Molybdenum catalyzed ammonia borane dehydrogenation: oxidation state specific mechanisms.

Though numerous catalysts for the dehydrogenation of ammonia borane (AB) are known, those that release >2 equiv of H2 are uncommon. Herein, we report the synthesis of Mo complexes supported by a para-terphenyl diphosphine ligand, 1, displaying metal-arene interactions. Both a Mo(0) N2 complex, 5, and a Mo(II) bis(acetonitrile) complex, 4, exhibit high levels of AB dehydrogenation, releasing over 2.0 equiv of H2. The reaction rate, extent of dehydrogenation, and reaction mechanism vary as a function of the precatalyst oxidation state. Several Mo hydrides (Mo(II)(H)2, [Mo(II)(H)](+), and [Mo(IV)(H)3](+)) relevant to AB chemistry were characterized.

There has been significant interest in transitioning from petroleum-based fuels to a “hydrogen economy,” with respect to both green energy and increased energy security.[1] A limitation to the implementation of hydrogen (H2) in transportation is its low energy density and complications arising from compressed gas storage.[1a,2] Numerous forms of chemical H2 storage from metal hydrides[1a] to metal organic frameworks[3] have been explored. A forerunner in this field is ammonia borane (AB, NH3BH3), a compound with a substantial gravimetric storage capability of 19.6 wt % H2 when dehydrogenated through the third equivalent (equiv).[4] AB shows promise for reversible H2 storage, with work toward efficient regeneration ongoing.[5]

Various AB dehydrogenation catalysts, including frustrated Lewis pairs[6] and ionic liquids,[7] have been investigated. Metal-based catalysts show the most potential for controlling both the rate and extent of H2 release[2] and have demonstrated high activities in the cases of Ir,[8] Ru,[9] and Pd.[10] Extensive H2 release is less common due to NH2BH2 oligomerization,[11] but examples are known for Ni (ca. 2.7 equiv),[12] Fe (ca. 1.7 equiv),[13] Pd (ca. 2.0 equiv),[10] Rh (ca. 2.0 equiv),[14] and Ru (ca. 2.3 equiv) (Table S1).[15] These catalysts either employ expensive metals (Ru[15] and Pd[10]) or suffer from instability (Pd,[10] Fe,[13] and Ni[16]). We report here the first examples of AB dehydrogenation catalysts based on Mo, an abundant and inexpensive metal. Our systems have demonstrated distinct behavior dependent on oxidation state, with isolated Mo0, MoII, and MoIV complexes capable of releasing over 2 equiv of H2 from AB under moderate conditions.

Transition metals ligated by para-terphenyl diphosphine 1 (Scheme 1) have been studied for new types of reactivity.[17] Mo complexes supported by diphosphine 1 were targeted to take advantage of the pendant arene acting as a versatile and hemilabile ligand for supporting different metal oxidation states and binding modes. Heating 1 in the presence of Mo(CO)3(MeCN)3 cleanly afforded the Mo para-terphenyl diphosphine complex 2 (Scheme 1). Through single crystal X-ray diffraction (XRD) analysis, η2-arene binding was observed with partial disruption of aromaticity in the central ring (Figure 1).

Scheme 1

Synthesis of Mo para-Terphenyl Diphosphine Complexes

Figure 1

Solid-state structures of 2–5 and 7–8. Selected bond distances are reported in Å. Solvent molecules, counteranions, and select hydrogen atoms are omitted for clarity.

Targeting open Mo coordination sites, decarbonylation was pursued. Oxidation of 2 with 2 equiv of silver trifluoromethanesulfonate liberated one of the carbonyl ligands, increasing the hapticity of the Mo–arene interaction and maintaining an 18-electron configuration at the metal. Compound 3 exhibits a uniform elongation of the arene C–C bonds in the solid state (Figure 1), consistent with η6-binding.

Irradiation of 3 with UV light at −78 °C in the presence of acetonitrile afforded a deep purple complex lacking C–O stretching bands in the IR spectrum. XRD confirmed complete decarbonylation to the η6-arene-bis(acetonitrile) complex 4 (Figure 1). Stirring 4 vigorously over Mg0 under an N2 atmosphere afforded the Mo0 dinitrogen complex, 5 (Scheme 1). The same species can be accessed upon treatment of 4 with LiHBEt3, albeit in lower yield. The 1H NMR spectrum displayed two central arene signals at 4.3 and 4.0 ppm, suggesting a pseudo-C symmetric structure in solution, similar to the solid state. The average Mo–C distances (Å) in 5 (2.255(1)) are shorter than those in 3 (2.379(1)) and 4 (2.312(2)), consistent with increased δ-backbonding from Mo0 compared to MoII (Figure 1).[18] The N–N IR stretching frequency, 2020 cm–1, is similar to previously characterized (C6H5Me)Mo(PPh3)2N2 (2000 cm–1).[19] Compounds 2–5 demonstrate the ability of diphosphine 1 to support Mo in multiple binding modes and oxidation states.

