Square planar silver(I) complexes: A rare but increasingly observed stereochemistry for silver(I)

Abstract

The square planar Ag(I) stereochemistry is generally acknowledged as rare, with only ∼2% of all reported silver complexes possessing this stereochemistry. Many researchers reporting such complexes often mistakenly believe that their example is one of only a handful of previously reported examples. This is despite the fact that there are currently around 65 well characterised complexes containing square planar Ag(I) ions, about half of which are coordination polymers. In this review, we critically examine each example and draw attention to trends that arise in their formation. The scope is limited to ‘traditional’ coordination complexes. Inorganic complexes containing extended mineral like structures, and complexes containing silver-π or silver-arene motifs are not considered in this review.

Introduction

In coordination chemistry, silver in the +1 oxidation state is found to adopt a wide variety of coordination geometries. A Cambridge Structural Database CSD (version 5.27) search yielded 3319 crystal structures containing silver ions coordinated to non-metal atoms [1]. Of these, 24.2% are two-coordinate, 22.7% are three-coordinate, 43.9% are four-coordinate, 4.7% are five-coordinate, 3.9% are six-coordinate, 0.3% are seven-coordinate, and 0.3% are eight-coordinate. This variety is in part due to the lack of stereochemical preference that arises from a d¹⁰ configuration. Furthermore, the weak nature of the silver-ligand bond [2] means that in the solid state, weak interactions and crystal packing forces have a greater influence on structure than may be expected in more robust metal–ligand systems. With the exception of the higher coordination numbers, the least common geometry is square planar, with around 65 of the 3319 reported crystal structures possessing this stereochemistry. This includes two different studies which report two [3] and three [4] isomorphous crystalline compounds, respectively, that have been characterised by X-ray crystallography. Several other crystal structures have analogues likely to be isomorphous based on spectral measurements. A comprehensive literature survey provided the examples presented herein. This survey included a CSD search where the coordination number of the Ag centre was fixed at four and bonds to donor atoms of any type were allowed. The trans-bond angles were specified to be in the range of 175 ± 5° and 165 ± 15°, respectively. These bond angle limits are sufficient to include occurrences where a reasonable degree of distortion towards tetrahedral geometry is present but the Ag(I) coordination geometry still has square planar character.

Structural methods such as X-ray crystallography generally provide an unambiguous assignment of the stereochemistry about a Ag(I) ion. However, the square planar stereochemistry has also been assigned in solution to Ag(I) ions from electronic spectral [5], [6] and electron spin-echo modulation studies [7].

With the rarity of the square planar geometry in Ag(I) chemistry, the examples which have been identified have always been of interest. More recently, with the development of supramolecular synthesis and the related rise of coordination polymer chemistry and crystal engineering, the prospect of employing this stereochemistry as a four-connected node or square planar secondary building unit (SBU) in the design of new network topologies has found some success which is likely to continue (Sections 5 Polymeric silver(I) moieties with no imposed symmetry, 6 Polymeric silver(I) moieties with crystallographically imposed symmetry).

Early workers who identified square planar Ag(I) complexes were often concerned to show that the Ag ion had not undergone oxidation to Ag(II) d⁹ for which a square planar geometry might be more likely. Typically ESR and UV–vis measurements were used in an attempt to probe the likely paramagnetic behaviour of such a Ag(II) square planar complex [8], [9], [10]. Improvements in analytical techniques and a better understanding of Ag(I) and Ag(II) coordination chemistry have reduced the ambiguity when determining oxidation numbers of Ag complexes.

