Interpreting the Disordered Crystal Structure of Sodium Naproxen

Jul 8, 2013 - Synopsis. The crystal structure of sodium naproxen tetrahydrate contains layers of naproxen molecules that are ordered and Na+/H2O regio...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/crystal

Interpreting the Disordered Crystal Structure of Sodium Naproxen Tetrahydrate Andrew D. Bond,*,† Claus Cornett,† Flemming H. Larsen,‡ Haiyan Qu,§ Dhara Raijada,† and Jukka Rantanen† †

Department of Pharmacy, University of Copenhagen, Universitetsparken 2, DK-2100 Copenhagen, Denmark Department of Food Science, University of Copenhagen, Rolighedsvej 30, DK-1958 Frederiksberg, Denmark § Department of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Niels Bohrs Allé 1, DK-5230 Odense, Denmark ‡

S Supporting Information *

ABSTRACT: The crystal structure of the tetrahydrate of the active pharmaceutical ingredient sodium naproxen is examined using single-crystal X-ray diffraction, supported by 13C and 23 Na solid-state NMR. The structure has previously been reported to be a heminonahydrate, Na+(naproxen−)·4.5H2O. The average structure in space group C2 contains layers of naproxen molecules that are ordered, except for two orientations of the carboxyl groups, and Na+/H2O regions that exhibit complex disorder. The atomic positions in the disordered regions are interpreted as Na(H2O)6 octahedra, alternately sharing edges and faces to define 1-D coordination polymers with translational periodicity twice that of the b axis in the average C2 structure. There is also one noncoordinated H2O molecule per two naproxen molecules, giving an overall formula of {Na2(H2O)7}2+(naproxen−)2(H2O). Two resonances seen for the naproxen methyl group in 13C CP/MAS SS-NMR are accounted for by the presence of two orientations along the doubled b axis for the carboxyl group. A single resonance in the 23Na SS-NMR is consistent with local 21/m symmetry in the Na+/H2O regions. The single-crystal X-ray diffraction pattern contains diffuse rods in positions consistent with the doubled b axis, indicating a disordered stacking sequence for the Na+/H2O sections.

1. INTRODUCTION Different solid forms of active pharmaceutical ingredients (APIs) can exhibit different physicochemical properties, which may have an impact on the safety or performance of drug products.1−3 According to regulatory guidelines,4,5 pharmaceutical companies are encouraged to search for alternative solid forms and to assess the risks associated with solid-state factors such as potential polymorphic transformations. In this context, hydrates have a special importance because of the ubiquitous nature of water in our environment, and solid APIs that can exist in different hydration states must be monitored and controlled particularly carefully in order to ensure consistent quality for the final medicinal products.6 Investigation and understanding of hydrates with pharmaceutical relevance, and especially the transformations that occur between different hydration states in anhydrate−hydrate systems, is therefore of significant interest within pharmaceutical materials science.7−9 Sodium naproxen (Scheme 1) is a nonsteroidal antiinflammatory drug (NSAID) that exhibits numerous hydration states. Anhydrate (AH), monohydrate (MH), polymorphic dihydrate (DH), and tetrahydrate (TH) phases are known.10,11 The properties and transformation pathways of the system have been investigated by several groups. Malaj et al. have reported a © 2013 American Chemical Society

Scheme 1. Chemical Structure of Sodium (S)-Naproxen

thorough study of the thermodynamic and kinetic aspects,10 and we have recently applied dynamic vapor sorption (DVS) analysis to map the solid-form landscape of the system and to identify optimal conditions for isolation of the various hydrate phases.11 One of our continued aims for sodium naproxen is to link the established transformation behavior more closely to the molecular-level structure in the solid state. To date, crystal structures have been reported for AH12 and MH,13 but not for the DH polymorphs. Recently, a crystal structure has been reported for TH, which was presented as a heminonahydrate, Na+(naproxen−)·4.5H2O, on the basis of refined site occupancy Received: May 3, 2013 Revised: June 4, 2013 Published: July 8, 2013 3665

