Rhenium-Mediated Conversion of Dinitrogen and Nitric Oxide to Nitrous Oxide.

Reductive splitting of N2 is an attractive strategy towards nitrogen fixation beyond ammonia at ambient conditions. However, the resulting nitride complexes often suffer from thermodynamic overstabilization hampering functionalization. Furthermore, oxidative nitrogen atom transfer of N2 derived nitrides remains unknown. We here report a ReIV pincer platform that mediates N2 splitting upon chemical reduction or electrolysis with unprecedented yield. The N2 derived ReV nitrides undergo facile nitrogen atom transfer to nitric oxide, giving nitrous oxide nearly quantitatively. Experimental and computational results indicate that outer-sphere ReN/NO radical coupling is facilitated by the activation of the nitride via initial coordination of NO.

Haber–Bosch related hydrocarbon steam reforming accounts for a substantial share of the global energy consumption and anthropogenic CO2 footprint. N2 fixation strategies for high‐ and mid‐valent nitrogenous products without initial, energy demanding over‐reduction are therefore highly attractive, but largely unexplored. For example, redox economic biological and plasma processes towards nitrous oxide have been extensively examined, but industrial production still relies on controlled NH4NO3 decomposition. A thermochemically viable alternative could be the comproportionation of N2 with nitric oxide [Eq. 1], but the demanding cleavage of the strong N2 triple bond requires efficient catalysis.

In recent years, thermally, electro‐, and photochemically driven N2 splitting into nitride complexes has been advanced as an entry to N2 fixation beyond ammonia. Synthetic strategies typically relied on nucleophilic nitride reactivity and, so far, enabled catalytic formation of silyl‐ and borylamines.[ , ] However, direct oxidative functionalization of nitrides that originate from N2 is unknown. Cummins and co‐workers demonstrated facile N2O splitting into nitride and NO complexes at ambient temperatures (Scheme 1). Later, the groups of Mayer and Caulton reported the reverse reaction, i.e., the release of N2O by coupling of NO with terminal OsVI and RuIV nitride complexes, respectively. Motivated by this precedence, we here disclose the generation of N2O from dinitrogen via electrochemical N2 splitting and subsequent nitride transfer to NO at ambient conditions (Scheme 1).

Scheme 1

Top: Nitrous oxide activation and generation reactions via triple bond metathesis reported in the literature. Bottom: Schematic reaction sequence for N2O generation from N2 and NO described here.

Our group has previously explored reductive N2 splitting by rhenium complexes with aliphatic pincer ligands into nucleophilic, terminal ReV nitrides. As the N‐terminus defines the electrophilic site of nitric oxide, a related platform was chosen to evaluate nitride/NO coupling. To avoid undesired pincer oxidation, we resorted to the robust diphenylamido ligand N(C6H3‐4‐CH3‐P Pr2)2 − (PNP−). Ison and co‐workers recently reported the synthesis of the ReV nitride [Re(N)Cl(PNP)] (1), yet not by N2 activation.

Heating [ReCl3(PPh3)2(MeCN)] and HPNP in benzene gives the ReIV amide complex [ReCl3(PNP)] (2) in 86 % isolated yield. Amide formation is supported by IR spectroscopy, while metal oxidation is attributed to H2 evolution, as confirmed by gas chromatography. N−H bond weakening in related pincer ligands by metal coordination is well established. The solution magnetic moment of 2 at room temperature (μ eff=1.63±0.1 μB) supports low‐spin ReIV. While an EPR signal could not be resolved at temperatures down to 5 K, the well resolved 1H NMR spectrum is in agreement with C 2 symmetry that arises from an axially chiral pincer ligand. Single‐crystal X‐ray diffraction confirmed a planar amide group with a short N−Re bond (1.994(3) Å) that indicates π‐bonding as origin for the low‐spin configuration.

