`
`Crystal growth, polymorphism and
`structure-property relationships in
`organic crystals
`
`Joel Bernstein
`Department of Chemistry, Ben-Gurion University of the Negev, PO Box 653,
`Beer Sheva 84105, Israel
`
`Abstract. Understanding the role of structure in determining the properties of
`materials is a crucial aspect of the design of new materials. The existence of
`polymorphic crystal structures provides a unique opportunity to study structure(cid:173)
`property relationships, since the only variable among polymorphic forms is that
`of structure, and variation in properties must be due lo differences in structure.
`Systematic characterization of the polymorphic forms and acquisition of the
`ability \o grow crystals of a desired form are additional elements in the design
`strategy of new materials. The conditions and techniques required to obtain a
`particular polymorph, combined with knowledge of the crystal structures, can also
`provide information on the relative stability of the different structures. Studies
`of representative systems which illustrate the intimate connections between
`polymorphism, structur&-property relations and crystal growth are presented.
`
`1. Introduction
`
`The design and preparation of materials with desired
`properties is one of the principal goals of chemists.
`physicists and structural biologists. Achieving that
`goal depends critically on understanding the relationship
`between the structure of a material and the properties
`in question. Systematic studies of structur~property
`relationships generally require eliminating as many as
`possible of the structural variables in an attempt to
`isolate the one or few structural parameters which play
`the most important role in determining the particular
`property under investigation. For organic molecules a
`typical strategy might involve, for instance. a systematic
`variation in the mode or type of substitution on one part
`of the molecule in order to test a particular hypothesis.
`Variations in substituents, while they do often result in
`changes in structure, and the corresponding changes in
`properties, also lead to perturbations in the electr.onic
`structure of the molecules in question. In such cases,
`changes in properties cannot always be correlated
`directly with changes in structure. The existence of
`polymorphic forms provides a unique opportunity for the
`investigation of structure-property relationships, since
`by definition the only variable among polymorphs is
`that of structure. For a polymorphic system, differences
`in properties among the polymorphs must be due
`to differences in structure. As a corollary to this
`principle, a constancy in properties for a polymorphic
`system indicates a lack of structural dependence on that
`property, at least within the limitations of the structural
`
`0022•3727/93/8B0066•1 1$07.50 @) 1993 IOP Publlshlng Ltd
`
`variations through that particular series of polymorphic
`structures.
`For organic materials, studies of structure-property
`relations fall into two broad categories.
`In the first,
`the properties under investigation are due to strong
`interactions between neighbouring molecules, and we
`wish to study the changes in bulk properties resulting
`from differences in the spatial relationships between
`mo)ecules in the crystal, i.e. the crystal structure. In
`the second category we seek information related to
`variations in molecular structure, generally molecular
`conformation. The existence of different molecular
`conformations
`in different polymorphic structures,
`known as conformational polymorphism. also provides
`an opportunity for the study of the influence of crystal
`forces on the m.olecular conformation, since variations
`in conformation must be a result of different crystalline
`environments.
`Our intention in this paper is to provide a brief
`overview, with suitable examples, of the utilization
`of polymorphism in organic materials to investigate
`structure-property relations.
`
`2. How widespread is polymorphism in organic
`materials?
`
`the phenomenon of polymorphism was
`Although
`recognized by :Mitscherlich (1 822, 1823) at the dawn of
`modem chemistry, relatively little effort was devoted
`to
`the systematic search for and
`investigation of
`
`Merck Exhibit 2049, Page 1
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`polymorphic forms of organic materials until the last
`twenty-fl ve years or sot. The extent of the phenomenon
`is evident in the fact that over 3000 of the entries in
`the Cambridge Structural Database (Allen et al 199()
`contain qualifying descriptions as being members of
`polymorphic systems.
`While most examples of polymorphism are still
`discovered through serendipity, there are a number
`of areas of chemical research and development where
`full characterization of solid materials is critical in
`the determination of their ultimate use.
