Single and double activation of acetone by isolobal B ^ N and B ^ B triple bonds †

B ^ N and B ^ B triple bonds induce C – H activation of acetone to yield a (2-propenyloxy)aminoborane and an unsymmetrical 1-(2-propenyloxy)-2-hydrodiborene, respectively. DFT calculations showed that, despite their stark electronic di ﬀ erences, both the B ^ N and B ^ B triple bonds activate acetone via a similar coordination-deprotonation mechanism. In contrast, the reaction of acetone with a cAAC-supported diboracumulene yielded a unique 1,2,3-oxadiborole, which according to DFT calculations also proceeds via an unsymmetrical diborene, followed by intramolecular hydride migration and a second C – H activation of the enolate ligand. (H1, B2, B1, C2) 105.7(9) (cid:2) ) and shows a bond length of 1.721(2) ˚ A typical of a diborane ( 5 ). The alkenylborane moiety around B1 is supported by a neutral cAAC ligand with a relatively short B1 – C4 bond (1.5295(19) ˚ A) and forms an angle of only ca. 19 (cid:2) with the plane of the B 2 C 2 O heterocycle (torsion (N1, C4, B1, C2) 14.4(2) (cid:2) ), which is indica-tive of p conjugation. The enoxyborane moiety around B2 bears a protonated cAAC ligand displaying clear sp 3 -hybridisation at C24 (B2 – C2 4 1.6045(18), C24 – N2 1.4878(16) ˚ A). The structure of 3a is reminiscent of the products obtained from the reduction of (SIMes)BBr 2 BAr 2 diborane ( 5 ) precursors (SIMes ¼ 1,3-Mes 2 -4,5-dihydroimidazol-2-ylidene, Mes ¼ 2,4,6-trimethylphenyl; Ar ¼ Mes, 9-anthryl). These display a central, m 2 -hydride-bridged, planar B 2 C 5 heterocycle, resulting from the C – H activation of one aryl substituent by an intermediate boraborylene, and coordinated on one side by a neutral SIMes ligand and on the other by the second aryl substituent. 18

Heating a suspension of IV in hexanes with excess acetone overnight at 70 C resulted in clean formation of the (2-propenyloxy)aminoborane 1 (Scheme 1A). 11 B NMR data of 1 showed a resonance at 24.8 ppm, while the 1 H NMR spectrum displayed a NH singlet at 3.49 ppm and two characteristic 1H resonances for the terminal methylidene protons of the enolate ligand at 4.36 and 4.11 ppm.
Whereas diboryne I proved unreactive towards acetone even under forcing conditions, diboryne II reacted rapidly with excess acetone in benzene at room temperature to yield the green-coloured 1,2-enol addition product 2 (Scheme 1). Compound 2 presents two broad 11 B NMR resonances at 38.1 and 19.3 ppm in a 1 : 1 ratio, attributable to the BH and the BOC 3 H 5 moieties of the unsymmetrical diborene, respectively. The 1 H NMR spectrum displayed two inequivalent SIDep ligands, as well as the inequivalent terminal methylene protons of the 2-propenyloxide ligand at 3.93 and 3.47 ppm. X-Ray crystallographic analysis of 2 showed a trans-1-alkoxy-2hydrodiborene with a B-B double bond of 1.599(4)Å similar to that of its dihydrodiborene relative, (SIDep)HB]BH(SIDep) (1.589(4)Å). 10 The SIDep ligand at the BH moiety is near coplanar with the diborene core (torsion (N4, C27, B2, B1) 12.1(5) ) and displays a short B2-C27 bond (1.523(4)Å), indicative of p backdonation. In contrast, the SIDep ligand supporting the BOC 3 H 5 moiety is twisted ca. 35.5 out of the diborene plane and displays a pure s-donor interaction (B1-C4 1.574(4)Å). The planar 2-propenyloxide ligand lies at a ca. 58 angle with respect to the diborene plane, and its bond lengths (O1-C1 1.352(3), C1]C2 1.316(4)Å) are similar to those of 1. With Kinjo and co-workers recently reporting the rst diborene with two different donor ligands 16 and our group having just published the rst fully unsymmetrical diborene, 17 compound 2 is only the second unsymmetrical diborene with respect to the anionic substituents.
