Mutual functionalization of dinitrogen and methane mediated by heteronuclear metal cluster anions CoTaC₂⁻

Abstract

The direct coupling of dinitrogen (N₂) and methane (CH₄) to construct the N–C bond is a fascinating but challenging approach for the energy-saving synthesis of N-containing organic compounds. Herein we identified a likely reaction pathway for N–C coupling from N₂ and CH₄ mediated by heteronuclear metal cluster anions CoTaC₂⁻, which starts with the dissociative adsorption of N₂ on CoTaC₂⁻ to generate a Ta^δ+–N_t^δ− (terminal-nitrogen) Lewis acid–base pair (LABP), followed by the further activation of CH₄ by CoTaC₂N₂⁻ to construct the N–C bond. The N≡N cleavage by CoTaC₂⁻ affording two N atoms with strong charge buffering ability plays a key part, which facilitates the H₃C–H cleavage via the LABP mechanism and the N–C formation via a CH₃ migration mechanism. A novel N_t triggering strategy to couple N₂ and CH₄ molecules using metal clusters was accordingly proposed, which provides a new idea for the direct synthesis of N-containing compounds.

---

Introduction

The construction of N–C bonds directly from N₂ offers the potential to reduce the enormous fossil-fuel consumption for the current synthesis of N-containing organic compounds through NH₃, which is mainly produced via the energy-intensive Haber–Bosch process (N₂ + H₂ → NH₃).¹ Considerable research efforts have been devoted to this subject since the initial observation of N–C bond formation in the reaction of N₂ complexes with organic halides.² A long-sought goal for this chemistry has been the direct coupling of N₂ with abundant carbon-based molecules that are viable in industry. However, activation of these stable molecules remains a challenge, and only CO,^3–5 CO₂,^6–8 and unsaturated hydrocarbons^9,10 have been reported to form N–C bonds with N₂ induced by appropriate reactive species. CH₄, the major component of widely available natural gas, is closely linked to industrial NH₃ synthesis in which H₂ is produced through the energy-intensive CH₄-reforming process. Therefore, the direct coupling of N₂ and CH₄ is of great importance for economic and environmental reasons. However, the appropriate species that can mediate the coupling of inert N₂ and CH₄ molecules to form an N–C bond has not been revealed.

Isolated gas-phase atomic clusters are important model systems for the precise design of desired active sites and the fundamental understanding of bond activation and formation processes.¹¹ The activation and individual functionalization of N₂ or CH₄,^12–18 and the formation of N–C bonds^19–25 are also extensively studied issues in the gas-phase field. Typical examples of N–C bond formation from N₂ or CH₄ in gas-phase studies can be classified as follows: (i) reactions of metal carbide clusters with N₂^22–25 or metal nitride clusters with CH₄,^26,27 and (ii) coupling reactions of CH₄ with NH₃^28,29 or N₂ with CO₂.⁶ In addition to reactions involving N₂ or CH₄, the N–C bond formation was also observed in a few other reaction systems.^30–32 Compared with these N–C formation processes, the direct coupling of N₂ and CH₄ is rather challenging and of particular interest.

Herein, we report the first example of N–C bond formation from the mutual functionalization of N₂ and CH₄ mediated by heteronuclear metal cluster anions CoTaC₂⁻ under thermal collision conditions. A terminal-nitrogen (N_t) triggering strategy for the coupling of N₂ and CH₄ was accordingly proposed (Scheme 1), which starts with the dissociative adsorption of N₂ on metal-based substrates to generate a M^δ+–N_t^δ− Lewis acid–base pair (LABP), followed by the H₃C–H cleavage of CH₄via the LABP mechanism and the N–C formation via a CH₃ migration mechanism. Considering that the activation of CH₄ by some metal nitrides follows the LABP mechanism,^26,27 we infer that the proposed strategy can be quite general, which is confirmed by the study on the FeTaC₂⁻/N₂/CH₄ reaction system. The important roles of producing substrate-N_t complexes in the coupling of N₂ and CH₄, as well as the strength of M–N_t bonds in the activity of the N₂ complexes were discussed.

