A tetranuclear nickel hydride cluster capable of reversible activation of aromatic and aliphatic C–H bonds Authors: Kohei Sunami, Tatsuya Higaki, Kanata Tanaka, Ryotaro Yamanashi, W M C Sameera, Ryo Takahata, Toshiharu Teranishi, Soichi Kikkawa, Seiji Yamazoe, Takuya Shiga, Masayuki Nihei, Tatsuhisa Kato, Hitoshi Izu, Yasuhiro Ohki Published: 2026 ABSTRACT AbstractMetal cluster complexes supported by multiple hydrides are useful for activating inert bonds and molecules, while nickel-based variants capable of activating external substrates remain scarce. Here, we report a tetranuclear Ni-hydride-phosphine cluster, [Ni4H4(PtBu3)4] (1), which we synthesize by reacting [Li(THF)x][Ni{N(SiMe3)2}3] (x ≈ 4.5 − 5) with pinacolborane (HBpin) in the presence of tri-tert-butylphosphine (PtBu3). Its molecular structure features a tetrahedral Ni4 core, with four face-capping μ3-hydrides and four terminally bound PtBu3 ligands. Cluster 1 is capable of reversible C–H and C–D bond cleavage of benzene-d6, toluene-d8, n-octane-d18, and its own PtBu3 ligands at 50 °C, leading to H/D exchange between the deuterated solvents and the ligands in 1. This reactivity is further utilized in the catalytic hydrogen isotope exchange reaction between benzene-d6 and toluene. The reversible dissociation of one PtBu3 ligand in 1 is proposed as the key step to trigger the C–H and C–D bond activation. Quantum chemical computations propose the reversible displacement of one of the PtBu3 ligands in 1 as the key step in triggering C–H and C–D bond activation.Subject terms: Reaction mechanisms, Chemical bonding, Chemical bonding Introduction Metal hydride clusters have attracted significant attention due to their unique reactivity, which arises from the cooperation between metal centers and hydrides1–16. For instance, [(CpLTi)3(μ-H)6(μ3-H)] ([Ti3]; CpL = C5Me4SiMe3) has been shown to react with N2 via reductive H2 elimination and N ≡ N triple bond cleavage17,18, forming nitride species that undergo hydroamination in the presence of alkenes19. Our group has demonstrated that [Fe6(μ-H)10(μ3-H)2(PMe3)6] ([Fe6]) can catalytically convert N2 to N(SiMe3)3 with a high turnover number (TON) of 183 ± 18 equiv. per cluster20. These examples highlight the potential of metal hydride clusters as platforms for activating and transforming inert molecules. Among transition-metal hydride clusters, nickel-based species are of interest due to their unique reactivity toward hydrocarbons21,22. For example, Johnson et al. reported [Ni5H6(PiPr3)5] ([Ni5]), which can cleave C = C bonds of alkenes such as ethylene and styrene at temperatures as low as −30 °C23. These reactions proceed with near-quantitative yields and high selectivity, even in the presence of other reactive functional groups such as alcohols and amines. Moreover, [Ni5] has been found to activate aromatic C–H bonds under mild conditions24,25. Despite these advances, nickel hydride clusters capable of activating external substrates remain scarce, partly because of the relatively weak Ni–C bonds compared to those of heavier metals26 and the limited structural diversity of nickel hydride clusters. Nevertheless, in a broader context, nickel-containing clusters have been explored for various applications, including alkyne semihydrogenation12, H2 activation27,28, cyclization reactions29,30, catalytic N2 reduction31,32, and the synthesis of electro-conductive nanosheets33, demonstrating the potential of nickel as a versatile functional element in cluster chemistry. Herein, we report a phosphine-protected nickel-hydride cluster, [Ni4H4(PtBu3)4] (1), featuring a tetrahedral Ni4 core coordinated by four terminal PtBu3 ligands and four face-capping hydrides. Interestingly, cluster 1 mediates H/D exchange between its hydride/phosphine ligands and deuterated solvents, such as C6D6 and n-octane-d18, via reversible C–H(D) bond activation. Heating a solution of 1 in a mixture of benzene-d6 and toluene resulted in the isotope exchange, with 1 serving as the catalyst. This work paves the way for further exploration of nickel-hydride clusters with structural diversity and unique reactivity. Results Synthesis and crystal structure of Ni4H4(PtBu3)4 (1) Typical synthetic methods for nickel-hydride clusters involve reactions of Ni complexes with H2 or hydride reagents34–41. For example, Johnson et al. synthesized [Ni5] by reacting an N2-bridged Ni complex [(iPr3P)2Ni]2(μ-N2) with H224, and a tetranuclear hydride-chloride cluster [(iPr3P)2Ni]4H4Cl2 from a related reaction of [(iPr3P)2Ni]2(μ-N2) with (iPr3P)2NiCl2 and H234. A cyclopentadienyl (Cp)-supported tetrahedral Ni hydride cluster, [(CpNi)4H3] was synthesized by reacting CpNi(NO) with AlCl3 and LiAlH435,36. Pasynkiewicz et al. reported the synthesis of [(CpNi)4H2] by reacting nickelocene and phenyllithium in the presence of 1-decene37. As such, nickel-hydride clusters prepared by conventional methods are typically stabilized by ligands originating from the starting Ni complexes. Although exceptions exist, such as the work by Schneider et al. on the synthesis of [(CpRNi)4Hn] (n = 1 or 3; CpR = C5H4tBu) by vaporizing nickel atoms into solutions of substituted Cp ligands38, these methods are less practical due to the harsh conditions required to generate atomic Ni. Recently, our group developed a facile method for the synthesis of metal hydride clusters by treating metal(II) bis-amide complexes M{N(SiMe3)2}2 with pinacolborane (HBpin) in the presence of phosphines. For example, [Fe6] was synthesized via the reaction of Fe{N(SiMe3)2}2 with HBpin and PMe320. This reaction proceeds through σ-bond metathesis, forming “Fe–H” intermediates for cluster growth and an inert byproduct (Me3Si)2N–Bpin42. This synthetic approach has also been successfully applied to the synthesis of a nanometer-sized iron-hydride cluster [Fe55H46(PtBu3)12]+43, as well as cobalt-hydride clusters including [Co6H8(PiPr3)6]5. Additionally, Jacobi von Wangelin et al. developed a related protocol, using diisobutylaluminum hydride instead of HBpin for iron-hydride cluster synthesis9,44. With these successful examples, we sought to extend this methodology to synthesize new nickel-hydride clusters. However, the instability of the