Marike
du Plessis
,
Vincent J.
Smith
and
Leonard J.
Barbour
*
Department of Chemistry and Polymer Science, University of Stellenbosch, Stellenbosch 7600, South Africa. E-mail: ljb@sun.ac.za; Fax: +27 21 808 3360; Tel: +27 21 808 3335
First published on 28th March 2014
Single crystals of a previously reported porous metallocycle [Ag2L2](BF4)2·2CH3CN (1), where L is the ligand 1,4-bis(2-methylimidazol-1-ylmethyl)benzene, were grown from acetonitrile and immersed in different organic solvents. The crystals thus treated were subjected to single-crystal X-ray diffraction analysis, which revealed that the acetonitrile guest molecules had been replaced by the solvent that the compound was exposed to, yielding five different solvates: [Ag2L2](BF4)2·2(CH3)2CO (2), [Ag2L2](BF4)2·2CHCl3 (3), [Ag2L2](BF4)2·C6H6 (4), [Ag2L2](BF4)2·C6H4F2 (5), [Ag2L2](BF4)2·C7H8 (6). Thermogravimetric analysis supports these findings.
One of the target architectures for porous crystals investigated by our group is the “doughnut-shaped” metallocycle. The “doughnut” shape of the metallocycles prevents them from packing efficiently and results in the formation of crevices, cavities or channels in the packing arrangement. Naturally, owing to close-packing requirements the occurrence of empty space in the crystal structure is energetically unfavourable and therefore the available “space” is usually occupied by solvent molecules. Ideally, the porous phase can be obtained by removing the solvent molecules from the channels without disrupting the host framework. In this regard we have conducted a further investigation of the porous metallocycle previously reported by Barbour et al.5 The formation reaction of the metallocyclic complex [Ag2L2](BF4)2·2CH3CN (1) is shown in Scheme 1. We have reported that the guest molecules of 1 can be removed from the channels in a SC–SC fashion to yield a porous, gas sorbing material (1apohost)5 and elsewhere a series of eight solvates of the copper analogue of 1 was reported in an investigation of the effect of solvent templation on the formation of metallocyclic complexes.6 Most of the solvates in the latter case were prepared by removing the solvent from the channels of the as-synthesised crystals under reduced pressure, followed by dissolving the resulting apohost complex in different solvents. Slow evaporation of the solvent afforded the previously reported solvates.
Scheme 1 Formation of [Ag2L2](BF4)2·2CH3CN (1) where L = 1,4-bis(2-methylimidazol-1-ylmethyl)benzene. |
We now show that 1 can incorporate different solvent molecules into its channels in SC–SC transformations as opposed to these solvates being obtained by dissolving and recrystallising the apohost from different solvents. Furthermore, the transformation is accompanied by subtle changes, which result in a change in the space group and/or crystal system. It is quite unusual for metallocycles to undergo a SC–SC phase transformation5–8 as only a single case has been reported to date.9
Scheme 2 Formation of solvates 2–6 from the original solvate 1 by means of solvent exchange in SC–SC fashion. |
Using a glass pipette the prism shaped crystals of 1 were carefully removed from the mother liquor and allowed to dry on filter paper. The dry crystals were then immersed in the target solvent for periods ranging from two to eight weeks. In each case the crystals showed signs of cracking and deterioration such that it was no longer possible to subject the original crystal to SCD. This is in sharp contrast to other reported guest exchanges where the individual crystals generally remained intact.7 Although the bulk integrity of the crystals in the study reported here was not maintained, small single crystals could still be isolated by breaking the original single crystal into smaller fragments. Structure elucidation by means of SCD, together with thermogravimetric analysis (TGA) revealed unambiguously that the guest exchange process was successful. This process occurs with very little alteration of the host structure. As a result, structures 1 to 6 and 1apohost‡ are very similar with respect to the arrangement of the host molecules in the crystal structure and selected parameters are presented for comparison in Table 1. TGA results can be found in the ESI† and they support the findings elucidated by SCD methods.
