2. Structural Commentary 3. Supramolecular Features 4. Analysis of the Hirshfeld Surfaces 5. Database Survey

A cinnamaldehyde Schiff base of S-(4-methylbenz-yl) dithiocarbazate: crystal structure, Hirshfeld surface analysis and computational study The title dithiocarbazate ester (I), C 18 H 18 N 2 S 2 [systematic name: (E)-4-methylbenzyl

The title dithiocarbazate ester (I), C 18 H 18 N 2 S 2 [systematic name: (E)-4methylbenzyl 2-[(E)-3-phenylallylidene]hydrazinecarbodithioate, comprises an almost planar central CN 2 S 2 residue [r.m.s. deviation = 0.0131 Å ]. The methylene(tolyl-4) group forms a dihedral angle of 72.25 (4) with the best plane through the remaining non-hydrogen atoms [r.m.s. deviation = 0.0586 Å ] so the molecule approximates mirror symmetry with the 4-tolyl group bisected by the plane. The configuration about both double bonds in the N-N C-C C chain is E; the chain has an all trans conformation. In the crystal, eightmembered centrosymmetric thioamide synthons, {Á Á ÁHNCS} 2 , are formed via N-HÁ Á ÁS(thione) hydrogen bonds. Connections between the dimers via C-HÁ Á Á interactions lead to a three-dimensional architecture. A Hirshfeld surface analysis shows that (I) possesses an interaction profile similar to that of a closely related analogue with an S-bound benzyl substituent, (II). Computational chemistry indicates the dimeric species of (II) connected via N-HÁ Á ÁS hydrogen bonds is about 0.94 kcal mol À1 more stable than that in (I).

Chemical context
A large number of studies have been carried out since 1974 on dithiocarbazate-derived Schiff bases of general formula NH 2 NHC( S)SR which are synthesized from the condensation reaction of S-alkyl or -aryl esters of dithiocarbazic acid with different types of aldehydes or ketones (Ali & Livingstone, 1974;Ravoof et al., 2010;Hamid et al., 2016). Recent work has reported electrochemical studies of conjugated copper(II) dithiocarbazate complexes that undergo an irreversible oxidation/reduction of Cu II /Cu I (Blower et al., 2003;Paterson et al., 2010). Dithiocarbazate Schiff bases have also been reported to show variable cytotoxicity against estrogen receptor positive human breast cancer cells  and other cell lines depending on their substituents (Pavan et al., 2010;Low et al., 2016). In fact, related 2-acetylpyridine Schiff bases of S-methyl-and S-benzyl-dithiocarbazate have better cytotoxic potential as compared to their complexes (Hamid et al., 2016). As part of an on-going study on the potential biological activities and structural chemistry of dithiocarbazate Schiff bases and their metal complexes (Yusof, Ravoof, Jamsari et al., 2015;Low et al., 2016), the synthesis of the title compound, (I), its crystal and molecular structures along with an analysis of its Hirshfeld surface and computational modelling are reported herein.
Further discussion on the molecular geometry of (I) is given in Computational chemistry calculations.
essence, the C-HÁ Á Á interactions connect molecules into layers in the bc plane and these are linked by the N-HÁ Á ÁS hydrogen bonds.

