The relevance of the amorphous state to …...Many pharmaceuticals, either by accident or design,...

The relevance of the amorphous state to pharmaceutical dosageforms: glassy drugs and freeze dried systems

Craig, D. Q. M., Royall, P. G., Kett, V., & Hopton, M. L. (1999). The relevance of the amorphous state topharmaceutical dosage forms: glassy drugs and freeze dried systems. International Journal of Pharmaceutics,179(2), 179-207. https://doi.org/10.1016/S0378-5173(98)00338-X

Published in:International Journal of Pharmaceutics

Document Version:Publisher's PDF, also known as Version of record

Queen's University Belfast - Research Portal:Link to publication record in Queen's University Belfast Research Portal

General rightsCopyright for the publications made accessible via the Queen's University Belfast Research Portal is retained by the author(s) and / or othercopyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associatedwith these rights.

Take down policyThe Research Portal is Queen's institutional repository that provides access to Queen's research output. Every effort has been made toensure that content in the Research Portal does not infringe any person's rights, or applicable UK laws. If you discover content in theResearch Portal that you believe breaches copyright or violates any law, please contact [email protected].

Download date:03. Oct. 2020

https://doi.org/10.1016/S0378-5173(98)00338-X

https://pure.qub.ac.uk/en/publications/the-relevance-of-the-amorphous-state-to-pharmaceutical-dosage-forms-glassy-drugs-and-freeze-dried-systems(7534d759-ecf0-4ecd-91e6-1f6ee73c3225).html

International Journal of Pharmaceutics 179 (1999) 179–207

The relevance of the amorphous state to pharmaceutical dosageforms: glassy drugs and freeze dried systems

Duncan Q.M. Craig *, Paul G. Royall, Vicky L. Kett, Michelle L. Hopton

Centre for Materials Science, School of Pharmacy, Uni6ersity of London, 29–39 Brunswick Square, London WC1N 1AX, UK

Received 25 May 1998; accepted 6 August 1998

Abstract

Many pharmaceuticals, either by accident or design, may exist in a total or partially amorphous state. Conse-quently, it is essential to have an understanding of the physico-chemical principles underpinning the behaviour of suchsystems. In this discussion, the nature of the glassy state will be described, with particular emphasis on the molecularprocesses associated with glass transitional behaviour and the use of thermal methods for characterising the glasstransition temperature, Tg. The practicalities of such measurements, the significance of the accompanying relaxationendotherm and plasticization effects are considered. The advantages and difficulties associated with the use ofamorphous drugs will be outlined, with discussion given regarding the problems associated with physical and chemicalstability. Finally, the principles of freeze drying will be described, including discussion of the relevance of glasstransitional behaviour to product stability. © 1999 Published by Elsevier Science B.V. All rights reserved.

Keywords: Amorphous; Crystallisation; Differential scanning calorimetry; Freeze drying; Glass transition; Lyophilisa-tion

1. Introduction

The majority of solid drugs and dosage formsare prepared in the crystalline state, characterisedby a regular ordered lattice structure. In practicalterms, the physical structures of these systems are

generally thermodynamically stable and are rela-tively simple to study using techniques such asdifferential scanning calorimetry and X-ray dif-fraction. However, it has been recognized for aconsiderable period of time that pharmaceuticalmaterials may also be prepared in an amorphousform (Haleblian, 1975; Byrn, 1982), where there isno long range order. The classic example of thisapproach is that of novobiocin which was shown

* Corresponding author. Tel.: +44-171-753-5863; fax: +44-171-753-5863; e-mail: [email protected].

0378-5173/99/$ - see front matter © 1999 Published by Elsevier Science B.V. All rights reserved.

PII: S 0378 -5173 (98 )00338 -X

D.Q.M. Craig et al. / International Journal of Pharmaceutics 179 (1999) 179–207180

to exhibit favourable dissolution properties whenprepared in an amorphous as opposed to crys-talline state (Mullins and Macek, 1960). Further-more, processes such as freeze and spray dryingmay lead to the generation of amorphous systems,while grinding or conventional drying may resultin materials which are partially or whollydisordered.

The amorphous state may arise as a result ofthree sets of circumstances. Firstly, the drug, ex-cipient or delivery system may be deliberatelyproduced in an amorphous form in order to en-hance product performance characteristics. Exam-ples of this strategy include the preparation ofglassy drugs for enhanced dissolution behaviouror freeze drying, both of which will be discussedin more detail in later sections. Secondly, thematerial may be intrinsically at least partiallyamorphous at room or body temperature, exam-ples including D/L polylactic acid,polyvinylpyrrolidone or polyethylene glycol. Con-sequently, dosage forms prepared using these ma-terials will inevitably be at least partiallyamorphous. Thirdly, the amorphous state may begenerated accidentally, examples includingmilling, drying and compression. In many waysaccidental production can prove to be the mostproblematic, as the levels of disordered materialgenerated may be sufficiently large to causechanges in product performance but also toosmall to be easily detected.

While the preparation of amorphous systemsmay be desirable, there are a number ofdifficulties associated with their use. Amorphousmaterials are thermodynamically unstable andwill tend to revert to the crystalline form onstorage (devitrification); such behaviour has beenwell documented for a number of drugs (Fukuokaet al., 1986; Yoshioka et al., 1994; Hancock et al.,1995). It should be stressed at this point that theonset of the devitrification process may be so slowso as to be effectively irrelevant within the storagetime of a product, although an understanding ofthe nature and characterisation of the glass transi-tional behaviour is nevertheless essential in orderto enhance predictability of stability.

The mechanical properties and vapour sorptionprofiles of amorphous systems may be markedly

different from the crystalline material (Hancockand Zografi, 1997), while the chemical reactivityof amorphous drugs may be greater (Pikal et al.,1978). In addition, the behaviour of the systembelow and above the glass transition temperature(Tg, at which the material changes on coolingfrom a liquid or rubbery state to a brittle state)will differ; the rate of crystallisation is muchgreater above Tg, while freeze dried products areless likely to physically collapse if stored belowTg. A further consideration which is particularlypertinent to commercial uses of amorphous mate-rials is the lack of a ‘comfort factor’ associatedwith such systems. The physical structure ofglassy materials is more difficult to characteriseand quality control than is that of crystallinesystems. Similarly, the knowledge base concerningthe relationship between Tg and pharmaceuticalproduct performance is not fully developed. Fur-thermore, considerable care must be taken withregard to the use of conventional accelerated sta-bility studies in order to predict chemical or phys-ical stability, as the behaviour above and belowTg is not directly comparable (Duddu and DalMonte, 1997). In common with phenomena suchas polymorphism, the characteristics of amor-phous drugs must also be considered in the lightof regulatory requirements (Byrn et al., 1995).These difficulties have led to a number of in-stances whereby the use of amorphous drug formshas been rejected as a formulation strategy bypharmaceutical companies, despite the fact thatthis route offers a potential means of considerablyenhancing product performance.

The subject of glassy systems has received con-siderable attention in other fields such as thepolymer and food sciences (Wunderlich, 1990;Slade and Levine, 1991, 1995) and the interestedreader is referred to these and related texts formore detailed information. Similarly, the reviewsby Hancock and Zografi (1997) and Kerc' andSrc' ic' (1995) are also recommended, the formercontaining a detailed discussion of the nature ofthe glassy state of pharmaceuticals and the latterplacing emphasis on published examples of amor-phous drug systems, with particular considerationof thermal analysis. Given the breadth that thetopic of amorphous pharmaceutical systems has

D.Q.M. Craig et al. / International Journal of Pharmaceutics 179 (1999) 179–207 181

now assumed, it is not possible to cover all as-pects of the subject in a single review. In thisdiscussion, therefore, the basic principles under-lining the behaviour of amorphous materials willbe described, with particular emphasis on charac-terisation of these systems using thermal analysis.In terms of applications, the discussion will beconfined to glassy drugs and freeze dried materi-als, as these are two of the most important sys-tems in which the amorphous state may beencountered.

2. The amorphous state and glass transitions

2.1. The generation of amorphous materials bycooling from the melt

While partially or wholly amorphous solidsmay be generated via a number of routes, mosttexts describing glassy systems work on the basisthat the material has been formed by rapid cool-ing from the melt. For simplicity, this will also beassumed in the forthcoming description of thetheoretical basis of glass formation, although inpractice processes such as solvent precipitation ormilling may also lead to the generation of amor-phous systems. A number of texts are availablefor more information on the nature of the glasstransition (Elliott, 1983; Wunderlich, 1990; An-gell, 1995a; Hancock and Zografi, 1997).

The essential differences between the formationof amorphous and crystalline systems may beillustrated with respect to Fig. 1. For crystallinesystems, decreasing the temperature from the liq-uid state to the melting point (Tm) results in atransition to the crystalline form (assuming nosupercooling), which below Tm is the thermody-namically stable state with respect to non-crys-talline forms. The exothermic crystallisationprocess leads to a sudden contraction of the sys-tem due to a decrease in free volume (defined asthe difference between the total volume and theactual volume displaced by the constituentmolecules). Consequently, both the enthalpy (H)and specific volume (V) decrease at Tm. Furtherless marked decreases in the above may be seen asthe temperature is lowered as a result of heat

capacity and thermal contraction effects. It shouldbe noted, however, that in the case of water,crystallisation leads to expansion rather thancontraction.

In the case of a glass-forming material, thecooling process is too fast for the crystallisationprocess to take place, either due to the use of arapid rate of cooling or the crystallisation processbeing unfavoured because of molecular size andshape (as is the case with the majority ofproteins). No discontinuity in enthalpy or volumeis seen on cooling the material below Tm and thesystem forms a supercooled liquid. As the mate-rial is cooled further, a point is reached at whichthe material becomes ‘frozen’ into the glassy state.At this temperature, the bonding betweenmolecules remains essentially the same as that ofthe liquid but the translational and rotationalmotions of those molecules are dramatically re-duced, with principally vibrational motions takingplace below Tg. The glass transition will thus becharacterised by a step change in heat capacity Cp

which is the derivative of enthalpy with respect totemperature (((H/(T)p), hence the transition isdependent on molecular mobility with no associ-ated heat transfer for the process. The transition

Fig. 1. Schematic illustration of the change in volume orenthalpy with temperature for a material undergoing crystalli-sation or a glass transition. The first-order crystallisation atTm is shown in conjunction with the discontinuity in volume/enthalpy at Tg. The short-dashed line indicates the behaviourof a system cooled at a slower rate than that corresponding tothe solid line. The long-dashed line shows the Tg values for thefast cooled (Tg1

) and slow cooled (Tg2) systems [adapted from

Elliott (1983)]


is rate dependent, with slower cooling rates result-ing in lower values for Tg, as indicated in Fig. 1.In practical terms, the material is considered to bein the liquid (or ‘rubbery’ in the case of somepolymers) and glassy states above and below Tg,respectively. In the case of many polymers themechanical properties of the system change fromthose of a pliable to a brittle material as thesystem is cooled through the glass transition,these changes arising as a result of the decrease inmolecular motions.

There are a number of theories associated withthe nature of the glass transition but, as yet, thereis no universally accepted single explanation forthe phenomenon. The observation that Tg is rep-resented by a change in the derivative of extensivethermodynamic parameters such as volume, en-thalpy and entropy suggests that the glass transi-tion is a second order thermodynamic phasetransition, where the term second order refers tothe order of the lowest derivative of the Gibbsfree energy which shows a discontinuity at thetransition point (processes such as crystallisationbeing first order). However, there are a number ofdifficulties associated with this explanation, notleast of which is the common observation that thevalue of Tg is dependent on cooling rate. As asecond order thermodynamic transition is by defi-nition rate independent, the glass transition is notideally second order.

It is also helpful to consider the entropy of thesystem, which at equilibrium is related to the heatcapacity via Cp=T ((S/(T)p The heat capacitiesof the glassy (BTg) and crystalline states for agiven material are essentially the same and ariseprincipally from vibrational contributions, whilethe higher values of Cp observed at temperatures\Tg are caused by the material possessing addi-tional configurational degrees of freedom when inthe rubbery state. Consequently, the Tg may beconsidered to occur at a given value of excessentropy (Elliott, 1983). Kauzmann (1948) ad-dressed the question of the temperature depen-dence of Tg, in particular whether there is a lowerlimit to the value of the glass transition at infi-nitely long cooling times. It was suggested thatsuch a lower limit does indeed exist and is gov-erned by the fact that if the Tg represents a loss of

excess entropy, that value may not exceed theentropy associated with the change from the liq-uid to the crystalline form (i.e. the entropy offusion). Otherwise, the entropy of the glass wouldbe lower than that of the crystal which wouldviolate the third law of thermodynamics. Thissituation is known as the Kauzmann paradox andis resolved by there being a lower limit to Tg forany system (termed either the calorimetric idealglass transition temperature Toc or the Kauzmanntemperature Tk). In practice, the experimental Tg

may occur 20 K or more above Tk. It maytherefore be postulated that, over and above thepossibility of a second order phase transition tak-ing place, the heat capacity will decrease withtemperature in any case as a result of equilibriumthermodynamics. On this basis, the glassy be-haviour of materials may be considered to be afunction of their configurational entropy, hence itshould theoretically be possible to relate the Tg

value to the ‘stiffness’ of the molecule in question.This has been discussed in detail for linear poly-mers (Gibbs and Di Marzio, 1958; Gibbs, 1960;Gee, 1970).

