Molecular symmetry and the design of molecular solids: The oxalamide functionality as a persistent hydrogen bonding unit

Article abstract of DOI:10.1021/ja961958q

A symmetry analysis based upon the structure of simple molecules and their anticipated intermolecular interactions can lead to successful predictions of molecular packing and crystal symmetry. As a demonstrated of these ideas an in-depth study of the oxalamide functionality as a persistent hydrogen bonding unit is presented. The synthesis and structural characterization of a series oxalamide dicarboxylic acids is presented and the structures compared with the analogous urea compounds. Both the urea and oxalamide dicarboxylic acids form designed two-dimensional hydrogen-bonded β-networks with a significant degree of reliability. The urea designs are quite reliable when there is a molecular 2-fold axis, but competing hydrogen bond patterns are found when less symmetrical molecules are studied. The oxalamide design based on inversion centers is also quite reliable, with the designed layer structure found in most cases.

Full text of DOI:10.1021/ja961958q

86
J. Am. Chem. Soc. 1997, 119, 86-93
Molecular Symmetry and the Design of Molecular Solids: The
Oxalamide Functionality as a Persistent Hydrogen Bonding Unit
Seth Coe, John J. Kane, Tam Luong Nguyen, Leticia M. Toledo, Eric Wininger,
Frank W. Fowler,* and Joseph W. Lauher*
Contribution from the Department of Chemistry, State UniVersity of New York,
Stony Brook, New York 11794
ReceiVed October 1, 1996X
Abstract: A symmetry analysis based upon the structure of simple molecules and their anticipated intermolecular
interactions can lead to successful predictions of molecular packing and crystal symmetry. As a demonstration of
these ideas an in-depth study of the oxalamide functionality as a persistent hydrogen bonding unit is presented. The
synthesis and structural characterization of a series oxalamide dicarboxylic acids is presented and the structures
compared with the analogous urea compounds. Both the urea and oxalamide dicarboxylic acids form designed
two-dimensional hydrogen-bonded â-networks with a significant degree of reliability. The urea designs are quite
reliable when there is a molecular 2-fold axis, but competing hydrogen bond patterns are found when less symmetrical
molecules are studied. The oxalamide design based on inversion centers is also quite reliable, with the designed
layer structure found in most cases.
Introduction
In our own work3 we have identified two-dimensional layered
solids as significant targets for supramolecular synthesis. Our
approach to a two-dimensionl layered solid has been to design
a two dimensional network of mutually hydrogen bonded
molecules. This approach has required the identification of
persistent and complementary hydrogen bond intereactions, but,
in addition, we have focused upon the problem of including
within our designs the necessary translational symmetry opera-
tors needed for a two-dimensional network.
The physical and chemical properties1 of a molecular solid
depend upon the nature of the constituent molecules as well as
the relative orientations and spacing of the molecules within
the solid. The preparation of a new solid, with specific
properties, is an example of supramolecular synthesis.2,3 As in
the case of molecular synthesis, a successful supramolecular
synthesis requires first a good design.
A successful design of a molecular crystal depends upon the
identification and synthesis of molecular functionalities that will
predictably and persistently lead to specific favorable intermo-
lecular interactions. By combining such chemical insight with
symmetry considerations, one may design and prepare molecules
that will self-assemble into crystalline solids that contain desired
structural features. However, this process of supramolecular
synthesis is extremely difficult due to a multitude of inherent
problems.
Intermolecular interactions are weak and relatively nondi-
rectional as compared to intramolecular bonds. This makes the
identification of predictable and persistent interactions difficult,
since alternate unanticipated structures may be more stable or
kinetically favored. Many designs that appear reasonable in
theory are never realized fact, because of an inadequate
accounting of the packing and symmetry requirements of a
crystalline solid. Despite these problems, much progress has
been made and many groups have completed successful
supramolecular syntheses.
A series of urea dicarboxylic acid derivatives have been
designed, prepared, and structurally characterized.3 The urea
functionality was found to be a reasonably reliable and persistent
unit that will self-assemble into a one-dimensional R-network4
held together via hydrogen bonds. With two side chains that
terminate in carboxylic acid or amide functionalities, urea
dicarboxylic acid derivatives form additional hydrogen bonds
in a second dimension, linking the one-dimensional R-networks
into two-dimensional â-networks.4 Simple symmetrical urea
dicarboxylic acid derivatives were found to form structures
consistent with this design, but various unsymmetrical racemates
formed structures that lacked the necessary translational sym-
metry and failed to form the anticipated â-network. However,
when a single enantiomer of an unsymmetrical chiral ureylene
was studied, the â-network formed in accordance with the
design.3b A discussion of the importance of chirality as a means
to control the translational symmetry needed for layer structures
was an important part of an earlier publication.
An examination of the racemate structures suggests that a
factor for the failure of the molecules to form the designed
â-networks was the tendency of these molecules to form discrete
centrosymmetric dimer structures in a manner inconsistent with
the layer design (Figure 1a). Centrosymmetric dimer formation
by asymmetric molecules does not generate the translational
X Abstract published in AdVance ACS Abstracts, December 15, 1996.
(1) There are many examples of physical properties that depend upon
molecular orientation and spacing as well as crystallographic symmetry.
Optical properties depend critically upon the relative orientation of molecular
dipoles in a crystal. Electronic properties depend upon molecular contacts
and spacing. Topochemical reactions depend upon molecular spacing and
relative orientations.
(2) Many groups are working on this problem. For a recent review see:
Lehn, J.-M. Supramolecular Chemistry. Concepts and PerspectiVes; VCH
Verlagsgesellschaft: Weinheim, 1995.
