Volume 16, Issue 17 e202202312
Review
Open Access

Renewable Hydrogen Production and Storage Via Enzymatic Interconversion of CO2 and Formate with Electrochemical Cofactor Regeneration

Eleftheria Sapountzaki

Eleftheria Sapountzaki

Biochemical Process Engineering, Division of Chemical Engineering, Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, SE-97187 Luleå, Sweden

Search for more papers by this author
Prof. Ulrika Rova

Prof. Ulrika Rova

Biochemical Process Engineering, Division of Chemical Engineering, Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, SE-97187 Luleå, Sweden

Search for more papers by this author
Prof. Paul Christakopoulos

Prof. Paul Christakopoulos

Biochemical Process Engineering, Division of Chemical Engineering, Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, SE-97187 Luleå, Sweden

Search for more papers by this author
Prof. Io Antonopoulou

Corresponding Author

Prof. Io Antonopoulou

Biochemical Process Engineering, Division of Chemical Engineering, Department of Civil, Environmental and Natural Resources Engineering, Luleå University of Technology, SE-97187 Luleå, Sweden

Search for more papers by this author
First published: 11 May 2023

Graphical Abstract

Formic acid (FA) is a promising hydrogen carrier. Formate dehydrogenase catalyzes the CO2 reduction to FA with need of a cofactor that can be regenerated efficiently in a bioelectrochemical system. Bacterial enzyme complexes are capable of dehydrogenating FA to produce hydrogen. This Review summarizes relevant research to propose a biocatalytic CO2 recycling cycle for H2 production, storage, and release.

Abstract

The urgent need to reduce CO2 emissions has motivated the development of CO2 capture and utilization technologies. An emerging application is CO2 transformation into storage chemicals for clean energy carriers. Formic acid (FA), a valuable product of CO2 reduction, is an excellent hydrogen carrier. CO2 conversion to FA, followed by H2 release from FA, are conventionally chemically catalyzed. Biocatalysts offer a highly specific and less energy-intensive alternative. CO2 conversion to formate is catalyzed by formate dehydrogenase (FDH), which usually requires a cofactor to function. Several FDHs have been incorporated in bioelectrochemical systems where formate is produced by the biocathode and the cofactor is electrochemically regenerated. H2 production from formate is also catalyzed by several microorganisms possessing either formate hydrogenlyase or hydrogen-dependent CO2 reductase complexes. Combination of these two processes can lead to a CO2-recycling cycle for H2 production, storage, and release with potentially lower environmental impact than conventional methods.

1 Introduction

Climate change and its already evident consequences create an urgent need to reduce CO2 emissions, which is also portrayed in international and European environmental regulations, aiming for carbon neutrality by 2050 and negative emissions from then on.1 Fossil fuels are mainly responsible for CO2 emissions, hence it is important to target activities that can establish independency and offer more sustainable solutions by using clean energy sources.2

Over the last decades, mitigation actions have been mainly focused on the capture and storage of CO2. Although this has helped to reduce the increasing amounts of greenhouse gas emissions, it has not tackled the root of the problem, which is the ever-increasing consumption of non-renewable resources and the subsequent release of CO2. To address this, new technologies to capture CO2 and transform it into valuable products need to be developed. CO2 transformations can be performed by chemical methods, photochemical reductions, chemical and electrochemical reductions, biological conversions, reforming and inorganic transformations.3-6

Hydrogen is emerging as a green energy carrier, as it produces no undesirable emissions when burned. It has a high gravimetric, but a low volumetric energy density under normal conditions (33.3 kW h kg−1 and 2.5 W h L−1 respectively). To increase its volumetric energy density, hydrogen is being either pressurized or liquefied. However, both these solutions are very energy intensive since high pressures and low temperatures are required. Moreover, the storage, transportation and handling of pressurized or liquid hydrogen can be dangerous, require expensive materials, and cannot directly utilize the existing infrastructure for storage and distribution of natural gas.7-10 Formic acid (FA) can provide a solution to these problems as a potential hydrogen carrier, since it contains 4.4 wt % H2 and has a volumetric density of 53 gurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0001  L−1, which corresponds to 1.77 kW h L−1, while being liquid in ambient conditions and therefore easier and safer to handle.11

The low solubility of CO2 in water (1.45 g L at 1 atm at 298 K) hampers its capture and conversion. CO2 mass transfer can be sped up by transforming the dissolved CO2 into other products. CO2 reduction to FA addresses this problem and can serve both CO2 sequestration and the development of hydrogen-related technologies. Moreover, formate is reported to have a higher economic value compared to other potential CO2 reduction products, such as ethanol, methanol or carbon monoxide.12, 13 Conventionally, CO2 is reduced to FA using both homogeneous and heterogeneous chemical catalysts. Although achieving high catalytic rates, they often require harsh conditions, such as temperatures between 120 and 200 °C and pressures of 5 to 6 MPa.14-16 On the other hand, the same reaction can be catalyzed by biocatalysts, namely formate dehydrogenase (FDH). Enzymatic catalysis is a promising alternative because it is highly selective and occurs at mild conditions, therefore being less energy intensive.17, 18

The same is the case for the release of H2 from formic acid, which is concomitant with the release of CO2.11, 19 Homogeneous catalysts have been reported to reach high H2 production rates, but often require expensive metals and high temperatures.20 Heterogeneous catalysts facilitate product separation but also involve precious metals, achieve lower rates and show lower specificity.19, 21, 22 Using bacteria which have been found to possess enzymes catalyzing this reaction can improve the process by bringing the aforementioned benefits of biocatalysis.23 More specifically, the enzyme complexes reported to be responsible for FA dehydrogenation are formate hydrogenlyase (FHL) and hydrogen dependent CO2 reductase (HDCR).

The first reversible hydrogen storage and release system reported utilizes a homogeneous iridium catalyst (more specifically a dinuclear Cp*Ir catalyst) which is being deprotonated at high pH and further reprotonated at low pH to store and release hydrogen.14 The reversible CO2 to formate cycle for hydrogen storage has been proposed with the use of various chemical catalysts.24 Recently, an alternative biological equivalent cycle has been reported by Schwarz et al.25 In that system, an enzyme complex from Acetobacterium woodii is used to convert H2 and CO2 to formate and vice versa. However, this is a hydrogen storage and release and not a hydrogen production system, since CO2 and H2 were supplemented as gases for conversion to FA.

The present literature review aims to collectively present the current advances on the electrochemical hydrogenation of CO2 via biocatalysis and the enzymatic production of H2 from formate, proposing that the two steps could be combined to create a biocatalytic cycle for H2 production, chemical storage, and release. Moreover, if this cycle would be implemented, industrial off-gas could be utilized as the input stream, leading to a highly pure H2 stream as an output. This concept is hence dependent on an efficient purification step, where the CO2 from the industrial off-gas is selectively captured. Carbonic anhydrase (CA), a remarkably fast enzyme catalyzing CO2 hydrogenation to bicarbonate, can be used for CO2 adsorption to isolate it from other industrial off-gas components.26 Subsequently, captured CO2 can either be desorbed and fed in the form of purified gas as a substrate for FA production, or used directly as substrate, in the solubilized form of bicarbonate. CA can also be employed in a CO2 capture step as a last part of the cycle. As mentioned above, H2 release occurs through FA oxidation to H2 and CO2. Therefore, enzymatic CO2 capture can allow for gas separation and recycling of the produced CO2 back into the cycle. The complete proposed cycle, presented in Figure 1, could provide an alternative solution for hydrogen storage and release, while recycling CO2.

Details are in the caption following the image

Proposed cycle for biocatalytic hydrogen production, storage, and controlled release. Steps 1 & 2 could be merged into one.

2 Reduction of CO2 to Formate via Enzymatic Electrosynthesis (EES) in Bioelectrochemical Systems (BES)

Electrocatalytic reactions, currently in a renaissance, offer a great potential for the sustainable production of fuels and chemicals from biomass. Electrochemical reduction of CO2 under abiotic conditions requires large overpotentials that reduce the conversion efficiency.

Several electrochemical systems that carry out this reaction have been studied, but no large-scale process has been developed. Regarding electrode design, gas-diffusion type electrodes appear more promising for such an application due to higher current efficiencies and better mass transfer.27 The catalyst also plays a major role in the efficiency of the system. Both metal containing and non-metal containing electrocatalysts have been reported for CO2 reduction to formate. The former can be based on either transition metals (e. g., Ni, Rh)28 or p-block metals (e. g., Sn, Bi).29, 30 Encouraging faradaic efficiencies have been calculated at a wide range of applied potentials, but low selectivity, as well as toxicity of the catalysts remain significant concerns.31

Bioelectrochemical systems (BES) have gained increased attention as a promising route, both to generate energy from biochemical reactions and to produce chemicals through the utilization of electricity and biocatalysts (microorganisms or enzymes). The use of oxidoreductases in enzymatic electrosynthesis (EES) takes advantage of the high specificity and mild operating conditions of enzymes, while allowing for the required redox equivalents to be provided directly or indirectly by renewable electricity.32

A BES in its basic form consists of an anode and a cathode placed in an electrolyte solution. The anodic and cathodic chambers are divided by a membrane that allows the bidirectional movement of ions. Another indispensable part of the system is the biocatalyst, which, in the case of EES, is present in the cathode. For the system to operate, electricity needs to be provided in the form of applied potential. The applied potential must exceed the difference between the potential of the substrate reduction reaction and the oxidation of the electron-donating species. Moreover, the applied potential affects the reaction kinetics, as higher overpotential motivates faster electron transfer.33

Electrons are generated in the anode through an oxidation reaction and then flow towards the cathode. The oxidation reaction occurring at the anode is mainly oxygen evolution reaction, described by the equation: 2 H2O→O2+4 H++4 e.34

After catalyzing the reaction in the cathode, the enzyme ends up with less electrons than its original form, which need to be replenished. In the most basic form of a BES, these electrons can be provided directly from the cathode electrode, by direct electron transfer (DET), so no additional component is needed in the system. This can be quite challenging and requires short distance between the two components and appropriate orientation to ensure accessibility of the electrons to the active site of the enzyme. Another alternative is mediated electron transfer, in which the required reducing equivalents are provided by the reversible oxidation/reduction of additional molecules. In that case, an enzyme cofactor and potentially also a mediator for cofactor regeneration are present in the system.35 Cofactors are defined as small non-protein molecules that aid enzymes in catalyzing reactions.36 Mediators are small organic compounds that act as electron carriers through their reversible oxidation and reduction. Therefore, they lower the applied potential required for cofactor regeneration.37

A schematic representation of an example of a BES for EES is presented in Figure 2.

Details are in the caption following the image

Basic structure of a BES for CO2 reduction to formate.

2.1 Main components of a BES

2.1.1 Formate dehydrogenase (FDH)

FDHs belong to the class of oxidoreductases and form a broad category of enzymes found in various organisms, including bacteria, yeasts and plants, that exhibit great heterogeneity in the number of redox-active centers and overall structure.38 They can be bound to cellular membranes or free in the cytoplasm or periplasm.39-42 Physiologically, FDHs provide energy to cells, being important for methylotrophic metabolism,43, 44 and the first step of the Wood–Ljungdahl pathway in acetogens.45 In pathogenic microorganisms and plants, they have been reported to protect the cell from stress.46 Most FDHs catalyze the oxidation of formate to CO2 [Eq. (1)], in a reaction that is, however, reversible.47
urn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0002(1)
urn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0003(2)
urn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0004(3)

FDHs that exhibit a preference for the reverse reaction [Eq. (2)] of CO2 reduction to formate have also been reported.50, 51 Overall, there have been indications that bacterial FDHs have a higher efficiency at reducing CO2 compared to eukaryotic FDHs, which could be attributed to structural differences, such as an additional loop at the N-terminal end of the former.39, 50 As seen in Equation (1), the reduction of CO2 to formate requires 2 electrons and one proton.52 These are provided by a redox partner, which is usually nicotinamide adenine dinucleotide (NADH), as seen in Equation (3). Some enzymes also utilize cytochromes, ferredoxins, Coenzyme F420 or membrane quinols.38, 51

Taking the active site structure, reaction mechanism and related cofactors into consideration, FDHs can be divided into two categories. The first is FDHs that require NAD+ for their function (EC 1.17.1.9), usually found in aerobic organisms and therefore oxygen tolerant. These are mostly homodimers and do not contain metals (except for the FDH of Pseudomonas oxalaticus, which is an oxygen sensitive Fe containing heterodimer). They are usually mentioned as NAD+ dependent, or non-metal containing FDHs. The other category includes enzymes that contain metals in their active site, namely molybdenum or tungsten, and can operate utilizing different electron acceptors. These FDHs are referred to as NAD+- linked (or NAD+-independent) or metal-containing (EC 1.17.98.4). They usually occur in anaerobic organisms, are oxygen sensitive, and have monomeric, dimeric or trimeric structures.53

NAD+-dependent FDHs do not exhibit great variety in their structure and overall characteristics, as opposed to the metal-containing ones, which can differ significantly from one another.54 Moreover, the fact that the latter contain metals in their active site contributes to their electroactivity.55 W-containing FDHs appear to have higher CO2 reducing ability, while Mo-containing FDHs prefer formate oxidation.56 It is also mentioned that the lower reduction potential of W4+ compared to Mo4+ makes the W-containing FDHs more competent in reducing CO2 compared to the Mo-containing FDHs.44

The active site structure of NAD+ dependent and metal containing FDHs differs, so the reaction mechanism also differs. Another distinction observed in the active site structure within metal containing FDHs (apart from whether they contain molybdenum or tungsten) is the presence of a SeCys or Cys residue. It is suggested that SeCys is preferable for CO2 reduction than Cys. Arginine and histidine residues are also regarded as key catalytic residues and are present in all metal-containing FDHs.38, 56

The mechanism proposed for metal-dependent FDH-catalyzed CO2 reduction is hydride transfer (Figure 3). The active site is initially in its reduced form, with the metal (Mo- or W-) having an oxidation number of 4+. Arginine and potentially also histidine residues are responsible for the attachment and proper orientation of the CO2 molecule in the active site, by forming hydrogen bonds with its oxygen atoms. Its carbon atom also forms a hydrogen bond with the H of the −SH group. Subsequently, the hydrogen atom attacks the CO2 molecule and hydride transfer occurs, with the metal of the active site being oxidized to is 6+ state. Formate is released, and the electrons provided externally by the system reduce the site metal to its initial 4+ state, which leads to protonation of the sulfur atom and the restoration of the catalytic center.38

Details are in the caption following the image

Proposed reaction mechanism of metal containing FDHs. Figure reproduced from Ref. [38] (copyright 2016, with permission from Elsevier).