The glass transition may also be considered interms of the relaxation processes that occur as theliquid is cooled. These relaxation times will betemperature dependent, with longer times beingobserved as the liquid is cooled; for hydrogen-bonded fluids, these relaxation processes will begoverned largely by reorganisation of those hy-drogen bonds. If the ‘structural’ relaxation time ofthe material (tr) is short with respect to the timeof observation t0 (which will be the case at tem-peratures above Tg) then the material will appear‘liquid-like’ as the sample will be able to respondto changes in temperature within the timescale ofthe cooling process. Consequently, above Tg thesample is in equilibrium with the cooling pro-gramme. Below Tg, however, t0 will be Btr, hencethe material will assume solid-like characteristicsdue to the relaxation process being slow withrespect to the timescale of the cooling pro-gramme, leading to reduced molecular mobility.Tg may therefore be considered to occur whentr: t0. A related approach to interpreting theglass transition is to consider the free volume ofthe system. This model was developed for fluids


(Cohen and Turnbull, 1959; Turnbull and Cohen,1961, 1970) and considers the liquid to be com-posed of a volume occupied by the constituentmolecules (Vocc) and a volume in which themolecules are free to move (the free volume Vf).The free volume is composed of voids of varyingsizes which are continuously redistributed throughthe system via random movements of the con-stituent molecules. It is further assumed that themolecules may diffuse through the system onlywhen the free volume is above a critical value. Asthe temperature is lowered, both volumes con-tract. For a glassy system, the free volume reachesa lower limit at Tg and is thereafter temperatureindependent, hence in essence the glass transitionoccurs when Vf falls below a critical value (Foxand Flory, 1950, 1951, 1954).

The relaxation behaviour of glassy systems maybe associated with molecular mobility and macro-scopic fluidity, which in turn determine many ofthe important properties of the amorphous mate-rials. These properties will therefore be outlinedseparately in a subsequent section. However, theabove discussion illustrates the difficulty in estab-lishing an exact definition of the glass transition.The process has been described as a second orderthermodynamic phase change (although, as previ-ously stated, the rate dependence of the processweakens this interpretation considerably), as anequilibrium process involving excess entropy or asa kinetic relaxation process. In practice, it is usualto work on the basis of the process being essen-tially kinetic, hence the Tg is considered to be afunction of the relaxation behaviour of the mate-rial. It is important to stress, however, that this isnot the only interpretation of this event.

2.2. The 6iscosity of glasses

At the Tg, the viscosity of the liquid attains avalue in the region of 1012–1014 Pa.s, while abovethe glass transition the material may show consid-erable temperature dependence in the viscosityvalue, as shown in Fig. 2. The extent and natureof this temperature dependence itself varies be-tween materials. ‘Strong’ glass formers such asSiO2 which form networks in the liquid state obeythe Arrhenius relationship over a wide tempera-

Fig. 2. Viscosities of a variety of glass-forming systems as afunction of reduced reciprocal temperature. The curves havebeen normalised such that the glass-transition temperature isdefined to be when h=1013 Poise (Reprinted from Glass:structure by spectroscopy, J. Wong and C.A. Angell, MarcelDekker, New York, Copyright (1976), with permission fromElsevier Science).

ture range, while ‘fragile’ glass formers such asorganic glasses (e.g. O-terphenyl and i-butyl bro-mide) show strong deviation from such behaviour(Wong and Angell, 1976). In the region immedi-ately above Tg, such materials may follow theVogel–Tammann–Fulcher (VTF) relationshipwhich may be expressed in terms of the relaxationtime t via:

t=t0 exp (DT0/T−T0) (1)

where T is the temperature and t0, D and T0 areconstants. As can be seen from Fig. 2, systemsobeying this relationship show a temperature-de-pendent activation energy as opposed to the con-stant value predicted by the Arrheniusrelationship. A related empirical expression is theDoolittle equation (Doolittle, 1951) which relatesthe viscosity to the free volume of the sample via:

h=A exp (BVocc/Vf) (2)


where A and B are constants, if the thermalexpansion of the system is linear with temperaturethen Eq. (1) may be obtained from the Doolittleequation. Alternatively, the Williams–Landel–Ferry (WLF) equation (Williams et al., 1955), aspecial case of the VTF equation, has been shownto be applicable to a wide range of glass formingsystems whereby:

h=hg exp [C1(T−Tg)/(C2+ (T−Tg))] (3)

where hg is the mean viscosity at Tg and C1 and C2

are constants. This equation allows calculation ofthe relaxation times of the system over a range oftemperatures and therefore allows prediction to bemade with regard to temperature-dependent be-haviour. As will be demonstrated later, these rela-tionships may be highly important inunderstanding the behaviour of pharmaceuticalproducts.

Strong glass-formers exhibit minimal molecularmobility changes at Tg, hence the shift in heatcapacity tends to be small. Conversely, fragileglass-forming systems exhibit marked changes inmobility through the Tg due to their non-direc-tional, non-covalent interactions, hence the heatcapacity alterations at Tg are larger (Angell, 1995a)and easier to detect. A ‘rule-of-thumb’ with regardto predicting the fragility has been proposed (Sladeand Levine, 1995; Angell, 1995a,b) whereby ifTm/Tg is less or greater than 1.5 the glass can beconsidered fragile or strong, respectively. Unfortu-nately, native globular proteins are generallystrong glass forming systems, hence the heat capac-ity changes associated with Tg tend to be smallwhich renders their measurement using thermalmethods more difficult. The glassy behaviour ofproteins has been the subject of considerable dis-cussion (Doster et al., 1989; Frauenfelder et al.,1991; Green et al., 1994) and, given the preponder-ance of proteinaceous drug molecules under devel-opment, this area may represent a subject ofconsiderable future interest within the pharmaceu-tical sciences.

2.3. A6ailable methods for the measurement of Tg

As Tg is associated with a change in molecularmobility, the transition will consequently affect

many of the material’s physical characteristics. Itis therefore possible to detect glass transitionalbehaviour by changes in volume/density, heat ca-pacity, viscoelastic moduli, electrical permittivityand refractive index. There are a number of avail-able methods by which the glass transition may bestudied and the interested reader is referred to thetext by Hancock and Zografi (1997) for a moredetailed discussion of these approaches along withother methods such as NMR. In the present textthe use of thermal methods will be outlined. Thesimplest method with which to measure the changein heat capacity associated with the glass transitionis differential scanning calorimetry (DSC). Thistechnique lends itself to the measurement of Tg

values for small quantities of powdered material,which is the most likely state in which an amor-phous pharmaceutical system will be presented tothe analyst. Other thermal approaches includethermomechanical analysis (TMA), whereby thedimensional changes of a sample are measured asa function of temperature. This method is usuallyemployed for the measurement of polymeric films,although in practice cast films of a range ofamorphous materials may be analysed (Hancock etal., 1995).

The change in the viscoelastic properties of anamorphous material with temperature can be mea-sured by dynamic mechanical analysis (DMA,Craig and Johnson, 1995; Haines, 1995). An oscil-latory stress is applied to the sample and both themagnitude and phase relationship of the stress andstrain values are measured as a function of eitherfrequency or temperature. When the sample isheated through the glass transition the shear mod-ulus decreases, often by several orders of magni-tude. The ratio of the loss and storage modulus isequal to tan d, and the glass transition is seen asa peak in tan d when plotted against temperature.Andronis and Zografi (1997) have described the useof DMA for the determination of the glass transi-tional behaviour for amorphous indomethacin.Dielectric analysis (DEA) is another oscillatorytechnique, but in this case a sinusoidal oscillatoryelectric field is applied to the sample. The complexdielectric permittivity is measured as a function oftemperature and may change abruptly through the


glass transition, thereby allowing identification ofthe Tg (McCrum et al., 1967).

2.3.1. The measurement of Tg by differentialscanning calorimetry

DSC is the most frequently used technique forthe measurement of glass transitional behaviour.The method involves the heating or cooling of asample and reference and the measurement of thedifferential heat flow (power) between them withrespect to temperature (or, less usually, time).There are two principal approaches to such mea-surements. Heat flux DSC involves the measure-ment of the temperature differential between thesample and reference and the subsequent calcula-tion of the equivalent heat flow, as given by:

DQ= (Ts−Tr)/RT (4)

where Q is heat, RT is the thermal resistance ofthe cell and Ts and Tr are the sample and refer-ence temperatures, hence at a given scanning ratethe heat flow (power, P) is obtained and displayedagainst temperature.

Power compensation DSC involves the applica-tion and measurement of a compensatory powerinput (P) to one or other pan in order to maintainboth at the same programme temperature, hence:

P=I2R (5)

where I is the current supplied to the heater ofresistance R in order to maintain equality oftemperature.

As power is energy per unit time and the sam-ple is being heated or cooled at a predeterminedrate, in effect the instrument is measuring thedifference in energy required to raise the tempera-ture of either sample or reference by a unitamount, i.e.

P=dQ/dt=Cp·dT/dt+ f(t, T) (6)

where f(t, T) is a function of temperature andtime and reflects kinetically controlled events suchas melting and crystallisation. In essence, there-fore, DSC is measuring the difference in apparentheat capacity between the two pans. Conse-quently, the glass transition is seen as a step in thebaseline which allows identification of both the

temperature and, with suitable calibration, quan-tification of the DCp value, although the latter hasnot been extensively employed in the study ofpharmaceutical systems. It should also be pointedout that, due to the scaling direction on theordinate, power compensation DSC shows en-dotherms and exotherms pointing upwards anddownwards respectively, while heat flux instru-ments show the opposite.

Fig. 3 shows an idealised sketch of a typicalglass transitional response measured using DSC,the response being expressed in terms of thechange in heat capacity. Tg is preferably specifiedas the temperature of half vitrification on cooling,i.e. the temperature at which the heat capacity ismidway between the liquid and glassy state (Wun-derlich, 1990). It is determined by the extrapola-tion of the Cp (or power) plots for the glass andthe liquid/rubber state, with Tg given as the mid-point between the two lines. In addition the glasstransition range and onset can be used to charac-terise this region. The beginning of the transitionis given by Tb, whereas the extrapolated onsettemperature is T1. Similarly, the extrapolated end-point is T2 and the transition end is given by Te.However, in practice the use of parameters suchas Tb and Te may be limited by the difficulties

Fig. 3. Schematic representation of the change in heat capacitythrough the glass transition, indicating the various parametersthat may be used to express the glass transitional behaviour(Reprinted from Thermal Analysis, B. Wunderlich, AcademicPress, London, Copyright (1990), with permission from El-sevier Science).


associated with distinguishing the beginning andend of the transition from the baseline.

It is important to note that the value of theglass transition depends on the heating and cool-ing rate. As illustrated in Fig. 1, a fast coolingrate produces a higher value for Tg than does theuse of slower rates (Richardson and Savill, 1975).This relationship may be described in terms of therelaxation behaviour of the system. The time scalefor the relaxation processes is higher at lowertemperatures, i.e. the relaxation will be slower(Moynihan et al., 1974). Consequently, at slowerrates the temperature at which the relaxation pro-cess becomes comparable with the time scale ofthe experiment will be lower, hence the measuredTg will be lower. This issue is a source of consider-able confusion when constructing quality controltests, as unlike melting which is independent ofheating rate, the glass transition is a response to aheating or cooling signal and hence will varydepending on the method of measurement.

The difficulty of scanning rate dependence maybe overcome by use of the fictive temperature Tf,which represents the temperature at which theextrapolated enthalpies above and below to theglass transition are equal. As described above, theglass transition is a kinetic event and thereforehighly dependent on the heating or cooling rate ofthe temperature programme. However, the truevalue for the glass transition temperature is onlydependent on the conditions of formation, e.g. onthe cooling rate of the sample in the case ofquenching. The heating rate dependence is seenbecause DSC is a dynamic technique, leading topossible differences in the experimental andmolecular time scales. Consequently, analysis ofthe transition on heating will not give the ‘true’ Tg

value, but a dynamic glass transition temperaturedependent on the underlying heating rate. Toremove this problem, Richardson and Savill(1975) have suggested measurement of the fictivetemperature which is independent of heating rate.This method involves raising the temperaturefrom a steady value below the glass transitionregion, T1, to a steady value above Tg, T2, and,using suitable calibration, recording the specificheat as a function of temperature, i.e.

