From Molecufes to Crystal Engineering}
`
`Chemical Reviews. 2001. Vol. 101. No. B
`
`1647
`
`*
`
`k
`
`_,
`
`_‘:
`.'
`._
`«£339?
`
`In
`
`§._
`.$s_=_-*-_*__~.‘E-..
`
`lllustrationof the crystal packing of the
`Figure 18.
`hydrogen-bonded ZD networks in {(BT('.'3 )(Nl l2[Cl*IzPhlz'l2|.
`The benzyl moieties preciude intertiigitation and facilitate
`reversihie sorption of aromatic molecules.
`
`
`
`‘
`-
`-in A.
`Figure 19. Illustration of the crystal packing in [{B'I'C2'J-
`(NH3(CHgPh]+J3], a prototypal example of the network
`structures formed by BTC dianions and primary am-
`monium cations.
`
`structure is illustrated in Figure 18, and as should
`be clear, there is no interdigitation oi‘ bcnzyl groups.
`The guest molecules interact with walls of the chan-
`nels only, and the asymmetric unit is unusual: 3:3: 1
`for hostzguestsolvent. In the presence of primary
`ammonium cations. similar structures are obtained
`
`they are more appropriately termed bilayer
`but
`architectures since there are alternating hydrophobic
`and hydrophilic regions. A typical structure is il-
`lustrated in Figure 19. Similar structures are ob-
`tained for both alkyl— and benzylammonium cations.
`It might be reasonable to describe such structures
`as being Cytuminlt-‘.I'.i(: since there is a resemblance to
`the type of supramolecular synthons that exist in
`phosphnlipid membranes and in the solid phases ol"
`surfactants. The ancillary organic groups orient in
`the same direction and interdigitate with adjacent
`layers to generate hydrophobic regions. The hydro-
`philic faces of adjacent bilayers also face one another
`and can incorporate water molecules. The thickness
`of hydrophilic layers range.-.: from 3.2 to 3.4 A, while
`the thickness of interdigitated layers increases with
`the size of the organic group.
`Stoichiornetry 1:3 (H3TMA:amine). In principle,
`1:3 stoichiometry offers the opportunity to generate
`honeycomb networks. As revealed by Figure 20a.
`motif A or B should be capable of propagating the
`trimesate anion into a honeycomb structure. Figure
`20b reveals that the crystal structure of [TM.=i.]—
`[tiic:ylol*n-zxylammonium]3 exists as the arititzipated
`honeycomb array."'9” The cyclohexyl moieties, which
`are omitted for the sake of clarity. effectively prevent
`interpenetration by tapping the 13 A cavities that
`are present within the honeycomb structure. If the
`
`<_:~
`
`-cu:«::~
`''1..'-!...i''L..-!'.12"'i.'.-g-.‘..r""
`
`
`
`
`
`
`
`
`
`(=1)
`
`(bl
`
`Figure 20. Modular honeycomb network sustained by
`BTC lrianiuns and secondary alkylamrnonium cations:
`(a)
`schematic representation of the hydrogeobonding pattern
`and (b) spacefilling illustration of the crystal structure of
`i(E-TC?) (NH3lCsH i zl+l2l-
`
`solvent is changed, a honeycomb network based upon
`the other supramolecular synthon is generated.’-"33
`and it has been reported that this form of the
`honeycomb network will self-assemble at the air-
`water interface. The modular nature of this structure
`
`permits replacement of the cyclohexyl moieties by
`other moieties. in this context. alkyl groups that are
`less sterically demanrling (e.g..
`I2—alkyl) have also
`been incorporated into the motif in Figure 20a.
`lnterpenetration occurs in these structures.
`Cuaniclinium Sulfonates. A series of related
`structures that are based upon two-dimensional
`layer.-; resulting from hydrogen bonding of the trigo-
`nal guanidinium cation, C(NHg)3*. and organic sul-
`fonate ions R303" has been extensively studied by
`the Ward group17B-233-24“-395‘3°2 (Scheme 14}.
`lnterdigitation of the organic substituent of the
`sulfonate ions on adjacent layers and ionic hydrogen-
`bonding predictably leads to a broad series of laminar
`architectures.
`it should be noted that there are
`several key dilTerences between guanidinium sul-
`fonates and alkylammonium trimesates.
