Ambient-Temperature Isolation of a Compound with a Boron-Boron Triple Bond

Stable compounds with homoatomic triple bonds, although well known for carbon and nitrogen, are particularly rare for the heavier main-group elements (groups 13 to 15) and are unknown for boron (1–3). The difficulties involved in synthesizing molecules with homoatomic triple bonds of the heavier main-group elements are well documented: Low-valent main-group centers are typically highly reactive in the absence of extreme steric protection by inert groups. In cases where these daunting synthetic difficulties have been overcome, such compounds of the heavier main-group elements are often severely trans-bent, reducing their triple-bond character (1).

For the first-row main-group element boron, which resists multiple bonding in general and forms triple bonds to other elements only very rarely (4–7), the synthetic challenge of preparing a stable B-B triple bond has thus far been prohibitive. In 2002 the molecule OCBBCO, ostensibly a B₂ molecule stabilized by two Lewis bases, was isolated in an argon matrix at 8 K by the reaction of laser-vaporized boron atoms with CO (8). This finding prompted a flurry of theoretical studies of molecules with B-B triple bonds from the groups of Mavridis, Jones, Frenking, and Michalak (9–13), considering L→E≡E←L (E = B, Al, Ga, In) systems in which the donors L were diatomic molecules (CO, CS, N₂, BO⁻), noble gases (Ar, Kr), phosphines (PCl₃, PMe₃), N-heterocyclic carbenes (NHCs), or tetracoordinate germylenes. In marked contrast to the heavier Al, Ga, and In analogs, NHC-stabilized L→B≡B←L compounds were predicted to be linear, to have very short B-B distances, and to possess, in effect, true boron-boron triple bonds.

Successful attempts at creating stable molecules with homoatomic multiple bonding between group 13 elements have been based on populating the empty π-bonding orbital between the atoms. Most often this has been accomplished by reduction, whereby the groups of Berndt and Power accessed mono- and dianionic diboranes(4) [e.g., (R₂BBR₂)^•−, 1; (R₂BBR₂)²⁻, 2; Fig. 1] with one- and two-electron π bonds (14–19). Similarly, Power and co-workers reported dialuminyne and digallyne dianions of the form [REER]²⁻ (E = Al, Ga), which, although trans-bent, show some triple-bond character (20–22). A second strategy, used by Robinson’s group, involved Lewis base stabilization of the boron atoms by NHC donors, allowing boron’s own valence electrons to fill the π-bonding orbital of an “HBBH” unit. Thereby, they were able to isolate the bis(carbene)-stabilized diborene 3 (Fig. 1) by reduction of an NHC-BBr₃ adduct (23). Were it not for the opportunistic abstraction of hydrogen, presumably from the reaction solvent, this reaction might have resulted in a triply bound species of the form L→B≡B←L. The stability and triple-bond character of such a compound was subsequently predicted computationally by the groups of Mitoraj, Jones, and Frenking (11, 12).

• Download high-res image
• Open in new tab
• Download Powerpoint

The hydrogen abstraction implicit in Robinson’s diborene synthesis led us to assume that during the reduction a boron-based radical intermediate was being formed, perhaps as the result of a slow boron-boron coupling step. If so, we reasoned that the synthesis of a triply bound L→B≡B←L compound could be promoted by the reduction of a molecule with a preformed B-B bond. After preparing tetrabromodiborane(4) (24, 25), simple addition of two equivalents of the free NHC 1,3-bis-(2,6-diisopropylphenyl)imidazol-2-ylidene (IDip) furnished the bis(carbene)-stabilized diborane(4), 4 (Fig. 1) (26). By treating 4 with either two or four equivalents of a tetrahydrofuran solution of the one-electron reducing agent sodium naphthalenide [Na(C₁₀H₈)], we were able to controllably generate and isolate the bis(NHC)-stabilized dibromodiborene 5 (48% yield) and bis(NHC)-stabilized diboryne 6 (57% yield) as green solids after removal of by-product naphthalene under high vacuum and extraction of the neutral compounds into pentane.

The ¹¹B nuclear magnetic resonance (NMR) signals of 4 (δ = −4.8 ppm), 5 (δ = 20 ppm), and 6 (δ = 39 ppm) show a clear progression toward low field, as expected for the stepwise decrease in the coordination number at boron. The ¹¹B NMR chemical shift of 6 matches those calculated using the gauge-independent atomic orbital density functional theory [GIAO-DFT (27–31)] method at the B3LYP/6-311G* level and referenced to BF₃ (δ = 36.54 and 36.01 ppm). The identity and purity of complexes 4 to 6 were confirmed by ¹H and ¹³C{¹H} NMR spectra and elemental analyses. Further support for the connectivity of compounds 5 and 6 was provided by a comproportionation reaction, in which mixing of an equimolar amount of tetrabromodiborane(4) 4 and diboryne 6 resulted in full conversion (monitored by ¹¹B NMR) to the dibromodiborene 5. Compound 6 exhibits impressive thermal stability, decomposing only at 234°C.