The square planar Ag(I) stereochemistry is generally acknowledged as rare and many researchers reporting such complexes often mistakenly claim that their example is one of only a handful of previously reported examples. For example, in 2003, Chowdhury et al. [11] claimed “Our complex 2 provides, most possibly, the third example of a square planar silver complex”; in 2002, Carmona et al. [12] claimed “As far as we know only three previous structures…have been described with a Ag(I) atom in a square planar environment.”; also in 2002, Suenaga et al. [8] claimed “However, a silver(I) coordination polymer having a square planar geometry has never been reported.” This is despite the fact that to our knowledge the first reported and structurally characterised square planar Ag(I) complex was of a flavin complex and was reported in 1972 by Charles J. Fritchie, Jr. [13]. Interestingly, this paper has received only about 23 citations (excluding self-citations) most of which are concerned with flavin chemistry and not the uniqueness of stereochemistry of the Ag(I) ion. To our knowledge, there are currently 65 well characterised square planar Ag(I) complexes of which about half are coordination polymers. Given that half are coordination polymers and most have been reported since 1996, it seems likely that the number of square planar Ag(I) containing complexes will continue to grow, if slowly.

As with the assignment of any stereochemistry, ambiguity results both from whether or not a particular auxiliary species, such as a counterion, is coordinated, and from issues with describing the specific geometry perhaps associated with bond angles being intermediate between well-defined archetypes. In order to distinguish between a formal Ag–ligand bond and a longer weak interaction a number of searches of the CSD [1] were undertaken for this review. General unrestricted searches were conducted for all Ag containing crystal structures with NO₃⁻, ClO₄⁻ and BF₄⁻ counter anions, respectively. These searches identified all structures that contained Ag⋯X distances (where X = O–NO₂⁻, O–ClO₃⁻, or F–BF₃⁻) that were less than 4 Å (Fig. 1). These distances represent a combination of formal Ag–ligand bonds, longer weak interactions, and also any other distance found to be shorter than 4 Å. The NO₃⁻, ClO₄⁻ and BF₄⁻ counter anions were selected for the CSD search because they are generally the most commonly used silver salts. The CF₃SO₃⁻ anion is frequently found to be disordered within crystal structures, and the PF₆⁻ anion was found to have too few occurrences to allow a reliable analysis. The resulting searches yielded 519 datasets with 1126 occurrences, 327 datasets with 569 occurrences and 131 datasets with 193 occurrences for NO₃⁻, ClO₄⁻ and BF₄⁻, respectively. For each counter anion, the profile of occurrences shows a skewed normal distribution. A tail leading off into the longer distances arises from the distances that are not due to formal Ag–ligand bonds [14]. This profile is well defined for NO₃⁻ and ClO₄⁻, but less so for BF₄⁻ due to the lower number of occurrences.

Limiting the above CSD searches to distances designated solely as formal Ag–ligand bonds allows the examination of the results according to coordination number. This affects the occurrence profiles by removing the tail at longer distances and leaves an approximate normal distribution of bond lengths. This search gave 370 datasets with 752 occurrences, 142 datasets with 249 occurrences and 16 datasets with 20 occurrences for NO₃⁻, ClO₄⁻ and BF₄⁻, respectively. For NO₃⁻ and ClO₄⁻ the median Ag–O bond lengths were found to be 2.51 and 2.53 Å, respectively. The anion BF₄⁻ is traditionally considered to be a more weakly coordinating anion and as such shows a median Ag–F bond length of 2.58 Å. Further limiting of the search parameters to include solely crystal structures containing four coordinate Ag moieties and non-metal donor atoms, both significantly reduced the number of occurrences and slightly compressed the range of formal Ag–ligand bond lengths observed, and the results are shown in Fig. 2. This search gave 123 datasets with 190 occurrences, 40 datasets with 63 occurrences and 5 datasets with 5 occurrences for NO₃⁻, ClO₄⁻ and BF₄⁻, respectively. Interestingly for NO₃⁻ and ClO₄⁻ the median Ag–O bond lengths were found to be 2.49 and 2.52 Å, respectively. The close similarity between the median bond length for a silver complex of unspecified coordination number and a four-coordinate silver complex is attributed to a combination of the high proportion of four-coordinate complexes compared to other coordination numbers and the fact that four lies at the midpoint between two and six thereby removing the extreme values from the distribution.

