Transient states in [2 + 2] photodimerization of cinnamic acid: Correlation of solid-state NMR and X-ray analysis

Article abstract of DOI:10.1021/ja0773711

¹³C-CPMAS and other solid-state NMR methods have been applied to monitor the solid-state reactions of trans-cinnamic acid derivatives, which are the pioneer and model compounds in the field of topochemistry previously studied by X-ray diffracti

Full text of DOI:10.1021/ja0773711

Published on Web 01/12/2008
Transient States in [2 + 2] Photodimerization of Cinnamic
Acid: Correlation of Solid-State NMR and X-ray Analysis
Mujeeb Khan, Gunther Brunklaus,* Volker Enkelmann, and Hans-Wolfgang Spiess
Max-Planck-Institut f u¨ r Polymerforschung, Postfach 31 48, D-55021 Mainz, Germany
Received September 24, 2007; E-mail: brunklaus@mpip-mainz.mpg.de
Abstract: 13C-CPMAS and other solid-state NMR methods have been applied to monitor the solid-state
reactions of trans-cinnamic acid derivatives, which are the pioneer and model compounds in the field of
topochemistry previously studied by X-ray diffraction, AFM, and vibrational spectroscopy. Single-crystal
X-ray analyses of photoirradiated R-trans-cinnamic acid where the monomers are arranged in a head-to-
tail manner have revealed the formation of a centrosymmetric R-truxillic acid photodimer. For a
centrosymmetric dimer, however, two cyclobutane carbon signals and one carbonyl carbon signal were
expected apart from other aromatic carbon signals. Instead, four cyclobutane and two carbonyl carbon
signals were observed suggesting the formation of a non-centrosymmetric photodimer. Removing hydrogen
bonds from the system by esterfication of R-truxillic acid yield a centrosymmetric photodimer. Careful analysis
of the obtained products via solid-state NMR clearly showed that the observed peak splittings in the 13C-
CPMAS spectra did not originate from packing effects but rather result from asymmetric hydrogen bonds
distorting the local symmetry. Further evidence of this rather dynamic hydrogen-bonding stems from high-
temperature X-ray data revealing that only the joint approach of both X-ray analysis and solid-state NMR
at similar temperatures allows for the successful characterization of dynamic processes occurring in
topochemical reactions, thus, providing detailed insight into the reaction mechanism of organic solid-state
transformations.
1
. Introduction
Numerous studies have been pursued on this class of compounds
for basic design, planning, and understanding of the solid-state
[2 + 2] Photodimerizations are among the most extensively
8
3
reactions. Schmidt et al. have classified the packing of trans-
cinnamic acid and its derivatives in three different forms: an
R-form (where monomers are arranged in a head-to-tail manner),
â-form (with the monomers arranged head-to-head), and a
γ-form (where the monomers are unfavorably aligned). While
in both the R- and â-forms the reactive double bonds approach
each other by less than approximately 4 Å and which are
photoactive yielding head-to-tail and head-to-head dimers,
respectively (Scheme 1), the γ-form is photostable. The topo-
tactic nature of the obtained reaction products has been revealed
by applying various techniques such as atomic force microscopy
1-2
studied organic solid-state reactions. A first systematic study
concerning the relationship between the chemical reactivity of
an organic solid and its crystal structure has been reported by
3
Schmidt and co-workers on the photodimerization of cinnamic
acid and its derivatives. The range of reactions that have since
then been reported to be topochemically controlled has increased
significantly but naturally cannot compete with synthetic
chemistry in solution. What has given particular impetus to the
study of organic solid-state reactions is the appreciation that
the “frozen” state of the reactant can allow a very detailed
4
exploration of the mechanistic aspect of the solid-state reactions.
9
10
11,12
(
AFM), vibrational spectroscopy, X-ray diffraction,
and
Other developments in the areas of morphology control, crystal
growth inhibition, crystal-to-crystal conversions,5 absolute
asymmetric synthesis, and, more recently, supramolecular
design and molecular recognition have all impinged on the
development of organic solid-state chemistry.
1
3-16
solid-state NMR.
Since AFM is primarily sensitive to
(8) Ghosh, M.; Chakrabarti, S.; Misra, T. N. J. Raman Spectrosc. 1998, 29,
63-267. (b) Atkinson, S. D. M.; Almond, M. J.; Bowmaker, G. A.; Drew,
2
6
M. G. B.; Feltham, E. J.; Hollins, P.; Jenkins, S. L.; Wiltshire, K. S. J.
Chem. Soc., Perkin Trans. 2 2002, 9, 1533-1537. (c) Davaasambuu, J.;
Busse, G.; Techert, S. J. Phys. Chem. A 2006, 110, 3261-3265. (d) Jenkins,
S. L.; Almond, M. J.; Atkinson, S. D. M.; Drew, M. G. B.; Hollins, P.;
Mortimore, J. L.; Tobin, M. J. J. Mol. Struct. 2006, 786, 220-226.
9) Kaupp, G. Angew. Chem., Int. Ed. Engl. 1992, 31, 592-595. (b) Kaupp,
G. Angew. Chem. 1992, 104, 606-609.
The [2 + 2] photodimerization of trans-cinnamic acid (and
7
many of its derivatives ) constitutes the classic example of a
solid-state reaction that conforms to the topochemical principle.
