Mechanisms of catalytic reduction of CO₂ with heme and nonheme metal complexes

This critical review discusses the thermal and photocatalytic mechanisms of one-, two-, four-, six- and eight-electron reduction of CO₂ with metal complex catalysts.

Abstract

The catalytic conversion of CO₂ into valuable chemicals and fuels has attracted increasing attention, providing a promising route for mitigating the greenhouse effect of CO₂ and also meeting the global energy demand. Among many homogeneous and heterogeneous catalysts for CO₂ reduction, this mini-review is focused on heme and nonheme metal complexes that act as effective catalysts for the electrocatalytic and photocatalytic reduction of CO₂. Because metalloporphyrinoids show strong absorption in the visible region, which is sensitive to the oxidation states of the metals and ligands, they are suited for the detection of reactive intermediates in the catalytic CO₂ reduction cycle by electronic absorption spectroscopy. The first part of this review deals with the catalytic mechanism for the one-electron reduction of CO₂ to oxalic acid with heme and nonheme metal complexes, with an emphasis on how the formation of highly energetic CO₂˙ is avoided. Then, the catalytic mechanism of two-electron reduction of CO₂ to produce CO and H₂O is compared with that to produce HCOOH. The effect of metals and ligands of the heme and nonheme complexes on the CO or HCOOH product selectivity is also discussed. The catalytic mechanisms of multi-electron reduction of CO₂ to methanol (six-electron reduced product) and methane (eight-electron reduced product) are also discussed for both electrocatalytic and photocatalytic systems.

1. Introduction

Solar-driven reduction of CO₂ has merited increasing attention due to the hike in the world-wide consumption of fossil fuels and the consequential escalation in the atmospheric CO₂ level.1–9 There have so far been extensive studies on solar-driven reduction of CO₂ (ref. 7–21) as well as electrocatalytic reduction22–36 and catalytic hydrogenation of CO₂ in aprotic solvents and also in water.37–51 Carbon dioxide can be reduced by one electron and one proton to produce a half equivalent of oxalic acid (H₂C₂O₄) with a standard reduction potential of –0.50 V vs. SHE [eqn (1)].26,52 The two-electron reduction of CO₂ with two protons affords formic acid (HCOOH) [eqn (2)] or CO and H₂O [eqn (3)] with the standard reduction potentials of –0.25 and –0.11 V, respectively.52 Carbon dioxide can be further reduced by four, six, and eight electrons with four, six, and eight protons to produce formaldehyde [HCHO: eqn (4)], methanol [CH₃OH: eqn (5)] or ethylene [C₂H₄: eqn (6)], and methane [CH₄: eqn (7)] with the standard reduction potentials of –0.07, +0.02 or +0.06 V, and +0.17 vs. SHE, respectively.52 Therefore, the standard reduction potential is anodically shifted with an increased number of electrons and protons for CO₂ reduction, indicating that the involvement of a higher number of electrons and protons favours the CO₂ reduction thermodynamically. However, because the kinetic barrier generally increases with an additional number of electrons and protons involved in the reaction, appropriate catalysts are required to facilitate turnovers.

CO₂ + e^– + H⁺ → (1/2)H₂C₂O₄, E⁰ = –0.50 V vs. SHE 1

CO₂ + 2e^– + 2H⁺ → HCOOH, E⁰ = –0.25 V vs. SHE 2

CO₂ + 2e^– + 2H⁺ → CO + H₂O, E⁰ = –0.11 V vs. SHE 3

CO₂ + 4e^– + 4H⁺ → HCHO + H₂O, E⁰ = –0.07 V vs. SHE 4

CO₂ + 6e^– + 6H⁺ → CH₃OH + H₂O, E⁰ = +0.02 V vs. SHE 5

CO₂ + 6e^– + 6H⁺ → (1/2)C₂H₄ + 4H₂O, E⁰ = +0.06 V vs. SHE 6

CO₂ + 8e^– + 8H⁺ → CH₄ + 2H₂O, E⁰ = +0.17 V vs. SHE 7

Among many catalysts for CO₂ reduction, metalloporphyrinoid complexes are suitable for mechanistic studies, because metalloporphyrinoids such as heme have intense absorption bands, which are sensitive to the oxidation states of metals and porphyrinoid ligands.53–58 Not only heme but also nonheme metal complexes have merited significant interest as bioinspired catalysts, which are studied in many redox reactions.26,58–73 This review is intended to focus on the mechanisms of both electrocatalytic and photocatalytic reduction of CO₂ with heme and nonheme metal complexes. The catalytic mechanisms are discussed not only for the one-electron/one-proton and two-electron/two-proton reductions of CO₂ [eqn (1) and (3)] but also for multi-electron/multi-proton pathways [eqn (4)–(7)]. The electrocatalytic and photocatalytic efficiencies are discussed based on the overpotentials and quantum yields, respectively.

2. Catalytic one-electron reduction of CO₂ to oxalic acid

The standard one electron reduction potential of CO₂ to CO₂˙ is as negative as –2.21 V vs. SCE (=–1.97 V vs. SHE) in N,N′-dimethylformamide (DMF)74 and –1.90 V vs. SCE in water.75 The dimerization of CO₂˙ affords oxalate dianions (C₂O₄^2–).76 Therefore, catalysts are required to avoid the formation of highly energetic CO₂˙ and secure a low energy pathway en route to oxalic acid.

Ag^II and Pd^II complexes of both 2,3,7,8,12,13,17,18-octaethylporphyrin (OEP) and 5,10,15,20-tetraphenylporphyrin (TPP) were reported to act as catalysts for one-electron reduction of CO₂ to produce oxalic acid in CH₂Cl₂ containing 0.10 M tetrabutylammonium fluoride (TBAF) at an applied potential of –1.65 and –1.80 V vs. SCE, respectively.77 In comparison with Ag^II(TPP) and Pd^II(TPP), neither Cu^II(TPP) nor Ni^II(TPP) showed catalytic activity for the electrochemical CO₂ reduction.77

In contrast to Cu^II(TPP), a dinuclear nonheme copper(ii) complex ([4]⁴⁺) can catalyse the one-electron reduction of CO₂ in acetonitrile (MeCN) in the presence of LiClO₄ to produce lithium oxalate at an applied potential of –0.03 V vs. NHE.78 The electrochemical reduction of [4]⁴⁺ at a cathodic peak potential (E_pc) of +0.06 V vs. NHE (=+0.06 V vs. SHE) produced a dinuclear copper(i) complex ([1]²⁺) that is oxidized in air selectively by CO₂ (rather than O₂) to yield a tetranuclear copper(ii) complex containing two bridging CO₂-derived oxalate groups ([2]⁴⁺) as shown in Scheme 1.78 The treatment of the copper(ii) oxalate complex in MeCN with a soluble lithium salt results in the quantitative precipitation of lithium oxalate.78 DFT calculations suggest the catalytic mechanism (Scheme 2) in which one CO₂ molecule is first reduced cooperatively by two Cu(i) metals to give a fully delocalized mixed-valence Cu^I/Cu^II(CO₂˙) radical anion intermediate, followed by further partial reduction of the metal-ligated CO₂ molecule and (metal-mediated) nucleophilic-like attack on the carbon atom of an incoming second CO₂ molecule to afford the dinuclear Cu(ii)-oxalate product ([2]⁴⁺).79

Scheme 1. Catalytic one-electron reduction of CO₂ to oxalate with a dinuclear copper(i) complex ([1]²⁺). Reprinted with permission from ref. 78. Copyright 2010, The American Association for the Advancement of Science.

Scheme 2. Proposed catalytic cycle for one-electron reduction of CO₂ to oxalate with a copper(i) complex. Reprinted with permission from ref. 79. Copyright 2017, American Chemical Society.