(3) (a) Zhao, X.; Chang, Y.-L.; Fowler, F. W.; Lauher, J. W. J. Am.
Chem. Soc. 1990, 112, 6627-6634. (b) Chang, Y.-L.; West, M. A.; Fowler,
F. W.; Lauher, J. W. J. Am. Chem. Soc. 1993, 115, 5991-6000. (c) Toledo,
L. M.; Lauher, J. W.; Fowler, F. W. Chem. Mater. 1994, 6, 1222-1226.
(d) Kane, J. J.; Liao, R. F.; Lauher, J. W.; Fowler , F. W. J. Am. Chem.
Soc. 1995, 117, 12003-12004
(4) Supramolecular assemblies can be divided into four groups dependent
upon the degree of translation symmetry.3b Discrete assemblies lack
translation symmetry and are characterized by their point group symmetry.
An R-network has one degree of translational symmetry and is characterized
by its rod group symmetry. A â-network has two degrees of translational
symmetry and is characterized by its layer group symmetry. A γ-network
has three degrees of translational symmetry and is characterized by its space
group symmetry. Lauher, J. W.; Chang, Y.-L.; Fowler, F. W. Mol. Cryst.
Liq. Cryst. 1992, 211, 99-109.
S0002-7863(96)01958-0 CCC: $14.00 © 1997 American Chemical Society

Molecular Symmetry and the Design of Molecular Solids
J. Am. Chem. Soc., Vol. 119, No. 1, 1997 87
Figure 1. (a) If an external center of symmetry is added to a chiral
molecule, then a centrosymmetric dimer with Ci point group symmetry
is produced. (b) If an external center of symmetry is added to a
centrosymmetric molecule, then the combination of the two centers
produces a translational symmetry operator and the resulting assembly
is an infinite one dimensional R-network with P1h rod group symmetry.
symmetry operations needed for layer production. The tendency
of racemates to form discrete centrosymmetric dimer structures
is well discussed in the crystallographic literature.5 The strategy
of using a single enantiomer is one approach to solving this
problem.
Another approach to the problem of discrete centrosymmetric
dimer production is to avoid the question of chirality all together
by choosing achiral molecules that will reliably maintain their
symmetry element and their achirality in a crystalline lattice.
The best approach is to choose molecules that possess and
maintain in the crystal a center of symmetry. If a molecule,
which is itself centric, forms a centrosymmetric dimer with
another molecule then the combination of two different centers
of symmetry, one internal to the molecule and the second
external to the molecule, will necessarily generate a translational
symmetry operator (Figure 1b). Chemically this will correspond
to the supramolecular assembly of a one-dimensional R-network.
In contrast to the urea functionality which is inherently
acentric and polar, the oxalamide functionality is inherently
centric. In crystals, symmetrical substituted ureas tend to
possess C2 molecular point group symmetry and form polar
R-networks with P2 rod symmetry, Figure 2. We anticipated
that centric oxalamide derivatives would maintain Ci symmetry
in a crystal and would form R-networks of P1h rod symmetry.
The necessary external center of symmetry would correspond
to the self-complementary oxalamide hydrogen bonds between
neighboring molecules in the R-networks, Figure 3.
In this paper we present an in-depth study of the oxalamide
functionality as a persistent hydrogen bonding unit. Oxalamides
have a chemical similarity to ureas and a similar propensity to
form hydrogen bonds. We report the synthesis and structural
characterization of oxalamide dicarboxylic acids and compare
them with the analogous urea compounds. For purpose of com-
parison, the synthesis and structural analysis of additional urea
dicarboxylic acid derivatives are also reported. Comparisons
will also be made with other known oxalamides, particularly
the work of Karle and Ranganathan,6 who have studied a series
of retro-bispeptides based upon the oxalamide functionality.
Figure 2. This diagram shows the self-assembly of a symmetric
dicarboxylic acid substituted urea molecule. As each intermolecular
interaction is considered, the corresponding symmetry operators can
be combined to yield the corresponding symmetry group. Single
molecules have C2 point group symmetry. Simple translation yields a
linear R-network of P2 rod symmetry via urea hydrogen bonds.
Neighboring R-networks self assemble into a â-network with P2/c
symmetry by forming cyclic carboxylic acid dimers about centers of
symmetry. Finally neighboring layers pack in an offset matter using a
centering translation to give a crystal with C2/c space group symmetry.
from aqueous solutions. In previously reported work, we
prepared and structurally characterized the lower members of
the urea series, compounds 1u-3u, the glycylglycine compound,
6u, as well a representative unsymmetrical urea, 7u.3a The
syntheses and structure determinations of the urea compounds
4u and 5u as well as oxalamide compounds 1o and 3o-7o have
now been completed. The crystal strucutrure of 2o has been
determined previously.7
Results
Symmetrical ureas (1u-6u) and oxalamides (1o-6o) were
prepared and crystals suitable for X-ray diffraction were grown
(5) Brock, C. P.; Schweizer, W. B.; Dunitz, J. D. J. Am. Chem. Soc.
1991, 113, 9811-9820.
(6) 6(a) Karle, I. L.; Ranganathan, D.; Sha, K.; Vaish, N. K. Int. J. Peptide
Protein Res. 1994, 43, 160-165. (b) Karle, I. L.; Ranganathan, D. Int. J.
Peptide Protein Res. 1995, 46, 18-23. (c) Karle, I. L.; Ranganathan, D.
Biopolymers 1995, 36, 323-331.
(7) Shkol’nikova L. M.; Gasparyan A. V.; Poznyak A. L.; Bel’skii V.
K.; Dyatlova N. M. Koord. Khim. 1990, 16, 50-54.

88 J. Am. Chem. Soc., Vol. 119, No. 1, 1997
Coe et al.
