Craig D.
MacKinnon
*a,
Shawna L. M.
Parent
a,
Robert C.
Mawhinney
a,
Abdeljalil
Assoud
b and
Craig M.
Robertson
b
aDepartment of Chemistry, Lakehead University, 955 Oliver Rd., Thunder Bay, ON, Canada P7B5E1. E-mail: mailto:craig.mackinnon@lakeheadu.ca; Fax: +807-346-7775; Tel: +807-343-8327
bDepartment of Chemistry, University of Waterloo, Waterloo, ON, Canada N2L3G1
First published on 17th October 2008
The two ligands 3,3′-dicyanobiphenyl (3) and 5,5′-dicyano-2,2′-bithiophene (4) are geometrically similar but give rise to different coordination compounds when reacted with silver(I) salts. The crystal structures of 3·AgClO4 and 3·AgBF4 have 1:
1 ligand
:
cation ratios, while 42·ClO4 and 42·BF4 have 2
:
1 ligand
:
cation ratios. These ratios for the solid-state materials were constant, even when different stoichiometric ratios were used in the reaction vessels. The structures of these four coordination compounds are presented and the reasons for the difference in the coordination environments discussed with reference to IR spectroscopic and DFT computational results.
Silver(I) is a typical d10 metal cation. In coordination compounds it can have virtually any geometry, from linear to octahedral. Ligands can range from the very strong (e.g., phosphine donors2–5) to the very weak (e.g., sulfonate donors6–8). The most common donor atom for silver(I), though, is nitrogen.9,10 Of this subset, the nitrile group is an especially favoured donor unit, because of the wide variety of organic nitriles available. Ligands range from very flexible alkyl-bridged dinitriles,11,12 to rigidly trigonal 1,3,5-trisubstituted phenyl trinitriles.13
We are interested in exploring rod-like dinitrile ligands with potential additional binding sites in the bridge. Rod-like ligands can be on-axis or off-axis, as demonstrated in Scheme 1 for the 4,4′-biphenyl 1 (on-axis) and 3,3′-biphenyl 3 (off-axis) bridges. On-axis ligands are more common, examples of which include the seminal dinitrile 1, and also 2, which contains a bipyridine bridge. Silver(I) compounds containing ligand 1 range from linear coordination polymers14 to three-dimensional diamondoid networks,15 depending on the solvents, counterions, stoichiometry, etc. For ligand 2, the dinitriles form a one-dimensional coordination polymer with Ag+, but in this case the bipyridyl moiety also binds to the cation to cross-link the polymer chains.16
![]() | ||
Scheme 1 |
In contrast, 3 and 4 (5,5′-dicyano-2,2′-bithiophene) are off-axis ligands. The sulfur atoms in the thiophene rings of 4 are potential secondary binding sites, although there are many examples where binding between metals and thiophene sulfurs does not occur, even when the ligand geometry should allow for such an interaction.2,17–19 Thus, we have embarked on the synthesis of silver(I) coordination compounds of the off-axis rod-like ligands 3 and 4. These two ligands are putatively the same shape, while having different donor characteristics. We present herein the crystal structures for several compounds, comparing the structures formed using ligands 3 and 4 in a variety of solvents and with a variety of silver(I) counterions.
Coordination compounds of silver(I) with the ligands 3 and 4 were generated by sealing the two solids (ligand and silver(I) salt, 1:
1 molar ratio) in a glass tube with an appropriate solvent. The contents were heated to 100 °C until the solids dissolved, and were then cooled to room temperature at the rate of 1° h−1 using a programmable tube furnace. Alcoholic and chlorinated solvents are unsuitable, as reaction between the silver(I) and solvent causes a black decomposition product. We therefore limited ourselves to aromatic solvents (benzene, toluene, nitrobenzene, etc.) and acetonitrile. The latter was also discovered to be unsuitable as the nitrile group on the acetonitrile competes with the ligand, leading to partial or no reaction between the cation and the ligand; this result is unsurprising considering the large number of alkyl nitriles that are known to bind to silver(I).11,26–30 Solids were isolated by cutting open the tubes and characterised by the infrared absorption frequency of the nitrile.