A binuclear metallacyclic copper complex was reported to be capable of selectively capturing CO₂ from air and reduce CO₂ to oxalate, in the form of an oxalate-bridged complex. The oxalate-bridged complex releases oxalic acid when it is treated with dilute mineral acid to regenerate the original copper complex.80

It has also been reported that the one-electron reduction of chalcogen-bridged tricopper cyclophanates (Cu₃EL: L^3– = tris(β-diketiminate; E = S, Se) (the one-electron reduction potential: E_red = –0.89 and –1.04 V vs. SCE, respectively) by CoCp₂* (one-electron oxidation potential: E_ox = –1.94 V vs. Fc/Fc⁺), FeCp(C₆Me₆) (E_ox = –2.07 V vs. Fc/Fc⁺), or KC₈ (E_ox < –3.7 V vs. Fc/Fc⁺) in the presence of KPF₆ in DMF afforded [Cu₃EL]^–, which reacts with CO₂ to yield exclusively C₂O₄^2– (95% yield, TON = 24) and regenerate Cu₃EL.81 Catalysis is observed employing KC₈ and FeCp(C₆Me₆) as reductants, but only in the presence of KPF₆.81

3. Electrocatalytic reduction of CO₂ to CO with metalloporphyrins

Iron tetraphenylporphyrin [Fe(TPP)] (Fig. 1) was reported to catalyse the electrochemical reduction of CO₂ at the Fe^I/Fe⁰ redox potential (–1.64 V vs. SCE) in DMF.82 However, the catalytic efficiency was very low and the catalytic activity of Fe(TPP) was rapidly lost during preparative-scale electrolysis.82 The addition of Mg²⁺ ions to the solution resulted in improved catalytic efficiency, resulting in a CO faradaic yield of ca. 60–70% and the rest in formate.83 The presence of Lewis acids, such as Li⁺, Na⁺, Ba²⁺, and Al³⁺ ions, also improved the catalytic efficiency. The addition of CF₃CH₂OH (1.47 M) resulted in a large increase of the Fe^I/Fe⁰ current to reach an i_p/i_p⁰ value of 131.82 The i_p/i_p⁰ value is given by eqn (8),

i_p/i_p⁰ = 2.24(k_app[CO₂]RT/Fv)^1/2 8

where k_app is the rate constant, R is the gas constant, T is the absolute temperature, F is the Faraday constant, and v is the scan rate.83 The i_p/i_p⁰ value was proportional to the concentration of CF₃CH₂OH at low concentrations, indicating that k_app exhibits second-order dependence on [CF₃CH₂OH]: k_app ∝ [CF₃CH₂OH]².83 Such second-order dependence of k_app on [CF₃CH₂OH] suggests that the rate-determining step in the electrocatalytic reduction of CO₂ to CO involves a two-step protonation of the CO₂ adduct of [PFe⁰]^2– ([PFe^IICO₂]^2–) by AH to produce [PFe^IICO] with dehydration as shown in Scheme 3, where AH = CF₃CH₂OH.83,84 The CO₂ adduct of an Fe⁰ porphyrin ([PFe^IICO₂]^2–) was characterised by vibration at 590 cm^–1 and CO bending mode at 806 cm^–1, which are both sensitive to ¹³C substitution.85 The [PFe^IICO₂]^2– complex, which could only be observed at –95 °C, was easily protonated by methanol even at –80 °C to produce [PFe^IIC(OH)O]^–.85

Fig. 1. Iron tetraphenylporphyrin derivatives used for electrocatalytic reduction of CO₂ to CO. Reprinted with permission from ref. 86. Copyright 2012, The American Association for the Advancement of Science.

Scheme 3. Catalytic cycle for two-electron reduction of CO₂ to CO with an iron porphyrin ([PFe^I]^–).84.

The electrocatalytic efficiency of CO₂ reduction to CO was improved by the introduction of phenolic groups in all ortho and ortho′ positions of the phenyl groups of iron tetraphenylporphyrin.86 The electrocatalytic reduction of CO₂ with iron 5,10,15,20-tetrakis(2′,6′-dihydroxyphenyl)porphyrin (FeTDHPP in Fig. 1) in the presence of 2 M H₂O in CO₂-saturated DMF ([CO₂] = 0.23 M) afforded a CO faradaic yield of 94% with a turnover number (TON) of 5.0 × 10⁷ for 4 hours of electrolysis at –1.16 V vs. SHE with no observed degradation.86 The average current density was 0.31 mA cm^–2 that corresponds to a turnover frequency (TOF) of 3.2 × 10³ s^–1 at an overpotential of 0.466 V, which is the most efficient among other CO₂ reduction catalysts.86–90 When the hydroxy groups in FeTDHPP were replaced by methoxy groups (FeTDMPP in Fig. 1), the catalytic activity of FeTDMPP was decreased by a factor of around 1 billion as compared to that of FeTDHPP with hydroxyl groups.86 Therefore, the enhanced catalytic activity of FeTDHPP was attributed to the high local concentration of protons associated with the phenolic hydroxy substituents.86

The introduction of four positively charged trimethylanilinium groups on the phenyl groups of iron tetraphenylporphyrin resulted in further improvement of the electrocatalytic activity of CO₂ reduction by means of coulombic stabilization of the initial Fe⁰–CO₂ adduct.88 When four positively charged trimethylanilinium groups were introduced at the ortho positions of the TPP groups (Fe-o-TMA in Fig. 2), the maximum TOF was as high as 10⁶ s^–1 at a low overpotential of 0.220 V.88 The catalyst standard potential (E0cat) of Fe-o-TMA was determined to be –0.944 V vs. SHE, which is the most positive ever reported for an iron porphyrin CO₂-reduction catalyst.88 The para-substituted analogue (Fe-p-TMA in Fig. 2) exhibited an E0cat value of –1.263 V vs. SHE, which is much more negative than that of Fe-o-TMA, due to smaller coulombic stabilization of the initial Fe⁰–CO₂ adduct.88 The importance of the coulombic stabilization for the electrocatalytic activity of CO₂ reduction was further demonstrated by the introduction of four negatively charged sulfonate groups on the phenyl groups of iron tetraphenylporphyrin (Fe-p-PSULF in Fig. 2), which resulted in a more negative E0cat value (–1.428 V vs. SHE).88

Fig. 2. Iron tetraphenylporphyrin derivatives with positive and negative charges used for electrocatalytic reduction of CO₂ to CO. Reprinted with permission from ref. 88. Copyright 2016, American Chemical Society.

In contrast to the electrocatalytic reduction of CO₂ in nonaqueous aprotic solvents such as DMF (vide supra), the electrocatalytic reduction of CO₂ in water always competes with the reduction of aqueous protons to hydrogen (H₂). The TOF of the electrocatalytic CO₂ reduction in an aqueous solution was reported to be much smaller compared to that in DMF solution because of the lower solubility of CO₂ in water.88 A successful example of selective electrocatalytic reduction of CO₂ to CO over proton reduction was reported by employing Fe-p-TMA (Fig. 2) as a catalyst. This preparative-scale electrolysis was performed in water at pH 6.7 (adjusted by KOH addition) with an applied potential of –0.97 V vs. NHE for 4 h.88 The average faradaic yields were CO (90%), H₂ (7%), acetate (1.4%), formate (0.7%) and oxalate (0.5%).88 When pH was adjusted to 3.7 by the addition of 0.1 M formic acid buffer, the electrolysis with Fe-p-TMA resulted in the exclusive proton reduction to produce H₂ over the CO₂ reduction.88

The selective electrocatalytic reduction of CO₂ to CO (over proton reduction) under acidic conditions was made possible with a cobalt(ii) chlorin complex, Co^II(Ch), adsorbed on multi-walled carbon nanotubes (MWCNTs) as a catalyst (Fig. 3).91 The electrolysis of a CO₂-saturated aqueous solution (pH = 4.6 with Na₂SO₄) employing a glassy carbon working electrode modified with Co^II(Ch)@MWCNT (0.01 μmol) afforded mainly CO with a maximum TON of 1500 and TOF of 100 h^–1 and little H₂ at –1.1 V vs. NHE (=–1.34 V vs. SCE).91 The faradaic yield of CO for the initial 2 h was determined to be as high as 89%, whereas that for H₂ production was only 11% at pH 4.6.91 The high selectivity for CO was maintained at pH 3.6, although H₂ became the main product at pH 2.0.91

Fig. 3. (a) Structure of Co^II(Ch) and (b) schematic image of Co^II(Ch) on MWCNTs. Reprinted with permission from ref. 91. Copyright 2015, Royal Society of Chemistry.