Analysis of the distributions indicates that distances lying 0.2 Å or less above the median bond length can be considered a formal Ag–ligand bond [15]. This range of formal bond lengths includes the vast majority of reported distances at least in the case of NO₃⁻ and ClO₄⁻ complexes. Therefore, we generally consider Ag(I)–donor atom distances above this limit of 2.70–2.72 Å to be weak interactions and not formal Ag–ligand bonds.

For many square planar complexes distortion within the square plane due to differing cis bond lengths is observed leading to two longer and two shorter bond lengths. In approximately half of the cases this is imposed by the central Ag(I) ion residing on a special position such as a centre of symmetry or two-fold axis. The question arises as to whether this observation is a consequence of the electronic make up of the primary coordination sphere of the silver or an artefact of crystal packing. In order to investigate this issue we undertook molecular orbital calculations using the Gaussian 03W program package. The simple linear model cation [Ag(NH₃)₂]⁺ has seven occurrences in the CSD with bond lengths in the range 2.11–2.16 Å. A DFT calculation at a B3LYP level of theory using LANL2DZ basis set calculated Ag–N bond lengths of 2.18 Å. A similar calculation was carried out using a more complex pseudo-tetrahedral model cation [Ag(pyridine)₄]⁺ which has five occurrences in the CSD with bond lengths in the range 2.27–2.43 Å. The calculated individual Ag–N bond lengths were found to be 2.36, 2.37, 2.39 and 2.44 Å. For both model compounds in the gas phase the calculated bond lengths tended to be at the higher end of the observed ranges for the corresponding cations in the solid state. The LANL2DZ basis set has been criticised because it freezes all core electrons and approximates their effect with an effective core potential. However, for the theoretical square planar [Ag(pyridine)₄]⁺ cation with pyridine rings orientated perpendicular to the square plane, this calculation yielded the same energies for the cation in both the D_4h and D_2h point groups. This suggests that electronic effects are not responsible for the variation in bond lengths. Further evidence can be obtained by considering the theoretical model compound [Ag(NH₃)₄]⁺. DFT calculations were carried out at a B3LYP level of theory using the 6–31G(d) basis set for the ligands and the LANL2DZ basis set for the Ag(I) ion. Beginning from a square planar C_4h point group and progressing through less symmetric point groups C_2h, C₄, C₂ and ending at C₁, identical molecular energies and Ag–N bond lengths were found for every point group. This result also suggests that there are no favourable molecular energy consequences for having two longer and two shorter trans-Ag bonds. The cyclic point groups were selected for these calculations instead of the dihedral point groups because the more symmetric dihedral point groups do not allow for ‘canting’ of the pyridine rings as has been observed in various crystal structures [16], [17]. In summary, it seems that the prevalence of two longer and two shorter trans-bonds is a consequence of crystal packing effects and crystallographically imposed symmetry.

The bond valence sum (BVS) model has been used to support the assignment of a square planar geometry in contrast to a linear geometry where pairs of bond distances vary significantly [18], [19]. This model relates the bond distances around a metal ion to its oxidation state. A BVS of 1.00 ± 0.20 unit represents complete valence satisfaction for a Ag(I) ion. The bond valence sum (BVS) model has been used to show that all four N atoms are required to satisfy the Ag(I) valence, in three reported square planar Ag–N systems. For {[Ag(pmit)]ClO₄}_∞ (pmit is shown in Fig. 66) using two Ag–N(imino) bond lengths of 2.247(5) Å a BVS of 0.70 unit is obtained [11]. This value was outside the stipulated error limit of ±0.20 unit for valence satisfaction. When two Ag–N(pyridyl) bonds are added the BVS increases to 1.02 unit. Similarly for {[Ag(psdbf)₂]ClO₄}_∞ (psdbf is shown in Fig. 61) assuming linear coordination, a BVS of 0.59 unit was calculated with two Ag–N distances of 2.308(2) Å [3]. Conversely, a BVS of 0.88 ± 0.20 unit was calculated when all four Ag–N distances were included. For the corresponding {[Ag(psdbf)₂]NO₃}_∞ the linear BVS and square planar BVS were calculated at 0.60 ± 0.20 unit and 0.95 ± 0.20 unit, respectively [3].