The catalytic mechanism for NAD+-dependent FDHs is not clearly described. However, investigation of the interactions of the enzyme with the substrate and NADH indicate a hydride transfer mechanism between the C4 atom of the pyridine ring of NADH to CO2 (Figure 4). The substrate and the cofactor are led to a favorable orientation for this transfer to occur, through hydrogen bond interactions with key active site residues, namely Ile122, Asn146, Thr282, Arg284, Asp308, His332, Ser334 and the pair His332-Gln313.57 Moreover, it has been reported that there is a specific order in the binding of the cofactor and the substrate to the active site of NAD+- dependent FDHs, with the former binding first, when the reaction happens in the direction of formate oxidation.58 The highest efficacy of metal independent FDHs in catalyzing formate oxidation compared to CO2 reduction has been attributed to the enzyme structure and the reaction mechanism. More specifically, studying the electrostatic characteristics of the system indicated that hydride transfer from the carbon atom of the substrate to the NAD+ is more favorable than the reverse reaction.54

Details are in the caption following the image

Proposed reaction mechanism for NADH-dependent FDHs.

Another characteristic of FDHs important for the design of a biochemical process is their oxygen tolerance. Overall, NAD+ dependent FDHs are regarded to be more likely to be functional under aerobic conditions. However, several metal containing FDHs have also shown indication of activity in the presence of oxygen, such as enzymes from Rhodobacter or Clostridium species, which have been assayed under aerobic conditions.59-61 Moreover, FDH from Desulfovibrio species have been reported to tolerate aerobic conditions, with their activity being improved by activation through pre-incubation in the presence of dithiothreitol (or other thiol-containing species, such as mercaptoethanol).62-64 Although the exact role of thiols is not clearly explained, their effect is attributed to the fact that they react with oxygen and lower the redox potential of the reaction.65

Parameters such as the catalytic rate and CO2 reduction efficiency, the ability to utilize other cofactors and oxygen tolerance are important for the selection of the biocatalyst in a BES. The properties of various FDHs that have been tested for their CO2 reducing activity are presented in Table 1.

Table 1. Properties of CO2-reducing FDHs.

Microorganism

Enzyme[a]

Category

Oxygen tolerance

CO2 reduction[b]

Formate oxidation

Reference

kcat [s−1]

Km [mM]

kcat/Km

[s−1 mM−1]

conditions

substrate[c]

cofactor

kcat

[s−1]

Km [mM]

Kcat/Km

[s−1 mM−1]

Candida boidinii

CbFDH

NAD+ dependent

tolerant

0.0091

27.3

0.0003

pH 7, 25 °C, anaerobic

NaHCO3

NADH

0.153

[66]

tolerant

0.015

31.28

0.0005

pH 7, 25 °C, aerobic

NaHCO3

NADH

1.081

8.55

0.126

[50]

Candida methylica

CmFDH

NAD+ dependent

tolerant

0.008

0.78

0.0103

pH 7.5, 25 °C, aerobic

NaHCO3

NADH

1.31

7.01

0.187

[67]

Chaetomium thermophilum

CtFDH

NAD+ dependent

tolerant

0.023

0.32

0.0719

pH 8, 25 °C, aerobic

NaHCO3

NADH

2.039

3.3

0.618

[67]

Thiobacillus sp. KNK65MA

TsFDH

NAD+ dependent

tolerant

0.318

9.23

0.0345

pH 7, 25 °C, aerobic

NaHCO3

NADH

1.796

16.24

0.111

[50]

Methylobacterium extorquens AM1

W-containing

tolerant

no kinetic characterization

pH 7, aerobic, −0.75 V vs Ag/AgCl

CO2 gas, purged

NADH

No kinetic characterization

[68]

Cupriavidus necator

FdsABG

Mo-containing

tolerant

11

2.7

4.07

pH 7, 30 °C, anaerobic conditions

CO2 gas, presaturated

NADH

201

0.31

649

[69]

Rhodobacter capsulatus

RcFDH

Mo-containing

tolerant

1.48

pH 9, 30 °C, aerobic conditions

NaHCO3

NADH

36.48

0.281

130

[59]

Rhodobacter aestuarii

RaFDH

Mo-containing

tolerant

0.805

pH 7, 30 °C, aerobic conditions

CO2 gas, purged

NADH

15.6

[60]

Clostridium ljungdahlii

ClFDH

Mo-containing

tolerant

0.012

7.3

0.0016

pH 9, 25 °C, aerobic

NaHCO3

NADH

0.246

1.4

0.176

[70]

5.66

66.2

0.0855

pH 7, 35 °C, aerobic

NaHCO3

NADH

3.93

19.91

0.197

[71]

Myceliophthora thermophila

MtFDH

NAD+ dependent

tolerant

0.1

0.4

0.25

pH 6.0, 25 °C, aerobic

NaHCO3

NADH

0.3

7.2

0.0417

[72]

Clostridium autoethanogenum

CaFDH

W-containing

tolerant

4

23.1

0.173

pH 7, 30 °C, aerobic

NaHCO3

NADH

1.04

4.51

0.231

[61]

Clostridium coskatii

CcFDH

W-containing

tolerant

5.62

59.7

0.0941

pH 7, 30 °C, aerobic

NaHCO3

NADH

0.62

5.57

0.111

Clostridium ragsdalei

CrFDH

W-containing

tolerant

3.28

31.2

0.105

pH 7, 30 °C, aerobic

NaHCO3

NADH

11.88

44.83

0.265

Clostridium formicoaceticum

Not characterized

no evidence

no kinetic characterization

pH 7, 37 °C, anaerobic

NaHCO3

MV+

No kinetic characterization

[73]

Clostridium carboxidivorans strain P7T

W-containing

tolerant

0.08

0.05

1.6

pH 6.8, 37 °C, aerobic

NaHCO3

NADH

>100

[74,75]

Desulfovibrio desulfuricans

DdFDH

Mo-containing

tolerant

46.6

15.7

2.97

pH 7.0, 22 °C, anaerobic

Na2CO3

MV+/BV+ (not specified)

543

57.1

9.51

[76]

Desulfovibrio vulgaris Hildenborough

DvH-FDH

W-containing

tolerant

315

0.42

750

pH 7, RT, anaerobic

NaHCO3

MV+

1310

0.017

77515

[62]

Shewanella oneidensis MR-1

Not characterized

tolerant

no kinetic characterization

pH 7, RT,

−0.75 V vs Ag/AgCl

CO2 gas, purged

MV+

No kinetic characterization

[77]

Escherichia coli

EcFDH

Mo-containing

no evidence

<1

pH 7.5, 23 °C, anaerobic

Na2CO3

MV+

4

[78]

Syntrophobacter fumaroxidans

SfFDH (FHD1)

W-containing

sensitive

112

pH 5.9, 37 °C, anaerobic, −0.8 V vs Ag/AgCl

Na2CO3

No kinetic characterization

[48]

pH 7.3, 37 °C, anaerobic

NaHCO3

MV+

930

0.04

23250

[56]

Pseudomonas oxalaticus

NAD+ dependent

sensitive

180

40

4.5

pH 6.2, 10 °C, anaerobic

14CO2 gas, absorbed by soda asbestos, purged

NADH

No kinetic characterization

[79]

  • [a] Name used for the respective FDH studied in each work. [b] kcat: Enzyme turnover number. It correspons to the maximum substrate-to-product conversions performed by a single active site per unit of time; Km: Michaelis constant. It is the substrate concentration at which the enzyme has half of its maximum velocity and it is a measure of affinity of the substrate to the enzyme (lower values of Km indicate higher affinity); kcat/Km: The ratio of the two kinetic constants is a measure of the catalytic efficiency of the enzyme. [c] Substrate used in the enzymatic activity assay.

2.1.2 Substrate form

An important aspect of the design of a bioprocess for the reduction of CO2 to formate is the substrate. There is contradictory evidence regarding whether CO2 or a hydrated species (H2CO3, HCO3 or CO32−) is a preferrable substrate for FDH. Although some studies report that HCO3 as a substrate results in better rates and higher formate production by FDH from Candida boidinii (CbFDH),80-82 several works claim the opposite,66 including the most recent electrochemical study by Meneghello et al.,83 according to which CO2 is the active species for FDH-catalyzed reduction. This view is shared by Miyaji and Amao,84 who observed that CbFDH only reduces CO2 to formate. The work of Zhang et al.85 has provided indications that both CO2 and HCO3 can be efficiently reduced by FDHs.

2.1.3 Enzyme cofactor

The cofactor that FDH physiologically uses is NADH. However, it is pricey and unstable. NADH stability, which is significantly affected by acidic pH, can be improved, for example with use of ionic liquids.86 However, for a process to be viable, NADH regeneration or the use of synthetic cofactors instead are required.52 NADH regeneration can be performed chemically, via homogeneous or heterogeneous catalysis,87, 88 photochemically,89, 90 enzymatically91 or electrochemically.92-94 Enzymatic regeneration is the most widely studied and preferrable to chemical, due to its high selectivity.88, 95 FDH is one of the enzymes that can regenerate NADH concomitantly with oxidizing formate to CO2.96 Other commonly used enzymes are glucose dehydrogenases, which can reduce NAD+ to NADH while oxidizing β-d-glucose to β-d-glucono-1,5-lactone.97, 98 This Review focuses on electrochemical regeneration, which can be direct or indirect (utilizing an electron mediator), since it is easy to control, through adjusting the applied potential, and could have a lower environmental impact if utilizing renewable electricity.99

A high concentration of NADH must be maintained in a biocatalytic system for the reaction equilibrium to be moved towards CO2 reduction. To achieve this through electrochemical regeneration, high overpotentials and therefore a high amount of energy are required.100 Regeneration might also not be efficient, due to the potential irreversible dimerization of NAD radicals towards the inactive NAD2, or the formation of the non-catalytically active isomer 1,6 NADH.93, 101, 102 It is reported that these unwanted side reactions can lead to a 40 % loss of cofactor in each regeneration cycle.103 The utilization of either redox mediators or artificial cofactors could help to overcome these problems. Synthesized electron-donating molecules with high substrate affinity are believed to have the potential to substitute NADH and improve the catalytic performance of FDH.104

Methyl viologen (1,1-dimethyl-4,4′- bipyridinium dichloride) radical cation (MV⋅+) has been proposed as a successful replacement of the natural cofactor NADH for CbFDH. Instead of being a hydride donor as NADH, MV⋅+ is an electron donor. Two MV⋅+ molecules are required to reduce one molecule of CO2 to formate, and they both appear to bind to the active site of FDH in molecular docking simulations. Its main advantages over NADH are that it can be directly regenerated at a carbon electrode, at a low reduction potential of −0.44 V (vs NHE) and without undesired byproducts. MV⋅+ also does not act as a cofactor for formate oxidation, thus not leading the reaction to that direction. Moreover, given the fact that the potential for the reduction of CO2 to formate is −0.42 V (vs NHE), the reaction utilizing MV⋅+ as a cofactor can be carried out at a remarkably low overpotential of only 20 mV.100 The first time that MV⋅+ as a cofactor for FDH was kinetically characterized, at pH 7.4 and with CbFDH being used, a Km of 213 μM and a kcat of 22.4 min−1 were calculated. These values are drastically improved compared to the respective ones of 2087 μM and 0.91 min−1 calculated for NADH under the same conditions.105 One concern regarding MV could be its toxicity for some organisms. However, that appears to depend on the respective redox state of the molecule and so far no reports of toxicity in its use as electron mediator for FDH-catalyzed CO2 reduction have been made.106

Sakai et al.107 also agree on MV being a thermodynamically favorable mediator, reporting that the MV2+/MV⋅+ couple has a midpoint potential very similar to the formal potential of the HCOO/CO2 couple (E°′CO2) around neutral pH, meaning that MV⋅+ can act as a mediator for the reaction at a low overpotential. Moreover, the standard reduction potential of the MV2+/MV⋅+ couple is more negative than that of the NAD+/NADH. couple (by 0.12 V), leading to a significantly higher equilibrium constant for the CO2 reduction reaction for MV⋅+ compared to NADH.100 Miyaji and Amao84 also report that utilizing 2,2’- or 4,4’-bipyridinium salts and their respective single electron reduced forms as redox couples favor the FDH-catalyzed reaction towards the direction of CO2 reduction to formate.

