Nanoemulsions Prepared by a Two-Step Low-Energy ProcessLijuan Wang, Kevin J. Mutch, Julian Eastoe, Richard K. Heenan, and Jinfeng Dong*,

College of Chemistry and Molecular Science, Wuhan UniVersity, Wuhan 430072, China, School ofChemistry, UniVersity of Bristol, Bristol BS8 1TS, U.K., and ISIS-STFC, Rutherford Appleton Laboratory,

Chilton, Oxon OX11 0QX, U.K.ReceiVed February 27, 2008. ReVised Manuscript ReceiVed March 20, 2008

A simple low-energy two-step dilution process has been applied in oil/surfactant/water systems with pentaoxyethylenelauryl ether (C12E5), dodecyldimethylammonium bromide, sodium bis(2-ethylhexyl)sulfosuccinate, sodium n-dodecylsulfate-pentanol, and hexadecyltrimethylammonium bromide-pentanol. Appropriate formulations were chosen forthe concentrate to be diluted with water to generate oil-in-water (O/W) emulsions or nanoemulsions. For the systemof decane/C12E5/water, bluish, transparent nanoemulsions having droplet radii of the order of 15 nm were formed,only when the initial concentrate was a bicontinuous microemulsion, whereas opaque emulsions were generated ifthe concentrate began in an emulsion-phase region. Nanoemulsions generated in the system decane/C12E5/water havebeen investigated both by dynamic light scattering (DLS) and contrast-variation small-angle neutron scattering (SANS).The SANS profiles show that nanodroplets exist as spherical core-shell (decane-C12E5) particles, which sufferessentially no structural change on dilution with water, at least for volume fractions down to 0.060. These resultssuggest that the nanoemulsion droplet structure is mainly controlled by the phase behavior of the initial concentrateand is largely independent of dilution. A discrepancy between apparent nanoemulsion droplet sizes was observed bycomparing DLS and SANS data, which is consistent with long-range droplet interactions occurring outside of theSANS sensitivity range. These combined phase behavior, SANS, and DLS results suggest a different reason for thestability/instability of nanoemulsions compared with earlier studies, and here it is proposed that a general mechanismfor nanoemulsion formation is homogeneous nucleation of oil droplets during the emulsification.

1. IntroductionNanoemulsions1 consist of small quite monodisperse droplets,

typically in the 20-200 nm size range. Although they are ofsimilar size to microemulsion droplets (1-100 nm), and appeartransparent or translucent, they are in fact distinctly quite differentfrom true microemulsions. Nanoemulsions are thermodynami-cally unstable, and their formation generally requires energyinput. The properties of nanoemulsions depend not only onthermodynamic conditions (i.e., composition, temperature, andpressure) but also on the preparation method, and, crucially, onthe order of component addition.2 However, the kinetic stabilityof nanoemulsions, and their transparent or translucent appearancedue to the presence of nanometer-sized droplets, makes na-noemulsions of interest for fundamental studies and practicalapplications (e.g., in chemical, pharmaceutical,3 and cosmetic4fields). To obtain nanodroplet emulsions, significant amounts ofmechanical energy are needed, making high-energy preparationmethods unfavorable for industrial applications.5,6 Therefore,the preparation of nanoemulsions with reproducible properties

and small droplet sizes using low-energy methods has been afield of growing interest.7,8

Low-energy methods7 make use of accessible phase transitionsoccurring during the emulsification process as a result of changesin surfactant film spontaneous curvature. This curvature transitionhas been achieved through different routes: (a) partitioning ofalcohol from the oil to the aqueous phase or diffusion of waterinto the initial droplet, both producing a shift from lipophilic tohydrophilic conditions;9,10 (b) chemical reactions which convertlipophilic surfactants to hydrophilic surfactants;11,12 (c) a suddendecrease of ionic strength with ionic surfactant systems;13,14 and(d) an increased hydration of poly(oxyethylene) chains of PEO-type nonionic surfactants.1522 The reversal of spontaneouscurvature from water-in-oil (W/O) to an oil-in-water configuration(O/W) reduces the solubilization capacity for oil to such an extentthat supersaturation may occur, leading to oil droplet nucleation.

* To whom correspondence should be addressed. E-mail: colloid@whu.edu.cn.

Wuhan University. University of Bristol. Rutherford Appleton Laboratory.(1) Forgiarini, A.; Esquena, J.; Gonzalez, C.; Solans, C. Langmuir 2002, 17,

20762083.(2) Esquena, J.; Solans, C. Prog. Colloid Polym. Sci. 1998, 110, 235239.(3) Nicolaos, G.; Crauste-Manciet, S.; Farinotti, R.; Brossard, D. Int. J. Pharm.

2003, 263, 165171.(4) Sonneville-Aubrun, O.; Simonnet, J.-T.; LAlloret, F AdV. Colloid Interface

Sci. 2004, 108-109, 145149.(5) Liedtke, S.; Wissing, S.; Muller, R. H.; Mader, K. Int. J. Pharm. 2000, 196,

183185.(6) Shi, R.; Hong, L.; Wu, D.; Ning, X.; Chen, Y.; Lin, T.; Fan, D.; Wu, K.

Cancer Biol. Ther. 2005, 4, 218224.(7) Solans, C.; Izquierdo, P.; Nolla, J.; Azemar, N.; Garcia-Celma, M. J. Curr.

Opin. Colloid Interface Sci. 2005, 10, 102110.(8) Pons, R.; Carrera, I.; Caelles, J.; Rouch, J; Panizza, P. AdV. Colloid Interface

Sci. 2003, 106, 129146.

