`Disappearing Polymorphs
`
`193
`
`JACK D. DUNITZ*>~ AND JOEL BERNSTEIN*>#
`Organic Chemistry Laboratory, Swiss Federal Institute of Technology, ETH-Zentrum, CH 8092 Zurich, Switzerland,
`and Department of Chemistry, Ben-Gurion University of the Negev, Beer Sheva, Israel 84965
`
`Received November 1, 1994
`
`Introduction
`When a compound exhibits polymorphism-the ex-
`istence of more than one crystal structure-it may be
`important to obtain a particular polymorph under
`controlled and reproducible conditions. However, this
`is not always easy to achieve. Tales of difficulties in
`obtaining crystals of a particular known form or in
`reproducing results from another laboratory (or even
`from one’s own!) abound. Indeed, there are cases
`where it was difficult to obtain a given polymorphic
`form even though this had previously been obtained
`routinely over long time periods. Several monographs
`contain explicit or passing references to these prob-
`lems,l but much of this lore has gone undocumented,
`especially in the last 30 years or so. In this Account
`we present and discuss old and new examples.
`Crystallization is a process taken for granted by
`most practicing chemists; the majority of the tech-
`niques were developed long ago and are described in
`all standard laboratory textbooks. It is the standard
`method for purifying solid compounds, and chemists
`generally believe that they can control the process, at
`least when it yields the desired product. What is
`disturbing about the phenomenon of disappearing or
`elusive polymorphs is the apparent loss of control over
`the process: we did the experiment last week and got
`this result, and now we cannot repeat it! This kind
`of statement can lead to raised eyebrows or even to
`outspoken expressions of disbelief. We have ourselves
`experienced the frustration of not being able to
`reproduce an experimental result that was undoubt-
`edly obtained earlier.
`
`Crystallization: Nucleation and Growth
`The process of crystallization of a compound from
`solution or from the melt is poorly understood. At
`least two stages must be distinguished: the formation
`of a critical nucleus and its subsequent growth. The
`first step is decisive in that it can be regarded as being
`associated with a free energy of activation and is
`therefore rate limiting. Under suitable conditions,
`that step may be delayed almost indefinitely. For
`
`Jack D. Dunitz was born in Glasgow, Scotland, in 1923 and studied chemistry at
`Glasgow University. Following a decade of postdoctoral studies at Oxford, Caltech,
`NIH, and the Royal Institution, London, he moved to the ETH in Zurich as professor
`of chemical crystallography, a post he held until his retirement in 1990. He is the
`author of X-Ray Analysis and the Structure of Organic Molecules (1979) and (with
`E. Heilbronner) Reflections on Symmetry in Chemist ry.... and Nsewhere (1993).
`Joel Bernstein was born in Cleveland, OH, in 1941. He received his B.A. in
`chemistry from Cornell University in 1962 and Ph.D. in physical chemist
`from Yale
`in 1967. Following postdoctoral stints in chemical crystallography at UrLA with K.
`N. Trueblood and organic solid state chemistry at the Weizmann Institute with G. M.
`J. Schmidt he moved to the Ben-Gurion University of the Negev in Beer Sheva
`Israel, where he is now professor of chemistry. His research interests include a
`variety of aspects of the chemistry of the organic solid state, including polymorphism,
`structurGactivity relationships, hydrogen bonding, and organic conductors.
`0001-4842/95/0128-0193$09.00/0
`
`instance, Faraday2 observed that molten sulfur in a
`flask cooled to room temperature did not entirely
`solidify. When a drop of the fluid material was
`touched, it immediately crystallized; untouched, some
`drops were retained for a week in the fluid state.
`Faraday noted that this supercooled state of sulfur is
`analogous to that of water cooled below its freezing
`point, although the temperature difference is much
`greater (the freezing point of sulfur is 119 “C); De
`Coppet found that samples of salol (phenyl salicylate)
`could be kept in the liquid state at room temperature
`for periods of several years.3 When nucleation is
`rapid, the formation of many nuclei leads to many
`crystals, whereas slow nucleation tends to produce a
`smaller number of larger crystals. Of course, stirring,
`shaking, or other disturbances of the liquid phase
`during the crystallization process can affect the out-
`come.
