Highly efficient electrocatalytic CO₂ reduction by a Cr^III quaterpyridine complex

Significance

CO₂reduction into CO, CH₄, etc., can lower the atmospheric CO₂concentration and provide carbon-neutral energy simultaneously, attracting scientists to design high-performance catalytic systems to facilitate this process. Molecular catalysis based on metal complexes is appealing for their well-defined structures which allow the rational optimizations by synthetic efforts and facile characterization of reaction intermediates. The chromium-based molecular catalysts for CO₂reduction are still rare, leading to their underexplored design strategies and mechanistic insights. Herein, we deliberately used a quaterpyridine ligand to support the Cr^IIIcenter, affording a very efficient molecular electrocatalyst which outperforms the Cr-based precedents. Based on this promising catalyst, we have characterized a series of key intermediates to reveal its mechanism by various experimental techniques and calculations.

Abstract

Design tactics and mechanistic studies both remain as fundamental challenges during the exploitations of earth-abundant molecular electrocatalysts for CO₂ reduction, especially for the rarely studied Cr-based ones. Herein, a quaterpyridyl Cr^III catalyst is found to be highly active for CO₂ electroreduction to CO with 99.8% Faradaic efficiency in DMF/phenol medium. A nearly one order of magnitude higher turnover frequency (86.6 s⁻¹) over the documented Cr-based catalysts (<10 s⁻¹) can be achieved at an applied overpotential of only 190 mV which is generally 300 mV lower than these precedents. Such a high performance at this low driving force originates from the metal–ligand cooperativity that stabilizes the low-valent intermediates and serves as an efficient electron reservoir. Moreover, a synergy of electrochemistry, spectroelectrochemistry, electron paramagnetic resonance, and quantum chemical calculations allows to characterize the key Cr^II, Cr^I, Cr⁰, and CO-bound Cr⁰ intermediates as well as to verify the catalytic mechanism.

CO₂ reduction is a versatile reaction for both relieving global warming and providing carbon-neutral, renewable fuels (1–3). To address the efficiency and selectivity issues in CO₂ reduction, molecular catalysts, mainly transition metal complexes, are promising candidates; their well-defined structures are amenable to rational optimizations via synthetic means as well as kinetic and spectroscopic investigations on the catalytic intermediates that govern their mechanisms (4–7). Moreover, the molecular catalysts based on earth-abundant metals are appealing for their ease in lowering the expense for large-scale applications (8–10). In this context, major development has been made in designing the molecular catalysts featuring 3d transition metal centers like Mn (7, 11), Fe (12, 13), Co (14, 15), Ni (16–18), and Cu (6), while the utilizations of Cr (19–22), along with its congeners Mo (23–25) and W (25), are rather underexplored. The pioneering works were mainly contributed by Machan’s lab, where a family of tetradentate ligands consisting of mixed N/O donors and polypyridyl backbones has been developed to chelate Cr^III for electrocatalytic CO₂ reduction (19–22). However, a more indicative design principle for the Cr-based molecular catalysts in CO₂ reduction is still elusive, demanding novel design strategies and in-depth mechanistic investigations. We herein employ a quaterpyridyl ligand to coordinate the Cr^III center, affording [Cr(qpy)(Cl)₂]Cl (CrQPY, qpy = 2,2′:6′,2″:6″,2‴-quaterpyridine; Fig. 1 A, Inset) as a high-performance molecular catalyst for CO₂ electroreduction to CO with near 100% Faradaic efficiency. Impressively, CrQPY exhibits a record-high turnover frequency (TOF) of 86.6 s⁻¹ at a 300 mV lower applied overpotential in comparison with the Cr-based precedents (Table 1). Detailed electrochemical, spectroscopic, and computational studies have been performed to characterize the electronic structures of the key reactive intermediates, including Cr^II, Cr^I, Cr⁰, and Cr⁰-CO species which were rarely investigated before for the Cr-based molecular electrocatalyst in CO₂ reduction. These mechanistic insights indicate that such a large TOF at this low overpotential of CrQPY is attributable to the use of N4 quaterpyridyl ligand which stabilizes low-valent intermediates and acts as an ideal electron reservoir, showing advantages over the Cr-based precedents with mixed N/O donors.