Fig. 4. Schematic representation of the fictive temperature,showing the extrapolation of the enthalpy curves below andabove Tg for a quenched (Tgq) and annealed (Tga) glass. Thesolid lines depict typical experimental paths, while the dashedline shows how the fictive Tg is defined [adapted from Richard-son and Savill (1975)].

Cpg=a+bt, Cpl=A+BT (7)

where Cpg and Cpl are the specific heat capacitiesof the glassy and liquid states and a, b, A and Bare constants determined by linear regression. In-tegration of these equations gives:

Hg(T)=aT+1/2 bT2+P,

Hl(T)=AT+1/2 BT2+Q (8)

where P and Q are constants. The fictive tempera-ture is defined as the intersection of the extrapo-lated enthalpy curves and is obtained by solvingfor the quadratic for T when Hg(T)=Hl(T). isshown schematically in Fig. 4. The value may alsobe determined via calculation of the areas underthe experimental heat capacity curves; interestedreaders are referred to the work by Moynihan etal. (1976) and Moynihan (1994) for more details.Use of the fictive temperature could have severaladvantages as far as pharmaceutical systems areconcerned, particularly in terms of establishingreliable quality control protocols.

A further, often considerable problem associ-ated with the measurement of Tg using DSC is thepresence of a relaxation endotherm; this feature isseen on heating as an endothermic response su-perimposed on the baseline shift which may ren-


der identification and quantification of the Tg

extremely difficult (an example of an amorphousdrug showing this behaviour is given in Fig. 5).Indeed, for complex or multicomponent samplessuch as freeze dried formulations it may be ex-tremely difficult to differentiate between a glasstransition and a melting response, although differ-ences in rate dependence of the relaxation andmelting endotherms may allow distinction be-tween the two. The relaxation endotherm mayarise as a result of one of two processes. Firstly, itmay reflect a mismatch in the rate of cooling andsubsequent heating of the sample. This may beexplained with reference to Fig. 6. When coolingis slow the glass that forms has long relaxationtimes ‘frozen in’. Subsequently, when this materialis heated quickly, with a faster rate through theglass transition than was the case on cooling, therelaxation times are slow with reference to theheating rate, thereby producing an overshoot inthe enthalpy curve. In other words, the moleculeswithin the glass cannot achieve the motion re-quired for the glass transition within the timescale of the heating rate and the glass brieflysuperheats (Wunderlich, 1990). Once the relax-

ation times lower to the order of the heating ratethe superheated glass reverts quickly towards theliquid line in the enthalpy curve. This overshootand subsequent recovery in the enthalpy curveproduces the characteristic endothermic peak inthe DCp (or power) curve. The second reason forthe appearance of the endothermic relaxation isthat because glasses are not in an equilibriumstate, they can relax over time, thereby decreasingthe enthalpy and volume of the material andincreasing the structural relaxation time. Conse-quently, such annealing also produces a relaxationendotherm at the glass transition for the samereasons as above, the magnitude of which may beused to calculate relaxation times for the sample(Montserrat, 1989; Hancock et al., 1995), this isdescribed in more detail in Section 3.2.

There are a number of difficulties associatedwith the presence of relaxation endotherms. Overand above problems in identifying glass transi-tions, the quantification of both the Tg value andthe relaxation endotherm may be difficult, as illus-trated in Fig. 5. It is not clear, for example,exactly where under the peak the Tg lies (althoughthe fictive temperature may be of use in this

Fig. 5. Glass transition of saquinavir, showing the relaxation endotherm associated with Tg (Reprinted from Pharm. Res., 15, P.G.Royall, D.Q.M. Craig and C. Doherty, Characterisation of the glass transition of an amorphous drug using modulated DSC, pp.1117–1121, Copyright (1988), with permission from Elsevier Science).


Fig. 6. Schematic representation of the origin of the relaxationendotherm caused by a mismatch in cooling and heating rates.The figure shows the enthalpy and corresponding heat capac-ity changes through the Tg region on slow cooling and fastheating.

true glass transition will be reproducible andshould be seen on both heating and cooling (giventhe provisos listed above). Furthermore, it is es-sential that adequate baseline calibration is per-formed to ensure as flat a baseline as possible; ifextensive bowing is present then quantitative iden-tification of the Tg will be extremely difficult.Baseline calibration is achieved by subtraction ofthe response from two empty DSC pans in thereference and sample positions. It is also possibleto use higher scanning rates to aid visualisation ofthe Tg. Inspection of Eq. (6), for example, indi-cates that at higher values of dT/dt, the heat flowwill be greater, hence the sensitivity of the mea-surement may be improved (although there willinevitably be a concomitant loss of resolution);the influence of the heating rate on the glasstransition value should, however, be noted. Greatcare is required if the DCp value at Tg is to bemeasured quantitatively, as it is essential to per-form adequate reference calibration using a stan-dard such as sapphire in order to obtainmeaningful data.

Sample preparation conditions must be care-fully controlled (and stated), particularly in termsof the choice of pans and level of residual mois-ture or other solvents, as otherwise changes to thesample either during or prior to analysis may havea profound influence on the measured Tg. A re-cent investigation (Hill et al., 1998) indicated thatthe measured Tg of spray dried lactose may varyby :35°C depending on pan type. This is due tothe retention of water in hermetically sealed panswhich acts as a plasticizer (discussed below),thereby reducing Tg compared to non-hermeti-cally sealed or open pans from which water is lostduring the heating run.

Tg measurements are, ideally, performed incooling rather than heating cycles, as in the for-mer the sample starts from the equilibrium liquidstate before entering the nonequilibrium glass,which is a reproducible route compared to start-ing from the glass in a heating cycle (Wunderlich,1990). However, as mentioned above, many phar-maceutical systems such as drugs and freeze driedsystems are heat sensitive and may not withstandcycling. Similarly, other systems are multicompo-nent, hence heating through the Tg of one compo-

respect), hence it is also difficult to calculate thevalue of the relaxation enthalpy. This may beovercome by controlling the thermal history ofthe sample, particularly by heat/cooling whichshould remove annealing effects. However, thismay not be appropriate for pharmaceutical sam-ples which are often multicomponent and hencemay not be reversibly temperature cycled. Analternative approach involves the use of modu-lated temperature DSC, which allows separationof the relaxation endotherm from the glass transi-tion, although Royall et al. (1988) have identifiedcertain provisos associated with such measure-ments. This technique is discussed in more detailbelow.

In addition to the above, there are also a num-ber of practical considerations associated with themeasurement of glass transitions using DSC. Thebaseline shifts may be small, particularly forstrong glasses, rendering differentiation frombaseline noise a non-trivial problem. However, a


nent may result in irreversible changes in thestructure of the system as a whole. Consequently,most pharmaceutical studies have involved mea-suring Tg upon heating.

The development of MTDSC has removed orlessened some of the difficulties associated withcharacterising glass transitions. This techniquerepresents a change in the software of the conven-tional DSC and, in the model marketed by TAInstruments (modulated DSC, MDSC) involvesthe application of a sinusoidal heating signal su-perimposed on the linear programme, hence infor-mation may be derived from both the sine waveresponse and the Fourier transformed total heatflow output (equivalent to conventional DSC).The advantage of the technique is that changes inheat capacity (i.e. glass transitions) may be seen inisolation from other events, particularly relax-ation endotherms, with a considerably enhancedsignal-to-noise ratio. An example of this is shownin Fig. 7 for the amorphous drug saquinavir.More details of this technique are available else-where (Royall et al., 1988; Reading et al., 1993;Coleman et al., 1996; Hill et al., 1998) and the

method is expected to make a major contributionto the assessment of glassy pharmaceuticals.

2.4. Plasticization of amorphous materials

A highly important consideration with regardto the glass transition of amorphous materials isthe effect of the presence of additional materials,particularly water, on the value of Tg. The studyof mixed systems has been of particular impor-tance in the polymer science field, whereby mate-rials may be multicomponent, with the degree ofmixing being estimated by observing the glasstransitions of the product in relation to those ofthe individual components. Mixing of amorphouspharmaceuticals is an approach that may be usedto raise the glass transition of a product in orderto improve stability (Fukuoka et al., 1989).

The work on which amorphous mixing theoryis based is that of Gordon and Taylor (1952),which was originally used to describe the be-haviour of polymer blends. The Gordon Taylorequation is based on free volume theory and givesthe glass transition, Tgmix

, of a binary mixture,

Fig. 7. Separation of the DSC response of amorphous saquinavir into the reversing and non-reversing signals, in which the glasstransition and relaxation endotherm may be seen in isolation (Reprinted from Pharm. Res., 15, P.G. Royall, D.Q.M. Craig and C.Doherty, Characterisation of the glass transition of an amorphous drug using modulated DSC, pp. 1117–1121, Copyright (1988),with permission from Elsevier Science).


assuming no specific interaction between the twocomponents, via:

Tgmix=81Tg1

+82Tg2(9)

where 8 is the volume fraction and the subscriptsrepresent the two components. The volume frac-tion can be described in terms of the weightfraction of the components w, as 8= (wDa)/r,where Da is the change in thermal expansivity atTg and r is the density of the material. RedefiningEq. (9) in terms of weight fraction gives:

Tgmix=

(w1Tg1)+ (Kw2Tg2

)w1+ (Kw2)

(10)

where:

K=(r1Da2)(r2Da1)

(11)

which can be simplified by application of theSimha–Boyer rule to:

K=r1Tg1/r2Tg2

(12)

K can be considered the ratio of the free volumesof the two components. The goodness of fit ofexperimental data to the Gordon–Taylor equa-tion gives an idea of the ideality of mixing of twocomponents, as well as providing a predictive toolfor assessing the effects of different levels of asecond material on Tg. In terms of pharmaceuticalsystems, the area in which this approach hasproved to be particularly important is in the studyof the effects of water on the glass transition. It iswell recognised that residual water levels may bean important factor in determining chemical sta-bility and mechanical properties. However, waterwill also have a profound effect on the glasstransition of amorphous pharmaceuticals, actingas a plasticizer by increasing the free volume ofthe material, hence leading to a decrease in Tg.Exactly the same principle applies to the inclusionof plasticizers in polymeric film coats, although inthe light of the systems discussed in the nextsection the question of water sorption will beemphasised here.

In the amorphous state considerably more wa-ter may be taken up relative to the crystallineform (Hancock and Zografi, 1997), effectively dueto absorption of water into the solid, hence in

Fig. 8. Variation in the glass transition of freeze dried lactosewith water content. Solid line is the predicted behaviour fromEq. (8)(Reprinted from Pharm. Res., 11, B.C. Hancock and G.Zagrafi, The relationship between the glass transition tempera-ture and the water content of amorphous pharmaceuticalsolids, pp. 1166–1173, Copyright (1994), with permission fromElsevier Science).

contrast to crystalline systems the uptake processtends to be dependent on sample mass rather thansurface area. A typical profile of Tg against wateruptake is shown in Fig. 8 (Hancock and Zografi,1994). As the concentration of water in the solidincreases, the Tg is seen to decrease according tothe Gordon–Taylor equation (or an approxima-tion of this relationship if the system is non-ideal).At any particular temperature, therefore, the sys-tem may change from the glassy to the rubberystate if water uptake takes place, with concomi-tant implications for chemical degradation or de-vitrification (discussed below). Ahlneck andZografi (1990) have emphasised that at higherstorage temperatures, a lower amount of water isrequired to lower the Tg to that particular temper-ature (Fig. 9). As a general principal, this hasprofound implications for accelerated stabilitytesting of amorphous pharmaceuticals. Hancockand Zografi (1994) have reported that water is apotent plasticizer for a wide range of pharmaceu-tical materials including PVP, lactose, starch, su-crose and others.

3. Relevance of the glassy state to amorphousdrugs

There are numerous ways in which the amor-phous state is of relevance pharmaceutically and afull discussion of all such considerations would be


prohibitively lengthy. Instead, this and the follow-ing section will focus on two of the most impor-tant aspects of glassy pharmaceuticals, namelyamorphous drugs and freeze dried systems.

