23Na magic-angle spinning and double-rotation NMR study of solid forms of sodium valproate

Article abstract of DOI:10.1139/cjc-2013-0442

Sodium valproate is a pharmaceutical with applications in the treatment of epilepsy, bipolar disorder, and other ailments. Sodium valproate can exist in many hydrated and acid-stabilized forms in the solid state, and it can be difficult to obtain precise structural information about many of these. Here, we present a ¹³C and ²³Na solid-state NMR study of several forms of sodium valproate, only one of which has been previously structurally characterized by single-crystal X-ray diffraction. ²³Na magic-angle spinning (MAS), double-rotation (DOR), and multiple-quantum magic-angle spinning (MQMAS) NMR spectra are shown to provide useful information on the number of molecules in the asymmetric unit, the local coordination geometry of the sodium cations, and the presence of amorphous phases. Two previously identified forms are shown to be highly similar, or identical, according to the ²³Na NMR data. The utility of carrying out both DOR and MQMAS NMR experiments to identify all crystallographically unique sites is demonstrated. ¹³C cross-polarization MAS NMR spectra also provide complementary information on the number of molecules in the asymmetric unit and the crystallinity of the sample.

Full text of DOI:10.1139/cjc-2013-0442

9
ARTICLE
23Na magic-angle spinning and double-rotation NMR study of
solid forms of sodium valproate
Nuiok M. Dicaire, Frédéric A. Perras, and David L. Bryce
Abstract: Sodium valproate is a pharmaceutical with applications in the treatment of epilepsy, bipolar disorder, and other
ailments. Sodium valproate can exist in many hydrated and acid-stabilized forms in the solid state, and it can be difficult to
obtain precise structural information about many of these. Here, we present a 13C and 23Na solid-state NMR study of several forms
of sodium valproate, only one of which has been previously structurally characterized by single-crystal X-ray diffraction. 23Na
magic-angle spinning (MAS), double-rotation (DOR), and multiple-quantum magic-angle spinning (MQMAS) NMR spectra are
shown to provide useful information on the number of molecules in the asymmetric unit, the local coordination geometry of the
sodium cations, and the presence of amorphous phases. Two previously identified forms are shown to be highly similar, or
identical, according to the 23Na NMR data. The utility of carrying out both DOR and MQMAS NMR experiments to identify all
crystallographically unique sites is demonstrated. 13C cross-polarization MAS NMR spectra also provide complementary infor-
mation on the number of molecules in the asymmetric unit and the crystallinity of the sample.
Key words: nuclear magnetic resonance, solid-state NMR, sodium-23, polymorphism, hydrates, pharmaceuticals, DFT, X-ray
diffraction, carbon-13.
Résumé : Le valproate de sodium est un composé pharmaceutique ayant des applications dans le traitement de l’épilepsie, les
troubles bipolaires ainsi que d’autres maladies. Le valproate de sodium peut aussi exister dans de nombreuses phases solides
hydratées ainsi que diverses formes stabilisées par acide conjugué pour lesquelles il est difficile d’obtenir de l’information
structurelle. Dans cet article nous présentons une étude de RMN a` l’état solide du 13C et 23Na de plusieurs formes du valproate
de sodium pour lesquelles seulement une structure cristalline est connue. Nous montrons que les spectres de RMN de rotation
a` l’angle magique (MAS), double-rotation (DOR) et de quantum multiples en rotation a` l’angle magique (MQMAS) du 23Na peuvent
être utilisés pour déterminer le nombre de molécules dans l’unité asymétrique et les polyèdres de coordination du sodium, ainsi
que pour identifier la présence de formes amorphes. Nos données de RMN du 23Na indiquent que deux des formes du valproate
de sodium ayant précédemment été identifiées sont en fait identiques ou très semblables. L’utilité d’accomplir des expériences
DOR ainsi que MQMAS pour identifier les sites cristallographiquement différents est démontrée. Les spectres MAS avec polari-
sation croisée du 13C fournissent aussi de l’information complémentaire concernant le nombre de molécules dans l’unité
asymétrique ainsi que la cristallinité de l’échantillon.
Mots-clés : résonance magnétique nucléaire, RMN a` l’état solide, sodium-23, polymorphisme, hydrates, produits pharmaceu-
tiques, DFT, diffractrométrie de rayons X, carbone-13.
for the interconversion of these forms. A single-crystal X-ray
Introduction
structure was successfully obtained for trisodium hydrogentetra-
The sodium salt of 2-propylvaleric (valproic) acid (see Fig. 1) is
valproate monohydrate (known as form C);11 however, detailed
used as an anticonvulsant in the treatment of epilepsy,1,2 as a
structural information is lacking for the other forms. Petruševski
mood stabilizer in the treatment of bipolar disorder,3,4,5,6 and for
treating various other ailments including migraines and even
et al. comment in their conclusions that poor crystallinity contrib-
uted to the difficulties in obtaining high-quality diffraction data
cancer.7 Petruševski et al. have presented a detailed study of var-
even with a high-intensity synchrotron X-ray source.8
ious solid-state forms of sodium valproate, including various
Solid-state nuclear magnetic resonance (SSNMR) can be used to
acid-stabilized forms and hydrates thereof.8 Much like true poly-
provide additional atomic-level information concerning the mo-
morphs, these hydrates (also known as pseudopolymorphs9 or
lecular structure of a substance and can serve as a powerful tool in
solvatomorphs10) can exhibit different physiological properties,
the identification of pharmaceuticals, including their polymorphs
and it is therefore of interest to fully characterize them. Previous
studies of the different solid forms of sodium valproate have used
methods such as differential scanning calorimetry, thermogravi-
metric analysis, infrared spectroscopy, and powder X-ray diffrac-
tion (PXRD) to characterize up to eight hydrates and polymorphs
of sodium valproate.8 This study carefully examined the condi-
tions required to prepare various forms, labelled A through H (see
Table 1 for labelling of the forms studied herein), and the routes
and hydrates12,13 in crystalline or amorphous states. SSNMR spec-
troscopy is routinely performed on spin-1/2 nuclei such as 13C.
