Single-Crystal X-ray Diffraction Structure of the Stable Enol Tautomer Polymorph of Barbituric Acid at 224 and 95K

Article abstract of DOI:10.1002/anie.201508078

The thermodynamically stable enol crystal form of barbituric acid, previously prepared as powder by grinding or slurry methods, has been obtained as single crystals by slow cooling from methanol solution. The selection of the enol crystal was facilitated by a density-gradient method. The structure at 224 and 95K confirms the enol inferred on the basis of powder data. The enol has bond lengths that are consistent with the expected bond order and with DFT calculations that include treatment of hydrogen bonding. In isolation, the enol is higher in energy than the tri-keto form by 50kJ mol^-1 which must be more than compensated by enhanced hydrogen bonding. Both crystal forms have four normal H-bonds; the enol has two additional H-bonds with O-O distances of 2.49?. Conversion into the enol form occurs spontaneously in the solid state upon prolonged storage of the commercial tri-keto material. Slurry conversion of tri-one to enol in ethanol is reversed in direction in ethanol-D₁.

Full text of DOI:10.1002/anie.201508078

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201508078
German Edition: DOI: 10.1002/ange.201508078
Polymorphism
Single-Crystal X-ray Diffraction Structure of the Stable Enol Tautomer
Polymorph of Barbituric Acid at 224 and 95 K
Madalynn G. Marshall, Valerie Lopez-Diaz, and Bruce S. Hudson*
Abstract: The thermodynamically stable enol crystal form of
barbituric acid, previously prepared as powder by grinding or
slurry methods, has been obtained as single crystals by slow
cooling from methanol solution. The selection of the enol
crystal was facilitated by a density-gradient method. The
structure at 224 and 95 K confirms the enol inferred on the
basis of powder data. The enol has bond lengths that are
consistent with the expected bond order and with DFT
calculations that include treatment of hydrogen bonding. In
isolation, the enol is higher in energy than the tri-keto form by
50 kJmolÀ1 which must be more than compensated by
enhanced hydrogen bonding. Both crystal forms have four
normal H-bonds; the enol has two additional H-bonds with O–
O distances of 2.49 Š. Conversion into the enol form occurs
spontaneously in the solid state upon prolonged storage of the
commercial tri-keto material. Slurry conversion of tri-one to
enol in ethanol is reversed in direction in ethanol-D1.
Figure 1. Relevant tautomeric forms of barbituric acid, keto (1) and
enol (2).
experiments in methanol, ethanol, and butanol showed that
this alternative tautomeric polymorph is the thermodynami-
cally stable form.[9]
In that work cooling and evaporation methods of crystal
growth and several solvents, initial temperatures, initial
concentrations, and cooling rates were employed. The density
of the enol tautomer from the X-ray powder data is
1.669 gcmÀ3.[9] The density of the tri-keto tautomer at room
temperature[4] is 1.589 gcmÀ3. This distinct density difference
permits easy detection of candidate tautomers using density
gradient analysis.[10] A mixture of tetrabromoethane and
tetrachloroethylene was used for the high-density bottom of
the gradient and a mixture of tetrachloroethylene and
benzene as the low-density upper limit. A standard gradient
maker consisting of two cylindrical glass containers connected
via a stopcock at the bottom with one containing a stir bar and
an effluent line via a second stopcock produces a linear
gradient.[10] The resolution of a density gradient can be
expanded by using a shallow gradient in a long column.
A solution of 1 g of our sample of commercial barbituric
acid (see below) in 175 mL of methanol was heated to 608C
and then cooled to 208C over four days and placed in
a refrigerator at 48C. This is 3–4-fold below saturation at
608C. Crystals of this material had only one location in the
density gradient indicating that only one polymorphic form
was present with a density close to the value of the enol
form.[9] Crystal growth from ethanol solution using a similar
procedure resulted in crystals with the same density as those
from methanol. They co-located in the same density gradient.
The single-crystal diffraction structure at 224 K and at
95 K were determined and found to be the enol.[9,11,12] The
224 K structure P21/n, Z = 4 has a density of 1.661 gcmÀ3 and
T
he enol tautomer of barbituric acid (Figure 1) is an example
of tautomeric polymorphism. Tautomeric polymorphism is
defined[1,2] as a pair of crystals each with a distinct tautomer.