Analysis of the Hirshfeld surfaces
The most closely related compound in the crystallographic literature is one with a benzyl substituent at the S2 atom (Tarafder et al., 2008) rather than a CH 2 (tolyl-4) group, that might be regarded as the 'parent' compound, hereafter referred to as (II). While detailed discussion on the comparison of their molecular geometries and computational modelling are given in Computational chemistry calculations, the present section focuses upon the study of intermolecular interactions formed by (I) and (II) in their respective crystals by Hirshfeld surface analysis in accord with the method described recently (Yeo et al., 2016).
Both (I) and (II) exhibit closely related topological interactions as evidenced by the relative distribution of similar contacts, Fig. 3, computed based upon the mapping of the contact distances at specific points on their Hirshfeld surfaces (Spackman & Jayatilaka, 2009 Fig. 4c, at approximately 2.7 Å , which is slightly shorter than the van der Waals radii of 2.9 Å . The decomposed fingerprint plots of SÁ Á ÁH/HÁ Á ÁS (Fig. 4d) and NÁ Á ÁH/HÁ Á ÁN contacts ( Fig. 4e) for (I) register contact distances of 2.47 and 2.90 Å , respectively, which is about 0.05 Å (1.7-2.0%) longer than those of (II). It is noteworthy that the HÁ Á ÁH contact of (I) is significantly shorter than the sum of their van der Waals radii, by 0.44 Å (22.4%) cf. (II), in which the difference is merely 0.04 Å (1.7%). Similarly, the SÁ Á ÁH/ HÁ Á ÁS contacts of both (I) and (II) exhibit shorter contact distances cf. the sum of their van der Waals radii by 0.53 and 0.58 Å , respectively (21.5 and 24.0%). As a result, those contacts display intense red spots on their Hirshfeld surface, Fig. 4d.
In view of the close structural similarity between (I) and (II), their physical properties such as molecular volume, surface area, shape, density and packing efficiency were computed either by Crystal Explorer (Wolff et al., 2012) or PLATON (Spek, 2009) and data are compared in Table 2. As expected, the molecule of (I), which has an additional methyl group cf. (II), exhibits a greater molecular volume and surface area, and is slightly less globular. This results in a lower surface-to-volume ratio and density for (I), and ultimately leads to reduced packing efficiency when compared to (II).

Database survey
As mentioned in the previous section, the 'parent' compound represents the most closely related analogue to (I) in the Cambridge Crystallographic Database (Groom et al., 2016) and hence, it is adopted for direct comparison in terms of their geometric parameters; selected data are collated in Table 3. All bond lengths are equal within experimental error and bond angles agree to within 1 . The influence, if any, upon the molecular conformation exerted by the tolyl substituent in (I) might be manifested in the twists about the C11-C12 bond as Relative percentage contributions of close contacts to the Hirshfeld surfaces of (I) and (II). Table 2 Comparison of some physical properties between (I) and (II).  Table 1 Hydrogen-bond geometry (Å , ).

Computational chemistry calculations
Both (I) and (II) were subjected to geometry optimization calculations assuming a gas-phase environment in order to compare the structural difference between the experimental and theoretical models. The corresponding theoretical models were first drawn using GaussView5 (Dennington et al., 2009) based on the geometrical conformation of the structure (trans-cis along C1 S1 and E, E along N2-C2, C3-C4) and pre-optimized using a semi empirical method (PM6) with a precise self-consistent field criterion. Subsequently, the geometries were further optimized at B3LYP/6-311+G(d,p) without imposing symmetry constraints. A frequency analysis was performed on each optimized structure using the same level of theory and basis set to validate that each structure was indeed the local minimum structure with no imaginary frequency. All calculations were performed using the Gaus-sian09 software package (Frisch et al., 2016). The results, as shown from the superposition of the experimental structure and theoretical model of (I) and (II), Fig. 5, indicate that there is not much difference between the experimental and optimized structures with the r.m.s. deviation of about 0.2110 Å in the case of (I) and 0.1747 Å in the case of (II). The key geometric parameters obtained from the calculations are also listed in Table 3. The energy-minimized structures have effective mirror symmetry whereby the Sbound aryl ring is bisected by the plane. The bond lengths and angles for optimized-(I) and -(II) are identical indicating no influence upon the electronic structure is exerted by the addition of a methyl group in (I). Indeed, the optimized geometries for (I) and (II) are superimposable, Fig. 5. Despite the close similarity between the optimized structures, some differences are noted between the experimental and optimized structures. For example, the C1-S2 and C11-S2 bond lengths have elongated by ca 0.02 and 0.03 Å , respectively. In the chain, the C1-N1 bond lengths have lengthened by ca 0.03 Å , a difference accompanied by a contraction in the N1-N2 bond length by about the same amount. Minor differences are also noted in bond angles with widening of S1-C1-S2 and the angles subtended at the nitrogen atoms by 2-3 with similar contractions in the C1-S1-C11 and S1-C1-N1 angles.
Apart from geometry optimization, both (I) and (II) were also subjected to computational modelling for calculation of their interaction energies. Briefly, the crystallographic coordinates of the experimental dimeric structures of (I) and (II) connected through N-HÁ Á ÁS interactions were used as the input without further optimization. In order to preserve the integrity of the structure for best possible estimation of the interaction energy from the experimental model, the positions  Table 3 Selected geometric parameters (Å , ) in (I) and (II) and in geometry-optimized-(I) and -(II).

