Exopolyhedral Ligand Orientation Controls Diastereoisomer in Mixed-Metal Bis(Carboranes) †

Heterobimetallic derivatives of a bis(carborane), [μ7,8-(1′,3′−3′-Cl-3′-PPh3-closo-3′,1′,2′-RhC2B9H10)-2-(p-cymene)-closo-2,1,8-RuC2B9H10] (1) and [μ7,8-(1′,3′−3′-Cl-3′-PPh3-closo-3′,1′,2′-RhC2B9H10)-2-Cp-closo-2,1,8-CoC2B9H10] (2) have been synthesised and characterised, including crystallographic studies. A minor co-product during the synthesis of compound 2 is the new species [8-{8′-2′-H-2′,2′-(PPh3)2-closo-2′,1′,8′-RhC2B9H10}-2-Cp-closo-2,1,8-CoC2B9H10] (3), isolated as a mixture of diastereoisomers. Although, in principle, compounds 1 and 2 could also exist as two diastereoisomers, only one (the same in both cases) is formed. It is suggested that the preferred exopolyhedral ligand orientation in the rhodacarboranes in the non-observed diastereoisomers would lead to unacceptable steric crowding between the PPh3 ligand and either the p-cymene (compound 1) or Cp (compound 2) ligand of the ruthenacarborane or cobaltacarborane, respectively.

Specifically, [8- [8]. Compounds of type IV were found to be active catalyst precursors for alkene isomerisation and the hydrosilylation of acetophenone.
Seeking to expand the scope of this chemistry, we have now investigated the reactions of III [M = Ru, L = (p-cymene); M = Co, L = Cp] with [Rh(PPh 3 ) 3 Cl] under different conditions and here present the results. Whilst these new reactions also afford compounds of type IV as minor co-products, the major species produced are unique heterometalated derivatives of bis(carborane), each isolated in only one of two possible diastereoisomeric forms.

Synthesis and Characterisation of Compound 1
We have previously reacted III [M = Ru, L = (p-cymene); III Ru ] following deprotonation, with [Rh(PPh 3 ) 3 Cl] under overnight reflux in tetrahydrofuran (THF) to afford IV Ru as a mixture of diastereoisomers with a total yield >50%. Repeating this reaction at room temperature, after workup involving preparative thin-layer chromatography (TLC), resulted in the isolation of a new species 1 as the main product, albeit with a modest yield (11%). Also isolated were trace amounts (<1%) of both diastereoisomers of the known species IV Ru , identified by multinuclear NMR spectroscopy [8].
The 1 H NMR spectrum of 1 reveals a number of interesting features. The resonances due to the p-cymene ligand demonstrated that the molecule is asymmetric, as expected, but the two multiplets (each due to 2H) for the aromatic protons are at very different chemical shifts, δ 6.40-6.36 and 4.97-4.92 ppm. Moreover, while there is clearly only one PPh 3 ligand present, the resonances due to it appear as two multiplets, one of which is at a relatively high frequency (δ 7.82-7.75 ppm) and integrates for 5H. These results are consistent with a p-cymene ligand locked in conformation and a PPh 3 ligand in which one ring is in a unique environment. Two C cage H resonances are present, one at δ 2.68 ppm assigned to the ruthenacarborane cage by analogy with the resonance in IV Ru [8], and the other at much higher frequency, δ 5.02 ppm, consistent with a non-isomerised 3,1,2-RhC 2 B 9 rhodacarborane cage.
There are three crystallographically independent molecules of 1 in the asymmetric fraction of the unit cell, a relatively rare occurrence (<0.5% of structures in the Cambridge Crystallographic Database (CCD) [10] have Z = 3), and all three molecules are practically superimposable. Figure 2 shows a space-filling representation of one molecule viewed from above the p-cymene ring. The C3−H3 bond of the p-cymene points towards the centre of one of the phenyl rings of the PPh 3 ligand, with H3 lying only 2.52-2.71 Å from the ring centre, and we believe that this effectively locks both the p-cymene ligand and the Ph group in fixed positions even in solution, consistent with the 1 H NMR spectrum discussed above.