nickel precursor Ni{N(SiMe3)2}2, which decomposes gradually at room temperature45, prevented its direct application. The tetrameric complex [Ni{N(SiMe3)2}]4 has been reported45, but we have been unable to synthesize it on a gram scale. We then tested an alternative two-coordinate complex reported by Don Tilley et al., Ni{N(SiMe3)(DIPP)}2 (DIPP = 2,6-diisopropylphenyl)46, but this approach was unsuccessful due to the steric hindrance of the DIPP groups, which slowed the approach of HBpin to the nickel center, preventing the formation of “Ni–H” intermediates at a sufficient rate. Under these circumstances, a tris-amide complex, [Li(THF)x][Ni{N(SiMe3)2}3] (x ≈ 4.5 − 5) (NiL3)47, reported recently by Werncke et al., offered us a potential solution. This tris-amide complex NiL3 was found to reversibly dissociate one of its amide ligands in toluene, generating Ni{N(SiMe3)2}2 and LiN(SiMe3)2. Inspired by this finding, we employed NiL3 for the in-situ generation of Ni{N(SiMe3)2}2 and subsequent reactions with HBpin in the presence of phosphines. Treatment of a toluene solution of NiL3 with a slight excess of PtBu3 (2.0 equiv.) and HBpin (3.3 equiv.) at room temperature provided a black solution, from which a tetranuclear nickel-hydride cluster [Ni4H4(PtBu3)4] (1) was isolated (Fig. 1). Initially, crystals of 1·PtBu3 containing an additional PtBu3 molecule incorporated in the crystal lattice were obtained as the primary crystalline product in 19% yield. Subsequent recrystallization from pentane yielded phosphine-free dark brown crystals of 1 in 9% yield. After separation of the first crop of 1·PtBu3, the supernatant was analyzed by mass spectrometry to reveal 1 (Supplementary Information Fig. S22), but attempts to obtain the second crop did not provide 1·PtBu3 in a pure form. An analogous reaction was also attempted using PiPr3 to explore the potential alternative synthesis of [Ni5]. However, [Ni5] was not detected from this reaction mixture. The 1H NMR spectrum of isolated 1 in C6D6 exhibited a broad signal of PtBu3 at 2.60 ppm (Supplementary Information Fig. S1), indicating the paramagnetic nature of 1 at ambient temperature. The hydride and phosphorus signals were not observed even at −80 °C, likely due to significant broadening induced by interactions with the magnetic (nickel) centers5,9,20,43. Nevertheless, the labile hydrogen atoms (hydrides) in 1 were quantified by thermolysis of crystalline 1 at 150 °C (Supplementary Information Tables S1–S2, Figs. S19–S21). The amount of released H2 was measured as 2.24(8) equivalents per molecule of 1, consistent with the presence of four hydrides. Synthesis of [Ni4H4(PtBu3)4] (1). Synthetic route to 1 from [Li(THF)x][Ni{N(SiMe3)2}3] (x = 4.5 − 5), HBpin, and PtBu3. The relative stoichiometry of the reactants is indicated. Single crystals of 1·PtBu3 were characterized by X-ray crystallographic analysis (Fig. 2A and Supplementary Information Table S3), and the molecular structure of phosphine-free 1 was similarly confirmed by X-ray analysis. In the crystal of 1·PtBu3, free PtBu3 was identified in addition to 1. The Ni–Ni bond distances in 1 range from 2.3857(3) to 2.4549(3) Å, which are shorter than the typical Ni–Ni covalent bond length (i.e., 2.48 Å)48. The formal oxidation state of metals in 1 is Ni(I), based on the chemical formula of [Ni4H4(PtBu3)4]. The observed Ni–P bond distances in 1, ranging from 2.1963(4) to 2.2128(4) Å, are consistent with those in formally Ni(I) phosphine complexes (Supplementary Information Table S4), further supporting the formal oxidation state of Ni(I) in 1. The Ni4 core is protected by face-capping μ3-hydrides that are placed on each triangle facet of the tetrahedral Ni4 structure. The positions of the hydrides were determined and refined from residual electron density during X-ray analysis. The presence of four hydrides was also confirmed by mass spectrometry (vide infra). Crystal structure of 1·PtBu3. A Crystal structure of 1·PtBu3 with atomic displacement parameters set at 50% probability. The co-crystallized PtBu3 molecule and hydrogen atoms on the PtBu3 ligands are omitted for clarity. B Selected bond distances (Å) and angles (°) in 1: Ni–Ni 2.3857(3)—2.4549(3), Ni–P 2.1963(4)—2.2128(4); Ni–Ni–Ni 59.101(7)—61.105(9). The Ni–Ni bond distances in 1 were compared with previously reported tetrahedral Ni4 hydride clusters. For example, the Ni–Ni bond distances in [(CpNi)4H3] range from 2.454(3)—2.490(3) Å, which are slightly longer than those in 1. [(CpNi)4H3] was also characterized by neutron diffraction, which revealed the positions of hydrides49. The neutron diffraction data showed Ni–Ni bond distances ranging from 2.478(3)—2.490(3) Å, consistent with those determined by X-ray crystallography. The Ni-H bond distances determined by neutron diffraction range from 1.661(9) to 1.720(8) Å. Magnetic properties and density functional theory calculations The broad signal observed in the 1H NMR spectrum of 1 suggests its paramagnetic nature at ambient temperature. To investigate its magnetic properties, the solution magnetism of 1 was analyzed using the Evans method50,51. The resulting magnetic moment of 1.8(3) μB at 298 K indicates partial population of the S = 1 state (expected spin-only moment: 2.83 μB) at ambient temperature. Further magnetic characterization of 1 was performed on solid-state samples using a superconducting quantum interference device (SQUID). Direct current (DC) magnetic susceptibility data collected on microcrystalline samples of 1 under a 1 T field revealed χMT values of 0.49 cm3 K/mol at 300 K and 0.0061 cm3 K/mol at 1.8 K, corresponding to μeff values of 2.0 and 0.22 μB, respectively. The μeff value at 300 K is consistent with the solution magnetic moment (μeff = 1.8(3) μB), while the value at 1.8 K indicates the S = 0 ground state (Supplementary Information Fig. S36). This S = 0 ground state assignment was further supported by the silent electron spin resonance (ESR) spectrum observed for microcrystalline samples at 4 and 8 K in the X-band mode (Supplementary Information Fig. S37). Density functional theory (DFT) calculations on 1, using the BP86-D3BJ functional52,53, align with the experimentally determined ground state. Further, the S = 