Sunteți pe pagina 1din 7

DOI: 10.

1021/cg100522f

Characterization and Structure Analysis of Thiamine Hydrochloride 2010, Vol. 10


Methanol Solvate 4414–4420
Paroma Chakravarty and Raj Suryanarayanan*
University of Minnesota, Department of Pharmaceutics, Minneapolis, Minnesota 55455

Received April 19, 2010; Revised Manuscript Received August 24, 2010

ABSTRACT: Thiamine hydrochloride (THCl) forms a monomethanolate (MM) upon exposure of crystalline thiamine phases
(thiamine hydrochloride hemihydrate/HH, nonstoichiometric hydrate/NSH, and anhydrate/AH) to anhydrous methanol (solvent
and vapor). Desolvation of MM at 50-80 C resulted in the formation of a poorly crystalline intermediate which crystallized to
AH at elevated temperatures (g150 C). When exposed to water vapor (11-75% RH, RT), MM transformed to HH (NSH was
detected at e40% RH), while exposure to polar solvent vapor resulted in direct formation of AH. MM was stable in the presence of
nonpolar (benzene and hexane) solvent vapor. Crystal structure analysis revealed a near identical H-bonding network in the two
solvate (MM, and NSH with 1 mol water/mol THCl) lattices. The structure of MM as well as the solvent properties were identified
as the key factors influencing both the kinetics of desolvation of MM and the nature of the product phase.

Introduction hydrate formation.6 Analogous to polymorphs, where a


change in lattice packing or conformation can bring about
Pharmaceutical solvates are molecular adducts containing
pronounced changes in physicochemical properties, desolva-
a parent compound (drug or excipient) and a solvent in the
same crystal lattice.1 Hydrates, where water is the solvent of tion can considerably alter pharmaceutically relevant proper-
crystallization, are the most abundant solvates due to the high ties, including solubility and stability.1
propensity of water to be entrapped in the lattice by virtue of Thiamine hydrochloride (THCl, vitamin B1), our model
its small size and tendency to form multidirectional hydrogen compound, exists in several solid-state forms. The hydrated
bonds.2 However, investigation of other nonaqueous solvates, forms are a nonstoichiometric hydrate (NSH) with ∼1 mol
such as in this study, is also of interest, as this provides water/mol anhydrous THCl and a hemihydrate (C12H18OCl2-
(i) insights into intermolecular interactions which lead to N4S 3 0.5H2O, HH). Both dehydrate to form an anhydrous
solvent inclusion in the lattice and (ii) offers new routes of form (AH) which is isostructural with NSH. The hydrated and
habit modification, polymorph formation, and control of anhydrous forms of THCl have been the subject of detailed
nucleation rates.3 Based on a recent analysis of the Cambridge discussion in our previous publications.7,8 Both the anhy-
Structural Database, it was evident that ∼7% of organic drous and hydrated forms of THCl, when placed in contact
compounds form nonaqueous solvates, with the most com- with methanol, form a monomethanolate (C12H18OCl2N4S 3
monly occurring solvents being methanol, acetone, benzene, CH3OH, MM). The preparation and the preliminary charac-
dichloromethane, and ethanol. In the last several years, there terization of the methanol solvate have been reported.9 How-
has been a dramatic increase in the relative number of organic ever, no attempt was made to relate the structure of THCl
compounds forming solvates.4 In light of the increased in- phases to their stability. Earlier, while evaluating the dehydra-
cidence of solvate formation, it is critical to obtain a thorough tion behavior of NSH, we hypothesized that, despite the lack of
understanding of their structure, physical stability (propensity continuous hydration channels in the crystal lattice, coopera-
to desolvate), and also the implication of desolvation on the tive deformation of the thiamine molecules allowed a nondis-
properties of the active pharmaceutical ingredient (API). ruptive departure of water molecules during dehydration. In
The structure of the solvate and the kinetics of desolvation contrast, the strong ion-dipole interactions (water-chlorine)
are expected to influence the nature of the product phase. The appeared to be responsible for the unusual physical stability of
three possibilities are (i) formation of an isostructural an- HH. In this continuing study, we have attempted to relate the
hydrate or “desolvated solvate” where the lattice packing is structure of the methanol solvate to its physical stability.7,8
retained upon solvent loss, (ii) recrystallization into a structu- The objectives of this study were to (1) characterize MM
rally distinct desolvate, and (iii) formation of a desolvate with and the desolvated product phases, (2) develop a mechanistic
complete lattice collapse.5 In (i) and (ii), the lattice may be understanding of MM physical stability based on its crystal
partially disordered. Loss of lattice solvent may also be structure, and (3) determine the influence of solvent (both
brought about by exposure of a solvate to another solvent polar and nonpolar) properties on the nature of its interac-
vapor, wherein an anhydrous product or a different solvated tion with MM. In order to accomplish the last objective, the
form may be obtained. This process, termed “solvent ex- desolvation kinetics of MM was evaluated following exposure
change”, has been widely observed in cephalosporins, for to the vapor phase of organic solvents. The resulting product
example in cefamandole sodium methanolate, where exposure phase was also characterized.
to water vapor resulted in rapid desolvation followed by
Experimental Procedure
*To whom correspondence should be addressed. Address: 9-157 Weaver
Densford Hall, 308 Harvard Street SE, University of Minnesota, MN 55455. Materials. THCl, as received from Sigma Aldrich (St.
Phone: 612-624-9626. Facsimile: 612-626-2125. E-mail: surya001@umn.edu. Louis, MO), was characterized to be a nonstoichiometric
pubs.acs.org/crystal Published on Web 09/08/2010 r 2010 American Chemical Society
Article Crystal Growth & Design, Vol. 10, No. 10, 2010 4415