However, quadrupolar nuclei represent approximately three quar-
ters of the periodic table and many quadrupolar nuclides such as
14N, 17O, 23Na, and 35/37Cl are often found in drugs. While 13C SSNMR
is a powerful tool for the identification of polymorphs, researchers
have also studied the possibility of using SSNMR of quadrupolar
nuclei to characterize polymorphism and hydration state.14,15,16,17,18
Received 25 September 2013. Accepted 11 October 2013.
N.M. Dicaire, F.A. Perras, and D.L. Bryce. Department of Chemistry and Centre for Catalysis Research and Innovation, University of Ottawa, 10 Marie
Curie Private, Ottawa, ON K1N 6N5, Canada.
Corresponding author: David L. Bryce (e-mail: dbryce@uottawa.ca).
Can. J. Chem. 92: 9–15 (2014) dx.doi.org/10.1139/cjc-2013-0442
Published at www.nrcresearchpress.com/cjc on 15 October 2013.

10
Can. J. Chem. Vol. 92, 2014
solid-state forms of sodium valproate using 23Na MAS, DOR, and
MQMAS NMR spectroscopy as well as 13C CP/MAS NMR and X-ray
diffraction.
Fig. 1. (a) Structural diagram of sodium valproate. (b) Local
environments of the three crystallographically distinct sodium
cations in the sodium valproate − valproic acid monohydrate
Na3(C8H15O2)3(C8H16O2)·H2O (compound C). OV indicates an oxygen
atom from a valproate − valproic acid moiety and OW indicates an
oxygen atom from water.
Experimental
Sample preparation
Sodium valproate (98%) was purchased from Aldrich and used
without further purification. Valproic acid was prepared by dissolv-
ing 4.0 g of sodium valproate in 32 mL of 1.9 mol/L hydrochloric acid.
The mixture was subsequently stirred and heated to 50 °C for
20 min. Valproic acid, which is liquid at room temperature, could
then be easily separated from the aqueous solution by extraction.
Compound C, reported to be Na3(C8H15O2)3(C8H16O2)·H2O,8 was
obtained by dispersing 2.49 g of sodium valproate in 15 mL of hot
acetone. The mixture was treated with 1.17 mL of valproic acid,
which caused the sodium valproate to fully dissolve when stirred
at approximately 50 °C for 5 min. The clear colourless solution
was cooled in an ice bath for 30 to 45 min and a voluminous white
precipitate formed.
To obtain compound F, reported to be Na3(C8H15O2)3(C8H16O2)·
2H2O,8 0.63 g of sodium valproate was dissolved in 6 mL of acetone
and the mixture was treated with 0.2 mL of valproic acid. The solu-
tion was stirred and heated at 50 °C for approximately 10 min, until
the solids dissolved, and left at room temperature for 20 min before
being cooled in an ice bath for 2 h. Small needle-shaped crystals
formed. Samples were covered with perforated Parafilm and then
stored at 4 °C.
Table 1. Various compounds studied in the present work.
Composition as reported by Petruševski
Label
et al.8,11
Commercial sodium
—
valproate
C
Na3(C8H15O2)3(C8H16O2)·H2O, a valproate −
valproic acid monohydrate
Compound E, reported to be Na(C8H15O2)(C8H16O2),8 was pre-
pared by adding 1.16 g (6.98 mmol) of sodium valproate and
0.73 mL (5 mmol) of valproic acid to 1.17 mL of hot acetone at 50 °C.
The solution was stirred at 50 °C for 10 min and then left at room
temperature for 20 min and subsequently cooled in an ice bath for
2 h. After being dried, the sample was placed under vacuum for a
few hours to dehydrate the compound. The resulting crystals were
not of sufficient size and quality for single-crystal XRD.
An additional form (compound D) has previously been reported
to form when the commercial compound is pressed into pellets
with KBr with a mechanical press in a 1:10 ratio. Due to the nature
of the obtained samples, very little information could previously
be obtained relating to this form, other than the fact that it pre-
sented largely different IR spectra. For our NMR experiments, the
pellets were then ground before being packed into an NMR rotor.
This form has been previously denoted as Na(C8H15O2)·yH2O;8
however, little is known about its structure or composition.
The anhydrous form of sodium valproate (compound H) was
obtained by heating the commercial compound in an oven grad-
ually to 400 °C followed by cooling to room temperature under
anhydrous conditions. This form is highly hygroscopic and needed
to be packed immediately after heating to perform the NMR ex-
periments.
D
E
Na(C8H15O2)·yH2O (y < 1)
Na(C8H15O2)(C8H16O2), an anhydrous
valproate − valproic acid compound
Na3(C8H15O2)3(C8H16O2)·2H2O, a valproate −
valproic acid dihydrate
F
H
Na(C8H15O2), anhydrous sodium valproate
In a powder sample, anisotropic spectral broadening of the ob-
served NMR central transition (m = 1/2 to −1/2) of half-integer spin
quadrupolar nuclei arises from the coupling between the electric
quadrupole moment of these nuclei and the electric field gradient
(EFG) at the nucleus.19 The second-order quadrupolar broadening
that affects the central transition of half-integer spin quadrupolar
nuclei such as 23Na cannot be completely averaged by magic-angle
spinning (MAS). However, valuable information concerning the
EFG tensor can be obtained by simulating the pattern obtained
with MAS NMR experiments that can be related to the crystalline
structure of the drug. To obtain high-resolution SSNMR spectra
of quadrupolar nuclei, either sophisticated sample rotation, i.e.,
double-rotation (DOR)20 NMR or sophisticated two-dimensional
approaches such as multiple-quantum magic-angle spinning
(MQMAS)21 are typically necessary. While MQMAS is used more
often and does not require specialized equipment, DOR NMR,
which relies on spinning the sample at two different angles simul-
taneously to more completely average the quadrupolar interac-
tion, can often provide similar information in a shorter amount of
time. In favourable cases, DOR can also be used to obtain a greater
amount of information than is possible with MAS-based experi-
ments through high-resolution two-dimensional experiments such
as homonuclear correlation experiments22,23,24 and quadrupolar −
chemical shift correlation experiments.25
SSNMR spectroscopy
23Na SSNMR experiments were performed in an external mag-
netic field of 9.4 T (␯L(23Na) = 105.85 MHz) on a Bruker AVANCE III
spectrometer. Chemical shifts were referenced to 1 M NaCl(aq)
using solid NaCl as an external reference (␦(NaCl(s) = 7.21 ppm).