Table 1 summarises the results for complexes using 3 as the ligand. The anions are arranged in the table from most coordinating to least coordinating (using an order determined by others31,32). All combinations of ligand and solvent gave a reaction, although it was only partial in some cases and in other cases decomposition (black powder) occurred. Crystalline solids were stable for short periods in the air, but did tend to pick up atmospheric water (at least on the surface) upon prolonged exposure. There is no photodegradation, even after several months under ambient light in the laboratory.
We have previously published the results of the reaction of 4 with some silver(I) cations.33Table 2 presents our new results. As was found in the earlier work, only some of the new combinations of anion and solvent yielded coordination compounds (the lack of reactivity is explained below).
Anion (X) | BF4− | ClO4− | PF6− | SbF6− | CF3SO3− | NO3− |
---|---|---|---|---|---|---|
a From ref. 33. b A vibrational frequency of 2205 cm−1 has been reported for this compound,33 which was a shift of +10 from the ligand frequency of 2195 cm−1; on the instrument used for this paper, the ligand frequency is 2220 cm−1, so the frequency in this table has been entered as 2220 + 10 = 2230 cm−1. c The same shift was observed for crystals of this compound grown in o-, m-, and p-xylene. | ||||||
Solvent | ||||||
benzene | insol. | 2250 | SM a | insol. | 2230ab | SM a |
toluene | 2230 | 2231 | SM a | SM a | SM a | SM a |
CH3CN | --- | SM | SM a | SM a | SM a | SM a |
CH2Cl2 | --- | decomp. | SM | SM | SM | SM |
DMSO | --- | decomp. | --- | decomp. | decomp. | decomp. |
CH3NO2 | --- | decomp. | --- | decomp. | decomp. | decomp. |
C6H5NO2 | 2229 | 2230c | --- | 2233 | high sol. | decomp. |
For both ligands, the IR stretching frequency increases upon coordination. The lone pair on the nitrile group is in a weakly σ* (antibonding) orbital, so donation from this orbital to the metal centre should cause such an increase. Since a metal-to-ligand charge transfer back-bond would populate a π* orbital causing a decrease in stretching frequency, the opposite of our observation, it would seem to be of little importance. This is confirmed by DFT calculations (see below).
We have solved the crystal structures of several combinations of ligand, solvent, and counterion. The crystallographic data is summarised in Table 3, and discussed in detail immediately below.
Compound | 4 | 3·AgClO4 (C6H6 solvate) | 3·AgBF4 (0.5 C7H8 solvate) | 4 2·AgClO4 | 4 2·AgBF4 |
---|---|---|---|---|---|
Chemical formula | C10H4N2S2 | C20H14AgClN2O4 | C17.5H12AgBF4N2 | C20H8AgClN4O4S4 | C20H8AgBF4N4S4 |
Mr/g mol−1 | 216.27 | 489.65 | 445.00 | 639.86 | 627.22 |
Crystal system |
P![]() |
P21/c | P21/c | P4/n | P4/n |
Space group | Triclinic | Monoclinic | Monoclinic | Tetragonal | Tetragonal |
μ(Mo Kα)/mm−1 | 0.531 | 1.226 | 1.275 | 1.358 | 1.285 |
a/D | 3.8090(2) | 7.4242(6) | 8.8779(14) | 13.834(5) | 13.6376(19) |
b/D | 7.4518(4) | 19.9968(17) | 14.465(2) | 13.834(5) | 13.6376(19) |
c/D | 8.2086(5) | 12.8576(10) | 13.422(2) | 6.162(4) | 6.2054(17) |
α/° | 87.9360(10) | 90 | 90 | 90 | 90 |
β/° | 85.4850(10) | 92.527(2) | 108.114(3) | 90 | 90 |
γ/° | 82.1260(10) | 90 | 90 | 90 | 90 |
V/D3 | 230.00(2) | 1907.0(3) | 1638.1(4) | 1179(1) | 1154.1(4) |
Z | 1 | 4 | 4 | 2 | 2 |
D c/g cm−1 | 1.561 | 1.705 | 1.821 | 1.802 | 1.805 |
T/K | 173(2) | 296(2) | 200(2) | 296(2) | 200(2) |
F(000) | 110 | 976 | 892 | 632 | 616 |
θ range/° | 3.67 to 26.46 | 2.93 to 29.99 | 2.79 to 30.56 | 3.62 to 29.99 | 3.28 to 27.99 |