(
(
10) Samantha, D. M.; Matthew, J. A.; Bruneel, J.; Amdrew, G.; Hollins, P.;
Mascetti, J. Spectrochim. Acta 2000, A56, 2423-2430.
(
(
(
(
1) Tanaka, K.; Toda, F. Chem. ReV. 2000, 100, 1025-1074.
2) Ramamurthy, V.; Venkatesan, K. Chem. ReV. 1987, 87, 433-481.
3) Schmidt, G. M. J. Pure Appl. Chem. 1971, 27, 647-678.
4) McBride, J. M.; Segmuller, B. E.; Hollingsworth, M.; Mills, D. E.; Weber,
B. A. Science 1986, 234, 830-835.
5) Tomislav, F.; MacGillivray, L. R. Z. Kristallogr. 2005, 220, 351-363.
6) Desiraju, G. J. Angew. Chem., Int. Ed. Engl. 1995, 34, 2311-2327.
7) Schmidt, G. M. J. J. Chem. Soc. 1964, 2014-2021.
(11) Enkelmann, V.; Wegner, G.; Novak, K. J. Am. Chem. Soc. 1993, 115,
10390-10391.
(12) Abdelmoty, I.; Buchholz, V.; Di, L.; Enkelmann, V.; Wegner, G.; Foxman,
B. M. Cryst. Growth Des. 2005, 17, 2210-2217.
(13) Stitchell, S. G.; Harris, K. D. M.; Aliev, A. E. Struct. Chem. 1994, 5, 327-
333.
(
(
(
(14) Hilgeroth, A.; Hempel, G.; Baumeister, U.; Reichert, D. Solid State Nucl.
Magn. Reson. 1999, 13, 231-243.
10.1021/ja0773711 CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 1741-1748
9
1741

A R T I C L E S
Khan et al.
Scheme 1. Diagram Illustrating the Conversion of the R- and
â-Forms of trans-Cinnamic Acid to Dimers via Topotactic
Reactions
provides in-depth information about the topotactic nature of the
reactions as well as structural changes that occurred at the
molecular level during the reaction.
In this paper, we return to certain “missing links” in the
evolving story of crystalline state [2 + 2] cycloadditions. We
have applied single-crystal X-ray analysis as well as solid-state
NMR to monitor the [2 + 2] photodimerization of R-trans-
cinnamic acid to R-truxillic acid. Crystal structures1 suggest
that R-truxillic acid is a centrosymmetric photodimer. During
1,12
1
3
the course of our study, however, C-CPMAS spectra were
reported for different conversion ratios ranging from R-trans-
1
5
cinnamic acid (0%) to R-truxillic acid (100%). In contrast to
an anticipated centrosymmetric molecular geometry of R-trux-
1
3
illic acid, the C-CPMAS spectra exhibited additional signals
for the cyclobutane ring and carbonyl region, which Hayes et
1
5
al. attributed to solid-state packing effects, e.g., distortions in
the cyclobutane ring and/or dihedral angle twists of both the
phenyl rings and carboxylic acid group. Alternatively, however,
the observed signal splittings may originate from (possibly
asymmetric) hydrogen bondings, which not only are known to
1
3
15
22,23
distinctly impact C or N spectra
but also more impor-
tantly are revealed by characteristic chemical shifts of up to 21
ppm in the respective solid-state proton NMR spectra.
surfaces, structural information cannot be obtained directly. In
contrast, vibrational spectroscopy monitors the topotactic nature
of organic solid-state reactions, but conformational details of
possible intermediate states remain vague.
23,24
Indeed, the comparison of both X-ray and NMR data turned
23,25
out to be very fruitful in many different cases,
care of the temperature settings must be taken.
but proper
X-ray crystallography is an excellent tool to investigate
photodimerization reactions of single crystals. However, many
crystals tend to shatter into microcrystalline particles as the
dimerization proceeds, rendering an X-ray analysis of obtained
products difficult. Previously, we reported the topochemically
controlled single-crystal-to-single-crystal transformation of
R-trans-cinnamic acid into R-truxillic acid. Utilizing a selective
reaction procedure where the photoactive crystals are irradiated
by light for which they have low absorptivity, even crystals
In order to clearly distinguish the two possible sources of
peak splitting, we have synthesized not only mixed crystals of
photodimers at various conversion ratios but also the methyl
ester of truxillic acid. Moreover, as the hydrogen bonding can
be of dynamic nature, the X-ray structures of photodimerized
R-trans-cinnamic acid obtained at low (e.g., liquid helium-
cooled) and high temperatures (350 K) are compared and
11
13
discussed with respect to the recorded C-CPMAS NMR
spectra.