`These
`include phannaceuticals (Haleblian and Mccrone 1969,
`Haleblian 1975, Clements 1976), dyes (Walker et al
`1972, Griffiths aod Monahan 1976, Etter et al 1984,
`Tristani-Kendra et al 1983, Morel et al 1984) and
`explosives (Karpowicz et al 1983). Various aspects
`of the subject have been treated in books (Varna and
`Krishna 1966, Byrn 1983, Kuhnert-Brandstatter I971)
`and a number of reviews (McCrone 1963, Haleblian and
`McCrone 1969, Haleblian 1975).
`The proliferation of examples of polymorphism in
`these areas would seem to lend credence to the widely
`quoted statement by Mccrone (1963) that 'Vutually
`all compounds are polymorphic and the number of
`polymorphs of a material depends on the amount of
`time and money spent in research on that compound'.
`However, it is still not possible to predict with any
`reasonable level of confidence the crystal structure
`of an organic materiaJ, much less the existence of
`polymorphism, although considerable progress is being
`made in these endeavours (Roberts 1992, Sato 1992,
`Desiraju 1991, Gavezzotti 1989, 1991, Maddox 1988,
`Scaringe and Perez 1987, Fagan and Ward 1992).
`Moreover, since scientists do not make a practice of
`reporting negative results, we are generally not aware of
`unsuccessful systematic searches for polymorphs. When
`required. a systematic search for polymorphic fonns
`should be carried out on solid materials, involving a
`wide variety of techniques and conditions for growing
`crystals and enlisting a diversified armoury of analytical
`techniques to detect them. The range and combinations
`of crystal growth conditions are virtually infinite, and
`there is no way to guarantee the preparation of additional
`polymorphs of a substance, much less the generation of
`'all' of them.
`In spite of the considerable efforts and progress in
`understanding and controlling polymorphic behaviour
`(Weissbuch et al 1987) the phenomenon still seems to be
`at least partially shrouded in mystery and folklore, with
`tales of metastable crystal forms 'disappearing' in favour
`of more stable ones (Woodard and McCrone 1975,
`1acewicz and Nayler 1979, Webb and Anderson 1978,
`Scheidt et al 1983). To fuel this debate there have been
`reports of successful attempts (Catti and Ferraris 1976,
`Czugler et al l 981, Bar and Bernstein 1982), after
`considerable efforts, to overcome the 'poisoning' of a
`laboratory or even vast areas of the earth (Woodard and
`McCrone 1975).
`
`t One notable exception is the relatively obscure but very useful
`compilation by Deffet (1942) of polymorphic organic materials.
`
`Structure-property relationships in organic crystals
`
`3. Comparing polymorphic crystal structures
`
`The crystal structure representation containing most
`information is a stereo diagram of the packing, with a
`suitably chosen view of a central reference molecule and
`its immediate surroundings. Comparison of polymorphic
`structures is best ma.de by preparing the same type of
`drawing on the same reference plane of the reference
`molecule for all of the polymorphs. While this pictorial
`approach is very helpful to a viewer it is difficult to
`translate the graphic information into a form which can
`be communicated to others. The traditional method of
`doing so involves the preparation of a listing of the
`'short' intermolecular contacts (i.e. those Jess than the
`sum of the van der Waals radii) including the symmetry
`operations relating the neighbouring molecules. This
`representation falls at the opposite end of the information
`spectrum: it contains all the necessary analytical data,
`but is virtually impossible for the reader to visualize.
`A solution to this dilemma may lie in the use
`of graph sets which have been developed recently to
`represent the patterns of hydrogen bonds (Etter 1990,
`Etter et al 1990), but which, in principle, may be
`applied to any type of intermolecular interaction.
