632-650
`
`CRYSTAL
`GROWTH
`g,aDEs1GN
`Perspective
`
`2011, Vol. 11
`
`D01: 10.1021/cg1013335
`
`Published as part of the Crystal Growth & Design 10th Anniversary
`Perspective.
`
`Polymorphism - A Perspective
`
`Joel Bernstein*
`
`D€p(.|t“tm£’fl! of Chem.".s'trv, Ben-Gmton UnrTv:>r.r.e'tv ofrlie Negev, Bee)‘ Slmva, Israel‘.
`'
`’
`’
`*Curr€.m‘ address.‘ Faculty of Sciemre. New York U!1I't'£’J‘.'i}ilAt‘ Aim Dlzrihi, A/nu Dlrahi,
`United Arab En1[i'ur9s. E-m.:uTl'.' j(J£’f.f7€6}‘fl.‘.'I£’l!'I’1 (EjJ.ii_i'ii.edu or _1.vJ9I(2_?'11gu.m?.ii.
`
`Received 0 ember I0, 2010,‘ Rew'.s'ed Manuscript Receitmi Nm-‘ember 16, 2010
`
`Case No. 2:10-cv-05954
`Janssen Products, L.P, at al.
`_
`_
`_
`v. Lupin Umned. et al.
`
`
`
`PTX926
`
`Introduction
`
`Pe1'.s'pecIr'vr» — “at particular evaluation ofa situation or facts, especially
`from one persons point of view“ - “a measured or objective assessment
`of a situation. giving all elements their coniparative point of view".1
`
`The continuing success and increasing impact of C'ry.s-ml
`Growth & Design in its first decade is due in large measure to
`the editorial leadership of the ournal, but no doubt it has also
`benefited from the increasing interest and activity in research
`on polymorphism in molecular crystals in all of its ramifica-
`tions. There has been an exponential increase in the number
`of publications exploring and exploiting the phenomenon
`of polymorphism in particular and crystal forms in general.
`A comprehensive review of the subject is beyond the scope of
`this Perspective, so in concert with the definition of the term in
`the epigraph I will try to give a view of where we are and some
`of the directions We might be headed. In particular, I will
`attempt to give “a measured or objective assessment of a
`situation. giving all elements their comparative point of view",
`but the reader must remember that after all, as also appears in the
`definition. this Per.3pectire is written “from one person's point of
`view”, and thus nwessaiily reflects personal scientific bias.
`
`Definitions and Terminology
`
`The launching of Crystal Growth & Design was nearly
`coincidental with the publication of Polymorphism in [Molec-
`ular Cr_1'.stal\'2 (hereafter cited as [PMCnn—mt] for specific
`page numbers). As noted in the preface to the latter, my intent
`was to provide an introduction and entry into the field for
`those encountering it for the first time, as well as to provide a
`basic body of literature that could be used as a starting point
`for keeping track of subsequent developments by suitable
`citation searches. In keeping with that spirit, and that of a
`Pe:‘.s'pecti1'e, most of the discussion here will relate to issues
`and developments in the past decade.
`That period has witnessed a continuing debate about the
`terminology and nomenclature of multiple crystal forms.3'7
`Time and experience have apparently not diluted the general
`agreement with McCronc’s definition of polymorphism in
`molecular crystals[PMC2"‘l vide infra. although some alterna-
`tives have been offered.7 However, the natural tendency of
`scientists to categorize and to pigeonhole phenomena has led
`to a number of definitions of related phenomena that are
`misleading at best and simply incorrect at worst.