`A striking case where nucleation was decisive in
`determining the result of a crystallization experi-
`ment has been described r e ~ e n t l y . ~ Sodium chlorate
`(NaC103) crystallizes in the chiral space group P213;
`that is to say, individual crystals of this substance may
`occur in enantiomorphic forms. Normally, crystal-
`lization from solution produces the enantiomorphs in
`roughly equal numbers. Kondepudi, Kaufman, and
`Singh5 found, however, that stirring an aqueous
`solution of this substance leads to a predominance of
`* Correspondence may be directed to either author.
`+ ETH.
`Ben-Gurion University.
`(1) Buckley, H. E. Crystal Growth; Wiley: New York, 1951. Tipson,
`R. S. Crystallization and Recrystallization. In Technique of Organic
`Chemistry; Weissberger, A., Ed.; Interscience Publishers, Inc.: New York,
`1956; Volume 111, Part I, Chapter 111, pp 395-562. Holden, A,; Singer,
`P. Crystals and Crystal Growing; Doubleday: New York, 1960.
`(2) Faraday, M . Experimental Researches in Chemistry and Physics;
`Taylor and Francis: London, 1853; p 212. On the following page, Faraday
`apologized for not having acknowledged observations along similar lines
`made earlier (in 1813) by M. Bellani: “I very hlly join in the regret ... that
`scientific men do not know more perfectly what has been done, or what
`their companions are doing; but I am afraid the misfortune is inevitable.
`It is certainly impossible for any person who wishes to spend a portion
`of his time to chemical experiment, to read all the books and papers
`that are published in connection with his pursuit; their number is
`immense, and the labour of winnowing out the few experimental and
`theoretical truths which in many of them are embarrassed by a very
`large proportion of uninteresting matter, of imagination, and of error,
`is such, that most persons who try the experiment are quickly induced
`to make a selection in their reading, and thus, inadvertently, at times,
`pass by what is really good.” Since Faraday’s times, these difficulties
`have multiplied out of all proportion, but we may still use his words to
`apologize to any scientists whose works we may similarly have over-
`looked.
`(3) De Coppet, M . L.-C.Ann. Chim. Phys. 1907,10,457. “La surfusion
`dure donc depuis bientBt 6 ans.” In another experiment, de Coppet
`reported that a sample of sodium sulfate, supersaturated with respect
`to the decahydrate, had still not crystallized aRer 25 years. In general,
`the higher the temperature to which the liquid was raised and the longer
`the time it was held at high temperature, the more resistant the liquid
`was to crystallization. Heating a liquid destroys residual order.
`(4) McBride, J. M.; Carter, R. L. Angew. Chem. 1991,103,298; Angew.
`Chem., Int. Ed. Engl. 1991,30, 293.
`(5) Kondepudi, D. K.; Kaufman, R. J; Singh, N. Science 1990, 250,
`975.
`0 1995 American Chemical Society
`
`Merck Exhibit 2168, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`194 Ace. Chem. Res., Vol. 28, No. 4, 1995
`crystals of one handedness, sometimes right, some-
`times left, but not depending on the direction of
`stirring. In checking this result, McBride and Carter4
`showed by video recording that a single nucleation
`event can produce almost all of the crystals formed:
`“...Crystals begin nucleating at random, but the first
`crystal to be struck by the stirrer clones hundreds or
`thousands of new nuclei. Growth of so many nuclei
`soon lowers the concentration of the solute below the
`threshold for spontaneous formation of primary nuclei,
`so that there is no way to begin crystallization of the
`enantiomer.”
`
`Seeding
`One way of influencing the crystallization process
`is by seeding, and here we need to differentiate
`between what we may term intentional and uninten-
`tional seeding.
