您当前的位置： 首页 > 资源详细信息
网页快照

来源机构： 《自然杂志》 news Aug 13, 2019 328.19KB 重要新闻

Selectivity control of CO versus HCOO ? production in the visible-light-driven catalytic reduction of CO 2 with two cooperative metal sites | Nature Catalysis

Abstract.It is highly desirable to discover molecular catalysts with controlled selectivity for visible-light-driven CO2 reduction to fuels. In the design of catalysts employing earth-abundant metals, progress has been made for CO production, but formate generation has been observed more rarely. Here, we report a binuclear Co complex bearing a bi-quaterpyridine ligand that can selectively reduce CO2 to HCOO− or CO under visible light irradiation. Selective formate production (maximum of 97%) was obtained with a turnover number of up to 821 in basic acetonitrile solution. Conversely, in the presence of a weak acid, CO2 reduction affords CO with high selectivity (maximum of 99%) and a maximum turnover number of 829. The catalytic process is controlled by the two Co atoms acting synergistically, and the selectivity can be steered towards the desired product by simply changing the acid co-substrate. Access provided by Main.In principle, CO2 can be used as a renewable feedstock in photochemical devices for solar energy storage in the form of synthetic fuels or as a fuel precursor, such as CO for Fischer–Tropsch chemistry, CH3OH and CH4 (refs. 1,2,3). However, selective reduction of CO2 is a challenging and energetically demanding process, and so suitable catalysts are required. The use of molecular catalysts is a promising approach due to the easy and ready fine-tuning of the ligand structures achieved by controlling steric and electronic effects, and this may lead to highly efficient and selective processes. Known molecular catalysts include noble metal complexes such as Ru4,5,6,7, Ir8,9,10 and Re11,12,13,14,15 and earth-abundant metal complexes such as Co16,17,18,19, Ni20,21, Fe22,23,24,25, Mn26,27,28,29 and Cu30,31. Of the earth-abundant molecular photocatalysts reported, most can reduce CO2 to CO, but relatively few of them can produce HCOO− with high yield, and there are even fewer examples of complexes that can generate highly reduced hydrocarbons (beyond 2H+/2e− reduction products)25,32,33,34. Although CO is a useful chemical for the Fischer–Tropsch reaction with H2, formic acid has recently attracted some attention as a hydrogen storage material for liquid formic acid fuel cells or as feedstock for bacteria to produce alcohols as liquid fuels35,36,37,38.We have recently developed the mononuclear [Co(qpy)]2+ (qpy = 2,2′:6′,2′′:6′′,2′′′-quaterpyridine) as an efficient and selective catalyst for both photo- and electrocatalytic reduction of CO2 (ref. 39). The product is predominately CO (selectivity of >98%), with only trace amounts of formate or H2. Formate has been found in other molecular Co40 and Fe19 photosystems, but the selectivity or yield has been very low. Better results have been obtained by using Mn-based diamine complexes. Others have reported fac-Mn(bpy)(CO)3Br (bpy = 2,2′-bipyridine) as a competent photocatalyst for the reduction of CO2 to formic acid with a turnover number (TON) of 149 and selectivity of 85% in dimethylformamide (DMF)/triethanol amine (TEOA) with 1-benzyl-1,4-dihydronicotinamide as a sacrificial reductant26. A similar system using fac-Mn(CN)(bpy)(CO)3 as catalyst was investigated previously, and the results showed that the product distribution is affected by the nature of solvent29. The HCOOH/CO ratio is 18 and 0.42 in DMF/TEOA and CH3CN/TEOA, respectively.Inspired by the ubiquitous existence of multi-metallic complexes as catalytic centres of metalloenzymes in nature, such as the Ni–Fe-based carbon monoxide dehydrogenases (CODHs) that reversibly transform the CO2 into CO, we became interested in the design of bimetallic complexes as catalysts for CO2 reduction. In the case of CODHs, a NiI centre binds to the C atom of the CO2 while a FeII binds to one O atom of the substrate, playing the role of a Lewis acid to assist cleavage of the C–O bond. In this reaction, the two metals act in synergy for high selectivity41,42. There are only a few examples of efficient molecular multimetallic CO2 reduction catalysts in the literature43. A recent one is a binuclear cobalt cryptate catalyst that affords CO in high yield44. Changing one of the Co centres to Zn strongly boosts the photochemical CO2-to-CO conversion45. An Fe carbonyl cluster has also been developed recently for the selective electrochemical reduction of CO2 to formate in pH neutral water solutions46.Here, a binuclear Co complex bearing a bi-quaterpyridine ligand [Co2biqpy]4+ (Fig. 1, biqpy = 4,4′′′′-(2,7-di-tert-butyl-9,9-dimethyl-9H-xanthene-4,5-diyl) di-2,2′:6′,2′′:6′′,2′′′-quaterpyridine) was prepared and used as a catalyst for the photoinduced reduction of CO2 in acetonitrile (MeCN) solutions. Formate could be produced with high TON and excellent selectivity provided no acid was added to the reaction mixture. Conversely, when phenol was used as a co-substrate, the selectivity for CO2 reduction remained high, but CO was almost exclusively produced, with high TON. It appears that [Co2biqpy]4+ is a highly efficient and selective molecular catalyst for the visible-light-induced reduction of CO2, with two cooperative metals controlling catalytic activity and selectivity.Fig. 1: Structure of compounds used in this study.Cobalt complexes [Co2(biqpy)]4+ (1) and [Co(qpy)]2+ (2), molecular sensitizers ([Ru(phen)3]2+, Pheno) and sacrificial reductant 1,3-dimethyl-2-phenylbenzimidazoline (BIH).Full size image .Results.Preparation and characterization of catalyst 1.Ligand 4,4′′′′-(2,7-di-tert-butyl-9,9-dimethyl-9H-xanthene-4,5-diyl) di-2,2′:6′,2′′:6′′,2′′′-quaterpyridine (biqpy) was synthesized by the coupling of 4-bromo-2,2′:6′,2′′:6′′,2′′′-quaterpyridine with 2,7-di-tert-butyl-9,9-dimethylxanthene-4,5-diboronic acid, using Pd(PPh3)4 as a catalyst. Treatment of biqpy with excess CoCl2 in MeCN affords the binuclear compound Co2(biqpy)Cl4 in high yield. Electrospray ionization mass spectrometry (ESI-MS) exhibits peaks at m/z 364.0, 564.6 and 1,161.5 (Supplementary Fig. 1), which are assigned to the triply-charged [Co2(biqpy)Cl]3+, doubly-charged [Co2(biqpy)Cl2]2+ and singly-charged [Co2(biqpy)Cl3]+ ions, respectively. Three of the Cl− ions in Co2(biqpy)Cl4 were easily replaced by ClO4− in methanol, affording [Co2(biqpy)Cl(MeOH)(H2O)](ClO4)3, which was structurally characterized by X-ray crystallography (Fig. 2 and Supplementary Tables 1 and 2). Both Co1 and Co2 cobalt atoms are coordinated by four N atoms of biqpy in the equatorial plane, with a bridging Cl− atom. The Co1–Cl–Co2 angle is 136.02(7). The other axial position around Co1 (Co2) is occupied by a water (methanol) molecule. The ESI-MS of [Co2(biqpy)Cl(MeOH)(H2O)](ClO4)3 in MeCN exhibits peaks at m/z 264.3 and 586.4, which