Bioinspired Binickel Catalyst for Carbon Dioxide Reduction: The Importance of Metal–ligand Cooperation

Abstract

Catalyst design for the efficient CO₂ reduction reaction (CO₂RR) remains a crucial challenge for the conversion of CO₂ to fuels. Natural Ni–Fe carbon monoxide dehydrogenase (NiFe-CODH) achieves reversible conversion of CO₂ and CO at nearly thermodynamic equilibrium potential, which provides a template for developing CO₂RR catalysts. However, compared with the natural enzyme, most biomimetic synthetic Ni–Fe complexes exhibit negligible CO₂RR catalytic activities, which emphasizes the significance of effective bimetallic cooperation for CO₂ activation. Enlightened by bimetallic synergy, we herein report a dinickel complex, Ni^IINi^II(bphpp)(AcO)₂ (where NiNi(bphpp) is derived from H₂bphpp = 2,9-bis(5-tert-butyl-2-hydroxy-3-pyridylphenyl)-1,10-phenanthroline) for electrocatalytic reduction of CO₂ to CO, which exhibits a remarkable reactivity approximately 5 times higher than that of the mononuclear Ni catalyst. Electrochemical and computational studies have revealed that the redox-active phenanthroline moiety effectively modulates the electron injection and transfer akin to the [Fe₃S₄] cluster in NiFe-CODH, and the secondary Ni site facilitates the C–O bond activation and cleavage through electron mediation and Lewis acid characteristics. Our work underscores the significant role of bimetallic cooperation in CO₂ reduction catalysis and provides valuable guidance for the rational design of CO₂RR catalysts.

Introduction

The reduction of CO₂ to CO is a promising strategy for utilizing CO₂ as a renewable C₁ feedstock and establishing an artificial carbon cycle.¹⁻⁵ Despite numerous studies on transition-metal-based CO₂RR molecular catalysts, including those based on Re,⁶⁻¹⁰ Rh,¹¹⁻¹³ Ru,¹⁴⁻²³ Fe,²⁴⁻³⁷ Ni,³⁸⁻⁴⁴ Co,⁴⁵⁻⁵⁵ Mn,⁵⁶⁻⁶³ Cr,^64,65 and so forth,⁶⁶⁻⁷⁰ challenge persists in bridging the gap between current advances and practical applications, and thus the design of efficient molecular catalysts remains a significant challenge.⁷¹⁻⁷⁴ In nature, carbon monoxide dehydrogenase (CODH) plays a pivotal role in biological carbon fixation processes. This enzyme proceeds the binding of CO and CO₂ to its Ni–Fe bimetallic active center, followed by electron transfer to the substrate, enabling the reversible conversion.⁷⁵⁻⁸⁰ NiFe-CODH exemplifies the importance of bimetallic synergy in CO₂ reduction catalysis: it activates CO₂ at only −0.6 V versus standard hydrogen electrode (SHE) through the Ni site, with a [Fe₃S₄] cluster serving as an electron mediator. Upon protonation of Ni-CO₂, the Fe site promotes the hydroxide elimination as a Lewis acid, thereby catalyzing the conversion of CO₂ to CO (Scheme 1a).⁸¹⁻⁸³ This mechanism indicates that dinuclear metal complexes are promising candidates for CO₂ reduction catalysis, as evidenced in several bimetallic catalytic systems.⁸⁴⁻⁹⁹

Scheme 1. (a) Structure of the NiFe-CODH Active Site; (b) Proposed Two Modes of Bimetallic CO₂ Activation; (c) CO₂ Reduction by the Designed Dinickel Catalyst in This Work.

Although structure similarities in bimetallic catalytic intermediates, such as Pd(μ₂-η²-CO₂)Pd^85,87 and Co(μ₂-η²-C,O:η¹-O)Co,⁹¹ suggest that the structure feature of NiFe-CODH could inform the development of artificial bimetallic catalysts, the synthesized Ni–Fe clusters and macrocyclic Ni–Fe complexes with closely positioned metals have yet to demonstrate the catalytic activity for CO₂ reduction.^100−109 These findings underscore that simply combining two metals does not replicate the catalytic function of NiFe-CODH and also highlight the necessity of integrating CO₂ complexation, effective electron transfer, stabilization of intermediates, and an appropriate geometric arrangement of the catalyst to facilitate C–O bond cleavage and CO release. Meanwhile, two primary mechanisms of bimetallic cooperative CO₂ activation have been proposed: M₁(μ₂-η²-CO₂)M₂ and M₁(μ₂–C(O)OC(O)O−κ²-C,O)M₂ (Scheme 1b).^{85,91,95,110,111} For the former, maintaining a close intermetallic distance (<3.0 Å) is crucial,¹⁰⁸ which imposes a higher geometric constraint. Conversely, the latter mechanism can accommodate various geometries, potentially allowing for closer positioning and enhanced interaction between the metal sites.

