N=N Bond Cleavage by Tantalum Hydride Complexes: Mechanistic Insights and Reactivity

The reaction of [TaCpRX4] (CpR = η5-C5Me5, η5-C5H4SiMe3, η5-C5HMe4; X = Cl, Br) with SiH3Ph resulted in the formation of the dinuclear hydride tantalum(IV) compounds [(TaCpRX2)2(μ-H)2], structurally identified by single-crystal X-ray analyses. These species react with azobenzene to give the mononuclear imide complex [TaCpRX2(NPh)] along with the release of molecular hydrogen. Analogous reactions between the [{Ta(η5-C5Me5)X2}2(μ-H)2] derivatives and the cyclic diazo reagent benzo[c]cinnoline afford the biphenyl-bridged (phenylimido)tantalum complexes [{Ta(η5-C5Me5)X2}2(μ-NC6H4C6H4N)] along with the release of molecular hydrogen. When the compounds [(TaCpRX2)2(μ-H)2] (CpR = η5-C5H4SiMe3, η5-C5HMe4; X = Cl, Br) were employed, we were able to trap the side-on-bound diazo derivatives [(TaCpRX)2{μ-(η2,η2-NC6H4C6H4N)}] (CpR = η5-C5H4SiMe3, η5-C5HMe4; X = Cl, Br) as intermediates in the N=N bond cleavage process. DFT calculations provide insights into the N=N cleavage mechanism, in which the ditantalum(IV) fragment can promote two-electron reductions of the N=N bond at two different metal–metal bond splitting stages.


■ INTRODUCTION
The study of N−N bond cleavage reactions is of great interest for the development of synthetic transformations using azo compounds as precursors of [NR] fragments 1−3 and, more importantly, for a mechanistic understanding of the industrial and biological dinitrogen (N 2 ) reduction to ammonia (NH 3 ). 4,5 Among the variety of metal compounds capable of promoting these reactions, 6 low-valent early transition metals have attracted significant attention because they lead to metal imide fragments, which are important intermediates for a variety of catalytic processes such as nitrogen transfer, hydroamination, and metathesis reactions. 7 The reduced metallic compounds required for the multielectron N−N scission process can be accessed by three different pathways. The first is the use of strong reductants such as alkaline and alkaline-earth metals, amalgams, alloys, and naphthalenides. 8 This is nicely illustrated by the chemical reduction of [V( iPr BPDI)Cl 3 }] ( iPr BPDI = 2,6-(2,6-iPr 2 -C 6 H 3 NCMe) 2 C 5 H 3 N) using sodium amalgam to form the vanadium dinitrogen complex [{V( iPr BPDI)(thf)} 2 (μ-N 2 )], which reacts with azobenzene to form a bis(imido) compound. 9 In general terms, this approach has important limitations in the isolation of pure low-valent metal compounds, which are typically contaminated by the corresponding inorganic salts and over-reduced impurities. Improving this strategy, Mashima 10 has reported a salt-free methodology employing a series of organic reducing reagents. This has been recently implemented in the formation of niobium and tantalum 2-pyridylimido compounds through the reaction of MCl 5 (M = Nb, Ta), 2,2′-azopyridine, and 1methyl-3,6-bis(trimethylsilyl)-1,4-cyclohexadiene as the reducing agent. 11 Although this synthetic route succeeds in producing organometallic compounds with high purity, its use has been limited to halide and cyclopentadienyl group 4 and 5 derivatives.
Alternatively, metal hydride compounds have proved to be efficient precursors of low-valent species, H 2 being the only byproduct generated. 12 −15 In line with the latter, Hou described the potential of his titanium hydride cluster [{Ti(η 5 -C 5 Me 4 SiMe 3 )} 3 (μ 3 -H)(μ-H) 6 ] in the fixation and functionalization of N 2 . 16 Notably, most of the investigations on N−N bond activation on azo compounds with low-valent early transition metals employ azobenzene as a model substrate and metallic reducing agents. 6 On consideration of the advantages of metal hydride species, it is surprising that there has been only one example reported with a mid-valent transition metal. In this contribution, Holland et al. explored the ability of a highspin iron(II) hydride dimer to break the NN double bond in azo aromatics. 17 Filling this gap for early transition metals, herein we report the NN double-bond cleavage of the azo aromatics azobenzene and benzo [c]cinnoline, mediated by the series of dinuclear tantalum(IV) hydride complexes [(TaCp R X 2 ) 2 (μ-H) 2 ] (Cp R = η 5 -C 5 Me 5 , η 5 -C 5 H 4 SiMe 3 , η 5 -C 5 HMe 4 ; X = Cl, Br). Structurally mapping these reactions shows how the tantalum(IV) hydrides partially or totally break the NN double bond with elimination of H 2 . In addition, insights into the electronic structures of dinuclear hydride tantalum(IV) compounds and a detailed atomistic description of the reaction mechanism are gained by theoretical studies.

