Article

Cite This: Acc. Chem. Res. XXXX, XXX, XXX−XXX pubs.acs.org/accounts

Strategy and Methodology in the Synthesis of Multicomponent

Molecular Solids: The Quest for Higher Cocrystals
Niyaz A. Mir,†,‡ Ritesh Dubey,† and Gautam R. Desiraju*,†
†
Solid State and Structural Chemistry Unit, Indian Institute of Science, Bangalore 560 012, Karnataka, India
‡
Higher Education Department, Government Degree College Pattan, Pattan 193 121, Jammu and Kashmir, India

CONSPECTUS: Crystal engineering is the art and science of

making crystals by design. Crystallization is inherently a
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

purifying phenomenon. Bringing together more than one

organic compound into the same crystal always needs
deliberate action. Cocrystals are important because they
offer a route to the controlled modulation of crystal
Downloaded via GUILFORD COLG on July 19, 2019 at 03:02:10 (UTC).

properties. The route to cocrystal synthesis was opened up

with the heterosynthon concept, which considers the
complementary recognition of chemical groups from different
molecules. Using this concept, binary cocrystals of enormous
variety have been generated, even as crystal engineering has
evolved into a form of solid-state supramolecular synthesis.
Introducing a third component (a component is somewhat
arbitrarily defined as an organic substance that is a solid at room temperature, mostly with the idea of excluding solvates) in a
stoichiometric manner requires substantially greater effort and a careful balance of intermolecular interactionstheir strengths,
directional properties, and distance falloff characteristics. The first systematic ternary cocrystal synthesis was reported around 15
years ago. Drawing in a fourth component in stoichiometric amounts is exceedingly difficult, and we reported such syntheses in
2016. To date, a limited number of ternary cocrystals have been realized (around 120 in all, with a half from our group) and an
even smaller number of quaternary cocrystals (around 30, all from our group, barring one). It is impressive that our experiments
largely yielded the intended higher cocrystal (three- or four-component) with very small traces of contaminating binaries and
pure compounds. A fifth or sixth component may be brought into the solid in the manner of a solid solution in that these
components are situated at one of the sites of the quaternary cocrystal. To date, five components have not been included
stoichiometrically within the same crystal. This is still an open challenge.
The merit in synthesizing (higher) cocrystals is that one can systematically engineer property modularity: Each component is
associated with a distinct property. This is important in the pharmaceutical industry, where each component can, in principle,
confer a different, desirable propertydrug action, solubility, or permeability. However, difficult synthetic targets are also
addressed in chemistry simply because they are there. The intellectual satisfaction in making something that is very difficult to
make renders the enterprise worthwhile in itself, and new chemistry usually gets uncovered in the process.
The development of synthetic organic chemistry can undoubtedly be credited to various reliable methods for chemical
transformations, and many difficult total syntheses were achieved by employing these methods over two centuries of research. In
contrast, supramolecular synthesis (of multicomponent cocrystals and other assemblies) is in no way at a similar level of
sophistication because the subject is still relatively young. Our group and others have reported the synthesis of many higher
cocrystals with reliable, reproducible, and robust design strategies. There is a general perception that the isolation of some of
these cocrystals is a matter of luck! The crux of this Account is that far from being a serendipitous matter, higher cocrystals may
only be made with a judicious combination of strategy and methodologythe essence of synthesis.

■ INTRODUCTION
Organic molecules have a surfeit of mutual recognition sites,
actually is that multicomponent crystalscrystals containing
two or more molecules, each of which crystallizes as a solid
regions of chemical or geometrical distinctiveness, so that the under ambient conditionsare even formed. In reality,
formation of a monocomponent crystal from solution is multicomponent crystals have been noted from the late 18th
unproblematic.1 Complementarity is the basic requirement for
molecular association. This is expressed in several contexts, and early 19th centuries.10,11 Reports in the last century were
such as lock-and-key,2 bumps-in-hollows,3 donor−acceptor sporadic at first12 but picked up significantly later on.13
interactions,4 hydrogen bonds,5,6 and halogen bonds.7,8
Crystals are usually built from just one type of molecule; Received: April 27, 2019
even solvent is not included, for the most part.9 The surprise

© XXXX American Chemical Society A DOI: 10.1021/acs.accounts.9b00211

Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
Figure 1. Early examples of binary cocrystals obtained with design principles. The final example contains segregated layers of molecules. The
principal author and year of publication are given in each case.18−22
We (GRD research group) have been studying multi-
component crystals,14 or cocrystals,15,16 for several decades,
mostly in the context of crystal engineering, the design of
functional solids. Noting the special nature of multicomponent
crystallization, we wrote,17 as far back as 1985, that “the
engineering of a crystal structure of a molecular complex is
inherently attractive for two reasons. First, dissimilar organic
molecules prefer to crystallize separately when [taken] together
in solution unless very specific electronic or geometrical factors
are optimized. Hence the very manifestation of co-crystal-
lization in a particular system implies that it is possible to
dissect and analyse a few significant molecular interactions
from amongst the larger number that actually determine the
stable crystal structure. This is true for all kinds of molecular
complexes, disordered or ordered and with fixed or variable
stoichiometry.”
An ability to identify characteristic heteromolecular
interactions was identified fairly early on as a strategic entry
B DOI: 10.1021/acs.accounts.9b00211
Article
point for the formation of binary cocrystals.18−22 Figure 1
illustrates some examples. X-ray crystallography23 rapidly
established itself as the method of choice for the character-
ization of intermolecular interactions, and by the late 1990s, it
was clear that the formation of a multicomponent crystal, and
in those times this meant just a two-component crystal, was
mediated by chemical or geometrical recognition between
molecules. Chemical recognition through directed interac-
tions24 is more specific and generally leads to stoichiometric
cocrystals, whereas geometrical recognition,3 which is based on
shape and size, leads to both stoichiometric solids and solid
solutions (alloys) that are inherently nonstoichiometric. We
will not enter into a philosophical discourse as to whether
these types of recognition are intrinsically the same,25 and we
merely use the chemical/geometrical yardstick in this Account
as a measure of subjective classification.