When MWCNTs were replaced by reduced graphene oxide (rGO), which is a planar π-system, as a support material of Co^II(Ch), the CO yield became significantly smaller (TON = 350 for CO and 250 for H₂ at 20 h) (Fig. 4b) compared to that with MWCNTs (Fig. 4a).91 Thus, the three dimensional assembly of MWCNTs with Co^II(Ch) (Fig. 3b) on the electrode surface is essential for the selective electrocatalytic reduction of CO₂ to CO.91 The π–π interaction between MWCNTs and Co^II(Ch) is presumed to provide a suitable hydrophobic environment for more selective binding of CO₂ over protons compared to the system with two-dimensional rGO as a support material.91

Fig. 4. Time profiles of the formation of CO and H₂ in the electrocatalytic reduction of CO₂ on a glassy carbon electrode modified with Co^II(Ch) (0.01 μmol) adsorbed on (a) MWCNTs (13 μg) and (b) rGO (13 μg) in a CO₂-saturated aqueous solution (pH 4.6) containing Na₂SO₄ (5.0 mM) at an applied potential of –1.34 V vs. SCE. Reprinted with permission from ref. 91. Copyright 2015, Royal Society of Chemistry.

FeTDHPP (Fig. 1), a known CO₂ reduction catalyst in DMF, also exhibits electrocatalysis in a CO₂-saturated aqueous solution (pH 7.3, NaHCO₃ 0.5 M) at an applied potential of –1.03 V vs. NHE (η = 480 mV), selectively producing CO over H₂.92 A surface immobilisation tactic was implemented for the FeTDHPP catalyst on carbon surfaces by removing one phenyl group and appending a pyrene unit through a short linker.92 This immobilisation boosted both selectivity (96 : 4 CO : H₂ ratio) and overall conversion (97% total faradaic yield).92

The cobalt phthalocyanine (CoPc) complex when adsorbed on MWCNTs was demonstrated to be an electrocatalyst for CO₂ reduction to CO at –0.63 V vs. RHE, with a remarkable TON of 9.7 × 10⁴ and a faradaic yield exceeding 90%.93 Compared to CoPc/MWCNT (2.5%), CoPc/rGO (2.2%) and CoPc/CB (3.3%) (CB: carbon black) exhibited less than one-third of the current density at –0.59 V vs. RHE with a 10% lower faradaic yield of CO and inferior stability.93 The higher graphitic nature of CNTs compared to those of either rGO or CB allows stronger π–π interactions with CoPc and higher electron conduction, resulting in superior electrocatalysis.93 A Pc/CNT hybrid without Co afforded a much lower faradaic yield for CO (only 19%), indicating that the Co centres in the CoPc/CNT act as the primary catalyst active sites.85 Low but non-zero conversion of CO₂ to CO on Pc/CNT is attributed to the catalytic activity of Pc itself.93

An iron tetraphenylporphyrin bearing six pendant OH groups in ortho and ortho′ positions on three of the phenyl rings and one carboxylic acid group in the para position of the fourth phenyl was covalently attached to MWCNTs (CATCO₂H, Scheme 4).94 The covalent grafting of an Fe porphyrin on MWCNTs also led to efficient electrocatalytic reduction of CO₂ to CO selectively in water (pH 7.3) at an overpotential of 0.5 V.94

Scheme 4. An iron tetraphenylporphyrin bearing six pendant OH groups in ortho and ortho′ positions on three of the phenyl rings and one carboxylic acid group in the para position of the fourth phenyl covalently attached to MWCNTs. Reprinted with permission from ref. 94. Copyright 2016, Royal Society of Chemistry.

Remarkably, the redox silent zinc(ii) complex of 5,10,15,20-tetramesitylporphyrin (ZnTPP) was also reported as an effective electrocatalyst that afforded faradaic efficiencies as high as 95% for CO₂ reduction to CO at –1.7 V vs. SHE in DMF/H₂O (9 : 1 v/v).95 The UV-vis spectrum of the reduced ZnTPP species exhibits absorption bands at 710, 820, and 920 nm (Fig. 5, red line), which are the characteristic features of a transiently generated Zn complex with reduced ligands (i.e., TPP˙).95 ZnTPP˙ can react with CO₂ to regenerate ZnTPP and unidentified species.95 The exact mechanism of the CO₂ reduction by ZnTPP˙ is yet to be clarified.

Fig. 5. Absorption spectra of ZnTPP (black line), its reduced species (red line) produced by the reduction of ZnTPP by sodium naphthalenide, and the reduced species after exposure to CO₂ (blue line) or air (green line) in THF. Reprinted with permission from ref. 95. Copyright 2017, American Chemical Society.

4. Electrocatalytic reduction of CO₂ to CO with nonheme metal complexes

A nonheme cobalt complex with a macrocyclic aminopyridine (I in Scheme 5) can also catalyse the electrochemical reduction of CO₂ to CO with a faradaic efficiency of 98 ± 2% and 1.22(1) million TON at an applied potential of –2.8 V in a DMF solution containing [ⁿBu₄N][PF₆] (0.1 M) and trifluoroethanol (1.2 M) under a CO₂ atmosphere.96 When the pendant N–H group was replaced by a N–Me group, TONs became 300 times lower than that of I, indicating that the presence of the pendant NH moiety of the secondary amine is crucial for catalysis.96,97 Moreover, the presence of NH groups leads to a positive shift in the reduction potential of the Co^I/0 couple, resulting in a decrease in the overpotential for CO₂ reduction. Complex I is reduced by one electron to generate the Co^I complex, which is further reduced to the Co⁰ form. CO₂ binds to the Co⁰ complex to produce the CO₂ adduct ([Co(CO₂)]⁰, III in Scheme 5), which is stabilised by intramolecular H-bonds of the pendant secondary amines.96 The successive protonation of the CO₂ complex affords CO and H₂O to regeneratecomplex I.96,97

Scheme 5. Catalytic cycle of two-electron reduction of CO₂ to CO with a nonheme cobalt complex with a macrocyclic aminopyridine (I). Reprinted with permission from ref. 96. Copyright 2016, American Chemical Society.

The [Co^II(qpy)(H₂O)₂]²⁺ (qpy = 2,2′:6′,2′′:6′′,2′′′-quaterpyridine) complex also efficiently catalyses the electrochemical CO₂-to-CO conversion in an MeCN solution in the presence of weak Brønsted acids.98 Controlled potential electrolysis (CPE) was performed at –1.1 V vs. SCE in the presence of 3 M PhOH, corresponding to 140 mV overpotential. CO was produced with 96% catalytic selectivity (a small amount of H₂ (4%) was obtained as the only byproduct) and 94% faradaic efficiency.98 The [Co^II(qpy)(H₂O)₂]²⁺ complex is an excellent catalyst, at a very low overpotential, better than active Mn complexes (vide infra),99–103 being only surpassed by the Fe tetraphenylporphyrin bearing four trimethylammonium groups in the ortho position of each phenyl (complex a in Fig. 6).88

Fig. 6. Comparison of the metal catalysts (a,88 b,98 c,88 d,99 e,100 f,101 g,102 and h103) for the CO₂-to-CO electrochemical conversion in DMF or MeCN by means of their catalytic Tafel plots (log TOF as a function of the overpotential (η = E⁰(CO₂/CO) – E). d′ dotted lines: data for complex d in the presence of 0.1 M Mg²⁺.99 ■: TOF value for b. Reprinted with permission from ref. 98. Copyright 2018, American Chemical Society.

