Crystal structure from X-ray powder diffraction data, DFT-D calculation, Hirshfeld surface analysis, and energy frameworks of (RS)-trichlormethiazide

The structure of the racemic form of the diuretic drug trichlormethiazide was determined from laboratory X-ray powder diffraction data: the extended structure features an intricate combination of N—H⋯O hydrogen bonds and π–π and C—Cl⋯π interactions.


Chemical context
Trichlormethiazide (TCMZ), systematic name 6-chloro-3-(dichloromethyl)-1,1-dioxo-3,4-dihydro-2H-1 6 ,2,4-benzothiadiazine-7-sulfonamide (C 8 H 8 Cl 3 N 3 O 4 S 2 ), is a diuretic drug derived from thiazide, the precursor of a classic family of diuretic compounds, discovered in the 1950s. The first approved drug of this class, chlorothiazide, was marketed under the trade name Diuril in 1958 (Beyer, 1993). The compound under study, trichlormethiazide, has a similar chemical structure to hydrochlorothiazide, the most prescribed member of the group (Hripcsak et al., 2020). The difference is the substitution of one hydrogen atom of the methylene group by a CHCl 2 dichloromethyl group. Thiazide diuretics and their derivatives are primarily used in the treatment of mild to moderate hypertension and oedema associated with Na + and K + retention and expansion of the extracellular fluid volume. They also increase Ca 2+ excretion, a potentially useful effect in patients with hypercalciuria, a condition that could lead to the formation of kidney stones (Menè, 2004). It is commonly used around the world under different brand names such as Achletin, Aquacot, Diu-hydrin, Diurese, Metahydrin, Naqua, Triflumen, as well as with the generic trichlormethiazide name. Given our interest in the structure of materials involved in pharmaceutical formulations or with potential pharmaceutical applications, it was decided to undertake the structure determination of the racemic form of this active pharmaceutical ingredient (API).

Structural commentary
The refinement of the final structural model using powder diffraction data recorded showed bond distances and angles within the range suggested in the statistical analysis performed with the Mogul geometry check (Bruno et al., 2004). Only two out of 56 distances and bond angles in the structure are classified in the analysis as 'unusual'. However, these two 'unusual' parameters are close to the values suggested by the Mogul geometry analysis, with Z-scores below 3. These parameters are similar to the values reported for the S-isomer (Cambridge Structural Database refcode KIKCUD; Fernandes et al., 2007) and for the DFT-D-optimized racemic structure.
The asymmetric unit contains one TCMZ molecule ( Fig. 1): the stereogenic centre C7 has an S configuration but crystal symmetry generates a racemic mixture. The thiazide ring (A) exhibits a conformation that could be described as distorted half-chair to distorted envelope at N3 (Spek, 2020). The substituents in the ring are in bisecting (S1-O3), axial (S1-O4, N3-H3A, C7-H7) and equatorial (C7-C8, C4-C3, C5-C6, N2-H2A) conformations. The almost planar benzosulfonamide ring (ring B) makes an angle of 8.2 (2) with the best plane through the thiazide ring. The molecule is oriented almost parallel to the a-axis as indicated by a 3.11 (8) angle (PLATON; Spek, 2020), which corresponds to the angle between the a-axis and the perpendicular to the normal of the best least-square plane defined by the atoms of the two rings. The angle between the corresponding A and B rings in S-TCMZ is smaller than in RS- TCMZ [4.7 (2) ]. Fig. 2 shows a superposition of the molecule with an S-configuration in racemic TCMZ with the molecule of the S-enantiomer in KIKCUD. When flexibility is allowed in the superposition (Fig. 2a), the r.m.s.d. deviation is 0.070 and the maximum deviation (max. D) is 0.146 Å . Without flexibility, the values for r.m.s.d. and max. D are 0.785 and 2.763 Å , respectively (Fig. 2b). The difference between the two conformations lies in the orientation of the sulfonamide group and leads to differences in the hydrogen-bonding patterns between the two compounds as discussed below.