`Intentional seeding is a common
`practice among chemists who wish to coax crystalliza-
`tion of a compound from solution or from the melt;
`small crystals or crystallites of the desired material
`(seeds) are added to the system. In this way, the rate-
`limiting nucleation step, which may be extremely slow,
`is circumvented. For this method to be applied, it is
`of course necessary that a sample of the crystalline
`material is available; that is, the compound must have
`been already crystallized in a previous experiment.
`When polymorphic forms of a substance are known
`to occur, intentional seeding with one of the poly-
`morphs is a useful and often the most successful way
`of preferentially producing it rather than the other.
`Seeding may also occur if small amounts of the
`crystalline material are present as contaminants:
`unintentional seeding.6 Unintentional seeding is often
`invoked as an explanation of phenomena which oth-
`erwise are difficult to interpret. We shall argue in
`favor of this explanation, although there is no con-
`sensus about the size and range of activity of such
`seeds, which have never actually been directly ob-
`~ e r v e d . ~ Estimates of the size of a critical nucleus
`range from a few tens of molecules to a few million
`molecules.6 With a size of about a million molecules,
`even a speck
`g) of a compound of molecular
`weight 100 contains approximately 10l6 molecules,
`sufficient to make 1O1O such nuclei. One can think of
`local seeding, where the contamination may apply to
`the experimentalist’s clothing, a portion of a room, an
`entire room, a building, or even, with increasing
`degrees of implausibility, to a district, a town, a
`country, a continent, and so on. In the limit we have
`what has been proposed as universal seeding (plan-
`etary seeding would be a more accurate expression),
`where the whole planet is assumed to be contami-
`nated.g A seed that promotes formation of a crystal-
`lization nucleus need not necessarily be composed of
`the same molecules as the compound that is to be
`crystallized. Specks of dust, smoke particles, and
`other small foreign bodies can act as seeds in promot-
`
`(6) It is well-known that it is oRen difficult to crystallize a newly
`synthesized compound. Subsequent crystallizations may be easier,
`because of the presence of suitable seeds.
`(7) Chemists and physicists have long become accustomed to postulat-
`ing models as explanations for phenomena that cannot be directly
`observed. The existence of atoms is perhaps the classic example.
`(8) Mullin, J. W. Crystdizution, 3rd ed.; Buttenvorth-Heineman
`Ltd.: Oxford,1993; pp 182-185.
`(9) The claim for “universal seeding”, taken literally, is obviously
`absurd. After all, the universe is estimated to contain about a millimole
`of stars, so one seed per star (per solar systemknot much-would need
`about 100 kg of the compound in question (MW = 100).
`
`Dunitz and Bernstein
`ing crystallization, which is the reason laboratory
`chemists often scratch the walls of a glass vessel with
`a glass rod to encourage a solute to crystallize.1°
`Polymorphism
`We have mentioned the phenomenon of polymor-
`phism, which is commonly understood as connoting
`the ability of a compound (or of an element) to
`crystallize in more than one distinct crystal structure.
`According to McCrone,’l “A polymorph is a solid
`crystalline phase of a given compound resulting from
`the possibility of at least two different arrangements
`of the molecules of that compound in the solid state.”
`Because polymorphs have different structures, they
`may differ greatly in density, hardness, solubility, and
`optical and electrical properties; e.g., diamond and
`graphite are two polymorphic forms (allotropes) of
`carbon. Many compounds are known to crystallize in
`polymorphic forms. In the inorganic and mineralogi-
`cal fields, these sometimes have different names, e.g.,
`ZnS, wurtzite and sphalerite; CaC03, calcite, arago-
`nite, and vaterite; TiO2, rutile, brookite, and anatase;
`but, more generally, different polymorphic forms are
`denoted by letters, A, B, C or a, p, y , etc., or by Roman
`numerals, I, 11, 111, etc., depending on the preference
`of the discoverer. McCronel’ has provocatively sug-
`gested that “every compound has different polymor-
`phic forms, and that, in general, the number of forms
`known for a given compound is proportional to the
`time and money spent in research on that compound.”