Herein, as depicted in Scheme 1c, we designed a well-structured dinickel complex with a nearly planar geometry. Taking inspiration from the [Fe₃S₄] cluster in NiFe-CODH, we employed a redox-active phenanthroline ligand¹¹² as an electron reservoir. This design facilitates electron transfer and enhances CO₂ binding at the coordinated Ni1 site. Furthermore, the secondary Ni2 site, which possesses a distinct coordination environment, aids in the activation and cleavage of the C–O bond. This is achieved through a [Ni(μ₂–C(O)OC(O)O−κ²-C,O)Ni] six-membered ring intermediate, which mitigates the geometry constraints associated with bimetallic centers. The binuclear [Ni^IINi^II(bphpp)(AcO)₂] complex (NiNi(bphpp)) demonstrates high reactivity in the catalytic reduction of CO₂ to CO and promotes the generation of CO₃^2– through ligand reduction and bimetallic synergy. This work holds the potential to make a substantial contribution to the development of novel strategies for CO₂ capture and utilization.

Results and Discussion

Synthesis and Characterization of NiNi(bphpp) and Ni(hbpp)

H₂bphpp (2,9-bis(5-tert-butyl-2-hydroxy-3-pyridylphenyl)-1,10-phenanthroline) and H₂hbpp (2,9-bis(5-tert-butyl-2-hydroxyphenyl)-1,10-phenanthroline) are synthesized by the modified procedures¹¹³ and used as the dinuclear and mononuclear ligands, respectively (Scheme S1, Figures S1–S15). The dinickel complex NiNi(bphpp) and mononickel complex [Ni^II(hbpp)] (Ni(hbpp)) are prepared by mixing the corresponding ligand with 2 equiv or 1 equiv Ni(AcO)₂·4H₂O, respectively, which are detailed in the Supporting Information (SI). The X-ray crystal structure of NiNi(bphpp) shows that the two Ni^II ions are both in an octahedral geometry configuration, which are bridged by two phenolic O atoms and connected by two acetates located above and below the ligand plane (Figures 1a and S16). The distance between the two Ni^II sites is 2.70 Å, which is similar to the distance of 2.8 Å between Ni and Fe in CODH,^81,83,108 indicating a suitable geometric structure required for CO₂ activation. In contrast, the X-ray crystal structure of Ni(hbpp) shows that the Ni^II ion has a square-planar configuration (Figures 1b and S17), consistent with low-spin Ni^II (S = 0) characterized by ¹H NMR (Figure S18). NiNi(bphpp) and Ni(hbpp) were also characterized by high-resolution mass spectrometry (HRMS) in methanol, which showed the molecular cation peak at m/z 372.0763 ([Ni^IINi^II(bphpp)]²⁺, Figure S19) and m/z 555.1557 ([Ni^II(hbpp) + Na]⁺, Figure S20), respectively. These results suggest that NiNi(bphpp) and Ni(hbpp) exist as di- and mononickel species in solution, respectively. In addition, the density functional theory (DFT) study showed that the dissociation energy of acetate in NiNi(bphpp) is 13.5 kcal/mol (Figure S41), indicating the consistency of the crystal structure with that in solution.

Figure 1.

Schematic structures and X-ray crystal structures of NiNi(bphpp) (a), (b) and Ni(hbpp) (c), (d), respectively. Ellipsoids are shown at 80% probability of atoms and 60% probability of bonds. Two acetate anions are drawn in wireframe style; the hydrogen atoms and solvent molecules are omitted for clarity.

