Interfacing Formate Dehydrogenase with Metal Oxides for the Reversible Electrocatalysis and Solar‐Driven Reduction of Carbon Dioxide

Abstract

The integration of enzymes with synthetic materials allows efficient electrocatalysis and production of solar fuels. Here, we couple formate dehydrogenase (FDH) from Desulfovibrio vulgaris Hildenborough (DvH) to metal oxides for catalytic CO₂ reduction and report an in‐depth study of the resulting enzyme–material interface. Protein film voltammetry (PFV) demonstrates the stable binding of FDH on metal‐oxide electrodes and reveals the reversible and selective reduction of CO₂ to formate. Quartz crystal microbalance (QCM) and attenuated total reflection infrared (ATR‐IR) spectroscopy confirm a high binding affinity for FDH to the TiO₂ surface. Adsorption of FDH on dye‐sensitized TiO₂ allows for visible‐light‐driven CO₂ reduction to formate in the absence of a soluble redox mediator with a turnover frequency (TOF) of 11±1 s⁻¹. The strong coupling of the enzyme to the semiconductor gives rise to a new benchmark in the selective photoreduction of aqueous CO₂ to formate.

Electrocatalytic‐ and solar‐driven fuel synthesis from the greenhouse gas CO₂ is a desirable approach to simultaneously produce sustainable energy carriers and combat increasing atmospheric CO₂ levels.1 Formate is a stable intermediate in the reduction of CO₂ and can be used as liquid energy carrier in fuel cells, as a hydrogen storage material, or feedstock for the synthesis of fine chemicals.2 Metals and synthetic molecular systems have been widely studied as electrocatalysts for CO₂ reduction to formate, but largely lack the required efficiency, selectivity or affordability to enable carbon capture and utilization technologies.3, 4

There is avid research into both biological and artificial CO₂ fixation. Semi‐artificial photosynthesis provides a common stage for these contrasting approaches as components from synthetic and biological origin can be combined in hybrid model systems.5 To date, enzyme‐based visible‐light‐driven CO₂ reduction to formate relies on diffusional mediators, such as methyl viologen (MV²⁺) and nicotinamide adenine dinucleotide (NAD⁺).6, 7 Mediated processes are inefficient as they consume energy, are kinetically slow, and cause short‐circuit reactions. MV²⁺ is toxic to microorganisms,8 and NAD⁺ is prohibitively expensive for fuel production.6

In this work, we selected wild‐type formate dehydrogenase (FDH) from Desulfovibrio vulgaris Hildenborough (DvH) as it has previously displayed robustness and high activity for the oxidation of formate in solution assays,10, 11 and the electrochemical reduction of CO₂.12 Initially, protein film voltammetry (PFV) was employed to study the interfacial electron transfer between FDH and porous indium‐doped tin oxide (ITO) and TiO₂ electrodes in the absence of a mediator. Immobilization and loading of FDH on TiO₂ were then investigated using a quartz crystal microbalance (QCM) and attenuated total reflection infrared (ATR‐IR) spectroscopy. FDH was finally coupled directly to dye‐sensitized TiO₂ nanoparticles for the selective photocatalytic reduction of CO₂ to formate in a diffusional mediator‐free colloidal system (Figure 1).

Figure 1.

Schematic CO₂ conversion with a dye–semiconductor–FDH photocatalyst system. Photoexcited electrons from the dye, RuP in (A) or DPP in (B), are transferred via the conduction band (CB) of TiO₂ across the enzyme–material interface through the intraprotein [4Fe–4S] relays to the W‐active site of FDH for the reduction of CO₂ to formate. The oxidized dye is regenerated by triethanolamine (TEOA). A protein structure homologous to DvH FDH is shown.9

The electrocatalytic activity of FDH on metal‐oxide electrodes was studied by PFV on mesoporous ITO (mesoITO) and TiO₂ (mesoTiO₂) electrodes with a film thickness of approximately 2.5 μm (Supporting Information, Figure S1).13 FDH (21.5 μm) was activated by incubation with the reducing agent dl‐dithiothreitol (DTT, 50 mm)9 and the resulting solution (2 μL) was drop‐cast on the electrode surface. The FDH‐modified electrode was placed in an electrolyte solution containing CO₂/NaHCO₃ and KCl at pH 6.5 under a CO₂ atmosphere.