Another study tested the performance of 4 methyl viologen derivatives with amino and carboxyl groups as FDH cofactors (namely 1-methyl-1′-aminoethyl-4,4′-bipyridinium salt (MA2+), 1,1′-diaminoethyl-4,4′-bipyridinium salt (DA2+), 1-methyl-1′-carboxymethyl-4,4′-bipyridinium salt (MC2+) and 1,1′-dicarboxyl-4,4′-bipyridinium salt (DC2+)), where the amino derivatives exhibited an improved formate production compared to MV. Among them, DA2+ in its reduced form (DA+⋅) exhibited a catalytic efficiency 560 times higher than NADH in the reduction of CO2, with a reduction potential of −0.6 V vs Ag/AgCl and Km and kcat values of 17 μM and 4.3 min−1 respectively, as opposed to 212 μM and 1.9 min−1 when MV⋅+ was used.108, 109

Zhang et al.85 have reached similar conclusions, testing MV2+, DA2+ and DC2+ with CbFDH, each at their peak reduction potential. The amino MV derivative (DA2+) yielded better results than the carboxyl derivative, highest overall formate production and a catalytic efficiency more than 500-fold higher than NADH. At a reduction potential of −0.4 V vs SHE, the kinetic parameters for CO2 reduction calculated for DA+⋅were: Km=2.8 mM, kcat=0.45 min−1, as opposed to 50 mM and 0.013 min−1 for NADH.

An artificial cofactor, namely 1,1′-bis(2-(dimethylamino)ethyl)-4,4′-bipyridinium bromine (BPNC2), with remarkably improved formate production yield and high affinity to CO2 has also been reported and shown the best results so far. The cofactor exhibits a much higher equilibrium constant for the CO2 reduction reaction (3.6×108) compared to the natural cofactor (1.7×10−7), indicating that it thermodynamically favors production of formate. Moreover, it results in a higher product yield. The cofactor has been utilized in a BES that achieved formate production of 4.7 mM formate in 1 h, operating at the reduction potential of the cofactor (−0.77 V vs SHE), which is among the highest productivities reported to date in such a system.104

2.2 Parameters to improve the efficiency of the BES

2.2.1 Use of a mediator

The efficiency of NADH regeneration has been reported to improve with the use of mediators, which help overcome the problems of direct electrochemical NADH regeneration and lower the required potential. A molecule is likely to be a suitable mediator for NADH regeneration if it can transfer two electrons or one hydride in one step, being highly selective towards NADH, without transferring electrons to the enzymatic substrate and also has an activation potential higher than −0.9 V, so that no simultaneous direct NAD+ reduction occurs.93 A common category of mediators involves rhodium and iridium complexes. The first study to report indirect electrochemical regeneration of NADH used Bis(2,2’-bipyridine) rhodium (I) [Rh(bpy)2]+.110 Other rhodium complexes reported are [Cp(Me)5Rh(bipy)Cl]+,111 (pentamethylcyclo-pentadienyl-2,2′-bipyridine aqua) Rh,112 Cp[Rh(5,5′-methyl-2,2′-bipyridine)], Cp[Rh(4,4’-methoxy-2,2′-bipyridine)].113 [Cp*Rh (bpy)Cl]Cl in particular, when immobilized onto the BES cathode, has been reported in several studies as more capable of facilitating NADH regeneration by lowering the required potential compared to other mediators.66, 114, 115 Immobilization onto the electrode can improve the stability of the complex and has been reported to lead to 87 % faradaic efficiency (FE) for NADH regeneration.116 Moreover, in a BES using CbFDH, the presence of the complex as an electron mediator allowed regeneration of NADH at a potential of −0.8 V, as opposed to −1.2 V when no mediator was present.117 Kim et al.115 screened different electrode materials (Au, Ag, In and Cu) and evaluated Cu as the most suitable for NADH regeneration. The concentration of the Rh complex is also a factor to be taken into consideration, since the hydride transfer between the complex and NADH is reversible, meaning that high concentrations of RhIII could lead to NADH oxidation. The highest NADH yield (96.5 %) was achieved for a concentration of 0.25 mM [Cp*Rh (bpy)Cl]Cl, pH 6 and an applied potential of −1.0 V (vs Ag/AgCl). Contrary to what would theoretically be expected, the formate production rate was higher for neutral pH instead of pH of 6 (6.5×10−4 and 2.7×10−4 μmol min−1 mgFDH−1 at 30 min, respectively). It is interesting that the reduced form RhI exhibited CO2 reducing activity (yielding 0.43 mM of formate after 330 min when NADH was also present).

Other compounds that have been used to regenerate NADH for enzymatic reactions include methyl viologen, dithiothreitol and flavin adenine dinucleotide.93 Methyl viologen and other bipyridinium salts, that have been previously mentioned as alternative FDH cofactors, have been successfully utilized as mediators for NADH regeneration in many FDH-related studies. Alizarin red S, anthraquinone-2-sulfonic acid, benzyl viologen, and methyl viologen have been tested in the same study by Choi et al.118 as FDH electron mediators and the latter was evaluated as the best, based on the catalytic rate constant calculated for CO2 reduction.

Neutral red (NR) is another molecule often used as an electron transfer mediator. It has been reported by Singh et al.119 to improve formate formation in a BES with an activated carbon-based iron phthalocyanine carbamide composite cathode and FDH from Escherichia coli (EcFDH) as a biocatalyst. More specifically, from ≈23 mg L−1 cm−2 without NR, production increased to ≈30 mg L−1 cm−2 at an applied potential of −1.0 V (vs Ag/AgCl). In addition, NR has been investigated both in solubilized form and electropolymerized onto a carbon felt electrode, for formate production in a BES containing whole-cell Methylobacterium extorquens.120 The study concluded that NR improves formate production rates by facilitating electron transfer. Comparing the different systems on 57-day operation basis, formate production rate was 93.7 μM day−1 for solubilized NR, as opposed to 81.7 μΜ day−1 for poly(NR) and the FE was 4 % in both cases, at an applied potential of −0.75 V (vs Ag/AgCl). The non-mediated system yielded a 68.6 μΜ day−1 formate production rate and a 2 % FE. The enhancing effect of NR has also been described in several works, which focus, however, on the construction of NR-containing biocathodes onto which FDH is immobilized and will be analyzed further below.81, 121, 122 Moreover, NR is mentioned as being more biocompatible than methyl or benzyl viologen and therefore less likely to inhibit enzymatic activity.123

A simple batch BES containing all the so-far mentioned elements has been described by Srikanth et al.121 This system uses bubbled CO2 as substrate, CbFDH as biocatalyst, NADH as cofactor and NR as mediator. Cofactor regeneration occurs on a plain graphite rod electrode. The highest formate production rates were yielded at an applied potential of −1.0 V. More specifically, the figures of merit calculated include a formate titer of 414 μg L−1, yield of 9.4 mg L−1 CO2 and volumetric, specific, and molar productivities of 226 mg L−1 h−1, 0.197 mg mgenzyme−1 h−1 and 15.5 μM mgenzyme−1 min−1 respectively. This study showed that such a system can effectively produce formate from CO2, but product removal to prevent the reverse reaction and enzyme immobilization were highlighted as two parameters to be optimized. Another similar system, consisting of the same type of elements, reported formate production at a rate of 6.3×10−3 μmol mgenzyme−1 min−1. CbFDH was used as a biocatalyst and NADH was regenerated with 67 % efficiency, using rhodium complex [Cp*Rh(bpy)Cl]Cl on a Cu nanorod glassy carbon (CuGC) electrode at an applied potential of −1.0 V.115

2.2.2 Potential synergy with CA

CA is an enzyme that has drawn a lot of attention due to its very fast rate of catalyzing the CO2 hydration towards bicarbonate and has therefore been investigated for multiple applications in CO2 capture and utilization technologies.124, 125 In this context, it has been reported to improve the FDH-catalyzed formate production from CO2, leading to a more than 4-fold increase of the formate production rate (from 4.2×10−4 to 2.2×10−3 μmol min−1) in a non-electrochemical system utilizing CbFDH and NADH. The ratio of the two enzymes (in units) has been identified as a major rate-controlling parameter. Highest formate production rate was achieved for a CA/FDH unit ratio of 1061. It is important to note that high concentration of CA leads to progressive reduction of the pH. That is likely the reason for the observed reduction of the reaction rate when CA exceeded the concentration of 1.0 mg mL−1, which was determined as optimum.80

Srikanth et al.81 have also investigated the effect of addition of CA in a batch BES for CbFDH-catalyzed CO2 reduction. NADH regeneration was possible at a lower potential compared to other works, −0.75 V (vs Ag/AgCl), due to the coating of polymerized NR on the working electrode. In this study, too, CA appeared to positively affect FA production both in free and immobilized enzyme experiments, by increasing the product titer by 20 and 25 % respectively. The final product titer when the two enzymes were immobilized onto the working electrode was 623 mg L−1, which was marginally lower compared to that achieved with the free enzymes (647 mg L−1) but provided good stability and reusability of the enzymes (no significant drop in productivity was observed after three reuse cycles).

The enhancing effect of CA in CO2 assimilation and FA production in BES was also confirmed by Addo et al.,82 in a study where methanol was produced by an enzyme cascade. NADH was used as a cofactor and a poly(NR modified electrode used for its regeneration. Applied potential was set at −0.8 V (vs SCE). Addition of CA improved FA productivity, with the specific activity of FDH being increased from 0.052 to 0.074 μM min−1 mg−1. The influence of CA on the system is portrayed in the amperometric measurements as well, with an 82 % current increase being observed when CA was added, indicating higher reduction rates.

However, another work provides indications that the addition of CA to a CO2 reducing FDH system has an undesirable effect, resulting in reduced FA production attributed to the fact thatCO2 and not HCO3 is the active substrate for FDH.79 It is worth mentioning that the temperature at which the experiments were conducted in this preceding study is different than the ones used in the afore mentioned ones (10 °C as opposed to 37 °C used in the most recent works) and that FDH from Pseudomonas oxalaticus was used, while the rest of the studies used commercially available CbFDH. FDH-II from Eubacterium acidaminophilum has also shown to be capable of reducing CO2, and no enhancement of its activity was observed upon addition of CA.126

2.2.3 Enzyme immobilization and design of biocathodes

Several reversible and irreversible immobilization methods have been studied to improve reusability and stability of FDH, including use of glutaraldehyde, epoxy or amino-epoxy supports, metal–organic frameworks and cross-linking.127-129 However, many of them refer to non-bioelectrochemical systems, so the focus of this work will be on studies researching enzyme immobilization for the design of effective biocathodes for bioelectrochemical reduction of CO2 to FA using FDH. This approach can also target application of direct electron transfer (DET) systems, since immobilization of the enzyme on the electrode surface could provide a solution to the cofactor-related challenges such as regeneration efficiency, interaction with the enzyme or catalysis of the reverse reaction. DET systems could combine the benefits of smaller required amount of catalyst and easier product separation. In order for a DET system to be effective, sufficient contact between the enzyme and the electron surface has to be ensured to enable adequate electron transfer rates.130, 131 So far, DET systems have almost exclusively been reported for metal containing enzymes,132 with the exception of one, to our knowledge, utilizing CbFDH.18

Lee et al.133 propose the co-immobilization of CbFDH and cofactor in a polydopamine film matrix electropolymerized onto a glassy carbon electrode. The thin film created (below 10 nm) reduces the distance the electrons need to cover to reach FDH. Moreover, the pi-stacking of the dopamine units create a secure network that retains the enzyme, while the indole rings of dopamine inhibit dimerization of NAD+ radicals. These characteristics contribute to a very stable biocathode, that is able to produce formate and regenerate NADH at a quite low overpotential (−0.5 V vs. Ag/AgCl) and retain 60 % of its initial electrocatalytic activity after 16 days. When the described biocathode was paired with a photoanode, the occurring CO2 reduction system exhibited an exceptionally high faradaic efficiency of 99 %.

