Bi(I)-Catalyzed Transfer-Hydrogenation with Ammonia-Borane.

A catalytic transfer-hydrogenation utilizing a well-defined Bi(I) complex as catalyst and ammonia-borane as transfer agent has been developed. This transformation represents a unique example of low-valent pnictogen catalysis cycling between oxidation states I and III, and proved useful for the hydrogenation of azoarenes and the partial reduction of nitroarenes. Interestingly, the bismuthinidene catalyst performs well in the presence of low-valent transition-metal sensitive functional groups and presents orthogonal reactivity compared to analogous phosphorus-based catalysis. Mechanistic investigations suggest the intermediacy of an elusive bismuthine species, which is proposed to be responsible for the hydrogenation and the formation of hydrogen.

For over half a century, the use of noble metal catalysis has revolutionized the way chemists assemble molecules via the construction of new bonds.[1] The enormous impact of these metals at industrial level has led to their exploitation, becoming less abundant and hence more expensive. In recent years, an increasing attention has been placed in unlocking the potential of more abundant first-row transition metals thus becoming powerful sustainable alternatives.[2] In parallel, approaches that depart from transition metals have also gained momentum; for example the use of alkali and alkaline,[3] main group,[4] and the use of Frustrated Lewis Pairs,[5] have become promising alternatives to transition metals (TM) thus emulating their behavior and reactivity. However, despite the wealth of literature in these areas, the quest for conferring TM-like catalytic properties to main group elements is still in its infancy.[6] In this regard, the ability of pnictogens to maneuver between distinct oxidation states represents a promising approach. Recently, Radosevich reported a variety of P(III)/P(V) redox platforms, which show catalytic activity toward a variety of transformations (Figure A).[7] This reactivity is based on highly strained P(III) compounds, where the lone pair becomes easily oxidizable and thus more prone to nucleophilic attacks and formal oxidative additions.[8] Such a groundbreaking approach opened the door to the possibility of performing catalytic redox processes beyond the TM block. With the aim of investigating unconventional catalytic redox processes of nontransition metals, we have recently started a program which focuses on the exploitation of the redox abilities of bismuth (Bi) to be applied in organic synthesis. Bi represents the last stable element in the periodic table,[9] with properties at the interface of metalloids and main group.[10] Importantly, Bi has been considered nontoxic and largely more abundant than commonly employed TM such as Pd, Rh or Ir,[11] thus highlighting its potential toward developing truly sustainable catalytic strategies. Yet, the use of bismuth in organic synthesis has been largely dominated by stoichiometric reactions based on Bi(V) or Bi(III) species, and catalytic strategies primarily focused on the soft Lewis-acid properties of Bi(III) salts (Figure B).[12] In addition, methods beyond the classic reactivity of Bi salts have recently attracted increasing attention.[13] In contrast to the wealth of methods using high-valent Bi species,[14] attention to its low-valent counterparts has been scarce. Low-valent Bi(I) compounds are known in the literature,[15] yet seldom monomeric Bi species have been isolated.[16] Generally, the formation of Bi(I) compounds is achieved through a highly unstable Bi(III) dihydride, which rapidly extrudes H2 upon ligand coupling.[16,17] Inspired by these results, herein we present a transfer hydrogenation of azoarenes and nitro compounds with ammonia-borane (AB) catalyzed by a well-defined and stable Bi(I) complex. Preliminary mechanistic investigations point out at a catalytic platform involving extremely reactive Bi(III) hydride intermediates (Figure C).[18] To the best of our knowledge, this is the first example of a catalytic redox cycle in the pnictogens group maneuvering in a Pn(I)/Pn(III) redox cycle.

Figure 1

(A) Catalytic redox-activity of pnictogens. (B) Typical reactivity of Bi(III) in catalysis. (C) Bi(I)/Bi(III) redox catalysis.

Initially, we attempted the transfer hydrogenation using a bismuthinidene complex (1)[16b] as catalyst in THF at 50 °C (Table , entry 1). To our delight, 2a was completely converted to 3a using 1.0 equiv of AB as reducing agent. When the catalyst loading and the reaction time were reduced, lower yields were obtained (entry 2), but the addition of 2.0 equiv of AB using 1 mol % of 1 resulted in good yields of 3a (entry 3). Yields were substantially diminished at lower temperatures (entry 4), but the reaction performed well in halogenated solvents (entry 5). Such reactivity is in stark contrast to the high reactivity of low-valent TM, which react with halogenated compounds leading to catalyst deactivation or decomposition. The use of other polar solvents (entries 6 and 7) was not beneficial, but noticeably, the addition of 1.0 equiv of H2O improved the yield and reduced reaction times (entry 8). Furthermore, the reaction did not proceed in absence of catalyst (entry 9) or AB complex (entry 10).