Fig. 1.

(A) CVs of 0.5 mM CrQPY under Ar (black), CO₂ (gold), or CO₂ with 1.9 M PhOH (violet). Inset: Chemical structure of CrQPY. (B) CVs of 0.5 mM CrQPY with varying [PhOH] under CO₂. (C) Plot of current at −1.95 V versus the square root of [PhOH]. (D) CVs of varying [CrQPY] with 1.9 M PhOH under CO₂. (E) Plot of current at −1.95 V versus [CrQPY]. General CV conditions: 3 mm GC disk working electrode, 0.1 M ⁿBu₄NPF₆ DMF solution at 0.05 V s⁻¹ scan rate. (F) CPE at −2.00 V with 0.5 mM CrQPY with 1.9 M PhOH at a GC plate (0.5 cm²) under stirring, giving Q = 2.05 C and a CO yield of 10.4 μmol during 30 min.

Table 1.

Performances of Cr-based molecular electrocatalysts for CO₂ reduction to CO^*

Entry	Catalyst	Medium	Eapp (V)	FECO (%)	TOF (s−1)†	TON†
1 (this work)	CrQPY	DMF/PhOH (1.9 M)	−2.00	99.8 ± 4.0	86.6	1.56 × 105
2 (19)	Cr(tBudhbpy)Cl(H2O)	DMF/PhOH (0.62 M)	−2.10	96 ± 8	n.a.	n.a.
3 (20, 22)		DMF/PhOH (0.6 M)	−2.30	111 ± 14	7.12	1.42 × 105
4 (21)	Cr(tpytbupho)Cl2	DMF/PhOH (0.6 M)	−2.30	93 ± 7	1.82	3.40 × 104
5 (20)	Cr(tBudhtBubpy)Cl(H2O)	DMF/PhOH (0.6 M)	−2.30	95 ± 8	9.29	1.86 × 105
6 (22)	Cr(tBudhphen)Cl(H2O)	DMF/PhOH (0.6 M)	−2.30	101 ± 3	4.90	9.80 × 104

^*(^tBudhbpy)²⁻ = 6,6′-([2,2′-bipyridine]-6,6′-diyl)bis(2,4-di-tert-butylphenolate), (tpy^tbupho)⁻ = 2-(2,2′:6′,2″terpyridin-6-yl)-4,6-di-tert-butylphenolate, (^tbudh^tbubpy)²⁻ = 6,6′-(4,4′-di-tert-butyl-[2,2′-bipyridine]-6,6′-diyl)bis(2,4-di-tert-butylphenolate), (^tbudhphen)²⁻ = 6,6′-(1,10-phenanthroline-2,9-diyl)bis(2,4-di-tert-butylphenolate). Error bars are the SD from three parallel experiments.

^†Calculated via Eqs. 1–3.

Results

Simple mixing qpy ligand with CrCl₃·6H₂O in hot ethanol afforded the gray microcrystalline CrQPY in high yield. Then, the redox properties of CrQPY were initially examined by cyclic voltammetry at a 3 mm glassy carbon (GC) disk electrode in dry DMF. The potentials were calibrated to ferrocenium/ferrocene (Fc⁺/Fc) unless otherwise stated. The cyclic voltammogram (CV) of 0.5 mM CrQPY under Ar features three redox couples at −1.07, −1.50, and −1.85 V (Fig. 1A), assignable to formal Cr^III/II, Cr^II/I, and Cr^I/0 redox events, respectively. The first and third couples are reversible, while the Cr^II/I one is quasi-reversible for its large peak-to-peak separation (>120 mV at 0.05 V s⁻¹ scan rate). But the scan rate–dependent CVs indicate that this couple is still freely diffusing (SI Appendix, Fig. S1) (26).