Barin et al.134, 135 have tested a bioelectrocatalytic process with immobilized CbFDH on electrospun polystyrene nanofibers and direct electrochemical regeneration of NADH on a Cu foam electrode at −1.1 V (vs Ag/AgCl), both on batch and semi-continuous modes. For the latter, removal of formate with ethyl acetate and recycling of the catholyte solution was initiated after 2 h. Polymeric nanofibers were chosen as an immobilization support due to high porosity and enzyme loading capacity, cost efficiency and ease in preparation, as opposed to alginate gel supports in which mass transfer can be limited, or widely investigated magnetic nanoparticles, which cannot be used in such a system due to the presence of electrical current. This immobilization strategy led to 43 % relative FDH activity, 57 % immobilization efficiency, 90 μgFDH cm−2 loading capacity and acceptable reusability and storage stability (53 % of initial enzyme activity was retained after 8 cycles, 41 % after 20 days of storage). It should be noted however, that the free enzyme system produced almost twice the amount of formate in 300 min (0.61 mM as opposed to 0.31 mM for the immobilized system), a difference attributed to the distance between the immobilized enzyme and the working electrode.134 Moreover, the semi-continuous mode exhibited a 42 % higher total formate production compared to the batch mode and reached a formate concentration of around 0.2 mM.135

Nanostructured hierarchical titanium nitride (TiN) has been selected as a support in a highly efficient BES using FDH from Thiobacillus sp. KNK65MA (TsFDH) and NR-mediated NADH regeneration. The advantages highlighted for this material as a support are its high surface area, pore size distribution and conductivity. The system exhibited a faradaic efficiency of 76 % and a maximum product yield of 44 μM mgFDH−1 h−1 at −0.15 V and −0.75 V (vs RHE) respectively, figures that are particularly encouraging, compared to other BES reported.136

Nafion is often used as binder to immobilize FDH onto the electrode to create a biocathode. Hernández-Ibáñez et al.117 propose an effective electrochemical FA-producing system where Nafion was used to co-immobilize CbFDH and a rhodium complex mediator on a nanoporous carbon electrode. This immobilization strategy appeared to be stable, exhibiting minimal leaching, and yielded higher formate production rates than the free enzyme at the same amount. 1.3×10−3 μmol mgCbFDH−1 min−1 of formic acid was produced when the enzyme was immobilized, as opposed to 3.0×10−5 μmol mgCbFDH−1 min−1 for the free enzyme. Biocathodes for the electrochemical CO2 reduction to FA were engineered and found to produce best results when 30 wt % Nafion was used (compared to the 50 : 50 carbon-to-Nafion ratio that was also investigated), achieving a coulombic efficiency of 46 %.

Zhang et al.122 describe the construction of another enzymatic cathode utilizing Nafion micelles modified with quaternary ammonium bromide salts (tetrabutylammonium bromide in particular) for the immobilization of CbFDH on a graphite disk electrode coated with polymerized NR. This immobilization method provides good stability of the enzyme, as it allows it to retain 88 % of its activity after 2 weeks. The system, operating at an overpotential of −0.8 V (vs. SHE), could reach a maximum formate production rate of 160 mg L−1 h−1 in the first 10 min, which then decreased, leading to a production of 130 mg L−1 formate in the first 2 h, with a FE of ∼80 %.

Taking these results one step further, the same group proposed a hybrid system consisting of an enzymatic biocathode combined with a microbial anode (composed of three MFCs connected in series) which collects electrons from degrading organic content of wastewater, resulting in an efficient system with low energy demands (requiring a very low overpotential of 0.10 V), which achieved a formate production rate of 60 mg L−1 h−1 with a 70 % FE. This rate was only slightly lower to that of a conventional BES operating at an applied potential of −1.0 V (65 mg L−1 h−1). NR electropolymerized onto a simple graphite rod electrode surface was used as an electron mediator for NADH regeneration and Nafion micelles were used to immobilize the enzyme onto the electrode surface.123

An indium-tin oxide electrode was used as the platform for immobilizing FDH from Saccharomyces cerevisiae and viologen with a long alkyl chain (CH3V(CH2)nCOOH), as an electron mediator. The system reached a formic acid production of 23 μM after 3 h, measured for anaerobic operating conditions, pH 7.4 and an applied potential of −0.55 V vs Ag/AgCl. Results showed that the length of the alkyl chain had an impact on the FA production, although seemingly not playing a role in the amount of FDH immobilized on the electrode nor the viologen reduction potential. The most efficient electron transfer between the enzyme and the alkyl chain and therefore the highest product concentration was yielded for n=9. Although it is mentioned that the electrode is developed with the intention of being paired with a photoanode for an artificial photosynthesis system, this work provides evidence of the efficiency of such a cathode for FA production from CO2 in a typical BES.137

A different arrangement for electrochemical CO2 reduction proposes the conjugation of the enzyme and its cofactor with a polyethylene glycol swing arm, so that the cofactor is close to the enzyme and allowed to move between the electrode surface and the enzyme active site to facilitate electron transfer. The electrode used in this system is based on Cu nanoparticle-coated carbon felt. The nanoparticles enable the immobilization of the enzyme, through the formation of Cu−S bonds. This system yields promising results compared to other BES, including highest formate concentration of 8.5 mM, reached after 4 h, FE of 23 % and molar productivity of 11.8 μM mU−1 h−1, achieved at an applied potential of −1.0 V (vs Ag/AgCl).138

An interesting type of electrode is one involving FDH from Clostridium ljungdahlii (ClFDH) cross linked with glutaraldehyde on a conductive polyaniline (PANi) hydrogel coated carbon cloth electrode. The PANi hydrogel consists of nanofibers that make the distance between the electrode and the enzyme short, therefore making effective and quick direct electron transfer possible. Compared to a normal carbon cloth electrode, the PANi electrode allowed for the immobilization of higher amount of FDH and more than 9-fold faster formate production. More specifically, the system reached a 93 % FE, with FDH converting CO2 to formate at a rate of 1.42 μmol h−1 cm−2 at −0.6 V vs Ag/AgCl and a 976 h−1 turnover frequency.131

Liu et al.139 have used polyethylenimine-functionalized SBA-15 as a support for the immobilization of CbFDH, in a system that exhibits good storage stability (63 % of activity retained after 20 d of storage, as opposed to 23 % for the free enzyme) as well as very promising reusability, with less than 20 % loss of activity after 12 reuse cycles. CO2 reduction was investigated in a BES operating with NADH and NR at −0.7 V (vs Ag/AgCl). The system produced 1.1 mM formate after 3 h, which corresponds to a productivity of 5.3 mM g−1 h−1 and is almost 3-fold higher than the formate production by the respective free enzyme system.

In search of alternative solutions for FDH immobilization on the electrodes, research has turned to metal–organic frameworks (MOFs). Chen et al.140 report the construction of an efficient biocathode utilizing a mesoporous MOF (NU-1006) that allows NADH regeneration and reaches a formic acid production rate of 79 mM h−1 at an applied potential of −1.1 V (vs Ag/AgCl). NU-1006 was selected because its pore size was comparable to the diameter of FDH, meaning that immobilization can be more efficient. The base of the biocathode is a fluorine-doped tin oxide (FTO) glass electrode coated with ZrO2. A rhodium complex (Cp*Rh(2,2′-bipyridyl-5,5′-dicarboxylic acid)Cl2) was deposited on the electrode to mediate hydride transfer for NADH regeneration. CbFDH encapsulated in NU-1006 was then casted onto the modified electrode to create an additional layer. This method appears promising, as NU-1006 contributes to FDH retaining its activity even in an acidic environment. Moreover, at the optimum pH 7, MOF-encapsulated FDH exhibited three times higher activity compared to the free FDH.

ZIF-8 is another MOF utilized for FDH immobilization, in a multi-enzymatic system that is not however bioelectrochemical. The arrangement proposed includes FDH co-immobilized with CA, NADH and glucose dehydrogenase, which serves enzymatic NADH regeneration, in ZIF-8. The formate yield was as high as 13.8 mM after eight 1 h reusing cycles, being 4.6-fold higher than the respective yield calculated for the free enzyme system.141

Another alternative is gas-diffusion type electrodes. The first gas-diffusion type electrode reported for bioelectrocatalytic CO2 reduction uses waterproof carbon cloth coated with Ketjen-black and polytetrafluoroethylene (the optimum content of which was determined to be 20 %), on which FDH from M. extorquens AM1 was then immobilized using glutaraldehyde. The enzyme (referred to as FoDH1) has been initially utilized by the same research group for the creation of a biocathode that was able to reduce CO2 with a quite high limiting current of density of 15 mA cm−2 at 60 °C and pH 6.6. The biocathode was based on a Ketjen-black glassy carbon electrode and glutaraldehyde was used for enzyme immobilization, while methyl viologen was used as an electron mediator.107 The gas-diffusion type electrode was the result of further investigation on the potential of FoDH1. This system, employing 1,1′-trimethylene-2,2′-bipyridinium dibromide as an electron mediator, achieved high current density (20 mA cm−2) at neutral pH, atmospheric pressure and room temperature, which translates to formate production rate of 0.1 μmol cmelectrode−2 s−1.142 Hydrophobicity is a parameter that the authors highlight as important for the enzyme-mediator interactions and should therefore be taken into consideration. It is also pointed out that FoDH1 can be adsorbed on mesoporous carbon electrodes and directly transfer the required electrons to catalyze CO2 reduction from the electrode surface utilizing its FeS clusters, while its flavin mononucleotide cofactor can act as a mediator for a mediated electron transfer system where the electrodes are planar and FMN is dissociated in the solution.143

Another such electrode, operating without the need for enzyme cofactors or redox mediators, employs FDH from Desulfovibrio vulgaris Hildenborough and a negatively charged polymer backbone with a reduction potential of −0.39 V vs SHE. This novel cathode consists of four layers, an outer carbon cloth layer, a polymer layer on the inner side of the carbon cloth, an intermediate viologen-modified polymer layer used for electrical wiring and an inner active viologen and enzyme-containing polymer layer. Current density expressed as a function of the enzyme load has been used as a measure of evaluating the system, and it was calculated to be 3.6 μA μgFDH−1, as opposed to 2.2 μA μgFDH−1 for direct electron transfer systems where no polymer was used.144

The same enzyme has been covalently immobilized on low-density graphite electrodes on which aminophenyl groups have been introduced. The occurring biocathode could produce formate from NaHCO3 at pH 6 and −0.66 V vs SHE, both with MV as mediator and through DET, with the mediated system exhibiting almost 3-fold higher catalytic rate (8.6 s−1 as opposed to 3.5). The operational stability of the electrode, which initially did not exceed 1.5 h, was improved by 2 h with crosslinking with 0.9 % glutaraldehyde for 30 min.145 Another study reports that immobilization of the enzyme on mesoporous TiO2 leads to successful formate production from CO2 at −0.6 V vs SHE in the absence of electron mediators. However, no specific formate productivity data are given, since the work focuses on the development of a photocatalytic system, by FDH immobilization on TiO2 nanoparticles sensitized with ruthenium tris-2,2′-bipyridine dye. That system efficiently reduced CO2 with a TOF of 4×104 h−1.146

FDH from Desulfovibrio desulfuricans has been immobilized onto a pyrolytic graphite electrode using a cellulose membrane and proved to have electrocatalytic activity in the absence of any cofactors or mediators, confirming that the reducing equivalents could be directly provided from the electrode. The study, however, focuses on the electrochemical behavior of the enzyme and does not provide formate productivity information.147

Another DET system employing FDH as part of the heterodisulfide reductase supercomplex (Hdr-SC) of Methanococcus maripaludis has been reported by Lienemann et al.148 The Hdr-SC complex contains a heterodisulfide reductase, a FDH and a NiFe-hydrogenase. The enzyme was adsorbed to the electrode to create a biocathode which allowed formate production with a FE of 90 % over 5days, when an overpotential of −0.8 V vs Ag/AgCl was applied. 13 mM formate accumulated after 6 days of operation, while the initial formate production rate was 3 μmol h−1 cmelectrode_surface−2.