Molecular structures of compounds 5 and 6 derived from single-crystal x-ray crystallography are shown in Fig. 1. The diborene 5 has a planar structure similar to that of dihydrodiborene 1 (Fig. 1, top; L = IDip) prepared by Robinson, and their B-B distances are also comparable (5, 1.546 ± 0.006 Å; 1, 1.561 ± 0.018 Å) (23). These distances compare well with calculated values from the groups of Mitoraj (1.57 to 1.60 Å) and Jones and Frenking (1.618 and 1.594 Å) for diborene systems (11, 12). The B-B distances of neutral, base-stabilized diborenes are slightly shorter than the corresponding distances of dianionic diboranes(4) containing B-B double bonds (1.623 to 1.706 Å). The structure of 6 shows an effectively linear C→B≡B←C core, as predicted by theory (11, 12), with a B-B distance of 1.449 ± 0.003 Å and very slight bending at the boron atoms (173.0 ± 0.2°, 173.3 ± 0.2°). The B-B distance of 6 fits very well with the calculated values for the highly comparable bis(carbene)- and bis(phosphine)-stabilized diborynes [1.45 to 1.46 Å in (11); 1.470 Å in (12)] and surprisingly well with both measured (1.453 to 1.468 Å) (8) and calculated (1.439 Å) (9) values of OCBBCO, as well as calculated values of SCBBCS (1.482 Å) (9), N₂BBN₂ (1.460 Å) (9), and [OBBBBO]^2– (1.481 to 1.504 Å) (10, 13). The difference in the B-B bond distances of the doubly (5) and triply bonded (6) species is statistically significant (0.097 Å) and corresponds to a contraction of ~6%.

A solid-state infrared spectrum of diboryne 6 showed a number of strong bands at 1467, 1386, 1363, and 1328 cm^–1. Although we expect these bands to be somewhat associated with the B₂ unit of 3, we calculated the true B≡B stretch of the molecule to be at 1339 cm^–1. This mode is highly symmetrical and is calculated to be infrared-inactive (see figs. S4 and S5). The color of both 5 and 6 can be explained by low absorption in the green region of their ultraviolet-visible (UV-vis) absorption spectra (5, ~530 nm; 6, 510 nm). Diborene 5 has absorption bands centered at 380 nm (extinction coefficient: 134,965 L mol⁻¹ cm⁻¹) and 695 nm (317,819 L mol⁻¹ cm⁻¹) (fig. S1), whereas diboryne 6 has a weak band at 335 nm (9852 L mol⁻¹ cm⁻¹) and intense bands at 385 nm (26,426 L mol⁻¹ cm⁻¹, with relatively strong shoulder peak) and 600 nm (17,685 L mol⁻¹ cm⁻¹) (Fig. 2).

• Download high-res image
• Open in new tab
• Download Powerpoint

The unusual color of diboryne 6 prompted us to turn to computation in order to find an explanation. Our computational results on the geometry and structure of the electronic ground state of diboryne 6 show no notable differences from those determined by Jones and Frenking (12) for a comparable analog; thus, our discussion focuses only on computational studies related to the spectral properties of 6. We sought to identify the states that are responsible for the observed absorption bands in the UV-vis spectrum. Dipole-allowed electronic excitation calculations were carried out on gas-phase optimized structures within the adiabatic local density approximation (ALDA) of time-dependent DFT (TD-DFT) at the B3LYP/6-311G* level.

Figure 2 shows the stick plot of the calculated oscillator strengths and the simulated spectra. The overall absorption profile (shape) of the bands is well reproduced; however, there are some small discrepancies. The narrow nonabsorbing domain centered at 510 nm in the experimental spectrum (which gives the compound its green color) is shifted some 40 nm to the red on an absolute energy scale; moreover, the relative intensity of the higher-energy band is somewhat lower than what is observed experimentally. DFT is well known to underestimate the highest occupied molecular orbital (HOMO)–lowest unoccupied molecular orbital (LUMO) gap of molecules, and thus tends to provide excitations that are too low in energy. Nonetheless, the calculations predict a green-yellow coloration for diboryne 6. The absorption band observed experimentally at 380 nm is not fully symmetrical and probably incorporates a side peak, making it appear steeply cambered in the 410- to 510-nm region. This effect is fully explained by calculations: The stronger band (494.2 nm) is a result of an excitation from the HOMO to the LUMO+5 and the LUMO+7, whereas the neighboring shoulder (527.9 nm) is a transition from the HOMO to the LUMO+1. All three of these transitions involve electronic promotion into orbitals centered on the aryl substituents. The calculated spectra also predict a very intense band at 599.2 nm. The high intensity of this signal is not surprising in light of the seemingly high overlap between the occupied B≡B-centered π system and the empty B-C π-like orbitals of 6. This signal is attributed to the HOMO→LUMO+9 and HOMO-1→LUMO transitions. The three major absorption bands in the UV-vis region all involve excitation of electrons from one or the other of the π bonds of the B≡B unit.

Compound 3 is stable at room temperature in the absence of air and water. The chromophore properties and robustness of 5 and 6 offer exciting possibilities for the further study of the reactivity and optical properties of B-B double and triple bonds, in line with the recent interest in boron-based functional materials (32, 33).