Laurie
Bodart
*a,
Maria
Prinzo
b,
Amélie
Derlet
a,
Nikolay
Tumanov
a and
Johan
Wouters
*a
aUniversity of Namur (UNamur)Namur Medicine and Drug Innovation Center – Namur Research Institute for LIfe Science (NAMEDIC-NARILIS), Namur Institute of Structured Matter (NISM), Department of Chemistry, University of Namur (UNamur), 61 Rue de Bruxelles, 5000 Namur, Belgium. E-mail: laurie.bodart@unamur.be; johan.wouters@unamur.be
bDrug Science Department, University of Catania, Viale Andrea Doria 6, 95125 Catania, Italy
First published on 17th November 2020
Non-steroidal anti-inflammatory drugs, such as diclofenac, are gaining attention as repurposed compounds for the treatment of multi-drug resistant tuberculosis. In this study, salts combining diclofenac with clofazimine, are prepared by solvent crystallization and by liquid-assisted grinding. Diclofenac anion possesses an H-bond acceptor which can strongly interact with protic solvent molecules. In this context, selected solvents (protic, aprotic and solvents with increasing molecular volume) are screened in order to investigate solvent impact on crystallization of solvated or unsolvated salt of clofazimine with diclofenac in 1:
1 ratio. Five solvated salts and one unsolvated salt were successfully crystallized. The ability of the diclofenac anion to interact with a protic molecule is also exploited in order to crystallize a cocrystal of salt with drug
:
drug ratio different from 1
:
1. Structures of two solvated cocrystal of salts (with acetonitrile and ethylacetate) and two polymorphs of an unsolvated cocrystal of salt combining clofazimine with diclofenac in 1
:
2 ratios are determined.
Several factors have been investigated for their impact on solvate formation. The two most important ones have been identified as solvent–solute affinity and molecular size and shape of the solvent.8,9 The first parameter, among other factors, depends on hydrogen bonding and aromatic interaction abilities of the molecule. The second parameter is in correlation with the concept of packing efficiency, since solvent presenting low affinity for a compound can still be incorporated into its structure if it allows a better packing.8,10 While various solvents, depending on their molecular volume, size and shape, may lead to different crystal packings, it is also possible that bulky molecules of the compound itself fail to form efficient packing and thus may incorporate solvent molecules in order to improve it. The presence of hydrogen bonding groups tends to promote inclusion of polar solvents by strong and specific interactions.11 Despite identification of certain factors contributing to solvate formation, theoretical solvate prediction is challenging and research in this area is still ongoing.9,10
In this study, we focus on the solvated salts of a drug–drug system comprising clofazimine and diclofenac. Clofazimine (CFZ), exhibits antimycobacterial and anti-inflammatory properties and has been recently reevaluated as a potential treatment for multidrug-resistant tuberculosis.12–17 Diclofenac (DCF) belongs to the class of non-steroidal anti-inflammatory drugs (NSAIDs) that has been recently proposed as a host-directed therapy in the treatment of tuberculosis owing to its antitubercular properties.18–22 Several studies showing the ability of clofazimine (pKa: 9.29 (ref. 23)) to form salts with organic and inorganic acids24–28 point that combining CFZ with DCF in a drug–drug salt should be achievable. Several solvates were previously reported for clofaziminium as well as diclofenac salts26,28 and strong H-bond interaction between DCF and solvent molecules have been reported.29 Solvate formation is thus also expected with protic solvents during the preparation the drug–drug CFZ–DCF salts.
In our experiments, as expected, diclofenac tended to bind protic solvents and incorporate them into structure, thereby resulting in the solvated form of the CFZ–DCF 1:
1 salt. Our further hypothesis was that diclofenac molecule, being also protic, should be able to compete with the solvent molecule for that binding site and thus instead of a solvated 1
:
1 form lead to a 1
:
2 CFZ–DCF cocrystal of salt, with diclofenac molecule being connected to diclofenac anion. To verify this and also to see whether an unsolvated 1
:
1 salt can also be obtained, we have selected a number of aprotic solvents with increasing molecular volume, which due to their chemical nature should have a lower probability of binding to diclofenac. Experimenting with various solvents, we obtained in total five solvated 1
:
1 forms, two solvated 1
:
2 forms, one unsolvated 1
:
1 CFZ–DCF salt and two unsolvated 1
:
2 cocrystal of salt (Fig. 1), thus proving that selecting a proper solvent is essential in crystallization processes and might be key to obtaining desired structures.
