Enhanced Bio‐Electrochemical Reduction of Carbon Dioxide by Using Neutral Red as a Redox Mediator

Abstract Microbial electrosynthetic cells containing Methylobacterium extorquens were studied for the reduction of CO2 to formate by direct electron injection and redox mediator‐assisted approaches, with CO2 as the sole carbon source. The formation of a biofilm on a carbon felt (CF) electrode was achieved while applying a constant potential of −0.75 V versus Ag/AgCl under CO2‐saturated conditions. During the biofilm growth period, continuous H2 evolution was observed. The long‐term performance for CO2 reduction of the biofilm with and without neutral red as a redox mediator was studied by an applied potential of −0.75 V versus Ag/AgCl. The neutral red was introduced into the systems in two different ways: homogeneous (dissolved in solution) and heterogeneous (electropolymerized onto the working electrode). The heterogeneous approach was investigated in the microbial system, for the first time, where the CF working electrode was coated with poly(neutral red) by the oxidative electropolymerization thereof. The formation of poly(neutral red) was characterized by spectroscopic techniques. During long‐term electrolysis up to 17 weeks, the formation of formate was observed continuously with an average Faradaic efficiency of 4 %. With the contribution of neutral red, higher formate accumulation was observed. Moreover, the microbial electrosynthetic cell was characterized by means of electrochemical impedance spectroscopy to obtain more information on the CO2 reduction mechanism.