A 2011 systematic survey of the CSD showed that as of that
date there were only 16 cases of this phenomenon.[1] There are
many more cases of two tautomeric forms appearing in the
same crystal. The related case of 2-thiobarbituric acid adds to
this tally with the same keto/enol tautomeric pair.[3]
The reported single-crystal structures of barbituric acid,
including its hydrates, are all in the tri-keto tautomer (1).[4–7]
Ball-mill grinding experiments showed that a tautomeric form
can be prepared in this fashion.[8] The resulting material was
investigated by solid-state NMR and vibrational spectrosco-
py. An obvious feature of the Raman spectrum of the new
polymorphic form was the relative absence of the character-
=
istic C O stretch vibrations. On this basis, it was concluded
that the most likely possibility was the tri-hydroxy aromatic
tautomer[8] as noted in Ref. [2]. It was, however, inferred from
X-ray and neutron powder diffraction data that this tauto-
meric form is the enol structure (2; Figure 1).[9] Slurry
[*] M. G. Marshall, V. Lopez-Diaz, B. S. Hudson
Department of Chemistry, 1-014 Center for Science & Technology
Syracuse University
Syracuse NY 13244-4100 (USA)
E-mail: bshudson@syr.edu
3
a unit cell volume of 512.19 Š in agreement with that
M. G. Marshall, B. S. Hudson
reported for the X-ray powder data with a density of
1.659 gcmÀ3. The bond lengths for the room temperature
data are shown along with those of the two powder data sets
in Table 1. The numbering system follows that of Ref. [8] and
Figure 1.
Syracuse University & Department of Chemistry
Iowa State University
1605 Gilman Hall, Ames, IA 50011-3111 (USA)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201508078.
Angew. Chem. Int. Ed. 2016, 55, 1309 –1312
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1309

Angewandte
Communications
Chemie
Table 1: Enol bond lengths from data obtained near room temperature.
Bond
SCX-ray[a]
XPD[b]
NPD[c]
N1-C2
=
C2 O
C2-N3
N3-C4
C4-O
=
C4 C5
C5-C6
=
C6 O
C6-N1
1.362
1.228
1.364
1.366
1.311
1.366
1.402
1.264
1.376
1.376
1.209
1.398
1.389
1.332
1.325
1.443
1.269
1.369
1.385
1.185
1.398
1.330
1.271
1.397
1.423
1.287
1.369
[a] Single-crystal X-ray diffraction. [b] X-ray powder diffraction. [c] Neu-
tron powder diffraction.
The esd values for the 224 K single-crystal X-ray data of
Table 1 are 0.0016 Æ 0.00015 Š. (Table S5) The corresponding
values for the powder-diffraction studies are reported Fig-
ure S5 the Supporting Information of Ref. [9] as “The statisti-
cally calculated standard deviations of all bond lengths are
0.001 Š, but the actual accuracy can be estimated to be in the
order of 0.01 to 0.02 Š only.” This estimate reflects appreci-
ation of the effect of systematic error which is evident in the
last two columns of Table 1 that show the X-ray and neutron
results. The bond angle esdꢀs are similarly estimated to be 10–
20 times the statistical values (Supporting Information of
Ref [9]).
Figure 2. Heavy-atom bond lengths of the enol form of barbituric acid:
black: observed at 95 K. Red: computed from DFT B3LYP-31G(2d,2p)
for a hydrogen bonded model (see Supporting Information). Blue: as
computed for an isolated enol tautomer. Atoms: C gray, O red, N blue,
H white.
The bond lengths involving the heavy atoms, obtained
from the 95 K data set, are given in Table S11 and are shown
in Figure 2 where they are shown along with the results of two
Table 2: Hydrogen-bonding heavy atom distances [Š] in barbituric acid
crystals. Note the difference in temperature.
Enol
this work
95 K
Tri-keto
this work
95 K
Bolton[4]
RT
Lewis[5]
150 K
2.861
2.74
2.871
2.855
2.903
2.795
2.856
2.8
2.497
Figure 3. Pentamer hydrogen-bonding (broken lines) pattern for barbi-
turic acid in its enol structure. The top two units are coplanar with the
central unit; the bottom two are out of plane with a dihedral angle of
134.548.
DFT calculations both with the B3LYP functional and the 6-
31G(2d, 2p) basis set. The experimental results from the 95 K
diffraction data are in black in Figure 2. The esd values for
these numbers are 0.0014 Š. The calculations in red in
Figure 2 include neighboring hydrogen-bonding groups- as
shown in Figure S1 of the Supporting Information. The values
in blue in Figure 2 are for the same computation without
added hydrogen-bonding groups that are optimized with the
barbituric acid structure.