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671

Crystal Growth & Design

Article

change was observed. The presented data are from the final measured scan, which was collected after ca. 4 h. 2.4. Solid-State NMR. Considering the susceptibility of TH toward dehydration under ambient conditions, special care was taken during preparation of the samples for NMR analysis. The TH sample generated at 25 °C/95% relative humidity (1 month) was transferred to the NMR tube while keeping the tube as far as possible within the desiccator, then the filled tube was kept in the closed desiccator at 25 °C/95% relative humidity for around half an hour in order to reestablish the equilibrium. The desiccator was then reopened, and the NMR tube sealed immediately. Solid-state NMR spectra were recorded at 313 K on a Bruker Avance 400 spectrometer operating at Larmor frequencies of 100.62, 105.85, and 400.13 MHz for 13C, 23Na, and 1H, respectively, using a double-tuned CP/MAS probe equipped for 4 mm (od) rotors. 13C CP/MAS spectra were recorded using a contact time of 8.5 ms, a spin rate of 12 kHz, a recycle delay of 8 s, 256 scans, and an acquisition time of 40.9 ms during which 1H TPPM decoupling (80 kHz rf-field strength) was employed.20 Cross-polarization was carried out using variable-amplitude CP21 with a maximum rf-field strength of 80 kHz for both 1H and 13C. 13C chemical shifts are referenced to an external sample of α-glycine (carbonyl group) at 176.5 ppm. Single-pulse 23Na MAS NMR spectra were recorded using a 1.8 μs pulse (69.4 kHz rffield strength), spin-rate of 12 kHz, a recycle delay of 16 s, 32 scans, and an acquisition time of 40.9 ms during which 1H TPPM decoupling (80 kHz rf-field strength) was employed. 23Na chemical shifts are referenced to an external sample of 1.0 M NaCl (aq). All data were processed using Topspin 2.122 then transferred to MATLAB23 to set up figures. 2.5. Computational Procedures. Energy minimization of the crystal structure was carried out using the CASTEP module24 in Materials Studio.25 The PBE functional was applied26 with a planewave cutoff energy of 520 eV and a dispersion correction according to Grimme.27 All other parameters were set to the “fine” defaults within Materials Studio. All atomic coordinates and unit-cell parameters were optimized, and symmetry constraints were not applied. This combination has been established to reproduce effectively the geometries of molecular crystal structures.28

factors for the O atoms within an extensively disordered Na+/ H2O region.14 The heminonahydrate formula is not consistent with previous studies, in particular thermogravimetric, sorption, and Karl Fischer data,10,11,15 all of which indicate a tetrahydrate. We present here a reinterpretation of the TH structure that is consistent with the expected tetrahydrate formula, and we show that the complex disorder in the structure can be interpreted either as stacking faults or as multiple twinning by 2-fold rotation around one of the crystallographic axes. “Nonaverage” structural details of this kind are being increasingly recognized in pharmaceutical materials,16−19 and it is hoped that a better understanding of these structures will facilitate improved solidstate structure−property relationships.