`In support of this, McCrone observes that many
`compounds of industrial importance (i.e., those on
`which a great deal of time and money are spent) are
`known to exhibit polymorphism: silica, iron, calcium
`silicate, sulfur, snap, pharmaceutical products, dyes,
`and explosives. Such compounds, unlike the vast
`majority of compounds that are isolated, are prepared
`and crystallized not just once but repeatedly, under
`conditions that may vary slightly from time to time.
`Similarly, in the biomolecular area, where much time
`and effort is invested in attempts to crystallize pro-
`teins under many slightly different conditions, poly-
`morphism is frequently observed.12 The universality
`suggested by McCrone’s statement may, however, be
`considerably tempered by the fact that fewer than 5%
`of the compounds in the Cambridge Structural Data-
`base (CSD) are known to be polymorphic (although it
`must be admitted that crystallographers typically
`choose one crystal specimen from their sample and
`leave it at that). Moreover, some very widely studied
`compounds have shown no evidence of polymorphic
`behavior, even though they have been crystallized and
`handled for many years under a far-ranging variety
`of conditions; naphthalene is an example that im-
`mediately comes to mind.
`Here we shall be concerned exclusively with molec-
`ular crystals, where the molecule may have the same
`shape in the two polymorphs or it may have a different
`shape, resulting in what has been termed “conforma-
`(10) “Auch das Reiben mit einem Glasstab an der Wandung des
`GefZisses schafff Keime, an deren Vorhandensein die Kristallisation
`gebunden ist.” Organikum; VEB Deutscher Verlag der WissenschaRen:
`Berlin, 1977; p 46.
`(11) McCrone,W. C. Polymorphism In Physics and Chemistry of the
`Organic Solid State; Fox, D., Labes, M. M., Weissberger, A., Eds.;
`Interscience: New York, 1965; Vol. 11, pp 726-767.
`(12) For example, according to the Protein Data Bank (distributed
`by Brookhaven National Laboratory, Upton, NY), the extensively studied
`human hemoglobin is known in monoclinic, orthorhombic, and tetragonal
`modifications; lysozyme in triclinic, monoclinic, orthorhombic, trigonal,
`tetragonal, and hexagonal ones.
`
`Merck Exhibit 2168, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`Disappearing Polymorphs
`
`ACC. Chem. Res., Vol. 28, No. 4, 1995 195
`
`I
`
`I
`
`Te mpe r a t u r e
`Temperature
`Figure 1. Free energy vs temperature diagrams for two polymorphs, with crossing points where their free energies cross: left,
`enantiotropic system;right, monotropic system.
`
`tional polymorphism”.13 McCrone’s criterionll is that
`polymorphs are different in crystal structure but
`identical in the liquid or vapor states. This implies
`that crystals containing molecules with different
`atomic arrangements are to be classed as polymorphs
`if the molecules concerned interconvert rapidly in the
`melt or in solution to give the same equilibrium
`mixture. Thus, this definition would encompass not
`only conformational isomers but all kinds of isomers
`in dynamic equilibrium. In phase-rule terminology,14
`the various polymorphs and the liquid obtained by
`melting them constitute a one-component system (or
`a two-component system if we consider solution of the
`polymorphs in a given solvent).
`Clearly, this definition is not completely satisfactory
`and leaves several kinds of borderline cases open: are
`syn- and anti-oximes in the solid state to be classed
`as polymorphs or as separate compounds? What
`about the various molecular species involved in the
`complex equilibria among open-chain and cyclic forms
`of saccharides (constitutional and configurational poly-
`morphs)? How long are we supposed to wait for
`equilibrium to be established? Should different hy-
`drates or solvates of a given compound be classified
`as polymorphs? (The term pseudopolymorphism has
`been proposed to cover such cases.) Definitive answers
`to these and similar questions cannot be given; they
`depend on one’s point of view. In the same way, there
`seems to be no unequivocal way of distinguishing
`between polymorphic transformations and solid-state
`chemical reactions. There are borderline cases that
`show characteristic features of both.