(9) Rang, M. J.; Miller, C. A. J. Colloid Interface Sci. 1999, 209, 179192.(10) Rang, M. J.; Miller, C. A. Prog. Colloid Polym. Sci. 1998, 109, 101117.(11) Nishimi, T.; Miller, C. A. J. Colloid Interface Sci. 2001, 237, 259266.(12) Sole`, I.; Maestro, A.; Pey, C. M.; Gonzalez, C.; Solans, C.; Gutierrez,

J. M. Colloids Surf., A 2006, 288, 138143.(13) Nishimi, T.; Miller, C. A. Langmuir 2000, 16, 92339241.(14) (a) Bataller, H.; Lamaallam, S.; Lachaise, J.; Graciaa, A.; Dicharry, C.

J. Mater. Process Tech. 2004, 152, 215220. (b) Lamaallam, S.; Bataller, H.;Dicharry, C.; Lachaise, J. Colloids Surf., A 2005, 270-271, 4451.

(15) Leaver, M. S.; Olsson, U.; Wennerstrom, H. J. Chem. Soc., FaradayTrans. 1995, 91, 42694274.

(16) Izquierdo, P.; Esquena, J.; Tadros, Th. F.; Dederen, C; Garcia, M. J.;Azemar, N.; Solans, C. Langmuir 2002, 18, 2630.

(17) Morales, D.; Gutierrez, J. M.; Garca-Celma, M. J.; Solans, C. Langmuir2003, 19, 71967200.

(18) Izquierdo, P.; Esquena, J.; Tadros, Th. F.; Dederen, J. C.; Feng, J.; Garcia-Celma, M. J.; Azemar, N.; Solans, C. Langmuir 2004, 20, 65946598.

(19) Izquierdo, P.; Feng, J.; Esquena, J.; Tadros, Th. F.; Dederen, J. C.; Garcia-Celma, M. J.; Azemar, N.; Solans, C J. Colloid Interface Sci. 2005, 285, 388394.

(20) Morales, D.; Solans, C.; Gutierrez, J. M.; Garca-Celma, M. J.; Olsoson.,U. Langmuir 2006, 22, 30143020.

(21) Forgiarini, A.; Esquena, J.; Gonzalez, C.; Solans, C. Prog. Colloid Polym.Sci. 2000, 115, 3639.

(22) Forgiarini, A.; Esquena, J.; Gonzalez, C.; Solans, C. Prog. Colloid Polym.Sci. 2001, 118, 184189.

6092 Langmuir 2008, 24, 6092-6099

10.1021/la800624z CCC: $40.75 2008 American Chemical SocietyPublished on Web 05/20/2008

In these cases, emulsions can be generated spontaneously.Therefore, the low-energy methods can be considered to generateemulsions with small droplet sizes through this spontaneousemulsification.

Particular attention has been directed to nonionic surfactants,for which PEO hydration can be changed both by temperature(phase inversion temperature (PIT) method)1620 and composition(emulsion inversion point (EIP) method).1,21,22 With the PITmethod, emulsions stabilized by appropriate nonionic surfactantsare obtained by rapid temperature changes passing through thehydrophile-lipophile balance (HLB) temperature. Nanoemul-sions can be obtained only when the oil is completely dissolvedin a single phase prior to the nanoemulsification.1620 Theformation mechanism has been investigated by 1H-pulsed-field-gradient-spin-echo NMR (PFGSE-NMR), indicating that athermally induced disruption of a bicontinuous microemulsionis necessary to generate nanoemulsion droplets.19 In the EIPmethod, water is added dropwise to a mixture of surfactant andoil at constant temperature.1,12,21,22 The formation of nanoemul-sions is generally attributed to phase instabilities duringemulsification, where the presence of lamellar crystallites and/orbicontinuous microemulsions are thought to play criticalroles.1,21,22 However, since the systems may pass through variousdifferent phases on the route to nanoemulsion formation, it is farfrom clear which phase (if any) is key.

Nanoemulsions (so-called miniemulsions) stabilized by ionicsurfactants have also been prepared by dilution of microemulsionswith excess water.8,14,23,24 In the work reported by Pons et al.8,nanoemulsions can be obtained either by dilution of a bicon-tinuous, a O/W microemulsion, or multiphase mixtures with watercontents intermediate between the two microemulsion regions.The flexibility of the surfactant film, its affinity for water, andthe interfacial tension were suggested as important factors in theformation of nanoemulsions in two complex systems studied byBataller et al.14 However, no systematic study into the mechanismof nanoemulsion formation has been conducted for ionic surfactantsystems. Therefore, further effort is required in order to betterunderstand the mechanisms of nanoemulsion formation underisothermal conditions, with both nonionic and ionic surfactants,this will help in optimizing nanoemulsification processes forindustrial and research applications.

An isothermal two-step process, which requires no heat orenergy input, has recently been demonstrated to generatenanoemulsions with an industrially relevant system composedof methyl decanoate/technical grade surfactant poly(oxyethylene)7-lauryl ether AEO-7/water.25 In this system, initial concentrateswith different water weight fractions were rapidly diluted withwater to reach the same final composition. Although the oil andsurfactant are industrial grade, these previous results25 suggesta relationship between the formation of nanoemulsions and theinitial concentrate equilibrium-phase behavior. Bluish transparentO/W nanoemulsions with a narrow size distribution were formedonly when the concentrate was located in a microemulsion(bicontinuous and/or oil-in-water) region. The formation of O/Wnanoemulsions was attributed to homogeneous nucleation of oilfrom the microemulsion phase upon dilution. The use of thistwo-step process allow not only generation of nano-emulsions, but also improved understanding of the mechanismof nanoemulsion formation.