2. EXPERIMENTAL SECTION 2.1. Materials and Crystallization. Sodium (S)-naproxen anhydrate (AH; USP grade) was received from Divi’s Laboratories Limited, India. A saturated solution of potassium sulfate was used to create static conditions of 95% relative humidity at 25 °C in a desiccator. Bulk samples of TH were generated by exposing AH to these conditions for 1 week. TH remains stable under these conditions for at least three months but dehydrates rapidly under ambient conditions. Single crystals of TH were obtained by dissolving AH in deionized H2O (Milli-Q, Millipore Corporation) at 60 °C, then insulating the vessel and allowing the solution to cool slowly to room temperature. Needle-like crystals were obtained overnight. The crystals dehydrate rapidly on removal from the mother liquor. A bulk sample of the solution-grown crystals suitable for powder X-ray diffraction was obtained by dissolving AH in deionized H2O at 60 °C, then sucking the hot solution into a glass capillary (0.5 mm od), which was immediately sealed. Crystals were grown inside the capillary on cooling to room temperature, and the capillary was used directly for powder X-ray diffraction. 2.2. Single-Crystal X-ray Diffraction. Single crystals were transferred rapidly from the mother liquor into perfluoropolyether oil, then into an N2 gas stream on the diffractometer, operating at 150 K. This was sufficient to stabilize the crystal for the duration of the data collection. Single-crystal X-ray diffraction data were collected using a Bruker-Nonius X8-APEXII instrument equipped with graphitemonochromated Mo Kα radiation (λ = 0.7107 Å). Structure solution and refinement were carried out using SHELXTL.19 Full details of the structure modeling are given in the Results section. H atoms of the water molecules in the disordered Na+/H2O region were not included in the refined model, so the atomic sites in the accompanying CIF are 32 H atoms short of the empirical formula per unit cell. The absolute structure could not be determined from the diffraction data, and Friedel pairs were merged as equivalent data. A summary of the crystallographic information is provided in Table 1. 2.3. Powder X-ray Diffraction (PXRD). PXRD data were collected using a PANalytical X’Pert Pro diffractometer equipped with a PIXcel detector. Non-monochromated Cu Kα radiation (average λ = 1.5418 Å) was used. Capillary samples were measured in transmission geometry with an incident-beam focusing mirror. Data were collected over the range 2−40° with a point resolution of 0.026° (1346 data points) and a total data collection time of 20 min. Multiple scans were measured to investigate possible dehydration, and the crystals within the sealed capillary were found to be stable over a period of at least 16 h. The data presented are an average of the 50 consecutive scans that were measured. Variable-humidity PXRD data were collected with the addition of an Anton Parr CHC-Plus chamber equipped with an MHG humidity generator. Height alignment of the stage was calibrated prior to each run by measurement of a standard AH pattern. The temperature was maintained at 25 °C, and the atmosphere was ramped to 90% relative humidity over a period of 1 h. Data were collected over the range 4−40° with a total data collection time of 10 min. AH transforms first to DH then to TH (in accordance with previous results10,11), and patterns were collected until no further

3. RESULTS AND DISCUSSION 3.1. Single-Crystal X-ray Diffraction. As for the previous report,14 initial treatment of our diffraction data for TH indicated space group C2 with unit-cell parameters (Table 1) comparable to those reported. In our case, the electron density map contained eight distinct peaks in the disordered Na+/H2O region, three of which appeared to be significantly larger than the other five. Inclusion of the three largest peaks as halfoccupied Na+ and the five other peaks as half-occupied O atoms gave uniform displacement parameters for all eight atoms, although this does not give charge balance. The disordered region appears to comprise {Na(H2O)x}+ coordination polymers extending along the b axis. Burgess et al. referred to these motifs as “zig-zag” chains.14 Two of the Na+ sites in our structure lie clearly within the coordination polymers. The third assumed Na+ site, however, is at the periphery of the polymer and only two-coordinated. This was reassigned as a fully occupied O atom (O5W), and atoms Na1, Na2, and O5W could then be refined to give satisfactory anisotropic displacement parameters. At this point, residual electron density peaks of ca. 1.2 e Å−3 remained on either side of atom O6W (Figure 1), so this was split into two closely spaced sites (O6W and O7W), each assigned site occupancy 0.5. With this model, it was possible to refine all atoms with anisotropic displacement parameters to give good final R-factors (Figure 1 and Table 1) and an overall tetrahydrate stoichiometry. The carboxyl group of the naproxen molecule is disordered over two orientations, 3666

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671

Crystal Growth & Design

Article

3.2. Structure Modeling. Transforming the C2 structure to a corresponding primitive unit cell (denoted by a1, b1, etc.; Figure 2) provides an equivalent structure containing only one

Table 1. Summary of the Crystallographic Data for the Average Model and the Corresponding Primitive Unit Cell Described in the Text average structure empirical formula formula weight T (K) cryst syst space group a (Å) b (Å) c (Å) α (deg) β (deg) γ (deg) vol (Å3) Z/Z′ calcd density (g cm−3) μ (mm−1) data collected unique datab Rint obsd data [I > 2σ(I)] R1 [I > 2σ(I)] wR2 (all data) S ρmin, ρmax (e Å−3)

C14H21NaO7 324.3 150 monoclinic C2 40.819(5) 5.5540(5) 7.0928(7) 90 98.086(4) 90 1594.0(3) 4/1 1.353 0.130 3473 1446 0.041 1026 0.052 0.135 1.01 −0.20, 0.26

primitive cella

triclinic 20.5976 5.5540 7.0928 90 98.012 97.748 797.0

Figure 2. Definition of the C-centered cell (thick lines) and corresponding primitive cell (thin lines) as defined in Table 1. The disordered Na+/H2O region is interpreted as two orientations of the proposed {Na2(H2O)7}2+ coordination polymer as shown. The ordered interpretation requires that the b axis is doubled compared with the average C2 structure.