`Io
`this representation all hydrogen bonds may be classified
`as belonging to one of only four different patterns:
`intramolecular, finite dimeric, finite ring, or infinite
`chain. A complete definition of the pattern is given by
`specifying the number of donors (hydrogens) and the
`number of acceptors (basic atoms) as well as the totaJ
`number of atoms contained in the pattern. The hydrogen(cid:173)
`bond networks may then be summarized in a concise
`shorthand way which carries a great deal of information
`that may be used to compare readily the structures of
`polymorphic systems. Examples of such a treatment are
`given for the trimorphic system of iminodiacetic acid
`(Bernstein et al l 9<JO) and the. dimorphic L-glutamic acid
`(Bernstein 1991a).
`Because of their highly directional nature, as
`compared with other types of interactioos, this approach
`bas been applied initially to hydrogen-bonded structures.
`Other intermolecular atom-atom interactions also exhibit
`characteristic patterns (Desiraju 1991) which are now
`much more amenable to study because of the ready
`accessibility of the structural data from the Cambridge
`Structural Database and the analytical and statistical
`software associated with it. The extension of the graph
`set approach for cataloguing structures from hydrogen(cid:173)
`bond systems to other types of interaction is a natural one
`which should allow for much more facile understanding
`and comparison of polymorphic systems.
`
`4. Examples of structure-property studies on
`polymorphic systems
`
`4.1. Bulk properties
`For historical reasons, the properties of organic materials
`have been considered in terms of the molecular structure.
`This was a natural result of the fact that most of the
`
`867
`
`Merck Exhibit 2049, Page 2
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`J Bernstein
`
`chemistry of these materials was carried out in solution,
`where molecules might interact or react in a pairwise
`fashion. Even three-body interactions were considered
`quite unusual. The last two decade.s have witnessed
`a revolution in this point of view, as the wealth of
`possibilities for designing and utilizing the solid state
`properties of organic materials have become evident.
`These properties intrinsically depend on both the nature
`of the molecules and the way in which they interact with
`each other in the solid. Important information about
`the role of these solid state interactions can be obtained
`from the study of polymorphic materials, and variation
`of physical properties from one polymorph to another.
`Moreover, the recognition of the conditions required
`for the growth of the different polymorpbs can provide
`additional information on the energetic relationships
`among them, and the crystallization conditions necessary
`to obtain the crystal architecture with the desired
`intermolecular interactions required
`for a particular
`physical property. Two examples from the realm of bulk
`properties will serve to illustrate these points.
`
`4.1. l. Electrical conductivity. Organic materials are
`traditionally considered to be electrical insulators, but the
`discovery twenty years ago of metallic conductivity in
`crystals of the :re molecular complex of tetrathiafulvalene
`I and tetracyanoquinodimethan II
`
`(>=<J
`
`s
`
`s
`
`s
`
`s
`
`N\_ /=\., _ _;°N
`NC~CN
`
`II
`(Ferraris et al 1973, Coleman et al 1973) led ·to a
`revolution in thinking about these materials in particular
`and the potential for organic materials in general,
`as the basis for the next generation of electronic
`In
`components (Wudl 1984, Williams et a/, 1985).
`contrast to the vast majority of known 1r: molecular
`complexes that crystallize with plane-to-plane stacks
`of alternating donors and acceptors (mixed stack.
`figure l(a)) (Herbstein 1972), the complex of I and
`Il crystallized with segregated stacks of molecules
`along the same crystallographic axis (but not 01utuaJly
`parallel), each stack containing only one type of
`molecule.
`This structural feature has been shown
`to be
`a necessary condition for electrical conductivity in
`these materials, although the mixed mode of stacking
`is generally
`thermodynamically preferred one
`the
`(Shaik 1982). What means do chemists have at their
`disposal to overcome the tendency to form mixed stacks
`rather than segregated stacks? One method is to choose
`either a donor or acceptor with a completely different
`molecular shape. · The principles of close packing in
`organic crystals (Kitaigorodskii 1973) then suggest that
`molecules will pack in segregated stacks in order to
`most efficiently fill the space. Such reasoning led, for
`instance, to the 2: 1 'Bechgaard salts' (Becbgaard et al
`1980, 1981 ) based on III as a donor, and a similar series
`based on IV ('ET') as a donor, in which the acceptors are
`
`B68
`
`A
`D
`A
`D
`A
`D
`
`D
`A
`D
`A
`D
`A
`
`¼ 3.2•3.S J,
`
`A
`A
`A
`A
`A
`A.