=
bonds NH-CO of 2.795 and 2.903 Š (Table 2) and C O-HN
of the same lengths[4] The BARBAC01 of Lewis et al. has two
[5]
H-bonds of length 2.856 Š and two of 2.800 Š If we
compare the H-bond lengths, it is clear that four of the H-
bonds are about the same in each case or slightly shorter and
thus stronger in the enol form and there is an additional
hydrogen bond in the enol form that is considerably shorter,
The hydrogen bonding of the enol tautomer of barbituric
acid in Figure 3 shows that each molecule forms four NH:O
hydrogen bonds and one hydrogen bond with a central
=
À
carbonyl group acting as a double acceptor. These are NH-O
2.497 Š. Based on its O O distance, this O-H-O hydrogen
C 2.861 and 2.740 Š, C O-HN 2.740 and COH-O C 2.497 Š
and the C O to HN (2.861 Š) and HO-C (2.497 Š) where
bond, is “strong”, as classified in standard Reviews.[13,14]
The structure determined at 95 K (Figure 2) differs
significantly from that obtained by optimizing the structure
=
=
=
bold denotes central molecule groups. In contrast the original
tricarbonyl structure of Bolton[4] BARBAC, has only four H-
in isolation. The major difference is the longer C O bond and
=
1310
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1309 –1312

Angewandte
Communications
Chemie
À
the shorter C N bond in the crystal compared to that for the
optimized structure without hydrogen bonding. This situation
is well established from previous work on peptides.[15]
A normal mode calculation for the Z = 4 unit cell shows
One of the authors of Ref. [9] has published a paper on the
subject of the relative stability of the tautomeric forms of four
ammonio benzene sulfonates.[11] The method used is the
plane-wave VASP code with the PBE functional using a cited
empirical dispersion correction and specified convergence
criteria. It is not specified if the lattice parameters are
optimized. These four materials crystallize in four different
crystal structures all in the zwitterionic tautomer form. In two
cases, the correct structure is not predicted to be the lowest by
amounts of 2.44 and 2.62 kJmolÀ1. If the same error limits
apply to the barbituric acid case, then the lattice energy
difference is uncertain to 50%. The actual uncertainty limits
must have added to the above, the difference in energy that
favors the correct structure, that is, a zero energy difference is
not predictive. There is, in addition, the systematic error in
comparison of the energy difference between two molecules
with a correlated method with the difference of two solid
forms with a DFT method.
The conversion of barbituric acid from the tri-keto into
the enol tautomeric crystal polymorph is spontaneous and
occurs slowly in the solid state. A commercial sample[17] that is
now 7 years old was found by Raman microscopy to be 80%
converted into the enol. This conversion is greatly facilitated
by grinding and slurry.[8,9] Slurry conversion does not occur if
the sample has its four exchangeable hydrogen atoms
replaced by deuterium atoms. This result suggests a role for
the zero-point level in the thermodynamics as has previously
been shown for polymorphic forms of glycine[18] and for other
cases.[19] More specifically, our commercial sample that is
predominantly the enol tautomer, if slurried in ethanol-D1 for
several days, shows conversion into the tri-keto form as
=
that the C O stretch vibrations in this structure are strongly
admixed over all four units and that the C O stretch is also
admixed with other of in-plane motions and is distributed
over a range of 150 cmÀ1. This is also the case if we use the
pentamer hydrogen-bonded structure of Figure 3. This
arrangement explains the absence of the usual characteristic
=
=
C O stretch activity in the vibrational spectra that formed the
basis of the original proposal as to the nature of this material.
The enol form of barbituric acid in isolation is less stable
than the tri-keto form. This energy difference is on the order
of 45–55 kJmolÀ1 (see Supporting Information and Ref. [9]).
This case is thus an unusual situation, especially for the lowest
energy polymorphic form. The dipole moment of the tri-one
form is 0.25 D while that for the enol is 5 D so this energy
difference will be decreased by high dielectric solvents. For
the case of a 6-31G(3d,3p) B3LYP calculation where the enol/
tri-keto difference is 44 kJmolÀ1 in vacuuo, this value drops to
37 kJmolÀ1 for a solvation computation simulating water.
This energy difference must be at least compensated by the
strength of the hydrogen bonds.