Figure 4. The â-network structures formed by compounds 4u (at the
top) and 5u (at the bottom). The urea functionality of each molecule
lies on a crystallographic 2-fold axis. The carboxylic acid functionalities
dimerize about inversion centers.
Figure 3. This diagram shows the self-assembly of a symmetric
dicarboxylic acid substituted oxalamide molecule. As each intermo-
lecular interaction is considered, the corresponding symmetry operators
can be combined to yield the corresponding symmetry group. Single
molecules have Ci point group symmetry. Simple translation yields a
linear R-network of P1h rod symmetry via oxalamide hydrogen bonds.
Neighboring R-networks self-assemble into a â-network with P1h
symmetry by forming cyclic carboxylic acid dimers about centers of
symmetry. Finally neighboring layers pack via simple translation to
give a crystal with P1h space group symmetry.
Symmetrical Urea Dicarboxylic Acids. Symmetrical di-
substituted ureas usually crystallize with retention of the
molecular 2-fold axis, thus in the crystal, they possess C2 point
group symmetry. This is illustrated by the structures of ureas
4u and 5u shown in Figure 4. Each of these molecules forms
a layer structure in accordance with the scheme outlined in
Figure 2. The presence of the urea functionality leads to a
characteristic spacing along the 2-fold axis of each layer (4.69
Å in 4u, 4.65 Å in 5u). The nearly planar layers of the two
structures are quite similar to the layer structure found earlier
for compound 3u.3a Compound 1u crystallizes in the same C2/c
space group and has the same intermolecular interactions and
the same supramolecular structure, but because of a different
molecular conformation, it forms distinctly pleated layers. The
pleating of the layers in 1u is likely due to steric interactions
between the amide and carboxylic acid functional groups that
force the molecular structure into a gauche conformation
incompatible with a more planar layer structure. The nearly
planar layer structure found for the higher members of the series,
3u-5u, would maximize the van der Waals interactions between
the layers. The remaining member of the series, 2u, is unusual.
Although it has a similar supramolecular structure and also
crystallizes in the space group C2/c, it has three independent
molecules in the unit cell. Thus, the first two members of the
1u-5u series have somewhat unusual structures, while the three
Figure 5. The â-network structures formed by compounds 4o (at the
top) and 5o (at the bottom). The oxalamide functionality of each
molecule lies on a inversion center. The carboxylic acid functionalities
dimerize about inversion centers.
higher members form a isostructural series. Since there is no
obvious problem with the molecular packing in these structures,
it is reasonable to predict that symmetrical ureylene dicarboxy-
lates with even longer side chains would have similar structures.
Symmetrical Oxalamide Dicarboxylic Acids. The struc-
tures of the five compounds 1o-5o also form an isostructural
series with one exception, compound 3o. The symmetrical
oxalamides are centrosymmetric molecules. In an isostructural
series they crystallize in the space group P1h and retain their
center of symmetry as they crystallize forming layers in
accordance with the scheme illustrated in Figure 3. The nearly
planar layer structures of the two higher members of the series
4o and 5o are shown in Figure 5. The first member of the series,
compound 1o, forms a similar layer, but like the first member
of the urea series, compound 1u discussed above, compound

Molecular Symmetry and the Design of Molecular Solids
J. Am. Chem. Soc., Vol. 119, No. 1, 1997 89
Figure 8. The â-network structure formed by compound 6o which
has glycylglycine derived side chains. The structure is similar to those
shown in Figures 5 and 6, but an extra set of amide hydrogen bonds
forms between the layers. These extra hydrogen bonds are not shown
in the figure but the protruding amide hydrogen and oxygen atoms
can be seen.
Figure 6. The â-network formed by compound 1o, polymorph A.
Compare this â-network to the designed â-network shown in Figure
3.
Figure 7. Compound 3o does not form the designed â-network. Instead
of the expected like to like, acid to acid, amide to amide hydrogen
bonds there is a like to unlike pattern that results in a large 20-member
ring held together by amide hydrogen to acid carbonyl hydrogen bonds.
1o forms a pleated layer structure due to its molecular
conformation, Figure 6.
Compound 3o is an exception. Instead of forming the
designed layer structure as outlined in Figure 3, it forms a
complex three-dimensional structure in which the anticipated
like to like, acid to acid, amide to amide hydrogen bonds are
replaced by like to unlike, acid to amide, and amide to acid
hydrogen bonds. The molecules are centrosymmetric, but their
characteristic hydrogen bond motif now includes a large 20-
membered ring held together by amide hydrogen to acid
carbonyl hydrogen bonds, Figure 7. These large rings are linked
together via acid hydrogen to amide carbonyl hydrogen bonds.
The Oxalamide of Glycylglycine. The urea of glycylglycine,
6u, was found to form a two-dimensional â-network in complete
accordance with the scheme outlined in Figure 2.3a The amide
functionality of each side arm turned sideways and formed
additional amide-amide hydrogen bonds between neighboring
â-networks and united the entire structure into a complex
γ-network. The oxalamide of glycylglycine, 6o, forms an
analogous arrangement, Figure 8, consistent with the scheme
outlined in Figure 3. Again the amide groups turns sideways
and form additional amide-amide hydrogen bonds to translation
related amide groups in the next layers of the structure.
An Unsymmetrical Oxalamide. Oxalamides, 1o-6o, all
have symmetrical substituents and crystallize with a molecular
inversion center. Oxalamide, 7o, is unsymmetrical with one
acetic acid substitutent and one propionic acid substituent, and
thus cannot have an inversion center. However, the molecule
still forms a layered â-network structure, Figure 9, one with
less symmetry than the layer shown in Figure 3. The oxalamide
R-network lacks an inversion center and has only P1 symmetry.