are respectively assigned to the quadruply-charged [Co2L]4+ and doubly-charged [Co2(biqpy)(OH)(ClO4)]2+ (Supplementary Fig. 2), indicating that the μ-Cl atom between Co1 and Co2 is labile in solution.Fig. 2: Crystal structure of [Co2(biqpy)Cl(MeOH)(H2O)]3+.ORTEP drawing of the complex. Thermal ellipsoids are drawn at 50% probability. Hydrogen atoms (except O–H) and solvent molecules are omitted for clarity.Full size image .Visible-light-driven reduction of CO2 into formate.Photocatalytic CO2 reduction experiments were initially performed in a three-component system containing 1 as catalyst, Ru(phen)3Cl2 (phen = 1,10-phenanthroline) as photosensitizer, a sacrificial donor (see below) and MeCN as solvent. Ru(phen)3Cl2 was employed due to the photochemical inactivity of [Ru(phen)2(H2O)2]2+ from the photolabilization of the phen ligand of [Ru(phen)3]2+ (note that [Ru(bpy)2(H2O)2]2+ originated from the [Ru(bpy)3]2+ decomposition may be a catalyst for CO2 reduction). Regarding the solvent, although DMF was found to favour the formation of formate by the aforementioned fac-Mn(CN)(bpy)(CO)3, it may be non-innocent due to possible hydrolysis producing formate during photocatalysis, especially in the presence of trace amounts of amine or water47,48, and MeCN was instead preferred. We started our investigation with triethylamine (TEA) as electron donor. In a typical run, a glass flask containing a mixture of 2 ml of CO2-saturated MeCN, 1 (0.05 mM), Ru(phen)3Cl2 (0.2 mM) and TEA (20%, vol%) was irradiated using a blue light-emitting diode (LED) strip (460 nm) at 24 °C. Formate was analysed by ion chromatography (IC) while the gaseous products and other volatile products were analysed by gas chromatography (GC, see Supplementary Methods).As shown in Fig. 3 and Table 1 (entry 1), formate was produced with a TON of 110 and selectivity of 92%. Only a small amount of H2 (TON of 9) was obtained (Fig. 3a). Control experiments under Ar only furnished H2 (TON of 22), while no products were detected in the absence of a catalyst (Supplementary Table 3, entries 2 and 3) as well as when the non-metallated ligand was used instead of 1 (Supplementary Table 3, entry 8). Note that the amount of formate using 2 as catalyst was only about one-third of that with 1 (Supplementary Fig. 3 and Supplementary Table 4, entry 1; see the mechanistic studies of the CO2 reduction section for further discussion). When CoCl2 was used as catalyst, neither formate nor CO was detected after irradiation for 23 h (Supplementary Table 3, entry 7), indicating that the catalytic activity is not due to demetallation of the Co complex. With 1,3-dimethyl-2-phenylbenzimidazoline (BIH) as an efficient two-electron donor, the catalytic performances could be optimized. In the presence of 0.025 M BIH, the TON and selectivity for formate were boosted to 386 and 96.5% respectively (Fig. 3b and Table 1, entry 2). Only trace amounts of CO and H2 were obtained (TONs of 8 and 6 for CO and H2, respectively). Control experiments conducted under Ar gave trace amounts of formate (TON of 7) and H2 (TON of 1), while a negligible amount of formate was detected in the absence of catalyst. The TON for formate could be further increased to 821 when TEA/BIH was replaced by TEOA/BIH (Supplementary Fig. 5 and Table 1, entry 3). However in this case the selectivity for formate then dropped to 75.9% while more CO and H2 were produced (TONCO of 221 and TONH2 of 40).Fig. 3: Photocatalytic CO2 reduction products.a–d, Reaction products as a function of time in CO2-saturated MeCN solution for reactions performed in the absence (a) and presence (b) of 0.025 M BIH for a solution containing 50 µM 1, 0.2 mM Ru(phen)3Cl2 and 20% TEA (λ = 460 nm); with 20 µM 1, 2.5 mg graphitic carbon nitride (g-C3N4), 20% TEOA and 0.05 M BIH (λ > 400 nm) (c); or with 50 µM 1, 0.2 mM Ru(phen)3Cl2, 0.1 M BIH and 1 M PhOH (λ = 460 nm) (d). Error bars indicate standard error of the mean, calculated from two to four runs.Full size image Table 1 Photoinduced CO2 catalytic reduction with catalyst 1 and [Ru(phen)3]2+ or Pheno or g-C3N4 as photosensitizerFull size table Under optimized photocatalytic conditions (Fig. 3), a dynamic light scattering (DLS) experiment showed that no particles were formed during the irradiation period (Supplementary Fig. 7). A mercury poisoning test indicated that the TON of formate decreased by 19% in the presence of 0.4 ml Hg after irradiation for 18 h, illustrating that the reaction system is mainly homogeneous (Supplementary Fig. 8). The ESI-MS spectrum of the product solution on saturating the solution with 13CO2 led to H13COO− formation, indicating that formate is generated from CO2 reduction exclusively (Supplementary Fig. 9). As can be seen from Fig. 3b, catalysis stopped after 18 h. Upon re-addition of 1, Ru(phen)3Cl2 or TEA to the inactive solution, only 5%, 21% and 22% of the catalytic activity was restored, respectively. However, 91% of activity could be recovered on adding BIH, indicating that the loss of activity is mainly due to the consumption of BIH itself (Supplementary Fig. 10). Finally, the quantum yield for formate production was determined to be 2.6% from ferrioxalate actinometry (at λ = 460 nm).Photochemical experiments with cheap sensitizers.The expensive Ru-based sensitizers could be replaced by an organic dye as well as by the semiconducting g-C3N4. First, the organic chromophore Pheno (Fig. 1), which is based on a phenoxazine motif, was used as a sensitizer due to its excellent photon absorption in the visible light spectrum, high triplet quantum yield and long triplet lifetime, as well as redox reversibility and strong reducing ability (the triplet state excited standard reduction potential E0(2Pheno•+/3Pheno) = −1.80 V versus saturated calomel electrode (SCE) in dimethyl acetamide)49. In a CO2-saturated solution containing 15 μM 1, 0.4 mM Pheno, 20% TEOA and 0.05 M BIH, a high TON of 565 was obtained for formate, with 59% selectivity; CO (29%) and H2 (12%) were also produced (Table 1, entry 5). On using g-C3N4, a well-known semiconductor that can absorb visible light and whose conduction band energy is negative enough to transfer electrons to 1 (the estimated flat band potential is ~−1.35 V versus SCE)50,51,52, excellent selectivity to formate was equally achieved. With a system containing 20 μM 1, 2.5 mg g-C3N4 and 20% TEOA + 0.025 M BIH, CO2 was reduced to formate with a TON of 493 (Fig. 3c and Table 1, entry 4) and a selectivity of 91% (1.7% quantum yield). H2 was obtained as a minor side product with a TON of 48. No CO and no hydrocarbons were detected. A control experiment in the absence of catalyst only produced H2 (TON of 53); no product was detected in the absence of catalyst, nor under Ar, and when replacing catalyst 1 by CoCl2 only small amounts of H2 and CO were provided (Supplementary Table 3, entries 30–32). Data obtained with