Metal-Ligand Cooperative Synthesis of Benzonitrile via Electrochemical Reduction and Photolytic Splitting of Dinitrogen

Thermal nitrogen fixation relies on strong reductants to overcome the extraordinarily large N–N bond energy. Photochemical strategies that drive N2 fixation are scarcely developed. Here, the synthesis of a dinuclear N2-bridged complex is presented upon reduction of a rhenium(III) pincer platform. Photochemical splitting into terminal nitride complexes is triggered by visible light. Clean Ntransfer with benzoyl chloride to free benzamide and benzonitrile is enabled by cooperative 2H+/2e– transfer of the pincer ligand. A threestep cycle is demonstrated for N2 to nitrile fixation that relies on electrochemical reduction, photochemical N2-splitting and thermal N-

The Haber-Bosch process consumes large amounts of energy, for the generation of H2. Electrocatalytic N2 reduction has therefore been targeted as an attractive alternative. 1,2 Nitrogen fixation at ambient conditions with molecular catalysts has seen remarkable progress, 3 in some cases also for electrochemical ammonia generation. 4 The direct synthesis of other compounds than NH3 from N2 remains a formidable challenge. Catalytic protocols are only known for trisilylamine. 5 Nitriles, 6 , 7 , 8 isocyanates, 9 silylamines, 10 , 11 and borylamines, 12 have been synthesized within stoichiometric, cyclic reaction sequences that allow for evaluating strategies to offset the extremely strong NºN bond (225 kcal/mol), enable E-N (E = C, Si, B) bond formation and deliver six reduction equivalents. All reported 'synthetic cycles' proceed through initial N2 cleavage into nitride complexes. Subsequent N-transfer typically requires strong electrophiles like alkyl triflates. The thermochemistry of N2 splitting must therefore be tuned to avoid nitride overstabilization and enable functionalization with reagents that are more compatible with reductive conditions. We have examined N2 activation and splitting, i.e. triggered by chemical or electrochemical reduction of pincer halide complexes. 13 , 14 The rhenium(III) precursor [ReCl2(PNP tBu )] (1; PNP tBu = N(CH2CH2PtBu2)2) exhibits a complex mechanism via rapid Re III /Re II -reduction, N2-binding, halide loss, Re II /Re Ireduction and Re I /Re III -comproportionation. 13b Splitting of the resulting dinuclear complex [(µ-N2){ReCl(PNP tBu )}2] (2) is ratedetermining and gives the nitride complex [ReNCl(PNP tBu )] (3). A simple electronic structure model for the Re-N-N-Re core of key intermediate 2 (p 10 d 4 -configuration; Scheme 1) provides a starting point to tune the thermochemistry. 15 We now report a modified platform that splits N2 photolytically into more reactive nitrides that can be transferred with benzoyl chloride. The pincer ligand serves as 2e -/2H + reservoir, enabling electrochemical N2 reduction, photochemical splitting and thermal transfer within a three-step cycle. Scheme 1. Schematic frontier orbital correlation diagram for the splitting of N2bridged complex 2 into two nitrides 3 (n.b. = non bonding). 13b,15 The Re III complex [ReCl3(HPNP iPr )] (4) was obtained in 70 % isolated yield. 16 In contrast to 1, the amide complex [ReCl2(PNP iPr )] could not be isolated in analytical purity. 4 displays strongly shifted NMR resonances, as exemplified by the 31 P{ 1 H}-NMR signal (dP = -1525.9 ppm). However, the sharp signals exhibit well-resolved J-coupling. This observation can be rationalized with a magnetic, energetically isolated ground state that results from excited state admixture via spin-orbit coupling. Contributions from temperature independent paramagnetism (TIP) are well documented for 3 rd row complexes with d 4 (octahedral Re III , Os IV ) and d 6 configuration (square-planar, Os II ). 17 Single-crystal X-ray diffraction confirmed meridional coordination of the neutral diphosphinoamine ligand (Figure 1a). Complex 4 was investigated by cyclic voltammetry (CV) under argon. Quasireversible, scan rate dependent reduction is observed at -1.84 V suggesting chloride loss on the CV timescale. 16,18 Chemical reduction with 1 equiv. CoCp * 2 in tetrahydrofuran (THF) under an N2 atmosphere (Scheme 2) gives an intensely blue product (5) in around 60% isolated yield. Complex 5 exhibits sharp and strongly shifted NMR signals suggesting an even-electron compound with a TIP contribution. Symmetric N2 binding was confirmed by a single signal in the 15 N{ 1 H}-NMR spectrum and a band at 1733 cm -1 ( 15 N2-5: 1675 cm -1 ) in the resonance Raman spectrum. 16 Two doublets in the 31 P{ 1 H}-NMR spectrum (dP = -370.6/-380.4 ppm) with mutual trans-coupling ( 2 JPP = 237 Hz), eight 1 H NMR signals assignable to CH3-groups and one amine NH-signal are in agreement with an N2 bridged, dinuclear compound with C2 symmetry, in analogy to 2. 13b Structural assignment as [(µ-N2){ReCl2(HPNP iPr )}2] was further substantiated by LIFDI mass spectrometry.