■ RESULTS AND DISCUSSION
The preparation of the dinuclear hydride tantalum(IV) compound [{Ta(η 5 -C 5 Me 5 )Cl 2 } 2 (μ-H) 2 ] (1) was first reported by Schrock and co-workers, via hydrogenation of both the bis(neopentyl) species [Ta(η 5 -C 5 Me 4 R)-(CH 2 CMe 3 ) 2 Cl 2 ] (R = Me, Et) and the propylene complex [Ta(η 5 -C 5 Me 4 R)(MeCHCH 2 )Cl 2 ]. 18 Later, its synthesis was revisited by Messerle and co-workers, who prepared the hydride complex from the reaction of SnHBu 3 with [Ta(η 5 -C 5 Me 4 R)Cl 4 ] (R = Me, Et) (Scheme 1). 19 Despite both groups providing a detailed characterization in solution, the determination of the solid-state structure was still elusive. We were delighted to observe that heating a toluene solution of SiH 3 Ph and [Ta(η 5 -C 5 Me 5 )Cl 4 ] at 100°C for 24 h affords, after slow cooling at room temperature, compound 1 as a crystalline solid in 92% yield (Scheme 1). Notably, this reaction is accompanied by a dramatic change in color from orange to dark blue.
This methodology is widely applicable, as evidenced by the synthesis of the series of hydride halo (chloro and bromo) tantalum species [(TaCp R X 2 ) 2 (μ-H) 2 ] (Cp R = η 5 -C 5 Me 5 , η 5 -C 5 H 4 SiMe 3 , η 5 -C 5 HMe 4 ; X = Cl, Br) (1−3 and 1Br−3Br) supported by different cyclopentadienyl derivatives. These compounds were isolated in high yields and characterized by X-ray diffraction analyses. For reasons of similarity in the synthetic protocols and the solid-state structures of the chloro and the bromo compounds, the experimental details and figures with the molecular structures of the latter derivatives are given in the Supporting Information.
The solid-state structures of compounds 1 and 2 are depicted in Figure 1, while those of 1Br−3Br can be found in Figures S1−S3 in the Supporting Information. Relevant bond distances and angles are given in Table 1. These products have a dinuclear structure with two bridging hydride units, consistent with the 1 H NMR spectra. Both tantalum centers exhibit a distorted-trigonal-bipyramidal geometry comprising the centroid of the cyclopentadienyl ring, two chlorine/ bromine atoms, and two hydride ligands, in some cases related by a crystallographic binary axis perpendicular to the Ta−Ta bond. The two tantalum atoms are separated by distances ranging from 2.753(2) to 2.813(1) Å, with the largest values being found for compounds 1 and 1Br containing the bulkier pentamethylcyclopentadienyl ligand, which impedes a closer approximation between both metal centers. In line with this observation, the absence of cyclopentadienyl ligands in previously reported systems containing bridging hydrides results in slightly shorter metal−metal bond lengths. 20 The Ta−Cl average bond length of 2.36(1) Å in complexes 1 and 2 is consistent with a Ta(IV)−Cl single bond, although it is slightly shorter than Ta−Cl distances found in the related ditantalum hydride complexes [{TaCl 2 (PMe 3 ) 2 } 2 (μ-H) 2 ] (average 2.418(3) Å) 20c and [{TaCl 2 (PMe 3 ) 2 } 2 (μ-H) 4 ] (2.461(5) Å). 20d Likewise, the Ta−Br average bond length of 2.505(4) Å in 1Br−3Br is typical of single-bond distances. Additionally, the hydride atoms in these species were characterized as asymmetrically bridging the two tantalum atoms (see Table 1).
