Investigation of the polymorphs of dimethyl-3,6-dichloro-2,5-dihydroxyterephthalate by 13C solid-state NMR spectroscopy

Article abstract of DOI:10.1021/ja003599b

Two of the three conformational polymorphs of dimethyl-3,6-dichloro-2,5-dihydroxyterephthalate are studied by solid-state NMR techniques. The structural differences between the polymorphs have previously been studied by X-ray. In these two polymorphs named white and yellow due to their color, the major structural difference is the torsional angle between the ester group and the aromatic ring. The yellow form has a dihedral angle of 4° between the plane of the aromatic ring and the plane of the ester group, while the white form has two different molecules per unit cell with dihedral angles of 70° and 85°. This change greatly affects the conjugation in the π-electronic system. In addition, there are differences in the hydrogen-bonding patterns, with the white form having intermolecular hydrogen bonds and the yellow form having intramolecular hydrogen bonds. In this work, the carbon isotropic chemical shift values and the chlorine electric field gradient (EFG) tensor information are extracted from the ¹³C MAS spectra, and the principal values of the chemical shift tensors of the carbons are obtained from 2D FIREMAT experiments. Quantum chemical calculations of the chemical shift tensor data as well as the EFG tensor are performed at the HF and DFT levels of theory on individual molecules and on stacks of three molecules to account for the important intermolecular interactions in the white form. The differences between the spectral data on the two polymorphs are discussed in terms of the known electronic and structural differences.

Full text of DOI:10.1021/ja003599b

J. Am. Chem. Soc. 2001, 123, 1713-1722
1713
Investigation of the Polymorphs of
Dimethyl-3,6-Dichloro-2,5-Dihydroxyterephthalate by C Solid-State
NMR Spectroscopy
13
Mark Strohmeier,† Anita M. Orendt,‡ D. W. Alderman,† and David M. Grant*,†
Contribution from the Department of Chemistry, UniVersity of Utah, 315 South 1400 East,
Salt Lake City, Utah 84112-0850, and Center for High Performance Computing, UniVersity of Utah,
155 South 1452 East, Salt Lake City, Utah 84112-0190
ReceiVed October 5, 2000. ReVised Manuscript ReceiVed December 12, 2000
Abstract: Two of the three conformational polymorphs of dimethyl-3,6-dichloro-2,5-dihydroxyterephthalate
are studied by solid-state NMR techniques. The structural differences between the polymorphs have previously
been studied by X-ray. In these two polymorphs named white and yellow due to their color, the major structural
difference is the torsional angle between the ester group and the aromatic ring. The yellow form has a dihedral
angle of 4° between the plane of the aromatic ring and the plane of the ester group, while the white form has
two different molecules per unit cell with dihedral angles of 70° and 85°. This change greatly affects the
conjugation in the π-electronic system. In addition, there are differences in the hydrogen-bonding patterns,
with the white form having intermolecular hydrogen bonds and the yellow form having intramolecular hydrogen
bonds. In this work, the carbon isotropic chemical shift values and the chlorine electric field gradient (EFG)
tensor information are extracted from the 13C MAS spectra, and the principal values of the chemical shift
tensors of the carbons are obtained from 2D FIREMAT experiments. Quantum chemical calculations of the
chemical shift tensor data as well as the EFG tensor are performed at the HF and DFT levels of theory on
individual molecules and on stacks of three molecules to account for the important intermolecular interactions
in the white form. The differences between the spectral data on the two polymorphs are discussed in terms of
the known electronic and structural differences.
Introduction
Scheme 1
Due to the conformational flexibility of organic molecules
distinct crystal structures can often be obtained by varying
crystallization conditions. This phenomenon is known as poly-
morphism. Conformational polymorphism, as encountered in
dimethyl-3,6-dichloro-2,5-dihydroxyterephthalate (Scheme 1),
is one of several categories of polymorphism. The analysis of
polymorphs is increasingly important in pharmaceutics, material
science, and industrial chemistry because solubility, color, and
mechanical properties in solids depend inherently on the crystal
structure and the conformation of the molecule in a given
polymorph.1,2
Solid-state NMR is known to be a powerful method to
investigate the effects of differences in conformation and
packing without the limitations of diffraction methods. Chemical
shifts, especially, are very sensitive to local conformations. Most
previous solid-state NMR investigations of polymorphs have
relied on differences in isotropic chemical shifts.3-11 Only a
few studies have utilized the chemical shift principal values to
investigate polymorphism.12,13 However, the chemical shift is
a second rank tensor and contains detailed orientational infor-
mation. Thus, the principal values and the orientations of the
chemical shift tensor may reveal a deeper insight into the
conformation of the compound. The full tensor is only accessible
by single crystal studies, but from measurements on powdered
samples the chemical shift principal values may be obtained.
(6) Schmidt, A.; Kababya, S.; Appel, M.; Khatib, S.; Botoshansky, M.;
Eichen, Y. J. Am. Chem. Soc. 1999, 121 11291.
(7) Singh, D.; Marshall, P. V.; Shields, L.; York, P. J. Pharm. Sci. 1998,
87, 5, 655.
(8) Harris, R. K.; Kenwright, A. M.; Say, B. J.; Yeung, R. R.; Fletton,
R. A.; Lancaster, R. W.; Hardgrove, G. L., Jr. Spectrochim. Acta 1990,
46A, 927.
(9) Fletton, R. A.; Harris, R. K.; Kenwright, A. M.; Lancaster, R. W.;
Packer, K. J.; Sheppard, N. Spectrochim. Acta 1987, 43A, 1111.
(10) Harris, R. K.; Yeung, R. R.; Lamont, R. B.; Lancaster, R. W.; Lynn,
S. M.; Staniforth, S. E. J. Chem. Soc., Perkin Trans. 2 1997, 8, 2653.
(11) Zell, M. T.; Padden, B. E.; Grant, D. J. W.; Chapeau, M.-C.; Prakash,
I.; Munson, E. J. J. Am. Chem. Soc. 1999, 121, 1372.
(12) Smith, J.; MacNamara, E.; Raftery, D.; Borchardt, T.; Byrn, S. J.
Am. Chem. Soc. 1998, 120, 11710.
* To whom the correspondence should be addressed.
† Department of Chemistry, University of Utah
‡ Center for High Performance Computing, University of Utah
(1) Threlfall, T. L. Analyst 1995, 120, 2435.
(2) Byrn, S. R.; Pfeiffer, R. R.; Stephenson, G.; Grant, D. J. W.; Gleason,
W. B. Chem. Mater. 1994, 6, 1148.
(3) Anwar, J.; Tarling, S. E.; Barnes, P. J. Pharm. Sci. 1989, 78, 4, 337.
(4) Kimura, K.; Hirayama, F.; Uekama, K. J. Pharm. Sci, 1999, 88, 4,
385.
(5) Stockton, G. W.; Godfrey, R.; Hitchcock, P.; Mendelsohn, R.;
Mowery, P. C.; Rajan, S.; Walker, A. F. J. Chem. Soc., Perkin Trans. 2
1998, 9, 2061.