`(1) There
`exists only one ancillary organic functional group per
`sulfonate ion compared to up to two ancillary Func-
`tional groups per ammonium cation. (2) The anion
`is lunctionalizecl rather than the cation. (3) In one
`sense. the alkylsullonates are more versatile since
`they can exhibit architectural (i.e., supramolecular)
`isornerism so as to generate either bilayer or clay-
`llke architectures. 'l‘o generate a clziyyike architec-
`ture. organic groups must orient above and below
`each layer as illustrated by Scheme 15. The steric
`demands of the organic group appear to determine
`whether they orient in the same direction {i.e., a
`bilaycr structure) or alternate above and below the
`layer (i.e.. a clay-like structure).
`
`(I. 3D Networks
`
`1. Self-Assembled Hydrogen-Bonded Diamondoid
`Networks
`
`A report by Ermer“ on the structural characteriza-
`tion of adamantane—1.3,5.7—tetracarboxyIic acid and
`its implications represented a watershed for crystal
`engineering. Ermer's study was followed by a [lorry
`of activity into design from first principles of both
`organic diamondoid networks and metal—organic
`diamondoid coordination polymers. The carboxylic
`
`Janssen Ex. 2043
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 19 of 30)
`
`

`
`1543 ichernica: Reviews. 2001. Vol. 1411. No. 5
`
`rulaulton and Zaworulku
`
`(a}
`Scheme 14. Schematic Illustration of the Key Structural Features of Guanidiniurn Sulfonates:
`Hexagonal Channels that Farm 21) Honeycomb Networks and (b) Two Extended Structures (2D and 3D)
`that Can Result from the Self-assembly of Guandinium Cations with Disulfonates Anions
`
`
`
`(a)
`
`(bl
`
`Scheme 15. Schematic Illustrating Two Possible Modes of [nterdigitation for Guanidinium Sulfnnates:
`Clay-like. Induced by Sterically Demanding Functional Groups and (ti) Bilayer Architecture, Typically
`Dbservorl l‘m- Smaller Functional Groups
`
`(3)
`
`
`
`('3)
`
`acid groups ol" adamantanc-1,3,5.7—tctracarl:loxy1ic
`acid are tetrahedrally oriented. It is therefore un-
`surprising that they self-assemble via the hydrogen-
`bonded carboxylic dimer supramolecular synthon to
`afford an infinite diamondoid network. Each network
`
`possesses cavities that could accommodate a large
`mu hly spherical guest, or guest aggregate, of roughly
`12
`in cliamctcr. However, those L:nvitic-es; are filled
`
`by five independent networks that interpenetratc in
`such a way that the crystal structure is densely
`packed, and consequently. guest inclusion is pre-
`cluded. As subsequent studies have revealed, inter-
`penetration is a widespread phenomenon in diamon-
`doid networks and occurs in many other organic and
`rnetal—organic structures that would otherwise have
`large cavities or channels. An intcrpenctratcd dia-
`mondold architecture is also exhibited by methane-
`t.etra::u:etir: acid, for which the Cavities generated are
`approximately 10 A in diameter?” As would be
`expected, methanetetraacetic acid exhibits a lower
`degree of interpenetration: 3-Fold. 2.fi—dimethylidine-
`adamantane-I,3,5,7—tetracarboxylic acid also forms
`a hydrogenlionded diarnondoid structure, but
`it
`exhibits a much lower degree of interpenetration
`
`than its unsubstituted precursor. The 2-fold "double
`diarnondoid" architecture is not as densely packed,
`and it can therefore act as a host and enclathrate
`
`guest molecules.“
`Wuest demonstrated that the pyridone moiety also
`generates a hydrogen-bonded supramolecular syn-
`thon that
`is suitable For building extended at-
`rays.”9'3°3~3“" Remarkably. mothanctctra{6-phenyl-
`ethynyJ—2-pyriclone) exhibits a diamondoid network.
`7-fold interpenetration, and cavities large enough to
`enclathrate butyric or valeric acid.“ Wuest intro-
`duced the concept of "tectons" to describe molecules
`that inherently possess the molecular structure and
`intermolecular recognition features to predictably
`sell"-assemble into crystalline networks. This study
`was followed by studies that demonstrated that there
`are several other examples of diamondoid networks
`that. can he sustained by tho pyridone rI1r>icty.3°5-3”“
`
`2. Modular Self-Assembly of Hydrogen-Banded
`Diamandoid Networks
`
`As discussed earlier, modular selilassembly relies
`upon two molecular components that are not indi-
`vidually capable of Self-assembly and can be invoked
`
`Janssen Ex. 2043
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 20 of 30)
`
`

`
`From Molecules to Crystal Engineering
`
`Chemical Reviews, 2001. Vol. 101. No. 6 1649
`
`
`
`[.\-ln(u3-OHJ{CU)3l4: a cubane like cluster
`Figure 21.