The electronic configuration of NiNi(bphpp) is further investigated. The Evans effective magnetic moment (μ_eff) was measured to be 4.16 μ_B at 290 K (Table S4, detailed in SI),^114,115 which is consistent with the spin multiplicity of S = 2. The magnetic properties of NiNi(bphpp) were obtained over the temperature range of 1.8–300 K by a superconducting quantum interference device (SQUID, Figure S21). The calculated effective magnetic moment, μ_eff, was 4.5 μ_B at the range 200–300 K, which is consistent with the measurement obtained using the Evans method, indicating a total spin state of S = 2. As the temperature decreases, the magnetic susceptibility increases, reaching a maximum of 5.3 μ_B at 10 K, which is close to the expected μ_eff for a two-spin-only S = 1 Ni^II system (μ_eff.s.o = 2.8 × 2 = 5.6 μ_B), suggesting the two Ni^II ions are weakly ferromagnetic coupled. These results agree well with the DFT calculation of the ground state electronic structure of NiNi(bphpp) that indicates a quintuple state (Figure S48). The μ_eff. decreases to 4.0 μ_B at 1.8 K, which could be ascribed to the zero-field splitting (zfs) of the ground state. The two Ni^II ions of NiNi(bphpp) are assumed to be equivalent with the same zfs parameters D₁, rhombicity E/D, and electronic g-tensors. VT-VH magnetic susceptibility was collected at 1, 4, and 7 T (Figure S21). Global simulation of the VT-VH magnetic susceptibility and magnetization data, based on the spin-Hamiltonian (Section 1 of SI) of the low-lying electronic states, yielded an exchange coupling J = 7.4 cm^–1, and D₁ = D₂ = −9.4 cm^–1 with strongly anisotropic g values of g_x = g_y = 2.10 and g_z = 2.33 for each Ni^II ion. The X-band EPR spectra of NiNi(bphpp) for orientations of the microwave field B₁ parallel and perpendicular to static field H were both obtained at 5 K in frozen DMF solution (Figure S22), and the valleys at g = 12.6 and g = 9.9 were observed, respectively. For this non-Kramers dinuclear Ni system, large D₁ values of each Ni^II ion were obtained from the magnetic data fitting and comparable to the exchange coupling, J, causing the S = 2 and S = 1 multiplets to mix,^116−118 and the zero-field absorption would origin from the energy levels with splitting in energy Δ ≈ 0.3 cm^–1. Simulation of the parallel mode and perpendicular mode spectra gives E/D ≈ 0.12 with Dstrain = 0.1 cm^–1 and Estrain = 0.05 cm^–1 for the broad lines.

Electrochemical Studies of NiNi(bphpp) and Ni(hbpp)

The redox behavior of NiNi(bphpp) was examined by cyclic voltammogram (CV) and differential pulse voltammetry (DPV) measurements in anhydrous degassed DMF. The CV and DPV of NiNi(bphpp) show four consecutive reduction waves at −1.68, – 1.80, – 2.14, and −2.44 V vs Fc^+/0 (Figures 2a and S25, all potentials below quoted vs Fc^+/0). The scan-rate-dependent CVs illustrate that the first wave is reversible with a peak-to-peak separation (ΔE_p) of 70 mV at a scan rate of 0.1 V/s (Figure S26), suggestive of a one-electron reduction process. The subsequent three waves are found to be irreversible (Figure S25). To elucidate the electrochemical processes associated with each reduction peak, a trizinc complex, Zn₃(bphpp)(AcO)₄, was synthesized (Figures S23 and S24). Its CV displayed two successive irreversible reduction waves at −1.86 and −1.98 V (Figure S27),¹¹⁹ indicating that the initial two reduction peaks in the CV of NiNi(bphpp) are attributable to the reduction of the phenanthroline moiety. DFT studies were employed to further investigate the electrochemical processes. The one-electron reduction of NiNi(bphpp) (1) leads to the formation of the Ni^IINi^II(bphpp^–•)(AcO)₂ species (2), which exhibits a spin population of 0.82 on the phenanthroline moiety, and the potential of this reduction process is calculated to be −1.65 V (eq 1, Figure S49). At the calculated potential of −1.69 V, Ni^IINi^II(bphpp^–•)(AcO)₂ can accept an additional electron at the phenanthroline moiety to generate Ni^IINi^II(bphpp^2–)(AcO)₂ (3_+2AcO, Figure S50). The two AcO^– ligands of 3_+2AcO readily dissociate with a Gibbs free energy change of −11.3 kcal/mol, which is followed by charge redistribution, leading to the formation of Ni^INi^I(bphpp) (3, eq 2). Moreover, even at a scan rate of 2.5 V/s, the second reduction wave remains irreversible, suggesting that the dissociation rate of the two AcO^– groups exceeds a certain threshold (Figure S28). Additionally, the reverse oxidation scan following the second reduction wave is tentatively ascribed to two-electron oxidation of the Ni^INi^I(bphpp) species. Subsequently, the phenanthroline unit of Ni^INi^I(bphpp) undergoes two successive one-electron reductions at −1.75 and −2.27 V, resulting in the generations of Ni^INi^I(bphpp^–•) (4, eq 3, Figure S51) and Ni^INi^I(bphpp^2–) (4_red, eq 4, Figure S52), respectively. These findings align with the observed experimental electrochemical behavior and highlight the role of the phenanthroline moiety as an electron reservoir. For Ni(hbpp), the CV and DPV show two reduction peaks at −1.83 and −2.46 V, which are more negative than those observed for NiNi(bphpp) (Figures 2a and S29). The scan-rate-dependent CVs for Ni(hbpp) also indicate that the first wave is reversible, with a ΔE_p value of 66 mV at 0.1 V/s (Figure S30), consistent with a one-electron reduction. These reduction potentials are similar to those reported for (^tbuthbpy)FeCl by Machan et al.,¹²⁰ which are proposed to represent two consecutive one-electron reductions, leading to the formation of Ni^II(hbpp^–•) and Ni^II(hbpp^2–), respectively. These findings are consistent with DFT results (see Figure S60). The CV comparisons demonstrate that both the phenanthroline moiety and the secondary Ni site in NiNi(bphpp) could help mitigate charge accumulation.