Given the low vapor pressure at room temperature (20–25 °C) of propiophenone and salicylaldehyde and the quite good solubility of CFZ and DCF in these solvents, crystallization by solvent evaporation is quite difficult. For this reason, the 1:
1 powder mixture of CFZ and DCF was not fully dissolved in these solvent (i.e. an excess powder was placed in the solvent and left at room temperature for crystallization) to give CFZNH+–DCF−–PPP (2
:
2
:
5) and CFZNH+–DCF−–SAL–H2O (4
:
4
:
9
:
1).
Variable-temperature PXRD (VT-PXRD) experiments were conducted on the same diffractometer equipped with an Anton-Paar TTK 450 system. Data were collected at 25 °C and then from 30 °C to 140 °C with data collection every 10 °C.
CFZNH
+
–DCF
−
–MeOH (1![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–EtOH (1![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–EtOH (1![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–MeCN–H
2
O (1![]() ![]() ![]() ![]() ![]() ![]() |
|
---|---|---|---|---|
Chemical formula | C27H23Cl2N4·C14H10Cl2NO2·0.741 (CH4O) | C27H23Cl2N4·C14H10Cl2NO2·C2H6O | C27H23Cl2N4·C14H10Cl2NO2·C2H6O | C27H23Cl2N4·C14H10Cl2NO2·C2H3N·2(H2O) |
Mr | 793.42 | 815.59 | 815.59 | 846.61 |
Crystal system, space group | Monoclinic, C2/c | Triclinic, P![]() |
Triclinic, P![]() |
Triclinic, P![]() |
Temperature (K) | 295(2) | 295(2) | 100(2) | 100(2) |
a, b, c (Å) | 16.7209(3), 17.2504(5), 27.8319(7) | 12.5818(4), 13.4486(3), 13.5318(3) | 12.5325(5), 13.1139(3), 13.1773(4) | 11.6515(3), 11.7149(3), 16.0793(4) |
α, β, γ (°) | 90, 93.438(2), 90 | 80.291(2), 71.638(2), 71.710(2) | 82.032(2), 71.833(3), 74.538(3) | 103.558(2), 90.461(2), 105.250(2) |
V (Å3) | 8013.4(3) | 2057.12(10) | 1979.61(12) | 2052.78(9) |
Z | 8 | 2 | 2 | 2 |
No. of measured, independent and observed [I > 2σ(I)] reflections | 23![]() |
27![]() |
21![]() |
18![]() |
R int | 0.040 | 0.024 | 0.046 | 0.019 |
R[F2 > 2σ(F2)], wR(F2), S | 0.059, 0.193, 1.02 | 0.050, 0.146, 1.06 | 0.057, 0.166, 1.07 | 0.031, 0.080, 1.03 |
CCDC number | 2032488 | 2032489 | 2032490 | 2032491 |
CFZNH
+
–DCF
−
–SAL–H
2
O (4![]() ![]() ![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–PPP (2![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–PPP (2![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
(1![]() ![]() |
|
---|---|---|---|---|
Chemical formula | 4(C27H23Cl2N4)·4(C14H10Cl2NO2)·9(C7H6O2)·H2O | 2(C27H23Cl2N4)·2(C14H10Cl2NO2)·5(C9H10O) | 2(C27H23Cl2N4)·2(C14H10Cl2NO2)·5(C9H10O) | (C27H23Cl2N4)·(C14H10Cl2NO2) |
Mr | 4195.16 | 2209.89 | 2209.89 | 769.52 |
Crystal system, space group | Triclinic, P![]() |
Triclinic, P![]() |
Triclinic, P![]() |
Triclinic, P![]() |
Temperature (K) | 100(2) | 295(2) | 100(2) | 295(2) |
a, b, c (Å) | 10.4132(5), 21.2266(7), 25.0657(7) | 10.7319(4), 13.6005(6), 21.5496(9) | 10.5086(4), 13.4783(5), 21.4224(8) | 12.4461(6), 13.2751(5), 13.3667(6) |