hydrate (NSH) by thermogravimetry (TGA) and X-ray Ambient Temperature X-ray Powder Diffractometry
powder diffractometry (XRPD). NSH, when exposed to (XRPD). Powder samples were exposed to Cu KR radiation
anhydrous CaSO4 at 40 C for 4 h (∼0% RH), yielded AH. (45 kV  40 mA) in a wide-angle diffractometer (D 5005,
HH was obtained by suspending NSH in water for ∼12 h and Bruker AXS). The angular range was 5-40 2θ, with a step
then drying the resultant slurry at 40 C in a forced air oven size of 0.05 2θ and a dwell time of 1 s. XRPD data analyses
for ∼4 h.7,8 NSH, AH, and HH were stored in contact with were performed using commercial software (JADE, version
anhydrous methanol and also exposed to methanol vapor for 8, Materials Data Inc., Livermore, CA).
24 h to obtain MM. Crystals of MM were obtained by Single-Crystal X-ray Diffractometry. A crystal was placed
dispersing ∼225 mg of HH powder over 15 mL of anhydrous on the tip of a 0.1 mm diameter glass capillary and mounted
methanol for 7 days at room temperature. on a diffractometer equipped with a CCD area detector
Exposure to Solvent Vapor. MM samples were stored at (Bruker SMART system, version 5.054, 2001, Bruker AXS
room temperature (24.0 ( 1 C) in chambers where the Inc., Madison, WI) for data collection (123 K). Data was
relative humidity (RH) was maintained at 11, 22, 32, 40, 58, collected using Mo KR radiation (graphite monochromator)
and 63% RH using saturated salt solutions.10 MM was stored with a frame time of 10 s and a detector distance of 4.8 cm. A
at 50 C for 5 h over anhydrous CaSO4 (∼0% RH), and the preliminary set of cell constants was calculated using reflec-
product phase, after analysis by XRPD, was exposed to tions harvested from three sets of 20 frames each. These initial
ethanol vapor for 1 week. For the “solvent exchange” studies, sets of frames were oriented in such a way that orthogonal
MM powder was exposed to solvent (ethanol, acetonitrile, wedges of reciprocal space were surveyed, which in turn
acetone, isopropyl alcohol, benzene, hexane, and 75% RH) produced initial orientation matrices determined from 85
vapor for 48 h at room temperature. Aliquots were with- reflections. A randomly oriented region of reciprocal space
drawn periodically and analyzed by TGA and XRPD. was surveyed up to one sphere and to a resolution of 0.77 Å.
Methods. Differential Scanning Calorimetry (DSC). A Four major sections of frames were collected with 0.30 steps
differential scanning calorimeter (MDSC, Model 2920, TA in ω at four different φ settings and a detector position of -28
Instruments, New Castle, DE), equipped with a refrigerated in 2θ. The intensity data were corrected for absorption and
cooling accessory, was used. The instrument was calibrated decay (SADABS).11 Final cell constants were calculated from
with tin. In three separate experiments, about 5 mg of MM 2973 strong reflections from the actual data collection after
was weighed into an open pan and subjected to three tem- integration (SAINT-Plus, version 7.34A, 2008).
perature programs, under dry nitrogen purge. The sample The structure was solved using SIR-9212 and refined using
was subjected to one of the following: (i) Heat treatment the Bruker SHELXTL suite of programs (version 6.14,
from RT to 250 at 10 C/min. (ii) Heat treatment from RT to 2000). The space group P1 was determined based on sys-
50 C and held isothermally for 5 h. The sample was then tematic absences and intensity statistics. A direct-methods