(and their corresponding X-ray structures) at intermediate stages
of the photodimerization were obtained. These can be character-
ized as substitutional mixed crystals in which the dimer and
monomer pair occupies the same lattice site with an occupancy
according to the conversion. A severe limitation of this technique
stems from the fact that it is not suitable for the characterization
of either amorphous or rather ill-defined solids that lack long-
range translational order. In addition, the localization of lighter
2. Experimental Section
trans-Cinnamic acid (Aldrich, 99%) was recrystallized from acetone
(Aldrich) to obtain the R-polymorph. The crystals were ground to small
microcrystalline particles, and about 400 mg of product were evenly
distributed in a thin layer on a 100 mm Petri dish and placed in the
focus of the lamp for irradiation. The powder sample was exposed to
broad band irradiation (material exposed to both tail irradiation and
broad band irradiation produced the same structure, except that samples
obtained by tail irradiation resulted in intact crystals suitable for single-
crystal X-ray analysis) using a 100 W high-pressure Hg lamp (Muller
electronic LXH100) for a period of 10 h to obtain 100% converted
1
1
atoms (e.g., hydrogen-bonded protons)17 may be difficult even
with sophisticated X-ray diffraction methods.18
Recently, it has been demonstrated that solid-state NMR can
also be successfully applied to monitor organic solid-state
photoreactions.1
3-16
Its unique selectivity allows for the dif-
(
22) Schilf, W.; Kamie n´ sky, B.; Szady-Chełmieniecka, A.; Grech E. J. Mol.
1
9
ferentiation of chemically distinct sites, including protons, on
the basis of the NMR chemical shift.20 Thus, solid-state NMR
Struct. 2004, 700, 105-108.
(23) Gobetto, R.; Nervi, C.; Chierotti, M. R.; Braga, D.; Maini, L.; Harris, K.
R.; Hodgkinson, P. Chem.sEur. J. 2005, 11, 7461-7471.
reveals molecular species present in the sample, including
products, reactants, and possible side products as well as
(24) (a) Li, S.; Zheng, S.; Su, Y.; Zhang, H.; Chen, L.; Yang, J.; Ye, C.; Deng,
F. J. Am. Chem. Soc. 2007, 129, 11161-11171. (b) Yates, J. R.; Pham, T.
N.; Pickard, C. J.; Mauri, F.; Amando, A. M.; Gil, A. M.; Brown, S. P. J.
Am. Chem. Soc. 2005, 127, 10216-10220. (c) Riedel, K.; Leppert, J.;
Ohlenschlaeger, O.; Goerlach, M.; Ramachandran, R. J. Biomol. NMR 2005,
2
1
polymorphs. Hence, the combination of solid-state NMR and
single crystal) X-ray analysis to study topochemical reactions
(
3
1, 331-336. (d) Brus, J.; Dybal, J. Macromol. 2002, 35, 10038-10047.
(
e) Brown, S. P.; Zhu, X. X.; Saalwaechter, K.; Spiess, H. W. J. Am. Chem.
(
15) Bertmer, M.; Nieuwendaal, R. C.; Barnes, A. B.; Hayes, S. E. J. Phys.
Soc. 2001, 123, 4275-4285. (f) Schnell, I.; Brown, S. P.; Low, H. Y.;
Ishida, H.; Spiess, H. W. J. Am. Chem. Soc. 1998, 120, 11784-11795.
(25) (a) Fyfe, C. A; Brouwer, D. H.; Lewis, A. R.; Villaescusa, L. A.; Morris,
R. E. J. Am. Chem. Soc. 2002, 124, 7770-7778. (b) Liu, H.; Roger, M.;
Hanson, J. C.; Grey, C. P.; Martin, J. D. J. Am. Chem. Soc. 2001, 123,
7564-7573. (c) Grey, C. P.; Poshni, F. I.; Gualtieri; A. F.; Norby, P.;
Hanson, J. C.; Corbin, D. R. J. Am. Chem. Soc. 1997, 119, 1981-1989.
(d) Schaller, T.; Buechele, U. P.; Klaerner, F. G.; Blaeser, D.; Boese, R.;
Brown, S. P.; Spiess, H. W. J. Am. Chem. Soc. 2007, 129, 1293-1303.
Chem. B 2006, 110, 6270-6273.
(
16) Harris, K. D. M.; Thomas, J. M. J. Solid State Chem. 1991, 94, 197-205.
17) Brown, S. P.; Spiess, H. W. Chem. ReV. 2001, 101, 4125-4155. (b) Harris,
R. K. Solid State Sci. 2004, 6, 1025-1037.
(
(18) Harris, K. D. M.; Cheung, E. Y. Chem. Soc. ReV. 2004, 33, 526-538.
(19) Brown, S. P. Prog. Nucl. Magn. Reson. Spectrosc. 2007, 50, 199-251.
(20) Harris, R. K. Solid State Sci. 2004, 6, 1025-1037.
(
21) Harris, R. K. The Analyst 2006, 131, 351-373.
1742 J. AM. CHEM. SOC.
9
VOL. 130, NO. 5, 2008

[
2
+
2] Photodimerization of Cinnamic Acid
A R T I C L E S
Figure 1. O-H group conformation in carboxylic acids and packing mode of R-trans-cinnamic acid in crystal lattice. (a) Synplanar conformation of
carboxyl group. (b) Antiplanar conformation. (c) Schematic view of the R-modification of cinnamic acid.
1
2
2
27
as-dimerized truxillic acid (P2
powder diffraction data provided in the Supporting Information.
Powder diffraction data are also given for the Donnay polymorph of
1
/n phase ), which was confirmed by
Static, variable-temperature H quadrupolar echo spectra were recorded
at 46.07 MHz using a home-built 7 mm probe and a Bruker Avance-II
300 spectrometer. The interpulse delay was set to 20 µs, and the π/2-
pulse length, to 2 µs. The temperature was varied from 240 to 400 K.
