2.1 Synthesis and characterisation

The reaction of the mono-ethynyl carborane 2 [22] with Co₂(CO)₆(μ-dppm) in refluxing benzene afforded dark red 4 (40%), which was isolated from the reaction mixture by preparative TLC (Scheme 1). The IR ν(CO) spectrum contained terminal ν(CO) bands between 2077 and 1977 cm^–1 in addition to the ν(BH) bands centred at 2656 cm^–1, while in the ¹H NMR the methylene protons of the dppm ligand were observed as partially resolved resonances centred at 3.27 and 3.40 ppm. The negative ion ES-mass spectrum of a solution in methanol containing NaOMe was characterised by isotopic envelopes from the [M–H]^– and [M–H–CO]^– ions [23]. We note here that 4 is not the first mixed carborane/organometallic dicluster of its kind; the complex [Co₂(CO)₆](μ-HC₂CB₁₀H₁₀CH) has been prepared from 1-ethynyl-ortho-carborane and Co₂(CO)₆ but was not structurally characterised [24].

The reaction of the diethynyl carborane 3 [13] with two equivalents of Co₂(CO)₆(μ-dppm) afforded the tricluster 5 in moderate yield (40%). Coordination of each ethynyl moiety in 3 by the Co₂(CO)₄(μ-dppm) fragment was clearly evidenced from the IR spectrum which contained both ν(BH) and ν(CO) stretches, but no ν(C≡C) band, the characteristic ¹H resonances in the NMR spectrum of the dppm ligand methylene protons and a single resonance from the protons of the SiMe₃ groups in the correct ratio for 5, and the FAB mass spectrum, which contained a series of fragment ions derived from 5 by loss of CO ligands.

For purposes of comparison, the complexes {Co₂(CO)₄(μ-dppm)}₂(μ-Me₃SiC₂C₆H₄C₂SiMe₃) (6) [25] and {Co₂(CO)₄(μ-dppm)}₂(μ-Me₃SiC₂C≡CC≡CC₂SiMe₃) (7) [26] were also obtained, using minor variations of the methods described previously by others.

2.2 Molecular structures

Single crystals of 4 (Fig. 1), 5 [21] and 6 (Fig. 2) suitable for X-ray diffraction were grown by slow diffusion of MeOH into a concentrated CH₂Cl₂ solution of each complex (Table 1). Significant bond lengths and angles are summarised in either the appropriate figure captions or Table 2, which also includes the appropriate parameters associated with 7 for comparison [26]. All three molecules sit in a symmetry element. Molecule 4 is bisected by a plane parallel to the longest molecular axis and located at (0 ¹/₄ 0), and molecules 5 and 6 lay in an inversion centre located at (¹/₂ 0 ¹/₂) and (¹/₄ ¹/₄ ¹/₄), respectively. The crystal packing is driven by the close-packing principle [27]. Molecules 5 and 6 present small cavities between molecules, where solvent molecules are trapped. The bulkier nature of 5 and 6, containing two dppm moieties, compared to only one dppm moiety in 4, is likely the cause of the reduced packing efficiency of molecules 5 and 6.

The bis(phosphine) ligand in 4 lies approximately cis to the C₂B₁₀H₁₁ icosahedron. The two ordered phenyl rings of the dppm ligand adopt an unusual co-planar arrangement, and this probably reflects the most efficient packing motif within the solid-state structure. The two Co₂C₂ tetrahedra of 5 are related by a crystallographic inversion centre at the core of the C₂B₁₀ icosahedron, and, in contrast to 4, the dppm ligands in 5 are found occupying positions approximately trans to the carborane cage. In the case of the phenyl-substituted derivative the dppm ligands adopt an anti arrangement.

Within the carborane cage portion of 4, the C(5)−B bonds are somewhat shorter [1.687(3)–1.703(3) Å] than for C(1)−B [1.714(3)–1.738(4) Å] while the B−B distances are experimentally equivalent [1.755(4)–1.776(4) Å]. The lengthening of C−B bonds in carborane cages caused by C-substitution is well-documented, although variation in B−B distances is relatively rare [13]. The C(1)−C(2) bond distance [1.510(3) Å] is considerably elongated in comparison to C(carborane)−C(ethynyl) distances in 3 [1.453(2) Å] and diethynyl carborane [1.449(2)/1.452(2) (two independent molecules)], but comparable with that found in 5 [1.497(2) Å]. The C(2)−C(3) [1.360(3) Å], C(2,3)−Co [1.977(2), 1.981(2) Å] and Co–Co [2.4737(4) Å] separations are within the normal limits [28].