(13) Christopher, E. A.; Harris, R. K.; Fletton, R. A. Solid State Nucl.
Magn. Reson. 1992, 1, 93.
10.1021/ja003599b CCC: $20.00 © 2001 American Chemical Society
Published on Web 02/02/2001

1714 J. Am. Chem. Soc., Vol. 123, No. 8, 2001
Strohmeier et al.
Figure 1. View of the molecule stacks chosen for the calculations on W-I a and Y-I, The stack used for the W-I b stack of three molecules has
the same configuration as the one for W-I a, but with the W-I b molecules sandwiched between two W-I a molecules.
The recently developed 2D FIREMAT14 technique with
TIGER15 processing is one such method which makes it possible
to obtain the chemical shift principal values for each resolved
isotropic chemical shift from a powdered sample.
Y-I to W-I at 385-410 K with optical microscopy and
differential scanning calorimetry, and reported the 35Cl NQR
frequencies. The most striking difference between W-I and Y-I
is the torsional angle between the planar aromatic system and
the ester group, resulting in different crystal packing and a
change in hydrogen bonding between the carbonyl and the
hydroxy group in the molecule. The torsional angle in Y-I is
4° and an intramolecular hydrogen bond exists. Due to this low
torsional angle the substituent sphere is sterically crowded
Another probe that is sensitive to conformation is the
quadrupolar coupling of nuclei with a quadrupole moment. The
magnitude of the coupling depends on the size of the quadrupole
moment of the nucleus and the size and orientation of the
electromagnetic field gradient (EFG) at the nucleus. The EFG
is generated by the electron distribution around the nucleus and
is therefore very sensitive to changes in the electronic environ-
ment caused by variations in conformation and packing in the
crystal. The quadrupolar coupling is described by a traceless
second rank tensor. It may be observed in solid-state MAS
resulting in longer C-Cl bonds and distorted bond angles. Y-I
1
has one
/ centrosymmetric molecule per asymmetric cell,
2
1
whereas W-I has two
/ centrosymmetric molecules per
2
asymmetric cell, with torsional angles of 70° (W-I a) and 85°
(W-I b) in these molecules, respectively. In addition, there is
intermolecular hydrogen bonding in W-I (Figure 1). The large
torsional angle in W-I has an extensive impact on the overall
electronic structure of the compound, since the π-π overlap
between the carbonyl group and the aromatic system is
negligible compared to that in Y-I where the low torsional angle
allows appreciable π-π overlap. Swiatkiewicz et al.23 assigned
the solid-state Raman and electronic photoexcitation and emis-
sion spectra for W-I and Y-I and used these methods to
investigate the phase transition at 380-400 K.
LY-I was discovered by Yang et al.,24 who determined that
it has a torsional angle of 40° and crystal packing similar to
Y-I. They performed a variable-temperature single-crystal X-ray
diffraction study and a combined optical microscopy and
differential scanning calorimetry study to investigate the con-
nection between internal molecular motion and conformational
isomerization between the different polymorphic forms.25 These
studies reveal a large temperature dependence of the torsional
angle of the ester group with respect to the aromatic plane. They
also found large anisotropic displacement parameters for Y-I
1
NMR spectra of common spin /2 nuclei as a dipolar coupling
between the quadrupolar nucleus (35Cl and 37Cl) and the ob-
served spin 1/2 nucleus (13C), which is not completely averaged
by magic angle spinning.16 Simulation of the resulting Pake
doublet-like resonance in the MAS spectrum allows the deter-
mination of most parts of the EFG-tensor and the internuclear
distance.16-19
In this paper solid-state NMR is used to investigate the
conformational polymorphism of dimethyl-3,6-dichloro-2,5-
dihydroxyterephthalate (I). The polymorphism of I was dis-
covered by Hantzsch20 over 100 years ago and has attracted
the attention of many researchers. I forms three different
conformational polymorphs, which are named by their color:
yellow (Y-I), light yellow (LY-I) and white (W-I). All three
forms are stable at room temperature and can be stored for years
without phase transition. Byrn et al.21,22 solved the crystal
structure of W-I and Y-I, investigated the phase transition from
(14) Alderman, D. W.; McGeorge, G.; Hu, J. Z.; Pugmire, R. J.; Grant,
D. M. Mol. Phys. 1998, 95, 6, 1113.
(15) McGeorge, G.; Hu, J. Z.; Mayne, C. L.; Alderman, D. W.; Pugmire,
R. J.; Grant, D. M. J. Magn. Reson. 1997, 129, 134.
(16) Menger, E. M.; Veeman, W. S. J. Magn. Reson. 1982, 46, 257.
(17) Olivieri, A. C.; Elguero, J.; Sobrados, I.; Cabildo, P.; Claramunt,
R. M. J. Phys. Chem. 1994, 98, 5207.
(18) Eichele, K.; Wasylischen, R. E.; Grossert, J. S.; Olivieri, A. C. J.
Phys. Chem. 1995, 99, 10110.
(19) Alacoron, S. H.; Olivieri, A. C.; Carss, S. A.; Harris, R. K.; Zuriaga,
M. J.; Monti, G. A. J. Magn. Reson., Ser A 1995, 116, 244.
(20) Hantzsch, A. Chem. Ber. 1915, 48, 785.
(21) Byrn, S. R.; Curtin, D. Y.; Paul, I. C. J. Am. Chem. Soc. 1971, 94,
890.
(22) Curtin, D. Y.; Byrn, S. R. J. Am. Chem. Soc. 1969, 91, 1865.
(23) Swiatkiewicz, J.; Prasad, P. N. J. Am. Chem. Soc. 1982, 104, 6913.
(24) Yang, Q.-C.; Richardson, M. F.; Dunitz, J. D. Acta Crystallogr.
1989, B45, 312.
(25) Richardson, M. F.; Yang, Q.-C.; Novotny-Bregger, E.; Dunitz, J.
D. Acta Crystallogr. 1990, B46, 653.

Conformational Polymorphism Studies by SS NMR
J. Am. Chem. Soc., Vol. 123, No. 8, 2001 1715
100.625 MHz. On the 400 MHz spectrometer TOSS26 sideband
compared to the other two forms and proposed a mechanism
for the thermal phase transition.24
suppression and TPPM27,28 decoupling with a phase angle of 16° was
1
used. The H T1 was measured to be 33 s for Y-I and 13 s for W-I at
In this work the 13C CPMAS spectra at 2.35, 4.7, and 9.4 T
and the FIREMAT spectra at 9.4 T are obtained on both Y-I
and W-I. From simulations of the 13C CPMAS spectra, dipolar
information as well as information on the EFG tensor are
obtained. The FIREMAT datasets provide the chemical shift
principal values of all carbon positions of the two polymorphic
forms. The chemical shift principal values and the EFG tensors
are calculated for the known low-temperature structures using
HF and DFT methods. To account for intermolecular interactions
these calculations are done on isolated molecules and on stacks
of three molecules. The theoretical values are compared to
experiment.