Na+/H2O region. Using the atomic sites obtained from the Xray refinement, we can contruct a plausible chemical model for the disordered Na+/H2O region by tracing reasonable Na−O and Na−Na distances along the coordination polymer. A search of the Cambridge Structural Database (CSD)29 identified mean values for such distances, according to Table 2. Atomic

a

The primitive cell is not reduced: the stacking axis is placed along the a-axis in order to maintain correspondence with the C-centered cell. The bc planes of the C- and P-cells are directly comparable. bFriedel pairs are treated as equivalent data.

Table 2. Distance Ranges Applied for Interpretation of the Disordered Na+/H2O Region, Based on Searches of the Cambridge Structural Databasea motif

distance

mean (Å)

Na(H2O)6

Na−O

2.44(8)

Na−(OH2)2−Na (edgesharing octahedra)

Na···Na

3.52(13)

Na−(OH2)3−Na (facesharing octahedra)

Na···Na

3.29(8)

refined values (Å) (average structure)b Na1O1W Na1O2W Na1O3W Na1O4W Na1O5W Na1O6W Na2O1W Na2O2W Na2O7W Na1Na2 Na1Na2i

2.392(10) 2.385(9) 2.440(8) 2.339(8) 2.511(6) 2.350(15) 2.417(8) 2.391(9) 2.414(13) 3.497(6) 3.345(8)

a

Details of the search criteria and results are provided in the Supporting Information. bSymmetry code (i): 1 − x, 1 + y, 1 − z.

Figure 1. Asymmetric unit in the average C2 structure. All atoms in the Na+/H2O region have site occupancy 0.5, except for O5W, which is fully occupied. The two disorder components of the carboxyl group are labeled O1/O2 and O1A/O2A. Atoms O6W and O7W are closely spaced but separated in the asymmetric unit (by application of symmetry) for clarity.

positions at reasonable distances can be picked out to define a chain of alternate edge-sharing and face-sharing Na(H2O)6 octahedra, with chemical formula {Na2(H2O)7}2+ (Scheme 2) and a periodicity corresponding to twice the b axis in the average C2 structure (Figure 2). We could identify a comparable coordination polymer in one other (nondisordered) structure in the CSD,30 so this motif appears to be chemically plausible. The Na+/H2O region also contains one water molecule that is not coordinated to Na+ so that the total

each with site occupancy 0.5, as was observed in the previously published structure.14 3667

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671

Crystal Growth & Design

Article

structures.31 The presence of one noncoordinated water molecule in an environment that is rich with Na+ cations is noteworthy, but again this is quite common in the CSD.30,32 Other chemical models might be envisaged for the Na+/H2O region: for example, two {Na(H2O)5}+ polyhedra sharing one edge could give a tetrahydrate unit, {Na2(H2O)8}2+, which is seen in the CSD.33 However, we could not identify any other ordered possibility for the Na+/H2O region that did not contain any unreasonable contacts and was consistent with the observed average structure, including the site occupancy factors. 3.3. Origin of the Disorder. The proposed model also provides a convincing explanation for the observed disorder. The atomic positions in the ordered {Na2(H2O)7}2+(H2O) region conform to approximate 21/m symmetry (Figure 3; see Supporting Information for the structure in electronic format). A 21 screw runs along the b axis (with b2 ≈ 11.1 Å), and centers of inversion lie between the edge-sharing Na(H2O)6 octahedra. The combination of these symmetry elements gives local mirror planes, which pass through the three water molecules of the face-sharing Na(H2O)6 octahedra and the noncoordinated water molecule. The proposed structure overall comprises Na+/H2O sections with 21/m symmetry and layer mesh b2 ≈ 11.1, c2 ≈ 7.1 Å, α2 = 90° sandwiched between layers of naproxen molecules that conform (aside from the carboxyl disorder) to space group symmetry C2 in the unit cell a ≈ 40.82, b ≈ 5.55, c ≈ 7.1 Å, β ≈ 98°. In a given Na + /H 2 O section, the {Na 2 (H 2 O) 7 } 2+ coordination polymer can be accommodated in two different orientations (Figure 2), related to each other by 2-fold rotation parallel to b (i.e., the rotational component of the 21 screw axis) or equivalently by translation of 5.55 Å parallel to b (i.e., the translational component of the 21 screw axis). For the structure containing a single Na+/H2O region in the primitive cell a2 ≈ 20.6, b2 ≈ 11.1, c2 ≈ 7.1 Å, α2 = 90, β2 = 98.0, γ2 = 97.7°, the space group symmetry is reduced overall to P1. All elements of the 21/m symmetry in the Na+/H2O region are broken by the arrangement of the surrounding naproxen molecules, while the 2-fold rotation relating the naproxen molecules is broken by the 21/m symmetry of the Na+/H2O region. This structure in space group P1, which corresponds to the DFT-D minimized structure, is therefore the best conventional representation of the structure.