Rhenium bipyridine complexes (Re(bpy)(CO)₃Cl and its derivatives)104–108 and manganese complexes (Mn(bpy)(CO)₃Br and its derivatives)109–114 are known to act as catalysts for electrocatalytic reduction of CO₂ to CO. The catalytic mechanisms have been investigated extensively using various spectroscopic methods including UV/Vis absorption and pulsed-EPR techniques (2P-ESEEM and HYSCORE) combined with DFT calculations.113 A key intermediate in the catalytic cycle of CO₂ reduction is demonstrated to be a low spin Mn^II–hydroxycarbonyl complex after the oxidative addition of CO₂ and H⁺ to a Mn⁰ carbonyl dimer (Scheme 6).113

Scheme 6. The catalytic cycle for two-electron reduction of CO₂ to CO with [Mn(bpy)(CO)₃]⁺. Reprinted with permission from ref. 113. Copyright 2014, WILEY-VCH Verlag GmbH.

The catalytic activity of a manganese complex fac-[MnBr(4,4′-bis(phosphonic acid)-2,2′-bipyridine)(CO)₃] (MnP) for CO production was improved when immobilized on a mesoporous TiO₂ electrode (TiO₂/MnP).114 Controlled potential electrolysis of TiO₂/MnP at E_appl = –1.7 V vs. Fc⁺/Fc (η = 0.42 V) under CO₂ in MeCN/H₂O (19/1) for 2 h passed an average charge of 1.10 ± 0.25 Coulombs with a faradaic efficiency of 67 ± 5% and a TON of 112 ± 17 for CO. The faradaic yield of H₂ for this system was 12.4 ± 1.4% and formate was not detected by ion chromatography.114 The high activity and low overpotential of TiO₂/MnP are suggested to result from temporary desorption of the catalyst, followed by dimerization and re-anchoring within mesoporous TiO₂, because phosphonic acid modified molecules, such as MnP, display some lability when bound to TiO₂.115 The high local concentration of MnP may also place the metal centres in an environment where they are predisposed to dimerization upon reduction.114

The lowest overpotential for the electrocatalytic reduction of CO₂ to CO was achieved by using a manganese complex, [Mn{4,4′-di(1H-pyrrolyl-3-propyl carbonate)-2,2′-bipyridine}(CO)₃-MeCN]⁺(PF₆)^– loaded onto conductive MWCNTs ([Mn-MeCN]/MWCNT), which exhibited catalysis at –0.21 V vs. SHE (overpotential of about 100 mV for CO production).116 The MWCNTs together with surface adsorbed K⁺ ions provided an environment to stabilize CO₂ adjacent to the Mn complex and significantly lowered the overpotential for CO₂ reduction in an aqueous solution.116

A case of selectivity tuning was observed as a function of ligand type when Ni(ii) complexes in Fig. 7 were employed as catalysts for CO₂ reduction.117 Controlled potential electrolyses were carried out with each member of the catalyst series (Fig. 7) in CO₂-saturated MeCN solutions containing 2% H₂O.117 The 15-membered macrocyclic complex (3-Ni) gave high selectivity for CO₂ reduction to CO with a faradaic efficiency of 87%, the remaining 11% faradaic yield resulted in H₂ generation at an applied potential of 2.44 V vs. Fc⁺/Fc.116 On the other hand, 1-Ni supported by a non-macrocyclic ligand gave a faradaic efficiency of 93% for H₂ and only 5% for CO, whereas the less rigid 16-membered macrocyclic complex (2-Ni) exhibited moderate selectivity for both CO (56%) and H₂ (43%) evolution in the presence of CO₂ and H₂O.117 Thus, increased rigidity of the redox-active macrocycle leads to enhanced selectivity for CO₂ reduction to CO over the competing hydrogen evolution reaction.117

Fig. 7. Nickel(ii) catalysts supported by non-macrocyclic and macrocyclic bipyridyl-NHC ligands. Reprinted with permission from ref. 117. Copyright 2018, Royal Society of Chemistry.

The first electron reduction of 1-Ni is calculated to be that of the metal center and the second electron reduction is calculated to be ligand-based to produce the [Ni^I(L˙)] species. The [Ni^I(L˙)] complex reacts with H⁺ to form a metal hydride, from which H₂ is produced through a reaction with another H⁺.117 In contrast, the first and second reductions of 3-Ni are calculated to be both ligand-centered, and the resulting [Ni^II(L²˙)] complex is not capable of metal hydride generation, and therefore leads to no H₂ production. The two-electron reduced species of 2-Ni is best described by resonance forms [Ni^II(L²˙)] ↔ [Ni^I(L˙)], which are capable of producing both CO and H₂.117

A nickel complex supported by a pincer-type carbene–pyridine–carbene ligand also exhibited high selectivity for the electrocatalytic reduction of CO₂ to CO over proton reduction.118 A series of Ni^II macrocycle complexes structurally similar to [Ni(cyclam)]²⁺ (cyclam = 1,4,8,11-tetraazacyclotetradecane) also act as selective electrocatalysts for CO production over H₂ at a mercury pool working electrode in an aqueous solution.119–121 At pH 5, with an applied potential of –0.96 V vs. NHE (overpotential of –0.55 V), the Ni complexes are efficient, having high faradaic efficiencies for the selective reduction of CO₂ to CO.118 Hg provides favorable noncovalent dispersive interactions with the cyclam ligand to destabilise the poisoned CO-bound form of the catalyst, leading to enhanced catalytic reactivity.122 The binding of CO₂ to Ni(i) complexes has been extensively studied by pulse radiolysis measurements to demonstrate that the formation of the Ni(i)–CO₂ adduct is in equilibrium between the Ni(i) complex and free CO₂.123,124

5. Electrocatalytic reduction of CO₂ to HCOOH with heme and nonheme metal complexes

The two-electron/two-proton reduction of CO₂ also affords formate (HCOO^–) as well as CO.125 The reduced metal complex (M^–) may react directly with CO₂ to produce the M–CO₂^– adduct that is further reduced with H⁺ to evolve CO and H₂O (Scheme 7) as in the case of the iron porphyrin in Scheme 3.125 Alternatively M^– can react with H⁺ to generate the hydride complex (MH) capable of direct reaction with a proton to produce H₂, or alternatively couple with CO₂ to produce HCOO^– (Scheme 7).125 Whether the catalytic two-electron/two-proton reduction of CO₂ affords CO or HCOO^– depends on the interplay of metals and ligands of the implemented catalyst.126

Scheme 7. The reaction pathway of two-electron reduction of CO₂ with a metal complex (M) to produce HCOO^– or CO. Reprinted with permission from ref. 125. Copyright 2017, American Chemical Society.

A series of metalloprotoporphyrins (MPPs in Fig. 8) were employed as catalysts for the electrochemical reduction of CO₂ in an aqueous solution. Formic acid was not obtained when CrPP, MnPP, CoPP and FePP were used as catalysts. In the case of CoPP, CO was produced selectively at pH 3 with high faradaic efficiency.126 Exercising the reaction with InPP, CrPP, SnPP and GaPP as catalysts, no gaseous products other than H₂ were observed. When SnPP, InPP and RhPP were employed as catalysts, however, significant amounts of formic acid were produced depending on the pH as shown in Fig. 9, where the faradaic efficiency of HCOO^– with InPP at pH 9.6 was optimal, reaching a value close to 70%.126 At very low pHs, hydrogen evolution dominated, resulting in little or no HCOOH production.126 At very alkaline pHs, the H₂ production also seemed to be dominant, leading to poor selectivity towards HCOO^–.126 The catalytic mechanisms for HCOO^– production with InPP, SnPP and RhPP have not yet been clarified.

Fig. 8. Chemical structure of metalloprotoporphyrins (MPP). Reprinted with permission from ref. 126. Copyright 2017, Elsevier.

Fig. 9. Faradaic efficiencies for the electrocatalytic reduction of CO₂ to HCOOH or HCOO^– with RhPP, InPP and SnPP, determined at t = 10 min, at E = –1.5 V vs. RHE as function of pH. Reprinted with permission from ref. 126. Copyright 2017, Elsevier.