Intermolecular hydrogen bonds
Given the number of potential hydrogen-bond donors and acceptors, the hydrogen-bonding pattern in RS-TCMZ is very rich and relevant geometric parameters are summarized in Table 1. Fig. 4  N2,N3 (H2A, H3A, VI) motifs, forming layers parallel to the ac plane, resulting in an intricate threedimensional hydrogen-bonded network. In the structure of S-TCMZ, N1 and N3 are also involved in hydrogen bonds based on the N-HÁ Á ÁO heterosynthon. However, N2 participates in the homosynthon N2-H2Á Á ÁN1. A C8-H8Á Á ÁO3 hydrogen bond is also important in the packing arrangement of S-TCMZ.
Cg is the centroid of the C1-C6 ring.
The structure of RS-TCMZ is a complex arrangement of hydrogen bonds,and C-ClÁ Á Á interactions as shown in Fig. 6a and 6b. It can be described in terms of chains of head-to-tail dimers connected byinteractions, which are further connected via C-ClÁ Á Á interactions, also in a head-to-tail fashion. These chains are connected by N-HÁ Á ÁO hydrogen bonds, producing layers parallel to the ac plane. The layers stack along the b-axis, connected by other N-HÁ Á ÁO hydrogen bonds. In contrast, the structure of S-TCMZ can be described as chains of S-TCMZ molecules connected by C-ClÁ Á Á interactions (Fig. 6c), which form columns along the baxis. These columns are further connected by N-HÁ Á ÁO and N-HÁ Á ÁN hydrogen bonds (Fig. 6d). entry. It corresponds to S-trichlormethiazide (refcode: KIKCUD; Fernandes et al., 2007), which reports the structure determined using single-crystal X-ray diffraction data. This enantiomer crystallizes in the orthorhombic space group P2 1 2 1 2 1 . The PDF-4/Organics database of the Powder Diffraction File (Gates-Rector & Blanton, 2019) contains two entries associated with this material, PDF 02-094-5865 and PDF 00-039-1828. The first report consists of a calculated pattern based on the CSD report described above. The PDF 00-039-1828 entry contains an experimental pattern with no structural information. The superposition of the recorded pattern and the simulated pattern contained in entry PDF 00-039-1828 (depicted in Fig. S1 of the additional supporting information) shows that they correspond to the same phase. It is worth mentioning that a broader search of the CSD resulted in 100 structures related to TCMZ, among them chlorothiazide and hydrochlorothiazide, their polymorphs, derivatives, solvates, and co-crystals.

Synthesis and crystallization
(RS)-Trichlormethiazide was kindly provided by Tecnoquímicas (Cali, Colombia). Based on the FT-IR spectra and the quality of the preliminary diffraction patterns, the present study was carried out on the sample as it was received. Crystallization experiments in different solvents, in search of possible polymorphs, are underway in our laboratories.
The fitting of the pattern was carried out with the Pawley algorithm by modelling the background, zero-point and sample displacement errors, cell parameters, peak shape parameters (including anisotropic broadening) using TOPAS-Academic (Coelho, 2018). A 20-term Chebyshev polynomial was used to model the background. The intermediate Gaussian-Lorentzian function was employed with a correction for axial divergence as proposed by the program. The Pawley refinement produced a good fit with residuals R p = 0.02480, R wp = 0.03280%, and GoF = 1.343, strongly supporting the correctness of the unit cell: all the diffraction maxima recorded were accounted for by the triclinic unit cell obtained with DICVOL14 (Louë r & Boultif, 2014). The initial molecular model, introduced as a '.mol' file, was obtained from the CIF of KIKCUD. With this model and the parameters obtained from the Pawley fit, the crystal structure was determined with DASH 3.4.5 (David et al., 2006). The refinement of the structure, carried out with TOPAS-Academic (Coelho, 2018), produced a reasonably good fitting with residuals R p = 0.0687, R wp = 0.0931, and GoF = 3.985. However, there were discrepancies between the calculated and measured intensity for a few of the most intense diffraction maxima of the pattern.