Structure | Guest | Host:guest | Guest volumea/Å3 | Solvent accessible volumeb/Å3 | Intra Ag⋯Agc/Å | Inter Ag⋯Agd/Å | Metallocycle tilt angle (ε)e/° | Phenylene tilt angle (η)f/° | Imidazolyl tilt angle (θ)g/° |
---|---|---|---|---|---|---|---|---|---|
a Calculated using a literature reported equation.11 b Solvent accessible void volume per unit cell as calculated by Platon, SQUEEZE.12 c Distance between Ag atoms belonging to the same metallocycle. d Distance between Ag atoms belonging to adjacent metallocycles. e Angle between the least squares plane through the metallocyclic complex and the stacking direction as calculated in Mercury.13 f Angle between the least squares plane through the benzene moiety of the metallocycle in 1 and that of the structure in question. g Angle between the least squares plane through the imidazole moiety of the metallocycle in 1 and that of the structure in question. | |||||||||
1apohost | — | — | — | 245 | 7.7 | 7.0 | 90 | 8.0 | 5.0 |
1 | Acetonitrile | 1:2 | 49 | 223 | 7.3 | 7.0 | 90 | 0 | 0 |
2 | Acetone | 1:2 | 67 | 235 | 7.5 | 7.1 | 87.5 | 8.5 | 5.0/4.5 |
3 | Chloroform | 1:2 | 72 | 237 | 7.7 | 7.1 | 86.7 | 4.8 | 4.0, 4.5/4.3 |
4 | Benzene | 1:1 | 81 | 238 | 7.4 | 7.1 | 88.6 | 2.8 | 1.7, 2.1/3.2 |
5 | 1,4-DFBz | 1:1 | 93 | 241 | 7.6 | 7.1 | 88.7 | 2.3/2.2 | 4.0/3.3 |
6 | Toluene | 1:1 | 99 | 208 | 7.5 | 7.0 | 84.1 | 7.6 | 2.1/6.3 |
Acetonitrile molecules are situated in anti-parallel pairs along the infinite channel. Each pair of guest molecules is centred within a metallocycle resulting in a host-to-guest ratio of 1:2. In order to measure/show the subtle differences between the host metallocycles of 1 and 2–6, we define a metallocycle tilt angle (ε) between the least squares plane trough the metallocycle, and the stacking direction. In 1ε is perpendicular to the stacking direction. Furthermore we define a phenylene tilt angle (η) and an imidazolyl tilt angle (θ) as the angle between the least squares plane through the aromatic moiety and its corresponding moiety in the reference structure 1. These planes were calculated and the angles measured after overlaying each of the metallocycles 2–6 and 1apohost with the reference structure 1 using the molecule overlay function in Mercury.13
The increase in guest volume from 49 Å3 for acetonitrile to 67 Å3 for acetone is accompanied by a corresponding decrease in ε. The tilt angles η and θ display a significant change in the shape of the host metallocycle, which is further adapted through increases in the inter- and intramolecular Ag⋯Ag distances. As expected, the increase in the volume of the guest gives rise to an increase in the guest accessible volume. Fig. 3 shows the one-dimensional solvent filled channel of 2.
Fig. 3 Perspective view perpendicular to a one dimensional channel of solvate 2. Acetone guest molecules are shown in space filling representation. |
Fig. 5 shows the hydrogen atoms of the guest protruding through the Connolly surface that maps the guest accessible volume of the infinite channel. This is an indication of a CH⋯π interaction between the chloroform hydrogen atom and the imidazole moiety of the ligand L. The distance between the carbon atom of the chloroform molecule and the centroid of the imidazole ring (2.59 Å) indicates the very close contact between these moieties.
Fig. 5 Perspective view perpendicular to a one dimensional channel of solvate 3. Chloroform guest molecules are shown in space filling representation. |
Although the guest volume increases from 72 Å3 for one chloroform molecule to 81 Å3 for a single benzene molecule, the solvent accessible volume calculated for solvates 3 and 4 are almost identical. While two molecules of chloroform occupy each cavity of the metallocycles in 3, only one benzene molecule occupies the corresponding space in 4. Therefore there is a relatively large amount of space available for the benzene guest molecules to orientate themselves within the one-dimensional channels and this may explain the disorder of the guest molecules. With reference to Fig. 6 and 7, benzene molecules are situated in the positions of the red molecules with a site occupancy of ca. 30% while the guest occupies the position of the green and orange molecules with occupancies of ca. 27 and ca. 13%, respectively. Note that the periodicity of the host in Fig. 7(C) and (D) is identical but there is a difference in the periodicity of the guest.