In contrast to iron porphyrins that catalyse the two-electron reduction of CO₂ to CO (Scheme 1), a nonheme iron(iii) chloride complex bearing a 6,6′-di(3,5-di-tert-butyl-2-hydroxybenzene)-2,2′-bipyridine (^tbudhbpy) ligand, Fe(^tbudhbpy)Cl, catalyses the electrochemical two-electron reduction of CO₂ to formate in the presence of phenol (0.50 M) as a proton source in DMF with a faradaic efficiency of 68 ± 4% together with H₂ as a minor product (30 ± 10% faradaic efficiency) and minimal CO (1.1 ± 0.3% faradaic efficiency) at –2.5 V vs. Fc/Fc⁺.127 The first one-electron reduction of Fe(^tbudhbpy)Cl (Fe^III/II; E⁰ = –0.89 V vs. Fc/Fc⁺) exhibited a Nernstian PhOH-dependent electrochemical response, which suggests that the reduction is coupled with protonation of a bound phenolate moiety of the ligand framework by a PhOH proton donor as shown in Scheme 8.127 At the second reduction potential (E⁰ = –2.09 V vs. Fc/Fc⁺), a second protonation event occurs at the metal centre to produce the iron(iii)–hydride complex. At the third reduction potential (E⁰ = –2.65 V vs. Fc/Fc⁺), the catalytic activity was observed in the presence of PhOH as a sacrificial proton donor when the Fe(ii)–hydride complex reacts with CO₂ to produce the formate complex from which formate is released to regenerate the Fe(ii) complex (Scheme 8).127 A kinetic isotope effect (KIE) of 4.8 ± 0.9 was observed when PhOD is used instead of PhOH, indicating that the insertion of CO₂ into the Fe(ii)–hydride complex is the rate-determining step.127 Fe(ii)–hydride complexes were reported to hydrogenate CO₂ to produce formate.128–130 An iron cluster hydride [HFe₄N(CO)₁₂]^– is also reported as an effective catalyst for the electrochemical reduction of CO₂ to formate under mild conditions of pH 7 buffered aqueous solution and an applied potential of –1.2 V vs. SCE.131,132

Scheme 8. Catalytic cycle for two-electron reduction of CO₂ to CO with a nonheme iron(iii) chloride complex bearing a 6,6′-di(3,5-di-tert-butyl-2-hydroxybenzene)-2,2′-bipyridine ligand (Fe(^tbudhbpy)Cl) via protonation by phenol. Reprinted with permission from ref. 127. Copyright 2018, American Chemical Society.

Cobalt complexes ([CpCo(PR2NR′2)I]⁺) containing diphosphine ligands (PR2NR′2) with two pendant amine residues catalyse the production of formic acid in DMF/water mixtures with a faradaic efficiency of 90 ± 10% at 500–700 mV overpotentials.133 The [CpCo(PR2NR′2)I]⁺ complex with R = cyclohexyl and R′ = benzyl exhibited the best catalytic activity with a TON for HCOOH of 23 and 15 in the presence of 1.1 and 0.56 M water, respectively, after electrolyses at –2.25 V vs. Fc/Fc⁺ for 1 h.133 The electrocatalytic mechanism for CO₂ reduction to formate is proposed as shown in Scheme 9, where a Co^II–hydride intermediate hydrogenates CO₂ to produce formate. After two successive electron transfers, a Co^I species (complex III in Scheme 9) is generated. This species is then protonated and further reduced to form a cobalt(ii)–hydride (Co^II–H) complex (complex V in Scheme 9).133 DFT calculations indicated that the protonation of the pendant amine in either the Co^II or Co^I states allows two other pathways via intermediates III′ and IV, respectively, through intramolecular proton transfer to the cobalt centre with the reduction of the metal center.133 The Co^II–H species V then reacts with CO₂, generating complex VI. Internal hydride transfer from cobalt to CO₂ then yields complex VII.133 The release of formic acid from complex VII regenerates the Co^II species (complex II in Scheme 9) and completes the catalytic cycle.133

Scheme 9. Catalytic cycle for two-electron reduction of CO₂ to CO with [CpCo(PR2NR′2)I]⁺ complexes containing diphosphine ligands (PR2NR′2) with two pendant amine residues. Reprinted with permission from ref. 133. Copyright 2017, American Chemical Society.

The product selectivity of electrocatalytic reduction of CO₂ to CO vs. HCOOH with the assembly of a [MnBr(2,2′-bipyridine)(CO)₃] complex anchored to a carbon nanotube electrode via a pyrene unit was examined, and reported to be tuned by variation of metal complex density on the nanotube surface, where CO was observed as the main product at high catalyst loadings, and formate was the dominant CO₂ reduction product at low catalyst loadings.134 The formation of a dimeric Mn⁰ species at higher surface loading preferentially leads to CO formation, whereas at lower surface loading the electrochemical generation of a monomeric Mn–hydride is suggested to enhance the production of formate (Scheme 10).134 The Mn–CNT hybrid catalyst exhibited much higher activity for electrocatalytic reduction of aqueous CO₂ with TONs of up to 1790 ± 290 for CO (η = 550 mV) and up to 3920 ± 230 for formate (η = 590 mV),134 compared to electrolyses in the absence of carbon nanotubes.135–137

Scheme 10. Two-electron reduction of CO₂ to CO vs. HCOOH with the [MnBr(2,2′-bipyridine)(CO)₃] complex anchored to a carbon nanotube electrode via a pyrene unit depending on the surface concentration. Reprinted with permission from ref. 134. Copyright 2017, American Chemical Society.

When a Ni(bis-dithiolene) complex with a quinoxaline-pyran-fused dithiolene ligand (qpdt^2–) (see the X-ray crystal structure in Fig. 10) was employed as a CO₂ reduction catalyst using a Hg/Au amalgam electrode, formate as the major product together with minor amounts of CO and H₂ was observed at reasonable overpotentials with good faradaic yield and notable stability.138 The onset potential of the catalytic wave with [Ni(qpdt)₂]^– was observed at approximately –1.7 V vs. Ag/AgCl (=–1.5 V vs. SHE), with a halfwave potential at –1.80 V and a peak at –2 V.138 The overpotential was determined to be ∼340 mV by comparing the experimental onset potential to the standard potential of the CO₂/HCOOH couple in MeCN in the presence of trifluoroethanol as a proton source.138 During the electrolysis, a complete reduction of the ligand occurred, followed by the subsequent pyran ring opening to produce the real catalyst ([Ni^II(L)₂]^2–; L is a pyran ring-opened ligand produced from qpdt^2– by electrolysis) as shown in Scheme 11.138 A Ni–hydride intermediate [Ni^III(H)(L)₂]^2– was proposed to be a catalytically relevant species for CO₂ reduction to formate, which was produced by one-electron reduction of [Ni^II(L)₂]^2–, followed by the reaction of [Ni^I(L)₂]^3– with H⁺.138 The [Ni^III(H)(L)₂]^2– species reacts with CO₂, followed by a release of HCOO^– and regeneration of [Ni^II(L)₂]^2– upon the one-electron reduction at –0.51 V vs. Ag/AgCl.138

Fig. 10. X-ray crystal structure of a Ni^III(bis-dithiolene) complex with a quinoxaline-pyran-fused dithiolene dianion ligand (qpdt^2–) ([Ni^III(qpdt)₂]^–) at 50% probability. Hydrogen atoms are omitted for clarity. Reprinted with permission from ref. 138. Copyright 2018, American Chemical Society.

Scheme 11. Electrochemical reduction of [Ni(qpdt)₂]^–, followed by the subsequent pyran ring opening to produce the real catalyst ([Ni^II(L)₂]^2–; L is a pyran ring-opened ligand produced from qpdt^2– by electrolysis) during electrolysis. Reprinted with permission from ref. 138. Copyright 2018, American Chemical Society.