Scheme 1 The proposed terminal-nitrogen (N_t) triggering strategy for the coupling of N₂ and CH₄. The key to this strategy is the predissociation of N₂ on an appropriate substrate to generate a N_t and a bridging-N (N_b).

Results

Cluster reactivity

The spectra in Fig. 1 have been obtained by using an online time-of-flight (TOF) mass spectrometer and show the results for the reactions of CoTaC₂⁻ (m/z = 264) with N₂ and CoTaC₂N₂⁻ (m/z = 292) with CH₄. Reference spectra with inert He as the reactant gas were also recorded (Fig. 1a and d). The CoTaC₂⁻ ions were generated by laser ablation of a mixed Co–Ta disk target (molar ratio Co/Ta = 2 : 1) in the presence of 0.05% CD₄ diluted with He carrier gas, and then mass-selected and thermalized to room temperature to react with N₂ in a linear ion trap (LIT) reactor. As shown in Fig. 1b, on pulsing 0.56 Pa N₂ into the LIT reactor, a strong product peak assigned as CoTaC₂N₂⁻ appeared, suggesting the following reaction channel:

CoTaC₂⁻ + N₂ → CoTaC₂N₂⁻ (1)

Fig. 1 Time-of-flight mass spectra of the reactions of CoTaC₂⁻ with (a) He, (b) N₂, and (c) ¹⁵N₂, and CoTaC₂N₂⁻ with (d) He, (e) CH₄, and (f) CD₄. The reaction times are 4.8 ms for (b and c), 1.8 ms for (e), and 1.9 ms for (f). Peaks marked with asterisks are due to water impurities in the gas handling system.

The isotopic labeling experiment using ¹⁵N₂ as the reactant gas (Fig. 1c) confirmed the above reaction channel. We should mention in passing that, a very tiny metal atom ejection channel (less than 1%) producing TaC₂N₂⁻ was also observed (Fig. S1a–c†). Based on a least-squares fitting procedure (Fig. S2a†), the rate constant k₁(CoTaC₂⁻/N₂) of the pseudo first-order reaction between CoTaC₂⁻ and N₂ was estimated to be (8.0 ± 1.6) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹, corresponding to a reaction efficiency (Φ = k₁/k_c) of about 0.1% relative to the theoretical collision rate (k_c).³³

To study the reaction of the N₂ association product CoTaC₂N₂⁻ with CH₄, a newly-developed double ion trap apparatus (Scheme S1 in the ESI†) was used to first generate the CoTaC₂N₂⁻ from the reaction of CoTaC₂⁻ with N₂ in the first LIT reactor and then mass select CoTaC₂N₂⁻ to interact with CH₄ in the second LIT reactor. Upon the interaction of CoTaC₂N₂⁻ with CH₄ (Fig. 1e), three product peaks assigned as CoTaC₂N₂CH₄⁻ (m/z = 308), CoTaC₂N₂CH₂⁻ (m/z = 306), and CoTaC₂N₂C⁻ (m/z = 304) were observed, suggesting the following reaction channels:

CoTaC₂N₂⁻ + CH₄ → CoTaC₂N₂CH₄⁻ 49% (2)

CoTaC₂N₂⁻ + CH₄ → CoTaC₂N₂CH₂⁻ + H₂ 41% (3)

CoTaC₂N₂⁻ + CH₄ → CoTaC₂N₂C⁻ + 2H₂ 10% (4)

The rate constant k₁(CoTaC₂N₂⁻/CH₄) for the reaction of CoTaC₂N₂⁻ with CH₄ was estimated to be (1.4 ± 0.3) × 10⁻¹¹ cm³ molecule⁻¹ s⁻¹, corresponding to a Φ of about 1.4%. Noticeably, the reaction channel of ejecting two D₂ molecules was negligible when using isotope-labeled CD₄ in place of CH₄ (Fig. 1f). The branching ratio of reaction channel (2) to reaction channel (3) changed to 80 : 20 in the CD₄ experiment, and the intermolecular kinetic isotopic effect (KIE) calculated using k₁(CoTaC₂N₂⁻/CH₄)/k₁(CoTaC₂N₂⁻/CD₄) was estimated to be 1.3 (Fig. S2b and c†). The loss of H₂ (D₂) in the reaction of CoTaC₂N₂⁻ with CH₄ (CD₄) suggests that activation of C–H bonds must have occurred. To determine the mechanisms of CoTaC₂⁻ + N₂ and CoTaC₂N₂⁻ + CH₄ reactions, structural characterization of the reactant cluster ions should be performed.