1

2

3

4

Figure 2.

(a) CVs of 0.5 mM NiNi(bphpp) and 0.5 mM Ni(hbpp) under Ar atmosphere; (b) CVs of 0.5 mM NiNi(bphpp) (red line) and 0.5 mM Ni(hbpp) (blue line) under saturated CO₂ atmosphere, and CV under saturated CO₂ without catalyst (black line) using a glassy carbon electrode at 100 mV/s in 0.1 M ⁿBu₄NPF₆/DMF electrolyte.

In CO₂-saturated anhydrous DMF solution (0.23 M⁴), the CV of NiNi(bphpp) shows a significant current enhancement, with a half peak potential (E_cat/2) of −2.25 V and peak potential at −2.40 V (Figure 2b).¹²¹ The overpotential is determined to be 1.52 V concerning the estimated potential for the CO₂/CO couple in anhydrous DMF. this value is comparable to those reported for past catalytic systems in aprotic solutions.^{60,95,122,123} In contrast, the E_cat/2 of Ni(hbpp) is ca. −2.43 V under the same condition (Figure 2b), where the overpotential is more negative by 0.18 V than that of the dinickel catalyst. This suggests that the introduction of a secondary Ni site decreases the overpotential for catalysis by facilitating charge accumulation. According to the CVs of NiNi(bphpp) under Ar and CO₂ atmosphere, the third irreversible reduction wave is attributed to the catalytic CO₂ reduction by the Ni^INi^I(bphpp^–•) species. For Ni(hbpp), the electrochemical properties indicate that the species formed after two-electron reduction, that is Ni^II(hbpp^2–) (Figure S60), is responsible for catalyzing CO₂ reduction.

To confirm the products of electrocatalytic CO₂ reduction, controlled potential electrolysis (CPE) of 0.5 mM NiNi(bphpp) at −2.20 V was carried out using a foamed graphite electrode (1.0 × 1.0 × 0.6 cm³). The electrolysis current remained stable at an absolute value of approximately 2.8 mA, with a total charge consumption of 53.6 coulombs over 5 h (Figures 3a and S31). Gas chromatography (GC) and ion chromatography (IC) were used to detect the products in the gas phase above the electrolysis cell and the electrolyte solution, respectively. The major products were identified as CO and CO₃^2– (Figures S32 and S33). The CO production was quantified at 250 μmol over 5 h by GC (Figure 3b), while the turnover number (TON) and the Faradaic Efficiency (FE) were calculated to be 50.2 and 91%, respectively (Table S5). Furthermore, the production of CO₃^2– was determined to be 267 μmol with a FE of 97% (Figure S33a). The measured quantity of CO₃^2– is marginally higher than that of CO, which can be attributed to the inherent deviation in the quantification of CO₃^2–.⁶ Additionally, negligible amounts of H₂ and HCOO^– were detected (Figures S32 and S34), which was presumed to be due to the trace amounts of H₂O in DMF. Gas chromatography–mass spectrometry (GC–MS) of CPE under a ¹³CO₂ atmosphere confirmed the formation of ¹³CO (Figure S35). These results indicate that the reductive disproportionation of CO₂ occurs without the presence of a Brönsted acid. In contrast, CPE of 0.5 mM Ni(hbpp) under the same conditions resulted in an electrolysis current of only about 0.5 mA, and product analysis indicated the formation of merely 45 μmol CO and 49 μmol CO₃^2– (Figures 3b and S33b) over 5 h CPE with FEs of CO and CO₃^2– are 90% and 99%, respectively. These results illustrate that Ni(hbpp) is less active toward the electrocatalytic reduction of CO₂ compared to NiNi(bphpp), suggesting that the secondary Ni site plays an important role in the catalytic process of CO₂ reduction disproportionation.

Figure 3.

