Crystal structure, conformation and vibrational characteristics of diethyl 4,4′-disulfanediylbis(6-methyl-2-phenylpyrimidine-5-carboxylate) – A new pharmaceutical cure

2.1 Synthesis

The general method for the synthesis of DSMP using ethyl 4-methyl-2-phenyl-6-sulfanylpyrimidine-5-carboxylate as a substrate was mentioned in our previous work (Cieplik et al., 2015). 4 g (0.010 mmol) of this substrate was dissolved in 50 cm³ of tetrahydrofuran. The mixture was refluxed for 8 h with the presence of phenylhydrazine, and then it was cooled and poured into 100 cm³ cold water. The obtained compound was purified by crystallization from methanol to give yellow crystals yielding 4.2 g (25%). In this procedure, only one polymorph of DSMP is formed – the orthorhombic form. The results of chemical analysis of DSMP (C₂₈H₂₆N₄O₄S₂) are : C 61.53% (theor. 61.51%), H 4.75% (4.79%), N 10.18% (10.24%). Its melting point: 129–130 °C and molecular weight: 546.65.

2.2 IR and Raman measurements

IR spectra in the 4000–40 cm⁻¹ range were recorded at room temperature in Nujol and Fluorolube suspensions and KBr pellet with a FTIR Biorad 575C spectrometer. They were identical in the ranges where the bands of Nujol or Fluorolube are absent, so the IR spectra measured in KBr are shown in this paper. The resolution was 2.0 cm⁻¹.

Raman spectra in the 4000–80 cm⁻¹ range were measured in back scattering geometry with a FT Bruker 110/S spectrometer. The resolution was 2.0 cm⁻¹. The YAG:Nd³⁺ (excitation wavelength 1064 nm) laser was used as an excitation source.

2.3 X-ray data collection

Crystallographic measurements for the DSMP were performed on a κ-geometry Oxford Diffraction Xcalibur PX KM-4-CCD diffractometer with graphite-monochromatized Mo Kα radiation (λ = 0.71073 Å) at 90(2) K, using an Oxford-Cryosystems cooler. Data collection, cell refinement, and data reduction and analysis were carried out with the Xcalibur PX software (Oxford Diffraction, Poland): CRYSALISCCD and CRYSALISRED, respectively (Oxford Diffraction Ltd., 2009).

The structure was solved by direct methods using the SHELXS-97 program (Sheldrick, 2008) and refined on F² by full-matrix least squares with anisotropic thermal parameters for all non-H atoms using SHELXL-97 (Sheldrick, 2008). All H atoms were found in difference Fourier maps and were refined isotropically. In the final refinement cycles, the C-bonded H atoms were positioned geometrically and treated as riding atoms, with C—H = 0.95–0.99 Å, and with U_iso(H) = 1.2U_eq(C) for CH and CH₂ or 1.5U_eq(C) for CH₃. The N-bonded H atom was refined with U_iso(H) = 1.2U_eq(N). The figures were made using the DIAMOND program (Brandenburg, 2005).

Crystal data for DSMP: C₂₈H₂₆N₄O₄S₂, M = 546.65, orthorhombic (No. 61), space group Pbca (D_2h¹⁵), a = 16.829 (5), b = 15.125 (4), c = 20.728 (6) Å, V = 5276 (3) ų, Z = 8, T = 90(2) K, 57595 measured reflections, 7652 independent reflections (R_int = 0.029), 6537 reflections with I > 2σ(I), S = 1.05, θ_max = 30°, R[F² > 2σ(F²)] = 0.033, wR(F²) = 0.086, Δρ_max = 0.40 e Å⁻³, Δρ_min = −0.25 e Å⁻³.

The powdered crystal of DSMP diffraction data were recorded at room temperature on a D8 ADVANCE powder diffractometer with nickel-filtered Cu Kα radiation (λ = 1.5418 Å) and a Vantec detector. The measurements were performed within 2θ range of 10–115° with a scan rate 0.008° per step and counting time of 4 s per step. X-ray powder diffraction pattern of DSMP was compared with the simulated XRD pattern of single crystal of the polymorph obtained byCunha et al. (2007). The simulated XRD pattern was obtained with a use of the CCDC’s Mercury program (Macrae et al., 2006) and compared to that of new polymorph obtained by us.

2.4 Quantum chemical calculations

The geometry optimization of the molecular structure of the studied compound was performed for the monomeric unit using Gaussian 03 program package (Frisch et al., 2003). In the calculations the atomic positions from X-ray studies were taken as input data. Only one stable conformer with zero vibrational potential energy was found in such an approach. All calculations were performed applying density functional three-parameter hybrid (B3LYP) methods (Becke, 1996; Lee et al., 1988; Parr and Yang, 1989) with the 6-311G(2d,2p) (McLean and Chendler, 1980; Krishnan et al., 1980) basis set, starting from the X-ray geometry. The calculated and experimental values were compared using two scaling factors (see Table 3) to correct the evaluated wavenumbers for vibrational anharmonicity and deficiencies inherent to the used computational level. The IR and Raman wavenumbers were calculated for a single molecule.