0, 1, and 2 spin states of 1 were optimized, revealing S = 0 as the ground state (Supplementary Information Table S6). Kohn-Sham frontier orbitals of the ground state are delocalized over nickel atoms (Supplementary Information Fig. S50), showing extensive orbital delocalization over the metal framework. Despite several attempts, we were unable to locate the broken symmetry singlet (or triplet) state, where the computations ultimately converged to the closed-shell singlet state. The S = 1 and 2 states were calculated to be 7.8 and 20.5 kcal mol−1 higher in energy than the ground state, respectively. The exact S = 0/S = 1 energy gap is sensitive to the computational method employed (vide infra); all calculations consistently predict the S = 0 state with low-lying excited spin states, in agreement with the observed magnetic properties of 1. The calculated value for the S = 1 state is 2.01, which is in close agreement with the theoretical value of 2.00. This result confirms the absence of spin contamination. The computed spin densities on the nickel centers in the S = 1 state are ⍴(Ni1) = 0.43, ⍴(Ni2) = 0.44, ⍴(Ni3) = 0.42, and ⍴(Ni4) = 0.26, indicating that the unpaired electrons are primarily delocalized across the four nickel atoms. X-ray absorption fine structure (XAFS) analysis To further verify the electronic structure of Ni atoms in 1, Ni K-edge X-ray absorption fine structure (XAFS) was measured at a synchrotron facility. Solid samples were sealed in plastic films, placed in aluminum bags, and their spectra were measured in transmission mode. For comparison, some standard samples such as Ni(0) foil, Ni(I)3P, and Ni(II)Cl2·6H2O were also measured. The onset of the X-ray absorption near edge structure (XANES) spectra, which corresponds to the electronic structure of Ni, shifts to higher energy from Ni(0) to Ni(II) (Fig. 3). The absorption onsets of Ni(0) foil and Ni(I)3P are very similar, indicating that Ni atoms in Ni3P have almost metallic electron density, as reported previously54,55. The similarity of the absorption onset between Ni3P and 1 further suggests a relatively high electron density in the Ni atoms of 1. The two XANES spectra of 1 recorded at 298 and 123 K are nearly identical, probably due to the negligible temperature dependence of the electronic structures within the observed temperature range (Supplementary Information Fig. S38). Ni K-edge XANES spectra of 1 and reference compounds. Ni K-edge XANES spectra of Ni foil, Ni3P, Ni(II)Cl2·6H2O and 1 recorded at 298 K. [Ni4] = [Ni4H4(PtBu3)4] (1). Extended X-ray absorption fine structure (EXAFS) analysis of the standard samples was compared with that of 1 to investigate the local environment of the nickel centers (Supplementary Information Fig. S39). The EXAFS spectra of Ni(0) foil and Ni(I)3P, which feature the Ni–Ni or Ni–P bonds, exhibited oscillations with similar periodicity. This similarity complicated the curve-fitting analysis of 1, which contains both Ni–Ni and Ni–P bonds (Supplementary Information Fig. S40). Nevertheless, the FT-EXAFS data of 1 at 298 K yielded the r values (bond lengths) of 2.37(4) Å for Ni–Ni and 2.20(5) Å for Ni–P bonds (Supplementary Information Table S5), in agreement with overlapping peaks in the FT-EXAFS spectrum. These r values remained constant at 123 K, with values of 2.37(3) and 2.19(4) Å for Ni–Ni and Ni–P bonds. These r values are consistent with the crystallographic Ni-Ni and Ni-P bond distances (Ni–Ni: 2.3857(3)—2.4549(3) Å, Ni–P: 2.1963(4)—2.2128(4) Å). ESI-MS analysis of 1 and deuterated 1 The molecular formula of 1, including four hydrides, was further confirmed by ESI-MS analysis. The positive-mode spectrum exhibited a signal at m/z = 1046.5, with an isotope pattern closely matching the simulated pattern for [Ni4H4(PtBu3)4]+ ([1]+, m/z = 1046.5) (Fig. 4A). To investigate deuterium incorporation and its origin, ESI-MS analysis was also performed for deuterated samples of 1. The deuterated sample synthesized using DBpin56 in toluene-d8 gave a signal at m/z = 1058.6 (Fig. 4B, purple, and Supplementary Information Fig. S24), indicating a shift of +8 units compared to 1. This shift suggests an average of 8 H/D exchanges during synthesis. As this shift exceeds the number of hydrides in 1, it indicates deuterium incorporation into both hydrides and the PtBu3 ligands. The isotope pattern of deuterated 1 was broader due to variations in the number of incorporated deuterium atoms. Additional experiments were performed using DBpin in non-deuterated toluene and yielded a signal at m/z = 1051.5 (Fig. 4B, blue), corresponding to an average of 5 H/D exchanges. This result indicates that deuterium atoms were derived from both pinacolborane and toluene. Thus, when HBpin was used in toluene-d8, the ESI-MS spectrum showed an average m/z shift of +1 unit (Fig. 4B, green), confirming that deuterium incorporation in this case was primarily from toluene through C–D bond activation. Positive-mode ESI-MS characterization of 1 and its deuterated analogue. A Positive-mode ESI-MS spectrum of 1. The inset shows the simulated isotope pattern. B Positive-mode ESI-MS spectrum of deuterated 1 synthesized using DBpin and/or toluene-d8. Tol-H = toluene and Tol-D = toluene-d8. Reversible C–H(D) bond activation of aromatic and aliphatic hydrocarbons mediated by 1 The ESI-MS analysis of deuterated 1 (Fig. 4) revealed evidence of C–H(D) bond activation of the PtBu3 ligands and toluene-d8. This finding prompted us to investigate the C–H(D) bond activation of non-functionalized aromatic and aliphatic hydrocarbons, such as C6D6 and n-octane-d18, through hydrogen isotope exchange reactions. Heating a C6D6 solution of 1 at 50 °C for 24 h resulted in a shift of the ESI-MS signals by +36 units, exhibiting a broad signal with the highest peak at m/z = 1082.7 (Fig. 5, orange and Supplementary Information Fig. S25). The shift of +36 units corresponds to 36 H/D exchanges between the hydrides and PtBu3 ligands in 1 and C6D6. Deuterium incorporation into the PtBu3 ligands was further verified by 2H NMR analysis of deuterated 1 prepared from the reaction with C6D6 (Supplementary Information Fig. S8). The H/D exchange was facilitated in toluene-d8, leading to 98 H/D