cooled to 25 C and reheated to 250 C. Both the heating and solution was calculated which provided most non-hydrogen
cooling rates were 10 C/min. (iii) Heat treatment from 25 to atoms from the E-map. Full-matrix least-squares/difference
50 C at 10 C/min (standard mode) and then in the modu- Fourier cycles were performed which located the remaining
lated mode, using an amplitude of (1 C/min and a fre- non-hydrogen atoms. All non-hydrogen atoms were refined
quency of 40 s, up to 220 C. with anisotropic displacement parameters. All hydrogen
Thermogravimetric Analysis (TGA). In a thermogravi- atoms were placed in ideal positions and refined as riding
metric analyzer (Model Q50 TGA, TA Instruments, New atoms with relative isotropic displacement parameters.
Castle, DE), ∼8 mg of sample was heated in an open alu- The final full matrix least-squares refinement converged to
minum pan from RT to 250 at 10 C/min under dry nitrogen R1 = 0.0280 and wR2 = 0.0744 (F2, obs data).
purge. The temperature calibration was performed with
Alumel and nickel. Weight calibration was performed with Results
standard weights of 100 mg and 1 g. The DSC and TGA data
were analyzed using commercial software (Universal Ana- Thermal Analysis. The chemical structure of thiamine
lysis 2000, TA Instruments, New Castle, DE). hydrochloride is shown in Figure 1. MM, obtained upon
Water Sorption and Desorption. About 10-15 mg of MM exposing NSH, AH, and HH to anhydrous methanol or
was placed in a quartz sample pan of an automated vapor methanol vapor, was characterized to be a monomethanolate.
sorption analyzer (DVS 1000, Surface Measurement Systems, Figure 2a shows the overlaid DSC and TGA profiles of MM.
U.K.). The sample was dried at 0% RH (25 C) using a nitrogen In the TGA profile, a weight loss of 8.4% was in good agree-
flow rate of 200 mL/min for 1000-1500 min and subsequently ment with the solvent stoichiometry in the lattice. The TGA
exposed to 20% RH for 1400 min. In a different experiment, a weight loss coincided with the desolvation and vaporization
dried sample was exposed to 50% RH for 3000 min. endotherm observed in the DSC profile at 50-125 C. Ad-
Variable Temperature X-ray Powder Diffractometry ditionally, the DSC profile revealed an exotherm at ∼160-
(VTXRPD). MM powder samples were exposed to Cu KR 180 C. The exotherm was also observed in the same tempera-
radiation (45 kV  40 mA) in a wide-angle diffractometer ture range for the sample held isothermally at 50 C, which did
(Scintag, XDS 2000) equipped with a variable temperature not show a desolvation/vaporization endotherm in the second
stage (Micristar, model 828D, R. G. Hansen & Associates, heating cycle. This indicated solvent loss during the isothermal
Santa Barbara, CA; working temperature range -190 to hold (data not shown). When a temperature modulation was
300 C) for variable temperature XRPD experiments. The applied, the reversing heat flow signal revealed an endotherm
sample was packed in an aluminum holder and heated from in the same temperature range of 160-180 C (Figure 2b).
RT to 200 at 10 C/min with an isothermal hold at 50 C for This indicated melting of the desolvated phase and its recrys-
5 h. Scans were taken at select temperatures over the angular tallization into another anhydrous form. In an effort to obtain
range 5-40 2θ, using a step size of 0.05 2θ and a dwell time phase information, MM was subjected to variable temperature
of 1 s. XRPD (details in the next section).
4416 Crystal Growth & Design, Vol. 10, No. 10, 2010 Chakravarty and Suryanarayanan