These features led us to examine the electronic structure of these complexes by DFT calculations. Full geometric optimization of complexes 1−3 led to calculated geometric parameters in excellent agreement with the experimental values (see Table S2). Figure 2 shows the analysis of the frontier molecular orbitals (MOs) for complexes 1 and 3, while Figure  S60 shows the MOs of 2. In all structures, the highest occupied molecular orbital (HOMO) consists of a bonding combination of atomic d-type orbitals centered at both tantalum centers, which is a clear indication of a σ Ta−Ta bonding between two Ta(IV) centers. The lowest unoccupied molecular orbital (LUMO) corresponds to a nonbonding orbital based on dtype orbitals centered at the Ta centers. The large energy gap between the HOMO and LUMO, ∼3 eV, further supports the experimental diamagnetic behavior.
To explore the potential of the hydride complexes in NN bond cleavage, we began our studies using azobenzene. Thus, treatment of a dark blue toluene solution of [{Ta(η 5 -C 5 Me 5 )Cl 2 } 2 (μ-H) 2 ] (1) with 1 equiv of azobenzene at room temperature yielded, after 12 h, a reddish orange solution formed by a mixture of products, from which only the mononuclear imide complex [Ta(η 5 -C 5 Me 5 )Cl 2 (NPh)] (4) could be unequivocally identified. Accordingly, single crystals grown from the latter reaction mixture allowed us to confirm the identity of compound 4. In an improvement of the synthesis, a thermal treatment at 50°C for 24 h of equimolar amounts of 1 and PhNNPh renders compound 4 in high yield and purity (Scheme 2).
The monomeric nature and the absence of hydride ligands in the solid-state structures of the complexes 4−6 and 4Br−6Br suggest a metal-based four-electron reduction of azobenzene proceeding via Ta−Ta bond scission and reductive elimination of hydrogen. Similar metal−metal bond metathesis reductions of arylazo reagents to give imido fragments have been previously reported in the literature. Thus, Cotton and co-workers described the reaction of the MM complexes [Cl 2 (R 2 S)M] 2 (μ-Cl) 2 (μ-SR 2 ) (M = Nb, 22 Ta 23 ) with PhN NPh to form dinuclear imido metal species. In a similar fashion, Tsurugi, Mashima, and co-workers reported the cleavage of the NN double bond present in benzo[c]cinnoline by reaction with a WW triply bonded compound, (tBuO) 3 WW(OtBu) 3 , affording dinuclear (imido)tungsten compounds. 24 In contrast, the hydrogen via has been only explored by Holland and co-workers in the treatment of the low-coordinate iron(II) hydrides [L MeEt Fe(μ-H)] 2 (L MeEt = bulky β-diketiminate ligand) with PhNNPh to give rise to an iron(III) imido dimer. 17  Encouraged by these results, we expanded our studies toward NN bond cleavage of the benzo[c]cinnoline. The treatment of complexes 1 and 1Br with this diazo cyclic reagent in toluene at room temperature resulted in a dramatic color change from dark blue-green to orange-yellow to yield the aryl-imido complexes 7 and 7Br, as outlined in Scheme 3.
The identity of complex 7 as an aryl-imido complex was determined by a single-crystal X-ray diffraction study. The molecular structure of 7 is depicted in Figure 3, and a selection of the most relevant bond distances and angles is given in Table 2 25 Intrigued by the formation of the intermediates 8, 8Br, 9, and 9Br, we monitored the reactions by 1 H NMR spectroscopy, which display the release of molecular hydrogen (δ 4.46) and the presence of a symmetrical cyclopentadienyl tantalum tricyclic aryl moiety. The selective reduction of benzo[c]cinnoline by these compounds is in contrast with other potential four-electron or greater reductant species acting through bond metathesis, where the complete reduction and cleavage give the only observed product. 24 In fact, these types of intermediates with benzo[c]cinnoline are usually accessed by employing two-electron-reducing derivatives such as those reported for niobium, 26 iron, 27 and zirconium. 28 Complexes 8Br and 9 were crystallized from hexane or toluene solutions, and the determination of their solid-state structures shows the geometry displayed in Scheme 3. Figure 4 depicts the molecular structure of 9, while that of 8Br can be found in Figure S5 in the Supporting Information. Relevant bond distances and angles are given in Table 2.