exchanges under identical conditions at 50 °C for 24 h (Fig. 5, red). Extending the reaction time to 72 h led to nearly complete deuteration, producing [Ni4D4(PC12D~27)4] (1-d~112) (Supplementary Information Fig. S26C). To further investigate the reaction behavior of 1 and evaluate the possible involvement of Ni nanoparticles in the H/D exchange process, several control experiments were performed in C6D6. Although a mercury poisoning test was initially attempted, direct reaction between Hg and 1 prevented meaningful interpretation of the experiment. A hot filtration experiment showed that the H/D exchange proceeded at a comparable rate after removal of any insoluble material, suggesting that the observed reactivity does not require insoluble heterogeneous species (Supplementary Information Fig. S45). In addition, time-course analysis revealed no observable induction period and a steady increase in H/D exchange over 48 h (Supplementary Information Figs. S29 and S30). Furthermore, the H/D exchange became slower at higher concentrations of 1 and was significantly suppressed in the presence of externally added PtBu3 (Supplementary Information Fig. S28). These observations collectively support a homogeneous molecular pathway involving reversible phosphine displacement in 1 under the reaction conditions and argue against nanoparticle-mediated reactivity as the dominant H/D exchange pathway. Deuterium incorporation into 1 upon reaction with deuterated hydrocarbons. ESI-MS spectra showing deuterium incorporation into 1 through H/D exchange with cyclohexane-d12, n-octane-d18, C6D6, and toluene-d8 after heating at 50 °C for 24 h. tBuD = partially deuterated tBu. As aliphatic C–H bond activation of PtBu3 occurs reversibly during H/D exchange, we extended this study to investigate the C–H(D) bond activation of external alkanes, which represent a challenging class of substrates due to the absence of metal-coordinating groups. Commercially available n-octane-d18 was used as the deuterated solvent instead of C6D6, and the solution of 1 was heated under similar conditions at 50 °C (Fig. 5, blue). Although the reaction proceeded more slowly than with C6D6, 11 H/D exchanges were observed after 24 h between 1 and n-octane-d18. In contrast, a similar reaction with cyclohexane-d12 at 50 °C for 24 h resulted in negligible exchange (Fig. 5, purple), indicating a preference for methyl C–H(D) bond activation. Although the origin of this substrate dependence remains unclear, the different reactivity of n-octane-d18 and cyclohexane-d12 may reflect the greater accessibility of primary C–H(D) bonds compared with the secondary C–H(D) bonds of cyclohexane. The ability of 1 to activate alkane C–H(D) bonds under relatively mild conditions at 50 °C is noteworthy, as representative rhodium and iridium complexes typically require temperatures of 100 °C or higher57–63. It is also notable that heterolytic C–H activation has proven effective with aliphatic C–H bonds at around 80 °C using a half-sandwich iridium complex64. Despite recent advances in C–H bond activation by nickel complexes22,65,66, in addition to the aromatic C–H activation at 45 °C catalyzed by a (diamine)Ni complex67, this study represents a rare example of thermally driven aliphatic C–H(D) bond activation of n-octane (and tert-butyl groups of the ligand) by a nickel complex under mild conditions. In contrast to mono-nuclear nickel complexes, multi-metallic redox behavior of the four nickel centers in 1 can cooperatively accommodate the 2e-oxidation required for C–H oxidative addition, thereby facilitating C–H(D) activation. The reversible activation of both aromatic and aliphatic C–H(D) bonds by 1 prompted us to explore its catalytic potential for H/D exchange between C6D6 and toluene. A toluene solution of 1 (0.2 mol%) was mixed with C6D6 (toluene/C6D6 = 1/5 by weight) and heated at 50 °C for 24 h (Fig. 6). The resultant 1H and 2H NMR spectra revealed deuterium incorporation into the methyl group of toluene (Supplementary Information Fig. S41), achieving a TON of 96 based on the intensity change in the methyl signal of toluene (19% methyl deuteration). When the reaction temperature was increased to 60 °C under otherwise identical conditions, the H/D exchange reaction was significantly accelerated, and the TON reached 211 after 24 h (42% methyl deuteration, Supplementary Information Fig. S44). Notably, the 1H NMR signal of 1 was still observed after completion of the reaction, indicating that the cluster remains intact under these conditions. GC-MS analysis of the reaction mixture revealed an increase in the m/z of toluene, further confirming the deuterium incorporation (Supplementary Information Fig. S43). A similar catalytic reaction was attempted using n-octane-d18 instead of C6D6, but no evidence of deuteration was observed under the same conditions, indicating that the C–H activation of n-octane by 1 is significantly slower than that of toluene. At higher temperatures (e.g. 100 °C), 1 gradually decomposed to form insoluble materials, preventing confirmation of its catalytic activity. Further attempts to functionalize hydrocarbons catalytically, such as the C–H borylation56, ended up with the decomposition of the nickel cluster and gave insoluble materials. Catalytic H/D exchange mediated by 1. A Catalytic H/D exchange between toluene and C6D6 mediated by 1. B Proposed reaction pathway involving initial H/D exchange between 1 and C6D6, followed by subsequent H/D exchange between the resulting deuterated 1 and toluene. [Ni4] = [Ni4H4(PtBu3)4]. Mechanistic insights and computed reaction pathways The 1H NMR spectrum of 1 (Supplementary Information Fig. S1) consistently exhibited a minor signal of free PtBu3, even in isolated phosphine-free crystalline samples. This observation is consistent with phosphine displacement during substrate coordination. To examine the possibility of reversible ligand dissociation, a ligand exchange experiment was performed using PCy3 (Cy = cyclohexyl), which possesses electronic and steric properties similar to those of PtBu3. In the presence of 3 equiv. of PCy3, a benzene solution of 1 was heated at 50 °C. After 1 h, the ESI-MS spectrum of the reaction mixture revealed