Both types of molecular recognition are complementary in
nature, and this brings with it the element of directionality,26
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
which is the critical property of the supramolecular synthon,27
the basic modular unit that encapsulates the key recognition
events that must precede crystallization (Figure 2). The
Figure 2. Synthon, homosynthon, heterosynthon, and retrosynthesis.
synthon has become an important strategic element in crystal
engineering because it lends itself to a retrosynthetic dissection
of a crystal structure, with the crucial interactions acting as the
disconnection sites, so that a crystal synthesis may be planned
from molecular precursors. Accordingly, the synthon has
proved itself to be particularly useful in planning the synthesis
of a multicomponent crystal wherein different molecules need
to be persuaded to come together within the same crystalline
lattice.
The synthon, while it was implied from earlier times, was
formally defined in 1995. Nearly a decade later, the
heterosynthon, which is a recognition unit between two different
functional groups, was defined by Almarsson and Zaworotko.28
For example, the common acid−amide heterosynthon is found
both in the single-component crystal of a monoamide of a
dicarboxylic acid and in a two-component acid−amide
cocrystal (Figure 2). The heterosynthon is critical to the
general design of a two-component crystal. Until its formal
definition, binary cocrystals were obtained mostly with just a
qualitative knowledge of hydrogen-bonding and donor−
acceptor interactions. This simple methodology was mostly
trial and error. Cocrystals were obtained, and one published
crystal structures without commenting too much on the large
numbers of unsuccessful trials.
The definition of the heterosynthon opened the floodgates
as far as the number and variety of binary cocrystals were
concerned.29 The preparation of individual two-component
solids (called molecular complexes at the time) evolved to the
synthesis of general families of binary cocrystals. The synthon
itself could be fine-tuned so that whole groups of related
compounds could be screened against groups of a different
chemical or geometrical diversity, literally in a combinatorial
C DOI: 10.1021/acs.accounts.9b00211
Article
fashion, so that enormous numbers of binary cocrystals could
be reliably obtained. An added impetus to such efforts was the
identification of the pharmaceutical cocrystal, a two-compo-
nent solid in which one or both components are active
pharmaceutical ingredients.30 These cocrystals offered many
beneficial properties beyond those exhibited by single-
component crystals of the drug molecules themselves and
contributed in no small way to the maturation of synthetic
approaches toward two-component crystals. In summary, the
synthesis became more general, logical, quantitative, and
analytical. The phase of using “general methods for the
construction of complex”31 crystalline supramolecular entities
had begun.
■ SYNTHON HIERARCHY
An important objective during the first phase of logic-based
crystal engineering, roughly 1995−2005, was the probing of
various homo- and heterosynthon preferences with an aim to
obtain desired crystal structure patterns. Hydrogen bonds
often form in a hierarchical mode, in that the best donor
connects to the best acceptor, the second best donor to the
second best acceptor, and so on.32 Such hierarchical patterns
are characteristic of kinetic factors in crystallization.33 Using
such an approach, if a ditopic molecule, or supramolecular
reagent (SR), contains either two different hydrogen bond
(HB) acceptor34 or two different HB donor35−37 sites, and if it
is taken with two different molecules having HB functionalities
complementary to SR, then a ternary cocrystal is most likely to
form (Figure 3).
Figure 3. Ternary cocrystal design using interaction hierarchy.
This design strategy was first used by the Aakeröy group,
who crystallized isonicotinamide with equimolar amounts of
two different carboxylic acids with a difference in acidity.34
Stoichiometric ternary cocrystals in high supramolecular yield
were formed, wherein the stronger acid (best donor)
hydrogen-bonds preferentially with the best acceptor (the
pyridine N atom) to form acid···pyridine heterosynthons, and
the weaker acid (second best donor) forms an acid···amide
heterosynthon. In a complementary fashion, a molecule may
be chosen with two distinct acidic (or hydrogen-bond donor)
sites and cocrystallized with two different bases to give a
ternary cocrystal.35−37
We developed this line of thinking using two different
methodological approaches. Noting that crystal structures of
phenols like γ-quinol, 4,4′-biphenol, and 4-hydroxybenzoic
acid are homologous with infinite cooperative O−H···O−H···
O−H··· hydrogen bonds, we cocrystallized a mixture of 4-
hydroxybenzamide, pyrazine, and an α,ω-dicarboxylic acid.
Pyrazine molecules are hydrogen-bonded to the phenolic
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
Figure 4. Strategy for ternary cocrystal synthesis based on synthon insulation and robustness.
hydroxy group and are insulated from the acid···amide synthon
to give ternary cocrystals (Figure 4).35,38
In the second approach, we employed halogen bonding as
the connecting interaction, retaining the acid···amide hetero-
synthon. The use of halogen bonding along with hydrogen
bonding in cocrystal synthesis needs some comment. The
halogen bond is similar to the hydrogen bond, and whereas it is
of an intermediate strength between strong and weak hydrogen
bonds, it is also of a different polarizability.8 All of this makes it
an interesting SR. If one seeks to homologate any two-
component system, AB, into a three-component system, ABC,
then the A···B interaction and the B···C (or A···C) interaction
must be of graded strength. If B···C competes too well with
A···B, then a two-component crystal BC will form. If B···C is
too weak, then AB will continue to be preferred. Controlled
interaction hierarchy is therefore a must: This line of argument
took us to the I···NO2 synthon.
In particular, I···NO2 is a well-insulated synthon in systems
containing carboxylic acid, pyridine, and iodo and nitro groups.
The cocrystallization of 4-nitrobenzamide with 1,4-diiodoben-
zene produced a 2:1 binary cocrystal containing the amide···
amide homosynthon insulated from the I···NO2 synthon.
Using this information and the reliability of carboxylic acid
insertion into the amide···amide homosynthon, we were able
to make three ternary cocrystals of the type 4-nitrobenzamide,
1,4-diiodobenzene, and a dicarboxylic acid (Figure 4). The
generality of this strategy was tested by using the molecules
with nitro and halogen groups exchanged, and four new ternary
cocrystals were obtained of the pattern 4-halobenzamide:di-
carboxylic acid:1,4-dinitrobenzene.38 The yields of these
supramolecular reactions were also estimated by listing the
failed cocrystallization experiments, with the eventual aim of
correlating these failures with better future methodologies.
This has been a consistent feature of our work in the synthesis
of multicomponent crystals. Each successful result is
accompanied by many failures and the fine-tuning of the
recrystallization solvent, the amounts of precursors chosen, the
cogrinding procedures, and temperature is a sine qua non of
D DOI: 10.1021/acs.accounts.9b00211
Article
this type of work, which in the end is painstaking and
laborious.