The inversion center at the carboxylic acid functionality still
exists giving an overall layer symmetry of P1h, but with fewer
Figure 9. The two independent â-networks formed by the two
independent molecules of compound 7o. The main differences between
the two layers are the dihedral angles within the propionic acid side
chains. These two layer alternate in the crystal.
inversion centers per molecule. There are actually two inde-
pendent molecules in the asymmetric unit of the crystal, with
each of the two molecules forming an independent â-network
of P1h layer symmetry. The only significant difference between
the two molecules is the orientation of the propionic acid side
chain, Figure 9. It should be noted that unsymmetrical urea,
7u, did not form a layered â-network, it formed a complex three-
dimensional γ-network with the expected complementarity of
the hydrogen bond network completely broken.8
The Unexpected Polymorph. One fascinating aspect of
molecular crystallography is the unexpected appearance of a
second polymorph of a well-studied compound. Chronologi-
cally the first compound studied among those reported here was
the simplest member of the oxalamide series, compound 1o.
The molecule was chosen for study to test the design scheme
shown in Figure 3. The fact that the observed structure was in
accordance with the design led us to prepare and determine
structures of the remaining members of the oxalamide series.
Subsequently in connection with another project, a second
polymorph of 1o was observed. It was found that while
crystallization from acidic aqueous solution gives the original
polymorph, A, crystallization from neutral or basic solutions
(8) In 7u the hydrogen of the carboxylic acid forms a hydrogen bond to
the oxygen atom of the urea carbonyl, while the hydrogen atoms of the
urea form hydrogen bonds to the acid carbonyl. See ref 3b.

90 J. Am. Chem. Soc., Vol. 119, No. 1, 1997
Coe et al.
designed interaction should be greater than alternate arrange-
ments. The second criterion is a crystallographic one. Mol-
ecules pack in a crystal only as allowed by the rules of
crystallographic symmetry.
An individual molecule in a crystal will have many different
intermolecular contacts. Commonly, there will be about twelve
nearest neighbors corresponding to a close-packed environment
of spheres. Each pair of adjacent molecules will be related to
each other by a symmetry operator. Usually about half of these
nearest-neighbor contacts will correspond to unique symmetry
operations. In most cases only about three of these distinct
symmetry operators are needed to generate the symmetry of
the crystal. The remaining symmetry operators can be derived
from these generating operators.
Figure 10. The â-network of polymorph B of compound 1o. This
layer structure is formed by acid to amide hydrogen bonds.
Various hydrogen bonding functionalities will be compatible
with certain symmetry operators and incompatible with others.
Indeed some functional groups strongly favor certain operators.
Consider the two secondary amides shown below, one a cyclic
molecule with a syn hydrogen and the second an acyclic
molecule with a anti hydrogen. With a syn hydrogen cen-
trosymmetric dimer formation about an inversion center will
be favored. But with an anti hydrogen an inversion center is
an incompatible operator and a one-dimensional R-network will
be favored using a translational symmetry operator. The urea
and oxalamide functionalities are each unique amide forms that
have their own symmetry characteristics as we will discuss
below.
Figure 11. The â-network of compound 8. This layer structure is
isostructural to that of polymorph B of compound 1o, Figure 10. This
figure was generated from the original literature coordinates.6c
gives instead a second polymorph, B. Polymorph B has a
completely unanticipated structure, Figure 10. The hydrogen
bond pattern in B consists of like to unlike, acid to amide, and
amide to acid hydrogen bonds instead of the like to like pattern
of polymorph A. The molecules of B have assembled into a
â-network, but it is quite different that the â-network of A. The
layer symmetry of B is P21/a with neighboring centrosymmetric
molecules related by a screw axis and a glide plane. Interest-
ingly, the â-network of B does not contain independent
R-networks. The â-network of A contained two independent
R-networks, one based upon the oxalamide hydrogen bonds,
the second roughly perpendicular to the first based upon the
carboxylic acid hydrogen bonds.
To design a supramolecular structure one must predict the
intermolecular interactions that will take place between the
molecules. If the symmetry operators corresponding to these
interactions can be anticipated as well, then these operators can
be combined following the rules of group theory to generate
the symmetry group of the anticipated supramolecular structure.
Knowing the most probable symmetry groups of the supramo-
lecular structure can lead to a correct prediction of the space
group of the crystal.
Ureas. The urea functionality has two hydrogen donor
nitrogen atoms and only one carbonyl acceptor, but the one
carbonyl commonly accepts both hydrogen bonds, one at each
lone pair position. Symmetrical ureas have a 2-fold axis through
the carbonyl and importantly can retain this 2-fold when they
form a one-dimensional hydrogen bonded R-network via simple
crystallographic translation, Figure 2. Using group theory one
would say that when a translation axis is added to the C2 point
group (the urea monomer), the P2 rod group (the urea
R-network) is generated. The carboxylic acid residues on the
ends of the side chains characteristically dimerize about crystal-
lographic inversion centers giving a two-dimensional â-network.
Adding an external inversion center to the P2 rod group
generates the P2/c layer group. The â-network layers stack on
top of each other to form the crystal; with no further hydrogen
bonds available presumably the layers are held together by
mainly van der Waals interactions. One could imagine two
simple ways for this stacking to occur. If the layers were
Discussion
A successful design of a molecular solid depends upon the
identification and synthesis of molecular functionalities that will
predictably and persistently lead to specific favorable intermo-
lecular interactions.9 Hydrogen bonds are ideal since they are
generally the strongest available intermolecular interactions in
an organic crystal and, with good design, they can be as highly
specific as Watson-Crick hydrogen bonds. With this principle
in mind we have been seeking to develop a library of
functionalities that will allow one to design and synthesize
supramolecular assemblies with desired properties and structure.