Cyclic voltammetry (CV) under Ar shows two reductions of 2 as a reversible (E 1 0=−0.95 V; all potentials vs. Fc+/0) and an irreversible feature (E 2,pc=−2.40 V; v=100 mV s−1), respectively (Scheme 2). The low ReIII/II potential is required for nitrogen activation as shown by chemical reduction with Na/Hg (2.5 equiv.) under N2 (Scheme 2), which led to 1 in 60 % isolated yield. Reductive N2 splitting was confirmed by isotopic labelling using 15N NMR (δ=363 ppm) and IR spectroscopy (ν(Re≡14N)=1080 cm−1; ν(Re≡15N)=1047 cm−1). The molecular structure of 1 in the crystal (Scheme 2) showed distorted square‐pyramidal coordination (τ 5=0.36) with a Re≡N bond length (1.645(6) Å) that is in the typical range of ReV nitride complexes.[ , , ] 1 was alternatively synthesized by in situ reduction of 2 with CoCp2 and subsequent reaction with Me3SiN3.

Scheme 2

Top left: CVs of 2 and 3 under Ar in THF. Top right: Molecular structure of 1 from single‐crystal X‐ray diffraction; ellipsoids are set at 50 % probability; hydrogen atoms are omitted for clarity. Selected bond lengths [Å] and angles [°]: Re1–N2 1.645(6), Re1–N1 2.071(4), Re1–Cl1 2.3987(17), Re1–P1 2.4152(11), Re1–P1#1 2.4121(11); N1‐Re1‐Cl1 132.09(5), P1‐Re1‐P1 153.73(3), N2‐Re1‐N1 116.7(2). Bottom: Reductive N2 splitting.

The reversibility of E 1 suggests that the unfavorably negative ReIII/II couple E 2 results from slow Cl− loss after reduction. We therefore moved to iodide as a better leaving group. [ReI3(PNP)] (3) is easily prepared from 2 by halide metathesis. As hoped for, the first reduction of 3 under Ar (E 1,pc=−0.79 V; v=100 mV s−1) becomes irreversible at scan rates up to 10 V s−1 (Scheme 2), and a particularly large anodic shift with respect to 2 is obtained for the irreversible ReIII/II wave (E 2,pc=−1.69 V; v=100 mV s−1). In consequence, the N2 splitting product [Re(N)I(PNP)] (4; Scheme 2) can be obtained with the mild reductant CoCp2 (82 %) or by controlled potential electrolysis (CPE) in 1,2‐difluorobenzene (E appl.=−1.67 V), with unprecedented near quantitative yield (94 %) and Faradaic efficiency (98 %).[ , , ]

Benzene solutions of 1 or 4 slowly react with nitric oxide (1 atm) at room temperature over the course of minutes (Scheme 3 A). Nitrous oxide formation was confirmed and quantified by IR spectroscopy (2219 cm−1), which is nearly quantitative in case of 1 (93.8±4.3 %). With 15N nitride labelled N‐1, 15N14NO was obtained as the only isotopomer and isotopologue by 15N NMR (Scheme 3 C), supporting N2O release via selective nitride coupling with NO. Accordingly, the paramagnetic {Re(NO)2}7 dinitrosyl complex [ReCl(NO)2(PNP)] (5) could be isolated in 88 % yield from the reaction of 1 with nitric oxide. The magnetic susceptibility (μ eff=1.69±0.1 μB) and X‐band EPR signal (g iso=2.01) of 5 in CH2Cl2 at room temperature are in agreement with a low‐spin ground state. The characteristic EPR line shape results from comparatively small, partially resolved 185/187Re hyperfine interaction (HFI, A iso=140 MHz), which is indicative of significant spin delocalization to the nitrosyl ligands. This was confirmed by DFT calculations, which locate over 99 % of the excess spin density on the nitrosyl ligands with satisfyingly reproduced g‐value (g iso DFT=2.03) and 185/187Re HFI (A iso DFT=191 MHz).