`
`Perhaps the centerpiece of this interchange has been
`pseudopolymorphisn-i[PMC4_fi] that prompted an exchange of
`letters in this journal. The self-styled “polemic” was initiated
`by Seddon3 and I was one of the later joining disputantsj
`In spite of the fact that the term has been included in
`CrystEngWiki (“a service provided by CrystEngCommunity,
`hosted by the UK's Royal Society of Chemistry") [httpzff
`prospectrsc.org/wikifrscfindex.php?title = CrystEngWiki:About]
`and defined there as “When different crystal types are the
`result of hydration or solvation, the phenomenon is called
`pseudopolymorphism”, it is mistaken English usage and a
`scientific misnomer. In response to my objection to the use
`of this term, apparently first coined by McCroneg and sub-
`sequently adopted by Bryng to encompass all solvates and
`hydrates, there appeared a rejoinder advocating continued use
`ofthe term — only because it is apparently part of the lingua
`firtznca (i.e., “...this widely accepted term").6
`Language is an aid to science only if we define our terms
`precisely and unambiguously; it is inappropriate and mislead-
`ing to propagate inaccurate. indeed incorrect, terminology.
`To repeat, in short, the conclusion of my earlier missive:
`“Pseudo means jrtlse. Appending “pseudo" before another
`noun is an abdication of the responsibility of naming. It is
`a failure to invent a proper name. Solvates and hydrates are
`not false anything. least of all polymorphs (by any accepted
`definition); moreover. any of those solvates and hydrates can
`be polymorphic, which would require their absurd description
`as polymorphs of pseudopolymorphs. Again, the existing
`literature cannot be corrected, but the future literature need
`not be polluted or corrupted with such imprecise, confusing,
`erroneous language." See, for example, the discussion on the
`pseudoasymmetric carbon atom and the confusion engendered
`by this “infelicitous term." '0
`Furthermore, pseudopolymorphism is regressive with respect
`to the history of chemistry nomenclature. Liebig labored at
`C/H analyses with his Kaiiapparat and pan balances he
`designed in order to differentiate discriminate compounds
`with different chemical compositions. From this we gained
`the science of organic chemistry and i'.s'amer.v, a term qualified
`later only when models of bonding required discriminating
`con.s'ri'mrioi1al isomers and .vter'coisomers. In science. we add
`language to distinguish. not to confuse. Pseudopczlymorphixm
`convolves that set of things with similar structures and
`different compositions and that set of things with very differ-
`ent structures and identical compositions.
`
`puhs.acs.org/crystal
`
`Published on Web OM19/2011
`
`2011 Anierican Cheinical Society
`
`Janssen Ex. 2016
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`lPR2015-01030
`
`(Page 1 of 19)
`
`

`
`Perspective
`
`Crystal Growth & Design. Vol. 11 , No. 3. 2011
`
`633
`
`In a 2008 paper entitled “Polymorphism: The Same and
`Not Quite the Same",7 Desiraju commented on many of
`the idiosyncrasies of the world of multiple crystal forms.
`the variety of compositions and structures observed. and
`the conceptual problems in defining the various situations
`encountered and the nomenclature to describe them. These
`
`Conundrums are not unique to crystal forms; they exist in all
`branches of chemistry. As I noted earlier,5 using the chemical
`bond as an example: in order to properly conceptualize them.
`we tend to make our definitions and our descriptions in terms
`of often idealized models. We then can attribute some or all of
`
`the idealized characteristics to any particular (often nonideal)
`situation. Much of chemistry - often the most interesting
`chemistry, because it does not fit the definitions precisely -
`deals with understanding the sources of the differences from
`the idealized situations. It is simply not necessary, nor is
`it particularly informative, to have a definition or a nomeii-
`clature for every imaginable situation. However, the same
`author seems to be somewhat uncomfortable with the term
`
`pseudopo{v:1i0rp}ii.mz. ln the abstract of a 2003 paper he co-
`authored “Five New Pseudopolymoiplis ofmi?-Tiiiiiti'obeii2ene“
`appears the following statement: "This study indicates that
`the use of terms such as .rriimIe. p.reu(lop0/'v1'm0rpii_ d0imr~
`acceptor coniplex. and ?‘.l“l'0fé,‘L‘l£f£II' conipiex is a subjective matter,
`and also that a better definition for the term p.re'i.rdopulymorph
`may be needed, especially because it occurs frequently in the
`pharmaceutical literature."' I There may be some strength in
`numbers, but the fact that the term “occurs frequently in the
`literature" does not make it correct or incorrect. The ques-
`tion is simply wlietlier it is a definition that iiicisively divides
`a set of things and is linguistically and scientifically appro-
`priate. To quote the late Jerry Donohue in another context,
`“It isn't". '3
`I simply do not agree that there is a “need for the term ‘false
`polymorphism’ or p.s'ezrdap0I_i'inorp/rism", which it is claimed
`arises. for instance, in the difficulties of describing crystals
`with more than one component. What is the difficulty‘? Just
`say that it has more than one component. The dilemma,
`determining whether two materials are polymorphic. can be
`resolved by McCrone‘s simple test of whether they have
`identical melts. Materials of different composition will
`not meet that crit.erion:_ they are not polymorphic [nor iso-
`morphous) and therefore not pseudopolymorphic. If the
`relative amount of solvent or guest varies. then simply call it
`a variable soivate.'3'M or a host with a variable guest. Both
`of these are simple, clear descriptions, requiring no further
`amplification.