Structural characterization

Photoelectron imaging spectroscopy (PEIS)³⁴ combined with quantum chemistry calculations was employed to characterize the structures of CoTaC₂⁻ and CoTaC₂N₂⁻. The structures were optimized at the density functional theory (DFT) level,³⁵ and their relative energies were then refined by high-level RCCSD(T) (partially spin-adapted open-shell coupled cluster method with single, double, and perturbative triple excitations) and DMRG-SC-NEVPT2 (density matrix renormalization group strongly-contract n-electron valence perturbation theory) methods.^36,37 The experimental spectrum of CoTaC₂⁻ recorded with 670 nm photons at 10 K reveals a sharp peak centered at 1.63 eV, followed by three discernible peaks with electron binding energies of 1.70, 1.73, and 1.76 eV (Fig. 2a, top). The calculated lowest-lying isomer (²IS1) of CoTaC₂⁻ features a Co–Ta double bond with a Wiberg bond index (WBI) of 2.04 and a C₂ ligand. The presence of two d–d bonding orbitals between Co and Ta atoms provides further evidence for the Co–Ta double bond (Fig. S12†). The Franck–Condon (FC)-simulated spectrum of the ²A → ³A vibrational transition for ²IS1 can reasonably reproduce the experimental spectrum from 1.70 to 1.76 eV (Fig. 2a, middle),³⁸ and the calculated adiabatic electron detachment energy (ADE) of ²IS1 is close to the experimental value (1.83 eV vs. 1.70 eV), suggesting that ²IS1 is the most probable species of CoTaC₂⁻ generated in the experiment. The first spectral peak centered at 1.63 eV might come from the minor population of the isomer ⁴IS2 with a relative energy of 0.59 eV higher than ²IS1, considering that its calculated ADE (1.59 eV) and simulated profile of the ⁴A → ³A transition match well with the first spectral peak (Fig. 2a, bottom). A more detailed discussion of the structural assignment of CoTaC₂⁻ is provided in the ESI (Fig. S3†).

Fig. 2 (a) Experimental and Franck–Condon (FC)-simulated photoelectron spectra of CoTaC₂⁻. (b) RCCSD(T)-calculated potential energy profile for the reaction of CoTaC₂⁻ (²IS1) with N₂. (c) Experimental and density of states (DOS)-simulated photoelectron spectra of CoTaC₂N₂⁻. The 670 nm (1.85 eV) and 410 nm (3.02 eV) photons were used for CoTaC₂⁻ and CoTaC₂N₂⁻, respectively. The relative energies, ADEs and VDEs (in brackets) are given in eV. The bond lengths are in pm. The superscripts indicate spin multiplicities. The simulated spectra of ²IS1, ²I1, ²I4 and ⁴IS2 are red shifted by 0.13, 0.05, 0.12 eV and blue shifted by 0.04 eV, respectively.

Reaction mechanism for the CoTaC₂⁻/N₂ couple

The reaction pathway of CoTaC₂⁻ (²IS1) with N₂ calculated at the RCCSD(T) level is shown in Fig. 2b. The N₂ molecule first approaches the Ta atom through a side-on (η²) mode to form the encounter complex ²I1 (−0.47 eV). Further binding of N₂ to Co is impeded by a high energy barrier with the doublet spin state (²TS1/0.10 eV); however, this process can be accomplished through transiting to the quartet spin state (⁴TS1/−0.13 eV) and then a more stable intermediate ²I2 (−0.84 eV) with a side-on-end-on (η²:η¹) bounded N₂ unit is formed. By surmounting ²TS2, N₂ is coordinated to the Co–Ta center in a distorted-η²:η¹ mode (²I3/−0.64 eV) with an N–N bond length of 135 pm. The N–N bond is disrupted entirely after overcoming a slight energy barrier of 0.11 eV (²TS3/−0.53 eV), yielding one terminal-N (N_t) and one bridging-N (N_b) in ²I4 (−1.39 eV). The subsequent N–C_cluster coupling would encounter a highly positive energy barrier (²TS17/0.10 eV, Fig. S4†) that has little chance of being surmounted under thermal collision conditions, and thus the ²I4 would be stabilized as the adsorption product CoTaC₂N₂⁻ through collisions with bath gas He. Note that the very minor reaction channel of generating TaC₂N₂⁻ in the reactivity experiment might come from the reaction of the high-lying isomer ⁴IS2 with N₂, in which two steps of N–C_cluster coupling take place before ejecting a neutral Co atom (Fig. S5†).