Synthesis and Characterisation of Compounds 2 and 3
Although deprotonation of III CoCp* with n BuLi followed by treatment with [Rh(PPh 3 ) 3 Cl] leads to the heterobimetallic IV CoCp* as a mixture of diastereoisomers [8], the same approach cannot be used with the Cp analog III CoCp because of the attack on the Cp ring by n BuLi. Accordingly, we have reverted to direct reaction between [HNMe 3 ][8- The presence of two diastereoisomers is evident from the observation of two Cp and two high-frequency C cage H resonances in the 1 H NMR spectrum, the latter assigned to the cobaltacarborane cage, and the diastereoisomeric ratio is approximately 1:1.5.
However, the major reaction product (2), fully analogous to compound 1. As was the case with 1, the 1 H NMR spectrum of 2 suggests that one Ph ring (δ 7.81-7.72 ppm) is in a unique environment, and the results of a crystallographic study support this. Although the structure of compound 2 is not particularly precise, it is unambiguous. There are four crystallographically independent molecules in the asymmetric fraction of the unit cell, again a relatively rare occurrence-somewhat surprisingly, the CCD reports slightly more structures with Z = 4 than Z = 3 (4861 c.f. 4654), but in a database of ca. 10 6 this is still <0.5%. All four independent molecules of 2 are very similar, and a representative example is shown in Figure 3. The structure of compound 2 bears a close similarity to that of compound 1. One Ph ring stands approximately perpendicular to the plane of the Cp ring, and there are close contacts between one Cp H atom and the Ph ring centroid, ca. 2.6-3.0 Å. A space-filling view of the molecule from above the Cp ring, as shown in Figure 4, is remarkably similar to the analogous representation of compound 1, and fully consistent with a crowded molecule in which there is contact between the Cp ligand and one Ph ring. Although at room temperature in solution the Cp ring is clearly able to rotate (a singlet observed in the 1 H NMR spectrum), in the solid state the Ph ring appears locked in conformation.   Because anion III, the precursor to compounds 1-3, exists as a mixture of diastereoisomers, products 1-3 would reasonably also be expected to be diastereoisomeric mixtures. Whilst this is true for compound 3, both 1 and 2 are only isolated as one diastereoisomer. There is no evidence for separation into two diastereoisomers on workup by TLC, nor is there evidence of diastereoisomers in the NMR spectra of 1 and 2. Both 1 and 2 crystallise with multiple independent molecules in the asymmetric fraction of the unit cell (three for compound 1 and four for compound 2), but all these multiple molecules are of the same diastereoisomeric form both within and between each compound. Figure 5 shows the structure of molecule 1 viewed from above the Rh atom. The observed diastereoisomer is defined by C at the 2 position of the rhodacarborane cage and B at the 4 position, and the "missing" diastereoisomer would be defined by B at 2 and C at 4 . Thus, identification of the correct diastereoisomer is dependent on the correct assignment of C and B vertices in the structural studies. Whilst distinguishing between C and B vertices crystallographically has traditionally sometimes been challenging, we have recently developed powerful new methods to overcome this problem, specifically the vertex-centroid distance (VCD) [12] and boron-hydrogen distance (BHD) [13] methods. Both approaches were used for compound 1, but the relative imprecision of the structural study of 2 meant that cage H atoms could not be reliably refined, restricting the structure of 2 to an analysis by only the VCD method. Nevertheless, in both compounds the non-linking cage C atoms were clearly identified in both cages in all seven crystallographically independent molecules, confirming the same single diastereoisomer in all cases. We believe that one reason for the complementary diastereoisomer not forming could be due to the preferred exopolyhedral ligand orientation (ELO) of metal-ligand fragments in metallacarboranes [14]. It is well established that in a carborane ligand, the C atoms in the open face contribute less to the frontier molecular orbitals of the ligand than the B atoms [15], resulting in the C atoms having a weaker structural trans effect (trans influence). Therefore, the preferred ELO is that in which the exopolyhedral ligand with the strongest structural trans effect lies trans to the cage C atoms. In the case of the rhodacarborane components of 1 and 2, the strongest exopolyhedral ligand will be PPh 3 , rationalising the ELO observed in the observed diastereoisomer. For the non-observed diastereoisomer, with C atoms at the 1 and 4 vertices, the preferred position of the PPh 3 ligand would be above B7 , trans to the 1 −4 connectivity. However, the Rh atom is bound to the ruthenacarborane (compound 1) or cobaltacarborane (compound 2) cage via the B7H7 unit as part of a Rh3 H7B7C8C1 cycle that is likely to restrict full orientational freedom of the ligand set on Rh3 . Consequently, the likely outcome for the non-observed diastereoisomer is simply that the PPh 3 and Cl ligands would effectively exchange places. The observed diastereoisomers are already crowded species, evidenced by the interactions between the p-cymene ligand in 1 and the Cp ligand in 2 with one of the phenyl rings of PPh 3 . We anticipate that the crowding in the complementary diastereoisomer, with the PPh 3 ligand in effectively the same position occupied by the Cl ligand (the green atoms in Figures 2 and 4) in the observed form, would simply be untenable.

General Considerations
All experiments were performed under an atmosphere of dry nitrogen using standard Schlenk techniques with some subsequent manipulations and purifications carried out in the air.
All solvents were freeze-pump-thawed three times before use.

Crystallographic Studies
For compound 1, cage C atoms were distinguished from B atoms by application of the VCD and BHD methods [12][13][14]. However, for compound 2, the fact that cage H atoms except H7 could not be positionally refined restricted us to the VCD method. Figures 1, 3 and 5 were drawn with OLEX2 [19]. Figures 2 and 4 were drawn with Mercury [21].

Conclusions
Two examples of species in which a 2,1,8-MC 2 B 9 metallacarborane is linked to a 3 ,1 ,2 -RhC 2 B 9 metallacarborane via a direct C8−C1 bond and a B7−H7 Rh3 B-agostic interaction are reported. Although such compounds could exist as diastereoisomers, only a single, common diastereoisomer is observed both spectroscopically and as a result of crystallographic studies involving a total of seven crystallographically independent molecules. In both species, there is clear evidence of steric congestion between one Ph ring of the PPh 3 ligand on Rh and the ligand η-bonded to M. Consideration of the preferred exopolyhedral ligand orientation about Rh in the non-observed diastereoisomer suggests that it would be too sterically crowded to form.