The survey of tautomeric polymorphism in the CSD[1]
used the energy difference between two reported tautomeric
forms as a criterion for detailed examination of the structures
involved. In this case a simplified DFT method (B3LYP/6-
311 ++ G**/PCM e = 3) was necessarily used. Using this
method the majority of the 16 pairs designated as tautomeric
polymorphs have energy differences less than 12 kJmolÀ1.
The statistics reported for the combined sample of structures
including both tautomeric polymorphs and the much more
numerous mixed tautomer structures shows that in this more
general sense the energy difference as computed by the above
method, extends to slightly over 30 kJmolÀ1. The value
computed by this above method for the current enol/keto
pair is 36 kJmolÀ1 and is thus just at the limit of other known
cases for this larger group of tautomeric materials but well
beyond any other tautomeric polymorph. The corresponding
values for the thiobarbituric acid keto/enol pair are about
10% smaller than for the barbituric acid pair.
A key point in the validation of the conclusions of the
powder diffraction studies of Ref. [9] was the inclusion in that
report of the results of a calculation indicating that based on
the lattice energy of the enol structure was 58.5 kJmolÀ1 more
stable than that for the tri-keto form. This is larger than
a 53.7 kJmolÀ1 value of the energy difference using CCSD-T/
cc-pvtz[9] of the two tautomers by 4.8 kJmolÀ1 and thus in
agreement with and supporting of the finding that the enol
forms the most stable crystal. The computational method
used is specified only as a periodic boundary dispersion
corrected DFT. Neither the functional or convergence criteria
were specified. The cited references[16] are not much help. The
optimized structure obtained from this calculation was not
reported, so it cannot be compared to the results of our
structure determination. Further, the use of a highly corre-
lated gas-phase reference is debatable.
=
indicated by the reappearance of the C O stretch intensity in
the Raman spectrum. This conversion indicates that there is
a significant zero-point energy and or vibrational enthalpy
contribution to the energetic difference between the keto and
enol forms.
Our work herein provides a reliable basis for full lattice
computations and harmonic dynamics. Specifically, following
the procedure used for the a- and g-polymorphs of glycine.[18]
The G-point computation of the harmonic lattice dynamics
will be used to simulate the inelastic neutron scattering
spectrum. Comparison of the observed and computed INS
spectra provides a full assignment of the vibrations as well as
a check on the structure used. The zero-point and thermal
energy difference between the tri-keto and enol forms and the
effect of H/D replacement can then be included in the
tabulation of the enthalpy. In this regard, the enthalpy
difference between these two forms as obtained from heats
of dissolution would be of considerable interest in providing
a test of the energy computation as has been done for two
polymorphic forms of glycine.[18]
Acknowledgements
This work was made possible by acquisition of a diffractom-
eter with funds from NSF grant CHE-1048703. M.M. received
support from NSF REU grant CHE 1263154 to Syracuse
Angew. Chem. Int. Ed. 2016, 55, 1309 –1312
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1311

Angewandte
Communications
Chemie
R1 = 0.0399, wR2 = 0.0991, largest diff. peak and hole, 0.206 and
À0.239 eŠÀ3, full-matrix least-squares refinement on F2, data/
restraints/parameters 1139/0/85, goodness-of-fit on F2 = 1.048.
CCDC 1418712 contains the supplementary crystallographic
data for this paper. These data can be obtained free of charge
from The Cambridge Crystallographic Data Centre. At 95 K:
a = 4.7799(4), b = 8.9363(8), c = 11.8023(11) Š, a = 90, b =
University. We thank Aurora J. Cruz-Cabeza for helpful
discussion.
Keywords: density gradient separation · deuterium effects ·
hydrogen bonding · tautomeric polymorphs · X-ray diffraction
96.133(3), g = 908, V= 501.25(8) Š3, V= 1.697 MgmÀ3
, m =
How to cite: Angew. Chem. Int. Ed. 2016, 55, 1309–1312
Angew. Chem. 2016, 128, 1331–1334
0.148 mmÀ1, qmax = 27.3978, 11533 reflections collected (1130
independent), Rint = 0.0208, final R indices [I > 2sigma(I)], R1 =
0.0306, wR2 = 0.0801, R indices (all data), R1 = 0.0344, wR2 =
0.0832, largest diff. peak and hole, 0.273 and À0.255 eŠÀ3, full-
matrix least-squares refinement on F2, data/restraints/parame-
ters 1130/0/85, goodness-of-fit on F2 = 1.079. CCDC 1418693
contains the supplementary crystallographic data for this paper.
These data can be obtained free of charge from The Cambridge
Crystallographic Data Centre.
