggestingthat hydrophobic anions are distributed in most of BLMs more easily than hydrophobiccations of hydrophobicities similar to the anions.

3. The peak current density with TPhB� or DPA� increased and that with CVþ orEVþ decreased when a BLM composed of dioleoylphosphatidylcholine (DOPC) and (Ch)was used instead of a BLM composed only of DOPC. The peak current density withTPhB� or DPA� decreased and that with CVþ or EVþ increased when a BLM composedof phosphatidylserine (PS) and Ch was used instead of a BLM composed only of PS.When a BLM of phosphatidylethanolamine (PE) and Ch was used instead of a BLM ofPE alone, peak current densities with all of the hydrophobic ions (TPhB�, DPA�,TPhAsþ, CVþ, or EVþ) increased. In the case of a BLM composed of sphingomyelin(Sph) and Ch, peak current densities for all of the hydrophobic ions (TPhB�, DPA�,TPhAsþ, CVþ or EVþ) were less than those with a BLM of PCþ Ch.

The various BLMs of different lipids have been discussed by several authors[35,36]. A review of these papers shows that they assumed that the positive potentialin the BLM near to the W/BLM interface or in the bulk of the BLM, which is inducedby functional groups oriented at the interface, and the preferable condensation of hydro-phobic anion, could be explained on the basis of the interaction between the hydro-phobic anion distributed in the BLM and the positive potential inside the BLM.However, no attention has been paid to the hydrophilic cation, which moves into theBLM as a counterion of the hydrophobic anion, and hence it is difficult to understandthe dependence of the peak current density on the kind or concentration of the hydro-philic ion in W1 or W2

Distinct from the explanations mentioned above, the results described in SectionIV.A indicate that the distribution of the hydrophilic ion in the BLM as the counterionof the hydrophobic ion is considered to be important to explain the ion-transfer reactionthrough the BLM as well as the condensation of both the hydrophilic and hydrophobicions into the BLM.

In particular, when the ion transfer is of Type B, the current flowing through theBLM is carried mostly by the hydrophilic ion (when Type C or D, the hydrophobic ionparticipates in the transfer in addition to the hydrophilic ion).

Based on the distribution of the hydrophilic ion, the characteristics (1) to (3) ofVITTMs observed with various BLMs can be explained as follows:

1. Since not only the distribution ratio of the hydrophobic ion from W to BLM,but also that of the hydrophilic ion are estimated to be larger when a hydrophobic ion ofsmaller �G0

tr is added to W in the presence of a hydrophilic ion, peaks in VITTM of TypeB, which depend on the concentration of hydrophilic ion in the BLM, are larger when amore hydrophobic ion (with smaller �G0

tr) is added to W1.


2. The short-range interaction (complex formation) between hydrophilic ionssuch as Mg2þ, Naþ, Kþ, SO2�

4 , or Cl� and functional groups, –OH or –C——O, inlipids composing BLMs is expected to be significant from the comparison of �G0

tr ofhydrophilic ions from W to various alcohols with those from W to NB or DCE in theabsence of the functional groups. Since the interaction for hydrophilic cations, whichdistribute in the BLM as the counterions of hydrophobic anions, is greater than thatfor hydrophilic anions, which distribute into BLM with hydrophobic cations, the con-centration of hydrophilic cation (and hence hydrophobic anion) in the BLM is greaterthan that of the hydrophilic anion (and hence hydrophobic cation). This explains thereason why the peak in the VITTM of Type B is larger when a hydrophobic anion isadded to W1 of the BLM system than when a hydrophobic cation is added, eventhough the hydrophobicity of the hydrophobic anion is identical with that of thehydrophobic cation.

3. One reason for the increase of the peak by the coexistence of Ch in a BLM isconsidered to be the stabilization of hydrophilic ions through the short-range interactionwith –OH [cf. (2)] of which concentration in BLM increases with the coexistence of Ch.Another reason may be the structural change of BLM caused by the coexistence of Chwhich may be considerable especially for the BLM composed of PE. The BLM of PE isreported to be highly structured, and hence it requires more energy to form a cavity toimmerse a bulky ion in the BLM than other weakly structured BLMs. This means that thedistribution of a hydrophobic ion and its counterion (hydrophilic ion) in the BLM of PEare smaller than that in other BLMs. When Ch coexists with the BLM of PE, the structureof the BLM is weakened and the distributions of a hydrophobic ion and its counterionbecome larger. Therefore, larger peaks were observed in the VITTM of Type B with theBLM in the presence of Ch. The effect of Ch observed when a hydrophobic cation (CVþ

or EVþ) or a hydrophobic anion (TPhB� or DPA�) was added to the system equippedwith a BLM of DOPC or PS, respectively, cannot be understood by the above-describedexplanation, and requires further consideration. The difference in current density betweena BLM of Sphþ Ch and a BLM of PCþ Ch may be also attributable to the difference inboth binding energies among lipids and short-range interactions.

V. CONCLUSION

The voltammetric concept and method have been demonstrated to be very useful forelucidation of the processes involved in membrane transport. The results introduced inthis chapter suggest that membrane transport is mainly determined by the complementaryion-transfer reactions at two aqueous/membrane interfaces if the membrane as well as twoaqueous phases contains sufficient electrolytes. This fundamental fact is important forunderstanding the influence of the ion-transfer reaction at one interface on that at anotherinterface under a constant membrane potential and for selecting an appropriate conditionfor ion separation by electrolysis with an applied membrane potential. The fundamentalfacts described in this chapter were successfully applied to the interpretation of the poten-tial of liquid membrane-type ion-selective electrodes [37,38], the elucidation of themechanisms of oscillations of the membrane potential and current [35,36,39], and thequantitative understanding of energetics in the coupling of ion transport with electrontransport through a membrane [41–43], although the details have not been introducedhere.


The findings with liquid membranes are also applicable to the elucidation of iontransfer through an extremely thin membrane such as a BLM, as far as the membranesystem contains sufficient electrolytes. As for ion transport through a BLM in the presenceof hydrophobic ions, the ion-transport process is affected largely by the concentration ofthe hydrophilic ion distributed spontaneously in the BLM as the counterion of the hydro-phobic ion added to the aqueous phase, and the concentration is determined by thehydrophobic/hydrophilic natures of both the added ions and counter ions and the char-acteristics of the lipid composing the BLM.

REFERENCES

1. HT Tien. Bilayer Lipid Membranes. New York: Marcel Dekker, 1974, chs 1–5.

2. RB Gennis. Biomembranes. Tokyo: Springer-Verlag, 1990, chs 1, 7–9. (Japanese edition.)

3. EA Liberman, VP Topaly. Biochim Biophys Acta 163:125–136, 1968.

4. YA Liberman, VP Topaly. Biofizika 14:452–461, 1969.

5. OH LeBlanc Jr. Biochim Biophys Acta 193:350–360, 1969.

6. R de Levie, NG Seidah. J Membr Biol 16:1–16, 1974.

7. CJ Bender. Chem Soc Rev 17:317–346, 1988.

8. R de Levie. J Electroanal Chem 69:265–297, 1976.

9. J Kutnik, HT Tien. Bioelectrochem Bioenerg 16:435–447, 1986.

10. CJ Bender, HT Tien. Anal Chim Acta 201:51–58, 1987.

11. CJ Bender, HT Tien. Anal Chim Acta 198:259–269, 1987.

12. HT Tien. Prog Surface Sci 19:169–274, 1985.

13. J Koryta. Ions, Electrodes, and Membranes. New York: John Wiley, 1982, pp 135–193.

14. HH Girault, DJ Schiffrin. In: AJ Bard, ed. Electroanalytical Chemistry. vol. 15. New York:


15. HH Girault. In: JO’M Bockris, BE Conway, RE White, eds. Modern Aspects of

Electrochemistry. vol 25. New York: Plenum Press, 1993, pp 1–58.

16. O Shirai, S Kihara, M Suzuki, K Ogura, M Matsui. Anal Sci 1991;7(suppl): 607–610.

17. O Shirai, S Kihara, Y Yoshida, M Matsui. J Electroanal Chem 389:61–70, 1995.

18. O Shirai, S Kihara, Y Yoshida, K Maeda, M Matsui. Bull Cehm 389:61–70, 1995.

19. GE Goldman. J Gen Physiol 25:37–60, 1943.

20. AL Hodgkin, B Katz. J Physiol 108:37-77, 1949.

21. P Henderson. Z Phys Chem 59:118–227, 1907.

22. N. Laksminarayanaiah. Equations of Membran Biophys. Orlando, FL: Academic Press, 1984,

pp 1–367.

23. V Horvath, G Horvai. Anal Chim Acta 273:145–152, 1993.

24. Z Samec, A Trojanek, J Langmaier, E Samcova. J Electroanal Chem 481:1–6, 2000.

25. MH Barker, L Murtomaki and K Kontturi. J Electroanal Chem 497:61–68, 2001.

26. T Kakiuchi. Electrochim Acta 44:171–179, 1998.

27. C Beriet, HH Girault. J Electroanal Chem 444:219–229, 1998.

28. VK Miyamoto, TE Thompson. J Colloid Interface Sci 25:16–25, 1967.

29. LJ Bruner. Biophysik 6: 241–256, 1970.

30. B Ketterer, B Neumcke, P Lauger. J Membr Biol 5:225–245, 1971.

31. A Parsegian. Nature 221:844–846, 1969.

32. G Szabo. Nature 252:47–49, 1974.

33. RF Flewelling, WL Hubbell. Biophys J 49: 541–552, 1986.

34. P Lauger, R Benz, G Stark, E Bamberg, PC Jordan, A Fahr, W Brock. Q Rev Biophys

14:513–598, 1981.

35. R Larter. Chem Rev 90:355–381, 1990.

36. S Kihara, K Maeda. Prog Surface Sci 47:1–54, 1994.


37. S Kihara, Z Yoshida. Talanta 31:789–797, 1984.

38. Y Yoshida, M Matsui, K Maeda, S Kihara. Anal Chim Acta 374:269–281, 1998.

39. K Maeda, S Kihara, M Suzuki, M Matsui. J Electroanal Chem 295:183–201, 1990.

40. K Maeda, M Nishihara, H Ohde, S Kihara. Anal Sci 14:85–88, 1998.

41. H Ohde, K Maeda, O Shirai, Y Yoshida, K Kihara. J Electroanal Chem 438:139–145, 1997.

42. H Ohde, K Maeda, Y Yoshida, S Kihara. Electrochim Acta 44:23–28, 1998.

43. H Ohde, K Maeda, Y Yoshida, S Kihara. J Electroanal Chem 483:108–116, 2000.

44. Z Samec, P Papoff. Anal Chem 62:1010–1015, 1990.


21Development of Structurally OrganizedPhotocatalytic Systems for PhotocatalyticHydrogen Evolution on the Basis of LipidVesicles with Semiconductor NanoparticlesFixed on Lipid Membranes

OXANA V. VASSILTSOVA and VALENTIN N. PARMON Boreskov Instituteof Catalysis, Novosibirsk, Russia

I. INTRODUCTION

Photocatalytic systems based on lipid vesicle suspensions are a vivid example of a possi-bility for designing artificially highly organized catalytic systems. An initial concern wasdictated by the desire to create synthetic systems, which model the function of a naturalplant’s photosynthesis on the basis of much more simple processes that entail a directtransformation of solar energy into the energy of chemical bonds. The most attractive goalfor such a transformation are the products of water splitting—hydrogen and oxygen.Indispensable components of respective photocatalytic systems are a photosensitizer(photocatalyst) for the process, reversible donors and acceptors of electrons, and catalystsof ‘‘dark’’ processes of formation of the target products (see Fig. 1). The photocatalyst(PhC) is a substance that is able to produce, after absorbing the light quanta, somechemical transformations of the reaction participants, repeatedly coming with them intointermediate chemical interactions and regenerating their chemical composition after eachcycle of such interactions [1].

Use of a microheterogeneous medium allows separation of spatially primary oxidi-zers and reducing agents, which can be generated at the first stage of the photoseparationof charges. As a result, it prevents the recombination of these primary energy-saturatedsubstances via their direct collisions. Indeed, such stabilization of primary redox-activereactants promotes the accomplishment of the target chemical process with the accumula-tion of light energy.

A cornerstone for the development of such microheterogeneous systems is the choiceof an optimal photosensitizer for the process. At present, a great many experimental datahave been obtained on the properties of ordered microheterogeneous photocatalytic sys-tems operating with molecular photosensitizers such as complexes of various metals [2,3].However, some important photochemical properties of the designed systems (e.g., thequantum yields of the charge photoseparation or the rates of the vectorial transmembraneelectron transfer) have not allowed, up to now, the design of composite photocatalytic


systems for water splitting in which the operation of the PhC would be conjugated with theoperation of two catalysts of the ‘‘dark’’ processes of water reduction to H2 and itsoxidation to O2 (Fig. 1).

Semiconductor nanoparticles, i.e., particles of a characteristic size of a few nan-ometers, which retain most basic physicochemical properties of bulk semiconductors[4,5], have been suggested as photosensitizers in structured microheterogeneous media.The expected advantages of such photosensitizers, as compared to the most widely usedmetallocomplexes (e.g., polypyridyl complexes of ruthenium(II), porphyrin, and othertetrapyrrole complexes [2,3], in the development of photocatalytic systems are evident.First, there is the well-known possibility of a broad variation of redox and optical proper-ties of the semiconductor particles. Second, it is possible to form semiconductor hetero-junctions (see Section III.C and Fig. 2) as well as to modify and activate catalytically theirsurfaces. Note that, in the last few years, semiconductor nanoparticles have become asubject of undivided concern mostly because of the development of methods of molecularand electronics engineering.

Dispersed semiconductor PhCs can be easily heterogenized on a polymeric or cera-mic support to create photocatalytic installations of practical interest for hydrogen gen-eration in sunlight illumination. Tests on pilot devices of � 0:25m2 area, based on thistype of system, which are capable of producing a few liters of hydrogen on a clear sunnyday, have been reported [6]. Undoubtedly, the expected serious demand for such systemscould create an impetus to develop the respective technologies to the level of their com-mercial application.

The research on dispersed semiconductors for splitting of water into hydrogen andoxygen anticipates a breakthrough in constructing a spatially well-organized system on thebasis of polymeric or some other dielectric membranes incorporating semiconductor par-ticles which penetrate the membrane (Fig. 3). By making an asymmetry in the membrane-separated solutions (e.g., by modifying asymmetrically the semiconductor particles or byvarying the composition of the solutions) one could contract a very efficient and reliable

FIG. 1 Simplified energy diagram of electron phototransfer in photocatalytic systems for water

splitting on the scale of reduction potentials Ee.


hydrogen-producing device. Until now, experiments in this field were restricted mostly tothe production of hydrogen from HBr or other electron-donating substrates [4,7].However, no principal restrictions are foreseen in improving these systems in order tosatisfy thermodynamic requirements for water cleavage.

A more elegant way to combine the advantages of PhCs on the basis of dispersedsemiconductors with those of membrane-structured systems seems to be the inclusion ofsemiconductor nanoparticles into microscopic vesicular systems with bilayer lipid mem-branes (vesicles are the microscopic bubbles, see Section II and Fig. 4). It is anticipatedthat semiconductor nanoparticles in such systems can serve the role of very efficient andstable integral photoreaction centers mimicking completely the spatially well-organized

FIG. 2 Energy diagram and scheme of photoseparation of charges and catalytic processes in a

suspended particle of a semiconductor with microheterojunction CuSx=ZnyCd1�yS, which is up to

now one of the most efficient photocatalysts for H2S cleavage in water solutions. Ee is the

electrochemical potential of the electron, Eg1 and Eg2 are the widths of forbidden gaps of the

semiconductor phases, and h�1 and h�2 are the quanta of exciting light corresponding to them.

(From Refs. 26 and 27).

FIG. 3 Schematic view of the macroscopic ‘‘monograin membrane’’ suggested in Ref. 4 as an

efficient tool for photocatalytic production of hydrogen at the expense of oxidation of some

electron-donating compounds. The thickness of the polymeric membranes in the experiments was

a few micrometers.


FIG. 4 Schematic view of hypothetical biomimetic devices for photocatalytic charge separation by

using semiconductor nanoparticles penetrating the bilayer membrane of a lipid vesicle. Such systems

combine the ideas of systems shown in Figs. 2 and 3 and are the main target of elaboration. (a)

Conventional version with a simple nanoparticle of CdS, CdSe, etc.; (b) Improved version with a

microheterojunction between two semiconductor nanoparticles of different nature.

reaction centers on the basis of chlorophyll molecules in photosensitizing organisms. Thefirst results on such systems confirm their potential [4,5,8–10], so that serious research isnow obviously essential. The main problem still to be overcome in designing such systemsis the elaboration of ways to imbed highly hydrophilic inorganic nanoparticles into thebody of the hydrophobic lipid bilayer. However, the greatest challenge will be the long-term stability of photocatalyzing substrates, since it is well known that most polymerlayers are degraded in intense sunlight.

The hydrogen-generation photoactivity of vesicle-stabilized and catalyst-coated col-loidal CdS was first demonstrated for dihexadecyl phosphate (DHP) vesicles with Rh asthe catalyst and thiophenol (PhSH) as a sacrificial electron donor [see Fig. 5(a)] [4].Although CdS could be located selectively at the inner or outer surfaces of the vesicles,the symmetrically organized systems were found to be the easiest to prepare most repro-ducibly. No significant effect of the CdS location on the photochemical activity for the H2

generation was observed.Positively charged vesicles prepared from dioctadecyldimethylammonium chloride

(DODAC) ½C15H31CO2ðCH2Þ2NþðCH3ÞðCH2C6H4CH——CH2Cl� [Fig. 5(b)], as well as

from its polymerized counterpart, were found to be better media for CdS-mediatedwater photoreduction than those prepared from DHP [Fig. 5(c)] [11,12]. Unfortunately,the quantum efficiency of the hydrogen formation was lower than 0.5% and the accumu-lated oxidized electron donor diphenyl disulfide PhSSPh led rapidly to the ultimate destruc-tion of the vesicles.

The optimization of the photosensitized hydrogen production in the surfactant vesi-cle-stabilized and catalyst-coated CdS nanoparticles was performed in several ways. First,via the incorporation of a recyclable electron donor as an integral part of the vesicularsystem, e.g., ðn-C18H37Þ2NþðCH3ÞCH2CH2SH Br� (RSH) [Fig. 5(d)] [11]. The band-gapexcitation of Rh-coated CdS, attached to the outer membrane of the DODAC–RSH vesi-cles, resulted in hydrogen formation at the expense of RSH oxidation to a disulfide RSSR.Reduction of the disulfide by NaBH4 led to the regeneration of the electron donor RSHthat could be subsequently reused for additional colloidal semiconductor photosensitizedwater splitting (thus making possible a cyclic process of hydrogen generation) [13]. Second,through the stabilization of vesicles via their polimerization [12]. Third, via the utilizationof more efficient sacrificial electron donors, e.g., benzyl alcohol [12]. A dramatic 40-foldenhancement in the hydrogen photogeneration rate was achieved by combining benzyl


FIG. 5 Idealized models for CdS-sensitized photoreduction of water in aqueous vesicle

suspensions. The vesicles are from: (a) DHP; (b) DODAC or DODAB; (c) polymerized n-

C15H31CO2ðCH2Þ2NþðCH3Þ½CH2C6H4CH ¼ CH2�Cl�; (d) DODAC and thiol-functionalized

surfactant. (From Refs. 4 and 11–13.)


alcohol at 1% concentration with cationic vesicles [12] compared to the previously reportedCdS/DHP/PhSH [14]. The overall quantum yield of hydrogen generation obtained in thelast two cases was � 10%. The maximum reported yield of hydrogen in the vesicle suspen-sions is 5.5 L of hydrogen per hour and per gram of the CdS/Rh catalyst [12].

Thus, although a lot of work has been done in this field, some problems still remain.For example, the quantum yield of the hydrogen generation remains to be rather low for acommercial utilization.

The purpose of this paper is to consider some experience in the field of engineering ofcomposite structurally organized microheterogeneous systems based on lipid vesicles withsemiconductor nanoparticles as photocatalysts.

II. SPATIAL ORGANIZATION OF A PHOTOSYSTEM FOR SOLAR ENERGYCONVERSION BASE ON LIPID VESICLES

Lipid vesicles, or liposomes, are spherical, self-closed structures composed of curved lipidbilayers, which entrap part of the solvent, in which they freely float, in their interior. Theymay consist of one or several concentric membranes; their sizes range from 20 nm toseveral dozens micrometers, while the thickness of the membrane is around 4 nm [15].The vesicles represent elementary entities with the developed bilayer membrane structureand are an appropriate microheterogeneous system for making ‘‘microreactors’’ in solarenergy conversion [2,3]. In fact, the vesicle systems have at least three areas accessible tothe arrangement in them of redox-active components of a sophisticated photosystem: (1)confining bilayer of the membrane as well as (2) outer and (3) inner aqueous phases. Itallows spatial organization of, e.g., the photocatalytic process of water photosplitting.

An example of a possible system for photocatalytic water decomposition is shown inFig. 6. The photocatalyst in Fig. 6 is a CdS nanoparticle, which is located, e.g., in the inneraqueous phase. A sacrificial electron, donor (D) is also located in the inner phase. In thepresence of a suitable water oxidation catalyst, the role of the donor could be served by themolecules of water. The molecular carriers of electrons (C in the figure) are built into thelipid membrane by the principle of a ‘‘cascade,’’ providing a certain gradient of redoxpotentials. In the outer aqueous phase, an electron acceptor and a catalytic agent of waterreduction to hydrogen are placed. Thus, at light quantum absorption by the semiconduc-tor PhC, the charge separation derives an ‘‘electron hole’’ which passes to the catalyst of

FIG. 6 Scheme of a photocatalytic system for direct electron phototransfer photosensitized by CdS.

The final acceptor of electrons is water, C and A are reversible electron carriers, D is an irreversible

donor of electrons, and ktH2 is a catalyst of hydrogen evolution.


the O2 evolution and oxidizes there water to oxygen while the generated electron escapesvia the system of electron carriers through the membrane to the catalyst of the H2 evolu-tion. The topology of such a system can be inverted, then the electron transfer will becarried out from the outer aqueous phase to the inner one. Some other modifications ofsystems of this type are also possible.

A. Lipid Vesicles and Spatial Organization of Redox- andPhotochemically Active Species of a Photocatalytic System in aVesicle

The methods of preparing lipid vesicles are well known [2,5,9]. Vesicles are made pre-dominantly from amphiphiles, a special class of surface-active molecules, which arecharacterized by having a hydrophilic (water soluble) and a hydrophobic (water inso-luble) group on the same molecule. A typical vesicle-forming molecule, such as lecithin(see Fig. 7), has two hydrocarbon chains, also called hydrophobic or nonpolar tails,attached to a hydrophilic group, often named the polar head. In general, most of thesemolecules are not soluble in water; however, instead of solutions they form colloidaldispersions [15].

At vesicle formation, lipid molecules are self-organized in such a way that theirhydrophilic polar moieties appear to be directed to an aqueous phase, while two longnonpolar hydrocarbon chains appear to be deeply immersed in the bilayer membrane. Itallows direct placement of the electron-transfer chain components into the bilayer by useof the hydrophilic–lipophilic properties of the molecular electron relays and as well as oftheir size.

Note that clarification of the spatial localization of the electron-transfer chain com-ponents inside the artificial bilayer membranes is of a key value for the development ofbiomimetic systems modeling natural photosynthesis. The direct methods of identificationof localization specificity of the functional molecules are usually quite laborious. For thisreason, in practice, in this particular research, some studies commonly make use of certainanalogs of molecular electron relays or of special molecules such as, e.g., paramagneticspin labeled ones [2,5,6].