Under CO₂, the first two redox couples remained unaltered upon scanning to −1.7 V (SI Appendix, Fig. S2), implying the absence of CO₂ binding. When scanned to −2.1 V, an enhanced current was initiated at the Cr^I/0 reduction wave, suggesting catalytic CO₂ reduction (15). Particularly, the formal Cr^II/I oxidation wave negatively shifted from −1.44 to −1.67 V (indicated in Fig. 1A), inferring the formation of a Cr^I-CO species and thus the CO evolution, reminiscent of an iron porphyrin catalyst (12). Further addition of phenol (PhOH) induced a much larger current at an onset potential of ca. −1.7 V, showing the CO₂ reduction facilitated by proton source (14). It can be noticed that the first redox couple underwent a positive shift with addition of PhOH under either Ar (SI Appendix, Fig. S3) or CO₂, indicating the interaction between PhOH and CrQPY (21). The catalytic current shows a linear correlation with the square root of [PhOH] until reaching a plateau with >ca. 1.6 M PhOH, indicating the typical saturation kinetics expected for catalytic reactions and the first-order kinetics of the catalysis in [PhOH] (Fig. 1 B and C) (11, 17). The catalytic current also increased linearly with higher [CrQPY], showing that the electrocatalysis is first order in the catalyst concentration (Fig. 1 D and E) (17).

The electrocatalytic performance of CrQPY was then evaluated by controlled-potential electrolysis (CPE) in DMF/PhOH (1.9 M) electrolyte, exhibiting 99.8% ± 4.0% Faradaic efficiency for CO formation at −2.00 V (Table 1, entry 1). A stable current density (j; Fig. 1F) of ca. 1.4 mA cm⁻² was observed, and a total turnover number (tTON) value of ca. 7 was calculated with the number of catalyst molecules in the bulk electrolyte within 30 min. This tTON value, however, underestimates the catalytic performance as only a small fraction of catalyst molecules that interact with the electrode could contribute to catalysis (27). Therefore, with the above CPE data (j = 1.4 mA cm⁻²; E = −2.00 V), we applied Eq. 1 (28) to calculate the intrinsic kinetic constant (k_cat = 86.8 s⁻¹; see SI Appendix for details) of CrQPY based on the number of catalyst molecules in the diffusion layer of the cathode. A record-high TOF of 86.6 s⁻¹ could then be calculated via Eq. 2 (28), which is almost one order of magnitude higher than those pioneering Cr^III catalysts in DMF with respective optimal [PhOH] (4.90 ~ 9.29 s⁻¹; Table 1). It is also notable that the reported Cr catalysts generally demand applied potentials at −2.30 V in CPE (Table 1). In comparison, the complete CO formation at −2.00 V by CrQPY demonstrates its much lower overpotential [as low as 0.19 V (19); see SI Appendix for details] by 0.3 V under similar conditions. Such a low applied overpotential is tentatively comparable to the reported overpotentials of some pioneering molecular electrocatalysts based on first-row transition metals for CO₂-to-CO conversion, albeit with different benchmarking standard potentials (E⁰(CO₂/CO), such as an Fe^II quaterpyridine complex (240 mV; E⁰(CO₂/CO) = −1.41 V) (29), a Co^II tripodal complex (200 mV; E⁰(CO₂/CO) = −1.29 V) (15), a cationic Fe^III porphyrin (220 mV; E⁰(CO₂/CO) = −1.43 V) (30), etc.