A small amount of CO observed during the electrolysis was proposed to be produced via the η¹-CO₂ species based on the DFT calculations.138 Although many η²-CO₂ nickel structures have been known to date,139–143 a rare η¹-κC binding mode of CO₂ to Ni was also reported by utilizing an anionic tridentate PNP ligand (PNP^– = N[2-PⁱPr₂-4-Me-C₆H₃]₂^–).144,145 The protonation of the η¹-CO₂ species affords the carboxylate intermediate ([Ni(C(O)OH)(L)₂]^2–), which undergoes a heterolytic C–O bond cleavage to generate [Ni(CO)(L)₂]^– upon protonation, followed by the release of CO.138,146

6. Further electrocatalytic reduction of CO to fuels with metalloporphyrins

The electrochemical reduction of CO₂ to CO and its further reduction to methane were made possible by using a simple Co protoporphyrin molecular catalyst immobilized onto a pyrolytic graphite (PG) electrode (CoPP-PG).147 In an aqueous solution with a pH value of 1, the faradaic efficiency for methane production is larger than that for CO, but the dominant product is H₂.147 At pH = 3, CO became a major product, especially at less cathodic potentials, where the faradaic efficiency for CO was 40%.147 The efficiency towards CO can be further increased by performing the experiment at higher CO₂ pressure (10 atm), which leads to a faradaic efficiency of 60% at a potential of –0.6 V (Fig. 11).147

Fig. 11. Faradaic efficiency (FE) for the electrocatalytic CO₂ reduction to (a) CH₄ and (b) CO in 0.1 M perchlorate solution saturated with CO₂ conducted at each applied potential for 1 h and 90 min at P_CO2 = 1 atm and 10 atm, respectively [yellow bars: pH = 1, P_CO2 = 1 atm; blue bars: pH = 1, P_CO2 = 10 atm; magenta bars: pH = 3, P_CO2 = 1 atm and black bars pH = 3, P_CO2 = 10 atm]. Reprinted with permission from ref. 147. Copyright 2015, Nature Publishing Group.

The mechanism of catalytic CO₂ reduction to CO by CoPP may be similar to that of iron porphyrins in Scheme 1. The reduction of Co^IIPP to [Co^IPP]^– is coupled with CO₂ binding to afford the CO₂ adduct that reacts with water to produce the Co^IIIPP–C(O)OH adduct, which is further reduced to produce the Co^IIPP–CO adduct as shown in Scheme 12.147 The two-electron/two-proton reduction of the Co^IIPP–CO adduct affords the Co^IIPP–HCHO adduct that is further reduced by four electrons with four protons to finally yield methane and water, accompanied by the regeneration of Co^IIPP (Scheme 12).147 The formation of CO is more favoured than H₂ at higher pHs.147 However, further reduction of CO is more favoured at a low pH.147 Thus, the faradaic yield of methane was higher than that of CO at pH 1, while H₂ remained the major product.147

Scheme 12. Catalytic cycle for two electron reduction of CO₂ to CO and further reduction to CH₄ with a Co protoporphyrin molecular catalyst (CoPP) immobilized onto a pyrolytic graphite (PG) electrode (CoPP-PG). Reprinted with permission from ref. 147. Copyright 2015, Nature Publishing Group.

Crystallized copper phthalocyanine supported on carbon black (CuPc/CB) was reported to exhibit high selectivity for C₂H₄ with a maximum faradaic efficiency of 25% under atmospheric pressure at –1.6 V vs. Ag/AgCl (=–1.4 V vs. SHE), while CH₄ and CO were also generated as minor products (Fig. 12).148 In contrast to the crystalline form of CuPc/CB, the noncrystalline CuPc/CB catalyst showed a much lower selectivity and reactivity for C₂H₄ production.148 When noncrystalline CuPc was treated with Milli-Q water and chloroform to completely restore its crystallinity, the restored crystalline CuPc/CB catalyst afforded faradaic efficiency and partial current density values as high as those of the original crystalline CuPc/CB catalyst.148 These results indicate that catalyst crystallinity is crucial for the selective conversion of CO₂ to C₂H₄, because the C–C bond forming step to produce C₂H₄ from CO is likely mediated by two metal centers.148

Fig. 12. Products derived from CO₂ electrochemical reduction by crystalline CuPc/C at –1.6 V vs. Ag/AgCl in a two-compartment H-cell. Reprinted with permission from ref. 148. Copyright 2017, American Chemical Society.

Heterobimetallic cavities formed by the face-to-face coordination of thiol-terminated iron porphyrins on copper electrodes via self-assembly of supramolecular cages (M = Fe in Fig. 13) have made it possible to convert CO to C₂ products via C–C bond formation in aqueous electrochemical CO reduction with high faradaic efficiency (83% total with 57% to ethanol) and current density (1.34 mA cm^–2) at a potential of –0.40 V vs. RHE.149 The cage-functionalized electrodes afforded an order of magnitude improvement in both selectivity and activity for electrocatalytic CO reduction compared to their parent copper surfaces.149 Control analogues that lack thiol binding groups as well as positional isomers favouring edge-on binding or direct van der Waals stacking exhibited reduced surface access and negligible CO over water reduction selectivity, suggesting the critical role of the three-dimensional pockets in catalysis, where the Fe centre can aid in the cooperative reduction of potential acetaldehyde intermediates.149

Fig. 13. Schematic illustration of supramolecular assembly of cages between molecular components and supramolecular assembly of cages between molecular components and a copper electrode. Reprinted with permission from ref. 149. Copyright 2017, American Chemical Society.

7. Photocatalytic reduction of CO₂ with heme and nonheme metal complexes

The electrocatalytic reduction of CO₂ can be replaced by photocatalytic reduction using photocatalysts instead of electrocatalysts. For example, the selective electrocatalytic reduction of CO₂ to CO with a glassy carbon electrode modified with a cobalt(ii) chlorin complex (Co^II(Ch)) adsorbed on MWCNTs (Co^II(Ch)/MWCNTs; see Fig. 3 for the schematic image) is replaced by the photocatalytic reduction of CO₂ with trimethylamine (TEA) using Co^II(Ch)/MWCNTs as a CO₂ reduction catalyst and [Ru^II(Me₂phen)₃]²⁺ (Me₂phen = 4,7-dimethyl-1,10-phenanthroline) as a sensitizer.150 The photocatalytic mechanism of the CO₂ reduction is shown in Scheme 13, where electron transfer from TEA to the excited state of [Ru^II(Me₂phen)₃]²⁺ ([Ru^II(Me₂phen)₃]²⁺*: * denotes the excited state) occurs to produce TEA radical cations and [Ru^I(Me₂phen)₃]⁺ that reduces Co^II(Ch) to [Co^I(Ch)]^–.150 [Co^I(Ch)]^– reacts with CO₂ to produce the [(Ch)Co^III(CO₂)]^– complex that is protonated to produce Co^II(Ch)COOH, from which water is released by the protonation to produce the [Co^III(Ch)CO]⁺ complex.150 After the CO release from the Co(iii)CO complex, the Co(iii) complex is reduced by TEA to regenerate Co^II(Ch). Hydrogen evolution simultaneously occurs as a side reaction, much like in the electrocatalytic case, because [Co^I(Ch)]^– can also react with H⁺ to produce the hydride complex ([Co^III(Ch)(H)]), which then can couple to a H⁺ to produce H₂.151 The TON of the photocatalytic reduction was determined to be 710 with the ratio of CO to H₂ of 2.4 : 1 in 20 h when 5.0 μM of Co^II(Ch) and 1.0 mg of MWCNTs were used.150 The π–π interaction between MWCNTs and Co^II(Ch) may provide a suitable hydrophobic environment for the selective binding of CO₂ over protons, because the binding of CO₂ to the Co(i) complex is required for the formation of CO (Scheme 13).150

Scheme 13. Photocatalytic cycle for two-electron reduction of CO₂ to CO with a cobalt(ii) chlorin complex (Co^II(Ch)) adsorbed on multi-walled carbon nanotubes (Co^II(Ch)/MWCNTs). Reprinted with permission from ref. 150. Copyright 2016, Royal Society of Chemistry.