an intense signal corresponding to the ligand-exchange product, [Ni4H4(PtBu3)3(PCy3)]+ (m/z = 1124.5), in addition to unreacted 1 (Fig. 7). This result confirms that one of the PtBu3 ligands of 1 exchanges with external ligands or substrates at 50 °C in solution. Ligand exchange between PtBu3 in 1 and PCy3. Positive-mode ESI-MS spectrum recorded after heating a benzene solution of 1 with PCy3 (3 equiv.) at 50 °C for 1 h. The H/D exchanges between hydrides in 1 and toluene-d8 or C6D6 occurred even at room temperature, as was indicated by the appearance of multiple signals of methyl and quaternary carbons in the 13C NMR (Supplementary Information Figs. S5 and S6), while this phenomenon was not observed in the 13C NMR measured in C6H6 (Supplementary Information Fig. S4). Heating of the solution of 1 in toluene-d8 at 50 °C resulted in the appearance of more complicated signals (Supplementary Information Fig. S6), indicating the deuterium incorporation into the tBu groups. These results indicate that the C–H activation barrier for the tBu groups is higher than those for toluene and benzene. The discussion above indicates that the rate-determining step of the reaction is not the PtBu3 ligand dissociation from 1 but is more likely associated with the cleavage of C–H(D) bonds. To support this hypothesis, we roughly estimated the kinetic isotope effect (KIE) for the overall reaction. This estimation was based on comparing the H/D exchange observed in the ESI-MS spectra for the reactions of 1 with C6D6 and [Ni4D4(PC12D~27)4] (1-d~112) with C6H6, as the paramagnetic nature of 1 at ambient temperature prevents accurate analysis of the H/D exchange kinetics by NMR. Heating a C6H6 solution of 1-d~112 at 50 °C for 24 h resulted in the −20 m/z shift in the ESI-MS signal (Supplementary Information Fig. S26, pink), while the reaction of 1 with C6D6 under the same conditions led to the +36 m/z shift (Fig. 5, orange). Based on these results, the preliminary KIE for this reaction can be approximately estimated as 36/20 ≈ 1.8. Although this observation is consistent with the involvement of C–H(D) bond activation in the overall reaction process, the present data do not allow unambiguous assignment of the rate-determining step. DFT calculations were performed to elucidate the mechanism of C–H bond activation of PtBu3 and benzene (Figs. 8 and 9). The free energy profile for the intramolecular C–H bond activation (cyclometallation) of PtBu3 is shown in Fig. 8, using the S = 0 state of 1 (11) as the reference point. The calculated S = 0 and S = 1 states for local minima and transition states in the profile are relatively close in energy. The spin-state energetics of 1 in Fig. 8 were obtained using ONIOM (BP86-D3BJ:GFN1-xTB), where the tBu groups of three PtBu3 ligands were treated in the ONIOM low-level to reduce computational cost. Energy gaps differ quantitatively from the full BP86-D3BJ calculations, although both methods predict the S = 0 state as the ground state. The S = 2 state was also calculated, but its contribution is negligible due to the significantly higher energy. The computed barrier for cyclometallation (11 → 1TS1) is 23.6 kcal mol−1, and overcoming this barrier is possible by heating the reaction mixture at 50 °C. In agreement with this calculation, an overall activation energy of 26.8 kcal mol−1 was estimated by preliminary kinetic analysis of the H/D exchange reaction based on ESI-MS monitoring at 40–70 °C (Supplementary Information Figs. S31–32). The computed KIE for this process is 2.3, which is in qualitative agreement with the preliminary experimental value of 1.8. The transition state in the S = 1 state (3TS1) lies only 0.5 kcal mol−1 above the S = 0 transition state (1TS1). The resulting intermediate after cyclometallation (1I1) lies 16.4 kcal mol−1 higher in energy than 11, suggesting that the reverse reaction to regenerate 1 is thermodynamically favored. These results support the feasibility of the proposed reaction pathway under the experimental conditions used in this study. Free energy profile for the intramolecular C–H bond activation of PtBu3. Relative free energies are given in kcal mol−1. Superscripts denote the spin multiplicities. Calculations were performed using the ONIOM method with BP86-D3BJ/ZORA-def2-TZVP. Free energy profile for C–H bond activation of benzene starting from 1·benzene. Relative free energies are given in kcal mol−1. Superscripts denote the spin multiplicities. Calculations were performed using the ONIOM method with BP86-D3BJ/ZORA-def2-TZVP. An initial step in substrate C–H activation likely involves displacement of a PtBu3 ligand, as suggested by the ligand exchange reaction of 1 (Fig. 7). Consistent with this interpretation, the H/D exchange between 1 and C6D6 became progressively slower upon increasing the amount of externally added PtBu3 from 1 to 10 equiv. under otherwise identical conditions (Supplementary Information Figs. S27C and S28). These observations are consistent with a mechanism in which substrate coordination is accompanied by displacement of a PtBu3 ligand prior to subsequent C-H activation. The computed relaxed potential energy surface (Supplementary Information Fig. S52) indicates that the spontaneous dissociation of one PtBu3 ligand from 11 has a significantly higher barrier than an associative pathway in which Ni-P bond elongation and benzene binding occur simultaneously. The computed free energy profile for the benzene C–H bond activation is summarized in Fig. 9. For these calculations, the reference state was chosen as 1·benzene (31·benzene in the S = 1 state), which is the pre-associated reactant complex, rather than 1 alone. This choice allows for a meaningful comparison with the intramolecular C–H activation of PtBu3 (Fig. 8), as it accounts for the significant entropy difference between intra- and inter-molecular processes. Displacement of a PtBu3 ligand upon substrate coordination, starting from 31·benzene and accompanied by a spin crossover from S = 1 to S = 0, affords the intermediate 1I2, which lies 6.0 kcal mol−1 above the entry point. The corresponding S = 1 state 3I2 is 5.3 kcal mol−1 higher in energy than 1I2. Subsequent C–H bond activation of benzene proceeds