Figure 8 presents the results of modeling of a simplest lipid bilayer by the easilyaccessible method of molecular mechanics. The method also enables one to imitate theprobable location in the bilayer of the molecule of some electron relays. Figure 7 demon-strates typical results of such imitation for a widely used lipophilic electron relay likecetylviologen bication (C16V

2þ). It is clearly seen that the positively charged bipyridylmoiety has to locate near the surface of the bilayer. Thus, this relay appears to be capableof diffusing through the membrane only in an uncharged doubly reduced state, as wassupposed in several papers [17–19], but not demonstrated by direct numerical modeling.

Note that, despite numerous studies on the mechanisms of the redox processes andtransmembrane electron transfer with participation of cetylviologen in these systems [17–19], the direct experimental evidence on the localization of C16V

2þ or C16Vþ in the bilayer

of lipid vesicles until now was not obtained. The first direct information on the localizationof paramagnetic electron relays in the bilayer lipid membranes has been recently obtainedby NMR spectroscopy via studies on the paramagnetic broadening of the 1H-NMR linesof a lipid [20]. This method has been applied to elucidation of the localization of the one-electron reduced state of cetylviologen, C16V

þ.Indeed, dl-dipalmitoyl-�-lecithin (DPL), being homogeneously dissolved in organic

solvents of low viscosity, exhibits in its 1H-NMR spectrum a series of narrow multiple


FIG. 7 Structural formulas of some lipophilic compounds mentioned in this chapter.

FIG. 8 Probable spatial localization of C16V2þ and H4SiW12O40 in the DPL lipid bilayer obtained

from molecular mechanics simulations. Carbon and oxygen atoms are gray; hydrogen atoms are

black.


peaks, with their integrate intensities reflecting a number of equivalent protons in themolecule. It is remarkable that the protons of the methylene groups at C4–C15 contributeto one common peak at 1.25 ppm with the integrate intensity of 48H.

The 1H-NMR spectrum of DPL in D2O suspended vesicles differs dramatically: all1H-NMR lines of DPL are strongly and unequally broadened, thus their multiplet struc-ture is not observed any more. Besides, the lines of the �NþðCH3Þ3 and �CH2 �Nþ

groups move to the strong field at 0.16 ppm, and the apparent integrate intensity of the1H-NMR signals of DPL does not reflect the number of the equivalent protons. The latterseems to be related to the coexistence in the lipid membrane of domains of the gel andliquid crystalline phases [21]. In this case the protons in the gel crystalline phase domainsare practically NMR inactive, giving very broad signals perceived in the 1H-NMR spec-trum as a reference background line. The fraction of the gel crystalline phase is enlargedwith the temperature rise in a broad temperature range above and below the point of thelipid phase transition (41:5�C for DPL) [21]. Thus, the �NþðCH3Þ3 groups have highmolecular mobility and exhibit a 1H-NMR line of integrate intensity equal to 9H evenat 20�C. The integrate intensities of the other 1H-NMR lines of DPL, especially those ofthe lines of the NMR inactive methylene groups of the aliphatic ‘‘tails,’’ grow with thetemperature rise (most sharply near to the phase transition point) and achieve their the-oretical values (48H for methylene groups at C4–C15) only at temperatures 10�–20�Cabove the phase transition point. The widths of all lines decrease gradually with increasein temperature.

Insertion of 1 mol% of the C16V2þ in the vesicular membranes does not noticeably

influence the 1H-NMR spectrum of DPL, and the signals from the C16V2þ protons are not

observed due to their low concentration.At the same time, the formation of paramagnetic cation radicals C16V

þ with achemical reduction of C16V

2þ, which are also bound with the membrane, change notice-ably the 1H-NMR spectrum of DPL. It was found that paramagnetic C16V

þ influencemostly the resonance line of protons of the �NþðCH3Þ3 group; the linewidth of this line atcetylviologen reduction is essentially enlarged. The other lines in the field of 2–4 ppm,belonging to protons of the other DPL groups, which are localized in the vicinity of thesurface of the membrane, also experience a small broadening. At the same time, the 1H-NMR lines, which belong to protons of the �CH3 and �CH2� groups of the aliphatic‘‘tails,’’ are not subjected to a noticeable broadening at cetylviologen reduction. Theintegrate intensities of all 1H-NMR signals of DPL remain without considerable changesat reduction.

Thus, the selective broadening of the lines of the protons localized at the surface ofthe membrane, testifies unambiguously to the localization of the paramagnetic moieties ofthe C16V

þ cation radicals near the surface of the lipid bilayer (Fig. 8).One can easily show that the observable influence of the C16V

þ cation radicals onthe 1H-NMR spectra of the DPL molecules which comprise the vesicular bilayers corre-spond to the model of ‘‘fast’’ (in the scale of the relaxation times of the spins of the DPLprotons in the vesicular membrane, which are � 10�6 s) movement of the DPL moleculesin the co-ordination sphere of the C16V

þ cation radical.Thus, the detection of the direct and selective influence of the paramagnetic lipo-

philic electron relay on the 1H-NMR spectra of the surrounding lipid molecules, whichconstitute the bilayer lipid membranes, appears to be direct evidence for the localization ofthe paramagnetic moiety of lipophilic cetylviologen cation radicals just near the surface ofthe vesicle membrane. Evidently, this coincides with the above data from molecularmechanics calculations.


III. CADMIUM SULFIDE NANOPARTICLES IN SUSPENSIONS OF LIPIDVESICLES, THEIR PROPERTIES, AND METHODS OF PREPARATION

The photosensitizing properties of CdS nanoparticles (i.e., the particles with a character-istic size of a few nanometers) depend dramatically on their size: the less the size of theparticles, the larger is the probability of the photogeneration outcome of the primarycharge carriers and, thus, the higher has to be the quantum yield of the process to besensitized. Accordingly, the main goal of the experiments described below was to developmethods to control the CdS nanoparticle size in the inner water-containing cavities of thelipid vesicles.

A. Size of CdS Nanoparticles in Inner Cavities of Vesicles

In our experiments, we formed the CdS nanoparticles by adding an aqueous Na2S solutionto a suspension of the DPL vesicles, which contained in the inner cavities various pre-cursors of CdS (see Fig. 9) [5,9,10]. This procedure allowed control of the growth of theCdS nanoparticles in the inner cavities of the lipid vesicles.

We used the potassium salt of the cadmium complex with ethylenediaminetetra-acetate ½CdEDTA�2� as the precursor of the CdS nanoparticles in the inner cavities of

FIG. 9 Schematic view of photocatalytic systems designed for CdS/lipid membrane interface

electron transfer, which is photosensitized by CdS nanoparticles attached to (a) the internal, (b)

outer, or (c) both inner and outer membrane surfaces.


the vesicles. The EDTA anion is also a well known electron donor in systems for vectorialelectrons’ phototransfer through the lipid vesicle membrane.

The absorption spectra of the CdS particles in the process of their growth in theinner cavities of the vesicles are presented in Fig. 10. The CdS precursors were: (a)K2½CdEDTA�, (b) CdCl2, and (c) CdðNO3Þ2. One can see from Fig. 10 that both theshape of the absorption spectrum of the CdS nanoparticles formed and its changes withtime depend strongly on the nature of the precursor.

In the case (a) the position of the CdS absorption edge suggests [22] the primaryformation of ‘‘large’’ particles (spherically shaped particles have diameters of � 7:4 nm);an individual peak at ¼ 360 nm s upposes the formation of smaller particles of diameter

FIG. 10 Evolution of the absorption spectra of CdS nanoparticles in the process of their growth

after addition of 0.5 M Na2S to the suspension of pure DPL vesicles, which contained initially (a) 0.3

M K2[CdEDTA], (b) CdCl2, or (c) CdðNO3Þ2 in the internal cavities and borate buffer wof pH 8.0 in

the outer solution. The spectra were recorded at (a) 150 s, (b) 100 s, and (c) 60 s intervals at room

temperature; the arrows show sequences in the changes. The absorption spectra are corrected for

light scattering by the vesicle suspension.


2.1 nm. The formation of two types of CdS particle was proved by transmission electronmicroscopy and from the luminescence spectra [10]. It is interesting to point out that, inthe process of CdS particle growth from K2½CdEDTA�, there is practically no shift ineither the absorption edge or the noted peak with time. It characterizes persistence of thesize of both small and large particles during their formation. Besides, the degree of con-version of cadmium ions into CdS in the vesicle suspensions with CdS in the inner cavitieswhen using [CdEDTA�2� as the precursor is several times less than that with CdCl2 andCdðNO3Þ2.

When the EDTA anions are removed from the system, and simple salts, CdCl2 orCdðNO3Þ2, are used as the precursors, the position of the CdS absorption edge during theparticle growth shifts to longer wavelengths with time, i.e., it behaves ‘‘normally,’’ asagainst the system containing EDTA (Fig. 10). In reality, such a shift evidences a con-secutive increase in the size of the primarily formed CdS particles in the process of theirgrowth. So, for CdCl2, growth is observed from 3.3 nm up to 4.8 nm, and, for CdðNO3Þ2,from 4.3 nm up to 6.3 nm.

The CdS nanoparticles formed in the absence of EDTA have the following averagediameter: 4.8 nm for CdCl2 and 6.3 nm for CdðNO3Þ2, while the diameter is 7.5 nm forK2½CdEDTA�.

It was shown [23] that addition of ligands that form complexes with Cd2þ influencesessentially the size of the CdS nanocolloids. Namely, an increase in the stability of acadmium complex precursor reduces the equilibrium size of the colloidal particles. TheEDTA anions appear to be an exception to this rule because they form strongly chelatedCd2þ complexes with a stability constant of 5� 1016 M�1 [24]; the presence of this liganddissolves CdS particles of a size less than a certain diameter of the CdS particles in ahomogeneous colloidal solution. Data on the CdS synthesis in the inner cavities of thelipid vesicles are in a good agreement with the results of the cited work.

Also, the final CdS nanoparticle size is influenced by the pH of the ‘‘outer’’ solution(the increase in pH decreases the size of the CdS particles formed) as well as by modifica-tion of the membrane with ionogenic surfactants such as SDS and CTAB (see Fig. 7). Thesize of the particles formed is influenced as well by the concentration of the CdS precursorin the inner cavities of the vesicles; the increase in concentration enlarges the final nano-particle size (in the case of the surfactant-unmodified membrane). At the same time, forthe CTAB-modified membrane, the relationship is reversed.

B. Growth Rate of CdS Nanoparticles in Inner Cavities of Lipid Vesicles

The kinetics of the CdS nanoparticles’ growth in the inner cavities of the vesicles dependson the nature of the CdS precursor (for K2½CdEDTA� the initial growth rate is two ordersless than for the two other cited precursors) as well as on the presence of modifying agentsin the membrane (i.e., on the permeability of the membrane) and on the pH of the outersolution. However, the kinetics has been found to be independent of the concentration ofthe precursor. Thus, the rate-controlling step in the growth of nanoparticles in suspensionsof the vesicles seems to be neither diffusion of the uncharged H2S molecules through themembrane, nor the diffusion collisions of the H2S molecules with the vesicles [10].

Apparently, the nature of the rate-controlling step in CdS formation in the innercavities of the vesicles is quite complicated and, probably, includes the preliminary for-mation of complexes of the S2� anions with either some molecules in the lipid membraneor the CdS precursors (because the growth rate of the nanoparticles depends on the natureof the CdS precursors).


Besides the synthesis of the CdS nanoparticles in the inner cavities of the vesicles, anapproach has been found for the synthesis of the CdS nanoparticles, which are attached tothe outer surface of the lipid membrane. This can be done by modification of the mem-brane by ionogenic surfactants such as SDS and CTAB.

Thus, the first approach to CdS nanoparticle synthesis in both the outer and innersurfaces of the lipid vesicles seems to have been found. The main factors that control thesizes of the CdS nanoparticles in the inner cavities of the lipid vesicles were determined andsome attempts to determine the nature of the rate-controlling step in the process of CdSparticle growth in these cavities were also undertaken.

C. Formation of CuxS Nanoparticles and CdS/CuxS Heterostructures inSystems Containing Lipid Vesicles

A contact between two semiconductor phases, e.g., a contact between CdS and CuxS,leads to the formation of heterojunctions (Fig. 2) [25]. The presence of the heterojunctionincreases considerably the quantum efficiency of the semiconductor-based PhC in com-parison with the separately taken semiconductor components. First, the presence of anelectric field in the area of the heterojunction improves considerably the conditions ofspatial separation of the photogenerated holes and electrons and prevents their recombi-nation. Second, the threshold wavelength for the PhC operation is shifted in the presenceof the heterojunction to longer wavelengths since the photogenerated reaction can beinitiated by light quanta absorbed not only by the wide-gap ones (semiconductor CdS)but also by the narrow-gap ones (CuxS) [25].

Therefore, of great importance seems to be the creation of CdS=CuxS heterojunc-tions, which penetrate the lipid membrane and function simultaneously as both photo-sensitizers of a photocatalytic process and as electron carriers through the vesiclemembrane. An intermediate step in the development of approaches to the formation ofsuch heterojunctions is the synthesis of CuxS nanoparticles in the outer or inner surface ofthe membrane.

Unfortunately, it was found that the presence of lipophilic bications, C16V2þ (which

are used as common hydrophobic molecular electron relays), in the membrane leads to thereduction of these bications even during the growth of the CuxS particles as well as ofCdS=CuxS heterostructures (obtained by consecutive sedimentation of CdS and CuxS onthe outside surface of the membrane) even in the absence of light. The reduction of C16V

2þ

seems to happen because of the oxidation of the S2� anions, which is known to becatalyzed by copper ions. Thus, the creation of a system for transmembrane electrontransfer with the participation of CdS=CuxS microheterojunctions seems to necessitate asearch for new electron donors and electron carriers. It would appear that systematicstudies in this direction are indispensable.

IV. PHOTOCHEMICAL PROPERTIES OF SYSTEMS BASED ON LIPIDVESICLES WITH CDS NANOPARTICLES

A final goal in the development of systems based on lipid vesicles with CdS nanoparticlesis to construct photochemically active systems with CdS as the photosensitizer. Of greatimportance also is determination of the influence of both the size of the synthesized CdSnanoparticles and their localization in the vesicles on the quantum yield of the interfacial


electron transfer. These studies were carried out with the use of lipophilic C16V2þ as the

primary electron acceptor [10]. The efficiency of the electron phototransfer was determinedvia measurement of the initial quantum yield of the viologen photoreduction. The testedcompositions of the described systems are presented in Fig. 9.

The quantum yield was found to be strongly dependent on the nature of the CdSprecursor. So, in the presence of C16V

2þ in the membrane (molar C16V2þ-to-lipid ratio in

the membrane was 0.01) and of CdS in the inner cavities of the vesicles, the measuredquantum yields for different precursors was as follows: 2.4% for K2½CdEDTA], 0.9% forCdCl2, and 0.7% for CdðNO3Þ2. The EDTA anions serve here, possibly, as additional andefficient electron donors, apart from S2�.

The quantum yield depends also on the C16V2þ concentration in the membrane. One

can assume that, in the case of a low mobility of both the CdS nanoparticles and theC16V

2þ molecules in the membrane, the increase in C16V2þconcentration in the membrane

has to result in an increase in the apparent quantum yield of C16Vþ because of the larger

number of the electron acceptor molecules in the ‘‘sphere of influence’’ of the CdS nano-particle.

It was found that the localization of CdS (the precursor CdCl2) on the inner surfaceof the lipid membrane provides a quantum yield lower than that provided by localizationon the outside surface of the membrane (compare �in ¼ 0:9% and �out ¼ 3:2% accord-ingly) when other conditions are equal. One can expect that this is because of a betteraccessibility of the electron donor, i.e., the S2� anion, to the CdS nanoparticles in the outeraqueous phase.

For a system where the CdS particles are localized both on the outer (the precursorCdCl2) and on the inner surface of the membrane (the precursor K2½CdEDTA�Þ, theapparent quantum yield was 3.2%.

It is essential that in a system with CdS on the inner surface of the membrane thedegree of C16V

2þ reduction constitutes only 24% of its maximum possible value in asystem with CdS on the outer surface of the membrane. On localizing the CdS nanopar-ticles simultaneously on both the outer and inner surfaces of the membrane, the degree ofC16V

2þ stationary reduction was equal to 45–49%. Therefore, one can suppose that, underthe particular experimental conditions, only 45–49% is the maximum possible degree ofreduction of the C16V

2þ cations, because of the considerable influence of disproportiona-tion of lipophilic viologen radicals (e.g., for cetylmethylviologen, C16MV2þ, the dispro-portionation process 2C16MVþ $ C16MV0 þ C16MV2þ has been reported) [19]. Thedisproportionation results in the appearance of uncharged molecules C16V

0, whichmigrate rapidly through the membrane.

Thus, when the CdS particles localize on the outer surface of the vesicle membranes,the quantum yield of the C16V

2þ photoreduction appears to be much larger than at thelocalization of the CdS nanoparticles in the inner cavities of the vesicles, due to a greateraccessibility of the CdS nanoparticles for the redox-active reagents. It was found that thenature of the CdS precursor also influences the quantum yield of the viologens reduction.For example, the presence of the EDTA anions, which serve as additional electron donors,enlarge the quantum yield. Besides, the quantum yield increases with increase in theC16V

2þ concentration inside the membrane.Note that the maximum value 3.3% of the registered quantum yield of the lipophilic

viologen photoreduction is evidently not sufficient for the design of an efficient electron-transfer chain of a photocatalytic system. So, there is a necessity for a further improve-ment of systems based on lipid vesicles sensitized with semiconductor nanoparticles.Possible approaches to this improvement seem to be, e.g., the creation of semiconductor


heterojunctions penetrating the membrane as well as a search for lipophilic electron relaysmore efficient than lipophilic viologens.

A. Menaquinone as Reversible Electron Carrier Through LipidMembranes

As a new type of electron relay, which is able to penetrate lipid membranes, we testedmenaquinone (MQ, Fig. 7). Compounds of this type were not utilized earlier for artificialvesicle-based systems. However, these mimick the functioning of the Z-scheme of naturalplant photosynthesis (see Figs 9 and 12). Indeed, the activity of MQ in the redox processesin a lipid bilayer membrane was revealed. However, the quantum yield of the transmem-brane electron transfer from a CdS nanoparticle in the inner cavity to a CdS nanoparticleon the outer membrane surface with the participation of MQ appeared to be very low anddid not exceed 0.2–0.4%.

B. Heteropolyacid Anions as Reversible Electron Carriers in LipidVesicle Suspensions

Note that a common approach in the creation of photocatalytic systems based on lipidvesicles utilized only organic lipophilic compounds as reversible electron relays.

In order to increase the quantum yield of electron phototransfer from semiconductornanoparticles attached to the lipid membrane, we tested an inorganic compound—12-silicotungstate heteropolyacid (HPA, H4SiW12O40), which can serve both as reversibleelectron carrier and as a catalyst of hydrogen evolution in vesicle suspensions [26]. TheHPA anions were chosen, because their structure and properties are similar to those ofiron–sulfur clusters that constitute the reaction centers of enzymes such as hydrogenasesand nitrogenases. Second, the reduced forms of HPA are capable of evolving hydrogen [6].Besides, it is known that some forms of HPA can be dissolved in low-polarity organicphases; thus, imbedding of HPA in lipid membranes is possible.

FIG. 11 Schematic view of the designed photocatalytic systems with (a) transmembrane and (b)

interfacial electron transfer, which is photosensitized by the CdS nanoparticles attached to the lipid

membrane surface. Menaquinone (MQ) and heteropolyanions (HPA, SiW12O4�40 ) are lipophilic

molecular electron relays. Palladium particles are attached to CdS and operate as ‘‘dark’’catalysts

of hydrogen evolution from water. MV2þ: methylviologen bication; Gl: glucose.


Indeed, modeling of the lipid membrane interface by a two-phase system ‘‘water–nonpolar organic solvent’’ has shown that HPA prefers to localize in lipid membranes nearthe membrane surface.

An efficient photocatalytic functioning of HPA in the vesicle suspension in thepresence of CdS nanoparticles was verified experimentally. Simultaneously, a significantinfluence of pH on such system behavior was found. For example, it was revealed thatthough the quantum yield of the HPA photoreduction increases with a decrease in thesuspension pH from 7 to 3 (up to 1.5%, in the presence of CdS nanoparticles), a con-siderable decrease in the stability of the system as a whole was observed. Is was also foundthat, with increasing pH (up to 5 and above), the HPA reduction occurs even in the darkdue to the reducing ability of surplus sulfide anions that are obviously present in thesystem.

It is of interest that substitution of the sulfide-containing CdS PhC by the highlyphotoactive trisbipyridylruthenium(II) complex results in the formation of a photoinactiveRuðbpyÞ3 HPA complex.

Indeed, the ‘‘dark’’ reduction by residual free sulfide anions used for the synthesis ofsemiconductor sulfide nanoparticles interferes strongly in the quantitative study of trans-membrane electron transfer in the systems under consideration. Nevertheless, it is possibleto formulate some approaches to eliminating the masking influence of residual sulfides,which will be an important part of future studies on new types of promising lipophilicelectron carriers like HPA anions.

It is of importance that, in the vesicle suspensions, 12-silicotungstate HPA appears tobe really capable of efficient functioning in integrated cycles of photocatalytic hydrogenproduction. It was also found that, in the pH range of existence of stable lipid vesiclescontaining CdS on the outer surface and simultaneously of HPA in the vesicle membrane,the photocatalytic production of hydrogen was not observed without introduction of anadditional ‘‘dark’’ catalyst for hydrogen formation. Introducing chloride complexes ofpalladium results in efficient photocatalytic evolution of hydrogen. For example, on expo-sure of the system to low energy UV radiation, hydrogen evolution starts after a 15-mininduction period; this period seems to correspond to activation of the evolution catalyst,accounted for by the formation of metal palladium particles (see Fig. 11) [26]. The sta-tionary photocatalytic activity of the process described was 0.3 mol H2 per hour per gram-atom of palladium. In the absence of HPA, the induction period of hydrogen evolutionwas 2 h and the stationary photocatalytic activity decreased by two order of magnitudes.

The system described seems to be the first example of an integrated system forphotocatalytic hydrogen evolution based on 12-silicotungstate HPA, CdS, and Pd, loca-lized on the outer surfaces of the lipid vesicle membranes. The system is operating effi-ciently under illumination with UV radiation with wavelengths longer than 300 nm.

V. CONCLUSIONS

The above data demonstrate the first experimental experience in developing photocatalyticsystems based on lipid vesicle suspensions with semiconductor nanoparticles as PhCs.First, this proves the possibility of targeted synthesis of such systems with controllabletopology of the arrangement of semiconductor nanoparticles with respect to the vesiclemembrane. Besides, the factors were found which permit control of the size of semicon-ductor nanoparticles attached to the lipid membranes and, as a result, the quantum yieldof the primary charge separation on electron transfer from semiconductor nanoparticles to


lipophilic reversible molecular electron carriers imbedded in lipid membranes. It is estab-lished that not only lipophilic organic molecules, but also inorganic compounds likeheteropolyanions can be used as efficient molecular electron carriers in artificial lipidvesicles. One can assume that the last class of compounds can appear even more promisingfor photocatalytic operation. Indeed, the possibility of creation, on this basis, of anefficiently operating photocatalytic system of hydrogen evolution from water is demon-strated.

All the above data allow us to suppose that composite highly organized photocata-lytic systems with semiconductor nanoparticles as photosensitizers can actually be con-sidered as a prospective class of objects for developing many functional models of naturalphotosynthesis.

ACKNOWLEDGMENTS

This work has been supported by grant no. 01-03-42730 from the Russian Fund for BasicResearch as well as by grant no. 00-15-97446 from the Program ‘‘Leading Science Schoolsof Russia.’’