$$j=\frac{D^{\frac{1}{2}}(2k_{\text{cat}}{)}^{\frac{1}{2}}F}{1+\text{exp}[\frac{F}{\mathit{RT}}(E-E_{\text{cat}})]},$$ [1]

$$\text{TOF}=\frac{k_{\text{cat}}}{1+\text{exp}[\frac{F}{\mathit{\text{RT}}}(E-E_{\text{cat}})]},$$ [2]

$$\text{TON}=\frac{k_{\text{cat}}}{1+\text{exp}[\frac{F}{\mathit{\text{RT}}}(E-E_{\text{cat}})]}\times \mathrm{t}.$$ [3]

Meanwhile, a remarkable TON of 1.56 × 10⁵ within 30 min could also be estimated via Eq. 3 (28), comparable to the reported instances (Table 1) and thus suggesting its good stability. However, the more sustained CPE was limited possibly by the CO poisoning of the catalyst suggested by the subsequent mechanistic studies. After CPE, the reuse of the used working electrode (no visible deposit) into a CO₂-saturated, catalyst-free electrolyte for CPE at −2.0 V could not produce any CO, confirming the molecular nature in electrocatalytic CO₂ reduction. Overall, the above results display the high performance and molecular nature of CrQPY in the CO₂ electroreduction.

To shed light on the catalytic mechanism of CrQPY, UV–Vis spectroelectrochemistry (UV–Vis-SEC) was applied to identify its catalytic intermediates in DMF. We collected spectra at the reduction potentials where Cr^III, Cr^II, Cr^I, and Cr⁰ should be the dominating species, under Ar, CO₂, and CO (SI Appendix, Figs. S4–S7), respectively. Interestingly, the spectra of the formal Cr⁰ state under CO₂ and CO are very similar, and different from the one obtained under Ar. This outcome suggests the formation of analogous species under both CO₂ and CO, most possibly CO-bound species. The CO-binding behavior is consistent with the CV of CrQPY under CO (SI Appendix, Fig. S8). A notable positive shift (60 mV) of the formal Cr^I/0 redox couple was observed in the CV under CO in comparison to that under Ar, giving a thermodynamic CO-binding constant (18) (K_CO) of 1.87 × 10³ M⁻¹ (see SI Appendix for details), confirming the CO binding (31) at the Cr⁰ species.

To further examine the CO-bound species, we conducted the differential infrared SEC (IR-SEC; Fig. 2 and SI Appendix, Fig. S9) on CrQPY in DMF under CO and CO₂. By applying −1.90 V, the formal Cr⁰ species under CO exhibits three CO stretching signals at 2,005, 1,895, and 1,831 cm⁻¹ in its IR spectrum (blue trace in Fig. 2B). The IR spectrum with CO₂ revealed similar signals at 2,004, 1,888, and 1,829 cm⁻¹ (red trace in Fig. 2B), further suggesting the formation of the same CO-bound species under either CO or CO₂. Moreover, the signals displayed an overall shift to lower wavenumbers when CO₂ was replaced with ¹³CO₂, verifying that these peaks were arising from the reaction with CO₂ (SI Appendix, Fig. S10). The small signal at 2,005/2,004 cm⁻¹ is attributable to Cr(CO)_x species (32). The other two CO signals can be assigned to the expected [Cr(qpy)(CO)₂]⁰ species. This assumption is well consistent with the DFT-calculated IR spectrum of the singlet [Cr(qpy)(CO)₂]⁰ intermediate, which reveals C=O stretching modes at 1,893 and 1,838 cm⁻¹ (violet traces in Fig. 2B; see SI Appendix for details), whereas its triplet analog shows peaks at 1,878 and 1,975 cm⁻¹ (SI Appendix, Fig. S11).

Fig. 2.

(A) UV–Vis-SEC spectra of 2.0 mM of CrQPY in DMF under Ar (green), CO (blue), and CO₂ (red) at the Cr⁰ state. (B) FT-IR-SEC spectra of 2.0 mM of CrQPY in DMF under CO (blue) and CO₂ (red), with DFT-simulated IR spectrum of singlet [Cr(qpy)(CO)₂]⁰ (violet). (C) TDDFT-simulated UV–Vis spectra of singlet [Cr(qpy)]⁰ (green) and singlet [Cr(qpy)(CO)₂]⁰ (blue). (D) X-Band EPR spectra of frozen 1.0 mM DMF solutions of Cr^III (gray), Cr^II (orange), Cr^I (green), Cr⁰ (red), and [Cr(qpy)(CO)₂]⁰ (blue) species of CrQPY at 10 K.