`I also see no reason to restrict polymorphism to a single
`compound. Why, for instance, can co-crystals not be poly-
`morphic (vide iiifra)? Surely among molecular complexes
`(lately reincarnated as co-crystals), there are numerous exam-
`ples of polymorphs as defined above[P“C'gq_'97] that meet
`McCrone"s criteria. If the composition varies, then clearly
`they do not meet the criterion of having identical melts and
`they are not polymorphs; they are something else.
`Other hitchhikers on the nomenclature bandwagon should
`no longer be carried. For instance, .s'rrm.'tural1Job=nrmyJiiirirz'5
`is simply redundant and belongs in the etymological trash
`heap. Lll(€WiSEN}‘iiIf!0r1p()f1.‘I?10F]J1tllW7‘t
`'6 HI‘lCl]?£i(‘ki1’lgp(Jl_)'f?’10ijJf1.$‘. '7
`And what criteria must a compound meet in order to be
`classified as a ,r)izarmc1ceim'm:’ pu1'ymorph?'8 Must
`it be a
`pharmaceutically active ingredient‘? Do excipients count‘?
`Wliat if the compound is taken off the market for some
`reason? Does it no longer qualify as a phm‘nia.'ceuri'arI
`
`polyniorp/1? To quote Stalily,” “There is really no reason
`to classify organic compounds as ‘pharmaeeutical‘ or
`‘non-pharmaceutical’.
`in discussing solid properties. Coin-
`pounds used in the pharmaceutical industry are quite structur-
`ally varied; there is not any specific chemical attribute that
`renders them pharmaceutically active or warrants the term
`p}iarmar*eui‘fmIpo{vrnorpl1_"2O
`On the other hand, there certainly are situations where a
`new term is helpful in recognizing and even describing a
`particular previously unobserved phenomenon. An example
`is the recently coined i.s'0i0pameric p0.’yr1i.0r'phism,2' describing
`a change in crystal structure upon changing t.lie isotopic
`identity of one or more of the atoms in a molecule. While
`seemingly an isolated incident when initially discovered_. at
`least one other example has been reported.22 There will
`undoubtedly be many others given the incredible sensititivity
`of molecular crystal structure to the positions of hydrogen
`atoms.” In a related development,
`the influence of the
`isotopic distribution of the solvent on the polymorphic out-
`come of the crystallization of glycine from aqueous solutions
`has also been observed.24
`Perhaps somewhere in between these extremes of appro-
`priate and nonappropriate definitions is the case of iairtanieric
`p()f}‘m0f‘phf.l'm.2S particularly of omeprazole. In keeping with
`the spirit of the McCrone definition alluded to earlier. the
`appropriate questions to ask would be essentially: (1) Are the
`crystal structures different‘? (2) Do they give the same melt‘?