The carborane cage in 5 is one of the most spherical structurally characterised examples of a para-carborane known to date. For example, the C(1)···C(1) separation [3.176(7) Å] and average tropical B−B bond lengths [1.772(4) Å] are in sharp contrast with the more ovoid shape of the carborane cages in 3 [12] and 1,12-bis(ethynyl)-para-carborane 8 [C^…C 3.104(2) Å; avg. tropical B−B 1.793(3) Å] [13]. The elongation of the C(1)−C(2) and C(2)−C(3) bond lengths in 5 compared to 3 are as expected, and reflect the different nature of the C(2) and C(3) centres in each complex. The parameters associated with Co₂C₂ clusters fall within the normal ranges.

The molecular parameters of the C₆H₄ spaced species 6 are comparable with those of other dicobalt complexes of this type derived from donor-substituted phenyl acetylenes [28]: the C(3)−C(2)−C(1)−C(4) torsion angle is 44.7(3)° and there is a small twist of the ligating groups around the Co(1)−Co(2) bond axis away from an ideal eclipsed geometry. The Co(1)−Co(2) bond length [2.488(1) Å] is similar to that observed for Co₂(μ-HC₂Ph)(CO)₄(μ-dppm), while the C(3)−C(2) bond length [1.357(2) Å] is at the longer end of the range usually encountered.

2.3 Electrochemical characterisation

Electrochemical studies have been used to demonstrate communication between two Co₂C₂ clusters of this type via aryl, alkynyl or ferrocenyl entities [18,29,30]. An examination of the electrochemical response of 4 using cyclic voltammetry revealed both a single oxidation process and a single reduction (E_1/2 = +0.82 V, –1.60 V, vs. SCE, decamethylferrocene/decamethylferrocinium (Fc*/Fc*⁺) = 0.084 V in THF containing 0.1 M [NBu₄]PF₆). These redox processes were chemically and electrochemically reversible at a moderate scan rate (0.100 V s^–1), with i_p^a/i_p^c ca. 1 and ΔE_p ca. 60 mV, identical to that of the Fc*/Fc*⁺ couple used as an internal standard. The reversibility of the redox processes is notable, with both the dppm ligand and the steric bulk of the carborane cage likely to be playing a role in the stabilisation of both the oxidised and reduced forms of 4.

A detailed electrochemical study of 5 revealed both oxidation and reduction processes at + 0.87, + 0.77 and –1.58, –1.66 V (vs. SCE, Fc*/Fc*⁺ = 0.084 V); ΔE⁰(ox) ca. 105 mV, K_C = 60; ΔE⁰(red) ca. 80 mV, K_C = 5], respectively [21]. No redox processes were found within the accessible potential window when 3 was subjected to a similar electrochemical study. For comparison, the phenyl-bridged complex [{Co₂C₂(H)(CO)₄(μ-dppm)}₂(μ-1,4-C₆H₄)] (6) also gave two oxidation and reduction processes with ΔE⁰(ox) 110 mV, ΔE⁰(red) 80 mV. The electrochemical response of the butadiyndiyl spaced derivative [{Co₂C₂(SiMe₃)(CO)₄(μ-dppm)}₂(μ-C₄)] (7) was complicated by the chemical irreversibility of the redox processes. Two chemically irreversible oxidations were observed with peak potentials +0.99 and +1.26 V, and single reduction at –1.32 V. On the basis of this electrochemical work it is not possible to identify or quantify the electronic interactions occurring between the Co₂C₂ redox centres via the C₄ bridge.

The difference in oxidation potentials observed in 5 and 6 is significant, especially given the > 6 Å separation of the cobalt cluster cores, and indicates a moderate degree of electronic communication between the cobalt clusters in the radical cations 5⁺ and 6⁺. The separation of the reduction events is somewhat smaller, but is still larger than the statistical limit (ca. 36 mV at 298 K) predicted for non-interacting redox centres [30]. The electrochemical studies clearly indicate that the bonding framework of the para-carborane cage and the para-C₆H₄ ring provides a conduit for electronic effects between the cluster substituents in the 1,12 and 1,4 positions, respectively, as a through space mechanism would perhaps be expected to give rise to more similarly spaced oxidation and reduction processes, although such process could be influenced by solvation and charge effects.