The molecular structure was confirmed by single crystal Xray diffraction (Figure 1b). 16 The asymmetric unit features two octahedrally coordinated, linearly N2-bridged Re-ions. The N-N bond length (1.169(5) Å) is slightly shorter compared with 2 (1.202(10) Å), 13b indicating moderate N2 activation in agreement with the Raman data. In contrast to 2, the pincer nitrogen atoms are located trans to the N2 bridge. The two {ReCl2(HPNP iPr )} fragments are twisted with respect to each other by 75.5° giving rise to idealized C2 symmetry.
The striking thermal stability of 5 vs. 2 can be rationalized by qualitative molecular orbital (MO) considerations. Thermal cleavage of a linearly N2 bridged M2N2 core proceeds via electron transfer into an MO with M-N-N-M s-s*-s-character (Scheme 1) within a zig-zag shaped transition state (TS). 15,13b,19 Ligands trans to the N2-bridge raise this MO in energy, thereby disfavoring N2splitting. A classic case for such geometry controlled reactivity might be the trisanilides [(RArN)3Mo}2(N2)], 20 which split into nitrides in contrast to analogous triamidoamine complexes. 3a , 6 (c) and 7 (d) in the crystal obtained by single crystal X-ray diffraction. Hydrogen atoms were omitted for clarity except NH protons. 16 To overcome kinetically hindered N2-splitting we sought to populate N-N antibonding MOs by electronic excitation. 9, 21 , 22 Photolysis of 5 in THF with a Xe arc lamp (λ > 305 nm) led to gradual color change to yellow over 2 h. 31 P and 1 H NMR spectra revealed the formation of cis-dichloro nitride complex 6 (Scheme 2) in 95 % spectroscopic yield upon comparison with an original sample. 16 Use of 15 N2-labelled 5 confirmed photolytic splitting of the N2 ligand by 15 N{ 1 H} NMR spectroscopy. A quantum yield below 1 % was estimated by actinometry. 16 The configuration of 6 requires isomerization before or after N2-splitting. Photolysis was therefore carried out in the presence of NHex4Cl (0-500 equiv.). 16 The independence of the reaction rate excludes isomerization by chloride (photo)dissociation prior to or as the rate determining step.  (depf = 1,1'bisdi(ethylphosphino)ferrocene), which also splits into nitrides upon photolysis at 580 > λ > 400 nm. 22d In the solid state (Figure 1c), 16 6 retains octahedral coordination with one significantly elongated Re-Cl bond (2.4309(7) Å vs. 2.6712(7) Å), reflecting the nitride trans-influence. In turn, the Re≡N bond (1.669(2)Å) is longer compared to fivecoordinate [Re(N)Cl(HPNP tBu )]Cl (1.642(4) Å). 13a Addition of NaBAr F 24 (BAr F 24 = tetrakis-(3,5-(trifluoromethyl)phenyl)borate) to 6 in THF results in two new 31 P NMR signals that are assigned to two isomers of [Re(N)Cl(HPNP iPr )] + . 16 In turn, 6 is fully restored upon addition of NHex4Cl.