`The answers to both are somewhat ambiguous. Bhatt and
`Desiraju obtained five different forms with varying ratios of
`two tautomers. Three forms have been patented, distinguish-
`able by their solid-state properties. Do they all give the same
`composition of tautomers in the melt"? As the authors point
`out, this may be a matter of time. until equilibrium is reached;
`that also may be a complicating factor. The problem
`seems closely related dynamic isomerism. also discussed by
`McCrone.R‘lPMCf‘]
`This actually brings us back to the question ofpolymorphism
`in molecular crystals. McCrone‘s definition first requires
`establishing the concept of molecularity, and in those cases
`the definition works very well. Even though McCrone’s
`definition is still very useful, the last half century has led to a
`vastly expanded view of solids. which flaunts this concept
`wonderfully. so that even molecularity is an inherently fuzzy
`concept. For instance. ai'e molecular solids limited to neutral
`molecules‘? Are inet.al organic frameworks molecular solids‘?
`At what point is a solid no longer i’i"l0fe'£‘1([[U‘?
`In the end, on the issue of nomenclature I would prefer
`pragmatism to dogmatism.26 What excites and motivates
`many of us about chemistry is the infinite vai'iability that is
`possible and often observed. That variability defies precise
`definitions in many cases. As noted above, we use definitions
`to define essentially ideal cases in order to create a conceptual
`framework, and we then describe any particular situation as
`oxliibitirig or embodying features from more llian one of those
`ideal situations. The example I gave earlier was that of the
`chemical bond — in many cases described as a covalent bond
`with a certain amount of polaiity or ionic character. All the
`terms are clear and the meaning is clear. This is the language of
`chemistry and we should use that same language in the realm
`of multiple crystal forms. When a particular situation defies
`a precise description on the basis of our definitional frame-
`work that does not necessary warrant the creation of a new
`descriptive term. The perfectly acceptable alternative for
`special situations is to describe it as it is; it does not necessarily
`
`Janssen Ex. 2016
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 2 of 19)
`
`

`
`634 Crystal Growth & Desigri. Vol. 11. No. 3. .2011
`
`database
`
`*Merek Index 13th ed. (2001)
`
`Table 1. Statistics“ on Multiple Crystal Forms from "Disinlcrested”* and “Interested”: Sources
`total number of entries
`(monomorphic) %
`(polymorphs) %
`(hydrates) %
`(10,150)
`
`l0.25IIl
`
`Bernstein
`
`lsolvates) %
`
`*C‘.a1nbridge Structural Database
`(November 2008)
`
`*Bcilstcin (2009)
`
`l‘Agroehemicals
`
`456,637
`
`1l.0(]0.(J00
`
`686
`
`:iP11urn1m'opoe:'o Europa 5.8
`
`960 drug compounds
`
`70.4”
`
`99.85
`
`78
`
`36
`
`(16,035)
`
`3.5
`
`18.8
`
`41
`
`(119,107)
`
`26.1
`
`42
`
`34
`
`2.2
`
`13
`
`" Data from U. Griesser.” except for entry from Cambridge Structural Database. I’ All the rest. These entries do not necessarily sum to 100%. since
`there are many cases for which there are solvates andlor hydrates andjor polymorphs for a single compound.
`
`require an inclusive moniker. Let tautomeric polymorphism
`..
`_
`.. 27-29
`pass and be foi gotten .
`In short. as one of n1y colleagues often reminds me. “words
`have meaning" and we should exercise cai'e and avoid
`tlippancy.
`
`Propensity to Formation of Multiple Crystal Forms
`
`How common is the appearance of multiple crystal forms
`(polymorphs. solvates, and hydrates) in molecular crystals‘?
`Even neophytes would no doubt encounter the widely quoted
`assessments by two of the historically most prominent scholars
`in the field of polymorphism:
`
`It is at least this author’s opinion that eveij‘ campmmd has different
`polymorphic forms and that. in general. the number of forms known
`for each cornpotmd is proportional to the time and money spent in
`research on that compound. (italics in original)
`Walter McCrone3
`
`Probably every substance is potentially polymorphic. The only
`question is whether it is possible to adjust the external conditions in
`such a way that polymorpllism can be realized or not.
