Access to cationic polyhedral carboranes via dynamic cage surgery with N-heterocyclic carbenes

Polyhedral boranes and heteroboranes appear almost exclusively as neutral or anionic species, while the cationic ones are protonated at exoskeletal heteroatoms or they are instable. Here we report the reactivity of 10-vertex closo-dicarbadecaboranes with one or two equivalents of N-heterocyclic carbene to 10-vertex nido mono- and/or bis-carbene adducts, respectively. These complexes easily undergo a reaction with HCl to give cages of stable and water soluble 10-vertex nido-type cations with protonation in the form of a BHB bridge or 10-vertex closo-type cations containing one carbene ligand when originating from closo-1,10-dicarbadecaborane. The reaction of a 10-vertex nido mono-carbene adduct with phosphorus trichloride gives nido-11-vertex 2-phospha-7,8-dicarbaundecaborane, which undergoes an oxidation of the phosphorus atom to P = O, while the product of a bis-carbene adduct reaction is best described as a distorted C2B6H8 fragment bridged by the (BH)2PCl2+ moiety.


S5
Synthesis of 10-NHC Dip -5,6-C2B8H10 (o-1)   Solution of closo-o-C2B8H10 (89 mg, 0.74 mmol) in diethyl ether (5 mL) was added dropwise to a stirred solution of NHC Dip (688 mg, 1.77 mmol) in diethyl ether (15 mL) at room temperature. The slightly yellow reaction mixture was stirred for one hour and then the volatiles were removed in vacuo. The crude product was washed with hexane (15 mL  Synthesis of 6,9-NHC Dip 2-5,7-C2B8H10 (m-2) Solution of NHC Dip (425 mg, 1.09 mmol) in diethyl ether (10 mL) was added dropwise to a stirred solution of closo-m-C2B8H10 (66 mg, 0.55 mmol) in diethyl ether (5 mL) at room temperature. The turbid reaction mixture was filtrated and the slightly yellow solution was left for crystallization for 5 hours. The crude product was washed with hexane (15 mL) to remove the excess NHC Dip . The slightly yellow powder of m-2 was dried in vacuo. Colorless single crystals of m-2

Supplementary Discussion
Position of the hydrogen bridge in o-1, o-2 and o-2a As it was mentioned in the main text, in the structure of o-1, the carbene moiety is coordinated to a B-vertice without a hydrogen atom, which moved to the bridging position between boron atoms B8 and B9 (observed by 1 H NMR, see Supplementary Figure 53). Addition of the second equivalent of NHC yielded compound o-2, however, we were not able to detect a hydrogen bridge in its structure (the same works for m-2 and p-2) in both NMR in solution (Supplementary Figure 53) or scXRD (m-2 and p-2) in the solid state. There is a possibility of the presence of an equilibrium between the isomer with two IPr→B-H vertices (Supplementary Figure 54A) and isomers with one or two hydrogen bridges (Supplementary Figure 54 B, C and D), which has not been proved experimentally by the NMR spectroscopy (not even in the range of -50 -+50 °C). Such an equilibrium could also explain the absence of any cross-peaks in 11  Thermal decomposition of o-2, m-2 and p-2 As it has been already mentioned in the paper, o-2 and m-2 decompose at 100 °C in thf to the parent carborane p. Both reactions were monitored by the help of 11 B NMR spectroscopy during one month (o-2 - Supplementary Figure 55 and -m-2 Supplementary Figure 56). In both cases the first step is formation of an intermediate, which we were, unfortunately, not able to isolate and fully characterize. However, both intermediates exhibit very similar patterns in 11 B NMR spectra as the parent compounds. Moreover, none of the intermediates is p-2, which decomposes under the same conditions to p in one day. The decomposition of m-2 is completed within one month while o-2 shows ca 5 % conversion after the same time. Compared to the rearrangement of the parent carboranes o and m to p, our method uses more gentle reaction conditions. However, considering the long reaction time and the presence of decomposition products of the carbene makes it inapplicable in bulk synthesis. The published method is much more simple, faster and high-yielding. Nevertheless, such kind of decomposition at lower temperatures has not been described yet and definitely deserves attention in the future studies.

Synthesis of o-1a and o-1b
We proposed a simple mechanism of the formation of o-1a cluster (Supplementary Figure 57). First, the phosphorus atom is bound to atoms B8 and B9 (originally the position of the hydrogen bridge) and atom C6 (Supplementary Figure 57A). Subsequently, the bonds of the B2 atom with other boron atoms are cleaved (Supplementary Figure 57B) and B2 is turned around the C-C axis forming three new bonds (Supplementary Figure 57D). Finally, one last bond between the phosphorus atom and B11 (numbering of o-1a) is formed and a mirror analogue of o-1a is obtained. In the crystal structure, both mirror forms of o-1a are present in the solid state according to the X-Ray analysis. Both can be obtained by the same mechanism from the corresponding isomer of o-1, which presence was not confirmed not excluded in the solid state by the X-Ray analysis, however, since the difference between both forms is only the position of hydrogen bridge in two, chemically equivalent positions, it is highly possible, that both isomers are present in the solution.
Supplementary Figure 57. Plausible mechanism of the formation of o-1a.
We have proposed a plausible mechanism for the rearrangement of the borane cage (Figure 4), which consists of oxidation as the initial step (A), followed by a simple cleavage of three skeletal bonds of atom B1 (B) with its subsequent turn around the C-C (C) bond, and finally the closure of the borane cage.
Supplementary Figure 58. A plausible mechanism for the atom rearrangement after the oxidation of o-1a to o-1b.