R, wR [I > 2σ(I)] | 0.0236, 0.0604 | 0.0422, 0.0894 | 0.0384, 0.0950 | 0.0299, 0.0715 | 0.0400, 0.0866 |
R, wR (all data) Reflections | 2744 | 18166 | 18925 | 10613 | 10648 |
collected | 947 | 5548 | 5016 | 1699 | 1398 |
unique | 889 | 4476 | 4475 | 1480 | 1233 |
observed |
![]() | ||
Fig. 1 Two views of the structure of 3·ClO4 (benzene solvate). Top: The full structure viewed down the crystallographic a axis. Bottom: The paired one-dimensional coordination polymer, bridged by ClO4− anions. Colour scheme: grey = C, blue = N, pink = Ag, red = O, green = Cl. |
![]() | ||
Fig. 2 Comparison of the structures of 3·AgBF4 and 3·AgClO4, emphasising the stepped structure of the coordination polymer. Top: Full structure of 3·AgBF4 (0.5 toluene solvate); the toluene is disordered. Bottom: One strand of 3·ClO4 with solvate and anion removed. Colour scheme is the same as in Fig. 1 (peach = B, yellow–green = F). |
The ligand 3 is in the anti (or transoid) conformation, leading to a stepped 1D coordination polymer. Fig. 2 is drawn to emphasize this feature. Adjacent polymers are not intertwined, which can occur, e.g., with the off-axis rod-like nitrile 3,3′-dicyanodiphenylacetylene.36 Both complexes of 3 contain disordered molecules of solvent. The N–Ag–N bond angles are not quite linear (172.8° for the BF4− species and 170.3° for the ClO4− species). The silver(I) centre in 3·AgClO4 is roughly square planar, which allows the near linear coordination to 3 (the remaining two vertices of the square are used by the two bridging ClO4− anions). The two phenyl rings in the biphenyl bridge are nearly coplanar in 3·AgClO4 (<2° torsion angle), but considerably more twisted in 3·AgBF4 (just under 30°). Presumably the angles are due to packing effects. In the only other crystal structure containing 3 of which we are aware, a cocrystal containing 3 and tetramethylimidazolium, the rings are essentially coplanar.20 This cocrystal contains isolated nitriles (not coordinated nor H-bonded) with a C–N bond distance of 1.172 Å. The coordinated C–N distance in 3 is 1.140 Å for the ClO4−salt. The BF4−salt has two crystallographically inequivalent nitriles with bond lengths 1.140 and 1.133 Å. The shortened distances are consistent with the increase in the IR stretching frequency noted above.
Although Ag+–Ag+ contacts are weak (certainly weaker than for aurophilic bonds37–39), they can exert a structural effect in the solid state.36,40–42 In our case, however, argentophilic interactions cannot be invoked; the closest Ag+–Ag+ contact in 3·ClO4 is 3.42 Å, which is the same as the sum of the van der Waals radii.43 The closest separation in 3·BF4 is slightly longer at 3.51 Å. The benzene rings of the ligand are packed in a slipped π-stack in 3·ClO4. In the BF4−salt, the benzene rings of the ligands are slipped more, so that each π-systems is over the CN–Ag–NC portion of the molecule below it.
As shown in Fig. 3, the structure contains pairwise hydrogen bonds. The two H-bonds are crystallographically equivalent, with a distance of 2.593 Å and a C–N⋯H bond angle of 127.28°. This pairwise H-bond is geometrically similar to those found in the higher-order oligothiophenes (C4H2S)n(CN)2 (n = 3–5)44 and in p-dicyanobenzene.45 The H-bond in 3 is shorter and presumably stronger than the average H-bond in these species (e.g., the CN⋯H distances are 2.688 and 2.711 Å in (C4H2S)4(CN)2 and 1,4-C6H4(CN)2, respectively); however, it is not the shortest ever observed for oligothiophene nitriles ((C4H2S)3(CN)2 has six crystallographically independent H-bonds, the shortest of which is 2.420 Å, although the average of these six bonds is longer than the H-bond in 3 at 2.64 Å).