Scheme 3

A) Reaction of 1 with NO in benzene and CH2Cl2/CCl4. B) Molecular structure of 5 with ellipsoids set at 50 % probability; hydrogen atoms are omitted for clarity. Selected bond lengths [Å] and angles [°]: Re1–N1: 2.096(2), Re1–N2: 1.835(5), Re1–N3: 1.811(7), Re1–Cl1: 2.3635(18), Re1–Cl2: 2.378(4), Re1–P1: 2.4411(7), O1–N2: 1.215(6), O2–N3: 1.153(12); Re1‐N3‐O2: 180.0, P1‐Re1‐P1#1: 155.50(3), O1‐N2‐Re1: 177.0(5), N1‐Re1‐N3: 180.0. C) 15N{1H} NMR spectrum of 15NNO in CD2Cl2 from the reaction of N‐1 with NO. D) Normalized transmittance ATR‐IR spectra of 5 and NO‐5. E) Molecular structure of 7 with ellipsoids set at 50 % probability; hydrogen atoms are omitted for clarity. Selected bond lengths [Å] and angles [°]: Re1–N2: 1.810(6), N2–O1: 1.178(6), Re1–N1: 2.050(3), Re1–Cl1: 2.358(2), Re1–Cl2: 2.3943(9); Re1‐N2‐O1: 178.2(5), N1‐Re1‐N2: 91.17(19), P1‐Re1‐P1#1: 157.17(3), N1‐Re1‐Cl2: 180.0.

Single‐crystal X‐ray diffraction confirmed the chemical composition of 5 (Scheme 3 B). While crystallographic disorder in space group P2/n prevented an unequivocal determination of the configuration, this was obtained by refinement in space group Pn and confirmed by vibrational spectroscopy. However, the disorder does not allow for detailed discussion of NO bonding. Three intense IR bands are observed in the characteristic region of NO stretching vibrations, both in solution and in the solid state (ν=1694, 1634, 1597 cm−1; Scheme 3 D). Assignment to NO stretching modes was assisted by reaction of 1 with 15NO, which leads to a bathochromic shift of only two bands in NO‐5 (ν15NO=1660 and 1599 cm−1). The observation of two strong NO bands supports a cis‐dinitrosyl configuration, as only one strong IR‐active NO stretching vibration is expected for the trans‐isomer. DFT calculations support this interpretation, placing the cis‐isomer of 5 energetically below the trans‐configuration by Δ0 G=18 kcal mol−1. Monitoring the reaction of 1 with NO by NMR (−20 °C) and IR (−15 °C) spectroscopies showed slow, clean conversion to 5 over the course of some hours. Besides the dinitrosyl product, small amounts of diamagnetic {Re(NO)2}8 complex [Re(NO)2(PNP)] (6) could be identified as side product by comparison with an authentic sample.

The mechanism was further probed by reacting 1 with NO in CH2Cl2/CCl4 (4:1) to trap possible low‐valent, coordinatively unsaturated rhenium intermediates via chlorine atom transfer. From this experiment, the mononitrosyl complex [ReCl2(NO)(PNP)] (7) was obtained in 84 % yield (Scheme 3 A). Concomitant formation of 15N14NO from N‐1 was verified by 15N{1H} NMR spectroscopy. X‐ray crystallography confirmed the cis‐dichloride configuration of {ReNO}5 complex 7 (Scheme 3 E) with a linearly coordinated nitrosyl ligand (Re‐N‐O: 178.2(5)°). SQUID magnetometric and X‐band EPR spectroscopic characterization support an S=1/2 ground state. However, the considerably larger 185/187Re HFI (A iso=782 MHz) as compared to 5, suggests negligible spin delocalization to the nitrosyl ligand. This interpretation is supported by DFT computations, which place the SOMO of 7 orthogonal to the Re‐NO axis.