100.625 MHz.
The following optimized experimental conditions were used: a 7
ms contact time and a 12 s recycle time for W-I and a 9 ms contact
time and 33 s recycle time for Y-I. The proton 90° pulse was
approximately 4.1 µs on all three spectrometer systems.
The 13C FIREMAT14 experiments were performed on the CMX400,
using the 7.5 mm probe with a feedback circuit to synchronize the
pulse sequence to the rotor position. The 5π pulse sequence29 was used
with TPPM decoupling using a phase angle of 36°. For W-I the spectra
window in the acquisition dimension was 78.125 kHz, the spectral width
in the evolution dimension was 13 kHz, and the spinning speed was
812.5 Hz. For Y-I the spectra width in the acquisition dimension was
63.291 kHz, in the evolution dimension it was 10.504 kHz, and a
spinning speed of 659 Hz was used. For both forms 16 complex
increments were taken with 384 scans each.
Experimental Section
The 2D data were processed on a Sun Enterprise 3500. Individual
spinning side-band patterns were obtained from the 2D dataset using
TIGER15 processing. TIGER processing requires a linear model for
the evolution dimension, which is obtained by fitting a guide spectrum
with Gaussian and Lorentzian line shapes. The line shapes encountered
in the 9.4 T spectra are well described by Gaussian and Lorentzian
line shapes, despite the pattern-like nature of the coupled shifts. The
model is then used to extract individual sideband patterns at the
maximum intensity for the different isotropic shifts.
All sideband patterns were fit with a single CSA sideband pattern
model; only when breakthrough from a nearby resonance was apparent
in the spectrum was a second or third sideband pattern included in the
model to obtain the best fit. The principal shift values of this additional
sideband patterns were taken from the best fits of the neighboring
resonances, and only their intensities were optimized.
The residual dipolar coupling apparent in the sideband manifold was
not incorporated into the sideband model. The effect of the residual
dipolar coupling resulting from quadrupolar interactions onto the shape
of the sidebands is well understood.30 The effect on the FIREMAT
spectrum, however, is more complicated and was not previously
investigated. A challenging problem arises from the large number of
parameters that must be considered in a complete description of the
dipolar-coupled sideband pattern since both Cl isotopes interact with
the 13C nuclei. Thus, it is only possible to fit the coupled sideband
pattern when both the chemical shift anisotropy, the 13C-35,37Cl dipolar
coupling constants and the 35,37Cl quadrupolar couplings are combined
in the computation of the line patterns.
Sample Preparation. Dicarbomethoxy-1,4-cyclohexanedione was
purchased from Arcos and used without further purification. HPLC-
grade methanol and benzene were stored over 4 Å molecular sieves
and used for recrystallization.
Dimethyl 2-5-Dichloro 3,6 dihydroxyterephthalate (I). Dicar-
bomethoxy-1,4-cyclohexanedione (3.6 g, 15.3 mmol) was dissolved in
60 mL of acetic acid (reagent grade), and chlorine was bubbled through
this solution for 1.5 h. It was then heated to 75 °C and stirred overnight.
Since the cyclohexadione shows blue while I shows green fluorescence,
the reaction progress can easily be monitored by TLC on silica with
benzene eluent. After completion of the reaction the solvent was
removed to precipitate a yellow solid (2.55 g). This crude product was
further purified on a column packed with silica gel (200-400 mesh)
with benzene eluent. The overall yield of I was 0.96 g (22%).
EI-MS [m/z (%)]: 293.9 (M+, 24.3), 261.9 (∆ ) 32, -CH3OH, 91.2),
229.9 (∆ ) 64, -2 (CH3OH), 100), 201.9 (∆ ) 92, -2 (CH3OH) -CO,
23.8)
13C NMR (500 MHZ, methanol): δ (ppm) ) 53.46 (CMe), 119.40
(CCl), 126.18 (Ci), 145.92 (COH), 166.93 (CO2).
Crystallization. The two different polymorphs studied were obtained
by recrystallization from different solvents by slowly cooling a saturated
refluxing solution of I. If the solution was not refluxed for several
minutes, a mixture of different polymorphic forms was usually obtained.
White needles are preferentially formed from a benzene solution, but
occasionally the formation of yellow plates is observed. In this case
the solution is reheated and refluxed for several minutes to remove all
yellow seed crystals from the system.
Y-I crystallizes preferentially from saturated methanol solutions. If
the white needles appear along with the yellow plates, the solution is
heated again to dissolve the crystals. Since W-I is more soluble in
methanol, the yellow crystals remain and become effective seeds. The
obtained crystals were then filtered off, washed, and dried in a vacuum
to remove residual solvent.
X-ray Powder Diffraction. The X-ray powder diffraction patterns
of the two polymorphic forms were recorded on a Rigaku diffractometer
with monochromated Cu KR radiation (λ ) 1.540598 Å). Data were
collected between a 2θ angle of 5° and 60° in a step scan mode with
0.05° steps and a collection time of 0.3 s. Approximately 20 mg of
sample was cautiously ground, placed on a greased microscope slide,
and carefully smoothed with a spatula.
Theory of Residual 35,37Cl-13C Dipolar Couplings and MAS
Simulations. Since the ratio of the quadrupolar coupling ø to the Larmor
frequency ν is larger than 1 for both 35Cl and 37Cl at commonly used
field strengths in solid-state NMR, first-order perturbation is not
applicable to describe the effect of the quadrupolar Hamiltonian. The
1
total Hamiltonian for an I,S spin system where I is spin /2 (C) and S
3
is spin /2 (Cl) is given by
H
total ) HIZ + HSZ+ HSQ + HD + HJ
Applying first-order perturbation to the coupling Hamiltonian, only the
combined Zeeman and quadrupolar Hamiltonian have to be diagonal-
ized.
The theoretical XRD patterns based on the room-temperature X-ray
single-crystal data21,24 were calculated using the Molecular Simulations
Inc. software package and compared to the collected data to confirm
the crystal form of the sample.
H0 ) HIZ + HSZ + HSQ H′ ) HD + HJ
Using the EFG principal axis system as the axis system with the
magnetic field direction specified by R and â (Figure 2), the complete
Solid-State Spectroscopy. The solid-state CPMAS spectra at 2.35
T were recorded on a Chemagnetics CMX100 spectrometer equipped
with a 7.5 mm Chemagnetics high-speed-spinning PENCIL probe with
(26) Geen, H.; Bodenhausen, G. J. Am. Chem. Soc. 1993, 115, 1579.
(27) Bennet, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V. J. Chem.
Phys. 1995, 103, 6951.
(28) McGeorge, G.; Alderman, D. W.; Grant, D. M. J. Magn. Reson.
1999, 137, 138.
(29) Hu, J. Z.; Alderman, D. W.; Ye, C.; Pugmuire, R. J.; Grant, D. M.
J. Magn. Reson. A, 1993, 105, 82.
(30) Zheng, Z.; Gan, Z.; Sethi, N. K.; Alderman, D. W.; Grant, D. M. J.