Samples were powdered and packed in 4 mm o.d. ZrO2 MAS rotors
or 4.3 mm o.d. Vespel DOR rotors. MAS NMR experiments were
typically performed at spinning rates of 10 kHz. A 2 s recycle delay
was used and sufficiently many scans to obtain good quality spec-
tra were performed. 23Na MAS experiments were carried out using
either a Hahn echo or a simple Bloch decay pulse sequence; all
We have recently demonstrated the utility of 23Na MAS and DOR
NMR spectroscopy as well as X-ray crystallography and gauge-
including projector-augmented-wave density functional theory
(GIPAW DFT) calculations to characterize anhydrous, monohydrate,
and methanol solvates of sodium naproxen, a nonsteroidal anti-
inflammatory pharmaceutical compound.26 23Na DOR NMR spec-
troscopy has also been used to identify a new hydrated form of the
nucleotide dCMP.27 In the present work, we characterize several
1
experiments used two-pulse phase modulation (TPPM) H decou-
pling.28 The MAS NMR spectra were fit using the WSolids software
package.29 The simulations were aided by the results obtained
from the DOR and MQMAS NMR experiments, vide infra.
Published by NRC Research Press

Dicaire et al.
11
23Na MQMAS NMR experiments were performed at 10 kHz MAS
using the three pulse experiment with soft-pulse-added mixing
signal enhancement.30 The excitation and conversion pulses lasted
4.0 and 1.5 ␮s, respectively, and the read pulse lasted 30 ␮s. A total
of 12 to 24 anti-echoes were acquired, with twice the number of
echoes; the t1 increments were 100 ␮s for rotor synchronization.
TPPM 1H decoupling was used in both dimensions for all MQMAS
experiments.
Fig. 2. Carbon-13 CP/MAS NMR spectra of the various sodium
valproate compounds acquired in a magnetic field of 9.4 T:
(a) as-received commercial sample of sodium valproate,
(b) compound C/F, (c) anhydrous compound H, and (d) compound E.
DOR experiments were performed using the same spectrometer
with a Bruker HP WB 73A DOR probe. Samples were packed into
4.3 mm rotors, which were inserted into a 14 mm outer rotor. The
outer rotor was spun at frequencies ranging from 700 to 1000 Hz.
Odd-numbered sidebands were removed by synchronizing the ex-
citation pulse with the outer rotor phase.31 Experiments were
carried out with SPINAL-64 1H decoupling32 and a recycle delay of
2 s. A double-frequency sweep pulse33 lasting 5 ms and sweeping
from 900 to 175 kHz was used to enhance the signal. DOR NMR
spectra were simulated with an extended Floquet theory pro-
gram34 written with the GAMMA spin dynamics libraries.35 The
simulations were useful to constrain the fits of the MAS NMR
spectra.
(d)
(c)
(b)
(a)
13C cross-polarization (CP) MAS SSNMR experiments were per-
formed with the same spectrometer (␯L(13C) = 100.62 MHz). The
chemical shifts were referenced to tetramethylsilane by using
solid glycine as an external reference (␦(13CO) = 176.4 ppm). Sam-
ples were spun at 5 or 10 kHz, and up to 2048 scans were obtained
with a recycle delay of 2 s. The ramped-amplitude CP experiment36
was used with a contact time of 2 ms and TPPM 1H decoupling.
200
175
75
50
25
0
δ(13C) (ppm)
Powder X-ray diffraction
PXRD measurements were performed using a Rigaku Ultima IV
diffractometer with Cu K␣ radiation (wavelength of 1.54184 Å) and
the Bragg−Brentano geometry. Intensity measurements were per-
formed for a 2␪ diffraction angle of 5° to 40° in steps of 0.02° at a
rate of 0.5°/min.
in the molecule, an indication that the asymmetric unit likely
contains two crystallographically distinct molecules. This is more
clearly seen by observing the carbonyl region of the spectrum,
which shows the greatest resolution. One-dimensional 23Na MAS
NMR alone was unsuccessful in resolving the sodium sites as can
be seen in Fig. 3a; however, the observed overlapping second-
order quadrupolar line shapes give a strong indication of more
than one site. Using MQMAS NMR (see Fig. 3c), it was possible to
clearly resolve two different sodium sites. This is consistent with
the number of peaks in the 13C NMR spectrum and further indi-
cates that there are likely two molecules in the asymmetric unit.
(Note, however, that spectral fitting required an intensity ratio of
1:1.4 for the two sites, suggesting alternatively that the commer-
cial sample could be a physical mixture of two forms.) Although
the two sites are not resolved with DOR NMR at 9.4 T (Fig. 3b), the
observed DOR shift (␦DOR = ␦iso − ((CQ2(1 + ␩2/3)/40)(106Hz/␯L)) was
used as a restraint when fitting the MAS NMR spectrum to deter-
mine the 23Na NMR parameters (Table 2). The DOR shifts, like the
MQMAS ones, are well known to be magnetic field-dependent and
greater resolution would be obtained at a different magnetic field
strength.24 The combined use of MQMAS and DOR improves the
chances of resolving all of the sodium sites. Interestingly, the
values of CQ for the two sites are identical within experimental
error (2.7 MHz) but the asymmetry parameters take on very differ-
ent values (0.17 and 1.00). By comparison with previous reports38
of 23Na NMR on organic systems, these sodium cations are likely
to be in a five-coordinate environment, coordinated to both car-
bonyl moieties and water molecules. Six-coordinate sodium cat-
ions typically have CQ values below 2 MHz.