`possessing perfect Ta symmetry that represents a prom-
`typal example of a tetrahedral hydrogen-bond donor.
`
`Scheme 16. Schematic Illustration of the Two
`Types or Diamoridoid Architecture: (a) Cubic and
`(b) Hexagonal
`
`Tihr
`
`l
`
`i
`
`[
`
`lb)
`
`to understand coordination polymers or multiple-
`component hydrogen-bonded networks.“ There are
`significant differences between the types of tetra-
`hedral moiety that can sustain networks that have
`been self-assembled from a single component vs
`networks that have been self-assembled from mul-
`tiple components. The most fundamental difference
`between the two types of structure is that
`the
`tetrahedral component that sustains single-compo-
`nent self-assembled architectures would not ordi-
`narily be able to sustain modular architectures and
`vice versa. In the case of the former, the tetrahedral
`moieties must be sell‘-complementary and there is
`only one component necessary for sell‘-assembly to
`occur. This means. for example. that both hydrogen-
`bond donors and hydrogen—bond acceptors must be
`present in the same molecule. In the case of the
`latter. the tetrahedral node must be either an accep-
`tor or a donor of hydrogen bonds and the linker or
`spacer must be complementary. Both components are
`necessary. and there must be a 1:2 ratio in order for
`the diamondoid architecture to self-assemble. Scheme
`16 illustrates the difference between the two types
`of diamondoicl networks.
`An example of a node that is suitable for modular
`sell'—assembly is the cubane cluster
`[Mn(u3-OH)-
`[CO);,]4,3°7 which possesses perfect Td symmetry and
`has four strong hydrogen-bond donors and no strong
`hydrogen-bond acceptors (Figure 21). This tetrahe-
`dral hydrogen-bond donor forms diamoncloid cocrys—
`tals with a wide range of obvious and. in some cases.
`not so obvious spacer molecules. A "not so obvious"
`
`
`
`Illustration of the crystal structure and
`Figure 22.
`dlarnondoid cavity generated in [.‘t/ln(u3rDH)(C0lzl4‘Zbenzene.
`
`structure is that formed when {Mn(.u;.-OH)(CO}3].; is
`cocrystallized with benzenefms A 2-fold diamondoid
`structure is sustained by OI-l---rr hydrogen bonds, and
`the tetrahedral symmetry of the node is observed in
`the crystallographic sense since [Mn (u3-OH)(C()}3]4-
`Zbenzene crystallizes in the cubic space group Pn-
`3m with z = 2. As revealed by Figure 22. which
`illustrates an adamantoid portion of the structure, a
`large cavity is generated and this facilitates inter-
`penetration of a second diamondoid network.
`The use of transition metals or transition—metal
`clusters to act as nodes for the modular selliassembly
`of diarnondoid networks that are sustained by coor-
`dinate covalent bonds is aiso well established. Such
`architectures are of more than aesthetic appeal. and
`they have resulted in a class of compounds with
`interesting bulk and functional properties. Metal-
`organic diarnondoid structures in which the spacer
`moiety has no center of inversion are predisposed to
`generate polar networks since there would not be an
`inherent center of inversion. Pyridine-4-carboxylic
`acid.
`isonicotinic acid,
`is such a ligand. and his-
`(isonicotinato)zinc exists as a 3-fold diamondoid
`structure that is both thermally stable and inherently
`polar.” lt exhibits SHC activity that is three times
`higher than the commercially relevant NLO material
`KDp_3o9.3io
`
`There are also supramolecular synthons that do not
`rely upon hydrogen bonds.
`in this context.
`l\.‘*--Br
`interactions were exploited to propagate a diamon-
`doid network in the cocrystal
`formed by carbon
`tetrahromide and hexarnethylenetetraaamine. This
`structure also represents a different but equally
`effective form of the modular approach:
`two tetra-
`hedral nodes with one possessing donor functionality
`and the other acceptor functionality only. The struc-
`ture of the cocrystai formed by carbon tetrabromide
`and hexamethylenetetraaamine exhibits 2-fold in-
`terpenetration and does not enclathrate solvent or
`guest?"