[1] A. J. Cruz-Cabeza, C. R. Groom, CrystEngComm 2011, 13, 93.
[2] J. Elguero, Cryst. Growth Des. 2011, 11, 4731 – 4738.
[3] M. R. Chierotti, L. Ferrero, N. Garino, R. Gobetto, L. Pellegrino,
D. Braga, F. Grepioni, L. Maini, Chem. Eur. J. 2010, 16, 4347 –
4358.
[4] W. Bolton, Acta Crystallogr. 1963, 16, 166.
[5] T. C. Lewis, D. A. Tocher, S. L. Price, Cryst. Growth Des. 2004, 4,
979 – 987.
[13] G. A. Jeffrey, An Introduction to Hydrogen Bonding, OUP,
Oxford, New York 1997, p. 12, Table 2.1 Strong AHB hydrogen
bonds are 2.2 – 2.5 Š for R(AB).
[6] G. A. Jeffrey, S. Ghose, J. O. Warwicker, Acta Crystallogr. 1961,
14, 881 – 887.
[7] A. R. Al-Karaghouli, B. Abdul-Wahab, E. Ajaj, S. Al-Asaff,
Acta Crystallogr. Sect. B 1977, 33, 1655 – 1660.
[14] T. Steiner, Angew. Chem. Int. Ed. 2002, 41, 48 – 76; Angew.
Chem. 2002, 114, 50 – 80.
[8] M. R. Chierotti, R. Gobetto, L. Pellegrino, L. Milone, P.
Venturello, Cryst. Growth Des. 2005, 5, 983 – 993.
[15] H. Guo, M. Karplus, J. Phys. Chem. 1992, 96, 7273 – 7287.
[16] One is to a commercial website that refers to multiple methods.
The other G. Kresse and J. Hafner, Phys. Rev. B 47, 558 –
561.10.1103/PhysRevB.47.558 presents the results of a study of
liquid metals using the LDA limit of DFT.
[17] Sigma–Aldrich product number 11709 lot S72013-029 specifica-
tion date 2008. The 50 gram sample was stored on a laboratory
shelf and infrequency opened until 2012 when it was used in
a course that involved Raman microscopy studies of the solid
material. The spectra observed were those reported in the
supplementary materials for Ref. [8], with and without the
[9] M. U. Schmidt, J. Bruening, J. Glinnemann, M. W. Huetzler, P.
Moerschel, S. N. Ivashevskaya, J. van de Streek, D. Braga, L.
Maini, M. R. Chierotti, R. Gobetto, Angew. Chem. Int. Ed. 2011,
50, 7924 – 7926; Angew. Chem. 2011, 123, 8070 – 8072.
[10] “Method and apparatus for separation of pharmaceutical
materials on the basis of their density”: B. S. Hudson, J.
Melton, U.S. Patent US 9,089,850 B2 (28 July, 2015.
[11] Interestingly, the absence of previous attempts to obtain enol
crystals from solution has been taken as evidence that barbituric
acid is subject to kinetic resistance to tautomer crystal formation
and thus as justification for the lack of appearance of a tauto-
meric form that has been predicted to be the stable form. S. L.
Bekç, C. Czech, M. A. Neumann, M. U. Schmidt, Z. Kristallogr.
2015, 230, 611 – 620.
=
strong C O stretching mode, changing from one to the other
when the region being examined was changed from one particle
to another. In some cases the particles showed a demarcation
line that had one or the other spectrum on either side of that line.
[18] S. A. Rivera, D. G. Allis, B. S. Hudson, Cryst. Growth Des. 2008,
8, 3905 – 3907.
[12] C4H4N2O3, FW 128.09, 0.6 ” 0.05 ” 0.05 mm3, monoclinic, P21/n,
Z = 4, 0.71073 Š X-ray. At 224(2) K: a = 4.8497(3), b = 8.9134-
(6), c = 11.8920(8) Š, a = 90, b = 94.892(2), g = 908, V= 512.19-
[19] K. Merz, A. Kupka, Cryst. Growth Des. 2015, 15, 1553 – 1558.
, , qmax = 27.2218, 11272
(6) Š3, 1.661 MgmÀ3 m = 0.145 mmÀ1
reflections (1139 independent), Rint = 0.0316, final R indices
[I > 2sigma(I)], R1 = 0.0335, wR2 = 0.0933, R indices (all data),
Received: August 31, 2015
Published online: December 14, 2015
1312
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1309 –1312