Figure 7
Rietveld plot obtained after the structure refinement of RS-TCMZ Cartesian displacement (RMSCD) of 0.539 Å . This value is beyond the limit of 0.35 Å considered acceptable for correct structures determined from powder diffraction data (van de Streek & Neumann, 2014). The examination of the structure with Mercury (Macrae et al., 2020) showed that the structure displayed short N-HÁ Á ÁCl-C contacts. The charge distribution on a C-Cl bond is such that at the 'tip' (the other side of the atom away from the bond to the carbon atom), the Cl atom has a positive charge. Surprisingly, a hydrogen-bond donor points to this region of the Cl atom. There are several OÁ Á ÁO contacts around 3.0 Å , which are possible, but surprising given the presence of four N-H hydrogen bond donors. By rotating the O-S(O)-N group around the C1-S1 bond axis by 120 all these inconsistencies disappeared and, therefore, this model was adopted. The refinement performed with TOPAS-Academic (Coelho, 2018), using the energy-minimized structure as the starting model, was very stable and proceeded smoothly. Fig. 7 shows the final Rietveld refinement plot. The refinement included an overall scale parameter, the background, the peak shapes (including anisotropic broadening), unit-cell parameters, atomic coordinates and, initially, an overall B iso parameter. The bond distances and angles were restrained based on the values of the energy-minimized structure. A planar restraint for the molecule with a standard deviation of 0.01 Å was also established for atoms C1-C6/Cl1/S1. The positions of the hydrogen atoms were refined with restrictions on bond lengths and angles to the atoms to which they are attached, as in the related hydrochlorothiazide form II structure (Florence et al., 2005). The standard uncertainties of the hydrogen atoms, calculated by TOPAS, reflect the propagation of statistical errors from the raw data and do not reflect contributions from systematic errors. More realistic values are somewhat larger than those reported. The isotropic atomic displacement parameters for S and Cl were constrained to be equal and those for C, N, and O were also constrained to be equal. For the hydrogen atoms, they were 1.2 times the U iso of the C or N atom to which they are attached.
In total, 177 parameters were refined against 3922 data points, 48 restraints and 2 constraints. The final whole pattern fitting converged with good figures of merit: R e = 0.02577, R p = 0.0512, R wp = 0.0694, R B = 0.0397, and GoF = 2.704. Table 2 shows the crystal data, experimental parameters, and the refinement parameters obtained. The DFT-D calculations of the new model led to an RMSCD of 0.126 Å , which is lower than the 0.35 Å value (van de Streek & Neumann, 2014), indicating that the structure determined can be assumed to be correct.