Typically, 8096 scans were accumulated.
(
1
2
truxillic acid (C2/c phase ) that was obtained by recrystallization of
small amounts of 100% converted as-dimerized truxillic acid from
acetone.) About 100 mg of product were taken out at each interval of
Solution NMR and Mass: The solution NMR spectra were recorded
on a Bruker AC 250. The residual H peak of the deuterated solvent
was used as an internal standard (Acetone, H, δ ) 2.05 ppm). The
1
2
h for solid-state NMR measurements. In order to obtain fast
1
conversion, the powder was agitated after every 30 min and was
simultaneously rotated using a home-built small rotating machine.
Single crystals of R-trans-cinnamic acid were obtained by slow
evaporation of an acetone solution; crystals were then irradiated by a
tail irradiation technique using a 345 nm edge filter to obtain intact
mass spectrum was obtained using a Varian Mat 7a (70e.v.).
DFT-Based Chemical Shift Calculations: Geometry optimizations
were performed based on the available crystal structures using the BLYP
functional28 and 6-311G split valence basis set augmented with diffuse
and polarization functions. Subsequently, C chemical shifts with
respect to tetramethyl silane (TMS) were computed at the B3LYP/6-
311++G (2df, 2pd) level of theory with the GIAO approach as
implemented in the Gaussian98 program package.30
29
13
crystals of P2
1
/n phase truxillic acid. A suitable single crystal was
picked up from the sample, and crystal structures were determined on
that crystal at different temperatures to investigate the temperature
dependency of the bond length. The deuterated samples of trans-
cinnamic acid and truxillic acid were obtained by using ethanol-d
Roth). About 100 mg of product were dissolved in 5 mL of solvent
and left for slow evaporation.
6
(Carl
Single-Crystal X-ray Analysis: All X-ray structures were recorded
on a Kccd diffractometer with graphite monochromated Mo KR
radiation. Lattice parameters were obtained by least-squares fits to the
scattering angles of reflections observed in several prescans. The
intensity data collection was performed by φ and ω scans. The raw
data were corrected for Lorentz and polarization effects. The structures
were solved by direct methods and refined by full matrix least-squares
analyses with anisotropic temperature factors for all atoms except H.
Positions of the H atoms were calculated using the known molecular
geometry and refined in the riding mode with fixed isotropic temper-
ature factors. Empirical absorption corrections were applied to the data.
Synthesis of Methyl Ester of Truxillic Acid: Methyl ester of
truxillic acid (dimethyl 2, 4-diphenylcyclobutane-1, 3-dicarboxylate)
was synthesized by a simple esterfication reaction. Truxillic acid (250
mg, 0.84 mmol) was dissolved in 30 mL of methanol in a 50 mL round
2
bottomed flask; to this solution about 0.5 mL of thionyl chloride (SOCl )
was added, and the mixture was allowed to reflux until TLC analysis
indicated a complete reaction. The mixture was then concentrated in
vacuo, and the residue was taken up in 50 mL of methylene chloride
and washed with 3 × 50 mL of water. The organic layer was dried
3. Results and Discussion
2 4
over Na SO . Removal of the solvent in vacuo afforded a white powder
which was then purified over a short silica gel column using acetone
and petrol ether as eluent in a 1:9 ratio. Evaporation of the solvent
afforded 248 mg of a white crystalline solid (methyl ester of truxillic
acid) in 90% yield. Single crystals of methyl truxillic acid for X-ray
analysis were obtained by the slow evaporation of an acetone solution.
The carboxyl group in the majority of mono- and dicarboxylic
acids in its crystalline phase adopt either a synplanar (1a) or
antiplanar (1b) OdCsOsH conformation as shown in Figure
31,32
1
.
The synplanar structure is the more stable conformer while
1
the antiplanar form usually occurs when the OsH bond
participates in an intramolecular OsH‚‚‚O bond. R-trans-
Cinnamic acid, a nonsubstituted acid, which does not involve
H NMR, δ: 7.38-7.26 (M, 10H, Ar); 4.47 (T, 2H, Ar-CH-CH);
4
3
.02 (T, 2H, OOC-CH-CH); 3.30 (S, 6H, O-CH ). EI MS: m/z 324
+
(M ). Mp ) 175 °C.
Solid-State NMR: All 13C-CPMAS spectra were collected at 125.77
(
(
26) Morkombe, C. R.; Zilm, K. J. Magn. Reson. 2003, 162, 479-486.
27) Furo, I.; Hedin, N. J. Magn. Reson. 2001, 152, 214-216.
MHz (Bruker Avance 500 spectrometer), with a CP-contact time of 5
ms coadding 4096 transients. The experiments were carried out using
a standard Bruker 2.5 mm double resonance MAS probe spinning at
(28) (a) Becke, A. D. Phys. ReV. 1988, A38, 3098-3100. (b) Lee, C.; Yang,
W.; Parr, R. G. Phys. ReV. 1988, A37, 785-789.