Moreover, the electronic absorption spectra of [Cr(qpy)]⁰ and [Cr(qpy)(CO)₂]⁰ intermediates were predicted at the time-dependent DFT (TDDFT; SI Appendix, Tables S1 and S2) level of theory. Notably, quantum chemical calculations on the two intermediates suggest a strong mixing of d(Cr) and π* orbitals of qpy ligand and thus the involvement of ligand-centered reduction, as exemplified by the electron density analysis on the singlet [Cr(qpy)]⁰ (SI Appendix, Fig. S12). The calculated UV–Vis spectrum of the singlet [Cr(qpy)]⁰ features a main metal-to-ligand charge transfer (¹MLCT) excitation into S₈ at 808 nm, in good agreement with the experimental data (green traces in Fig. 2 A and C). Meanwhile, the computed spectrum of the singlet [Cr(qpy)(CO)₂]⁰ is also in good agreement with the experimental data, both featuring a broad absorption feature at ca. 630 nm, which mainly stems from three MLCT transitions (S₆, S₇ and S₈; blue traces in Fig. 2 A and C). On the other hand, the possible open-shell, triplet species of [Cr(qpy)]⁰ and [Cr(qpy)(CO)₂]⁰ were excluded computationally, as their calculated vibrational and UV–Vis signals (SI Appendix, Figs. S11 and S13) are in stark contrast to the measure ones.

The electronic structures of Cr^III and the electrochemically generated species at different oxidation states were further studied using X-band electron paramagnetic resonance (EPR) spectroscopy in frozen DMF solutions. The EPR spectrum of 1.0 mM initial Cr^III at 10 K shows a rhombic anisotropic signal, matching a high-spin S = 3/2 octahedrally coordinated Cr^III complex (Fig. 2D) (33, 34). This is in accordance with the DFT-calculated spin density of the initial Cr^III species which was determined as [Cr^III(qpy)Cl₂]⁺ by the mass spectrum of CrQPY in DMF solution (SI Appendix, Fig. S14), where the three unpaired α-electrons are distributed among the nearly degenerate t_2g orbitals, respectively (SI Appendix, Fig. S15A). This is also in accordance with the literature regarding the electronic configuration of related Cr^III complexes (35). The experimental EPR spectrum was simulated (see SI Appendix and SI Appendix, Table S3 for simulation details) with g-values of g_x,y,z [1.989, 2.0303, 1.964], a zero-field splitting D = 0.68 cm⁻¹ and a rhombicity factor E/D = 0.18 (SI Appendix, Fig. S16), within the range of reported systems (33, 34). Upon 1e⁻ reduction, the EPR signal vanished as expected for a formal Cr^II system with an even number of electrons. Further applying a potential at −1.6 V enabled a sharp EPR fingerprint of the formal Cr^I species with good agreement with the simulations in SI Appendix, Fig. S17; this can be attributed to an S = 1/2 signal located at a g-value of 1.984, close to the g-values of reported Cr^I systems (36). This S = 1/2 feature can be interpreted as a low-spin configuration in the metal center. Notably, the DFT results for this Cr^I intermediate, predicted as [Cr^I(qpy)]⁺·2Cl⁻, reveal a d⁴ species instead of a d⁵ character, as one unpaired β-electron is localized within a π* orbital (HOMO-1; SI Appendix, Fig. S15B). Finally, the EPR spectra of the Cr⁰ systems are silent under either N₂ or CO. Presumably, it is less likely to form high-spin Cr⁰ species as it demands additional energy for the spin transition from the low-spin Cr^I intermediate, as supported by the quantum chemically predicted IR and UV–Vis signals of the various spin species. Overall, the EPR measurements on CrQPY reveal the isolations of the EPR-silent Cr^II, Cr⁰ and [Cr(qpy)(CO)₂]⁰ intermediates as well as the EPR-active Cr^III and Cr^I species.