`Maria Kuhnert-Brandstéittersn
`
`Clearly such statements from two of the doyens in the field
`might lead one to expect to find polymorphs or multiple
`crystal forms for any compound. The accumulated facts
`and experience leave one a bit less sanguine. First of all. two
`extremely common compounds that have been crystallized
`repeatedly and in huge quantities have never shown any evi-
`dence of polymorphism: sucrose and naphthalene. The widely
`used analgesic ibuprofen. developed in the early l96tls has
`annual production of -15.000 tons. and until very recently
`had never exhibited any evidence of polymorphism. The
`caveat here is that the frame of reference is that the vast
`
`majority of these crystal.1izations have been carried out under
`“normal” conditions — atmospheric pressure and close t.o
`ambient temperature or up to the boiling points of various
`solvents employed. As the case of ibuprofen demonstrates. a
`wider exploration of phase space can reveal the existence of
`other polymorphic forms. A second form was obtained by
`specific recrystallization of the supercooled liquid. from which
`a protocol was developed to reproducibly obtain that form.3 I
`The structure of phase II, determined by powder diffraction
`methods, was recently reportedxll
`What is the reality‘? Statistics on crystal forms are not easy
`to determine. The choice of the database for the statistical
`
`study in essence biases the result. There are many reports
`
`(or hints) of multiple crystal forms in the primary literature
`that are not included in information recorded in databases.
`
`Claimed examples of polymorphism may turn out not to be so
`and vice versa. The results of the statistical survey may
`actually be dependent on the intent of the survey. This is not
`meant to be a criticism. but simply a statement of fact. One
`may distinguish, for instance. between :1 statistics based on
`“disinterested sources" and “interested sources". Some exam-
`
`ples from both categories appear in Table I.
`From these data, it is clearly not possible to affirm either of
`the two quotations above. The entries for “Agroche1nicals"
`and the Phczrinrzcopoezu Europa might be considered to pro-
`vide some support for these claims, but in the end in both cases
`more entries do not exhibit polymorphism than those that do,
`and in no case are more than half of the cases polymorphs.
`hydrates. or solvates.
`is informative to examine the
`it
`At the other extreme,
`statistics from an “interested“ source. Stal1ly'9 summarized
`the results on 245 compounds that had been specifically
`.vrr'eened by SSCI Aptuit. Inc. in the search for multiple crystal
`forms. Even on the basis of an intentional and concerted
`
`experimental search. 18% of the compounds did not exhibit
`multiple crystal forms, although some of those did exhibit
`noncrystalline forms. It might be pointed out here that these
`data, and those of TlSCl’1l€1'.:M apparently suggest that salts
`tend to form more hydrates than neutral compounds. while
`polymorphism was more common for nonsalts than for
`salts. 111 any event, none of these statistical analyses truly
`fulfill the prophecies of the two “giants” ofthe field,“ and it
`is clear that even when actively sought by some of the most
`experienced practitioners in the field. true polymorphs
`(as defined above) are found in barely 50% of the com-
`pounds studied.
`
`The Experimental Seareh for Polymorphs
`
`Crystallization from solution is one of the lirst laboratory
`skills that chemists acquire, and applying variations to the
`conventional methods has been the traditional strategy in the
`Search for polymorphs. However, the increasing recognition
`of the desire, indeed the need, to explore crystal form space as
`thoroughly as possible in the search for improved materials
`and,’ or because of intellectual property considerations has led
`to the development of a dazzling panoply of new techniques
`for growing crystals. with the aim of obtaining new forms of
`crystals. A number of excellent reviews of these techniques
`have appeared reoently. and offer the potential for developing
`exciting strategies for searching for new forms.” It is illus-
`trative to list some of the traditional methods for growing
`crystals along with those that have been developed recently.
`
`Janssen Ex. 2016
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 3 of 19)
`
`

`
`Perspective
`
`Crystal Growth & Design. Vol’. II. No. 3. 2011
`
`635
`
`
`
`e
`
`313 333 353 373 393 413
`
`xsoivent
`
`0.1
`
`0.01
`
`T/K
`
`Figure 2. Demonstration ofcontrol over polymorphic Form obtained
`through crystallization via superc1'iticr1lCO3ai1 ::11‘lllC'cl1‘lC€t‘ quinazo-
`line derivative. Forms I-III were known prior to the experiment.