only TEOA as sacrificial donor gave essentially similar results although with slightly lower formate selectivity (74.5%, see Supplementary Table 3, entries 24–28). This highlights that a catalytic system combining an earth-abundant metal complex with g-C3N4 can selectively catalyse the reduction of CO2 to formate.Switching catalytic selectivity to CO.Surprisingly, when phenol (PhOH) was used as a co-substrate to boost the reactivity, the selectivity for CO2 reduction remained high but CO became the major product, with a maximum TON of 829 and catalytic selectivity ranging from 90% to 99%, depending on the catalytic conditions. In the presence of 1 M PhOH, 0.1 M BIH, Ru(phen)32+ and 1, CO was obtained with 96% catalytic selectivity (TON of 829) upon 1 h irradiation (Fig. 3d and Table 1, entry 6). In similar conditions, with Pheno as sensitizer, a TON of 518 (89.3% selectivity) in MeCN (Table 1, entry 7) and a TON of 380 (99% selectivity) in a DMF solution were obtained (Supplementary Table 3, entry 20). Selectivity towards CO or HCOO− formation could thus be tuned and controlled. It is noticeable that the rate for CO production was faster than the rate for HCOO− formation, suggesting different catalytic pathways (compare, for example, entries 2–3 and 6 in Table 1). To shed light on the catalytic mechanisms, we thus performed electrochemical experiments as well as density functional theory (DFT) calculations.Mechanistic studies of the CO2 reduction.The electrochemical properties of 1 (noted [CoIICoIIL]4+ in the following) were investigated by cyclic voltammetry (CV), electrolysis and spectro-electrochemistry. As shown in Supplementary Fig. 14, the first CV reversible wave is observed at ~−0.65 V versus SCE. It is a two-electron, metal-centred wave that leads to [CoICoIL]2+ formation. The second wave at −0.81 V versus SCE can be assigned to the reduction of each ligand moiety, giving rise to [CoICoIL2−]. The more negative waves probably correspond to ligand-centred processes. The number of electrons for each wave was assigned upon comparison with the one-electron reversible wave of the monomeric complex 2 (Supplementary Fig. 15)53,54,55. Following CO2 saturation of the solution, a large catalytic wave is observed on the multi-electronic wave at −1.75 V versus SCE, while the other waves remain unchanged (Fig. 3a). In the presence of PhOH, the catalytic wave is shifted towards positive potential with an onset potential at ~−1.25 V versus SCE (Supplementary Fig. 16); a bulk electrolysis at −1.4 V versus SCE (3 h, with 0.5 M PhOH) furnished CO with a faradaic yield of 99.5% and trace H2 (Supplementary Fig. 17), in accordance to the results obtained in photochemical experiments (Table 1, entries 6 and 7).On adding TEA (0.5 M) under an Ar atmosphere, the first two reduction waves merged into a single one at a more cathodic potential (−0.92 V versus SCE, Supplementary Fig. 18), suggesting TEA binding to the CoII centres, which are then reduced with a C+E mechanism (a chemical reaction follow-up electrode electron transfer mechanism):$$\left[ {\mathrm{Co}}^{\mathrm{II}}{\mathrm{Co}}^{\mathrm{II}}( {\mathrm{L}} ) \right]^{4 + } + 2{\mathrm{TEA}} \rightleftarrows \left[ {\mathrm{Co}}^{\mathrm{II}}{\mathrm{Co}}^{\mathrm{II}}( {\mathrm{TEA}} )_2{\mathrm{L}} \right]^{4 + }$$ (1) $$\left[ {\mathrm{Co}}^{\mathrm{II}}{\mathrm{Co}}^{\mathrm{II}}({\mathrm{TEA}})_2\left( {\mathrm{L}} \right) \right]^{4 + } + 4{\mathrm{e}}^ - \rightleftarrows \left[ {\mathrm{Co}}^{\mathrm{I}}{\mathrm{Co}}^{\mathrm{I}}\left( {\mathrm{L}}^{2 - } \right) \right] + 2{\mathrm{TEA}}$$ (2) The disappearance of the [CoIICoII(L)]4+ oxidation wave of 1 (Supplementary Fig. 18) also points towards binding of the complex to the amine. When the solution was purged with CO2, the first reduction peak shifted towards positive potential by 24 mV, indicative of binding to CO2 (Supplementary Fig. 18). A new catalytic wave was observed at ~−1.5 V versus SCE (Fig. 4a) that corresponds to reduction of the CO2-catalyst adduct, in addition to a catalytic process observed at more negative potential. Bulk electrolysis of the CO2-saturated MeCN solution containing 0.5 mM 1, 0.5 M TEA and 0.1 M nBu4NPF6 at −1.5 V versus SCE gave CO, formate and H2 with faradaic yields of 90%, 7% and 0.5%, respectively (Supplementary Fig. 19). Increasing the concentration of TEA to 1.44 M (20% in volume ratio) resulted in an increase in faradaic yield for formate and H2 to 15.8% and 3.8%, respectively, while the value of CO decreased to 78.3%. Formate production could thus be observed in electrochemical conditions in the presence of a large concentration of amine, even if there was competition with CO formation, due to several catalytic pathways occurring in parallel, as evidenced by the multiple catalytic waves observed in CV. Spectro-electrochemistry (SEC) in the infrared region was performed at various potentials in CO2-saturated solutions and in the presence of 0.5 M TEA (Fig. 4b; a Pt grid placed in front of the beam was used as a working electrode and a Ag wire served as a pseudo-reference electrode, see Supplementary Methods for further details). When the potential was scanned from −0.35 V to −0.85 V (versus the Ag pseudo-reference), a potential window within which the catalyst is reduced, an absorption band at 1,635 cm−1 built up (Fig. 4b). This corresponds to the formation of a stable adduct between CO2 and the four-electron-reduced complex. DFT calculations were carried out at B3LYP and M06 levels for two different coordination modes of CO2 with [CoICoI(L2−)] (noted [CoCo(L)] in the following): (1) CO2 sandwiched between two Co centres and (2) CO2 coordinated to one of the Co only (see Supplementary Data File for the cartesian coordinates). For (1), the scaled vibrational frequencies56 of asymmetric C=O stretching are 1,612 (B3LYP) and 1,615 (M06) cm−1. For (2), the scaled vibrational frequencies56 of asymmetric C=O stretching are 1,866 (B3LYP) and 1,987 (M06) cm−1. These results indicate that a CO2 molecule binding to only one Co from the outside of the complex cavity is very unlikely, while an adduct with the C atom from CO2 binding to one Co atom, and one O atom interacting with the second Co atom (Fig. 4b, inset), gives a C–O stretching frequency in good agreement with the experimental value. Natural bond orbital (NBO) analysis of the adduct between [CoCo(L)] and CO2 (noted [CoCoCO2(L)]) shows that the Co atom bound to the C of the CO2 has approximately nine electrons in its valence orbitals (oxidation state 0) while the Co bound to the O atom has only eight electrons in its valence orbitals (oxidation state +1) (see Supplementary Tables 7 and 8 for details). This indicates that in the four-electron-reduced compound [CoCoCO2(L)], two electrons are located on one Co atom, which acts as a nucleophile on binding to the C of CO2, one electron is on the second