TDDFT calculations performed upon the optimised geometry of 2 at the (smd: n-pentane)lc-uPBE/6-311+g(d) level of theory provided a maximum UV-vis absorbance at 592 nm (see Table S1 and Fig. S24 in the ESI †), which is in good agreement  with the experimentally measured absorbance maximum in pentane at 605 nm (Fig. S16 †). This corresponds to the HOMO-LUMO transition from the p-bonding orbital of the B]B double bond into the empty p z orbital of the carbene carbon of the SIDep ligand supporting the BOC 3 H 5 moiety, and is responsible for the blue-green color of the compound. Surprisingly, the 1 : 1 reaction of diboracumulene III with acetone did not yield the expected cAAC analogue of 2. Instead, 11 B NMR data revealed a 92 : 8 mixture of two sp 2 -sp 3 diborane products, the major one (3a) showing two broad singlets at 42.8 (full width at half maximum: fwhm z 370 Hz) and À1.9 ppm (fwhm z 130 Hz), and the minor (3b) presenting a very broad resonance at 63.0 ppm (fwhm z 630 Hz) and a broad BH doublet at À15.0 ppm ( 1 J B-H ¼ 50.8 Hz), suggesting a nonbridging hydride. The 1 H NMR spectrum of the mixture showed very similar sets of resonances for 3a and 3b, which strongly suggests an isomeric relationship. Both compounds display one neutral cAAC ligand and one C1-protonated cAAC ligand (d ¼ 3a 4.02, 3b 4.24 ppm) as well as a single 1H alkene resonance (d ¼ 3a 3.50, 3b 3.78 ppm) (Scheme 2).
Single-crystal X-ray crystallography revealed a unique planar 2,3-dihydro-5-methyl-1,2,3-oxadiborole heterocycle displaying an endocyclic C1]C2 double bond (1.3301 (18) Although single crystals of the minor species in solution were never obtained, the propensity for cAAC-supported hydroboranes to undergo 1,2-hydrogen shis from boron to an adjacent cAAC carbene centre, which has been demonstrated both experimentally and computationally, 19 rst prompted us to identify the second isomer as compound 3 taut , a tautomeric form of 3a, in which the neutral cAAC ligand coordinates to the enoxyborane moiety, and the protonated cAAC ligand coordinates to the alkenylborane moiety (Fig. 3). DFT optimisations at the ONIOM(M06-2X/6-311+G(d):PM6) level (see ESI † for details) showed, however, that 3 taut is 8.4 kcal mol À1 higher in energy than 3a and that its calculated 11 B NMR shis (d ¼ 45.1, 6.1 ppm) do not t the experimental data (d ¼ 63.0, À15.0 ppm). Since 3a presents two stereocentres, one at B2, which is locked by the B 2 C 2 O ring and the asymmetrically bridging hydride, and one at the protonated cAAC carbon atom, the other possibility is that 3a and 3b could be diastereomers. This would also t the observation that they do not exchange in solution even at high temperatures. To test this, the geometries and 11 B NMR chemical shis of the possible diastereomeric pairs derived from 3a were computed (Fig. 3).
The predicted 11 B NMR chemical shis for the (R C ,R B )/ (S C ,S B )-3a pair (d calc ¼ 44.0, À4.4 ppm) adequately match the experimentally-observed shis (d exp ¼ 42.8, À1.9 ppm; D(d) z AE2 ppm). Calculations on the diastereomeric pair showed that a form with a non-bridging hydride is the most likely. This also correlates well with the observation that, unlike 3a, which shows two very broad 11 B NMR resonances typical for a m 2 -hydride-bridged diborane, 3b shows a doublet at À15.0 ppm ( 1 J 11B-1H ¼ 50.8 Hz), indicating a terminal hydride rather than a bridging one. The predicted 11 B NMR chemical shis for the (R C ,S B )/(S C ,R B )-3b pair (d calc ¼ 65.4, À18.9 ppm) are comparable to the experimental ones (d exp ¼ 63.0, À15.0 ppm; D(d) z AE3 ppm). The relative energy of (R C ,S B )/(S C ,R B )-3b, at 3.1 kcal mol À1 above (R C ,R B )/(S C ,S B )-3a, is consistent with the experimentally observed ratio of 92 : 8.
