A 1:1:1 co-crystal solvate comprising 2,2′-dithiodibenzoic acid, 2-chlorobenzoic acid and N,N-dimethylformamide: crystal structure, Hirshfeld surface analysis and computational study

The three-component title compound contains a molecule each of 2,2′-dithiodibenzoic acid (DTBA), 2-chlorobenzoic acid (2CBA) and dimethylformamide (DMF). The molecules are connected via O—H⋯O hydrogen bonds between DTBA and 2CBA molecules, and O—H⋯O hydrogen bonds between the second carboxylic acid of DTBA and the carbonyl group of the DMF molecule.


Chemical context
Recent bibliographic reviews have highlighted the rich coordination chemistry based on ligands derived from 2-mercaptobenzoic acid (2-MBA) (Wehr-Candler & Henderson, 2016) and its 3-and 4-isomeric analogues (Tiekink & Henderson, 2017). By contrast, co-crystal formation with these molecules is quite limited with the only co-crystal of an n-MBA molecule being that formed between 2-MBA and its oxidation product 2,2 0 -dithiodibenzoic acid (DTBA) (Rowland et al., 2011). One reason for the scarcity of cocrystals containing 2-MBA is the propensity for the acid to be oxidized, to generate DTBA, during co-crystallization experiments with bipyridyl-type molecules (Broker & Tiekink, 2007) and with other carboxylic acids (Tan & Tiekink, 2019a). Another, less common, outcome of crystallization experiments with 2-MBA is the sulfur extrusion product, 2,2 0 -thiodibenzoic acid (Tan & Tiekink, 2018;Gorobet et al., 2018). Herein, ISSN 2056-9890 another unexpected product from a co-crystallization experiment involving 2-MBA is described. While the now anticipated coformer DTBA was observed after the co-crystallization of 2-MBA with 2-chlorobenzoic acid (2CBA), and recrystallization from a toluene/dimethylformamide solution (50:50 v/v), a solvent dimethylformamide molecule was also found in the resultant co-crystal solvate. In this threecomponent crystal, one of the carboxylic acid groups of the DTBA molecule forms hydrogen bonds to DMF rather than to 2CBA. Herein, the crystal and molecular structures of the title co-crystal solvate are described along with an analysis of the calculated Hirshfeld surfaces and a computational chemistry study.
As for DTBA, the confirmation that 2CBA exists as a carboxylic acid is readily ascertained by the difference observed in the C1-O1, O2 bond lengths of 1.222 (4) and 1.320 (4), respectively. The carboxylic acid group is almost coplanar with the phenyl ring (C2-C7) as seen in the dihedral angle of 4.4 (4) between their planes. Similarly, co-planarity is also noted between the chloride atom and benzene ring plane with the r.m.s deviation from the least-squares plane through the seven non-hydrogen atoms being 0.027 Å .

Supramolecular features
The geometric parameters characterizing the interatomic contacts in the crystal of (I), as identified in PLATON (Spek, 2009), are given in Table 1. Some of the main contacts in the molecular packing provide direct links between DTBA, 2CBA and DMF molecules, in that hydrogen bonds are formed between one of the terminal carboxylic groups of DTBA and 2CBA, and between the other carboxylic acid terminus with the carbonyl group of DMF. The former interaction leads to a 476 Tan  The molecular structures of (a) 2,2 0 -dithiodibenzoic acid, (b) 2-chlorobenzoic acid and (c) dimethylformamide in (I), showing the atomlabelling scheme and displacement ellipsoids at the 50% probability level. Table 1 Hydrogen-bond geometry (Å , ). Symmetry codes: (i) x À 1; y; z; (ii) x þ 1; y; z; (iii) Àx þ 1; Ày; Àz þ 1.