The Potential Energy Distribution (PED) of the normal modes among the respective internal coordinates was calculated using the BALGA program (M.J. Nowak, L. Lapinski), BALGA computer program for PED calculations (Rostkowska et al., 2009).

Vector displacements of the atoms from their equilibrium positions during vibration and the pictures of these displacements were prepared using the ChemCraft program that also visualizes particular modes in an animated way (Zhurko and Zhurko, ChemCraft 1.7).

A linear correlation was used for scaling the theoretical wavenumbers to compare them with the experimental values (Palafox and Rastogi, 2002). The scaling of the calculated wavenumbers improves this result to 2.4 cm⁻¹ for the IR and the Raman spectra. 0.93 scaling factor was used for the range 3500–2500 cm⁻¹ and 0.98 for the range 2499–0 cm⁻¹ of the spectra. The root mean square deviation (RMSD) between the experimental and calculated unscaled wavenumbers for DSMP was 48.4 for the IR and 48.3 for the Raman spectra. The scaling of the calculated wavenumbers improves this result to 6.5 cm⁻¹ for the IR and 5.6 for the Raman spectra.

The theoretical Raman intensities were calculated using the RAINT computer program (Michalska, 2002) reported in Michalska and Wysokiński (2005).

3.1 Crystal structure

The formation of the crystalline DSMP in the new polymorph (orthorhombic form) of single crystals was confirmed by XRD measurements. The simulated XRD pattern of DSMP from the single crystal data of the triclinic form obtained by Cunha et al. (2007) (red), and of the orthorhombic form obtained by us (black), and the experimental XRD patterns of the powder orthorhombic form (blue) are presented in Fig. 1. We can see clear differences between simulated XRD patterns obtained for the single crystal data of the triclinic and orthorhombic forms. On the other hand, a good agreement appears between the simulated XRD patterns obtained for the orthorhombic single crystal and the experimental pattern of the powdered sample. The small difference between both patterns, particularly at high angles, can be explained as the results of the measurements at different temperatures (90 K and RT), but it is not caused by the phase transition. Such differences, a result of changes in unit cell parameters due to the lowering of temperature, are well known (Li et al., 2014; Wang et al., 2014; Yoshii et al., 2015).

Close

Figure 1 Simulated XRD pattern from the single crystal data of DSMP obtained for the triclinic (red) and orthorhombic (black) forms, and the experimental XRD patterns of the powdered crystals of the orthorhombic form (blue).

Export to PPT

DSMP crystallizes in Pbca space group, with one molecule in the asymmetric unit (Fig. 2). The primitive unit cell is built of eight molecules. Selected bond distances, bond angles and principal torsion angles are presented in Table 1. Geometrical parameters were compared with those derived from the DFT calculations and also with the data on the other polymorph – the triclinic form reported by Cunha et al. (2007). The former data are presented for verification of the theoretical model used in the calculations (three-parameter hybrid B3LYP methods with the 6-311G(2d,2p) basis set).

Close

Figure 2 X-ray structure of the DSMP orthorhombic form, showing the atom-numbering scheme. Displacement ellipsoids are drawn at the 50% probability level. H atoms are shown as small spheres of arbitrary radii.

Export to PPT

DSMP contains the S—S bond linking two identical fragments which are built of the phenyl ring, the methyl and the ethyl ester groups attached to the pyrimidine ring. The bond lengths S1—S1′ [2.0473(5) Å], S1—C4 [1.7952(11) Å] and S1′—C4′ [1.7927(12) Å] are compared with corresponding values observed in the previously described triclinic form of DSMP (Table 1) (Cunha et al., 2007). On the whole, bond distances and bond angles in the orthorhombic form are in accordance with the corresponding values of the triclinic form (Cunha et al., 2007). However, pronounced differences between both polymorphs are apparent in the conformation of the pyrimidine rings in relation to each other and also in the conformation of ethyl ester groups and phenyl rings in relation to pyrimidine rings. A comparison of the molecular structures of both polymorphs of DSMP is shown in Fig. 3.

Close

Figure 3 A comparison of structures: the orthorhombic (solid line) and triclinic (dashed line) forms of DSMP. The common reference points are N1, C2, C3, C4, C5 and C6 atoms of the pyrimidine ring.