With a precedent for base–metal catalysts effecting extensive H2 release,[12] the reactivity of complex 5 with AB was tested. Addition of 1 equiv of AB led to partial conversion to a new species over several hours at room temperature. Although X-ray quality crystals have not been obtained, the 1H NMR spectrum for this compound displays a single central arene signal at 4.81 ppm, indicating pseudo-C2 symmetry, and a triplet at −4.10 ppm integrating to two protons, consistent with a Mo dihydride, 6. Complex 6 was independently synthesized via addition of H2 to complex 5. Under excess N2, 6 quantitatively reverts to 5, suggesting 6 as an intermediate in the LiHBEt3 induced formation of 5 from 4. The T1(min) (78 ms, 233 K, C7D8, 500 MHz; Figure S23) of 6 is inconsistent with a dihydrogen complex (T1(min) ca. 20 ms), but is shorter than a typical dihydride relaxation, suggesting intermediate character.[20] The hydride-deuteride isotopolog, 6-HD, displays coupling (JHD = 10.75 Hz, Figure S24) consistent with an H–D distance of 1.25[21] to 1.36[22] Å, further supporting this assignment. The Mo center in 6 is more electron rich than that of a similar Mo–dihydrogen complex bearing arene and CO ligands,[20h] facilitating conversion toward a dihydride structure.

Extending this stoichiometric reaction to a catalytic system, a 0.25 M AB solution in diglyme was treated with 5 mol % 5 at 70 °C and gas evolution was monitored via eudiometery (Figure S30). This system produced 2.5 equiv of H2 within 15 h, with the first 2 equiv liberated in 6.5 h (Figure 2). Such extensive H2 release, 2.5 equiv, is rare.[12] Dehydrogenation attempts with Mo0 powder showed no change from the uncatalyzed control, and in the presence of elemental mercury, catalysis still proceeds (Figure S30),[23] consistent with homogeneous catalysis.[24] Analysis of the final reaction mixture by 11B NMR spectroscopy showed a broad signal at 30 ppm corresponding to polyborazylene (PB),[11,25] in agreement with production of >2 equiv of H2.

Figure 2

Eudiometry of AB dehydrogenation catalyzed by 4, 5, and 11. Mo0 powder and catalyst-free controls are included for reference.

Other Mo0 complexes were tested for AB dehydrogenation activity for comparison (Table S1). Mo(N2)2(dppe)2, 9, proved ineffective, releasing less H2 than the control and instead forming the stable tetrahydride complex Mo(dppe)2(H)4.[26] A pyridinediphosphine-supported N2 complex, 10,[27] showed similar dehydrogenation activity to 5, though it ultimately provided less H2 (Figure S32). Reports of efficient dehydrogenative coupling of amino boranes by group 6 metal carbonyl species under thermo- or photolytic conditions[28] prompted the investigation of Mo(1,3,5-trimethylbenzene)(CO)3, 11,[29] which demonstrated a similar initial rate of AB dehydrogenation to 5, but yielded less H2 (Figure 2). Overall, precatalyst 5 is superior to other Mo0 species in terms of rate and extent of AB dehydrogenation (Table S1).

For comparison, catalytic trials were performed with MoII compound 4 under the aforementioned conditions, resulting in the release of 2 equiv of H2 in 8.5 h (Figure 2). Though complex 4 provided less extensive H2 release, the initial rate was significantly faster than that of 5. Similar to 5, addition of elemental mercury had no effect on the rate of dehydrogenation (Figure S30). The Mo oxidation state (Mo0 vs MoII) significantly affects the efficacy of the dehydrogenation catalysis, a phenomenon also observed for Fe-based systems.[13]

Interest in the disparate rate and extent of H2 release catalyzed by 4 and 5 prompted a closer investigation of their respective reactivity. Monitoring stoichiometric reactions of 4 with AB at 70 °C showed the formation of a single new species with a peak at 92 ppm in the 31P NMR spectrum. 1H NMR spectroscopy displayed two central arene signals suggesting pseudo-C symmetry and a triplet with a relative integration of one at −0.5 ppm, consistent with a Mo monohydride (Scheme 1). XRD analysis confirmed the structure as a cationic MoII hydride, 7, with an acetonitrile ligand completing the metal coordination sphere (Figure 1).

Treatment of 7 with AB resulted in partial conversion to another species (Scheme 1). The 1H NMR spectrum of the mixture showed a new hydridic triplet at −4.6 ppm, integrating to three protons with respect to a single central arene peak at 5.6 ppm, suggesting a more oxidized and symmetric complex: the MoIV trihydride cation, 8. Selective 1H decoupling of the 31P NMR spectrum (Figure S21) and independent synthesis via treating 7 with H2 further supported this assignment. The hydridic resonance of 8 has a T1(min) of 235 ms at 223 K (CD2Cl2, 500 MHz; Figure S23), on the order of reported Mo(H)3 complexes, indicating classical trihydride character.[20d,20e,30] Precipitation in the absence of acetonitrile allows for the isolation of 8. Acetonitrile promotes H2 loss and the formation of 7, with an equilibrium constant (Keq) of 0.3 at 25 °C, as determined from solution concentrations.