The square planar geometry often shows some bond angle distortion towards the tetrahedral geometry and in this case an unambiguous stereochemical assignment can be difficult. For example, the complex bis(miax)silver(I) nitrite tetrahydrate [20] is shown in Fig. 3. Each flavin ligand binds one Ag(I) ion which lies on a two-fold axis and is chelated through a N and O donor. Benno and Fritchie describe the Ag(I) ion as exhibiting coordination halfway between tetrahedral and square planar. Numerous other complexes containing a similar intermediate Ag(I) stereochemistry have been reported by Bowmaker et al. [21], Blake et al. [22], and Wu et al. [23] An intriguing helical structure containing 16 unique Ag(I) ions with stereochemistries that show a progression from near tetrahedral and becoming increasingly planar has also been reported by Baxter et al. [24].

To our knowledge, no robust system for classifying square planar geometry has been developed, however some models have been suggested [97]. Constable et al. [25] use a model in which the sum, Σ, of the six inter-bond angles for a four-coordinate complex expresses the degree of planarity (Σ = 720° for a perfectly planar complex and Σ = 657° for perfectly tetrahedral system). Their CSD search is restricted to Ag–N non-macrocyclic systems and they state two reported examples of complexes with Σ > 700° exist. For [Ag₂(ptp)₂](CF₃SO₃)₂ where ptp is 3,6-bis(2-pyridyl)-1,2,4,5-tetrazine, they report Σ = 701.4°. Unfortunately, this model is not robust enough to apply to all four-coordinate coordination environments that arise. For example, a commonly observed centrosymmetric environment is presented in Fig. 4 below. The calculated Σ = 720° but the coordination environment is severely distorted away from square planar. In this case, Σ indicates the planarity but not whether the four donor atoms lie on the corners of a square. Furthermore, if one or both of the two sets of bonds related trans about a four-coordinate Ag(I) centre are not parallel (shown in Fig. 4), at least one trans-bond angle will be less than 180°. Such systems can be planar [26] with a Σ calculated of 659.8°. In some situations, this degree of planarity model is not applicable.

More than half of the reported square planar Ag(I) complexes have some degree of diaxial interaction with counterions. This gives rise to a tetragonally distorted environment about the Ag(I) ion. At some point, the diaxial interactions become short enough to be classified as formal bonds, whereby the coordination of the Ag(I) ion becomes distorted octahedral. Our analysis for light donor atoms suggests this point is around 2.70–2.72 Å. For example, in 1979 the complex H[Ag(RH)₂], where RH₂ = o–aminobenzenesulfonyl glycine, was reported [27]. This crystal structure has an R-factor of 10.2% given that it was solved from photographic data, and is shown in Fig. 5. The complex contains two RH⁻ ligands arranged centrosymmetrically about a Ag(I) ion forming nine-membered chelate rings. Ray and Saha claim an approximate square planar arrangement about Ag(I), by two N and two O donor atoms at 2.280(12) and 2.599(12) Å, respectively. Crystallographically, the ligands are identical as the Ag(I) ion resides on a centre of inversion. The authors are uncertain about the position of the H⁺ counterion, suggesting that it remains virtually associated with the complex through resonance of the two modes of bonding between Ag and the bound carboxyl O atoms. The non-coordinated carboxyl O atom has an interaction of 2.698 Å with the apical site of a Ag(I) ion of a neighbouring complex. The authors consider the possibility of distorted octahedral coordination about the Ag(I) ion arising from the favourable arrangement of this other carboxyl O atom. Given the quality of the structure and the errors associated with the bond lengths it would seem that all the Ag–O bonds are of similar length. Thus in this structure octahedral coordination is perhaps more appropriate and would in turn give rise to a 1D polymeric chain. This is a good example of the ambiguity of the identity of square planar in the context of additional axial interactions.