Therefore, this low-energy two-step process was employed inthe present work for several academically relevant systems,

formulated from research grade components and well-knownsurfactants: decane/pentaoxyethylene lauryl ether (C12E5)/water,decane/dodecyldimethylammonium bromide (DDAB)/water, de-cane/sodium bis(2-ethylhexyl)sulfosuccinate (AOT)/water (with100 mM NaCl), dodecane/sodium n-dodecyl sulfate (SDS)/pentanol/water, and dodecane/hexadecyltrimethylammoniumbromide (CTAB)/pentanol/water. In this work, both contrastvariation small-angle neutron scattering (SANS) and dynamiclight scattering (DLS) were employed to gain insight into thestructure and stability mechanisms of model nanoemulsionsformulated from decane/C12E5/water mixtures at 25 C. SANShas been used previously to study silicone oil-in-water na-noemulsions, which were generated by a high-energy extremeshear method and stabilized by the anionic surfactant SDS, overa range of nanoemulsion volume fractions (0.008 < < 0.6).26As increases, the primary structure factor peak increases insize, indicating stronger interactions and deformation of thedroplets at high above the jamming point. The systems studiedhere contrast with those published previously19,26 in three essentialways: (a) they are formed by a low-energy dilution method,rather than PIT method19 or high-energy shear;26 (b) a directmethod SANS was used to gain insight into the equilibriumstructure and stability mechanisms of the resultant nanoemulsions,rather than the dynamic 1H-PFGSE-NMR technique;19 and (c)nonionic, rather than ionic, surfactants have been used,26minimizing electrostatic structure factor effects which couldcomplicate the detailed analyses of SANS data.

Therefore this new work contributes to the studies onnanoemulsions in three important ways: (1) the generality of thelow-energy dilution method has been explored for systems withionic, as well as nonionic surfactants; (2) internal droplet structuresof nanoemulsions can be assigned with better confidence, basedon model fitting of contrast variation SANS data; and (3) studieswith complementary scattering techniques SANS and DLS, bothsensitive to different relevant length scales, provide new insightinto the instability mechanism, suggesting it is flocculationdominated, at least up to 4 h after the initial nanoemulsionpreparation.

2. Experimental Section2.1. Materials. C12E5 (>98% purity), AOT (>98% purity), and

DDAB (98% purity) were purchased from Sigma-Aldrich, and SDS(99% purity) and CTAB (98% purity) from Alfa Aesar. n-Decane(Alfa Aesar, >99% purity), n-decane-d22 (Fluka, >99% D atom),n-dodecane (Aldrich, >99% purity), 1-pentanol (Aldrich, >99%purity), and D2O (Aldrich, 99.9% D atom) were used as received.Milli-Q ultrapure water with resistivity no less than 18.4 M cmwas used.

2.2. Phase Diagrams. All components were weighed, sealed inampules, and homogenized with a vibromixer. The samples wereequilibrated at 25 C. Optically anisotropic liquid crystalline phaseswere identified by using polarizing light microscopy (OptipHot-2,Nikon, Japan). The boundary lines were found by consecutive additionof one component to mixtures of the other components.

2.3. Preparation Method. Emulsions were prepared for severaldifferent systems. For the decane/C12E5/water system, the partialternary phase diagram is shown in Figure 1, and sample preparationpathways, outlined below. The nanoemulsion volume fraction, givenby the oil volume fraction plus the surfactant volume fraction ( )decane + C12E5) was varied from 0.006 to 0.120. Two formulationsA (decane ) 0.020, C12E5 ) 0.010) and B (decane ) 0.017, C12E5) 0.013) were chosen as the final systems. There are two importantcomposition parameters in the work, the overall droplet volume

(23) Taylor, P.; Ottewill, R. H. Colloids Surf., A 1994, 88, 303316.(24) Taylor, P.; Ottewill, R. H. Prog. Colloid Polym. Sci. 1994, 97, 199203.(25) Wang, L. J.; Li, X. F.; Zhang, G. Y.; Dong, J. F.; Eastoe, J. J. Colloid

Interface Sci. 2007, 314, 230235.(26) (a) Graves, S.; Meleson, K.; Wiliking, J.; Lin, M. Y.; Mason, T. G J. Chem.

Phys. 2005, 122, 134703/1134703/6. (b) Mason, T. G.; Graves, S.; Wiliking,J. N.; Lin, M. Y. J. Phys. Chem. B 2006, 110, 2209722102.

Nanoemulsions Prepared by a Two-Step Process Langmuir, Vol. 24, No. 12, 2008 6093

fraction (defined above) and the weight fraction of water in theinitial concentrate w. Note, with reference to Figure 1, sample Aends up in an emulsion region, whereas sample B, at slightly highersurfactant concentration, is in a microemulsion region. The followinglow-energy dilution pathways were followed to arrive at a suite ofsamples A and B: (I) the one-step process, where components neededto generate final compositions A and B were mixed by gentle magneticstirring at 25 ( 0.1 C (LTD 6G, Grant, England); (II) the two-stepprocess, where, first, oil, surfactant, and an appropriate amount ofwater were mixed to generate a sequence of concentrates containingwater weight fractions w ) 0, 0.1, 0.2, . . . , 0.9 (these concentratesare marked in Figure 1 and denoted as A0, A1, A2, . . . , A9 and B0,B1, B2, . . . , B9, respectively), and then, a certain amount of oneof these concentrates was injected into an appropriate amount ofwater under gentle stirring to achieve the final composition, eitherA or B. As indicated in Figure 1 the dilution steps were arrangedso that A0 w A, A1 w A, A2 w A, . . . , A9 w A and B0 w B, B1w B, B2 w B, . . . , B9 w B. The temperature was kept constantat 25 ( 0.1 C.

Other systems with different ionic surfactants, decane/DDAB/water,27 decane/AOT/water (100 mM NaCl),28 dodecane/SDS/pentanol/water,29 and dodecane/CTAB/pentanol/water,29 were for-mulated to prepare emulsions by the two-step dilution process asdescribed above. The appropriate formulations were chosen asconcentrates from literature2729 and then injected into water toyield emulsions with final volume fraction of ) 0.030.