2.2.4 Protein engineering

Methylobacterium extorquens AM1 mutants expressing recombinant FDH have been constructed using gene knockout. Mutant F1A−P1 exhibited 2.5-fold higher productivity compared to the wildtype (2.53 mM g−1wet_cell h−1 in a whole cell BES) under an applied potential of −0.75 V (vs Ag/AgCl) and in the presence of tungstate, which was found to enhance product formation.149

Another strategy towards FDH variants with higher CO2 reducing ability is directed evolution. Çakar et al. applied a directed evolution strategy to produce a variant of FDH from Chaetomium thermophilum (G93H/I94Y), with 5.4 times higher catalytic efficiency for HCO3 reduction compared to the wildtype and no formate oxidation activity.150

Although not incorporated in BES, more engineered FDHs with improved catalytic efficiency, thermal stability and substrate specificity have been reported. For example, change of the Val313 residue of FDH from Ceriporiopsis subvermispora to Pro results in a remarkable increase of the CO2 reduction rate, with the enzyme reaching a 0.39 s−1 mM−1 catalytic efficiency at an optimum pH of 6.5, as opposed to 0.001 s−1 mM−1 for the wild type. These results were the best among different mutations tested for residues in the vicinity of the active site for CtFDH and CsFDH. The enhancement of catalytic activity is attributed to the hydrophobicity of Pro, which appears to play a role in the substrate orientation in the active site, even if not directly involved in substrate binding.151 Another study investigates potential mutations for CtFDH and CmFDH, pointing out the potential effect of the mutation of Asn120 residue of CtFDH to Cys, which decreases substrate affinity to the enzyme (therefore making its release easier) while increasing the catalytic rate, resulting in an overall similar catalytic efficiency compared to the wild type.152 Site-directed mutagenesis has been performed on CtFDH to replace Asp188 with Arg, a residue located in the interface between the two enzyme subunits. The potential effect of the mutation on the protein structure was initially investigated with molecular dynamics. The obtained mutant showed an almost 3-fold higher kcat (0.42 s−1 as opposed to 0.16 s−1 for the wildtype) and lower Km (0.91 mM as opposed to 2.71 mM) for bicarbonate reduction in the presence of NADH at optimum pH 8.153 Moreover, mutants of EcFDH with improved stability have been investigated. In particular, a rationally designed mutant with modified surface (occurring by replacing hydrophobic surface residues Tyr3, Phe26, Ile140 and Glu190 with hydrophilic Glu, Ser, Arg and Lys respectively) and additional salt bridges (created by substitution of Ser5 and Ile163 with Lys, Val168 and Gln70 with Asp, Leu150 with Arg and Leu155 with Glu), showed improved formate production rates, although no kinetic constants were calculated and hydrogenase and H2 were also present in the reaction.154

As seen from the above, there are several studies that have proved the possibility to use FDH in a BES to reduce CO2 to formate. The base for the development of a such a BES is the selection of a competent FDH, use of appropriate substrate form and an efficient cofactor. Strategies to optimize the system include use of an electron mediator, enzyme immobilization, addition of CA to increase CO2 solubility or protein engineering to improve FDH properties. A collective presentation of the BESs for CO2 reduction to formate reported to date can be found in Table 2.

Table 2. Reported bioelectrochemical systems for CO2 reduction to formate using FDH.

Microorganism

pH

Aerobic/

anaerobic conditions

Substrate

Applied potential [V]

Electrode type

Working volume [mL]

Enzyme immobilization

Cofactor

Mediator

Formate production

Formate productivity

FE11 [%]

Ref.

Candida boidinii

6

NS

CO2 gas, bubbling

−0.7 vs. Ag/AgCl

GCE[c]

10

immobilized on a polyethylenimine support modified on mesoporous silica

NADH

NR[j]

1.12 mM in 3 h

5.3 mmol gFDH−1 h−1

[139]

Candida boidinii

7.4

Anaerobic

CO2 gas, bubbling

−0.75 vs. Ag/AgCl

Cold-rolled electrode based on graphite powder

10

immobilized onto the electrode along with carbonic anhydrase

NADH

NR

6.99 mg mg FDH−1 h−1

[81]

Candida boidinii

7.4

NS

CO2 gas, bubbling

−0.8 vs. SCE

GCE with NR electropolymerized on it

NS

co-immobilized onto the electrode with CA

NADH

NR

[82]

Candida boidinii

7

Anaerobic

CO2 gas, bubbling

−0.5 vs. Ag/AgCl

GCE

NS

co-immobilized with NADH onto a thin film of polydopamine electropolymerized onto the electrode

NADH

0.074 μΜ mgFDH−1 min−1

99

[133]

Candida boidiniii

6.2

Anaerobic

CO2 gas, bubbling

−1.0 vs. Ag/AgCT.l

Cu foil

203

free

NADH

[Cp*Rh(bpy)Cl]Cl

≈16 μΜ in 25 h

3.2×10−4 μmol mgFDH−1 min−1

[66]

Candida boidiniii

6.6

NS

NaHCO3

−0.44 vs NHE

porous carbon fiber paper electrode

7.5[f]

free

MV∫+[g]

44 mM in 30 h

96

[100]

Candida boidiniii

7

NS

CO2 gas, pre-bubbled

−0.39 vs. Ag/AgCl

CC[d]

10

free

DA2+[h]

3.5 mM h−1

[85]

Candida boidiniii

7

NS

CO2 gas, pre-bubbled

−0.773 vs. SHE

CC

10

free

BPNC2[i]

4.7 mM in 1 h

[104]

Candida boidinii

7.4

Anaerobic

CO2 gas, bubbling

−0.8 vs. Ag/AgCl

GCE

65

co-immobilized with mediator on the electrode

NADH

[Cp*Rh (bpy)Cl]Cl

-

1.33×10−3 μmol mgFDH−1 min−1

46

[117]

Candida boidinii

7.4

NS

CO2 gas, bubbling

−1.0 vs. Ag/AgCl

Graphite rod

10

free

NADH

NR

413.97 mg L−1 in 3 h

15.5 μM mgFDH−1 min−1

0.2

[121]

Candida boidinii

7

NS

CO2 gas, bubbling

−1.1 vs. Ag/AgCl

Cu foam electrode

15

immobilized on electrode with modified electrospun polystyrene nanofibers

NADH

-

0.31 mM in 300 min

[134]

Candida boidinii

6

anaerobic

CO2 gas, pre-bubbled

−1.0 vs. Ag/AgCl

GCE with Cu electrochemically deposited

203

free

NADH

Cp*Rh(bpy)Cl]Cl

1.2×10−2μmol mgFDH−1 min−1

[115]

Candida boidinii

6

NS

NaHCO3

−0.8 vs. SHE

rotating graphite disk with a layer of electropolymerized NR

203

immobilized in modified Nafion micelles, onto the electrode

NADH

NR

160 mg L−1 h−1

77

[122]

Candida boidinii

7

NS

CO2 gas, pre-bubbled

−1.1 vs. Ag/AgCl

fluorine-doped tin oxide glass electrode

NS

immobilized in NU-1006

NADH

Cp*Rh(2,2′-bipyridyl-5,5′-dicarboxylic acid)Cl2

79 mM h−1

[140]

Methylobacterium extorquens AM1 wholecell

6

microaerobic

CO2 gas, bubbling

−0.75 vs. Ag/AgCl

Cu plate

10

free

MV⋅+

60 mM in 80 h

[68]

Methylobacterium extorquens AM1

6

anaerobic

CO2 gas, pre-bubbled

−0.75 vs. Ag/AgCl

Cu plate

10

free

MV⋅+

2.53 mM gwet_cell−1 h−1

[149]

Methylobacterium extorquens AM1

6.5

anaerobic

CO2 gas, bubbling

−0.8 vs. Ag/AgCl

gas-diffusion type electrode (ketjen black mixed with PTFE)[e]

NS

immobilized on electrode using glutaraldehyde

TQ[k]

0.1 μmol cm2 s−1

-

[142]

Methylobacterium extorquens AM1

6.6

anaerobic

HCO3

−0.8 vs. Ag/AgCl

GCE functionalized with 9,10-phenanthrenequinone[e]

NS

co-immobilized on the electrode with a viologen-functionalized polymer

MV2+

[107]

Methylobacterium extorquens AM1

7

NS

CO2 gas, pre-bubbled

−0.75 vs. Ag/AgCl

carbon felt electrode

NS

biofilm developed onto the electrode

NR

93.7 μM day−1

4

[120]

Escherichia coli

7.5

anaerobic

Na2CO3

150 mV overpotential

graphite-epoxy[e]

NS

absorbed onto the electrode surface

[78]

Escherichia coli, wholecell

7

anaerobic

NaHCO3

−1.0 vs. Ag/AgCl

activated carbon fiber electrode modified with silicon carbide-derived carbon and iron phthalocyanine

80

immobilized on the electrode

NR

30 mg L−1 h−1

58

[119]

Escherichia coli

6

anaerobic

NaHCO3

−0.66 vs. SHE

GCE

NS

immobilized on the electrode with the cobaltocene-poly(allylamine)

431 nmol in 90 min

99

[130]

Thiobacillus sp. KNK65MA

6

anaerobic

NaHCO3

−1.0 vs. Ag/AgCl

carbon fiber electrode with electrodeposited CU nanoparticles[e]

53

co-immobilized with cofactor in a polyethylene glycol swing arm

NADH

8.5 mM in 4 h

11.8 μM mU−1 h−1

23

[138]

Thiobacillus sp. KNK65MA

6.1

aerobic

CO2 gas, bubbling

−0.75 V vs. RHE

hierarchical titanium nitride nanostructured electrode (HTNE)[e]

40

immobilized on the TiN support of the electrode

NADH

NR

44 μmol mgFDH−1 h−1

76

[136]

Rhodobacter capsulatus

7

aerobic

CO2 gas, bubbling

−0.7 vs. Ag/AgCl

GCE

10

free

MV⋅+

6 mM in 5 h

[118]

Pseudomonas oxalaticus

7

anaerobic

NaHCO3

−0.7 vs. SCE

GCE

5

free

MV⋅+

7.3 μmol in 20 h

91

[155]

Desulfovibrio vulgaris Hildenborough

6

anaerobic

CO2 gas, bubbling

−0.590 vs. SHE

CC-based gas-diffusion type electrode[e]

30

immobilized onto the electrode with poly(4-styrene sulfonate-co-glycidyl methacrylate-co-butyl acrylate)

amino-modified viologen

1.9–2.6 mg L−1 in 48 h

[144]

Desulfovibrio vulgaris Hildenborough

6

anaerobic

NaHCO3

−0.66 vs. SHE

modified low-density graphite electrode

33

immobilized onto the electrode

3.3 μM in 1.5 h

100

[145]

Desulfovibrio vulgaris Hildenborough

6.5

anaerobic

CO2/NaHCO3

−0.6 V vs. SHE

mesoporous TiO2 electrode

NS

immobilized onto the electrode

92

[146]

Syntrophobacter fumaroxidans

5.9

anaerobic

Na2CO3

−0.8 vs. Ag/AgCl

pyrolytic graphite edge electrode[e]

NS

adsorbed to the electrode

MV⋅+

97

[48]

Clostridium ljungdahlii

6.5

NS

NaHCO3 in CO2 saturated solution

−0.6 vs. Ag/AgCl

CC electrode modified with conductive polyaniline (PANi) hydrogel[e]

NS

immobilized onto the electrode

1.42 μmol h−1 cm−2

93

[131]

Saccharomyces cerevisiae

7.4

anaerobic

CO2 gas, pre-bubbled

−0.55 vs. Ag/AgCl

indium-tin oxide glass electrode[e]

NS

immobilized on the electrode with long alkyl chain viologen CH3V(CH2)nCOOH, n=9

23 μmol in 3 h

[137]

Shewanella oneidensis MR−1, wholecell

7

anaerobic

CO2 gas, bubbling

−0.75 vs. Ag/AgCl

Cu plate

10

free

MV⋅+

137 mM in 72 h

3.8 mM h−1 gwet_cell−1

65

[77]

Methanococcus maripaludis strain MM1264[a]

NS[b]

anaerobic

CO2 gas

−0.8 vs. Ag/AgCl

graphite rod

1253

free

13 mM in 6 d

3 μmol h−1 cm−2

90

[148]

  • [a] Formate dehydrogenase as part of the heterodisulfide reductase supercomplex (Hdr-SC). [b] Not specified. [c] Glassy carbon electrode. [d] Carbon cloth electrode. [e] The electrode is not part of an H-cell system. [f] The volume corresponds to the cathodic chamber volume, it is not specified whether that equals the working volume. [g] Methyl viologen. [h] 1,1′-diaminoethyl-4,4′-bipyridinium salt. [i] 1,1′-bis(2-(dimethylamino)ethyl)-4,4′-bipyridinium bromine. [j] Neutral red, 10: 1,1′-trimethylene-2,2′-bipyridinium dibromide. [k] Faradaic efficiency

3 Hydrogen Release from Formate

3.1 Formate decomposition catalyzed by formate hydrogen lyase (FHL)

3.1.1 The FHL complex

Hydrogen production from formate can be performed by the formate hydrogenlyase (FHL) complex, a bienzymatic complex found in facultative anaerobes.23 The main enzymes forming an FHL complex are a hydrogenase and a FDH. FHLs are not well characterized enzymes and a lot of information about them has been derived by studying the two putative E. coli complexes FHL-1 and FHL-2. The physiological role of the FHL-1 complex is to regulate intracellular pH through formate metabolism. The role of FHL-2 is uncertain, but is presumably related to fermentative CO2 production in alkaline conditions.156 Both complexes comprise of a catalytic, soluble part and a membrane bound part. In FHL-1, the former is made of 5 subunits and the latter of 2. It is believed that the catalytic soluble part of FHL-2 is similar to that of FHL-1, but the membrane bound part is larger, involving 5 subunits at a minimum.157 Moreover, the genes encoding the hydrogenase of FHL-2 are organized in a different operon, while FDH is the same in both complexes. It has been observed that FHL-1 is responsible for H2 production at slightly acidic pH, while FHL-2 at slightly alkaline pH. The expression of the respective genes and formation of either of the two forms of E. coli FHL depends on pH and formate concentration.156 There is also evidence that both complexes require F0F1-ATPase to function.158

The isolation and analysis of the FHL-1 complex from E. coli showed its main constituents being a two-subunit [NiFe] hydrogenase (Hyd-3, encoded by the gene HycE), a molybdenum-dependent FDH−H (encoded by the gene FdhF) and three iron-sulfur proteins (corresponding to the genes HycB, HycF, and HycG). The hydrogenase and the FDH are the catalytic sites of the complex, while the FeS clusters create a pathway for the necessary electron transfer (Figure 5).159 Moreover, two membrane proteins, HycC and HycD could be associated with the binding of the complex to the cytoplasmic side of the cell membrane.160-162 The aforementioned proteins are encoded by the operon hycABCDEFGHI, which additionally encodes transcriptional repressor hycA and proteins hycH and hycI, involved in the maturation of Hyd-3.163 The only exception is FdhF, which is a separate gene. The function of FDH−H is to oxidize formate towards CO2 and protons, which are then reduced by Hyd-3 to H2. The conversion is described by the two following reactions:
urn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0005(4)
urn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0006(5)
Details are in the caption following the image

Structure of FHL-1 complex from E. coli (Reproduced under terms of the CC-BY license.165 Copyright authors 2016). Schematic representation of the subunit arrangement (right) and crystal structure of the complex (PDB ID: 7Z0T). The color of each subunit is the same in the two versions.

