Silylamido supported dinitrogen heterobimetallic complexes: syntheses and their catalytic ability

Abstract Molybdenum dinitrogen complexes supported by monodentate arylsilylamido ligand, [Ar(Me3Si)N]3MoN2Mg(THF)2[N(SiMe3)Ar] (5) and [Ar(Me3Si)N]3MoN2SiMe3 (6) (Ar = 3,5-Me2C6H3) were synthesized and structurally characterized, and proved to be effective catalysts for the disproportionation of cyclohexadienes and isomerization of terminal alkenes. The 1H NMR spectrum suggested that the bridging nitrogen ligand remains intact during the catalytic reaction, indicating possible catalytic ability of the Mo-N=N motif.

In these systems, dinitrogen (N 2 ) as a weakly π -accepting ligand to stabilize highly reactive and low valence-electron species, was proved not to be involved in the catalytic processes. Actually, there are only a few examples of M-N 2 units as active sites in catalytic organic transformations. In 2004, Hidai reported that Ti-W heterobimetallic dinitrogen complexes were excellent precursors for copolymerization of ethylene and 1-hexene, in which the W-N 2 fragment acted as a unique spectator ligand to the catalytically active titanium center [49]. The intriguing results hint at potential reactivity of the coordinated N 2 units in organometallic catalysis. Herein, we synthesized and structurally characterized molybdenum-nitrogen complexes supported by monodentate arylsilylamido ligand (L = [N(SiMe 3 )Ar]). Meanwhile, we observed catalytic reactivity of the Mo-N 2 unit as a key motif in disproportionation of cyclohexadienes and isomerization of terminal alkenes where the -N 2 ligands unusually remain intact. In this catalytic reaction, the Mo-N=N motif was considered as a possible catalytic site to advance the hydrogen transfer (Scheme 1).