`
`3. Other 30 Hydrogen-Bondedllvelwarlcs
`
`Although there are many examples of organic
`crystals that can be defined as 3D networks. few of
`
`Janssen Ex. 2043
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`lPR2015-01030
`
`(Page 21 of 30)
`
`

`
`155!) Chemical Reviews. 2001, Vol. 101, No. 6
`
`Moulinn and Zaworolko
`
`Scheme 17. Schematic Illustrating Five Supramolecular Synthons Possible for Pyramellitate Dianions
`0.
`0
`.
`0'
`O
`-_
`H“
`
`GICi
`
`o
`H
`.0"
`
`'
`
`‘o
`H
`o
`
`.0
`
`'
`
`0’
`0'
`
`D‘
`
`H;
`
`*0"
`
`‘o
`o
`
`HA3
`
`o’
`T
`To
`0'
`
`__
`'3~._
`
`(3)
`
`(b)
`
`(C)
`1
`
`2‘
`
`.
`
`ah
`
`‘I .
`
`‘Tn
`
`‘ ‘-
`
`‘bx
`
`-~-A A'-0 G--A—<:%% A...
`...Aj{j{_,,,....;,. z=2;,;....A u<:\:|_Ay
`._
`[L
`A
`"pa
`...o g...A_&'—;_A..- n...
`...-|J—£$;'-—|:;--;\~ A__a n_.
`
`__A
`
`.‘
`
`uh
`
`.
`
`".
`
`D.
`
`A.
`
`5‘
`
`‘-
`
`A.‘
`
`(<1)
`
`(8)
`
`them are predictable or even rational in the same
`sense as diamondoid networks. Trimesic acid, 1-13-
`TMA, is an interesting exception and was discussed
`earlier in the context of 2D structures. H3Tl\/IA
`represents a prototypal molecule in the context of
`hydrogen bonding and generates extended structures
`when pure, partially deprotonated, in coordination
`polymers or in cocrystals. Anionic derivatives of
`H3'l'MA sell"-assemble into honeycomb grids via
`O—ll"'O hydrogen bondS.““”"z32-313-315 Pyrumellitic
`acid, 1,2,4,5-benzenetetracarboxylic acid.
`I-l4PMA,
`has been less widely explored than I-I3Tl\/IA. it has
`been utilized as a ligand in coordination polymer
`networks?“ and very few organic structures contain-
`ing l*l4Pl\/IA or its derivatives are known. We anticl-
`pated that doubly deprotonated H.;P?i/IA would self-
`assemble via dicarboxylate hydrogen bonds to form
`OD (two intramolecular hydrogen bonds). 1D (one
`intramolecular and one interrntllecular hydrogen
`bond). or BDIISD (two intermolecular hydrogen bonds)
`networks (Scheme 17). HzPMA"’ anions exhibit all
`Four of these suprarnolecular isomers depending upon
`the polymorph or the counterion.3“‘ The 3D structure
`occurs because HEPZVIA3’ moieties orient in such a
`manner that they form hydrogen bonds to the next
`layer. The network can be described as a framework
`built from square building blocks that alternate
`parallel and perpendicular with respect to one an-
`other. This network can therefore be regarded as an
`organic analogue of Nb0.“
`It should be obvious that all 2D networks must. in
`the absence of solvent or intercalated guest. engage
`in noncovaient interactions with the layers above and
`below. The possibility of exploiting a combination of
`molecular recognition modes represents a particu-
`larly attractive approach to control crystal packing
`since it places no restrictions in terms of the type of
`chemical components that can be rationally incorpo-
`rated into crystalline phases. In this context, Fowler
`and Lauher‘9'3‘5‘32° illustrated how it is possible to
`use hydrogen hondirtg to control interactions between
`ID networks. or a-networks. and to thereby yield
`
`predictable 2!.) networks or (1-networks. If the sheets
`are designed to he selficornplementary in the third
`dimension, then a predictable 3D network or y-net-
`work can result. Fowler and Lauher demonstrated
`
`not only that such a strategy is viable, but that it
`can offer a degree of control over stacking of layers
`such that the interlayer components are positioned
`within the limits of the topochemical principle- They
`were thereby able to i::fl"oct solid-state reactions upon
`appropriate perturbation. When coupled with other
`advances in this context, including recent reports1""9
`that demonstrate how discrete aggregates may also
`afford components that are positioned within the
`topochemical boundaries,
`it should be clear that
`crystal engineering involving multiple types of mo-
`lecular recognition offers significant implications for
`solid-state synthesis and solvent-free, green chem-
`istry.