`In molecular crystals, free energy differences be-
`tween polymorphs are usually quite small, a matter
`of a few kilocalories/mole at most,15 and depend on
`temperature, mainly because of the entropic contribu-
`tion to the free energy. Because of the thermodynamic
`relation G = H - TS, the form with the higher entropy
`will tend to become the thermodynamically more
`stable form as the temperature is raised (Figure 1).
`Thus, over a small temperature range, and particu-
`larly between room temperature and the melting
`point, one polymorph or another can change from
`being the stable form to being metastable. If the
`
`(13)Bernstein, J.; Hagler, A. T. J. Am. Chem. Soc. 1978, 100, 673.
`Bernstein, J. Conformational Polymorphism In Organic Solid State
`Chemistry; Desiraju, G., Ed.; Studies in Organic Chemistry, Vol. 32;
`Elsevier: Amsterdam, 1987; pp 471-518.
`(14) See, for example: Findlay, A,; Campbell, A. N.; Smith, N. The
`Phase Rule and its Applications, 9th ed.; Dover: New York, 1951.
`(15) Kitaigorodskii, A. I. Adu. Struct. Res. Diffr. Methods 1970,3,173.
`
`thermodynamic transition temperature is below the
`melting point, the polymorphic system is known as
`enantiotropic (not to be confused with enantiotopic, a
`term applied to atoms or groups in a molecule that
`are related by an improper symmetry operation but
`not by a proper one, e.g., the two methylene H atoms
`in ethanol) and the transition is in principle reversible;
`if the transition temperature is above the melting
`point, then the system is monotropic and the transi-
`tion can take place only in one direction. A metastable
`form can persist for years, or it can undergo spontane-
`ous transformation to the stable form.
`
`Mechanisms of Polymorphic Transformations
`The title of this section promises more than it can
`deliver, because the mechanisms of polymorphic trans-
`formations in molecular crystals are largely unknown.
`The one type of transformation for which some level
`of understanding can be claimed is order-disorder
`transformations, where the high- temperature phase
`has essentially the same molecular arrangement as
`the low-temperature one and differs from it only by
`an increase in the crystallographic site symmetry of
`the structural units. This increase in apparent mo-
`lecular symmetry is due to an increase in crystal
`disorder such that the space-averaged, time-averaged
`distribution of matter has a higher symmetry than the
`instantaneous distribution in an individual unit cell.
`The reverse transformation corresponds to the onset
`of an ordering process. Such transitions are usually
`classified as “second-order” from the thermodynamic
`point of view, and, since they are virtually the only
`ones that can be handled on a theoretical basis, they
`receive the most attention in textbooks. From reading,
`one might even get the impression that order-disorder
`transformations are the prototype of phase transitions
`in general, but this is not the case.
`Presumably, as in the primary crystallization proc-
`ess, the mechanisms of most solid-solid transforma-
`tions involve the formation of critical nuclei of the new
`phase, followed by their growth. According to My-
`nukh,16 the nucleation step is critically dependent on
`the presence of “suitable” defects. Depending on the
`nature of these defects, nuclei of the new phase may
`be formed at different temperatures and grow at
`different rates. Thus, defects in the initial crystal
`structure may be necessary for initiating (or cata-
`(16) Mynukh, Yu. V. J. Cryst. Growth 1974,38, 284; Mol. Cryst. Lig.
`Cryst. 1979, 52, 467, 505.