It is interesting to note that the toluene solvate deviates significantly from the overall trend of an increase in the solvent accessible volume with an increase in guest volume. Even though the van der Waals volume of toluene is the largest of the series of guest molecules, it has the smallest solvent accessible volume. Furthermore, ε in 6 is the largest of the deviations observed and the neighbouring metallocycles are stacked closest together, as is evident from the inter Ag⋯Ag distances. Despite the significant deviations in 6, there are no significant interactions between the host framework and the guest molecules.
When comparing the six solvates we find that the metallocyclic host frameworks are very similar. Upon closer inspection of the structural parameters it is evident that the host undergoes very subtle changes upon guest exchange. In the series of structures presented here, the volume of the guest molecule is gradually increased from 2–6 (67–99 Å3). A consequence of increasing the size of the guest molecule is that at some point the host:guest ratio has to decrease from 1:2 to 1:1 as observed by comparing the smaller, non-aromatic guests with the larger, aromatic guests. If we consider the aromatic- and non-aromatic guest molecules separately the intramolecular Ag⋯Ag distance displays a trend. There is an increase in this distance with an increase in the guest volume with the toluene solvate (6) being the exception. A similar observation is made when comparing the solvent accessible volume of the structures. It is interesting to note that the empty host structure has the largest void volume. It can be rationalised that the metallocycles of the apohost expand significantly since there is no guest to embrace. The other extreme is that the metallocycle collapses in on itself. Interestingly this may occur as a SC–SC transformation.7,9ε adjusts between 1 and 6 degrees while the intermolecular Ag⋯Ag distance changes only slightly (7.0–7.1 Å). η adjusts by as much as 8.5 degrees in 2. The largest values for η and θ are observed for 2 and 6, which can be rationalised since these are the solvates with the smallest and largest guests respectively in the series 2–6. The N–Ag–N angle ranges between 175.5° and 179.0° and deviates from linearity as the metallocycle adapts its shape to accommodate the guest. Further adjustments are observed in the corners of the metallocycles (the N–C–C angle ranges between 109.9° and 111.7°).
It could be postulated that limited weak interactions or rather, close contacts as observed in some of these solvates are responsible for the orientation of the guest molecules with respect to the host framework. However, none of the structures show host–guest or guest–guest interactions significant enough to direct and determine the position of the guest molecules in the host framework unequivocally. In fact, the contributions of these weak contacts are insignificant when compared to the many other factors that determine the crystal structure as Gavezzotti outlined in a recent article.14 We believe that incorporation of the various types of guest molecules into the channels of the host takes place via cooperative movement.15a,b During this process the host framework adapts to the shape of the guest molecules in a shape-fit manner.
Very little data exist in the literature of single-crystal replacement of guest molecules in zero-dimensional coordination compounds. It is envisioned that a large database of discrete coordination complexes incorporating different solvent molecules may provide sufficient information to assist in gaining more insight into the mechanism of guest exchange. Often the large solvent accessible space in MOFs presents difficulties with modeling guest molecules from SCD experimental data. In this regard, metallocyclic compounds are ideal candidates for SC–SC transformation studies with the aim of investigating structure–property relationships. We can, in effect, take “snapshots” of small organic molecules as they are captured by an appropriate crystalline host.
Footnotes |
† Electronic supplementary information (ESI) available: TGA results for 1apohost and 1–6, selected structure parameters, crystallographic data tables. CCDC 984255–984259. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4ce00244j |
