Comment

Metal complexes incorporating benzimidazole derivatives may mimic the behaviour of metal-ion sites in biological systems, in both structure and reactivity (Alagna et al., 1984; Rijn et al., 1987), and this fact has rendered their study increasingly attractive. One such derivative, namely 2,6-bis(benzimidazol-2-yl)pyridine (BzimpyH₂), is a potentially active ligand which binds through one pyridine and two benzimidazole N atoms in a typical tridentate mode (a comprehensive review has been published recently; Boca et al., 2011). In particular, a common pattern has two tridentate ligands bound to a transition metal cation (Tr), with the planar ligands at right angles to each other, thus shielding the cation from interaction with other species.

In these molecules, the ligand can appear as the neutral unit (BzimpyH₂), with both uncoordinated imidazole N atoms protonated, in which case there is a counter-ion balancing the [Tr(BzimpyH₂)₂]²⁺ charge. Many structures of this sort appear in Version 5.33 of the Cambridge Structural Database (CSD; Allen, 2002), viz. DURWOJ (Huang et al., 2010) and DURWOJ01 (Wu, Huang, Yuan, Kou, Chen et al., 2010) for Ni^II, EYINAB (Harvey et al., 2004) for Zn^II, NETBUJ (Boca et al., 1997) and PAFZIF (Ruttimann et al., 1992) for Fe^II, and WUXBUN (Yan et al., 2010), EZEXOX (Wu, Huang, Yuan, Kou, Jia et al., 2010), OYAKEF (Guo et al., 2011) and BAHJOL (Wu et al., 2011) for Mn^II. There are also a number of complexes in which one of these H atoms is lost, giving a monoanion (hereinafter BzimpyH) which forms neutral Tr(BzimpyH)₂ units, viz. PANXAE (Shi et al., 2003), PANXAE01 (Bai & Zhang, 2009) and TAWZOG (Rajan et al., 1996) for Mn^II, TIBGUH (Zhang et al., 2007) for Co^II, WICJOH (Wang et al., 1994) and WICJOH01 (Yue et al., 2006) for Cd^II (see footnote¹), and EJEBOK (Harvey et al., 2003) and EJEBOK01 (Yue et al., 2006) for Zn^II (see footnote¹).

We present here the structure of the title complex, Zn(BzimpyH)₂, (I), where the ligand displays the latter behaviour. The compound appeared serendipitously in tiny amounts as a by-product of the frustrated synthesis of a Zn + BzimpyH₂ + tetrathionate complex (see Experimental). In addition to (I), the same crystallization batch produced a second, also unexpected, compound which proved to be a known polymorph of (I) [CSD refcode EJEBOK (Harvey et al., 2003), (II)], which presents a number of noteworthy similarities to (I) but some interesting differences as well.

Compound (I) crystallizes in the tetragonal space group P4₃2₁2 (No. 96), while (II) crystallizes in P4₂2₁2 (No. 94), although the c axis of (I) is doubled with respect to that of (II). The point group (422) is the same. There is a clear group-subgroup relationship, as P4₃2₁2 (c' = 2c) is a maximal non-isomorphic subgroup of P4₂2₁2. Unfortunately, the scant amount of material obtained precluded any serious attempt to detect any potential phase transition linking the two structures.

Table 1 presents a comparison of significant parameters in (I) and (II), while the slight differences introduced into the structure by symmetry relaxation will be presented below.

The structural building block in (I) is a Zn(BzimpyH)₂ monomer (Fig. 1) lying on a single twofold axis which traverses the Zn^II cation and relates the two N,N',N''-tridentate BzimpyH^- anions; thus, half of the molecule is independent. In the previously reported structure of (II), the monomer is bisected by a second independent twofold axis, passing through Zn^II but also bisecting the BzimpyH^- anion, thus rendering just one quarter of the monomer independent. In addition, in (II), there is a third symmetry-required twofold axis perpendicular to the other two diads. The symmetry differences between the two structures can be seen in Fig. 2, which shows a schematic representation of the symmetry elements at the origin in both space groups, where the molecules lie.

The BzimpyH^- anion in (I) is nearly planar, with a mean deviation of 0.063 (2) Å (maximum deviation for atom N5 of 0.1684 Å); the dihedral angle between the mean planes of the symmetry-related ligands is 75.7 (2)°, compared with an angle of 75.4 (3)° for (II). The similarities - metric as well as orientational - can be seen in Fig. 3, which shows an overlay of (I) and (II), with neither least-squares fitting nor rotations having been performed and with their relative original orientations in the unit cells preserved. The almost perfect overlap is apparent, with a mean unweighted deviation of 0.14 (8) Å for all atoms.

The double tridentate bite with five-membered chelate rings imposes a distorted geometry on the Zn coordination octahedron in (I), with `cis' N-Zn-N angles spanning the broad range 74.93 (7)-107.91 (7)° and `trans' angles spanning the range 141.35 (15)-173.98 (9)°. The strain in the ligand due to the triple (N,N',N'') bite is evidenced by the N1N5 distance [4.220 (4) Å], which is significantly shorter than those reported for three (unstrained) free BzimpyH₂ entities (Freire et al., 2003), which have a range of 4.550 (3)-4.580 (3) Å. Comparable values were observed for (II).