`
`Merck Exhibit 2168, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`196 ACC. Chem. Res., Vol. 28, No. 4, 1995
`lyzing) nucleation of the new phase. Indeed, in some
`cases, the transformation can be induced by mechani-
`cally introducing defects, for example, by scratching
`the surface of the crystal with a pinpoint. On the
`other hand, there are also examples where the trans-
`formation is virtually instantaneous (and in one case
`even reversible), causing the crystals to "jump".17
`Solid-state transformations in molecular crystals
`often show a high degree of hysteresis. It may be
`necessary to heat the low-temperature form to a
`temperature well above the thermodynamic transition
`temperature before signs of phase transformation can
`be detected. Even when no solid-solid transformation
`of the low-temperature form occurs below the melting
`point, this is not sufficient proof that the system is
`monotropic; the transformation may simply be too
`sluggish to be observed. Similarly, transformations
`in the reverse direction, produced by cooling the high-
`temperature form, are also invariably accompanied by
`hysteresis. This can be so severe that a high-temper-
`ature form can sometimes be kept indefinitely at
`temperatures well below the transition point. Thus,
`X-ray structure analyses at 100 K have been made of
`crystal phases more than 200 K below their thermo-
`dynamic range of stability.18
`Vanishing Polymorphs
`Woodard and McCronelg described several cases
`where, after nucleation of a more stable crystal form,
`a previously prepared crystal form could no longer be
`obtained. Other examples were described by Webb
`and Anderson,20 who wrote, "Within the fraternity of
`crystallographers anecdotes abound about crystalline
`compounds which, like legendary beasts, are observed
`once and then never seen again." In a sober comment
`on these views, Jacewicz and NaylerZ1 criticized some
`of the more exaggerated claims. While admitting the
`role of seeding in promoting nucleation, they argue
`that the disappearance of the metastable form is a
`local and temporary phenomenon and conclude that
`"any authentic crystal form should be capable of being
`re-prepared, although selection of the right conditions
`may require some time and trouble."
`In most of the examples cited by these authors,
`relevant questions are left unanswered. Many chem-
`ists remain skeptical about a subject that calls into
`question the criterion of reproducibility as a condition
`for acceptance of a phenomenon as being worthy of
`scientific inquiry. Nevertheless, there are well-
`documented cases of crystal forms that were observed
`over a period of time but not thereafter, having been
`apparently displaced by a more stable polymorph. The
`relevant literature is scattered and almost impossible
`to find by subject searches. In the remaining space
`
`(17)Gigg, J.; Gigg, R.; Payne, S.; Conant, R. J. Chem. SOC., Perkin
`Trans. 1, 1987, 2411. Ding, J.; Herbst, R.; Praefke, K.; Kohne, B.;
`Saenger, W. Acta Crystallogr., Sect. B 1991,47,739. Steiner, T.; Hinrichs,
`W.; Saenger, W.; Gigg, R. Ibid., in press. Zamir, S.; Bernstein, J.;
`Greenwood, D. J . MoE. Cryst. Liq. Cryst. 1994, 242, 193. Etter, M. C.;
`Seidel, A. R. J . Am. Chem. SOC. 1983,105,641. Kohne, B.; Praefke, K.;
`Mann, G. Chimia 1988,42, 139.
`(18) For example, the white high-temperature modification of dimethyl
`3,6-dichloro-2,5-dihydroxyterephthalate, unstable below about 340 K,
`crystal structure analysis at 98 K. Yang, Q.-C.; Richardson, M. F.; Dunitz,
`J. D. Acta Crystallogr., Sect. B 1989,45, 312. Richardson, M. F.; Yang,
`Q.-C.; Novotny-Bregger, E.; Dunitz, J. D. Ibid. 1990,46, 653.
`(19) Woodard, G. D.; McCrone, W. C. J. Appl. Crystallogr. 1975, 8,
`342.
`(20) Webb, J.; Anderson, B. J . Chem. Educ. 1978,55, 644.
`(21) Jacewicz, V, W.; Nayler, J. H. C. J . Appl. Crystallogr. 1979, 12,
`396.
`
`Dunitz and Bernstein
`we review published examples, present some new
`results, and try to put the subject into perspective. We
`begin with one of the best-studied examples.