3.1. Amorphous drugs and dissolution beha6iour

It is well recognised that, for many drugs, disso-lution within the gastrointestinal tract may be therate limiting step to absorption, hence improve-ment in the dissolution rate may enhance thebioavailability of that drug. One approach to suchimprovement is to prepare the drug in an amor-phous form, the higher molecular mobility of thisform compared to the equivalent crystalline mate-rial may lead to enhanced dissolution rate andbioavailability, although there are important dis-advantages to the approach which will be dis-cussed below.

From the discussion given above, it is clear thatthe basic physico-chemical parameter that may beused to characterise amorphous drugs is the glasstransition temperature. Kerc' and Src' ic' (1995)have prepared a highly useful compilation of Tg

and melting point (Tm) data for a number ofdrugs which is reproduced in Table 1. It should beborn in mind that the conditions under which the

various samples were prepared and measured arehighly unlikely to be uniform, hence absolutecomparisons between data points may not beapplicable. Furthermore, it is extremely difficult inpractice to remove all of the water associated withsuch samples, hence plasticization may have takenplace in some cases. Nevertheless, a number ofuseful trends may be extracted from the data. Inparticular, the authors have discussed the signifi-cance of the ratio between the glass transitionaland melting parameters, as Tg/Tm tends to be:0.5 for symmetrical polymers and 0.7 for asym-metrical (Gujrati and Goldstein, 1980), this refersback to the concept of fragile and strong glassesoutlined in Section 2.2, although here the ratio isinverted. Kerc' and Src' ic' (1995) report values be-tween :0.6 and 0.8 for low molecular weightpharmaceuticals, which are slightly higher thanthose found for polymeric systems. According tothe ‘rule of thumb’ of Tm/Tg ratios outlined ear-lier, this indicates that low molecular weight phar-maceutical systems are more fragile glass-formingsystems than polymers. The knowledge of thisratio is of use as it does allow the formulator tomake a rough estimate of where a glass transitionis likely to be found for a drug with a knownmelting point.

A number of examples of enhanced dissolutionfor amorphous drugs are available in the litera-ture. For example, 9,3%%-diacetylmidecamycin(MOM), a 16-membered macrolide antibiotic, wasprepared in the crystalline and amorphous formsand the dissolution behaviour of both at a rangeof temperatures compared (Sato et al., 1981).Figs. 10 and 11 show the corresponding dissolu-tion profiles for the two forms; the amorphousform clearly shows higher dissolution rates, al-though at longer dissolution times the concentra-tion in water decreases due to the formation ofcrystalline MOM from the supersaturated solu-tion. Similarly, the dissolution rate of amorphousindomethacin has been reported to be greaterthan for the crystalline material (Fukuoka et al.,1986) as shown in Fig. 12 for water–ethanol andwater systems. Interestingly, these authors alsostudied the relaxation behaviour of the glassysystem on storage below Tg, showing that whilequench cooled indomethacin was stable over a 2year period at room temperature, pulverised

Fig. 9. Effect of relative humidity in the glass transitiontemperature of PVP K30. The box illustrates conditions typi-cally used in accelerated storage testing (Reprinted from Int J.Pharm., 62, C. Ahlneck and G. Zografi, The molecular basisof moisture effects on the physical and chemical stability ofdrugs in the solid state, pp. 87–95, Copyright (1990), withpermission from Elsevier Science, data compiled from Ok-sanen (1992)).


Table 1Glass transition (Tg) and melting point (Tm) data for a rangeof pharmaceuticals [reproduced from Kerc' and Src' ic' (1995)]

Tg/TmTm/KTg/KPharmaceutical

180 291Glycerol 0.620.59408Aspirin 243

246 336Dibucaine 0.73247 340Mephenesin 0.73

380 0.67256Antipyrine360 0.73Ribose 263

0.70384Sorbitol I 270270 421Methyltestosterone 0.64271 367Sorbitol II 0.74

0.73377Phenylbutazone 277278 362Quinine ethylcarbonate 0.77

0.70399Progesterone 279408 0.69Pentobarbital 279379 0.74Atropine 281

282 398Ethacrynic acid 0.710.72283Citric acid 432

283 426Xylose 0.66403 0.70Tolbutamide 284

286 423Hexobarbital 0.68286 432Amobarbital 0.66

373 0.77Fructose 2860.75384Tolnaphtate 287

389 0.74Nimodipine 288430 0.67Tartaric acid 289

290 406Flufenamic acid 0.71290 434santonin 0.67290 376Ergocalciferol 0.77

0.73403Proxyphylline 295295 447Acetaminophen 0.66296 352Cholecalciferol 0.84

0.86347Paracetamol 297297 378Eserine 0.79

0.70427Nialamide 297393 0.76Chlorotrianisene 298349 0.86Chlorimphenicol I 301

302 441Acetaminophen 0.69303 419Glucose 0.72303 429Nitrendipine 0.71

0.67460Sulphisoxazole 306414 0.74Chloramphenicol II 306439 0.70Stilbestrol 308

309 425Estradiol-17b-cypionate 0.730.72310Dextrose 432

438 0.72Diphylline 315315 452Phenobarbital 0.70

375 0.84316Maltose407 0.76Felodipine 316

0.72443Phenobarbital 3210.72453Norethynodrel 3240.73445Quinidine 326

Sucrose 329 0.73453

Table 1 (Continued)

Pharmaceutical Tg/K Tm/K Tg/Tm

331 478 0.69Spironolactone333Salicin 466 0.71

Sulphathiazole 0.71471334334 483Chlormadinone acetate 0.69336b-Estradiol-3-benzoate 472 0.71337Amlodipine besylate 467 0.72

Sulphadimethoxine 339 465 0.73344Glibenclamide 447 0.77

0.69502Dehydrocholic acid 348350Cellobiose 498 0.70

Trehalose 350 0.744760.8044517b-Estradiol 354

Nicardipin hydrochloride 358 0.81440Griseofulvin I 362 0.86422

0.81Brucine 365 451370 0.74497Griseofulvin II371Chenodeoxycholic acid 436 0.85

0.84447Deoxycholic acid 3770.79477Ursodeoxycholic acid 378

Cholic acid 393 0.83473

amorphous drug crystallised over a period ofmonths.

Amorphous drugs may be prepared in a num-ber of ways such as rapidly cooling from the meltor precipitation from suitable solvent systems.Drying or grinding may deliberately or acciden-tally induce amorphous characteristics. A groundmixture of griseofulvin and microcrystalline cellu-lose significantly improved both the dissolutionrate and bioavailability of the drug compared to amicronised griseofulvin powder preparation (Ya-mamoto et al., 1974). This was ascribed to anincrease in amorphous content of the drug as aresult of the grinding process. Grinding of pheny-toin with microcrystalline cellulose was also foundto enhance drug dissolution rate, this again beingattributed to the formation of an amorphousform of the drug (Yamamoto et al., 1976).

Spray drying may also be used to prepareamorphous systems. A review of the physicalstructure of spray dried products, with specialreference to thermal analysis, has been given byCorrigan (1995). The formation of amorphouspharmaceuticals via this process has been demon-strated for a number of drugs such as digitoxin(Nurnburg, 1976) and a range of thiazide diuretics


Fig. 10. Concentration-time curves for crystalline MOM indistilled water at various temperatures [reproduced from Satoet al. (1981)].

(MAT) were spray dried using a range of inlettemperatures (Yamaguchi et al., 1992). The au-thors performed storage studies on the samples attemperatures BTg, finding that materials preparedusing inlet temperature between Tg and Tm weremore stable than those prepared at lower or highertemperatures, even though the storage conditionswere identical for the various samples.

These studies touch on one of the questionswhich, to date, does not appear to have been fullyanswered, namely to what extent can the amor-phous ‘form’ of a drug vary depending on manu-facturing or processing conditions? It is predictablefrom basic theory that changing the preparationconditions such as cooling rate from the melt or thepresence of trace plasticisers (particularly water)may alter the Tg (and hence ‘structure’) of thematerial. For example, Kerc' et al. (1991) havereported (comparatively small) changes in the Tg

value of felodipine on altering the cooling rate fromthe melt. However, marked changes in the dissolu-tion rate were observed depending on coolingconditions (Fig. 13), even though the glass transi-tion values for systems cooled at 1°C/min and thosecooled at the maximum set rate of the instrumentshowed differences in Tg of B2°C. It is conceivablethat other factors such as the macroscopic integrityof the sample may have contributed to this effect.However, the change in dissolution profile is clearlyof potential practical significance. Angell (1995a)has discussed the concept of ‘polyamorphism’,whereby materials may exist in distinct disorderedforms, as has been reported for water (Mishima etal., 1984, 1985, 1991). The generation of such formsappears to take place under fairly extreme condi-tions and hence the direct relevance to pharmaceu-tical materials is far from certain. However, thesuggestion that polyamorphism may exist at all isof interest.

3.2. The physical stability of amorphous drugs

Given the potential advantages of preparingdrugs in an amorphous form, the question arises asto why this approach is not used more often. Thesingle most important reason is undoubtedly theproblems associated with stability, both physicaland chemical. The amorphous state is, by defini-

(Corrigan et al., 1984). In addition, Hill et al.(1998) have recently studied the glass transitionalbehaviour of spray dried lactose using modulatedDSC. Interestingly, Corrigan (1995) has describedhow changing the spray drying conditions mayresult in the formation of materials with differingTg values and physical properties. For example,Matsuda et al. (1992) prepared amorphousfrusemide using two spray drying protocols andreported that the resultant materials had differingTg values (44.2 and 54.4°C). These changes may bea result of differences in moisture content but thestudy clearly demonstrates the dependence of Tg onspray drying protocol. Similarly, dichloromethanesolutions of 4%%-O-(4-methoxyphenyl)acetyltylosin

Fig. 11. Concentration-time curves for amorphous MOM indistilled water at various temperatures [reproduced from Satoet al. (1981)].


Fig. 12. Dissolution of crystalline and amorphous indomethacin in water-ethanol (A) and water (B) � glass � crystalline material[reproduced from Fukuoka et al. (1986)].

tion, metastable with regard to the crystallinematerial, hence amorphous drugs will tend torevert to the crystalline form over a period oftime. Prediction of the timescales involved isclearly critical and yet may be difficult to achieve.However, with some knowledge of the fundamen-tal principles of the crystallisation process and thephysico-chemical properties of the drug in ques-tion it is possible to make some assessment of thelikelihood of devitrification and to recommendstorage conditions which will minimise the risk ofthe process occurring.

Crystallisation from the amorphous state isgoverned largely by the same factors that deter-mine crystallisation from the melt and has beendiscussed in more detail by Hancock and Zografi(1997). In the case of amorphous materials, thereare two conflicting phenomena associated with thecrystallisation process. As the temperature is low-ered, the rate of nucleation may be expected toincrease. However, the molecular mobility de-creases as the temperature decreases, particularlybelow Tg, thereby slowing the molecular diffusionand reducing the rate of crystallisation. This issummarised in Fig. 14 which shows that the max-imum rate of crystallisation will take place be-tween Tm and Tg. If a sample is stored below Tg

the risk of devitrification is considerably reduceddue to the molecules having lower mobility withwhich to undergo crystal growth; studies on thecrystallisation behaviour of sucrose and lactoseappear to confirm the importance of Tg in thisrespect (Makower and Dye, 1956; Saleki-Ger-

hardt and Zografi, 1994). However, it should bestressed that storage below Tg is far from a guar-antee of physical stability. Hancock et al. (1995)have shown that the molecular mobility below Tg

may be sufficient to result in devitrification overthe typical storage periods encountered for phar-maceutical products. The relationship between Tg

and the stability of glassy pharmaceuticals istherefore complex; studies comparing the stabilityof glassy indomethacin and phenobarbital, bothof which have similar Tg values, showed veryconsiderable variation in stability profiles eventhough both were stored below their Tg values

Fig. 13. Dissolution of felodipine in water-ethanol. � crys-talline felodipine � glassy felodipine prepared by coolingunder ambient conditions o glassy felodipine prepared byquenching in liquid nitrogen (Reprinted from Int. J. Pharm.,68, J. Kerc' , S. Src' ic' , M. Mohar and Smid-Korbar, Somephysiochemical properties of glassy felodipine, pp. 25–33,Copyright (1991), with permission from Elsevier Science).


