Organic Process Research & Development 2000, 4, 384−390

Crystallisation of Polymorphs: Thermodynamic Insight into the Role of Solvent

Terry Threlfall
Chemistry Department, UniVersity of York, Heslington, York, YO10 5DD, U.K.

Abstract: Experiments reported in the literature for crystallisation

Dependent on the conditions, crystallisation of polymorphs from of polymorphs are frequently unrepeatable. Detailed com-
solvent may be under kinetic or thermodynamic control. In the parison of the stated polymorphic outcomes of crystallisation
latter case the nature of the solvent will be immaterial in respect of much investigated substances often reveals the lack of
of the polymorph produced. The conditions under which each consistency between accounts. The author recently undertook
of these factors may apply are analysed in detail. The transition to crystallise 20 well-known pharmaceutical polymorphic
point between two dimorphs may not present a sharp divide in pairs, using apparently well-described recipes, often from
which crystallisation above and below the transition tempera- well-respected groups, but failed to obtain the expected
ture produces the high melting and the low melting polymorph, outcome in respect of the form obtained in 10 of these cases.
respectively. It is shown that even in those cases where the Others have recently encountered the same problem with
choice of solvent appears to be critical this may be a secondary respect to one of these compounds, carbamazepine.7 This is
effect related to the concentration attainable in that solvent at not a criticism of the veracity of previous work but a
a certain temperature rather than a specific effect dependent reflection of the complexity of polymorph behaviour, a matter
on solvent-solute interaction. A corollary to these consider- which is often overlooked. For example it appears no longer
ations is the necessity to determine solubility curves and possible to crystallise sulphathiazole I reliably from n-
metastable zone widths in order to be able to control polymorph propanol8,9 which has been a standard method for half a
crystallisation. century. Other sulphathiazole polymorphs also cannot be
prepared reproducibly by direct crystallisation, despite
intensive investigation and thus need to be prepared by
Introduction maturation or solvate decomposition routes.10 The sheer
There are numerous accounts in the literature of the effect confusion surrounding the reported polymorphs of mannitol
of conditions and of additives on the growth rate of crystals has been recently highlighted and clarification of its poly-
forming from solution and of the consequent variation in morphic behaviour attempted.11 A major part of the problem
crystal habit, face perfection, and crystal size distribution.1 of repeating the literature on the preparation of polymorphs
This knowledge in some cases allows control of the crystal- would appear to stem from the widespread but erroneous
line product in these respects.2 By contrast less has been belief that the solvent is the unique determinant of the
written about the rational control of different polymorphs3 polymorphic outcome, so that other essential parameters such
and even less from the point of view of understanding the as concentration, cooling rate, and temperature of nucleation
general factors underlying the observed results. Most of the are not recorded. The object of this paper is to indicate the
accounts which purport to address this latter issue prove on circumstances under which the solvent will not and cannot
close examination to be plausible deductions from a limited affect the polymorphic outcome and circumstances under
set of specific experimental observations, but unrelatable to which it may do so.
the general problem of the interaction between thermody-
namic and kinetic factors and of the relative importance of
Discussion
nucleation, early crystal growth and subsequent transforma-
tion. This state of affairs is all the more surprising because Instinctively, it would be expected that slow crystallisation
of the high profile enjoyed by polymorphism and related from dilute solution would produce the form stable at the
solid-state phenomena4,5 in recent years in the area of temperature of nucleation and crystallisation, whereas rapid
pigments, explosives, electronics, food, agrochemical, and, crystallisation from concentrated solution in which the
above all, in the pharmaceutical industry, in which regulatory kinetics could dominate would generate metastable forms.
controls necessitate the close examination of all products Whilst there may be some tendency for solutions to behave
under development for their solid-state behaviour.6 in this way, the effects of solution concentration are more