`1,2,3,5-Tetra-O-acetyl-/3-~-ribofuranose
`(I). The
`early history of this compound reads like a mystery
`story. As first prepared in 1946 in Cambridge, Eng-
`land, by Howard, Lythgoe, and Todd,22 the compound
`had melting point 58 "C.
`
`AcO
`
`OAc
`
`AcO OAc
`I
`
`Virtually the same melting point was measured for
`material prepared by a different method in Jena by
`Bredereck and H ~ e p f n e r . ~ ~ When several batches of
`the same material were prepared soon afterward
`(1949) in a different laboratory on the other side of
`the Atlantic, in New York, by Davoll, Brown, and
`
`V i s ~ e r , ~ ~ the first three preparations had melting point
`56-58 "C, but the fourth run yielded material with a
`distinctly higher melting point, 85 "C. Around the
`same time, in Jena, by direct acetylation of ribose,
`Zinne9 obtained a mixture of two tetraacetyl deriva-
`tives, one the ribopyranose and the other the ribo-
`furanose, with a melting point of 82 "C for the latter.
`The two high-melting compounds appeared to be
`identical, although the nature of the structural dif-
`ference between them and the low-melting form was
`unknown. So far, so good; innumerable examples of
`polymorphism are known. The low-melting form can
`be called A, the high-melting one B.
`After some time, however, the melting points of the
`early New York preparations had risen to 85 "C, and
`it was no longer possible to prepare the A form.24 A
`sample of A was sent from Cambridge, but when it
`was exposed to the air in New York, in a laboratory
`that contained samples of B, the crystals of A rapidly
`became opaque and transformed to B. In the mean-
`time, transformation of A to B was also found to have
`taken place in Cambridge. Since the A form could no
`longer be obtained in the New York laboratory, further
`experiments involving this form were moved to distant
`Los Angeles, where it was shown that when 1 g of A
`(melting point 57 "C) was inoculated with 1 mg of B
`(melting point 85 "C), the melting point of the sample
`was raised to 75-77 "C within 2 h and to 77-79 "C
`overnight.24 Similar phenomena were observed in
`Manchester.26 Low-melting A was first obtained, but
`when B was introduced into the laboratory, the whole
`of the material had the higher melting point and the
`low-melting form could no longer be prepared.27
`The scene now changes to Philadelphia, where
`Patterson and GroshensZ8 (the same Patterson as in
`the Patterson function used in crystallography) took
`on the task of measuring X-ray diffraction data for the
`two crystalline forms. Low-melting A was found to
`be monoclinic, space group P21, and the crystal was
`(22) Howard, G. A,; Lythgoe, B.; Todd, A. R. J. Chem. SOC. 1947,1052.
`(23) Bredereck, H.; Hoepfner, E. Chem. Ber. 1948,81, 51.
`(24) Davoll, J.; Brown, B. B.; Visser, D. W. Nature (London) 1952,
`170 GA
`- . - I (25) Zinner, H. Chem. Ber. 1950, 83, 153.
`(26) Farrar, K. R. Nature (London) 1952,170, 896.
`
`Merck Exhibit 2168, Page 4
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`Disappearing Polymorphs
`
`ACC. Chem. Res., Vol. 28, No. 4, 1995 197
`
`I1
`
`c
`
`Figure 2. Stereoviews of the two forms of I. In both cases the view is on the plane of Cl-O-C4 of the furanose ring: upper,
`monoclinic A form; lower, orthorhombic B form. For clarity, only carbon atoms are labeled.
`
`sufficiently stable to last for 7 weeks. At the end of
`this time, crystals of B were introduced into the room.
`After three days, the A crystal was unchanged, but
`when powdered B was sprinkled over the A crystal,
`the latter transformed completely to B in a few
`minutes. The transformed material still had the
`external shape of the original A crystal, but it was
`opaque and polycrystalline with no preferred orienta-
`tion of the crystallites. Crystals of B were found to
`be orthorhombic, space group P212121, with quite
`different cell dimensions from A. Patterson and
`Groshens noted that the molecular volume increased
`by about 2% during the A to B transformation (A,
`383.9 A3; B, 392.5 A3).