`Form X was discovered in the course of the experiment ( from ref 42).
`Form 11 was most prized.
`
`niicroporous membranes, sonocrystallizations. light field induced
`control of cry stallizationfm
`
`In addition. there have been sotne notable developments
`that warrant attention and further development and exploita-
`tion. Many of these are techniques that were designed to gain
`control over the crystallization process. It is important to
`establish that control even in the exploratory stage for crystal
`forms, since it can greatly aid in scale up and can play a role
`in preventing the appearance of undesired forms or the dis-
`appearance of the desired form somewhere down line, even
`beyond launch of a product.
`One technique that promises a fairly high degree of control
`over the polymorphic form obtained is crystallization by super-
`critical fluids ~ in particular supercritical C 03. For instance,
`Bouchard et al.“ demonstrated that they could obtain
`)5’-glycine exclusively by control of the conditions of super-
`critical CO3 crystallization. One potential advantage of this
`technique is that it is engineering based, offering considerable
`control over most crystallization conditions.
`This degree of control is demonstrated by the sample of
`an anticancer quinazoline derivative studied by Kordikowski
`and York.“ The compound exhibited five polymorphs. of
`which the difficult to obtain metastable Form 1] was desired.
`
`The variation of the conditions revealed a small processing
`window for Form II that was achievable with the added
`
`benefit of particle size control. Also in the course of determin-
`ing the conditions for obtaining the various polymorphs. a
`new metastable Form X was discovered. which was not
`obtained by conventional crystallizations.
`
`Which Polymorph Will We Obtain? Some Comments about
`Z’, and :1 “Crystal on the Way”
`
`Many chemical crystallographers have been fascinated by
`the phenomenon of Z’ > 1 - more than one molecule in the
`asymmetric unit - and the explanations for its appearance
`and its meaning have generated considerable debate, with
`Steed and Desiraju among the principal protagonists.43‘44
`Desiraju has contended that all the reasons suggested for the
`existence of structures with Z’ > 1 can be attributed to meta-
`
`stable forms obtained under kinetic (i.e.. nonequilibrium)
`
`Table 2. Percentage ol'For1us from Polyrnorph Screening"
`
`all compounds
`[count (°/cl]
`
`salts
`[count (%)]
`
`nonsalts
`[count ("/5)]
`
`multiple forms”
`multiple crystalline forms‘
`polymorphs"
`hydrates
`solvates
`noncrystalline
`total compounds
`
`220 (89)
`200 (82)
`118 (43)
`94 (38)
`78 (33)
`llfl (48)
`245
`
`so (91)
`77 (81)
`37 (39)
`46 (48)
`34 (36)
`5] (54)
`95
`
`I16 (91)
`I05 (82)
`71 (55)
`38 (30)
`36 (28)
`55 (43)
`128
`
`" Reproduced from ref 19. with permission. Copy right 2007 American
`Chemical Society. "Crystalline polymorph. hydrate. and solvates plus
`noncrystalline forms. "Crystalline polymorphs. hydrates. and solvates.
`“Crystalline polymorphs.
`
`E 1
`
`”!»-i_;t)u(rimr, fanny
`
`Figure 1. Time frames for various crystallization techniques (from
`ref 36. with permission. Copyright 2008 Elsevier).
`
`Traditional solution crystallizations generally allow for varia-
`tion in the following parameters:
`
`Solution crystallization. solvent(s) (including solvent mixtures).
`temperature, stirring. cooling rate. seeding, antisolvent. slurrying
`
`Llanas and Goodmatfs useful chart summarizes the time
`
`scales of solution crystallization techniques as well as the
`qualitative relationship between the stability of the form
`obtained and the time required to obtain those forms. Many
`pharmaceutical companies seek to identify a.nd then develop
`the most stable form of an active pharmaceutical ingredient
`(API).37 but there are sometimes reasons, for instance materi-
`al handling properties or intellectual property issues.