In general terms, the interaction between the orbitals of a M₂L₆ fragment with those of an acetylenic fragment gives rise to a pronounced bending of the acetylenic chain and to a splitting of both the π and π* C–C orbitals [31]. If these orbitals are located in the HOMO–LUMO region, or mix with the frontier orbitals of a spacer group, they will govern the chemistry of this type of complex, and in this context it is important to note that the Co₂C₂ tetrahedron offers a high-lying fragment orbital of the correct symmetry to interact with aryl [28] and ethynyl [19c] groups. In light of the electrochemical properties of 5, 6 described above and elsewhere, and the ongoing interest in redox-active organometallic species spanned by C₄ [32], it was of interest to probe the mechanism through which electronic effects are transmitted between Co₂C₂ fragments via the 1,12-para-carborane cage, the 1,4-C₆H₄ ring and the butadiyndiyl fragment in more detail.

2.4 Theoretical description

Density functional-theory (DFT) molecular-orbital calculations were carried out on dicobalt complexes of 1,12-diethynyl-para-carborane (5-H), 1,4-diethynylbenzene (6-H) and 1,3,5,7-octatetrayne (7-H) in order to further the understanding of their electronic properties¹. In order to verify the accuracy of the computational method, ‘free’ 1,12-diethynyl-para-carborane of D_5d symmetry (3-H) was studied in the first instance, although we note that the electronic structure of 3-H has been calculated at the ab initio Hartree–Fock (HF) level of theory on a previous occasion [12]. Pertinent results from the geometry optimisations are shown in Table 3, together with the experimentally observed values. Both HF and DFT optimised structures of 3-H compare well with the crystallographically determined structure of 8 [13]. Interestingly enough, the C≡C distances are slightly better reproduced at the HF level since the largest deviation is only 0.005 Å, whereas a deviation of 0.030 Å is observed at the DFT level. On the other hand the DFT computed B–B and B–C distances of the carborane cluster are in a better agreement with the X-ray structure. The maximum deviation concerns some B–B distances computed 0.007 Å shorter than the experimental values.

The hydrogen-substituted model complex [Co₂(CO)₄(PH₃)₂]₂(μ-H₃SiC₂CB₁₀H₁₀CC₂SiH₃) (5-H) of C_i symmetry, was used to mimic complex 5. Model 5-H quite satisfactorily mimics the metric distances of the crystallographically determined structure (Table 2). The computed Co–Co distance is slightly overestimated (2.527 vs. 2.4804(4) Å), mainly due to the use of the Becke–Perdew non-local corrections. The important geometry variations between 3 and 5 noted experimentally are reproduced in the computed results and, for example, coordination of the Co₂ groups results in a large increase in the ethynyl C–C bond length and a smaller increase in the C(sp)–C(sp³) separation. The calculated B–B distances are virtually unaffected by coordination of the ethynyl moieties to the metal centres.

2.5 Electronic structure

The DF molecular orbital diagram of the neutral complex 5-H is shown on the left-hand side of the Fig. 3. A large energy gap (1.966 eV) separates the HOMOs and a set of two LUMOs (57a_g and 58a_g) which are also separated from the other vacant orbitals by a gap of around 1 eV. It is informative to look at the composition of the HOMOs of 5-H before considering the structural and electronic changes induced by oxidation. To this end a Mulliken atomic orbital population analyses for selected MOs of 5-H is given in the Table 4. It appears that the orbitals of the HOMO–LUMO region present a preponderant metallic character. For instance, the HOMOs 57a_u and 56a_g are 71% and 73% Co in character, respectively, with a weak participation of the carbon atoms C(3) and C(2). The rest of the electronic contribution is localised on the PH₃ and CO ligands bound to the metallic atoms. These MOs are mainly of π-type Co–Co antibonding. These results are displayed graphically in Figs. 3 and 4a.