RESULTS AND DISCUSSION
The reaction of MoCl 3 (THF) 3 (1) with 1.5 equivalents of lithium N-(trimethylsilyl)anilide (2) in Et 2 O for 5 h afforded the corresponding tris-anilide complex of Mo[N(SiMe 3 )Ar] 3 (3) in moderate yield (Scheme 2, a). X-ray diffraction on single crystals revealed that three-coordinate 3 was mononuclear with silylamino substituents arrangement above the trigonal plane of MoN 3 core ( Fig. 1, a). Compared with complex Mo[N( t Bu)Ar] 3 [73,74], a long Mo-N1 distance (1.985Å) and small Mo-N-Si bond angles (126 • ) might arise from the slightly different steric hindrance and electronic pattern of silylamido ligand around the Mo center. We then carried out reaction 3 with N 2 in the condition for conversion of Mo[N( t Bu)Ar] 3 to N≡Mo[N( t Bu)Ar] 3 (1 atm of N 2 , d 8 -toluene, −35 • C). With less electron negativity and the poor electron donating ability of silicon (Si), 3 was proved unreactive with N 2 molecules even at −35 • C for 5 days. Lengthening the reaction time of MoCl 3 (THF) 3 and lithium amide, the 1 H NMR spectrum of the crude product mixture showed that Mo III 3 had disappeared. The Mo IV -Cl complex 4 was observed as the only product, along with a small amount of free ligand HN(SiMe 3 )Ar (Scheme 2, b). A similar result was reported by the Fürstner group to isolate complex ClMo[N( t Bu)Ar] 3 (Ar = 3,5-dimethoxyphenyl) [86].
Reduction of the Mo IV -Cl reaction mixture with magnesium powder under N 2 atmosphere (1 atm) produced the Mo-N 2 complex (Scheme 2, c). Diamagnetic signals in the proton NMR spectrum indicated a high oxidation state of the Mo center. X-ray study gave the unambiguous structure of [Ar(Me 3 Si)N] 3 MoN 2 Mg(THF) 2 [N(SiMe 3 )Ar] (5), in which Mo and Mg were both supported by silylamido ligand, and bridged by dinitrogen ligand to form the heterobimetallic dinitrogen complex (Fig. 2, a). The bond length of N-N is 1.194Å, indicating the possible feature of  d). An X-ray analysis of this complex showed a long N-N bond (1.214Å) (Fig. 2, b). An analogous Mo-N=N-Mg complex supported by other silylamido ligand -N(SiMe 2 t Bu)Ar was prepared using similar procedures and characterized by X-ray chromatography (Fig. S6). Unfortunately, attempts to obtain analytically pure material failed.
According to those structural features, we envisioned that the hydrazine-like Mo-N=N-Mg backbone (5) with synergetic effects of both alkalineearth metal and transition metal might be a good precursor to transfer two nitrogen atoms to organic molecules through [4 + 2] cycloaddition [87,88]. However, after treatment of 5 with a stoichiometric amount of 1,3-cyclohexadiene (7) in C 6 D 6 at 100 • C for 24 h, it was found that featured signals of the complex remained in the 1 H NMR spectrum of the reaction mixture, implying the remarkable stability of 5 under the reaction conditions. The substrate 7 was completely consumed and two new sets of 1 H NMR signals appeared, which were a perfect fit for the disproportionation products benzene (8) and cyclohexene (9) (Fig. S7). Such disproportionation has been studied with different transition-metal catalysts [89][90][91][92][93][94][95][96], but the catalytic ability of complexes with metal-dinitrogen moiety, particularly from N 2 gas, has not yet been observed. This study could stimulate new avenues to develop efficient catalysts directly from N 2 . On the other hand, identification of the active centers (Mo or Mg) of the bimetallic complex and their exact behaviors also attracted us to unveiling of the catalytic disproportionation.
To prove the catalytic reactivity of 5, two experimental protocols were followed. Firstly, having confirmed the stoichiometric conversion of 1,3-cyclohexadiene, we added an additional amount of the substrate up to 60 equivalents in portions. Indeed, 90% of 7 converted to 8 and 9 after 18 days (Fig. S8, catalyst/substrate = 0.016/1), showing that the catalyst remained active. We also conducted the reaction by adding 60 equivalents of 7 into the C 6 D 6 solution of 5 in one portion. After 180 h, the conversion of 7 was 99%, monitored by in situ NMR (Fig. S9). A small amount of cyclohexane (10) was detected during the process, suggesting potential hydrogen-transfer reduction of non-conjugate alkene with this catalyst. Isomerized
1,4-cyclohexadiene (11) was also observed, which could also further transform into 8 and 9 by catalytic disproportionation (Fig. S13). The reaction pathway was different from a previously reported case in which a cationic molybdenum nitride species transferred a hydrogen atom from 11 to afford molybdenum imide complex and 8 in a stoichiometric manner [97]. Attempts were made to gain insight into the catalytic reaction (Table 1). Complex 5 exhibited comparable activity towards the disproportionation under 100 • C for 24 h on a 0.30 mmol scale (en-try 1), while 6 displayed high catalytic reactivity with 94% conversion within 12 h (entry 2). It was noteworthy that those catalysts remained intact after complete conversion of 7 (Figs S10 and S14), suggesting that the activated -N 2 units were retained during catalysis. These observations were different from previously reported catalytic transformations in which the electrically neutral and weakly activated N 2 units were substitutable ligands for substrate binding [31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48]. When performed under an Ar atmosphere or 10 atm of N 2 , reactions also smoothly occurred (entry 3 and 4). Unlike Mo-N 2 complexes, featured signals of 3 could not be detected after heating at 100 • C for a short time (< 5 h) despite its low catalytic competence (entry 5). Related [N] 3 Mo IV -Cl complex failed to promote such disproportionation efficiently (entry 6). Magnesium and lithium N-(trimethylsilyl)-3,5dimethylanilides were also tested, but failed (entry 7 and 8). Therefore, these results indicate that Mo-N=N moiety was a key structure and the active site was located at the Mo center. Kinetic studies showed that the initial rate of disproportionation dependent on the concentration of catalyst 5 was first order (Fig. S25), further evidence that 5 was not a precatalyst in the transformation.
Based on these observations, we proposed a plausible catalytic pathway shown in Scheme 3. After coordination of diene 7 to the Mo center (A), the resulting activated allylic hydrogen was transferred from 1,3-cyclohexadiene to N α atom through ligand-to-ligand hydrogen transfer (LLHT) [98][99][100][101], to form a cyclohexadienyl-Mo complex (B) with hydrazine as a ligand. β-hydride elimination of B released benzene and afforded the key intermediate Mo-H species (C), which coordinated with another molecule of 1,3-cyclohexadiene with subsequent insertion (or hydromolybdation) to generate a cyclohexenyl-Mo species (E), further undergoing reverse LLHT to produce cyclohexene and regenerate the catalyst.
According to the above proposal, complex 5 should be suitable to catalyze alkene isomerization through LLHT-reverse LLHT process. We examined allylbenzene (12) as substrate and found that the isomerization indeed took place at high efficiency to afford thermodynamic trans-adduct 13 as the product (Scheme 4, eq 2). 1-Hexene (14) was also submitted to the reaction system and internal alkenes were produced, albeit with poor site-and regio-selectivity (Scheme 4, eq 3). Kinetic studies indicated that C-H cleavage was not involved in the rate-determining step (Scheme 4, eq 4-6), consistent with the feature of hydrogen transfer between ligands [99].

CONCLUSIONS
In summary, we demonstrated that monodentate silylamido substituents could serve as excellent ligands for supporting low-valence complexes [N] 3 Mo and heterobimetallic dinitrogen complexes [N] 3 Mo-N 2 -Mg[N], which showed great catalytic ability in the disproportionation of cyclohexediene and isomerization of terminal alkenes with -N 2 ligands intact. Preliminary mechanistic studies indicated that the active catalytic center was the Mo-N 2 moiety through a ligand-to-ligand hydrogen transfer process. The detailed mechanism and new catalytic applications of these M-N 2 complexes in organic transformations are currently under consideration.

DATA AVAILABILITY
The X-ray crystallographic coordinates for the structures of 3, 4, 5, 6 and Mg(THF) 2 [N(SiMe 3 )Ar] 2 reported in this paper have been deposited at the Cambridge Crystallographic Data Centre (CCDC) under deposition numbers 1963795, 1963701, 1963702, 1963703 and 1963706. These data can be obtained free of charge from http://www.ccdc.cam.ac.uk/data request/cif.

SUPPLEMENTARY DATA
Supplementary data are available at NSR online.