α, β, γ (°) | 66.238(3), 89.969(3), 86.658(3) | 73.502(4), 77.404(4), 73.540(4) | 73.629(3), 76.913(3), 73.271(3) | 84.423(3), 67.963(4), 66.662(4) |
V (Å3) | 5060.5(3) | 2859.7(2) | 2752.92(19) | 1876.21(16) |
Z | 1 | 1 | 1 | 2 |
No. of measured, independent and observed [I > 2σ(I)] reflections | 53![]() ![]() ![]() |
28![]() ![]() |
28![]() |
17![]() |
R int | 0.069 | 0.051 | 0.038 | 0.037 |
R[F2 > 2σ(F2)], wR(F2), S | 0.060, 0.171, 1.03 | 0.052, 0.160, 1.02 | 0.035, 0.090, 1.03 | 0.048, 0.141, 1.03 |
CCDC number | 2032492 | 2032493 | 2032494 | 2032495 |
CFZNH
+
–DCF
−
–DCF (1![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–DCF–MeCN (1![]() ![]() ![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–DCF–EtOAc (1![]() ![]() ![]() ![]() ![]() ![]() |
CFZNH
+
–DCF
−
–DCF (1![]() ![]() ![]() ![]() |
|
---|---|---|---|---|
Chemical formula | C27H23Cl2N4·C14H10Cl2NO2·C14H11Cl2NO2 | C27H23Cl2N4·C14H10Cl2NO2·C14H11Cl2NO2·2(C2H3N) | C27H23Cl2N4·C14H10Cl2NO2·C14H11Cl2NO2·C4H8O2 | C27H23Cl2N4·C14H10Cl2NO2·C14H11Cl2NO2 |
Mr | 1065.66 | 1147.77 | 1153.76 | 1065.66 |
Crystal system, space group | Triclinic, P![]() |
Triclinic, P![]() |
Triclinic, P![]() |
Triclinic, P![]() |
Temperature (K) | 295(2) | 295(2) | 295(2) | 295(2) |
a, b, c (Å) | 10.8837(2), 14.4737(3), 17.7519(4) | 15.0694(11), 15.2735(10), 15.5991(13) | 15.0569(4), 15.3773(4), 15.4247(4) | 11.9326(3), 15.2549(4), 16.0042(4) |
α, β, γ (°) | 101.568(2), 105.173(2), 94.286(2) | 74.882(7), 61.685(8), 64.227(7) | 76.043(2), 61.975(3), 63.541(3) | 89.161(2), 70.570(2), 69.693(2) |
V (Å3) | 2620.06(10) | 2839.6(4) | 2820.07(16) | 2559.79(12) |
Z | 2 | 2 | 2 | 2 |
No. of measured, independent and observed [I > 2σ(I)] reflections | 26![]() |
27![]() ![]() |
27![]() |
25![]() |
R int | 0.022 | 0.032 | 0.022 | 0.025 |
R[F2 > 2σ(F2)], wR(F2), S | 0.050, 0.141, 1.05 | 0.050, 0.159, 1.08 | 0.042, 0.122, 1.05 | 0.045, 0.129, 1.04 |
CCDC number | 2032496 | 2032497 | 2032498 | 2032499 |
These dimers are further stacked in a head-to-tail fashion (Fig. 3(b), centroid–centroid distance of 3.623(2) Å, orthogonal projection distance of 3.378(1) Å and horizontal displacement of 1.311 Å). This salt can also be prepared by liquid-assisted grinding of CFZ with DCF (in 1:
1 molar ratio) and with MeOH as solvent (Fig. S3(a)†). TG/DSC analysis of the powder indicates a weight loss of 2.6% between 30 and 150 °C (calculated MeOH content: 2.99%) (Fig. S4†). The powder of CFZNH+–DCF−–MeOH (1
:
1
:
0.74) melts at 118 °C (as confirmed by an analysis performed on a Koffler apparatus). Complete desolvation is difficult to achieve before melting of the powder (at 100 °C, the solvate is still present) and the crystalline phase obtained upon heating of CFZNH+–DCF−–MeOH (1
:
1
:
0.74) at 110 °C could not be identified (Fig. S5†).