Using a C_i restrained geometry, one electron was removed from the structure 5-H and the structure re-optimised. For comparison, calculations were also carried on the broken symmetry geometry [5-H]⁺ (C₁ symmetry). A comparison of the structural parameters calculated for 5-H and [5-H]⁺, both with full symmetry (FS) and broken symmetry (BS) reveals in a decrease in the metal-metal bond length, a contraction of C(3)–C(2) bond and a lengthening of the various C–Co separations upon oxidation. The weak antibonding and bonding character of the HOMO with respect to these connections provides a simple rationalisation of these structural variations. Interestingly, a slight asymmetry is computed for [5-H(BS)]⁺ (see Table 2), which is less stable than [5-H(FS)]⁺ by 4.23 kcal mol^–1. The latter result suggests than such a geometry, which would favour some valence-localised structure character, is unlikely with respect to the valence-delocalised structure.

A study of the spin density of the complex [5-H]⁺ indicates a net localisation of the single electron on the Co atoms with some uneven distribution over Co1 and Co2 (see Table 5). The DFT calculations indicate that the π-orbitals on the carborane entity do not participate much in the partially occupied MO and therefore should not play an important role in the communication between the metallic clusters in the complex 5. Rather, the separation of the electrochemical events observed voltammetrically must be attributed to inductive effects transmitted by polarisation of the σ-bonding framework.

Attention is next turned to the structures and spin states of the dicationic species [5-H]²⁺. Oxidation of 5-H to the high-spin (triplet) state [5-H(T)]²⁺, which was calculated to be some 5.63 kcal mol^–1 more stable than the low-spin (singlet) state [5-H(S)]²⁺, does not result in any notable structural modification of the carborane moiety or the C(2)–C(1) distance. However, the intra-cluster Co₂C₂ bond distances are more significantly affected. The metal–metal distance decreases by some 0.116 Å (4.5%), while there is a smaller contraction of the C(3)–C(2) bond length (2.5%). The C–Co distances increase between 3% and 4% overall. Again, these calculated results are consistent with the nodal properties of the HOMOs (vide supra) and with an interpretation of the electrochemical response in terms of Co₂C₂ centred oxidation processes.

2.6 Influence of the nature of the spacer entity

In order to compare the electronic properties of the complex 5 with those of the complexes [Co₂C₂(SiMe₃)(CO)₄(μ-dppm)]₂(μ-1,4-C₆H₄) (6) and [Co₂C₂(SiMe₃)(CO)₄(μ-dppm)]₂(μ-C≡CC≡C) (7), DFT calculations have been carried out on the model complexes [Co₂C₂(H)(CO)₄(μ-dppm)]₂(μ-1,4-C₆H₄) ([6-H]ⁿ⁺) and [Co₂C₂(H)(CO)₄(μ-dppm)]₂(μ-C≡CC≡C) ([7-H]ⁿ⁺) (n = 0–2). The optimised structural characteristics of these two model complexes are given in Table 2. The optimised distances of the neutral complexes 6-H and 7-H are in good agreement with the crystallographically determined structures, with some over-estimation of the Co–Co distances attributed to the previously mentioned methodological reasons. Both the observed and calculated C(3)–C(2)–C(1) angles vary across the series of complexes 5 (5-H), 6 (6-H) and 7 (7-H) being 139.2(2) (139.2°), 136.5(3) (145.8°) and 144.4(9) (142.8°), respectively (see Figs. 1 and 2 for a representative atom numbering scheme). As might be expected, the bulkier the spacer entity, the more pronounced the bending of the chain, although the observed deviation from linearity is greater in the case of 6 than 6-H.

The DF molecular orbital diagrams for the neutral complexes 6-H and 7-H are shown in Fig. 3, and illustrate the large HOMO–LUMO gap calculated in each case (1.623 and 1.417 eV, respectively). The HOMO and HOMO-1 of the complexes 6-H and 7-H are shown graphically in Fig. 4, with the energy and composition of these orbitals given in Table 5. The HOMO 49a_g and HOMO-1 48a_u of the neutral complex 6-H are mainly localised on the Co atoms (61% and 72%, respectively). The HOMO also contains a small contribution from the benzene spacer (12% C(1) and 8% C(4)) which is not present in the HOMO-1. In much the same way as has been described for 5-H, the strong antibonding character of the Co–Co and C(2)–C(3) interactions in these frontier orbitals helps explain the reduction of the Co–Co bond length calculated to occur upon oxidation, while the bonding nature of the Co–C (2, 3) interaction is consistent with the elongation of these bonds (Table 2). The elongation of the C(1)–C(4) bond and contraction of C(4)–C(4′) give some indication of a quinodal form, and suggests a more intimate role of the 1,4-C₆H₄ π-system in the interactions between the cluster centres than was found for the carborane cage.