Export to PPT

The conformation of DSMP is best defined by the torsion angles C4—S1—S1′—C4′, S1′—S1—C4—N3 and S1′—S1—C4—C5 of −81.78(5)°, 8.51(8)° and −171.06(8)°, respectively (Table 1). Two pyridine moieties are in the nearly gauche position, with the dihedral angle between the mean planes of both pyrimidine rings of 86.01(4)°; the corresponding angle is 78.13(5)° in the triclinic form (Cunha et al., 2007). Contrary to the triclinic form, the phenyl rings are nearly coplanar with the pyrimidine rings. The angle between the least-squares planes through N1—C6 and C21—C26 atoms is 3.54(4)°, and through N1′—C6′ and C21′—C26′ atoms is 3.13(6)°, whereas in the case of the triclinic form, the corresponding angles are 6.57(5)° and 9.29(5)°, respectively (Cunha et al., 2007). Additionally, two ethyl ester groups are twisted in relation to the pyrimidine rings, as shown by torsion angles around C5—C51 and C5′—C51′ bounds (Table 1).

The crystal structure analysis revealed different arrangement of molecules and intermolecular interactions between them to distinguish both polymorphs. There are no-direction-specific interactions between adjacent molecules in the triclinic form; the shortest distance H⋯A is longer than 2.7 Å (Cunha et al., 2007). In the case of the orthorhombic form, the molecules are linked by weak C—H⋯O interactions (Table 2), with C23 and C24′ phenyl atom as donors and ester atoms O51′(−x + 1, y−1/2, −z + 3/2) and O51 (x−1/2, y, −z + 1/2) as acceptors in these linkages (Fig. 4). There are also weak C—H⋯π interactions between molecules to complete the three-dimensional connectivity of the structure, but they are no considered in this paper.

Close

Figure 4 Part of the crystal structure of the DSMP orthorhombic form, showing C—H⋯O (orange dashed lines) interactions. Symmetry codes as in Table 2.

Export to PPT

3.2 IR and Raman spectra

The PED data show that three types of vibrations could be identified for the studied compound. The normal modes, in which mainly one structural unit contributes to the vibration, form the first class of vibrations. The vibrations in which the separate CH₃, C₂H₅, C⚌O units participate belong to this type. The second type of normal modes constitutes of the skeleton vibrations activating mainly the benzene ϕ or pyrimidine θ rings. The normal modes originating from the coupled vibrations, both the skeleton and substituted groups, i.e. mixed modes, form the third type of vibrations. This classification has been used in Table 3 that lists the calculated and observed band wavenumbers together with their assignments to the respective normal modes.

3.2.1

3.2.1 Benzene (ϕ) ring vibrations

The vibrational characteristics of the benzene ring vibrations in its mono- and polysubstituted derivatives are well known (Lin-vien et al., 1991; Socrates, 2001; Varsányi and Szőke, 1969). In the studied compound, both benzene rings are substituted by polar chromophore – pyrimidine ring. Therefore, we expect appearance of the bands originating from vibrations of ten CH bonds and two benzene rings in the IR and Raman spectra (Fig. 5 and Table 3). For the latter bonds the characteristic vibrations are observed in the following ranges (Raman lines in parentheses): νCH: 3089 (3090), 3062 (3070, 3064), 3056 (3057, 3044), 3033 (3035) cm⁻¹; δCH_ϕ + νϕ: 1600, 1586 (1599); δCH_ϕ + νθ: 1511, 1495 (1495) cm⁻¹; δϕ + δθ + δCH_ϕ: 1303, δCH_ϕ: 1172, 1161 (1166), 1083, 1061 (1081), δϕ: 1026 (1028), ν_sϕ: 1002 (1001); γCH_ϕ: 986, 949, 938 (988), 851 (851); γθ + γCH_ϕ: 816 (816); γCH_ϕ + ωOCO: 751, 745 (748), γCH_ϕ: 702, 696; δϕ + δθ: 675, 670 (672); γϕ + γCH_ϕ + γθ: 662, 654 (660), δϕ: 619 (619) cm⁻¹; γϕ + δC_θ—O—Et: 463; γϕ + γC_θ—CH₃ + γC_θ—O—Et: 449 cm⁻¹. These bands fit well the literature data (Lin-vien et al., 1991; Socrates, 2001; Varsányi and Szőke, 1969). The wavenumbers observed in the spectra are described as benzene (ϕ) ring modes. It should be noted that the vibrations originating form the benzene ring vibrations produce the characteristic doublet contours due to the appearance of two benzene rings in the molecule. This is observed for the bands at about 1590, 1510, 1490, 1170, 1080, 1025, 1000, 990, 940, 850, 815, 780, 750, 700, 675, 660 and 620 cm⁻¹. Because the experimental wavenumbers fit well those calculated in the quantum chemical procedure it confirms that the used basis set and B3LYP functional have been chosen properly.

Close

Figure 5 Experimental and calculated IR (top) and Raman (below) spectra of DSMP in the 3250–80 cm⁻¹ range. IR spectrum was measured in KBr pellet.

Export to PPT