Fig. 14. Schematic representation of the temperature depen-dence of the crystallisation process for an amorphous system[Hancock and Zografi, 1997, after Jolley (1970)].

reasonable guide would be that, if possible, thesample should be stored at least 50°C below theTg value. Clearly, this may not always be practi-cally possible and, in addition, the stability willinevitably be system dependent. However, the es-timate affords the operator some prediction as tothe likelihood of there being a physical stabilityproblem, which may be of considerable benefitwhen at the early stages of planning a formulationstrategy. The authors describe the measurement ofthe magnitude of the relaxation endotherm afterstorage at different temperatures as a means ofcalculating the relaxation time of the sample.More specifically, the maximum enthalpy recov-ery (DH�) is calculated via:

DH�= (Tg−T)·DCp (13)

where Tg and T are the glass and experimentaltemperatures and DCp is the change in heat capac-ity at Tg. The extent of relaxation ft is thencalculated at any time (t) and temperature (T)conditions via:

ft=1− (DHt/DH�) (14)

where DHt is the measured enthalpy of recoveryunder those conditions. The relaxation time maythen be calculated via the Williams–Watts equa-tion (Williams and Watts, 1970):

ft=exp (− t/t)b (15)

where t is the mean relaxation time constant andb is an empirical relaxation time distributionparameter (05b51). By non-linear regression itis possible to calculate the relaxation times over arange of conditions, thereby allowing an insightinto the relationship between molecular mobilityand temperature, with clear implications for stor-age stability. There are a number of difficultiesassociated with this approach, including the ne-cessity to accurately calibrate for DCp and thedifficulty in reliably measuring the magnitude ofthe relaxation endotherm; given the associatedshift in baseline, modulated temperature DSCmay in the future prove to be highly useful inboth respects. In addition, it is interesting to notethat in the dielectrics field, from which theWilliams–Watts equation was obtained, there isgrowing emphasis on non-empirical analysis of

(Fukuoka et al., 1989). Glassy indomethacin wasreported to be stable over a 2 year period at roomtemperature, while phenobarbital devitrifiedwithin 1 week. These differences were ascribed tovariations in steric structure and hydrogen bond-ing within the two materials. It is also importantto emphasise that even if one is storing the mate-rial below the Tg of the dry materials, the pres-ence of trace levels of water may plasticize thesample sufficiently to bring the Tg value to that ofthe storage temperature. Consequently, it is rec-ommended that if possible the material should bekept as dry as possible. It has been shown, forexample, that the crystallisation temperature ofsalbuterol sulphate may be reduced by the pres-ence of sorbed water (Ward and Schultz, 1995).The converse also applies; inclusion of materialssuch as PVP, which has a high Tg value (:180°C), may improve the physical stability ofglassy drugs by raising the Tg value of the binarysystem (Fukuoka et al., 1986; Yoshioka et al.,1995).

Hancock et al. (1995) have argued that the keyissue is that of the temperature in relation to theTg at which a glassy drug should be stored inorder to provide a pharmaceutically acceptablestability profile. From the above discussion it isclear that storage below Tg is advantageous but isnot a guarantee of stability due to the molecularmobility which is present below the glass transi-tion temperature. The authors have argued that a


non-Debye relaxation behaviour, particularly interms of the Dissado–Hill theory (Dissado andHill, 1979) which describes the relaxation be-haviour in terms of many-body interactions ratherthan a spread of relaxation times which is in itselfa controversial concept. As yet there has been noextensive cross-referencing of this theory with re-gard to glass transition measurements; such anapproach could prove to be highly useful.

The recrystallisation behaviour of the sampleon heating above Tg may also be characterised.This is commonly seen as an exotherm above theglass transition, as shown in Fig. 15 for spraydried lactose. This approach has been studied byKerc' et al. (1991), who describe two main meth-ods for obtaining kinetic data from the exotherm.The heat evolution method (Caroll and Manche,1972; Torfs et al., 1984) involves the measurementof the rate constant k at any temperature duringthe crystallisation process via:

k=dH/dt ·[(DHtot(DHrem/DHtot))n]−1 (16)

where dH/dt is the heat flow, DHtot is the totalenthalpy, DHrem is the enthalpy corresponding to

the non-crystallised fraction at temperature T andn is the order of the reaction. An alternativeapproach is the variable heating rate method(Ozawa, 1975) whereby the exothermic peak max-imum temperature is measured as a function ofheating rate. These approaches are of use in esti-mating the kinetics of the crystallisation above Tg,although they are not applicable to the predictionof the behaviour at temperatures below the glasstransition.

3.3. The chemical stability of amorphous drugs

Numerous reports have shown that rates ofdrug degradation may be enhanced in the amor-phous state compared to the crystalline material.A series of papers in the 1970s explored the drugdegradation of amorphous and crystalline formsof drugs. For example, the temperature dependentdegradation of cefoxitin sodium was found to bemarkedly enhanced when the drug was preparedin an amorphous form (Fig. 16, Oberholtzer andBrenner, 1979), while Pikal et al. (1977) showedthat the degradation rate of a range of b-lactam

Fig. 15. DSC response for spray dried lactose, showing a water loss endotherm (a), a glass transition with accompanying relaxationendotherm (b), recrystallisation (c) and melting with decomposition (d) (Reprinted from Int. J. Pharm., 161, V.L. Hill, D.Q.M. Craigand L.C. Feely, Characterisation of spray-dried lactose using modulated differential scanning calorimetry, pp. 95–107, Copyright(1998), with permission from Elsevier Science).


Fig. 16. Typical plots of decomposition of (a) crystalline and (b) amorphous cefoxitin sodium at various temperatures � 40°C, 60°C and � 80°C [reproduced from Oberholtzer and Brenner (1979)].

antibiotics was enhanced by approximately anorder of magnitude for amorphous as opposed tocrystalline drugs, even when sorbed water levelswere very low. Interestingly, these authors alsoreported a discoloration of amorphouscephalothin sodium prior to the development of ameasurable potency loss.

The mechanism and kinetics of such degrada-tion reactions have been the subject of somespeculation. As one might expect, the presence ofsorbed water has been reported to increase thedegradation of a range of drugs including insulin

(Strickley and Anderson, 1996) and the aforemen-tioned b-lactam antibiotics (Pikal et al., 1977). Inthe latter study, the authors discussed the kineticsof the degradation process, finding that in generalthe antibiotics showed what appeared to be firstorder kinetics as opposed to the sigmoidal curveexpected for solid state reactions (although thiswas not the case for all the drugs under study).However, the authors also reported that the tem-perature dependence of the decomposition rateindicated that the activation energy for the pro-cess was decreasing as the temperature was raised.


Pikal et al. (1977) suggested that this may be afunction of the rate limiting step of the decom-position being associated with molecular reorien-tation processes, whereby certain stericconditions need to be satisfied in order for thereaction to occur. In other words, the degrada-tion may be a function of molecular mobilityand relaxation behaviour. The concept of orien-tation-specific degradation has been discussed byseveral authors. Sukenik et al. (1975, 1977) sug-gested that the solid state reactivity of aminoben-zoate derivatives may be associated with theirorientation within the crystal, while Hageman etal. (1992) have suggested that protein intermolec-ular reactivity may be higher in the solid statedue to a higher ‘effective concentration’ com-pared to the solution. Similarly, Strickley andAnderson (1996) demonstrated that the forma-tion of the [desamidoA21–GlyA1] dimer was en-hanced in glassy insulin compared to theaqueous solution. Pikal et al. (1977) suggestedthat as non-Arrhenius behaviour (involving theactivation energy decreasing with temperature) ischaracteristic of glassy behaviour in the vicinityof Tg then the kinetics of degradation may beassociated with relaxation behaviour of theamorphous solid. This is also consistent with theobservation that the presence of water enhancedthe decomposition rate, as over and above thepresence of a chemical reactant, the water maybe plasticizing the amorphous material and de-creasing the relaxation time at the temperature ofstorage, several other studies have demonstrateda correlation between the chemical reactivity ofan amorphous material and the glassy behaviour(Roy et al., 1992; Levine and Slade, 1993).

4. Freeze dried systems

4.1. Introduction

Freeze-drying has been used as a pharmaceuti-cal unit operation for a number of years for thelow temperature drying of injectable systems.However, the approach has recently becomemore prominent due to the necessity of preparingnovel peptide and proteinaceous drugs in a dry,

stable form which may be easily reconstitutedprior to parenteral administration. The chal-lenges associated with preparing such formula-tions have necessitated re-examination of thephysico-chemical principles underlying the freezedrying process, hence research is ongoing in or-der to enhance predictability of the chemical andphysical stability of such products. One aspect ofthis technology which has been the subject ofconsiderable study is the glass transitional be-haviour of both the frozen system prior to dry-ing and of the finished product. A number ofexcellent texts are available on the subject offreeze drying to which the interested reader isreferred to for more details (Holdsworth, 1987;Pikal, 1990a,b; Nail and Gatlin, 1993; Pikal,1993; Franks, 1994). In the interests of brevity,the description of the freeze drying process givenhere is intended as an outline of the amorphousnature of these systems and is not meant as acomprehensive discussion of all aspects of theprocess.

Freeze drying involves the desiccation of asubstance by crystallisation of ice, followed bysublimation of water vapour from the solid stateat reduced pressure. Drying at low temperaturestheoretically avoids extensive chemical degrada-tion and the resulting solid tends to have anopen porous structure that facilitates the rehy-dration process, this may be essential in lifethreatening conditions whereby rapid reconstitu-tion may be critical. The process has found con-siderable application in the processing ofpharmaceutical products of biological origin suchas serums, vaccines, peptide drugs and liposomes,as the stability of freeze-dried solids is usuallymuch higher than the equivalent aqueous solu-tion or suspension. In addition, there has alsobeen considerable interest in the use of freezedrying as a means of producing rapidly dissolv-ing oral dosage forms (Corveleyn and Remon,1997). However, there may be substantialdifficulties associated with both the chemical andphysical stability of freeze dried products, hencethere is considerable interest in establishing thenature of the interplay between the formulation,the process used and the stability of freeze driedsystems.


4.2. The freeze drying process

4.2.1. Initial FreezingFreezing occurs upon cooling aqueous solutions

to temperatures below 0°C and takes place via icenucleation (Korber, 1988). This process involvesthe generation of water aggregates with suffi-ciently long lifetimes so as to allow interactionswith surrounding water molecules. As the temper-ature is lowered, molecular mobility decreases andthe lifetimes of the aggregates increase, leading tonucleation and subsequent crystal growth. It isimportant to note that, depending on factors suchas the cooling rate, water may undergo consider-able supercooling. Ice formation may thereforetake place at temperatures down to −20°C usingrates which are practical in most freeze dryers (forhomogeneous nucleation, undercooling to −40°Cis possible in microlitre volumes, although inpractice heterogeneous nucleation will inevitablyplace at higher temperatures during the freezedrying process). The degree of supercooling inturn determines the size distribution and mor-phology of the ice crystals formed, with faster

Fig. 18. Schematic representation of the state diagram of aglass forming binary system (Reprinted from In: Avis, K.E.,Lieberman, H.A., Lachman, L. (Eds.), Pharmaceutical DosageForms: Parenteral Medications. Marcel Dekker, New York,S.L. Nail and L.A. Gatlin, pp. 163–233, Copyright (1993),with permission from Elsevier Science).

freezing rates giving rise to smaller crystals whichpermit more rapid sublimation in the primarydrying step (although secondary drying may beslower). The resulting product will have a finepore structure that facilitates rehydration.

The removal of water to form ice has the effectof increasing the solute concentration, whichwhen coupled with cooling causes an increase inviscosity of the solute phase. This concentrationeffect is responsible for many of the deleteriouseffects of freeze drying. Nucleation of the solute(s)may occur leading to crystallisation and the for-mation of a eutectic system (Fig. 17). For exam-ple, sodium chloride is expected to form a eutecticwith water at −21°C, although the rate of crystalgrowth is slow and complete crystallisation is notgenerally observed. If nucleation does not occurwithin the timescale of the cooling process theneventually the remaining ‘solution’ will form aglass with a viscosity sufficiently high (typically:1013 Pa.s) to inhibit further ice formation. Thecharacteristics of this glass may be of considerableimportance in determining the freeze drying pro-

Fig. 17. Schematic representation of the phase diagram of theNaCl–water eutectic system (Reprinted from In: Avis, K.E.,Lieberman, H.A., Lachman, L. (Eds.), Pharmaceutical DosageForms: Parenteral Medications. Marcel Dekker, New York,S.L. Nail and L.A. Gatlin, pp. 163–233, Copyright (1993),with permission from Elsevier Science).


tocol and hence will be discussed in more detailwith reference to Fig. 18.