The oxidation of 6-H to [6-H]²⁺ leads to two spin states (HS and LS), which are isoenergetic (less than 1 kcal mol^–1 in favour of the triplet state). As would be expected on the basis of the orbital composition of the HOMO and HOMO-1 in 6-H, in the case of the LS state the structural modifications take place on the whole Co₂C₂–C₆H₄–C₂Co₂ chain, and lead to a further emphasis of the quinoidal character of the bridge. In the case of the HS state, the greatest structural changes of significance are restricted to the intracluster Co₂C₂ distances (Table 2).

The HOMO (46a_g) and HOMO-1 (45a_g) of the complex 7-H, which lie close together in energy, are slightly separated from the other occupied orbitals, are of π-type and mainly localised on the Co atoms and on the C atoms of the spacing group. Whereas the HOMO is rather localised on the Co atoms, with a small, but significant contribution from the C₄ bridge (56% Co against 4%, 15% and 11% for the C(2), C(1) and C(4) atoms), the HOMO-1 presents an even more delocalised character with 25%, 11%, 7%, 32% and 22% on the Co, C(3), C(2), C(1) and C(4) atoms, respectively. Moreover, the bonding-antibonding properties of these orbitals are not totally equivalent and, for example, the Co–Co antibonding character of the HOMO is not observed in the HOMO-1.

In a manner entirely analogous to the monooxidation of 5-H and 6-H, oxidation of 7-H to [7-H]⁺ results in contraction of the Co–Co, C(1)–C(2) and C(2)–C(3) bonds reflecting the composition of the redox-active orbital. The structural parameters of the C₄ bridge are rather less affected than is generally the case of σ-bonded diyndiyl complexes [32], with a small elongation of the formal C(3)≡C(4) triple bond and contraction of the C(4)–C(4′) single bond (Table 4).

The structural parameters of the Co₂C₂–C₄–C₂Co₂ chain are further affected by oxidation to [7-H]²⁺ (Table 2). In the LS state of [7-H]²⁺, which is be calculated to be slightly more stable than the HS state by 2.62 kcal mol^–1, the Co–Co, C(3)–C(2), C(2)–C(1) and C(4)–C(4') distances decrease by 0.075, 0.016, 0.031 and 0.026 Å, respectively, when compared with 7-H, whereas the C(1)–C(4) bond length increases by 0.014 Å. The intra-cluster Co–C and the intra-bridge C(1)–C(4) bond lengths are further increased by this second oxidation process. Overall, the structural and electronic parameters of the oxidised forms of 7-H support a significant contribution from the C₄ π-system in promoting delocalisation of electron density between the Co₂C₂ centres, and an evolution from a formally diyndiyl C(1)≡C(4)–C≡C description to a more cumulenic C(1)=C(4)=C=C form. This is further supported by the increasing linearity of the C(3)–C(2)–C(1) fragment as oxidation proceeds.

The atomic spin densities calculated for [6-H]⁺ and [7-H]⁺ (Table 5) show that the unpaired electron is also mainly located on the Co atoms, as noted previously for [5-H]⁺. Nevertheless, the amount of spin density localised on the metal centres drops from 85% for [5-H]⁺ to 80% and 78% for [6-H]⁺ and [7-H]⁺ with an increasing amount of the unpaired electron spin found on the organic bridge, indicating a tendency for a somewhat higher delocalisation for the latter. The trend is even higher in the case of the dicationic high-spin species, in particular with complex 7-H, since 84%, 83%, and 59% of the two unpaired electrons are located at the cobalt centres for [5-H]⁺, [6-H]⁺, and [7-H]⁺, respectively.

2.7 Ionisation potentials

The first and second ionisation potentials have been computed for the three complexes 5-H (6.64 and 9.04 eV), 6-H (6.31 and 9.08 eV) and 7-H (6.33 and 9.14 eV) and are not strongly influenced by the nature of the bridging group, strongly supporting the concept that the oxidation processes will mainly affect the common part of these molecules i.e. the Co₂C₂ cluster. However, the larger difference between the first and second IPs for the complexes 6-H and 7-H with respect to 5-H (2.77, 2.81 vs. 2.40 eV), which is in agreement with the experimental electrochemical results from the real systems, gives some evidence for greater thermodynamic stability of the mono-oxidised forms [6-H]⁺ and [7-H]⁺, which may be due in part to the better electronic interaction between the Co₂C₂ clusters when linked by the aryl or ynyl moieties than the carborane.