■
SHAPE-SIZE MIMICRY
The origins of this synthetic strategy go back to the early days
of crystal engineering. The chloro-methyl exchange,39 phenyl-
thienyl40 exchange, and other similar rules have been used
often enough to create structural mimics, in effect non-
stoichiometric binary solids, organic alloys as it were. In the
context of stoichiometric solids, 1,3,5,7-tetrabromoadamantane
(AdBr4)41 may be taken as representative. Here molecules are
connected, with six identical Br···Br contacts constituting a
tetrahedral Br4 cluster in a diamondoid network. This Br4
supramolecular cluster is topologically similar to the
hexamethylenetetramine (HMT) molecule. Thus, by applying
volume effects, AdBr4 is homologated to give the 1:2
AdBr4:HMT cocrystal. Interestingly, HMT molecules exist in
two crystallographic environments in the cocrystal, and volume
effects were further applied to make a stoichiometric ternary
cocrystal, wherein the more loosely bound HMT (poorer C−
H···N hydrogen bonds) is replaced by CBr4, a molecule of
roughly the same shape and size, so that the result is the 1:1:1
ternary cocrystal, HMT:CBr4:AdBr4. This one-off example of
crystal engineering is unique in that it was reported by us as far
back as 1994. Similarly, in 1999, Smolka et al. obtained a 2:3
2,2′-dihydroxybiphenyl (DHB):phenazine (PHE) cocrystal
and homologated it to a 2:2:1 DHB:PHE:ACR stoichiometric
ternary cocrystal.42
The simultaneous use of chemical and geometrical factors
offers greater synthetic control toward higher cocrystals
(Figure 5). The 1994 AdBr4 strategy was applied to binary
cocrystals of 4,4′-bipyridine (BPY) with either 2-methylresor-
cinol (MRES) or 5-methylresorcinol (ORC). Cocrystallization
of BPY with ORC gave two 2:3 cocrystal polymorphs, form I
and form II.43 Form I has the O−H···N hydrogen-bonded
closed tetramer B, and an additional loosely C−H···N-bound
BPY is sandwiched between these tetramers. We may refer to
the latter as free BPY. Form II has an O−H···N open finite
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
Figure 5. Shape-size mimicry for higher cocrystals.
chain A between ORC and BPY (Figure 6). In both of these
cases, the larger assemblies or Long-range Synthon Aufbau
Modules (LSAM),26 A and B, are more characteristic of the
eventual crystal structures formed. Both modules contain the
same O−H···N hydrogen bond, but the latter single interaction
is too small and lacks the critical information to be
representative of the entire crystal structure. MRES gives a
2:3 cocrystal with BPY, wherein the components, while they
are connected with O−H···N hydrogen bonds in an extended
topology, still have a free BPY, analogous to ORC:BPY form I
(Figure 7). In the next step, this free BPY is replaced by its
shape-size mimics 2,2′-bisthiophene (BTP), biphenyl (BP),
PHE, ACR, anthracene (ANT), and pyrene (PYR) to produce
stoichiometric ternary cocrystals (Figure 7).43 This strategy is
both specific, with respect to binding sites in the original
structure, and general, with respect to the shape/size mimic.
This augurs well for future homologation strategies.
In MRES:BPY, the introduction of some mimic molecules
(BP, BTP) produced pseudopolymorphic ternary cocrystals
with either O−H···N hydrogen bonding or both O−H···N and
O−H···O hydrogen bonding dictating the crystal packing.
Except that the latter type of structures is more flexible with
respect to hydrogen-bonding options, this difference did not
really affect the design strategy. The generality of this volume
replacement strategy was exemplified by the fact that as many
as 10 ternary cocrystals were formed on the pattern
resorcinol:mimic:BPY. This generality was found to be of
great utility in the ternary → quaternary homologation that we
undertook later.43
Chakraborty et al. showed that free PHE in ternary 4,4′-
bishydroxyazobenzene (BHB):ACR:PHE can be replaced by
only those π-stacking coformers (like quinoline and quinoxa-
line) that maintain the crucial closed C−H···N hydrogen-
bonded synthon present between two PHE molecules in the
original ternary.44 This function is served by N atoms present
in the one- and four-positions of the ring only. It is apparent
Figure 6. Synthons and LSAMs in the two polymorphs of the 2:3 cocrystal orcinol:4,4′-bipyridine. Similar patterns are obtained in related systems
derived from bases like 1,2-bis(4-pyridyl)ethylene (DPE-I), 1,2-bis(4-pyridyl)ethane (DPE-II), tetramethylpyrazine (TMP) and PHE, and phenols
like ORC and phloroglucinol (PGL).
E DOI: 10.1021/acs.accounts.9b00211
Article
Figure 7. Homologation to ternary cocrystals with mimics. The
parent binary is in the red box.
from these examples that shape-size mimicry can be applied for
homologation provided that the volume and directional
stabilizing interactions exhibited by the substitutable molecule
are met by the surrogate to a substantial extent.
■
STRUCTURAL INEQUIVALENCE
This strategy parallels the shape-size argument in that if a
molecule is situated in two distinct crystallographic environ-
ments in the same crystal structure, then it becomes
susceptible to substitution by a new molecule at the site
where it is weakly bound, if it is enthalpically advantageous to
do so. Therefore, homologation is achieved. Accordingly, a
binary cocrystal acts as a stepping stone to a ternary and so on
(Figure 8). However, structural inequivalence is distinct from
the shape-size strategy in the sense that it embodies the
essence of the graded interaction hierarchy of strong,
moderate, and weak interactions in the homologation exercise.
In this strategy, both chemical (interaction) and geometrical
(shape/size) complementarities are integral to crystal design.
MRES is the prototype molecule in this methodology, and
TMP was selected as a coformer.45,46 The 1:1 stoichiometric
cocrystallization of TMP and MRES provided an O−H···N-
mediated AB1B2-type binary solid, with B1 and B2 representing
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
Figure 8. Structural inequivalences lead to higher cocrystals. A, B, C, and D denote chemically distinct molecules, whereas 1 and 2 denotes
crystallographically distinct sites.