`
`D
`D
`0
`D
`D
`0
`
`Hixtd stacks
`
`Segregated stack~
`
`(b I
`la l
`Figure 1. Upper part: schematic diagram of the mixed
`stack and segregated stack motifs for packing of molecular
`charge-transfer complexes; lower part: views of the two
`polymorphic structures of 11:111. In both cases the view
`is on the plane of the TCNO molecule (II), (a) The red,
`transparent, mixed stack complex, a semiconductor; (b)
`the black, opaque, segregated slack complex, a conductor.
`(From Bernstein (1991b) wilh permission.)
`
`nead y spherical (i.e. octahedral or tetrahedral) or short
`linear anions that fill the voids between stacks of donors.
`H,CXSo
`Se)(CH3
`I )=( I
`Se
`Se
`
`H..,C
`
`1-1,
`
`HI
`IV
`The 2: 1 stoichiometry ensures the partial charge on
`the donor stack, another necessary condition for
`conductivity, and many of these salts are conductors or
`superconductors.
`Proof of the relative stability of the mixed and
`segregated stack packing motifs and a recipe for
`obtaining crystals of the latter came with the discovery
`of a pair of polymorphic 1:1 complexes of II with m
`(Bechgaard et al 1977, Kistenmacher et al 1982). The
`red, transparent, mixed suick form of the complex is
`a semiconductor, while the black, opaque, segregated
`This finding
`stack is a conductor (figure I (b )).
`demonstrated conclusively that the segregated stacks
`are a necessary condition for electrical conductivity.
`Reflecting the relative stabilities for the two stacking
`modes noted above, crystals of the red semiconductor
`form are obtained by a 'thermodynamic' or 'equilibrium'
`crystallization: equimolar solutions of the donor and
`acceptor in acetonitrile are mixed and allowed to
`evaporate slowly. Crystals of the black form are
`obtained by a 'kinetic' or nonequil.ibriam crystallization:
`hot equimolar solutions of the donor and acceptor
`io acetonitrile are mixed and cooled rapidly. Some
`microcrystals of the resulting black powder are then used
`
`Merck Exhibit 2049, Page 3
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`lal
`
`t b)
`Rgure 2. Networ1< of ET (V) molecules in two phases of
`the salt (ET)t 13. The unit cell is included in each figure
`and thin lines indicate short intermolecular S- • ,S contacts.
`(a) {3 form; (b)" form. (From Williams et al (1991) with
`permission.)
`
`as seeds to obtain Jarger crystals of the mixed sracJc
`black form. Non-equilibrium crystallization methods,
`in particular electrochemical techniques, have become
`standard procedure for obtaining crystals of organic
`conductors, in part because of the ability to control and
`reproduce the crystallization conditions. However, as
`we indicate below, control and reproducibility do not
`guarantee obtaining a single or unique crystal phase.
`The role of polymorphism in understanding the
`connection between structure and properties is most
`poignantly represented by the BT2X salts of IV. No
`fewer than 20 have been reported to be superconductors,
`having among them the highest known Tc values
`(Williams et al 1991). For X=l3 alone, there are at
`least fourteen known phases, and learning to understand
`the crystal '.;fOwth conditions in these
`and control
`polymorphic systems is one of the greatest challenges
`facing workers in this rapidly expanding field. As above,
`crystals are typically obtained by electrocrystallization
`methods, using Pt electrodes and an H tube, and a
`constant current of ~ 1 µ,A cm-2• Crystal growth
`experiments can extend up to periods of months, with
`a number of polymorphs simultaneously appearing on
`the same electrode, in many cases with indistinguishable
`colours or crystal habits. For instance, Kobayashi et al
`(1987) identified the presence of the a, /3, y , .5, 0 and
`IC polymorphs of (ET)!I; in the same crystallization
`experimenl
`Intense activity in this field has led to the definition
`of some conditions for preferentially obtaining one, or
`a limited number, or phases (Williams et al 1991), but
`very often the isolation and identification of the various
`phases requires characterization of each individual
`crystal (Kato et al 1987).