The spontaneous formation of 3a/b is particularly remarkable in view of the fact that there is seemingly no literature precedent for a one-step, uncatalysed, 100% atom-efficient double C-H activation of acetone or other enolisable ketones. We were therefore keen to investigate the mechanism of the formation of 3a/b and compare it to that of the boron enolates 1 and 2. While the reaction of dimeric iminoboranes with Scheme 2 Double activation of acetone by diboracumulene III. enolisable ketones always yielded the 1,4-enol addition products, Paetzold and co-workers showed that with acetophenone, which is less prone to enolisation, a [2 + 4] cycloaddition product can also be isolated. 15 However, it remained unclear whether or not the latter is an intermediate to the former. For comparison, nonpolar disilenes are known to rst undergo [2 + 2] cycloaddition with acetone and acetophenone to form the corresponding 1,2,3-oxadisiletane heterocycles, which then rearrange to the 1,2-enol addition products. 20 In our case, however, careful monitoring of the reaction of iminoborane IV and diboryne II with acetone showed no evidence of [2 + 2] cycloaddition products or intermediates.
DFT calculations carried out at the D3-PBE0/6-31G(d) level for IV and at the ONIOM(M06-2X/6-311+G(d):PM6) level for II and III showed that acetone activation does not proceed via 1,2enol addition, as the enol form of acetone lies 15.3 kcal mol À1 higher than the reactants, well above the activation energy for direct acetone addition (Fig. 4, see ESI † for details on the methodology and the optimised structures of all reactants, products, intermediates and transition states).
For iminoborane IV two plausible mechanisms were investigated, the rst via a 4,4-dimethyl-1,3,2-oxaboretidine [2 + 2] cycloaddition product (A), the second via concerted acetone coordination-deprotonation (Fig. 4). Although the cycloaddition product A is calculated to be more stable than 1 by 2.8 kcal mol À1 , the energy barrier for the formation of A is slightly higher than for 1. § Furthermore, as there is no thermodynamically viable reaction path from A to 1, a [2 + 2] cycloaddition mechanism followed by rearrangement to 1 can be ruled out.
Instead, for compounds II-IV the rst reaction step involves coordination of the carbonyl oxygen atom to one boron centre to form the acetone adducts I1 1 , I1 2 and I1 3 (DG ‡ 1 ¼ 7.6 (IV), 20.6 (II), 10.1 (III) kcal mol À1 ), respectively ( Fig. 4 and 5). This step is followed in all three cases by C-H activation of one of the pendant methyl groups of the coordinated acetone by either the nitrogen atom (for IV) or the electron-rich, second boron centre (for II and III), to yield the cis-aminoborane I2 1 , and the SIDepand cAAC-supported cis-diborenes I2 2 and I2 3 , respectively (DG ‡ 2 ¼ 4.0 (IV), 14.1 (II), 14.9 (III) kcal mol À1 ). Finally, the transaminoborane 1 and the trans-diborenes 2 and I4 3 are obtained by rotation around the B-N and B-B bond, respectively. Overall, the formation of 2 presents the highest energy barrier and is also the most exergonic (DG ¼ À43.2 kcal mol À1 ), followed by that of 1 (DG ¼ À26.6 kcal mol À1 ) and I4 3 (DG ¼ À27.4 kcal mol À1 ).