GIPAW DFT calculations
For form C, the single-crystal X-ray structure of which is known,
GIPAW DFT calculations were performed using the CASTEP pro-
gram (version 4.1).37 The proton positions were first optimized
using a kinetic energy cutoff of 450 eV and the EFG and magnetic
shielding tensors were subsequently calculated using a kinetic
energy cutoff of 550 eV. The default 2 × 2 × 1 k-point grid was used
for both the geometry optimization and the NMR calculation.
Results and discussion
In addition to following the reported sample preparation pro-
tocols, correspondences of the compounds prepared herein to the
previously identified forms8 were determined as follows. For com-
pound C, a single-crystal X-ray diffraction (SCXRD) structure was
acquired and shown to be identical to that previously reported.
For compounds E and F, powder X-ray diffractograms were re-
corded (see Supplementary Material section), which matched
those reported previously. IR spectroscopy provided additional
supporting evidence. Anhydrous sodium valproate, form H, was
identified with IR spectroscopy, from which a band for water is
conspicuously absent (see Supplementary Material).
Further atomic-level characterization of the samples was pur-
sued by 23Na and 13C MAS NMR spectroscopy. 13C CP/MAS NMR
spectra are presented in Fig. 2 and 13C chemical shifts are tabu-
lated in the Supplementary Material. DOR and MQMAS 23Na NMR
spectra were also acquired to aid with spectral simulation. Each
compound is discussed separately in the following sections.
Compounds C and F
Shown in Fig. 4 are the 23Na MAS NMR spectra recorded for
compound F. Since the MAS spectrum alone was not sufficient to
distinguish between all crystallographic sites, MQMAS and DOR
NMR were applied to try to identify all nonequivalent sodium
sites. Interestingly, the 23Na NMR spectra of compound C are
identical to those of compound F (Fig. 4, inset). As noted in the
experimental section, SCXRD and PXRD experiments were used to
Commercial form of sodium valproate
23Na and 13C SSNMR experiments were carried out directly on
the commercial sample. As shown in Fig. 2, 13C NMR spectroscopy
revealed two resonances for each chemically distinct carbon atom
Published by NRC Research Press

12
Can. J. Chem. Vol. 92, 2014
Fig. 3. Sodium-23 (a) MAS, (b) DOR, and (c) MQMAS NMR spectra of
commercial sodium valproate acquired at 9.4 T. Two nonequivalent
sites are resolved in the MQMAS NMR spectrum.
Fig. 4. Sodium-23 (a) MAS, (b) DOR, and (c) MQMAS NMR spectra of
compound F acquired at 9.4 T. Three nonequivalent sites are
resolved in the DOR NMR spectrum. Shown in the inset are
experimental 23Na MAS NMR spectra of compound F (green, bottom)
and compound C (blue, top), providing evidence that these two
forms are identical (colour in the online version only).
(c) MQMAS
site 1
(c) MQMAS
site 2
(b) DOR
sim.
exp.
(b) DOR
sim.
exp.
(a) MAS
sim.
exp.
(a) MAS
50
25
0
-25
-50
-75
-100
sim.
exp.
δ(23Na) (ppm)
Table 2. Sodium-23 NMR parameters for compounds studied in this
work.
50
25
0
-25
-50
-75
δ(23Na) (ppm)
Compound
C
Q (MHz)
␩
␦
iso (ppm)
0.7 (0.2)
Commercial sodium
valproatea
C/F
Site 1
Site 2
Site 1
Site 2
Site 3
Site 1
Site 2
Site 1
2.70 (0.03)
2.71 (0.03)
2.76 (0.06)
2.52 (0.03)
1.55 (0.05)
1.50 (0.04)
1.50 (0.05)
3.55 (0.03)
1.00 (0.01)
0.17 (0.04)
0.25 (0.07)
0.47 (0.08)
0.33 (0.10)
0.15 (0.10)
0.68 (0.03)
1.00 (0.02)
Fig. 5. Powder X-ray diffraction patterns for compound F
(experiment, blue) and compound C (simulated, black), providing
evidence that these two forms are very similar or identical.
−3.6 (0.3)
−2.1 (0.8)
−0.2 (0.6)
5.5 (0.2)
−1.4 (0.2)
−1.5 (0.5)
−4.0 (0.4)
Eb
H
aThe best fit is obtained with relative intensities of ϳ1.4:1 for sites 1 and 2,
respectively. The 13C NMR spectrum (Fig. 2a) is also consistent with two sets of
resonances in approximately this ratio.
bThe 23Na MAS spectrum of compound E also showed a broad peak attributed
to an amorphous hydrate whose line width remained the same with DOR.
5
10
15
20
25
30
35
40
confirm the identities of compounds C and F by reproducing the
literature SCXRD structure of the former and the PXRD pattern of
the latter. A comparison of PXRD patterns is shown in Fig. 5 and
suggests a strong similarity between the two forms. The NMR
results also show that the two forms are either extremely similar
or rapidly interconvert. The latter explanation seems less plausi-
ble given that the X-ray data are consistent with the identification
of the two different forms. The SCXRD structure of compound C
shows three nonequivalent sodium sites and these are readily
identified in the 23Na DOR NMR spectrum in Fig. 4b. The 23Na
MQMAS NMR spectrum at 9.4 T (Fig. 4c) does not provide sufficient
2θ
resolution to distinguish between the three sites although their
individual powder patterns can be partially identified in the two-
dimensional contour plots.
