Previous studies have shown that the global energy minimum of a fully flexible [bis(pyridine)iodine(I)]⁺ halogen bond complex is centrosymmetric with a nitrogen-nitrogen distance (r_NN) of 4.51 Å.²² The symmetric halogen bond geometry is also preferred in slightly strained systems, such as 2 (Figure 1, middle), allowing r_NN ≤ 4.88 Å, whereas above this distance the asymmetric [N–I⋯N]⁺ arrangement becomes energetically most favorable. The preference of halonium ions for forming symmetric over asymmetric halogen bonds has been demonstrated in a geometrically restricted bis(pyridine) ligand 3, (Figure 1, bottom) that enforces a long, r_NN = 6.1 Å, nitrogen-nitrogen distance.²² This ligand prefers to form a dimer with two symmetric three-center halogen bonds over the monomeric complex with asymmetric halogen bonds. The former is an entropically disfavored dimer that has enthalpically favored strong and symmetric halogen bonds. In contrast, the latter, is an entropically favored monomer that has enthalpically disfavored weaker, asymmetric halogen bonds. The halogen bonds of conformationally flexible ligands that may allow symmetric three-center halogen bond (r_NN ≤ 4.88 Å) complexes to form only at the cost of significant strain have not yet been investigated. Such complexes balance entropic and enthalpic factors, and may accordingly either form monomeric species with a strained three-center halogen bond or higher-order aggregates possessing strong, symmetric three-center halogen bonds. The geometry of the halogen bond complex of such a ligand is expected to be inherently challenging to predict due to the delicate (im)balance of entropic and enthalpic contributions determining its preferred geometry. Herein, we investigate ligand 1 which has r_NN ∼ 4.7 Å in its monomeric [N⋯I⋯N]⁺ complex, allowing formation of a slightly destabilized yet symmetric halogen bond. As the core of 1 is more adaptable than the previously studied bidentate bis(pyridine)-type donors,^15,22 it may allow for the formation of not just monomeric, but also dimeric or polymeric halogen bond assemblies. Thus, energetically favorable short and symmetric three-center [N⋯I⋯N]⁺ bonds might compensate for the unfavorable entropic consequence of the dimer or oligomer formation.

The properties of three-center, four-electron halogen bonds are typically compared to analogous [D⋯H⋯D]⁺ hydrogen and [D⋯Ag⋯D]⁺ metal coordination bonds.^11,12,23 The former prefers an asymmetric geometry, with a distinct covalent and a distinct secondary bond, while the latter is centrosymmetric. Three-center halogen bonds have occasionally been referred to as coordinative bonds.²⁴ However, halogen bonds of halonium ions differ from metal coordinative bonds concerning their solvation and counterion coordination,^25,26 which leads to important structural differences, for example in supramolecular assemblies.¹⁷ Based on this knowledge, comparison of the silver(I) and iodine(I) complexes of 1 is expected to provide deeper understanding of three-center complexes.

Bis(pyridine) complexes of iodine(I), bromine(I) and chlorine(I) have been demonstrated to exist in comparable geometries and have analogous properties, despite the altering extent of electrostatic and charge-transfer components of their halogen bonds.²⁷ Accordingly, most fundamental studies and supramolecular applications so far have focused on the investigation of iodine(I) complexes in order to draw general conclusions about the halogen bonds of halonium ions.^{11,15–20,22,24–27} Moreover, the three-center [N⋯Ag⋯N]⁺ bond of silver(I) in the bis(pyridine) complex has been shown to be similar to the analogous Au(I) complex in solution, the solid state and in silico.²³ The analysis of statistical trends from X-ray crystallographic observations, using data obtained from the Cambridge Structural Database (CSD) indicated that the bis(pyridine) complexes of other transition metals, such as Hg(II), Cd(II), Te(III), Er(III), Zn(II), Gd(III), Mn(II), Fe(II), Ni(II), Cr(II), and Rh(I) also form linear and symmetric bis(pyridine)metal(I-III) three-center complexes with comparable geometries.²³ Accordingly, the analyses of iodine(I) and silver(I) complexes are expected to provide general conclusions to compare the bonding of halonium and transition metal complexes. Previous work suggests^{11,12,25–27} that the main origin of the difference between the behavior of the halogen and transition metal complexes is the availability of empty d-orbitals in the latter. This allows coordination of Lewis bases to the metal center. In contrast, the s²(d¹⁰)p⁴ electron configuration of halogens makes them inert for coordination of Lewis bases, such as counter anions and solvents.

