Hydrogen bonds and π–π interactions in two new crystalline phases of methylene blue

Two unprecedented solid phases of methylene blue (MB +), viz. 3,7-bis(dimethylamino)phenothiazin-5-ium chloride dihydrate and 3,7-bis(dimethylamino)phenothiazin-5-ium bisulfite, have been obtained and structurally characterized. The effective absence of hydrogen-bond donors in the second compound has important consequences on the stacking geometry and supramolecular interactions of the MB+ ions, which are analysed by Hirshfeld fingerprint plots.


Chemical context
The 3,7-bis(dimethylamino)phenothiazin-5-ium ion, better known as methylene blue cation (MB + ), is a renowned compound with important applications in medicine (Hanzlik, 1933;Wendel, 1935;Wischik et al., 1996), biology (Jung & Metzger, 2013;Fä rber et al., 1998) and chemistry (Bergamonti et al., 2015;Kim et al., 2014). MB + , with formula C 16 H 18 N 3 S + , consists of three condensed six-membered rings with two heteroatoms in the central one, and two terminal dimethylamine groups. The delocalization of the +1 charge, which involves the whole molecule with the exception of the four peripheral methyl groups, causes an overall planarity and the typical intense blue colour exhibited by MB + solutions in many solvents. The formal resonant structures are shown in the Scheme.
The MB + chloride salt is the first fully synthetic drug to be used in medicine, originally as an antimalarial agent (Coulibaly et al., 2009), an antidepressant (Erog lu & Ç ag layan, 1997), an antihemoglobinemic (Cawein et al., 1964) and as a disinfectant (Lo et al., 2014). In chemistry, it has various colourimetric and photocatalytic uses (Hang & Brindley, 1970;Kim et al., 2014;Bergamonti et al., 2015), which rely on its capability of undergoing a reduction process in the presence of weak reducing agents, turning into the colourless leukomethylene blue. The latter, in turn, can be oxidized to restore the original MB + cation, and this feature makes it a valid redox agent in biochemistry where it plays relevant roles in the study of enzyme-catalysed redox reactions. Recently, despite the cationic nature of MB + , we found that its peculiar electronic ISSN 2056-9890 situation enables ligand behaviour towards MCl 2 fragments (M = Cu and Ag) through the central aromatic nitrogen atom (Canossa et al., 2017), thus proving that some properties of this common and widespread molecule are still to be discovered.
Commercial MB is a pentahydrate chloride salt, whose structure was reported in 1973 (Marr et al., 1973). Recently, Rager et al. (2012) reinvestigated its crystalline states at variable temperatures, which led to the observation of five different hydrates with clearly distinct structures, as shown by powder X-ray diffraction analyses. However, no structural data are available and, to date, only the structure of the commercial pentahydrate form is known. Herein, we report and discuss the molecular and crystal structures of the unreported dihydrate phase of MB + chloride (I), one of those predicted by Rager et al. (2012), and the crystal structure of a new anhydrous form of MB + bisulfite (II).

Structural commentary
The molecular structures of compounds (I) and (II) are illustrated in Figs. 1 and 2, respectively. Details of the hydrogen bonding in the crystals of compounds (I) and (II) are given in Tables 1 and 2, respectively. In compound (I), the asymmetric unit is composed of one MB + cation, a chloride anion, and two water molecules. The latter are linked head-totail by O-HÁ Á ÁO hydrogen bonds which, in turn, are linked by O-HÁ Á ÁCl hydrogen bonds, forming chains propagating along [001], as shown in Fig. 1 (see also Table 1). The asymmetric unit of compound (II) consists of an MB + cation and a bisulfite anion. In both compounds, the MB + cations display a typical resonance structure, as evidenced by the values of the C-C bond lengths in the rings, which range from 1.352 (3) to 1.447 (5) Å . This bond-length distribution range is the same as that observed in other reported structures containing MB + cations, for example, as for its chloride pentahydrate form (Marr et al., 1973). The two C-S bond lengths, S1-C7 and S1-C9 [respectively, 1.731 (4) and 1.734 (4) Å in (I) and 1.732 (2) and 1.727 (2) Å in (II)], are very similar and in agreement with analogous data reported in the literature. The MB + cations are planar considering the three condensed sixmembered rings [atoms S1/N1/C3-C14; r.m.s. deviations are 0.011 Å for (I) and 0.01 Å for (II)] and the external dimethylamine groups, with the only exception being the aliphatic hydrogen atoms. In compound (II), one of the four  The molecular structure of compound (I), with the atom labelling. Displacement ellipsoids are drawn at the 50% probability level. Hydrogen bonds are shown as dashed lines (see Table 1).