Figure 9. Synthetic routes to higher cocrystals based on structural inequivalences.
a breakdown of O−H···N hydrogen-bond degeneracy at the
chemical site B in the crystal structure. (The two O−H···N
metrics are different.) In the next homologation step, this
AB1B2-type binary solid was cocrystallized with polycyclic
aromatic hydrocarbons (PAHs) that can form C−H···π
interactions with the methyl groups of TMP, but rather than
the desired ABC1C2, what was actually obtained was an ABC-
type stoichiometric ternary where the two TMP molecules
flanking the MRE are in the same chemical and crystallo-
graphic environment. Typical PAH coformers in this structure
type are ANT, hexamethylbenzene (HMB), and PYR, and we
refer to them as the first set of coformers. In all of these ternary
crystal structures, there are no inequivalences at any of the
chemically identical sites; therefore, the ternary cannot be
exploited further for homologation. Accordingly, we term them
synthetic dead ends (Figure 9).46
Nevertheless, the same AB1B2-type binary solid, when
cocrystallized with other PAH coformers like BP, BPY, BTP
or ACR, perylene (PER), PHE, and tolan (TOL), leads to
ABC1C2-type stoichiometric ternary solids. These PAHs are
referred to as second- and third-set coformers, respectively.
The second set of PAHs leads to ternaries in which structural
degeneracy breakdown occurs at MRES, with respect to O−
H···N and O−H···O hydrogen bonds at different crystallo-
graphic sites (Figure 9). These can be homologated to
F DOI: 10.1021/acs.accounts.9b00211
Article
quaternaries ABCD, which, however, are synthetic dead ends.
In the third set of PAHs, the ternaries exhibit degeneracy
breakdown at TMP in that alternate TMPs are involved in C−
H···π stacking with the third component (Figure 9). These can
be homologated into ABCD1D2 quaternaries and are, in
principle, capable of higher homologation to quintinaries. In
summary, the ABC1C2-type solid is arguably the most critical
stage in our logic-driven methodology toward higher
cocrystals. The generality of our strategy led to more than
10 quaternary solids: The synthesis relies partly on certain
critical molecules like TMP but also on the principle of
structural inequivalences.46
Going forward from ABCD1D2-type solids, we attempted to
homologate them to ABCDE quinitaries. Noting that MRES
was disordered in the quaternary, we selected trimethylhy-
droquinone (TMHQ) as a shape-size mimic. The idea was that
TMHQ replaces the disordered MRES molecule completely.
However, TMHQ got oxidized to trimethybenzoquinone
(TMBQ) during the crystallization, and the resulting quinone
was found with partial occupancy in the quintinary crystal so
obtained. At this stage, we realized that there are limits to
higher homologation beyond the quaternary cocrystal using
the strategy of inequivalences. The operating interactions in
molecular crystals are so feeble (approach kT) that further
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
Figure 10. Toward senary solid solutions.
Figure 11. Combinatorial crystal synthesis strategy. Color coding represents chemically distinct molecules.
synthon modulation in the reaction cascade is simply not
possible.46
In our search for quintinary and higher cocrystals, we
therefore reverted to the age-old shape-size strategy.47 We
noted that when 2-chlororesorcinol (CRES) was taken in place
of MRES along with the three other components, the
quaternaries obtained were isomorphous to those reported
with MRES. On the basis of this validation of the chloro-
methyl exchange rule, it was anticipated that MRES and CRES
are mutually miscible and can share the same site in the crystal
structure to give quintinary solids of the type ABCD1−xEx.48
Accordingly, when TMP, BPY, and DPE-I (or DPE-II) were
ground with variable proportions of MRES and CRES and
then crystallized from solution, a series of quintinary solid
solution cocrystals were obtained in which MRES and CRES
share the same crystal site with the same ease. Using a similar
bromo-methyl analogy, 2-bromoresorcinol (BRES) was found
to behave similarly. Finally, and as expected, a mixture of
MRES, CRES, BRES, TMP, BPY, and DPE-II yielded a six-
component (or senary) molecular crystal of the type
ABC[D1−(x+y)ExFy]2, in which all resorcinols share the same
crystal site in a disordered manner with site occupancy factors
for CH3, Cl, and Br groups being 0.36, 0.24, and 0.40,
respectively. This six-component crystal is truly a synthetic
benchmark for a few years at least. An independent challenge is
to work out a successful route to a stoichiometric quintinary
cocrystal. The entire sequence is schematically illustrated in
Figure 10.
■ COMBINATORIAL CRYSTAL SYNTHESIS
So far, our synthetic strategies have been based on stepwise
buildup from precursors with heteromolecular recognition.
The methodologies were based on interaction hierarchies of
varying strengths and directionalities and led us logically to
stoichiometric quaternary solids. Because the precursors have
been selectively chosen to facilitate these hierarchies, these
syntheses may be considered to be outcomes of chemist
selection. In contrast, combinatorial crystallization depends on
the fact that if a particular molecule exists in many crystal
G DOI: 10.1021/acs.accounts.9b00211
Article
forms due to possibilities of multiple competing interactions,
then the crystallization of a cocktail of such molecules might
amplify some interactions only resulting in new and hitherto
unanticipated multicomponent solids. Unlike chemist selec-
tion, the system itself selects the fittest interactions or sets of
interactions. Accordingly, such syntheses might be taken as
examples of system selection (Figure 11).
The combinatorial method follows three steps: (i) selecting
molecular precursors; (ii) identifying a competing pool of
intermolecular interactions or synthon library; and (iii)
providing a route to amplify a particular interaction and
ultimately fix its expression in the solid state.49,50 The first step
is strategic in nature, and the two next steps are
methodological. Accordingly, we selected a polyhydric phenol
(ORC, PGL), a heterocyclic base (TMP, DPE-I, DPE-II), and
PAHs as precursors.51 In the second step, O−H···N is the
preferred hydrogen bond, along with other weaker variants of
interactions available in solution. In the third step, a skilful
selection of a template molecule is applied to quench certain
interaction possibilities, effectively leading to the fittest set.
Because the competing interactions in solution are weak and
reversible, they provide a sufficient scope for proof correction,
avoiding structural confusion; in practice, the ternaries and
quaternaries we isolated were largely free of contaminants of
lower cocrystals.