Figure 1. Chemical structure of thiamine hydrochloride.

Figure 2. (a) DSC and TGA profiles of MM. (b) Modulated DSC
of MM. The total (;), reversing (- - -), and nonreversing ( 3 3 3 )
curves are shown.

Phase Transformation. Figure 3a shows the XRPD patterns


of AH, NSH, HH, and MM. The characteristic peaks of AH (8.9 Figure 3. (a) XRPD patterns of AH, NSH, HH, and MM. The
and 11.9 2θ), NSH (8.5 and 11.4 2θ), HH (8.1 and 10.3 2θ), characteristic peaks of AH (8.9 and 11.9 2θ, *), NSH (8.5 and 11.4
and MM (8.0 and 9.8 2θ) are indicated. In an effort to interpret 2θ, þ), HH (8.1 and 10.3 2θ, #), and MM (8.0 and 9.8 2θ, ∧) are
the DSC results, MM was subjected to variable temperature indicated. (b) XRPD patterns obtained when MM was subjected to a
controlled temperature program from 26 to 200 C. Only a few
XRPD, wherein it was heated to 200 C (Figure 2b). The re- representative XRPD patterns are shown. The characteristic peaks
moval of solvent, which was complete by ∼80 C (based on of AH appeared in the patterns collected at 160 and 180 C and are
TGA), resulted in a poorly crystalline lattice, which is evident indicated by f. The reference pattern of AH is included for reference
from the broad halo as well as the broad poorly resolved peaks (uppermost pattern). (c) contains the XRPD patterns at some inter-
over the angular range of 15-30 2θ (Figures 3b and c). Upon mediate temperatures wherein the broad, poorly resolved peaks in the
further heating, the characteristic peaks of AH appeared at pattern reveal the partially crystalline nature of the intermediate.
160 C (Figure 3b).
Figure 4 shows the water sorption profiles of MM follow- loss indicated expulsion of sorbed water - a consequence of
ing exposure of individual samples to 20 (panel a) and 50% recrystallization. XRPD (of sample withdrawn at point D)
RH (panel b) at 25 C. When first exposed to 0% RH (region revealed a mixture of NSH and HH. When the sample was
A-B; top panel), ∼7% weight loss was observed (panel a). exposed to 50% RH after initial drying at 0% RH (panel b),
Upon exposure to water vapor (20% RH), there was in- the weight loss was more pronounced and abrupt (1.3%).
stantaneous sorption of ∼3.3% w/w water (region B-C; top XRPD of the sample withdrawn at point D revealed forma-
panel). Following water sorption, there was a small but tion of HH.
gradual weight loss of ∼0.8%. Exposure to 0% RH caused Upon storage of MM at different relative humidites
desolvation, and the lattice of the product phase is expected (11-63%, RT; 1 week), it transformed to HH (Figure 5).
to be substantially disordered (based on variable tempera- At lower RH values (11-40%), NSH was detected, while the
ture XRPD; Figure 3b). This material has a strong tendency transformation appeared to be complete in 1 week at higher
to sorb water (B-C). The pronounced but gradual weight RH (g58%).
Article Crystal Growth & Design, Vol. 10, No. 10, 2010 4417