The first criterion in choosing a hydrogen bonding functional-
ity is a chemical one. The number of hydrogen donor positions
should match the number of acceptors and the strength of the
(9) Etter, M. C. Acc. Chem. Res. 1990, 23, 120-126.

Molecular Symmetry and the Design of Molecular Solids
J. Am. Chem. Soc., Vol. 119, No. 1, 1997 91
directly on top of each other using a simple translation operator,
then the space group of the final molecular assembly and the
crystal would be P2/c. However, it seems more likely that the
layers would be offset such that the molecules of one layer fall
over the groves between the molecules of the layer below. This
would generate a centering operation and the final space group
would be C2/c.
Five of the six ureas 1u-6u do indeed crystallize in the space
group C2/c in exact agreement with the above analysis. The
urea functionalities lie on a 2-fold axis of the C2/c space group,
and the carboxylic acid groups dimerize about inversion centers.
The exception, 2u, has three independent molecules in the unit
cell, but the molecular packing is very similar and still in the
C2/c space group.
Compound 7u has differing side arms and lacks the 2-fold
axis of molecules 1u-6u. It thus cannot follow the C2/c
scheme, but could instead form a â-network with lower
symmetry. Computer modeling suggested a â-network of P1
layer symmetry formed only by the simple translation operator.
However, the actual structure showed a complete breakdown
of the design and a complex three-dimensional γ-network of
hydrogen bonds.8 Several other ureas lacking the 2-fold axis
were also studied in our earlier work.3b These molecules were
chiral, but racemic, and they uniformly failed to form a layer
structure. Instead they tended to form centrosymmetric dimers
about crystallographic inversion centers via unpredicted hydro-
gen bond patterns. However, when a single enantiomer was
used, the inversion center was symmetry forbidden and a simple
translation based â-network was formed as designed.
hydrogen, the one on the carboxylic acid, and the most basic
carbonyl, the one on the amide. Obviously this cannot be an
absolute rule since all compounds that follow the design in
Figure 3 violate the rule, but it does point out the nature of one
competing force.
The one example of an unsymmetrical oxalamide, 7o, forms
a â-network with simple translation along the oxalamide
R-network and inversion centers along the carboxylic acid
R-network. In this case the oxalamide design is more successful
than the urea design, since the urea 7u did not form a layered
structure.
The Second Polymorph. A crystal of a second polymorph
B of 1o was prepared and its structure determined. Like
compound 3o, polymorph B contains a network of acid to amide,
like to unlike hydrogen bonds, Figure 10. The oxalamide
functionality retains its inversion center and forms two planar
nine-member hydrogen-bonded rings. The molecules form a
two-dimensional â-network of P21/a layer symmetry. The
structure is generated by the addition of an external 21 screw
axis to single molecules of Ci point group symmetry.11 The
structure appears to be compact and has a higher calculated
density than the original structure.
These two polymorphs can be prepared selectively and
reproducibly. Polymorph A, the original structure with like to
like hydrogen bonds, Figure 6, precipitates from acidic solutions.
Polymorph B, the second polymorph with like to unlike
hydrogen bonds, precipitates from pure water or basic solutions.
The existence of the 1o polymorphs illustrates both a
difficulty and the power of crystal engineering. We are dealing
with simple functional groups. They can in general be expected
to form certain hydrogen bond patterns. However, there will
always be competing structures that are only slightly different
in energy. Since crystallization is governed by kinetics as well
as thermodynamics, predictions of crystal structures will always
be difficult. The existence of polymorphism both compounds
and illustrates the problem. The preparation of the second
polymorph of compound 1o was not anticipated. It has a
structure that differs from the design of Figure 3. On the other
hand when we look at compound 3o, the lone exception in the
oxalamide series, we can conjecture that one might also be able
to prepare a second polymorph of that compound, a polymorph
that would have a layered structure in accordance with the design
of Figure 3. Unfortunately, there is no universal scheme for
preparing polymorphs.12
Thus, with ureas we can conclude that the 2-fold axis is an
important part of the layer design. All the symmetrical urea
molecules with a 2-fold axis formed â-networks in accordance
with the design. Ureas with differing side chains, lacking the
2-fold axis, tend to form alternate hydrogen bond patterns about
centers of inversion unless forbidden by symmetry.
Oxalamides. With the oxalamide functionality the inversion
center is part of the design. The oxalamide functionality is
centrosymmetric and with symmetrical side arms the molecular
symmetry will be Ci. Most importantly the oxalamide-
oxalamide hydrogen bond also forms about an inversion center.
Adding an external inversion center to a molecule that already
has one generates a translation axis, Figure 3, and an R-network.
Adding a second external inversion center corresponding to the
carboxylic acid dimerization produces a â-network. The two
independent R-networks have P1h rod group symmetry, the layer
has P1h layer group symmetry, and the crystal has P1h space group
symmetry. Since the inversion center seemed to be so favorable
in the urea compounds, even to the extent of destroying the
expected hydrogen bond complementary, one might expect that
the oxalamide hydrogen bonds based upon the inversion center
would be highly persistent.