from 1I2, also involving a spin crossover (1I2 → 3TS2). The calculated free energy barrier for the C–H bond activation of benzene starting from 1I2 is 20.0 kcal mol−1. Since this step corresponds to the H/D exchange process, we calculated the KIE and obtained a value of 1.8, which is in good agreement with the roughly estimated experimental value. The resultant intermediate 3I3 lies 18.3 kcal mol−1 above 1I2. After C–H activation of benzene, migration of the terminal Ni–H hydride to a bridging position (3I3 → 1I4) occurs with a low energy barrier of only 1.4 kcal mol−1. The energy of 1I4 is comparable to that of 3I3, suggesting that reversible migration between terminal and bridging hydrides is facile. Based on these computations, we conclude that the C–H activation of benzene from 1·benzene proceeds via a pre-equilibrium involving 1I2, while we have been unable to locate an optimized transition state. This pre-equilibrium is followed by a barrier of 20.0 kcal mol−1 (1I2 → 3TS2), which is lower than that for the intramolecular cyclometallation of PtBu3 (Fig. 8). The overall barrier of 26.0 kcal mol−1 (31·benzene → 3TS2) is slightly higher than that for the cyclometallation. Nevertheless, direct quantitative comparison between unimolecular and bimolecular processes is challenging because the entropy contributions to bimolecular reactions are often difficult to estimate accurately within standard DFT thermochemical treatments. Therefore, small differences in the calculated free energy barriers should be interpreted cautiously. Importantly, the experimental results clearly demonstrate that benzene C–H activation occurs under the conditions employed. Notably, C–H bond activation by nickel cluster systems possessing singlet ground states has also been discussed previously68. This provides additional context for the present findings. For comparison, 1·benzene was also used as the reference state in calculations of the C–H activation of benzene without dissociation of PtBu3 (Supplementary Information Fig. S53), although this speculative pathway is inconsistent with experimental observations, which indicate that ligand exchange from PtBu3 to benzene occurs prior to C–H activation. The resultant free energy barrier is 26.4 kcal mol−1, which is somewhat higher than that proceeding through the pre-equilibrium intermediate 1I2 (20.0 kcal mol−1) or an overall barrier from 1·benzene (26.0 kcal mol−1). From 1I2, one could also envision an alternative pathway for intramolecular C–H activation of PtBu3 (Supplementary Information Fig. S54). However, this route exhibits a higher barrier of 29.0 kcal mol−1 (1I2 → 1TS5), rendering it an unlikely contributor to the observed reactivity. Discussion In this study, we synthesized and fully characterized a tetranuclear Ni(I)-hydride cluster, [Ni4H4(PtBu3)4] (1), which exhibits reversible intra- and inter-molecular C–H(D) bond activation. While previous studies, including seminal work by Johnson and co-workers on [Ni5], have demonstrated the potential of nickel-hydride species for C–H bond activation, cluster 1 expands this reactivity to unactivated aliphatic C–H bonds, such as those in n-octane, under mild conditions (50 °C). Unlike mono-nuclear Ni(I) complexes, which typically require challenging oxidative addition to access Ni(III) species, cluster 1 takes advantage of cooperative redox behavior of four Ni(I) centers to facilitate C–H(D) bond activation, leading to a formal Ni(I)2Ni(II)2 oxidation state. These findings overturn the conventional view of the limited capability of nickel in hydrocarbon activation, and provide a conceptual foundation for the functionalization of C-H bonds under mild conditions. Reversible H/D exchange was observed not only at hydrides but also within the supporting ligands, as revealed by mass spectrometry, providing mechanistic insights beyond those accessible via NMR alone. Moreover, ligand exchange experiments and DFT calculations indicate that replacement of one of the PtBu3 ligands with benzene occurs prior to the C–H(D) activation of benzene. The lability of PtBu3, together with the synthetic modularity of our system, distinguishes it from previously reported nickel cluster chemistry and opens new directions for the design of nickel clusters. Thus, our findings complement and extend prior works in the field, offering a versatile platform for exploring the reactivity and mechanistic diversity of nickel clusters in C–H activation chemistry. Methods General procedures Unless otherwise noted, all reactions were performed in an inert Ar atmosphere using Schlenk techniques and a glove box (typical O2 level, <1 ppm). Hexane, toluene, acetonitrile and tetrahydrofuran (THF) were purified over columns of activated alumina and supported copper catalyst purchased from Hansen & Co., Ltd. C6D6 and toluene-d8 for NMR measurements was dried over sodium metal and distilled under reduced pressure before use. [NnBu4][PF6] was recrystallized from THF prior to use. Ni3P powder (NII22PB, >99%(2N), fine powder) was purchased from Kojundo Chemical Laboratory Co. Ltd. for use as the Ni(I) reference sample for XAFS analysis. Nickel(II) tris-amide complex ([Li(THF)x][Ni{N(SiMe3)2}3], x ~ 4.5 − 5)47 and DBpin56 were prepared according to the literature procedure. All other chemicals were purchased from commercial sources and used without further purification. 