`
`IV. Supramoiecuiar isomerism and Polymorphism
`
`The existence of supramolecular isomerisrn might
`be seen as a problem from a design perspective since
`it necessarily implies that there will be superstruc-
`tural diversity for a given molecular building block.
`However. there is another way to look at this n1atl:or.
`It is also possible to view supramolccular isomerism
`as an opportunity to gain a better fundaniental
`understanding of the factors that influence crystal
`nucleation and growth. Such a linkage can be justi-
`fied as follows. {1} If one invokes the concept of
`supramolecular isomerism, then it should become
`apparent that it represents a significant limitation
`on the number of possible super.strui:turcs {i.e.,
`discrete structures or 1D. 2D. or SD networks) that
`can occur for a given molecular building block.
`Therefore. one can invoke a study on supramolecular
`isomerism or polymorphism with the assumption
`that selfiassembly means that there will only he a
`finite number of architectures that are feasible for a
`
`given molecular species. This assumption will be
`based upon crystals being the result of directional
`
`Janssen Ex. 2043
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 22 of 30)
`
`

`
`From Molecules to Crystal Engineering
`
`Scheme 18. Schematic Illustration of the Six
`Supramolecular Isomers Reported for T-Shaped
`Nodes Linked by Linear Bifunctiona] Exodentate
`Ligands:
`(al ID Ladder. (bl 3D Lincoln Logs, (C)
`ZD Herringbone. ((1) 2D Bilayer, (e) ED Brick Wall.
`and
`3D Frame
`
`.l...
`
`-
`
`l-—-ll--ll--I
`
`l--_ll--ll--ll
`-
`—-T
`l-
`-l
`—-I
`F -l
`J.
`:
`i-——-I
`l-
`-I
`
`l----I
`(3)
`
`.L
`
`.L
`
`>‘
`J’
`3'
`V‘
`‘¢““4"1
`A
`A
`‘*3-y‘
`-r:_l-L—_'ili.?_-—*::l.~,-
`7
`V
`T
`T
`4:
`'
`.4:
`4
`‘C
`V
`"
`'
`"
`'1
`'
`X.
`>.
`1)!‘
`i K “>,‘:.
`Y
`V _>*
`
`lb)
`
`(0)
`
`rr ‘*+‘—*—t'r’*
`‘i:'-fin-ll.
`1* 4- -r -‘-'-r-‘‘* ‘P
`1-‘ by ‘ll
`rr
`* -r -'-
`—_
`‘
`..
`4-
`-'-
`_l:.H'
`-
`4.
`I
`.i.c. : .34. TLT_TAT
`(d)
`ll:J
`
`_.'L_
`_
`_l-,-J.“qT.‘.-'5uIF-
`qt’
`'
`‘as:-T ~-tr
`,
`~—L,.b—_:T_1-5“.-.~_—T
`fr]
`
`supramolccular synthons, the fundamental precept
`of crystal engineering. (2) The ability of the crystal
`engineer to design a molecular building block that is
`predisposed toward the formation of supramolecular
`isomers provides an ideal opportunity for design of
`supramolecular isomers and, perhaps more impor-
`tantly, for learning how to control supramolecular
`isomers. At the very least.
`it will be possible to
`develop "recipes" that invoke crystallization concil-
`tions, templates. andfor solvents to favor or disfavor
`a particular supramolecular isomer.“-333 {3} The
`concepts of self-assembly, crystal engineering. and
`networking provide clear implications for gaining a
`better fundamental understanding of polymorphism
`since polymorphism can be regarded as a subset of
`supramolecuiar isomerism. {4} It should be clear that
`to gain a better understanding of supramolecular
`isomerism and polymorphism. it is a requirement
`that full structural characterization of compounds be
`conducted. This means an in-depth analysis of crystal
`packing and intermolecular contacts as well as
`measurement of physical properties.
`To illustrate the linkage between crystal engineer-
`ing, supramolecular isomerism, and polymorphism,
`we shall consider three types of supramolecular
`isomorism and Llemoristrate how analogies can be
`readily drawn between coordination polymers and
`organic networks.
`
`A. Structural Supramolecular lsomerism
`
`That structural supramolccular isomerism can
`have profound implications for structure and proper-
`ties is exemplified by the range of structures that has
`thus Far been observed in coordination polymers that
`are generated by one of the simplest building blocks
`and stoichiometries:
`l:1.5 stoichiometry, mer—metal.