Of the six symmetrical oxalamides studied, 1o-6o, five of
them form P1h structures in accordance with the design in Figure
3. The exception, 3o, has a breakdown in the hydrogen bond
complementary and forms a large centrosymmetric ring with
hydrogen bonds occurring between the oxalamide and the
carboxylic acid functionalities. The occurrence of this like to
unlike hydrogen bond is not in accordance with the design, but
perhaps not totally surprising either. Others10 have examined
the question of like to like versus like to unlike hydrogen bonds
in acid-amide molecules. They have concluded that the
strongest hydrogen bond should occur between the most acidic
Other Oxalamide Structures. An examination of the
Cambridge Structural Data Base13 and the recent literature
reveals nine additional examples of well-characterized, metal
free, acyclic oxalimides. The three examples with simple
symmetrical substituents, N,N′-dimethyloxalamide,14 N,N′ -bis-
(pentafluorophenyl)oxalamide,15 and N,N′-bis(3-nitratopropyl)-
oxalamide,16 all crystallize to form the characteristic primary
P1h R-network discussed above and shown in Figure 3. This
(11) It is interesting to note that in polymorph A of 1o the â-network
can be considered to be an assembly of independent R-networks, the
R-networks in turn being assemblies of discrete molecules. However, in
polymorph B of 1o the â-network is assembled directly from discrete
molecules, and there are no symmetry independent R-networks.
(12) We have tried the obvious experiment of growing crystals of 3o
from solutions with various pH values, but we have seen no evidence of
polymorphism.
(13) Allen, F. H.; Taylor, R.; Kennard, O. Acc. Chem. Res. 1983, 16,
146-153.
(14) Klaska, K. H.; Jarchow, O.; Scham, W.; Widjaja, H.; Voss, J.;
Schmalle, H. W. J. Chem. Res. 1980, 104, 1643-1700.
(15) Yamaguchi, K.; Matsumura, G.; Haga, N.; Shudo, K. Acta Crys-
tallogr., Sec. C 1992, 48, 558-559.
(10) Berkovitch-Yellin, Z.; Leiserowitz, L. J. Am. Chem. Soc. 1983, 105,
765-767. Leiserowitz, L.; Nader, F. Acta Crystallogr., Sect. B 1977, 33,
2719-2723. Leiserowitz, L. Acta Crystallogr., Sect. B 1976, 32, 775-
802.
(16) Bhattacharjee, S. K.; Ammon, H. L. Acta Crystallogr., Sec. B 1982,
38, 2503-2505.

92 J. Am. Chem. Soc., Vol. 119, No. 1, 1997
Coe et al.
Table 1. Crystallographic Parameters
1o (form A) 1o (form B)
3o
4o
5o
6o
7o
4u
5u
emperical formula C6H8N2O6 C6H8N2O6
C10H16N2O6
260.2
C12H20N2O6 C14H24N2O6 C10H14N4O8 C7H10N2O6 C11H20N2O5 C13H14N2O5
formula wt
204.1
204.1
288.3
316.4
318.2
218.2
260.3
278.3
a (Å)
b (Å)
c (Å)
4.446(2)
5.020(2)
10.853(6) 10.192(5)
94.19(4)
91.19(3)
115.72(3)
218.2(4)
5.037(2)
7.911(3)
5.877(5)
9.864(4)
10.556(6)
5.1205(4) 5.183(2)
5.3932(3) 6,701(3)
12.692(7) 12.354(3) 16.445(7) 24.194(9) 10.144(13)
97.567(8) 102.79(1) 96.69(3)
92.090(10) 98.72(1) 90.94(3)
94.249(8) 102.43(1) 118.02(2) 83.88(3)
346.2(1) 399.6(3) 342.4(5) 969.1(3)
4.761(2)
5.004(2)
5.069(3)
8.026(3)
26.903(24)
4.6919(6)
13.206(2)
4.648(3)
25.084(2)
R
83.63(3)
84.74 (3) 99.90(4)
â
100.28(3)
104.13(11)
593.4(3)
103.73(1)
1496(1)
γ
volume (Å3)
space group
Z
399.6(3)
1261.3(5)
P1h(No. 2) P21/c (No. 14) P21/c (No. 14) P1h(No. 2) P1h(No. 2) P1h(No. 2) P1h(No. 2) C2/c (No. 15) C2/c (No. 15)
1
1.553
refined
915
2
2
1
1
1
4
4
4
d
calc (g/cm3)
1.696
refined
1306
1.456
refined
1242
1.383
refined
1368
1.315
fixed calc refined
1735
1.543
1.495
fixed calc fixed calc
1612
1.371
1.236
fixed calc
hydrogens
observations
collected (total)
observations
(I > 3σ)
variables
R
1337
813
891
1018
921
1139
984
808
634
665
81
0.051
0.074
81
.065
.081
115
0.034
0.056
132
0.055
0.067
100
0.064
0.093
129
0.054
0.065
271
0.066
0.069
83
0.045
0.055
92
0.071
0.087
Rw
gives further support to the primary oxalamide R-network as a
reliable design element. However, due to a variety of problems
none of the remaining examples of oxalamides from the
literature were found to form this simple R-network. The first
of these exceptions is the sterically hindered N,N′-bis(2,2,2-
trinitratoethyl)oxalamide17 which forms only intramolecular
hydrogen bonds. The next is an unsymmetrical oxalamide with
a large benzodiazepin substituent.18 It forms an alternate
hydrogen bonded network involving the substituent.
The remaining four literature oxalamides were reported in a
series of papers by Karle and Ranganathan.6 They report the
structures of three dimethyl esters of oxalamido retropeptides,
MeO-Aib-COCO-Aib-OMe, MeO-Leu-COCO-Leu-OMe, and
MeO-Ser-Leu-COCO-Leu-Ser-OMe. Each of these three struc-
tures form alternate networks that involve hydrogen bonds
between an oxalamide unit and the hetero atoms of side arm
substituents. The latter two compounds are of course chiral
and are prevented by symmetry from forming a P1h R-network.
The fourth compound from Karle and Ranganathan6c is the
one most closely related to the present work. This is compound
8, the oxalamide derived from the amino acid dimethylglycine.