`As an example of the type of variation observed, we
`can compare the /3 and IC phases of (EThI; (figure 2).
`The former, apparently favoured by thermodynamic
`crystallization conditions (e.g. low corrent densi ty) is a
`centrosymmetric triclinic structure with one fonnula unit
`of the salt in the unit cell. The symmetry arguments
`require that the anion lie on a crystallographic inversion
`centre, and that the donor molecules all be parallel, as
`shown in figure 2(a). The structure is thus characterized
`by stacks along the diagonal of the unit cell, with strong
`intennolecular S· • •S interactions between stacks. Tc for
`this phase is 1.4 K.
`The IC phase may be obtained, together with the
`et and 0 phases, in a THF solution under N2 with
`
`Structure-property reJationships in organic crystals
`
`a mixed supporting electrolyte of (n-C4~)4Nl3 and
`(n-CJ{9),tNAuh at 20 °C and constant current of 1.0 µ,A
`(Kato et al 1987). It is characterized by the fonnatioo
`of dimers, but a number of different salts exhibit more
`variability in this type of structure thao in the /3 phase
`(Williams et al 1991). 1C-(ET)!I; is centrosymmetric,
`with one Jayer per unit cell in a P2t/c structure (Kato
`et at 1987) (figure 2(b)) and bas a Tc of 3.6 K.
`Understanding the structure-property relationship in
`these materials is crucial to the rational development of
`organic conductors and superconductors with increas(cid:173)
`ingly high Tc values. The plethora of polymorphic
`structures can easily lead the unwary investigator astray,
`but it provides an opportunity not available in many
`other systems for isolating the structural characteristics
`required for a very specific physical property. Williams
`et al (1991) have also pointed out that the isostructural
`series of salts are important for the information they
`can yield. In this case, the structural parameter is kept
`nearly fixed (or only slightly perturbed) and the effect
`of chemical perturbations can then be evaluated.
`
`4.1.2. Aggregation of dyes. The aggregation of
`organic dye molecules has held the attention of dye
`chemists since its discovery over half a century ago
`(Smith 1974, Herz 1974). A lack of understanding of
`the phenomenon did not prevent extensive use beiog
`made of it, for instance, in the spectral 'tuning' of
`the response of photographic silver halide emulsions
`(Nassau 1983).
`Jelley {1936) found that when the
`solution conceotration of many dyes is increased, the
`intensity of the characteristic molecular absorption band
`decreases in favour of the growth of a new intense,
`narrow absorption band on t.he long-wavelength side of
`the original molecular absorption band {figure 3). The
`new absorption, called a J band after its discoverer,
`is commonly attributed to the formation of aggregates
`of dye molecules, and results from the coupling of
`transition dipoles within the aggregate structure.
`Since the aggregates are generally believed to contain
`only a few tens of molecules (Smith 1974) it has been
`difficult to study directly the relationship between their
`structure and spectral properties, While there is no
`certainty that a crystal structure represents the strucrure
`of the aggregate, the existence of polymorphic structure
`of a dye does allow one to probe directly the relationship
`between structure and spectral properties. For many
`years great difficulty was encountered in obtaining
`crystaJs of most dyes, to say nothing of polymorphlc
`forms. However, as a result of the energy crises during
`the 1970s, considerable research effort was expended on
`many dyes with potential applications as photovoltaic
`materials (Morel et al 1978, Morel 1979, Forster and
`Hester 1982). True to McCrone's dictum some were
`found to be polymorphic, and one, the squarylium
`dye V was the subject of a combined crystallographic
`and spectroscopic study to investigate the relationship
`between the structure and spectral properties of such
`dyes.