The PEIS characterization of CoTaC₂N₂⁻ at different reaction temperatures and photon energies supported the assignment of ²I4 as the adsorption product and also proved the reliability of our RCCSD(T) calculations. As shown in Fig. 2c, only one spectral peak (B) was observed in the experimental spectrum recorded with 410 nm photons at 250 K (note that the residual water in the room-temperature ion trap will deplete the CoTaC₂⁻ signal under the condition of a long trapping time of 80 ms, so the room-temperature reaction was not performed in the PEIS experiment); however, two spectral peaks (A and B) could be observed when the reaction temperature decreased to 200 K and 170 K. This implies that an additional intermediate was stabilized as the adsorption product CoTaC₂N₂⁻ at lower temperatures and accounted for the appearance of peak A. Considering that the process of ²I1 → ⁴TS1 → ²I2 with the highest energy barrier and a spin crossing³⁹ is the rate-limiting step (Fig. 2b), ²I1 is the most probable species that could be stabilized at lower temperatures. When the temperature is decreased to a certain value, all of the I1 might be stabilized and cannot transform into I4. However, this situation is not the focus of this work because we aim to study the reactions of CoTaC₂⁻/N₂ and CoTaC₂N₂⁻/CH₄ couples at room temperature. The simulated spectra of CoTaC₂N₂⁻ isomers based on density of states (DOS) simulations⁴⁰ confirmed the contribution of ²I1 to peak A and indicated that peak B in the 250 K spectrum was contributed by ²I4 (Fig. 2c). Moreover, the employment of 365 nm photons for the PEIS characterization of CoTaC₂N₂⁻ provides further evidence for the agreement of the DOS-simulated spectrum of ²I4 with the experimental one (Fig. S6b and d†). Other CoTaC₂N₂⁻ isomers with the N–N or N–C bonds should be excluded due to their mismatched VDE values and spectral patterns with the experimental spectrum (Fig. S6e–k†) and their theoretically predicted inertness toward CH₄ (Fig. S7†). Therefore, a N_t-containing complex was successfully prepared through the dissociative adsorption of N₂ on CoTaC₂⁻ at room temperature.

Reaction mechanism for the CoTaC₂N₂⁻/CH₄ couple

As shown in Fig. 3, the CoTaC₂N₂⁻ (²I4) interacts with CH₄ by first anchoring it on the Ta atom to form ²I5 with a binding energy of 0.44 eV. The first H₃C–H bond cleavage preferably proceeds via the cooperative mechanism of a LABP composed of Ta (natural charge: 1.31e) and N_t (natural charge: −0.79e) atoms, generating a N_t–H bond and a Ta–CH₃ moiety in ²I6 (−2.52 eV). After that, the CH₃ group tends to migrate from Ta to N_b to liberate the Ta site (²I6 → ²TS5 → ²I7). On the basis of Rice–Ramsperger–Kassel–Marcus (RRKM) theory,⁴¹ the conversion rate of ²I6 → ²TS5 is estimated to be 6.9 × 10⁴ s⁻¹, which is one order of magnitude smaller than the collision rate (6.7 × 10⁵ s⁻¹) that a cluster experiences with the bath gas He in the LIT reactor. This suggests that only a small part of ²I6 could overcome ²TS5 to form ²I7, while most of the ²I6 would be stabilized as the adsorption product CoTaC₂N₂CH₄⁻ through collisions with bath gas. If the barrier height of ²TS5 is decreased by 0.1 eV, which could be the uncertainty of RCCSD(T) calculations,⁴² the rate of ²I6 → ²TS5 is increased to 3.3 × 10⁵ s⁻¹, which is of the same order of magnitude as the collision rate and can lead to the stabilization of about half of the ²I6 (P1) as CoTaC₂N₂CH₄⁻. This agrees well with the ratio of CoTaC₂N₂CH₄⁻ to CoTaC₂N₂CH_2,0⁻ (49 : 51) in the reactivity experiment (Fig. 1).