![]() | ||
Fig. 3 Structure of 4. Left: Viewed to illustrate the intermolecular hydrogen-bond network. Right: Viewed to show π-stacks. Colours as in Fig. 1 (yellow = S). |
Nitrile-substituted oligomers have been considered as potential n-type semiconducting materials for use in transistor devices.46,47 In this context, the structure of 4 is interesting as the molecules pack in a slipped π-stack, considered the preferred packing motif for maximum mobility.48–50 Unfortunately, the intrinsic conductivity of bithiophenes is normally too low for devices, and the closest S⋯S contacts in 4 are over 3.8 Å, well outside the van der Waals contact distance of 3.6 Å.
![]() | ||
Fig. 4 Structure of 42·AgX (X− = ClO4− or BF4−) viewed down the crystallographic c axis. The counterions are represented as the large green spheres. Other colours are as in Fig. 1 (yellow = sulfur). |
Neither anion is coordinating, nor is there any solvent encapsulated in the lattice. There are no Ag+–S contacts with the thiophene sulfurs, nor are there any argentophilic bonds. This structure is remarkably reproducible from a number of different solvents. Crystals of 42·AgClO4 grown in nitrobenzene, and in o-, m- and p-xylene, all formed this structure (crystallographic diffraction data were collected on crystals grown from each solvent, but were not fully refined once it became clear the crystals were identical). Attempts to generate a similar 2:
1 complex with ligand 3, by using a 2
:
1 stoichiometric ratio in the reaction vessel, failed. The only crystalline material recovered were 1
:
1 complexes that were, by IR and combustion analysis, the same complexes generated in the 1
:
1 reactions.
Density functional theory (DFT) calculations were performed on benzonitrile- and cyanothienyl-ligated silver(I) molecules to understand the bonding in these complexes. Organonitrile–silver(I) complexes have been the subject of a previous computational analysis52 and we have employed the same computational approach with identical results for the simplest benzonitrile-silver(I) complexes. The results for the model compounds (PhCN)2Ag+ and (TCN)2Ag+ (Ph = phenyl and T = thienyl) are given in Table 4. Fig. 5 shows the localised orbitals and their electron populations, obtained using the NBO method,53 in the silver(I)-nitrile complexes along with change in population from the isolated species. An analysis of the orbitals and interaction energies verifies that the nitrile–silver(I) bond is predominantly a donation from the nitrile lone pair into the s and pz orbitals on the silver(I), with practically no backbonding to the nitrile π* orbital. The orbital “CN lp” (Fig. 5, top) has a node between the C and the N of the nitrile (i.e. is antibonding) and loses 0.10 e− upon complexation. The computed nitrile stretching frequencies (Table 4) reflect this decrease in orbital population. The frequency is calculated to increase upon coordination to Ag+. For (PhCN)2Ag+ it increases 20 cm−1 relative to PhCN, and 16 cm−1 for (TCN)2Ag+ over TCN. This matches the experimental trend that the increase in IR frequency is larger for coordination compounds of 3 than for compounds of 4.
L = PhCN | L = TCN | |
---|---|---|
a The change in value from the uncomplexed ligand in parenthesis. b Ag+ + 2L 6 L2Ag+ complexation reaction in the gas phase at 298 K. c See Fig. 5 for graphical representation of localized interacting orbitals. | ||
r CN/Å | 1.162 (−0.004)a | 1.164 (−0.003)a |
r n –Ag+/Å | 2.163 | 2.162 |
ν CN/cm−1 | 2342 (+20)a | 2329 (+16)a |
q Ag+ /e− | 0.822 | 0.824 |
ΔGcomplex/kcal mol−1b | −68.74 | −67.59 |
NBO Orbital Interaction energies kcal−1 mol−1c | ||
CN lp → Ag+ 5s | 29.88 | 29.71 |
CN lp → Ag+ pz | 9.18 | 9.17 |
Ag+ dxz → CN π2* | 1.74 | 1.70 |
Ag+ dyz → CN π1* | 1.88 | 1.93 |
![]() | ||
Fig. 5 Graphical representation of the localized natural bond orbitals in L2Ag+ obtained at the B3LYP/dzvp model chemistry corresponding to the interactions given in Table 4 along with the occupancy and change in occupancy, in parentheses, upon complexation. |
The relative weakness of 4 as a ligand is also illustrated in the coordination compounds that have been crystallographically characterised. In 3·AgBF4 only two nitrile-Ag+ contacts are needed to satisfy the Ag+ centre, and there are no solvent or anion coordinations. In 42·AgBF4, on the other hand, the Ag+ requires four nitriles to satisfy its coordination sphere, even though the stoichiometry of the reaction from which the crystals were grown was not 2:
1.
