C–H bond formation with CO2 to selectively form products such as formate (HCOO–) is an important step in harnessing CO2 emissions as a carbon-neutral or carbon-negative renewable energy source. In this report, we show that the iron carbonyl cluster, [Fe4N­(CO)12]−, is an electrocatalyst for the selective reduction of CO2 to formate in water (pH 5–13). With low applied overpotential (230–440 mV), formate is produced with a high current density of 4 mA cm–2 and 96% Faradaic efficiency. These metrics, combined with the long lifetime of the catalyst (>24 h), and the use of the Earth-abundant material iron, are advances in catalyst performance relative to previously reported homogeneous and heterogeneous formate-producing electrocatalysts. We further characterized the mechanism of catalysis by [Fe4N­(CO)12]− using cyclic voltammetry, and structurally characterized a key reaction intermediate, the reduced hydride [HFe4N­(CO)12]−. In addition, thermochemical measurements performed using infrared spectroelectrochemistry provided measures of the hydride donor ability (hydricity) for [HFe4N­(CO)12]− in both MeCN and aqueous buffered solution. These are 49 and 15 kcal mol–1, respectively, and show that the driving force for C–H bond formation with CO2 by [HFe4N­(CO)12]− is very different in the two solvents: +5 kcal mol–1 in MeCN (unfavorable) and −8.5 kcal mol–1 in water (favorable).

The design of electrocatalysts that will selectively transfer hydride equivalents to either H(+) or CO2 to afford H2 or formate is a long-standing goal in molecular electrocatalysis. In this Forum Article, we use experimentally determined thermochemical parameters, hydricity and pKa values, to rationalize our observations that the carbide-containing iron carbonyl cluster [Fe4C(CO)12](2-) reduces H(+) to H2 in the presence of CO2 in either acetonitrile (MeCN), MeCN with 5% water, or buffered water (pH 5-13), with no traces of formate or other carbon-containing products observed. Our previous work has shown that the closely related nitride-containing clusters [Fe4N(CO)12](-) and [Fe4N(CO)11(PPh3)](-) will also reduce H(+) to H2 in either MeCN with 5% water or buffered water (pH 5-13), but upon the addition of CO2, they selectively generate formate. The thermochemical measurements on [Fe4C(CO)12](2-) predict that the free energy for transfer of hydride, in MeCN, from the intermediate [HFe4C(CO)12](2-) to CO2 is thermoneutral and to H(+) is -32 kcal mol(-1). In water, these values are less than -19.2 and -8.6 kcal mol(-1), respectively. These results suggest that generation of both H2 and formate should be favorable in aqueous solution and that kinetic effects, such as a fast rate of H2 evolution, must influence the observed selectivity for generation of H2. The hydride-donating ability of [HFe4N(CO)12](-) is lower than that of [HFe4C(CO)12](2-) by 5 kcal mol(-1) in MeCN and by at least 3 kcal mol(-1) in water, and we speculate that this more modest reactivity provides the needed selectivity to obtain formate. We also discuss predictions that may guide future catalyst design.

Molecular approaches to the electrocatalytic reduction of CO2 to formate are varied and versatile in their methods. We discuss recent efforts to catalyse this reaction including significant progress made in the last 5 years. This Feature Article begins with a survey of molecular electrocatalysts that produce CO or H2, but have been observed under certain conditions to afford some formate. These examples are included because they provide valuable mechanistic insight for design of catalysts that produce hydrogenated products selectively from CO2. The subsequent discussion describes catalyst properties that favour C-H bond formation with CO2 and this is followed by recent advances that have been made in developing these catalysts. The focus on specific catalyst systems includes recently reported Ir PCP-type pincer complexes and Fe carbonyl clusters, such as [Fe4N(CO)12](-), that selectively produce formate from CO2 in aqueous solution. A discussion of the relevant thermochemical properties of the catalysts in the context of formate production is included.

Thermochemical insights are often employed in the rationalization of reactivity and in the design of electrocatalysts for CO2 reduction reactions targeting C–H bond-containing products. This work identifies experimental methods for assessing kinetic aspects of reactivity. These methods are illustrated using [Fe4N­(CO)12]−, which produces formate from CO2 at −1.2 V versus SCE in either a MeCN/H2O solvent (95:5) or pH 6.5 buffered water. Elementary rates for each reaction step are identified along with the rate-determining step (RDS) as C–H bond formation. Transition state kinetics were determined from an Eyring analysis for the rate-determining C–H bond formation step using temperature-dependent electrochemical measurements. A lower measured ΔG ⧧ (298 K, 12.3 ± 0.1 kcal mol–1) in a pH 6.5 aqueous solution, compared with a ΔG ⧧(298 K) of 15.0 ± 0.1 kcal mol–1 in a MeCN/H2O solvent (95:5), correlates with faster observed reaction rates and provides a kinetic rationalization for the solvent-dependent chemistry. Taken together, the experimentally determined kinetic insights highlight that enhancement of the local concentration of CO2 at catalyst–hydride sites should be a primary focus of ongoing catalyst design. This will both enhance reaction rates and increase selectivity for C–H bond formation over competing H–H bond formation, because that step is fast in H2 evolution reactions.