`In the early 1950s it would have been a major
`undertaking to determine the atomic arrangement in
`these noncentrosymmmetric crystals by X-ray analy-
`sis, and it was only some 20 years later that the crystal
`structure of form B was determined.29 The authors
`made no mention of the other polymorph. Essentially
`the same structure was found by P ~ p p l e t o n , ~ ~ who
`commented that an attempt to prepare the “rare” A
`form by application of high pressure was unsuccessful.
`Comparison of the structures of the two forms only
`became possible when the elusive A form was obtained
`
`(27) The state of affairs was summarized by Brown et al. (Brown, G.
`B.; Davoll, J.; Lowy, B. A. Biochem. Prep. 1955,4, 70) as follows: “The
`form first reported melted at 58” or 56” and the form melting at 84” was
`initially termed the B form. A number of laboratories have observed the
`transformation of the low melting into the high melting form and once
`the latter is obtained the former is not encountered.” For another
`contemporary account of the confusion, see: Overend, W. G.; Stacey, M.
`In The Nucleic Acids; Chargaff, E., Davidson, J. N., Eds.; Academic
`Press: New York, 1955; Vol. 1, p 44.
`(28) Patterson, A. L.; Groshens, B. P. Nature (London) 1954,173, 398.
`(29) James, V. J.; Stevens, J. D. Cryst. Struct. Commun. 1973,2, 609.
`(30) Poppleton, B. J. Acta Crystallogr., Sect. B 1976, 32, 2702.
`
`in Budapest and its crystal structure determined.31
`There is no simple structural relationship between the
`two polymorphs; the crystal packing is quite different,
`and although the ribose ring and its directly attached
`atoms are nearly superimposable, the molecules adopt
`different conformations with respect to the orienta-
`tions of the acetyl groups about the bonds C2-02,
`C3-03, and C5-05 (Figure 2).
`According to force-field calculation^^^ the intra-
`molecular nonbonded potential energy of the form A
`conformation is lower than that of the B conformation
`by 15.7 kJ mol-$ that is, the more stable molecular
`structure is found in the low-melting polymorph. This
`is reasonable, because, as mentioned earlier, the
`thermodynamic stability of a high-temperature form
`must be due to its higher entropy rather than to its
`lower potential energy (see Figure 1). The increase
`in molecular volume on going from the A to the B form
`is consistent with this.
`In spite of all the work done on this system, we still
`do not know the thermodynamic transition point,
`where the two free energy curves cross. From the
`many instances where A has been reported to trans-
`form spontaneously to B, we can infer that the
`transition point lies somewhat below normal labora-
`tory temperature. Thus, form A is likely to have been
`present as a metastable species during most of its
`existence. In spite of its thermodynamic instability
`with respect to form B, it may have tended to crystal-
`lize first from solution because of a more rapid rate
`of nucleation, a kinetic factor. Once formed, the
`crystals of A may endure for a longer or shorter period,
`depending on the local temperature and other factors.
`(31) Czugler, M.; Kalman, A,; Kovacs, J.; Pinter, I. Acta Crystallogr.,
`Sect. B 1981, 37, 172.
`
`Merck Exhibit 2168, Page 5
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`198 ACC. Chem. Res., Vol. 28, No. 4, 1995
`The solid-state transformation to B may take place
`spontaneously, or it may be catalyzed by the presence
`of seeds of B. In subsequent crystallization experi-
`ments in the same laboratory the presence of B seeds
`will circumvent the kinetic advantage of form A, once
`such seeds are present in the laboratory atmosphere,
`the lower solubility of thermodynamically stable B
`must tip the balance in its favor, resulting in the
`virtual “disappearance” of metastable A from labora-
`tories “contaminat