2.4. Dynamic Light Scattering. DLS ( 500 nm) is commonlyemployed to study droplet size distributions in emulsions andnanoemulsions.1,1225 Analysis of the autocorrelation function decayobtained by DLS30 from a dilution series of nanodroplets yieldsinformation on the effective diffusion coefficient D. In the limit ofinfinite dilution, the apparent hydrodynamic radius rh can be estimatedfrom

rh )kT

6D (1)where is the viscosity of the medium. The apparent hydrodynamicradius includes solvation effects. DLS experiments were carried outusing a Malvern 4800 Autosizer (Malvern Instruments, England).

In self-assembling systems such as emulsions, nanoemulsions,and microemulsions, DLS is particularly sensitive to colloidal

interactions on a length scale of 2-1000 nm; hence, the apparentsize can become obscured if the samples are highly polydisperse,insufficiently dilute, or experiencing attractive/repulsive interactions.Hence, there are potential limitations to structural studies ofnanoemulsions by DLS alone.

2.5. Small-Angle Neutron Scattering Experiments. In com-parison to DLS, SANS covers length scales which are associatedwith the internal structure of nanoemulsion droplets (1-10 nm).31An additional advantage of SANS over DLS is that contrastvariation can be used to obtain detailed structural information onthe droplets by selective deuteration of the various components.

The time-of-flight LOQ instrument at ISIS, U.K., was used; thedata acquisition, treatment, and normalization procedures have beenoutlined elsewhere.3234 Samples were held in Hellma quartz cellsand thermostated at 25 C. Following standard procedures, raw datawere treated to yield normalized scattering intensities I(Q) in cm-1,where the momentum transfer is related to the scattering angle, ,by Q) (4/) sin(/2). The observed Q range was 0.007-0.23 -1.Scattering data were also corrected for wavelength-dependenttransmission factors, as well as cell, background, and any incoherentscattering. The overall nanoemulsion drop contrast was generatedwith decane-h/C12E5-h/D2O, and the external surfactant shellcontrast was highlighted with decane-d/C12E5-h/D2O. Data wereanalyzed using standard Guinier limiting laws, Porod analysis, andthe multimodel FISH fitting program.35 Details of these analysescan be found in the Supporting Information.

3. Results and Discussion3.1. Samples Prepared by Methods I and II for the Decane/

C12E5/Water System. The partial ternary-phase diagram of thesystem decane/C12E5/water at 25 C is shown in Figure 1. Ascan be seen from the figure, samples A and B with volumefraction ) 0.030 are located in the emulsion (A) andmicroemulsion regions (B) of the phase diagram, respectively.As the actual phase behavior is quite complicated,15 the exact

(27) Blum, F. D.; Pickup, S.; Chen, S. J.; Evans, D. F. J. Phys. Chem. 1985,89, 711713.

(28) Chen, C. H.; Chang, S. L.; Strey, R.; Samseth, J.; Mortensen, K. J. Phys.Chem. 1991, 95, 74277432.

(29) Sripriya, R.; Muthu Raja, K.; Santhosh, G.; Chandrasekaran, M; Noel,M. J. Colloid Interface Sci. 2007, 314, 712717.

(30) Pecora, R. Dynamic Light Scattering; Plenum: New York, 1985.(31) Eastoe, J. Surfactants; Wuhan University Press: Wuhan, China, 2005; pp

96-134.(32) Dupont, A.; Eastoe, J.; Murray, M.; Martin, L.; Guittard, F.; Givenchy,

E. T.; Heenan, R. K. Langmuir 2004, 20, 99539959.(33) Eastoe, J.; Dominguez, M. S.; Wyatt, P.; Orr-Ewing, A. J.; Heenan, R. K.

Langmuir 2004, 20, 61206126.(34) Summers, M.; Eastoe, J.; Davis, S.; Du, Z.; Richardson, R.; Heenan,

R. K.; Steytler, D. C.; Grillo, I. Langmuir 2001, 17, 53885397.(35) Heenan, R. K. Fish Data Analysis Program; Report RAL-89-129,

Rutherford Appleton Laboratory, CCLRC: Didcot, U.K., 1989.

Figure 1. Partial ternary phase diagram of the decane/C12E5/water system at 25 C. Samples A and B, generated from A5 and B5, respectively, areshown in the inset photographs. The preparation protocol II is outlined in section 2.3. (Om, inverse micellar solution or W/O microemulsion; Lc,optically anisotropic phase; MLc, multiphase region including liquid crystal phase; ME, bicontinuous microemulsion or O/W microemulsion; Em,multiphase region).

6094 Langmuir, Vol. 24, No. 12, 2008 Wang et al.

microstructures of the liquid crystal phases have not been fullyidentified.

The appearance of samples A and B prepared by the gentlestirring method I, at different times, are shown in Figure 2. Itcan be seen that a coarse, white emulsion was obtained by gentlestirring for sample A (emulsion region). On the other hand sampleB became transparent over time, which is consistent with eventualmicroemulsion formation, as expected on the basis of the phasediagram shown in Figure 1. Samples A and B prepared by methodI were studied by DLS 24 h after preparation. System A is awhite coarse emulsion with radii 290 nm, whereas thetransparent system B gave rh ) 9.3 nm, exactly the same as foran equivalent B sample but formulated by the dilution methodII (see below). The apparent droplet size for the presumedmicroemulsion sample B was constant over 6 months.

Samples A and B prepared by method II were investigated byDLS. The relationship between the final droplet radius and thewater weight fraction w in the concentrate is shown in Figure3. It can be seen that, with system A, the droplet radii vary withwater level w in the concentrates. Bluish transparent nanoemul-sions with droplet radii as low as about 15 nm were obtained ondilution of concentrates with w ) 0.4 and 0.5 (A4 and A5; seeFigure 1). For concentrates with w less than 0.3 (A0-A3; referto Figure 1) or more than 0.6 (A6-A9; refer to Figure 1), whitecoarse emulsions with radii larger than 100 nm were obtained.On the other hand, the droplet radii for systems B remainessentially constant at about 9 nm, regardless of water level inthe concentrate. It can be seen that microemulsions formspontaneously, and the scattering and visual properties areindependent of the preparation pathway. However, appearanceand the droplet radii of the emulsions from pathway A are clearly

dependent on the preparation procedure, and crucially the waterlevel in the concentrate.