Hirshfeld surface analysis
To better understand the nature of the intermolecular interactions in the crystal of (I), the individual molecules comprising the asymmetric unit as well as the contents of the asymmetric unit were subjected to Hirshfeld surface analysis using Crystal Explorer 17 (Turner et al., 2017) and based on the procedures described in the literature (Tan et al., 2019).
The d norm maps of the respective molecules in the aggregates are shown in Fig. 3. DTBA exhibits several intense red spots on the d norm map signifying close contacts which DTBA-C OÁ Á ÁH(hydroxyl-2CBA) and DTBA-C OÁ Á ÁH(DMF). Other red spots are observed through the d norm map, albeit with relatively weak intensity. The contacts are consistent with those identified above except for some additional interactions such as DTBA-C OÁ Á ÁH(phenyl-DTBA), 2CBA-ClÁ Á ÁH(phenyl-DTBA) as well as acontact between the delocalized eightmembered {Á Á ÁHOC O} 2 carboxylic dimer and the phenyl ring of 2CBA, Fig. 3(b). To validate the non-conventionalcontact, the interacting molecules were subjected to electrostatic potential (ESP) mapping using Spartan'16 (Spartan'16, 2017) by treating the DTBA dimer as a single entity through a DFT-B3LYP/6-311+G(d,p) level of theory. The ESP mapping shows that the dimeric ring ranges from electropositive to neutral within the centre of the ring while the phenyl ring of 2CBA is mainly neutral indicating that the interaction is mainly diffusive in nature, Fig. 3(c) and (d). As for the 2CBA and DMF molecules, the corresponding d norm maps (not shown) are reflective of their interactions with the DTBA molecule.
The two-dimensional fingerprint plots were generated to quantify the close contacts identified on the Hirshfeld surfaces.      The overall fingerprint plot of (I) and the corresponding plots of the individual components are shown in Fig. 4. In general, (I) exhibits a shield-like profile in the overall fingerprint plot without any obvious spikes unlike the individual components. This indicates the discrete nature of the three-molecule aggregate sustained by hydrogen bonding. Decomposition of the full fingerprint plots of (I) shows that the contacts are mainly dominated by HÁ Á ÁH (34.3%; 72 Å ) and other contacts (14.0%). Almost all of these contacts are shorter than their corresponding sum van der Waals radii, with HÁ Á ÁH, OÁ Á ÁH, CÁ Á ÁH, SÁ Á ÁH and ClÁ Á ÁH being $2.4, $2.72, $2.9, $3.0 and $2.95 Å , respectively.
The DTBA and 2CBA molecules display similar fingerprint patterns having a claw-like profile in the respective full fingerprint plots, implying the existence of nearly identical interactions between the molecules which is expected considering the similarity of their molecular structures. Detailed analysis of the decomposed fingerprint plots shows that HÁ Á ÁH is the most prevalent contact for the molecules, with the percentage contribution to the overall contacts of 29.7 and 25.0% and minimum d i + d e contact distance of $2.18 and $2.24 Å for DTBA and 2CBA, respectively. The OÁ Á ÁH/ HÁ Á ÁO contacts are the second most dominant contact for the individual molecules which lead to the distinctive spikes in the corresponding decomposed fingerprint plots with a contribution of 26.4% for DTBA and 22.2% for 2CBA. Further delineation of the contact shows that DTBA possesses about 11.1% of (internal)-HÁ Á ÁO-(external) and 15.3% (internal)-OÁ Á ÁH-(external) compared to 2CBA with 10.9 and 11.2% of the equivalent contacts, both with approximately the same