Figure 2
The molecular structure of compound (II), with the atom labelling. Displacement ellipsoids are drawn at the 50% probability level. Hydrogen bonds are shown as dashed lines (see Table 2).
S-O bond lengths of the bisulfite anion [S2-O1 = 1.575 (3) Å ] is longer than the other three, which vary from 1.439 (2) to 1.468 (2) Å , thus confirming the identity of the OH group in this anion. The anions are linked by a pair of O-HÁ Á ÁO hydrogen bonds forming an inversion dimer ( Fig. 2 and Table 2).

Supramolecular features
In the crystal packing of the two compounds, illustrated in Figs. 3 and 4, the planar MB + cations are stacked in an antiparallel mode, with the sulfur atom disposed alternatively on opposite sides. The aromatic systems exhibit offsetinteractions and form infinite layers as shown in Figs. 5 and 6. The average interplanar distances are 3.326 (4) Å in (I) and 3.550 (3) Å in (II). This disposition differs from the one observed in the pentahydrate form where the MB + species are stacked together while adopting the same orientation, so that the sulfur atoms of all of the molecules lie on the same side along the stacking column. Moreover, as evidenced in Fig. 5iiiii and Fig. 6ii-iii, the stacking geometry of MB + differs significantly in the two phases. In fact, in the case of (I), the antiparallel mode is accompanied by a mutual shift of the cations, resulting in the formation of a zigzag chain with an inter-centroid distance between central thiazine rings of 3.734 (3) Å (Fig. 5iii). On the other hand, in (II) the stacked molecules are almost eclipsed and the equivalent intercentroid distances are 3.912 (4) and 3.956 (5) Å (Fig. 6iii).

Hirshfeld surface analysis
An evaluation of the Hirshfeld fingerprint plots (Spackman & Jayatilaka, 2009)  The crystal packing of compound (I) viewed along the c axis, with the unit cell highlighted in the upper left-hand corner. Displacement ellipsoids are drawn at the 50% probability level.
stacking. The first involves MB + as a donor by means of aromatic and aliphatic C-H bonds (Table 1), and as an acceptor by means of the central N atom, whose lone pair pointing out of the molecule is readily exploited by a water molecule to form a strong hydrogen bond [NÁ Á ÁO distance = 2.936 (4) Å ; see Fig. 1 and Table 1). The presence of hydrogenbond donors surrounding MB + , i.e. water molecules, is therefore able to satisfy the region of the cation with the most prominent partial negative charge (the nitrogen atom).
On the other hand, considering the fingerprint plot of compound (II), it can be seen that the strongest interactions arestacking and C-HÁ Á ÁO contacts (Table 2) between MB + and the oxygen atoms of the bisulfite inversion dimer. Since no available hydrogen-bond donor is present near MB + , no interaction is able to exploit the electron density concentrated on the central N atom. This has important consequences, since, on one side, it allows a better alignment of the MB + cations in their stacking arrangement, as clearly shown in Fig. 6. However, although there is an improved geometrical match, the stacking distance increases as a consequence of the charge repulsion between the mono-cationic molecules.
This evidence constitutes an exception to a general trend in the packing preferences of organic species. Indeed, in cases where both hydrogen bonds and stacking can be found in the solid phase, the two interactions compete to maximize their efficiency. This competition is usually in favour of the more directional supramolecular interactions, i.e. hydrogen bonds (Gospodinova & Tomšík, 2015). In the present case, however, the cationic nature of the aromatic molecules does not favour the stacking disposition that is usually better (in energetic terms), and in the case where there are no strong hydrogen bonds involving MB + , as in compound (II), the molecule is able to adopt a theoretically more stabilizing stacking geometry, which in this case is a less stabilizing one.