Magn. Reson. 1991, 95, 509.
a
13C frequency of 25.152 MHz. The CPMAS spectra at 4.7 T were
recorded on a Varian VXR-200 spectrometer using a 5 mm high-speed-
spinning probe from Doty Scientific Inc. at a 13C frequency of 50.318
MHz. The high-resolution spectra at 9.4 T were collected on a
Chemagnetics CMX400 spectrometer equipped with a 7.5 mm Che-
magnetics high-speed-spinning PENCIL probe at a 13C frequency of

1716 J. Am. Chem. Soc., Vol. 123, No. 8, 2001
Strohmeier et al.
neighboring 13C where the unique EFG tensor axis is directed along
the C-Cl bond as generally found. This configuration makes it possible
to retrieve the orientational information from the coupling of 35,37Cl to
the carbons in â-position (Figure 2).
The 35,37Cl-coupled 13C line shapes were calculated by diagonalizing
the appropriate Hamiltonian and applying the calculated wave functions
on H′ using software written in Fortran 77. Different spinning-axis
orientations were produced using the POWDER33 approach, and the
MAS line shapes were then calculated by averaging over the magic
angle. It was found that 47 steps per rotation cycle and 4096 crystal
orientations were sufficient to obtain all expected line shape features
in the simulation. The program was tested and found to produce
simulations in good agreement with simulations and MAS spectra from
the literature.17-19 The calculations for a 35,37Cl-13C spin-pair required
about 1 min on a Sun Enterprise 3500.
The quadrupolar coupling constants previously found in NQR
experiments21 were used in the simulations. However, an average value
for the two molecules per asymmetric cell was used to simplify the
simulations for W-I. The small difference between the two coupling
constants and the good agreement between simulation and experimental
spectra show that this assumption is correct within the experimental
resolution. The asymmetry used in the simulations was systematically
improved, starting from values found for similar compounds.32 Initial
values for the dipolar coupling constants for the different forms were
calculated using the C-Cl distances found in the room-temperature
crystal structure.21 Care was taken that the geometrical restraint im-
posed by the molecular geometry onto the four-body system was
conserved when the spatial orientation of the three internuclear vectors
with respect to the EFG principal axis system was determined (Figure
2). Since the MAS spectra of W-I is relatively insensitive to J coupling
due to overlap with the Ci resonance, the same scalar coupling found
in Y-I was also assumed for W-I. The same EFG-tensor parameters
were used for all three different field strengths. The resulting spectra
were then line-broadened using a Gaussian function and compared to
the collected data. Agreement was judged to be excellent by visual
inspection, as can be seen in Figures 3 and 4 for Y-I and W-I,
respectively.
Quantum Mechanical Calculations. DFT (density functional
theory) and HF (Hartree-Fock) calculations were performed with the
Gaussian 98 computer program34 using the GIAO35 (gauge invariant
atomic orbitals) method and the D95** basis set.36 The DFT calculations
use the B3LYP37,38,39 exchange correlation functional. The shift tensor
calculations were done on isolated molecules as well as on stacks of
three molecules to account for intermolecular interactions using the
known low-temperature structures.24 The C-H bond distances were
lengthened to 1.085 Å as C-H distances found by X-ray diffraction
are known to be too short. As the chemical shift is known to be very
sensitive to the geometry, in particular the C-H bond lengths, the
standard procedure has often been to optimize all proton positions.40
However, in this case due to the size of the system (76 atoms) it was
Figure 2. Relationship between the C-Cl bond vector, the spinning
axis S and the magnetic field Bo in the EFG principal axis system. The
angles θ and φ specify the orientation of the internuclear vector, and
the angles R and â specify the orientation of the magnetic field vector
in the EFG principal axis system, respectively. The magic angle is
abbreviated by ma.
quadrupolar Hamiltonian31 is given by:
e2Q
4S(2S - 1)
HSQ
)
‚ q0 ‚ [(3Sˆ2X - Sˆ2) + η(SˆX2 + Sˆ2Y)]
Diagonalizing the combined quadrupolar and Zeeman Hamiltonian
yields the wave function as an expansion of Zeeman IS eigenstates.
I
S
Ψomn(R,â) )
am(R,â)an(R,â)|Φm |Φn
∑∑
-I -S
Once all am(R,â) and an(R,â) coefficients are determined, the combined
dipolar and scalar I,S coupling terms may be calculated numerically.
For the sake of simplicity, the same principal axis system is assumed
for the dipolar and scalar coupling. The 35,37Cl-coupled 13C resonance
is then calculated by:
∆ν1/2n(R,â) ) Ψo-1/2n(R,â)|H′|Ψ-o 1/2n(R,â) -
Ψo1/2n(R,â)|H′|Ψ1o/2n(R,â)
The powder average is obtained by collecting the resonance fre-
quencies for all R and â; the MAS line shapes can then be obtained by
averaging over the magic angle.
In this approach no assumptions regarding the symmetry and the
orientation of the EFG tensor are made. The common assumption of
an axially symmetric EFG tensor for chlorine is applicable for most
cases, but as soon as the axial symmetry of the three free-electron pairs
on chlorine is broken, the asymmetry
(33) Alderman, D. W.; Solum, M. S.; Grant, D. W. J. Chem. Phys. 1986,
84, 7, 3717.
(34) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.;
Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A.
D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi,
M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.;
Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick,
D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.;
Ortiz, J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,
P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-
Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe,
M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.;
Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople J. A. Gaussian 98,
revision A.7; Gaussian, Inc.: Pittsburgh, PA, 1998.
η ) (qxx - qyy)/qzz
becomes appreciable different from 0. That is the case when chlorine
is attached to a sp2 hybridized carbon atom with an electron deficiency
in the p-orbital, resulting in a polarization of the chlorine lone pairs
parallel to the electron-deficient p-orbital. In these cases asymmetries
as high as 0.26 are observed.32
Since no assumptions regarding the symmetry and orientation of
the EFG tensor are made, it is possible to obtain not only its magnitude
and asymmetry but also its orientation with respect to the molecular
frame even when the tensor is not axially symmetric. To be able to
retrieve this information it is necessary that the unique axis is not
parallel to the unique dipolar coupling axis, that is, internuclear vector
(θ * 0). In the investigated compound the 35,37Cl is coupled to three
(35) Ditchfield, R. Mol. Phys. 1974, 27, 789.
(36) Dunning, T. H., Jr.; Hay, P. J. In Modern Theoretical Chemistry;
Schaefer, H. F., III, Ed.; Plenum: New York, 1976; pp 1-28.
(37) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(38) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(39) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett. 1989,
157, 200.
(31) Slichter, C. P. Principles of Magnetic Resonance, 3rd ed.;
Springer: New York 1996; p 496.
(32) Lucken E. A. C. Nuclear Quadrupole Coupling Constants; Academic
Press: New York 1969; p 167.
(40) Liu, F.: Phung, C. G.; Alderman, D. W.; Grant, D. M. J. Am. Chem.
Soc. 1996, 118, 10629.

Conformational Polymorphism Studies by SS NMR
J. Am. Chem. Soc., Vol. 123, No. 8, 2001 1717
Figure 3. 13C CPMAS spectra of Y-I and simulations of the coupled 13C resonances at 2.35, 4.7, and 9.4 T. Spinning sidebands are marked by
asterisks.