‡ Crystal data for 1apohost: C32H36Ag2B2F8N8, M = 922.05, colourless prism, 0.25 × 0.20 × 0.10 mm3, monoclinic, space group C2/m (no. 12), a = 14.8307(10), b = 20.5641(13), c = 7.0449(5) Å, β = 90.1180(10)°, V = 2148.5(3) Å3, Z = 2, Dc = 1.425 g cm−3, F000 = 920, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 56.5°, 6760 reflections collected, 2555 unique (Rint = 0.0199). Final GooF = 1.071, R1 = 0.0267, wR2 = 0.0639, R indices based on 2430 reflections with I > 2sigma(I) (refinement on F2), 120 parameters, 0 restraints. Lp and absorption corrections applied, μ = 0.977 mm−1. Crystal data for 1: C36H42Ag2B2F8N10, M = 1004.16, colourless prism, 0.25 × 0.20 × 0.15 mm3, monoclinic, space group C2/m (no. 12), a = 14.9110(10), b = 20.1383(14), c = 7.0209(5) Å, β = 90.2930(10)°, V = 2108.2(3) Å3, Z = 2, Dc = 1.582 g cm−3, F000 = 1008, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 56.4°, 6675 reflections collected, 2520 unique (Rint = 0.0203). Final GooF = 1.062, R1 = 0.0257, wR2 = 0.0625, R indices based on 2413 reflections with I > 2sigma(I) (refinement on F2), 139 parameters, 0 restraints. Lp and absorption corrections applied, μ = 1.004 mm−1. Crystal data for 2: C38H48Ag2B2F8N8O2, M = 1038.20, colourless prism, 0.17 × 0.11 × 0.10 mm3, triclinic, space group P (no. 2), a = 7.1272(12), b = 12.249(2), c = 12.943(2) Å, α = 71.973(2)°, β = 89.308(2)°, γ = 89.483(2)°, V = 1074.4(3) Å3, Z = 1, Dc = 1.605 g cm−3, F000 = 524, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 55.1°, 13534 reflections collected, 4907 unique (Rint = 0.0276). Final GooF = 1.174, R1 = 0.0310, wR2 = 0.0753, R indices based on 4733 reflections with I > 2sigma(I) (refinement on F2), 275 parameters, 0 restraints. Lp and absorption corrections applied, μ = 0.990 mm−1. Crystal data for 3: C34H38Ag2B2Cl6F8N8, M = 1160.78, colourless prism, 0.21 × 0.21 × 0.19 mm3, monoclinic, space group P21/c (no. 14), a = 7.0910(19), b = 20.479(6), c = 14.855(5) Å, β = 91.038(4)°, V = 2156.9(11) Å3, Z = 2, Dc = 1.787 g cm−3, F000 = 1152, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 56.7°, 13383 reflections collected, 5366 unique (Rint = 0.0483). Final GooF = 1.061, R1 = 0.0573, wR2 = 0.1403, R indices based on 3989 reflections with I > 2sigma(I) (refinement on F2), 273 parameters, 0 restraints. Lp and absorption corrections applied, μ = 1.352 mm−1. Crystal data for 4: C38H42Ag2B2F8N8, M = 1000.16, colourless prism, 0.20 × 0.19 × 0.18 mm3, triclinic, space group P (no. 2), a = 7.1020(18), b = 12.388(3), c = 12.794(3) Å, α = 72.237(4), β = 89.284(4), γ = 89.239(4)°, V = 1071.8(5) Å3, Z = 1, Dc = 1.550 g cm−3, F000 = 502, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 61.7°, 15794 reflections collected, 6181 unique (Rint = 0.0438). Final GooF = 1.005, R1 = 0.0372, wR2 = 0.0868, R indices based on 5125 reflections with I > 2sigma(I) (refinement on F2), 288 parameters, 249 restraints. Lp and absorption corrections applied, μ = 0.986 mm−1. Crystal data for 5: C38H40Ag2B2F10N8, M = 1036.14, colourless prism, 0.13 × 0.13 × 0.10 mm3, triclinic, space group P (no. 2), a = 7.0814(11), b = 12.445(2), c = 12.837(2) Å, α = 71.505(2), β = 89.185(2), γ = 89.430(2)°, V = 1072.7(3) Å3, Z = 1, Dc = 1.604 g cm−3, F000 = 518, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 61.5°, 15820 reflections collected, 6139 unique (Rint = 0.0331). Final GooF = 1.153, R1 = 0.0459, wR2 = 0.1125, R indices based on 5607 reflections with I > 2sigma(I) (refinement on F2), 291 parameters, 14 restraints. Lp and absorption corrections applied, μ = 0.994 mm−1. Crystal data for 6: C39H43Ag2B2F8N8, M = 1013.17, colourless prism, 0.17 × 0.13 × 0.11 mm3, triclinic, space group P (no. 2), a = 6.9746(10), b = 12.0647(17), c = 13.1454(18) Å, α = 72.422(2), β = 87.762(2), γ = 86.363(2)°, V = 1052.1(3) Å3, Z = 1, Dc = 1.599 g cm−3, F000 = 509, Bruker APEX-II CCD, MoKα radiation, λ = 0.71073 Å, T = 100(2) K, 2θmax = 56.3°, 4849 reflections collected, 4849 unique (Rint = 0.0354). Final GooF = 1.026, R1 = 0.0412, wR2 = 0.0911, R indices based on 4131 reflections with I > 2sigma(I) (refinement on F2), 274 parameters, 0 restraints. Lp and absorption corrections applied, μ = 1.005 mm−1. |
This journal is © The Royal Society of Chemistry 2014 |