(
29) Krishnan, R.; Binkley, J. S.; Seger, R.; Pople, J. A. J. Chem. Phys. 1980,
2
5 kHz, with a typical π/2-pulse length of 2.5 µs and a recycle delay
72, 650-654.
of 15 s. The spectra are referenced with respect to tetramethyl silane
(30) Frisch, M. J., et al. Gaussian 98, revision A.7; Gaussian Inc.: Pittsburgh,
PA, 1998.
26
(
TMS) using adamantane as a secondary standard (29.46 ppm for 13C).
(
31) Leiserowitz, L. Acta Crystallogr. 1976, B32, 775-802.
If not stated otherwise, all spectra were acquired at room temperature.
(32) Miyazawa, T.; Pitzer, K. S. J. Am. Chem. Soc. 1959, 81 (1), 74-79.
J. AM. CHEM. SOC. VOL. 130, NO. 5, 2008 1743
9

A R T I C L E S
Khan et al.
Figure 2. Solid-state 13C-CPMAS spectra ranging from R-trans-cinnamic acid to R-truxillic acid: (a) monomer prior to irradiation, (b) ∼25% converted
13
mixture, (c) ∼50% converted mixture, (d) dimer after complete conversion (as-dimerized, P21/n phase). C-CPMAS spectrum of another polymorph of
1
2
truxillic acid, i.e., recrystallized truxillic acid C2/c phase, was also measured and is provided in the Supporting Information. Surprisingly there is not a very
1
3
distinct difference between the C-CPMAS spectra of P21/n and C2/c phases which can be easily distinguished by their powder patterns provided also in
the Supporting Information.
in any kind of intramolecular hydrogen bonding adopts a
synplanar conformation with respect to its carboxyl group.
However, units of two monomers which are approximately in
the same plane and connected via intermolecular hydrogen
bonds form highly overlapped parallel stacks getting monomers
within certain distances and relative orientations into close
contact. In the R-form, the overlap offset is so large that the
closest monomer pairings in a reactive orientation are not in
the same stack but in neighboring stacks. Thus, while neighbors
within the same stack are related by mirror symmetry and are
over 5.5 Å apart, the reacting pairs are only 4 Å apart. Therefore,
in an R-lattice head-to-tail centrosymmetric dimers are formed
cyclobutane ring region. This prompted us to consider the
consequences of (possibly dynamic) hydrogen bonding present
in the crystals. Indeed, each R-truxillic acid photodimer is
hydrogen bonded to two of its neighboring molecules forming
1
2
a chainlike structure. Since X-ray methods suffer from the
fact that heavy atoms tend to dominate the observed diffraction
pattern, it is difficult to locate lighter atoms (e.g., hydrogen)
accurately. In this case, solid-state NMR can provide comple-
mentary information. Figure 2 shows a series of 13C-CPMAS
NMR spectra ranging from R-trans-cinnamic (0% conversion)
1
2
to as-dimerized R-truxillic acid (P2 /n phase, 100% conver-
1
sion), where Figure 2b and 2c are spectra for partial mixtures
11-12,33
with the least lattice disturbance.
containing both monomer and dimer at ∼25% and ∼50%
A series of 13C-CPMAS NMR spectra for R-trans-cinnamic
acid and the corresponding [2 + 2] photodimerization products
conversion, respectively.
As the dimerization proceeds, the corresponding intensities
of both the olefinic carbon and carbonyl carbon signals assigned
to the monomer decreased with a gradual rise of the new carbon
signals attributed to the cyclobutane ring indicating the suc-
cessful formation of a photodimer. Additionally, the chemical
shifts of the neighboring carbons, i.e., the ipso (non protonated)
carbon and the carbonyl carbon, display a small downfield shift
upon conversion of the monomer into a dimer, possibly due to
differences in inductive effects of neighboring groups (vinyl
vs cyclobutane). The complete assignments of the carbon sites
in all samples are given in Table 1. Two signals in the carbonyl
region and four different signals in the cyclobutane ring region
1
5
after various irradiation times were reported by Hayes et al.
They showed that solid-state NMR can be successfully applied
to study this class of photodimerization reactions because of
the good resolution of individual resonances and the large
separation of the signals of reactant and product functional
groups. In contrast to an anticipated centrosymmetric molecular
geometry of R-truxillic acid, the 13C-CPMAS spectra exhibited
additional signals for the cyclobutane ring and carbonyl region,
1
5
which Hayes et al. referred to as solid-state packing effects,
e.g., distortions in the cyclobutane ring and/or dihedral angle
twists of both the phenyl rings and carboxylic acid group.
Careful observation of the crystal structures of mixtures as well
as fully converted truxillic acid did not reveal any distortions
or packing defects in the crystal lattice.
are observed in both partially and fully converted dimer spectra
13
(
Figure 2b-d). In contrast, the solution C NMR spectrum
exhibits only two peaks in the cyclobutane ring region and a
single peak in the carbonyl region (as expected from centrosym-
metric molecular geometry). The latter resonance is downfield
shifted from 173 ppm in solution to about 179 ppm in the solid
state, probably due to hydrogen bondings that are absent in
solution.
Even computed 13C chemical shifts of the R-truxillic acid
dimer based on an optimized X-ray geometry revealed neither
any splitting in the carbonyl region nor any extra signals in the
(
33) Cohen, M. C.; Schmidt, G. M. J.; Sonntag, F. I. J. Chem. Soc. 1964, 2000-
2
013.