In 2014, we reported six distinct cocrystal polymorphs of
PGL:DPE-I.51 When DPE-I was replaced by PHE, the
resulting PGL:PHE cocrystal was obtained in seven distinct
crystal forms. Cocrystal polymorphism is not very common,
and so the appearance of multiple crystal structures in these
systems hints at the presence of competitive and kinetically
favored synthons in solution. In the next step, PHE was
diffused into a solution of PGL and DPE-I; the result was a
stoichiometric PGL:DPE-I:PHE ternary solid based on closed
synthon B (Figure 12). Similarly, when DPE-I was diffused
into a solution of PGL and PHE, an identical result was
obtained. Conceptually, this is a synthetic convergence going
from two binaries to a single ternary system.
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
Figure 12. Combinatorial crystal synthesis methodology. Note that DPE-I converges the synthetic cascade, whereas TMP diverges it.
When TMP is used as a replacement for DPE-I, we noticed,
perhaps for the first time, synthon polymorphism in the ternary
solid system, namely, PGL:TMP:PHE (Figure 12).52 Form I is
isostructural to PGL:DPE-I:PHE, the only difference being
that TMP now acts as a connector, unlike DPE-I. On the
contrary, form II is quite distinct in its supramolecular features:
PGL, TMP, and PHE are O−H···N hydrogen-bonded and are
topologically similar to open LSAM A. The appearance of
polymorphism, in other words, structural divergence, in a
ternary solid suggests multiple binding possibilities available in
solution, suggesting routes for further homologation. Following
this hypothesis, when ANT was used as a template with PGL,
PHE, and TMP in solution, the system selected module A in
such a way that both N-coformers participate in distinct O−
H···N hydrogen-bonding patterns, whereas ANT is sand-
wiched between these modules. This is the combinatorial step.
In this way, we obtained the stoichiometric PGL:PHE:TM-
P:ANT and also PGL:PHE:TMP:DPE-II quaternary cocrystals
(Figure 12).52 The selection by the template is likely to be the
H DOI: 10.1021/acs.accounts.9b00211
Article
dominant factor that amplifies module A and amends its
intermolecular signature to accommodate a fourth distinct
compound in the crystal. A feature of combinatorial synthesis
is the precise compound selection during crystallization. Stable
minima appear after the system screens out other putative
kinetic minima in solution. This approach is dynamic,
reversible, and instructive in nature.49
To validate our template-directed homologation strategy in
more complex systems, we chose the polymorphic flavonoid
quercetin (QUE) because the flexible single bonds in the
molecular structure increase the conformational space in
solution, in other words, the level of complexity of molecular
recognition. High-throughput screening of QUE with dibasic
coformers enabled us to localize kinetically favored situations
having robust molecular conformations and supramolecular
synthons.53 Using shape-size mimicry, the free BPY was
replaced by BTP, and a QUE:BPY:BTP ternary solid was
isolated. Using a similar approach, larger molecules like
tetrathiofulvalene, PYR, and ANT were used as templates
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
and incorporated as third components in binary homologa-
tions. In summary, PGL and QUE provided us with substantial
confidence toward the reliability of this overall design strategy.
QUE may well be a starting point in opening up more
structural space in the quest for higher cocrystals because of
the plethora of hydroxyl groups in the molecule and their
propensities for conformational variations.
■
STRATEGY AND METHODOLOGY
We have outlined four synthetic strategies, namely, synthon
hierarchy, shape-size mimicry, structural inequivalence, and
combinatorial synthesis, for the design or engineering of higher
cocrystals. There are some regions of intersection of these four
approaches. The first and second have been used by others as
well, and ternary cocrystals are gradually becoming more
common, but there is a great deal that separates a putative
strategy and actually isolating a higher cocrystal in the
laboratory. This is the gulf of methodology: What compounds
does one select, how does one conduct the cocrystallization
experiment, how is the required temperature chosen, and so
on. It is difficult to give a sense of the many trials and
(eventually) failed experiments in our laboratory; many
crystallization attempts toward a quaternary cocrystal resulted
in ternaries, which, while in themselves were interesting, are
still failures in terms of what was planned.
In terms of solvents used, a typical experiment would involve
taking selected compounds and using liquid-assisted cogrind-
ing before dissolving the mixture in a range of solvents and
solvent mixtures.54 All of our experiments involved a
polyhydric phenol and a heterocyclic base. For this
combination, solvents such as nitromethane, THF, dioxane,
EtOAc, and MeCN are typical. Higher boiling solvents such as
DMF and DMSO rarely provided desired and isolable
cocrystals. Even when a solvent combination and other
experimental conditions were optimized, a small perturbation
in the stoichiometric ratios used might have an adverse effect
on the planned outcome. We did not greatly alter the
crystallization temperatures across runs. However, this is an
aspect that undoubtedly needs more attention in the future,
especially because crystallization is an inherently kinetic
phenomenon.
We have sampled an infinitesimal amount of the structural
space potentially available for the synthesis of ternary,
quaternary, and other higher cocrystals. Undoubtedly, more
of them will be isolated with systematic variations in the
constituent compounds. In this context, the role of the
Cambridge Structural Database in being able to zero in on
compounds that have a higher probability of forming
heterosynthons should not be underestimated. Eventually,
with such cheminformatics and artificial intelligence ap-
proaches, it should be possible to identify libraries of small-
molecule compounds that will have an enhanced ability to
yield higher cocrystals.
■
QUO VADIS?
The synthesis of higher cocrystals may be likened to the
supramolecular equivalent of making highly functionalized
molecules in synthetic organic chemistry. The inherent
problem, unlike in synthetic chemistry, is that a single-
component crystal is obtained so easily and that its formation
from a solution containing many compounds is so efficient.
The introduction of a second component implies that one or
I DOI: 10.1021/acs.accounts.9b00211
Article
more heteromolecular interactions is stronger or more specific
than the homomolecular interactions. This in itself is not
routine because the interactions involved in the crystallization
of both single- and two-component crystals, hydrogen
bonding, π···π, and ionic, are essentially the same. Solvent
and temperature effects are therefore important, even in the
synthesis of two-component crystals, because these factors
might bring about some discrimination between possible
crystallization routes.