Characteristic Spacing. An approach to supramolecular
synthesis based upon a combination of chemical insight and
symmetry considerations has a high probability of success. The
urea functionality has a characteristic repeat spacing of about
4.65 Å along the urea R-network, Table 1. This means that
the molecules in a â-network will be spaced at regular intervals
equal to this approximate distance. For example, in Figure 4
the molecules 4u and 5u are spaced at 4.69 and 4.65 Å,
respectively. The oxalamide functionality also has a charac-
teristic spacing, about 5.1 Å, longer than the value found in the
ureas. The longer distance is due to the fact that the C-O‚‚‚H
bond angle is nearly linear in the oxalamide case, but is bent in
the urea case. These two different characteristic spacings are
quite reliable and can serve as the basis for controlling
topochemical reactions such as diacetylene polymerizations.3d
Conclusions. A symmetry analysis based upon the structure
of simple molecules and their anticipated intermolecular interac-
tions can lead to successful predictions of molecular packing
and crystal symmetry. Both the urea and oxalamide dicarboxy-
lic acids form designed â-networks with a significant degree
of reliability. The urea designs are quite reliable when there is
a molecular 2-fold axis, but competing hydrogen bond patterns
are found when less symmetrical molecules are studied. The
oxalamide design based on inversion centers is also quite
reliable, with the designed layer structure found in most cases.
In the exceptions the designed like to like patterns of acid-
acid and amide-amide hydrogen bonds are replaced by like to
unlike acid-amide and amide-acid hydrogen bonds. The urea
and oxalamide functionalities are thus both good entries in our
library of functionalities useful for designed supramolecular
syntheses. They have different symmetry properties and give
different characteristic spacings, 4.6 Å for the urea and 5.1 Å
for the oxalamide functionality.
Compound 8 is a direct analogue of compound 1o, the
oxalamide of glycine. Most interestingly the crystal structure
of 8 is isostructural with that of polymorph B of 1o. It
crystallizes in the same P21/c space group with somewhat similar
cell constants.19 The compound forms a â-network based upon
acid to amide hydrogen bonds completely analogous to the one
found in polymorph B of 1o. The â-network of 8 is analogous
to the â-network of polymorph B of 1o and not to the designed
â-network found in polymorph A. It is possible that the steric
hindrance of the four methyl groups of 8 would perhaps preclude
a â-network structure analogous to the one in polymorph A.
Experimental Section
X-ray Diffraction Studies. Crystals were obtained as described
below, selected, and mounted on glass fibers using epoxy cement. The
crystals were optically centered on an Enraf Nonius CAD4 diffracto-
meter and X-ray data were collected using graphite-monochromated
Mo radiation. The unit cells were determined by a least-squares analysis
of the setting angles of 25 high-angle reflections. Data were collected
as indicated in Table 2, and the structures were solved and refined
using the TEXSAN crystallographic program package of the Molecular
Structure Corporation. The quality of the structures varied due to the
quality of the available single crystals. In the best cases, hydrogen
atoms were located in difference maps and refined. In the cases with
(17) Chen, R.; Rheingold, A. L.; Brill, T. B. J. Crystallogr. Spectrosc.
1991, 21, 173-177.
(18) Fryer, R. I.; Earley, J. V.; Blout, J. F. J. Org. Chem. 1977, 42,
2212-2219.
(19) Compound 8 was found (ref 6c) to crystallize in the space group
P21/c with Z ) 2 and cell constants of a ) 6.342(2) Å, b ) 9.910(2) Å,
c ) 10.020(2) Å, and â ) 91.65(2)°. These values are somewhat similar to
those found for 1o (polymorph B) as shown in Table 1.

Molecular Symmetry and the Design of Molecular Solids
J. Am. Chem. Soc., Vol. 119, No. 1, 1997 93
poorer data sets, the hydrogen atoms were placed in calculated positions
and added as fixed contributions.
4,4′-(Oxalyldiimino)dibutyric Acid (3o). Using general method
B with 4-aminobutyric acid gave 4,4′-(oxalyldiimino)dibutyric acid as
colorless needles in 20% yield: mp 214-217 °C (lit.24 mp 210-215
°C).
General Method A. The following procedure, adapted from the
literature,20 was used to prepare the ureylene dicarboxylic acids
described in this paper. To a solution of 20 mmol of the amino acid
in 4 mL of water was added 0.40 mL of 6.25 N NaOH. This solution
was placed in an ice bath and 0.88 g (3.3 mmol) of triphosgene in 5
mL of toluene was added. The reaction mixture was stirred in an ice
bath for 30 min and at room temperature for 2 h. The reaction mixture
was cooled in an ice bath, 5 mL of water was added, and the pH was
adjusted to approximately 2 with 6 N HCl. The urea precipitate was
collected by filtration and washed with a small amount of cold water.
5,5′-Ureylenedipentanoic Acid (4u). Using general method A with
5-aminopentanoic acid, 5,5′-ureylenedivaleric acid was prepared in 20%
yield: mp 188-189 °C (lit.21 mp 189 °C). Recrystallization from hot
water gave needles of good quality for X-ray crystallography.
6,6′-Ureylenedihexanoic Acid (5u). Using general method A with
6-aminohexanoic acid, 6,6′-ureylenedihexanoic acid was prepared in
56% yield: mp 162-163 °C (lit.22 mp 162-163 °C). Recrystallization
from hot water gave needles of adequate quality for X-ray crystal-
lography.
General Method B. The following procedure, adapted from the
literature,23 was used to prepare the symmetrical oxalamides described
in this paper. To a stirred solution of 3.73 mL (27.5 mmol) of ethyl
oxalate in 50 mL of water at 40 °C was added 55 mmol of the amino
acid. The reaction mixture was stirred for 2 h, allowed to cool to room
temperature, and acidified to pH = 2 with concentrated HCl. The
precipitated product was collected by filtration and air dried. Crystals
suitable for crystallographic analysis were obtained by recrystallization
from water, unless otherwise specified.