Co centre, which behaves as an electrophile to stabilize one O atom of the CO2, and the fourth electron is delocalized over the ligand core. The two metals thus act synergistically to reductively bind and stabilize the CO2.Fig. 4: CV plots and infrared SEC spectra.a, CVs in MeCN solution containing 0.3 mM 1 and 0.1 M nBu4NPF6 under an Ar atmosphere (blue), in the absence (black) and presence (red) of 20% TEA under CO2 saturation. b, Infrared SEC experiment on a 1 mM solution of 1 in MeCN (0.1 M nBu4NPF6, 0.5 M TEA) under CO2; the potential was set at the first reduction wave (see main text). Inset, calculated structure of the CO2-reduced complex adduct (see main text).Full size image On setting the potential at the foot of the catalytic wave (−1.25 V versus the Ag pseudo-reference electrode), the band at 1,635 cm−1 quickly disappeared and new bands at 1,649 cm−1 and 1,680 cm−1 appeared, showing the formation of bicarbonate (Supplementary Fig. 29). A low intensity band (shoulder) at ~1,616 cm−1 also appeared. Although we do not know exactly its nature, this may be related to a formato-complex or free formate. These observations are in line with photochemical experiments for which high selectivity for formate production was achieved (Table 1, entries 1–5). That the reactivity is a direct consequence of metal cooperativity is further illustrated by photochemical experiments with monometallic complex 2. We found that at low catalyst concentration (5 µM) and in the presence of [Ru(phen)3]2+ and TEOA, CO was produced with high selectivity (91%) and TON (1989), as we have already observed with various sensitizers, but on increasing the catalyst concentration to 150 µM, the selectivity for formate increased to 50%, with CO and H2 being produced in 33% and 17% yield, respectively (Supplementary Fig. 20a and Supplementary Table 5). From 5 to 150 µM of catalyst, the CO/HCOO− ratio was reduced from 18.1 to 0.65, in accordance with a mechanism for formate production involving two Co centres for binding CO2, which is favoured at a high concentration of 2.All these observations, together with preliminary theoretical calculations (Supplementary Fig. 30), converge to a possible mechanism for the visible-light-driven reduction of CO2 with catalyst 1 that is shown in Fig. 5. Initially, catalyst 1 undergoes 4e− reduction by the photosensitizer to afford the neutral [CoCo(L)] species, which reductively binds to CO2 with one Co centre linked to the C atom of CO2 and the other to an O atom, forming a sandwich type CO2 adduct [CoCoCO2(L)] (Figs. 5 and 4b, inset). Remarkably, the OCO angles decrease from 180° to 126° and the C–O bond lengths increase from 1.116 Å to 1.243 and 1.286 Å, in line with CO2 reduction. After a one-electron reduction of this adduct, the catalytically active species [CoCoCO2(L)]− leads to product formation. In acidic conditions, protonation of the CO2 adduct takes place efficiently at the O atom stabilized by one of the two Co atoms, and C–O bond cleavage results in the subsequent release of CO as product. A second reaction pathway was identified, involving C to O linkage isomerization of the reduced CO2 adduct to [CoCoO2C(L)]− (Fig. 5). This isomerization step is endowed with a barrier of 18.5 kcal mol−1 and may become competitive in media with low proton availability. [CoCoO2C(L)]− is then quickly protonated at the basic C atom with release of formate, in accordance with experimental results observed under basic conditions. Cooperativity between the two Co centres of reduced catalyst 1 is thus the key for not only CO2 binding but also for driving the reaction towards HCOO− evolution. The alternative formation of a hydride species bridging between the two Co atoms that would further insert CO2 to give HCOO− could not be ruled out, although the absence of formate evolution in acidic conditions and the low amount of H2 produced in all experiments makes this mechanism unlikely. However, further mechanistic investigation will shed more light on the process, starting from Fig. 5, which stands as a working model.Fig. 5: Proposed mechanism for visible-light-driven catalytic reduction of CO2 into CO and HCOO− with catalyst 1.Complex 1 undergoes 4e− reduction by the photosensitizer to afford the neutral [CoCo(L)] species, which reductively binds to CO2 with one Co centre linked to the C atom of CO2 and the other to an O atom, forming a sandwich-type CO2 adduct [CoCoCO2(L)]. After a one-electron reduction of this adduct, the catalytically active species [CoCoCO2(L)]− leads to product formation, either CO or formate.Full size image .Conclusions.The binuclear cobalt complex 1 bearing a bi-quaterpyridine ligand can selectively catalyse visible-light-driven CO2 reduction to formate with high TON and a maximum quantum efficiency of 2.6%. Various sensitizers could be used to achieve this reaction, including a Ru complex, an organic phenoxazine compound as well as semiconducting g-C3N4. By simply adding a weak acid such as phenol to the solution, the catalytic process could be switched to selective CO formation. This dual reactivity was made possible by the two metals acting in synergy towards CO2. Such control of the two-electron/two-proton reduction of CO2 illustrates the remarkable potential of multimetallic molecular catalysts, as performed by enzymes. Controlling the formation of CO versus formate, two early intermediates in the CO2 catalytic reduction, may also open new pathways for the selective production of highly reduced products such as methanol or hydrocarbons.Methods.Chemicals.Ru(phen)3Cl2 (ref. 57), g-C3N4 (ref. 58), BIH (ref. 59), 6-bromo-2,2′:6′,2′′-terpyridine60 and 2,7-di-tert-butyl-9,9-dimethylxanthene-4,5-diboronic acid61 were synthesized according to literature methods. Cobalt(ii ) chloride hexahydrate (Acros Organics, 98%), sodium formate (Sigma-Aldrich, 99.998%), acetonitrile (MeCN, ACS, 99.9%), triethanolamine (Sigma-Aldrich, 99.5%), 2-phenylbenzimidazole (J&K, 98%), iodomethane (Energy Chemical, 99.5%), sodium borohydride (Acros, 99%), phenol (Sigma-Aldrich, ≥94%), hexamethyldistannane (Aldrich, 99%) and Pd(PPh3)4 (J&K, 98%) were used as received. Triethylamine (Acros Organics, 99.7%) was distilled with KOH. Synthesis of Pheno has been described elsewhere62. See Supplementary Methods for the synthesis and characterization of Co2(biqpy)Cl2, Ru(phen)3Cl2 and g-C3N4.Photocatalytic CO2 reduction.Photocatalysis was conducted in a glass tube sealed with a rubber septum. The headspace of the tube was 10.4 ml. A 2 ml volume of reaction solution was bubbled with CO2 for 20 min and then irradiated with blue LED light (4.3 W) with the wavelength centred at 460 nm for the experiments with Ru(phen)3Cl2 as photosensitizer. The glass tube was placed into a constant-temperature water bath (24 °C) during irradiation. The 18-module blue LED light strip was wrapped around the water bath