An iron tetraphenylporphyrin complex functionalized with trimethylanilinium groups (Fe-p-TMA in Fig. 2), which is an efficient and selective molecular electrocatalyst for converting CO₂ to CO,88 can also act as an efficient catalyst in photocatalytic reduction of CO₂ to methane using Ir(ppy) (ppy = phenylpyridine) as a dye under visible light irradiation at ambient temperature and pressure.152,153 The catalytic system, which was evaluated in an acetonitrile solution containing a photosensitizer (Ir(ppy)), sacrificial electron donor (TEA) and proton source (trifluoroethanol), operates stably under irradiation over several days to produce CO, CH₄, and H₂ with turnover numbers (and selectivities) of 367 (78%), 79 (17%) and 26 (5%), respectively in 107 h.152 These values correspond to a methane production rate of 763 μmol per hour per gram of catalyst (μmol h^–1 g^–1), which is larger than those obtained using other catalysts154–159 that generate methane from CO₂. Trifluoroethanol facilitates the C–O bond cleavage step as observed in the electrocatalytic reduction of CO₂ (Scheme 3).84 No other gaseous product, methanol, formaldehyde or formate was produced.152 Isotopic labelling experiments conducted under a ¹²CO₂ or a ¹³CO₂ atmosphere confirmed that methane was produced from CO₂. A two-step procedure, that first reduces CO₂ to CO and then reduces CO to further reduced species of CO, generates methane with a selectivity of up to 82% and a quantum yield (light-to-product efficiency) of 0.18%.152

The photocatalytic mechanism for CO₂ reduction to methane is proposed as shown in Scheme 14, where the starting Fe^III porphyrin is reduced with three electrons by Ir(ppy)* to the catalytically active Fe⁰ species.152 The Fe⁰ species reduces CO₂ to CO through a two-step protonation of the Fe⁰ species by CF₃CH₂OH with dehydration (right-hand iron cycle) as in the case of the electrocatalytic reduction of CO₂ to CO in Scheme 3, where AH = trifluoroethanol.84 The CO produced binds to Fe^II and is further reduced by a total of six electrons by electron transfer from Ir(ppy)* and six protons to generate CH₄ (and H₂O), through a postulated Fe^I-formyl (Fe^ICHO) intermediate (left-hand iron cycle).152

Scheme 14. Photocatalytic cycles for two-electron reduction of CO₂ to CO and further reduction to CH₄ with Fe-p-TMA and Ir(ppy)₃. Reprinted with permission from ref. 152. Copyright 2017, Macmillan Publishers Limited, part of Springer Nature.

Not all photocatalytic systems employ separately a catalyst and a photosensitizer. Fe-p-TMA can also catalyse the photocatalytic reduction of CO₂ to CO with 1,3-dimethyl-2-phenyl-2,3-dihydro-1H-benzo[d]-imidazole (BIH) selectively under visible light irradiation without the assistance of an additional photosensitizer.160 BIH was reported to enhance the photocatalytic formation of CO from CO₂,161,162 due to its high reductive ability (E_ox = +0.33 V vs. SCE = +0.57 vs. SHE),163 its fast deprotonation to BI˙ once oxidized and its overall two-electron donation capacity (BI˙ could be in turn easily oxidized to BI⁺).

BIH was also used as a sacrificial reductant together with triethanolamine (TEOA) for visible-light driven CO₂ reduction in MeCN using [Ru(bpy)₃]²⁺ (bpy = 2,2′-bipyridine) as a photosensitizer and a Cu^II complex bearing 2,2′:6′,2′′:6′′,2′′′-quaterpyridine (qpy) ([Cu(qpy)]²⁺) as a selective reduction catalyst.164 The photocatalytic reaction is greatly enhanced by the presence of H₂O (1–4% v/v), and a TON of >12 400 for CO production has been achieved with 97% selectivity, which is among the highest of molecular 3d-metal CO₂ reduction catalysts.164 It was confirmed that the photocatalyst remains homogeneous based on results from Hg poisoning and dynamic light scattering experiments.164

BIH alone acts as an electron donor in the photocatalytic reduction of CO₂ to CO with a Ni(ii) complex bearing an S₂N₂-type tetradentate ligand, [Ni^II(bpet)]²⁺ (bpet = bis(2-pyridylmethyl)-1,2-ethanedithiol), as a CO₂ reduction catalyst and [Ru(bpy)₃]²⁺ as a photosensitizer in a DMA/H₂O solution mixture (9 : 1 v/v) under a CO₂ atmosphere at 298 K.165 In this photocatalytic system, the [Ni^II(bpet)]²⁺ complex showed a high TON exceeding 700 with a very high CO selectivity of >99% and a quantum yield of 1.42%.165

Composite electrodes formed by a combination of an organic-semiconductor with a metal-porphyrinoid catalyst were also evaluated as photocatalysts for CO₂ reduction. Graphite carbon nitride (g-C₃N₄), which has frequently been used as an organic semiconductor photocatalyst,166–169 was modified with a carboxyl group of tetra(4-carboxyphenyl)porphyrin iron(iii) chloride (FeTCPP) to prepare a g-C₃N₄ nanosheets/FeTCPP (g-C₃N₄/FeTCPP) heterogeneous catalyst. The material was then implemented for the photoreduction of CO₂ to CO with TEOA under visible light illumination in MeCN : H₂O : TEOA (3 : 1 : 1) (Scheme 15).170 Similar to the case shown in Scheme 13, Fe^IIITCPP was reduced by the photo-induced electrons produced in g-C₃N₄ nanosheets up to Fe⁰TCPP that reduced CO₂ to CO coupled with proton transfer, whereas the holes left in g-C₃N₄ nanosheets were reduced by TEOA.170 A maximum rate of 6.52 mmol g^–1 in 6 h and a selectivity up to 98% for CO production have been achieved using the g-C₃N₄/FeTCPP heterogeneous catalyst.170

Scheme 15. Photocatalytic reduction of CO₂ to CO with TEOA using g-C₃N₄ nanosheets/FeTCPP (g-C₃N₄/FeTCPP). Reprinted with permission from ref. 170. Copyright 2018, Elsevier.

To further improve the catalytic performance of composite electrodes, rutile TiO₂ nanoparticles were employed as modifiers to enhance interfacial charge transfer between semiconducting carbon nitride nanosheets (NS–C₃N₄) and a catalyst (supramolecular Ru(ii)–Re(i) binuclear complex (RuRe)).171 The RuRe/TiO₂/NS-C₃N₄ hybrid photocatalyzed CO₂ reduction to CO with high selectivity under visible light (λ > 400 nm) irradiation, exhibiting higher catalytic activity compared to an analogue without TiO₂ by a factor of 4, in terms of both the CO formation rate and the TON.171 The enhanced photocatalytic activity was attributed mainly to the prolonged lifetime of free and/or shallowly trapped electrons generated in TiO₂/NSC₃N₄ under visible-light irradiation, as revealed by transient absorption spectroscopy.171

A case of metal porphyrinoid catalysis was reported in conjunction with an inorganic semiconductor photoanode, completing a full photoelectrochemical cell. The photoelectrochemical reduction of CO₂ was performed in a two-compartment cell composed of an FeO(OH)/BiVO₄/FTO photoanode and a Co^II(Ch)/MWCNT cathode, which are connected with conducting wire as an external circuit and separated by a Nafion membrane (Fig. 14).172 A photo-driven oxidation reaction (of water) occurs at the photoanode, and the generated electrons are transported through the external circuit, being supplied as reducing equivalents to the Co^II(Ch)/MWCNT catalyst at the cathode. The photocatalytic controlled potential electrolysis of a CO₂-saturated aqueous solution (pH 4.6) using a photoelectrochemical cell in Fig. 14 at an applied bias voltage of –1.3 V at the cathode versus the photoanode resulted in the formation of CO and H₂ as shown in Fig. 15a, where the CO yield is significantly higher than the H₂ yield.172 The maximum current efficiency for CO production for the initial 2 h was 83% at pH 4.6.172 The amount of O₂ produced in the photoelectrochemical oxidation of water is one-half of the amounts of the sum of CO and H₂ produced in the electrochemical reduction of CO₂ and H₂O on the cathode (Fig. 15b).172

Fig. 14. Schematic illustration of a photoelectrochemical cell composed of an FeO(OH)/BiVO₄/FTO photoanode for the photocatalytic oxidation of water to O₂ and a Co^II(Ch)/MWCNT cathode for the catalytic reduction of CO₂ to CO. Cathode and anode compartments are separated by a Nafion membrane. Reprinted with permission from ref. 172. Copyright 2017, American Chemical Society.