Obtaining a three-component crystal represents a higher
level of difficulty because there needs to be a trade-off between
stability and lability. Unlike the solution synthesis of a target
molecule, which is accomplished in a stepwise manner,
supramolecular synthesis in crystal engineering needs to be
done in a single step. The reactants and reagents are all the
same. If many binary interactions are strong, then the tendency
to form a ternary crystal is diminished, and there might be a
competition among putative binaries. If the binary interactions
are too weak, then only a conglomerate of single-component
crystals might be obtained. Accordingly, some kind of
gradation or hierarchy of interactions seems to be required,
and this, in turn, implies that certain types of molecules might
be more predisposed to the formation of three-component
crystals: Isonicotinamide occurred fairly often in early
examples of ternary cocrystal synthesis. This corresponds to
the first stage of synthetic developmentartistic, qualitative,
and unpredictable.55 Subsequently, it was found that more
general strategies of synthesis could be employed: Several 4-
substituted benzamides could be used instead of just
isonicotinamide, and with this development, interaction types
outside of strong hydrogen bonding came to be employed.
This might be said to correspond to the next stage of
synthesisscientific, logical, and predictable.56 Accordingly,
the 120 or so ternary cocrystals that have been reported to date
contain a nice diversity of molecules and molecular
functionalities, but making a three-component crystal today
is still not routine; we have a way to go.
The synthesis of a quaternary cocrystal is definitely in the
artistic domain today. The balancing of interactions and
experimental conditions is so very delicate that one is
necessarily restricted to a small structural and supramolecular
space. Around 20 four-component crystals have been reported
so far, in which the four compounds are present in a
stoichiometric ratio. All of them contain TMP and some kind
of resorcinol. In the next, more scientific stage, one predicts
that more general protocols will become available for the
synthesis of quaternary cocrystals, and with this a greater
molecular diversity will come to be used. This progression
from art to science occurred in synthetic organic chemistry 50
years ago, and Corey,31 while introducing it, described the
previous phase as one in which “the uncertain and severe limits
on the applicability of any given reaction all combine to create
an impression that the design of a synthesis is apt to be
tenuously hypothetical and is mainly a f unction of the unique
circumstances in each particular case and, moreover, that
considerations of a highly general nature are neither dominant
nor very useful.” All of the present indications are that the
crystal engineering of higher cocrystals will see this progression
too.
At this time, the synthesis of stoichiometric five-component
and higher crystals would appear to be very difficult simply
because there is very little room to fit in four or more
intermolecular interactions in a decreasing energy sequence if
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
the first interaction is strong hydrogen bonding with an energy
of, say, just ∼40 kJ/mol and the lower limit is obviously a little
above kT. It is entirely possible that a stoichiometric five-
component crystal is made, but this might be the natural limit
for the number of separate compounds in a single crystal unless
one adopts the solid solution route.57
Mechanistically, the synthesis of higher cocrystals offers a
scope for further study into the nature of crystallization itself.
For example, do some of these crystallizations go through a
single-step nucleation mechanism, whereas others take the
two-step, solvent-mediated, route?58 Why are the crystalliza-
tions so efficient in terms of supramolecular yield once one
arrives at the correct combinations of coformers? It is easy to
understand that crystallization outcomes are solvent-sensitive,
but why does a particular solvent lead to a particular outcome?
Along these lines, would a mechanical or sonochemical
approach to cocrystal synthesis favor the formation of higher
cocrystals? The syntheses we have described are linear in the
sense that we have retroanalyzed each extra component in
terms of a homologation. Would it be possible in the future to
conceive of a convergent approach to higher cocrystals by
stabilizing smaller entities? Accordingly, could higher cocrys-
tals be grown from lower ones? These and other matters
continue to place higher cocrystal synthesis at the very
frontiers of crystal engineering.
In the end, any synthetic activity at the highest level strains
both the scientific imagination and the artistic imagery of the
practitioner. In crystal engineering, as in organic synthesis, art
and science overlap. The practice and use of supramolecular
synthons, which play a key role in cocrystal synthesis, need
considerable ability and knowledge. The artistic component is
bound by the laws of science, even as it seems to operate in a
different frame of reference. It is possibly most appropriate to
quote Corey,31 who said, “The proposition can be advanced
that many of the most distinguished synthetic studies have
entailed a balance between two different research philosophies,
one embodying the ideal of a deductive analysis based on
known methodology and current theory, and the other
emphasizing innovation and even speculation.”
The parallelisms between synthetic organic chemistry and
crystal engineering continue to unfold. Nearly 20 years ago,
one of us coauthored a comment59 in which it was stated, “..the
development of organic chemistry during the last several
decades shows that synthesis can be taken to a higher plane
only if it is accompanied by analysis. We are now at the
threshold of similar developments in crystal engineering, where
imagination will be fuelled by reason to yield supramolecular
structures, the variety and utility of which will only be limited
by the ingenuity of the practitioner.” Recent developments in
the synthesis of higher cocrystals show that we have almost
certainly arrived at this point in the development of the subject
of crystal engineering.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: gautam.desiraju@gmail.com.
ORCID
Niyaz A. Mir: 0000-0002-9247-6385
Ritesh Dubey: 0000-0003-1201-420X
Gautam R. Desiraju: 0000-0002-7708-9176
J DOI: 10.1021/acs.accounts.9b00211
Article
Notes
The authors declare no competing financial interest.
Biographies
Niyaz A. Mir received his B.Sc., M.Sc., and Ph.D from Aligarh Muslim
University (AMU). He received his Ph.D. in 2013 under the
supervision of Prof. M. Muneer. He did postdoctoral work with Prof.
G. R. Desiraju from 2014 to 2015. He was briefly an assistant
professor at AMU before taking up his current assignment as assistant
professor at Jammu & Kashmir.
Ritesh Dubey obtained his Ph.D under the supervision of Prof. G. R.
Desiraju in 2015. He then undertook a postdoctoral position in the
group of Prof. M. Fujita at The University of Tokyo, working in the
area of crystalline sponges. His current research interests include
metal coordination and hydrogen-bonded self-assemblies and their
applications.
Gautam R. Desiraju served on the Editorial Advisory Board of
Accounts of Chemical Research between 1998 and 2005 and as a
consulting editor between 2005 and 2009. He has been published in
Accounts of Chemical Research in 1986, 1991, 1996, 2002, 2014, and
2018.
■
ACKNOWLEDGMENTS
G.R.D. thanks the DST, Government of India for the award of
a J. C. Bose fellowship.