Compound | Transition onset (°C) | Melting onset (°C) |
---|---|---|
a Value obtained from desolvated powder of CFZNH+–DCF−–DCF–MeCN (1![]() ![]() ![]() ![]() ![]() ![]() |
||
CFZ | — | 218 |
DCF | — | 177 |
CFZNH
+
–DCF
−
–EtOH (1![]() ![]() ![]() ![]() |
Desolvation at melting | 113 |
CFZNH
+
–DCF
−
(1![]() ![]() |
— | 186 |
CFZNH
+
–DCF
−
–DCF (1![]() ![]() ![]() ![]() |
— | 171 |
CFZNH
+
–DCF
−
–DCF–MeCN (1![]() ![]() ![]() ![]() ![]() ![]() |
93 | 157 |
CFZNH
+
–DCF
−
–DCF–EtOAc (1![]() ![]() ![]() ![]() ![]() ![]() |
93 | 157 |
CFZNH
+
–DCF
−
–DCF (1![]() ![]() ![]() ![]() |
— | 157a |
Structure | Packing coefficient | Angle between N4–C9–C8–N3 and O1–C28–O2–C29 planes (°) |
---|---|---|
a Angle between planes passing through N8–C49–C50–N9 atoms of second clofaziminium cation and O3–C69–O4–C70 atoms of second diclofenac anion. | ||
CFZNH
+
–DCF
−
–MeOH (1![]() ![]() ![]() ![]() |
0.661 | 83.0 |
CFZNH
+
–DCF
−
–EtOH (1![]() ![]() ![]() ![]() |
0.658 | 73.3 |
CFZNH
+
–DCF
−
–EtOH (1![]() ![]() ![]() ![]() |
0.675 | 72.2 |
CFZNH
+
–DCF
−
–MeCN–H
2
O (1![]() ![]() ![]() ![]() ![]() ![]() |
0.677 | 64.9 |
CFZNH
+
–DCF
−
–SAL–H
2
O (4![]() ![]() ![]() ![]() ![]() ![]() |
0.700 | 74.5; 83.2a |
CFZNH
+
–DCF
−
–PPP (2![]() ![]() ![]() ![]() |
0.698 | 81.3 |
CFZNH
+
–DCF
−
–PPP (2![]() ![]() ![]() ![]() |
0.696 | 80.9 |
CFZNH
+
–DCF
−
(1![]() ![]() |
0.653 | 14.3 |
CFZNH
+
–DCF
−
–DCF (1![]() ![]() ![]() ![]() |
0.642 | 66.8 |
CFZNH
+
–DCF
−
–DCF–MeCN (1![]() ![]() ![]() ![]() ![]() ![]() |
0.665 | 63.4 |
CFZNH
+
–DCF
−
–DCF–EtOAc (1![]() ![]() ![]() ![]() ![]() ![]() |
0.657 | 62.5 |
CFZNH
+
–DCF
−
–DCF (1![]() ![]() ![]() ![]() |
0.660 | 46.4 |
Melting point of this unsolvated salt was determined as 186 °C by DSC analysis (Fig. S4† and Table 2). Interestingly, this unsolvated salt has a smaller packing coefficient than the solvated 1:
1 salts (Table 3). This is in accordance with the presence of voids (15.41 Å3, 0.8% of the unit cell volume). This lower packing efficiency could explain solvent insertion (and so solvate crystallization) when diclofenac anion–solvent or clofaziminium cation–solvent interactions are possible.