FIG. 12 Simplified energy diagram of electron phototransfer in the process of natural

photosynthesis in (a) plants, and in an artificial model [shown in Fig. 11(a)]; (b) on the scale of

reduction potentials Ee (versus the NHE).


REFERENCES

1. VN Parmon. Catal Today 39:137–144, 1997.

2. SV Lymar, VN Parmon, KI Zamaraev. Top Curr Chem 159:1–65, 1991.

3. VYa Shafirovich, SV Lymar, VN Parmon, AE Shilov, KI Zamaraev. In: KI Zamaraev, VN

Parmon, eds. Fotokatalliticheskoe Preobrazovanie Solnechnoi Energii. Novosibirsk: Nauka,

1991, pp 18–63 (in Russian).

4. JH Fendler. Chem Rev 87:877–899, 1987.

5. MI Khramov, VN Parmon. J Photochem Photobiol A: Chem 71:279–284, 1993.

6. YA Gruzdkov, EN Savinov, LL Makarshin, VN Parmon. In: KI Zamaraev, VN Parmon, eds.

Fotokatalliticheskoe Preobrazovanie Solnechnoi Energii. Novosibirsk: Nauka, 1991,

pp 186–228 (in Russian).

7. D Meissner, R Memming, B Kastening. Chem Phys Lett 96:34–39, 1983.

8. Y Tian, CH Wu, JH Fendler. J Phys Chem 98:4913–4918, 1994.

9. TI Igumenova, OV Vasil’tsova, VN Parmon. J Photochem Photobiol A: Chem 94:205–213,

1996.

10. OV Vassiltsova, AL Chuvilin, VN Parmon. J Photochem Photobiol A: Chem 125:127–134,

1998.

11. R Rafaeloff, Y-M Tricot, F Nome, JH Fendler. J Phys Chem 89:553–537, 1985.

12. Y-M Tricot, A Emeren, JH Fendler. J Phys Chem 89:4721–4726, 1985.

13. R Rafaeloff, Y-M Tricot, F Nome, P Tundo, JH Fendler. J Phys Chem 89:1236–1238, 1985.

14. Y-M Tricot, JH Fendler. J Am Chem Soc 106:2475–2476, 1984.

15. DD Lasic. Liposomes: from Physics to Applications. Amsterdam: Elsevier, 1993, pp 3–43.

16. EI Kuznetsov, SV Lymar, VI Bolshakov, VN Parmon, YuI Fyedorov, KI Zamaraev. Khim

Fiz 10:1370–1375, 1983 (in Russian).

17. L Hammarstrom, M Almgren, J Lind, G Merenyi, T Norrby, B Akermark. J Phys Chem

97:10083–10091, 1993.

18. L Hammarstrom, H Berglund, M Almgren. J Phys Chem 98:9588–9593, 1994.

19. SV Lymar, KJ Hurst. J Phys Chem 98:989–996, 1994.

20. OV Vassiltsova, DE Babushkin, VN Parmon. Dokl Akad Nauk 361:489–494, 1998 (in

Russian).

21. VG Ivkov, GN Berestovskii. Dinamicheskaya Structura Lipidnogo Bisloya. Moscow: Nauka,

1981, pp 21–221 (in Russian).

22. A Henglein. Chem Rev 89:1861–1873, 1989.

23. DB Bavikin, IN Martyanov, EN Savinov, VN Parmon. Izv AN Ser Khim 9:1739–1746, 1995.

24. YuYu Lurye. Spravochnik po analiticheskoi khimii M: Khimiya. 1979, pp 328–348 (in

Russian).

25. OV Vassiltsova, VN Parmon. Dokl Akad Nauk 360:61–65, 1998 (in Russian).

26. VN Parmon, KI Zamaraev. In: N Serpone, E Pelizzetti, eds. Photocatalysis. Fundamentals

and Applications. New York: John Wiley, 1989, pp 565–601.

27. KI Zamaraev. In: JW Hightower, WN Delgass, E Iglesia, AT Bell, eds. Studies in Surface

Science and Catalysis. vol. 101, part A. Amsterdam: Elsevier, 1996, pp 35–50.


22Catalysis and Photocatalysis at PolarizedMolecular Interfaces: An ElectrochemicalApproach to Catalytic Processes Based onTwo-Phase Systems, Self-OrganizedMicroheterogeneous Structures, andUnsupported Nanoparticles

RIIKKA LAHTINEN University of Liverpool, Liverpool, United Kingdom

HENRIK JENSEN and DAVID J. FERMIN Ecole Polytechnique Federale deLausanne, Lausanne, Switzerland

I. INTRODUCTION

The main objective of this chapter is to illustrate how fundamental aspects behind catalytictwo-phase processes can be studied at polarizable interfaces between two immiscible elec-trolyte solutions (ITIES). The impact of electrochemistry at the ITIES is twofold; first,electrochemical control over the Galvani potential difference allows fine-tuning of theorganization and reactivity of catalysts and substrates at the liquidjliquid junction.Second, electrochemical, spectroscopic, and photoelectrochemical techniques provide fun-damental insights into the mechanistic aspects of catalytic and photocatalytic processes inliquid|liquid systems. We shall describe some fundamental concepts in connection withcharge transfer at polarizable ITIES and their relevance to two-phase catalysis. In sub-sequent sections, we shall review catalytic processes involving phase transfer catalysts,redox mediators, redox-active dyes, and nanoparticles from the optic provided by electro-chemical and spectroscopic techniques. This chapter also features a brief overview of theproperties of nanoparticles and microheterogeneous systems and their impact in the fieldsof catalysis and photocatalysis.

II. FUNDAMENTAL ASPECTS OF CHARGE TRANSFER AT THE ITIES

Electrochemical processes at the ITIES involve two basic types of elementary reactions:ion transfer and electron tunneling across the liquidjliquid boundary. Depending on theproperties of the ionic species and the solvents, these two processes can be accompanied bya variety of phenomena such as solvent exchange, interfacial complexation, adsorption,photoexcitation, acid–base dissociation, etc. There are conceptual as well as practical


issues concerning charge-transfer phenomena and catalytic processes in two-phasesystems. In this section, we shall outline an electrochemical framework in which experi-mental results may be described and discussed.

A. Ion Transfer Reactions at the ITIES

The difference in the activity of an ionic species in a system with two immiscible electro-lytes manifests itself in terms of an electrical potential. This term, in analogy to cellmembranes, has profound influence on the reactivity and composition of the system. Byconvention, we shall define �w

o� as the difference in the Galvani potential of the aqueouselectrolyte, �w, and the organic phase, �o, respectively.

�wo� ¼ �w � �o ð1Þ

Let us consider an ion, i, with charge zi present in both liquid phases at equilibrium:

izðwÞ Ð izðoÞ ð2ÞAt constant temperature and pressure, the equilibrium condition is determined by

the equality of the electrochemical potentials in water, ��wi , and in the organic phase, ��o

i :

��oi ¼ ��w

i ð3Þwhich can be further developed to obtain the Galvani potential difference as a function ofthe concentration ratio of the ionic species:

�wo� ¼

��;oi � ��;wi

ziFþ RT

ziFln

aoiawi

ð4Þ

where ai and ��i correspond to the activity and the standard chemical potential of the ion

in each electrolyte phase. The standard Gibbs energy of ion transfer from water to oil,�G�;w!o

tr;i , is given by the difference in chemical potentials. This term is determined by thedifference in the bulk solvation energies of the ionic species. By defining the standardGalvani potential for the ion transfer from water to oil as

�wo��i ¼ �

��;wi � ��;oi

� �ziF

¼ �G�;w!otri;i

ziFð5Þ

the corresponding Nernst expression can be obtained [1]:

�wo� ¼ �w

o��i þ

RT

ziFln

aoiawi

ð6Þ

From this relatively simple expression, a variety of useful considerations can bedeveloped. First, if we consider an ion that is soluble in both phases, the Galvani potentialdifference can be tuned by changing the concentration ratio of the ionic species. Thiscondition is commonly referred to as an ideally nonpolarizable interface. On the otherhand, if the ionic species is strongly insoluble in one of the electrolyte solutions, thecorresponding standard transfer potential [see Eq. (5)] will be rather positive or negative,depending on the sign of the charge zi. Consequently, the Galvani potential difference canbe varied over a certain range without affecting the concentration ratio of the ionic species.This condition defines ideally polarizable interfaces. We will be employing this terminol-ogy throughout this chapter. It should also be considered that when more than one ionicspecies is involved in the equilibrium condition, and when the phase volume ratio is


significantly different from unity, estimation of the interfacial potentials can be complex[2,3]. These questions are dealt with in a separate chapter [chapter by Prof. Kakiuchi].

In the case where the ionic species in the aqueous electrolyte are fairly hydrophilicand the organic phase features hydrophobic ions, the liquid|liquid junction behaves simi-larly to an ideally polarizable metal electrode. Under this condition, the Galvani potentialdifference can be effectively controlled by a four-electrode potentiostat [4,5]. A schematicrepresentation of a typical electrochemical cell is shown in Fig. 1 [6]. Cyclic voltammo-grams illustrating the potential window for the waterj1,2-dichloroethane (DCE) interfacefor various electrolytes are also shown in Fig. 1. In the presence of bis(triphenylpho-sphoranylidene)ammonium hexafluorophosphate (BTPPAþ PF�6 ) as the supporting elec-trolyte in DCE, the potential window is limited to less than 200 mV due to thehydrophilicity of the anion. Wider polarizable potential ranges are obtained on replacing

FIG. 1 Typical electrochemical cell employed for electrochemical studies at the polarizable

waterj1,2-dichloroethane interface. Cyclic voltammograms in the presence of various supporting

electrolytes are also illustrated. Polarizable potential windows close to 1 V can be obtained

employing Li2SO4 and BTPPAþTPFB as supporting electrolytes in the aqueous and organic

phase, respectively. (From Ref. 6.)


PF�6 by either tetrakis-(4-chlorophenyl)borate (TPBCl�) or tetrakis(pentafluorophenyl)-borate (TPFB�). Figure 2 exemplifies a voltammetric response for a quasireversible trans-fer of an ionic species taking place within the polarizable window [7]. The position of thevoltammetric signal allows determination of the formal transfer potential, from whichvaluable information such as the Gibbs energy of transfer [Eq. (6)] can be obtained. Inthe case of a reversible ion-transfer process, the charge of the crossing ion can be calcu-lated from the peak-to-peak separation ð�Ep � 59mV=ziÞ. In the particular case of Fig. 2,the transferring ion zinc meso-tetrakis-(N-methylpyridyl)porphyrin (ZnTPPS) has charge�4, and the�Ep approaches 15 mV. Furthermore, the concentration ratio of the ion at theinterfacial region is readily accessible as a function of the Galvani potential difference, Eq.(5). Dynamic parameters such as diffusion coefficients and the phenomenological rateconstant of ion transfer can also be evaluated.

The measured current due to an ion transfer from one phase to the other can beexpressed as [8–12]

I ¼ ziFA k�!cai � k

�cbi

�ð7Þ

where k�!

and k �

are the forward and backward local ion transfer rate constants in units ofcm s�1. This expression describes the flux of ions across two reaction planes ‘‘a’’ and ‘‘b’’located at each side of the liquid|liquid boundary. The concentration of the ion at thereaction planes can be expressed in terms of the bulk concentrations, cwi and coi , and thepotential drop across the corresponding diffuse layers, �a

w� and �bo�,

cai ¼ cwi e

�ziF�aw�

RT ð8Þ

cbi ¼ coi e

�ziF�bo�

RT ð9ÞSubstitution of Eqs (8) and (9) into Eq. (7) yields:

I ¼ ziFA k�!

cwi e

�ziF�aw�

RT � k �coi e

�ziF�bo�

RT

264

375 ð10Þ

FIG. 2 Cyclic voltammogram of the water-soluble ZnTPPS4� transfer at the waterjDCE junction.

The composition of the cell is as in Fig. 1, employing Li2SO4 and BTPPATPFB as supporting

electrolytes. The voltammograms were recorded at 10, 20, 40, 60, and 80 mV s�1. The formal

transfer potential as well as the diffusion coefficient in the aqueous phase can be readily extracted

from the voltammetric signal. Reprinted with permission from Ref. 7. Copyright (1999) American

Chemical Society.


If we consider that the potential drop between the reaction planes is negligible [13–15], i.e.,the applied potential is effectively developed within the diffuse layer in each phase, Eq.(10) can be rewritten as

I ¼ ziFA k�!

cwi e

�ziF�wo�

RT � k �

coi e

�ð1� �ÞziF�wo�

RT

24

35 ð11Þ

where the term � corresponds to the fraction of the applied potential developed in theaqueous phase. In analogy to the classical expression in electrodynamics, the standard rateconstant for ion transfer, k�i , can be defined from the condition:

k�!

cwi e

�zoF�wo�� 0

RT ¼ k �

coi e

�ð1� �ÞziF�wo�� 0

RT ¼ k�i ð12Þallowing Eq. (12) to be expressed as

I ¼ ziFAk�i cwi e

�ziFð�wo��w

o�� 0 Þ

RT � coi e

�ð1� �ÞziFð�wo��w

o�� 0 Þ

RT

264

375 ð13Þ

This general expression has been confirmed experimentally by different groups [8,16–19], although deviations have also been observed [20–23]. Questions remain open concern-ing the physical meaning behind this phenomenological ion-transfer rate constant. Forinstance, it is still a matter of debate whether ion transfer can be treated as an activatedphenomenon or as a transport process through an inhomogeneous media. We have verylittle knowledge on what the correlations are between the observed rate constant of iontransfer and properties such as solvation distribution and dynamics, surface tension,specific adsorption, local viscosity, and friction forces [24].

It should also be mentioned that the Gibbs energy of ion transfer can be affected bycomplexation phenomena at the liquid|liquid boundary. A classical example is given bythe work of Koryta [25], who studied the transfer of Kþ from water to nitrobenzeneassisted by dibenzo-18-crown-6. The complexation step decreases the energy of solvationof the cation in the organic phase, decreasing the formal transfer potential as defined inEq. (5). Various mechanisms have been proposed for assisted ion-transfer processes,namely, aqueous complexation followed by transfer, transfer followed by complexationin the organic phase, transfer by interfacial complexation, or transfer by interfacial dis-sociation [26,27].

Despite the lack of a comprehensive understanding of the physical aspects under-lying the phenomenological expressions derived previously, they allow characterization ofthe dynamics of charge transfer employing electrochemical techniques. Conventional tech-niques such as cyclic voltammetry, chronoamperometry, and a.c. impedance have beencomplemented by more modern approaches including voltammetry at microinterfaces [28–33] and dynamic spectroelectrochemistry [34–43]. All these methodologies can providevaluable insights into the mechanistic aspects of two-phase catalysis involving chargedspecies. As discussed in Section III, electrochemical techniques can provide information onthe transfer rate of a phase transfer catalyst by means of Eq. 13. Furthermore, the parti-tioning of an ionic catalyst can affect the Galvani potential difference as indicated by theNernst expression, inducing concentration polarization of ionic reactants and substrates(Eqs. 8 and 9).


B. Electron-Transfer Reactions at the ITIES

In many biological systems, electron transfer takes place between redox couples present inmedia with different dielectric properties. Electrochemical studies at the ITIES enable oneto address systematically the effect of polarization and specific properties of the electrolytemedium on the dynamics of electron transfer. This knowledge has particular relevance inprocesses involving redox phase transfer catalysis.

Let us consider a heterogeneous electron-transfer reaction of the form:

Ow1 þRo

2kobsÐ Rw

1 þOo2 ð14Þ

for which the equilibrium condition can be described by the corresponding Nernst equa-tion:

�wo� ¼ �w

o�oET �

RT

nFln

awO1aoR2

awR1aoO2

" #ð15Þ

where �wo��ET is the standard potential for heterogeneous electron transfer, and n is the

number of electrons transferred in the redox process. The standard potential for hetero-geneous electron transfer is determined by

�wo��ET ¼ E�;oO2=R2

� E�;wO1=R1ð16Þ

where the bulk phase standard potentials, E�;oO2=R2and E�;wO1=R1

, are related to the samereference state.

By employing transition-state theory a phenomenological Butler–Volmer equationcan be derived for the heterogeneous electron-transfer rate [12,24,44]:

I ¼ Io e�nF �wo ��w

o ��ð Þ=RT � e�ð1��ÞnF �w

o ��wo ��ð Þ=RT

h ið17Þ

where

Io ¼ nFAk�ETcwO1coR2

ð18ÞIn Eq. (17), � is the fraction of the applied potential acting at the transition state. Theexchange current, Io, is determined by the concentrations of the reacting redox species, cwO1

and coR2, and the standard rate constant for the heterogeneous electron transfer, k�ET.

Similarly to electrochemical studies of the ion-transfer process, experimental evidenceseems to support the behavior described by Eq. (18) [45–48]. However, the significanceof the parameter � is yet to be clarified, as a simplified model of the electrical double layersuggests that the potential drop developed between the redox species at the interfaceshould be relatively insignificant. However, the basic physics in connection with the phe-nomenological rate constant of electron transfer are better understood than in the case ofion-transfer phenomena. Contributions by Marcus [49–52], Kharkats and Volkov [53,54],and Schmickler [55] in this area can be highlighted.

From the perspective of two-phase catalysis, the relevance of these concepts can bediscussed in similar terms to ion-transfer phenomena. For instance, by introducing ionswith the appropriate transfer potential, the Galvani potential difference between twoimmiscible liquids can be adjusted in order to enhance the driving force for interfacialredox processes. Furthermore, a combination of electrochemical and spectroscopic tech-niques allows unraveling of the interaction between redox catalysts and substrates in theinterfacial region [56].


C. Photoinduced Charge-Transfer Processes

Photocatalytic processes in two-phase systems involve either a homogeneous photoreac-tion followed by the transfer of intermediate species across the interface, or aheterogeneous electron transfer between the photoactive species and the substrate. Atthe polarizable ITIES, both processes would manifest themselves by an increase in thecurrent on illumination at constant potential, i.e., a photocurrent response. Indeed, photo-current measurements have been recorded for the transfer of photogenerated ions at aliquidjliquid boundary, as well as for heterogeneous redox quenching. We shall reviewsome of these studies in this section.

According to the Frank–Condon approximation, light absorption by moleculesleads to electronic transitions in which the momentum is essentially preserved. The‘‘hot’’ excited state readily relaxes to thermally equilibrated excited states within picose-conds. Further relaxation phenomena can involve radiative transitions between states ofthe same spin multiplicity (fluorescence) or different multiplicity (phosphorescence).Excited states can also undergo intersystem crossing, where the system changes from astate of high to low spin multiplicity. A schematic diagram of the photophysical phenom-ena and characteristic lifetimes for zinc tetraphenylporphyrin is displayed in Fig. 3 [57].From the kinetic point of view, all these phenomena are in competition with unimolecularphotochemical reactions such as isomerization and bond cleavages, as well as bimolecularprocesses involving energy and electron transfers.

From Fig. 3, it is clear that the reactivity of molecules in the excited state isdirectly linked to lifetimes. In general, the redox potentials of an excited state, E�Sþ=S and E�S =S� , can be expressed in terms of the ground state redox potentials, E�Sþ=S andE�S=S� , as

E�S=S ¼ E�Sþ=S � E ð19ÞE�S =S� ¼ E�S=S� þ E ð20Þ

FIG. 3 Jablonski energy level diagram for a regular porphyrin, illustrating the photophysical

transitions and corresponding time scales for excitation at the Soret ðS0 ! S2) and Q-bands

(S0 ! S1).


where E is the characteristic energy of the excited state. If electron transfer can effectivelycompete with vibrational relaxations, leading to thermally equilibrated excited states(‘‘hot’’ electron transfer), E corresponds to the wavelengths of excitation. These phenom-ena have been observed in dye-sensitized metal oxides [58–63]. However, electron transfercommonly involves relatively long-living excited states such as the first singlet (S1) ortriplet states (T1). Figure 4 shows the redox potentials associated with the ground stateand T1 for the dye ruthenium (II)-trisbipyridine RuðbpyÞ2þ3 [57].

As we mentioned previously, photoinduced electron transfer at the polarizableliquid|liquid junction manifests itself by photocurrent responses under potentiostatic con-ditions. The nature of the photoelectrochemical processes is reflected in the basic featuresof the photocurrent transient. For instance, a homogeneous photochemical reaction fol-lowed by the transfer of the products is characterized by a slow increase in the photo-current on illumination. A typical example can be extracted from the work of Kotov andKuzmin shown in Fig. 5(a) [64–66]. In this case, protoporphyrin is located in the organicphase in the presence of benzoquinone. On illumination, the quinone is reduced and theradical anion transfers to the water phase. The increasing photocurrent is connected withthe flux of the radical anion from DCE to water.

The rather slow transients observed in the previous case can be contrasted with therelatively fast photocurrent response arising from heterogeneous redox quenching [see Fig.5(b)] [7,24,56,67–69]. In this case, the ion pair formed by zinc meso-tetrakis-(p-sulfonato-phenyl)porphyrin (ZnTPPS4�) and zinc meso-tetrakis-(N-methylpyridyl)porphyrin(ZnTMPyP4þ) located in the aqueous phase is reductively quenched by decamethylferro-cene (DCMFc) present in the DCE phase [7]. In this case, the observed photoresponseresults from competition between the decay of the excited state and the rate of hetero-geneous electron transfer. The magnitude of the photocurrent increases as the Galvanipotential difference is shifted to more positive values.

The general mechanism associated with heterogeneous photoinduced electron trans-fer is highlighted in Fig. 6. Considering the efficiency of intersystem crossing in zincporphyrins, it is expected that the electron transfer takes place from the excited stateT1. The dependence of the photocurrent on the applied potential indicates that the elec-tron transfer step is in competition with the decay of the excited state. The relationshipbetween the phenomenological ket and the driving force for electron transfer has beendescribed in terms of the Marcus expression for electron transfer between redox speciesseparated by a sharp change in the dielectric medium [70].

Depending on the redox properties of the quencher and the Galvani potential dif-ference, back electron-transfer reactions can occur [56,69–71]. The back electron transfermanifests itself as relaxation after the initial photocurrent. The rather small relaxation

FIG. 4 Latimer diagram for RuðbpyÞ2þ3 . The excited state corresponds to a triplet metal-to-ligand

charge transfer state.


observed in the photocurrent transient of Fig. 5(b) reveals an efficient separation of thephotoproducts. Dynamic photoelectrochemical measurements including photocurrenttransients and intensity modulated photocurrent spectroscopy (IMPS) have providedvaluable insights into the dynamics of photoinduced electron transfer as well as backcharge-transfer phenomena [69,71].

Photoelectrochemical studies have also revealed information on the molecular orga-nization of dyes at interfaces. Photoresponses are essentially connected to the heteroge-neous quenching of specifically adsorbed dyes at the liquid|liquid boundary [72,73]. Forinstance, zinc meso-tetracarboxyphenylporphyrin self-assembles at the water|DCE junc-tion via the formation of co-operative hydrogen bonds. The coverage as a function of theapplied Galvani potential difference has been estimated from the photocurrent depen-dence on the concentration [68]. Furthermore, photocurrent measurements employinglinearly polarized light in total internal reflection allowed estimation of the average orien-tation of the transition dipole of the adsorbed photoactive dye.

The detailed information about organization and reactivity of dyes obtained fromphotoelectrochemical approaches has been complemented by a variety of spectroscopictechniques including second-harmonic generation [74], dynamic fluorescence [42,43], andquasielastic laser scattering [75]. Indeed, one of the most appealing aspects of modelingphotocatalytic processes in a system with two liquid phases is the diversity of experimentaltechniques at hand. These techniques not only allow characterization of the kinetics ofcharge-transfer phenomena, but also the interfacial organization of photocatalysts.