Finally, comparing the DLS results for system A, prepared bythe two separate routes I and II (rh ) 290 and 15 nm if generatedfrom A5), shows that different states are achieved. This indicatesthat the formulation pathway to nanoemulsion A is an importantfactor.

3.2. Relationship between the Formation of Nanoemulsionsand Equilibrium Phase Behavior in the Decane/C12E5/WaterSystem. Electrical conductivity measurements on concentratesof systems A with initial water weight fractions w in the range0.34-0.57 are consistent with the presence of bicontinuousmicroemulsions (see Supporting Information). It can be seenfrom Figure 1 that concentrates A4 and A5 are located in thebicontinuous microemulsion region, while other concentratesare located in different regions of the phase diagram. Nanoemul-sions with small droplet radii are formed only when theconcentrate starts as a bicontinuous microemulsion, with w of0.4 and 0.5 (samples A4 and A5) (Figure 3). This suggests a closerelationship between the equilibrium phase behavior of the initialconcentrate and the final droplet radius of the resulting emulsionsafter they have been diluted to achieve A.

These different phases are consistent with changes in thesurfactant hydration and solubility of decane as a function ofdilution. It is well-known that the preferred curvature of nonionicsurfactant layers depends on hydration of the PEO, which withsurfactant-rich concentrates can be increased by dilution withwater15 (EIP method). At low water contents the preferredcurvature may be around water, generating reverse W/Ostructures. With increased water dilution, hydration of the PEOheadgroups may drive the curvature about oil to favor O/W phases.The solubilization capacity for oil is decreased during the phasetransition, leading to local supersaturation and then oil dropletnucleation: therefore, spontaneous emulsification occurs. For lowwater level systems (A0-A3), oil is the continuous phase in theseinitial concentrates. The existing oil droplets may act as nucleiand trigger heterogeneous nucleation,36 which result in dropletswith larger sizes and polydispersity. On the other hand inconcentrates A4 and A5, the oil was completely solubilized in abicontinuous microemulsion phase (see Supporting Information).The dilution of these concentrates with excess water convertsthem to oil-in-water systems and decreases the concentration ofsurfactants, which also inevitably leads to a decrease in oilsolubilization. Similar to concentrates A0-A3, these systemsbecame supersaturated in oil, leading to nucleation of oil. Thepoint is that supersaturation in the single phase results inhomogeneous nucleation, which favors the formation of small-sized oil droplets with narrow polydispersity.36,37 Consideringconcentrates located in a region with higher water, the surfactantlayer is saturated with water and the mean curvature can beconsidered as water concentration independent.15 ConcentratesA6-A9 are located in the emulsion (Em) region, in which theoil droplets coexist with other phase(s). Highly polydisperseemulsions were obtained by dilution of water because no curvature(structural) change is induced by added water; the emulsion justbecome less concentrated.

Therefore, nanoemulsions were obtained by dilution ofbicontinuous microemulsion, as a result of homogeneousnucleation during a spontaneous emulsification. These resultsare consistent with previous reports that nanoemulsions were

(36) Vincent, B.; Kiraly, Z.; Obey, T. M. In Modern Aspects of EmulsionScience; Binks, B. P., Ed.; The Royal Society of Chemistry: Cambridge, U.K.,1998; pp 100-114.

(37) Morris, J.; Olsson, U.; Wennerstrom, H. Langmuir 1997, 13, 606608.

Figure 2. Appearance of A and B prepared by method I at differenttimes.

Figure 3. Droplet radii for samples A (0) and B ()) with volume fraction of 0.030 at 25 C as a function of w, the weight fraction of water inthe initial concentrate. Inset: appearance of emulsion diluted from differentconcentrates.

Nanoemulsions Prepared by a Two-Step Process Langmuir, Vol. 24, No. 12, 2008 6095

obtained only when the oil was completely dissolved in a singlephase prior to the nanoemulsification through the PIT meth-od.1620 Furthermore, similar results have also obtained by anisothermal two-step process in a system with technical grade oiland surfactant.25

3.3. Emulsions Prepared by Method II in Systems withIonic Surfactants. To explore the generality of the mechanismof nanoemulsion formation, a range of concentrates wereformulated with different surfactants. Literature was used toguide the formulation of appropriate concentrates withDDAB,27 AOT,28 and CTAB-pentanol and SDS-pentanol.29

For systems with DDAB, all concentrates includingbicontinuous microemulsions do not disperse in water at all;coarsely dispersed oil droplets were observed during stirringwhich phase-separated as soon as stirring was stopped. Thisis consistent with the published phase behavior of the decane/DDAB/water system:27 addition of an appropriate amount ofwater transforms the bicontinuous concentrate into a discon-tinuous discrete droplet water-in-oil microemulsion, and thenfurther dilution with water places the samples in an undefined(emulsion) phase region. There is no spontaneous curvaturetransition to an oil-in-water structure in this system. Therefore,it may be reasonable to assume that spontaneous emulsificationis also required for nanoemulsion formation as well as thecomplete solubilization of oil in a bicontinuous microemulsion.

Crash dilution of concentrates with AOT, CTAB-pentanol,and SDS-pentanol in water, to yield final volume fractions of) 0.030, resulted in emulsions having droplet radii of the order70 nm, when the concentrates were located in a W/O orbicontinuous microemulsion phase region (see the SupportingInformation, Table S1). Phase separation was observed in thesesystems after 3 days. This means nanoemulsions can be preparedby dilution of concentrates even when they are not bicontinuousmicroemulsions. Due to the decrease of ionic strength13,14 andpartitioning of alcohol,9,10 as reported elsewhere, dilution of theseconcentrates with water allows a change of spontaneous curvaturefrom water-in-oil to a bicontinuous microemulsion and then toan oil-in-water configuration.