Table 1

Optimization of the Bi(I)-Catalyzed Transfer Hydrogenation

entry	X	Y	solvent	T (°C)	t (h)	3a, yield (%)a
1	4	1	THF	50	24	99
2	1	1	THF	50	16	57
3	1	2	THF	50	16	86
4	1	2	THF	35	16	53
5	1	2	DCE	50	16	76
6	1	2	1,4-dioxane	50	16	87
7	1	2	acetone	50	16	45
8b	1	1	THF	35	3	99 (99)c
9b	–	1	THF	35	16	traces
10b	1	–	THF	35	16	traces

Yield calculated by 1H NMR using 1,3,5-trimethoxybenzene as internal standard.

With 1.0 equiv of H2O.

Isolated yield.

With these optimal reaction conditions in hand, we explored the influence of different substituents in the azoarene (2). As shown in Table , the protocol boded well with substrates bearing ester groups, without reduction of the carbonyl moiety (3b). Electron-rich azoarenes can also be reduced (3c), as well as substrates containing electron-withdrawing functionalities, such as fluoride (3d) and trifluoromethyl (3e). The presence of a bromide group at the ortho position of the reactive functionality did not inhibit the reactivity (3f). Azoarenes bearing substituents at the meta position also reacted smoothly under the optimized conditions (3g). Interestingly, unsymmetrical azoarenes in a push–pull electronic situation were also tolerated (3h). It is noteworthy that the presence of the iodide group did not affect the reaction outcome, further indicating the stability of 1 toward oxidative additions to labile bonds. Although cyclic compounds can also be completely reduced (3i), aromatic azoarenes has proven.[19]

Table 2

Scope of the Transfer Hydrogenation of Azoarenesa

Isolated yields.

Yield calculated by 1H NMR using 1,3,5-trimethoxybenzene as internal standard.

At this point, we decided to expand the protocol to other unsaturated functionalities such as nitroarenes (4, Table ). Contrarily to the majority of TM-catalyzed reductions, this protocol is highly selective toward the formation of N-arylhydroxylamines.[20] With a slight modification from the optimized protocol,[21] a variety of electronically distinct nitroarenes could be reduced in excellent yields. For example, simple nitrobenzene was reduced to N-phenylhydroxylamine (5a) in 89% yield. Electron-rich nitroarenes are also amenable to this reactivity (5b), as well as substrates containing carbon–halogen bonds such as bromide (5c) and iodide (5d). Compounds bearing unsaturated functionalities such as alkyne (5e), nitrile (5f) and alkene (5g) were also obtained in excellent yields. Interestingly, when 2-phenylnitrobenzene (4h) was subjected to the reaction conditions, excellent yields were obtained of the corresponding N-hydroxylamine (5h). This result is complementary to P(III) catalysis, with which intramolecular Cadogan-type reactions en route to carbazole have been observed.[7i] Finally, sterically congested nitroarenes delivered the corresponding N-hydroxylamines (5i) albeit in lower yields.

Table 3

Scope of the Transfer Hydrogenation of Nitroarenesa

Isolated yields.

The unprecedented catalytic activity of such Bi(I) complexes led us to explore the operative mechanism governing this homogeneous transformation (Scheme ).[22] First, the ability of 1 for catalytic dehydrogenation of AB was tested. Thus, 1.0 equiv of AB was mixed with 0.2 equiv of Bi(I) complex in THF-d8 and the formation of H2 in solution was monitored (Scheme A, red). In absence of Bi(I) complex, no H2 was observed after 150 min,[23] indicating that Bi(I) promotes a slow dehydrogenation of AB (Scheme A, blue and yellow). Furthermore, during our optimization studies we noticed a dramatic change in rate when water was added (Table , entry 8). With this result in mind, we tested the effect of H2O in the dehydrogenation of AB catalyzed by Bi(I). Indeed, the addition of 1.0 equiv of H2O caused an increase in rate for the formation of H2 (Scheme A, green). On the basis of Dostál observations,[16] we speculated that the formation of H2 is derived from a highly unstable bismuthine (6).[24] The positive effect of H2O is proposed to arise from H-bonding from AB and water, thus facilitating a plausible oxidation of 1 to 6, which upon rapid H2 extrusion regenerates species 1. Indeed, to further evaluate the effect of H2O in the reaction, a series of experiments with alkylated derivatives of AB were carried out (Scheme B). When the reaction was performed with NMe3BH3 as reducing agent, only a 10% of 3a was obtained after 16 h. Moreover, when NH2MeBH3 and NHMe2BH3 complexes were employed, a 55% and 36% of 3a was obtained, respectively. These results clearly indicate protons of AB play a key role in the transfer hydrogenation. Subsequently, the same reactions were performed in the presence of 1.0 equiv of H2O. Interestingly, with NMe3BH3 31% of 3a was observed, improving the yield of the anhydrous reaction. With NHMe2BH3 and 1.0 equiv H2O the yield was improved to 90% after 16 h, and with NH2MeBH3 complex the reaction time was dramatically reduced to fully convert 2a to 3a. These results support the experimental observations in Scheme A. As control, the reaction in the presence of NH3BEt3 did not lead to conversion of the starting material even after 16 h, which indicates the relevance of the hydride source of AB.