Database survey
In the Cambridge Structural Database (CSD, version 5.38, last updated May 2017; Groom et al., 2016) the crystal structure of the 3,7-bis(dimethylamino)phenothiazin-5-ium hydrogen sulfate dihydrate can be found as a Private Communication (XUVROW; Lynch, 2009  Views of the stacking geometry of MB + in compound (I): (i) displayed orthogonally to the stacking pillar axis by showing a tetramer of stacked molecules; (ii) the same group of MB + cations is shown along the stacking direction; (iii) view along the MB + longer dimension, highlighting the mutual shifts of the cations in the zigzag columns.

Figure 6
Views of the stacking geometry of MB + in compound (II): (i) displayed orthogonally to the stacking pillar axis by showing a tetramer of stacked molecules; (ii) the same group of MB + cations is shown along the stacking direction; (iii) view along the MB + longer dimension, highlighting the nearly completely eclipsed superposition of the cations in the antiparallel columns.
the H atom of the inorganic moiety, nor those of the water molecules, because of the poor data quality due to problematic twinning affecting the solid phase. Considering the overall crystal packing of this phase, which features MB + as both a hydrogen-bond donor and acceptor towards the water molecules and the anion, the interactions of the organic cation are much more similar to those observed for compound (I), than those observed for compound (II).
A search of the CSD found 30 compounds containing the aromatic unit 3,7-bis(dimethylamino)phenothiazin-5-ium cation. The anions present in the crystal structures include inorganic halogenide, nitrate, perchlorate, thiocyanate, triiodide, hydrogen sulfate and different metallates. The geometrical parameters of the cations (bond lengths, bond angles and torsion angles) are in the normal range for condensed ring systems.

Preparation of compound (I)
For the crystallization of compound (I), the commercial reagent 3,7-bis(dimethylamino)phenothiazin-5-ium chloride was used without any preparative treatment. 50 mg of 3,7bis(dimethylamino)phenothiazin-5-ium chloride pentahydrate (0.156 mmol) were transferred to a 10 ml glass vial containing 5 ml of dichloromethane. The container was then closed and placed in an ultrasound bath for 5 min. to reach the saturation limit of the compound. The mixture thus obtained was filtered into another 5 ml glass vial, and the resulting solution was left partially open for slow evaporation of the solvent. After 24 h, metallic dark-green needle-shaped crystals of compound (I), suitable for X-ray diffraction analysis, were obtained.
Preparation of compound (II) 306 mg of 3,7-bis(dimethylamino)phenothiazin-5-ium chloride pentahydrate (0.957 mmol) were transferred to an agate mortar, together with 284 mg of HgSO 4 (0.957 mmol). The two powders were subsequently mixed and ground for 30 min, resulting in a dark-green powder. X-ray powder diffraction analysis was performed on the as-obtained product. The resulting pattern showed peaks clearly belonging to the final compound (II) (see Fig. S1 in the supporting information). An excess of the powder was then placed in a glass vial, together with 3 ml of N,N-dimethylformamide. The container was closed and placed in an ultrasound bath for 5 min. to reach the saturation limit of the compound. The mixture obtained was filtered into another 5 ml glass vial, and the resulting solution was left partially open for slow evaporation of the solvent. After one week, metallic dark-green needlelike crystals of compound (II), suitable for X-ray diffraction analyses, were obtained.

Refinement details
Crystal data, data collection and structure refinement details are summarized in Table 3. For both compounds, the H atoms were positioned geometrically and refined using a riding model: C-H = 0.99 Å with U iso (H) = 1.2U eq (C). The H atoms of the water molecules in (I) and the bisulfite anion in (II) were located in difference-Fourier maps and refined freely. Compound (I) was refined as a merohedral twin with twin matrix, 1 0 0, 0 1 0, 0 0 1, with a refined BASF value of 0.185 (3).
Diffraction data for compound (I) were collected using a Bruker D8 Venture diffractometer, equipped with a CMOS PhotonII detector, a Mo High brilliance microsource (Incoatec) working at 50 KV and 1 mA. For compound (II), the data were collected at the ELETTRA Synchrotron facility (CNR Trieste) using monochromated 0.7 Å wavelength radiation and a Pilatus 2M Detector (Dectris).

3,7-Bis(dimethylamino)-phenothiazin-5-ium chloride dihydrate (I)
Crystal data Extinction correction: SHELXL2014 (Sheldrick, 2015b), Fc * =kFc[1+0.001xFc 2 λ 3 /sin(2θ)] -1/4 Extinction coefficient: 0.068 (12) 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. Refinement. Refined as a 2-component twin.