Figure 5
Structural overlay between the crystal and optimized structures of (I) (red image), (Io) (green), (II) blue) and (IIo) (purple). of all hydrogen atoms obtained during crystal refinement were kept unchanged, despite that this method (riding-model approximation) is commonly known to induce deviations by as much as 0.1 to 0.2 Å shorter C-H bond lengths. The respective input structures were submitted to single point interaction energy calculation by long-range corrected !B97XD functional combining the D2 version of Grimme's dispersion model and the 6-31G(d,p) basis set. It has been demonstrated that the long-range corrected hybrid method can greatly reduce self-interaction errors (Chai & Head-Gordon, 2008) and gives a better accuracy in binding energy as compared to coupled cluster calculations (Andersen et al., 2014). The computed interaction energy (i.e. the energy difference between the dimer and the sum of energies for the corresponding monomers) was obtained upon the correction of basis set superposition error (BSSE) by counterpoise correction. All calculations were performed in gas phase using Gaussian09 software (Frisch et al., 2016). The dimeric species of (I) and (II) possesses the interaction energy (E BSSE int ) of À12.92 and À13.86 kcal mol À1 , respectively. The range is approximately 3.89 to 5.23 kcal mol À1 less than the energy computed for a pair of thiourea dimers at the RIMP2/cc-pVDZ and cc-pVTZ levels of theory (AlDamen & Sinnokrot, 2014). Apparently, the corresponding E BSSE int energies were overestimated due to the use of the splitvalence double basis set as an necessary compromise between accuracy and computational cost since the calculations involve a rather large molecular system with over 80 atoms. Despite the difference, the dimer of (II) is lower in energy (ca 0.94 kcal mol À1 ) cf. (I), indicating that the former is connected by relatively stronger N-HÁ Á ÁS interactions and hence, the dimeric aggregate in (II) is more stable. The theoretical result is in accord with the experimental data, in which the HÁ Á ÁS

Synthesis and crystallization
The following procedure was adapted from the literature (Ravoof et al., 2010): S-4-methylbenzyldithiocarbazate (2.12 g, 0.01 mol) was dissolved in hot acetonitrile (100 ml) and added to an equimolar amount of cinnamaldehyde (Merck, 1.32 g) in absolute ethanol (20 ml). The mixture was heated for about 2 h and was then allowed to stand overnight. The pale-brown crystals that formed were filtered and washed with absolute ethanol at room temperature. Yield: 70%. M.p. 463-466 K. Analysis: Calculated for C 18 H 18 N 2 S 2 : C, 66.22;H,5.56;N,8.58. Found: C,65.87;H,5.77;N,9.

Computing details
Data collection: CrysAlis (Agilent, 2011); cell refinement: CrysAlis (Agilent, 2011); data reduction: CrysAlis (Agilent, 2011); program(s) used to solve structure: SHELXS97 (Sheldrick, 2008); program(s) used to refine structure: SHELXL2014/7 (Sheldrick, 2015); molecular graphics: ORTEP-3 for Windows (Farrugia, 2012), DIAMOND (Brandenburg, 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.