The exergonic isomerisation step leading from the cisdiborenes I2 2 and I2 3 to the trans-diborenes 2 and I4 3 , respectively, was further investigated to determine the rotation barrier in each case. Interestingly, DFT calculations showed two distinct mechanisms at work for the SIDep-stabilised and the cAAC-stabilised diborene, respectively (Fig. 6). For the SIDep analogue I2 2 , rotation about the B-B bond is facilitated by shiing the p-electron density of the B]B double bond into the p backbonding to the unsaturated carbene ligands. The resulting transition state TS3 2 now displays a B-B single bond, which allows facile rotation. The isomerisation process from I2 2 to 2 occurs with a low barrier of 9.7 kcal mol À1 . In contrast, the lowest energy pathway for the cAAC analogue I2 3 proceeds via a 1,2-hydride shi from boron to the adjacent cAAC carbene carbon to yield the intermediate diborene I3 3 (DG ‡ 3 ¼ 8.9 kcal mol À1 ), in which the boron bearing the now protonated cAAC ligand is sp-hybridised. Rotation about this B-C cAACH single bond and a second 1,2-hydride shi back to the boron centre then yield the trans-diborene I4 3 with a low barrier of 9.3 kcal mol À1 . This pathway is assisted on the one hand by the facile 1,2-hydride shuttling chemistry displayed by cAAC hydroboron compounds 19 and on the other hand by the very strong p acceptor properties of cAAC, 12 which enable the stabilisation of the coordinatively saturated intermediate I3 3 .
For cumulene III, however, the reaction does not stop at trans-diborene I4 3 (Fig. 3). The latter undergoes hydride migration from B1 to B2 to form the (alkoxy)hydroboryl-(alkylidene)borane I4 3 (DG ‡ 3 ¼ 19.8 kcal mol À1 ). Coordination of the pendant terminal alkene to the two-coordinate boron yields adduct I5 3 , which is 6.6 kcal mol À1 more stable. Subsequent C-H activation of the methylidene moiety yields the bis(cAAC)stabilised 1,2,3-oxadiborole I6 3 (DG ‡ 5 ¼ 21.2 kcal mol À1 ). This is the highest energy barrier in the entire reaction mechanism. I6 3 then tautomerises to compound 3a by concomitant migration of the hydride on B1 to the adjacent cAAC carbene centre and bridging of the hydride on B2 (DG ‡ 6 ¼ 11.7 kcal mol À1 ). Overall the formation of 3a from III and acetone is exergonic by 61.7 kcal mol À1 , which explains why the intermediate diborene cannot be isolated.
To conclude, we have shown that three linear, isolobal, multiply bonded boron compounds, iminoborane IV, diboryne II and cumulene III, all activate acetone via a similar acetone coordination-deprotonation mechanism, regardless of their polar or nonpolar nature. For the iminoborane-based reaction, an enol addition mechanism and a mechanism proceeding via a [2 + 2] cycloaddition intermediate, as would normally be Fig. 4 Mechanisms of acetone addition to iminoborane IV to yield aminoborane 1 (straight lines in black) and alternative [2 + 2] cycloaddition to yield A (dashed lines in blue), as well as energy level of the enol form of acetone (green) calculated at the D3-PBE0/6-31G(d) level of theory. Gibbs free energies (kcal mol À1 ) in brackets. expected for such a polar compound, were both ruled out. For diboron compounds II and III the addition of acetone rst yields a cis-diborene intermediate which isomerises to the thermodynamic trans-diborene product through a low energy barrier. Calculations showed that this isomerisation process heavily relies on the p-accepting nature of the carbene ligands, coupled, in the case of the cAAC-supported diborene, with a hydride shuttling mechanism from boron to the carbene carbon and back. These cAAC-specic properties also enable an unprecedented second C-H activation of the enolate ligand to yield a novel 1,2,3-oxadiborole heterocycle, demonstrating once again the unique reactivity of cAAC-supported low-valent boron compounds.
Overall this study should act as a reminder that the parallels all too eagerly drawn between organic compounds and their isoelectronic/isolobal inorganic p-block counterparts only rarely translate into actual organomimetic behaviour when it comes to reactivity or reaction mechanisms. Furthermore, this rst example of reactivity overlap between polar and nonpolar boron-based triple bonds opens up new avenues for attempting reactions that may have been previously disregarded, such as the addition of nonpolar small molecules to iminoboranes or, alternatively, of polar molecules to diborynes.

Conflicts of interest
There are no conicts to declare.