■ REFERENCES
(1) The Crystal as a Supramolecular Entity (Perspectives in
Supramolecular Chemistry); Desiraju, G. R., Ed.; John Wiley & Sons:
Chichester, U.K., 1996; Vol. 2.
(2) Fischer, E. Einfluss der Configuration auf die Wirkung der
Enzyme. Ber. Dtsch. Chem. Ges. 1894, 27, 2985−2993.
(3) Kitaigorodskii, A. I. Molecular Crystals and Molecules; Academic
Press: New York, 1973.
(4) Foster, R. Organic Charge-Transfer Complexes; Academic Press:
London, 1969.
(5) Jeffrey, G. A. An Introduction to Hydrogen Bonding; Oxford
University Press: Oxford, U.K., 1997.
(6) Desiraju, G. R.; Steiner, T. The Weak Hydrogen Bond: In
Structural Chemistry and Biology; Oxford University Press: Oxford,
U.K., 2001.
(7) Metrangolo, P.; Neukirch, H.; Pilati, T.; Resnati, G. Halogen
bonding based recognition processes: A world parallel to hydrogen
bonding. Acc. Chem. Res. 2005, 38, 386−395.
(8) Mukherjee, A.; Tothadi, S.; Desiraju, G. R. Halogen bonds in
crystal engineering: Like hydrogen bonds yet different. Acc. Chem. Res.
2014, 47, 2514−2524.
(9) Nangia, A.; Desiraju, G. R. Pseudopolymorphism: Occurrences
of hydrogen bonding organic solvents in molecular crystals. Chem.
Commun. 1999, 605−606.
(10) Davy, H. I. The Bakerian Lecture. On some of the
combinations of oxymuriatic gas and oxygene, and on the chemical
relations of these principles, to inflammable bodies. Philos. Trans. R.
Soc. 1811, 101, 1−35.
(11) Faraday, M. On hydrate of chlorine. Quart. J. Sci. 1823, 15, 71−
75.
(12) Pfeiffer, P. Organische Molekülverbindungen; F. Enke: Stuttgart,
Germany, 1922.
(13) Herbstein, F. H. Crystalline Molecular Complexes and
Compounds: Structures and Principles; Oxford University Press:
Oxford, U.K., 2005; Vols. 1 & 2.
(14) Bond, A. D. What is a co-crystal? CrystEngComm 2007, 9, 833−
834.
(15) Desiraju, G. R. Crystal and co-crystal. CrystEngComm 2003, 5,
466−467.
Acc. Chem. Res. XXXX, XXX, XXX−XXX
Accounts of Chemical Research
(16) Dunitz, J. D. Crystal and co-crystal: a second opinion.
CrystEngComm 2003, 5, 506−506.
(17) Sarma, J. A. R. P.; Desiraju, G. R. The novel 1:1 donor−
acceptor complex, 3,4-dimethoxycinnamic acid−2,4-dinitrocinnamic
acid. Crystal engineering, structure, and anomalous lack of solid-state
topochemical reactivity. J. Chem. Soc., Perkin Trans. 2 1985, 1905−
1912.
(18) Huang, C.-M.; Leiserowitz, L.; Schmidt, G. M. J. Molecular
packing modes. Part XI. Crystal structures of the 2:1 complexes of
benzamide with succinic acid and furamide with oxalic acid. J. Chem.
Soc., Perkin Trans. 2 1973, 503−508.
(19) Etter, M. C.; Frankenbach, G. M. Hydrogen-bond directed
cocrystallization as a tool for designing acentric organic solids. Chem.
Mater. 1989, 1, 10−12.
(20) Seto, C. T.; Whitesides, G. M. Self-assembly based on the
cyanuric acid-melamine lattice. J. Am. Chem. Soc. 1990, 112, 6409−
6411.
(21) Biradha, K.; Sharma, C. V. K.; Panneerselvam, K.; Shimoni, L.;
Carrell, H. L.; Zacharias, D. E.; Desiraju, G. R. Solid-state
supramolecular assembly via C−H···O hydrogen bonds: crystal
structures of the complexes of 1,3,5-trinitrobenzene with dibenzyli-
deneacetone and 2,5-dibenzylidenecyclopentanone. J. Chem. Soc.,
Chem. Commun. 1993, 1473−1475.
(22) Reddy, D. S.; Goud, B. S.; Panneerselvam, K.; Desiraju, G. R.
C−H···N mediated hexagonal network in the crystal structure of the
1:1 molecular complex 1,3,5-tricyanobenzene-hexamethylbenzene. J.
Chem. Soc., Chem. Commun. 1993, 663−664.
(23) Mak, T. C. W.; Gong-Du, Z. Crystallography in Modern
Chemistry: A Resource Book of Crystal Structures; John Wiley & Sons:
New York, 1992.
(24) Desiraju, G. R. Crystal Engineering: The Design of Organic Solids;
Elsevier: Amsterdam, 1989.
(25) Mukherjee, A.; Desiraju, G. R. Synthon polymorphism and
pseudopolymorphism in co-crystals. The 4,4′-bipyridine−4-hydrox-
ybenzoic acid structural landscape. Chem. Commun. 2011, 47, 4090−
4092.
(26) Ganguly, P.; Desiraju, G. R. Long-range synthon Aufbau
modules (LSAM) in crystal structures: systematic changes in
C6H6−nFn (0 ⩽ n ⩽ 6) fluorobenzenes. CrystEngComm 2010, 12,
817−833.
(27) Desiraju, G. R. Supramolecular synthons in crystal engineering-
a new organic synthesis. Angew. Chem., Int. Ed. Engl. 1995, 34, 2311−
2327.
(28) Almarsson, Ö .; Zaworotko, M. J. Crystal engineering of the
composition of pharmaceutical phases. Do pharmaceutical co-crystals
represent a new path to improved medicines? Chem. Commun. 2004,
1889−1896.
(29) Gunawardana, C. A.; Aakeröy, C. B. Co-crystal synthesis: fact,
fancy, and great expectations. Chem. Commun. 2018, 54, 14047−
14060.
(30) Bolla, G.; Nangia, A. Pharmaceutical cocrystals: walking the
talk. Chem. Commun. 2016, 52, 8342−8360.
(31) Corey, E. J. General methods for the construction of complex
molecules. Pure Appl. Chem. 1967, 14, 19−38.