Taking the simplest scenario, the solution iscooled and ice crystals form, hence the remainingsolution becomes increasingly concentrated until asaturation value is reached and no further in-crease in concentration is possible (Cg% ), the sys-tem is said to be in the maximally freezeconcentrated state (Franks, 1990). The systemundergoes a glass transition and the temperatureat which this occurs is denoted Tg% . However,rapidly cooled systems containing very low soluteconcentrations may exhibit a transition at temper-atures lower than Tg% because the glass formed willinclude more water than the ideal case; this prob-lem is prevented by the use of lower cooling rates.It should be noted that Tg% is a very low energytransition which may be obscured by the ice melt-ing peak and by the accompanying relaxationendotherm (if present). Clearly, such a measure-ment is prone to considerable variation, which isreflected by the relatively poor consistency of datafor Tg% and Cg% available in the literature, with arange of values reported for materials such assucrose. The corresponding values for this mate-rial are now generally accepted to be −32°C and83% sucrose (Slade and Levine, 1988; Franks,1990).

Identification of Tg% may be further complicatedby the presence of a second transition at slightlyhigher temperatures. One system which has beenwidely studied in this respect is sucrose, which isimportant not only as a pharmaceutical excipientbut in frozen foods. DSC studies of frozen carbo-hydrate solutions (Ablett et al., 1992; te Booy etal., 1992) have revealed two transitions below theonset of melting of ice. DSC studies have shownthat annealing at temperatures immediately belowTg% yielded a characteristic double transition belowTm of water, while annealing at the estimated Tg%gave only one; it was concluded that the twotransitions had coalesced at this temperature(Ablett et al., 1992). Current thinking is that thelower transition is the true glass transition whilethe second is caused by the irreversible collapse ofthe glass and is denoted Ts (softening point, Sha-laev and Franks, 1995).

4.2.2. Primary dryingDuring this stage the water separated from the

solute phase in the form of ice during freezing isremoved by sublimation under vacuum. Theproduct temperature remains relatively constantand drying follows a pseudo-steady-state rate withheat removal by sublimation at the same rate asthat of the heat input supplied by the shelves.This can be expressed by (Pikal, 1993):

DHs·dmdt

=dQdt

(17)

where DHs is the heat of sublimation, dm/dt is thesublimation rate and dQ/dt is the rate of heatinput. The rate of sublimation can be expressedas:

dmdt

=(P0−Pc)(Rp+Rs)

(18)

where (Po−Pc) is the thermodynamic driving-force for sublimation where P0 is the vapourpressure of ice in the frozen sample and Pc is thetotal pressure in the chamber. (Rp+Rs) is thetotal resistance to sublimation, where Rp is theproduct resistance and Rs is the resistance of thestopper in the vial. The rate of heat input can beexpressed as:

dQ/dt=Av·Kv (Ts−Tp) (19)

where Av is the cross-sectional area of the vial, Kv

is the vial heat transfer coefficient and (Ts−Tp) isthe heat difference between the shelf and theproduct, Ts being the shelf temperature and Tp theproduct temperature. P0, the vapour pressure ofthe ice in the product, increases exponentiallywith temperature, so that an increase in producttemperature will cause a large increase in rate ofsublimation. This also implies that the chamberpressure should not be set to below that of thestandard vapour pressure of water over ice at theproduct temperature since this will decrease therate of sublimation. In addition, the dimensionsof the cake influence the effectiveness of dryingsince the product resistance will increase as thethickness of dried cake increases, hindering diffu-sion of water to the solute/vapour interface. Thishas the effect of reducing the rate of sublimationwhich will give rise to a temperature gradient in


the product. At the end of primary drying, sec-ondary drying will begin spontaneously in someregions of the product where all the ice has sub-limed. It is generally assumed that primary dry-ing has ended when the product temperature isat the shelf temperature.

Knowledge of the glass transitional behaviourof the frozen system is extremely useful forchoosing the most appropriate processingparameters. If during primary drying the producttemperature is raised above Tg% then ice will meltback into the product, causing softening whichmay lead to collapse (discussed below), henceidentification of Tg% and Cg% allows greater pre-dictability of the effects of drying temperature onthe physical integrity of the sample, allowingdrying to occur at the highest possible tempera-ture without compromising the physical integrityof the sample. Given that a temperature increaseof 1°C may lead to a reduction in primary dry-ing time of 13% (Pikal, 1985), the economic im-portance of choosing the drying temperature ona rational basis is clear. Product collapse duringdrying has been discussed in detail by Bellowsand King (1972), Pikal and Shah (1990) andmore recently Sun (1997) and is associated withviscous flow of the amorphous material; this phe-nomenon leads to an inelegant product but mayalso increase reconstitution times and residualwater levels. It is therefore essential to dry thematerial below the collapse temperature (Tc)which, for most practical purposes, is :20°Cabove Tg% (Sun, 1997). A range of Tg% and Cg% forcommonly used freeze drying excipients has beengiven by Franks (1990).

4.2.3. Secondary dryingThis is the term given to the process during

which ‘unfrozen’ water is removed from thefreeze concentrate. Secondary drying is usuallyaided by increasing the shelf temperature, usuallyto 25–60°C. For this reason it is very importantthat all ice has been removed by sublimationbefore the temperature is raised in order to pre-vent melt-back. In the case of amorphous prod-ucts, the water remaining after primary drying istrapped in the glassy phase. Initially the rate ofwater loss (the first few hours) will be great but a

plateau level is reached beyond which furtherwater removal (below :2%) is very slow. Therate of water removal is not proportional to theconcentration of water in the freeze concentrate,but is controlled by the rate of diffusion to thesolute/vapour interface and the subsequent evap-oration. As the solid dries, the motion in theconcentrate becomes much slower and conse-quently diffusion becomes more difficult. Diffu-sion is a function of the porosity of theconcentrate; small pores caused by extreme su-percooling before ice formation during the initialfreeze hinder the diffusion of water to the solid/vapour interface, although chamber pressure (Pc)and dried cake thickness have little effect on therate of secondary drying. The removal of plasti-cizing water during secondary drying has the ef-fect of raising the glass transition and hence thecollapse temperature, therefore an ideal sec-ondary drying protocol would follow the Tg in-crease of the sample.4.2.4. Product stability during storage

The glass transition of the freeze dried productmust be well above the subsequent storage tem-perature in order to prevent collapse, hence de-termination of Tg for the freeze dried producthas important implications for storage stability(Nail and Gatlin, 1993). Since collapse occurswhen the sample is unable to hold its ownweight, its apparent value will depend on samplemass and period of observation (and appliedpressure if mechanical methods of detection areused).

As yet a limited number of reports have dis-cussed the significance of product collapse andimplications for stability of pharmaceutical prod-ucts, although numerous studies may be foundlinking the glass transition and stability in thefood science literature, describing phenomenasuch as volumetric shrinkage at collapse (Leviand Karel, 1995), caking and stickiness (Chuyand Labuza, 1994) and increased rates of nonen-zymic browning (Buera and Karel, 1995). Earlystudies have indicated that modulated DSC maybe particularly useful for measuring transitionsof the dried product in terms of identifying andquantifying the value of Tg in complex systems(Kett et al., 1988; Van Winden et al., 1998).


4.3. Freeze drying of biological products

4.3.1. The effects of water on stabilityThe principal challenge associated with freeze

drying proteins and other biological products isloss of activity during freezing, drying or storage;for example, lactate dehydrogenase may showcomplete loss of enzymatic activity after freezedrying unless protective measures are taken (Pikal,1993). A number of texts (Franks, 1985; Baffi andGarnick, 1991; Constantino et al., 1995; Anchor-doquy and Carpenter, 1996; Chang et al., 1996a,b;Carpenter et al., 1997; Colaco et al., 1997) areavailable which outline the current thinking in thisarea.

One highly important issue associated with thestability of freeze dried proteins is the optimalresidual water of the dried product. Aggregation ofexcipient free human growth hormone (hGH) hasbeen attributed to overdrying (Pikal et al., 1991a),while excipient free tissue type plasminogen activa-tor (tPA) was found to aggregate at an enhancedrate if the residual water level was low (Hsu et al.,1991a). However, these observations are not thenorm, as in general lower water contents improvestorage stability. For example, the rate ofhaemoglobin oxidation at room temperature isdoubled if the residual water content is increasedfrom 1 to 4% (Pristoupil et al., 1985), while the rateof degradation of human growth hormone atelevated temperatures is increased tenfold if thewater content is increased from :1 to 2.5% (Pikalet al., 1991b). The detrimental effect of water isusually attributed to the increased mobility andhence reactivity of solute species. It has beenpostulated (Hsu et al., 1991b) that above mono-layer levels of water the protein has increasedconformational mobility and that additional watercan mobilise reactant species in the amorphousphase. It should be emphasised that the distribu-tion of the water throughout the product is alsoimportant. A low overall %w/w water content maybe misleading due to the possibility of localisedregions containing higher water concentrations.

4.3.2. CryoprotectantsCryoprotectants are materials which are com-

monly added during the freeze drying process in

order to afford protection of the drug from degra-dation. The term literally refers to protection dur-ing freezing or freeze-thawing rather than dryingand hence the term lyoprotectant is preferred if theadditive is capable of preventing degradation dur-ing the lyophilisation process as a whole. Thisdistinction may be important as the freeze-dryingprocess is much harsher than freeze-thawing due tothe additional sublimation stage. For example, theenzyme L-aspariginase can be freeze-thawed with-out any loss of activity, but retains only 20%activity after being exposed to freeze-drying (Hell-man et al., 1983). Indeed, many cryoprotectantsthat are useful additives during freeze-thawinghave little effect during freeze-drying, examples ofsuch additives being proline and trimethylamineN-oxide (Carpenter et al., 1991).

The most commonly used cryoprotectants aresugars, although polymers and amino acids mayalso be used. The use of sugars as a means ofprotection against freezing or dehydration is wellknown in nature. For example, certain organismspossess the ability to survive complete dehydrationand exist in a state of anhydrobiosis for tens ofdecades, without loss of activity upon rehydration.An example of such an organism is brewers’ yeast,which is fully active only minutes after rehydration.This is possible because the organism accumulateslarge amounts of saccharides, usually either sucroseor trehalose (Crowe et al., 1984), and their abilityto survive dehydration can be correlated with theconcentration of sugar present. Both of thesesugars have been investigated as additives to freezedried protein formulations (Carpenter and Crowe,1989; Hall et al., 1995) and liposomes (Van Windenet al., 1998).

It has been postulated that carbohydrates andsome amino acids stabilise proteins during freezingby their thermodynamically favourable exclusionfrom the protein surface (Arakawa et al., 1991).This situation arises when it is more favourable forthe stabilizer molecules to interact with each otherthan with the protein. In this case the native stateof the protein is favoured because unfolding (orbreaking up of multi-unit proteins to sub-units)would give greater opportunity for such stabilizer–protein interactions to occur, and so raise theGibbs free energy of unfolding (Arakawa and


Timasheff, 1982). In contrast, it has been sug-gested that polymers such as polyvinylpyrrolidone(PVP) and maltodextrins are believed to exerttheir stabilising influence by raising the averagemolecular weight, and so increasing the viscosityof the formulation (Blond, 1994; Anchordoquyand Carpenter, 1996).

There are two main theories that have been putforward to account for the stabilising effect ofcryoprotectants during drying and storage,namely the ‘water substitute’ and the ‘vitrification’theories. The water substitute theory is thermody-namically based (Franks et al., 1991; Slade andLevine, 1991) and assumes that the stabiliser in-teracts with the protein in the same way as doeswater, i.e. that the native state is stabilised be-cause water lost during the drying process isreplaced by the stabiliser molecules (Prestrelski etal., 1993a). Studies using FTIR spectroscopy (Pre-strelski et al., 1993b,c) have indicated that theconformation of proteins that have bonded sugarsare similar to those in aqueous solution.

The vitrification hypothesis states that the ac-tivity of the stabilisers stems from their glass-forming properties. It is assumed that in theglassy state molecular motion is effectively re-moved, hence degradation is slowed. Evidencesupporting this theory includes the observationthat mobility and reactivity are both often de-creased as the product temperature is reducedbelow the glass transition temperature region(Roozen et al., 1991). This hypothesis has alsobeen used to explain the cryoprotective ability ofhigh molecular weight polymers in some in-stances, which are considered to raise the glasstransition temperature of a formulation. How-ever, a clear link between the Tg of the formula-tion and the stability of the product has not yetbeen established.

5. Conclusions

This discussion has attempted to tie togethersome of the concepts and challenges facing theformulation scientist with respect to the under-standing and handling of amorphous materials. Inparticular, the principles underpinning the glassy

state have been described, along with a consider-ation of thermal analysis as a means of character-ising such materials. Two applications have beendiscussed, namely the behaviour of amorphousdrugs and freeze dried systems. In all cases, theneed to understand the glass transitional and re-laxation behaviour of the systems under study hasbeen stressed.