4.1 General conditions

All reactions were carried out using standard Schlenk techniques under dry high-purity nitrogen. Solvents were dried and distilled prior to use. Preparative TLC was carried out on 20 × 20 cm glass plates coated with silica gel (Merck G₂₅₄, 0.5 mm thick). Reagents were purchased and used as received. Compounds 3 [13], 6 [25] and 7 [26] were prepared according to the literature methods with minor modifications. Compound 2 was made using a similar procedure to the reported synthesis of 3. IR spectra were recorded on a Nicolet Avatar spectrophotometer using solution cells fitted with CaF₂ windows. NMR spectra were obtained from solutions in CDCl₃ using Varian VXR-400 (¹H 399.97, ¹³C 100.57, ³¹P 161.1 MHz) or Bruker DRX-400 (¹H 400.13, ¹³C 100.61, ³¹P 162.05 MHz) spectrometers. FAB-MS were recorded at the EPSRC National Mass Spectrometry Service Centre (Swansea). Electrochemical experiments were conducted in THF solution containing 0.1 M [NBu₄][PF₆] using a simple three electrode cell and an EcoChemie Autolab PGSTAT-30, or an EG&G 273 potentiostat/galvanostat. Working electrodes were glassy carbon, with Pt wire counter and pseudo-reference electrodes. Potentials were corrected to SCE using an internal decamethylferrocene/decamethylferrocinium couple as standard (Fc*/Fc*⁺ = 0.084 V).

4.2 Preparation of [Co₂(CO)₄(μ-dppm)](μ-Me₃SiC₂CB₁₀H₁₀CH) (4)

A solution of [Co₂(CO)₆(μ-dppm)] (100 mg, 0.15 mmol) was treated with 2 (36 mg, 0.15 mmol) and heated at reflux point for 3 h. The solution was cooled, the solvent removed and the residue purified by preparative TLC (70:30 hexane/CH₂Cl₂). The major band was collected and crystallised (CH₂Cl₂/MeOH) to afford 4 (50 mg, 40%). ES-MS (m/z) 855, [M]⁺; 827, [M–CO]⁺. IR (cyclohexane}: v (BH) 2656; v(CO) 2027 m, 2002 s, 1977 s cm^–1. NMR (CDCl₃, δ) ¹H{¹¹B} 7.38–7.20 (m, 20H, Ph), 3.51, 3.37 (2 × dt, J_HP = J_HH = 12 Hz, 2 × 1H, CH₂), 2.61 (br, 6H, B7-11H + carboranyl CH), 2.15 (br, 5H, BH), 0.27 (s, 9H, SiMe₃); ¹¹B –10.8 (5B, B2-6), –15.6 (5B, B7-11); ³¹P 32.7.

4.3 Preparation of [Co₂(CO)₄(μ-dppm)]₂(μ-Me₃SiC₂CB₁₀H₁₀CC₂SiMe₃) (5)

A solution of 3 (83 mg, 0.24 mmol) in benzene (10 ml) was treated with [Co₂(CO)₆(μ-dppm)] (400 mg, 0.60 mmol) and heated at reflux for 2.5 h. The solvent was removed in vacuo and the dark brown residue obtained was purified by preparative TLC (CH₂Cl₂/hexane, 3:7) and crystallisation (CH₂Cl₂/MeOH) (150 mg, 40%). NMR (CDCl₃, δ) ¹H{¹¹B} 7.10–6.30 (m, 20H, Ph), 3.09, 3.25 (2 × dt, J_HP = J_HH = 11 Hz, 2 × 1H, CH₂), 1.28 (br, 5H, BH), 0.00 (s, 9H, SiMe₃); ¹¹B –11.0; FAB-MS (m/z) 1454, 1426, 1342 [M–nCO]⁺ (n = 4, 5, 7). IR (cyclohexane): ν(BH) 2659; ν(CO) 2026 m, 2002 s, 1976 s, 1964 m cm^–1. Found: C, 52.15%; H 4.56%. C₇₀H₇₂B₁₀O₈Si₂P₄Co₄.0.5CH₂Cl₂ requires C, 52.64%; H 4.42%.