Both hydrides 7 and 8 were observed in catalytic AB dehydrogenation by 4, within 20 min, via 31P NMR spectroscopy. Consumption of AB was observed in the 11B NMR spectrum concurrent with the initial appearance of new signals at −20 and −12 ppm (B-(cyclodiborazanyl)aminoborohydride, BCDB, Figure S26)[11] followed by two resonances near 30 ppm appearing after 45 min (borazine and PB).[25,31] These observations suggest generation of NH2BH2 as a dehydrogenation intermediate on the way to borazine and PB.[11,25] The formation of NH2BH2 was corroborated by cyclohexene trapping (Figure S26). Catalysis in the presence of excess cyclohexene afforded a major peak by 11B NMR spectroscopy at 47 ppm, assigned to the hydroborylation product.[11]

The terminus of activation was investigated using substituted amine–borane adducts.[9c] Compound 4 reacts with NMe3BH3 to form 7 (Scheme 2) and not with Et3BNH3, suggesting B–H activation as an initial step. This is reminiscent of the reaction of 4 with LiHBEt3 (Scheme 1). Reactions of the isolated monohydride 7 with these AB analogs at 70 °C showed no change after 12 h even in the presence of excess substrate, providing no insight as to the site of reactivity with AB. Analogous experiments with 8 (Scheme 2) show reaction with Et3BNH3, but not NMe3BH3, in agreement with protic N–H hydrogens reacting with MoIV–H moieties. Catalytic trials with 4, 7, and 8 show that all are kinetically competent. Monitoring the rate of AB consumption by 11B NMR showed first-order kinetics through three half-lives. Isotope effects of 1.7, 1.6, and 3.3 were determined for ND3BH3, NH3BD3, and ND3BD3 dehydrogenation by 4, respectively (Figure S41). Though this may be consistent with H2 evolution as the rate-determining step, the observed equilibrium between 8 and H2 and potential H/D scrambling complicate the interpretation of these data. A mechanism consistent with the present findings involves initiation from precatalyst 4 to generate MoII-monohydride 7. Reaction of 7 with AB leads to MoIV-trihydride 8 and NH2BH2 which can undergo further dehydrogenation events.[11] Activation of AB by 8 can occur directly at the N terminus with subsequent release of H2 or via initial H2 release followed by reaction with AB.

Scheme 2

Proposed Mechanism for MoII-Initiated Catalytic AB Dehydrogenation

Similar studies of the catalytic behavior of 5 demonstrated the formation of 6 and disappearance of 5 within minutes. The 11B NMR spectrum showed the consumption of AB and the appearance of the intermediate BCDB. Borazine and PB were observed, in accordance with dehydrogenation past 1 equiv. Unlike complex 4, precatalyst 5 reacts with Et3BNH3, albeit not cleanly. Complex 6, isolated under argon, did not react with either terminus blocked substrate but did demonstrate AB dehydrogenation catalysis. The kinetics of AB dehydrogenation initiated by 5 were complicated, likely due to H2 inhibition. A catalytic cycle consistent with these results involves reaction of 5 with AB via oxidative addition of an N–H bond (Scheme 3). Dihydride 6 may be accessible via β-hydride elimination, releasing NH2BH2. Complex 6 could eliminate H2 followed by reaction with AB. Alternatively, 6 may react with AB directly and release H2. In both instances, 1 equiv of H2 is generated in an on-metal process, with the remainder derived from dehydrooligomerization of NH2BH2. Although NH2BH2 is generated from both 4 and 5, the difference in dehydrogenation extent is presently not well understood. It may be due to further metal-based reactivity of AB dehydrogenation products, in accordance with disparate byproduct distributions.

Scheme 3

Proposed Mechanism for Mo0-Initiated Catalytic AB Dehydrogenation

In summary, a series of Mo complexes have been shown to effectively catalyze the extensive dehydrogenation of AB, releasing ca. 2 equiv of H2 in four cases. The Mo0 compound 5 is a rare example of a transition metal precatalyst capable of dehydrogenating AB through 2.5 equiv of H2. These well-defined systems supported by the para-terphenyl diphosphine ligand, 1, have been studied mechanistically and exhibit different reaction pathways as a function of metal oxidation state. A series of isolated Mo-hydrides (MoII(H)2, [MoII(H)]+, and [MoIV(H)3]+) were found to support catalysis for the dehydrogenation of AB. Elucidation of the respective mechanisms and investigation of additional Mo-based AB dehydrogenation catalysts are ongoing.