Figure 2 A shows the electrochemically reversible interconversion of CO₂ and formate by FDH immobilized on a conductive mesoITO electrode (mesoITO|FDH). The onset potential for both CO₂ reduction and formate oxidation was observed close to the thermodynamic potential (E ^0′=−0.36 V vs. standard hydrogen electrode (SHE), pH 6.5),14 demonstrating that interfacial electron transfer by the [4Fe–4S] relays and catalysis at the W‐active site are highly efficient.15 Similar electrochemically reversible characteristics have previously only been reported for FDHs from Escherichia coli and Syntrophobacter fumaroxidans on graphite electrodes.14, 16, 17

Figure 2.

PFV (ν=5 mV s⁻¹) showing A) reversible reduction of CO₂ to formate by mesoITO|FDH (blue trace) and B) CO₂ reduction by mesoTiO₂|FDH before (blue) and after 2 h CPE (black). Inset: CPE at −0.6 V vs. SHE. Conditions for A and B: 43 pmol FDH (amount drop‐cast), 100 mm CO₂/NaHCO₃, 50 mm KCl, 20 mm formate (only present in A), 1 atm CO₂, pH 6.5, 25 °C, Pt counter electrode. Dashed traces show control experiments of FDH‐free electrodes.

When FDH was immobilized on a semiconducting mesoTiO₂ electrode (mesoTiO₂|FDH), a similar onset potential for CO₂ reduction (−0.4 V vs. SHE) was observed and the current density reached −100 μA cm⁻² at −0.6 V vs. SHE (Figure 2 B). Formate oxidation could not be observed for mesoTiO₂|FDH electrodes as TiO₂ behaves as an insulator at the required potentials. Controlled‐potential electrolysis (CPE) at −0.6 V vs. SHE for 2 h produced formate with a Faradaic efficiency of (92±5)% (Figure 2 B, inset). Comparison of PFV scans before and after CPE showed that approximately 90 % of the initial FDH activity remains after 2 h, demonstrating the excellent stability of the immobilized enzyme.

The interaction of FDH and TiO₂ was quantitatively investigated with a previously described QCM cell.18, 19 Upon flowing an FDH‐containing solution over a planarTiO₂‐covered quartz chip (12 nm in 100 mm TEOA), the surface of TiO₂ reached saturation after 1 h, resulting in approximately 3.5 pmol cm⁻² of adsorbed FDH (planarTiO₂|FDH, Figure 3 A). The strength of the enzyme–TiO₂ interaction was probed by exposing the planarTiO₂|FDH electrode to buffer solutions with different ionic strengths. Rinsing the QCM cell with an enzyme‐free solution for 1 h desorbed only 6 % of the preloaded FDH. Upon increasing the KCl concentration to 0.5–3.0 m KCl, 70–60 % of FDH remained adsorbed on the TiO₂ surface. The finding that 60 % FDH remained adsorbed on TiO₂ after multiple rinsing steps with high KCl concentrations suggests a contribution from chemisorption to the attachment of the enzyme.20, 21 Amino‐acid residues exposed on the FDH surface are likely involved in binding. For example, aspartic and glutamic acid have previously been suggested to form a strong interaction with TiO₂.22, 23

Figure 3.

A) QCM analysis of the adsorption process of FDH on a planarTiO₂‐coated quartz chip. Conditions: 12 nm FDH, 100 mm TEOA, open circuit potential of −0.1 to 0.0 V vs. SHE, pH 6.5, 25 °C, N₂ atmosphere, circulation (0.141 mL min⁻¹). Inset: Desorption of FDH by replacing the solution with fresh solution (100 mm TEOA) and subsequent increase of the ionic strength (each condition was held for 1 h). Error bars correspond to standard deviation (N=3). B) ATR‐IR absorbance spectra of the amide‐band region of FDH during the adsorption process over time onto a planarTiO₂‐coated Si prism (100 nm thickness). Arrows indicate successively recorded spectra of every 1.5 min up to 7.5 min and subsequently every 30 min. Conditions: 1.0 μm FDH, 100 mm TEOA, total volume: 150 μL, open circuit potential, pH 6.5, 25 °C.