`and linear spacer ligand. These building blocks can
`be regarded as being based upon self-assembly of
`T—shaped nodes. There already exists a surprisingly
`diverse range (ifstructl1J't'.S that have been iiliserved
`in this context. Scheme 18 illustrates the supremo-
`
`Chemfral Reviews. 2001, Vol. 101. No. 6
`
`1651
`
`lecular isomers that have thus far been observed:
`ladcler9‘v”‘* (A), brick wall“ (B), 3D frame or “Lincoln
`Logs"235"’-37 (C), hilayerm (D). herringbone?” {E),73-Z“-“Z
`and another version of a 3D frame“ (F). Three of
`the isomers A,”3 D?” and Fm have been observed
`for the same asymmetric unit for metal — Co{l\lOr;]2
`and ligand -— bipy. and the other three have been
`seen in similar compuuntls which use bipy or ex-
`tended analogues as “spacer ligands".
`The following points should be noted about such
`structures. {1} These compounds are not true poly-
`morphs since guest or solvent molecules are present
`in the lattice. However, neither are they solvates in
`the conventional sense. (2) The diversity of network
`structures and hence bulk properties is remarkable.
`(3) None of those arclutectures occurs naturally in
`minerals. [-£1} The network structures themselves are
`entirely predictable based upon simple structural
`considerations. (5) Some of these structures can occur
`from the same building blocks under almost identical
`crystallization conditions.
`It is possible to draw direct analogies with poly-
`morphism in organic crystals. For example, a similar
`approach based upon networks can be used to ana-
`lyze the packing in organic Compounds. A recent
`paper highlighted this situation in the context of
`2—amino—5—nitropyrimidine, a compound that exhibits
`three readily available polymorphs, all of which have
`distinct hyrlrogen—bonded networks?“ Etter’s study
`concerning carboxylic acids“ and how they can sell"-
`asscrnblc to form either head-to-tall chains or con-
`
`trosymmetrlc dimers also illustrates how one can
`rationalize polymorphism based upon supramolecular
`isomers and networks.
`
`In summary, it seems likely that use of appropriate
`templates or guest molecules facilitates recipes that
`can be used to reliably generate all supramolecular
`isomers that are possible for a given node. Therefore,
`one might assert that there are a finite number of
`superstructures possible for a given molecular moiety
`and that it will eventually be possible to determine
`the crystallization conditions under which each one
`will occur.
`
`B. Conformational Supramolecular lsomerism
`
`Flexibility in ligands can lead to subtle or dramatic
`changes in architecture. For example, 1.2-bis[py-
`ridyllethane. bipy—eta, can readily adapt gauche or
`a11tz'—conlormatiuns. In the case of lcolbipy-eta)”-
`(NO3}g|,,. which contains a T-shaped node. infinite
`molecular ladders which contain six molecules of
`
`chloroform per cavity exist as the most commonly
`encountered architecture {Figure 23a).‘35 In such a
`situation. all "spacer ligands" are necessarily anti’.
`However. under certain crystallization conditions
`(e.g., solvent MeCN or dioxane). a bilayer architec-
`ture is obtained with two an.li- and one gauchevspacer
`ligand per metal atom (Figure 23b).
`The bilayer architecture can contain solvent mol-
`ecules such as MeCI\' or can collapse on itself in the
`absence of solvent.“ This more subtle form of su-
`
`pramolecular isomerism occurs if [Co(blpy-eta}1_5—
`(NO3)g],, is crystallized in the absence of 21 suitable
`guest or solvent.” Figure 24 reveals how [(io(hipy—
`
`Janssen Ex. 2043
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 23 of 30)
`
`

`
`1ii52 Chemical Reviews, 2001, Vol. 101. No. 6
`
`Moulion and Zaworotko
`
`
`
`
`
`illustrations of two crystal structures of the
`Figure 25.
`square grid coordination polymer [Ni(bipy-eta]g[N(]-_;]2|:
`(a)
`with included guest molecules and (hi
`intcrpenetraied
`networks.
`
`cavities are generated within a network. A thorough
`review of this subject in the context of coordination
`polymers was recently published by Batten and
`Robson.“ The existence oi" interpenetration has been
`regarded as a factor that mitigates strongly against
`the generation of stable open framework sl;I'uCt11I'c3,
`llowever. it is becoming clear that appropriate use
`of templates can afford either open framework or
`interpenetrated structures for the same network.