Other reported examples have been poorly or incorrectly assigned. The T-shaped coordination polymer {[Ag(4-cyanopyridine)₂]BF₄}_∞ is shown in Fig. 6. Carlucci et al. [28] describe the structure of the polymer as consisting of Ag(I) ions linearly coordinated to two pyridine N donors which are further weakly associated with two cyano N donors. One of the N(cyano)–Ag bond distances is extremely long at 3.06(1) Å and the Ag(I) ion coordination is described by the authors as distorted square planar. A better description of this structure would be to consider it as T shaped. Interestingly, the authors neglect to mention a similar axial Ag(I)⋯F–BF₃⁻ distance of 3.01 Å.

The discrete six coordinate complex Ag(tap)₂NO₃ (tap = 1,4,5,8-tetra-azaphenanthrene) is shown in Fig. 7. Nasielski et al. [29] describe the Ag(I) ion as having a strongly folded and twisted square planar stereochemistry. The angle between the normals to the two tap planes is 60° and this deviation from co-planarity is assigned to the close proximity of a NO₃⁻ ion. The authors comment that the complex displays two long and two short bonds to two crystallographically independent tap ligands (2.337(5)–2.570(4) Å). However, despite these two long Ag–N bond lengths, the authors claim that the comparable Ag⋯O₂NO bond lengths of 2.541(10) and 2.656(16) Å are too long for fully covalent but too short for vigorously independent bonding. Also they are reluctant to assign a formal Ag–O bond as the Ag(I) valence shell already contains 18 electrons. Nonetheless, a significant interaction clearly exists between Ag(I) and NO₃⁻ anion, and the complex should almost certainly be described as six coordinate and not pseudo-square planar as claimed.

Within this review, we consider only traditional coordination complexes. We do not consider silver–π complexes where the Ag(I) ion is bound in a π fashion to an arene ring or alkene moiety [30], [31]. In these systems, the exact stereochemistry is often ambiguous and their use as directed nodes can be more difficult [32]. A collection of inorganic complexes obtained by solvo-thermal means also display Ag(I) ions in square planar environments. Such complexes are not considered because of their extended mineral-like structures or the absence of any carbon containing ligands [33], [34], [35], [36], [37], [38], [39]. Currently, there is considerable debate regarding the existence and nature of d¹⁰–d¹⁰ interactions between neighbouring Ag ions [40], [41], [42], [43], [44], [45], [46], [47]. In the main we have disregarded silver complexes where the square plane is completed by Ag–Ag bonds [48], with the exception being where the authors specifically state their Ag(I) complex is square planar as a consequence of a Ag–Ag bond.

In general, a huge variety of donor atoms and ligands are observed in these square planar Ag(I) complexes. These donors range from aromatic, aliphatic and imine N as well as, O, Cl, P and thioether S atoms. Despite this ligand variety some common factors among these complexes can be noted. First, terpyridine-based ligands often enforce pseudo-square planar geometry due to their rigidity and donor atom arrangement. Second, many complexes contain flexible ligands with non-bonding aromatic groups in close proximity to the donor atoms. This means that upon coordination the non-bonding aromatic group shields coordination sites of the central Ag(I) ion.

In this review, the square planar Ag(I) complexes are divided into two categories according to whether they are discrete or polymeric. Within these categories, the complexes are further subdivided in terms of whether or not the square planar Ag(I) ions reside on crystallographic special positions. Finally, in the class of discrete complexes, a special section has been devoted to the complexes containing terpyridine-based ligands, due to their specific geometric requirements.