and connected to a 12 V power supply. Control experiments were conducted using the same irradiation conditions. The gaseous products in the headspace were analysed by GC-thermal conductivity detection (TCD) (HP 5890) equipped with a Chrompack 5 Å molecular sieve column (30 m × 0.32 mm × 10 µm). Calibration curves were obtained by filling pure gases into a tube with a graduated gas-tight syringe. The liquid products were analysed by GC-MS (Agilent 6890-5975 with DB-5MS column, 30 m × 0.25 mm × 0.25 µm) and ionic chromotography (ICS 1600 system with a Dionex IonPac AS22 column). The 13C-labelled experiment was conducted following the same procedure except 13CO2 was used and the gaseous products were analysed by GC-TCD and ESI-MS. The TON was based on the moles of [Co2L]4+ catalyst used.For photocatalysis with g-C3N4, a 400 W Xe-Hg lamp with a 400 nm optical filter was used as the light source. Other procedures were the same as described in the first paragraph of this section.For photocatalysis with phenoxazine, photocatalytic experiments were performed in a quartz reactor sealed with a rubber septum. The headspace and solution of the reactor were 3 ml and 1 ml, respectively. A solar simulator equipped with a 100 W Xe lamp and a 400 nm optical filter were used as irradiation source. Other procedures were the same as described in the first paragraph of this section.Electrochemical studies.CV was performed using a CHI 660C instrument with a three-electrode system. The working electrode was a glassy carbon electrode (3 mm diameter), the counterelectrode was a Pt wire and SCE was used as the reference electrode. The electrolyte solution was purged with acetonitrile-saturated Ar or CO2. All reported potentials are versus SCE. Bulk electrolysis was performed with a Princeton Applied Research Potentiostat (PARSTAT 2273), using a glassy carbon millimetric electrode (2 cm2) as working electrode, a Pt grid as counterelectrode and an aqueous SCE electrode as reference electrode. The acetonitrile solution in the counterelectrode compartment contained 0.4 M Et4N(CH3CO2) and 0.1 M nBu4NPF6. The volume of the electrolysis solution was 6 or 6.5 ml and the solution was purged with CO2 for 20 min before electrolysis.Infrared SEC studies.The infrared SEC experiments were performed with a home-made cell (thickness of 0.3 mm), equipped with two KBr windows. The working electrode was a platinum grid, placed on the infrared beam. A Ag wire was used as pseudo-reference and a Pt grid as counterelectrode. Two spectra (one spectrum every 1 min, four scans) were registered at each potential step.Quantum yield determination.Quantum yields were measured with a 500 W Xe lamp equipped with a monochromator. The number of photons was determined by ferrioxalate actinometry using monochromic light at 458 nm. The UV–vis absorption spectra of the actinometer solutions were obtained with an Agilent 8453 UV–vis diode-array spectrophotometer. A quartz cell (path length of 1 cm) containing 0.05 mM 1/0.2 mM Ru(phen)3Cl2/20% (vol%) TEA/0.025 M BIH/MeCN was sealed with a rubber septum and irradiated at 458 nm. The average light intensity taken before and after the photochemical reaction was 3.24 × 10−9 einstein s−1. The gaseous products were analysed by GC-TCD and the solution products were analysed by IC. The quantum yield for CO2–formate conversion after irradiation for 12 h was calculated using equation (3):$$\begin{array}{{20}{l}}\varPhi_{\mathrm{formate}} & = & \left({{\mathrm{number}}\,{\rm{{of}}}\,{\rm{{formate}}}\,{\rm{{molecules}}}} \right)/\left({{\mathrm{number}}\,{\rm{{of}}}\,{\rm{{incident}}}\,{\rm{{photons}}}}\right)\\&& \times 100\%\end{array}$$ (3) The quantum yield of the 0.02 mM 1/2.5 mg C3N4/20% (vol%) TEOA/MeCN system was measured following the same procedure. The average light intensity taken before and after the photochemical reaction was 2.63 × 10−9 einstein s−1. The quantum yield of CO2–formate conversion after irradiation for 4.7 h was calculated to be 3.42% using equation (4):$$\begin{array}{{20}{l}}{\varPhi}_{\mathrm{formate}} = \left({{\mathrm{number}}\,{\rm{{of}}}\,{\rm{{formate}}}\,{\rm{{molecules}}}}\right)\hfill{} \\ \qquad\qquad\quad \times 2/\left({{\mathrm{number}}\,{\rm{{of}}}\,{\rm{{incident}}}\,{\rm{{photons}}}} \right) \times 100\%\end{array}$$ (4) Finally, for the 0.05 mM 1/2.5 mg g-C3N4/20% (vol%) TEOA/0.05 M BIH/MeCN system, the average light intensity taken before and after the photochemical reaction was 9.36 × 10−10 einstein s−1. The quantum yield of CO2–formate conversion after irradiation for 9 h was calculated to be 1.7% from equation (3).ESI-MS.ESI-MS results were obtained on a PE SCIEX API 150 mass spectrometer. The solution was continuously infused with a syringe pump at a constant flow rate into the pneumatically assisted electrospray probe with nitrogen as the nebulizing gas.DLS.A Zetasizer Nano ZS instrument (Malvern Instruments) equipped with a HeNe gas laser emitting 632.8 nm vertically polarized light and operating at 4 mW was used to measure the DLS spectrum. Data were collected at 23 °C with a scattering angle of 175°.Fluorescence quenching.Various concentrations of BIH or TEA were added to [Ru(phen)3]2+ (0.05 mM) in MeCN and the solutions were degassed by at least four freeze–pump–thaw cycles. The lifetimes of the characteristic emission of the Ru sensitizer at 600 nm were measured. The quenching rate constant (kq) was calculated according to the Stern–Volmer equation (5):$$I_0/I\,{\mathrm{or}}\,\tau _0/\tau = 1 + k_{\mathrm{q}} \times \tau _0 \times \left[ Q \right]$$ (5) where I0 and I are the emission intensity in the absence and presence of quencher, τ0 and τ are the excited-state lifetime in the absence and presence of quencher, kq is the bimolecular quenching rate constant and [Q] is the molar concentration of the quencher.Theoretical calculations.The mechanism of CO2 reduction to CO and formate by 5[CoCo (L)] (four-electron-reduced catalyst 1, Fig. 5) was partly investigated at the B3LYP level of theory. The molecular structures were optimized using the LanL2DZ basis set (Co) and 6-31G(d) basis set (non-metals). All calculations were performed with the GAUSSIAN09 package63. Solvent effect in acetonitrile was treated implicitly by the integral-equation-formalism polarizable continuum model (IEFPCM)64,65 (single-point energy calculations) at the B3LYP level of theory with the LanL2DZ basis set (Co) and 6–31+G(d) basis set (non-metals). Calculations of the Gibbs free energy change for reactions (ΔG298º) and barrier heights (ΔG298‡), unless specified, include the contributions from ΔG298ºCH3CN(H+) = −266.48 kcal mol−1 in acetonitrile, where ΔG298ºCH3CN(H+) is the sum of ΔG298ºgas(H+) = −6.28 kcal mol−1 (ref. 66) and the solvation energy change of H+ in acetonitrile (value of −260.2 kcal mol−1 taken from ref. 67).We carried the vibrational frequencies calculations at the