Figure 5. XRPD patterns of MM following exposure to RH values


ranging from 11 (bottom pattern) to 64% (top pattern), each for 7
days at RT. While both NSH (þ) and HH (#) were observed at lower
RH conditions (11-40%), only HH was observed at RH >40%.

respectively. The single crystal data and the hydrogen bond-


ing interactions in the lattice are tabulated in Tables 2 and 3,
respectively. The hydrogen bonding network, involving the
two chloride ions, is identical in the NSH and MM lattices.8
Unlike NSH, where each water molecule forms two hydrogen
bonds (with the chloride ion and the hydroxyl group of the
ethylene side chain), the methanol molecule is associated with
THCl by means of a single hydrogen bond of the hydroxyl
group to the chloride ion (Figure 8b and Table 3). As in the
case of NSH, the solvent molecules are lined along the a-axis
in the lattice.

Discussion
As mentioned earlier, our objective was to develop a mecha-
nistic understanding of the physical stability of MM based on
its crystal structure. MM has been referred to as an “un-
stable” methanolate, since it desolvates readily even under
ambient conditions.13 This can be explained by the hydrogen
Figure 4. (a) Water sorption profile of MM at 25 C. The sample bonding environment of the solvent molecule in the lattice. In
was first dried at 0% RH for ∼1000 min (region A-B) and then the MM lattice (Figure 8b), the methanol molecule is linked
exposed to 20% RH (region B-D) and analyzed by XRPD at point to THCl molecules by a single hydrogen bond (ion-dipole
D [inset: mixture of NSH (þ) and HH (#)]. (b) In this case, the dried
interaction) via the chloride ion [Cl(2)]. Since Cl(2) acts as a
sample (region A-B) was exposed to 50% RH (region B-D).
XRPD (inset) revealed only HH (#). double acceptor (hydrogen bonding with the hydrogen atoms
of O(2) and N(2)), the two donors form “two approximately
“Solvent Exchange” Studies. Figure 6 shows the XRPD parallel dipoles which repel each other”.8,14 Such a pair of
and TGA profiles of MM following exposure to different hydrogen bonds may therefore be considered anticoopera-
solvents. Upon exposure to water vapor (75% RH), HH was tive, which effectively cancel out or “reduce the strength” of
formed in 14 h. AH was the final product phase following each other. Thus, the weakly bonded solvent molecule may be
exposure to all the other polar solvents. XRPD also revealed readily removed from the lattice, explaining the poor physical
AH formation upon exposure of desolvated MM to an- stability of MM.
hydrous ethanol for 1 week (data not shown). The desolva- Based on the continuous and unified dehydration model
tion kinetics in the polar nonaqueous solvents can be rank- of Petit and Coquerel (1996), the desolvation of MM is a
ordered as follows: ethanol > IPA > acetone > acetonitrile. “destructive process”, where the loss of solvent leads to partial
Upon exposure to the nonpolar solvents hexane and ben- lattice collapse.15 The physical form of the product phase will
zene, there was negligible MM desolvation in 48 h. The XRPD depend on the water vapor pressure in the atmosphere. In the
profile of MM, upon exposure to hexane, revealed the desol- absence of water vapor, the collapsed lattice recrystallized to
vated intermediate phase (described earlier; Figure 3b), as indi- AH. This was evident from DSC (melting and recrystalliza-
cated by the peak at 8.4 2θ (Figure 6a). Based on the known tion events in Figure 2b) and XRPD (Figure 3b). On the other
stoichiometric methanol content in MM of 8.4%, the % MM hand, exposure to 20% RH, in a water sorption analyzer,
in the sample was calculated based on the weight loss observed resulted in crystallization of a mixture of NSH and HH.
in the TGA (Figure 6b, Table 1). Figure 7 summarizes the phase Exposure to 50% RH resulted in the formation of HH.
transitions in THCl. The dynamic nature of the water sorption analyzer facili-
Single Crystal Structure. Parts a and b of Figure 8 contain tated the rapid attainment of equilibrium. Samples were also
the thermal ellipsoid plot and the packing diagram of MM, stored in constant humidity chambers, wherein, because of the
4418 Crystal Growth & Design, Vol. 10, No. 10, 2010 Chakravarty and Suryanarayanan

