Unprecedented formation of polycyclic diazadiborepine derivatives through cage deboronation of m-carborane

Icosahedral carboranes form an exciting and highly important class of inorganic cage compounds due to their numerous important applications, including energy storage, optoelectronics, nanomaterials, radiopharmaceuticals, medicine, and boron neutron capture therapy. Deboronation, i.e. the removal of a vertex from a carborane cage under formation of an anionic nido-C2B9 cluster, is one of the most important and longest known reactions of icosahedral carboranes. Deboronation of closo-carboranes is normally achieved by treatment with strong Lewis bases (neutral or anionic) such as fluorides, alkoxides, amines, or N-heterocyclic carbenes. Deboronation rates have been reported to decrease from oto p-carborane. According to the generally accepted mechanism (Scheme 1, illustrated for o-carborane (1)), deboronation starts with nucleophilic attack at one of the most positive boron vertices, which is followed by attack of a second nucleophile at the same B atom. Protonation of the resulting negatively charged nido-cluster leads to elimination of BH(Nu)2 which can undergo further reactions with nucleophiles. For example, in fluoride ion-initialized deboronation reactions, the fluoroborate anion HOBHF2 has been detected as monoborane by-product. On rare occasions, it has been possible to isolate type I (e.g. Nu = (Me2N)3PNH, 8a,10

Icosahedral carboranes form an exciting and highly important class of inorganic cage compounds due to their numerous important applications, including energy storage, optoelectronics, nanomaterials, radiopharmaceuticals, medicine, and boron neutron capture therapy. 1,2 Deboronation, i.e. the removal of a vertex from a carborane cage under formation of an anionic nido-C 2 B 9 cluster, is one of the most important and longest known reactions of icosahedral carboranes. 3,4 Deboronation of closo-carboranes is normally achieved by treatment with strong Lewis bases (neutral or anionic) such as fluorides, 5 alkoxides, 3b amines, 6 or N-heterocyclic carbenes. 7 Deboronation rates have been reported to decrease from o-to p-carborane. 5b According to the generally accepted mechanism (Scheme 1, illustrated for o-carborane (1)), deboronation starts with nucleophilic attack at one of the most positive boron vertices, which is followed by attack of a second nucleophile at the same B atom. 4 Protonation of the resulting negatively charged nido-cluster leads to elimination of BH(Nu) 2 which can undergo further reactions with nucleophiles. 7,8 For example, in fluoride ion-initialized deboronation reactions, the fluoroborate anion HOBHF 2 À has been detected as monoborane by-product. 9 On rare occasions, it has been possible to isolate type I (e.g. Nu = (Me 2 N) 3 PNH, 8a,10 pyridine, 8b N-heterocyclic carbene 7 ) intermediates, thereby backing the proposed mechanism experimentally. In all these intermediates, the BH unit bearing the nucleophile(s) remains attached to the original vertex of the carborane cage.
Recently, carboranylamidinates, a novel type of o-carboranebased chelating ligands, have been discovered in our lab. 11 Carboranylamidinate anions combine the versatile characteristics of carboranes and the widely employed amidinate anions 12 into one system. The general synthetic route to lithium carboranylamidinates involves in situ-preparation of monolithio-o-carborane, followed by treatment with a N,N 0 -diorganocarbodiimide, R-NQCQN-R (R = i Pr, cyclohexyl (Cy)) (Scheme 2). Quite unique and typical for the new carboranylamidinates is the C,N-chelating coordination mode while retaining an NH-functionality. 11 Following our initial report, the chemistry of carboranylamidinates has been rapidly extended to various main group element derivatives as well as early and late transition metal complexes, some of them showing promising catalytic activities. 13 It was also demonstrated that the carboranylamidinate system can be modified by introducing thiolate or selenolate functional groups or by designing novel nido-carboranylamidinate ligands. 14 Thus far, the synthetic protocol leading to carboranylamidinates Here we report about the highly surprising and unexpected outcome of initial experiments in that direction. m-Carborane (3) was mono-metalated in the usual manner 11 using 1 equiv. of n-butyllithium followed by treatment of the intermediate lithio-m-carborane with 1 equiv. of N,N 0 -diisopropylcarbodiimide. Crystallization from either THF or diethyl ether afforded large, colorless single crystals in estimated yields around 30-40%. However, under these conditions, fairly large amounts of unreacted m-carborane could be detected in the reaction mixtures. Initially, analytical and spectroscopic data did not allow unambiguous elucidation of the isolated products. Fortunately, in both cases well-formed single crystals were readily available. X-ray diffraction studies revealed the presence of the unexpected products 4a and 4b as illustrated in Scheme 3. Accordingly, the outcome of these reactions can be described as follows: Unlike the formation of carboranylamidinates from o-carborane (1) according to Scheme 2, in situ-prepared monolithio-m-carborane reacts with 2 equiv. of N,N 0 -diisopropylcarbodiimide. This accounts for the presence of unreacted m-carborane in the original 1 : 1 reaction mixture. Instead of a simple addition of the carborane anion to the central carbodiimide carbon alone, subsequent deboronation occurs and a BH unit is detached from the carborane cage. The following step is unprecedented in the chemistry of icosahedral carboranes, in that the BH unit is not eliminated as a monoborane by-product but retained in a newly formed sevenmembered diazadiborepine ring system which is assembled in the periphery of the resulting nido-carborane cage through coupling of two carbodiimides. The result is the formation of an unusual nido-carborane-anellated diazadiborepine ring system which is isolated in the form of its lithium salt solvated with either THF (4a) or diethylether (4b) after recrystallization from the respective solvent (Scheme 3). Fig. 1 displays the molecular structure of the THF-adduct 4a. The molecular structure of 4b differs only in the coordinated solvent at Li (see ESI ‡ for full structural details of both 4a and 4b). The newly formed seven-membered diazadiborepine ring (C1-C3-N2-B11-C10-B6) is anellated to a B-C bond of the nido-carborane cage. With C3-N1 1.2891(19) Å and C10-N4 1.2773(19) Å the two exocyclic CQN bonds clearly have double bond character, whereas the bonds within the ring are single bonds (C3-N2 1.4106(19), C10-N3 1.427(2) Å). Lithium interacts with B5 (2.710(3) Å) and N1 (1.999(3) Å), thereby forming a five-membered C 2 BNLi chelate ring. Two THF ligands complete the distorted tetrahedral coordination geometry around Li. After the crystal structures of 4a and 4b had been unequivocally established, it was shown that by using the proper 1 : 2 stoichiometry, the reaction reproducibly provided 4a and 4b in moderate (4b: 51%) to good (4a: 67%) isolated yields. Moreover, the compounds were thoroughly characterized by spectroscopic and analytical methods (see ESI ‡).
In a subsequent experiment it was shown that the analogous reaction with N,N 0 -dicyclohexylcarbodiimide proceeds in the same manner (Scheme 4). In this case, X-ray quality single crystals could be obtained from DME (=1,2-dimethoxyethane). Full spectroscopic characterization of the DME adduct 5 already indicated a similar outcome of this reaction as with i Pr-NQCQN-i Pr.
Scheme 3 Formation of the polycyclic diazadiborepines 4a and 4b from m-carborane.