Equation (5) was balanced, Genes fdhD, fdhE. fdnH, fdoG, foCb and fhlA encoding a protease (Fdh-D), a membrane-binding protein (Fdh-E), two FDHs (nitrate inducible Fdh-N and aerobic Fdh-O), a formate permease (FoCb) and an FHL transcriptional activator (FhlA) respectively are also reported to play a role in the activity of the FHL complex.23 The expression of all these proteins is necessary for a functional FHL complex, as they regulate changes in oxygen, nitrate and pH.164 Expression is induced upon accumulation of formate and subsequent reduction of extracellular pH.165 Moreover, Steinhilper et al.159 report that the expressed FHL complex is different under aerobic and anaerobic conditions, with the former lacking hycC and hycD subunits and the latter lacking fdhF. Moreover, they point out that aerobic expression has detrimental effect to the function of Hyd-3 and the nearby FeS cluster. Fur protein, a ferric ion uptake regulator, is another protein found to be crucial for H2 production in E. coli, since its absence led to 80–90 % lower Hyd-3 activity, as tested in a respective mutant.166

FHL-1 is the most thoroughly studied FHL complex while it has higher H2 production and lower product inhibition rates compared to other [NiFe] hydrogenases, which is why it is preferred for biotechnological applications.161

Bacterial and archaeal FHLs are slightly different.FHL complexes of methanogenic archaea, such as Methanococcus maripaludis or Methanobacterium formicicum, consist of a formate dehydrogenase and a F420-reducing hydrogenase.167-169 Another distinct example is FHL from Thermococcus onnurineus, which consists of a group 4B hydrogenase and a FDH (as opposed to a group 4 A hydrogenase in FHL complexes from E. coli). This hydrogenase is structurally closer to the Hyd-4 hydrogenase of FHL-2 complex of E. coli.157

The dehydrogenation of formate, although often a part of anaerobic metabolism, is considered unable to provide the energy required for microorganism growth. However, Kim et al.170 provide evidence that addition of formate in Thermococcus onnurineus NA1 cultures led to the production of ATP which could provide the energy necessary for cell growth. Further investigation of Thermococcus sp. provided four additional strains that are capable of formate-based growth and H2 production, namely T. gammatolerans, T. barophilus Ch5, Thermococcus sp. DS-1 and Thermococcus sp. DT-4. Formate concentration and pH are two parameters that play a role in the ATP synthesis, which has been reported to be coupled to reactions with a ΔG of −5 kJ mol−1 or lower.171

In addition, there have been cases of syntrophic microorganism growth on formate, involving communities of Moorella sp. strain AMP and a hydrogen-consuming Methanothermobacter species, and of Desulfovibrio sp. Strain G11 and Methanobrevibacter arboriphilus strain AZ1.172

3.1.2 H2 production with FHL in whole cell anaerobic biocatalytic systems

Most of the reported biocatalytic systems for H2 production from formate employ whole pre-grown E. coli cells. E. coli DJT135 has been evaluated as the most productive whole cell catalyst for H2 production from formate, when compared to Citrobacter amalonaticus Y19, Escherichia coli K-12 MG1655 and Enterobacter aerogenes, although the latter exhibited the highest H2 production rate in a growth-coupled system with glucose as substrate. Formate functioned more effectively as a substrate (at a concentration of 20 mM), leading to a maximum production rate of 195 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0007  mgcells−1 h−1. E. aerogenes, C. amalonaticus Y19, and E. coli K-12 were also able to produce H2 under the same conditions, with lower rates of approximately 180, 160 and 100 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0008  mgcells−1 h−1 respectively.173 Another study utilized C. amalonaticus Y19 whole cells used to produce H2 from a lower formate concentration as substrate (2 mM) and achieved a H2 productivity of 0.43 μmol mgDCW−1 min−1. This work also provided evidence that formate was only decomposed to form H2 and CO2 and was not utilized as a carbon source for cell growth.174

Immobilization of E. coli S13 whole cells in agar led to a production of 4 mL gwet_cell−1 h−1 H2 from formate, as opposed to 7 mL gwet_cell−1 h−1 for the free washed cells. It was also observed that supplementation of the reaction mixture with 28 mM glucose benefited production. Although the hydrogen production rate was lower for the immobilized system, it exhibited higher stability and considerable recycling efficiency, retaining 50 % of its activity after four 24 h cycles.175

Metabolic engineering of E. coli to improve hydrogen production has been the focus of several works, which mostly experiment with deletion of genes which encode proteins that either consume the produced hydrogen or lead to the degradation of formate through other, non-H2 producing pathways. Mutants in which uptake hydrogenase genes are deleted are common. One such strain is E. coli SH1, for which a H2 production rate of 0.97 μmol mgDCW−1 min−1 was determined in an assay with 20 mM formate as substrate.176 Pre-grown cells of E. coli RT1 strain, lacking every hydrogenase and FDH apart those of the FHL complex, catalyzed formate-based H2 production at 118 % of the rate of parental strain CP734, for which a production rate of 37.8 nmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0009  min−1 mg−1 was calculated, with 15 mM formate as substrate.165 Yoshida et al.160 tested E. coli recombinant strain SR13 (in which the FHL repressor and activator protein genes are deleted and enhanced respectively), which showed 2.8-fold higher specific H2 production rate compared to wildtype strain W3110. Using the wildtype strain (W3110), which has also shown H2 production potential from formate-producing yeast culture spent medium,177 they optimized reaction conditions towards H2 formation at 42 °C and pH 6.5 and studied both sodium formate and free FA as reaction substrates. Formate resulted in higher initial production rates at a concentration of 100 mM (99 mmol gDCW−1 h−1), which then decreased, however, making it unfavorable for continuous hydrogen production. On the contrary FA could sustain continuous hydrogen production at a concentration below 25 mM, exhibiting a peak H2 production rate of 82 mmol gDCW−1 h−1. Strain SR13 could efficiently be utilized in a 100 mL bioreactor for continuous H2 production with 25 mM free FA as substrate, reaching a volumetric H2 production rate of 23.6 gurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0010  L−1 h−1 at a cell density of 93 gDCW L−1. The same group further investigated H2 production with E. coli SR13 and reached the conclusion that accumulation of metabolites such as lactate, succinate or acetate can inhibit product formation. In this context, they developed E. coli strain SR14, where the anaerobic pathways that lead to the formation of lactate and succinate are disrupted. Moreover, to tackle the issue of low cell growth rates under anaerobic conditions, they propose the initial aerobic cultivation of cells, and then incubation in an anaerobic environment to induce FHL, which is otherwise not produced. This two-step method, combined with a metabolite excretion system operating at a dilution rate of 2.0 h−1, led to a maximum hydrogen production rate of 144.2 mmol h−1 g−1.178

Maeda et al.179 report an E. coli strain BW25113 mutant, devoid of two uptake hydrogenase, one FDH and FHL repressor protein genes. The engineered strain produced hydrogen from formate at a rate 141 times higher than the wild type (113 μmol mgprotein−1) in a partial pressure batch reactor that allowed for the removal of the produced hydrogen. Similarly, mutant E. coli strain SH5 originated from the suppression of genes encoding two uptake hydrogenases, lactate and fumarate dehydrogenases and FHL repressor. The occurring strain produced hydrogen at a rate of 0.87 μmol min−1 mgDCW−1.176 Strain SH5 was further utilized in an immobilized cell system. Agar was evaluated as better immobilization support than agarose and sodium alginate, due to its mechanical stability and gas product permeability. Temperature, pH, cell load for immobilization, substrate and agar concentration were optimized, allowing the system to reach a maximum hydrogen production rate of 2400 mL L−1 h−1 at 37 °C, pH 6.5 and 350 mM formate concentration. The system managed to sustain stable hydrogen production for 10 h by replenishing the substrate and cells exhibited better reusability when washed before substrate re-addition. Supplementation with nutrients such as glucose was an alternative strategy to retain product yield while avoiding impractical cell washing.180 A recombinant strain of E. coli SH5 overexpressing FHL activator protein and having the iscR gene deleted, which negatively regulates the iron-sulfur clusters, showed improved hydrogen production of 2.8 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0011  mgdcw−1 min−1, as opposed to 1.4 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0012  mgdcw−1 min−1 by the parental strain SH5.181

Crude extract of E. coli cells (strain MC4100) has also shown evidence of H2 production, reaching 1.5 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0013  mL−1 in 18 min, when incubated in small tubes with 80 mM formate as substrate.

Pre-grown cells of Enterobacter asburiae SNU-1 were able to produce hydrogen with formate as substrate optimally at pH 6.1, 37 °C and 350 mM formic acid concentration in a batch operation, with a hydrogen production rate of 491 mL L−1 h−1 and total hydrogen production of 6668 mL L−1. In fed-batch mode, with stable pH and FA concentration (300 mM), the total hydrogen production increased to 9468 mL L−1, although the production rate was lower.182

The effect of partial H2 pressure on H2 production was also pointed out by Lupa et al., who performed experiments using methanogen Methanococcus maripaludis. When larger headspace was used for the culture, resulting in lower of partial H2 pressure, H2 production was significantly improved (reduction of pressure from 220 to 80 kPa led to increase of H2 producing activity from 0.12 to 0.41 U mgDCW−1). Formate was used as a substrate for both cell growth and H2 production, in separate stages.183

Clostridium paraputrificum is another candidate for whole cell formate-based biocatalytic H2 production. Pre-grown cells in a glucose-containing medium were incubated for 6 h with 50 mM formate as substrate and reached a maximum hydrogen production of 40 mmol gDCW−1. When the bacterium was metabolically engineered to overexpress FDH gene fdh1 from C. boidinii, the amount of H2 produced increased by 3.2 %.184

Apart from utilizing the FHL complex of a single microorganism, hydrogen production can also be achieved in a bienzymatic system, combining a FDH from one microorganism and a hydrogenase from another, as reported by Klibanov et al.185 More specifically, FDH from P. oxalaticus and hydrogenase from C. necator, with MV2+ as a cofactor, were able to produce hydrogen from formate. Moreover, whole cells of C. necator strain 337 cells also stably produced H2 from 0.5 M formate. The catalytic activity of the cells was further improved by immobilization in kappa-carrageenan gel. Increased cell density led to increased H2 production, while increased substrate concentration appeared to inhibit FDH and therefore overall H2 production. Maximum H2 productivity was around 4.5 mLurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0014  gcells−1 h−1, achieved with 2 g of gel containing 11 % w/w cells and 0.5 M formate.

A summary of the whole cell biocatalytic systems producing H2 from formate is presented in Table 3.

Table 3. Properties of reported systems utilizing pre-grown cells of FHL-producing microorganisms as whole cell catalysts for H2 production. All the systems operate under anaerobic conditions.