Figure 6. (a) XRPD patterns of MM following exposure to solvent vapor for up to 48 h. In the presence of water (75% RH) and anhydrous
ethanol, MM (∧) converted to HH (#) and AH (*), respectively, in 14 h. While MM f AH conversion was observed following exposure to polar
solvents, the kinetics can be rank ordered as follows: IPA > acetone > acetonitrile. The physical stability of MM was unaffected when exposed
to nonpolar solvents (benzene and hexane) for 48 h. The peak at 8.4 2θ, following exposure to hexane, can be attributed to “desolvated MM”.
(b) TGA profiles of MM exposed to solvent vapor for up to 48 h. The % MM remaining is provided in Table 1.

Table 1. % MM Retained Following Exposure to Solvent Vapor for up eventually transform to HH. We had pointed out earlier that,
to 48 ha at higher RH values (g58%), the transformation of MM to
solvent time (h) MM remaining, % w/w HH was very rapid.
water (75% RH) 14 0 It is also important to note that, upon exposure of NSH to
ethanol 14 0 water vapor, there is considerable resistance to lattice rear-
isopropanol (IPA) 14 51 rangement to form HH. This is because the transformation
24 13 involves conversion of one monoclinic form (P21/n, NSH) to
48 0
acetone 14 65 another (C2/c, HH). Te et al. detected HH formation only
24 61 after 30 days of exposure of AH to 60% RH at RT.16 In the
48 8 case of MM, the conversion to HH occurred at a much lower
acetonitrile 24 100 RH of 11% and in just 7 days. One possible explanation is that
48 74
benzene 48 100
the poor lattice symmetry (triclinic cell, P1) in MM, coupled
hexane 48 98 with the affinity of water to methanol, facilitates lattice re-
a
arrangement and HH formation even at RH values as low as
The quantification was based on TGA (Figure 6b). 11% (Figure 5). It has recently been suggested that the affinity
of the solvent in the lattice to the solvent in the vapor phase
much longer experimental time scale, metastable phases can will dictate the ease of desolvation.17
often be detected. When stored over the RH range of 11 to Exposure of MM to polar solvent vapor (ethanol, isopro-
40%, the formation of NSH was discernible. This can be panol, acetone, and acetonitrile) revealed conversion to AH
explained by the structural similarity between MM and NSH. within 48 h. Such polar solvents are reported to function as
In this RH range, since HH is the stable phase, MM will “molecular looseners” and break the hydrogen bond network
Article Crystal Growth & Design, Vol. 10, No. 10, 2010 4419

Table 2. Single Crystal Data for THCl Monomethanolate (MM)