3.4. Nanoemulsion Droplet Structure and Stability De-termined by SANS. 3.4.1. Nanoemulsion Droplet Structure 30min after Preparation. Because the nanoemulsions preparedwith decane/C12E5/water have droplet sizes ideally matchedto SANS (up to 60 nm), with the additional benefit of anabsence of interdroplet electrostatic interactions, this neutronscattering method can be used to generate detailed structuralinformation.

Figure S2 in the Supporting Information shows the SANSprofile from concentrate A5, exhibiting a correlation peak at Q 0.03 -1, consistent with a bicontinuous microemulsion at ) 0.50. Nanoemulsions (sample A prepared by dilution fromA5), at drop contrast decane-h/C12E5-h/D2O, and shell contrastdecane-d/C12E5-h/D2O were studied by SANS 30 min afterpreparation. These two contrasts can be clearly distinguished ona log-log plot, as can be seen in Figure 4, and there is an absenceof any S(Q)-type correlation peak.

The drop contrast data were initially analyzed by Guinier andPorod approximations and then finally fitted using the multimodelFISH program (details given in the Supporting Information).The Guinier plots show significant linearity at low Q for spheres(ln I(Q) vs Q2, Figure S3 in the Supporting Information); nolinearity in an appropriate region of Q space was discerned forcylinders ln [I(Q)Q] vs Q2 or discs (ln [I(Q)Q2] vs Q2). Therefore,a Guinier analysis suggests the nanoemulsions exist in the formof spherical (globular) droplets. Estimates for the nanoemulsion

radii R were obtained from the Guinier plot, which is valid onlyfor very low Q (QR < 1); at high Q, the SANS intensity issensitive to scattering from local interfaces. In this regime,the Porod approximation was used to estimate particle radii andthe droplet concentration Cdrop, via the total specific area (see theSupporting Information). These values were used as startingpoints for a more detailed analysis, using a model of Schulzpolydisperse spheres.38,39 The data are described very well bythis model, with radii quite close to those obtained by the Guinierand Porod analyses (see Table 1). These analyses were consistentwith the view that the droplets exist in the form of oil spheres.

A core-shell model35 was employed to fit the shell contrastdecane-d/C12E5-h/D2O. This model described the profile verywell, which suggests that the nanoemulsion with an oil core issurrounded by a surfactant shell of apparent film thickness, ts.

The respective model fits, and associated parameters, are shownin Figure 4 and Table 1. It is shown that the overall radius ofthe nanoemulsion (sample A) is 10.2 nm and the thickness ofthe surfactant layer is 1.1 nm. This outer layer thickness reflectsreasonably well the dimension of a C12H25-h chain (1.7 nm) andwould be consistent with minimal contrast difference betweenthe outer hydrated EO-5 groups and the external D2O solvent,as would be expected under highly solvated conditions. The

(38) Eastoe, J.; Hetherington, K. J.; Sharpe, D.; Dong, J.; Heenan, R. K.;Steytler, D. C. Langmuir 1996, 12, 38763880.

(39) Nave, S.; Eastoe, J.; Heenan, R. K.; Steytler, D.; Grillo, I. Langmuir 2000,16, 87418748.

Figure 4. SANS profiles of the drop and shell contrasts of thenanoemulsion (sample A, at ) 0.030) at 30 min and 25 C. Error barsare shown. Lines are fits to a polydisperse sphere model described inthe Supporting Information. Inset: Schematic outlines of the contrastarrangements.

Table 1. Values Obtained from the SANS Data fromNanoemulsions at Different Times (Sample A at O ) 0.030)a

sample RGuinier/nm RPorod/nm Cdrop/(mol dm-3) RFit/nm ts/nmA, 30 min 12.7 11.9 15 10.2A shell, 30 min 10.1 1.1A, 120 min 12.7 11.9 15 10.0A, 240 min 12.3 11.6 15 9.9

a /Rc ) 0.23 ( 0.03, hs ) decane + surfactant, Rhs ) RPorod ( 0.8 nm.Parameters: RGuinier and RPorod are radii estimated by the Guinier and Porodapproximations; Cdrop is the calculated nanoemulsion droplet concentrationas described in the Supporting Information; RFit is the average radius givenby Schulz polydisperse spheres fitting routine; ts is the apparent shell thicknessgiven by the core-shell model; /Rc is the width of the Schultz distributionfunction.

6096 Langmuir, Vol. 24, No. 12, 2008 Wang et al.

overall droplet radius determined by SANS matches quite thevalue determined by DLS (15 nm) considering the need toinclude solvation of the surfactant shell. The polydispersityvalue is about 0.23, which reflects a narrow distribution ofnanoemulsion droplet radii, and typical of microemulsiondroplets using this model.38,39

Because of preexisting literature,40 the structure of dilutemicroemulsion B was not investigated by SANS here. Comparisonof the composition for system B [) 0.030, andR) 0.50) (massfraction wtdecane)/(wtdecane + wtC12E5)]40c) with those investigatedpreviously40a,b clearly suggests spherical decane-in-water mi-croemulsion droplets of radius 9 nm. Therefore, the DLS radiifor the microemulsion system B shown in Figure 3 are consistentwith previously published data.40 Note the microemulsion dropletradius is consistently lower than for the nanoemulsion A (Figure3 and ref 40), suggesting different underlying structures for thetwo systems A and B.

