Isolation and Characterization of an Organobismuth Dihydride

Abstract

We report the synthesis, isolation, and structural characterization of an elusive organobismuth­(III) dihydride (Ar*–Bi­(III)–H₂, 1). The complex features a bulky and rigid ^t Bu-M^sFluind ligand that permits complete spectroscopic characterization and SC-XRD. The compound is thermally unstable and decomposes to quantitatively form H₂ and Ar*–Bi­(I) in a chemoselective intramolecular process. In addition to H₂ formation, the presence of a Bi–H bond is supported by comparative spectroscopy (NMR and IR) with its deuterated analogue 1-d ₂.

Recently, the field of bismuth redox catalysis has emerged as an orthogonal alternative to light main group and transition metal catalysis, − thus offering a plethora of redox manifolds that can be harnessed in various contexts of synthesis. An in-depth analysis of the reported catalytic protocols to date reveals that a large portion invokes the presence of organobismuth hydrides as putative intermediates; for example, in transfer hydrogenations, , reduction of nitrous oxide or azides, amide reductions, hydrodefluorinations, dehydrogenative O–Si coupling or HER (hydrogen evolution reactions), among others. Due to the high reactivity of the Bi–H bond, these intermediates are highly fleeting and have posed severe difficulties when attempting their isolation and characterization. Indeed, almost all evidence gathered so far has been largely restricted to their identification via HRMS or in solution, using NMR at low-temperature (Figure A). , In addition to the relevance in catalytic strategies, bismuth hydrides have also spurred the minds of synthetic organometallic chemists for many years due to the intriguing Bi–H bond, − as it represents a hydride species originating from the bond between two elements with dramatic differences in radius. − Bi–H species are prone to rapid decomposition, generating H₂ and generally unknown Bi byproducts. For example, BiH₃ has been synthesized and its structure studied in silico and gas phase, before decomposing to Bi metal and H₂. On the other hand, organobismuth monohydrides have been synthesized and studied in solutionboth diaryl and dialkyl , and used in various organometallic endeavors. Power et al. reported the first and only thermally stable diorganylbismuth hydride, whose structure was confirmed by both SC-XRD (single crystal X-ray diffraction) and IR, and later NMR. The use of bulky Ar ligands provided the necessary steric congestion to stabilize the Bi–H bond. It was shown that upon heating, this mono hydride undergoes formation of dibismuthene (Ar–BiBi–Ar) ,, and H₂. This and other bismuth hydrides have been shown to undergo hydrobismuthation to alkenes, alkynes and CN bonds. Albeit the formation of H₂ from Bi­(III) halides is a common strategy to access bismuthinidenes, not much is known about the putative organobismuth dihydrides. Since the main decomposition pathway is the formation of H₂ and a low-valent “RBi”, it is not surprising that kinetic stabilization of two hydrides attached to Bi­(III), inasmuch as RBiH₂, would pose severe difficulties. Whereas MeBiH₂ has indeed been prepared and used in several inorganic recipes, the structure of organobismuth dihydrides still remains elusive. In this communication, we report the synthesis and characterization of the unique organobismuth­(III) dihydride (1), bearing a bulky and rigid ^t Bu-M^sFluind backbone (Ar*, Figure B). The compound is stable below −20 °C, and was characterized by NMR, IR and SC-XRD. This molecule extrudes H₂ above 0 °C and permits the study of the reductive elimination of dihydrogen from Bi­(III) to Bi­(I). Differently than in Power’s monohydride, kinetic isotope effect (KIE) and intermolecular experiments with the corresponding deuterated analogue suggest this process to be intramolecular and H₂-selective, thus avoiding reductive demetalation of the C–Bi bond.

1.

Bismuth hydrides: state-of-the-art.

For some years, our group has been interested in the chemistry and reactivity of low-valent Bi­(I) compounds, as a means to unlock catalytic redox processes spanning from small molecule activation to organic synthesis. ,− ,− ,,,− Within this context, we reported the synthesis of the first stable monocoordinated organobismuthinidene featuring a bulky and rigid ^t Bu-M^sFluind backbone (2). During the investigation of its chemical reactivity, we realized that 2 remained reluctant to H₂ cleavage under 1 atm at 25–60 °C, as judged by the absence of bismuth dihydride formation. This reluctance contrasts with the rapid H₂ cleavage of the analogous Ar*–Sb­(I) or a similarly bulky Ar–N­(I), , which rapidly react with H₂ to form the stable Ar*–SbH₂ and Ar–NH₂ respectively.