1H, 2H, 13C, 31P, and 11B NMR spectra were recorded on Bruker AV400M or AV800 spectrometers, where observed signals were referenced to the residual peaks of the deuterated solvents, TMS (tetramethylsilane), or BF3·Et2O, respectively. Fourier transform-infrared spectroscopy (FT-IR) was conducted using a Bruker Optics ALPHA FT-IR equipped with Diamond ATR. Electrospray ionization mass spectrometry (ESI-MS) was carried out using a Bruker SolariX FT-ICR-MS or compact spectrometers. UV-vis spectra were measured by a JASCO V-770 spectrophotometer. The deuterated products were analyzed by GC-MS using a Shimadzu 2010 Plus coupled with a QP2010 SE. Elemental analyses were performed using a J-SCIENCE MICRO CORDER JM11 for C, H and N determinations. Synthesis of [Ni4H4(PtBu3)4] (1) [Li(THF)x][Ni{N(SiMe3)2}3] (12.0 g, 11.6 mmol; x ~ 4.5 − 5) was dissolved in toluene (10 mL) in a 100-mL round-bottom flask and stirred for 1 h at room temperature. To the solution, PtBu3 (4.70 g, 23.2 mmol, 2 equiv. to Ni) dissolved in 4 mL of toluene was added. The vial of the PtBu3 solution was rinsed with 1.0 mL of toluene, which was also added to the reaction mixture. The solution was stirred for 3 h, and the color of the solution gradually changed from dark red to dark green. To the solution, HBpin (5.5 mL, 38.3 mmol, 3.3 equiv. to Ni) was added over 20 mins while stirring at room temperature. During the addition, bubbling was observed due to the formation of gaseous H2. Then, the solution was stirred for 24 h and filtered through a Whatman syringe filter (pore size: 0.45 μm). The supernatant was stored in the freezer at −35 °C for three days, yielding dark brown block crystals of 1·PtBu3 (816 mg, 19%), which contains additional PtBu3 for crystal packing. Further purification was attained by rinsing with 1 mL of cold pentane, followed by extraction and recrystallization from pentane (15 mL) at −35 °C for 24 h, yielding dark brown needle-like block crystals of phosphine-free 1 (276 mg, 9%). 1H NMR (in C6D6, 400 MHz, ppm): δ 2.60 (P(CMe3)3). 13C{1H} NMR (in C6H6 with a C6D6 capillary, 100 MHz, ppm): δ 65.74 (P(CMe3)3), 113.90 (P(CMe3)3). Note that no signal was observed in the 31P NMR, likely due to significant broadening induced by direct interactions with magnetic centers (nickel atoms). UV-Vis (in toluene, λmax, nm (ε, cm−1 M−1)): 391 (1.9 × 104), 531 (5.9 × 103), 664, (3.5 × 103). Anal. Calcd. for C48H112Ni4P4: C, 55.01; H, 10.77; N, 0.00. Found: C, 54.96; H, 10.83; N, 0.00. FT-IR (ATR, cm−1): ν 2992, 2966, 2950, 2889, 2863, 1479, 1469, 1442, 1396, 1387, 1356, 1172, 1019, 925, 807, 589, 562, 492, 424. X-ray crystallographic analysis Crystallographic data and refinement parameters of 1·PtBu3 and 1 are listed in Table S3. Single crystals were covered with oil (Immersion Oil, type-B: code 1248, Cargile Laboratories, Inc.) and mounted on a MiTeGen MicroLoop. Diffraction data were collected under a cold N2 stream (–150 °C), on a Rigaku RA-Micro 7 equipped with a PILATUS 200 K detector, using graphite-monochromated Mo Kα radiation (λ = 0.71075 Å). Eighteen preliminary images were taken at 0.5o increments of ω to evaluate the crystal quality and to determine pre-refined unit cell parameters, followed by full data collection also with 0.5o increments of ω. The raw frame data was processed using CrystalClear program, and the data sets were corrected for absorption using the REQAB program. The structure was solved with Intrinsic Phasing using SHELXT and refined by full-matrix least-square on F2 using SHELXL in Olex2. All the non-hydrogen atoms were refined anisotropically. All the H atoms were refined isotropically. For the structural analysis of 1, the twin law (−1 0 0; 0 −1 0; 0 0 −1) was applied with a 48.8(1.2)% contribution from a second domain. SIMU, RIGU, DELU, SADI, and FLAT were also applied to the disordered structural units. AFIX, DFIX, and SADI were applied to model hydrides due to significant structural disorder. The solvent molecules in 1 were SQUEEZED from the lattice using the Olex2 solvent mask command due to unresolvable disorder. A solvent mask was calculated, and 142 electrons were found in a volume of 714 Å3 in the void per unit cell. This is consistent with the presence of one pentane, 1·[C5H12], per Asymmetric Unit which accounts for 168 electrons per unit cell. The cif file can be found at the Cambridge Crystallographic Data Centre (CCDC; https://www.ccdc.cam.ac.uk) under accession numbers CCDC 2422079 (1·PtBu3) and 2422080 (1). Ni K-edge X-ray absorption fine structure (XAFS) The Ni K-edge XAFS measurements were conducted at beamlines BL01B1 of the SPring-8 facility of the Japan Synchrotron Radiation Research Institute (JASRI). The incident X-ray beam was monochromatized by a Si(111) double-crystal monochromator. As a reference, Ni foil, Ni3P, and Ni(II)Cl2·6H2O were measured in transmission mode. XAFS spectra of 1 were measured in transmission mode at room temperature and 123 K using an ionization chamber as a detector. Low temperatures were controlled and maintained using a UNISOKU CoolSpeK in conjunction with a temperature controller and a gas flow controller. Obtained EXAFS spectra were analyzed by the xTunes program69. The k3-weighted EXAFS spectra in the k range 3.0–16.0 Å for the Ni K-edge were Fourier transformed into r space for structural analysis. In the curve fitting analysis, the phase shifts and backscattering amplitude functions of Ni–Ni, Ni–Cl, and Ni–P were calculated using the FEFF8.5 L program. The results of the curve fitting analysis are summarized in Table S5. Electrochemical measurements Cyclic voltammograms were recorded on a BAS ALS-660A electrochemical analyzer. All potentials were measured with reference to Ag/AgCl (Innovative Instruments, Inc.), and ferrocene (Fc) was used as an external standard. Each measurement was performed under the following conditions: sample concentration, 1 mM; solvent, THF; working electrode, glassy carbon; counter electrode, Pt; reference electrode, Ag/AgCl (Innovative Instruments, Inc.); supporting electrolyte, [nBu4N][PF6] (0.2 M); starting potential, rest potentials. Magnetic measurements The microcrystalline sample of 1 (49.5 mg) was sealed in a quartz tube with quartz wool, and the tube was flame-sealed before measurement. Susceptibility data were collected using a Quantum Design MPMS-5XL SQUID magnetometer. Magnetic data were corrected for the diamagnetism of the sample using Pascal’s constants70. The diamagnetism of the sample holder and quartz wool was corrected using a blank sample tube without a compound. DC magnetization data for 1 were collected under applied dc magnetic fields at 1 T. The solution magnetic susceptibility of 1 was determined by the Evans method, where 12 vol% cyclohexane in C6D6 was used as the solvent. Using this solvent, the 1H NMR spectrum of 1 (2.3 mM) was measured at 298 K, in the presence of a sealed capillary containing the same solvent. The cyclohexane signals from the solution of 1 and from the capillary were observed, and their difference in the magnetic field was used to