3.4.2. Effect of Time on Nanoemulsion Droplets. The evolu-tion of droplet size is an important aspect of emulsion stability.Nanoemulsions were characterized by SANS, at different timesafter the initial dilution to generate system A. No obvious changesin the shape or intensity of the SANS were observed. Similardetailed analyses as outlined above in section 3.4.1 wereperformed, and the resulting values are shown in Table 1 whilethe fitting lines are displayed in Figure 5. It can be seen that nonotable changes in the structure or radius of the droplets wasobserved over the test time of 4 h after the initial dilution. Thisis most obviously seen in the inset to Figure 5, which magnifiesthe supersensitive high Q Porod region: increases in droplet sizewould result in shifts of the maxima/minima to lower Q values.The constant size demonstrates that the nanoemulsion dropletsin this system are quite stable during the test period againstcoalescence41 and Ostwald ripening,8,42 both of which would beexpected to give rise to increases in droplet size.

3.4.3. Structure of Nanoemulsions as a Function of Dilution.Systems were prepared by dilution with different water contents

from concentrate A5, located in the microemulsion region of thephase diagram.

Figure 6 shows SANS data for the nanoemulsions as a functionof volume fraction , and it can be seen that the shape of theSANS profile changes little over this range. The increase inintensity with is an indication of an increase in the nanoemulsiondroplet number density. This can be seen clearly from the near-linear relationship between the droplet concentration Cdrop(calculated from the limiting high-Q intensity on the Porod plot)and (details can be seen in Supporting Information, Figure S7)and further verified by the linear relationship between the fittedscale factor (fitted as polydisperse spheres) and (can be seenin Figure S8 in the Supporting Information). Fitted and derivedparameters are summarized in Table 2 along with the resultsobtained from the Guinier and Porod approximations. As is shownin Table 2, there is no significant change in the droplet size with (except for the R obtained from the Guinier approximation at) 0.120, which may be attributed to the more prominent S(Q),which is exhibited by the downward slope at low Q). Nanoemul-sions with radius of the order of 10 nm were obtained, independentof , and polydispersity indices were about 0.23, indicating anarrow size distribution. It can then be said that nanoemulsionsmay be obtained when the initial concentrates are located in a(40) (a) Menge, U.; Lang, P.; Findenegg, G. H.; Strunz, P. J. Phys. Chem. B

2003, 107, 13161320. (b) Menge, U.; Lang, P.; Findenegg, G. H. J. Phys. Chem.B 1999, 103, 57685774. (c) The composition parameter used in refs 40a and 40bis R is an oil mass fraction ) wt decane/(wt decane + wtC12E5). Translating nowto total droplet volume fraction used in this paper, spherical droplets wereobserved in ref 40b for ) 0.0141-0.141 and above R ) 0.25: the compositionof system B studied here is ) 0.030 and R ) 0.5.

(41) Deminiere, B.; Colin, A.; Calderon, F. L.; Bibette, J. In Modern Aspectsof Emulsion Science; Binks, B. P., Ed.; The Royal Society of Chemistry: Cambridge,U.K., 1998; pp 261-291.

(42) Tadros, T.; Izquierdo, P.; Esquena, J.; Solans, C. AdV. Colloid InterfaceSci. 2004, 108-110, 303318.

Figure 5. SANS profiles for nanoemulsions at ) 0.030 at differenttimes t ) 30 (0), 120 ()), and 240 min (). Error bars are shown. Linesare model fits to a polydisperse sphere model described in the SupportingInformation. Inset: shown is the corresponding Porod plot.

Figure 6. SANS profiles for nanoemulsions at various volume fractions, ) 0.120 (0), 0.060 ()), 0.030 (), 0.015 (O), and 0.006 (b). Errorbars are shown. Lines are model fits to a polydisperse sphere modeldescribed in Supporting Information. Inset: shown is the Porod plotcorresponding to the SANS profiles.

Table 2. Values Obtained from the SANS Data fromNanoemulsions at Various Volume Fractions Oa

RGuinier/nm RPorod/nm Cdrop/(mol dm-3) RFit/nm0.120 7.3 11.3 69 9.40.060 11.2 12.2 28 9.60.030 12.7 11.9 15 10.20.015 13.0 11.3 7 10.30.006 12.5 11.3 2 9.7

a /Rc ) 0.23 ( 0.03, HS ) decane + surfactant, RHS ) RPorod ( 0.8 nm.Parameters: RGuinier and RPorod are radii estimated by the Guinier and Porodapproximations; Cdrop is the calculated nanoemulsion droplet concentrationas described in the Supporting Information; RFit is average radius given bySchulz polydisperse spheres fitting routine; /Rc is the width of the Schultzdistribution function.

Nanoemulsions Prepared by a Two-Step Process Langmuir, Vol. 24, No. 12, 2008 6097

microemulsion region, regardless of the water content used fordilution. Eventual droplet sizes are mainly controlled by thestructure of the concentrate and are independent of dilution. Thisis different from the results reported by Mason et al.26 in thatthe structure of nanoemulsions deforms from hard sphere to aglassy structure at high volume fraction (with larger than 0.3)through screened surface charge repulsions, perhaps becausethose systems were stabilized by anionic SDS. Due to the lowervolume fraction and zero effective surface charge in the systeminvestigated here, the droplets of nanoemulsions are discretespheres at all test volume fractions, and excess water appears toact only as a dilution medium without having any effect on thestructure.

3.5. Stability of Nanoemulsions Determined by DLS. Phaseseparation of nanoemulsion sample A (prepared from A5) wasobserved after 3 days (see inset to Figure 7), whereas themicroemulsion samples B showed no phase change when left forup to 10 weeks. These results suggest that nanoemulsions areactually thermodynamically unstable.

To follow the breakdown process, nanoemulsion droplet sizeswere determined by DLS as a function of and time; theseresults are shown in Figure 7. When measured 15 min afterpreparation, similar droplet radii were obtained for the differentvolume fractions, agreeing with the SANS results described above.However, the DLS-determined droplet sizes show a markedincrease with time; the higher is, the faster the droplet sizeincreased. These results are contradictory to those obtained bySANS which suggest no change in the underlying dropletstructure, at least 4 h after preparation from the stock concentrates(see Figure 5).