Microorganism

System

Conditions

Formate concentration [mM]

H2 productivity

Cell concentration

Reference

Escherichia coli RT1

reaction in small tubes (1.7 mL w.v[a])

pH 7

15

49.9 nmol mgprotein−1 min−1

8.3 gcell_protein L−1

[165]

Escherichia coli DJT135

batch fermentation in 165 mL serum bottles (50 mL w.v.)

pH 6.0, 45 °C

20

195.2 μmol mgcell−1 h−1

0.3 g L−1

[173]

Escherichia coli SH1

reaction in small cuvettes

pH 6.8, 37 °C

20

0.97 μmol mgDCW−1 min−1

[176]

Escherichia coli S13

cells immobilized in agar gel, 80 mL reaction mixture

pH 7.2, 37 °C

1180

4 mL gwet_cell−1 h−1

2 % w/w in 2–2.5 % agar gel (740 mgwet_cell)

[175]

Escherichia coli SR13

100 mL bioreactor for continuous H2 production

pH 6.5, 37 °C

25

300 L L−1 h−1

(23.6 g L−1 h−1)

93 gDCW L−1

[160]

Escherichia coli SR14

bioreactor with metabolite excretion system

pH 6.3, 37 °C

25

144.2 mmol g−1 h−1

0.41 gDCW L−1

[178]

Escherichia coli SH5

batch system with periodic supplementation of formate and glucose (165 serum vial with 30 mL w.v.)—cells immobilized in agar

pH 6.5, 37 °C

350

2.4 L L−1 h−1

100 OD600 in 3.5 % agar gel (30 gDCW L−1)

[180]

Escherichia coli SH5 ΔiscR/pBKLA

10 mL serum bottle (2 mL w.v.)

pH 6.5, 37 °C

50

2.8 μmol mgDCW−1 min−1

OD600 0.5–0.7

[181]

Escherichia coli MC4100

reaction in small 5 mL tubes (0.75 mL w.v.)

pH 7, 37 °C

80

1.5 μmol mL−1 in 18 min

[166]

Escherichia coli BW25113 hyaB hybC hycA fdoG/pCA24 N-FhlA

Batch reactor (60 mL vial with 30 mL w.v.) maintaining low hydrogen headspace pressure

37 °C

100

113 μmol mgprotein−1 h−1

[179]

Enterobacter asburiae SNU-1

batch operation (1 L fermentor with 500 mL w.v.)

pH 6.1, 37 °C

350

0.491 L L−1 h−1

1 g L−1

[182]

Citrobacter amalonaticus Y19

batch reaction in 12 mL serum bottles or 4 mL cuvettes

pH 7.3, 30 °C

2

0.43 μmol mgDCW−1 min−1

[174]

Methanococcus maripaludis S2

reaction in 2.8 mL cuvette

pH 6.9, 37 °C

20

0.6 mM in 6 min

0.1 gDCW L−1

[183]

Clostridium paraputrificum

batch reaction in 70 mL serum bottles (20 mL w.v.)

pH 6.8, 37 °C

50

40 mmol gDCW−1 in 6 h

OD600 1.0

[184]

Cupriavidus necator

immobilized cells in 3.3 % kappa-carrageenan gel

pH 7.2, 25 °C

500

4 mL gcell −1 h−1

2 g gel with 11 % cells (w/w, wet cell weight)

[185]

  • [a] Working volume.

3.1.3 H2 production with FHL coupled with bacterial growth

Apart from whole cell biocatalysis with harvested cells, many studies have also studied H2 production by simultaneous cell growth and catalysis with formate as substrate (Table 4). Bakonyi et al.186 have optimized conditions of a stirred anaerobic bioreactor system producing H2 from growing E. coli strain XL1-BLUE and found that formate concentration was the key parameter and was optimum at 30 mM, where the highest yield of 0.41 molurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0015  molformate was observed. The system achieved a hydrogen productivity of 426 mLurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0016  L−1 d−1. However, increasing formate concentration (up to 75 mM), led to reduction of the yield, but a higher productivity of 725 mL L−1 d−1. E. coli MG1655 cells are also capable of both fermentative H2 production from glucose, in a pathway were formate is synthesized as an intermediate, and conversion of exogenously added formate to hydrogen in microaerobic conditions (10–20 % oxygen).187

Table 4. Properties of reported systems coupling H2 production with bacterial growth in formate-supplemented media. All the systems operate under anaerobic conditions unless otherwise stated.

Microorganism

System

Conditions

Formate

concentration [mM]

H2 productivity

Maximum cell concentration&ek

Ref.

Escherichia coli XL1-BLUE

batch experiments in a stirred 3.5 L bioreactor (3.1 L w.v.[a])

pH 6.5, 37 °C

30

426 mL L−1 d−1

0.8 gDCW L−1

[186]

75

725 mL L−1 d−1

Escherichia coli MG1655

batch culture in 250 mL flasks

&bkpH 6.5, 37 °C[b]&ek

20

6 mM in 16 h

OD600=1.7

[187]

Escherichia coli JW2701-1(pVSC133)

batch culture in vials (10 mL w.v.)

37 °C

20

7 μmol mgprotein−1 h−1

[189]

Escherichia coli BL21

batch culture in 125 mL serum bottles (100 mL w.v.)

pH 6.8, 37 °C

10

12.5 mL in 20 h

[190]

Enterobacter aerogenes

batch 2.5 L stirred reactor

pH 5.8, 37 °C

20

16 mmol gcell−1 h−1

0.4 g L−1

[192]

Thermococcus onnurineus NA1

batch culture

pH 6.5, 80 °C

10c

1.91 mmol L−1 h−1

1.2×108 cells mL−1

[193]

Thermococcus onnurineus NA1

7 consecutive batch cultures with cell recycling in 30 L fermentor (15 L w.v.)

pH 6.1–6.2, 80 °C

400

236 mmol L−1 h−1

OD600=1.7

[194]

Thermococcus onnurineus NA1

pH stat batch culture in 3 L fermentor (1 L w.v.)

pH 6.1–6.2, 80 °C

10c

404 mmol gcell−1 h−1, 2.82 mol L−1 h−1

OD600=18.6

[171]

Thermococcus onnurineus NA1

batch culture in 3 L bioreactor

pH 6.1–6.2, 80 °C

400

323 mmol gcell−1 h−1,

102 mol L−1 h−1

OD600=0.8

[195]

Pyrococcus furiosus (with FHL gene from T. onnurineus)

batch culture

pH 8, 80 °C

50

4.2 mmol L−1 h−1

10 μgcell_protein mL−1

[196]

Clostridium diolis JPCC H-3

batch culture in a 30 mL bottle (5 mL w.v.)

pH 6.8, 40 °C

3.0[c] (FA)

2.91 mL in 48 h

OD660=0.10

[197]

Desulfovibrio vulgaris

3 L anaerobic stirred tank reactor (1.5 L w.v.)

pH 7.2, 37 °C

40

7 mmol gDCW −1 h−1, 15 mL L−1 h−1

39 mgDCW L−1[d]

[198]

Desulfovibrio vulgaris

sparging column 830 mL reactor in batch mode &bk(340 mL w.v.)&ek

pH 7.2, 41 °C

80

2500 mL gDCW−1 h−1, 125 mL L−1 h−1

50 mgDCW L−1[d]

[199]

  • [a] Working volume. [b] Microaerobic conditions. [c] g L−1, d: initial cell load.

E. coli mutants have been used in these types of systems as well. For example, Redwood et al.188 report 37 % improvement of hydrogen production for mutant E. coli strain where the two uptake hydrogenases were deleted. The same work also claims that H2 production is possible in aerobically grown cells when formate is present in the growth medium. Random mutagenesis techniques have also led to E. coli strains with enhanced FHL activator protein gene, exhibiting 9 times faster H2 production rates. More specifically, a production rate of 7 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0017  mgprotein−1 h−1 was calculated by Sanchez-Torres et al.189 for strain JW2701-1(pVSC133), as opposed to 0.8 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0018  mgprotein−1 h−1 for the parental strain. E. coli BL21 grown on a formate-supplemented glucose-containing medium under microaerobic conditions also appeared capable of H2 production, only when it was engineered to express [NiFe] hydrogenase-1. Moreover, addition of nickel and iron to the medium showed to improve enzymatic activity. The H2 production achieved was 12.5 mL L−1 h−1, which is almost three times higher than the respective H2 production in a non-formate supplemented medium.190

Facultative anaerobe Enterobacter asburiae SNU-1 has also been reported to produce hydrogen from formate, with a yield of 0.43 molurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0019  molformate−1. Production began at the end of the exponential phase, but mainly occurred in the stationary phase. However, formate was only an intermediate product of glucose metabolism, and no extrinsically added formate was used. Glucose concentration of 25 g L−1 in the medium and neutral pH were determined as the optimum conditions for maximum H2 productivity, resulting in 398 mL L−1 h−1 and 174 mL L−1 h−1 maximum and overall productivities respectively.191 Another member of the Enterobacter genus, namely Enterobacter aerogenes, has been evaluated for its H2 producing ability in a culture with a glucose-containing medium supplemented with 20 mM formate. Compared to non-formate containing control cultures, the hydrogen evolution rates observed were faster, while cell growth was slowed down in the presence of formate. The system was tested at two pH values of 5.8 and 6.3. The former exhibited higher extrinsic formate decomposition rates, while hydrogen evolution rates were similar in both cases, reaching a maximum value of 16 mmol gcells−1 h−1.192

Thermococcus onnurineus NA1 has been reported to grow on formate while producing hydrogen at promising rates. In particular, this hyperthermophilic bacterium is the reported as the first microorganism to show evidence of H2 production coupled to cell growth, which was optimum at a pH of 6.5 and reached a maximum rate of 3.8 mmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0020  L−1 h−1.193 Sodium formate was also tested as a substrate, with a 400 mM concentration being evaluated as optimum. In such a system, an OD600 of 0.87 and a H2 production of 354.7 mmol L−1 were achieved after 5 h of culture. The optimized conditions were implemented in a consecutive batch culture process, with cell recycling, initially tested with 3 L bioreactors and then scaled up to 30 L. Both systems yielded encouraging hydrogen production rates at the end of the seventh cycle, 205 mmol L−1 h−1 and 236 mmol L−1 h−1 respectively.194

In a different study, T. onnurineus (strain NA1) has been reported to produce H2 in a cell density-dependent rate that reached a maximum of 2.82 mol L−1 h−1 (equivalent to a specific production rate of 404 mmol g−1 h−1) at OD600 of 18.6. The specific production rate reached at that cell density was the same with that observed for a cell density of OD600=1.3, in which case the volumetric production rate was rate were 233 mmol L−1 h−1.171

Interestingly, an engineered strain of T. onnurineus overexpressing the frhAGB hydrogenase gene has exhibited oxygen tolerance, reaching the same maximum cell density when grown under anoxic and oxic conditions and showing only a slight decrease in maximum hydrogen production rate in the latter case (364.8 and 323.3 mmol g−1 h−1 respectively). Both growth and H2 production (with 400 mM sodium formate as substrate), however, were delayed by a lag phase of approximately 6 h the presence of oxygen.195

Belonging to the same family as T. onnurineus, Pyrococcus furiosus is a hyperthermophile that cannot produce H2 from formate on its own but can be effectively metabolically engineered to accommodate the FHL expressing gene from T. onnurineus. The recombinant cells produced hydrogen from 50 mM formate substrate at a maximum rate of 4.2 mmol L−1 h−1, which is slightly higher than the respective value reported above for T. onnurineus. The potential of recombinant P. furiosus as a catalyst for formate conversion to hydrogen lies in its additional ability to produce H2 from sugars as well, through an independent pathway that does not hinder formate-based production, making the simultaneous utilization of both substrates possible. Growth was similar when cells were grown on media with and without formate, but H2 production almost twice as high in the former case.196

Clostridium diolis JPCC H-3 is another member of the Clostridiaceae family reported to oxidize formate to hydrogen and CO2, optimally at pH 6.8 and 40 °C, with a production of 2.91 mL H2 per 5 mL medium solution containing bacterial cells, after 48 h.197

There is evidence that sulfate-reducing bacteria can produce hydrogen from formate in the absence of sulfate. More specifically Desulfovibrio vulgaris has shown preference for H2 production from formate (when also compared to lactate and ethanol). When used in an anaerobic stirred-tank reactor system, for which optimum conditions were determined to be pH 7.2, 44 mM formate, presence of Fe, Ni and Se and initial cell load of 39 mgDCW L−1, formate was 100 % converted to H2 at a specific rate of 7 mmol g−1 h−1, corresponding to 560 mLurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0021  L−1medium. The addition of metal cofactors Fe, Ni and Se to the medium led to nearly 40 % increased H2 production.198 The same research group have found that D. vulgaris can grow on formate in an anaerobic stirred tank reactor at a rate of 0.078 h−1, reaching a maximum hydrogen production rate of 40 mL L−1 h−1. The sparging of the reactor with gas to keep the partial pressure of hydrogen low was found to be necessary for growth to be coupled to H2 production. Production rate was further improved to reach 125 mL L−1 h−1, in a column reactor with optimum conditions of 41 °C and initial cell load of 50 mgDCW L−1.The flow of argon was also an affecting parameter and was found to be optimum at 80 mL min−1. The reactor was operated in fed-batch mode as well, with formate being periodically added to maintain the production rate high. Results showed that cells are capable of maintaining their activity in consecutive cycles, reaching almost identical maximum hydrogen production rates in all formate addition cycles.199

3.2 Formate decomposition catalyzed by hydrogen dependent CO2 reductase (HDCR)

3.2.1 The HDCR complex

The HDCR complex is an enzyme complex isolated from acetogens Acetobacterium woodii and Thermoanaerobacter kivui, which, similarly to FHL, possesses a FDH and a [FeFe] hydrogenase as its main catalytic subunits.200 As seen in the enzyme from A. woodii, the complex additionally includes two smaller proteins to serve electron transfer. The part of the A. woodii genome encoding the enzyme complex includes 7 genes, encoding FDHs fdhF1 and fdhF2, FDH maturation protein FdhD, hydrogenase hydA2 and electron transfer proteins hycB1, hycB2 and hycB3, the four out of which participate in the formation of the complex. When FDH and hydrogenase activity were measured separately, with MV as a cofactor, the respective rates for formate and hydrogen oxidation were ∼600 μmolformate min−1 mg−1 and 10 800 μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0022  min−1 mg−1. 10 μg of the purified enzyme complex produced 25 μmol H2 in 80 min from 50 mM formate. This amount was measured at 30 °C, although the optimum temperature for formate oxidation was determined to be 40 °C.201 The structure of the HDCR from T. kivui is the same, also containing 4 subunits, namely a FDH, a [FeFe] hydrogenase and two electron transfer units. Cryon-electron microscopy provided evidence that the enzyme forms filaments bound to the plasma membrane, in which the smallest repeated unit is a hexamer. It comprises of FdhF and two hydA2 subunits, electron transfer between which is made possible through binding to three Fe–S proteins (one hycB3 and two hycB4), as seen in Figure 6 (PDB ID: 7QV7).202 This complex exhibits remarkable catalytic activity, as portrayed in the calculated turnover frequency of 9 892 000 h−1, which surpasses that of the HDCR from A. woodii at 30 °C, even though T. kivuii is thermophilic.203 Gene clusters similar to those encoding HDCR of A. woodii and T. kivuii have been found in several other microorganisms. However, no other isolated HDCR complex has been reported.