B3LYP and M06 levels for two different coordination modes of CO2 with 5[CoCo(L)] (four-electron-reduced catalyst 1, Fig. 5) in 5[CoCoCO2(L)]: (1) CO2 sandwiched between two Co(qpy) and (2) CO2 coordinated to one of the Co(qpy) only (see Supplementary Data File for the cartesian coordinates of both structures). For (1), the scaled vibrational frequencies56 of asymmetric C=O stretching are 1,612 (B3LYP) and 1,615 (M06) cm−1. For (2), the scaled vibrational frequencies56 of asymmetric C=O stretching are 1,866 (B3LYP) and 1,987 (M06) cm−1.NBO analysis of 5[CoCoCO2(L)] shows that the Co core bound to the C atom of CO2 (noted Co2 in Supplementary Tables 7 and 8) bears approximately nine electrons in its valence orbitals (eight in 3d orbitals, Supplementary Table 7) while the Co atom bound to an O atom (noted Co1 in Supplementary Tables 7 and 8) has approximately eight electrons (seven in 3d orbitals, Supplementary Table 8). Thus the oxidation states of the two Co atoms are 0 (Co2) and +1 (Co1). This indicates that, upon CO2 binding on 5[CoCo(L)], one electron localizes on the Co that binds to the O atom and two electrons localize on the Co bound to the C atom. The fourth electron is delocalized over a quaterpyridine ligand.The calculations (Supplementary Fig. 30) indicate that the adduct, 4[CoCoCO2(L)]− (obtained from the one-electron reduction of 5[CoCoCO2(L)] as sketched in Fig. 5), may undergo protonation on one O of CO2, via a transition state (TSCO), where the barrier height (ΔG298‡) is 23.5 kcal mol−1. This step yields an intermediate, INTCO with hydrocarboxyl moiety sandwiched between two Co(qpy). Then, a C–OH bond cleavage occurs, followed by dissociation of CO from Co(qpy) to generate CO/OH− and 4[CoCo(L)]+ (or CO/H2O in the presence of a second proton). Meanwhile, the 4[CoCoCO2(L)]1− adduct can undergo a coordination change from Co–CO2 to Co–OCO, via TSformate with ΔG298‡ = 18.5 kcal mol−1 (relative to INTCO), to form an intermediate (INTformate), with each O of CO2 coordinated with one Co metal centre (noted [CoCoO2C(L)]− in Fig. 5). On protonation at the C of CO2, formate can be readily released from INT1formate. Data availability. Crystallographic data for [Co2(biqpy)Cl(MeOH)(H2O)](ClO4)3 have been deposited at the Cambridge Crystallographic Data Centre (CCDC no. 1858669. The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request.References.1. Aresta, M., Dibenedetto, A. & Angelini, A. Catalysis for the valorization of exhaust carbon: from CO2 to chemicals, materials and fuels. Technological use of CO2. Chem. Rev. 114, 1709–1742 (2014).CAS.Article. Google Scholar.2. Jhong, H.-R. M., Ma, S. & Kenis, P. J. A. Electrochemical conversion of CO2 to useful chemicals: current status, remaining challenges and future opportunities. Curr. Opin. Chem. Eng. 2, 191–199 (2013).Article. Google Scholar.3. Seh, Z. W. et al. Combining theory and experiment in electrocatalysis: insights into materials design. Science 355, eaad4998 (2017).Article. Google Scholar.4. Fujita, E. Photochemical carbon dioxide reduction with metal complexes. Coord. Chem. Rev. 185–186, 373–384 (1999).Article. Google Scholar.5. Tamaki, Y., Morimoto, T., Koike, K. & Ishitani, O. Photocatalytic CO2 reduction with high turnover frequency and selectivity of formic acid formation using Ru(ii ) multinuclear complexes. Proc. Natl Acad. Sci. USA 109, 15673–15678 (2012).CAS.Article. Google Scholar.6. Rosas-Hernández, A., Junge, H. & Beller, M. Photochemical reduction of carbon dioxide to formic acid using ruthenium(ii )-based catalysts and visible light. ChemCatChem 7, 3316–3321 (2015).Article. Google Scholar.7. Tamaki, Y., Koike, K. & Ishitani, O. Highly efficient, selective, and durable photocatalytic system for CO2 reduction to formic acid. Chem. Sci. 6, 7213–7221 (2015).CAS.Article. Google Scholar.8. Sato, S., Morikawa, T., Kajino, T. & Ishitani, O. A highly efficient mononuclear iridium complex photocatalyst for CO2 reduction under visible light. Angew. Chem. Int. Ed. 52, 988–992 (2013).CAS.Article. Google Scholar.9. Reithmeier, R. O. et al. Mono- and bimetallic Ir(iii ) based catalysts for the homogeneous photocatalytic reduction of CO2 under visible light irradiation. New insights into catalyst deactivation. Dalton Trans. 43, 13259–13269 (2014).CAS.Article. Google Scholar.10. Reithmeier, R. O., Meister, S., Siebel, A. & Rieger, B. Synthesis and characterization of a trinuclear iridium(iii ) based catalyst for the photocatalytic reduction of CO2. Dalton Trans. 44, 6466–6472 (2015).CAS.Article. Google Scholar.11. Hawecker, J., Lehn, J.-M. & Ziessel, R. Efficient photochemical reduction of CO2 to CO by visible light irradiation of systems containing Re(bipy)(CO)3X or Ru(bipy)3 2+–Co2+ combinations as homogeneous catalysts. J. Chem. Soc. Chem. Commun. 536–538 (1983).12. Takeda, H., Koike, K., Inoue, H. & Ishitani, O. Development of an efficient photocatalytic system for CO2 reduction using rhenium(i ) complexes based on mechanistic studies. J. Am. Chem. Soc. 130, 2023–2031 (2008).CAS.Article. Google Scholar.13. Takeda, H. & Ishitani, O. Development of efficient photocatalytic systems for CO2 reduction using mononuclear and multinuclear metal complexes based on mechanistic studies. Coord. Chem. Rev. 254, 346–354 (2010).CAS.Article. Google Scholar.14. Agarwal, J., Fujita, E., Schaefer, H. F. & Muckerman, J. T. Mechanisms for CO production from CO2 using reduced rhenium tricarbonyl catalysts. J. Am. Chem. Soc. 134, 5180–5186 (2012).CAS.Article. Google Scholar.15. Yamazaki, Y., Takeda, H. & Ishitani, O. Photocatalytic reduction of CO2 using metal complexes. J. Photochem. Photobiol. C 25, 106–137 (2015).CAS.Article. Google Scholar.16. Lehn, J.-M. & Ziessel, R. Photochemical generation of carbon monoxide and hydrogen by reduction of carbon dioxide and water under visible light irradiation. Proc. Natl Acad. Sci. USA 79, 701–704 (1982).CAS.Article. Google Scholar.17. Varma, S. et al. Efficient photocatalytic hydrogen production in water using a cobalt(iii ) tetraaza-macrocyclic catalyst: electrochemical generation of the low-valent Co(i ) species and its reactivity toward proton reduction. Phys. Chem. Chem. Phys. 15, 17544–17552 (2013).CAS.Article. Google Scholar.18. Chan, S. L.-F., Lam, T. L., Yang, C., Yan, S.-C. & Cheng, N. M. A robust and efficient cobalt molecular catalyst for CO2 reduction. Chem. Commun. 