View Article Online
Once again, deboronation and carbodiimide coupling under formation of a diazadiborepine had occurred, underlining the fact that this novel reaction is quite general in nature. The NMR spectra of 5 ( 1 H, 7 Li, 11 B, 13 C) showed far-reaching similarity with those of 4a and 4b (apart from the different substituents). For example, in the 1 H NMR spectrum of 5, the BH unit in the seven-membered ring gives rise to a broad singlet at d 4.03 ppm. However, a single-crystal X-ray diffraction study of 5 revealed interesting structural differences ( Fig. 2 and ESI ‡). One of these differences is that Li interacts with two cage boron atoms (Li-B5 2.663(3), Li-B6 2.637(3) Å). Moreover, the electron distribution within one of the amidinate moieties (N3-C4-N4) differs significantly from that in 4a and 4b. Whereas the diazadiborepine ring in 4a and 4b comprises two exocyclic CQN double bonds (vide supra), there is a bond between N4 and B10 in 5 which leads to a seven-membered open face of the nido-carborane cluster. As a consequence, bond delocalization in the amidinate unit is observed which results in equilibrated C-N distances (C4-N3 1.3400(16) and C4-N4 1.3178(16) Å). Whether or not these structural differences are a consequence of the different steric demand of the substituents (isopropyl vs. cyclohexyl) remains speculative at this stage.
Based on the established knowledge about carboranylamidinate formation 11 and deboronation reactions 4 it appears reasonable to propose the mechanism illustrated in Scheme 5 for the formation of the novel diazadiborepines 4 and 5. Monolithiation of m-carborane (3) with 1 equiv. of n-butyllithium affords intermediate A which will add to the carbodiimide carbon to afford carboranylamidinate B. B-N bond formation could then activate the BH unit in the hydridoborate intermediate C for deboronation and diazadiborepine formation initialized by the second equivalent of carbodiimide.
In summarizing the results reported here, we have discovered an unprecedented deboronation reaction of icosahedral carboranes in which a BH group is detached from the cage and incorporated into a nido-carborane-anellated diazadiborepine ring. The reactions illustrated in Schemes 3 and 4 clearly open up a new field of carborane chemistry. Future work will show if and how the readily accessible lithium salts 4 and 5 can be employed as starting materials for novel carborane-based polycycles incorporating other main group or transition metals.
Financial support by the Otto-von-Guericke-Universität Magdeburg is gratefully acknowledged. We also thank Professor Edgar Haak for valuable discussions.