FIG. 5 Photocurrent transient responses obtained at a dye-sensitized waterjDCE junction for two

types of mechanisms: (a) homogeneous photoinduced electron transfer followed by ion transfer, and

(b) photoinduced heterogeneous electron transfer. Curve (a) was obtained in the presence of ZnTPP

and benzoquinone in the DCE phase (Reprinted from Ref. 64 with permission from Elsevier

Science). Panel (b) depicts the heterogeneous quenching of the heterodimer ZnTPPS–ZnTMPyP

by decamethylferrocene at two different Galvani potential differences. Reprinted with permission

from Ref. 7. Copyright (1999) American Chemical Society.


III. CATALYTIC REACTIONS AT LIQUIDjLIQUID INTERFACES

One of the main aspects behind two-phase catalysis is the synergism of homogeneous andheterogeneous catalysts. Homogeneous catalysis has the advantages of high activity andselectivity, mild reaction conditions, low sensitivity to catalyst poisoning, effective masstransfer, and the possibility of varying the steric and electronic properties of the catalyst[76]. Heterogeneous catalysis features long service life of the catalyst and the ease ofseparating the catalyst from the products. An effective combination of the two approachesconsists of separating the catalyst and the substrates by a liquidjliquid boundary [77]. Inprinciple, this configuration can overcome the difficulty often found in homogeneouscatalysis of separating and recovering the catalyst after the reaction. A common config-uration features the catalyst in the aqueous phase, while the substrate is located in theorganic phase. A great deal of recent interest has been directed towards synthesizingwater-soluble ligands and metal complexes to transfer the traditional homogeneous cata-lysts into an aqueous environment [76,78].

The use of a two-phase system can also have some other advantages in addition tothe catalyst-separation issues. A liquidjliquid system can be used to catalyze a reactionwhich would be very unfavorable in a homogeneous medium due to large differences in thesolvation energies of substrates and products. Substantial changes in heterogeneous reac-tion kinetics can be brought about either by an appropriate choice of solvents, tuning theGalvani potential difference along with redox properties of the reactants [79], or byinhibiting undesired processes [80].

A. Catalytic Studies at Polarizable LiquidjLiquid Interfaces

As we described in Section II, control over the Galvani potential difference allows tuning ofthe distribution, organization, and reactivity of species at the liquid|liquid boundary. This

FIG. 6 General mechanism for photoinduced heterogeneous electron transfer at dye-sensitized

liquid|liquid interfaces. The electron transfer from the excited state is in competition with the

relaxation process. In addition, the intermediate generated after the electron-transfer step can

undergo back electron transfer or dissociate at the interface. In the case of water-soluble

porphyrins at the waterjDCE junction, the orders of magnitude of the phenomenological rate

constants are indicated in the figure. The photoinduced electron transfer is denoted as a pseudo-

first order process with respect to the excited state (kel). Reprinted with permission from Ref. 7.

Copyright (1999) American Chemical Society.


property is useful in the particular case where the interfacial catalyst is charged. Classicalexamples of ionic catalysts include phase transfer catalysts, which shall be discussed in moredetail in the next section. Another powerful aspect concerning the electrochemical approachto two-phase catalysis is the possibility of real-time monitoring of the rate of the interfacialreaction, employing a variety of techniques such as chronoamperometry, chronopotentio-metry, cyclic voltammetry, a.c. impedance, and more recently, scanning electrochemicalmicroscopy (SECM) [81]. Electrocatalysis in this case is defined as a catalytic reactioninvolving electron or ion transfer, which takes place at the liquidjliquid interface betweenreagents located in different phases, and which requires an interfacial potential.

Despite the usefulness of the ITIES concepts, only very few studies have been pub-lished on mechanistic evaluations of catalytic processes at liquid|liquid interfaces. One ofthe few was presented by Cheng and Schiffrin [82], who employed cyclic voltammetry forstudying the behavior of redox mediators at a polarizable waterjDCE junction. This workdescribed the effect of modifying the liquid|liquid boundary by an adsorbed monolayer ofphospholipids on the electron-transfer rate between a hexacyanoferrate couple in waterand tetracyanoquinodimethane (TCNQ), lutetium bisphthalocyanine [LuðPCÞ2], andbis(pyridine)-meso-tetraphenylporphyrinato–ruthenium(II) [RuðTPPÞðpyÞ2] in DCE. Theelectron-transfer rate to TCNQ was noticeably decreased, while the electron transfer toLuðPCÞ2 or RuðTPPÞðpyÞ2 appeared totally hindered in the presence of the phospholipidlayer. The hindrance effect of the surfactant was rationalized in terms of an increase in thetunneling distance between the redox couples at the interface. However, for a systemcontaining both TCNQ and RuðTPPÞðpyÞ2 in the organic phase, a voltammetric responsewith two waves was observed. This behavior suggests that the electron transfer betweenRuðTPPÞðpyÞ2 and the hexacyanoferrate couple is effectively mediated by TCNQ. Theredox mediation is connected to the ability of TCNQ to penetrate the phospholipidlayer, and act as an electron shuttle between both phases.

Another form of two-phase redox electrocatalysis was discussed by Shao et al. [83],who employed the ITIES as a model system to investigate the mechanism and kinetics ofcomplex catalytic microemulsion reactions. The problem in studying electrochemicalkinetics in microemulsions is the lack of a well-defined interfacial area, and this can becircumvented by the use of the ITIES. The SECM technique was used to probe thereaction between the electrochemically generated aqueous Co(I) form of vitamin B12

and trans-1,2-dibromocyclohexane in benzonitrile. The effects of reactant concentration,Galvani potential difference, and surfactant adsorption on the electron-transfer kineticswere investigated. Kong and Kakiuchi [84] studied the nucleophilic substitution reactionof 2,4-dinitrofluorobenzene with hydroxide ions. Previously, this reaction has been studiedin homogeneous and micellar systems. They successfully used d.c. polarography and cyclicvoltammetry at a polarized water|nitrobenzene interface to obtain mechanistic informa-tion on the reaction.

A recent work by Chung and Anson [85] revealed interesting catalytic effects on theoxygen reduction on graphite electrodes modified by thin benzonitrile layers containingmeso-tetraphenyl cobalt porphyrin (CoTPP). As depicted in Fig. 7, the current associatedwith oxygen reduction increases on deposition of a thin layer of benzonitrile on thegraphite electrode. This increment of the current is associated with the larger solubilityof oxygen in the organic layer. Furthermore, a net enhancement of the four-electronoxygen reduction over the two-electron pathway was observed in the presence of thebenzonitrile layer. These results were initially rationalized by postulating that the organiclayer effectively protects active sites on the graphite electrode from electroreduction pro-cesses that may take place in aqueous media. The results in voltammograms (b) and (c)


indicate that the CoTPP can effectively catalyze the oxygen reduction when dissolved inthe benzonitrile thin layer. The CoTPP in the acidic environment is involved in a multi-electron transfer cycle that catalyzes the four-electron process. Although some of theseinterpretations may require independent experimental verification, the results do open newareas for two-phase redox electrocatalysis.

B. Electrochemical Approach to Phase Transfer Catalysis

Several chapters in this book are devoted to phase transfer catalysis (PTC), highlightingconventional as well as novel approaches to this important process. In this section, we shalllook at these phenomena from the viewpoint of electrochemistry in two-phase systems.

A key point to realize is that quaternary ammonium salts commonly employed as PTcatalysts exhibit a finite solubility in aqueous as well as in a variety of organic solvents.According to Eq. (5), the partitioning of an ion induces a Galvani potential differencebetween the electrolyte phases, which is determined by the difference in the solvationenergies of the ion. Similar ions have been used for electrochemical studies at the ideallynonpolarizable ITIES [86,87].

Based on these ideas, Cunnane et al. [88] compared the oxidation of tin diphthalo-cyanine [SnðPCÞ2] in the DCE phase by aqueous ferri/ferrocyanide redox couple underexternal polarization and in ideally nonpolarizable conditions. Good correlation for theformal redox potential measured in each case was observed. One of the main conclusionsof this work is that the role of ‘‘PT catalysts’’ can be simply associated with polarization ofthe two-phase system, resulting in an enhancement of the interfacial concentration of the

FIG. 7 Cyclic voltammograms of oxygen reduction at modified pyrolitic graphite electrodes.(a)

Voltammograms for the naked graphite electrode (curve 1) and in the presence of a thin layer of

benzonitrile (BN) (curve 2) are shown for oxygen-saturated HClO4 electrolyte solution. The dashed

line corresponds to Ar-saturated solutions in the presence of the BN layer. (b) Voltammograms

obtained for a 6� 109 mol cm�2 layer of CoTPP deposited on the graphite electrode, in the presence

(full line) and absence (dashed line) of oxygen. Finally, the cathodic current (plotted upwards for all

curves) is further increased on depositing a thin layer of BN on the CoTPP-modified graphite

electrode. Reprinted from Ref. 85 with permission from Elsevier Science.


reactants or an increase in the driving force for the charge-transfer process. Tan et al. [89]successfully applied the above formalism in the study of the Williamson ether synthesisand concluded that the PT catalyst salt established a Galvani potential difference, which inturn acted as the driving force for transferring the reactive aqueous ions to the organicphase.

Kong et al. [90] applied the electrochemical approach to the study of a two-phase azocoupling facilitated by reverse PTC. Cyclic voltammetry and chronoamperometry wereemployed to evaluate quantitatively the rate constants for the reaction. The process wasinterpreted in terms of an EC mechanism, i.e., diffusion-controlled electrochemical chargetransfer followed by a homogeneous chemical reaction. The authors highlighted the use-fulness of this approach based on the factors that enable the estimation of the contribu-tions of the chemical reaction, mass transfer, partitioning, and the adsorption of reactantsat the interface to the overall two-phase reaction.

In a similar line of research, Forssten et al. [91] investigated the formation of 2-hydroxy-3-methylbutyric acid (ROH) from 2-bromo-3-methylbutyric acid (RBr) at awaterjDCE interface. This Sn2 reaction in two-phase systems can be represented as

ðCH3Þ2CHBrCOOHðoÞ þOH�w Ð ðCH3Þ2CHOHCOOHðoÞ þ Br�ðwÞ ð21Þwhich can be catalysed in the presence of I� by following the cycle:

ðCH3Þ2CHBrCOOHðoÞ þ I�WÐ ðCH3Þ2CHICOOHðoÞ þ Br�ðwÞ ð22Þ

ðCH3Þ2CHICOOHðoÞ þOH�w Ð ðCH3Þ2CHOHCOOHðoÞ þ I�ðwÞ ð23ÞFigure 8 illustrates the effect of increasing the concentration of RBr in the organic

phase on the cyclic voltammogram responses in the presence of I� in the aqueous phase.As the concentration of RBr is increased, the voltammetric signal associated with thetransfer of I� is shifted toward more negative potentials. Indeed, in an excess of RBrthe transfer signal approaches the transfer potential of Br� (trace 5). In order to rationa-lize this behavior, a combination of two ECE mechanisms was postulated. The firstsequence involves the transfer of I� (electrochemical step), followed by the homogeneous

FIG. 8 Effect of 2-bromo-3-methylbutyric acid concentration (RBr) on I� transfer at a waterjDCE

interface. The compositions for the curves were: ½I�� ¼ 1mM, [RBr� ¼ 0ð1Þ; [I�� ¼ 1mM,

[RBr� ¼ 1mM (2); [I�� ¼ 1mM, [RBr� ¼ 2mM (3); [I�� ¼ 1 mM, [RBr� ¼ 4mM (4); [I�� ¼ 0,

[Br�� ¼ 1mM (5); [I�� ¼ ½RBr� ¼ 0 (6). It is observed that the ion-transfer signal shifts from the

transfer potential of I� in (1) to the transfer potential of Br� (5), with increasing concentration of

RBr. Reprinted from Ref. 91 with permission from Elsevier Science.


reaction (22) (chemical step) and the second electrochemical step is the transfer of Br� tothe aqueous phase. The other ECE mechanism corresponds to the assisted transfer ofOH� to the organic phase, homogeneous substitution of the I� [step (23)], and transferof the last to the aqueous phase. Simulation of this mechanism yielded results compatiblewith the experimental trends. Furthermore, the reaction was also induced by partitioningof the tetraphenylphosphonium ion, which establishes the appropriate Galvani potentialdifference for the reaction to take place according to the voltammetric studies, and theformation of the final products was confirmed by NMR spectroscopy.

More recently, Forssten et al. [92] employed a similar approach to study the oxida-tion of cis-cyclo-octene by permanganate at a waterjDCE junction. As in the previouscase, the redox species in the aqueous phase is driven in and out of the organic phase by alinear sweep of the Galvani potential difference. However, the homogeneous reaction israther slow, and the amount of permanganate consumed during the potential cycleappears insignificant. In order to increase the transient time of permanganate in theorganic phase, a pre-electrolysis step was applied, in which the potential was held for afixed period of time in the region where the transfer of permanganate was diffusioncontrolled. Subsequently, the potential was cycled to positive values, transferring theremaining permanganate back to the aqueous phase. From the current associated withthe back transfer of permanganate and the pre-electrolysis time, the rate of the homo-geneous reaction was estimated. Indeed, the recent works by Forssten et al. [92,93] andKong and coworkers [84,90] provide an elegant illustration of how electrochemical ana-lysis can deliver valuable kinetic information on PTC processes.

IV. PHOTOCATALYSIS AT DYE-SENSITIZED MOLECULAR INTERFACES

In Section II.C, we described the reactivity of adsorbed dye species at liquid|liquid junc-tions in heterogeneous photoredox reactions. The properties of these systems can be usedto catalyze electron-transfer processes. The behavior of dyes at interfaces has been vigor-ously studied in micelles and microemulsion systems, and many excellent reviews andbooks are available on this subject [94–97]. In this section, we shall consider some basicaspects of photoprocesses in microheterogeneous systems that are relevant to polarizableITIES. This is not intended to cover comprehensively the recent developments in the activearea of photochemistry at organized assemblies, but to highlight how spatial confinement,hydrophilic–hydrophobic forces, and local potentials can affect the course of a photoche-mical process. We shall also revise some recent developments in photocatalysis and photo-synthesis at polarizable liquid|liquid interfaces, highlighting advantages and limitations inrelation to two-phase catalysis.

A. Photocatalysis Employing Dyes at Organized Assemblies

Due to the tremendous impact of photocatalysis in the area of environmental chemistry,photocatalysts are mainly associated with colloidal suspensions of metal oxides, in parti-cular TiO2 [98–101], or with iron chloride solutions in the presence of hydrogen peroxide,also known as Fenton reagent [102–104]. Dye species can also behave as photocatalystseither by generating singlet oxygen via energy transfer to molecular oxygen, or by directelectron transfer involving excited states. For the latter mechanism, the ground state of thedye species must be regenerated after the redox step; therefore, this approach is essentiallyconstrained to single electron-transfer reactions. Semiconducting nanoparticles with high


dielectric permittivity, e.g., TiO2, are best suited for multielectron transfer processes due totheir ability of storing charge within the nanocrystal structure. We shall come back to thecatalytic properties of nanoparticles in the next section. However, the possibility of fine-tuning the hydrophilicity of dye species provides an effective tool for designing photo-catalytic processes at water|oil systems.

Photoredox reactions at organized assemblies such as micelles and microemulsionsprovide a convenient approach for modeling life-sustaining processes. Micelles are spon-taneously formed in solutions in the presence of surfactants above a certain criticalconcentration. In aqueous solutions, the hydrophobic tails of the surfactant form aggre-gates with the polar head facing toward the aqueous environment, as depicted in Fig. 9.The hydrophobic core in micelles is amorphous and exhibits properties similar to aliquid hydrocarbon. The polar heads are also randomly oriented, generating an electricaldouble layer around the micelle structure. In this respect, surface properties of micellescan be somewhat correlated with the polarized ITIES. The structure of micelles is indynamic equilibrium, in which monomers are exchanged between bulk solution and theassembly.

The general features of micelles and reversed micelles (polar head buried in, with thehydrophilic tail pointing out to a low dielectric medium) can enhance the separation ofphotoproducts in different fashions. One of the best known examples is the effect of thesurfactant cetyltrimethylammonium chloride (CTAC) on the photoreduction of C14V

2þ byRuðbpyÞ2þ3 [105]. As schematically depicted in Fig. 10(a), the radical C14V

þ is extractedinto the inner core of the micelle due to its hydrophobic character. The oxidizedRuðbpyÞ3þ3 is electrostatically repelled away from the positively charged micelle, decreasing

FIG. 9 Schematic representation of a micelle, illustrating the characteristic dimensions of the

double-layer structure.


the probability of back electron transfer. As a result, the product lifetime is increased by afactor of 20 with respect to the bulk homogeneous process.

The viologen reduction by EDTA in reverse micelles in the presence of RuðbpyÞ2þ3is another example of vectorial photoinduced electron transfer [106]. The accumulationof photoproducts is associated with the catalytic cycles depicted in Fig. 10(b). Theoxidative quenching of the ruthenium complex occurs at the micelle outer boundary,while the regeneration of the dye takes place by the oxidation of EDTA in the inner coreof the micelle. The reduction of the final product 4-dimethylaminoazobenzene is furthermediated by the acceptor 1-benzylnicotinamide (BNAþ). In Fig. 10(c), the photocatalyticreduction of methyl benzoylformate (MBF) by thiosulfate is described in the presence ofthe porphyrin ZnTPPS and the mediator quinolinium-3-carboxiamide (DCAþ) [107].This sequence of reactions occurs only in micelles such as those formed by hexadecyl-trimethylammonium bromide, which contain in the interior the ultimate donor acceptor.Under illumination, ZnTPPS photoreduces DCAþ to DCQ, which is subsequentlyextracted into the micelle core. Within the microenvironment, DCAþ is regeneratedvia reduction of MBF, while the oxidized porphyrin is reduced by thiosulfate outsidethe micelle.

Efficient photocatalytic processes have also been studied in water-in-oil (W/O) andoil-in-water (O/W) microemulsions. W/O microemulsions are readily prepared in C5–C8

FIG. 10 Schematic representation of photocatalytic processes in the presence of micelles and

reversed micelles.


n-alkanes in the presence of surfactants with a rather hydrophobic character such assodium bis-(2-ethylhexyl) sulphosuccinate. The basic structure can be regarded as asmall water pool entrapped by the surfactant in the alkane solvent. The differentiationbetween reversed micelles and W/O emulsions could be ambiguous, but in general theradii of the former are of the order of 10–20 A while for the latter the radii are above200 A. Willner and Joselevich [108] have compared the oxidation of tributylamine byFeðCNÞ3�6 in the presence of eosin (Eo2�) and ethyl eosin (EoEt�). The optical transientresponse at 460 nm, which corresponds to the oxidized form of the dyes, is depicted inFig. 11(b). It can be observed that the concentration of the oxidized Eo2� sharplydecreases on illumination, while a steady-state concentration of the EoEt is clearlyobserved. These observations are rationalized in terms of the mechanism depicted inFig. 11(a), in which the back electron transfer from FeðCNÞ4�6 to the oxidized eosin is incompetition with the transfer from the water inner core to the organic phase. In the caseof EoEt�, the radical formed after the photoinduced electron transfer is rather hydro-phobic and is efficiently extracted from the water pool prior to the back electron-transferphenomenon. In the organic medium, EoEt� is regenerated by tributylamine followed bytransfer back to the water pool.

Through these well-known examples the effect of three basic features of organizedassemblies can be visualized: (1) inhomogeneous solvation properties, (2) interfacial poten-tials, and (3) spatial confinement. Apart from the last, the other aspects can be finelycontrolled at the polarizable ITIES. As we have seen in Section II, the Galvani potentialdifference not only affects the dynamics of photoinduced electron transfer, but also theconcentration ratio of ionic species.

FIG. 11 (a) Schematic representation of electron transfer across a W/O microemulsion based on

the ‘‘shuttle photosensitizer’’ mechanism. (b) Transient absorption of the oxidized form of the

sensitizer eosin (Eo2�) and ethyl eosin (EoEt�) after a 9 ns pulse in the presence of

FeðCNÞ3�6 =FeðCNÞ4�6 and tributylamine (Bu3N). The faster decay observed for Eo2� is connected

to back electron transfer phenomena within the hydrophilic emulsion core. Reprinted with

permission from Ref. 108. Copyright (1999) American Chemical Society.


B. Photosynthesis and Photocatalysis at the Polarizable ITIES

The previous example illustrated how microheterogeneous systems can affect the efficiencyof charge separation following a photoredox process. These processes cannot be unam-biguously defined as photocatalytic or photosynthetic unless the overall change in theGibbs free energy is dealt with in a quantitative fashion. This can be complex as notonly the redox potentials, but also the changes in solvation energy associated with theionic and neutral species crossing the interfacial boundary have to be accounted for. Froman academic point of view, the possibility of tuning the Galvani potential difference andaccurate knowledge of the transfer potential of the ionic species provide a unique frame-work for characterizing the energy balance of photosynthetic and photocatalytic reactions.

FIG. 12 Photocurrent transient responses associated with the heterogeneous quenching of the

dimer ZnTPPS/ZnTMPyP by tetracyanoquinodimethane (TCNQ) at a waterjDCE interface. The

redox couple FeðCNÞ3�6 =FeðCNÞ4�6 was used as supersensitizer in the aqueous phase. The back

electron transfer reaction, responsible for the photocurrent decay in the on-transient, is

significantly quenched in the presence of the supersensitizer. According to the redox diagram in

(b), the overall process at �wo� ¼ �0:11V corresponds to the reduction of TCNQ by the redox

couple in the aqueous phase photocatalyzed by the porphyrin complex. Reprinted from Ref. 109

with permission from Elsevier Science.


The photoinduced reduction of TCNQ by the porphyrin heterodimer ZnTPPS–ZnTMPyP provides a good illustration of these concepts [56,109]. Figure 12(a) displaysphotocurrent transients at the waterjDCE junction in the presence and absence of anequimolar ratio of FeðCNÞ3�6 =FeðCNÞ4�6 . The photocurrent relaxation in the absence ofthe aqueous redox couples is associated with the back electron transfer from TCNQ� tothe oxidized porphyrin complex. The substantial decrease in back electron transfer onaddition of FeðCNÞ3�6 =FeðCNÞ4�6 is associated with the supersensitization phenomenonschematically depicted in Fig. 12(b). The back electron transfer from the radicalTCNQ� to the oxidized porphyrin complex is in competition with the regeneration ofthe dye by ferrocyanide. In the absence of back electron transfer, the overall reactioninvolves electron transfer from the redox species in the aqueous phase to TCNQ. In thiscase, the energetic balance is determined by the Galvani potential difference across theinterface.

From the redox potentials illustrated in Fig. 12(b), it is clear that, for the photo-responses obtained at potentials more negative than �0:07V, the overall reaction can beconsidered as a photocatalytic process. The heterogeneous reduction of TCNQ by ferro-cyanide across the waterjDCE interface is a quasireversible process featuring a phenom-enological bimolecular electron transfer of the order of 10�22 cm4 s�1 [47,48]. On the otherhand, the heterogeneous redox quenching of the porphyrin heterodimer by TCNQ can beestimated to be close to 10�20 cm4 s�1 [7,56,109]. Considering that the rate-determiningstep is the heterogeneous electron transfer step, the net increase in the reaction rate is dueto the larger driving force for the forward electron in the presence of the adsorbed dye.

Despite the simplicity of the electrochemical approaches to photoinduced reactionsat the ITIES, very few studies can be found dealing with dye-sensitized interfaces. Aninteresting point to be revisited is the dynamics and energetics associated with the classicalprocesses exemplified in Fig. 10. As we mentioned earlier, the distribution of ionic speciescan be tuned by the Galvani potential difference, providing a highly effective tool formanipulating the rate of the overall process.