(1) Mullin, J. W. Crystallization, 3rd ed.; Heinemann-Butterworth: London, (7) Rusticelli, C.; Gamberini, G.; Ferioli, V.; Gamberini, M. C.; Ficari, R.;
1993. Tommasini, S. J. Pharm. Biomed. Anal. 2000, 23, 41-54.
(2) Weissbuch, I.; Popowitz-Biro, R.; Lahav M.; Lieserowitz, L. Acta Crys- (8) Blagden, N.; Davey, R. J.; Liebermann, H. F.; Williams, L.; Payne, R.:
tallogr. 1995, B51, 115-148. Roberts, R.; Rowe, R.; Docherty, R. J. Chem. Soc., Faraday Trans. 1998,
(3) Rodriguez-Hornedo, N.; Murphy, D. J. Pharm. Sci. 1999, 88, 651-660. 94, 1035-1045.
(4) Threlfall, T. L. Analyst 1995, 120, 2435-2460. (9) Apperley, D. C.; Fletton, R. A.; Harris, R. K.; Lancaster, R. W.; Tavener,
(5) Yu, L.; Reutzel, R. M.; Stephenson, G. Pharm. Sci. Technol. Today 1998, S.; Threlfall, T. L. J. Pharm. Sci. 1999, 88, 1275-1280.
1, 118-127. (10) Lancaster, R. W.; Threlfall, T. L. manuscript in preparation.
(6) Byrn, S.; Pfeiffer, R. R.; Ganey, M.; Hoiberg, C.; Poochikian, G. Pharm. (11) Burger, A., Henck, J.-O., Hetz, S., Rollingere, J. M., Weissnicht, A. A.
Res. 1995, 12, 945-954. and Stoettner, H., J. Pharm. Sci. 2000, 89, 457-68.

384 • Vol. 4, No. 5, 2000 / Organic Process Research & Development 10.1021/op000058y CCC: $19.00 © 2000 American Chemical Society and The Royal Society of Chemistry
Published on Web 09/15/2000
 dimorphic system in Figure 2, using the desirable nomen-
clature in which polymorph I is the high melting form and
polymorph II is the low melting one.4 This representation is
capable of subsuming the monotropic case by altering the
temperature scale so that the transition point X lies below
absolute zero or above the melting point of either of the
forms. Indeed enantiotropic systems in which the transition
point is far removed from room temperature behave for all
practical purposes as monotropic systems. For clarity the
initial diagram has been restricted to the dimorphic case, as
a trimorphic or polymorphic representation would have
rendered the diagram unintelligible. No issue of the general
principles to be discussed is lost thereby, but some further
considerations relating to solvate formation and highly
unstable monotropic forms are presented in Figures3 and 4.
Let us now consider the effect of cooling hot, undersatu-
rated, solutions of various concentrations, A, B, C, D, E, F,
and G as shown in Figure 2.
Figure 1. System of two polymorphs. Solution at point of (A) If a solution of initial concentration A is cooled, it
spontaneous nucleation with initial supersaturation (ci - cI)/cI will reach saturation and then pass through the metastable
with respect to polymorph I and supersaturation (cII - cI)/cI zone to a point A1 at which it will spontaneously nucleate
of polymorph II with respect to polymorph I. After Cardew
and Davey.10 and crystallise. If the rate of cooling is controlled so that
the combined regime of cooling and desaturation due to the
complicated than the above simplistic description would crystallisation does not take the concentration to the left of
suggest, as has been shown by Cardew and Davey.12 They the solubility curve of polymorph II (and it cannot lie to the
analysed theoretically the effect of crystallisation of a simple right of the solubility curve of polymorph I), the crystalline
dimorphic system at a constant temperature, illustrated in product must consist entirely of polymorph I at this stage
Figure 1. By consideration of the supersaturation of the initial (A2), having followed the path from A1 to A2. The product
solution with respect to the two forms and by making could be filtered off, or it could be cooled further, relying
reasonable assumptions about the interfacial tension they on the massive area of crystal surface of polymorph I and
were able to derive relative nucleation rates. They were also of nuclei of polymorph I in solution to bring down the rest
able to formulate equations describing the relative growth of the product as polymorph I. Provided that the transforma-
rates. Three types of behaviour were recognised, dependent tion of I f II is not rapid, this procedure will reliably produce
on the total variation of nucleation and crystal growth rate. polymorph I. If under these circumstances, polymorph I does
These were (a) the more stable form would crystallise transform to polymorph II, then there is no polymorphism
preferentially at all concentrations, (b) the less stable form issue as only polymorph II can be obtained and kept.
would crystallise preferentially only at high concentrations, The ratio of the solubility of two polymorphs in any
(c) the less stable form would crystallise preferentially only solvent is a constant at any given temperature, provided the
at intermediate concentrations. Presumably, this intermediate solutions are ideal, as this solubility ratio is a thermodynamic
concentration could be moved at least marginally towards invariant, being a measure only of the relative thermody-
the lower concentration region by a suitable choice of namic stability (Gibbs energy) of the polymorphs at that
parameters. Hence, there is the whole range of possible temperature.4,13 Therefore, the result of changing the solvent
behaviours with concentration in respect of the polymorphic will only be to transform the concentration axis linearly. If
expectations. Cardew and Davey describe the system as the solutions are non-ideal, then the concentration axis will
monotropic, because of their subsequent considerations of need to be re-scaled in a nonlinear fashion. The temperature
the transformation kinetics, but it is in fact completely axis and the diagram itself will remain precisely the same.
general, and indeed the diagram as drawn could represent The result remains the same for any solvent. The solvent
an enantiotropic situation. plays no part in the polymorphic outcome other than in
The system discussed applies either to an evaporative determining the numerical values on the ordinate.
crystallisation which has reached a given concentration or (B) When a solution of concentration B is cooled to B1
to a cooling crystallisation which has reached a given and seeded with polymorph I, it will behave exactly as
temperature (most probably room temperature) and then described under A above. The difference between A and B
begins to crystallise. To generalise the analysis and to make is that B cannot reach the spontaneous crystallisation zone
it more applicable to common practice it is necessary to of polymorph I before passing into the metastable zone of
consider a solution being cooled through the nucleation and polymorph II. Thus, for any solution crystallising within the
crystallisation temperature. This is set out for an enantiotropic area jkXn of the diagram, between the solubility curves for