1744 J. AM. CHEM. SOC.
9
VOL. 130, NO. 5, 2008

[
2
+
2] Photodimerization of Cinnamic Acid
A R T I C L E S
Table 1. Resonance Assignments for R-trans-Cinnamic Acid,
R-Truxillic Acid, and R-Truxillic Acid Methyl Ester Based on ab
Initio Chemical Shift Calculations
sample
δ
(ppm)
assignment
R-trans-cinnamic acid 173.1
carbonyl carbon (C9)
vinylic carbon (C7)
ipso carbon (C1)
1
1
1
1
46.9
34.3
31.2-127.8 aromatic carbons (C2-C6)
19.4 vinylic carbon (C8)
-
R-truxillic acid
182.3
carbonyl carbon (COO )
carbonyl carbon (COOH)
ipso carbons (C1, C13)
aromatic (C2, C3, C14,
and C18)
1
1
1
78.9
38.0
29.6
1
1
26.6
aromatic (C4, C5, C15,
and C17)
aromatic (C6, C16)
Figure 4. Interconversion of the molecular structure of a carboxylic acid
dimer.34
25.5
5
4
4
4
2.1
6.2
5.1
1.3
cyclobutane ring carbon (C8)
cyclobutane ring carbon (C10)
cyclobutane ring carbon (C7)
cyclobutane ring carbon (C11)
carbonyl carbon (C11, C19)
Dynamics of Hydrogen Bonds in Carboxylic Acid Dimers:
After concluding that the distortion of the symmetry is mainly
caused by hydrogen bonds, we were concerned with the possible
dynamics of the hydrogen-bonded pair. In the past, numerous
studies have been published about the dynamics of hydrogen
atoms in carboxylic acid dimers suggesting various models to
R-truxillic acid
methyl ester
172.7
1
1
39.4
ipso carbon (C5, C13)
26.1-128.9 aromatic carbons
5
4
4
2.8
7.1
0.0
methyl carbons (C22, C24)
alphatic carbons (C2, C4)
aliphatic carbons (C1, C3)
describe the motion of the hydrogen along the hydrogen
34-38
bond.
The best suited approach in the case of R-truxillic
3
8
acid is given by Horsewill and Aibout. It states that (di-)-
carboxylic acids not only commonly yield centrosymmetric
dimers in the solid state where a ringlike structure is formed
by two hydrogen bonds (cf. Figure 4) but also arrange as
continuous polymeric chains of molecules stabilized by hydro-
gen bonds. This also exists in truxillic acid crystals.12 Within a
single dimer unit two conformations, labeled 1 and 2, are
possible. The interconversion between the two configurations
1
and 2 should involve a simultaneous proton exchange across
a symmetric transition structure X. Due to this effect the
potential energy of the system with respect to the hydrogen atom
position can conveniently be described with a double minimum,
while in the solid state the influence of neighboring molecules
breaks the symmetry of the potential so that the two potential
wells may in general be of unequal depth. It is possible for the
hydrogen atoms to move in a coordinated fashion within the
ring established by the two hydrogen bonds resulting in a change
of the electronic structure and atomic arrangement within the
dimer; e.g., a CdO bond turns into a CsO bond and Vice Versa.
In the case of R-truxillic acid, proton dynamics studies3
suggest that either a thermally activated proton transfer takes
place or tunneling occurs. This could be addressed in detail by
either T1 relaxation measurements or (quasi)inelastic neutron
scattering.
4-38
Figure 3. Solid-state 13C-CPMAS of truxillic acid methyl ester (dimethyl-
2
,4-diphenylcyclobutane-1,3-dicarboxylate).
To further explore possible influences of hydrogen bonding
in the solid compounds, we prepared the methyl ester of
R-truxillic acid. The corresponding 13C-CPMAS spectrum (cf.
Figure 3) unambiguously reflects a centrosymmetric photodimer;
that is, there are three signals in the aliphatic region: one is
assigned to methyl carbon and the other two signals belong to
the cyclobutane ring. Symmetry renders the two carbons
attached to the carbonyl groups or phenyl rings equivalent.
Structural Correlations from Solid State and Crystal
Structure Analysis: Considering the double minimum potential
approach, one might expect two different acidic proton peaks
1
in the H MAS NMR spectrum of the truxillic acid, but the
correponding spectrum (cf. Figure 5) displays only two sig-
nals: a broad signal with a small shoulder represents both the
aliphatic and aromatic protons, while the single peak at ∼14
ppm (fwhh ) 1000 Hz) is assigned to the strongly hydrogen-
13
Moreover, the C chemical shifts of the ester compound are
1
3
close to those C shift values obtained for R-truxillic acid in
solution (i.e., ∼173 ppm for carbonyl carbon). Taken together,
this strongly suggests that any observed excess signals are not
due to solid-state packing effects but rather due to asymmetry
of the (dynamic) hydrogen bonds which temporarily destroys
the center of symmetry of the R-truxillic acid photodimer.
(
(
34) Graf, F.; Meyer, R.; Ernst, R. R. J. Chem. Phys. 1981, 75, 2914-2918.