We have computationally investigated the potential [N⋯X⋯N]⁺ complexes of ligand 1 (see ESI for details). Our main goal was to analyze the structure and relative stability of the monomeric and dimeric forms of these complexes in solution phase. The counter anions were not included in our calculations, as we only model complexes with weakly coordinating anions. Computations predict that dimeric structures for both silver(I) and iodine(I) complexes of 1 are favored thermodynamically as compared to the monomeric structures (Figure 2). The closed (cl) forms of monomeric complexes possessing symmetric three-center, four-electron [N⋯X⋯N]⁺ bonds (1-Ag-cl and 1-I-cl) are strained, as indicated by their distorted binaphthyl framework and the increased r_XN bond distances. The open (op) structure of monomeric iodine(I) complex (1-I-op) is predicted to be slightly more favored over 1-I-cl, supporting the strained nature of 1-I-cl. For silver(I), the open monomeric form (1-Ag-op) involves Ag⁺-arene coordination, however, this structure is computed to be less stable than 1-Ag-cl. Two types of dimeric species were identified computationally for both 1-Ag and 1-I. The A-type structures are characterized by large separation of X⁺ ions, whereas the cations are much closer in the B-type forms. The [N⋯X⋯N]⁺ unit is symmetrical in these dimeric forms and their structures are analogous to those identified computationally for iodine(I) complexes of a phenanthrene-based ligand.²²

Figure 2. Computed monomeric and dimeric structures of a) 1-Ag and b) 1-I. Relative stabilities (in terms of solution phase Gibbs free energies) with respect to the most stable monomeric form are shown in parentheses (in kcal/mol). For dimeric species the values are given per monomer. Selected bond lengths (r_XN and r_XC) are given in Å. Hydrogen atoms are omitted for clarity.

3,3′-di(pyridin-2-yl)-1,1′-binaphthalene 1 was prepared following the synthetic route shown in Scheme 1 (see ESI for details). Compound 4 was synthesized by addition of bromine to a CCl₄ solution of naphthalene while irradiating the reaction mixture with a 1000 W Xe ARC lamp. Compound 4 was dehydrobrominated using potassium tert-butoxide. Compound 5 was then converted to 6 via Stille coupling with 2-(tributylstannyl)pyridine. Compound 7 was synthesized from 6 by a palladium(II)-catalysed reaction with bis(pinacolato)diboron. Suzuki coupling between 6 and 7 yielded ligand 1. Ligand 1 was converted into silver(I) coordination complexes by addition of Ag(I) salts to CH₂Cl₂ solutions of 1, yielding 1-Ag-BF₄, 1-Ag-PF₆, 1-Ag-OTf, 1-Ag-CO₂CF₃, and 1-Ag-OTs. Formation of an [N⋯Ag⋯N]⁺ coordination complex was confirmed by the Δδ¹⁵N_coord coordination shift of 44 ppm (1-Ag-OTf). Halogen bond complexes were prepared following a reported procedure,¹⁵ by addition of I₂ into a CD₃CN or CD₂Cl₂ solution of the Ag(I) complex of ligand 1, yielding the complexes 1-I-BF₄, 1-I-PF₆, 1-I-OTf, 1-I-CO₂CF₃, and 1-I-OTs. The formation of [N⋯I⋯N]⁺ complexes was confirmed by the Δδ¹⁵N_coord of 100–105 ppm, which is comparable to the coordination shifts reported for analogous complexes.^15,16,22,26

Scheme 1. Synthetic route to 3,3′-di(pyridin-2-yl)-1,1′-binaphthalene 1. Reagents and conditions: (a) Br₂, CCl₄, hv, 2.5 h; (b) K-tBuO, THF, r.t., overnight; (c) 2-(tributylstannyl)pyridine, PdCl₂(PPh₃)₂, DMF, µW, 120 °C, 40 min; (d) bis(pinacolato)diboron, Pd(dppf)Cl₂, KOAc, DMF, 80 °C, overnight; (e) Pd(PPh₃)₄, Cs₂CO₃, 1,4-dioxane/H₂O, 80 °C, 2 d.