(12) Cardew, P. T.; Davey, R. J. Tailoring of Crystal Growth; Institution of (13) Grant, D. J. W.; Higuchi, W. I. Solubility Behavior of Organic Compounds.
Chemical Engineers: Rugby, England, 1982, ISBN 0-9066-3623-X. In Techniques in Chemistry; Wiley: New York, 1980; Vol. XXI.

Vol. 4, No. 5, 2000 / Organic Process Research & Development • 385

Figure 2. Polymorphic system of two enantiomorphically related polymorphs I and II. Solubility curves, full lines; metastable
zone limits, dashed lines. Transition point X at temperature Tx. A-G, initial state of hot, undersaturated solutions; A1-G1 and B2,
E2 state of solution at point of initial crystallisation.

Figure 3. Solubility curves for a polymorphic system with two enantiomorphically related polymorphs and a solvate. Tx, transition
temperature between polymorphs. Ts, transition temperature between solvate and polymorph I.

polymorph I and polymorph II, the solvent does not influence polymorph I immediately, then on passing the metastable
the outcome. And for any solution crystallising initially and zone curve of polymorph II it reaches the spontaneous
substantially within this area, but then moving into another crystallisation region of polymorph II whilst still above the
region of the diagram, the solvent is unlikely to influence transition temperature. Which polymorph will be obtained
the outcome. will depend on the relative nucleation and crystallisation rates
(C) When a solution of concentration C is cooled, it of the two polymorphs. Since polymorph I on the premise
crosses the solubility curve of polymorph II before leaving just set out does not nucleate readily, it is likely that
the metastable zone of polymorph I. If it fails to nucleate as polymorph II will preferentially crystallise despite the
386 • Vol. 4, No. 5, 2000 / Organic Process Research & Development
Figure 4. Solubility curves for a polymorphic system with two accessible enantiomorphically related polymorphs I and II plus two
possible but highly unstable polymorphs III and IV. Parts of the metastable zone limits have been drawn in as dashed lines. III and
IV are monotropic in relation to I and II, but may be enantiotropically related to each other.