`to
`identify and occasionally use less stable forms. provided they
`can be stabilized to prevent conversion to the stable foi‘1n.33'39
`Other conventional (although not as widely used) methods
`of crystallization (with typical variable parameters) include
`the following:
`
`Sublimation (pressure. temperature gradient). crystallization from
`melt (temperature program). freeze drying. spray drying
`
`Some (but by no means all) of the additional methods of
`crystallization that have been added to the armory ofcrystalliza-
`tion tools?“
`
`High throughput (solution) crystallizations. confinement crystal-
`lizations (capillary. Contact line. nano). electroclieniical crystalliza-
`tjons. gel crystallizations. vapor dilfusion ciystalliiations. use ol‘(“tailor
`made”) additives. use of templates (polymers. inert surfaces. etc),
`mechanical grinding (cocrystals), solvent drop grinding (cocrystals).
`
`Janssen Ex. 2016
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 4 of 19)
`
`

`
`636 Crystal Growth & Design. V0]. 11. No. 3. .2011
`
`conditions. The implication of this postulate is that if a
`structure. or a number of structures, are determined to have
`Z’ > 1 they must be metastable forms and there exists. or will
`exist (if it has not yet been prepared) a more stable form with
`Z’ = I or lower. Such postulates indeed generate discussion,
`new theories. and in the best cases definitive experiments (I am
`inclined to include “computer experiments“ in this category).
`However. as the adage goes. “Theory guides; experiment
`decides", and until there is a body of experimental evidence
`to prove that the higher Z’ structures of a polymorphic system
`are all (or almost always all - we should always allow for the
`possibility of exceptions) of higher energy than the Z’ = I
`structures it seems prudent to reserve judgment on this issue.
`Brock and Duncan“ in their study of alcohols with Z’ > 1
`concluded that packing (i.e.. space filling) considerations
`better account for the Z’ > 1 than energetic or crystallization
`conditions. Moreover, we exhaustively studied benzidinelfi
`and found only four polymorphs ofwith Z’ = 4.5. 3. 1.5, and
`4.5. In addition to our own numerous crystallizations (many
`in the attempts to prepare co-crystals), the system has been
`studied intensively by hot stage methods by two previous
`groups4""3 with no evidence for an additional lower energy
`form (with or without Z’ = 1). Similarly, for cholesterol,
`another “classic” molecule first studied by Bernal. two 1nono-
`clinic Pl forms have been reported. one with 16 molecules in
`the asymmetric 1.ll"lll1.49 and the second with 8 molecules in the
`symmetric unit.” However. we are certainly aware of the
`caveat noted by Revel and Ricard, “But not to be able to find
`something is no proof ofits nonexistence."5 ’ While the failure
`to obtain a form with Z’ = l
`is certainly no proof of its
`nonexistence. until such a form is found. this case along with
`numerous other examplesm argues against the kinetic and
`thermodynamic rationale for the preference for Z’ = 1.
`It has been argued that crystal structures with high Z’ > 1
`should be considered as a “crystal on the vvay".44 While
`intuitively this notion may have some appeal. it defies the
`very essence of a crystal structure. Upon determining a crystal
`structure. we have become accustomed to producing an
`ORTEP diagram of “the molecule" and a packing diagram
`to portray “the crystal structure". For many current practi-
`tioners of crystallography and users of crystallographic data,
`the routine generation of these diagrams apparently belies the
`fact they represent. the space average and time average of
`~10” unit cells, all of which must be essentially identical - or
`nearly so on the atomic scale - in order for the diffraction
`experiment to be at all viable. That is. it is precisely a crnvral
`.v.rrucmre from which we generate those pictures. It is not on
`the way to anything; it is already a crystal structure. Other-
`wise, we could never have done the experiment to determine
`that structure. However. if the intention is to describe such
`
`a crystal structure with Z’ > I as one structure at a local
`minimum on the multidimensional potential energy surface
`on the reaction coordinate toward a proposed (or supposed)
`structure with Z’ = 1 (presumably. according to this model at
`the global minimum). then it might be possible to consider this
`concept as a working hypothesis. Such a notion harkens back
`to the classic studies of Biirgi and Dunitz on reaction path-
`ways from structures in the CSDSQ There may be other crystal
`structures. but they need not per force be on the same reaction
`coordinate along which the Z’ > 1 structure crystallized. This
`is not to deny that it could be the case, but it must be proren
`exprarinzeimzliy before the concept of "a crystal on the way”.