The crystallographic environments of the three sites are de-
picted in Fig. 1b. The site with the smallest CQ (1.55 MHz) has an
approximately tetrahedral coordination environment, while the
other two have distorted square pyramidal coordination (CQ val-
ues of 2.52 and 2.76 MHz). The higher CQ values for the latter two
Published by NRC Research Press

Dicaire et al.
13
Fig. 6. Comparison between experimental EFG tensor components
for compound F and PAW DFT calculated values for compound C.
The good correlation suggests that the two forms are identical.
sites are consistent with the lower site symmetry and previous
literature reports38,39 and are also corroborated by GIPAW DFT
calculations. A range of asymmetry parameters, from 0.25 to 0.47,
is observed. A good linear correlation is apparent when the pa-
rameters obtained for form C by GIPAW DFT calculations are
plotted against the experimental parameters of form F, as pre-
sented in Fig. 6. The systematic overestimation of the EFG tensor
parameters is consistent with previous correlations of 23Na qua-
drupolar coupling tensors for compounds with known structures.40
The somewhat larger overestimation noted here is likely caused by
rapid motions of the alkyl chains of the valproate molecule, which
would reduce the apparent 23Na CQ values.41
0.20
0.15
0.10
0.05
0.00
-0.05
Koller et al.39 have presented a detailed interpretation of 23Na
quadrupolar interaction parameters and chemical shifts in a
range of solids where oxygen atoms are exclusively the nearest
neighbours. Although their work highlighted mainly inorganic
compounds, a useful qualitative correlation between the value of
CQ(23Na) and the local geometry and coordination number can be
applied to other systems, such as those studied presently, featur-
ing oxygen coordinated to sodium cations. The above assignments
of the three values of CQ measured for compound F are consistent
with the correlation described by Koller et al.,39 where values of
ϳ1 MHz are expected for slightly distorted tetrahedral environ-
ments and values around 3 MHz are expected for square planar
environments.
calc
expt
Vii
= 1.3113Vii
-0.10
-0.15
R² = 0.9586
-0.10
-0.05
0.00
0.05
0.10
0.15
V expt. (compound F) (a.u.)
ii
Fig. 7. Sodium-23 (a) MAS, (b) DOR, and (c) MQMAS NMR spectra of
compound E acquired at 9.4 T. An unassigned peak is denoted with
an asterisk. A broad peak due to hydration of the sample is centred
at ϳ1 ppm in the MAS and DOR NMR spectra.
Compound E
This compound was synthesized using a higher ratio of valproic
acid to sodium valproate, as described in the experimental sec-
tion. Its PXRD pattern corresponds to that of the previously re-
ported Na(C8H15O2)(C8H16O2) (see Supplementary Material). The
23Na NMR spectra are shown in Fig. 7. At 9.4 T, the 23Na DOR NMR
spectrum did not conclusively reveal multiple sites, as illustrated
in Fig. 7b. However, the observed DOR shift could be determined
to help constrain the fit of the MAS NMR spectrum. An MQMAS
NMR experiment was carried out and suggested two overlapping
sodium sites. Simulation of the MAS NMR spectrum required a
two-site model, consistent with the MQMAS result. Both sites have
relatively small CQ values of 1.50 MHz and identical chemical
shifts within experimental error. They differ by their value of ␩
(0.15 and 0.68, respectively). Only coordination environments
with near tetrahedral and octahedral symmetry can lead to such
small 23Na CQ values. Compound C has a tetrahedrally coordi-
nated sodium cation that features a CQ value of the same magni-
tude; however, the chemical shift is very different for form E. The
chemical shift is in agreement with other 23Na NMR studies that
found that chemical shifts near zero are obtained for octahedrally
coordinated sodium sites with six oxygen ligands;38 lower chem-
ical shifts are typically expected as the coordination number in-
creases. On the basis of a comparison of these quadrupolar data
with those for compound C (the crystal structure of which is
known) and from previous reports,38,39 compound E likely fea-
tures two sodium sites in pseudo-octahedral coordination envi-
ronments (data in Table 2).
(c) MQMAS
(b) DOR
sim.
exp.
*
The compound appeared somewhat stable at ambient condi-
tions but quickly transformed into a wet paste when packed into
a rotor. Due to the hygroscopic nature of the sample, the 23Na
MAS and DOR NMR spectra show a broad peak centred near 1 ppm
attributed to the degradation of the compound by water absorp-
tion, as illustrated in Fig. 7a. This peak is also present in the DOR
spectrum (Fig. 7b); however, since its broadening originates from
a distribution of chemical shifts, its line width remains unchanged
(1 kHz) and is much broader than the DOR peak for the crystalline
sodium sites (120 Hz).
(a) MAS
sim.
exp.
30
20
10
0
-10
-20
-30
δ(23Na) (ppm)
KBr-pressed compound (compound D)
Due to its pronounced hygroscopic properties, 13C CP/MAS NMR
of compound E resulted in broad, unresolved peaks, making it
difficult to distinguish between the multiple resonance peaks (see
Fig. 2d).