circumstance that the temperature lies above the transition- that a mixture of polymorphs (“concomitant polymorphs”)14
point. Solvent is important in this case as it can accelerate is likely to be formed. Apart from the solvent, the temper-
the formation of one polymorph at the expense of the other. ature is rightly regarded as the most significant parameter
(D) Cooling a solution of concentration D to the point controlling polymorph formation, but it is clear from this
D1 produces an even more surprising situation. Although analysis that the transition temperature cannot be regarded
the temperature corresponding to point D1 lies below the as a sharp watershed for the determination of polymorph
transition temperature, polymorph II is still in its metastable formation. Rather, there is a broad temperature range either
zone, whilst polymorph I has already reached its spontaneous side of the transition point within which kinetic effects driven
crystallisation zone. Hence, the expectation is for polymorph by solvent specifics and external conditions, such as stirring
I to be favoured, although the role of solvent could again be and material of vessel construction, are likely to dominate.
critical. The crystallisation behaviour implied by circum- It can be seen that the analysis presented by Cardew and
stances C1 and D1 needs to be contrasted with the result of Davey refers to the behaviour of solutions at points B2 or
equilibrating the products by heating an excess in solvent at C1.
those temperatures. Polymorph I will always be formed on (F) F mirrors the situation presented under B above, so
equilibrating above the transition temperature, whilst poly- that seeding will reliably produce polymorph II, irrespective
morph II will always be formed below it, and this will be of the solvent of crystallisation. It is worth noting that seeding
irrespective of the solvent used. The reason for this potential of polymorphs is by no means always a reliable procedure.10
discrepancy between crystallisation and equilibration behav- Seeding may not produce the desired polymorphic result
iour is that the latter is under thermodynamic control, whilst when carried out in the regions described under C, D, and E
the former may be dominated by the kinetics, which in some above.
cases, as just detailed, can even lead to the reversal of (G) Cooling a solution of concentration G is absolutely
formation of the expected polymorph with temperature. safe in terms of producing polymorph II, irrespective of any
(E) Cooling a solution E to any point (E1, E2) within the kinetic considerations. At no point does the horizontal line
area klmX, within which both polymorphs remain within cut the solubility curve for polymorph I. The remarks about
their metastable zones, will again lead to a situation in which change of solvent in respect of concentration A apply here
the polymorphic outcome is dependent on kinetics and also, but with even more certainty. Since no transformation
especially on accidental seeding. The specific relation of the is possible, any solvent will produce polymorph II.
temperature of solution to the transition temperature is likely The impression may be gained that the differences in the
to be of little import. The polymorphic result need not be behaviour of the solutions of different concentrations A-G
erratic, although it could be, but it is almost certainly and particularly of C-E are an artifact of the drawing of
unpredictable. The solutions considered under C and D will the diagram and that for example the solubility differences
move into this region soon after the onset of crystallisation, indicated by the curves for the two polymorphs are exag-
which may lead to changed driving forces for the formation (14) Bernstein, J.; Davey, R. J.; Henck, J.-O. Angew. Chem., Int. Ed. 1999, 38,
of each of the polymorphs. Hence, it is just in this region 3440-3461.

