The inter­play of solvation, mol­ecular conformation and supra­molecular assembly in 1,1′-({[(ethane-1,2-diyl)dioxy](1,2-phenyl­ene)}bis­(methanylyl­idene))bis­(thio­semicarbazide) and its N,N-di­methyl­formamide disolvate

Thio­semicarbazones are a class of small molecules that have been evaluated over a number of years as anti­cancer agents (Finch et al., 1999) and as anti­tubercular agents (Shucla et al., 1984; Desai et al., 1984), as well as for their parasiticidal action (Wilson et al., 1974; Du et al., 2002; Greenbaum et al., 2004) against Plasmodium falciparum and Trypanasoma cruzi which are the causative agents of malaria and Chagas' disease, respectively. On the other hand, bis­(thio­semicarbazone) derivatives have been used as ligands in metal complexes designed to act as site-specific radiopharmaceuticals (Holland et al., 2007). This very wide diversity of application has thus prompted the present study in which we report the synthesis and the molecular and supra­molecular structures of two bis­(thio­semicarbazones), namely 1,1'-({[(ethane-1,2-diyl)di­oxy](1,2-phenyl­ene)}bis­(methanylyl­idene))bis­(thio­semicarbazide), (I) (Fig. 1), and its N,N-di­methyl­formamide disolvate, (II) (Fig. 2).

For the synthesis of compound (I), a mixture of 2,2'-[ethane-1,2-diylbis(­oxy)]dibenzaldehyde (1 mmol) and thio­semicarbazide (2 mmol) in ethanol (15 ml) containing a catalytic qu­antity of acetic acid was heated under reflux for 4 h. The mixture was cooled to ambient temperature and then poured into ice/water. The resulting solid product, (I), was collected by filtration, washed with water and dried in air to give colourless crystals suitable for single-crystal X-ray diffraction. ¹³C NMR (δ, p.p.m.): 67.34 (CH₂), 112.91, 120.97, 122.65, 126.21, 131.27, 138.10, 157.04 (aryl and alkenyl), 178.77 (C═S). In an alternative synthesis, a similar mixture, in methanol rather than ethanol, was subjected to microwave irradiation at 80 W for 20 min, after which the product (I) was isolated as before. Recrystallization by slow evaporation, at ambient temperature and in the presence of air, of a solution in N,N-di­methyl­formamide gave the bis­(solvate), (II).

Crystal data, data collection and structure refinement details are summarized in Table 1. It was apparent from an early stage in the refinement of (II) that one of the N,N-di­methyl­formamide molecules, that containing atom N41 (Fig. 2), was disordered over two sets of atomic sites having unequal occupancies. For the minor disorder component, the bonded distances and the one-angle non-bonded distances were restrained to be identical to the corresponding distances in the major disorder component, subject to uncertainties of 0.005 Å and 0.01 Å, respectively. In addition, the anisotropic displacement parameters for the atom sites N41 and N51 were constrained to be identical, as were those for the sites O41 and O51. Under these conditions, the refined site occupancies were 0.839 (3) and 0.161 (3). The H atoms, apart from those in the disordered solvent molecule, were all located in difference maps and then treated as riding atoms in geometrically idealized positions, with C—H = 0.95 Å (alkenyl, aromatic and formyl), 0.98 (CH₃) or 0.99 Å (CH₂) and N—H = 0.88 Å, and with U_iso(H) = kU_eq(carrier) where k = 1.5 for the methyl groups, which were permitted to rotate but not to tilt, and 1.2 for all other H atoms. The H atoms of the disordered solvent molecule were included in calculated positions in a similar manner. For (II), the bad outlier reflection (−5, 3, 4) was omitted from the final refinements. Also for (II) there was a large value of K, 3.198, for the group of very weak reflections having F_c/F_c(max) in the range 0.000 < F_c/F_c(max) < 0.005.

In the unsolvated form, (I), the molecules lie across a twofold rotation axis in space group C2/c, and the reference molecule was selected as one lying across the axis along (1/2, y, 1/4). The molecular components of the solvated form, (II), all lie in general positions and one of the two molecules of N,N-di­methyl­formamide is disordered over two sets of atomic sites having occupancies 0.839 (3) and 0.161 (3). For the following discussion, it is convenient to denote the molecule of (I) and that of the principal component in (II) in the form ArOCH₂CH₂OAr.