Figure 4. 13C CPMAS spectra of W-I and simulations of the coupled 13C resonances at 2.35, 4.7, and 9.4 T. Spinning sidebands are marked by
asterisks.
felt sufficient to adjust all C-H bond lengths to a single value. The
structures used for stacks of three molecules are shown in Figure 1.
Since the inversion symmetry of the system is broken when calculations
are performed on the stack chosen for Y-I, only the half of the molecule
that is exposed to the intermolecular interactions is considered. The
shielding values were converted to the ppm scale by using the slope
and intercept of the least-squares fit of the correlation between
experimental data and the calculated shielding values.
The electrostatic field properties were calculated using the same self-
consistent field wave functions employed for the NMR calculations.
The EFG values were converted to the nuclear coupling constant for
35Cl in frequency units by:
respectively. The chemical shift data obtained from the simula-
tions of the residual dipolar coupling and the FIREMAT
experiment along with solution isotropic chemical shifts obtained
in different solvents are given in Table 1. The parameters used
for the simulations of the CPMAS spectra are given in Tables
2 and 3.
Assignment of the 13C CPMAS Spectra. Previously de-
scribed features17 of 13C MAS spectral patterns of carbons
coupled to 35,37Cl are observed for the carbons in R (CCl) and â
(COH and Ci) positions relative to chlorine. While the spectra
obtained at 4.7 and 2.35 T show the dominating effect of the
13C-35,37Cl residual dipolar coupling, the simpler spectra ob-
tained at 9.4 T provides directly assignable resonances. At 9.4
T the strong coupling to the 35,37Cl at the CCl position results in
a doublet-like resonance in both forms with a splitting of
approximately 500 Hz. The aromatic COH resonance is found
at lower field in the aromatic shift region than the Ci resonance
for both polymorphic forms. The COH and Ci resonances also
show the effect of the residual coupling but with a smaller
coupling as a result of the larger distance and a different
e2Q‚qZZ
øZZ
)
h
With a quadrupole moment for 35Cl of Q ) -0.08165 × 10-28 Q/m2,
a conversion factor of 1 au ) -19.184484 MHz is obtained.41
Results and Discussion
For the purpose of the following discussion the nomencla-
ture shown in Scheme 1 is used. The CPMAS spectra of W-I
and Y-I at different fields are shown in Figures 3 and 4,
(41) Pyykko, P. Z. Naturforsch. 1992, 47a, 189.

1718 J. Am. Chem. Soc., Vol. 123, No. 8, 2001
Strohmeier et al.
Table 1. Experimental Chemical Shift Tensor Principal Values for
W-I and Y-I/ppm
shift difference of ∆δ ) 1.6 ppm. The excellent agreement
between simulation and experimental CPMAS spectra indicate
that the two molecules per asymmetric cell are indeed resolved
for the Ci position.
δ11
δ22
δ33
δiso δCPMAS δMeOD δDMSO δCDCl3 δC6D6
CO2
260.4 138.3 108.4 169.0 168.5a 166.9 164.3
251.1 149.4 110.0 170.2 169.8
At 4.7 T field strength the 35,37Cl quadrupolar interaction is
stronger than the Zeeman interaction, and the expected triplet
pattern with a 2:1:1 intensity ratio is observed for the CCl
position in both spectra. Due to overlap with the Ci reso-
nance only two peaks are resolved in either spectrum. The Ci
and COH resonance are broadened, and shoulders are formed
by the residual coupling at this field. At 2.35 T the chlorine
is predominately quantized by the quadrupolar interaction, and
four patterns result for CCl from the residual coupling, each
corresponding to a chlorine eigenstate. The extensive overlap
in the broadened Ci and COH resonances makes it difficult to
recognize individual features of the overlapping peaks at 2.35
T. The simulations of the carbon-chlorine residual dipolar
coupling in both polymorphic forms reproduce all of the
observed features in the spectra and are in excellent agree-
ment at all three field strengths as shown in Figures 3 and
4. The actual isotropic chemical shifts obtained from simu-
lating the MAS spectra are given in the δCPMAS column in
Table 1.
W-I
Y-I
167.5 167.8
148.3 149.3
119.6 120.3
COH
W-I
Y-I
213.9 157.5 60.0 143.9 143.6a 145.9 143.4
224.3 171.2 61.5 152.3 152.1
CCl
W-I 189.8 118.6 47.9 118.8 118.5a 119.4 117.9
Y-I
183.9 127.4 56.4 122.6 122.3a
b
Ci
W-I a 189.5 144.9 41.7 125.4 125.1a 126.2 125.7
W-I b 191.4 146.0 42.9 126.8 126.7a
Y-I
177.8 142.6 28.4 116.3 116.4
119.3 119.8
53.4 52.9
CMe
86.2 69.1 13.9 56.4 56.3
83.8 66.5 10.6 53.6 54.1
W-I
Y-I
53.5 52.9
a The isotropic chemical shift is determined by simulating the
CPMAS spectra. b The two molecules per asymmetric cell in the white
form are resolved for the Ci position
Table 2. Coupling Parameters Used in the CPMAS Spectral
Simulations
Effect of Conformation and Intermolecular Interactions
on the Quadrupolar Coupling. The parameters used for the
simulations in Figures 3 and 4 are shown in Tables 2 and 3.
The accuracy in the asymmetry parameter η is approximately
0.03. The asymmetries and quadrupolar coupling constants are
comparable to quadrupolar coupling constants found for other
six-membered aromatic systems with electronegative substitu-
ents.32 The orientation of the EFG tensor relative to the
molecular frame is only significant for the coupling to the
carbons in â-position to the chlorine, since the z-axis is directed,
as expected, along the C-Cl bond. The error in φ is about 20°
for Y-I because of the small magnitude of the dipolar coupling
to the carbons in â-position. The simulated spectra of W-I are
insensitive to changes in the angle φ, since the asymmetry is
small. The dipolar coupling constants obtained are smaller by
3.5% for Y-I and 2% for W-I than the dipolar coupling constants
predicted from the X-ray diffraction data. However, it is known
that dipolar coupling constants calculated from X-ray distances
are 2-5% larger than the values determined experimentally by
NMR.42
D /Hz
CCl
Ci
COH
J /Hz
W-I
Y-I
559
549
147
137
150
156
-20
-20
Table 3. Calculated and Experimental Quadrupolar Coupling
Constants for 35Cl and Orientations
a
a
øzz
øxx
øyy
θCCl
deg
φCI
φCOH
deg
MHz
MHz MHz
η
deg
experiment
-
-
W-I
Y-I
-70.65b
-73.30b
-
-
0.04
0.15
0
0
90c
90
90c
90
B3LYP D95**
W-I a -71.32
W-I b -69.92
Y-I
33.54 37.63 0.06
30.91 37.86 0.07
29.24 46.82 0.23
0
0
0
-119
-108
-91
71
74
89
-76.06
B3LYP D95** Three Molecules
33.34 38.57 0.07
33.28 38.29 0.08
W-I a -71.92
W-I b -71.56
Y-I
0
1
1
-101
-95
-92
82
87
88
-76.41
29.12 47.3
0.24
As discussed by Lucken,32 the coupling constant ø and the
asymmetry η are significantly influenced by the electronegativity
of the neighboring group and the nature of the chemical bond.