`This is exemplified by the prototypal diamondoid and
`square grid networks Cd(CN)g and M(bipy);-,Xg. Both
`of these compounds have been prepared as inter-
`penetrated and noninterpenetrated forms. Further-
`more. some interpenetratcd structures can also be
`regarded as open framework since if interpenetration
`will not 1|eccssa1‘ily afford close-packing. interpen-
`etrated structures can still contain channels large
`enough to hold, for example, aromatic guests. Such
`is the case for square grid networks based upon
`ligands such as bipy-eta. Figure 25 reveals how either
`open Framework square grid or interpenetrated square
`grid structures can be readily generated for the same
`square grid network.2°" Both compounds contain
`square grids of formula INi(bipy‘-eta)2(NO3)l,,. As
`would be expected.
`the compound illustrated in
`Figure 253.
`lNi(bipy-eta)2(;\i0s)ln. exhibits clay-like
`properties and can desert) and adsorb guests. Crys-
`tallinity is lost, but the square grid architecture
`retains its integrity below 220 “C. The interpen-
`etrated analogue is illustrated in Figure 25b and is
`effectively a 3D architecture that is built by inter-
`penetration of square grids. This compound has a
`more rigid structure than its nor1inl:erpenetrated
`form and behaves like a zeolitic solid rather than a
`
`clay-like solid. Both compounds are stable to loss of
`guest, but the former loses crystallinity upon loss of
`guest.
`Organic networks are also capable of exhibiting
`interpenetration. and in this context I-I:iTi\'iA repre-
`sents a prototype! example. As discussed earlier, H3-
`TMA is predisposed to generate honeycomb sheets
`with large cavities. Three phases of I-l3TMA have
`thus far been characterized. In two of these phases.
`one of which is pure l—l3TMA, 3-fold interpenetration
`oi‘ the individual hexagonal cavities occurs?“ In pure
`H3TMA. the honeycomb networks are puckered in
`such a manner that small cavities are generated.“”
`These cavities can hold small molecules. including
`
`(bl
`
`Figure 23. Illustrations of two structures observed for [Co-
`(hipy-eta)1_5[?\lO3)2]:
`(a} ladder in which all bipy-eta ligands
`adapt an arrtr‘-orientation and (bi bilayer in which bipy-
`cta ligands adapt anti-« and gauche-orientations in a 2:1
`ratio.
`
`
`
`Figure 24. Illustrations of crystal structures of [Co(bipy-
`etai1_=_-.(NO3)2l cocrystallized with (a) MeCN and (b) no
`solvent of crystallization. Note how the cavities collapse
`in the absence of the adsorbed solvent.
`
`eta);,5(N03)g],, collapses to close the cavity that exists
`when crystallized from MeC—.\i.’” Note the difference
`in torsion angles between the two CDT'flp0llnd:~;_ A
`similar situation occurs in compounds based upon a
`longer bis-pyridyl ligand. 1.4-bis((4—pyridyl)methyli-
`2,3,5,6-tetralluorophenylene. 2. When 2 is complexed
`to Cd to form compounds of formula [Cd(2}g{NO3l2l.
`three very different supramolecular isomers have
`been observed depending upon the nature of guests
`or templates:
`ID chains. 21) sheets. and 3D diamon-
`doid neLworks.33“
`.
`A dramatic illustration of how conformational
`
`variability can influence crystal packing in organic
`compounds is illustrated by the compound 5-methyl-
`2-[[2—nitrophenyl)aInino]—3—thiophenecarbonitile. This
`compound exists in at least six polymorphic: pliases.
`The primary differeiice between the six phases lies
`with the torsion angle between the thiophene moiety
`and the 0-nitroaniline fragment. which varies from
`21.7” to l04.?°.55
`
`C. Catenane Supramoiecular Isomerism
`
`interpenetrating
`independent
`The existent:e of
`networks is surprisingly common if relatively large
`
`Janssen Ex. 2043
`
`Lupin Ltd. v. Janssen Sciences Ireland UC
`|PR2015-01030
`
`(Page 24 of 30)
`
`

`
`From Molecules to Crystal Engineering
`
`Chernical Reviews. 2001. Vol. 101. No. B
`
`1653
`
`
`
`Figure 28. Illustration of the or-polymorph of BTC that
`illustrates how BTC can sustain the enclathration of
`p-nitroaniline:
`(a) view of the 10-fold interpenetrated
`puckered honeycomb networks, lb] view of a single BTC
`network with 3-fold interpenetration per cavity and incor-
`poration ofp-nitroaniline in “pockets”. and (C) perspective
`view of two p-nitroanilincs situated in a single cavity,
`surrounded by the three interpenetrating networks.