(a) CPE current curves of 0.5 mM NiNi(bphpp) (red line), 0.5 mM Ni(hbpp) (blue line) in anhydrous 0.1 M ⁿBu₄NPF₆/DMF, and blank solution (gray line) at −2.20 V; (b) CO production curves of 0.5 mM NiNi(bphpp) (red dot), 0.5 mM Ni(hbpp) (blue dot), and no catalyst (gray dot) during CPE at −2.20 V in 0.1 M ⁿBu₄NPF₆/DMF CO₂-saturated electrolyte.

Confirmation of Homogeneous Catalysis

To confirm the homogeneous nature of the catalysis, stability tests of the catalysts were conducted to determine whether (a) the catalyst gradually decomposes throughout the CPE; and (b) the active heterogeneous materials form and attach to the electrode surface. Due to the large current when using the foamed graphite electrode, it is difficult to observe the difference before and after electrolysis. Consequently, a plate glassy carbon electrode was chosen as the working electrode. After 4 h CPE at −2.20 V, CV of 0.1 M ⁿBu₄NPF₆/DMF blank solution without either NiNi(bphpp) or Ni(hbpp) showed no obvious current change by using the plate glassy carbon electrode after rinse (Figure S36). Additionally, CVs of 0.5 mM NiNi(bphpp) or Ni(hbpp) solution after 4 h of CPE with the rinsed electrode displayed only minor differences compared to the CVs obtained before electrolysis with a fresh electrode (Figure S36). Further examination of the surface of the electrolyzed electrode through X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM) revealed that the morphology and composition of rinsed plate glassy carbon electrode remained consistent with an unused electrode, with no visible heterogeneous materials detected (Figures S37 and S38). These findings strongly support the assertion that the catalysis processes involving NiNi(bphpp) and Ni(hbpp) are stable and homogeneous.

Kinetic Analysis

To intuitively assess the catalytic efficiency of NiNi(bphpp) and Ni(hbpp), the foot of wave analysis (FOWA) method is applied to determine the turnover frequency (TOF).^124−126 Our results show that the calculated TOF of NiNi(bphpp), derived from catalytic CV was 20.5 s^–1, while that of Ni(hbpp) was determined to be 4.4 s^–1 (Table S6, Figures S39 and S40).¹²⁷ The improved TOF_calc obtained by the introduction of the secondary Ni site demonstrates the significant enhancement of catalytic activity through bimetallic cooperation.

Furthermore, the kinetics of electrocatalytic CO₂ reduction of NiNi(bphpp) and Ni(hbpp) were explored to understand the specific role of the secondary Ni site. The catalytic peak current (i_cat) of NiNi(bphpp) showed a linear relationship with its concentration (Figure S41), indicating the first-order kinetics of NiNi(bphpp). Similarly, a linear relationship was observed between i_cat and the concentration of Ni(hbpp) (Figure S42). These findings suggest that both complexes facilitate CO₂ reduction as single molecules within the concentration range of 0–1.0 mM. However, the kinetic orders of CO₂ in the catalytic processes of the two catalysts differed markedly. Within the saturated range of CO₂ in DMF, the catalytic process of NiNi(bphpp) exhibited two distinct kinetic order regimes (Figures 4a and S43). At CO₂ concertation ([CO₂]) below 0.04 M, i_cat was directly proportional to the concentration of CO₂, indicating second-order kinetics. In contrast, i_cat correlated with the square root of CO₂ concentration ([CO₂]^1/2) between 0.04 and 0.23 M (Inset of Figure 4a), suggesting first-order kinetics. In comparison, in the saturated range of CO₂, the catalytic process of Ni(hbpp) followed zero-order kinetics (Figures 4b and S44). Considering the three key steps in the catalytic process of CO₂ reduction disproportionation (i.e., CO₂ binding, C–O bond activation, and C–O bond cleavage, eqs S5–S7),^8,111,123 it appears that NiNi(bphpp) presents a relatively low barrier of C–O bond cleavage; the CO₂ binding and C–O bond activation processes collectively dictate the catalytic rate at lower CO₂ concentration (eq S12), while C–O bond activation, which has a higher barrier, becomes more dominant as the concentration of CO₂ increases (eq S13). Conversely, the zero-order kinetics for Ni(hbpp) concerning [CO₂] implies an extraordinarily high barrier of C–O bond cleavage, as CO₂ is not involved in this step (eq S14). Integrating with DFT study (discussed below and in Section 5 of the SI), the kinetic results strongly indicate that the secondary Ni site plays a private role in facilitating C–O bond activation and cleavage, thereby lowering the barriers in the catalytic cycle and potentially alerting the rate-determining step of the reaction. The augmented catalytic capability conferred by the well-integrated secondary Ni site was corroborated by the combined product analysis and the TOF and TON values. Additionally, a mixing experiment was conducted for further validation. Mixing Ni(hbpp) with an equivalent of Ni(AcO)₂ under identical electrochemical conditions does not result in any significant enhancement of the catalytic current (Figure S45).