`
`B69
`
`Merck Exhibit 2049, Page 4
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`J Semstein
`
`s
`
`4
`
`.,.
`' 0 • 3
`..
`
`1
`
`H
`
`G
`
`F
`
`E
`
`0
`
`lal
`
`(b l
`
`17
`
`21
`
`18
`19
`20
`Wavenumber (l!r cm"1J
`Figure 3. Solution spectra of a dye exhibiting J banding.
`Solutions numbered 1-9 range in concentration from
`8.8 x 10-7 M to 1. 4 x 1 o M. The molecular absorption band
`is on the left; the J band on the right (see lext). (From
`Bernstein (1991b) with permission.) A, 8.81 x 10-9 M;
`B, 1.76 X 10-6 M; C, 2.20 X 10-6 M, 0, 2.64 X 10-6 M;
`E, s.oa x 10-6 M; F, 3.53 x 10- 6 M; G, 4.40 x 10--e M;
`H, 6.16 X 10-6 M; I, 1.44 X 10-3 M.
`
`lei,--- - - - - - - - . - - - - ------,,
`
`Figure 4. Stereo views of the crystal structure of the
`squarylium dye V: (a} triclinic structure, (100) face; (b)
`monoclinic structure, (100) face (from Tristani-Kendra
`et al (1983) with permission.); (c) view of the molecular
`plane, triclinic structure; (d) view of the molecular plane,
`monoclinic structure. (From Bernstein and Chosen (1988)
`wilh permission.)
`
`properties must be an expression of the differences of
`intermolecular relationships in the two structures. The
`polarized normal incidence .reflection spectra of the two
`crystals are gi ven in figure 5, and it can be readily
`seen that the spectra are significantly different between
`the two forms for light polarized along the long axis
`of the molecules. This must be a consequence of
`the difference in the interaction of a single molecule
`with its surroundings. The reflection spectrum of the
`band in the monoclinic form has been interpreted as
`being composed of two oscillators, while that of the
`triclinic form contains three, or possibly four, oscillators
`
`
`
`El -Q t -b -0 ' HO El
`'N
`/J
`{
`'
`f
`a1
`-
`
`. 2
`
`\
`
`Et
`
`OH o·
`V
`The material is dimorphic, fornting well developed
`crystals of both a violet cricJioic structure and a green
`triclinic structure. Representative views of the crystal
`structures are given in figure 4. Here again, · the
`crystal growth process is instructive in understanding
`the energetic relationships between the two observed
`phases. The two polymorpbs appear simultaneously in
`the same beaker, and hence they grow under identical
`conditions, indicating that they are of very similar
`energies. The energetic similarity would appear to be
`coincidental since the difference jn space groups and
`crystal structures indicated by figure 4(a) and (b) would
`suggest at least some difference in lattice energy and
`hence different crystallization conditions. The key to
`this conundrum lies, in part at least, in the way the
`crystal struccures are viewed. When the reference plane
`is the molecular plane (figure 4(c) and (d)) it is clear
`that the plane-to-plane stacking in the two structures is
`essentially identical. This apparently is the dominant
`interaction, and the one that governs the crystallization
`process. The only difference between the triclinic
`and monoclinic structures, then, is the relationship
`between the stacks, a translation in the triclinic structure,
`and a screw ax.is in the monoclinic structure. These
`must be either subsidiary to the plane-to-plane stacking
`interaction or equal to each other for the crystals to
`appear simultaneously (Bernstein and Chosen 1988).
`What are the spectroscopic manifestations of these
`structural differences? The molecule is essentially flat
`in both structures, so differences in the spectroscopic
`
`B70
`
`Merck Exhibit 2049, Page 5
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`\,.
`
`~
`
`100
`
`!al
`
`80
`~
`!- 60
`·;
`.:.