Amorphous systems are clearly an integralcomponent of the development of effective dosageforms and the importance of these materials islikely to increase both due to the need to developoral dosage forms which show favourablebioavailability profiles and the growing emphasison freeze drying for the preparation of proteina-ceous drugs. By understanding the relationshipbetween the glassy behaviour and the productperformance characteristics of these materials it istherefore possible to approach formulation strate-gies on a more rational basis.

Acknowledgements

We wish to thank and GlaxoWellcome andRhone-Poulenc Rorer for financial support of VKand MH, respectively.

References

Ablett, S., Izzard, M.J., Lillford, P.J., 1992. Differential scan-ning calorimetry study of frozen sucrose and glycerolsolutions. J. Chem. Soc. Farad. Trans. 88, 789–794.

Ahlneck, C., Zografi, G., 1990. The molecular basis of mois-ture effects on the physical and chemical stability of drugsin the solid state. Int. J. Pharm. 62, 87–95.

Anchordoquy, T.J., Carpenter, J.F., 1996. Polymers protectlactate dehydrogenase during freeze drying by inhibitingdissociation in the frozen state. Arch. Biochem. Biophys.332, 231–238.

Andronis, V., Zografi, G., 1997. Molecular mobility of super-cooled amorphous indomethacin, determined by dynamicmechanical analysis. Pharm. Res. 14, 410–414.

Angell, C.A., 1995. The old problems of glass and the glasstransition and the many new twists. Proc. Nat. Acad. Sci.USA 92, 6675–6682.

Angell, C.A., 1995. Formation of glasses from liquids andbiopolymers. Science 267, 1924–1935.

Arakawa, T., Timasheff, S.N., 1982. Stabilization of proteinstructure by sugars. Biochemistry 21, 6536–6544.


Arakawa, T., Kita, Y., Carpenter, J.F., 1991. Protein solventinteractions in pharmaceutical formulations. Pharm. Res. 8,285–291.

Baffi, R.A., Garnick, R.L., 1991. Quality control issues in theanalysis of lyophilized proteins. Dev. Biol. Stand. 74, 181–184.

Bellows, R.J., King, C.J., 1972. Freeze drying of aqueoussolutions: maximum allowable operating temperature. Cry-obiology 9, 559–561.

Blond, G., 1994. Mechanical properties of frozen model solu-tions. J. Food Eng. 22, 253–269.

Byrn, S.R., 1982. Solid-State Chemistry of Drugs. AcademicPress, London.

Byrn, S.R., Pfeiffer, R., Ganey, M., Hoilberg, C., Poochikian,G., 1995. Pharmaceutical solids: a strategic approach toregulatory considerations. Pharm. Res. 12, 945–954.

Buera, M.P., Karel, M., 1995. Effect of physical changes on therates of nonenzymic browning and related reactions. FoodChem. 52, 167–173.

Carpenter, J.F., Crowe, J.H., 1989. An infrared spectroscopystudy of the interactions of carbohydrates with driedproteins. Biochemistry 28, 3916–3922.

Carpenter, J.F., Arakawa, T., Crowe, J.H., 1991. Interactionsof stabilizing additives with proteins during freeze-thawingand freeze drying. Dev. Biol. Stand. 74, 225–239.

Carpenter, J.F., Pikal, M.J., Chang, B.S., Randolph, T.W.,1997. Rational design of stable lyophilized protein formula-tions: some practical advice. Pharm. Res. 14, 969–975.

Caroll, B., Manche, E.P., 1972. Kinetic analysis of chemicalreactions for non-isothermal processes. Thermochim. Acta3, 449–459.

Chang, B.S., Beauvais, R.M., Dong, A., Carpenter, J.F., 1996.Physical factors affecting the storage stability of freezedried interleukin-1 antagonist: glass transition and proteinconformation. Arch. Biochem. Biophys. 331, 249–258.

Chang, B.S., Kendrick, B.S., Carpenter, J.F., 1996. Surface-in-duced denaturation of proteins during freezing and itsinhibition by surfactants. J. Pharm. Sci. 85, 1325–1330.

Chuy, L.E., Labuza, T.P., 1994. Caking and stickiness ofdairy-based food powders as related to glass transitions. J.Food Sci. 59, 43–46.

Cohen, M.H., Turnbull, D., 1959. Molecular transport inliquids and glasses. J. Chem. Phys. 31, 1164–1169.

Colaco, C.A.L.S., Smith, C.J.S., Sen, S., Roser, D.H., New-man, Y., Ring, S., Rosers, B.J., 1997. In: Cleland, J.L.,Langer, R. (Eds.), Formulation and Delivery of Proteinsand Peptides. Pp. 222–240

Coleman, N.J., Craig, D.Q.M., 1996. Modulated temperaturedifferential scanning calorimetry: a novel approach to phar-maceutical thermal analysis. Int. J. Pharm. 135, 13–29.

Constantino, H.R., Griebenow, K., Mishra, P., Langer, R.,Klibanov, A.M., 1995. Fourier-transform infrared spectro-scopic investigation of protein stability in the lyophilizedform. Biochim. Biophys. Acta 1253, 69–74.

Corrigan, O.I., Holohan, E.M., Sabra, K., 1984. Amorphousforms of thiazide diuretics prepared by spray-drying. Int. J.Pharm. 18, 195–200.

Corrigan, O.I., 1995. Thermal analysis of spray dried products.Thermochim. Acta 248, 245–258.

Corveleyn, S., Remon, J.P., 1997. Formulation and productionof rapidly disintegrating tablets by lyophilisation usinghydrochlorothiazide as a model drug. Int. J. Pharm. 152,215–225.

Craig, D.Q.M., Johnson, F.A., 1995. Pharmaceutical applica-tions of dynamic mechanical analysis. Thermochimica Acta248, 97–115.

Crowe, J.H., Crowe, L.M., Chapman, D., 1984. Infraredspectroscopic studies on interaction of water and carbohy-drates with a biological membrane. Arch. Biochem. Bio-phys. 232, 400–407.

Dissado, L.A., Hill, R.M., 1979. Non-exponential decay indielectrics and dynamics of correlated systems. Nature 79,685–689.

Doolittle, A.K., 1951. Studies in newtonian flow, 2: the depen-dence of the viscosity of liquids on 3-space. J. Appl. Phys.22, 1471–1475.

Doster, W., Cusack, S., Petry, W., 1989. Dynamical transitionof myoglobin revealed by inelastic neutron scattering. Na-ture 337, 754–756.

Duddu, S.P., Dal Monte, P.R., 1997. Effect of glass transi-tional temperature on the stability of lyophilized formula-tions containing a chimeric therapeutic monoclonalantibody. Pharm. Res. 14, 591–595.

Elliott, S.R., 1983. Physics of Amorphous Materials. Long-man, London.

Fox, T.G., Flory, P.J., 1950. Second order transition tempera-tures and related properties of polystyrene, 1: influence ofmolecular weight. J. Appl. Phys. 21, 581–591.

Fox, T.G., Flory, P.J., 1951. Further studies on the meltviscosity of polyisobutylene. J. Phys. Chem. 55, 221–234.

Fox, T.G., Flory, P.J., 1954. The glass temperature and relatedproperties of polystyrene—influence of molecular weight.J. Polym. Sci. 14, 315–319.

Franks, F., 1985. Biophysics and Biochemistry at low temper-atures, 1st ed. Cambridge University press, Cambridge.

Franks, F., 1990. Freeze drying: from empiricism to pre-dictability. Cryo-Letters 11, 93–110.

Franks, F., 1994. Effective freeze drying: a combination ofphysics, chemistry, engineering and economics. Proc. Inst.Refrig. 91, 32–39.

Franks, F., Hatley, R.H.M., Mathias, S.F., 1991. Materialsscience and the production of shelf stable biologicals.Biopharmacy 4, 38.

Frauenfelder, H., Sligar, S.G., Wolynes, P.G., 1991. The en-ergy landscapes and motions of proteins. Science 254,1598–1603.

Fukuoka, E., Makita, M., Yamamura, S., 1986. Some physic-ochemical properties of glassy indomethacin. Chem.Pharm. Bull. 34, 4314–4321.

Fukuoka, E., Makita, M., Yamamura, S., 1989. Glassy state ofpharmaceuticals 11: thermal properties and stability ofglassy pharmaceuticals and their binary glass systems.Chem. Pharm. Bull. 37, 1047–1050.

Gee, G., 1970. Glassy state in polymers. Contemp. Phys. 11,313–334.


Gibbs, J.H., 1960. In: Mackenzie, J.D. (Ed.), Modern Aspectsof the Vitreous State. Butterworth, chapter 7.

Gibbs, J.H., Di Marzio, E.A., 1958. Nature of the glasstransition and the glassy state. J. Chem. Phys. 28, 373–383.

Gordon, M., Taylor, J.S., 1952. Ideal copolymers and thesecond-order transitions of synthetic rubbers, 1: non-crys-talline copolymers. J. Appl. Chem. 2, 493–498.

Green, J.L., Han, J., Angell, C.A., 1994. The protein–glassanalogy: some insights from homopeptide comparisons. J.Phys. Chem. 98, 13780–13790.

Gujrati, P.D., Goldstein, M., 1980. Viscous liquids and theglass transition, 9: nonconfigurational contribution to theexcess entropy of disordered phases. J. Phys. Chem. 84,859–863.

Hageman, M.J., Bauer, J.M., Possert, P.L., Darrington, R.T.,1992. Preformulation studies orientated towards sustaineddelivery of recombinant somatotropins. J. Agric. FoodChem. 40, 348–355.

Haines, P.J., 1995. Thermal Methods of Analysis. BlackieAcademic and Professional, Glasgow.

Haleblian, J.K., 1975. Characterisation of habits and crys-talline modifications of solids and their pharmaceuticalapplications. J. Pharm. Sci. 64, 1269–1288.

Hall, D.R., Jacobson, M.P., Winzor, D.J., 1995. Stabilizingeffect of sucrose against irreversible denaturation of rabbitmuscle lactate dehydrogenase. Biophys. Chem. 57, 47–54.

Hancock, B.C., Zografi, G., 1994. The relationship betweenthe glass transition temperature and the water content ofamorphous pharmaceutical solids. Pharm. Res. 11, 1166–1173.

Hancock, B.C., Shamblin, S.L., Zografi, G., 1995. Molecularmobility of amorphous pharmaceutical solids below theirglass transition temperatures. Pharm. Res. 12, 799–806.

Hancock, B.C., Zografi, G., 1997. Characterisation and signifi-cance of the amorphous state in pharmaceutical systems. J.Pharm. Sci. 86, 1–12.

Hellman, K., Miller, D., Cammack, K., 1983. The effect offreeze drying on the quaternary structure of L-asparaginasefrom erwinia-carotovora. Biochim. Biophys. Acta 749,133–142.

Hill, V.L., Craig, D.Q.M., Feely, L.C., 1998. Characterisationof spray-dried lactose using modulated differential scan-ning calorimetry. Int. J. Pharm. 161, 95–107.

Holdsworth, S.D., 1987. In: Thorne, S. (Ed.), Developments inFood Preservation—4. Elsevier Applied Science, NewYork, pp. 153–204.

Hsu, C.C., Ward, C., Pearlman, R., Ngiyen, H., Yeung, D.,Curley, J., 1991. Hemoglobin lyophilized with sucrose—the effect of residual moisture on storage. Dev. Biol. Stand.74, 255–271.

Hsu, C.C., Ward, C.A., Pearlman, R., Nguyen, H.M., Yeung,D.A., Curley, J.G., 1991. Detwermining the optimal resid-ual moisture in lyophilised protein pharmaceuticals. Devel.Biol. Stand. 74, 255–271.

Jolley, J., 1970. The microstructure of photographic gelatinebinders. Photogr. Sci. Eng. 14, 169–177.

Kauzmann, W., 1948. The nature of the glassy state and thebehaviour of liquids at low temperatures. Chem. Rev. 43,219–256.

Kerc' , J., Src' ic' , S., Mohar, M., Smid-Korbar, 1991. Somephysicochemical properties of glassy felodipine. Int. J.Pharm. 68, 25–33.

Kerc' , J., Src' ic' , S., 1995. Thermal analysis of glassy pharma-ceuticals. Thermochim. Acta 248, 81–95.

Kett, V., Craig, D.Q.M., Taylor, K.M.G., Deutsch, D., 1988.The use of modulated differential scanning calorimetry inthe study of freeze dried systems. In: Proceedings of TheSeventeenth Pharmacy Tech. Conference, Dublin, pp.104–108.

Korber, C., 1988. Phenomena at the advancing ice-liquidinterface, solutes particles and biological cells. Q. Rev.Biophys. 21, 229–298.