35) Meier, B. H.; Graf, F.; Ernst, R. R. J. Chem. Phys. 1982, 76, 767-774.
(36) Stoeckli, A.; Furrer, A. Physica 1986, 136B, 161-164.
(
(
37) Skinner, J. L.; Trommsdorff, H. P. J. Chem. Phys. 1988, 89, 897-907.
38) Horsewill, A. J.; Aibout, A. J. Phys.: Condens. Matter 1989, 1, 9609-
9622.
J. AM. CHEM. SOC.
9
VOL. 130, NO. 5, 2008 1745

A R T I C L E S
Khan et al.
Figure 5. 1H MAS NMR spectrum of R-truxillic acid measured at room
temperature.
Figure 7. A plot of the ∆(CsO) values as a function of temperature
showing that the truxillic acid CsO and CdO become less distinct with
an increase of temperature. The plot shows an approximate linear fit. CsO
and CdO distance variations vs temperature along with ∆(CsO) values
are explicitly listed in the Supporting Information.
Figure 6. Solid-state 2H spectrum of R-truxillic acid dimer at room
temperature (R ) (C6H5)2, C4H4). The quadrupolar splitting of NQCC
)
155.9 kHz suggests a two site jump in the fast limit.
bonded carboxylic acid protons. In addition, variable-temper-
ature NMR (T ) 225 K to T ) 335 K) did not reveal any
differences in the spectra indicating “constant” dynamics. Either
to freeze-in possible motions very low temperatures (e.g., 30
K) would be necessary to clearly distinguish two different proton
positions or the underlying motion of the proton is much faster
than the NMR time scale yielding an averaged peak position.
To unambiguously identify local mobilities, we focused on
static, variable-temperature deuterium NMR measurements.
Larsen et al.39 reported a perfect linear variation of the H
nuclear quadrupolar coupling constant (NQCC) with the asym-
metry of the hydrogen bond rendering NQCC a very useful
parameter for determining the asymmetry of a given hydrogen
Figure 8. View of methyl ester of truxillic acid (dimethyl 2,4-diphenyl-
cyclobutane-1,3-dicarboxylate) along b-axis.
NQCC ) 155.9 kHz, which indicates the presence of certain
dynamics, most probably reflecting a two site jump of a proton
along the hydrogen bond in the fast limit. The sharp center peak
is attributed to mobile deuterons that undergo pseudo-isotropic
reorientations. The extracted NQCC parameter of 155.9 kHz is
quite large and suggests the presence of long and asymmetric
hydrogen bonds. This leads to the assumption that at a given
time each truxillic acid molecule possesses two different
carbonyl groups: one is an original carboxylic acid (COOH),
2
2
bond. Generally the NQCC parameter for H increases with
-
and the other represents a carboxylate anion (COO ). In
increasing asymmetry of the hydrogen bond, and for a long and
asymmetric hydrogen bond it is about three times larger than
compounds that have asymmetric hydrogen bonds the two
carboxy groups are different and give rise to different chemical
2
40
that of H in a short symmetric hydrogen bond. Thus, for
1
3
shifts. The δ ( C) is close to 180 ppm for carboxylate groups
2
asymmetric hydrogen bonds, the NQCC for H should be quite
39
and close to 170 ppm for a carboxylic acid group. The isotropic
large.
1
3
C chemical shift moved to higher frequencies as it passes from
2
The static H spectrum (cf. Figure 6) of an R-truxillic acid
2
3
the carboxylic acid to the carboxylate form, as noticed for
truxillic acid as well, where the COOH signal appears at 178.9
ppm and the COO signal comes at a slightly higher frequency
dimer exhibits a Pake pattern with a quadrupolar splitting of
-
(
39) Kalsbeek, N.; Schaumberg, K.; Larsen, S. J. Mol. Struct. 1993, 299, 155-
70.
40) Mayas, L.; Plato, M.; Winscom, J.; Mobius, K. Mol. Phys. 1978, 36, 753-
64.
1
at 182.3 ppm. Hence, this results in a splitting of the cyclobutane
ring signals. The intensities of the carbonyl signals are not equal,
(
7
1746 J. AM. CHEM. SOC.
9
VOL. 130, NO. 5, 2008

[
2
+
2] Photodimerization of Cinnamic Acid
A R T I C L E S
Figure 9. Conformations of a carboxyl moiety in a carboxylic acid: (a) synplanar conformation adopted by dimer; (b) antiplanar conformation adopted by
monomer; (c) schematic representation of carboxyl moiety in the highest temperature (350 K) crystal structure of truxillic acid. Here, the proton is located
in a double-well potential so that two half occupied positions for H are observed.
Å) yielding a ∆(CsO) less than 0.1 Å (0.081 Å). When we
remeasured the crystal structure of the as-dimerized truxillic
acid at different temperatures (120 K to 350 K), we observed a
small gradual decrease of ∆(CsO) values at higher temperatures
(cf. Figure 7). This trend is more prominent in the case of
R-trans-cinnamic acid monomer where ∆(CsO) is far less than
0
.1 Å (∼0.038 Å) pointing toward the presence of an asym-
metric and a dynamic hydrogen bond in the carboxyl moiety
of the monomer. We have also determined the crystal structure
of truxillic acid methyl ester (the projection is given in Figure
8
). The ∆(CsO) value amounts to 0.135 Å and clearly
emphasizes an absence of carboxylates. This is quite expected
since the ester cannot be involved in dynamic hydrogen bonding.