empirical formula C13H22Cl2N4O2S
formula weight 369.31
temperature 123(2) K
crystal system triclinic
space group P1
unit cell dimensions a = 7.2946(8) Å R = 105.550(1)
b = 11.0874(13) Å β = 90.819(1)
c = 11.3981(13) Å γ = 101.825(1)
volume 866.92(17) Å3
Z 2
density (calculated) 1.415 mg/m3
goodness-of-fit on F2 1.056
final R indices R1 = 0.0280, wR2 = 0.0715
[I > 2σ(I)]
Figure 7. Schematic representation of phase transformations R indices (all data) R1 = 0.0327, wR2 = 0.0744
(upon solvation/desolvation) in THCl. The focus here is on the
methanol solvate (MM). The phase transformations of NSH and
HH were presented earlier.7,8 þ, solution mediated; þþ, vapor Table 3. Hydrogen Bonding in the MM Lattice, Determined from the
phase mediated. Single-Crystal Structure, with the Specific Hydrogen Bonds of Interest
Marked (A) and (B)
D-H 3 3 3 A d(D-H)a d(H 3 3 3 A)b d(D 3 3 3 A)c — (DHA)d
O1-H1A 3 3 3 Cl1 0.84 2.31 3.1275 (12) 165.0
N2-H2A 3 3 3 Cl2#1(B) 0.88 2.20 3.0760 (12) 171.2
N4-H4A 3 3 3 Cl1#2 0.88 2.36 3.2250 (12) 168.5
N4-H4B 3 3 3 Cl1 0.88 2.38 3.2190 (12) 158.4
O2-H2D 3 3 3 Cl2(A) 0.84 2.36 3.1991 (13) 175.2
a
Donor-hydrogen distance (Å). b Hydrogen-acceptor distance (Å).
c
Donor-acceptor distance (Å). d Hydrogen-bond angle (deg).

The MM f AH conversion occurs more rapidly in the pre-


sence of alcohols than the vapor phases of the other solvents
(Figure 6 and Table 1). Possibly, the solubility difference of
MM in these solvents (water vs alcohols vs acetone and
acetonitrile) influences the kinetics of nucleation and growth
of AH and therefore contributes to the differences in the
observed desolvation kinetics. While we do not have a mecha-
nistic explanation for this observation, the hydroxyl groups in
the solvent vapor may be effective in disrupting the methanol-
THCl hydrogen bonding in the lattice.18 Although there
have been several approaches to rank order the polarity and
the hydrogen bonding ability of solvents (dielectric constant
values, Kosower’s Z scale, acceptor number/AN), these can-
not fully explain the differences in the kinetics of MM deso-
lvation in the presence of the solvents tested.19-21 This is
because the dielectric constant and Z value of acetonitrile are
higher than those of acetone, indicating its higher polarity and
increased hydrogen bonding propensity. However, the desol-
vation was more rapid in acetone. The affinity of the lattice
solvent to the environmental solvent (vapor phase) is also
expected to influence the kinetics of MM desolvation.17
Figure 8. (a) Thermal ellipsoid plot of MM. (b) The hydrogen Conclusions
bonding network of methanol is shown in the simplified packing
diagram. The hydrogen bonds of interest (as indicated in Table 3) In the absence of water vapor, the methanol solvate (MM)
are marked. (A) denotes the hydrogen bonding between the chloride of thiamine hydrochloride desolvates to form a poorly crystal-
ion and the methanol molecule whereas (B) denotes the hydrogen line, anhydrous intermediate which recrystallizes to an-
bonding of the chloride ion with the -NH group of the thiamine
hydrochloride molecule. hydrous THCl (AH). In the presence of water vapor, MM
converts to HH. The concomitant formation of NSH at lower
RH (11-40%) was attributed to the near identical H-bonding
in the MM lattice, thereby facilitating conversion to the in the two solvates. Vapor phase mediated transformation of
thermodynamically stable phase, AH.18 The role of hydrogen MM f AH occurred following exposure to polar solvents.
bonding, in facilitating phase conversion, is also confirmed by MM did not undergo phase transformation in the presence of
the stability of MM in the presence of nonpolar solvent vapor nonpolar solvents.
(following exposure to hexane and benzene for 48 h). These
solvents, in light of their inability to hydrogen bond with the Acknowledgment. We thank Dr. Ramprakash Govindarajan
MM lattice, are unable to cause phase transformation in the for helpful discussions and suggestions for the manuscript.
experimental time scales. Dr. Victor G. Young Jr., X-ray Crystallographic Laboratory,
4420 Crystal Growth & Design, Vol. 10, No. 10, 2010 Chakravarty and Suryanarayanan