`..
`~ 40
`"'
`
`zo
`
`... ·--...
`,s
`
`0
`10
`
`,
`
`10
`
`25
`
`30
`
`,,;{
`
`l --..
`lit
`
`, .
`,,
`' . ,,
`
`.. -. . ...... ..... -~ .
`ZS
`20
`30
`Energy 003 cm· 1)
`Figure 5. Normal incidence reflection spectra of the two
`forms of V. In each case there are two spectra measured
`with the Jight polarized along each of the two directions
`(the so-called principal directions), as indicated in the
`upper right-hand corner, which also shows the projection of
`the molecule(s) onto the crystal face studied. (a) Triclinic
`polymorph, (100) face; (b} monoclinic polymorph, (100)
`face. (From Tristani-Kendra et al (1983) with permission.)
`
`'
`
`10
`
`IS
`
`(Triscani-Kendra and Eckhardt 1984, Tristani-Kendra er
`al 1985).
`
`4.2. Molecular properties
`At the other extreme in the range of interaction between
`molecules, organic crystals may be considered as simply
`a matrix for holding molecules in a specific juxtaposition
`with respect to each other. The interactions among
`molecules are ignored, at least to a first approximation,
`and the properties of the crystaJ reflect to a large extent
`the properties of the molecule, albeit with the three(cid:173)
`dimensional ordering imposed by the very fact that it
`is a crystal.
`When the molecule in question is conformationally
`flexible with a number of energetically accessible
`conformations (differing by less thao ~ 2 kcal mo1- 1)
`then different cryscallization conditions may lead to
`conformational polymorphs. An almost classic example
`is the dichloro-benzylideneaniline VI (X=Y=Cl).
`
`x--QJct;t L.F\_
`
`N~Y
`
`VI {X, 't' = 0, Bt, CH)}
`
`The molecule has only two degrees of conformational
`freedom, designated a and /3. The planar confonnation
`
`Structure-properly relationships in organic crysta)s
`
`of the molecule is known to be more energetic than
`the conformation in which a :::::: 50°, /J =:: 0° (Buergi
`and Dunitz 1971, Bernstei11 et al 1981) by about
`1.5 kcal mo1- 1
`• Two polymorphs are known: a triclinic
`form in which the molecule adopts a planar confonnation
`(a = {3
`:::::: 0°) (Bernstein and Schmidt 1972), and
`an orthorhombic fonn in whlch the conformation is
`a ~ 25°, p :::::: - 25° (Bernstein and Izak 1976).
`Hence this system is an example of conformational
`polymorphism where the crystal environment of the
`triclinic form is sufficient to stabilize the higher-energy
`planar conformation of the molecule (Bernstein and
`Hagler 1978).
`The treatment of the energetics of this system
`have been described in detail elsewhere (Bernstein
`and Hagler l 978, Bernstein 1987). However. in
`the present context it is of interest to review the
`crystallization behaviour.
`The triclinic form was
`discovered serendipitously in the course of a search
`for dicbloro-substituted molecules which might have
`a 4 A translation axis (Leser et al 1969, Green and
`Schmidt 1971, Desiraju 1989). A single oscillation
`photo of one crystal unambiguously indicated the
`presence of a 4 A axis, but that crystal did not diffract on
`the second photo, and aJJ subsequent attempts to obtain
`such crystals were unsuccessful. RecrystaJlizations
`carried out under equilibrium conditions yielded only
`the pale yellow orthorhombic form.
`The triclinic
`fonn was finally obtained under kinetic conditions by
`rapidly cooling a boiling saturated ethanolic solution in
`a desiccator freshly charged with CaCI2. The white
`(actually colourless when viewed individually under the
`microscope) needles, characteristic of a structure with
`a 4 A ax.is, proved to be metastable, and often upon
`standing for a few days wollld spontaneously revert to
`a pale yellow powder. The transfonnation could also
`be initiated by cleaving the crystaJs perpendicular to the
`needle axis, and sometimes could be foUowed visually
`by a gradual clouding of the crystal accompanied by
`a failure to extinguish polarized light. All of these
`phenomena are consistent with a metastable crystal fonn,
`and it is clear why somewhat unconventional methods
`were required to grow crystals of that form.