For (I), the 14 non-H atoms forming the asymmetric unit (Fig. 1) are very nearly co-planar; the r.m.s. deviation from the mean plane through these atoms is 0.0508 Å and the maximum deviation from this plane, for atom C1, is 0.0891 (10) Å. In the analogous component of (II), the non-H atoms in the fragment from C1 to N13 and S1 have an r.m.s. deviation from their mean plane of 0.0631 Å with a maximum deviation, for atom N13, of 0.1643 (10) Å; the corresponding values for the fragment from C2 to N23 and S2 are 0.0467 Å and 0.1106 (7) Å, for atom S2. In both compounds the two ArO fragments adopt a synplanar conformation relative to the central C—C bond, as confirmed by the O—C—C—O torsion angles (Tables 2 and 4). Although the molecule of (I), and the corresponding component in (II), could both, in principle, lie across centres of inversion with maximum molecular symmetry 2/m (C_2h), neither in fact does so, with just a twofold rotation axis present in (I) and no inter­nal symmetry in (II).

Despite the conformational similarities noted above, there is one significant conformational difference between the two compounds, namely the orientation of the thio­semicarbazide fragments relative to the adjacent aryl rings, which are trans-planar in (I) but cis-planar in (II), as indicated by the relevant C—C—C—N torsion angles (Tables 2 and 3) which are close to zero in (I) but almost 180° in (II). The bond distances in (I) and (II) show no significant deviations from the normal values (Allen et al., 1987), but we note that the inter-bond angles at atoms O1 and O2 (Tables 2 and 3) are significantly larger than the usual values for such C—O—C angles.

In each of (I) and (II) there are short N—H···N contacts (Tables 4 and 5) within the thio­semicarbazide units, but these have very small N—H···N angles and so must be regarded as adventitious contacts rather than structurally significant hydrogen bonds. A combination of two independent N—H···S hydrogen bonds (Table 4) links the molecules of (I) into complex sheets, but the formation of these sheets can readily be analysed in terms of two simple sub-structures (Ferguson et al., 1998a,b; Gregson et al., 2000). In the first of these two sub-structures, the N—H···S hydrogen bond having atom N2 as the donor links molecules lying across the twofold rotation axes along (1/2, y, n + 1/4), where n represents an integer, to form a chain of rings running parallel to the [001] direction and consisting of R²₂(8) (Bernstein et al., 1995) rings, which lie across the twofold axes along (1/2, y, n − 1/4), where n again represents an integer (Fig. 3). In the second sub-structure, the N—H···S hydrogen bond having atom N3 as the donor links molecules lying across the twofold rotation axes along (-n/2 + 1/2, y, n/2 + 1/4) into a chain of centrosymmetric R²₂(8) rings running parallel to the [101] direction, in which the hydrogen-bonded rings are centred at (-n/2 + 3/4, 1/4, n/2), where n represents an integer in each case (Fig. 4). The combination of these two chain-of-rings motifs gives rise to a sheet lying parallel to (010), in the domain 0 < y < 1/2, which contains R²₄(38) rings as well as two types of R²₂(8) rings (Fig. 5). A second sheet of this type, related to the first by inversion, lies in the domain 0.5 < y < 1.0, but there are no significant direction-specific inter­actions between adjacent sheets.