N,N′-Oxalyldiglycine (1o, polymorphs A and B). Using general
method B with glycine gave N,N′-oxalyldiglycine in 17% yield as
colorless plates from the original reaction mixture: mp (with decom-
position, turning red) 250-255 °C (lit.23 mp 250-255 °C). These
needles were subjected to single crystal X-ray analysis and are given
the designation polymorph A. Recrystallization of polymorph A (plates)
from water gave crystals with a block-like habit. Single-crystal X-ray
diffraction of these crystals showed that they were a second polymorph,
polymorph B: mp (with decomposition) 245-250 °C. Subsequent
experiments showed that samples of either form of 1o recrystallized
from solutions of pH 7 or higher (0.3 M NaOH or aqueous NH3)
consistently gave polymorph B, while recrystallization from acidic
solutions (0.3 M HCl) gave polymorph A. IR (KBr): polymorph A,
3302 (N-H), (1712, 1660) (CdO), 1434, 1407, 1237, 858; polymorph
B, 3264 (N-H), 1671 (CdO), 1406, 1191, 841.
5,5′-(Oxalyldiimino)dipentanoic Acid (4o). Using general method
B with 5-aminopentanoic acid gave 5,5′-(oxalyldiimino)dipentanoic acid
1
in 29% yield as colorless needles: mp 209-211 °C dec; H NMR
(DMSO-d6) δ 12.01 (bs, 1H), 8.74 (t, J ) 6.0 Hz, 1H), 3.12 (d, J )
5.7 Hz, 2H), 2.20 (2H), 1.45 (4H); IR (KBr) 3298 (N-H), 1699 (CdO),
1648 (CdO) cm-1
.
6,6′-(Oxalyldiimino)dihexanoic Acid (5o). Using general method
B with 6-aminohexanoic acid gave 4,4′-(oxalyldiimino)dihexanoic acid
1
as colorless needles in 30% yield: mp 178-180 °C dec; H NMR
(DMSO-d6) δ 11.9 (bs, 1H), 8.71 (t, 1H), 3.09 (qt, J ) 5.0 Hz, 2H),
2.18 (q, J ) 7.2 Hz, 2H), 1.48 (m, 4H), 1.25 (tt, J ) 8.0 Hz, 2H); IR
(KBr) 3300 (N-H), 1703 (CdO), 1647 (CdO) cm-1
.
N, N′-Oxalyldiglycylglycine (6o). Using general method B with
glycylglycine gave N,N′-oxalyldiglycylglycine colorless needles in 32%
1
yield: mp 262-267 °C dec; H NMR (DMSO-d6) δ 8.82 (t, J ) 6.0
Hz, 1H), 8.28 (t, J ) 5.5 Hz, 1H), 3.81 (d, J ) 6.0 Hz, 2H), 3.76 (d,
J ) 5.6 Hz, 2 H); IR (KBr) 3289 (N-H), 1708 (CdO), 1649 (CdO)
cm-1
.
Oxalyl-N-glycyl-N′-â-alanine (7o). To 80 mL of THF was added
ethyl oxalylchloride (5.60 mL, 50 mmol) and glycine ethyl ester
hydrochloride (6.98 g, 50 mmol). The solution was stirred and allowed
to reflux for 90 min. The solvent was removed in Vacuo to give the
intermediate diester as an oil. A basic solution was prepared from 10
mL of water and 3.2 mL of 6.25 N NaOH. To this solution was added
4.06 g (20 mmol) of the above diester and 1.78 g (20 mmol) of
â-alanine. The reaction mixture was stirred for 1.3 h. The intermediate
monoester was directly hydrolyzed by addition of 2 mL of 6.25 N
NaOH. The solution was acidified with an ion exchange resin in its
acid form (Rexyn 101H, Fisher Scientific) until pH = 4. Filtration of
the resin followed by removal of the solvent gave 3.15 g (32%) of a
colorless precipitate which was purified by recrystallization from
1
methanol: mp 228-229 °C; H NMR (DMSO-d6) δ 8.92 (t, J ) 6.0
Hz, 1H), 8.78 (t, J ) 6.0 Hz, 1 H), 3.80 (d, J ) 7.0 Hz, 2 H), 3.35 (q,
J ) 7.0 Hz, 2 H), 2.48 (t, J ) 7.0 Hz, 2 H); IR (KBr) 3300 (N-H),
1711 (CdO), 1650 (CdO) cm-1
.
Acknowledgment. This work is supported by the National
Science Foundation under Grant No. CHE-9307947.
Supporting Information Available: X-ray crystallographic
data for the nine compounds listed in Table 1 including tables
of atomic coordinates, temperature factors, bond distances and
angles, intermolecular contacts, and hydrogen-bond parameters
(26 pages). See any current masthead page for ordering and
Internet access instructions.
(20) Kondo, K.; Murata, K.; Miyoshi, N.; Murai, S.; Sonoda, N. Synthesis
1979, 735-736.
(21) McKay, A. F.; Tarlton, E. J.; Petri, S. I.; Steyermark, P. R.; Mosley,
M. A. J. Am. Chem. Soc. 1958, 80, 1510-1517.
(22) Giesemann, H.; Oertel M. J. Prakt. Chem. 1959, 4(8), 292-297.
(23) Hearn, W. R.; Hendry, R. A. J. Am. Chem. Soc. 1957, 79, 5213-
5217.
JA961958Q
(24) Wieland; Hasse, Kurt, Chem. Ber. 1960, 1686-1692.