V. CATALYTIC AND PHOTOCATALYTIC PROPERTIES OFNANOPARTICLES AT THE ITIES

A. Catalytic Properties of Metal Nanoparticles and Colloids in Solution

Decrease in the size of a metal particle below a critical dimension results in dramaticchanges in the electronic properties of the bulk metal. Properties like conductivity, mag-netism, light absorption, luminescence, electrochemical, and catalytic activity depend onthe particle size. Many heterogeneous catalysts are based on finely divided metal particleson various supports. However, this section deals with the catalytic properties of unsup-ported nanoparticles.

Metal nanoparticles can be prepared in a myriad of ways, e.g., by pulse radiolysis[110], vapor synthesis techniques [111], thermal decomposition of organometallic com-pounds [112], sonochemical techniques [113,114], electrochemical reduction [115,116],and various chemical reduction techniques. Some of the most frequently used reducingagents include alcohols [117,118], citrate [119,120], H2 [121], borohydrides [122], and, morerecently, superhydride [123]. The chosen experimental conditions determine the size, sizedistribution, shape, and stability of the particles. Because naked metal particles tend toaggregate readily in solution, stabilizing the nanoparticles is the key factor for a successfulsynthesis. Sometimes the solvent can act as a stabilizer, but usually polymers and surfac-


tants and, more recently, ligands and monolayers, have been used for this purpose. Animportant point is that the catalytic properties of the nanoparticles, including selectivityand activity, could be affected by the choice of the stabilizer [124]. Also, the solubility ofthe nanoparticles depends on the nature of the stabilization. Generally, hydrophilicligands yield water-soluble colloids and hydrophobic ligands yield colloids in organicsolvents.

Catalysis involving unsupported colloids has been referred to as ‘‘heterogeneouscatalysis in solution’’ [125]. In addition to the size-dependent electronic properties, thefraction of catalytically active surface atoms in nanoparticles is exceptionally large.Furthermore, reactions at the surface of small particles can be surface controlledwhereas reactions at a large plane surface may be diffusion limited [126]. Differencesbetween one metal colloid and another arise because of a number of factors: the parti-cular energy and symmetry of orbitals of the surface atoms and the cohesive energy ofthe particle [127]. In general, colloidal metals can be regarded as convenient pools into,and from, which electrons may be readily transferred. They can be described as nanoe-lectrodes in solution.

Most of the studied reactions take place in a ‘‘one-phase’’ environment, i.e., thecatalyst colloid and other reactants are dissolved in the same solvent or solvent mixture.Hydrogenation reactions are the most extensively investigated organic reactions catalyzedby unsupported metal, especially polymer-stabilized Pt [128–130] or Pd [131–133] colloids.Polymer-stabilized Pt has been found to have more than five times higher activity than acarbon-supported Pt catalyst with increased selectivity for partial hydrogenation of allylalcohol [134]. The regioselective [135] and enantioselective [136] properties of these cata-lysts in hydrogenation reactions have gained attention. Very recently, it was shown withparahydrogen-induced polarization NMR that, in the case of the hydrogenation of phe-nylacetylene mediated by colloidal Pd catalyst, the reaction follows a homogeneous path-way [137].

In the last few years, a significant amount of research has been carried out on Pdcolloids as catalysts for Heck reactions, and the most recent literature is focused on thecorrelations between the structure of the colloidal catalyst and the catalytic performance[138–141]. Other reactions for which the metal colloids have proven to be useful includehydrosilylations [142], isomerizations [128], and Suzuki reactions [124].

The excellent electron-transfer mediator properties of nanoparticles find special usein the different oxidation [126] and reduction [143,144] reactions catalyzed by noble metalcolloids. Recently, Ung et al. [145] showed how Ag particles coated with a thin layer ofsilica act as redox catalysts, and how the control of the rate of the catalyzed hydrogenevolution reaction was possible by tuning the silica shell thickness. It was concluded thatthe shell acts as a size-selective membrane, which can be used to alter the chemical yieldsfor competing catalytic reactions. This kind of tailoring of the catalyst properties opens upvery interesting prospects in future catalyst planning.

Successful tailoring of the metal nanoparticle catalyst has also been achieved by theuse of discrete, well-defined polymers called dendrimers as stabilizers. Dendrimers offereffective stabilization against agglomeration, and due to the steric nature of the stabiliza-tion a substantial fraction of the particle surface is unpassivated and available for catalyticreactions. Dendrimer branches can be used as selective gates to control access of substratesto the nanoparticle surface, and the terminal groups on the dendrimer can be modified tocontrol the solubility of the catalyst. The hydrogenation reaction rate could be controlledby using dendrimers with different porosities. The use of different stabilizing dendrimersalso makes selective catalysis possible [146].


Bimetallic nanoparticles have interesting scientific and technological properties inrelation to catalysis. Bimetallization can improve catalytic properties of monometalliccatalysts or create completely new assets. The new effects can often be explained withensemble or ligand effects in catalyses [147]. The chemical preparation of bimetallic par-ticles can be achieved by two strategies: coreduction or successive reduction of two metalsalts. The latter is used for preparing ‘‘core–shell’’ structured bimetallic nanoparticles.Unsupported colloidal bimetallic nanoparticles have been used to study hydrogenation[148,149] and hydration [147] reactions. The activity and selectivity of the catalyst wasimproved compared to that of a monometallic catalyst.

Colloidal catalysts tend to precipitate in homogeneous processes [150]. This can beobserved by the appearance of black metal residues in the system. This causes catalystlosses and decrease in the catalyst activity. Another significant problem in ‘‘one-phase’’colloid catalysis is the recovery of the catalyst. Although in some cases the nanoparticlecatalyst can be recycled several times by membrane filtration [151], most authors do notreport on catalyst recycling. One possible solution is the use of two-phase systems. Forexample, aqueous Rh colloids were shown to be effective hydrogenation catalysts in a two-phase system, where the water immiscible phase was olefin [152]. The colloidal catalystcould be readily separated and recycled. It was shown that surfactants could be used tolower the interfacial tension to improve the efficiency of the two-phase system [153,154].

Another possibility, taking advantage of the biphasic environment, is to use fluorousorganic solvents as the catalyst phase instead of water [155]. Crooks and coworkers pre-pared dendrimer-stabilized colloid catalysts soluble in the fluorous phase and used thecatalysts in hydrogenation [156] and in a Heck reaction [157]. In both cases the colloidalcatalyst in the fluorous phase was recyclable and showed some interesting selectivities andproducts unique to the nanoenvironment in the dendrimer interior.

The interface between two immiscible electrolyte solutions offers the means to com-bine two-phase catalysis, colloid catalysts, and electrocatalysis. In the study of Lahtinen etal. [158] citrate-stabilized palladium and gold colloids were prepared by a traditionalchemical reduction method. The voltammetric response of a system with an aqueouscolloid and an electron donor in the organic phase revealed an irreversible voltammetricwave as the potential was swept positive. The response was detected only in the presence ofboth the colloid and the electron-donor DCMFc. The response was concluded to resultfrom heterogeneous charging of the colloid with electrons from DCMFc.

The electrophoretic mobility of the particles was determined to confirm the buildupof negative charge on the colloid. In these measurements, the potential difference acrossthe liquid|liquid interface was controlled by potential-determining ions. It was shown thatthe charge on the colloid was dependent on the concentration of the electron-donorDCMFc. The results clearly showed that the metal colloid was charged in the two-phase process.

Finally, catalytic experiments were conducted in order to establish whether thecharged colloids can be used as an electron source in organic reactions. Dehalogenationof 2-bromoacetophenone (BrAc) was used as the model reaction. The experiments werecarried out in a similar fashion to the charging experiments. In addition to the electron-donor DCMFc and the supporting electrolyte, the organic phase contained BrAc. Themixture was stirred for 2 h to achieve a conversion to acetophenone. These results showedthe usefulness of this approach as a new type of two-phase catalysis. Figure 13 presents atentative catalytic cycle where the catalyst can be separately charged, brought in contactwith the substrate, stirred effectively, and, after the reaction, the catalyst can be easilyseparated from the reaction mixture and regenerated for another cycle.


B. Electrochemically Generated Particles as In Situ Catalysts at the ITIES

In the standard chemical preparation methods, the properties, especially the size and sizedistribution of the nanoparticles, are defined by the choice of the reaction conditions,reactant concentrations, etc. The use of electrochemical techniques to generate nucleihas the advantage that the supersaturation is determined by the applied potential orcurrent density. Thus, the size of the particles can be controlled by electrochemical instru-mentation rather than by changing the experimental conditions. Reetz and Helbig [115]demonstrated how electrochemical methods can be used to produce metal colloids ofnanometer size and more importantly how particle size can be controlled in a simplemanner by adjusting the current density [159]. First, a sacrificial anode was used as thesource of the metal ions, which were then reduced at the cathode. Later, a more generalapproach was introduced, where metal salts were used as the starting material [160]. Theparticles were stabilized by alkylammonium or betaine salts. With a suitable choice ofsurfactants, the electrochemical method can be applied in the preparation of differentshapes of particles, e.g., nanorods [161].

Metal nanoparticles can also be synthesized at a polarized liquid|liquid interface. Asa matter of fact, the first experimental evidence for heterogeneous electron transfer at anexternally biased ITIES featured the electrodeposition of copper and silver [162]. Morerecently, Cheng and Schiffrin [163] demonstrated the formation of gold nanoparticles atthe ITIES by reducing tetraoctylammonium tetrachloroaurate dissolved in DCE by aqu-

FIG. 13 Schematic representation of a two-phase reaction using colloids as electrocatalysts. The

cycle features as the first step the charging of the catalyst and as a second separate step the two-phase

reaction.


eous ferrocyanide. Spectroscopic measurements confirmed the generation of gold particles.Unfortunately, the transfer and adsorption of the tetrachloroaurate ions complicated thevoltammetric response, and systematic analysis of the nucleation reaction at the ITIESwas not possible.

The electrodeposition of palladium at the ITIES has been studied by Johans et al.[164]. The advantage over the gold system is the absence of ion-transfer signals in thepolarizable potential window. A model was presented for diffusion-controlled potentio-static electrodeposition at the ITIES, taking into account the development of diffusionfields in both phases. It should be emphasized that the experimental system has to bechosen with care. Ammonium tetrachloropalladate was used as the aqueous substrate andbutylferrocene as the reducing agent in the organic phase. In a cyclic voltammogram of thesystem, an irreversible reduction peak was formed at positive potentials, featuring a typicalnucleation loop. Comparison of experimental and simulated current transients providedgood correlations with classical concepts such as progressive and instantaneous nuclea-tion, and gave information on the number densities of the particles at the interface. Johanset al. [165] have also investigated galvanostatic electrodeposition. The authors developed ageneral model for three-dimensional nucleation. In this approach they incorporated theeffect of kinetics of the growth reaction into the model. The same model was used toinvestigate the influence of interfacial tension controlled by surfactants on nucleationkinetics by cyclic voltammetry and amperometry [166]. The position of a nanoparticleat a liquid|liquid interface was considered by classical thermodynamics. The theoreticalmodel indicated that while large particles preferentially reside in the interfacial region,small particles are expelled. This introduces a second critical radius in the nucleationexperiments. The experiments were in line with that prediction. There still are problemsto be solved concerning the use of the ITIES for preparing nanoparticles. In particular, thequestion of stabilization has to be carefully considered.

The catalytic properties of the deposited particles are yet to be systematically stu-died. In one rare account, Schiffrin and Cheng [167] reported the catalytic dehalogenationof 2-bromoacetophenone to acetophenone by an organic reductant in the presence ofelectrogenerated Pd particles. It was observed that the Pd particles exhibited interestingselective properties.

C. Two-Phase Photocatalysis in Presence of Metal and SemiconductingNanoparticles

As we mentioned in Section IV, current trends in photocatalysis are strongly biasedtoward photo-oxidation of organic substrates in the presence of semiconducting nanopar-ticles. The complete or partial mineralization of organic pollutants in aqueous media byTiO2 nanoparticles has proved to be commercially viable, and a significant impact on themultibillion A worth ‘‘clean technology’’ appears imminent. Beyond photo-oxidation pro-cesses, other reactions such as dehydrogenation and metal deposition as well as removal ofpollutants in the gas phase have been studied for a variety of semiconducting nanoparticles[101,168]. The literature in this area can be traced to numerous disciplines, and a com-prehensive review would be outside the scope of this section. We shall concentrate on somebasic aspects of photoinduced redox processes involving nanoparticles and their relevancein photocatalysis at liquid|liquid junctions.

One of the outstanding features of TiO2 as a photoactive material is the possibility ofwater splitting [169]. As indicated in Fig. 14, radiation with energies greater than the bandgap (3 eV) generates electron–hole pairs, which subsequently dissociate in the conduction


and valence bands, respectively. Dynamic photochemical studies indicate that valenceband holes are readily captured by adsorbed OH� groups, generating surface OH thatbehaves as an intermediate in photo-oxidation processes. Hole capture effectively com-petes with the nonradiative recombination with conduction band electrons. Furthermore,the electron-capture cross-section of the surface radical OH is rather small, allowing thepossibility of H2O2 formation at the surface or the oxidation of a species in solution.Photogenerated electrons can reduce molecular oxygen, although this reaction is ratherslow. The initial steps in photoredox processes can be represented by

X h�! hþVB þ e�CB Photoionization of a nanocrystal site ð24ÞhþVB þOH�S�!OH:

S Hole capture by surface OH� groups ð25Þe�CB þOHS�!OH�S Surface recombination via OH ð26Þ2OHS�! H2O2ð Þs Surface OH coupling ð27Þ

FIG. 14 (a) Redox potentials for valence and conduction bands of TiO2 in comparison with the

potentials for hole capture by water and electron capture by oxygen. (b) Schematic diagram of the

initial stages of photoinduced water splitting at TiO2 nanoparticles modified by RuO2 and Pd

clusters.


e�CB þO2�!O�2 Electron capture by oxygen ð28ÞOHS þRed�!OH�S þOxþ Oxidation step via surface OH ð29ÞThe surface-confined H2O2 can further react with OH ultimately to generate oxygen [170–173]. Recent studies on rutile [174] and anatase [175] single-crystal photoelectrodes demon-strate that surface recombination mainly occurs via step (26). In addition, the recombina-tion step appears strongly inhibited in the presence of organic species such as formic acidand methanol, further confirming that photo-oxidation processes occurs via OH.

In colloidal suspensions, the overall efficiency of photoreactions at the TiO2 surfaceis determined by the removal of conduction band electrons. In order to catalyze step (28)or even hydrogen evolution, Pd and Pt clusters have been deposited at the surface of TiO2

nanoparticles [99,176,177]. The basic principle of catalytic centers on TiO2 particles is alsoillustrated in Fig. 14. Photosplitting of water has been achieved by further depositingclusters of RuO2 on TiO2 particles, which is able to accelerate steps (25), (27), and even-tually O2 evolution. However, under certain conditions the metal–semiconductor bound-ary can also play the role of recombination center.

Metal oxide nanoparticles are usually stabilized in aqueous media by electrostaticinteractions. By fixing the pH at values higher or lower than the characteristic pH of zerozeta potential, the particle surface exhibits a net positive or negative charge, respectively.Recent studies have demonstrated that interfacial concentration of TiO2 nanoparticles atwater|DCE junctions can be effectively tuned by the Galvani potential difference [178]. AtpH 2, the particles are positively charged and interfacial accumulation is achieved byapplying a positive Galvani potential difference with respect to the organic phase.Under these conditions, heterogeneous photo-oxidation of species located in the organicphase can be induced under UV illumination. Photocurrent responses originating from thephoto-oxidation of ferrocene (Fc) by TiO2 at the waterjDCE interface have been observed[178]. These photoresponses exhibit features similar to those observed for dye-sensitizedliquidjliquid junctions (see Fig. 5), but with negligible back electron transfer. The photo-oxidation of ferrocene can be interpreted in terms of Eqs (24) to (29), considering that thelatter step is in competition with the recombination and the radical coupling processes. Atbasic pH, the accumulation of particles takes place at negative potentials, and the hetero-geneous transfer of conduction band electrons toward acceptors in the organic phase isreadily observed. This exciting approach opens the possibility of using photoelectrochem-ical means to study a variety of photocatalytic reactions where the substrate is insoluble inpolar media.

Recent results have also shown that the photoreactivity of TiO2 at liquidjliquidjunctions can be extended into the visible region via dye sensitization [178]. Dyes speciessuch as alizarin exhibit a strong affinity for metal oxide surfaces, and ultrafast injection ofelectrons into the particles has been observed on photoexcitation [58–63]. In the case ofheterogenous reactions at waterjDCE junctions, the photoinduced electron injection isfollowed by electron transfer from the donor in DCE to the oxidized dye. The regenerationof the dye, also known as supersensitization, manifests itself as a photocurrent response.The basic principle is illustrated in Fig. 15. These processes are somewhat analogous to thephenomena responsible for the photoeffects in dye-sensitized nanocrystalline solar cells(DSNC) [179,180]. The main difference is that the photoresponses in DSNC arise from thetransport of injected electrons across the TiO2 mesoporous film, while in this case thephotocurrents originate from the supersensitization step.

Sensitization phenomena have also been observed in the presence of metallic nano-particles. In this case, the nanoparticles act as an electron-transfer relay, enhancing the


charge separation due to the favorably situated energy levels [181]. Colloidal noble metalcatalysts have been used in the reduction of carbon dioxide [182,183] and decompositionof water [184–187] as well as in various hydrogenations [188]. Bimetallic nanoparticlesshowed higher catalytic activity in the decomposition of water compared to the mono-metallic colloids [189]. Figure 16 illustrates the mechanism for visible light-induced hydro-gen generation catalyzed by a bimetallic colloid. In this case, the excited state of theruthenium complex injects an electron into the methyl viologen ion ðMV2þ), which issubsequently transferred to the nanoparticle where hydrogen evolution takes place.

FIG. 15 Schematic diagram of the photoinduced electron transfer reaction at liquid|liquid

interfaces featuring dye-sensitized semiconducting nanoparticles.

FIG. 16 Schematic illustration of the mechanism of light-induced hydrogen generation catalyzed

by Au–Pt bimetallic nanoparticles. (From Ref. 189.)


We have also demonstrated that metallic nanoparticles can act as redox relays forphotoinduced heterogeneous redox processes at waterjDCE interfaces [190]. Electro-generated particles prepared by heterogeneous reduction of tetrachloropalladate asdescribed in Section IV.B exhibit catalytic effects toward electron transfer from photo-excited porphyrins to redox species located in the organic phase. This catalytic effect of Pdnanoparticles is illustrated in Fig. 17(a), where photocurrent responses are measured in thepresence of the water-soluble porphyrin ZnTPPC and a DCE solution containing theredox donor ferrocene and the acceptor TCNQ [190]. The positive photocurrent in theabsence of the palladate salt indicates that the excited state of the dye is more effectivelyquenched by the donor species, especially at positive potentials. Even in an excess ofTCNQ, the negative photoresponses are small and located at rather negative potentials.On addition of the palladate complex, the photocurrent responses became negative over alarge range of potentials, indicating a preferential photoreduction process. This sign rever-sal is only observed in the presence of TCNQ; therefore, it is not directly connected withthe nucleation process involving the palladate ion and ferrocene.

FIG. 17 Photocurrent responses originating from the heterogeneous quenching of the water-

soluble porphyrin ZnTPPC by TCNQ and Fc at the waterjDCE junction. The reductive

quenching by Fc (positive photocurrent) is more efficient than the oxidative quenching by TCNQ

(negative photocurrent), even for 10 times greater concentration of the latter. On addition of PdCl2�4 ,

the sign of the photocurrent is reversed over a wide potential range (a). The in-situ generated Pd

nanoparticles act as mediators for the photoinduced electron transfer from the porphyrin excited

state to the redox acceptor, as illustrated in (b). Reprinted from Ref. 190 with permission from

Elsevier Science.


The photocatalytic effect of dye-sensitized Pd nanoparticles is schematically depictedin Fig. 17(b). At positive potentials, the ferrocene reduces the palladate complex in thedark to generate Pd nanoparticles at the liquid|liquid boundary. These nanoparticles act aselectron-capture sites at the interface, which are subsequently transferred to the electronacceptor in the organic phase. This behavior is surprising in the sense that injected elec-trons in metallic particles are expected to have a very short lifetime due to fast backelectron transfer to the oxidized dye, or even hydrogen generation at the particle surface.A possible rationale for the mechanism highlighted in Fig. 17(b) is linked to adsorption ofCl� at the particle surface, which competes with the adsorption of Hþ, decreasing the rateof hydrogen evolution. Furthermore, the Cl� adsorption introduces a negative surfacecharge that can electrostatically repel the anionic dye.

VI. CONCLUSIONS

The various catalytic and photocatalytic processes in two-phase systems highlighted in thischapter allows evaluation of the potential impact of electrochemistry at the ITIES in thisarea. In a general sense, the possibility of tuning the interfacial concentration of chargedspecies and the reactivity of the whole system by controlling the electrical potential acrossthe interface opens an effective way of controlling reaction mechanisms and rates.Furthermore, electrochemical techniques provide direct access to the rate of charge trans-fer across the interfacial region. In conjunction with surface-sensitive spectroscopic tech-niques and modern computational methods [55,191–194], our understanding of thestructure and reactivity at liquid|liquid interfaces at the molecular level has remarkablyincreased in the last few years. These developments provide the ideal platform for under-standing and developing catalytic processes at molecular junctions. In order fully to realizethe potential of the electrochemical methods it would be advantageous in the future to seea more multidisciplinary approach, bringing together electrochemists and synthetic che-mists.

ACKNOWLEDGMENTS

R.L. is grateful for the Marie Curie Fellowship of the European Community program‘‘Improving Human Research Potential & Socio-Economic Knowledge Base,’’ contractnumber HPMF-CT-2000-00804. H.J. and D.J.F. also acknowledge the support by theEcole Polytechnique Federale de Lausanne and the Fonds Nationale Suisse de laRecherche Scientifique (Project 20-55692.98). The Laboratories of Electrochemistry ofthe University of Liverpool and EPFL are part of the European TMR networkSUSANA (Supramolecular Self-Assembly of Interfacial Nanostructures).

REFERENCES

1. LQ Hung. J Electroanal Chem 115:159–174, 1980.

2. LQ Hung. J Electroanal Chem 149:1–14, 1983.

3. T Kakiuchi. Anal Chem 68:3658–3664, 1996.

4. Z Samec, V Marecek, J Weber. J Electroanal Chem 96:245–247, 1978.

5. LQ Hung, P Vanysek. Chem Listy 74:869–873, 1980.


6. Z Ding PhD thesis. Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland, 1999.

7. DJ Fermın, H Duong, Z Ding, P-F Brevet, HH Girault. J Am Chem Soc 121:10203–10210,

1999.

8. C Gavach, B D’Epenoux, F Henry. J Electroanal Chem 64:107–115, 1975.

9. C Gavach, F Henry, R Sandeaux. CR Acad Sci, Ser C 278:491–494, 1974.

10. C Gavach, B D’Epenoux. J Electroanal Chem 55:59–67, 1974.

11. C Gavach. J Chim Phys Physicochim Biol 70:1478–1482, 1973.

12. Z Samec. J Electroanal Chem 99:197–205, 1979.

13. HHJ Girault, DJ Schiffrin. J Electroanal Chem 170:127–141, 1984.



16. Y Shao, HH Girault. J Electroanal Chem 282:59–72, 1990.

17. Y Shao, JA Campbell, HH Girault. J Electroanal Chem 300:415–429, 1991.

18. T Osakai, T Kakutani, M Senda. Bull Chem Soc Jpn 58:2626–2633, 1985.

19. T Wandlowski, V Marecek, Z Samec. J Electroanal Chem 242:291–302, 1988.

20. T Kakiuchi, J Noguchi, M Kotani, M Senda. J Electroanal Chem 296:517–535, 1990.

21. T Kakiuchi, J Noguchi, M Senda. J Electroanal Chem 327:63–71, 1992.

22. T Kakiuchi. J Electroanal Chem 322:55–61, 1992.

23. T Kakiuchi. Denki Kagaku oyobi Kogyo Butsuri Kagaku 63:358–363, 1995.

24. DJ Fermın, H Jensen, HH Girault. In: AJ Bard and M Stratmann, ed. Encyclopaedia of

Electrochemistry. 2. in press.