The structure of complexes 8Br and 9 consists of two [TaCp R X 2 ] (Cp R = η 5 -C 5 H 4 SiMe 3 , X = Br; Cp R = η 5 -C 5 HMe 4 ; X = Cl) units bridged by a diazene ligand bound in a μ-η 2 ,η 2 fashion. The two-electron reduction of the NN bond is reflected by the much larger bond distance (8Br, 1.450(3) Å; 9, 1.46(1) Å) in comparison to that found for benzo[c]cinnoline (1.292(3) Å). 29 A further analysis reveals that incorporation of the diamido fragment in the tantalum coordination sphere affects the Ta−Ta bond, which becomes slightly longer (2.917(1) Å (8Br) and 2.960(1) Å (9)) than   Computational Characterization of the Mechanism of NN Double-Bond Cleavage. To gain further insights into the four-electron NN double-bond cleavage of azobenzene by the dinuclear Ta(IV) hydride complexes, we performed DFT calculations on the transformation of 1 to the imide complex [Ta(η 5 -C 5 Me 5 )Cl 2 (NPh)] (4). Figure 5 depicts the potential free energy profile of the proposed mechanism as solid black lines, and Figure 6 shows the most relevant structures. To facilitate the discussion, the mechanism is divided into three stages: (i) an initial two-electron reduction of azobenzene by Ta(IV)−Ta(IV) bond splitting and oxidation to yield the dinuclear Ta(V) hydride intermediate coordinating a (μ-η 2 :η 2 -PhN−NPh) 2− fragment, (ii) H 2 reductive elimination to recover the Ta(IV) oxidation state and metal−metal bond, and (iii) a second two-electron reduction and N−N bond cleavage by virtue of the Ta(IV)− Ta(IV) bond scission and oxidation to form the Ta(V) imido complex 4. This sequence is analogous to the mechanism proposed by Holland and co-workers for the azo NN bond cleavage by an iron(II) hydride dimer, in which some intermediates were characterized computationally. 17 For the first step, we have been able to identify several adducts resulting from the coordination of the diazobenzene to complex 1; however, in all cases coordination required the cis configuration of the azobenzene, in which both phenyl substituents are pointing outside of the coordination sphere of the Ta dimer, to minimize the substrate−complex steric congestion. The trans−cis isomerization can occur photo-  . Gibbs free energy profile (kcal mol −1 ) for the reaction mechanism of 1 to 4 (solid black lines). An alternative, higher-energy mechanism is represented by blue lines.
Inorganic Chemistry pubs.acs.org/IC Article chemically and thermally, and the mechanisms have been extensively studied. 30 In our case, we computed a free energy penalty of 12.7 kcal mol −1 for the isomerization, which is in good agreement with the experimental value reported, ΔH = 12−13 kcal mol −1 . 31,32 Among the variety of identified adducts, the most stable 1A shows the azobenzene η 1 coordinated to a single tantalum formed in a slightly endergonic (+5.0 kcal mol −1 ) process (see Figure 5). From 1A, the azobenzene can undergo a two-electron reduction by Ta(IV)−Ta(IV) splitting and oxidation to Ta(V), overcoming a low free-energy barrier of 6.9 kcal mol −1 (ΔG ⧧ , 1A → TS 1A-1B ). Overall, the first reduction toward formation of intermediate 1B is an exergonic process with 25.3 kcal mol −1 , which balances the energy penalty of trans−cis isomerization of diazobenzene and coordination (∼24 kcal mol −1 ). Intermediate 1B is a tantalum(V) complex with a hydrazido ligand coordinated in a (η 2 -PhN−NPh) 2− fashion to only one of the metal centers. In addition, the calculated geometrical parameters for 1A to 1B reflect the partial cleavage of the NN double bond to a single N−N bond and a change in the oxidation state from Ta(IV) to Ta(V) with concomitant cleavage of the metal−metal bond. Thus, the N−N and the Ta−Ta distances lengthen from 1.25 to 1.42 Å and from 2.81 to 3.51 Å, respectively. Further rearrangement gives rise to the isomer 1B′ that is more stable by 10.6 kcal mol −1 , in which the hydrazide bridges the two metal centers (see Figures 5 and 6). It is interesting to note that metal hydrazide complexes are important intermediates in the development of dinitrogen activation 33,34 and high-energy-density materials that release energy through dinitrogen evolution. 