Scheme 2. Chemical Scheme for the Proposed {Na2(H2O)7}2+ Coordination Polymer

formula can be written as {Na2(H2O)7}2+(naproxen−)2(H2O), consistent with the expected tetrahydrate. With the b axis doubled to accommodate the periodicity of the {Na2(H2O)7}2+ coordination polymer, the primitive unit cell (Figure 2) has dimensions a2 ≈ 20.6, b2 ≈ 11.1, c2 ≈ 7.1 Å, α2 = 90, β2 = 98.0, γ2 = 97.7°. A single orientation of the {Na2(H2O)7}2+(H2O) region was defined within this unit cell, and H atoms were added so as to form a reasonable H-bond network. Minimizing this structure using dispersion-corrected density functional theory (DFT-D) methods resulted in only a minimal change to the unit cell and atomic positions (RMS Cartesian displacement for 88 non-H atoms = 0.11 Å; see Supporting Information). In the minimized model, the carboxyl groups of neighboring naproxen molecules along the b axis adopt two different orientations, consistent with the observed two-component carboxyl disorder in the average structure. Reaveraging of the minimized model into the cell with b ≈ 5.55 Å reproduces very closely the atomic positions in the C2 structure. In the averaged model, each of the Na and O atoms has site occupancy 0.5, but two of the O atoms (O3 and O16 in the energy-minimized structure provided as Supporting Information) are essentially superimposed, thereby producing the observed fully occupied atom O5W. The successful minimization does not guarantee that the model is the only possible atomic arrangement, but the fact that the deviation on minimization is very small supports that the model is plausible. In the minimized structure, most H atoms of the water molecules are involved in clear O−H···O hydrogen bonds. For one water molecule, forming part of the shared face between Na(H2O)6 octahedra, one of the H atoms points toward the naphthalene ring of the naproxen molecule. This type of O−H···π contact is also commonly seen in crystal

Figure 3. Local 21/m symmetry in the ordered model for the Na+/H2O region (b2 ≈ 11.1 Å). The noncoordinated water molecules (highlighted) are obscured in this projection. 3668

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671

Crystal Growth & Design

Article

Since a given Na+/H2O section can adopt two orientations, two Na+/H2O sections in sequence along the a axis can adopt two different relative positions (Figure 4). The DFT-D

Figure 5. Two orientations of the primitive unit cell (red and blue) related by a 2-fold rotation around the b axis. The positions of the naproxen molecules are identical for both orientations. The dashed lines correspond to the C-centered cell of the average structure. Figure 4. Alternative positions for neighboring Na+/H2O sections along the a axis. The two positions (red and blue) are related by a translation of 1/2b (∼5.55 Å) and define the two primitive unit cells shown.