To further elucidate the composition of the 1-Ag complexes with different anions, single crystals suitable for X-ray diffraction studies were obtained via slow diffusion of hexane into CH₂Cl₂ solutions of 1 with Ag-PF₆ (CCDC 2025121), Ag-CO₂CF₃ (CCDC 2025119), and Ag-OTs (CCDC 2025120). The three different complexes formed in the solid state highlight distinct anion coordination to the silver(I) atoms which, when paired with the conformational flexibility of 1, produce different dimeric complexes and a coordination polymer.

1-Ag-CO₂CF₃ crystallizes in the P2₁/n space group with two silver(I) complexes in the asymmetric unit. Each complex consists of a collapsed dimer that incorporates two silver(I) ions and two molecules of 1 (Figure 3a). The dimer resembles the A-type dimers described in the theoretical discussion. Interestingly, the collapsed dimer allows facial π-stacking between the naphthalene ring of one ligand and the pyridine ring of the other ligand. One main difference between the X-ray and computed structures is the presence of anions. In the X-ray structure, each silver(I) ion has a pseudo-trigonal planar coordination environment consisting of one nitrogen from each ligand and one oxygen from a CO₂CF₃⁻ anion. We note that the X-ray structure is well-reproduced by the calculations, if the CO₂CF₃⁻ anions are involved in the molecular models (see ESI for details). The anion is a monodentate donor and does not bridge silver(I) ions, which is in contrast with the PF₄⁻ and OTs⁻ containing structures described below.

Figure 3. Single crystal X-ray structures 1-Ag-CO₂CF₃ (a), 1-Ag-OTs (b) and of 1-Ag-PF₄ (c), highlighting the different silver(I) clusters that are formed depending on the counterion present. Carbons are colored gray, nitrogens dark blue, oxygens red, fluorines light blue, phosphorus orange, sulfur yellow and silver atoms are green (hydrogens not shown for clarity).

1-Ag-OTs crystallizes in the P-1 space group with two silver(I) complexes in the asymmetric unit. Here, the complex is a dimer consisting of two molecules of 1 and a cluster of four silver(I) ions bridged by four OTs⁻ anions (Figure 3b). The four nitrogen atoms, from two ligands of 1, coordinate to different silver(I) centers. Each of the silver(I) ions has a pseudo trigonal pyramidal coordination geometry occupied by three separate oxygens from three OTs⁻ anions and one nitrogen from ligand 1. The four OTs⁻ anions utilize one, two or three oxygens to coordinate to a single silver(I) ion or bridge between two or three different silver(I) ions.

1-Ag-PF₄ crystallizes in the P2₁/n space group as a coordination polymer. Interestingly, even though AgPF₆ was used in the crystallization process, electron density corresponding to a tetrahedral anion was observed and was best modeled as PF₄⁻. The coordination polymer is comprised of di-nuclear silver(I) complexes linked in a linear fashion by two bridging anions. Each complex is comprised of a dimer of 1 and the silver(I) centers are coordinated by two pyridine nitrogen atoms from different molecules of 1. One fluorine atom from two separate anions also coordinates to the silver centers resulting in a tetrahedral coordination geometry around each silver(I). Each anion also coordinates to an adjacent silver(I) complex linking the coordination polymer in the solid state.

The crystal structures highlight that 1 forms different dimeric complexes with silver(I) depending on the counter anion that is present. All three silver(I) complexes in the crystal structures deviate from the DFT calculations presented herein. Clearly, the anion plays an important role in the speciation of the silver(I) complexes. The anions were not computationally modeled in our present work. Unfortunately, due to their low stability, no crystals suitable for X-ray diffraction were obtained for the 1-I complexes. Thus, we turned to NMR spectroscopy to further elucidate the differences between the 1-Ag and 1-I complexes.