calculate the magnetic moment50,51. Electron spin resonance (ESR) measurements For ESR measurements, a powder of 1 was loaded into quartz tubes (o.d. 3.0 mm) in an Ar-filled glovebox and sealed with a J-Young valve. Continuous wave (CW) X-band (9.4 GHz) spectra were collected using a Bruker E580 spectrometer equipped with a super high Q resonator (ER 4122SHQ). Cryogenic temperatures were controlled and maintained using an ESR900 liquid helium cryostat in conjunction with a temperature controller and a gas flow controller. H/D exchange between 1 and deuterated solvents General procedure 1 (2.0 mg) was dissolved in 600 mg of a deuterated solvent (benzene-d6, n-octane-d18, cyclohexane-d12, or toluene-d8). The solution was transferred to a glass vessel, sealed, and heated at 50 °C for 24 h. After cooling to room temperature, an aliquot was diluted with THF and analyzed by ESI-MS (positive mode) to monitor the reaction progress. Investigation of concentration effects Samples were prepared by dissolving 1 (1.0, 2.0, 3.0, or 4.0 mg) in 600 mg of C6D6. Each solution was transferred to a glass vessel, sealed, and heated at 50 °C for 6, 12, 24, 36, or 48 h. After cooling, aliquots were diluted with THF and analyzed by ESI-MS. Investigation of temperature effects Samples were prepared by dissolving 1 (2.0 mg) in 600 mg of C6D6. Each solution was then transferred to a glass vessel and sealed. After heating the sample at 40, 50, 60, and 70 °C for 24 h, an aliquot was taken and diluted with THF, and the reaction progress was monitored by ESI-MS. Hot filtration test A mixture of 1 (5.0 mg, 5.0 μmol) and C6D6 (1500 mg) was prepared and divided equally into two glass vessels, which were sealed. Both samples were heated at 50 °C for 12 h, and the reaction progress was monitored by ESI-MS. One sample was then filtered through a Whatman syringe filter (0.45 μm pore size). Portions (600 μL) of both the filtered and unfiltered solutions were transferred to new glass vessels, sealed, and heated at 50 °C for an additional 12 h. After cooling, aliquots were diluted with THF and analyzed by ESI-MS. H/D exchange between 1 and C6D6 in the presence of PtBu3 1 (2.0 mg) was dissolved in 600 mg of C6D6. PtBu3 (1 or 10 equiv relative to 1) was added, and the solution was transferred to a glass vessel, sealed, and heated at 50 °C for 24 h. After cooling, an aliquot was diluted with THF and analyzed by ESI-MS. H/D exchange between C6D6 and toluene catalyzed by 1 Method A, at 50 °C An NMR tube was charged with 1 (2.0 mg, 2.0 μmol), toluene (100 mg, 1 mmol, 500 eq.), and C6D6 (500 mg). A capillary containing a C6D6 solution of SiMe4 was inserted into the tube as an external standard. The 1H and 2H NMR spectra were measured before starting the reaction. After heating the NMR tube at 50 °C for 24 h, the 1H and 2H NMR spectra were recorded. The intensity of the methyl group of toluene in the 1H NMR decreased by 19.2% after heating (Fig. S41 top). Therefore, at least 96 times (500 eq. x 19.2%) of H/D exchange occurred for the methyl group of toluene. The volatile material was separated from the mixture by distillation and subjected to GC-MS analysis. Method B, at 60 °C An NMR tube was charged with 1 (2.0 mg, 2.0 μmol), toluene (100 mg, 1.0 mmol, 500 equiv), C6D6 (500 mg), and cyclododecane (16.8 mg, 100 μmol) as an internal standard. 1H and 2H NMR spectra were recorded prior to heating. The tube was then sealed and heated at 60 °C for 24 h. After cooling to room temperature, a capillary containing a C6D6 solution of SiMe4 was inserted for helping d-locking of the measurement and as a chemical shift reference. Then, the 1H and 2H NMR spectra were recorded. The intensity of the methyl signal of toluene in the 1H NMR spectrum decreased by 42.3% after heating. On this basis, the extent of H/D exchange at the methyl position of toluene was estimated to be at least 211 turnovers (500 eq. ×42.3%). Computational methods Ground state geometry optimizations were performed for the S = 0, 1, and 2 states of the Ni cluster (1) using density functional theory (DFT), as implemented in the ORCA (Version 6.0)71. The ZORA scalar relativistic method was employed72. The BP86-D3BJ functional52,53, including the Grimme’s dispersion and Becke-Johnson damping73,74, ZORA-def2-TZVP basis sets and SARC/J auxiliary basis sets were used for all atoms75–77. An implicit solvent model, specifically a conductor-like continuum polarization model (C-PCM) was used78, where toluene was used as the solvent. Starting from the BP86-D3BJ optimized ground-state structures, vertical excitation energies were computed using the time-dependent density functional theory (TDDFT), based on the BP86-D3BJ optimized geometries. The B3LYP functional79,80, employing the Grimme’s dispersion and Becke-Johnson damping, ZORA-def2-TZVP basis sets and SARC/J auxiliary basis, C-PCM were used for TDDFT calculations. To investigate the mechanism for the C–H bond activation of benzene and PtBu3 unit, two-layer Our own N-layered Integrated molecular Orbital and Molecular mechanics (ONIOM) method81,82, as implemented in the ORCA program, was used. Partitioning of the molecular systems is shown in Fig. S48. For the ONIOM-high layer, BP86-D3BJ functional, the ZORA-def2-TZVP basis sets, and SARC/J auxiliary basis sets were employed. The Geometries, Frequencies, and Non-covalent Interactions Tight Binding (GNF1-xTB) method83 was applied for the ONIOM-low layer. All reported free energies include thermal corrections evaluated at 298.15 K and 1 atm. Potential energy of the optimized structures was calculated as the single-point energy, employing the BP86-D3BJ functional, ZORA-def2-TZVP basis sets, SARC/J auxiliary basis, and C-PCM for all atoms. The Kinetic Isotope Effect (KIE) was computed using the following formula: where: kH = rate constant for the reaction involving hydrogen kD = rate constant for the reaction involving deuterium ΔGH‡ = Gibbs free energy of activation for the reaction involving hydrogen ΔGD‡ = Gibbs free energy of activation for the reaction involving deuterium R = universal gas constant T = temperature in Kelvin Supplementary information Supplementary Information Transparent Peer Review file Source data Source Data