It has been hypothesized that a new solid form with the formula
Na(C8H15O2)·yH2O (y < 1) is obtained when sodium valproate is
pressed into KBr pellets for IR analysis. We have investigated the
nature of this sample via 23Na NMR. Due to their inherently low
Published by NRC Research Press

14
Can. J. Chem. Vol. 92, 2014
sensitivity, 13C NMR measurements were not productive for this
dilute sample. Its 23Na NMR spectrum is presented in Fig. 8 and
shows a sharp peak at −7.1 ppm along with a broad peak centred at
1.1 ppm. Since no discernible second-order quadrupolar line
shapes can be identified, it can be seen that the sample is no
longer crystalline and becomes amorphous when it is pressed
with KBr. The very sharp resonance at −7.1 ppm would, however,
point to a small quantity of a crystalline substance with cubic
symmetry. This chemical shift, however, does not correspond to
the chemical shift of any known sodium salt. One possibility that
may lead to the loss of crystallinity that we observe may be the
inclusion of hydration water, as in compound E. This is unlikely,
since no such peak was observed in the pure sodium valproate
(Fig. 3). A second possibility may be that the loss of crystallinity
arises from a cation-exchange reaction between KBr and sodium
valproate. This reaction would lead to an amorphous product with
the formula KxNa1−x(C8H15O2)·yH2O, (x < 1) and a small quantity of
NaxK1−xBr, in agreement with the broad resonance at 1.1 ppm and
the sharp peak at −7.1 ppm, respectively. This hypothesis was
tested by compressing a mixture of NaBr and KBr (in a 1:9 ratio)
into a pellet with a mechanical press. The 23Na NMR spectrum of
this sample exhibited two sharp resonances, one of which is at-
tributed to NaBr, and the other appearing at −7.1 ppm which may
be attributed to a mixed-alkali bromide of the form NaxK1−xBr. We
then speculate that the exchange of Na+ with K+ produces a mixed
sodium−potassium valproate to which the broad peak at 1.1 ppm
is assigned. This compound is amorphous as shown by the broad
featureless peak observed in the 23Na NMR spectrum. It is inter-
esting to note that KBr, which is routinely used to form KBr pellets
for IR analysis, is not entirely inert and can in fact exchange ions
with the compound of interest. Care should then be taken when
performing IR spectroscopy of alkali metal salts so as to avoid the
exchange of ions.
Fig. 8. Sodium-23 MAS NMR spectrum of compound D. The
dominant broad peak centred near 1 ppm suggests that the sample
is amorphous. The sharp peak at −7.1 ppm is attributed to Na+ ions
in a cubic environment resulting from grinding with KBr (see text).
75
50
25
0
-25
-50
-75
δ(23Na) (ppm)
Fig. 9. Sodium-23 (a) MAS, (b) DOR, and (c) MQMAS NMR spectra of
compound H (anhydrous sodium valproate).
(c) MQMAS
(b) DOR
Anhydrous compound (compound H)
Heating the commercial form of sodium valproate resulted in
an anhydrous form that was characterized by 13C CP/MAS, 23Na
MAS, DOR, and MQMAS NMR. IR spectroscopy performed on the
sample before packing into a rotor for NMR experiments revealed
that this compound is indeed anhydrous. Given the method used
to obtain this compound, we were unable to produce single crys-
tals suitable for SCXRD; the NMR spectroscopy is, however, quite
informative. The 23Na MAS NMR spectrum shows two broad reso-
nances as presented in Fig. 9a. One peak is broad and featureless
and originates from the partial rehydration of the sample, as in
compound E (Fig. 7). However, the other peak has a second-order
quadrupolar line shape and is identified as a crystallographic
sodium site characterized by a large CQ of 3.55 MHz and an
asymmetry parameter of 1. If the sample is left under ambient
conditions for brief periods of time, the featureless peak increases
in intensity and eventually leads to the complete depletion of the
crystalline resonance. Both the two-dimensional MQMAS spec-
trum and the DOR spectrum clearly resolve the two sites in this
case. Interestingly, the featureless peak, unlike that due to the
crystalline site, does not get sharper under DOR conditions, indi-
cating that the line width originates from a distribution of chem-
ical shifts in the partially rehydrated form. Comparison of the
quadrupolar parameters for compound H with those of the other
compounds suggests that the single sodium site in the former is in
a highly asymmetric environment.
sim.
exp.
(a) MAS
100
sim.
exp.
-150
50
0
-50
-100
23
δ( Na) (ppm)
diphenylacetate.42 In its crystal structure, there is a single sodium
site in a distorted seesaw coordination environment. A PAW DFT
calculation of the 23Na EFG tensor parameters was performed
for this compound that predicted that the CQ value would be
4.17 MHz and the ␩ value would be 0.52. The highly distorted
four-coordinate environment that originates from the anion’s
large size gives rise to a large quadrupolar coupling and high
asymmetry parameter, as was also observed in anhydrous sodium
valproate. It is then likely that, like sodium naproxen and sodium
diphenylacetate, the sodium cation in anhydrous sodium val-
proate is located in a highly distorted four-coordinate environ-
ment. The rehydration of the sample enables the possibility of
increasing the coordination number and forming octahedral
sodium sites, in agreement with the amorphous peak that has
very small quadrupolar coupling, as evidenced by its lack of DOR
sidebands.
The large size of the anions, when compared with the cations,
indicates that only low cation coordination numbers are possible
in the anhydrous form. A similar situation is also found in anhy-
drous sodium naproxen where large CQ values and high asymme-
try parameters were observed.26 The sodium ions in that case
are found in highly distorted four-coordinate environments. The
compound that is most similar to anhydrous sodium valproate
whose crystal structure has been solved is anhydrous sodium
Published by NRC Research Press

Dicaire et al.
15
(8) Petruševski, G.; Naumov, P.; Jovanovski, G.; Bogoeva-Gaceva, G.;
Ng, S. W. ChemMedChem 2008, 3, 1377. doi:10.1002/cmdc.200800112.
(9) Nangia, A. Cryst. Growth Des. 2006, 6, 2. doi:10.1021/cg050343e.
(10) Brittain, H. G. J. Pharm. Sci. 2012, 101, 464. doi:10.1002/jps.22788.
(11) Petruševski, G.; Naumov, P.; Jovanovski, G.; Ng, S. W. Inorg. Chem. Commun.
2008, 11, 81. doi:10.1016/j.inoche.2007.10.024.