A common approach to speeding up proton transfer (PT) by molecular catalysts is manipulation of the secondary coordination sphere with proton relays and these enhance overall reaction rates by orders of magnitude. In contrast, heterogeneous electrocatalysts have band structures that promote facile PT concerted with electron transfer (ET), known as the Volmer mechanism. Here, we show that [Co13C2(CO)24]4–, containing multiple Co–Co bonds to statistically enhance observed rates of PT, promotes PT on the order of 2.3 × 109 M–1 s–1 which suggests a diffusion-limited rate. The fast ET and PT chemistry is attributed to the delocalized electronic structure of [Co13C2(CO)24]4–. Electrochemical characterization of [Co13C2(CO)24]4– in the presence and absence of protons reveals ET kinetics and diffusion behavior similar to other small clusters such as nanomaterials and fullerenes.

The role of low-lying π* orbitals in dye-sensitized solar cells based on mesoporous thin films of anatase TiO(2) nanocrystallites remains unknown. Herein we report three ruthenium compounds, cis-Ru(dcbq)(2)(NCS)(2), cis-Ru(dcbq)(bpy)(NCS)(2), and cis-Ru(dcb)(bq)(NCS)(2), where bpy is 2,2'-bipyridine, dcb is 4,4'-(CO(2)H)(2)-2,2'-bipyridine, bq is 2,2'-biquinoline, and dcbq is 4,4'-(CO(2)H)(2)-2,2'-biquinoline, that were synthesized, characterized, and contrasted with the well-known N3 compound (i.e., cis-Ru(dcb)(2)(NCS)(2)) in dye-sensitized solar cells. These compounds maintain the same cis-Ru(NCS)(2) core with a systematic variation in the energy of the π* orbitals of the diimine ligand: bpy > dcb > bq > dcbq. The lowered π* orbitals resulted in enhanced red absorption relative to N3. With HCl pretreated TiO(2) in regenerative solar cells, sensitization from 400 to 900 nm was realized with cis-Ru(dcb)(bq)(NCS)(2) and global power conversion efficiencies as high as 6.5% were achieved under 1 sun of AM 1.5 irradiation. The energy conversion efficiency was found to be acutely sensitive to the presence of p-tert-butylpyridine (TBP) in a 0.5 M LiI/0.05 M I(2) acetonitrile electrolyte. Nanosecond transient absorption studies revealed that the addition of TBP decreased the excited-state injection yield for the compounds with biquinoline ligands. Spectro-electrochemical studies showed that the HCl pretreatment lowered the effective density of TiO(2) acceptor states and confirmed that the presence of TBP raised them toward the vacuum level. There was no spectroscopic data to support the hypothesis that the π* levels of the diimine ligand mediate back-electron transfer to the oxidized dye or the redox mediator was found.

Fe 4 N(CO) 12 ] − is a first-row transition element electrocatalyst that selectively produces C−H bonds to give formate from CO 2 in water at −1.2 V vs SCE. We present a thermochemical analysis which probes the possibility that [H-Fe 4 N(CO) 12 ] 2− ((H-1) 2− ) is an intermediate in this process: we show that (H-1) 2− is accessible at −1.2 V vs SCE, but if it were formed, we predict that it would generate H 2 .[Fe 4 N(CO) 12 ] 3− and (H-1) 2− were interrogated spectroscopically, and the product of CO loss, [Fe 4 N(CO) 9 (μ-CO) 2 ] 3− , was synthesized and characterized. Ultimately, we demonstrate that (H-1) 2− is an unlikely participant in the catalytic transformation of CO 2 to formate.

The metal-to-ligand charge transfer (MLCT) excited states of two related heteroleptic Ru(ii) compounds [Ru(bpy)2(deeb)](2+) and [Ru(bpy)2(deebq)](2+), where bpy is 2,2'-bipyridine, deeb is 4,4'-(CO2CH2CH3)2-2,2'-bipyridine and deebq is 4,4'-(CO2CH2CH3)2-2,2'-biquinoline, were characterized in fluid acetonitrile by temperature dependent photoluminescence spectroscopies as well as quenching by iodide ions. Photoluminescence emanates from a manifold of thermally equilibrated excited states referred to as the thexi states. Evidence for activated internal conversion to a 4(th) MLCT excited state was garnered from an Arrhenius analysis of temperature dependent lifetime data. The activation energy was found to be 550 cm(-1) for [Ru(bpy)2(deeb)](2+)* and 1200 cm(-1) for [Ru(bpy)2(deebq)](2+)*. The pre-exponential factor abstracted from the Arrhenius analysis of the [Ru(bpy)2(deebq)](2+)* data suggested that ligand field excited states might be populated, however there was no evidence for ligand loss photochemistry under the conditions studied. The excited states were found to quench iodide by a dynamic process in good agreement with the Stern-Volmer model. Transient absorption data showed that the quenching mechanism was electron transfer to generate an iodine atom and a reduced ruthenium compound as products. The quenching rate constants abstracted from temperature dependent Stern-Volmer quenching data were corrected for diffusion and activated complex formation to yield electron transfer rate constants that were found to increase markedly with temperature. An Arrhenius analysis of the electron transfer data revealed that electron transfer from iodide to the d-orbitals of the excited state was an activated process with an Ea of 2400 cm(-1) for [Ru(bpy)2(deeb)](2+) and 3300 cm(-1) for [Ru(bpy)2(deebq)](2+).