51, 7799–7801 (2015).CAS.Article. Google Scholar.19. Chen, L. et al. Molecular catalysis of the electrochemical and photochemical reduction of CO2 with earth-abundant metal complexes. selective production of CO vs HCOOH by switching of the metal center. J. Am. Chem. Soc. 137, 10918–10921 (2015).CAS.Article. Google Scholar.20. Tinnemans, A. H. A., Koster, T. P. M., Thewissen, D. H. M. W. & Mackor, A. Tetraaza-macrocyclic cobalt(ii ) and nickel(ii ) complexes as electron-transfer agents in the photo(electro)chemical and electrochemical reduction of carbon dioxide. Recl. Trav. Chim. Pays-Bas 103, 288–295 (1984).CAS.Article. Google Scholar.21. Thoi, V. S., Kornienko, N., Margarit, C. G., Yang, P. & Chang, C. J. Visible-light photoredox catalysis: selective reduction of carbon dioxide to carbon monoxide by a nickel N-heterocyclic carbene–isoquinoline complex. J. Am. Chem. Soc. 135, 14413–14424 (2013).CAS.Article. Google Scholar.22. Bonin, J., Robert, M. & Routier, M. Selective and efficient photocatalytic CO2 reduction to CO using visible light and an iron-based homogeneous catalyst. J. Am. Chem. Soc. 136, 16768–16771 (2014).CAS.Article. Google Scholar.23. Takeda, H., Ohashi, K., Sekine, A. & Ishitani, O. Photocatalytic CO2 reduction using Cu(i ) photosensitizers with a Fe(ii ) catalyst. J. Am. Chem. Soc. 138, 4354–4357 (2016).CAS.Article. Google Scholar.24. Alsabeh, P. G. et al. Iron-catalyzed photoreduction of carbon dioxide to synthesis gas. Catal. Sci. Technol. 6, 3623–3630 (2016).CAS.Article. Google Scholar.25. Rao, H., Schmidt, L. C., Bonin, J. & Robert, M. Visible-light-driven methane formation from CO2 with a molecular iron catalyst. Nature 548, 74 (2017).CAS.Article. Google Scholar.26. Takeda, H., Koizumi, H., Okamoto, K. & Ishitani, O. Photocatalytic CO2 reduction using a Mn complex as a catalyst. Chem. Commun. 50, 1491–1493 (2014).CAS.Article. Google Scholar.27. Torralba-Peñalver, E., Luo, Y., Compain, J.-D., Chardon-Noblat, S. & Fabre, B. Selective catalytic electroreduction of CO2 at silicon nanowires (SiNWs) photocathodes using non-noble metal-based manganese carbonyl bipyridyl molecular catalysts in solution and grafted onto SiNWs. ACS Catal. 5, 6138–6147 (2015).Article. Google Scholar.28. Fei, H., Sampson, M. D., Lee, Y., Kubiak, C. P. & Cohen, S. M. Photocatalytic CO2 reduction to formate using a Mn(i ) molecular catalyst in a robust metal–organic framework. Inorg. Chem. 54, 6821–6828 (2015).CAS.Article. Google Scholar.29. Cheung, P. L., Machan, C. W., Malkhasian, A. Y. S., Agarwal, J. & Kubiak, C. P. Photocatalytic reduction of carbon dioxide to CO and HCO2H using fac-Mn(CN)(bpy)(CO)3. Inorg. Chem. 55, 3192–3198 (2016).CAS.Article. Google Scholar.30. Guo, Z. et al. Photocatalytic conversion of CO2 to CO by a copper(ii ) quaterpyridine complex. ChemSusChem 10, 4009–4013 (2017).CAS.Article. Google Scholar.31. Liu, W. J. et al. A copper(ii ) molecular catalyst for efficient and selective photochemical reduction of CO2 to CO in a water-containing system. Chem. Eur. J. 24, 4503–4508 (2018).CAS.Article. Google Scholar.32. Boston, D. J., Xu, C., Armstrong, D. W. & MacDonnell, F. M. Photochemical reduction of carbon dioxide to methanol and formate in a homogeneous system with pyridinium catalysts. J. Am. Chem. Soc. 135, 16252–16255 (2013).CAS.Article. Google Scholar.33. Takeda, H., Cometto, C., Ishitani, O. & Robert, M. Electrons, photons, protons and earth-abundant metal complexes for molecular catalysis of CO2 reduction. ACS Catal. 7, 70–88 (2017).CAS.Article. Google Scholar.34. Rao, H., Bonin, J. & Robert, M. Toward visible-light photochemical CO2-to-CH4 conversion in aqueous solutions using sensitized molecular catalysis. J. Phys. Chem. C. 122, 13834–13839 (2018).CAS.Article. Google Scholar.35. Mellmann, D., Sponholz, P., Junge, H. & Beller, M. Formic acid as a hydrogen storage material—development of homogeneous catalysts for selective hydrogen release. Chem. Soc. Rev. 45, 3954–3988 (2016).CAS.Article. Google Scholar.36. Yu, X. & Pickup, P. G. Recent advances in direct formic acid fuel cells (DFAFC). J. Power Sources 182, 124–132 (2008).CAS.Article. Google Scholar.37. Li, H. et al. Integrated electromicrobial conversion of CO2 to higher alcohols. Science 335, 1596–1596 (2012).CAS.Article. Google Scholar.38. Loges, B., Boddien, A., Gärtner, F., Junge, H. & Beller, M. Catalytic generation of hydrogen from formic acid and its derivatives: useful hydrogen storage materials. Top. Catal. 53, 902–914 (2010).CAS.Article. Google Scholar.39. Guo, Z. et al. Highly efficient and selective photocatalytic CO2 reduction by iron and cobalt quaterpyridine complexes. J. Am. Chem. Soc. 138, 9413–9416 (2016).CAS.Article. Google Scholar.40. Matsuoka, S. et al. Efficient and selective electron mediation of cobalt complexes with cyclam and related macrocycles in the p-terphenyl-catalyzed photoreduction of carbon dioxide. J. Am. Chem. Soc. 115, 601–609 (1993).CAS.Article. Google Scholar.41. Jeoung, J.-H. & Dobbek, H. Carbon dioxide activation at the Ni,Fe-cluster of anaerobic carbon monoxide dehydrogenase. Science 318, 1461–1464 (2007).CAS.Article. Google Scholar.42. Appel, A. M. et al. Frontiers, opportunities and challenges in biochemical and chemical catalysis of CO2 fixation. Chem. Rev. 113, 6621–6658 (2013).CAS.Article. Google Scholar.43. Wang, J.-W., Zhong, D.-C. & Lu, T.-B. Artificial photosynthesis: catalytic water oxidation and CO2 reduction by dinuclear non-noble-metal molecular catalysts. Coord. Chem. Rev. 377, 225–236 (2018).CAS.Article. Google Scholar.44. Ouyang, T., Huang, H.-H., Wang, J.-W., Zhong, D.-C. & Lu, T.-B. A dinuclear cobalt cryptate as a homogeneous photocatalyst for highly selective and efficient visible-light driven CO2 reduction to CO in CH3CN/H2O solution. Angew. Chem. Int. Ed. 56, 738–743 (2017).CAS.Article. Google Scholar.45. Ouyang, T. et al. Dinuclear metal synergistic catalysis boosts photochemical CO2-to-CO conversion. Angew. Chem. Int. Ed. 57, 16480–16485 (2018).CAS.Article. Google Scholar.46. Taheri, A., Thompson, E. J., Fettinger, J. C. & Berben, L. A. An iron electrocatalyst for selective reduction of CO2 to formate in water: including thermochemical insights. ACS Catal. 5, 7140–7151 (2015).CAS.Article. Google Scholar.47. Paul, A., Connolly, D., Schulz, M., Pryce, M. T. & Vos, J. G. Effect of water during the quantitation of formate in photocatalytic studies on CO2 reduction in dimethylformamide. Inorg. Chem. 51, 1977–1979 (2012).CAS.Article. Google Scholar.48. Kuramochi, Y., Kamiya, M. & Ishida, H. Photocatalytic CO2 reduction in N,N-dimethylacetamide/water as an alternative solvent system. Inorg. Chem. 53, 3326–3332 (2014).CAS.Article. Google Scholar.49. Du, Y. et al. Strongly reducing, visible-light organic photoredox catalysts as sustainable alternatives to precious metals. Chem. Eur. J. 23, 10962–10968 (2017).CAS.Article. Google Scholar.50. Fang, Y. & Wang, X. Photocatalytic CO2 conversion by polymeric carbon nitrides. Chem. Commun. 54, 5674–5687 (2018).CAS.Article. Google Scholar.51. Wang, X. et al. A metal-free polymeric photocatalyst for hydrogen production from water under visible light. Nat. Mater. 8, 76–80 (2008).Article. Google Scholar.52. Maeda, K., Sekizawa, K. & Ishitani, O. A polymeric-semiconductor–metal-complex hybrid photocatalyst for visible-light CO2 reduction. Chem. Commun. 