Hydrogen-bond propensity analysis
As several donor and acceptor groups are present in trichlormethiazide, which could form different hydrogen bonding schemes, it was considered of interest to carry out a hydrogen-bond propensity (HBP) analysis for this molecule. The HBP analysis was carried out with Mercury (Macrae et al., 2020) for RS-TCMZ and for the S-TCMZ enantiomer using data from the CSD entry KIKCUD.
The HBP tool provides an insight into the expected intraand intermolecular hydrogen bonds in the structures. For the analysis, the donor atoms considered were N1 (sulfonamide), N2 (secondary amine) and N3 (next to the sulfonyl group). The acceptors were Cl1 (aryl chloride), Cl2/Cl3 (alkyl chloride), N2 (secondary amine), O1/O2 (sulfonamide), and O3/O4 (sulfonyl). The area under the receiver operating characteristics (ROC) curve was 0.863, indicating good statistical discrimination in the analysis. The results of the calculations are presented in the supporting information.
The intramolecular hydrogen bond with the highest propensity is N1-H1BÁ Á ÁCl1 (0.60). This hydrogen bond is observed only in the S-enantiomer. The intramolecular interaction involving N3-H3AÁ Á ÁCl3, observed in the two structures, has the second highest propensity value (0.48).
Regarding the intermolecular interactions, two hydrogen bonds involving the hydrogen atoms bonded to the nitrogen of the sulfonamide group and the two oxygen atoms of the sulfonyl group (N1-H1BÁ Á ÁO3 and N1-H1AÁ Á ÁO4) have the highest propensities (0.69). They are present in the structure of RS-TCMZ (motifs II and III). However, only one of them (N1-H1AÁ Á ÁO4) is present in the S-enantiomer. The next two interactions with highest propensities (0.68) are between the H and O atoms of the sulfonamide groups of two neighboring molecules. One of them (N1-H1BÁ Á ÁO2) is observed only in the S-enantiomer.
The CSD statistics predicts hydrogen bonds for the sulfonyl nitrogen atom (N3-H3A) and for the secondary amine (N2-H2A) with the sulfonyl O atoms (propensity values are 0.44 and 0.42, respectively), which are not present in either structure. However, N3-H3AÁ Á ÁO1 and N3-H3AÁ Á ÁO2 contacts with 0.42 propensities are displayed in S-TCMZ and RS-TCMZ (motifs I and IV), respectively. In addition, the hydrogen bond N2-H2AÁ Á ÁO1 is present in RS-TCMZ (motifs V and VI) but not in S-TCMZ. The hydrogen bond N2-H2AÁ Á ÁN1 was not predicted because the N1 atom was not considered an acceptor. The hydrogen-bond patterns found in the two structures are consistent with the hydrogenbond propensity analysis results. Every donor and acceptor in RS-TCMZ and in S-TCMZ has a hydrogen-bond coordination with a high likelihood. Figure S2 of the additional supporting information shows the putative landscape for trichlormethiazide. The two structures fall in the high propensity and hydrogen-bond coordination zone.
Hirshfeld surface analysis and energy frameworks The software CrystalExplorer21 (Spackman et al., 2021) was used to produce fingerprint plots of the intermolecular interactions occurring in RS-TCMZ and in CSD entry KIKCUD. The parameter d norm , mapped onto the Hirshfeld surface (Spackman & Jayatilaka, 2009) is useful to visualize the atoms involved in intermolecular contacts and the strength of such contacts. Energy frameworks were also calculated with Crys-talExplorer21.
Fingerprint plots representing d e /d i interactions were calculated for RS-TCMZ and are shown in Fig. 8. For comparison, plots for S-TCMZ were also calculated and shown in the same figure, along with the contributions of all contacts in RS-TCMZ and S-TCMZ to the Hirshfeld surface area. As can be seen, there are significant differences between the fingerprint plots for both compounds. The full set of parameters calculated are presented in Fig. S3 of the additional supporting information. Fig. 8a-8l show that the most important interactions in RS-TCMZ are the HÁ Á ÁO and HÁ Á ÁCl contacts, which represent 32.2 and 21.7% of the surface, respectively. In S-TCMZ, they are also the most important contacts (Fig. 8m-8x) with 36.0 and 16.9%, respectively. The next interaction, OÁ Á ÁCl, is slightly less important in RS-TCMZ than in S-TCMZ (8.7% versus 9.6%). The remaining interactions differ in order of 146 Toro et al.  importance. For example, the HÁ Á ÁH interaction is more important (8.5%) in RS-TCMZ than in S-TCMZ (7.2%). It is worth noting that the fingerprint plot delineated into the HÁ Á ÁH interaction for RS-TCMZ shows a tip at d e + d i = 2.20 Å , which is less than 2 times the van der Waals radii of hydrogen. In contrast, in S-TCMZ this interaction is dispersed over a range of d e + d i values. Weaker interactions such ascontacts are present only in racemic TCMZ and they represent 1.8% of the contribution to the Hirshfeld surface. The ClÁ Á Á interaction is more important in S-TCMZ, contributing 9.1% in contrast to RS-TCMZ where it represents 2.8%. This is the result of two interactions in the S-enantiomer that lead to layers parallel to the ab plane. In RS-TCMZ, the ClÁ Á Á interactions alternate withcontacts to produce chains nearly along [101]. Another interesting feature is displayed by the HÁ Á ÁN contacts. There is a lower degree of directionality and strength of this interaction in RS-TCMZ (2.1%) than in S-TCMZ (3.3%) as a result of the additional N2-H2AÁ Á ÁN1 interaction in the latter.
In addition, the electrostatic (E ele ), dispersive (E dis ), and total energies (E tot ) for the intermolecular interactions in RS-TCMZ and S-TCMZ were calculated with CrystalExplorer21 Energy frameworks calculated for RS-TCMZ viewed down the b-axis: (a) E ele , red; (b) E dis , green; (c) E tot , blue. Energy contributions are represented within 4 Â 4 Â 4 unit cells. The cylinder radii were scaled to 50 arbitrary units with a cut-off value of 10 kJ mol À1 . Energy frameworks calculated for S-TCMZ viewed down the b-axis represented within 2 Â 2 Â 2 unit cells: (d) E ele , red; (e) E dis , green; and (f) E tot , blue. The cylinder radii were scaled to 80 arbitrary units with a cut-off value of 5 kJ mol À1 . (Spackman et al., 2021). They are represented in Fig. 9. The summary of calculated energy values is presented in the supporting information and the detailed interactions are collected in Table S1 of the supporting information.
As depicted in Fig. 9, in RS-TCMZ the topologies of the electrostatic (E ele , Fig. 9a) and dispersive (E dis , Fig. 9b) components are similar although their contributions are quite different. They result in an offset tile arrangement for E tot when viewed down the b-axis direction (Fig. 9c). In the structure of S-TCMZ, E ele and E dis make similar contributions to E tot and their topology is similar (Fig. 9d and 9e). The pattern viewed down the b-axis direction resembles a herringbone arrangement (Fig. 9f).