The adsorption of FDH was also probed by surface‐selective ATR‐IR spectroscopy using a Si prism coated with a planar or a mesoTiO₂ layer (100 or 400 nm thickness, respectively). After the addition of FDH to the buffer solution covering the planarTiO₂ (Figure 3 B) or mesoTiO₂ (Supporting Information, Figure S2) coated prism, the two characteristic amide I and amide II bands of the protein backbone structure were detected at 1650 cm⁻¹ and 1545 cm⁻¹, respectively.24 The protein adsorption was monitored in situ over 2 h of incubation time and no (in the case of planarTiO₂) or slight (in the case of mesoTiO₂) changes to the band features in the amide‐band region were observed, suggesting a mainly retained backbone structure of FDH on the surface of TiO₂. During the adsorption process, amide I and amide II band intensities showed an increase over time (Figure 3 B). The majority of FDH remained adsorbed on the surface of planarTiO₂ (Supporting Information, Figure S3) upon increasing the ionic strength of the buffer, which agrees with the QCM experiments (Figure 3 A, inset) and supports a stronger than purely electrostatic interaction between FDH and TiO₂.

After establishing the strong interface between FDH and TiO₂, visible‐light‐driven CO₂ reduction to formate was investigated with FDH immobilized on dye‐sensitized TiO₂ nanoparticles (dye|TiO₂|FDH, Figures 1 and 4). The colloidal system was self‐assembled by adding FDH (pre‐activated with DTT) to a suspension of TiO₂ nanoparticles containing TEOA and a phosphonate group‐bearing dye, either a ruthenium tris‐2,2′‐bipyridine complex (RuP) or a diketopyrrolopyrrole (DPP) at pH 6.5 and 25 °C under N₂ atmosphere (to protect the enzyme from aerobic damage). Both dyes are known to adsorb onto TiO₂ via their phosphonate‐anchoring groups and DPP provides a precious‐metal‐free alternative to RuP.25 CO₂ was introduced to the solution via the addition of NaHCO₃. Upon UV‐filtered irradiation, the photoexcited dye injects electrons into the conduction band (CB) of TiO₂ (E _CB(TiO₂)=−0.67 V vs. SHE at pH 6.5),25 whereupon the electrons are conveyed to the catalytic W‐center of FDH to drive CO₂ reduction. The oxidized dye is regenerated by the sacrificial electron donor (Figure 1).26

Figure 4.

Photocatalytic CO₂ reduction to formate with FDH in a colloidal dye‐sensitized TiO₂ system. Conditions: 12 nm FDH, 10 mm DTT, 0.83 mg mL⁻¹ TiO₂, 16.7 μm dye (RuP or DPP), 100 mm TEOA, 100 mm NaHCO₃, pH 6.5, 25 °C, total volume: 1.0 mL, assembled in an anaerobic glove box, UV‐filtered simulated solar‐light irradiation: 100 mW cm⁻², AM 1.5G, λ>420 nm. Error bars correspond to standard deviation (N=3). Dashed traces show control experiments in the absence of FDH.

The dye|TiO₂|FDH systems showed stable formate production for approximately 6 h (Figure 4). The formation of gaseous or dissolved side‐products was not detected by gas chromatography, ion chromatography, and ¹H NMR spectroscopy. The activity of RuP|TiO₂|FDH was not limited by the amount of dye or the light intensity (Supporting Information, Figures S4 and S5). A solution assay monitoring the activity of FDH by UV/Vis spectroscopy (via formate oxidation in presence of 2 mm MV²⁺) showed that approximately 36±7 % FDH remained active after 24 h of photocatalysis (Supporting Information, Figure S6), suggesting that inactivation of FDH is likely the main reason for activity loss. The addition of MV²⁺ as a soluble redox mediator to RuP|TiO₂|FDH showed that not all FDH present in the system is accessible to direct electron transfer across the enzyme–material interface (Supporting Information, Figure S7). Control experiments demonstrated that all components are essential for formate production (Supporting Information, Figures S8 and S9) and support oxidative quenching and “through‐particle” electron transfer as depicted in Figure 1 (Supporting Information, Figures S10 and S11).26 Isotopic‐labeling studies confirmed that formate was produced from CO₂ (Supporting Information, Figure S12).