To avoid presence of toxic solvent, several options can be considered. The most evident ones, would be salt preparation without solvent (neat grinding) or with non-toxic/pharmaceutically accepted solvents (water, ethanol40) or desolvation of solvated salts. Grinding clofazimine with diclofenac in 1:
1 molar ratio in absence of solvent as well as in presence of water (LAG H2O) resulted in a physical mixture of clofazimine with diclofenac (Fig. S3(h)†), thus not yielding any reaction. Moreover, crystallization experiment in only water as solvent was unsuccessful because of the low aqueous solubility of clofazimine. Since water as a solvent turned out to be unsuccessful and complete desolvation of CFZNH+–DCF−–MeOH (1
:
1
:
0.74) is difficult to achieve before melting of the powder, another pharmaceutically accepted solvent, namely ethanol, was chosen. In this case, solvent crystallization experiment as well as liquid-assisted ball milling, led to the same solvated salt: CFZNH+–DCF−–EtOH (1
:
1
:
1), in which, as expected, ethanol interacted with DCF− through H-bond.
To further investigate the parameters influencing the crystallization of (un)solvated salts, another protic solvent of higher molecular volume, salicylaldehyde (111 Å3 (ref. 38)) was selected. Despite their high molecular volume, salicylaldehyde molecules are included in the structure of CFZNH+–DCF−–SAL–H2O (4:
4
:
9
:
1). Interestingly more than one salicylaldehyde molecule as well as one water molecule have been accommodated in the structure probably to allow a good packing (this structure turned to have the best packing coefficient, Table 3). The existence of this solvate proves that the ability of solvent to interact with solute via weak interactions may be one of the factors explaining why even bulky solvents can still be trapped in the crystal.
High molecular volume did not hamper solvent insertion in the structure, at least for protic solvents that are able to interact with the solute through H-bond. We then focused on aprotic solvents. First, we selected acetonitrile (molecular volume of 46 Å3 (ref. 38)). However crystals obtained in these conditions corresponded to CFZNH+–DCF−–MeCN–H2O (1:
1
:
1
:
2), a solvated hydrated salt, unstable at room temperature. Three other aprotic solvents (N,N-dimethylformamide, 78 Å3, ethylacetate, 91 Å3 and propiophenone, 136 Å3 (ref. 38)) with larger molecular volumes were then selected to perform crystallization experiments of CFZ and DCF in 1
:
1 molar ratio. A larger molecular volume could potentially hamper solvent inclusion in the structure. However, we still observed solvate formation: crystals obtained from DMF corresponded to the known clofazimine solvate CFZ–DMF (CSD refcode: CEKTER41); crystallization experiments performed in propiophenone led to a solvated salt, CFZNH+–DCF−–PPP (2
:
2
:
5), despite the absence of H-bond donor on propiophenone. The formation of the propiophenone solvate highlights again the fact that the molecular volume is not the only parameter affecting solvate formation, and that the interactions between the solvent and the solute are of crucial importance as well. In this case, propiophenone interacts with clofaziminium through π–cation interaction. Very interestingly, crystals of CFZNH+–DCF−(1
:
1) unsolvated salt could however be obtained in ethylacetate, an aprotic molecule with a smaller molecular volume than propiophenone. The absence of the aromatic ring in ethylacetate, in contrast to propiophenone, results in a smaller number of potential interactions that it can form with clofazimine and thus may explain why ethylacetate is not incorporated into the structure. This unsolvated salt does not correspond to the crystalline phase appearing upon desolvation of CFZNH+–DCF−–MeOH (1
:
1
:
0.74), which could indicate the existence of another polymorph of CFZNH+–DCF−(1
:
1).
All previously described interactions stabilize CFZNH+–DCF−–DCF assemblies. Two types of dimers of these assemblies, which are respectively stabilized by C31–H31⋯O3 and C3–H3A⋯O3 weak H-bonds (Table S2 † and Fig. 9(b) and (c)), are also observed in the structure of CFZNH+–DCF−–DCF (1:
1
:
1) cocrystal of salt. π–π stacking interactions are also observed between clofaziminium cations. The first one occurs between the phenazine core of two CFZNH+, with a centroid–centroid distance of 3.669(1) Å, an orthogonal projection distance of 3.413(1) Å and an horizontal offset of 1.345 Å (planes and centroid calculated using C5 C6 C11 and C12 atoms, Fig. 9(e)). This stacking is further stabilized by C3–H3A⋯O3 (Fig. 9(c)) weak H-bonds. A second π–π stacking interaction is also observed between two C19 C20 C21 C22 C23 C24 aromatic rings of clofaziminium with a centroid–centroid distance of 3.801(1) Å, an orthogonal projection distance of 3.357(1) Å and an horizontal offset of 1.781 Å (Fig. 9(f)). Finally a C–H⋯π interaction is observed between C18–H18 of clofaziminium and diclofenac molecule (C44 C45 C46 C47 C48 C49) with a H⋯centroid distance of 2.59 Å, a C⋯centroid distance of 3.443(3) Å and a C–H⋯centroid angle of 153° (Fig. 9(g)).