Spectroscopic and thermal characterization
The FT-IR spectrum shows the absorption bands expected for TCMZ (Fig. S4 of the supporting information). The stretches for the secondary N-H grouping of the sulfonamide group appear at 3387 and 3322 cm À1 followed by the stretching bands of the S-N-H and N-H groups of the amine on the dihydrothiadiazine at 3281 and 3232 cm À1 , respectively. The stretches of the Csp 2 -H (3150-3100 cm À1 ) and Csp 3 -H (3000-2900 cm À1 ) bonds are observed as weak bands. The Csp 2 -Csp 2 stretch of the aromatic ring appears at 1596 cm À1 while the C-N and S-N stretches overlap at 1351 and 1332 cm À1 . The stretches of the two S O groups appear as strong absorptions at 1176 and 1157 cm À1 . Table S2 summarizes the assignment of the most important absorptions for RS-TCMZ.
The TGA curve (Fig. S5a) recorded indicates the material is stable up to 240 C. A series of weight loss events occur from 240 C to 450 C. A 24.2% weight loss (2.270 mg) between 245 and 301 C coincides with the first two transitions in the DSC (Fig. S5b). The endotherm at 281.1 C (ÁH = 81.19 J g À1 ) is associated with melting of the material. This transition is followed by an exotherm with peak temperature 287.9 C (ÁH = 103.70 J g À1 ). The TGA curve shows two continuous weight loss processes at 302-384 C (1.354 mg, 14.4%) and 384-448 C (1.032 mg, 11.0%), associated with ill-defined transitions in the DSC. The total weight loss due to decomposition is 49.6%.

Special details
Experimental. The FT-IR spectrum was registered in a IS50 FT-IR Nicolet Thermo Scientific spectrophotometer, in the 4000-400 cm -1 range with 32 scans at an optical speed of 0.4747 cm s -1 . Thermogravimetric and differential scanning calorimetry measurements (TGA/DSC) were performed in a thermal analyzer DTA/DSC Instrument, Serie Discovery, under a dynamic nitrogen atmosphere at 50.0 mL min -1 . The instrument was equilibrated at 28.00 °C. A heating ramp of 10.00 °C min -1 up to 450.00 °C was used. RS-Trichlormethiazide was gently ground in an agate mortar. Small portions of the material were dusted on top of a flat plate low-background Si single crystal specimen holder. Powder diffraction patterns were registered at room temperature on a Bruker D8 ADVANCE diffractometer working in the Bragg-Brentano geometry using CuKa radiation, operating at 40 kV and 30 mA, equipped with a LynxEye position-sensitive detector. The pattern used in the structure determination was recorded from 5.007 to 60.006° (2θ) in steps of 0.01526°, at 1.5 sec/step. The standard instrument settings (Ni filter of 0.02 mm, Soller slits of 2.5°, Divergence Slit of 0.2 mm, scatter screen height of 3 mm) were employed. Geometry. Bond distances, angles etc. have been calculated using the rounded fractional coordinates. All su's are estimated from the variances of the (full) variance-covariance matrix. The cell esds are taken into account in the estimation of distances, angles and torsion angles