Fig. 3 Potential energy profile for the reaction of CoTaC₂N₂⁻ (²I4) with CH₄. The structures are optimized at the DFT level. The zero-point vibration corrected energies in eV relative to the separated reactants are calculated at the RCCSD(T) level. The C atoms from CoTaC₂⁻ and CH₄ are shown in different colours.

After the formation of ²I7, the reaction proceeds with a H atom of the CH₃ group transferred to the Ta atom to form ²I8. The resultant CH₂ unit then forms a chemical bond with the Co atom, enabling the consecutive activation of the remaining two C–H bonds via the transfer of H atoms to Co (²I9 → ²I10 → ²I11). Along with two steps of structural rearrangements (²I11 → ²I12 → ²I13), two H atoms on Co make a H₂ unit to generate the lowest-lying isomer (²IS3) of CoTaC₂N₂CH₂⁻ concomitant with loss of H₂. The RRKM-theory calculated rates of traversing ²TS6–²TS11 and H₂ desorption from ²I13 are at least two orders of magnitude larger than the collision rate (ESI Table 3†), indicating the impossible stabilization of intermediates ²I7–²I13 and the facile formation of ²IS3. The generation of P2 (²IS3 + H₂) is highly exothermic (−2.71 eV), so ²IS3 has enough internal energy to undergo further transformation to evaporate the second H₂ molecule and yield the lowest-lying isomer (²IS4) of CoTaC₂N₂C⁻ (Fig. S8†). The most favorable pathway to generate P3 (²IS4 + H₂) from ²IS3 involves a Co–NH bond-forming process (²I14 → ²I15) and a Co–CN bond-breaking process (²I15 → ²I16), followed by the transfer of a H atom from N to Ta to generate ²I17, from which a H₂ molecule can be evaporated. The above high-level RCCSD(T) calculations indicate that the N–C_CH4 bond is formed in the dehydrogenation product CoTaC₂N₂CH_2,0⁻. Note that it is very difficult to perform the PEIS characterization of CoTaC₂N₂CH_4,2,0⁻ with the current apparatus due to the very weak signals of these species and the mass overlap with other species such as CoTaC₄H₂O⁻ and CoTaC₂ON₂⁻ (Fig. S9†). Quantum chemistry calculations confirmed that the formation of the N–C_CH4 bond is necessary for the experimentally observed reactions and the migration of the CH₃ group from Ta to N_b is the most favorable pathway.

Discussion

As shown in Fig. 4, the coupling reaction of N₂ with CH₄ mediated by CoTaC₂⁻ starts with cleavage of the N≡N bond by CoTaC₂⁻ to generate CoTaC₂N₂⁻ with a Ta^δ+–N_t^δ− LABP, followed by further activation of CH₄ by CoTaC₂N₂⁻, in which the first C–H bond is cleaved via a Ta^δ+–N_t^δ− LABP mechanism and the N–C coupling is subsequently achieved via a CH₃ migration mechanism. The order of activation of the two molecules is crucial to this coupling reaction, as reflected by the low reactivity of CoTaC₂⁻ toward CH₄ to produce CoTaC₃H₂⁻ (Fig. S1d and e,†k₁ = 1.2 × 10⁻¹⁴ cm³ molecule⁻¹ s⁻¹) and the theoretically predicted impossibility of functionalizing N₂ by CoTaC₃H₂⁻ (Fig. S10c†). Comparative studies on CoTaC₂⁻/CH₄ and CoTaC₂N₂⁻/CH₄ reaction couples indicate that the approach of CH₄ to CoTaC₂N₂⁻ is overall barrierless, while the approach of CH₄ to CoTaC₂⁻ encounters a positive energy barrier of 0.05 eV (Fig. S10a†). Moreover, cleavage of the H₃C–H bond by CoTaC₂⁻ follows the oxidative addition mechanism, which is kinetically less favorable than the LABP mechanism in the CoTaC₂N₂⁻/CH₄ couple (Fig. S10b†). Therefore, the design of first activating N₂ not only constructs a Ta^δ+–N_t^δ− LABP to facilitate the initial activation of CH₄, but also generates a sufficiently reactive N_b atom that can accept the migrating CH₃ group to form the N–C bond.