(12) Potrzebowski, M. J. Crystallography and NMR: Applications to Organic and
Pharmaceutical Chemistry; In Encyclopedia of Magnetic Resonance; John Wiley &
Sons: New York, 2008; doi:10.1002/9780470034590.emrstm1005.
(13) Harris, R. K. Polymorphism and Related Phenomena; In Encyclopedia of Magnetic
Resonance; John Wiley & Sons: New York, 2007; doi:10.1002/9780470034590.
emrstm0410.
(14) Xu, M.; Harris, K. D. M. Cryst. Growth Des. 2008, 8, 6. doi:10.1021/cg701077p.
(15) Hamaed, H.; Pawlowski, J. M.; Cooper, B. F. T.; Fu, R.; Eichhorn, S. H.;
Schurko, R. W. J. Am. Chem. Soc. 2008, 130, 11056. doi:10.1021/ja802486q.
(16) Bryce, D. L.; Bultz, E. B.; Aebi, D. J. Am. Chem. Soc. 2008, 130, 9282. doi:10.1021/
ja8017253.
(17) (a) Widdifield, C. M.; Bryce, D. L. J. Phys. Chem. A 2010, 114, 2102. doi:10.1021/
jp909106j; (b) Widdifield, C. M.; Bryce, D. L. J. Phys. Chem. A 2010, 114, 10810.
doi:10.1021/jp108237x.
(18) Johnston, K. E.; Griffin, J. M.; Walton, R. I.; Dawson, D. M.; Lightfoot, P.;
Ashbrook, S. E. Phys. Chem. Chem. Phys. 2011, 13, 7565. doi:10.1039/
c1cp20258h.
(19) Fernandez, C.; Pruski, M. Top. Curr. Chem. 2012, 306, 119. doi:10.1007/
128_2011_141.
(20) (a) Samoson, A.; Lippmaa, E.; Pines, A. Mol. Phys. 1988, 65, 1013. doi:10.1080/
00268978800101571; (b) Samoson, A.; Pines, A. Rev. Sci. Instrum. 1989, 60,
3239. doi:10.1063/1.1140558.
(21) (a) Frydman, L.; Harwood, J. S. J. Am. Chem. Soc. 1995, 117, 5367. doi:10.1021/
ja00124a023; (b) Medek, A.; Harwood, J. S.; Frydman, L. J. Am. Chem. Soc. 1995,
117, 12779. doi:10.1021/ja00156a015.
(22) Hung, I.; Howes, A. P.; Anupõld, T.; Samoson, A.; Massiot, D.; Smith, M. E.;
Brown, S. P.; Dupree, R. Chem. Phys. Lett. 2006, 432, 152. doi:10.1016/j.cplett.
2006.10.085.
(23) Brinkmann, A.; Kentgens, A. P. M.; Anupõld, T.; Samoson, A. J. Chem. Phys.
2008, 129, 174507. doi:10.1063/1.3005395.
(24) Perras, F. A.; Korobkov, I.; Bryce, D. L. CrystEngComm 2013, 15, 8727. doi:10.
1039/C3CE40875B.
(25) Hung, I.; Wong, A.; Howes, A. P.; Anupõld, T.; Samoson, A.; Smith, M. E.;
Holland, D.; Brown, S. P.; Dupree, R. J. Magn. Reson. 2009, 197, 229. doi:10.
1016/j.jmr.2009.01.005.
(26) Burgess, K. M. N.; Perras, F. A.; Lebrun, A.; Messner-Henning, E.;
Korobkov, I.; Bryce, D. L. J. Pharm. Sci. 2012, 101, 2930. doi:10.1002/jps.23196.
(27) Perras, F. A.; Korobkov, I.; Bryce, D. L. Phys. Chem. Chem. Phys. 2012, 14, 4677.
doi:10.1039/c2cp40273d.
(28) Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.; Griffin, R. G.
J. Chem. Phys. 1995, 103, 6951. doi:10.1063/1.470372.
Conclusions
We have presented the results of the solid-state NMR study of
various solid forms of sodium valproate and its hydrates. Charac-
terization of pure forms of the various hydrates can be challeng-
ing due to their ability to interconvert. Nevertheless, several
insights into the number of distinct sodium cations, and thus the
number of valproate molecules, in the asymmetric unit, the sym-
metry of their local environment, and the degree of overall crys-
tallinity were obtained for five different samples. As only one of
these forms has been characterized by SCXRD, these insights ob-
tained through SSNMR are particularly valuable.
Through a combination of X-ray and NMR experiments, com-
pounds C and F were shown to be very similar or identical. Three
nonequivalent sodium sites in the asymmetric unit were clearly
identified with 23Na SSNMR spectroscopy. Differences in the
quadrupolar parameters for the sodium in a pseudo-tetrahedral
coordination environment and those in distorted square planar
environments were conclusive. The spectra measured for the
commercial sample of sodium valproate suggest the presence of
two nonequivalent sodium sites, and valproate molecules, and
that the sample is highly crystalline. Comparison of the quadru-
polar parameters for these sites with those of compound C, and
other samples from the literature, suggests that the sites in the
commercial sample are in distorted square planar environments.
As previously described, a distinct form of sodium valproate is
obtained by pressing one of the other forms with KBr; this sample
has been shown via 23Na SSNMR to be a largely amorphous prod-
uct of an ion-exchange reaction with potassium. The quadrupolar
and shielding data for compound E suggest that the former fea-
tures two nonequivalent sodium sites in pseudo-octahedral co-
ordination environments. Finally, the quadrupolar parameters
measured for anhydrous compound H suggest that there is a single
sodium site in a crystalline environment with a local, distorted
four-coordinate geometry. This compound is, however, highly hy-
groscopic and an amorphous, hydrated phase quickly forms when
exposed to moisture.
(29) Eichele, K.; Wasylishen, R. E. WSolids1 NMR Simulation Package; Universität
Tübingen, version 1.19.11, 2009.