perpendicular direction (e.g., precession image hk2 in Figure 6), these peaks are revealed to be diffuse rods. Thus, the total diffraction pattern comprises sharp Bragg peaks that correspond to the C-centered lattice with b ≈ 5.5 Å plus rods of diffuse intensity halfway between the Bragg peaks along b*. This indicates that the structure contains layers in the bc planes with unit mesh b ≈ 11.1, c ≈ 7.1, α = 90° but that these layers adopt a disordered sequence along the a axis. Along the diffuse rods, two sets of peaks can be identified that correspond to the two orientations of the unit cell discussed in the previous section (Figures 4 and 5). The diffuse nature of the rods (rather than just two sets of isolated Bragg peaks) indicates also the presence of regions where the sequence is less regular. 3.5. Solid-State NMR. The 13C CP/MAS SS-NMR spectrum of TH obtained at 68 MHz has been reported previously by Di Martino et al.15 Our spectrum at 100 MHz (Figure 7) is comparable but provides improved resolution, particularly in the aromatic region (ca. 120−140 ppm) and for the split methyl group resonance (ca. 15 ppm). The paper of Burgess et al. (with the heminonahydrate crystal structure) focuses principally on 23Na SS-NMR, but those authors were not able to obtain a spectrum for TH due to sample dehydration.14 Thus, the 23Na SS-NMR spectrum for TH (Figure 7) has not been published previously. We were able to prevent dehydration of TH for the duration of the SS-NMR measurement by taking the precautions noted in the Experimental Section. We note that the SS-NMR spectra were recorded at 313 K while the single-crystal X-ray data were

minimized structure is one possibility. The other possibility has one Na+/H2O section offset by ±1/2b. This gives an apparent unit cell of a3 ≈ 20.6, b3 ≈ 11.1, c3 ≈ 7.1 Å, α3 = 90, β3 = 97.5, γ3 = 82.3° (Figure 4), obtained from a3 = a2 + 1/2b2, b3 = b2, c3 = c2. Alternatively, the situation can be viewed as twinning of the primitive triclinic unit cell by 2-fold rotation around the b axis (Figure 5). The two views are equivalent; they correspond to application of either the translation or rotation component of the local 21 screw axis present within the Na+/H2O section. The translation/rotation does not affect the positions of the naproxen molecules because both are inherent symmetry operations for the naproxen molecule arrangement. Thus, the naproxen molecules appear to be ordered in the average C2 structure (Figure 5), apart from the carboxyl groups, whose positions are linked to the specific orientation of the neighboring Na+/H2O section. 3.4. Evidence for Disorder in the Diffraction Pattern. Closer examination of the single-crystal X-ray diffraction pattern revealed the presence of diffuse rods of intensity running parallel to a* in positions appropriate for the doubled b axis (Figure 6). Looking along the diffuse rods (reconstructed precession image 0kl), the diffraction pattern looks normal, and sharp peaks indicating the doubled b axis are clearly seen. In the 3669

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671

Crystal Growth & Design

Article

consistent with the proposed local 21/m symmetry, in which all Na atoms are equivalent by (local) symmetry. Despite the overall reduction in the crystallographic symmetry as described, SS-NMR detects the local 21/m symmetry with a single 23Na site. This is a good example of the complementarity of X-ray diffraction and SS-NMR for characterizing disordered structures of this type. 3.6. Powder X-ray Diffraction. Since the disorder corresponds formally to twinning of the primitive triclinic structure by 2-fold rotation, it should not have any influence on the PXRD pattern (aside from any weak diffuse features). The PXRD pattern simulated from the primitive structure is therefore identical to that of the average C2 structure (Figure 8). Burgess et al. noted previously that there is good agreement

Figure 6. Reconstructed precession images from the single-crystal Xray data: 0kl viewed along the diffuse rods; hk2 showing the diffuse rods (horizontal). The grid lines indicate the reciprocal lattice for the primitive cell with b ≈ 5.55 Å (Table 1). In the hk2 image, two sets of Bragg peaks are clearly evident along the diffuse rods.

Figure 8. PXRD patterns simulated from the average C2 structure and the P1 energy-minimized structure (using Mercury,34 contribution from H atoms not included) and measured for the bulk TH sample. The patterns correspond formally to different temperatures, and the influence of thermal contraction is evident in the systematic shift of the peak positions. For the TH bulk sample, the peak at 2θ ≈ 3.9° is due to trace dihydrate.