The formation of 7 in the presence of CCl4 suggests that the five‐coordinate species [ReCl(NO)(PNP)] might be an intermediate in the reaction. We turned to DFT computations to evaluate electronic prerequisites for the formation of this species from 1. Direct attack of NO at the nitride atom in 1 gives a highly endergonic intermediate (Δ0 G=+25 kcal mol−1) with a prohibitively high reaction barrier (Δ≠ G=30 kcal mol−1, Scheme S2). Instead, initial coordination of NO to the metal ion leads to nitrosyl nitride complex D8 in an essentially thermoneutral and kinetically undemanding step (Figure 1 A). The computed Re‐N‐O bond angle of D8 (160.7°) is a structural manifestation of NO‐centered radical character. NO coordination induces significant spin polarization (0.26α spin at Nnitride and 0.24β spin at Re). Furthermore, natural bond orbital (NBO) analysis reveals concomitant weakening of the Re≡N interaction as expressed by the Wiberg bond index (WBI), which reduces from 2.55 (1) to 2.31 (D8, Figure 1 B).

Figure 1

A) Lowest energy reaction pathway computed for the reaction of 1 with NO; alternative but less favorable pathways are presented in the Electronic Supporting Information (Δ0 G values in kcal mol−1 and selected bond lengths in Å; only truncated PNP ligands shown for clarity; S=singlet, D=doublet, TS=transition state). B) Selected NPA atomic charges and Wiberg bond indices (in parentheses) for 1, D8, and S9. C) Spin density plot (isosurface at ±0.005 a0 −3) with natural spin populations for selected atoms and magnetic corresponding orbitals of TS2 with spatial overlap S (isosurfaces at ±0.05 a0 −2/3, hydrogen atoms omitted for clarity).

Subsequently, a second NO molecule can directly add to the nitride ligand via TS2 with significantly reduced kinetic barrier (Δ≠ G=23 kcal mol−1). Singlet TS2 exists only within the broken spin symmetry framework (⟨S2⟩=0.79) and affords the initial formation of a covalent σ bond by spin pairing of the singly occupied π* orbital of NO (β orbital in Figure 1 C) and the in‐plane non‐bonding π* orbital (α orbital, WBI N‐N in TS2: 0.53). Hence, mixing of the dxy and Re≡N π* orbitals with the π* orbitals of the initially coordinating NO results in a singly occupied molecular orbital (SOMO) with significant nitride contribution (Figure 2), rendering N‐N coupling via radical recombination more efficient than via a nucleophilic pathway.

Figure 2

A) Qualitative MO diagram illustrating the mixing of selected MOs of 1 with the π* orbitals of NO. B) SOMO of D8 (isosurfaces at ±0.05 a0 −2/3). Hydrogen atoms omitted for clarity.

NBO analysis of the resulting closed‐shell intermediate S9 reveals formation of an N≡NO triple bond and, in turn, substantial electron transfer to the metal ion, which is stabilized by increased Re→NO π backbonding within the {ReNO}6 core. Coordination of the first NO ligand thus mediates electronic reorganization associated with Re≡N to N≡NO bond transformation. As a consequence, nitrous oxide is only weakly bonded to the Re ion. Release of N2O and subsequent NO addition gives cis‐5 in two barrierless steps (Figure 1). The overall reaction from 1 to cis‐5 is strongly exergonic by −51 kcal mol−1. The lowest computed path in Figure 1 emphasizes the role of initial NO binding for the activation of 1. However, further mechanistic analysis will be required to account, e.g., for the formation of 6.

In conclusion, rhenium mediated nitrogen atom transfer from dinitrogen to nitric oxide was described. Reductive N2 activation with ReIV trihalides [ReX3(PNP)] (X=Cl, I) gives the respective ReV nitride complexes. More favorable iodide dissociation kinetics enable electrochemically driven N2 splitting at relatively mild potential with unprecedented yields. The nitride products react with NO at ambient conditions to release near quantitative amounts of N2O. Isotopic labelling confirmed nitride transfer over nitride oxygenation as the preferred pathway. Computational analysis indicates that direct N≡NO triple bond formation by NO attack at nitride 1 is not kinetically facile. Instead, initial coordination of NO spin polarizes the strong Re≡N bond and makes it more susceptible to radical N‐N coupling. Importantly, our results offer a new strategy towards oxidative nitrogen fixation at ambient conditions via N2 splitting into nitride complexes, which are otherwise notorious for thermodynamic overstabilization.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.