Figure 4.

Relationship between i_cat and [CO₂] of NiNi(bphpp) (a) and Ni(hbpp) (b) in the [CO₂] range of 0–0.23 M. Inset: the relationship between i_cat and [CO₂]^1/2 of NiNi(bphpp) for the one-order area in (a).

Computational Studies

We further investigated the role of the secondary Ni site in enhancing the catalytic reduction of CO₂ using DFT calculations. The Gibbs free energy diagram (Figure S46) illustrates that CO₂ binding and subsequent chemical steps are facilitated following the formation of Ni^INi^I(bphpp^–•) species 4 (Figure 5a). Under our inference, CO₂ binds readily to the Ni1 site, which is coordinated by reduced phenanthroline, thereby exhibiting increased nucleophilic character. This significant property is demonstrated by the low energy barrier associated with the reaction. Specifically, the Ni1–C1 bond forms via transition state TS1, with a remarkably low barrier of 3.8 kcal/mol, while the formation of the Ni2–C1 bond has a considerably higher barrier of 14.8 kcal/mol relative to species 4 (Figure S52), resulting in the generation of Int1 (Figures 5b and S53). Interestingly, the Ni-CO₂ species within Int1 presents a unique geometry, with a C1–O1 and C1–O2 distance of 1.21 Å, a Ni1–C1 distance of 2.094 Å, and an O1–C1–O2 angle of 146.9°. These parameters differ markedly from the previously reported structures of Ni^II-η¹-CO₂^2–-κC species which exhibit longer C–O distance (1.24–1.25 Å), shorter Ni–C distance (1.91–1.95 Å), and smaller angle of O–C–O (128–129°).^130,131 Hence, Int1 represents a transitional configuration between Ni^II–CO₂ and Ni^II–CO₂^2–, presumably a Ni^II–CO₂^–• species. Moreover, during the formation of Int1, the spin population on Ni1 increases from 0.98 to 1.17, and the overall spin population on CO₂ increases to −0.363. This antiferromagnetic coupling with Ni1 gives rise to the distinct Ni^INi^II(bphpp^–•)(CO₂^–•) species. Subsequently, Int1 can attract another CO₂ molecule due to the impending nucleophilic attack of the O1 on C2, which carries an emerging negative charge, leading to the generation of intermediate Int2 (Figures 5c and S53), where the O1–C2 distance is 2.79 Å.

Figure 5.

Spin population analysis (isovalue of 0.01) of 4 (a), Int1 (b), Int2 (c), TS2 (d), and Int3 (e) (Phen = phenanthroline) by Multiwfn¹²⁸ and VMD.¹²⁹ Green and blue isosurfaces represent α and β electrons, respectively. Spin population values are shown in italics. (f) The bimetallic cooperation mode of C–O bond activation and cleavage regulated by NiNi(bphpp).

Int2 subsequently initiates the critical C2–O1 bond formation between the CO₂^–• moiety and the second CO₂ with the notable assistance of Ni2. Within this cooperative context, the incoming CO₂ molecule can readily approach the reduced CO₂^–• and activate the C1–O1 bond through transition state TS2 (Figures 5d and S54). The spin population of Ni1 increases from 1.21 to 1.59, while the spin population of phenanthroline decreases from 0.74 to 0.10, indicating an electron transfer from the phenanthroline ligand to the CO₂^–• moiety. Consequently, the resulting CO₂^2– unit can nucleophilically attack the adjacent CO₂ molecule with a relatively low barrier of 12.4 kcal/mol to Int2 (Figure 6). Following this cooperative binding of activated CO₂ with additional CO₂, the OCOCO₂^2– intermediate (Int3) is formed (Figures 5e and S55). Notably, the spin population of Ni2 increases from 0.97 in TS2 to 1.73, whereas that of phenanthroline reverts to 0.56. This marked change in spin populations indicates an electron transfer from the Ni2(I) site to the phenanthroline, resulting in the formation of Ni2(II), which serves to stabilize the OCOCO₂^2– anion with its Lewis acid character and culminates in the formation of the corresponding Ni^IINi^II(bphpp^–•)(OCOCO₂^2–) species (Int3). In addition, the C1–O1 bond length gradually increases from 1.21 Å (Int1) to 1.22 Å (Int2), and finally to 1.44 Å (Int3), indicative of a typical C–O single bond. This bond activation process comprehensively corroborates the essential cooperative functions of the secondary Ni site (Figure 5f), which (a) acts as an electronic reservoir by storing the reduction equivalents, (b) facilitates the spatial approach of another CO₂ molecule as an electron mediator, thereby enabling electron transfers from reductive phenanthroline moiety to CO₂^–• moiety, and also from Ni2 to phenanthroline, and (c) stabilizes the OCOCO₂^2– anion as a Lewis acid site.