between their single-crystal structure in space group C2 and experimental PXRD patterns.14 The agreement is equally good for our bulk TH samples generated either in desiccators or in situ on the diffractometer. (The structure of Burgess et al. is included in the CSD as refcode HEGFUW. The structure stored in the CSD has all site occupancy factors in the disordered region reset to unity. This is a consequence of the way in which disordered structures are stored in the CSD. In this case, it artificially amplifies the electron density in the Na+/ H2O region and produces a misleading PXRD pattern with a very intense peak at 2θ ≈ 4.5°.) To examine the possibility that there could be some differences between solution-grown TH (which yielded the single crystals) and TH generated directly from the solid sample, we examined also the PXRD for the crystals grown directly in a capillary. The resulting pattern was subject to severe preferred orientation, due to needle-like growth aligned along the capillary axis. Nonetheless, good agreement is observed with simulated data when a preferred orientation contribution is included (see Supporting Information), so we conclude that the single-crystal data are

Figure 7. (a) 13C CP/MAS SS-NMR spectra recorded at 100.6 MHz for TH, with the atom assignments indicated (labels refer to those in Scheme 1). (b) 23Na SS-NMR recorded at 105.8 MHz for TH. Both spectra are recorded at 313 K.

measured at 150 K, which could mean that some features of the SS-NMR spectra are subject to dynamic averaging. For the 13C CP/MAS spectrum, two distinct resonances are observed for C13, indicating two distinct chemical environments for the methyl group. This is associated with the two distinct orientations of the carboxyl group along the b axis. A slightly larger line width observed for C11 is also attributable to this feature. The single resonance for C12 is not affected by the carboxyl group orientation. The 23Na SS-NMR spectrum shows only one resonance (CQ = 840 kHz, ηQ = 0.4, and δiso = 0.3 ppm), indicating that all of the Na atoms in the Na+/H2O regions have identical chemical environments. This is 3670

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671

Crystal Growth & Design

Article

(6) Grieser, U. In Polymorphism in the Pharmaceutical Industry; Hilfiker, R., Ed.; Wiley-VCH: Weinheim, Germany, 2006. (7) Zhang, G. G. Z.; Law, D.; Schmitt, E. A.; Qiu, Y. Adv. Drug Delivery Rev. 2004, 56, 371. (8) Reutzel-Edens, S. M.; Bush, J. K.; Magee, P. A.; Stephenson, G. A.; Bryn, S. R. Cryst. Growth Des. 2003, 3, 897. (9) Roy, S.; Goud, N. J.; Babu, N. J.; Iqbal, J.; Kruthiventi, A. K.; Nangia, A. Cryst. Growth. Des. 2008, 8, 4243. (10) Malaj, L.; Censi, R.; Di Martino, P. Cryst. Growth Des. 2009, 9, 2128. (11) Raijada, D.; Bond, A. D.; Larsen, F. H.; Cornett, C.; Qu, H.; Rantanen, J. Pharm. Res. 2013, 30, 280. (12) Kim, Y.-S.; VanDerveer, D.; Rousseau, R. W.; Wilkinson, A. P. Acta Crystallogr. 2004, E60, m419. (13) Kim, Y. B.; Park, I. Y.; Lah, W. Y. Arch. Pharm. Res. 1990, 13, 166. (14) Burgess, K. M. N.; Perras, F. A.; Lebrun, A.; Messner-Herning, E.; Korobkov, I.; Bryce, D. L. J. Pharm. Sci. 2012, 101, 2930. (15) Di Martino, P.; Barthélémy, C.; Joiris, E.; Capsoni, D.; Masic, A.; Massorotti, V.; Gobetto, R.; Bini, M.; Martelli, S. J. Pharm. Sci. 2007, 96, 156. (16) Bond, A. D.; Boese, R.; Desiraju, G. R. Angew. Chem., Int. Ed. 2007, 46, 618. (17) Chen, E. J.; Welberry, T. R.; Heerdegen, A. P.; Goosens, D. J. Acta Crystallogr. 2010, B66, 696. (18) Bond, A. D. CrystEngComm 2012, 14, 2363. (19) Sheldrick, G. M. Acta Crystallogr. 2008, A64, 112. (20) Bennett, A. E.; Rienstra, C. M.; Auger, M.; Lakshmi, K. V.; Griffin, R. G. J. Chem. Phys. 1995, 103, 6951. (21) Peersen, O. B.; Wu, X. L.; Kustanovich, I.; Smith, S. O. J. Magn. Reson., Ser. A 1993, 104, 334. (22) Topspin 2.1, Bruker BioSpin GmbH, Rheinstetten, Germany, 2008. (23) MATLAB, The MathWorks Inc., Natick, MA, 2000. (24) Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. J.; Refson, K.; Payne, M. C. Z. Kristallogr. 2005, 220, 567. (25) Materials Studio 6.0, Accelrys Inc., San Diego, CA, 2011. (26) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (27) Grimme, S. J. Comput. Chem. 2006, 77, 3865. (28) van de Streek, J.; Neumann, M. A. Acta Crystallogr. 2010, B66, 544. (29) Allen, F. A. Acta Crystallogr. 2002, B58, 380. (30) Li, Z.; Liu, S.; Wu, J.; Li, G. Huaxue Yanjiu (Chin.) 2008, 19, 51 (CSD, LULSAN). (31) Braga, D.; Grepioni, F.; Tedesco, E. Organometallics 1998, 17, 2669. (32) For example, Dinoi, C.; Taban, G.; Sozen, P.; Demirhan, F.; Daran, J.-C.; Poli, R. J. Organomet. Chem. 2007, 692, 3743 (CSD, XIJFOM). (33) For example, Lennartson, A.; Håkansson, M. Acta Crystallogr. 2008, C64, m13 (CSD, MISFUQ). (34) Macrae, C. F.; Bruno, I. J.; Chisholm, J. A.; Edgington, P. R.; McCabe, P.; Pidcock, E.; Rodriguez-Monge, L.; Taylor, R.; van de Streek, J.; Wood, P. A. J. Appl. Crystallogr. 2008, 41, 466.