`The triroorphic dimethyl derivative VI (X=Y= CH3)
`represents an example of the ' disappearing' crystaJ form.
`Apparently unaware of the e:xistence of polymorphism,
`Buergi el al (1968) reported cell constants for the
`compound (form I). We had no trouble preparing the
`compound and repeating their determination CTzak 1973).
`After an interim of about eight months these crystaJs did
`not diffract well. Over a three-year period, subsequent
`recrystallization experiments, often preceded by a fresh
`synthesis of the material, resulted in the discovery of two
`previously unknown polymorphs (forms II and ill) (Bar
`and Bernstein 1977, 1982, Bernstein et al 1976), but
`failed to yield form I. Suspecting that our laboratory had
`been 'poisoned' by forms lI and m we took advantage
`of the opening of a new laboratory about a kilometre
`away to try again from scratch, using new reagents,
`new glassware, and a 'new' student-this time with
`
`B71
`
`Merck Exhibit 2049, Page 6
`Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.
`IPR2020-00040
`
`
`
`J Bernstein
`
`X (nm)
`
`,oo
`
`30
`
`2S
`
`20
`
`0
`
`f"
`•
`..
`
`1S
`
`10
`
`s
`
`300
`r,
`• , I
`\
`!
`'
`I
`I
`I
`\
`.t-.
`;
`/ 1 ·,
`. ,
`;': \
`I
`i
`i
`\
`,,
`ii
`'
`:,
`:,
`ii
`/ /
`: f
`f I
`'' . ' :
`
`;
`
`I
`
`l
`
`2SO
`
`\
`
`.. ,
`I
`;,.-. ..,_\
`,: \'"
`
`/'
`
`,, ,¥'\,
`
`•,.•
`
`40
`
`i.s
`
`20
`
`2S
`
`30
`
`3S
`(kKJ111
`
`Figure 6. Solution absorption spectra of benzylideneaniline
`(VI, X=Y=H) (- ) , azobenzene VII (- - - )and stilbene
`VIII (- • -) in ethanolic solutions. (From Bernstein et al
`(1979) with permission.)
`
`success. While the 'entire universe' was not poisoned
`by forms II and ill, there clearly is some basis to
`the claims of McCrone and Woodard regarding the
`difficulties in crysta11izing metastable polymorphic forms
`in an environment in which a more stable form has been
`obtained.
`The slight, but ce1tainly visually distinguishable,
`difference in the colour of the two polymorphs of the
`dichloro--derivative suggested using the two forms to
`make a direct el(amination of a problem which had
`occupied spectroscopists for nearly three decades (Hasel(cid:173)
`bach and Heilbronner I 968, and references therein).
`Benzylideoeaniline VI (X-= Y =H) is isoelectronic with
`azobenzene VII and stilbene VIIl
`
`o~-o
`
`CH ~ h
`
`VIII
`Vil
`but its solution absorption spectrum differs significantly
`from them (figure 6}. The difference in the absorption
`spectra between stilbene and azobenzene on the one
`hand, and benzylideneaniline on the other, has been
`attributed to a difference in molecular confonnation in
`solution:
`the former were believed to be essentially
`planar on the average, while the latter is not (Hasselbach
`and Heilbronner 1968). The non-planarity of VI is dtie to
`a repulsion between the hydrogen on the bridge and one
`of the ortho-hydrogens on the aniline ring. The bridge
`hydrogen is absent from VII and the increased length of
`the C=C bond in VIII alleviates that steric effect. For
`benzylideneaniline, the tendency towards nonplanarity
`due to the hydrogen repulsion is balanced in part by the
`,r-electron conjugation, leading to the minimum energy
`confor