Hence, six-coordinate [Re(N)Cl2(HPNP iPr )] is the dominant species in solution. Reduced steric shielding facilitates chloride coordination and thereby weakening of metal nitride bonding.
Next, re-reduction of the imine pincer ligand was examined to evaluate the amine/imine redox couple as cooperating ligand reservoir for 2e -/2H + proton coupled electron transfer (PCET). 7 does not react with H2 under thermal or photolytic conditions. However, chemical reduction is possible with stepwise addition of Li[HBEt3] and diphenylammonium chloride (Scheme 2). On this route, 4 is obtained in 62 % spectroscopic yield due to the formation of rhenium hydrides as byproducts, which reform 7 upon hydrolytic quenching.
We therefore turned to electrochemical regeneration of 4. The CV of 7 exhibits quasi-reversible reduction at -1.70 V. Titrating in benzoic acid (0-15 eq.; Figure 3a) results in a pronounced increase of the cathodic current and buildup of a second, quasi-reversible reductive feature at -1.84 V, which is assigned to parent 4. The strong increase of the first wave is indicative of multielectron reduction in the presence of acid. In fact, the viable intermediate [ReCl3(PNP iPr )] (8) exhibits a quasireversible reduction at -1.16 V, 16 confirming a potential inversion after the first e -/H + -transfer to 7. The electrochemical data in the presence of acid is therefore rationalized with two subsequent PCET steps at around -1.7 V that regenerate 4.
Controlled potential electrolysis (CPE) of 7 at the half peak potential of the first reduction feature (E = -1.65 V) did not give appreciable amounts of 4 in the presence of benzoic acid (10 eq). 2,6-Dichlorophenol (DCP) was therefore employed, which exhibits about same pKa as benzoic acid (pKa THF = 25.1) 25 yet a conjugate base that is less prone to metal coordination. In fact, bulk electrolysis of 7 at E = -1.65 V in the presence of DCP (10 eq.) for 7 h results in full consumption of 7 and formation of 4 as the only electroactive species. 1 H NMR spectroscopy confirmed a yield of 99 % and coulometry the transfer of 1.96 eper 7. The quantitative regeneration of 4 motivated the examination of in situ electrochemical N2 activation. In a CPE experiment, 7 was electrolyzed with DCP (10 eq.) under N2 for 8 h at E = -1.65 V, followed by 5 h at E = -1.85 V, i.e. the half potential of the reduction of 4 (Figure 3b). Overall, 3.32 eper Re atom were transferred. Formation of complex 5 during the second electrolysis step was indicated by the deep blue color and confirmed by 1 H and 31 P{ 1 H} NMR spectroscopy. UV/Vis spectroscopic quantification gave a yield around 70 % (Scheme 3). In situ photolysis (390 nm LED) of this mixture revealed N2 splitting to nitride 6 as the only detectable product by 31 P{ 1 H} and 1 H NMR spectroscopy. However, the yield in 6 dropped to 14 % with respect to parent 7, i.e. considerably lower compared to photolysis of isolated 5.
In summary, a three-step cycle for the generation of benzamide/benzonitrile from N2 in overall 61 % yield (with respect to N) was established that relies on electrochemical N2 activation, photochemical splitting into nitrides and thermal nitrogen transfer (Scheme 3). This model reveals some basic principles in comparison to our previously reported system. 7,13 Use of a sterically less encumbered pincer ligand stabilizes higher coordination numbers. In consequence, thermal N2 splitting becomes less favorable which can be overcome by photolysis with visible light. Based on the experimental and theoretical data, the photochemical reactivity is associated with the population of a dissociative state with Re→N2 MLCT character. The higher coordination number weakens nitride bonding, thus enabling the use of a weaker electrophile than alkyl triflates. Importantly, the cooperating pincer ligand serves as a reservoir for nitrogen hydrogenolysis upon 2e -/2H + -PCET and electrochemical rehydrogenation. Our results demonstrate how metal-ligand cooperativity and photo-and electrochemical approaches can facilitate the design of platforms for N2-fixation. Scheme 3. Optimized, three-step synthetic cycle.