(32) Etter, M. C. Encoding and decoding hydrogen-bond patterns of
organic compounds. Acc. Chem. Res. 1990, 23, 120−126.
(33) Desiraju, G. R. Crystal engineering: A holistic view. Angew.
Chem., Int. Ed. 2007, 46, 8342−8356.
(34) Aakeröy, C. B.; Beatty, A. M.; Helfrich, B. A. Total synthesis”
supramolecular style: Design and hydrogen-bond-directed assembly of
ternary supermolecules. Angew. Chem., Int. Ed. 2001, 40, 3240−3242.
(35) Tothadi, S.; Desiraju, G. R. Synthon modularity in 4-
hydroxybenzamide−dicarboxylic acid cocrystals. Cryst. Growth Des.
2012, 12, 6188−6198.
(36) Bolla, G.; Nangia, A. Multicomponent ternary cocrystals of the
sulfonamide group with pyridine-amides and lactams. Chem. Commun.
2015, 51, 15578−15581.
(37) Adsmond, D. A.; Sinha, A. S.; Khandavilli, U. B. R.; Maguire, A.
R.; Lawrence, S. E. Design and synthesis of ternary cocrystals using
K DOI: 10.1021/acs.accounts.9b00211
Article
carboxyphenols and two complementary acceptor compounds. Cryst.
Growth Des. 2016, 16, 59−69.
(38) Tothadi, S.; Sanphui, P.; Desiraju, G. R. Obtaining synthon
modularity in ternary cocrystals with hydrogen bonds and halogen
bonds. Cryst. Growth Des. 2014, 14, 5293−5302.
(39) Desiraju, G. R.; Sarma, J. A. R. P. The chloro-methyl exchange
rule and its violations in the packing of organic molecular solids. Proc.
- Indian Acad. Sci., Chem. Sci. 1986, 96, 599−605.
(40) Thallapally, P. K.; Chakraborty, K.; Carrell, H. L.; Kotha, S.;
Desiraju, G. R. Shape and size effects in the crystal structures of
complexes of 1,3,5-trinitrobenzene with some trigonal donors: The
benzene−thiophene exchange rule. Tetrahedron 2000, 56, 6721−
6728.
(41) Reddy, D. S.; Craig, D. C.; Desiraju, G. R. Organic alloys:
diamondoid networks in crystalline complexes of 1,3,5,7-tetrabro-
moadamantane, hexamethylenetetramine and carbon tetrabromide. J.
Chem. Soc., Chem. Commun. 1994, 1457−1458.
(42) Smolka, T.; Boese, R.; Sustmann, R. Design of a three
component crystal based on the cocrystal of phenazine and 2,2′-
dihydroxybiphenyl. Struct. Chem. 1999, 10, 429−431.
(43) Tothadi, S.; Mukherjee, A.; Desiraju, G. R. Shape and size
mimicry in the design of ternary molecular solids: towards a robust
strategy for crystal engineering. Chem. Commun. 2011, 47, 12080−
12082.
(44) Chakraborty, S.; Rajput, L.; Desiraju, G. R. Designing ternary
co-crystals with stacking interactions and weak hydrogen bonds. 4,4′-
bis-hydroxyazobenzene. Cryst. Cryst. Growth Des. 2014, 14, 2571−
2577.
(45) Mir, N. A.; Dubey, R.; Tothadi, S.; Desiraju, G. R.
Combinatorial crystal synthesis of ternary solids based on 2-
methylresorcinol. CrystEngComm 2015, 17, 7866−7869.
(46) Mir, N. A.; Dubey, R.; Desiraju, G. R. Four- and five-
component molecular solids: crystal engineering strategies based on
structural inequivalence. IUCrJ 2016, 3, 96−101.
(47) Lusi, M. Engineering crystal properties through solid solutions.
Cryst. Growth Des. 2018, 18, 3704−3712.
(48) Paul, M.; Chakraborty, S.; Desiraju, G. R. Six-component
molecular solids: ABC[D1−(x+y)ExFy]2. J. Am. Chem. Soc. 2018, 140,
2309−2315.
(49) Lehn, J.-M. Dynamic combinatorial chemistry and virtual
combinatorial libraries. Chem. - Eur. J. 1999, 5, 2455−2463.
(50) Rowan, S. J.; Cantrill, S. J.; Cousins, G. R. L.; Sanders, J. K. M.;
Stoddart, J. F. Dynamic covalent chemistry. Angew. Chem., Int. Ed.
2002, 41, 898−952.
(51) Dubey, R.; Desiraju, G. R. Combinatorial crystal synthesis:
structural landscape of phloroglucinol:1,2-bis(4-pyridyl)ethylene and
phloroglucinol:phenazine. Angew. Chem., Int. Ed. 2014, 53, 13178−
13182.
(52) Dubey, R.; Mir, N. A.; Desiraju, G. R. Quaternary cocrystals:
combinatorial synthetic strategies based on long-range synthon
Aufbau modules (LSAM). IUCrJ 2016, 3, 102−107.
(53) Dubey, R.; Desiraju, G. R. Combinatorial selection of molecular
conformations and supramolecular synthons in quercetin cocrystal
landscapes: a route to ternary solids. IUCrJ 2015, 2, 402−408.
(54) Trask, A. V.; Motherwell, W. D. S.; Jones, W. Solvent-drop
grinding: Green polymorph control of cocrystallisation. Chem.
Commun. 2004, 890−891.
(55) Anand, N.; Bindra, J.; Ranganathan, S. Art in Organic Synthesis;
Holden Day: San Francisco, CA, 1970.
(56) Warren, S.; Wyatt, P. Organic Synthesis: The Disconnection
Approach; John Wiley & Sons: Hoboken, NJ, 2011.
(57) Nangia, A. K.; Desiraju, G. R. Crystal engineering: An outlook
for the future. Angew. Chem., Int. Ed. 2019, 58, 4100−4107.
(58) Davey, R. J.; Schroeder, S. L. M.; ter Horst, J. H. Nucleation of
Organic Crystals-A Molecular Perspective. Angew. Chem., Int. Ed.
2013, 52, 2166−2179.
(59) Nangia, A.; Desiraju, G. R. Supramolecular structures-Reason
and imagination. Acta Crystallogr., Sect. A: Found. Crystallogr. 1998,
54, 934−944.
Acc. Chem. Res. XXXX, XXX, XXX−XXX