`
`halogens” and pnitroaniline.335 The structure of the
`inclusion compound formed between H3TMA and
`p-nitroaniline is illustrated in Figure 26. The other
`interpenetrated phase of H3T?vlA contains flat sheets.
`and it therefore contains infinite 1D channels?“ The
`noninterpenetratcd phase oil-I-_-.TlVl1\, illustrated in
`Figure 14,
`is formed when long-chain all-tanes are
`used during the crystallization process.
`A honeycomb network is also generated when H3-
`TMA is cocrystallized with bipy. This structure
`exhibits interpenetration in 2D to form a novel
`carpeolike architecture?“ However, the noninter—
`penetrated form remains to be isolated.
`The results described above all suggest that inter-
`penetration can be avoided if appropriate templates
`are used during the crystallization. It
`is therefore
`reasonable to see interpenetration as another ex-
`ample oi" supramolecuiar isomerism. one that can be
`controlled by use of guest or template molecules
`during crystallization.
`
`1/. Potential Applications
`A considerable amount oi" research into understand-
`
`ing the nature and predictability of supramolccular
`synthons remains to be conducted. An enhanced
`database concerning supramolecular synthons in the
`broad context would also assist our fundamental
`understanding of solution chemistry and biochemis-
`try. Nevertheless, it is clear that there are a number
`of applications of crystal engineering that could be
`realized in the short term. Several of‘ those are
`summarized below. (1) From a supramolecular per-
`spective. binary compounds represent an illustration
`of how one might exploit the modular approach to
`design new supermolecules, especially in the solid
`state, it is reasonable to assert that supramolecular
`synthesis oi‘ new classes of cocrystal and modular
`solid offers potential to increase the known range of
`crystalline tnateI‘ials by two or
`three orders of
`
`magnitude and to facilitate combinatorial approaches
`to materials science. For example, if one were to only
`consider cocrystals that are sustained by hydrogen
`bonding, a wide range of compositions exists that
`remain to be explored. it is perhaps sobering to
`realize that. at least in principle. molecules that are
`deficient in hydrogen-bonci acceptors arc. inherently
`prone to form supermolecuies with molecules that
`contain excess hydrogen-bond acceptors. Even if one
`considered only simple examples such as pyridines.
`there are many permutations for formation of binary
`compounds. If one were to study. for example. 20
`pyridines and 20 carboxylic acids, then one would
`expect I100 new binary compounds with predictable
`composition and structure. Such a strategy could be
`important in the context of supramoiecular deriva-
`tives of drugs and functional materials (i.e.. modifica-
`tion of bulk properties without changing the molec-
`ular structure of the active species) or they could
`serve as precursors to covalent products. including
`polymers. Such an approach has already been effec-
`tive in formulation of polaroid Film?“ There also
`exists the possibility ofrationalizlng certain types of
`host—guest structures as being based upon topologi-
`cally complementary networks?“-243 Such compounds
`are also based upon self—assembly and might be
`prototypal for large numbers of related structures.
`{2} It is already established that solvent-free synthe-
`sis, green chemistry, offers many potential advan-
`tages, including cost and environmental benefits. '1“
`Cocrystallization of substrates and subsequently
`conducting reactions in the solid state offers the
`opportunity for very careful control over regio— and
`stereochemistry. It
`is also possible that supra-
`molecular arrays could act as precursors to new
`classes of 2D and 3D covalent polyvmersfi"-*7 (3) New
`classes oi" adsorbent. "organic and metal—organic
`clays and zeolites" represent an area in which
`considerable progress has already been made. Such
`compounds offer clear potential for the following:
`efficient. cost-effective alternatives to current uneth-
`ods of enantiomerii: separations. new materials for
`separation of gases, liquids, and solutes, new indus-
`trial heterogeneous catalysts. new drug delivery
`matrixes (e.g., matrix for oral delivery of otherwise
`unstable drugs), a new generation of chemical sen-
`sors, and new storage matrixes for gases such as
`methane. Recent results indicate that synthetic
`metal organic polymers can offer high lcvels of
`thermal stability and can supersede zeolites in terms
`of surface area and Capacity for small guest mol-
`ecules."'”‘ (4) The rational design of polar materials
`for use in materials science also represents an aspect
`of crystal engineering that has alrea