35 Next, the reaction continues through H 2 reductive elimination in 1B′ to form the intermediate 1C, similar to the experimentally isolated dinuclear μ-η 2 ,η 2 -benzo[c]cinnoline complexes (8 and 9). Indeed, the computed distances for Ta− Ta and N−N bonds in 1C (2.98 and 1.46 Å, respectively) are very close to those determined by an X-ray study for complex 9 (2.960 and 1.46 Å, respectively). For the hydrogen evolution, we consider both homolytic and the heterolytic H−H bond formation as shown in Figure 5 (black and blue lines, respectively). 36 With regard to the homolytic pathway (black line in Figure 5), it proceeds via the transition state TS 1B′-1C , involving coupling of one terminal and one bridging hydride ligand of 1B′ to extrude H 2 and reduce the metals to Ta(IV). In the resulting complex 1C, the Ta−Ta distance is within the range of a metal−metal bond (2.98 Å) and the HOMO corresponds to a bonding orbital analogous to those represented in Figure 2. Accordantly, the Ta−Ta distance in the transition state TS 1B′-1C (3.17 Å) is significantly shorter that than in the reactant 1B′ (3.42 Å) (see Figure 6). Although this step has the largest free energy barrier along the proposed mechanism (29.5 kcal mol −1 ), the transformation of 1B′ to 1C is an exergonic process driven by the very favorable thermodynamic nature of the H 2 release process (ΔG(1B′ → 1C) = −17.1 kcal mol −1 ).
Alternatively, we have also examined the heterolytic reactivity across a polarized tantalum−hydride bond, which resembles that described for polynuclear titanium complexes activating strong bonds (N−H, CO, or C−H) via addition across bridging Ti−X junctions (X = N, C). 37−40 Moreover, a recent computational study proposes that the addition of dihydrogen to dinitrogen on a dimeric Ti(III)−hydride complex occurs without change in the oxidation state of titanium, while the subsequent reduction of the N−N bond involves oxidation to Ti(IV). 41 In our case, a similar proposal implies the migration of a terminal hydride in 1B′ to one N atom of the hydrazide moiety to afford 1E. On consideration that intermediate 1E is 30.1 kcal mol −1 higher in energy than 1B′ and the subsequent heterolytic H−H bond formation, by reaction between the acidic proton of the NHPh fragment and the hydride ligand, requires an extra 10.4 kcal mol −1 (total value >40 kcal mol −1 ), this heterolytic pathway can be ruled out. We have also characterized another alternative heterolytic pathway, but it resulted in a more highly energy demanding process in comparison to that through the homolytic H 2 elimination, and it did not explain the formation of the μη 2 ,η 2 -aza intermediate observed experimentally (see Figure S61 in the Supporting Information).
The last stage of the reaction consists of a second twoelectron reduction of the N−N bond of the μ-η 2 ,η 2 -hydrazide by the ditantalum(IV) fragment in the intermediate 1C. The process occurs with a moderate free energy barrier (ΔG ⧧ (1C → TS 1C-1D ) = 16.8 kcal mol −1 ) and causes the complete cleavage of the N−N bond to give the bis-imido Ta(V) complex 1D. This bond cleavage is reflected in the evolution of N−N distances along the reaction coordinate, varying from 1.46 Å in 1C to 2.15 Å in TS 1C-1D and then to 3.17 Å in 1D ( Figure 6). The formation of intermediate 1D is notably exergonic (ΔG(1C → 1D) = −26.5 kcal mol −1 ), whereas its dissociation into two Ta(V) imido complexes 4 provides the thermodynamic driving force of the whole process, with product 4 lying 97.1 kcal mol −1 below the reactants.