Vol. 4, No. 5, 2000 / Organic Process Research & Development • 387

gerated. Some of the concentrations discussed indeed have primarily concerned. The curve for form III has been drawn
been drawn so as to minimise clutter on the diagram. But with the intention that the virtual transition points with
this does not alter the fundamental messages being presented. polymorphs I and II lie at high temperatures, above the
In fact the diagram can be scaled as required and if this melting points of those forms. For form IV the implication
results in a very small temperature range for any of the is that the virtual transition points lie below absolute zero.
features then it is a warning of the keen control required The diagram as set up is now a universal representation of
and of the how easy it is to slip into near-chaotic conditions. a polymorphic system (except that the main protagonists are
Solution B, for example, if taken to point B2 will behave as still represented only as dimorphic). The addition of the two
does solution C1, indicating the considerable width of the very unstable forms III and IV on the diagram is realistic,
zone of ambiguous behaviour. Of course, either polymorph whether these can or cannot be observed in practice.
is capable of crystallising anywhere above the line jkXm, Computational studies generate many very unstable forms,
the area which represents solutions supersaturated with for which the question of the reality of their existence always
respect to both polymorphs, but we are concerned here with arises.16
inevitable and with probable rather than with possible Let us consider the implications of the presence of such
behaviour, see further discussion under H. The metastable unstable forms as III and IV on the polymorphic outcome.
zone width is interpolated to zero cooling rate by definition. (H) Suppose it is possible to cool rapidly and prevent the
In practice it can be considerably changed in real circum- crystallisation of polymorphs I and II and reach the saturation
stances, for example at practical cooling rates or when curve and the metastable curve of polymorph III or even of
different-sized vessels are used. This can complicate further polymorph IV as shown in Figure 4. If the nucleation kinetics
the control of polymorphs. What these considerations do of these forms are favourable, for example because of
reveal is why it is often so difficult to control polymorph structural similarities between the conformation in solution
crystallisations. and in the crystal, it may be possible to crystallise out these
The only exception to the certainties provided by cases forms. However, the improbability of being able to do so
A, B, F, and G is the possibility that a solvate can be formed needs to be emphasised. Certainly it will ultimately become
and that the solvate, because of structural resemblances or impossible to reach even less stable forms V or VI lying
favourable transformation pathways, can transform to poly- further to the left. The difficulties of preparing such forms
morph I after removal from the solvent. As shown in Figure has always been described in terms of the instability of the
3, the solvate solubility curve lies to the right of the crystal structure and of its potential transformation to a more
polymorphs. The solvate must be less soluble than the stable form. One sees from the diagram further reasons for
polymorphs, otherwise there would be no thermodynamic the difficulty of preparing such forms: the less stable they
incentive for its crystallisation.13 Above the temperature at are, the greater the distance on the diagram from the stable
which the curve crosses that of polymorph I the solvate polymorphs I and II. Hence, the solution spends more time
becomes unstable. There is often such a temperature limit in cooling, there are more competing forms to crystallise,
to solvate formation. This behaviour of the formation from and there is a greater temperature range over which other
a solvate of a polymorph which is otherwise inaccessible or forms can crystallise. The significance of this latter point is
difficultly accessible is shown by sulphathiazole. Many that molecular mobility at higher temperature can lead to
sulphathiazole solvates (e.g., from acetone, acetonitrile, more rapid transformation in competition with the lowered
n-propanol, pyridine) are formed in cooled solutions (20- supersaturation: at some point there will be a maximum rate
50 °C) and decompose readily in the absence of solvent, of crystallisation.1 Particle-free solutions, viscosity, and rapid
mainly to polymorph IV (pharmaceutical nomenclature, cooling will favour the formation of polymorphs such as III
equivalent to polymorph II of the Cambridge Crystal- and IV by minimising the possibility of pre-crystallisation
lographic Data Base) at temperatures at which polymorph of polymorphs I and II. In the case of polymorph IV, working
IV is not the thermodynamically stable form.10,12 Wirth and at low concentrations will be favourable. For polymorph III
Stephenson15 have described a situation in which it is there are competing factors such that it is not ascertainable
necessary to obtain a polymorph via a solvate. From the theoretically whether higher or lower concentrations will be
trajectories of the curves of the solvate and polymorph II it desirable.
would appear that in some circumstances there could also The diagrams have been presented as fixed patterns with
be a lower limit to the stability of a solvate. Although not a variable concentration scale, to emphasise the underlying
of direct consequence in the system discussed here, one can thermodynamic invariance. In practice and conceptually it
foresee circumstances in which it could confuse investigation is normal to consider, and easier to think of, a numerical
of polymorph crystallisation. The formation of unstable concentration scale and place the curves appropriately.
solvates as intermediates in polymorph crystallisation may However, this viewpoint can lead to illusions about the role
be more common than has been supposed,10 and provide a of the solvent. When a crystallisation experiment is under-
source of unexpected solvent influence. taken, a solution of a given concentration is cooled and
In Figure 4 have been drawn two further curves of forms crystals are observed to form at a certain temperature. The
which are thermodynamically very unstable in respect of the supposed control of polymorph formation by solvent may
two polymorphs I and II with which we have been so far
(16) Gavezotti, C. A. Theoretical Aspects and Computer Modelling of the
(15) Wirth, D. D.; Stephenson, G. Org. Process Res. DeV. 1997, 1, 55-60. Molecular Solid State; Wiley: Chichester, 1997; Vol. 2.