`even in this limited context. can be seriously considered.
`
`Bernstein
`
`The Crucial Role of Nucleation
`
`The process of crystallization is generally considered to
`involve two steps - nucleation. followed by crystal growth.
`Scanning probe microscopies have provided a great deal of
`insight and understanding into the structural and kinetic
`aspects of the second step. Of course. once the growth process
`has begun the structure oft.he crystal form has essentially been
`determined. In most cases. that will determine which poly-
`morph results from the process. However. if the first form is
`indeed metastable, there may be a subsequent change to a
`mo1'e stable form even during the crystallization process_.
`either as a solid -' solid transition, or via a solvent mediated
`transformation. for example. as observed in the case of
`be1izamide.53‘5" Upon cooling an aqueous solution, this com-
`pound. arguably the first polymorphic molecular compound.
`initially studied by Liebig and ‘Wohler.55 a metastable form
`appears first and subsequently transforms in situ to the stable
`form. both with Z’ > 1.
`Recent work has shown that we may have to reconsider the
`simple. and perhaps naive, notion that once a crystal form
`nucleates that form will continue to grow. The concept was at
`least part of the basis for rationalizing the existence of
`concomitant po1ymorphs:53‘56 they nucleate essentially simul-
`taneously and the growth rates are sufficiently similar so that
`they coexist in the time frame of the crystallization. In the case
`ofbenzamide. the transformation to the stable form continues
`at the expense of the metastable form. but there is a period of
`time when both may be observed simultaneously — hence
`concomitant. Some recent molmular dynamics calculations,
`albeit 011 simpler systems, suggested the possibility of the
`cross-nucleation of a metastable polymorph on the stable
`57
`polymorph Moreover, Yu and co-workers have demon-
`strated that such cross-nucleation can occur in the qui11tes-
`sential polymorphic ROY syrstemsmgl and has also shown that
`for 1.-glutamic acid the polymorph that nucleated in the early
`stages of crystallization was capable of nucleating another.
`faster-growing polymorph.(”’ He concluded that the selective
`crystallization of a polymorph depends not only on the initial
`nucleation but also on the cross-nucleation between poly-
`morphs and the relative growth rates of polymorphs.
`Clearly. the understanding and control of nucleation is one
`of the most challenging aspects of current polymorphism
`research. and recently there has been increasing activity and
`some impressive experiments in the efforts to develop and
`control the nucleation step. which of course is the ultimate
`means of controlling which polymorph is obtained. For
`example. Meyerson and colleagues’‘’ have used nonphoto-
`chemical laser-induced nucleation on aqueous solutions of
`urea, and the technique was recently applied for the selective
`crystallization of 0t- and }'-glycine.(’3
`The debate over the nature of the nucleation of glycine
`demonstrates some of the questions that need to be resolved.“
`One question intimately related to the nucleation is whether
`the dominant form of glycine in solution is the cyclic Ritltlj
`dimer (analogous to the R§(8) cyclic dimers in benzoic acid) or
`monomers — in other words. what is the basic synthon
`(tecton). In the 1990s. evidence argued in favor of dimers in
`glycine solution .644” However. Yu et al. have recently carried
`out freezing point depression and diffusion measurements of
`supersaturated aqueous solutions of glycine, and both were
`consistent with the fact that the solutions are mainly (but not
`exclusively) monomeric glycine.“ Moreover. the fact that the
`diffusion of glycine does not slow as the solution ages contradicts
`
`Janssen Ex. 2016
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`IPR2015-01030
`
`(Page 5 of 19)
`
`

`
`Perspective
`
`Crystal Growth & Design. Vol. 11 , No. 3. 2011
`
`637
`
`previous models for th