Levi, G., Karel, M., 1995. Volumetric shrinkage (collapse) infreeze dried carbohydrates above their glass transitiontemperature. Food Res. Int. 28, 145–151.

Levine, H., Slade, L., 1993. In: Blanchard, J.M.Y., Hillford,P.J. (Eds.), The Glassy State in Foods. Nottingham Press,Nottingham, pp. 35.

Makower, B., Dye, W.B., 1956. Equilibrium moisture contentand crystallisation of amorphous sucrose and glucose. J.Agr. Food Chem. 4, 72–77.

Matsuda, Y., Otsuka, M., Onoe, M., Tatsumi, E., 1992.Amorphism and physicochemical stability of spray driedfrusemide. J. Pharm. Pharmacol. 44, 627–633.

McCrum, N.G., Read, B.E., Williams, G., 1967. Anelastic anddielectric effects in polymeric solids. John Wiley and Sons,London.

Mishima, O., Calvert, L.D., Whalley, E., 1984. Melting ice 1at 77 K and 10 kbar: a new method of making amorphoussolids. Nature Lond. 310, 393–395.

Mishima, O., Calvert, L.D., Whalley, E., 1985. An apparentlyfirst-order transition between two amorphous phases of iceinduced by pressure. Nature Lond. 314, 76–78.

Mishima, O., Takemare, K., Aoki, K., 1991. Visual observa-tions of the amorphous–amorphous transition in H2Ounder pressure. Science 254, 406–408.

Montserrat, S., 1989. Physical ageing studies in epoxy resins:1, kinetics of the enthalpy relaxation process in a fullycured epoxy resin. J. Polym. Sci. Polym. Phys. Edn. 32,509–522.

Moynihan, C.T., Easteal, A.J., Wilder, J., 1974. Dependenceof the glass transition temperature on heating and cooling.J. Phys. Chem. 78, 2673–2677.

Moynihan, C.T., Easteal, A.J., DeBolt, M.A., 1976. Depen-dence of the fictive temperature of glass on cooling. J. Am.Ceram. Soc. 59, 12–21.

Moynihan, C.T., 1994. In: Segler, R.J. (Ed.), Phenomenologyof the structural relaxation process and the glass transition:assignment of the glass transition ASTM STP 1249. Amer-ican Society for Testing and Materials, Philadelphia, pp.32–49.

Mullins, J., Macek, T., 1960. Some pharmaceutical propertiesof novobiocin. J. Am. Pharm. Assoc. Sci. Ed. 49, 245–248.


Nail, S.L., Gatlin, L.A., 1993. In: Avis, K.E., Lieberman,H.A., Lachman, L. (Eds.), Pharmaceutical Dosage Forms:Parenteral Medications. Marcel Dekker, New York, pp.163–233.

Nurnburg, E., 1976. Kolloids Verteilungszustands in der phar-mazeutischen Technologie: herstellung und Eigenschaftenpharmazeutischer Praparate durch Spruhtrocknung. Prog.Colloid Polym. Sci. 59, 55–69.

Oberholtzer, E.R., Brenner, G.S., 1979. Cefoxitin sodium:solution and solid state chemical stability. J. Pharm. Sci.68, 863–866.

Oksanen, C.A., 1992. Molecular mobility in mixtures of ad-sorbed water and poly(vinyl pyrrolidone), PhD Thesis.University of Wisconsin-Madison.

Ozawa, T.J., 1975. Critical investigation of methods for kineticanalysis of thermoanalytical data. J. Therm. Anal. 7, 601–617.

Pikal, M.J., Lukes, A.L., Lang, J.E., 1977. Thermal decompo-sition of amorphous b-lactam antibacterials. J. Pharm. Sci.66, 1312–1316.

Pikal, M.J., Lukes, A.L., Lang, J.E., Gaines, K., 1978. Quan-titative crystallinity determinations for b-lactam antibioticsby solution calorimetry: correlation with stability. J.Pharm. Sci. 67, 767–772.

Pikal, M.J., 1985. Use of laboratory data in freeze dryingprocess design: heat and mass transfer coefficients and thecomputer simulation of freeze drying. J. Parenteral Sci.Technol. 39, 115–138.

Pikal, M.J., 1990a. Freeze drying of proteins. part 1: processdesign. Biopharmacy (September issue) 18–20.

Pikal, M.J., 1990b. Freeze drying of proteins. part 2: Formula-tion selection. Biopharmacy (September issue) 26–30.

Pikal, M.J., Shah, S., 1990. The collapse temperature in freezedrying: dependence on measurement methodology and rateof water removal from the glassy phase. Int. J. Pharm. 62,165–186.

Pikal, J., Dellerman, K., Roy, M.L., 1991. effects of moistureand oxygen on the stability of freeze dried formulations ofhuman growth hormone. Dev. Biol. Stand. 74, 21–38.

Pikal, M.J., Dellerman, K.M., Roy, M.L., Riggin, R.M., 1991.The effect of formulation variables on the stability offreeze-dried human growth hormone. Pharm. Res. 8, 427–436.

Pikal, M.J., 1993. Freeze-drying of proteins. In: Cleland, J.L.,Langer, R. (Eds.), Formulation and Delivery of Proteinsand Peptides, Chapter 8. ACS Symposium series 567,American Chemical Society, Washington.

Prestrelski, S.J., Arakawa, T., Carpenter, J.F., 1993. Separa-tion of freezing and drying induced denaturation oflyophilized proteins using stress specific stabilization. Arch.Biochem. Biophys. 303, 465–473.

Prestrelski, S.J., Tedeschi, N., Arakawa, T., Carpenter, J.F.,1993. Dehydration induced conformational transitions inproteins and their inhibition by stabilizers. Biophys. J. 65,661–671.

Prestrelski, S.J., Arakawa, T., Carpenter, J.F., 1993c. Struc-ture of proteins in lyophilized formulations using Fourier

Transform infrared spectroscopy. In: Cleland, J.L.,Langer, R. (Eds.), Formulation and Delivery of Proteinsand Peptides. ACS Symposium Series 187, AmericanChemical Society, Washington.

Pristoupil, T.I., Kramlova, M., Fortova, H., Ulrych, S., 1985.Haemoglobin lyophilised with sucrose: the effect of resid-ual moisture on storage. Haematologia 18, 45–52.

Reading, M., Elliott, D., Hill, V.L., 1993. MDSC, a newapproach to the calorimetric investigation of physical andchemical transitions. J. Therm. Anal. 40, 949–955.

Richardson, M.J., Savill, N.G., 1975. Derivation of accurateglass transition temperatures by differential scanningcalorimetry. Polymer 16, 753–757.

Roozen, M., Hemminga, M., Walstra, P., 1991. Molecularmotion in glassy water maltooligosaccharide (maltodex-trin) mixtures as studied by conventional and saturationspin probe ESR spectroscopy. Carbohydr. Res. 215, 229–237.

Roy, M.L., Pikal, M., Richard, E.C., Maloney, M., 1992. Theeffects of formulation and moisture on the stability of afreeze dried monoclonal antibody–vinca conjugate: a testof the WLF theory. Dev. Biol. Stand. 74, 323–340.

Royall, P.G., Craig, D.Q.M., Doherty, C., 1988. Characterisa-tion of the glass transition of an amorphous drug usingmodulated DSC. Pharm. Res. 15, 1117–1121.

Saleki-Gerhardt, A., Zografi, G., 1994. Non-isothermal andisothermal crystallisation of sucrose from the amorphousstate. Pharm. Res. 11, 1166–1173.

Sato, T., Okada, A., Sdekiguchi, K., Tsuda, Y., 1981. Differ-ences in physico-pharmaceutical properties between crys-talline and noncrystalline 9,3%-diacetylmidecamycin. Chem.Pharm. Bull. 29, 2675–2682.

Shalaev, E.Y., Franks, F., 1995. Structural glass transitions inaqueous carbohydrate solutions and their relevance tofrozen food stability. Thermochim. Acta 280/281, 449–464.

Slade, L., Levine, H., 1988. Thermomechanical properties ofsmall carbohydrate-water glasses and rubbers—kineticallymetastable systems at subzero temperatures. J. Chem. Soc.Farad. Trans. 84, 2619.

Slade, L., Levine, H., 1991. Beyond water activity: recentadvances based on an alternative approach to the assess-ment of food quality and safety. Crit. Rev. Food Sci. 30,115–360.

Slade, L., Levine, H., 1995. Glass transitions and water–foodstructure interactions. Adv. Food Nutr. Res. 38, 103–269.

Strickley, R.G., Anderson, B.D., 1996. Solid-state stability ofhuman insulin 1: mechanisms and the effects of water onthe kinetics of degredation in lyophiles from pH 2–5solutions. Pharm. Res. 13, 1142–1153.

Sukenik, C.N., Bonapace, J.A.P., Mandel, N.S., Bergman,R.G., Lau, P.-Y., Wood, G., 1975. Enhancement of achemical reaction rate by proper orientation of reactingmolecules in the solid state. J. Am. Chem. Soc. 97, 5290–5291.

Sukenik, C.N., Bonapace, J.A.P., Mandel, N.S., Lau, P.-Y.,Wood, G., Bergman, R.G., 1977. A kinetic and X-raydiffraction study of solid state rearrangement of methyl


P-dimethylaminobenzene sulfonate: reaction rate en-hancement due to proper orientation in a crystal. J. Am.Chem. Soc. 99, 851–852.

Sun, W.Q., 1997. Temperature and viscosity for structuralcollapse and crystallisation of amorphous carbohydratesolutions. Cryo-Letters 18, 99–106.

te Booy, M.P.W.M., de Ruiter, R.A., de Meere, A.L.J.,1992. Evaluation of the physical stability of freeze-driedsucrose-containing formulations by differential scanningcalorimetry. Pharm. Res. 9, 109–114.

Torfs, J.C.M., Deij, L., Dorrepaal, A.J., Heijens, J.C., 1984.Determination of Arrhenius constants by differentialscanning calorimetry. Anal. Chem. 56, 2863.

Turnbull, D., Cohen, M.H., 1961. Free-volume model ofamorphous phase glass transitions. J. Chem. Phys. 34,120–125.

Turnbull, D., Cohen, M.H., 1970. On free-volume model ofliquid–glass transitions. J. Chem. Phys. 52, 3038–3041.

Van Winden, E.C.A., Talsma, H., Crommelin, D.J.A., 1998.Thermal analysis of freeze-dried liposome-carbohydratemixtures with modulated differential scanning calorime-try. J. Pharm. Sci. 87, 231–237.

Ward, G.H., Schultz, R.K., 1995. Process-induced crys-tallinity changes in albuterol sulphate and its effect onpowder physical stability. Pharm. Res. 12, 773–779.

Williams, M.L., Landel, R.F., Ferry, D.J., 1955. The temper-ature dependence of relaxation mechanisms in amorphous

polymers and other glass-forming systems. J. Am. Chem.Soc. 77, 3701–3707.

Williams, G., Watts, D.C., 1970. Non-symmetrical dielectricrelaxation behaviour arising from a simple empirical de-cay function. Trans. Farad. Soc. 66, 80–85.

Wong, J., Angell, C.A., 1976. Glass: structure by spec-troscopy. Marcel Dekker, New York.

Wunderlich, B., 1990. Thermal Analysis. Academic Press,London.

Yamaguchi, T., Nishimura, M., Okamoto, R., Takeuchi, T.,Yamamoto, K., 1992. Glass formation of 4%%-O-(4-methoxyphenyl) acetyltylosin and physicochemical stabil-ity of the amorphous solid. Int. J. Pharm. 85, 87–96.

Yamamoto, K., Nakano, M., Arita, T., Nakai, Y., 1974.Dissolution rate and bioavailability of griseofulvin from aground mixture with microcrystalline cellulose. J. Phar-macokinet. Biopharm. 2, 487.

Yamamoto, K., Nakano, M., Arita, T, Takayama, Y.,Nakai, Y., 1976. Dissolution behaviour of phenytoinfrom a ground mixture with microcrystalline cellulose. J.Pharm. Sci. 65, 1484–1488.

Yoshioka, M., Hancock, B.C., Zografi, G., 1994. Crystallisa-tion of indomethacin from the amorphous state belowand above its glass transition temperature. J. Pharm. Sci.83, 1700–1705.

Yoshioka, M., Hancock, B.C., Zografi, G., 1995. Inhibitionof indomethacin crystallisation in poly(vinylpyrrolidone)coprecipitates. J. Pharm. Sci. 84, 983–986.

.

The relevance of the amorphous state to …...Many pharmaceuticals, either by accident or design,...

Documents

Transcript of The relevance of the amorphous state to …...Many pharmaceuticals, either by accident or design,...