For photocatalytic experiments, an enzyme loading of approximately 0.03 pmol cm⁻² was calculated assuming that all FDH is adsorbed on TiO₂ with a surface area of 50 m² g⁻¹. Saturation of the TiO₂ surface with FDH in the QCM experiment was only observed when two orders of magnitude higher amounts of FDH were adsorbed (Figure 3 A). As QCM and ATR‐IR spectroscopy indicate stronger than purely electrostatic interactions, close‐to‐quantitative adsorption of FDH on the TiO₂ nanoparticle in the colloidal system is likely. A turnover frequency (TOF) of 11±1.0 and 5±0.6 s⁻¹ (based on CO₂ conversion after 6 h) and approximately 4.9±0.2 and 2.0±0.2 μmol formate (after 24 h) were observed from CO₂ using RuP and DPP‐sensitized TiO₂, respectively (Figure 4). The results of all photocatalysis experiments are presented in Tables S1 and S2 in the Supporting Information.

Table 1 shows a comparison of state‐of‐the‐art catalysts (enzymatic and synthetic) in combination with dye‐sensitized TiO₂ nanoparticles without diffusional mediators for CO₂ reduction and H₂ evolution. Previous studies showed that enzymes outperform the synthetic systems in terms of TOF.30 Among the compared systems, the presented RuP|TiO₂|FDH system exhibits the highest TOF for CO₂ reduction. The DPP|TiO₂|FDH system shows that comparable activities can also be achieved in an entirely precious‐metal‐free system. In semi‐artificial systems, rapid electron transfer from TiO₂ to the enzyme was previously found to be essential for efficient catalysis,22, 31 suggesting that the strong interfacial interaction plays an important role for the high activity and stability of dye|TiO₂|FDH. Previously reported photocatalyst systems employing NAD⁺‐dependent FDHs for CO₂ reduction to formate rely on soluble redox mediators and only produced TOFs in the range of 10–20 h⁻¹.32

Table 1.

Comparison of TOFs for dye‐sensitized TiO₂ systems with enzymatic and synthetic catalysts for CO₂ reduction and H₂ evolution.

reaction	dye	catalyst	TOF [h−1]	ref.
CO2 → HCO2 −	RuP	DvH FDH [a]	4.0×104	this work
	DPP	DvH FDH [a]	1.8×104	this work
CO2 → CO	RuP	Ch CODH I[b]	5.4×102	27
	dye[c]	Re [d]	8.6	28
H+ → H2	RuP	Db [NiFeSe]‐H2ase[e]	1.8×105	22
	DPP	Db [NiFeSe]‐H2ase[e]	8.7×103	25
	CNx [f]	Db [NiFeSe]‐H2ase[e]	2.8×104	23
	RuP	Ni [g]	3.2×102	29

[a] W‐FDH from DvH. [b] Carbon monoxide dehydrogenase (CODH) I from Carboxydothermus hydrogenoformans (Ch). [c] (E)‐2‐cyano‐3‐(5′‐(5′′‐(p‐(diphenylamino)phenyl)thiophen‐2′′‐yl)thiophen‐2′‐yl)‐acrylic acid. [d] Synthetic rhenium catalyst (Re) in N,N‐dimethyl formamide (DMF) and water. [e] [NiFeSe]‐hydrogenase from Desulfomicrobium baculatum (Db). [f] Polyheptazine carbon nitride polymer melon (CN_x). [g] Synthetic nickel(II) bis(diphosphine) catalyst (Ni).

In summary, FDH immobilized on metal‐oxide electrodes is established as a reversible electrocatalyst for the selective conversion of CO₂ to formate. The porous metal‐oxide scaffolds allow for high FDH loading and consequently high current densities, which makes the protein‐modified electrodes not only a relevant model system for CO₂ utilization, but also for formate oxidation in formate fuel cells. An excellent interface between TiO₂ and FDH is confirmed by QCM analysis and ATR‐IR spectroscopy. The direct (diffusional mediator‐free) electron transfer across the enzyme–metal‐oxide interface is exploited for visible‐light‐driven CO₂ reduction to formate. These results underline the importance of characterizing the interactions at the enzyme–material interface and future improvements in performance may arise from more controlled immobilization and more efficient electron transfer with the directly wired FDH.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

M. Miller, W. E. Robinson, A. R. Oliveira, N. Heidary, N. Kornienko, J. Warnan, I. A. C. Pereira, E. Reisner, Angew. Chem. Int. Ed. 2019, 58, 4601.