Scheme 1

(A) Dehydrogenation of AB with Bi(I) and (B) Transfer Hydrogenation Using Different Amine Borane Complexes

To further evaluate the formal oxidation en route to bismuthine 6, primary kinetic isotope effects were measured using deuterium labeled AB (Scheme A). When the reaction was performed with 1.0 equiv ND3BH3 and 1.0 equiv of D2O (rapid D-exchange between H2O and ND3BH3 would result in misleading KIE values) a primary KIE value of 1.63 was obtained. With 1.0 equiv of labeled NH3BD3, a higher KIE value of 3.94 is observed. Finally, when the reaction was performed with 1.0 equiv ND3BD3, a large KIE value of 7.05 was obtained. These results suggest a mechanistic scenario in which both N–H and B–H bonds are broken in the rate-determining step (RDS).[25] Furthermore, a competition experiment between two-electronically distinct azoarenes was performed (Scheme B). When 1.0 equiv of 2a and 1.0 equiv of 2d were mixed with 1.0 equiv of AB in the presence of 1 mol % of 1, a 1:1 mixture of 3a and 3d was obtained after 1 h, suggesting that azoarenes are not participating in the RDS of this transformation.[26] These experiments point out to an scenario in which Bi(I) and AB are both involved in the RDS. However, the rate acceleration observed when H2O is present in the system suggests that H2O might interact with AB through H-bonding, and also participate in the RDS.[27]

Scheme 2

(A) Kinetic Isotope Effect Experiments; (B) Competition Experiment between Two Electronically Different Azoarenes

Bi(III)–H compounds are known to be highly unstable and reactive, which complicates their characterization as potential intermediates.[16a] However, a series of experiments were designed to elucidate the presence of such elusive species (Scheme A). Hence, when the reaction was performed with fluorobismuthine 7 and Ph3SiH, 3a was obtained in 71% yield. Equally, when 8 was mixed with 2.0 equiv K-Selectride or NaBH3CN, 3a was obtained in 95% and 98% yield respectively, together with Bi(I),[19] indicating that both H in 3a derive from the hydridic sources. Importantly, all these reactions afforded Bi(I) (1) and H2 when no azobenzene was present in the mixture.[16] While efforts to detect these intermediates by NMR spectroscopy were unsuccessful, we decided to investigate the dehydrogenation of AB by HRMS techniques. Indeed, when 1 is mixed with 5.0 equiv of AB, a peak corresponding to [6–H]+ (C16H24BiN2+, experimental: m/z = 453.1738; simulated: m/z = 453.1737) was observed (Scheme B), thus suggesting the formation of Bi(III)-hydride species.[28] Similarly, when AB was replaced by its deuterated analog, a mass for the Bi(III) deuteride was detected (C16H23DBiN2+, experimental: m/z = 454.1802; simulated: m/z = 454.1801. Additionally, the same Bi(III) hydride species was detected when the transfer hydrogenation of 2a was performed under catalytic conditions and analyzed by HRMS (Scheme C). Taken together, these results indicate formation of Bi(III) hydrides in the dehydrogenation of AB,[16] which further react in the presence of azoarene 2a to obtain 3a. Although different scenarios could be foreseen from such hydridic intermediate, further computational and spectroscopic evidence is needed to fully understand its role in the formation of 3a. Indeed, these studies are now being pursued in our laboratory.

Scheme 3

(A) Stoichiometric and (B and C) Mass Spectrometry Studies

2.0 equiv of reducing agent.

In conclusion, this work demonstrates the capacity of bismuth compounds to be engaged in catalytic redox transformations. The described protocol, which is a unique example of Bi(I) catalysis, resulted useful for the transfer hydrogenation of azoarenes and nitroarenes with AB as hydrogen surrogate. Preliminary mechanistic investigations suggest the intermediacy of highly reactive and elusive Bi(III) hydrides. These results constitute a unique proof-of-concept of a pnictogen operating between oxidation states I and III to mimic transformations typically performed by TM-catalysts.