Supplementary Computational Data
All the calculations (except GIAO-MP2, see bellow) were performed with the Gaussian 16 program. 6 The geometries of the compounds (o, m, p, o-1, m-1, p-1, o-2, m-2, p-2) were fully optimized at B3LYP-D3/cc-pVDZ level of theory 7 without any simplifications. The structures of complexes obtained by X-ray diffraction were used as the initial data. The polarizable continuum model (PCM) 8 was employed for the solvation effects (diethyl ether). All the structures are minima on the potential energy surface, as confirmed by the frequency calculations at the same level of theory and transition states by only one imaginary frequency. The topological analysis of the theoretical function ρ(r) was performed using the AIMALL program package. 9 Thus structures (o-1, o-2, o-2a, o-2b, m-2, m-2a, p-2, p-2a) were fully optimized at B3LYP/cc-pVTZ level of theory. The numbering of atoms corresponds to the numbering for XRD.

NMR Shifts
Magnetic shielding was calculated using the GIAO-MP2 method incorporated into Gaussian 16 10 utilizing the IGLO-II basis with the MP2/TZVP geometries and frozen core electrons. In order to reduce the GIAO-MP2 jobs to manageable dimensions, we approximated the Dipp groups with hydrogen atoms. Earlier work 11 provided a very good back-up for such an assumption. The best-fit assignments reveal very good accords between the GIAO-MP2 and experimental values (Supplementary p-2a Comp. C -13.9 -11.0 -9.6 -9.6 -12.6 -7.9 -9.6 -14.9 C X Exp.
-13.6 -11.6 -11.6 -11.6 -11.6 -11.6 -11.6 -13.6 Reactions of o, m and p with :IPr To understand the differences in the reactions of o, m and p with :IPr, the geometries of the ground states were optimized, and then the corresponding free energies were calculated (Supplementary Figure 62). The ΔGs calculated in order to shed more light on the reactivity of parent carboranes are consistent with the experiment. The low energy of the first reaction with o (−27.19 kcal/mol) and the higher ΔG of the second one (-13.36 kcal/mol) are explained both by the simultaneous presence of compounds o-1 and o-2 in the reaction mixture and by the willingness of o to react. Contrastingly, the high value of the ΔG of the reaction of p with the first :IPr (8.98 kcal/mol) and the negative ΔG value for the subsequent :IPr addition (−32.57 kcal/mol) explain the reluctance of p to react and the impossibility to detect p-1. The reactions of m exhibit highly negative values of both ΔGs (-24.10 and -17.2 kcal/mol), which also confirm reaction progress with no intermediate.
Supplementary Figure 62. The DFT-estimated (B3LYP-D3/cc-pVDZ theory level) relative Gibbs free energy comparison for reactants and products (kcal/mol). (Notes: the lines connecting the red, green and blue energy levels are shown only as a guide to the reader's eye. The relative Gibbs free energies of parent carboranes o, m and p are obviously not 0, but 38.3, 20.2 and 0 kcal/mol for o, m and p -the comparative approach is used here in order to illustrate differences between reactants and products in all series more clearly).

QTAIM
Within the framework of R. Bader's theory "Atoms in Molecules", the atomic charges in compounds o-1, o-2, m-2, p-2 were calculated (Supplementary The atomic charges also shed more light on the diversity in the reactions of o-2, m-2 and p-2 with hydrogen chloride. The elimination of the dihydrogen in the case of p-2 is most probably caused by the impossibility to form stable compound containing a hydrogen bridge. In o-2 and m-2, the hydrogen bridge is placed between the least positively charged B-atoms in the upper rim of the carborane cage (0.68 and 0.665 e for o-2; 1.086 and 0.75 e for m-2). On the contrary, p-2 does not have such low atomic charges in the upper rim (1.124-1.306 e). Moreover, the distant position of the C-atoms also lowers the number of plausible locations for the hydrogen bridge. Therefore, the p-2H + adduct releases imidazolium and the newly formed p-1 (which is not thermodynamically very stable, see above) reacts with another equivalent of hydrogen chloride releasing dihydrogen and closing the carborane cage to p-2a.
An analysis of the distribution the Laplacian of the electron density ( 2 ρ(r)) of o-2b in the B6-P11-B8 plane demonstrates the regions of electron density (ED) concentration between those atoms. Taking into account the other topological parameters at these bond critical points (BCP, Supplementary Table 13 programs. 12,13 The ESP color range is in kcal/mol. The negative surface is always located on Cland HCl2parts, respectively. The positive surface is always found on whole carborane moiety, except of one BH vertex next to the carbene(s). The areas with the most positive ESP, which interact with counter-anion, are the CH vertexes of carborane cages and CH=CH fragment of the NHCs.