Fig. 4 Key events and mechanisms for the coupling reaction of N₂ with CH₄ mediated by CoTaC₂⁻ cluster anions.

Natural charge analysis of the CoTaC₂N₂⁻/CH₄ reaction system reveals that the two N atoms in CoTaC₂N₂⁻ exhibit strong charge buffering ability throughout the activation of CH₄: they store a large number of negative charges (ΔQ = −0.47e) during the processes of H₃C–H cleavage (N_t–H formation) and N_b–C formation, while releasing all of the stored negative charges (ΔQ = 0.52e) during the activation of the remaining three C–H bonds (Fig. S11†). Such charge buffering behavior largely reduces the kinetic barriers of rate-limiting steps and thus drives the N–C bond formation. Noticeably, the N–C bond formation from N₂ outlined here differs fundamentally from that in previous gas-phase studies. Generally, the carbon ligands in transition metal carbide clusters, such as FeTaC₂⁻, FeV₂C₂⁻, V₃C₄⁻, and Ta₂C₄⁻, were employed to construct N–C_cluster bonds after the N≡N bond was completely cleaved by the metal center.^22–25 This study provides a likely pathway for N–C bond formation from the product of N₂ cleavage through migrating a CH₄-derived CH₃ group to a N₂-derived N atom, which opens a new window for the reactions of N₂-derived metal nitrides.

Finally, we infer that N₂-derived metal nitrides with a M–N_t LABP might be promising species for further activation of CH₄. As expected, the previously reported FeTaC₂N₂⁻ with a similar Ta^δ+–N_t^δ− LABP (natural charge: 1.33e/−0.89e) is demonstrated to react with CH₄ to produce FeTaC₂N₂CH₄⁻ and FeTaC₂N₂CH₂⁻ (ratio = 70 : 30) by using a newly-designed ship-lock-type reactor (Fig. S13†).⁴³ More interestingly, the reaction rate of the FeTaC₂N₂⁻/CH₄ couple (k₁ = 2.8 × 10⁻¹⁵ cm³ molecule⁻¹ s⁻¹) is about four orders of magnitude smaller than that of CoTaC₂N₂⁻/CH₄ couple, indicating a large difference in the initial activation of CH₄ by the Ta^δ+–N_t^δ− LABP. A key difference between the two systems is that the degree of N₂ reduction by CoTaC₂⁻ is properly reduced compared to FeTaC₂⁻, as reflected by fewer negative charges on N atoms and smaller electron occupancy on N_2p orbitals in CoTaC₂N₂⁻ than those in FeTaC₂N₂⁻ (Fig. S14†). This leads to a slightly weaker Ta–N_t bond in CoTaC₂N₂⁻ (WBI: 2.4) than that in FeTaC₂N₂⁻ (WBI: 2.6), which has more possibility to buffer the charge variation during the initial H₃C–H cleavage process. This agrees with the fact that it is difficult for N₂-derived nitrides with strong M–N bonds to activate other molecules. Further research on the optimal design of highly reactive M^δ+–N_t^δ− LABP during the activation of N₂ to trigger efficient N–C coupling in the further activation of CH₄ is in progress.

Conclusions

A possible N–C bond formation from the mutual functionalization of N₂ and CH₄ mediated by heteronuclear metal cluster anions CoTaC₂⁻ was studied by using mass spectrometry, photoelectron spectroscopy and a modelled reaction pathway. As verified by employing mass spectrometry, photoelectron imaging spectroscopy, and quantum chemistry calculations, the coupling reaction of N₂ with CH₄ starts with the dissociative adsorption of N₂ on CoTaC₂⁻ to generate a Ta^δ+–N_t^δ− Lewis acid–base pair, which then triggers the H₃C–H cleavage (N_t–H formation) and N–C_CH4 formation in the reaction of CoTaC₂N₂⁻ with CH₄. Cleaving the N≡N bond to generate two N atoms with large charge buffer capacity underlies the ability of CoTaC₂⁻ to mediate the coupling of N₂ and CH₄. Based on this finding, a universal N_t triggering strategy was proposed to couple inert N₂ and CH₄ molecules, which may inspire the rational design of catalysts to produce N-containing organic compounds.