Figure 5
Energy framework of (I) as viewed down along the b axis, showing the energy framework comprising (a) electrostatic potential force, (b) dispersion force and (c) total energy. The cylindrical radii are proportional to the relative strength of the respective energies and they were scaled by a factor of 80 with a cut-off energy value of 5 kJ mol À1 within 4 Â 4 Â 4 unit cells.
benzene ring gives an energy of À15.9 kJ mol À1 which is considered weak in nature. This indicates the energy is mainly dominated by dispersive forces, Table 2, which validates the previous finding on ESP mapping. Interestingly, a recent study demonstrated that the presence of external agents such as Lewis acids may either increase or decrease the strength of resonance assisted hydrogen bonds (RAHB) depending on the position of interaction of the external agent with a carboxylic acid dimer (Grabowski, 2008). The E int for other interactions present in the crystal were also calculated and the results are summarized as in Table 2. Generally, the energies for these interactions range between À23.4 to À6.5 kJ mol À1 which can be considered weak. The energy frameworks of (I) were also generated. The results of the calculations show that the molecular packing is mainly governed by electrostatic forces which can be attributed to the strong O-HÁ Á ÁO interactions, Fig. 5. The interactions coupled with the near orthogonal arrangement of the two carboxylic acid moieties of DTBA lead to a discrete, directional V-shape electrostatic energy topology which is arranged in an alternate array along the c-axis direction. A relatively weaker dispersion force co-exists along with the main energy framework due tointeractions which help to sustain the overall molecular packing of (I).
A structural analogue of (I) in the literature is the 2:1 cocrystal composed of two DTBA molecules and the isomeric 3-chlorobenzoic acid (3CBA) molecule, (II) (Tan & Tiekink, 2019b). Unlike (I), in which hydrogen bonds are formed between DTBA, 2CBA and DMF to result in a three-molecule aggregate, Fig. 2(a), in (II) the two DTBA molecules (DTBA-IIa and DTBA-IIb) form hydrogen bonds with each other, to yield a non-symmetric homosynthon, and with the two remaining carboxylic acid groups being hydrogen bonded to two 3CBA molecules to give rise to a four-molecule aggregate.
A molecular cluster of (I) and (II) containing 20 molecules was subjected to molecular packing analysis using Mercury (Macrae et al., 2006), with the geometric tolerances being set to the default values (20% for distance and 20 for angle tolerance); molecular inversions were allowed during the comparison. The study shows that there are five pairs of DTBA molecules from (I) and (II) which exhibit close similarity in the molecular packing with an r.m.s. deviation of 0.4 Å , Fig. 6. Both (I) and (II) also exhibit similarity in terms of their close contacts as evidenced from the percentage contribution of the corresponding contacts obtained through Hirshfeld surface analysis for the DTBA molecules in (I) and (II), 2CBA in (I) or 3CBA in (II), Fig. 7. In general, the variations in contributions between those DTBA molecules as well as 2CBA and 3CBA are relatively small: these differences range from 0.2 to 2.9% and 1.0 to 2.7% respectively. Exceptions are noted in the CÁ Á ÁH/ HÁ Á ÁC contacts which contribute about 17.5% of the overall contacts in DTBA-I, that is about 7.4 and 3.4% higher than the contacts in DTBA-IIa and DTBA-IIb, respectively. On the other hand, a relatively higher contribution is observed for the CÁ Á ÁC contacts in 3CBA (ca 12.4%) which is approximately 6% greater than 2CBA in (I) (ca 6.3%).

Database survey
There are over 200 structures included in the Cambridge Structural Database (version5.40; Groom et al., 2016) featuring hydrogen bonds between carboxylic acid residues and DMF. The most relevant structure is that of the 1:2   DTBA:DMF solvate (Cai et al., 2006). Here, both carboxylic acid residues engage in hydrogen bonding interactions with DMF molecules akin to that seen in (I). There are approximately 250 structures where (non-coordinated) DMF and a carboxylic acid residue are present in the same crystal but no hydrogen bonding is evident between them. This suggest a 40% likelihood of hydrogen bonding between carboxylic acids and DMF, a percentage higher than for the formation of the eight-membered {Á Á ÁHOCO} 2 synthon in carboxylic acid structures, i.e. 33%, emphasizing that this particular synthon can be readily disrupted in the presence of competing synthons (Allen et al., 1999).

Synthesis and crystallization
All chemical precursors were of reagent grade and used as received without further purification.

Refinement
Crystal data, data collection and structure refinement details are summarized in Table 3. The carbon-bound H atoms were placed in calculated positions (C-H = 0.93-0.96 Å ) and were included in the refinement in the riding-model approximation, with U iso (H) set to 1.2-1.5U eq (C). The oxygen-bound H atoms were located from difference Fourier maps and refined without constraint. Owing to poor agreement, one reflection, i.e. (4 2 2), was omitted from the final cycles of refinement.

Funding information
The support of Sunway University for studies in co-crystals, through Grant No. INT-FST-RCCM-2016-01, is gratefully acknowledged.  CrysAlis PRO (Rigaku OD, 2018); program(s) used to solve structure: SHELXT (Sheldrick, 2015b); program(s) used to refine structure: SHELXL (Sheldrick, 2015a); molecular graphics: ORTEP-3 for Windows (Farrugia, 2012), OLEX2 (Dolomanov et al., 2009) and Mercury (Macrae et al., 2006); software used to prepare material for publication: publCIF (Westrip, 2010). Special details Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.