These discrepancies are at least consistent with the lengthscale resolutions of the two scattering techniques: SANS is ideallymatched to examine the discrete droplet structure for dimensionsup to 30 nm or so; DLS being particularly sensitive to collectivemotions resulting from attractive interactions of particles, certainlywith dimensions 30 nm or so and above.

To explore the mechanism of instability with nanoemulsions,the results of Figure 7 were replotted as shown in Figure 8. Thechange in droplet size with time may follow eq 2 if the instabilitymechanism is coalescence:41

1r) r0

2(83 )t (2)where r is the average droplet radius after time t, r0 is the dropletradius at t ) 0, and is the frequency of rupture per unit surfaceof the film. If the mechanism is Ostwald ripening,43 the dropletradius should vary with time as

) dr3

dt )89 )

C()VmDFRT (3)

where r is the average droplet size after time t, C() is the bulkphase solubility (the solubility of the oil in an infinitely largedroplet), is the interfacial tension, Vm is the molar volume ofthe oil, D is the diffusion coefficient of the oil in the continuousphase, F is the oil density, and R is the gas constant.

Figure 8 suggests that there is no linear variation upon plotting1/r2 or r3 as a function of time. These results indicate that neithercoalescence nor Ostwald ripening are the underlying mechanismsfor the nanoemulsion instability, which are consistent with theresults obtained from SANS in section 3.4.2.

Comparing the DLS and SANS results, it is reasonable toassume that droplet-droplet hydrodynamic interactions existwithout droplet deformation or rupture of the surfactant filmduring the test period. Therefore, the mechanism for the instabilityin this system may be attributed to flocculation,43 where dropscluster without significant rupture of the stabilizing interfaciallayer. The droplet radius determined by DLS may be attributedto an effective cluster radius, which increases as more dropletsaggregate with time through flocculation. These aggregates riseunder gravity because of the density difference between thedispersed oil and the continuous phase, resulting in creamingand eventually phase separation, presumably with accompanyingcoalescence. At least over the time scale studied here, up to 4 h,it can be said that the individual droplet radius remains constant:SANS data are inconsistent with growth of individual droplets,which would be an inevitable consequence of both coalescenceand/or Ostwald ripening.

The rate of flocculation depends on the product of a frequencyfactor (how often drops encounter each other) and a probabilityfactor (how long they stay in contact).43 Thus it is easy tounderstand that the higher volume fraction (in which dropshave a higher frequency of encounters), the faster flocculationoccurs, which results in the faster increase in droplet radius asdetermined by DLS (as can be seen in Figure 7). Tests wereperformed by DLS, swapping H2O for the higher density D2Oto assess if this would affect the flocculation (Figure S10), butno significant differences were observed using this technique.

4. ConclusionsAn isothermal two-step process (method II) has been applied

to oil/surfactant/water systems with C12E5, DDAB, AOT,SDS-pentanol, and CTAB-pentanol. Oil-in-water nanoemul-sions have been obtained in the decane/C12E5/water system, bycrash-dilution of a bicontinuous microemulsion into a largevolume of water. The structure of nanoemulsions obtained thisway is mainly controlled by the structure of the initial concentrate,being apparently independent of the dilution factor with water.Coupled with the results gained with systems stabilized by DDAB,AOT, SDS-pentanol, and CTAB-pentanol, a general mech-anism for nanoemulsion formation may be postulated ashomogeneous nucleation of oil droplets during spontaneousemulsification.

(43) Binks, B. P. In Modern Aspects of Emulsion Science; Binks, B. P., Ed.;The Royal Society of Chemistry: Cambridge, U.K., 1998; pp 17-38.

Figure 7. Dependence of apparent droplet radii of nanoemulsions againsttime, determined by DLS, at various volume fractions, ) 0.120 (0),0.060 ()), 0.030 (), 0.015 (O), and 0.006 (b). Inset: the observedeventual phase separation of a nanoemulsion (sample A) after 3 days.

6098 Langmuir, Vol. 24, No. 12, 2008 Wang et al.

The droplet sizes of nanoemulsions have been determinedboth by DLS and SANS. The two techniques show no strikingdifferences in the nanoemulsion radii if newly prepared. However,droplet structure determined by SANS shows no change over4 h, while the apparent DLS droplet size increases with time.This discrepancy suggests flocculation may be responsible forthe nanoemulsion instability (at least up to emulsion ages of 4 h),which is different from the mechanisms widely reported before,such as coalescence and Ostwald ripening.8,41,42

This work gains insight into the general mechanism ofnanoemulsion formation, which can be used to guide thepreparation of nanoemulsions in a wide range of systems. Thistwo-step process is easy to scale up and has low energyconsumption, which is of great interest for practical applications.The findings verify that the two-step process can be used for thepreparation of nanoemulsions, which are stable without an

increase in the radii of individual droplets over several hours.This new nanoemulsification route paves the way to new potentialapplications of these easy-to-prepare systems in fields such asfoods, pharmaceuticals, and agrochemicals.

Acknowledgment. We acknowledge the National NaturalScience Foundation of China (Grant NSFC 20573079) and theMinistry of Science and Technology (Grant 2006 BAE01A07-5) for financial support. STFC (U.K.) is thanked for provisionof beam time at ISIS and for financial support for travel andconsumables.

Supporting Information Available: Details of the conductivitydata, SANS analysis, and theoretical background (pdf). This materialis available free of charge via the Internet at http://pubs.acs.org.

LA800624Z

Figure 8. DLS derived radii, 1/r2 as a function of time (a) and r3 as a function of time (b) for nanoemulsions at various volume fractions, ) 0.120(-0-), 0.060 (-)-), 0.030 (--), 0.015 (-O-), and 0.006 (-b-).

Nanoemulsions Prepared by a Two-Step Process Langmuir, Vol. 24, No. 12, 2008 6099