representative of the bulk TH sample prepared by any of the routes that we have applied.

4. CONCLUSIONS The crystal structure previously reported to be sodium naproxen heminonahydrate can be reinterpreted as a tetrahydrate, represented as {Na2(H2O)7}2+(naproxen−)2(H2O). The tetrahydrate formula is consistent with other data established for the sodium naproxen system. The best conventional description of the structure is in space group P1 with unit cell parameters a ≈ 20.6, b ≈ 11.1, c ≈ 7.1 Å, α = 90, β = 98.0, γ = 97.7°. The proposed structure is chemically plausible and consistent with the complete single-crystal X-ray diffraction pattern, including diffuse rods that reveal consistent two-dimensional sections in the structure, disordered along the stacking direction. The previously published structure in space group C2 (derived only from the Bragg reflections) represents an average of the P1 structure in two orientations related by a 2-fold rotation around the b axis. The description in terms of stacking disorder is equivalent to a description of multiple twinning by application of this 2-fold rotation. This example adds to the growing number of pharmaceutical compounds that are established to exhibit crystal structures more complex than the average representation. It is hoped that further investigation and better understanding of such structures will lead to more effective solid-state structure−property correlations.

■

ASSOCIATED CONTENT

S Supporting Information *

Electronic structure files (CIF and SHELX format), details of the CSD search, details of the DFT-D structure minimization, additional PXRD data, and additional reconstructed precession images. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank the Danish Natural Sciences Research Council for provision of the X-ray equipment. The Lundbeck Foundation (Grant Numbers 479/06, R31-A2630, R49-A5604) and Department of Pharmacy, University of Copenhagen, are acknowledged for financial support. Support from the Danish Council for Independent Research (Technology and Production Sciences, Project Number 09-066411) is also acknowledged.

■

REFERENCES

(1) Singhal, D.; Curatolo, W. Adv. Drug Delivery Rev. 2004, 56, 335. (2) Gardner, C. R.; Walsh, C. T.; Almarsson, Ö . Nat. Rev. Drug Discovery 2004, 3, 926. (3) Datta, S.; Grant, D. J. W. Nat. Rev. Drug Discovery 2004, 3, 42. (4) International Conference on Harmonization guidance Q6A specifications. Test procedures and acceptance criteria for new drug substances and new drug products − chemical substances. U.S. Food and Drug Administration: Rockville, 1999. (5) Guidance for Industry. ANDAs. Pharmaceutical solid polymorphism. Chemistry, manufacturing and controls information. U.S. Food and Drug Administration: Rockville, 2007. 3671

dx.doi.org/10.1021/cg400688c | Cryst. Growth Des. 2013, 13, 3665−3671