The conformational differences between the two polymorphs
have a great effect on the electronic properties of the molecule.
In Y-I the ester group is in plane with the aromatic ring allowing
conjugation, resulting in a higher electronegativity of the
aromatic system and therefore to a larger quadrupolar coupling.
The larger asymmetry is observed due to the electron depletion
of the lone-pair in the chlorine p-orbital perpendicular to the
aromatic plane. Conversely, the large torsional angle between
the ester group and the aromatic system in W-I prevents the
mutual π-systems from overlapping, thereby reducing the
electron depletion at the chlorine. The resulting electron
distribution leads to a lower quadrupolar coupling constant and
a lower asymmetry in W-I as observed.
HF D95**
35.44 38.67 0.04
35.67 38.72 0.04
W-I a -74.11
W-I b -74.39
0
0
1
-122
-108
-91
61
74
89
Y-I
-79.95
32.35 47.6
0.19
HF D95** Three Molecules
W-I a -74.92
W-I b -74.13
35.76 39.14 0.05
35.33 38.8 0.05
32.4 48.13 0.20
0
0
1
-107
-97
-92
75
85
89
Y-I
-80.53
a Because of the inversion symmetry of the EFG tensor, the residual
dipolar coupling observed in the CPMAS spectra is independent of
the sign. b Taken from ref 22. c The angle φ has only a small effect on
the simulation because of the small asymmetry η.
interaction angle in the EFG principal axis system between the
nuclei. In the case of Y-I the couplings between the 35,37Cl and
the COH and Ci position result in broadened resonances at 9.4
T. For W-I distorted doublet and triplet resonances are observed
for COH and Ci, respectively. Simulations of the residual coupling
in W-I reveal that doublet resonances are expected for these
two positions at 9.4 T. The triplet resonance for Ci is best
explained by two overlapping doublets resulting from the two
Ci of the two molecules per asymmetric cell having an isotropic
The Towns and Dailey analysis can be used to calculate the
fractional π-bond population Fπ of the C-Cl bond from the
quadrupolar coupling:
(42) Zilm, K. W.; Grant, D. M. J. Am. Chem. Soc. 1981, 103, 2913.

Conformational Polymorphism Studies by SS NMR
J. Am. Chem. Soc., Vol. 123, No. 8, 2001 1719
2 e2Qq
Fπ ) |
| ‚ η
2
3
e Qqo
where e2Qqo/h ) 109.6 MHz is the quadrupolar coupling of
the chlorine atom.32 For W-I and Y-I Fπ is calculated to be 1.7
and 6.7%, respectively.43
Isotropic 13C Chemical Shifts. The conformational differ-
ences of the two polymorphic forms result in substantial
isotropic chemical shift differences ranging from 3.8 to 10.3
ppm for the aromatic carbons COH, CCl and Ci (Table 1, Figure
5). The isotropic chemical shifts of CO2 and CMe on the other
hand show only a small effect (<2.2 ppm) due to the changes
in conformation. The isotropic chemical shifts in solution exhibit
a strong solvent effect, and individual shifts in different solutions
differ by several ppm. When the 13C shifts in different solvents
are compared to the CPMAS data of the two polymorphic forms,
it is readily seen that solutions of I in the nonpolar solvents
such as chloroform and benzene and Y-I have similar isotropic
chemical shifts. Solutions of I in polar hydrogen-bond-forming
solvents such as methanol and DMSO exhibit isotropic chemical
shifts similar to those measured in W-I. This observation reveals
similarities between the conformation of the two polymorphic
forms and the conformations and hydrogen bond arrangements
present in solution. In nonpolar solvents, where the solvent does
not allow intermolecular hydrogen bonds, the planar conforma-
tion with an intramolecular hydrogen bond is lowest in energy.
On the other hand, hydrogen-bond-forming solvents promote
the formation of intermolecular hydrogen bonding between
solvent and solute. Because of steric crowding in the aromatic
plane, the conformation with a large torsional angle should then
be favored, resulting in a conformation similar to W-I. A
thorough interpretation of the isotropic chemical shifts in
solution, however, would require equilibrium between several
species that exceeds the scope of this investigation. Byrn et al21
has already mentioned similar hydrogen-bond patterns in the
different crystalline and solution species. However, the previous
findings were based on spectroscopic data on different solutions
and on the X-ray crystal structure.
13C Chemical Shift Principal Values. The FIREMAT14
experiment is a 2D solid-state NMR method that provides the
isotropic spectrum in one dimension and the CSA information
in the other dimension. Since the chlorine-coupled resonances
are split into multiplets, a complication arises when the chemical
shift principal values are determined from the FIREMAT
dataset. For each peak of the multiplets a set of principal values
is obtained. As already mentioned, doublet-like resonances are
observed for COH and CCl in W-I and CCl in Y-I due to residual
dipolar coupling. The averaged principal values for these
resonances are obtained by averaging each component over the
two obtained values. The situation is more complicated for the
Ci position in W-I, where a triplet is observed. The simulation
of the residual dipolar coupling to the chlorine shows that this
triplet is due to a overlap of two doublet-like resonances
resulting from two different 13C positions, each of which
represents one of the two molecules per unit cell in W-I. To
obtain the principal shift values for the different molecules per
unit cell both the high-field principal values and the low-field
principal values are averaged with the center principal values
(Table 1 and Figure 5).
Figure 5. Experimental chemical shift principal values and isotropic
chemical shifts for W-I and Y-I. Solid lines represent the chemical
shift principal values and the isotropic chemical shifts are represented
by dashed lines. The isotropic chemical shift and δ22 of CCl in W-I
overlap.
π-system. Consequently, all chemical shift values are affected,
as the chemical shift is a very sensitive probe of the local
electronic environment. Exceptionally large shift differences up
to 14 ppm between the two forms are observed for δ22 at the
COH position and for δ33 at the Ci position. Even in the less
sensitive methoxy-carbon shifts, differences of 2-3 ppm are
measured. From Figure 5 it is readily seen that the conforma-
tional differences of the two polymorphic forms are better
resolved by the chemical shift principal values than by the
isotropic chemical shift alone. In particular, the substantial
differences in the CO2 environment are not well characterized
by the isotropic chemical shift, with a difference of only 1.3
ppm between W-I and Y-I. However, the δ11 and δ22 principal
values differ by -9.3 and +11.1 ppm, respectively.
The effect of hydrogen bonding on the 13C chemical shift
principal values of CO2H groups of amino acids has been
previously studied.44,45 It was found that the strength of the
hydrogen bond affects the δ11 and δ22 chemical shift principal
values, with a stronger hydrogen bond shifting δ11 to lower and
δ22 to higher chemical shift values. The δ33 value is found to
be unaffected by changes in hydrogen-bond strength. The δ33
values reported in the literature (av 109 ppm) agree with the
δ33 values of CO2 determined for W-I and Y-I. Therefore, the
lower δ11 and higher δ22 chemical shift for CO2 in Y-I may be
explained by the shorter OCdO‚‚‚H distance and stronger
intramolecular hydrogen bond in Y-I than in W-I.