There is no compelling energetic reason why 3 should not also form a 4-coordinate (2:
1) compound. DFT calculations (Table 5) suggest that there is little difference in energy between two-coordinate coordination compounds based on the nature of the aromatic ring. Certainly, (TCN)2Ag+ has a slightly lower formation energy but the difference is small. Similarly, although the stabilisation per nitrile is less for a four-coordinate species, the overall formation energy for (TCN)4Ag+ and (PhCN)4Ag+ are very similar.
n | L = PhCN | L = TCN |
---|---|---|
1 | −38.94 | −37.72 |
2 | −29.80 | −29.87 |
3 | −4.37 | −3.05 |
4 | −8.80 | −10.23 |
Thus, we conclude the four-coordinate equivalents using ligand 3 are not formed for steric reasons. Crowding at the metal centre is not the problem, as four phenyl nitriles can fit around silver(I), which is demonstrated for the complex 12·AgPF6.15 There is so much room in that structure, in fact, that it has several interpenetrating adamantoid networks. Similarly, compounds of 4 can fit four aromatic nitriles around the Ag+, although now there is only enough room for one independent network. There is enough space in the lattice to incorporate the (non-coordinating) counterion in the channels formed by the distorted-diamondoid network. Thus, the space in the lattice has decreased, as a consequence of the collapsing of the off-axis angle (the angle of intersection between the centroid–centroid axis and the nitrile CN axis, Scheme 2). On changing from ligand 1 to ligand 4, the angle changes from 0 to 33°, pinching off the potential for open space in the structure. In 3 the angle is even sharper at 60°, and presumably this wouldn't even allow space for counterions in a tetragonal structure with a 4-coordinate Ag+ centre.
![]() | ||
Scheme 2 |
Nuclear magnetic resonance (NMR) spectra were recorded on a Varian Unity Inova 500 at room temperature; chemical shifts are reported in ppm referenced to TMS added to the solvent. Infrared absorption spectra were recorded as nujol mulls on a Nicolet 380 FT-IR spectrometer; vibrational frequencies are reported as wavenumbers (cm−1), with an experimental resolution of 1 cm−1. Combustion analyses were performed on a CEC (SCP) 240-XA Analyzer. Gradient vacuum sublimations and crystal growth for X-ray crystallography were performed on an Applied Test Systems model 3210 split tube furnace attached to a series 2404 3-zone temperature control system. Melting points were recorded on an Electrothermal 9100 apparatus and are uncorrected.
The space groups for all compounds were determined on the basis of the systematic absences using symmetry and space group of the APEX II package. Structure solution and refinement using the SHELXTL package of APEX II were successful in the chosen centrosymmetric space groups. The structure solution via direct method led to the identification of the heavy atoms and the refinement on F2 using least-square method resulted in good R values. Full details of the structures have been deposited with the Cambridge Crystallographic Data Centre as CDCC 692771–692775. This information may be obtained free of charge viahttp://www.ccdc.cam.ac.uk/data_request/cif.
Full data sets were collected for 42·AgClO4 on crystals grown from o-, m-, and p-xylene, and from nitrobenzene. The unit cells for all four crystals were identical; the data set from p-xylene was chosen for a full refinement. This structure has a disordered ClO4− (O4 partially occupies two sites O4A and O4B); six restraints are used in the refinement, i.e., O4A and O4B have the same temperature factors Uij (U11, U22, U33, U12, U13, U23). Similarly for 42·AgBF4, the fluorine atom F4 is disordered and F4A and F4B have the same temperature factors Uij (U11, U22, U33, U12, U13, U23).
Footnote |
† CCDC reference numbers 692771–692775. For crystallographic data in CIF or other electronic format see DOI: 10.1039/b810479d |
This journal is © The Royal Society of Chemistry 2009 |