Fig. 15. Time profiles of (a) the formation of CO (red circles) and H₂ (black circles) and (b) the formation of O₂ (blue circles) and CO plus H₂ (red circles) in the photo-assisted CPE with the Co^II(Ch)-modified cathode at an applied bias voltage of –1.3 V at the cathode versus the FeO(OH)/BiVO₄/FTO photoanode in a CO₂-saturated aqueous solution containing Na₂SO₄ (5.0 mM) at pH 4.6 under simulated 1 sun (AM 1.5G) illumination at 298 K. Reprinted with permission from ref. 172. Copyright 2017, American Chemical Society.

A direct covalent linkage between a photosensitizer and a catalyst is a popular strategy due to the ease of electron transfer induced by proximity, and also one that is well suited for metal porphyrinoid catalysts owing to their synthetic versatility and flexibility. Such a case is shown in Scheme 16, where a Ru-based dye and a Re-based catalyst pair is synthesised and immobilised on a semiconductor surface. The photoelectrochemical reduction of CO₂ to CO was performed by using a RuRe/CuGaO₂ electrode under irradiation (λ_ex > 460 nm), which can be selectively absorbed by the Ru photosensitizer unit of RuRe, in an aqueous solution containing NaHCO₃ (50 mM) saturated with CO₂ as shown in Scheme 16.173 The difference in the observed current between the irradiation and dark conditions indicated that the cathodic photoresponse of the RuRe/CuGaO₂ electrode started at +0.30 V vs. Ag/AgCl (=+0.50 V vs. SHE), which was approximately 0.4 V more positive (due to the photovoltage supplied by the underlying semiconductor) than that of RuRe/NiO (ca. –0.1 V vs. Ag/AgCl = +0.1 vs. SHE).173 The TON for the formation of CO was 125 based on the RuRe loading, and the total faradaic efficiency for the production of CO plus H₂ was 81%.173 The wavelength dependence of the incident-photon-to-current efficiencies of the RuRe/CuGaO₂ electrode agreed well with the absorption spectrum of the electrode, whereas the CuGaO₂ electrode alone (without the dye and catalyst) exhibited almost no photoresponse under irradiation (λ_ex > 460 nm).173 The photocurrent was generated by the injection of the electrons from the CuGaO₂ electrode into the excited Ru photosensitizer unit of RuRe, as shown in Scheme 16, because the flat band potential of the CuGaO₂ electrode in the reaction solution (+0.47 V vs. Ag/AgCl = +0.67 V vs. SHE) is less positive than the one-electron reduction potential of the excited state of RuRe (E_red = +0.49 V vs. Ag/AgCl = +0.69 V vs. SHE).173

Scheme 16. Schematic illustration for the photocatalytic reduction of CO₂ to CO with the RuRe/CuGaO₂ electrode. Reprinted with permission from ref. 173. Copyright 2017, Royal Society of Chemistry.

8. Conclusions

The one-electron reduction of CO₂ to oxalic acid is catalysed not only by metalloporphyrins such as Ag^II(TPP) an Pd^II(TPP) but also by nonheme dicopper and tricopper complexes. Iron porphyrins act as efficient catalysts for both the electrocatalytic and photocatalytic reduction of CO₂ to CO in competition with the proton reduction to H₂. When four positively charged trimethylanilinium groups were introduced at the ortho positions of the TPP phenyls of an iron tetraphenylporphyrin (Fe-o-TMA in Fig. 2), the lowest overpotential of 0.220 V for CO₂ reduction to CO was achieved with a maximum TOF of 10⁶ s^–1. A variety of metal complexes such as cobalt and nickel complexes also act as effective catalysts for two-electron reduction of CO₂ to CO. When indium protoporphyrin was employed as an electrocatalyst, formate (HCOO^–) was the main product instead of CO and the faradaic efficiency of HCOO^– at pH 9.6 was close to 70%. Whether CO or HCOOH is produced depends on the metals and ligands of metalloporphyrins.

Further reduction of CO to methane has been made possible by using a Co protoporphyrin molecular catalyst immobilized onto a pyrolytic graphite (PG) electrode (CoPP-PG) in a purely aqueous electrolyte solution. The reduction of CO to C₂H₄ was also made possible by using crystallized copper phthalocyanine supported on carbon black (CuPc/CB) with a maximum faradaic efficiency of 25% under atmospheric pressure at –1.4 V vs. SHE, while CH₄ and CO were produced as minor products. The photocatalytic reduction of CO₂ to CO occurs in competition with the proton reduction to H₂ using triethylamine as a reductant, Co^II(Ch)/MWCNTs as a CO₂ reduction catalyst, and [Ru^II(Me₂phen)₃]²⁺ as a photosensitizer. When Fe-p-TMA with four positively charged trimethylanilinium groups introduced at the para positions of the TPP phenyls was employed as a catalyst, the photocatalytic reduction of CO using triethylamine as a reductant and Ir(ppy) afforded methane with a selectivity of up to 82% and a quantum yield of 0.18% under visible light irradiation at ambient temperature and pressure. Water can be used as an electron and proton source for the photocatalytic reduction of CO₂ to CO using a two-compartment cell composed of an FeO(OH)/BiVO₄/FTO photoanode and a Co^II(Ch)/MWCNT cathode. There are many reports on the photocatalytic reduction of CO₂ with water as an electron and proton source to produce CO, HCOOH and CH₄ using heterogenous catalysts.174–184 It is highly desired to develop photocatalytic systems of CO₂ reduction to fuels such as ethylene, methanol and methane using water as an electron and proton source using homogeneous molecular catalysts as well.

Conflicts of interest

There are no conflicts to declare.

Biographies

Shunichi Fukuzumi

Shunichi Fukuzumi earned a bachelor's degree and a Ph.D. degree in applied chemistry from the Tokyo Institute of Technology in 1973 and 1978, respectively. After working as a postdoctoral fellow from 1978 to 1981 at Indiana University in USA, he became an Assistant Professor in 1981 at Osaka University where he was promoted to a Full Professor in 1994. His research has been focused on electron transfer chemistry, particularly artificial photosynthesis. He is currently a Distinguished Professor at Ewha Womans University, a Designated Professor at Meijo University, and a Professor Emeritus at Osaka University.

Yong-Min Lee

Yong-Min Lee received his M.S. and Ph.D. degrees in Inorganic Chemistry from Pusan National University, Republic of Korea, under the supervision of Professor Sung-Nak Choi in 1999. After he worked at the Magnetic Resonance Centre (CERM) at the University of Florence, Italy, as a Postdoctoral fellow and Researcher under the direction of Professors Ivano Bertini and Claudio Luchinat from 1999 to 2005, he joined the Centre for Biomimetic Systems at Ewha Womans University, as a Research Professor in 2006. He is currently a Special Appointment Professor at Ewha Womans University since 2009.

Hyun S. Ahn

Hyun S. Ahn earned his undergraduate degrees in chemistry and chemical engineering from the University of Washington in Seattle (2008). He then pursued his Ph.D. studies at the University of California at Berkeley, in the field of inorganic chemistry under the supervision of Prof. T. Don Tilley. Upon conferral of his Ph.D. in 2013, Hyun continued his research as a postdoctoral scholar at the University of Texas at Austin with Prof. Allen J. Bard. He is currently an assistant professor of chemistry at Yonsei University. Hyun's research interest spans across the fields of catalysis, photoelectrochemistry, bio-electrochemistry, and energy conversion technologies.

Wonwoo Nam

Wonwoo Nam earned his B.S. (Honours) degree in Chemistry from California State University, Los Angeles, and his Ph.D. degree in Inorganic Chemistry from the University of California, Los Angeles (UCLA) under the supervision of Professor Joan S. Valentine in 1990. After working as a postdoctoral fellow at UCLA for one year, he became an Assistant Professor at Hong Ik University in 1991. In 1994, he moved to Ewha Womans University, where he is currently a Distinguished Professor. His current research has been focused on the dioxygen activation, water oxidation, and important roles of metal ions in bioinorganic chemistry.