49, 10127–10129 (2013).CAS.Article. Google Scholar.53. Cometto, C. et al. Highly selective molecular catalysts for the CO2-to-CO electrochemical conversion at very low overpotential. Contrasting Fe vs Co quaterpyridine complexes upon mechanistic studies. ACS Catal. 8, 3411–3417 (2018).CAS.Article. Google Scholar.54. Wang, M., Chen, L., Lau, T.-C. & Robert, M. A hybrid Co quaterpyridine complex/carbon nanotube catalytic material for CO2 reduction in water. Angew. Chem. Int. Ed. 57, 7769–7773 (2018).CAS.Article. Google Scholar.55. Cometto, C. et al. Molecular electrochemical catalysis of the CO2-to-CO conversion with a Co complex: a cyclic voltammetry mechanistic investigation. Organometallics https://doi.org/10.1021/acs.organomet.8b00555 (2018).Article. Google Scholar.56. Kesharwani, M. K., Brauer, B. & Martin, J. M. L. Frequency and zero-point vibrational energy scale factors for double-hybrid density functionals (and other selected methods): can anharmonic force fields be avoided? J. Phys. Chem. A 119, 1701–1714 (2015).CAS.Article. Google Scholar.57. Palmer, R. A. & Piper, T. S. 2,2’-Bipyridine complexes. I. Polarized crystal spectra of tris (2,2’-bipyridine)copper(ii ), -nickel(ii ), -cobalt(ii ), -iron(ii ) and -ruthenium(ii ). Inorg. Chem. 5, 864–878 (1966).CAS.Article. Google Scholar.58. Martha, S., Nashim, A. & Parida, K. M. Facile synthesis of highly active g-C3N4 for efficient hydrogen production under visible light. J. Mater. Chem. A 1, 7816–7824 (2013).CAS.Article. Google Scholar.59. Lee, I.-S. H., Jeoung, E. H. & Kreevoy, M. M. Marcus theory of a parallel effect on α for hydride transfer reaction between NAD+ analogues. J. Am. Chem. Soc. 119, 2722–2728 (1997).CAS.Article. Google Scholar.60. Stefankiewicz, A. R. et al. Self-assembly of transition metal ion complexes of a hybrid pyrazine-terpyridine ligand. Dalton Trans. 42, 1743–1751 (2013).CAS.Article. Google Scholar.61. Okamura, R., Wada, T., Aikawa, K., Nagata, T. & Tanaka, K. A platinum−ruthenium dinuclear complex bridged by bis(terpyridyl)xanthene. Inorg. Chem. 43, 7210–7217 (2004).CAS.Article. Google Scholar.62. Pearson, R. M., Lim, C.-H., McCarthy, B. G., Musgrave, C. B. & Miyake, G. M. Organocatalyzed atom transfer radical polymerization using N-aryl phenoxazines as photoredox catalysts. J. Am. Chem. Soc. 138, 11399–11407 (2016).CAS.Article. Google Scholar.63. Frisch. M. J. et al. Gaussian 09, Revision A.02 (Gaussian, 2016).64. Miertus, S., Scrocco, E. & Tomasi, J. Electrostatic interaction of a solute with a continuum. A direct utilizaion of AB initio molecular potentials for the prevision of solvent effects. Chem. Phys. 55, 117–129 (1981).CAS.Article. Google Scholar.65. Miertus, S. & Tomasi, J. Approximate evaluations of the electrostatic free energy and internal energy changes in solution processes. Chem. Phys. 65, 239–245 (1982).CAS.Article. Google Scholar.66. Tawa, G. J., Topol, I. A., Burt, S. K., Caldwell, R. A. & Rashin, A. A. J. Calculation of the aqueous solvation free energy of the proton. Chem. Phys. 109, 4852–4863 (1998).CAS. Google Scholar.67. Kelly, C. P., Cramer, C. J. & Truhlar, D. G. Single-ion solvation free energies and the normal hydrogen electrode potential in methanol, acetonitrile, and dimethyl sulfoxide. J. Phys. Chem. B 111, 408–422 (2007).CAS.Article. Google Scholar. Download references .Acknowledgements.The work described in this project was supported by the National Science Foundation of China (grant no. 21703034), Hong Kong University Grants Committee Area of Excellence Scheme (grant no. AoE/P-03–08), Hong Kong Research Grants Council (N_CityU115/18) and the French National Agency for Research (ANR-16-CE05-0010-01). G.C. acknowledges start-up grants from Dongguan University of Technology for high-level talents (grant nos G200906-47, GC200109-17 and KCYKYQD2017016). K.C.L. and M.R. acknowledge partial financial support from CityU Strategic Research Grant no. 7004819 and from the Institut Universitaire de France (IUF), respectively. PhD fellowships to C.C. from Université Sorbonne Paris Cité (USPC) and to B.M. from the China Scholarship Council (CSC student no. 201707040042) are acknowledged. G. Thoraval (Université Paris Diderot) is thanked for the design and preparation of the glassy carbon electrode (3 mm diameter) used during CV experiments. Finally, we thank G. Miyake (Colorado State University) for the sample gift of phenoxazine (Pheno).Author information.Author notes.These authors contributed equally: Zhenguo Guo, Gui Chen, Claudio Cometto, Bing Ma.Affiliations.School of Environment and Civil Engineering, Dongguan University of Technology, Dongguan, Guangdong, China.Zhenguo Guo., Gui Chen., Lingjing Chen., Hongbo Fan. & Shek-Man Yiu.Department of Chemistry and Institute of Molecular Functional Materials, City University of Hong Kong, Kowloon Tong, Hong Kong, China.Zhenguo Guo., Hongyan Zhao., Kai-Chung Lau. & Tai-Chu Lau.Université de Paris, Laboratoire d’Electrochimie Moléculaire, CNRS, Paris, France.Claudio Cometto., Bing Ma., Thomas Groizard. & Marc Robert.Department of Chemistry, Hong Kong Baptist University, Kowloon Tong, Hong Kong, China.Wai-Lun Man.Authors.Search for Zhenguo Guo in:.PubMed • . Google Scholar .Search for Gui Chen in:.PubMed • . Google Scholar .Search for Claudio Cometto in:.PubMed • . Google Scholar .Search for Bing Ma in:.PubMed • . Google Scholar .Search for Hongyan Zhao in:.PubMed • . Google Scholar .Search for Thomas Groizard in:.PubMed • . Google Scholar .Search for Lingjing Chen in:.PubMed • . Google Scholar .Search for Hongbo Fan in:.PubMed • . Google Scholar .Search for Wai-Lun Man in:.PubMed • . Google Scholar .Search for Shek-Man Yiu in:.PubMed • . Google Scholar .Search for Kai-Chung Lau in:.PubMed • . Google Scholar .Search for Tai-Chu Lau in:.PubMed • . Google Scholar .Search for Marc Robert in:.PubMed • . Google Scholar .Contributions.G.C., K.-C.L., M.R. and T.-C.L. conceived and supervised the project. G.C., L.C. and H.F. designed and synthesized the catalysts. W.-L.M. and S.-M.Y. characterized the structure of catalyst 1. Z.G., C.C. and B.M. carried out the CO2 reduction experiments. C.C. performed the spectro-electrochemistry experiments. H.Z. and T.G. carried out the DFT calculations. All authors discussed the results and assisted during manuscript preparation.Corresponding authors.Correspondence to Kai-Chung Lau or Tai-Chu Lau or Marc Robert.Ethics declarations. Competing interests. The authors declare no competing interests.Additional information.Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary information. Supplementary Information.Supplementary methods, Supplementary Figs. 1–30, Supplementary Tables 1–8, Supplementary references.Compound 1.Crystallographic data for compound 1.Supplementary Data File.Rights and permissions.Reprints and Permissions.About this article.Received.29 December 2018Accepted.26 June 2019Published.12 August 2019DOI.https://doi.org/10.1038/s41929-019-0331-6.

增加监测目标对象/主题，请 登录 直接在原文中划词！