The ¹H NMR spectra of the five silver(I) complexes, obtained in CD₂Cl₂ and shown in Figure 4, were broadened to various degrees depending on the coordination strength of the anion.²⁸ The spectra of complexes possessing strongly coordinating anions, such as CO₂CF₃⁻, OTs⁻, and OTf⁻, were not as broad compared to those of the weakly coordinating BF₄⁻ and PF₆⁻ complexes. The differences in the spectra could be explained by the strongly coordinating anions forming one well-defined geometry for their silver(I) complexes, whereas the weakly coordinating anions promote a dynamic mixture of several geometries. Alternatively, the weakly coordinating anions could promote the formation of larger aggregates, as the empty silver(I) coordination sites may allow for silver(I)-silver(I) interactions, which are known in the literature.²⁹ The direct coordination of strongly coordinating anions has previously also been reported for analogous [bis(pyridine)silver(I)]⁺ complexes.²⁶ The silver-silver interactions are further corroborated by the observation of several sets of signals at low temperature for the 1-Ag-BF₄ complex (Figure S41, ESI). The flexibility of ligand 1 makes it possible to form silver(I) complexes with various geometries as compared to the conformationally restricted complexes reported previously.²⁶

Figure 4. Aromatic region of the ¹H NMR spectrum of (a) 1-Ag-CO₂CF₃, (b) 1-Ag-OTs, (c) 1-Ag-OTf, (d) 1-Ag-PF₆, (e) 1-Ag-BF₄ and (f) 1 in CD₂Cl₂.

To assess the size of the complexes in solution, translational diffusion rates were acquired using ¹H NMR detection, following previous protocols.^15,26 The translational diffusion coefficients of 1-Ag-CO₂CF₃ (D = 7.49 × 10⁻¹⁰ m²/s), 1-Ag-OTf (D = 7.00 × 10⁻¹⁰ m²/s), and 1-Ag-OTs (D = 7.20 × 10⁻¹⁰ m²/s) were significantly lower than that of ligand 1 (D = 11.21 × 10⁻¹⁰ m²/s), indicating that these have 1.6-times larger molecular radii. This is in agreement with the dimeric geometry predicted for these complexes by DFT. The dimers were further supported by high-resolution mass spectrometry providing the m/z signal corresponding to complex [1₂ · Ag₂]²⁺ for 1-Ag-BF₄, 1-Ag-PF₆, 1-Ag-OTf, 1-Ag-CO₂CF₃, and 1-Ag-OTs (ESI). Hence, the solution structure of the 1-Ag complexes shows a counterion dependent dynamic behavior, and are in agreement with the speciation computed by DFT and observed by X-ray.

¹H NMR signals for all iodine(I)-centered complexes were slightly broadened, as compared to ligand 1 (Figure 5). ¹H DOSY was measured to assess the size of 1-I complexes in solution. The translational diffusion coefficients D = 10.67 × 10⁻¹⁰ m²/s (1-I-BF₄), D = 10.51 × 10⁻¹⁰ m²/s (1-I-PF₆), D = 11.09 × 10⁻¹⁰ m²/s (1-I-OTf), D = 10.86 × 10⁻¹⁰ m²/s (1-I-OTs), and D = 10.65 × 10⁻¹⁰ m²/s (1-I-CO₂CF₃) were comparable to that of ligand 1, suggesting that these complexes are monomeric. Neither ¹H NMR nor diffusion NMR indicates a counterion dependence of the structure of these 1-I complexes, which is in agreement with previous observations that the halonium ions of three-center [N⋯I⋯N]⁺ complexes do not coordinate anions.²⁶

Figure 5. The aromatic region of the ¹H NMR spectrum of (a) 1-I-CO₂CF₃, (b) 1-I-OTs, (c) 1-I-OTf, (d) 1-I-PF₆ and (e) 1-I-BF₄ in CD₃CN.