The strong hydrogen bond in Y-I also results in higher δ11
and δ22 values for COH than in W-I, whereas the δ33 is again
unaffected by the difference in hydrogen bonding. However, a
The differences in the torsional angle of the ester group, the
hydrogen bonding, and the change in bond length (especially
the C-Cl bond) affect the conjugation for the entire aromatic
(44) Gu, Z.; Zambrano R.; McDermott, A. J. Am. Chem. Soc. 1994, 116,
6368.
(43) Bersohn, R. J. Chem. Phys. 1954, 22, 2078.
(45) Facelli, J. C.; Gu, Z.; McDermott, A. Mol. Phys. 1995, 86, 4, 865.

1720 J. Am. Chem. Soc., Vol. 123, No. 8, 2001
Strohmeier et al.
Table 4. Correlation between Experiment and Calculated Chemical Shift Principal Values
W-I a
three molec.
W-I b
three molec.
Y-I
combineda
one molec.
one molec.
one molec.
three molec.
one molec.
three molec.
std. dev./ppm
slope
8.2
-0.9530
183.7
6.0
0.9832
185.9
0.9931
8.6
-0.9624
183.7
5.9
-0.9890
186.0
5.7
-0.9595
183.0
4.5
-0.9951
188.5
7.3
-0.9583
183.5
5.3
-0.9891
186.8
intercept/ppm
R2
0.9865
0.9856
0.9934
0.9936
0.9962
0.9885
0.9941
a The experimental and theoretical values for all three structures are correlated together.
systematic study of the effect of hydrogen-bond strength on the
chemical shift principal values of alcohol carbons is not
available. It is difficult to attribute other differences in the
chemical shift principal values to specific structural differences
between the two polymorphic forms, because of the extensive
conformational differences and the lack of systematic investiga-
tions of the effect of conformation on the chemical shift principal
values.
The above analysis clearly shows that the chemical shift
principal values are very useful for the investigation of poly-
morphism, as they provide a more accurate characterization of
differences in conformation and intermolecular interactions such
as hydrogen bonding than the isotropic chemical shifts. Inves-
tigating the chemical shift principal values provides comple-
mentary data to X-ray methods, especially when the difference
between polymorphic forms constitutes of hydrogen bonding,
Chemical Shift Tensor Calculations. In Table 4 the slopes
and intercepts as well as the standard deviation are given for
the correlation of the various calculations with the experimental
data. The slopes (-0.9530 to -0.9951), intercepts (183.0 to
188.5 ppm), and standard deviations (4.5 to 8.2 ppm) of the
fits are close to those previously found for B3LYP D95**.46,47
The intercepts compare favorably with the calculated shielding
of methane at the same level of theory and basis set (192.4
ppm) when properly adjusted to the TMS scale by subtracting
the 7 ppm difference between gas-phase methane and TMS
resulting in an intercept of 185.4 ppm.48 Figure 6 shows the
correlation plot between the experimental principal shift values
and the calculated shielding values using the B3LYP exchange
functional on a single W-I a, W-I b, and Y-I molecule and on
stacks of three molecules. In Table 5 the converted chemical
shift principal values for the calculations on stacks of three
molecules are given to allow comparison with the experimentally
determined chemical shifts.
Figure 6. Correlation between calculated shielding principal values
and experimental chemical shift principal values. The theoretical values
are obtained with the B3LYP D95** method. The correlation is done
on the experimental and theoretical values of W-I a, W-I b, and Y-I
combined.
Table 5. Calculated Chemical Shift Tensor Principal Values for
W-I a, W-I b, and Y-1 Using the B3LYP D95** Method on a
Stack of Three Molecules/ppm
δ11
δ22
δ33
δiso
δ11
δ22
δ33 δiso
CO2
Ci
W-I a 258.5 140.7 103.3 167.5 W-I a 187.3 150.2 40.7 126.1
W-I b 256.7 136.7 103.9 165.8 W-I b 193.8 150.3 43.3 129.1
Y-I
249.6 153.9 100.5 168.0 Y-I
175.5 143.3 25.5 114.8
COH
CMe
W-I a 211.7 155.2 59.4 142.1 W-I a 88.5 75.8 2.9 55.7
W-I b 214.4 154.6 60.2 143.1 W-I b 88.3 76.9 2.5 55.9
Y-I
222.0 170.3 59.5 150.6 Y-I
86.2 66.5 2.4 51.7
The agreement between experiment and theory is improved
by increasing the size of the calculated system to a stack of
three molecules (Table 4). The molecules were chosen to include
the intermolecular hydrogen bonds in the case of W-I. For Y-I
where there are only weak intermolecular interactions, the
nearest neighboring molecules were chosen to obtain a similar
molecular arrangement (Figure 1). The correlation and the
standard deviation between experiment and calculated values
for W-I a and W-I b are considerably improved when the
calculations are done on a stack of three molecules. The
correlation of the calculated data obtained on one molecule for
Y-I already give a satisfactory correlation with the experimental
data, and the improvement in going to the three-molecule stack
is considerably smaller than for W-I a and W-I b. An
improvement due to basis set superposition cannot account for
the substantial better correlation observed for W-I a and W-I
b when the calculations are done on stacks of molecules, as
CCl
W-I a 184.0 131.6 53.2 123.0
W-I b 189.7 132.3 53.0 125.0
Y-I
180.1 134.6 60.0 124.9
typical basis set superposition errors only account for changes
less than 2 ppm in the components.49,50
In Figure 7 the improvement in the B3LYP D95** results
obtained by including the hydrogen bonding is shown graphi-
cally, allowing for a more refined discussion of the observed
improvements. Comparison of the calculations done on isolated
molecules and stacks of molecules reveals that substantial
improvement of 5 to 16 ppm is achieved for those chemical
shift principal values (δ11 and δ22 of CO2) which are known to
be affected by intermolecular hydrogen bonding as encountered
in the white polymorphic form. In addition, an improvement of
7 and 10 ppm is achieved for the δ22 values of COH in W-I a
(46) Harper, J. K.; McGeorge, G.; Grant, D. M. J. Am. Chem. Soc. 1999,
212, 6488.
(47) Barich, D. H.; Orendt, A. M.; Pugmire, R. J.; Grant, D. M. J. Phys.
Chem. A 2000, 104, 8290, and previous papers in series.
(48) Jameson, A. K.; Jameson, C. J.. Chem. Phys. Lett. 1987, 134, 461.
(49) Clementi, E.; Corongin, G.; Chakravorty, S. In Modern Techniques
in Computational Chemistry; Clementi, E., Ed.; Escom: Leiden, 1990; p
343.
(50) Facelli, J. C. Chem. Phys. Lett. 2000, 322, 91.