25. J Koryta. Electrochim Acta 24: 293–300, 1979.

26. M Senda, T Kakiuchi, T Osakai. Electrochim Acta 36:253–262, 1991.

27. HH Girault. Mod Aspects Electrochem 25:1–62, 1993.

28. PD Beattie, A Delay, HH Girault. J Electroanal Chem 380:167–175, 1995.

29. PD Beattie, A Delay, HH Girault. Electrochim Acta 40:2961–2969, 1995.

30. JA Campbell, HH Girault. J Electroanal Chem 266:465–469, 1989.

31. JA Campbell, AA Stewart, HH Girault. J Chem Soc, Faraday Trans 1 85:843–853, 1989.

32. B Liu, MV Mirkin. Electroanalysis 12:1433–1446, 2000.

33. B Quinn, R Lahtinen, K Kontturi. J Electroanal Chem 436:285–290, 1997.

34. T Kakiuchi, Y Takasu, M Senda. Anal Chem 65:1123, 1993.

35. T Kakiuchi, Y Takasu, M Senda. Anal Chem 64:3096–3100, 1992.

36. T Kakiuchi, Y Takasu. J Electroanal Chem 381:5–9, 1995.

37. T Kakiuchi, Y Takasu. Anal Chem 66:1853–1859, 1994.

38. T Kakiuchi, K Ono, Y Takasu, J Bourson, B Valeur. Anal Chem 70:4152–4156, 1998.

39. N Nishi, K Izawa, M Yamamoto, T Kakiuchi. J Phys Chem B 105:8162–8169, 2001.

40. DJ Fermın, Z Ding, PF Brevet, HH Girault. J Electroanal Chem 447:125–133, 1998.

41. ZF Ding, F Reymond, P Baumgartner, DJ Fermın, PF Brevet, PA Carrupt, HH Girault.

Electrochim Acta 44:3–13, 1998.

42. H Nagatani, RA Iglesias, DJ Fermın, PF Brevet, HH Girault. J Phys Chem B 104: 6869–6876,

2000.

43. H Nagatani, DJ Fermın, HH Girault. J Phys Chem B 105:9463–9473, 2001.


45. M Tsionsky, AJ Bard, MV Mirkin. J Phys Chem 100:17881–17888, 1996.

46. M Tsionsky, AJ Bard, MV Mirkin. J Am Chem Soc 119:10785–10792, 1997.

47. J Zhang, PR Unwin. J Phys Chem B 104:2341–2347, 2000.

48. ZF Ding, DJ Fermin, PF Brevet, HH Girault. J Electroanal Chem 458:139–148, 1998.

49. RA Marcus. J Phys Chem 94:4152–4155, 1990.

50. RA Marcus. J Phys Chem 94:7742, 1990.



53. YI Kharkats. Sov Electrochem 12:1257–1263, 1976.

54. YI Kharkats, AG Volkov. J Electroanal Chem 184:435–442, 1985.


55. W Schmickler. J Electroanal Chem 428:123–127, 1997.

56. DJ Fermın, R Lahtinen. In: A. Volkov, ed. Liquid Interfaces in Chemical, Biological, and

Pharmaceutical Applications. 95. Boca Raton, FL: Marcel Dekker, 2001, pp 179–228.

57. K Kalyanasundaram. Photochemistry of Polypyridine and Porphyrin Complexes. London:

Academic Press, 1992.

58. Y Tachibana, JE Moser, M Gratzel, DR Klug, JR Durrant. J Phys Chem 100:20056–20062,

1996.

59. JE Moser, MWolf, F Lenzmann, M Gratzel. Z Phys Chem—Int J Res Phys Chem Chem Phys

212:85–92, 1999.

60. JE Moser, M Gratzel. Chimia 52:160–162, 1998.

61. R Huber, S Sporlein, JE Moser, M Gratzel, J Wachtveitl. J Phys Chem B 104:8995–9003,

2000.

62. JR Durrant, Y Tachibana, JE Moser, M Gratzel, DR Klug. Proc Indian Acad Sci—Chem Sci

109:411–414, 1997.

63. JR Durrant, Y Tachibana, I Mercer, JE Moser, M Gratzel, DR Klug. Z Phys Chemie—Int J

Res Phys Chem Chem Phys 212:93–98, 1999.

64. NA Kotov, MG Kuzmin. J Electroanal Chem 285:223–347, 1990.

65. NA Kotov, MG Kuzmin. J Electroanal Chem 341:47–60, 1992.

66. NA Kotov, MG Kuzmin. J Electroanal Chem 338: 99–124, 1992.

67. DJ Fermın, Z Ding, HD Duong, PF Brevet, HH Girault. J Chem Soc, Chem Commun

1125–1126, 1998.

68. DJ Fermın, Z Ding, H Duong, PF Brevet, HH Girault. J Phys Chem B 102:10334–10341,

1998.

69. DJ Fermın, H Duong, Z Ding, PF Brevet, HH Girault. Phys Chem Chem Phys 1:1461–1467,

1999.

70. N Eugster, DJ Fermın, HH Girault. J Phys Chem B 106:3428–3433, 2002.

71. R Lahtinen, DJ Fermın, K Kontturi, HH Girault. J Electroanal Chem 483:81–87, 2000.

72. H Jensen, DJ Fermın, HH Girault. Phys Chem Chem Phys 3:2503–2508, 2001.

73. H Jensen, JJ Kakkassery, H Nagatani, DJ Fermın, HH Girault. J Am Chem Soc

122:10943–10948, 2000.

74. H Nagatani, A Piron, PF Brevet, DJ Fermın, HH Girault. Langmuir in press.

75. H Nagatani, DJ Fermın, Z Samec, HH Girault. In preparation.

76. B Cornils, WA Herrmann. Aqueous-Phase Organometallic Catalysis. Concepts and

Applications. Weinheim: Wiley-VCH, 1998.

77. PJ Dyson, D Ellis, T Welton. Platinum Met Rev 42:135–140, 1998.

78. WA Herrmann, CW Kohlpaintner. Angew Chem, Int Ed Engl 32:1524–1544, 1993.

79. AG Volkov. Electrochim Acta 44:139–153, 1998.

80. AG Volkov, DW Deamer, DI Tanelian, VS Markin. Liquid Interfaces in Chemistry and

Biology. New York: John Wiley, 1998.

81. AG Volkov, ed. Liquid Interfaces in Chemical, Biological, and Pharmaceutical Applications.

Boca Raton, FL: Marcel Dekker, 2001.


83. Y Shao, M Mirkin, JF Rusling. J Phys Chem B 101:3202–3208, 1997.

84. YT Kong, T Kakiuchi. J Electroanal Chem 446:19–23, 1998.

85. TD Chung, FC Anson. J Electroanal Chem 508:115–122, 2001.

86. T Kakiuchi, M Senda. Bull Chem Soc Jpn 56:1753–1760, 1983.

87. T Kakiuchi, M Senda. Bull Chem Soc Jpn 56:2912–2918, 1983.

88. VJ Cunnane, DJ Schiffrin, C Beltran, G Geblewicz, T Solomon. J Electroanal Chem 247:203–

214, 1988.

89. SN Tan, RA Dryfe, HH Girault. Helv Chim Acta 77:231–242, 1994.

90. YT Kong, S Imabayashi, T Kakiuchi. J Am Chem Soc 122:8215–8219, 2000.

91. C Forssten, K Kontturi, L Murtomaki, HC Hailes, DE Williams. Electrochem Commun 3:

379–383, 2001.


92. C Forssten, J Strutwolf, DE Williams. Electrochem Comm 3:619–623, 2001.

93. C Forssten, DEWilliams, J Strutwolf. Phase transfer catalysis across liquidjliquid interfaces in

organic reactions. Proceedings of Euroconference on Modern Trends in Electrochemistry of

Molecular Interfaces, Kirkkonummi, Finland, 1999, pp P-13.

94. JN Robinson, DJ Cole-Hamilton. Chem Soc Rev 20:49–94, 1991.

95. K Kalyanasundaram. Photochemistry of Michoheterogeneous Systems. London: Academic

Press, 1987.

96. GJ Kavarnos. Fundamentals of Photoinduced Electron Transfer. New York: VCH

Publishers, 1993.

97. JK Hurst, RF Khairutdinov. In: V. Balzani, ed. Electron Transfer in Chemistry. 4. Weinheim:

Wiley-VCH, 2001, pp 578–618.

98. JM Herrmann. Catal Today 53:115–129, 1999.

99. B Levy. J Electroceram 1:239–272, 1997.

100. JM Herrmann. Catal Today 24:157–164, 1995.

101. A Fujishima, K Hashimoto, T Watanabe. TiO2 Photocatalysis. Fundamentals and

Applications. Tokyo: BKC, 1999.

102. F Herrera, C Pulgarin, V Nadtochenko, J Kiwi. Appl Catal B: Environ 17:141–156, 1998.

103. J Kiwi, C Pulgarin, P Peringer. Appl Catal B:Environ 3:335–350, 1994.

104. R Bauer, G Waldner, H Fallman, S Hager, M Klare, T Krutzler, S Malato, P Maletzky. Catal

Today 53:131–144, 1999.

105. P-A Brugger, M Gratzel. J Am Chem Soc 102:2461–2463, 1980.

106. I Willner, WE Ford, JW Otvos, M Calvin. Nature 280:823–824, 1979.

107. I Tabushi, S Kugimya, T Mizutani. J Am Chem Soc 105:1658–1659, 1983.

108. I Willner, E Joselevich. J Phys Chem B 103:9262–9268, 1999.

109. DJ Fermın, H Doung, Z Ding, PF Brevet, HH Girault. Electrochem Commun 1:29–32, 1999.

110. A Henglein. J Phys Chem 97: 5457–5471, 1993.

111. RW Devenish, T Goulding, BT Heaton, R Whyman. J Chem Soc, Dalton Trans 673–679,

1996.

112. VF Puntes, KM Krishnan, AP Alivisatos. Science 291:2115–2117, 2001.

113. N Arul Dhas, H Cohen, A Gedanken. J Phys Chem B 101:6834–6838, 1997.

114. N Arul Dhas, A Gedanken. J Mater Chem 8:445–450, 1998.

115. MT Reetz, W Helbig. J Am Chem Soc 116:7401–7402, 1994.

116. YY Yu, SS Chang, CL Lee, CRC Wang. J Phys Chem B 101:6661–6664, 1997.

117. H Hirai, Y Nakao, N Toshima. Sci Chem A13:727–750, 1979.

118. T Teranishi, M Miyake. Chem Mater 11:3414–3416, 1999.

119. J Turkevich, G Kim. Science 169:873–879, 1970.

120. G Schmid, H West, JO Malm, JO Bovin, C Grenthe. Chem Eur J 2:1099–1103, 1996.

121. TS Ahmadi, ZL Wang, TC Green, A Henglein, MA El-Sayed. Science 272:1924–1926, 1996.

122. Y Okuno, Y Chiba, Y O. Chem Lett 893–896, 1983.

123. S Sun, CB Murray. J Appl Phys 85:4325–4330, 1999.

124. Y Li, MA El-Sayed. J Phys Chem B 105:8938–8943, 2001.

125. HA Wievenga, L Soethout, IW Gerritsen, BEC van de Leemput, H van Kempen, G Schmid.

Adv Mater 2:482–484, 1990.

126. YH Chen, U Nickel. J Chem Soc, Faraday Trans 89:2479–2485, 1993.

127. JM Thomas. Pure Appl Chem 60:1517–1528, 1988.

128. LN Lewis. Chem Rev 93:2693–2730, 1993.

129. HP Choo, KY Liew, HF Liu, CE Seng. J Mol Catal A 165:127–134, 2000.

130. W Tu, H Liu, Y Tang. J Mol Catal A 159:115–120, 2000.

131. HY Ha, MA Anderson. J Environ Eng—ASCE 122:217–221, 1996.

132. H Hirai, H Chawanya, N Toshima. Bull Chem Soc Jpn 58:682–687, 1985.

133. ABR Mayer, JE Mark. J Macromol Sci A 33:451–459, 1996.

134. C Chen, T Serizawa, M Akashi. Chem Mater 11:1381–1389, 1999.

135. W Yu, M Liu, Y Liu, X Ma, Z Liu. J Colloid Interface Sci 209:439–444, 1998.


136. H Bonnemann, GA Braun. Chem Eur J 3:1200–1202, 1997.

137. A Eichhorn, A Koch, J Bargon. J Mol Catal A 174:249–274, 2001.

138. M Beller, H Fischer, K Kuhlein, CP Reisinger, WA Herrmann. J Organomet Chem

520:257–259, 1996.

139. MT Reetz, G Lohmer. Chem Commun 1921–1922, 1996.

140. A Biffis, M Zecca, M Basato. J Mol Catal A 173:249–274, 2001.

141. H Bonnemann, RM Richards. Eur J Inorg Chem 2455–2480, 2001.

142. LN Lewis, N Lewis. J Am Chem Soc 108:7228–7231, 1986.

143. X Jiang, Y Xie, J Lu, L Zhu, W He, Y Qian. Langmuir 17:3795–3799, 2001.

144. TK Sau, A Pal, T Pal. J Phys Chem B 105:9266–9272, 2001.

145. T Ung, LM Liz-Marzan, P Mulvaney. J Phys Chem B 103:6770–6773, 1999.

146. RM Crooks, M Zhao, L Sun, V Chechik, LK Yeung. Acc Chem Res 34:181–190, 2001.

147. N Toshima, T Yonezawa. New J Chem 22:1179–1201, 1998.

148. N Toshima, T Yonezawa, K Kushihashi. J Chem Soc, Faraday Trans 89:2537–2543, 1993.

149. N Toshima, M Harada, Y Yamazaki, K Asakura. J Phys Chem 96:9927–9933, 1992.

150. H Bonnemann, GA Braun, W Brijoux, R Brinkamann, A Schulze, K Tilling, K Seevogel, K

Siepen. J Organomet Chem 520:143–162, 1996.

151. Y Wang, H Liu, N Toshima. J Phys Chem 100:19533–19537, 1996.

152. C Larpent, F Brisse-le Menn, H Patin. J Mol Catal 116: L35–L40, 1991.

153. C Larpent, E Bernard, F Brisse-le Menn, H Patin. J Mol Catal 116:277–288, 1997.

154. C Larpent, E Bernard. J Chem Soc, Chem Commun 535–536, 1992.

155. IT Horvath, J Raba. Science 266:72–75, 1994.

156. V Chechik, RM Crooks. J Am Chem Soc 122:1243–1244, 2000.

157. LK Yeung, RM Crooks. Nano Lett 1:14–17, 2001.

158. R Lahtinen, CRC Johans, D Coleman, K Kontturi. Electrochem Comm 4:479–482, 2002.

159. MT Reetz, W Helbig, SA Quaiser, U Stimming, N Breuer, R Vogel. Science 267:367–369,

1995.

160. MT Reetz, SA Quaiser. Angew Chem, Int Ed Engl 34:2240–2241, 1995.

161. JJ Zhu, XH Liao, XN Zhao, HY Chen. Mater Lett 49:91–95, 2000.

162. M Guainazzi, G Silvestri, G Serravalle. Chem Commun 200–201, 1975.


164. CRC Johans, R Lahtinen, K Kontturi, DJ Schiffrin. J Electroanal Chem 488:99–109, 2000.

165. CRC Johans, K Kontturi, DJ Schiffrin. J Electroanal Chem 526:29–35, 2002.

166. CRC Johans, P Liljeroth, K Kontturi. Phys Chem Chem Phys 4:1067–1071, 2002.

167. DJ Schiffrin, Y Cheng. Metallic Clusters Electrodeposited at the Waterj1,2-dichloroethaneinterface. Proceedings of Charge Transfer at Liquid|Liquid and Liquid|Membrane Interface,

Kyoto, 1996, pp 61–63.

168. X Fu, WA Zeltner, MA Anderson. Stud Surs Sci Catal 103:445–461, 1997.

169. A Fujishima, K Honda. Nature 238:37, 1972.

170. D Tafalla, P Salvador. Ber Bunsen—Ges Phys Chem Chem Phys 91:475–479, 1987.

171. D Tafalla, P Salvador, RM Benito. J Electrochem Soc 137:1810–1815, 1990.

172. P Salvador, MLG Gonzalez, F Munoz. J Phys Chem 96:10349–10353, 1992.

173. P Salvador. J Phys Chem 89:3863–3869, 1985.

174. DJ Fermın, EA Ponomarev, LM Peter. Dynamics of photo-processes at the n-TiO2jaqueouselectrolyte interface. ECS Proceeding of the Symposium on Photoelectrochemistry, Paris,

1997, pp 62–71.

175. OA Semenikhin, VE Kazarinov, L Jiang, K Hashimoto, A Fujishima. Langmuir 15:3731–

3737, 1999.

176. T Sakata. In: N Serpone, E Pelizzetti, ed. Photocatalysis: Fundamentals and Applications.

New York: John Wiley, 1989, pp 311.

177. M Gratzel. Heterogeneous Photochemical Electron Transfer. Boca Raton, FL: CRC Press,

1988.

178. H Jensen, DJ Fermın, JE Maser, HH Girault. J Phys Chem B submitted.


179. B O’Reagan, M Gratzel. Nature 353:737–739, 1991.

180. M Gratzel. Platinum Met Rev 38:151–159, 1994.

181. A Henglein. Ber Bunsengen Phys Chem 101:1562–1572, 1997.

182. I Willner, R Maidan, D Mandler, H Durr, G Dorr, K Zengevle. J Am Chem Soc 109:6080–

6086, 1987.

183. H Durr, HP Trierweiler, I Willner, R Maidan. New J Chem 14:317–320, 1990.

184. J Kiwi, M Gratzel. J Am Chem Soc 101:710–711, 1979.

185. Y Degani, I Willner. J Chem Soc, Chem Commun 710–711, 1983.

186. S Yanagida, T Ogata, Y Kuwana, YWada, KMurakoshi, A Ishida, S Takamuka, M Kusaba,

N Nakashima. J Chem Soc, Perkin Trans 2:1963–1969, 1996.

187. S Matsuoka, H Fujii, T Yamada, C Pac, A Ishida, S Takamuku, M Kusaba, N Nakashima, S

Yanagida, K Hashimoto, T Sakata. J Phys Chem 95:5802–5808, 1991.

188. D Mandler, I Willner. J Phys Chem 91:3600–3605, 1987.

189. N Toshima, K Hirakawa. Appl Surface Sci 121/122:534–537, 1997.

190. R Lahtinen, DJ Fermın, H Jensen, K Kontturi, HH Girault. Electrochem Commun

2:230–234, 2000.

191. I Benjamin. ACS Symp Ser 568:409–422, 1994.

192. I Benjamin. Chem Rev 96:1449–1475, 1996.

193. I Benjamin. Annu Rev Phys Chem 48:407, 1997.

194. W Schmickler. J Electroanal Chem 426:5–9, 1997.


23Photosensitizers at Interfaces of ModelMembranes

SARAH A. GERHARDT and JIN Z. ZHANG University of California at SantaCruz, Santa Cruz, California, U.S.A.

I. INTRODUCTION

Micelles and liposomes are classified as association colloids. They are composed of sur-face-active agents, otherwise known as surfactants, which are amphiphillic moleculescontaining both hydrophobic and hydrophilic moieties. Some of the unique propertiesthat make such colloidal systems interesting and useful include increased microviscosity,micropolarity, slower solvation dynamics, enhancement of molecular excited-state life-times, changes in reaction rates, and solubilization of particles [1–10]. The applicationsof colloidal systems span many fields from physics and chemistry to biology and medicine.In particular, micelles and liposomes are used to model biological membrane interfaces[11–18]. These model studies are crucial for providing a fundamental understanding ofhow molecules, such as various drugs, incorporate and localize into membranes and how amembrane interface affects the functionality of the drug molecules. They also provide abasis for development of useful drug delivery systems [19–27].

Photodynamic therapy (PDT) agents are among the many drugs studied in micelleand liposome model membrane systems. Photodynamic therapy is a novel medical tech-nique currently used to treat various cancers [28–31]. It is also used as a blood sterilizationmethod to remove viruses such as hepatitis and HIV from blood for transfusion purposes[32–41]. The methodology of PDT involves at first selective uptake and retention ofphotosensitizers in a tumor and then excitation of the localized photosensitizers by illu-mination with visible laser light, creating excited singlet-state sensitizer molecules. A sen-sitizer molecule in the excited singlet state can decay back to its ground state by emitting aphoton or by crossing to a triplet state, followed by energy transfer from the excited tripletstate of the sensitizer to the ground state of oxygen to generate singlet oxygen. Theresulting singlet-state oxygen is believed to be responsible for the therapeutic action ofPDT [42–44].

Interaction between sensitizers and membrane interfaces can dramatically alter theirlocalization and photophysical properties, which in turn significantly affects drug efficacy.Studies using model membranes, such as micelles and liposomes, provide importantinsight into the effect of biological membranes on the properties of sensitizers as well asthe generation and reaction of singlet oxygen [45–53]. This chapter focuses on the uniqueinterfacial properties and applications of micelles and liposomes as model membranesystems used to probe the interaction of PDT sensitizers and membranes. The effect of


model membrane interfaces on the photophysics of photosensitizers will be illustrated withexamples, including the equilibrium between aggregates and monomers, excited-state sing-let and triplet lifetimes of the photosenitizers, location of photosensitizer molecules nearthe interface, reaction rate between the triplet-state sensitizer and ground-state tripletoxygen, and the rate as well as yield of singlet oxygen generation.

II. PHOTOPHYSICS OF PHOTOSENSITIZERS IN MODEL MEMBRANES

A. Influence of Model Membranes on Sensitizer Aggregation

Optimum sensitizers for PDT have the following qualities: high quantum yield of triplet-state formation, high singlet-oxygen quantum yield, and preferential tumor cell selectivity[54]. Critical solvation issues for the photosensitizers, such as localization and monodis-persion versus aggregation, are important to understand since they are directly related tomeeting the requirements for optimum PDT activity. In particular, understanding sensi-tizer interactions with membrane interfaces is a necessary first step toward developinguseful photodynamic therapy agents.

For instance, zinc phthalocyanine tetrasulfonate (ZnPcS4) has been investigated asa potential PDT photosensitizer in micelles, reverse micelles, and liposomes to under-stand how model membranes would affect its location and photophysical properties [55–57]. The ground-state absorption and emission spectra provide insight into the locationof ZnPcS4 in micelle systems. Figure 1 shows the effect of cetyltrimethylammoniumbromide (CTAB) micelles and human serum albumin (HSA) on the ground-state absorp-tion of ZnPcS4. Compared to the absorption spectrum in pure water, a dramatic redshift of the Q band of ZnPcS4 has been observed with CTAB. A similar, but lessdramatic, effect was observed for HSA. Likewise, while the fluorescence spectra ofZnPcS4 showed no observable signal in pure water, its fluorescence was significantlyenhanced when CTAB was added at concentrations above the critical micelle concentra-tion. It has been established that ZnPcS4 exists mainly as aggregates in pure water,especially at high concentration, while primarily as monomers in organic solvents[55,58]. The aggregates have a red-shifted Q band and low fluorescence yield. Thiscorresponds to low triplet sensitizer yield and thereby low singlet oxygen yield that isundesirable for PDT applications. Therefore, the red shift of the Q band of the absorp-tion spectrum and enhancement of fluorescence with CTAB or HSA have been attrib-uted to monomer formation or disaggregation as a result of sensitizer–micelleinteraction. In this particular case, the interaction is mainly electrostatic since theCTAB micelles are positively charged and the sensitizers are negatively charged. Thiseffect was not observed when negatively charged micelles were used.