Figure 6.

Calculated Gibbs free-energy diagrams of NiNi(bphpp) (blue line) and Ni(hbpp) (black line) starting at Int1 and Int1_S at the B3LYP-D3/6-311+G(2df,2p)||B3LYP-D3/6-31G(d,p) level.

Subsequently, the OCOCO₂^2– unit within Int3 can be converted into more thermodynamically stable species CO and CO₃^2– due to the cleavage of the C1–O1 bond. The CO–CO₃^2– cleavage occurs via a highly favorable transition state TS3 (Figure S56), which is characterized by a comparatively lower barrier of 8.3 kcal/mol. This leads to the formation of a transient intermediate, termed Int4 (Figure S57). The spin populations on Ni1, Ni2, and phenanthroline (1.26, 1.69, and 0.46, respectively) remain relatively unchanged from those of Int3, indicating a direct C–O cleavage process that results in the formation of the Ni^IINi^II(bphpp^–•)(CO)(CO₃^2–) complex. This facile cleavage of the C–O bond is attributed to the Lewis acid character of Ni2, which also stabilizes the CO₃^2– unit (Figure 5f). Two pathways for the release of CO and CO₃^2– were considered, and both are considered plausible (Figure S46). Initially, the direct release pathway of either CO₃^2– or CO is endergonic by 17 or 0.6 kcal/mol, leading to the generation of Int5 or Int5′ (Figure S58), respectively. However, when coupled with a one-electron reduction, both processes become exergonic by −21.8 and −8 kcal/mol, resulting in the formation of Int6 and Int6′ (Figure S59), respectively. Although the prior release of CO₃^2– after a one-electron reduction is thermodynamically favorable, the subsequent reductive release of CO to complete the catalytic cycle is challenging due to the reduction potential of −2.12 V. In contrast, the reductive release of CO₃^2– from Int6′ is facilitated by a lower potential of −1.52 V and returns to species 4.

The importance of the secondary Ni site is further illustrated by the catalytic process of Ni(hbpp), denoted as 1_S. In contrast to 1, the phenanthroline in Ni(hbpp) undergoes two consecutive one-electron reductions, resulting in the generations of 2_S and 3_S, respectively (Figure S60). The spin population of phenanthroline rises from 0 to 0.79 and then to 1.51, while that of Ni remains almost unchanged. The binding of CO₂ occurs via transition state TS1_S after the formation of 3_S with a low barrier of 2.2 kcal/mol (Figures S47 and S61). Similar to NiNi(bphpp), this process involves the one-electron activation of CO₂ with the generation of Ni^II(hbpp^–•)(CO₂^–•), labeled as Int1_S, which has a relatively high reduction potential of −2.56 V (Figure S62). Thus, we inferred that Int1_S directly forms the C2–O1 bond with the second CO₂ molecule via transition state TS2_S (Figures 6 and S63), resulting in the generation of the five-membered ring intermediate, Int2_S (Figure S64). Without the cooperation of the secondary Ni site, the barrier of this bonding process is calculated to be 18.7 kcal/mol, which is 2.8 kcal/mol higher than that of NiNi(bphpp) (Figure 6). Subsequently, Int2_S undergoes C1–O1 cleavage via transition state TS3_S (Figure S65), followed by the departure of CO to generate Ni^II(hbpp)(CO₃^2–) intermediate Int3_S (Figure S66), which is exergonic by 23 kcal/mol (Figures 6 and S47). The barrier calculated for TS3_S is as high as 21.3 kcal/mol, which shows an obvious difference from the dinickel catalyst. For the transformation of Int3 to Int4 via TS3, Ni1(II) coordinated with reductive phenanthroline exhibits a certain nucleophilic character, which stabilizes the CO unit during C–O cleavage. In addition, Ni2(II) also stabilizes the CO₃^2– unit as a Lewis acid. This bimetallic cooperation results in a much lower barrier for the formation of Int4. However, compared with the only C–O bond cleavage in TS3, TS3_S involves both C–O bond cleavage and CO dissociation on a single Ni^II site, which is undoubtedly more kinetically unfavorable. The high C–O bond cleavage barrier for TS3_S also supports the zero-order kinetics of the catalytic process toward [CO₂]. Furthermore, the energetic span model (detailed in SI) is utilized to better describe the overall catalytic rate through the Gibbs free energy diagrams (Figures S46 and 47).^50,132,133 According to the model equation eq S15, the overall kinetic barrier (δE_span) for the NiNi(bphpp) catalytic cycle is calculated to be 19.3 kcal/mol, which is significantly lower than the δE_span value of 22.1 kcal/mol for the catalytic cycle of Ni(hbpp). Using eq S16, the calculated TOF for NiNi(bphpp) catalysis is expected to be significantly higher than that of Ni(hbpp) catalysis. These theoretical results are completely consistent with the electrochemical properties, product characterizations, and catalytic kinetics of NiNi(bphpp) and Ni(hbpp).