Details are in the caption following the image

a) Schematic presentation of the structure and subunits of HDCR of A. woodii (*: The FDH may or may not contain selenium, depending on which of the two FDH encoding genes is expressed) b) Crystal structure of the HDCR from T. kivuii in filamentous form (PDB ID: 7QV7).

3.2.2 Systems for H2 production from formate using HDCR

In the first work where HDCR from Acetobacterium woodii was reported, the activity of the purified complex was tested with 50 mM formate as substrate. H2 production reached 27 μmol in 80 min.201 Another study used A. woodii whole cells to catalyze H2 production from formate. A. woodii cells exhibited growth on formate alone, but both doubling time and H2 productivities were higher when cells were grown on fructose and then supplied with formate to produce H2. Harvested fructose-grown cells with 100 mM formate as substrate reached a 30.5 mmol g DCW−1 h−1 specific H2 production rate at a cell density of 0.5 mg mL−1, and a 79 mmol L−1 h−1 volumetric production rate at a concentration of 4 mg mL−1. However, it was observed that highest specific production rate corresponded to the lowest volumetric production rate and vice versa. H2 production was also possible in a single fermentation vessel, in two steps: initial growth of the cells on 100 mM formate and subsequent addition of formate to 300 mM to initiate the production. The highest specific H2 production rate achieved was 19 mmol gCDW−1 h−1. Formation of high amounts of byproduct acetate appeared to cause inhibition, which could however be prevented with the addition of sodium ionophore ETH2120.204

Schwarz et al.205 also utilized fructose-grown A. woodii resting cells at a concentration of 1 mg/mL for formate based H2 production. Initial experiments in serum bottles showed higher specific H2 production rates when 200 mM K-formate was used as substrate (71 mmol g−1 h−1), in an imidazole containing buffer. However, when production was upscaled in a 2 L bioreactor, using 150 mM formic acid as substrate and a K-phosphate buffer, the calculated specific production rate was 27.6 mmol g−1 h−1. The lower rates were attributed to the different substrate and buffer, which were selected, however, to prevent salt build-up and high drop of pH upon substrate addition respectively. After 3.5 h of fermentation, FA was added to the system to maintain substrate concentration and pH value, converting it to a fed-batch mode. However, in that case hydrogen production rate started to decrease, dropping to 50 % in 5 h. This decrease was attributed to the reduction of the optical density of the biocatalyst cells. However, a high yield of 0.95 mol H2 per mol FA was calculated for the whole system.205

When the first-time reported purified HDCR complex of T. kivuii was evaluated for its H2 productivity, the highest specific production rate reported was 930 μmol mgenzyme−1 min−1, with 150 mM formate as substrate and with an enzyme concentration of 3 μg mL−1. The enzyme exhibited good storage stability at room temperature, losing 50 % of its activity in 52 days (as opposed to 9 h at 60 °C), as well as tolerance to freeze and thaw cycles.203 Harvested cells of T. kivuii grown on glucose have been efficiently used as a whole-cell biocatalysts for H2 production from formate, reaching high production rates (685 mmol g−1 h−1 for 0.3 mg mL−1 cells and 300 mM sodium formate, 999 mmol L−1 h−1 for 4 mg mL−1 cells and 600 mM sodium formate), while producing only marginal amounts of acetate as a byproduct. H2 production was not favored by alkaline pH and reached the highest rate at pH values between 5.5 and 7. Optimum temperature was determined to be 70 °C and the microorganism could maintain H2 production in the presence of up to 0.79 % oxygen. The microorganism was also efficient in a 2 L batch stirred bioreactor system, achieving a yield of 0.86 molurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0023 per mol formate when pH was kept constant.206 The data regarding each of the to-date reported HDCR systems producing H2 from formate is presented in Table 5.

Table 5. H2 producing systems utilizing HDCR as a biocatalyst.

Microorganism

Form of biocatalyst

Reaction volume[a]

Conditions

Substrate

Concentration [mM]

H2 production

[mmol gprotein−1 h−1]

Volumetric productivity [mmol L−1 h−1]

Enzyme/cell concentration [mg mL−1]

Ref.

A. woodii

purified complex

1 mL (in 7 mL serum vials)

&bkpH 7.5, 30 °C&ek

formate

50

27[c]

n. d.

0.010

[201]

A. woodii

whole cell

50 mL (in 115 mL glass bottles)

pH 7, 30 °C

formate[b]

100

19[d]

1.2

0.25–0.3[f]

[204]

pH 7, 30 °C

formate

100

71

≈35

0.5

pH 7, 30 °C

formate

100

≈35

79

4

A. woodii

whole cell

2 L (stirred tank bioreactor)

pH 6.2, 30 °C

FA

150

27.6

n. d.

1

[205]

T. kivuii

purified complex

1 mL (in 9 mL serum vials)

pH 7.0, 60 °C

formate

150

930[e]

n. d.

0.003

[203]

T. kivuii

whole cell

10 mL (in 120 mL serum bottles)

pH 7, 70 °C

formate

300

685

205

0.3

[206]

600

250

999

4

  • [a] All systems were operated in batch mode. [b] Cells were also grown on formate supplemented with fructose. [c] μmol of hydrogen produced in 80 min, no specific production rate determined. [d] Only determined in mmol gCDW−1 h−1. [e] μmolurn:x-wiley:18645631:media:cssc202202312:cssc202202312-math-0024  min−1 mgenzyme−1. [f] OD600 n. d.: not detected.

Overall, whole-cell catalysts possessing the FHL or HDCR complexes can be employed for formate-based hydrogen production. The latter have exhibited remarkably higher production rates, but have been less studied. Moreover, there is a differentiation in processes which use pre-grown cells as biocatalysts or couple growth with H2 production. In the second case, no separate step of bacterial growth would have to be incorporated in the process, but the reaction solution would be more complex and contain additional compounds needed for bacterial growth.

4 Challenges

Apart from all the aforementioned parameters affecting the two processes for FA production and H2 release, which need to be investigated and optimized, it is important to consider the challenges regarding the combination of these two processes. One of those is the scale up of the bioelectrochemical CO2 reduction. Flow through electrochemical reactors have been appointed as more efficient for scale up of FA production from CO2 when compared to conventional batch and three compartment cell reactors.207 This comparison, however, was made for non-biocatalytic systems. Apart from increasing the size of the individual reactor, stacking multiple reactors can be another strategy to develop a larger scale process.208

Another aspect to investigate is if and how formate would need to be separated from the electrosynthesis reaction mixture to be fed as a substrate to the next step of its dehydrogenation. Common methods for such product separation include extraction, ion exchange resins or crystallization,209 but the parameters of such a step would have to be determined after the ones for the two main steps of the cycle.

Lastly, investigation on whether CO2 or HCO3 is a preferrable substrate for FA production is necessary, as it determines if a separate step for CO2 capture by CA would have to be developed, or if it could be combined with the FDH catalyzed reaction.

5 Summary

Research interest has turned to biocatalysts in search of more environmentally viable solutions for energy production and storage and reduction of CO2 emissions. As seen from several studies, there are enzymes with different properties, from several microorganisms, that can catalyze both the CO2 reduction to formate and the decomposition of the latter to CO2 and H2. A biocatalyst that can show preference for CO2 reduction instead of formate oxidation is a key component of a CO2 to FA conversion process, and protein engineering can play a role in optimizing that. The matter of oxygen tolerance is also to be taken into consideration, as it affects how easily a large-scale industrial bioprocess could be designed. Another major parameter of a biocatalytic system as described above is the cofactor. Research has shown that there are promising artificial alternatives to NADH, as well as mediators that can make NADH regeneration efficient. Apart from optimizing the reaction conditions, immobilization of the enzymes on the cathode is a solution that could contribute to stability, reusability, facilitated electron transfer as well as product separation.

Various parameters must be considered for optimizing the H2 production from formate as well. This is a process that requires anaerobic conditions, something that perplexes the process design. Moreover, potential ability to grow with formate as a sole carbon and energy source can affect the selection of biocatalyst, as well as the setup of the system and whether pre-grown or growing cells are used as biocatalysts. Bacterial inoculum, temperature, pH, hydrogen partial pressure and bioreactor operational mode also need to be optimized.210

Overall, several factors need to be further investigated, and so far, few of the reported systems have studied the enzymes with a larger scale process approach. Nevertheless, studies provide evidence to support the proposal of an entirely biocatalytic cycle for hydrogen production form CO2, with formic acid as an intermediate storage molecule. The assessment of the environmental impact of this cycle compared to CO2–FA interconversion through chemical catalysis or conventional methods for hydrogen production such as water electrolysis is complicated and would require the calculation of specific metrics, such as the environmental factor E, and a life cycle analysis.211, 212 Although the use of biocatalysts alone is not enough to make a process environmentally friendly, their properties, including non-toxicity, mild operating conditions, use of water-based non-hazardous solvents and high specificity resulting in fewer by-products, are in alignment with principles of sustainability and green chemistry.213 Moreover, the conventional catalysts used for CO2–FA interconversion involve noble metals,214 the mining process of which can be strenuous for the environment,215 and operate at higher temperatures, which leads to higher energy demands. In particular, LCA analyses for CO2 conversion to FA have shown that thermal catalysis has the highest environmental impact among the processes used and that the supply of heat is the greatest contributor to the global warming impact of processes.207, 216 This data justifies further investigation of the proposed biocatalytic cycle as a more sustainable solution to the issue of hydrogen transport and storage that could also contribute to the ongoing efforts towards carbon neutrality.

Acknowledgments

The Centre for Hydrogen Energy Systems Sweden – CH2ESS at Luleå University of Technology and the Swedish strategic research environment – Bio4Energy are acknowledged for financial support.

    Conflict of interest

    The authors declare no conflict of interest.

    Biographical Information

    Eleftheria Sapountzaki graduated in 2021 from National Technical University of Athens, Greece, with a Master's degree in Chemical Engineering and a specialization in biotechnology. She is currently a PhD student in the Biochemical Process Engineering group of Luleå University of Technology, Sweden, working on bioelectrosynthesis and whole-cell catalysis within a project for the development of a biocatalytic cycle for chemical hydrogen storage and release via CO2–formate interconversion.

    Biographical Information

    Ulrika Rova is a Professor in Biochemical Process Engineering at Luleå University of Technology, Sweden. Prof. Rova's research is focused on lignocellulosic biomass pretreatment/fractionation and conversion to high-added value and sustainable products, using enzymes and various microbial platforms. Research activities also includes biocatalytic CO2 capture and conversion where development of bioelectrochemical processes plays a central role.

    Biographical Information

    Paul Christakopoulos has been appointed Chair Professor of Biochemical Process Engineering at Luleå University of Technology, Sweden, from February 2012. He has previously served as a Professor of Industrial Biotechnology at the School of Chemical Engineering, National Technical University of Athens, Greece. His research is focused on the development of biochemical (green chemistry) processes for the production and refinement of chemicals, fuels and material from CO2, either captured before it is emitted to the atmosphere (non-biomass route) or by recovering it from the atmosphere via photosynthesis in the form of biomass (biomass route).

    Biographical Information

    Io Antonopoulou is currently an Assistant Professor at the Division of Chemical Engineering at Luleå University of Technology, Sweden. She received her Master's degree in Chemical Engineering from the National Technical University of Athens in Greece and conducted a Ph.D. in Biochemical Process Engineering at Luleå University of Technology. Her research interests are in the application of enzymes for plant biomass valorization, carbon capture utilization and storage, and hydrogen production and storage.