(30) (a) Gan, Z.; Kwak, H.-T. J. Magn. Reson. 2004, 168, 346. doi:10.1016/j.jmr.2004.
03.021; (b) Amoureux, J. P.; Delevoye, L.; Steuernagel, S.; Gan, Z.;
Ganapathy, S.; Montagne, L. J. Magn. Reson. 2005, 172, 268. doi:10.1016/j.jmr.
2004.11.001.
(31) Samoson, A.; Lippmaa, E. J. Magn. Reson. 1989, 84, 410. doi:10.1016/0022-
2364(89)90389-2.
Supplementary material
Supplementary material is available with the article through
the journal Web site at http://nrcresearchpress.com/doi/suppl/
10.1139/cjc-2013-0442.
(32) Fung, B. M.; Khitrin, A. K.; Ermolaev, K. J. Magn. Reson. 2000, 142, 97. doi:10.
1006/jmre.1999.1896.
(33) (a) Kentgens, A. P. M.; Verhagen, R. Chem. Phys. Lett. 1999, 300, 435. doi:10.
1016/S0009-2614(98)01402-X; (b) Perras, F. A.; Viger-Gravel, J.;
Burgess, K. M. N.; Bryce, D. L. Solid State Nucl. Magn. Reson. 2013, 51–52, 1.
doi:10.1016/j.ssnmr.2012.11.002.
(34) Hung, I.; Wong, A.; Howes, A. P.; Anupold, T.; Past, J.; Samoson, A.; Mo, X.;
Wu, G.; Smith, M. E.; Brown, S. P.; Dupree, R. J. Magn. Reson. 2007, 188, 246.
doi:10.1016/j.jmr.2007.07.009.
(35) Smith, S. A.; Levante, T. O.; Meier, B. H.; Ernst, R. R. J. Magn. Reson., Ser A
1994, 106, 75. doi:10.1006/jmra.1994.1008.
Acknowledgements
N.M.D. is grateful to the University of Ottawa for an Undergrad-
uate Research Scholarship. F.A.P. and D.L.B. thank the Natural
Sciences and Engineering Research Council of Canada for a schol-
arship and for research funding, respectively. We thank Dr. Henry
J. Stronks and Bruker Biospin for their support of our DOR re-
search program.
(36) Metz, G.; Wu, X.; Smith, S. O. J. Magn. Reson. Ser. A 1994, 110, 219. doi:10.1006/
jmra.1994.1208.
References
(1) (a) Aldenkamp, A.; Vigevano, F.; Arzimanoglou, A.; Covanis, A. Acta Neurol. Scand.
2006, 114, 1. doi:10.1111/j.1600-0404.2006.00666.x; (b) Ben-Menachem, E.;
Schmitz, B.; Tomson, T.; Vajda, F. Acta Neurol. Scand. 2006, 114, 14. doi:10.1111/
j.1600-0404.2006.00667.x; (c) Perruca, E.; Aldenkamp, A.; Tallis, R.;
Krämer, G. Acta Neurol. Scand. 2006, 114, 28. doi:10.1111/j.1600-0404.2006.
00668.x.
(37) (a) Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. I. J.;
Refson, K.; Payne, M. C. Z. Kristallogr. 2005, 220, 567. doi:10.1524/zkri.220.5.
567.65075; (b) Segall, M. D.; Lindan, P. J. D.; Probert, M. J.; Pickard, C. J.;
Hasnip, P. J.; Clark, S. J.; Payne, M. C. J. Phys. Condens. Matter 2002, 14, 2717.
doi:10.1088/0953-8984/14/11/301; (c) Pickard, C. J.; Mauri, F. Phys. Rev. B 2001,
63, 245101. doi:10.1103/PhysRevB.63.245101; (d) Profeta, M.; Mauri, F.;
Pickard, C. J. J. Am. Chem. Soc. 2003, 125, 541. doi:10.1021/ja027124r.
(38) Wu, G.; Zhu, J. Prog. Nucl. Magn. Reson. Spectrosc. 2012, 61, 1. doi:10.1016/j.
pnmrs.2011.06.002.
(39) Koller, H.; Engelhardt, G.; Kentgens, A. P. M.; Sauer, J. J. Phys. Chem. 1994, 98,
1544. doi:10.1021/j100057a004.
(40) Perras, F. A.; Bryce, D. L. J. Phys. Chem. C 2012, 116, 19472. doi:10.1021/
jp308273h.
(2) Löscher, W. CNS Drugs 2002, 16, 669. doi:10.2165/00023210-200216100-00003.
(3) Smith, L. A.; Cornelius, V. R.; Azorin, J. M.; Perugi, G.; Vieta, E.; Young, A. H.;
Bowden, C. L. J. Affect. Disord. 2010, 122, 1. doi:10.1016/j.jad.2009.10.033.
(4) Bowden, C. L.; Singh, V. Acta Psychiatr. Scand. 2005, 111, 13. doi:10.1111/j.1600-
0447.2005.00522.x.
(5) Azorin, J. M.; Findling, R. L. CNS Drugs 2007, 21, 1019. doi:10.2165/00023210-
200721120-00005.
(6) Bowden, C. L. Neuropsychopharmacology 1998, 19, 194. doi:10.1016/S0893-
133X(98)00025-6.
(7) Lagace, D. C.; O’Brien, W. T.; Gurvich, N.; Nachtigal, M. W.; Klein, P. S. Clin.
(41) Draèínský, M.; Hodgkinson, P. CrystEngComm 2013, 15, 8705. doi:10.1039/
C3CE40612A.
(42) Paixão, J. A.; Nascimento, F. V.; Matos Beja, A.; Ramos Silva, M. Acta Cryst. C
2000, 56, 168. doi:10.1107/S0108270199014080.
Neurosci. Res. 2004, 4, 215. doi:10.1016/j.cnr.2004.09.013.
Published by NRC Research Press