The targeted organobismuth dihydride was synthesized as illustrated in Figure A. The arylbismuth dibromide 3 reacted with an excess LiAlH₄ at −40 °C generating ^t Bu-M^sFluind-BiH₂ 1 as an off-white powder. Maintaining the temperature <−40 °C was crucial to prevent decomposition of 1 (vide infra). Compound 1 could also be obtained using a neutral aluminum hydride, DIBAL-H in toluene or THF. From this latter reaction, colorless crystals suitable for SC-XRD analysis were obtained by layering pentane over the toluene solution of 1. The solid state structure of 1 is shown in Figure B and C. In late refinement cycles, traces of positive residual electron density were found at a distance of about 1.8 Å from the central Bi atom. Based on their relative position to the heavy atom, these were assumed to be the positions of the two H atoms. Their coordinates had to be fixed during refinement in order to achieve a convergence. Due to the significant difference in scattering power between Bi and H, the reliability of such assignments is limited. The obtained structure exhibited a monomeric form, similar to our previously reported monosubstituted Bi­(I) 2, but with some distinct structural features. The Bi–C bond (2.265(1) Å) was slightly shorter than that in 1 (2.2783(9) Å), indicating differences in the electronic and steric environments. The presence of Bi–H bonds is evident however in various geometrical observations. For example, the increased distance between the fluorene moiety of the ligandmeasured between the centroids of the flanking six-membered ringsis 6.960 Å and 7.096 Å in 1, while it shortens to 6.777 Å and 6.762 in 2. In addition, structure 1 is isomorphous to the Ar*Sb–H₂, recently reported by our group, which shows a similar distortion of the ligand. The presence of hydrogen atoms was also confirmed by NMR spectroscopy. The ¹H NMR spectrum of 1 showed a broad signal at 8.05 ppm, attributed to the Bi–H₂. The pronounced broadening at 20 °C (w_1/2,293 K = 22.2 Hz) sharpens upon cooling (w_1/2,223 K = 5.5 Hz). This behavior contrasts with the Ar*Sb–H₂ complex, which displays a sharp doublet (⁵ J _HH = 0.4 Hz; w_1/2,298 K = 0.4 Hz) at 1.24 ppm. We therefore attribute the broadening to the large quadrupole moment of the ²⁰⁹Bi nucleus (100%, I = 9/2) and the low local symmetry at the bismuth center. The Bi–H ¹H NMR chemical shift falls between the two reported values for a bismuth hydride: the highly deshielded (2,6-diMesPh)₂Bi–H (19.6 ppm) reported by Power and the highly shielded (TMS₂CH)₂Bi–H from Breunig (3.24 ppm). To further confirm the Bi–H signal, the deuterated analogue 1-d ₂ was synthesized from 3 and LiAlD₄, and the disappearance of the 8.04 ppm signal was observed by ¹H NMR (Figure D, middle). Also, the ²H NMR spectrum of 1-d ₂ confirmed the deuterated nature of the signal (Figure D, bottom). The ¹H NMR signals of the ligand scaffold remained sharp from 298 to 193 K. This is in contrast to 3, which showed a broadening of these signals at low temperatures. This indicates a faster rotation of the hydrides around the Bi–C bond compared to the bulkier halides. All of the ¹H and ¹³C chemical shifts are observed in the diamagnetic region. Compared to 2, the ¹³C_ipso–Bi chemical shift in 1 (125.7 ppm) lies in the normal diamagnetic range, whereas the corresponding carbon in the triplet bismuthinidene 2 resonates at a very uncommon frequency (−203.3 ppm). A similarly pronounced difference is observed for the proton at C4 (7.43 ppm in 1 vs −1.04 ppm in 2). To assess whether the broad Bi–H ₂ signals originate from a Bi­(III)-dihydride rather than a putative singlet Bi­(I)–dihydrogen complex, we determined the T ₁ relaxation time. In the case of dihydrogen complexes, the protons normally display T ₁ values below 100 ms, whereas dihydride complexes have similar relaxation times to other protons in the complex. The T ₁ relaxation time of Bi–H ₂ was 1.32 s at 223 K, thus strongly suggesting that 1 is a dihydride. Further confirmation of the Bi–H bonding was provided by IR spectroscopy. A Bi–H stretching absorption was observed at 1717 cm^–1, which was in good agreement with the reported bismuth monohydride and calculated bismuth trihydride. In contrast, the IR spectrum of 1-d ₂ showed no absorption at this frequency. Instead, an increasing intensity of the absorption at around 1250 cm^–1 was observed. We think that Bi–D stretching absorption would be observed in this region due to the overlap with other signals derived from the ^t Bu-M^sFluind ligand. The calculated Bi–D stretching absorption of 1-d ₂ was 1224.56 and 1225.53 cm^–1, in the range of previously reported values.

2.