The solvent components of (II) are linked to the ArOCH₂CH₂OAr component by two independent N—H···O hydrogen bonds (Table 5, Fig. 2), but neither of the N—H bonds involved in these inter­actions, nor the solvent molecules themselves, play any further direct role in the supra­molecular assembly. The ArOCH₂CH₂OAr components of (II) are linked by two independent three-centre N—H···(S)₂ hydrogen bonds (Table 5), both of which are effectively planar, to form complex sheets which can, like those in (I), be readily analysed in terms of two simpler sub-structures. In one sub-structure, the ArOCH₂CH₂OAr components which are related by translation along the [001] direction are linked by two N—H···S hydrogen bonds to form a C(19)C(19)[R²₂(8)] chain of rings. In the second sub-structure, molecules related by inversion are again linked by two independent N—H···S hydrogen bonds to form a chain of edge-fused R²₂(38) rings running parallel to the [100] direction, in which the rings involving atoms of type N13 as the hydrogen-bond donors are centred at (n, 1/2, 1/2), while those involving atoms of type N23 as the donors are centred at (n + 1/2, 1/2, 1/2), where n represents an integer in each case. The combination of these two motifs generates a sheet lying parallel to (010) and containing four types of ring, namely R²₂(8), R²₄(8) and two types of R²₂(38) rings, all of which, apart from the R²₂(8) rings, are centrosymmetric (Fig. 6). Because of the conformation of the ArOCH₂CH₂OAr components discussed above, the hydrogen-bonded sheets in the structure of (II) are deeply puckered and they enclose substantial voids, each of which accommodates four solvent molecules in two inversion-related pairs, one ordered and one disordered (Fig. 7). Two such sheets, related to one another by the translational symmetry elements of the space group, pass through each unit cell, and they are arranged such that the wide parts of one sheet are adjacent to the narrow parts of the two adjacent sheets (Fig. 7), although there are no significant direction-specific inter­actions between adjacent sheets.

It is of inter­est to compare the structure of the N,N-di­methyl­formamide disolvate, (II), reported here with that of the di­methyl­sulfoxide disolvate, (III), which was briefly reported a number of years ago (Zhu et al., 1999), although with no description or discussion whatever of the supra­molecular assembly; indeed, the presence in the structure of hydrogen bonds was not mentioned. In (III), the ArOCH₂CH₂OAr component lies across a twofold rotation axis in space group C2/c with an O—C—C—O torsion angle of −69.5 (3)°, very similar to those found in (I) and (II), and the overall conformation of this component corresponds to that observed in disolvate (II) rather than that found in unsolvated form (I). The unique solvent molecule in (III) is disordered over two sets of atomic sites having occupancies 0.708 (2) and 0.292 (2), and the molecular components within the selected asymmetric unit are linked by N—H···O hydrogen bonds. Examination of the deposited atomic coordinates for (III) shows, in fact, that the ArOCH₂CH₂OAr components in (III) are linked into sheets by N—H···S hydrogen bonds which contain a similar range of hydrogen-bonded rings to those in (II). Overall, however, the puckering of the sheets in (III) is far less that that found in (II) such that there are no cavities within the sheets, so that the solvent molecules are attached to the faces of the sheets, rather than being located in the inter­ior of deeply puckered sheets as in (II). The two solvent species, N,N-di­methyl­formamide in (II) and di­methyl­sulfoxide in (III), have fairly similar steric requirements, and both act as excellent acceptors of hydrogen bonds but poor donors. It is thus of inter­est that (II) and (III) crystallize in different space groups, P2₁/c and C2/c, respectively, with the non-solvent component lying in a general position in (II) and across a twofold rotation axis in (III), and thus also with different numbers of independent solvent molecules.

Closely related to the bis­(thio­semicarbazide) components in (I)–(III) is that in (IV) (see Scheme 2), which differs from the previous compounds in having a three-carbon spacer unit, –(CH₂)₃–, as opposed to a two-carbon spacer, –CH₂—CH₂–. Compound (IV) crystallizes from N,N-di­methyl­formamide as a partial solvate, but the solvent component was so badly disordered that no chemically sensible model could be developed (Mague et al., 2015). The non-solvent molecules in (IV) lie across twofold rotation axes in space group C2/c, and the conformation of the thio­semicarbazide fragment closely resembles that found in (II) and (III). Thus, the torsion angle corresponding to the angle C11—C12—C17—N1 in (I) and to Cx1—Cx2—Cx7—Nx1 (x = 1 or 2) in (II) has a value of 177.76 (17) Å in (IV), close to the values in (II) (Table 4) and (III) [−161.3 (3); Zhu et al., 1999] but very different from that in unsolvated (I) (Table 2). The conformation in (IV) is thus also consistent with the view that the orientation of the thio­semicarbazide fragments in this series of compounds takes one or the other of only two possible forms, depending upon whether or not the compound is solvated.