Department of Chemistry, University of Minnesota, is acknowl- (7) Chakravarty, P.; Berendt, R. T.; Munson, E. J.; Young, V. G., Jr.;
edged for the MM structure solution. Paroma Chakravarty is Govindarajan, R.; Suryanarayanan, R. J. Pharm. Sci. 2010, 99 (2),
the recipient of the Doctoral Dissertation Fellowship from 816–827.
(8) Chakravarty, P.; Berendt, R. T.; Munson, E. J.; Young, V. G., Jr.;
the Graduate School, University of Minnesota. Part of this
Govindarajan, R.; Suryanarayanan, R. J. Pharm. Sci. 2010, 99 (4),
work was carried out in the I.T. Characterization facility of 1882–1895.
the University of Minnesota, which receives partial support (9) Watanabe, A.; Nakamachi, H. J. Pharm. Soc. Jpn. 1976, 96 (10),
from the NSF through the NNIN program. 1236–40.
(10) Nyqvist, H. Int. J. Pharm. Technol. Prod. Manuf. 1983, 4 (2), 47–48.
Supporting Information Available: X-ray crystallographic infor- (11) Blessing, R. H. Acta Crystallogr., Sect. A 1995, A51 (1), 33–8.
mation file (CIF) for MM. This material is available free of charge (12) Altomare, A.; Cascarano, G.; Giacovazzo, C.; Guagliardi, A.
via the Internet at http://pubs.acs.org. The crystal structures of J. Appl. Crystallogr. 1994, 27 (6), 1045–50.
NSH and HH (CCDC deposition numbers 725982 and 725983) are (13) Griesser, U. J. 8th International Conference on the Application of
deposited in the Cambridge Crystallographic Data Center and are Physical Chemistry in Pharmacy, Switzerland, 2004.
available upon request. (14) Steiner, T. Angew. Chem., Int. Ed. 2002, 41 (1), 48–76.
(15) Petit, S.; Coquerel, G. Chem. Mater. 1996, 8 (9), 2247–2258.
References (16) Te, R. L.; Griesser, U. J.; Morris, K. R.; Byrn, S. R.; Stowell, J. G.
Cryst. Growth Des. 2003, 3 (6), 997–1004.
(1) Khankari, R. K.; Grant, D. J. W. Thermochim. Acta 1995, 248, (17) Fujii, K.; Ashida, Y.; Uekusa, H.; Hirano, S.; Toyota, S.; Toda, F.;
61–79. Pan, Z.; Harris, K. D. M. Cryst. Growth Des. 2009, 9 (2), 1201–
(2) Gillon, A. L.; Feeder, N.; Davey, R. J.; Storey, R. Cryst. Growth 1207.
Des. 2003, 3 (5), 663–673. (18) Yoshinari, T.; Forbes, R. T.; York, P.; Kawashima, Y. Int. J.
(3) Brychczynska, M.; Davey, R. J.; Pidcock, E. New J. Chem. 2008, Pharm. 2002, 247 (1-2), 69–77.
32 (10), 1754–1760. (19) Marcus, Y. In The Properties of Solvents; Ed.; John Wiley and Sons:
(4) Griesser, U. J. In Polymorphism; Hilfiker, R., Ed.; Wiley-VCH: West Sussex, England, 1998; pp 160-163.
Weinheim, 2006; pp 211-233. (20) Reichardt, C. Monographs in Modern Chemistry, Vol. 3: Solvent
(5) Garner, W. E. Chemistry of the solid state; Academic Press: Effects in Organic Chemistry; Verlag Chemie: Weinheim, 1979;
New York, 1955; pp 213-231. Vol. 3, pp 14-19, 48-51.
(6) Pikal, M. J.; Lang, J. E.; Shah, S. Int. J. Pharm. 1983, 17 (2-3), (21) Wypych, G. Handbook of Solvents; ChemTec Publishing: Toronto,
237–62. 2001; pp 66-69.

S-ar putea să vă placă și