Both controlled experiments and DFT calculations have substantiated the pronounced enhancement of redox-active phenanthroline in facilitating CO₂ activation through coordinated Ni site, which avoids the generation of high-energy Ni⁰ and contributes to the one-electron reduction of CO₂ to CO₂^–•. Furthermore, the distinct roles of both Ni sites have been elucidated. Notably, the secondary Ni site actively aids in the activation of the C–O bond through “pre-coordination” with another CO₂ molecule as an important electron mediator. Additionally, the Lewis acid character of the secondary Ni site markedly facilitates the C–O bond cleavage, thereby promoting the entire catalytic process. This work emphasizes that bimetallic cooperation and redox-active ligand assistance are effective design strategies for the development of CO₂ reduction catalysts, even for the multielectron/proton redox catalysis of the other small molecules.^134−141

Conclusions

Inspired by the catalytic mechanism of NiFe-CODH, this work explores the development of a binuclear Ni catalyst, NiNi(bphpp), which enables bimetallic synergistic catalysis for CO₂ activation and catalytic conversion. By incorporating redox-active phenanthroline and a secondary Ni site, the bimetallic NiNi(bphpp) catalyst enables a three-electron reduction at a lower potential, leading to the generation of Ni^INi^I(bphpp^–•) species. Subsequently, facile one-electron activation of CO₂ occurs, followed by activation and cleavage of the C–O bond through the coordination of a second CO₂ molecule. The involvement of the secondary Ni(I) site acts as an effective electron mediator, promoting electron transfer to the second CO₂ molecule and facilitating C–O bond activation. Furthermore, after formation of the [OCOCO₂^2–] species, the secondary Ni(II) site exhibits Lewis acid characteristics, significantly aiding in C–O bond cleavage. These findings shed light on the distinct roles played by bimetallic sites in CO₂ reduction catalysis, providing insights into the nature of bimetallic cooperation. This work proposes new design strategies for catalysts in bimetallic synergistic multielectron catalytic processes. Based on this strategy, the development of the Ni–Fe heterobimetallic catalyst is in progress for efficient CO₂ reduction.

Methods

All of the electrochemical experiments were conducted with a three-electrode system using a CHI-660E electrochemical workstation at room temperature. A glassy carbon electrode (GCE, 0.07 cm²) was used as the working electrode, the counter electrode was a Pt wire, and Ag/AgNO₃ (0.1 M AgNO₃ in DMF) was used as the reference electrode and calibrated versus Fc^+/0. All of the reduction potentials were obtained from the DPVs of corresponding complexes. The gas above the electrolytic cell was drawn by a 1 mL Hamilton syringe and injected into the GC. CPE was conducted with 10 mL of 0.1 M TBAPF₆/DMF solution containing NiNi(bphpp) or Ni(hbpp) (0.5 mM) in a sealed standard three-electrode cell with stirring. A foamy carbon electrode (1.0 × 1.0 × 0.6 cm) was used as the working electrode. The Faraday efficiencies were calculated by the equation below:

where CO_exp represents the CO content above the electrolytic cell which was determined by gas chromatography, and CO_CPE represents the CO content calculated from the total charge transfer (Q) during CPE (Figure S31).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.4c00047.

• Additional experimental details, materials, methods, complex synthesis (¹H NMR and ESI-HRMS), magnetic property characterizations (Evans magnetic moment, EPR, and SQUID), X-ray crystallography, electrochemical studies, kinetic analysis, DFT calculations, and so on (PDF)

Accession Codes

CCDC 2257038, 2257041, 2257045, and 2257048 contain supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, by emailing data_request@ccdc.cam.ac.uk, or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: + 44 1223 336033.

The authors declare no competing financial interest.