The powder pattern calculated from SCXRD data corresponds to the one of the batch powder prepared by LAG EtOH (Fig. S3(d)†). This cocrystal of salt melts at 171 °C (Fig. S4† and Table 2).
TG/DSC analysis of CFZNH+–DCF−–DCF–MeCN (1 : 1 : 1 : 2) (powder obtained by liquid-assisted grinding experiment) reveals a 6.2% weight loss between 50 and 140 °C on the TG curve (Fig. S4†) which is associated with an endothermic event on the DSC curve (onset: 93 °C) (Fig. S4† and Table 2). These events can be attributed to desolvation (calculated MeCN content of 7.2% in CFZNH+–DCF−–DCF–MeCN (1:
1
:
1
:
2)). The endothermic event observed at 157 °C corresponds to the melting of the desolvated phase and suggests the formation of either an eutectic mixture or of an unsolvated crystalline phase combining CFZ and DCF (Fig. S4† and Table 2). A variable-temperature PXRD experiment performed on CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2) confirmed the formation of a new crystalline phase upon heating which does not correspond to CFZNH+–DCF−–DCF (1
:
1
:
1) polymorph I (Fig. S6†). Results from variable-temperature PXRD experiment and DSC/TG analyses suggest the formation of a second polymorph of CFZNH+–DCF−–DCF (1
:
1
:
1).
Despite a quite good overlay of the molecules constituting their asymmetric unit (Fig. 11(e)), CFZNH+–DCF−–DCF–MeCN (1:
1
:
1
:
2) and CFZNH+–DCF−–DCF (1
:
1
:
1) are not isostructural as revealed by the crystal packing comparison performed with Mercury (only 9 molecules over 15 overlay despite distance and angle tolerance of 30% and 30°). Diclofenac anion orientation is slightly modified in the unsolvated cocrystal of salt in comparison to the one observed in CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2). Indeed, the angle between the planes passing through N4–C9–C8–N3 atoms of CFZNH+ and through O1–C28–O2–C29 of DCF− is different in the two structures (63.44° in CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2)vs. 46.41° in CFZNH+–DCF−–DCF (1
:
1
:
1)).
VT-PXRD and PXRD data measured after heating CFZNH+–DCF−–DCF–MeCN (1:
1
:
1
:
2) and CFZNH+–DCF−–DCF–EtOAc (1
:
1
:
1
:
1) indicate a phase transformation from the solvated cocrystal of salts to the CFZNH+–DCF−–DCF (1
:
1
:
1) unsolvated cocrystal of salt (polymorph II) (Fig. S3(g) and S6†). TG/DSC data obtained from CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2) indicate that CFZNH+–DCF−–DCF (1
:
1
:
1) polymorph II melts at 157 °C (Fig. S4† and Table 2). The difference in melting point observed between polymorph I (171 °C) and polymorph II (157 °C), can be partially explained by the fact that there are more π interactions in polymorph I while intermolecular H-bonds are of comparable strength in both polymorphs. It is interesting to notice that polymorph II has however a better packing (higher density) than polymorph I (packing coefficient of 0.660 vs. 0.642, density of 1.383 g cm−3vs. 1.351 g cm−3 (data collected at 295 K for both structures)) (Table 3).