Author contributions

L.-H. M. and S.-G. H. conceived the ideas. L.-H. M. carried out the experiments and calculations and wrote the original manuscript. Y. L. and H. C. assisted with high-level quantum calculations. G. P. W. and Q.-Y. L. assisted with the experiments. S.-G. H., Z.-Y. L. and H. C. supervised the results and commented on the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grants 21833011, 22173111 and 92161205), the Youth Innovation Promotion Association CAS (No. 2020034), and the K. C. Wong Education Foundation.

References

1. S. Kim, F. Loose and P. J. Chirik, Chem. Rev., 2020, 120, 5637–5681.
2. J. Chatt, A. A. Diamantis, G. A. Heath, N. E. Hooper and G. J. Leigh, J. Chem. Soc., Dalton Trans., 1977, 7, 688–697.
3. D. J. Knobloch, E. Lobkovsky and P. J. Chirik, Nat. Chem., 2010, 2, 30–35.
4. S. P. Semproni and P. J. Chirik, J. Am. Chem. Soc., 2013, 135, 11373–11383.
5. Z. J. Lv, Z. Huang, W. X. Zhang and Z. Xi, J. Am. Chem. Soc., 2019, 141, 8773–8777.
6. M. Wang, L.-Y. Chu, Z.-Y. Li, A. M. Messinis, Y.-Q. Ding, L. Hu and J.-B. Ma, J. Phys. Chem. Lett., 2021, 12, 3490–3496.
7. Y. Nakanishi, Y. Ishida and H. Kawaguchi, Angew. Chem., Int. Ed., 2017, 56, 9193–9197.
8. D. J. Knobloch, H. E. Toomey and P. J. Chirik, J. Am. Chem. Soc., 2008, 130, 4248–4249.
9. S. F. McWilliams, D. L. J. Broere, C. J. V. Halliday, S. M. Bhutto, B. Q. Mercado and P. L. Holland, Nature, 2020, 584, 221–226.
10. L. Morello, J. B. Love, B. O. Patrick and M. D. Fryzuk, J. Am. Chem. Soc., 2004, 126, 9480–9481.
11. S. M. Lang and T. M. Bernhardt, Phys. Chem. Chem. Phys., 2012, 14, 9255–9269.
12. L.-H. Mou, G.-D. Jiang, Z.-Y. Li and S.-G. He, Chin. J. Chem. Phys., 2020, 33, 507–520.
13. Y.-X. Zhao, Z.-Y. Li, Y. Yang and S.-G. He, Acc. Chem. Res., 2018, 51, 2603–2610.
14. H. Schwarz, S. Shaik and J. Li, J. Am. Chem. Soc., 2017, 139, 17201–17212.
15. D. V. Fries, M. P. Klein, A. Steiner, M. H. Prosenc and G. Niedner-Schatteburg, Phys. Chem. Chem. Phys., 2021, 23, 11345–11354.
16. F. Mafune, Y. Tawaraya and S. Kudoh, J. Phys. Chem. A, 2016, 120, 4089–4095.
17. G. Liu, I. R. Ariyarathna, S. M. Ciborowski, Z. Zhu, E. Miliordos and K. H. Bowen, J. Am. Chem. Soc., 2020, 142, 21556–21561.
18. N. Levin, J. Lengyel, J. F. Eckhard, M. Tschurl and U. Heiz, J. Am. Chem. Soc., 2020, 142, 5862–5869.
19. R. Kretschmer, M. Schlangen and H. Schwarz, Understanding Organometallic Reaction Mechanisms and Catalysis, ed. V. P. Ananikov, Wiley-VCH, Weinheim, 2014, pp. 1–16.
20. S. Zhou, J. Li, M. Schlangen and H. Schwarz, Acc. Chem. Res., 2016, 49, 494–502.
21. R. Kretschmer, M. Schlangen, M. Kaupp and H. Schwarz, Organometallics, 2012, 31, 3816–3824.