Conformational Polymorphism Studies by SS NMR
J. Am. Chem. Soc., Vol. 123, No. 8, 2001 1721
methods. The fact that the calculated EFG changes by about
0.4-0.6 MHz in øzz and by about 0.07-0.13 in η, when the
calculations are done on stacks of molecule, suggests that lattice
effects have a substantial effect on the EFG tensor and should
not be neglected. The large differences for the orientation of
the x and y principal axis with respect to the molecular frame
for W-I for different levels of theory and different system sizes
is due to the fact that the asymmetry is small and differences in
theory and system size have, therefore, a significant effect on
the orientation of the tensor. These differences, however, have
a small impact on the observed spectra, because of the relatively
small asymmetry.
Conclusions
Figure 7. Improvement achieved when the system size is increased
from a single molecule to a stack of three. The improvement is defined
as: Improvement ) |PV one molecule - PV experiment| - |PV three
molecules - PV experiment|. PV one molecule: Chemical shift
principal values obtained from B3LYP D95** calculations done on
one molecule. PV three molecules: Chemical shift principal values
obtained from B3LYP D95** calculations done on a stack of three
molecules. PV experiment: Experimentally determined chemical shift
principal values.
The carbon isotropic chemical shift values and the chlorine
electric field gradient (EFG) tensor information are extracted
from the 13C MAS spectra, and the principal values of the
chemical shift tensors of the carbons are obtained from 2D
FIREMAT experiments on two of the three known polymorphs
of dimethyl-3,6-dichloro-2,5-dihydroxyterephthalate. These solid-
state NMR parameters are shown to be very sensitive to the
changes in geometry, electronic structure, and hydrogen bonding
between the two polymorphs. The same data are obtained from
quantum chemical calculations of the chemical shift and the
EFG tensors. The 13C MAS isotropic chemical shift values are
compared to isotropic chemical shifts of solutions of dimethyl-
3,6-dichloro-2,5-dihydroxyterephthalate in different solvents.
When I is dissolved in nonpolar solvents, the isotropic chemical
shifts are similar to those observed for Y-I. Solutions of I in
polar hydrogen-bond-forming solvents have isotropic chemical
shifts similar to the isotropic shifts of W-I.
Differences in the isotropic chemical shifts between the two
polymorphs from 1 to 9 ppm are observed, with the largest
differences found for the carbon of the aromatic ring. However,
the differences in the individual chemical shift tensor compo-
nents exhibit better structural sensitivity and are more specific
when conformational differences are investigated. For the ester
carbon, the isotropic chemical shift difference is only 1 ppm,
while the differences in the δ11 and δ22 are each about 10 ppm,
but in opposing directions. The difference in the chemical shift
tensor principal values for this carbon between the two forms
is well understood in terms of the hydrogen-bonding interaction
that is much stronger in the white form.
The calculations of the chemical shift tensor were completed
using the X-ray geometries with adjusted C-H distances on
both isolated molecules and on stacks of three molecules chosen
such that the intermolecular hydrogen bonding present in the
white form was included. In going from a single molecule to
the stack of three in the white form there was a significant
improvement in agreement between theory and experiment. For
the yellow form, in which there are no significant intermo-
lecular interactions, the two results were fairly similar. More
importantly, the improvement observed in the white form was
found in those components of the ester carbon, which are known
to be sensitive to hydrogen bonding, and the δ22 of the COH
carbon.
Due to the presence of the chlorine atoms in this molecule a
residual dipolar coupling between 35,37Cl and the neighboring
carbons is observed in the 13C MAS spectra. The simulation of
the MAS spectra of both polymorphic forms gives accurate
isotropic chemical shift values, data on the EFG tensors of the
chlorine and the dipolar coupling constants. Spectra were
recorded at three different magnetic field strengths, and the
spectral parameters were varied to achieve an excellent fit on
and W-I b, respectively. In the yellow polymorphic form,
however, no significant improvements for the chemical shift
principal values of any carbons, except the δ22 value of CMe, is
obtained with the addition of the neighboring molecules,
implying that the intermolecular interactions are negligible.
The calculated shielding values of carbons adjacent to chlorine
are also strongly affected by the size of the calculated system
in both polymorphic forms, but the δ22 value of CCl in the white
polymorph deteriorates recognizable. Considerably different
quadrupolar coupling constants are obtained for the isolated and
stacked molecules, indicating that the lattice substantially
influences the polarizable chlorine electron cloud (Table 3). The
systematic uncertainty in the experimental CCl chemical shift
principal values, however, does not allow a highly refined
interpretation.
The improvement in W-I is exceptional, because of the strong
intermolecular interactions present. Hence, the agreement
between calculated and experimental chemical shifts may be
substantially improved by explicitly including strong intermo-
lecular interactions in the calculations.51-54
EFG Tensor Calculations. The calculated tensors for W-I
employing the B3LYP functional and a stack of three molecules
show good agreement with the orientations and values found
to give the best fit between simulation and recorded spectra
(Table 3). The values calculated for Y-I, however, differ by a
significant amount from the experimentally determined qua-
drupolar coupling constant and asymmetry. The agreement
between experimental and theoretical data may be improved
by a correlation. However, a reliable correlation on extensive
experimental and theoretical data is only available for MP2
TZVP calculations,55 which are not feasible on the relatively
large stacks of three molecules of I.
The trends, however, observed for the differences between
the two forms in the quadrupolar coupling are reproduced by
the calculations for both the HF D95** and B3LYP D95**
(51) Orendt, A. M.; Facelli, J. C.; Grant, D. M. Chem. Phys. Lett. 1999,
302, 499.
(52) Hu, J. Z.; Facelli, J. C.; Alderman, D. W.; Pugmire, R. J.; Grant,
D. M. J. Am. Chem. Soc. 1998, 120, 9863.
(53) Facelli, J. C.; Pugmire, R. J.; Grant, D. M. J. Am. Chem. Soc. 1996,
118, 5488.
(54) de Dios, A. C.; Oldfield, E. Solid State NMR 1996, 6, 101.
(55) Palmer, M. H.; Blair-Fish, J. A. Z. Naturforsch. 1998, 53a, 370.

1722 J. Am. Chem. Soc., Vol. 123, No. 8, 2001
Strohmeier et al.
all three. The larger asymmetry found for the yellow polymor-
phic form can be explained by the small torsional angle between
the ester group and the aromatic ring, resulting in a electron
depletion in the chlorine lone-pair p-orbital perpendicular to the
aromatic ring. The trends observed for the differences in the
EFG tensor between the two polymorphic forms are also found
in the quantum chemical calculation of the EFG tensor at the
chlorine. However, the absolute agreement between calculations
and experiment is only moderate and can certainly be improved
by correlation between calculated experimental data, when more
theoretical data at the level of theory used are available.
Acknowledgment. This research was supported by the
Institute of General Medical Sciences, NIH, under Grant No.
GM08521-39 and by the office of Basic Energy Sciences, DOE,
under Grant No. DE-FG-94ER14452. An allocation of computer
time from the Center for High Performance Computing at the
University of Utah is gratefully acknowledged. CHPC’s SGI
Origin 2000 system is funded in part by the SGI Supercomputing
and Visualization Center Grant.
JA003599B