Based on the above mechanistic insights, a plausible mechanism for the CO₂-to-CO conversion catalyzed by CrQPY can be proposed in Fig. 3. The catalytic cycle is initiated by three times of 1e⁻ reduction of [Cr(qpy)Cl₂]⁺, forming the active [Cr(qpy)]⁰ species for CO₂ binding. The CO₂-adduct presumably experiences a proton-coupled electron transfer to generate a Cr-COOH species with PhOH as the proton source, which then undergoes C-OH cleavage to afford [Cr(qpy)(CO)]⁺ (5, 10). This Cr^I-CO adduct is not thermodynamically stable as it was suggested by the negatively shifted Cr^II/I oxidation wave under dry CO₂ (Fig. 1A) but not revealed by the SEC. Thus, [Cr(qpy)(CO)]¹⁺ can release CO to recover [Cr(qpy)]⁺ for the next cycle. Meanwhile, it can also bind to another CO with 1e⁻ reduction to afford the detected [Cr(qpy)(CO)₂]⁰ intermediate in which two CO molecules will dissociate to give [Cr(qpy)]⁰ into another catalytic cycle.

Fig. 3.

Proposed catalytic cycle for CO₂ reduction catalyzed by CrQPY.

Discussion

In summary, we here disclose the evaluations and mechanistic studies on a Cr^III quaterpyridine complex for electrocatalytic CO₂ reduction, which achieves 99.8% ± 4.0% Faradaic efficiency for CO production in DMF/PhOH medium. Impressively, a record-high TOF among the documented Cr^III catalysts has been achieved at an overpotential of merely 190 mV. Such a low overpotential is also comparable to some pioneering molecular catalysts featuring earth-abundant metals. These excellent performances can be ascribed to the use of four N donors instead of the mixed N/O ones in most reported Cr^III examples, which should better stabilize the low-oxidation-state species (37) for catalysis at lower overpotentials. Notably, the electronic structures of the key intermediates involved in CO₂ reduction have been revealed by electrochemistry, SEC, EPR, and quantum chemical calculations, which are scarcely documented but desirable for understanding the mechanisms of Cr-based molecular catalysts. The ligand-radical-based reactive Cr⁰ species further emphasizes the merits of the quaterpyridine ligand as a versatile electron reservoir with its extended π-delocalized network. The mechanistic results also indicate the CO-binding behavior which may induce the catalyst poisoning, while it can be circumvented by surface immobilization (38) within our future efforts. We believe that these findings provide a promising noble-metal-free molecular electrocatalyst for CO₂ reduction, along with valuable mechanistic insights for further rational design on the Cr-based catalysts.

Materials and Methods

Materials.

CrQPY was prepared following the previously reported method (39), and other chemicals were commercially available and used without further purification.

Methods.

Electrochemical measurements were carried out using an electrochemical workstation (CHI 660D). The generated gas samples in the head space were analyzed by an Agilent 7890B Series GC custom chromatograph equipped with thermal conductivity detector (TCD) and flame ionization detector (FID), and the products in the solution were analyzed by ¹H NMR (Bruker Avance III 500 MHz) following the protocol reported by Nocera et al. (40). UV–Vis spectra were collected on a Shimadzu UV-3600 spectrophotometer. FT-IR spectra were obtained on a Nicolet iS50 FT-IR spectrometer (Thermo Fisher). All experiments were operated at room temperature (24 to 25 °C) unless otherwise stated. All quantum chemical calculations determining structural and electronic properties were performed using the Gaussian 16 program (41).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.