Coordination chemistry of anions through halogen-bonding interactions

While an IUPAC definition for hydrogen bonding was only released in 2011 after decades of discussions in the scientific community (Arunann et al., 2011), it did not take such a long time to come up with an analogous definition of halogen bonding (Desiraju et al., 2013), following a revival of this interaction in the literature, which can be traced back to the early 1990s. The identification of this interaction is, however, not recent, as illustrated by the isolation of the H₃N·I₂ adduct (Colin, 1814). Two centuries later, several review articles (Cavallo et al., 2016; Gilday et al., 2015; Fourmigué, 2009) have gathered together most available data. The halogen-bonding interaction (noted XB) is essentially described as an electrostatic interaction between a charge concentration (Lewis base) and a charge-depleted area, called an σ-hole, that a covalently bound halogen atom exhibits in the extension of this bond. Note that a charge-transfer covalent contribution can be found in the strongest halogen bonds, as illustrated for example in the triiodide anion following: I₂ + I⁻ → (I—I—I)⁻. The presence of a halogen bond is characterized by a shortening of the interatomic distance, relative to the sum of the van der Waals (or ionic) radii of the interacting atoms, which is defined as the reduction ratio. The halogen-bond donor character of halogenated molecules is modulated by the nature of the halogen with I > Br >> Cl >> F, by the hybridization of the substituted C atom with C≡C—I > C=CH—I > C—CH₂—I, and by the electron-withdrawing ability of the carbon substituents, with a strong activation in aliphatic and aromatic perfluorinated substrates such as F_{2n + 1}C_n—I or C₆F_6 − nI_n. Besides, dihalogens (I₂, Br₂, Cl₂), interhalogens (I—Cl, I—CN) and N-iodoimides (N-iodosuccinimide, N-iodosaccharin) are also strong halogen-bond donors. All Lewis bases are potentially XB acceptors and among them, anionic species, and particularly halide anions (Cl⁻, Br⁻, I⁻) thanks to a maximal charge concentration (Metrangolo et al., 2008a; Cavallo et al., 2010).

The paper by Szell et al. (2017) in this special issue of Acta Crystallographica Section B deals with such a crystal engineering approach through the formation of co-crystals associating ammonium or phosphonium halide salts with neutral, halogenated, activated halogen-bond donors such as sym-triiodotrifluorobenzene (1) or more specifically here its extended analog, 1,3,5-tris(iodoethynyl)-2,4,6-trifluorobenzene) (2). While ditopic halogen-bond donors such as diiodoacetylene (I—C≡C—I) or para-diiodotetrafluorobenzene most often lead to one-dimensional systems in the presence of halide anions, it was expected that threefold symmetric molecules such as (1) or (2) would provide higher symmetry structures (trigonal, hexagonal, cubic) that would express the molecular threefold symmetry of the halogen-bond donor.

Earlier reports (Metrangolo et al., 2008b) have indeed shown that `the mutual induced fitting process elicits the tridentate coordination profile of both 1,3,5-trifluoro-2,4,6-triiodobenzene (1) and I⁻ ions resulting in cation-templated anionic (6,3) networks', with the small cations (Me₃S⁺, Me₄P⁺, Et₄N⁺) sitting within the pores of the honeycomb network. With Et₄N⁺I⁻, Et₄P⁺I⁻ and solid solutions thereof, the adducts even crystallize in the trigonal space group , while with larger cations (Metrangolo et al., 2008b; Triguero et al., 2008), the (6,3) networks were maintained but strongly corrugated. In a search for larger pores, the extended halogen-bond donor (2) was recently designed (Lieffrig et al., 2013) and further used here with another set of cations of varying size and shape (Szell et al., 2017). Remarkably, besides the above-described honeycomb networks with tricoordinated halide anions (Fig. 1a), the adduct with the larger EtPh₃P⁺Br⁻ salt crystallizes in the cubic space group , with sixfold coordination around the bromide anion (Fig. 1c), as already observed with the smaller Et₃BuN⁺Br⁻ salt (Lieffrig et al., 2013) which crystallizes in the cubic space group . Furthermore, the XB interaction of acetonitrile acting as XB acceptor toward one iodine atom of (2) leaves only two I atoms (out of three) to interact with the halide anion, and leads to another set of structures with a four-coordinated bromide anion (Fig. 1b).

Figure 1 Detail of the coordination spheres of the bromide anion in three adducts of (2) with: (a) Et3BuN^+Br^−, (b) MePh3P^+Br^− and (c) EtPh3P^+Br^−.

These beautiful examples illustrate a charge-inverted coordination chemistry, where the metal cation is replaced by the halide anion and where the ligands are replaced by the halogen-bond donors. The essentially electrostatic nature of the XB interaction has important consequences on the coordination number and geometry. Indeed, in classical coordination chemistry, the overlap interactions of the ligand's orbitals with the nd, (n + 1)s and (n + 1)p orbitals of the metal center, together with the electron count, determines the favored coordination number and geometry. This does not hold in the present situation, where many examples exhibit very acute I⋯X⁻⋯I angles, as observed for example in the dicoordinated bromide anion (74°) in the one-dimensional chains formed from para-diiodotetrafluorobenzene adduct with Me₄N⁺Br⁻ (Grebe et al., 1999), or with the tricoordinated bromide anion in the honeycomb network formed from Bu₄N⁺Br⁻ and (2) (Lieffrig et al., 2013), where the sum of the three I⋯Br⁻⋯I angles does not exceed 246°. It appears that in all these systems, the sizeable dispersion forces associated with the indirect I⋯I contacts can favor these apparently surprising arrangements around the halide anion. Another remarkable point is also the coordination number around the halide anion, often found at 2 or 3, but easily extended to 6 as observed in the cubic structures described here (Szell et al., 2017; Lieffrig et al., 2013), and even to 8 in the co-crystal formed with CBr₄ as halogen-bond donor and Me₄N⁺Br⁻ (Lindeman et al., 2003). Note that in the latter and despite an apparent crowding around the bromide anion, the reduction ratio for the (Br₃C—)Br⋯Br⁻ distance still amounts to 0.88, indicating a sizeable halogen-bond interaction.

Last but not least, the paper by Szell et al. (2017) in this issue also combines these crystal structures analyses with complementary solid-state NMR investigations. The formation of the honeycomb network from Et₃BuN⁺Br⁻ and (2) is indeed associated with a clear and substantial shift (+18/35 p.p.m.) in the resonance of the carbon directly bonded to iodine, as a result of halogen-bond formation, in agreement with established trends in ¹³C shifts of aromatic ipso C atoms upon halogen bonding (Moss & Goroff, 2005; Viger-Gravel et al., 2013). Furthermore, ^35/37Cl and ^79/81Br solid-state NMR spectroscopy has also been used successfully to discriminate different coordination patterns of the Cl⁻ and Br⁻ anions, from approximately linear to distorted square planar to octahedral (Viger-Gravel et al., 2014). These last results illustrate how the halogen-bonding interaction, whose renewal in the 1990s was essentially rooted in crystal engineering studies, is now diffusing into all areas of chemistry (Cavallo et al., 2016), from theoretical and analytical investigations (in gas, liquid and solid states) to the optical, magnetic and conducting properties of crystalline materials, from soft matter issues (liquid crystals, polymers and gels) to biomolecular systems and their relevance to the pharmaceutical field.