22. L.-H. Mou, Y. Li, Z.-Y. Li, Q.-Y. Liu, H. Chen and S.-G. He, J. Am. Chem. Soc., 2021, 143, 19224–19231.
23. L.-H. Mou, Y. Li, Z.-Y. Li, Q.-Y. Liu, H. Chen and S.-G. He, J. Phys. Chem. Lett., 2020, 11, 9990–9994.
24. Z.-Y. Li, Y. Li, L.-H. Mou, J.-J. Chen, Q.-Y. Liu, S.-G. He and H. Chen, J. Am. Chem. Soc., 2020, 142, 10747–10754.
25. Z.-Y. Li, L.-H. Mou, G.-P. Wei, Y. Ren, M.-Q. Zhang, Q.-Y. Liu and S.-G. He, Inorg. Chem., 2019, 58, 4701–4705.
26. S. Zhou, J. Li, M. Schlangen and H. Schwarz, Angew. Chem., Int. Ed., 2016, 55, 14863–14866.
27. S. Zhou, J. Li, M. Schlangen and H. Schwarz, Angew. Chem., Int. Ed., 2016, 55, 11678–11681.
28. K. Koszinowski, D. Schroder and H. Schwarz, J. Am. Chem. Soc., 2003, 125, 3676–3677.
29. M. Diefenbach, M. Brönstrup, M. Aschi, D. Schröder and H. Schwarz, J. Am. Chem. Soc., 1999, 121, 10614–10625.
30. R. Kretschmer, M. Schlangen and H. Schwarz, Angew. Chem., Int. Ed., 2011, 50, 5387–5391.
31. R. Kretschmer, M. Schlangen and H. Schwarz, Angew. Chem., Int. Ed., 2012, 51, 3483–3488.
32. M. Schlangen, J. Neugebauer, M. Reiher, D. Schröder, J. P. López, M. Haryono, F. W. Heinemann, A. Grohmann and H. Schwarz, J. Am. Chem. Soc., 2008, 130, 4285–4294.
33. G. Gioumousis and D. P. Stevenson, J. Chem. Phys., 1958, 29, 294–299.
34. Q.-Y. Liu, L. Hu, Z.-Y. Li, C.-G. Ning, J.-B. Ma, H. Chen and S.-G. He, J. Chem. Phys., 2015, 142, 164301.
35. J. Kohanoff and N. I. Gidopoulos, Handbook of Molecular Physics and Quantum Chemistry, ed. S. Wilson, John Wiley & Sons, Ltd, Chichester, 2003, pp. 532–568.
36. J. D. Watts, J. Gauss and R. J. Bartlett, J. Chem. Phys., 1993, 98, 8718–8733.
37. S. Guo, M. A. Waston, W. F. Hu, Q. M. Sun and G. K. L. Chan, J. Chem. Theory Comput., 2016, 12, 1583–1591.
38. V. Mozhayskiy and A. I. Krylov, ezSpectrum v3.0, see https://iopenshell.usc.edu/downloads.
39. D. Schroder, S. Shaik and H. Schwarz, Acc. Chem. Res., 2000, 33, 139–145.
40. D. J. Tozer and N. C. Handy, J. Chem. Phys., 1998, 109, 10180–10189.
41. J. I. Steinfeld, J. S. Francisco and W. L. Hase, Chemical Kinetics and Dynamics. Prentice-Hall, Upper Saddle River, NJ, 1999, pp. 231–313.
42. W. Jiang, N. J. DeYonker and A. K. Wilson, J. Chem. Theory Comput., 2012, 8, 460–468.
43. G.-P. Wei, Q.-Y. Liu, Y. Ren and S.-G. He, Rev. Sci. Instrum., 2021, 92, 104104.

---

Footnotes

† Electronic supplementary information (ESI) available: Method details and additional experimental and theoretical results (spectra, data analysis, and calculated structures and reaction mechanisms). See https://doi.org/10.1039/d2sc02416k

‡ Present address. University of Chinese Academy of Sciences, Beijing 100049, P. R. China.

---