■ CONCLUSION
Combining SiH 3 Ph and the mononuclear tantalum(V) derivatives [TaCp R X 4 ] (Cp R = η 5 -C 5 Me 5 , η 5 -C 5 H 4 SiMe 3 , η 5 -C 5 HMe 4 ; X = Cl, Br), we have developed a one-step methodology for the synthesis of dinuclear tantalum(IV) hydrides [(TaCp R X 2 ) 2 (μ-H) 2 ] (Cp R = η 5 -C 5 Me 5 , η 5 -C 5 H 4 SiMe 3 , η 5 -C 5 HMe 4 ; X = Cl, Br) in high yields. We have shown the ability of these species to promote a four-electron reduction and complete NN bond cleavage in their reactions with azobenzene and benzo[c]cinnoline. Moreover, we were able to trap a side-on-bound μ-η 2 ,η 2 -azo species in the reaction with the cyclic diazo reagent benzo[c]cinnoline as a plausible intermediate before cleaving the NN bond. Inorganic Chemistry pubs.acs.org/IC Article A DFT analysis of the electronic structures of dinuclear tantalum(IV) hydrides complexes supports the metal−metal bonding between the two tantalum atoms, as manifested by the HOMO nature, consisting of a bonding combination of atomic d-type orbitals centered at both tantalum centers.
We have proposed a plausible reaction mechanism on the basis of DFT calculations that consists of three main steps: (i) initial two-electron reduction of diazabenzene by the hydride ditantalum(IV) fragment to yield the intermediate hydride ditantalum(V) coordinated by a (μ-η 2 :η 2 -PhN-NPh) 2− fragment, (ii) homolytic H 2 evolution via reductive elimination to recover a dinuclear tantalum(IV) complex, and (iii) a second two-electron reduction by the ditantalum(IV) fragment, causing the complete cleavage of the N−N bond and yielding the tantalum(V) imido products.
The structures were solved by using the WINGX package 43 (Olex2 for [Ta(η 5 -C 5 HMe 4 )Br 4 ]), 44 by direct methods (SHELXS) 45 or intrinsic phasing (SHELXT, in the case of [Ta(η 5 -C 5 HMe 4 )Br 4 ], 1Br, 2Br, and 11) 46 and refined by least-squares against F 2 (SHELXL). 47 All non-hydrogen atoms were anisotropically refined, while hydrogen atoms were placed at idealized positions and refined using a riding model, except for the bridging hydride atoms in complexes 1, 1Br, 2, 2Br, and 3Br that were located in the difference Fourier maps and isotropically refined; after several cycles of refinement, the coordinates and thermal factors were fixed most of the times. Cyclopentadienyl carbon atoms presented some dynamic disorder in some complexes; thus EADP restraints were applied for 1, 1Br, 2, and 2Br, while SIMU gave better results in the case of compound 4.
Complex 3Br crystallized with two deuterated benzene solvent molecules. On the other hand, one toluene molecule could be modeled in the case of 9, although voids for two more molecules were found using the Platon Squeeze procedure. 48 By using this program, the contribution of these disordered toluene molecules to the structure factors was removed.
Computational Details. All calculations were performed using the Gaussian16 program package. 49 The geometries, energies, and electronic structures of all the investigated species were calculated within the density functional theory (DFT) 50 framework using the B3LYP functional. 51−53 A standard double-ξ Lanl2dz 54 basis set with an f polarization function for tantalum atoms and with a d polarization function for silicon and chlorine atoms was used. 55,56 For bromine atoms a Lanl2dz basis set with polarization and diffuse functions was employed. 57−61 The rest of the atoms such as nitrogen, carbon, and hydrogen were treated with a standard 6-31G(d,p) basis set. 62−64 We also applied a GD3 dispersion correction 65 as implemented in Gaussian 16. The transition states were characterized by a single imaginary frequency, whose normal mode corresponded to the expected motion. Moreover, an intrinsic reaction coordinate (IRC) Inorganic Chemistry pubs.acs.org/IC Article analysis was performed for almost all transition states to confirm their nature and the intermediates that were connected by the transition state. Finally, in order to change the reference from the ideal gas standard state of 1 atm (0.041 mol L −1 ) to 1 mol L −1 in the condensed state, we applied a correction on Gibbs free energy values of 1.89 kcal mol −1 .
■ ASSOCIATED CONTENT * sı Supporting Information The following files are available free of charge. (PDF). (XYZ) The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.1c03152.