The measurement of ZnPcS4 excited-state dynamics in micelle and liposome systemsprovides further insight into the sensitizer–micelle interaction. Transient absorption stu-dies of ZnPcS4 in pure water have shown that excited-state dynamics are fast (< 80 ps) anddependent on the sensitizer concentration, faster at higher concentration [57]. This isconsistent with aggregate formation in pure water. In organic solvents such as dimethylsulfoxide (DMSO), the excited singlet-state lifetime is much longer (> 400 ps), in accordwith the strong room-temperature fluorescence observed. Interestingly, studies in micelles[57] and liposomes [59,60] show a dramatic effect on the solvation and excited-statedynamics as well as singlet oxygen generation. The excited singlet-state lifetime ofZnPcS4 was found to increase significantly. Figure 2 shows the effect of CTAB on theZnPcS4 ultrafast transient absorption signal. The signal shows normal transient absorp-


FIG. 1 Ground-state absorption spectra of 0.12 mM ZnPcS4 in pure water, 10 mM HSA, and 1

mM CTAB. The top panel shows the change in the Q band. The bottom panel shows the absorption

spectrum from 300 to 800 nm. (From Ref. 57.)

FIG. 2 Time-resolved transient absorption profile of 0.12 mM ZnPcS4 with CTAB excited at 390

nm and probed at 720 nm: (a) water only (&), (b) 0.5 mM CTAB (*), (c) 2.0 mM CTAB (*), (d)

3.0 mM CTAB (&). (From Ref. 57.)


tion in pure water with a lifetime < 100 ps (a). When CTAB is added, the lifetime increases(up to > 1 ns) with increasing CTAB concentration (b–d) [61]. This lifetime is comparableor even slightly longer than that in DMSO. This dramatic increase in the excited singlet-state lifetime of the sensitizer with CTAB was attributed to monomer formation inducedby CTAB–sensitizer interaction. These dynamics results support the conclusion of CTAB-induced disaggregation based on electronic absorption and fluorescence studies. Theobserved short excited-state lifetime in water, as a consequence of aggregation, is unfavor-able for triplet-state sensitizer generation and singlet oxygen production. Therefore,ZnPcS4 would seem unfavorable for PDT application based on the results in pure water[61]. However, the monomer formation as a result of disaggregation induced by modelmembranes such as CTAB indicates that ZnPcS4 is still potentially useful for PDT appli-cations if similar disaggregation can be induced by biological tissues. Since a similarobservation, though less dramatic, was made for ZnPcS4 in the presence of HSA, itdoes seem likely that biological membranes may have a similar effect [57]. Further studiesinvolving biological tissues are clearly necessary to establish how general this observationis and whether the results obtained from model membranes studies can indeed be extendedto real biological systems.

Similar observations have been made for other porphyrin and porphyrinoid photo-sensitizers. For instance, several groups have observed that tetra-(4-sulfanatophenyl) por-phyrin (TPPS4), a porphyrin of interest as a promising sensitizer for PDT, has somecritical solvation issues [47,51,62–65]. In particular, when a typical free-base porphyrin,such as TPPS4, is dissolved as a monomer, the fluorescence lifetime is of the order of 10–18 ns, but, when it is aggregated, the fluorescence lifetime can drop to as low as 0.1–0.3 ns[66]. Gandini et al. [51] observed that TPPS4 interactions with several different types ofmicelles could induce monomerization. A similar interaction between TPPS4 and biologi-cal membranes was believed to be responsible for inducing monomer formation, which is apossible reason for the PDT activity of TPPS4 in vivo.

Furthermore, another viable PDT agent, bacteriochlorin a (BCA), was found toexhibit monomer formation on incorporation into liposomes, as studied by electronspin resonance (ESR) and ground-state absorption and emission spectroscopy. Hoebekeet al. [67] observed concentration and time-dependent shape modifications of the UV–visible spectra in methanol and phosphate buffer solutions. Decreasing the solution pHinduced a red shift of all absorption peaks. Incorporation of BCA into dimyrstoyl-l-�-phosphatidylcholine (DMPC) liposomes caused an increase in fluorescence intensity,which continued until a maximum loading of the liposomes was reached. Nitroxide spinlabels incorporated into BCA-treated DMPC liposomes measured by ESR resulted in achange in signal shape and a significant change in stearic acid spin probe concentrationover time. The results of ground-state absorption and emission spectra, combined with theESR data, suggest that BCA is solubilized as a monomer and is incorporated into DMPCmembrane bilayer at the outer surface. Preliminary data have indicated that BCA inliposomes has a triplet quantum yield of 0.6, a reasonable value for use as a PDT sensi-tizer.

B. Influence of Model Membranes on the State of Existence and Locationof Porphyrins

The photochemistry and photophysics of porphyrins have been widely studied and theirrole as photosensitizers in PDT has been explored [42,43,66,68–74]. In particular, tetra-phenylporphyrins (TPPs) have been considered as PDT agents [47,68,74–76]. Despite the


vast research on porphyrins, a complete understanding of the affect of porphyrin mod-ification, such as substituent type and location on the macrocycle, on membrane localiza-tion, and membrane interaction is still under development. Substituent type and locationaffect critical factors such as the characteristics of porphyrin hydrophobicity/lipophilicity,solubilization as aggregates or monomers, and location in membranes and compartmen-talization [66,77,78].

Solubilization and location of TPPs with different substituents in model systems areof strong interest. For example, studies of para-substituted TPPs such as 5-(p-hydrox)phe-nyl-10,15,20-triphenylporphyrin [(p-OHÞ1TPPH2], 5-(p-carboxy)phenyl-10,15,20-triphe-nylporphyrin [(p-COO�Þ1TPPH2], and 5-(p-amino)phenyl-10,15,20-triphenylporphyrin[(p-NH2Þ1TPPH2] have been conducted in various solvents and model membranes.Solvents used to study the localization and photophysics of the para-substituted TPPsinclude cyclohexane and aqueous micellar solutions of sodium dodecyl sulfate (SDS),tetradecyltrimethylammonium bromide (TTAB), and poly(ethyleneglycol)-p-t-octylphenol(TX-100). Each TPP, (p-OHÞ1TPPH2, (p-NH2Þ1TPPH2, and (p-COO�Þ1TPPH2, was spon-taneously solubilized by diffusion into cyclohexane, TTAB, and TX-100, but not intoSDS. The solubility for (p-COO�Þ1TPPH2 was comparable in all solvents, but for (p-NH2Þ1TPPH2 the solubility was higher in cyclohexane, followed by TX-100, then TTAB.

The state of existence, i.e., whether the TPPs are dissolved as aggregates or mono-mers, and the location of the TPP in the micelle model membranes can be determined inpart by ground-state absorption and emission spectra. In the UV–visible spectra of (p-COO�Þ1TPPH2, (p-NH2Þ1TPPH2, and (p-OHÞ1TPPH2, there are typical, intense peaks inthe Soret region at 418 nm in cyclohexane, 420 nm in TX-100, and 420 nm in TTAB, inaddition to the four weaker peaks corresponding to the Q bands absorbing at 512–650 nm.The similarity in the absorption profiles of the porphyrins in cyclohexane and in micellesindicates that the porphyrins on average are monodispersed and remain dissolved innonpolar environments [18]. In the fluorescence spectra there is a characteristic emissionband near 650 nm for the porphyrins in all solvents. Typical fluorescence quenching andwavelength shifts, which are seen in aggregates, were not observed for these porphyrinsstudied. In agreement with the conclusion drawn from the absorption profiles, the emis-sion spectra suggest that the para-substituted TPPs exist as monomeric species in a non-polar environment.

Besides the ground-state absorption and emission spectra, NMR provides anotherpowerful, and perhaps more direct, tool to probe the location of the photosensitizer inmembranes [46,51,79], which is one of the most important issues in PDT [43]. For exam-ple, Simonis’ group [18] used NMR to characterize systematically the location of severalTPPs including (p-OHÞ1TPPH2, (p-NH2Þ1TPPH2, and (p-COO�Þ1TPPH2 in micelles suchas TTAB and TX-100. Figure 3 shows proton NMR shifts for different regions of theTTAB surfactant molecule in the presence of (p-OHÞ1TPPH2. The interaction between theporphyrin molecule and different regions of the surfactant molecule, in this case TTAB,provides information about the location of the porphyrin in the micelle. The NMR datapoint to intercalation of (p-OHÞ1TPPH2 among the surfactant chains in TTAB and TX-100 with an average orientation of the hydroxy group pointed toward the micelle waterinterface. Hydrogen-bond formation of the OH group to water at the interface may be afactor in the average preferential orientation of the porphyrin substituent. NMR data for(p-NH2Þ1TPPH2 in TTAB are not available due to the low concentration of porphyrinincorporated into the micellar solution. Data for (p-NH2Þ1TPPH2 and (p-COO�Þ1TPPH2

in TX-100 suggest that the porphyrins tend to localize in the polar domains of the micellesand to intercalate among the surfactant chains. The charged substituent of (p-


COO�Þ1TPPH2 does not seem to affect the orientation of the porphyrin towards theTTAB interface, as may be expected. Rather, the porphyrin is more deeply embeddedin TTAB than in TX-100. The hydrophobicity of the porphyrin macrocycle may be themajor factor affecting the location of (p-COO�Þ1TPPH2.

C. Effect of Model Membranes on Excited-State Dynamics of Porphyrins

As mentioned earlier, the exited states of sensitizers, as well as the generation of singlet-state oxygen, are responsible for therapeutic action in PDT. Time-resolved transientabsorption spectroscopy is a useful tool for measuring excited-state lifetimes and bimole-cular quenching rates with oxygen. For instance, time-resolved spectroscopy was used tomeasure the para-substituted TPP triplet lifetimes and bimolecular quenching reactionrates with oxygen in organic solvent as well as in micelle model systems. Significant life-time lengthening was observed for the above-discussed TPPs in micelles such as TTAB andTX-100 in the presence or absence of oxygen [80]. Similar observation has been made withvarious sensitizers in other studies [81,82]. Triplet-state lifetimes of p-TPP in aerobiccyclohexane were shorter (210–235 ns) than in micellar solutions (2.1–2:2�s). Despitetheir different locations within the micelles systems, the photophysical effect of themicelles was similar for all three porphyrins, which may be expected since many mono-meric porphyrins have similar photophysical properties [66].

Like the triplet-state lifetimes of the photosensitizer, the bimolecular quenchingconstants with oxygen are also affected by the presence of micelles. The experimentallydetermined quenching constants for the para-substituted TPPs in micelles(5:9� 108 M�1 s�1) were smaller than in cyclohexane (1:9� 109 M�1 s�1) despite thelonger triplet-state lifetimes in micelles. A possible reason for the smaller rate constantsin micelles is restricted molecular motion of the porphyrins within the micelles due toporphyrin–micelle interaction [7]. This suggestion agrees with NMR relaxation measure-ments, which clearly reveal that the porphyrins in micelles are less mobile than in organic

FIG. 3 1H NMR spectra of TTAB micellar solutions in the absence (bottom row) and in the

presence of increasing concentrations of (p-OHÞ1TPPH2. (From Ref. 18.)


solvents as evidenced in T1 and T2 relaxation times [18]. Another possible reason for lowerTPP quenching rates with oxygen in micelle systems is higher local viscosity of the micellarinterior. The internal local environment of the micelles may be effectively more viscousthan cyclohexane, which would reduce the bimolecular quenching rates. A study on thedirect influence of solvent viscosity on oxygen-quenching properties of excited states ofmesoporphyrin IX by Kapinus et al. [83] showed that bimolecular oxygen-quenching rateconstants decrease with increasing viscosity. Lower bimolecular quenching rates withoxygen are expected to be undesirable for PDT, since it could reduce singlet oxygengeneration. However, research has shown that sensitizers in micelles and liposomes haveincreased quantum yield of singlet oxygen [60,84]. This could be because the longer sensi-tizer triplet lifetime is a more dominant factor than bimolecular quenching rates in deter-mining singlet oxygen production.

III. INFLUENCE OF MODEL MEMBRANES ON SINGLET OXYGENLIFETIMES

One of the main requirements for the effective photodynamic action of sensitizers is highyield of singlet oxygen generation. Monitoring the production of singlet oxygen [O2 (

1�g)]in biological systems is very difficult because it has a short lifetime and is highly reactivetowards lipids, nucleic acids, and membrane proteins. The behavior of singlet oxygen inbiological systems is well modeled using micelles and liposomes, and indeed, most singletoxygen studies have been carried out model systems. Several methods are available formonitoring singlet oxygen production and lifetimes including direct infrared luminescence[85,86], thermal lensing [87], and indirect photochemical probing [87,88]. These methodshave been applied to the study of the effect of membrane interfaces on photosensitizerproduction of and lifetime of singlet oxygen.

Studies of singlet oxygen production in the presence of model membranes with aphotosensitizer inside or outside of the membrane have shown that singlet oxygen diffu-sion is not restricted by charged or neutral water–lipid interfaces and that it is incapable ofbeing compartmentalized [89–91]. The lifetime of singlet oxygen in micelles and liposomesis longer (30�s) than in pure water (3�s), cyclohexane (17�s) [92], and dodecane (24�s)[89]. Studies of singlet oxygen quenching by empty micelles have generated conflictingresults. Miyoshi and Tomita [93] found that the rate constants for singlet oxygen quench-ing by empty micelles were surprisingly large compared to those by sodium azide, and Jori[94] found that the empty micelles only slightly affected the reactivity of singlet oxygentowards substrates in the aqueous phase. Similarly, Usui et al. [91] observed that sensitizerand singlet oxygen deactivation was not affected by an increase in the number of emptymicelles, which could potentially act as quenchers.

IV. PHOTOSENSITIZERS AS PROBES FOR MICELLE AND LIPIDINTERFACIAL AND INTERNAL ENVIRONMENTS

While photosensitizers for PDT have been mostly studied in model membrane systems tounderstand how a membrane interface affects localization, photophysics, and reactionrates with oxygen, they also turn out to be useful as probes for the microenvironmentof the model membrane systems. The properties of micelles, reverse micelles, and lipo-somes are of special interest in understanding biological membrane systems and in the


synthesis of novel nanomaterials. Many of the same techniques, such as fluorescencedepolarization spectroscopy, NMR and ESR, time-resolved fluorescence, and transientabsorption spectroscopy, which are used to study sensitizer photophysics, can be used tostudy the local environment of model interfacial membrane systems. In particular, probingphotosensitizer interactions helps to gain insight into interfacial as well as interior solva-tion dynamics [9,10], micropolarity and microviscosity [6,95–97], and phase transitions ofthe lipid domains of liposomes [14,15,98].

In particular, liposomes are valuable model systems for biological membranes, espe-cially since they can be incorporated with proteins and biological lipids or completelycomprised of biological surfactants. Mixed liposome systems and the influence of addedmoieties can be studied with photosensitizers such as hematoporphyrin IX (HP) andprotoporyhrin IX (PP), which have been shown to be interesting in PDT as well asvaluable in probing various regions of liposomes [15]. For example, HP and PP distribu-tion in liposomes of dimirystoyl-l-�-phosphatidylcholine (DMPC) and dipalmitoylpho-sphatidylcholine (DPPC), enriched with increasing concentrations of cholesterol (Chol)and cardiolipin (Card), were studied by following the temperature dependence of theporphyrin fluorescence polarization and the quenching of porphyrin fluorescence bymethyl viologen [98]. In all the liposomal systems, HP was found distributed within thevery polar regions of the inner monolayer, and PP was found to interact preferentially inthe most nonpolar regions of the lipid bilayer. Changing the HP concentration in DPPCliposomes allowed for modulation of the dye occupation between the lipid headgroup andlipid–water interfaces. No modulation was observed for changing concentrations of HP inDMPC liposomes, instead, it remained partitioned at the lipid headgroup–water inter-faces. The effect of Card, which acts as a strong fluidifier, is to shift the occupation of bothporphyrins toward the external monolayer. Increasing the Chol content in the liposomeshas the effect of increasing the rigidity of the liposomes, but it does not cause a shift in HPor PP distribution at high concentrations.

Increasing the temperature of liposomal systems causes a phase transition to occur inthe lipid domains, but the effect of Chol and Card as monitored by HP and PP, can causechanges in transition temperatures. For example, Fig. 4 shows changes in the fluorescencepolarization for HP as a function of increasing temperature in the presence of varyingChol concentrations (0–55%) in DPPC liposomes. A linear dependence of fluorescencepolarization on temperature was observed in DPPC with 20% Chol, but at increasingChol concentrations two transitions were observed. The first transition is typical of HP inDPPC vesicles, but the second transition at T ¼ 45�C with 37% Chol and T ¼ 49�C withChol ¼ 47% are not typical. At the highest concentration of Chol no phase transitionoccurs.

The temperature-dependent fluorescence polarization data for HP in DPPC lipo-somes loaded with increasing concentrations of Chol have an interesting interpretation. Athigh concentrations, HP was distributed close to the inner polar headgroups, which wasconfirmed by a low critical value of phase transition for DPPC at 31�C rather than thetypical temperature at T ¼ 41�C. However, when 20% Chol was added a depolarizationeffect was observed because HP was at the aqueous interface and was not sensitive tochanges in the lipid domain. Increasing the Chol concentration further caused a redistri-bution of HP into the lipid domains and phase transitions were observed. The HP redis-tribution is attributed to a shift in Chol distribution in DPPC. As expected and observed,further increase in Chol concentration (55%) results in total inhibition of phase transitionsin DPPC due to increased rigidity. These studies show that photosensitizers such as HPand PP are important probes in liposomes and mixed liposome systems.


Similarly, photosensitizers have been used to probe micelles and reverse micelles. Paland coworkers [6,9] used 4-(dicyanomethylene)-2-methyl-6-(p-dimethylaminostyryl) 4H-pyran (DCM), which is insoluble in water and remains in the micellar core, and 3,3 0-diethyloxadicarbocyanine iodide (DODCI), which is highly water soluble and remains in asmall region of the micelle water interface, to probe the interior and exterior of micelles.The solvation dynamics of DCM using picosecond time-resolved Stokes shift and thephotoisomerization of DODCI, using picosecond time-resolved emission spectroscopy,were studied in CTAB, TX-100, and SDS. The observed time-resolved spectra of DCMreport on the interiors of TX-100, CTAB, and SDS. The data show that the fluorescencedecays of DCM are wavelength dependent and the spectra at the red and blue ends differsignificantly. The solvent relaxation time in TX-100 (2100 ps) is much slower than inCTAB (400 ps) and SDS (1400 ps). The observations are interpreted to mean thatDCM occupies the highly polar Stern layer of the micelles and not the hydrophobiccore, for two reasons. First, the emission maximum of DCM in nonpolar solvent isblue-shifted and does not exhibit wavelength dependence as was observed in micelles.Second, the core of the micelles is not expected to contribute to the observed solvationdynamics. The observed differences in solvation relaxation time are attributed to differ-ences in micellar structure for TX-100, CTAB, and SDS. Also, since TX-100 is a neutralmicelle, solvent relaxation dynamics can be due to restricted motion of the water moleculesin the Stern layer and not solely due to ionic relaxation.

While the photosensitizer DCM serves as a probe into the micellar interior, DODCIprovides a probe of a small region near the water–micelle interface. The time-resolvedemission spectra for DODCI in TX-100, CTAB, and SDS above the critical micelle con-centration showed a marked increase in the emission quantum yield (two to three timesgreater than in water). Also, a significant increase in the excited-state lifetime of DODCIwas observed, from 0.70 ns in water to 2.25 ns in SDS, 2.36 ns in CTAB, and 2.55 ns inTX-100. The observations of increased fluorescence quantum yield are attributed to adecrease in the photoisomerization rate of DODCI as a result of the micellar environment,

FIG. 4 Excitation fluorescence polarization (p) of 5�M HP in DPPC liposomes as a function of

temperature observed at 520 nm with emission at 626 nm. HP in liposomes of DPPC (&), DPPC–

Chol (20%) (~), DPPC–Chol (37%) (~), DPPC-Chol (47%) (*), and DPPC–Chol (55%) (*).

(From Ref. 98.)


which offers higher friction than bulk water. The decrease in isomerization rate is attrib-uted to a higher microviscosity at the water–micelle interface. At the Smoluchowski limit,the microviscosities, as compared to n-decanol (� ¼ 14 cP), of TX-100, CTAB, and SDSwere calculated to be 26:0� 2 cP, 70:0� 20 cP, and 24:5� 2 cP, respectively.

Similar techniques, such as time-resolved fluorescence Stokes shift (TRFSS), havebeen applied to reverse micelle systems to probe solvation dynamics. For example,Levinger [10] used TRFSS to probe the dynamics of polar solvation in several reversemicelle environments including sodium di-2-ethylhexyl sulfosuccinate, lecithin, Triton X-100, and Brij-30. All of the systems studied showed a dramatic slowing of the solvationdynamics occurring inside the reverse micelles, in agreement with studies done by otherresearchers [6,9,99]. Another observation was an ultrafast water relaxation, which wasfound to depend on micelle morphology and gave some insight into micellar structure. Thesource of slower relaxation components is still unclear, but was thought to be possibly dueto solvent bound to the supermolecule assembly or to the immobilization of the solvent ina confined environment, or both.

The above studies have demonstrated that various photosensitizers, in conjunctionwith many available experimental techniques, can be used to probe different regions of thecolloidal model membranes systems. Careful choice of sensitizers is important in deter-mining different regions of the micelles, reverse micelles, or liposomes, and their differentdynamic and structural features.

V. SUMMARY

In summary, we have provided a review of some of the unique properties of micelles,reverse micelles, and lipsomes with emphasis on their application as model interfacialmembrane systems in the study of the photophysics of photosensitizers useful for PDT.Compared to homogeneous environments such as solutions, heterogeneous systems suchas interfaces modeled with micelles, liposomes, or biological molecules, e.g., proteins, havesignificant influence on the equilibrium as well as dynamic properties of photosensitizersdue to interaction between sensitizer molecules and the interface. The sensitizer–mem-brane interaction leads to changes in the equilibrium between aggregation and disaggrega-tion, excited-state singlet and triplet lifetimes of the photosensitizers, location of thephotosensitizer molecules near the interface, reaction rate between the triplet-state sensi-tizer and ground-state triplet oxygen, and the rate as well as yield of singlet oxygengeneration. As a result, the overall functionality of photosensitizers and efficacy forPDT critically depend on the interaction between the photosensitizer molecules and thelocal environment of the membranes. Understanding this interaction thus has importantimplications in the development of new and more effective photosensitizers for PDTapplications.