(A) Synthesis of organobismuth dihydride 1. Isolated yield; in brackets, ¹H NMR yield using 1,2,4,5-tetramethylbenzene as internal standard. (B) Molecular structure of 1 (front view) with thermal ellipsoids at 50% probability and all hydrogen atoms except Bi bounded hydrogen are omitted for clarity. (C) Molecular structure of 1 (top view) with thermal ellipsoids at 50% probability, and all hydrogen atoms except Bi bounded hydrogen are omitted for clarity. (D) Snippet of the NMR data: (i) ¹H NMR spectrum of 1, (ii) ¹H NMR spectrum of 1-d ₂, (iii) ²H NMR spectrum of 1-d ₂. (E) IR spectra of 1 (blue line) and 1-d ₂ (red line).

Heating a THF solution of 1 led to the formation of the corresponding Bi­(I) 2 through reductive H–H coupling (Figure 3A). H₂ evolution is clearly observed by ¹H NMR. VT-NMR analysis of 1 revealed that the elimination process begins at 0 °C. At least two plausible mechanismsintermolecular or intramolecular couplingare considered for the H–H bond formation. Reported dialkyl bismuth monohydrides decompose through H₂ evolution, and the resulting dialkyl bismuth moieties combine to form dibismuthane. Such homocoupling of a potential dialkyl bismuth radical suggests that an intermolecular mechanism is being operative for dialkyl bismuth hydrides. For the highly sterically encumbered diaryl bismuth hydrides, however, such an intermolecular pathway is not feasible. Instead, an intramolecular C–H coupling occurs, producing Ar–H and dibismuthenethe dimerization product of the potential bismuthinidene intermediate. To determine the mechanism operating in the reductive elimination from 1, H–D scrambling tests were performed. When 1 and 1-d ₂ were mixed in a 1:1 ratio at −50 °C (223 K), no evidence for scrambling of the Bi–H and Bi–D was observed. This mixture was then allowed to convert to Bi­(I) over 1 h at 25 °C in THF. The ¹H NMR spectrum of the mixture confirmed the presence of H₂ and HD albeit in >50:1 ratio (Figure B). This suggests that intermolecular H–D exchange did not occur and points to an intramolecular process. To further investigate the reaction mechanism, the decomposition of 1 and 1-d ₂ was independently monitored by ¹H NMR spectroscopy, and the KIE was measured. The decay of 1 and 1-d ₂ followed pseudo-first-order kinetics, with rate constants k _H = (2.98 ± 0.01) × 10^–1 h^–1 and k _D = (3.79 ± 0.01) × 10^–2 h^–1. A large normal KIE of k _H/k _D = 7.9 ± 0.1 was measured from two independent reactions with 1 and 1-d ₂ (Figure C). Although interpretations of large KIEs have been previously discussed for organometallic complexes, the origin of such a large value is still unclear at present. Quantitative data on the activation parameters could be obtained from an Eyring plot analysis, from which a negative entropy of activation (ΔS ^‡ = −12.4 ± 1.1 cal K^–1 mol^–1) is apparent. An enthalpy barrier of ΔH ^‡ = 18.9 ± 0.3 kcal mol^–1 and an activation energy of ΔG ^‡ = 22.6 ± 0.6 kcal mol^–1 at 25 °C were also obtained. It should be noted that an activation energy of +23.7 kcal mol^–1 was obtained when computing the intramolecular H–H bond formation leading to 1 (Figure S41).

3.

(A) Reductive H–H or D–D coupling was performed from 1 or 1-d ₂. (B) H–D scrambling experiment. (C) KIE profiles (left) and an Eyring plot for the thermal decomposition of 1 (right).

In summary, we successfully synthesized, isolated, and characterized the first well-defined organobismuth­(III) dihydride complex (1). The use of a bulky and rigid ^t Bu-M^sFluind ligand enabled comprehensive spectroscopic analysis and X-ray diffraction studies. The complex exhibits thermal instability, undergoing regioselective decomposition to generate H–H and 2. The presence of a Bi–H bond is supported by a side-by-side NMR and IR study of its deuterated analogue 1-d ₂. The results provided herein represent a significant step forward in the understanding of organobismuth chemistry, and provides key insights for the design of future bismuth-catalyzed redox transformations.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c09023.

• Experimental procedures and analytical data (¹H and ¹³C NMR, HRMS, IR, and theoretical calculations) for new compounds. (PDF)

Financial support for this work was provided by Max-Planck-Gesellschaft, Max-Planck-Institut für Kohlenforschung, the Fonds der Chemischen Industrie (VCI-FCI) and Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy - EXC 2033-390677874-RESOLV. This project has received funding from European Union’s Horizon 2020 research and innovation program under the agreement 850496 (ERC Starting Grant, J.C.). S.K. thanks the Alexander von Humboldt Foundation for a postdoctoral fellowship. Y.P. thanks China Scholarship Council for a Fellowship. N.M. thanks JSPS KAKENHI for a JSPS Research Fellowship for Young Scientists (Grant Number 24KJ1634). Open access funded by Max Planck Society.

The authors declare no competing financial interest.