The Zn-N coordination distances also show the effect of symmetry relaxation (Table 1). Those in (II) are divided into two groups: Zn-N_central and Zn-N_lateral. In (I), a very similar Zn-N_central value is found, but the fourfold degeneracy of Zn-N_lateral is broken, splitting into two groups. It is interesting to note that the average of these latter bond distances [2.1775 (14) Å] agrees fairly well with those in (II) [2.181 (3) Å].

The symmetry restrictions on the disordered imidazole N-H groups impose differences on the pattern of protonation. In the case of (II), the two N atoms per ligand which can be protonated are related by symmetry, so H-atom occupancy is forced to be 0.5 per N atom to give a total charge of -1 per ligand. In the case of (I), there are two independent N atoms to accommodate one or two H-atom sites in such a way that their populations sum to 1. In order to check for differences, F syntheses were plotted in an orientation suitable for viewing the electron density in the neighbourhood of the imidazole N atoms (Fig. 4). The expected symmetric distribution in (II) contrasts with the asymmetric pattern in (I), notably biased towards atom N4. When allowed to refine, the occupancies reflected these results [0.59 (3) and 0.41 (3) for atoms N4 and N2, respectively]. These different disorder patterns for the imidazole H atoms are linked to the internal symmetry and surroundings of the molecule. There are examples in the literature (CSD refcode WICJOH01; Yue et al., 2006) of Tr analogues with the monomers lying on general positions for which there is no disorder in the N-H groups, with one of the two imidazole N atoms fully protonated and the second `naked' and acting as a hydrogen-bond acceptor. This leads to an ordered distribution of hydrogen bonds in space, defining a homogeneous three-dimensional hydrogen-bonded structure.

Entries 1 and 2 in Table 2 reflect the two different ways in which the disordered hydrogen bond in (I) is formed. The first entry corresponds to the major fraction, with the H atom linked to N4, while the second, minor, component has the H atom attached to N2. This contact links monomers in two (not three) directions parallel to the tetragonal base, to form broad two-dimensional nets on (001). Fig. 5(a) shows a packing view of one of these nets, while Fig. 5(b) presents a perpendicular view showing the way in which these planes stack. Interplanar interactions consist of much weaker C-H interactions (Table 2, entries 3 and 4). No - bonds linking aromatic groups are present in the structure, the rings being too far apart to have any kind of interaction.

A final difference observed between (I) and (II) is the enantiopurity revealed by the two refinements. While (II) refines with a Flack (1983) parameter of 0.48 (3), pointing to the presence of inversion twinning with almost equal populations of both absolute structures, (I) can be described as an almost enantiopure compound, with a Flack parameter of 0.087 (14).

As stated in the footnote¹, the analysis of a third Zn(BzimpyH)₂ polymorph (CSD refcode EJEBOK01) has been published, but the structure as reported presents serious formal errors which mitigate against its use for detailed comparison. However, the fact that there is an isomorphous Cd complex (refcode WICJOH01) reported in the same work and apparently error-free might suggest that the analogous Zn complex does in fact exist, possibly with space group Cc, and with its Zn cation on a general position. This would be a nonsymmetric Zn(BzimpyH)₂ unit, metrically similar but different in crystallographic symmetry from the two variants discussed here. Unfortunately, for the time being this is only speculative and this (potentially interesting) comparison must be postponed until better data are available.

Experimental

In a frustrated attempt to obtain zinc tetrathionate [the main final product happened to be Zn(BzimpyH₂)(acetate) monohydrate], tiny amounts of pyramidal crystals of the title compound, (I), and bipyramidal crystals of the previously published polymorph, (II), were obtained.

An aqueous solution of zinc acetate dihydrate and potassium tetrathionate was allowed to diffuse slowly into a solution of BzimpyH₂ in dimethylformamide (DMF), with all solutions equimolar (0.080 M). After the intial formation of a solid conglomerate, spontaneous dissolution occurred. When the process seemed to have finished, the diffusion system was disassembled and the resulting solution allowed to evaporate slowly. On standing (for about three weeks), three different phases were present in different amounts, viz. an overwhelming majority of the main product, Zn(BzimpyH₂)(C₂H₃O₂)₂·H₂O, and minor quantities of (I) and (II).

All H atoms were visible in a difference Fourier map. Those attached to C atoms were added at their expected positions (C-H = 0.93 Å) and allowed to ride. The single H atom of the BzimpyH^- anion was found to be distributed unequally over the two potential sites at the N atoms of different imidazole units. Their locations were further idealized and their occupancies refined to final values of 0.59 (3) and 0.41 (3). In all cases, H-atom displacement parameters were assigned as U_iso(H) = 1.2U_eq(host). Similar to what was observed for polymorph (II), where H-atom disorder was present, the outermost part of the pyridine group presents elongated displacement ellipsoids normal to the plane of the ring, due either to genuine vibration or to an uncharacterized disorder.

Data collection: CrysAlis PRO (Oxford Diffraction, 2009); cell refinement: CrysAlis PRO; data reduction: CrysAlis PRO; program(s) used to solve structure: SHELXS97 (Sheldrick, 2008); program(s) used to refine structure: SHELXL97 (Sheldrick, 2008); molecular graphics: SHELXTL (Sheldrick, 2008); software used to prepare material for publication: SHELXL97 and PLATON (Spek, 2009).

---

Supplementary data for this paper are available from the IUCr electronic archives (Reference: FA3291 ). Services for accessing these data are described at the back of the journal.

---