REFERENCES

1. SC Wallace, M Gratzel, JK Thomas. Chem Phys Lett 23:359–362, 1973.

2. M Chen, M Gratzel, JK Thomas. Chem Phys Lett 24:65–68, 1974.

3. M Gratzel, JJ Kozak, JK Thomas. J Chem Phys 62:1632–1640, 1975.

4. RE Riter, JR Kimmel, EP Undiks, NE Levinger. J Phys Chem B 101:8292–8297, 1997.

5. D Pant, RE Riter, NE Levinger. J Chem Phys 109:9995–10003, 1998.


6. SK Pal, A Datra, D Mandal, K Bhattacharyya. Chem Phys Lett 288:793–798, 1998.

7. SK Pal, D Mandal, D Sukul, K Bhattacharyya. Chem Phys 249:63–71, 1999.

8. SK Pal, D Mandal, D Sukul, K Bhattacharyya. Chem Phys Lett 312:178–184, 1999.

9. SK Pal, D Sukul, D Mandal, S Sen, K Bhattacharyya. Chem Phys Lett 327:91–96, 2000.

10. NE Levinger. Curr Opin Colloid Interface Sci 5:118–124, 2000.

11. JH Fendler. Membrane Mimetic Chemistry. New York: Wiley-Interscience, 1982, pp 522.

12. E Reddi, M Beltramini, F Ricchelli, G Jori. Photochem Photobiol 41:15S, 1985.

13. F Ricchelli, G Jori. Photochem Photobiol 44:151–158, 1986.

14. F Ricchelli, D Stevanin, G Jori. Photochem Photobiol 48:13–18, 1988.

15. F Ricchelli, G Jori, S Gobbo, M Tronchin. Biochim Biophys Acta 1065:42–48, 1991.

16. M Hoebeke. J Photochem Photobiol 28:189–196, 1995.

17. G Valduga, E Reddi, S Garbisa, G Jori. Photochem Photobiol 65:46S, 1997.

18. M Vermathen, S Reid, U Simonis. Langmuir 16:210–221, 2000.

19. F Ricchelli, G Jori. Utilization of liposomes as porphyrin carriers in the photodynamic

therapy of tumors. Proceedings of Bulgarian Academy of Sciences (ED). FECS Third

International Conference on Chemistry and Biotechnology of Biologically Active Natural

Products, V. 5. Communications: Bioorganic Chemistry and Structural Elucidation and

Chemical Transformation of Natural Products, Sofia, Bulgaria, 1987, pp 241–244.

20. VP Torchilin. Mol Med Today 2:242–249, 1996.

21. VP Torchilin. J Lipos Res 7:433–454, 1997.

22. V Weissig, C Lizano, VP Torchilin. J Lipos Res 8:391–400, 1998.

23. J Taillefer, MC Jones, N Brasseur, JE van Lier, JC Leroux. J Pharm Sci 89:52–62, 2000.

24. VP Torchilin. Eur J Pharm Sci 11:S81–S91, 2000.

25. V Torchilin, J Babich, V Weissig. J Lipos Res 10:483–499, 2000.

26. VP Torchilin. J Contr Release 73:137–172, 2001.

27. J Taillefer, N Brasseur, JE van Lier, V Lenaerts, D Le Garrec, JC Leroux. J Pharmacy

Pharmacol 53:155–166, 2001.

28. TJ Dougherty. Photochem Photobiol 45: 879–889, 1987.

29. TJ Dougherty, SL Marcus. Eur J Cancer 28A:1734–1742, 1992.

30. TJ Dougherty. Adv Photochem 17:275–311, 1992.

31. D Dolphin. Can J Chem 72:1105–1110, 1994.

32. E Ben-Hur, B Horowitz. Photochem Photobiol 62:383–388, 1995.

33. E Ben-Hur, S Rywkin, I Rosenthal, NE Geacintov, B Horowitz. Transfusion 35:401–406,

1995.

34. H Margolis-Nunno, R Robinson, B Horowitz, NE Geacintov, E Ben-Hur. Photochem

Photobiol 62:917–922, 1995.

35. P Gottlieb, H Margolis-Nunno, R Robinson, LG Shen, E Chimezie, B Horowitz, E Ben-Hur.

Photochem Photobiol 63:562–565, 1996.

36. E Ben-Hur, AC Moor, H Margolis-Nunno, P Gottlieb, MM Zuk, S Lustigman, B Horowitz,

A Brand, J Van Steveninck, TM Dubbelman. Transfusion Med Rev 10:15–22, 1996.

37. S Lustigman, E Ben-Hur. Transfusion 36:543–546, 1996.

38. H Margolis-Nunno, E Ben-Hur, P Gottlieb, R Robinson, J Oetjen, B Horowitz. Transfusion

36:743–750, 1996.

39. BZ Zmudzka, AG Strickland, JZ Beer, E Ben-Hur. Photochem Photobiol 65:461–464, 1997.

40. E Ben-Hur, G Barshtein, S Chen, S Yedgar. Photochem Photobiol 66:509–512, 1997.

41. E Ben-Hur, WS Chan, Z Yim, MM Zuk, V Dayal, N Roth, E Heldman, A Lazo, CR Valeri, B

Horowitz. Devel Biol Standard 102:149–155, 2000.

42. E Zenkevich, E Sagun, V Knyukshto, A Shulga, A Mironov, O Efremova, R Bonnett, SP

Songca, M Kassem. J Photochem Photobiol B:Biol 33:171–180, 1996.

43. NL Oleinick, HH Evans. Radiat Res 150:S146–S156, 1998.

44. WM Sharman, CM Allen, JE van Lier. Methods Enzymol 319:376–400, 2000.

45. G Valduga, E Reddi, G Jori. Photochem Photobiol 43:78S, 1986.

46. S Mazumdar. J Phys Chem 94:5947–5953, 1990.


47. IE Borissevitch, TT Tominaga, H Imasato, M Tabak. J Lumin 69:65–76, 1996.

48. F Mallamace, N Micali, S Trusso, L Monsu’ Scolaro, A Romeo, A Terracina, RF Pasternack.

Phys Rev Lett 76:4741–4744, 1996.

49. BW Gregory, D Vaknin, TM Cotton, WS Struve. Interfacial complexation of phospholipid

Langmuir monolayers with water-soluble porphyrins and phthalocyanines: an X-ray reflectiv-

ity study. Proceedings of Seventh International Conference on Organized Molecular Films,

Numana, Italy, 1996, pp 849–853.

50. SCM Gandini, IE Borissevitch, JR Perussi, H Imasato, M Tabak. J Lumin 78:53–61, 1998.

51. SCM Gandini, VE Yushmanov, IE Borissevitch, M Tabak. Langmuir 15:6233–6243, 1999.

52. RF Khairutdinov, N Serpone. J Phys Chem B 103:761–769, 1999.

53. H Xuezhong, X Guangming, Z Yalin, Z Manhua, S Tao. Spectrochim Acta, Part A: Mol

Biomol Spectrosc 55A:873–880, 1999.

54. ED Sternberg, D Dolphin, C Bruckner. Tetrahedron 54:4151–4202, 1998.

55. G Valduga, E Reddi, G Jori. J Inorg Biochem 29:59–66, 1987.

56. M Shopova, V Mantareva, K Krastev, D Hadjiolov, A Milev, K Spirov, G Jori, F Ricchelli. J

Photochem Photobiol B: Biol 16:83–89, 1992.

57. L Howe, JZ Zhang. Photochem Photobiol 67:90–96, 1998.

58. A Harriman, M Richoux. J Photochem 14:253, 1980.

59. G Valduga, E Reddi, G Jori, R Cubeddu, P Taroni, G Valentini. J Photochem Photobiol B:

Biol 16: 331–340, 1992.

60. G Valduga, S Nonell, E Reddi, G Jori, SE Braslavsky. Photochem Photobiol 48:1–6, 1988.

61. L Howe, JZ Zhang. J Phys Chem A 101:3207–3213, 1997.

62. JM Wessels, W Strauss, HK Seidlitz, A Ruck, H Schneckenburger. J Photochem Photobiol

B:Biol 12:275–284, 1992.

63. RM Ion, A Planner, K Wiktorowicz, D Frackowiak. Acta Biochim Polon 45:833–845, 1998.

64. IE Borissevitch, N Rakov, GS Maciel, CB de Araujo. Appl Optics 39:4431–4435, 2000.

65. E Weizman, C Rothmann, L Greenbaum, A Shainberg, M Adamek, B Ehrenberg, Z Malik.

J Photochem Photobiol B:Biol 59:92–102, 2000.

66. F Ricchelli. J Photochem Photobiol B:Biol 29:109–118, 1995.

67. M Hoebeke, X Damoiseau, HJ Schuitmaker, A Van de Vorst. Biochim Biophys Acta

1420:73–85, 1999.

68. R Bonnett, RD White, UJ Winfield, MC Berenbaum. Biochem J 261:277–280, 1989.

69. R Bonnett, AN Nizhnik, SG White, MC Berenbaum. J Photochem Photobiol B:Biol 6:29–37,

1990.

70. R Bonnett, A Harriman, AN Kozyrev. J Chem Soc, Faraday Trans 88:763–769, 1992.

71. K Kalyanasundaram. Photochemistry of Polypyridine and Porphyrin Complexes. San Diego,

CA: Academic Press, 1992, pp 626.

72. R Bonnett. Chem Soc Rev 24:19–33, 1995.

73. K Berlin, RK Jain, C Richerts. Biotech Bioeng Combinat Chem 61:106–118, 1998.

74. L Danaila, M Pascu, A Popescu, M Pascu, R Ion. Spectrophotometric characterization of

useful dyes in laser photodynamic therapy. Proceedings of SPIE—International Society of

Optical Engineering, Bucharest, Romania, 2000, pp 712–720.

75. VV Sapunov. Zh Prikl Spektrosk 62:171–178, 1995.

76. R Bonnett, BD Djelal, PA Hamilton, G Martinez, F Wierrani. J Photochem Photobiol B:Biol

53:136–143, 1999.

77. B Boyle, D Dolphin. Photochem Photobiol 64:458–485, 1996.

78. M Oschner. J Photochem Photobiol 39:1–19, 1997.

79. S Mazumdar, OK Medhi, S Mitra. Inorg Chem 30:700–705, 1991.

80. SA Gerhardt, J Lewis, D Kilger, U Simonis, JZ Zhang. Langmuir, submitted April 2002.

81. R Brennetot, J Georges. Spectrochim Acta, Part A:Mol Biomol Spectrosc 55:229–231, 1999.

82. M Havelcova, P Kubat, I Nemcova. Dyes Pigments 44:49–54, 2000.

83. EI Kapinus, VA Ganzha, BM Dzhagarov, EI Sagun, VB Pavlenko. Teor Eksp Khim 25:161–

167, 1989.


84. MG Shaposhnikova, VI Binyukov. Biofizika 20:1153–1154, 1975.

85. D De La Pena, C Marti, S Nonell, LAMartinez, MAMiranda. Photochem Photobiol 65:828–

832, 1997.

86. S Nonell, SE Braslavsky. Methods Enzymol 319:37–49, 2000.

87. F Wilkinson, WP Helman, AB Ross. J Phys Chem Ref Data 24:663–1021, 1995.

88. IE Kochevar, RW Redmond. Methods Enzymol 319:20–28, 2000

89. PC Lee, MAJ Rodgers. J Phys Chem 87:4894–4898, 1983.

90. MAJ Rodgers, PC Lee. J Phys Chem 88: 3480–3484, 1984.

91. Y Usui, M Tsukada, H Nakamura. Bull Chem Soc Jpn 51:379–384, 1978.

92. AA Gorman, MAJ Rodgers. Chem Phys Lett 55:52–54, 1978.

93. N Miyoshi, G Tomita. Z Naturforsch 33b:622–627, 1978.

94. G Jori. In: D Kessel, TJ Dougherty, eds. Porphyrin Photosensitization, vol. 160. New York:

Plenum Press, 1981, pp 193–212.

95. M Hasegawa, T Sugimura, K Kuraishi, Y Shindo, A Kitahara. Chem Lett 7:1373–1376, 1992.

96. M Hasegawa, T Sugimura, Y Shindo, A Kitahara. Colloids Surfaces A:Physicochem Eng

Aspects 109:305–318, 1996.

97. M Osman, H Hoiland, H Holmsen. Colloids Surfaces B 11:167–175, 1998.

98. F Ricchelli, S Gobbo. J Photochem Photobiol 29:65–70, 1995.

99. A Datta, D Mandal, SK Pal, K Bhattacharyya. Chem Phys Lett 278:77–82, 1997.


INTERFACIAL C ATALYS I s

EDITED BY

ALEXANDER G. VOLKOV Oakwood College Huntsville, Alabama, U.S.A.

M A R C E L

MARCEL DEKKER, INC. a% D E K K E R

NEW YORK BASEL


ISBN: 0-8247-0839-3

This book is printed on acid-free paper.

Headquarters

Marcel Dekker, Inc.

270 Madison Avenue, New York, NY 10016

tel: 212-696-9000; fax: 212-685-4540

Eastern Hemisphere Distribution

Marcel Dekker AG

Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland

tel: 41-61-260-6300; fax: 41-61-260-6333

World Wide Web

http://www.dekker.com

The publisher offers discounts on this book when ordered in bulk quantities. For more information,

write to Special Sales/Professional Marketing at the headquarters address above.

Copyright # 2003 by Marcel Dekker, Inc. All Rights Reserved.

Neither this book nor any part may be reproduced or transmitted in any form or by any means,

electronic or mechanical, including photocopying, microfilming, and recording, or by any informa-

tion storage and retrieval system, without permission in writing from the publisher.

Current printing (last digit):

10 9 8 7 6 5 4 3 2 1

PRINTED IN THE UNITED STATES OF AMERICA


http://www.dekker.com/sdek/home

Preface

Interfacial catalysis plays a key role in many chemical, physical, and biological processes.The past decade has witnessed a huge increase of research interest in the study of inter-facial catalysis at liquid interfaces. Processes taking place at the interface between twoimmiscible liquid phases are fundamental to life since virtually all energy conversionprocesses in living organisms occur at liquid interfaces. The properties of liquid–liquidinterfaces are very important for a variety of industries, including biotechnology, organicsynthesis, nanochemistry, catalysis, pharmaceuticals, cosmetics, paints, detergents, oilextraction processes, and mining.

The interface between two immiscible liquids with immobilized photosynthetic pig-ments serves as a convenient model for investigating photoprocesses that are accompaniedby spatial separation of charges. The efficiency of charge separation is defined by thequantum yield of any photochemical reaction. Heterogeneous systems in which the oxi-dants and the reductants are either in different phases or sterically separated are the mosteffective in this regard. Different solubilities of the substrates and reaction products in thetwo phases of heterogeneous systems can alter the redox potential of reactants, making itpossible to carry out reactions that cannot be performed in a homogeneous phase.

The book is organized into five parts. Part I consists of seven chapters and deals withfundamental aspects of interfacial phenomena such as catalytic properties of liquid inter-faces, electrochemistry at polarized interfaces, ion solvation and resolvation, interfacialpotentials, separations, and interfacial catalysis in metal complexation and in enhanced oilrecovery.

Part II contains four chapters about history, theory, molecular mechanisms, synth-esis, and experimental systems in phase transfer catalysis.

Part III deals with micellar catalysis, enzymology, and photochemical reactions inreversed micelles.

The chapters in Part IV discuss biological aspects of interfacial and membranecatalysis including bioelectrocatalysis, ion channels, mechanisms of respiration and photo-synthesis, membrane catalysis, and ion transport processes.

Part V, which is about interfacial photocatalysis, includes such topics as nano-chemistry, nanoparticles, self-organized microheterogeneous structures, photosensitizers,


and photocatalytic oxygen evolution. The experimental systems and theoretical analysis ofinterfacial photocatalytic systems are also discussed in Chapters 14, 15, and 18.

I would like to extend my thanks to the authors for the time they spent on thisproject and for teaching us about their work on nanochemistry and interfacial catalysis. Ialso thank our Acquisitions Editor, Anita Lekhwani, and our Production Editor, JosephStubenrauch, for their friendly and courteous assistance.

Alexander G. Volkov


Contents

PrefaceContributors

Part I. Interfacial Phenomena

1. Interfacial Catalysis at Oil/Water InterfacesAlexander G. Volkov

2. Electrochemistry of Chemical Reactions at Polarized Liquid–Liquid InterfacesTakashi Kakiuchi

3. Interfacial Catalysis in Metal ComplexationHitoshi Watarai

4. The Role of Water Molecules in Ion Transfer at the Oil/Water InterfaceToshiyuki Osakai

5. Interfacial Potential and Distribution Equilibria Between Two ImmiscibleElectrolyte SolutionsLe Quoc Hung

6. Use of Cyclodextrins or Porous Inorganic Supports to Improve Organic/Aqueous Interfacial TransfersMartine Urrutigoıty and Philippe Kalck

7. Ultrathin Films: Their Use in Enhanced Oil Recovery and in InterfacialCatalysisLu Zhang, Sui Zhao, Jia-Yong Yu, Angelica L. Ottova, and H. Ti Tien

Part II. Phase Transfer Catalysis

8. Phase Transfer CatalysisMieczysiaw Makosza and Michai Fedorynski

9. Liquid–Liquid Phase Transfer Catalysis: Basic Principles and SyntheticApplicationsDomenico Albanese

10. Phase Transfer Catalysis: Fundamentals and Selected SystemsJing-Jer Jwo


11. Interfacial Mechanism and Kinetics of Phase-Transfer CatalysisHung-Ming Yang and Ho-Shing Wu

Part III. Micellar Catalysis

12. Enzymes in Reverse Micelles (Microemulsions): Theory and PracticeAndrey V. Levashov and Natalia L. Klyachko

13. Micellar CatalysisVincent C. Reinsborough

14. Multiple Effects of Water Pools and Their Interfaces Formed by ReversedMicelles on Enzymic Reactions and PhotochemistryAyako Goto, Yuko Ibuki, and Rensuke Goto

Part IV. Interfacial Biocatalysis and Membrane Catalysis

15. Supported Planar BLMs (Lipid Bilayers): Formation, Methods of Study,and ApplicationsAngelica L. Ottova and H. Ti Tien

16. BioelectrocatalysisKenji Kano and Tokuji Ikeda

17. Energetics and Gating of Narrow Ionic Channels: The Influence of ChannelArchitecture and Lipid–Channel InteractionsPeter C. Jordan, Gennady V. Miloshevsky, and Michael B. Partenskii

18. Biocatalysis: Electrochemical Mechanisms of Respiration and PhotosynthesisAlexander G. Volkov

19. New Types of Membrane Reactions Mimicking Biological ProcessesSorin Kihara

20. Ion-Transport Processes Through Membranes of Various Types: LiquidMembrane, Thin Supported Liquid Membrane, and Bilayer Lipid MembraneOsamu Shirai and Sorin Kihara

Part V. Interfacial Photocatalysis

21. Development of Structurally Organized Photocatalytic Systems forPhotocatalytic Hydrogen Evolution on the Basis of Lipid Vesicles withSemiconductor Nanoparticles Fixed on Lipid MembranesOxana V. Vassiltsova and Valentin N. Parmon

22. Catalysis and Photocatalysis at Polarized Molecular Interfaces: AnElectrochemical Approach to Catalytic Processes Based on Two-PhaseSystems, Self-Organized Microheterogeneous Structures, andUnsupported NanoparticlesRiikka Lahtinen, Henrik Jensen, and David J. Fermın

23. Photosensitizers at Interfaces of Model MembranesSarah A. Gerhardt and Jin Z. Zhang


Contributors

Domenico Albanese Dipartimento di Chimica Organica e Industriale, Universita degliStudi di Milano, Milan, Italy

Michał Fedorynski Faculty of Chemistry, Warsaw University of Technology, Warsaw,Poland

David J. Fermın Laboratoire d’Electrochimie Physique et Analytique, EcolePolytechnique Federale de Lausanne, Lausanne, Switzerland

Sarah A. Gerhardt Department of Chemistry, University of California at Santa Cruz,Santa Cruz, California, U.S.A.

Ayako Goto School of Informatics, University of Shizuoka, Shizuoka, Japan

Rensuke Goto Institute for Environmental Sciences, University of Shizuoka, Shizuoka,Japan

Le Quoc Hung Institute of Chemistry, National Center for Natural Science andTechnology, Hanoi, Vietnam

Yuko Ibuki Institute for Environmental Sciences, University of Shizuoka, Shizuoka,Japan

Tokuji Ikeda Division of Applied Life Sciences, Graduate School of Agriculture, KyotoUniversity, Kyoto, Japan

Henrik Jensen Laboratoire d’Electrochimie Physique et Analytique, Ecole PolytechniqueFederale de Lausanne, Lausanne, Switzerland

Peter C. Jordan Department of Chemistry, Brandeis University, Waltham,Massachusetts, U.S.A.


Jing-Jer Jwo Department of Chemistry, National Cheng Kung University, Tainan,Taiwan, Republic of China

Takashi Kakiuchi Department of Energy and Hydrocarbon Chemistry, KyotoUniversity, Kyoto, Japan

Philippe Kalck Laboratoire de Catalyse, Chimie Fine et Polymeres, Ecole NationaleSuperieure des Ingenieurs en Arts Chimiques et Technologiques, Toulouse, France

Kenji Kano Division of Applied Life Sciences, Graduate School of Agriculture, KyotoUniversity, Kyoto, Japan

Sorin Kihara Department of Chemistry, Kyoto Institute of Technology, Kyoto, Japan

Natalia L. Klyachko Department of Chemical Enzymology, Faculty of Chemistry,Moscow State University, Moscow, Russia

Riikka Lahtinen Department of Chemistry, University of Liverpool, Liverpool, UnitedKingdom

Andrey V. Levashov Department of Chemical Enzymology, Moscow State University,Moscow, Russia

Mieczysław Makosza Institute of Organic Chemistry, Polish Academy of Sciences,Warsaw, Poland

Gennady V. Miloshevsky Department of Chemistry, Brandeis University, Waltham,Massachusetts, U.S.A.

Toshiyuki Osakai Department of Chemistry, Faculty of Science, Kobe University, Kobe,Japan

Angelica L. Ottova Department of Physiology, Michigan State University, East Lansing,Michigan, U.S.A.

Valentin N. Parmon Boreskov Institute of Catalysis, Novosibirsk, Russia

Michael B. Partenskii Department of Chemistry, Brandeis University, Waltham,Massachusetts, U.S.A.

Vincent C. Reinsborough Department of Chemistry, Mount Allison University,Sackville, New Brunswick, Canada

Osamu Shirai Department of Nuclear Energy System, Japan Atomic Energy ResearchInstitute, Ibaraki, Japan

H. Ti Tien Department of Physiology, Michigan State University, East Lansing,Michigan, U.S.A.


Martine Urrutigoıty Laboratoire de Catalyse, Chimie Fine et Polymeres, EcoleNationale Superieure des Ingenieurs en Arts Chimiques et Technologiques, Toulouse,France

Oxana V. Vassiltsova Boreskov Institute of Catalysis, Novosibirsk, Russia

Alexander G. Volkov Department of Chemistry, Oakwood College, Huntsville,Alabama, U.S.A.

Hitoshi Watarai Department of Chemistry, Graduate School of Science, OsakaUniversity, Osaka, Japan

Ho-Shing Wu Department of Chemical Engineering, Yuan-Ze University, Taoyuan,Taiwan, Republic of China

Hung-Ming Yang Department of Chemical Engineering, National Chung HsingUniversity, Taichung, Taiwan, Republic of China

Jia-Yong Yu Research Center for Enhanced Oil Recovery, Technical Institute of Physicsand Chemistry, Chinese Academy of Sciences, Beijing, People’s Republic of China

Jin Z. Zhang Department of Chemistry and Biochemistry, University of California atSanta Cruz, Santa Cruz, California, U.S.A.

Lu Zhang Research Center for Enhanced Oil Recovery, Technical Institute of Physicsand Chemistry, Chinese Academy of Sciences, Beijing, People’s Republic of China

Sui Zhao Research Center for Enhanced Oil Recovery, Technical Institute of Physicsand Chemistry, Chinese Academy of Sciences, Beijing, People’s Republic of China


Interfacial Catalysisimage.sciencenet.cn/olddata/kexue.com.cn/blog/admin//images/upfiles/...Hydrogen...

Documents

Transcript of Interfacial Catalysisimage.sciencenet.cn/olddata/kexue.com.cn/blog/admin//images/upfiles/...Hydrogen...