388 • Vol. 4, No. 5, 2000 / Organic Process Research & Development

not be due to any specific feature of the solvent (although predicting polymorphic outcomes of crystallisation of novel
this is sometimes strikingly the case)10 but to the attainment materials for several reasons. First, prediction cannot be
of saturation at a certain point in the diagrams at the particular carried out until the solubility and the metastable curves have
concentration attainable in that solvent. For example, if in been determined. By such time, the polymorphic behaviour
solvent 1 a solution of concentration corresponding to A is of the system may have already become apparent. Second,
cooled to A1 and at which point it behaves in an exemplary there is no means of knowing at the preliminary stage
fashion and crystallises rapidly, then polymorph I will be whether the supposedly dimorphic system may be tri- or
obtained. If the same numerical value of concentration is polymorphic, so that working at a higher or lower concentra-
employed in solvent 2 in which the polymorphs are twice tion may simply generate these other forms. This does not
as soluble, then this will now correspond for example to alter any of the principles set out above but may render some
concentration F because of the expanded or offset abscissa. of the detail irrelevant. Then there is always a phenomenon
Consequently, the polymorph produced will now be II. which renders the forecasting of polymorph crystallisation
Although the observations might be interpreted as a specific difficult, namely that of a solution well within the spontane-
solvent effect, it is clear that the same result as in solvent 2 ous crystallisation zone which hangs for hours up to weeks
could be achieved by working in solvent 1 at half the before crystallising either in a controlled manner or by
concentration. Without the benefit of the diagram this might suddenly crashing out.
be interpreted as a specific solvent effect related to the What the above analysis may emphasise is that, once the
solvent-solute interactions and to bulk effects (e.g., inter- solubility diagram for one solvent has been constructed, very
facial tension) of the concentrated solution. Of course the few measurements will be needed to construct the whole
latter might be the case, but in the absence of the solubility diagram for any other solvent. The current philosophy of
diagram it will not be possible to begin to judge. Specific screening for polymorphs would appear to be to crystallise
solvent effects are important in polymorph formation, but from as many solvents as possible, without concern for
by definition this is not open to the generalised considerations concentration. Thermodynamic realities would indicate that
presented here. Two nonspecific but solvent related effects concentration issues ought also to be considered at an early
needs to be emphasised, namely that there is a minimum stage of investigation, even before the solubility diagram has
solubility required in a given solvent to be able to reach all been drawn. In particular the temperature of first crystalli-
the points on the diagram and that the greater the concentra- sation ought to be noted. Intervention by solvates and
tion above A, the easier it is to control the crystallising hydrates in the process of crystallisation of polymorphs may
conditions with respect to the coordinates on the diagram. be more common than is generally admitted. Because water
The structure of solutions has been little investigated. is such a small molecule even a low concentration in the
However, conformational effects in solution, alluded to solvent may be capable of producing a hydrate, especially
above, could obviously favour kinetically the formation of one involving a fractional mole of water. Therefore control
specific solid forms, as could dimer or chain formation of the dryness of solvents must also be a consideration in
through hydrogen bonds and specific solvation patterns. It the control of polymorph crystallisation. Another important
is well-known, but not in the scientific literature, that issue, the question of practical seeding, is too large a topic
polymorphism is a persistent feature of final products but to be considered in detail here. Attention is drawn to the
rarely of intermediates. One reason for this is undoubtedly article by Beckmann in this issue.18 It is worth emphasising
McCrone’s dictum17 that the number of polymorphs is that seeding can only be effective if the required form lies
proportional to the time spent searching for them. Another within a thermodynamically allowed area of the concentra-
reason is surely the elaboration from simple intermediates, tion-temperature diagram and is undertaken before compet-
often by concatenation, of multiply conformationally flexible ing nucleation has begun.
molecules with myriad opportunities for packing. The
increasing compexity of modern pharmaceutical products
could also account for the increasing frequency of the Conclusions
observation of polymorphic behaviour as well as for the great
frequency of the occurrence of polymorphism amongst When crystallisation of a dimorphic compound from its
pharmaceutical products in contrast with other categories of solution is conducted sufficiently above or sufficiently below
chemicals and for the increasing frequency of more elaborate the transition point, the solvent used for the crystallisation
crystal structures with multiple non-equivalent molecules in is immaterial, provided that the solubility is adequate to allow
the unit cell. the prescribed concentrations to be reached. Irrespective of
the kinetics, the outcome is under total thermodynamic
control.
Practical Implications
Full knowledge of the solubility and metastable curves
This discussion is presented as an aid to understanding is required to be sure that the required points in crystallisation
the thermodynamic and kinetic factors and the role of solvent space are reached.
in the crystallisation of polymorphs. It is of less value in When crystallisation takes place near a transition point,
the choice of solvent may or may not be critical, but this
(17) McCrone, W. C. In Physics and Chemistry of the Organic Solid State;
Fox, D., Labes, M. M., Weissberger, A., Eds.; Interscience: New York,
1965; Vol. II, p 725. (18) Beckmann, W. Org. Process Res. DeV. 2000, 4, 372-383.

Vol. 4, No. 5, 2000 / Organic Process Research & Development • 389

will be determined by the relative kinetics of formation, relative rates of nucleation of the dimorphs in a given solvent
growth, and transformation of the two polymorphs in the and by the rate of transformation of the polymorphs.
various solvents.
The temperature of nucleation and crystallisation in a Acknowledgment
given solvent, whether just above or just below the transition I thank Dr. Bob Lancaster of Glaxo Wellcome R&D,
point, may not, in contrast to what has always been supposed, Stevenage, for his support and continuing encouragement.
be significant. The outcome will often be under kinetic Received for review June 1, 2000.
control and in that case will be determined solely by the OP000058Y

390 • Vol. 4, No. 5, 2000 / Organic Process Research & Development