TG/DSC analysis of these two solvated cocrystal of salts (CFZNH+–DCF−–DCF–MeCN (1:
1
:
1
:
2), CFZNH+–DCF−–DCF–EtOAc (1
:
1
:
1
:
1)) revealed a weight loss between 50 and 100 °C and a melting point around 157 °C (Fig. S4†). A phase transformation, probably associated with desolvation (and formation of an unsolvated cocrystal of salt combining CFZ and DCF), was identified by PXRD experiments (Fig. S6 and S3(g)†). The powder diffraction pattern obtained after desolvation of CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2) or CFZNH+–DCF−–DCF–EtOAc (1
:
1
:
1
:
1) does not correspond to the one of CFZNH+–DCF−–DCF (1
:
1
:
1) (polymorph I) (Fig. S6 and S3(g)†), which could be the indication of the existence of a second polymorph of CFZNH+–DCF−–DCF (1
:
1
:
1). To verify this hypothesis, we have tried to obtain single crystals of this new unsolvated crystalline phase using other solvents that cannot be incorporated into the structure. We selected short chain polymers such as PEG (average MW: 200 g mol−1 and 400 g mol−1) and polycaprolactone triol (average MW: 300 g mol−1) as recrystallization solvents because of their higher molecular weight and also because of their molecular ‘flexibility’ in comparison to salicylaldehyde and propiophenone which previously led to solvated salts. Few single crystals of CFZNH+–DCF−–DCF (1
:
1
:
1) (polymorph II) cocrystal of salt were obtained in PEG (average MW: 200 g mol−1). Comparison of the calculated powder pattern of CFZNH+–DCF−–DCF (1
:
1
:
1) (polymorph II) with the one obtained after variable-temperature PXRD experiment performed on CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2) confirmed desolvation and phase transformation that were inferred from VT-PXRD and TG/DSC experiments.
When facing probable solvate formation, the potential solute–solvent interactions can be more relevant than the size of the solvent. For example, the unsolvated CFZNH+–DCF−(1:
1) salt crystallized only in ethylacetate, an aprotic solvent having a lower molecular volume than another selected solvent, propiophenone. Despite its high molecular volume, propiophenone was incorporated into the structure owing to π interactions, resulting in the corresponding solvated salt. A particular feature of the CFZNH+–DCF−(1
:
1) unsolvated structure is that the carboxylate of DCF− is almost coplanar to CFZNH+. It is, to the best of our knowledge, the first time that this specific carboxylate orientation is reported in a clofaziminium salt. These results highlight the importance of considering potential solute–solvent interactions not only in terms of H-bond interactions, but also other weak interactions such as probable π interactions.
Changing the clofazimine to diclofenac ratio from 1:
1 to 1
:
2 allowed an extra diclofenac molecule to compete for the binding site with the protic solvent molecules. More particularly, diclofenac anion successfully beated EtOH for the H-bond interaction on diclofenac anion resulting in an unsolvated cocrystal of salt (CFZNH+–DCF−–DCF (1
:
1
:
1) polymorph I) with the 1
:
2 clofazimine to diclofenac molar ratio. The second polymorph of CFZNH+–DCF−–DCF (1
:
1
:
1) emerged from desolvation of CFZNH+–DCF−–DCF–MeCN (1
:
1
:
1
:
2) and CFZNH+–DCF−–DCF–EtOAc (1
:
1
:
1
:
1). The latter underlines that undesired solvated forms can potentially give access to other desired structures. Recrystallization using unconventional solvents such as short chain polymers exhibiting good fluidity, high molecular flexibility and high molecular weight can also be an option to avoid solvent inclusion into the structure, as illustrated by the successful growth of CFZNH+–DCF−–DCF (1
:
1
:
1) (polymorph II) in PEG 200.
Changing the drug:
drug ratio in the structure expectedly affects the physico-chemical properties of the corresponding solid forms, for instance, the melting point of CFZNH+–DCF−(1
:
1) is 186 °C whereas those of CFZNH+–DCF−–DCF (1
:
1
:
1) polymorphs I and II are 171 °C and 157 °C respectively.
Footnote |
† Electronic supplementary information (ESI) available: Extended table of experimental details for the described structures, table of H-bond parameters, ellipsoids plots of the described structures, powder diffraction patterns of the salts prepared by liquid-assisted grinding, TG/DSC data and variable-temperature powder X-ray diffraction data. CCDC 2032488–2032499. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0ce01400a |
This journal is © The Royal Society of Chemistry 2021 |