1
3
Hence, the C-CPMAS NMR spectrum truly reflects cen-
trosymmetry showing no peak splitting.
Figure 10. Plot of computed 13C chemical shifts for carboxylate anion of
truxillic acid vs observed chemical shifts.
Leiserowitz has already discussed molecular packing modes
of carboxylic acids. He stated that in the crystal structures of
3
1
R, â-saturated carboxylic acids RsCH2sCH2sCOOH, the
and this is more prominent at lower conversion; the carboxylate
signal is less intense than its neighboring carboxylic acid signal,
and it gradually increases as the conversion increases and
becomes more or less equal to the carboxylic carbon signal after
complete conversion. Dynamic hydrogen bonding also exists
in the trans-cinnamic acid monomer as suggested by the high
NQCC value of about 149.1 kHz, but this does not affect the
â
R
synplanar C sC sCdO arrangement as shown in Figure 9a
is adopted without any known exceptions, whereas the synplanar
rule does not apply to the R, â-unsaturated acids RsCHdCHs
â
COOH, with several compounds showing an antiplanar C s
R
C sCdO form. Concerning the photodimerization of R-trans-
cinnamic acid, the monomer (an R, â-unsaturated acid) adopts
1
3
an antiplanar conformation while the resulting photodimer
C-CPMAS spectra (cf. Figure 2a); e.g., the carbonyl signals
1
1,12
adopts a synplanar conformation.
Thus, upon increasing
do not split. Most likely, this is due to the fact that R-truxillic
acid possesses two carbonyl groups on the same molecule while
trans-cinnamic acid has one carbonyl atom. Since NMR
measures an averaged chemical shift no splitting is observed in
the case of the monomer.
The apparent discrepancy between NMR and X-ray analysis
in the case of truxillic acid may result from two reasons: the
carboxylic acid protons experience motion (dynamics) on the
NMR time scale but appear rather rigid when observed by X-ray
at very low temperature (30 K). Indeed, the presence of a
carboxylate group in the compound is indicated by nearly equal
conversion ratios the molecule slowly changes from an antiplanar
to a synplanar conformation, possibly facilitated by proton
exchange since reactions in the solid state commonly occur with
a minimum amount of atomic or molecular movement. This
strongly supports the presence of dynamic disorder in the
carboxyl moiety of truxillic acid where the proton undergoes
rapid oscillations across the OsH‚‚‚‚O bond accompanied by
interconversion between the carbonyl CdO and hydroxyl Cs
O(H) bond. The temperature dependence of ∆(CsO) in truxillic
acid (cf. Figure 7) supports this assumption. Indeed, at the
highest temperature (350 K) the search for the H-atoms by
difference Fourier analysis revealed two carboxyl H atoms each
occupied by 50%.
C-O distances, with typical C-O values of 1.25-1.26 Å
corresponding to a very small ∆(C-O) values.2
3,31,39
The
carboxylic acid group, COOH, with a CsO single bond (≈1.31
Å) and a CdO double bond (≈1.21 Å) has a ∆(CsO) value of
around 0.1 Å. In the crystal structure of truxillic acid we
observed that ∆(CsO) values are less than 0.1 Å, and a single
CsO value is 1.312 Å, which is typical but the double bond is
slightly larger than that in the carboxylic acid group (about 1.231
Finally, the 13C-CPMAS data were compared with ab initio
C chemical shift computations of both R-truxillic acid and its
1
3
anion (created by arbitrarily removing one proton from the
carboxyl group). Notably, the truxillic acid structure did not
reveal any splitting, but the anionic structure yields a similar
J. AM. CHEM. SOC.
9
VOL. 130, NO. 5, 2008 1747

A R T I C L E S
Khan et al.
signal splitting for the carbonyl and cyclobutane ring atoms as
observed experimentally. Figure 10 shows the plot of computed
C isotropic chemical shifts vs the observed values; the
agreement is good in the range of (5 ppm.
a majority of carboxylic acid based dimeric compounds. Only
the joint approach of both X-ray analysis and solid-state NMR
successfully revealed the dynamic processes occurring in these
topochemical reactions and, thus, provided detailed insight into
the reaction mechanism of organic solid-state transformations.
1
3
4. Conclusions
Supporting Information Available: Powder diffraction data
We have demonstrated that dynamic hydrogen bonding results
1
3
of both P21/n and C2/c phase, the C CPMAS spectrum of
C2/c phase, a table of the CsO and CdO distance variations
vs temperature along with ∆(CsO) values, full ref 30, and
crystal structures provided as cif files. This material is available
free of charge via the Internet at http://pubs.acs.org.
in the appearance of additional signals in the 13C CPMAS
spectra of photodimerized R-trans-cinnamic acid. The transient
asymmetry of the hydrogen bonds in the crystal temporarily
destroys the photodimer’s center of symmetry so that the truxillic
acid molecules may exist in a noncentrosymmetric state in their
crystal lattice. Indeed, this is a common phenomenon found for
JA0773711
1748 J. AM. CHEM. SOC.
9
VOL. 130, NO. 5, 2008