Introduction
Over the past decades, atmospheric carbon dioxide (CO 2 )c oncentration has been increasingc ontinuously and it is regarded as am ajor greenhouseg as. [1][2][3] Thus, the reductiono fa tmospheric CO 2 hasa ttracted lots of interest as the carbon capture and utilization (CCU) processes. [4] The CO 2 ,a sac arbon source, can be convertedt os everal value-added products, [5,6] such as carbon monoxide (CO), formic acids, acetic acid, methane, methanol and ethanol, through biological, [7][8][9] electrochemical [10][11][12][13] and photo(electro)chemical approaches. [14][15][16][17][18] However, CO 2 conversion requires al arge energy input due to the fact that CO 2 is in ah ighlys table and low energy state. According to the thermodynamic reduction potentials, ao ne-electron reaction from linear CO 2 to its bent anionic form (CO 2 ·À )r equires ah igh potential of À1.90 Vv ersus standard hydrogene lectrode (SHE). Compared to the single-electron process, the proton-coupled multi-electron reactions requires ubstantially lower potentials. [16] For example, the reduction potentialo ft he CO 2 reduction to formic acid (HCOOH) involving two electrons and two protons, is À0.61 Vv ersus SHE. [16] Nonetheless, the reduction potentials observed experimentally are much more negative than what thermodynamics predicts due to overpotentials. To overcomet hese high energy barriers, several approaches using catalysts were introduced. [19][20][21][22][23][24][25][26][27] Among variousC O 2 reduction systems, the biological approach using microorganisms and enzymes as catalysts has been investigated due to the accessibility of catalysts from the biosphere compared to synthetically obtained catalysts. Moreover,b iocatalysts provide high selectivity towards adducts and products and can be used under mild, ambient conditions without high temperature or high pressure. [28][29][30] In the enzymatic approach, dehydrogenase enzymesh ave been reported as efficient catalystsf or the CO 2 conversion to alcohols, aldehydes andother hydrocarbons. [31] Generally,these dehydrogenase enzymes only catalyzes pecific reactions with the requirement of as acrificial cofactor. [31][32][33] For example, formate dehydrogenase (FDH) is known to catalyze the reduction of CO 2 to formate with the aid of nicotinamide adenine dinucleotide (NADH) as ac ofactor. [34,35] Instead of using the sacrificial cofac-Microbial electrosynthetic cells containing Methylobacterium extorquens were studied for the reduction of CO 2 to formate by direct electron injection and redoxm ediator-assisted approaches, with CO 2 as the sole carbon source. Theformation of ab iofilm on ac arbon felt (CF) electrode was achieved while applying ac onstant potential of À0.75 Vv ersus Ag/AgCl under CO 2 -saturatedc onditions. During the biofilm growth period, continuousH 2 evolution was observed. The long-term performance for CO 2 reduction of the biofilm with and without neutralr ed as ar edoxm ediatorw as studied by an applied potentialo fÀ0.75 Vv ersus Ag/AgCl.T he neutral red was introduced into the systemsi nt wo different ways:h omogeneous (dissolved in solution)a nd heterogeneous (electropolymerized onto the working electrode). The heterogeneous approach was investigated in the microbial system, fort he first time, where the CF working electrode was coated with poly(neutral red) by the oxidative electropolymerization thereof. The formationo f poly(neutral red) was characterized by spectroscopict echniques.D uring long-terme lectrolysis up to 17 weeks, the formation of formate was observed continuously with an average Faradaic efficiency of 4%.W itht he contribution of neutral red, higher formate accumulation was observed. Moreover,t he microbial electrosynthetic cell wasc haracterized by means of electrochemical impedance spectroscopy to obtain more information on the CO 2 reduction mechanism.
tors, the possibility of direct electron injection to dehydrogenase enzymes has been reported recently for electrochemical methods. [36][37][38][39] However, the main limitation of using enzymes is that the enzymes are isolated from specific microorganism strains and the purification processes requiret rained personnel and specific equipment.
In contrastt oe nzymes, living biocatalysts (microorganisms) provide ag reat advantage in terms of sustainable systemsd ue to their ability to reproduce. [40] Compared to FDH, which is very often sensitivet oa no xygen-containing environment, the structureo fl iving microorganisms preserves the activity of enzymes inside due to their membrane and sincet hey are able to reproduce themselves, new activee nzymesa re regularly generated. Additionally,t he conversion efficiency of FDH for converting CO 2 to formate is usually extremelyl ow as compared to that for FDH converting formatet oC O 2 . [41] Several microorganisms have been already investigated in the field of CO 2 reduction for their capability of capturing and/or converting CO 2 to valuable products by using electrons from electron carriers,s uch as hydrogen (H 2 ). [42][43][44] In the 1930s, the first acetogenic bacteria (Clostridium aceticum)w as investigated and the authors reported that CO 2 andH 2 could be converted to acetate, witht he mechanism being describedl ater on as the reductivea cetyl-CoA or the Wood-Ljungdahl pathway. [45][46][47][48] Moreover,t he selectivity towards the products of CO 2 reduction can be tuned corresponding to the local dehydrogenase enzymes of the microbial strains. Furtheri nvestigationo ft hese microorganisms reported that they are able to receive electrons either directly from the cathode or indirectly from the redox mediators of which the mechanism was called as extracellulare lectron transfer. [44] Recently,H wang et al. firstly reported the capability of Methylobacteriume xtorquens,w hich is known to be able to grow on reduced C1 compounds, to reduce CO 2 to formate under aerobic conditions by using electrons supplied from an electrode through am ediator-assisted approach. [49][50][51] One possible approach to enhancet he microbial electrosynthesis system is to introduce ar edox mediator in order to facilitate the electron transfer from the electrode towards microorganisms' cytoplasmic membrane. Accordingt op reviouss tudies, 3-amino-7-dimethylamino-2-methylphenazine hydrochloride (neutralr ed), known as ap Hi ndicator and as tainingd ye, was used as an efficient redox mediator in several bio-electrochemicals ystemsb ecause its redoxp otential is close to that of the NAD + /NADH redox couple, which is one of the major electron carrieri nt he microbial electron transport chain. [52][53][54] Moreover,i ti sk nown that neutral red can be polymerized electrochemically on various substrates, especially carbonbased ones, yielding ap oly(neutral red) coating on the electrodes. [55] As reported in previous studies, the resulting poly(neutral red) film was electrochemically active and chemically stable under biological conditions, and it has been widely used for bio-electrochemical applications similar to those of the monomer,s uch as sensors [56] andN AD + /NADH regeneration. [57] From ap ractical point of view,as ystem containing ar edox mediatorc oated or directly deposited onto the electrode is preferred, rathert han the homogenous approach (a redox me-diator dissolved in the electrolyte). This is firstly because product separation becomes easier and secondly because the electron transfer is greatlye nhanced. [58][59][60] Furthermore, such a system will have the advantage of long-term operation and lower mediatorcosts.
In this study,w ef urther investigated the long-term performance of the microbial electrosynthetic cell (MEC)o fM. extorquens by monitoring the products produced during the biofilm growth and CO 2 reduction period. Additionally,t he aid of neutral red as ar edox mediatori na nM EC system was investigated in two differenta pproaches:h omogeneousa nd heterogeneous ones. In the first approach studied, the cathode electrolytec ontained soluble neutral red, whereas in the second approacht he poly(neutral red) was directly coated onto the electrode.

Results and Discussion
Inoculation of microorganisms M. extorquens biocathodes were developed in the two-compartment electrochemical three-electrode set-up as shown in Figure1A. The cathodic compartmentc onsisted of the mediumc ontaining 10 %( v/v)o fM. extorquens pure culture suspension equipped with ac arbon felt (CF) working electrode and aA g/AgCl( 3m KCl) reference electrode. The anodicc ompartment contained a0 .2 m phosphate buffer solution at pH 7.0 equipped with aP tp late counter electrode. Ap otential at À0.75 Vw as appliedc onstantly for am onth under CO 2 -saturated atmosphere. The cathodic compartment was purged with CO 2 approximately 2h every week to keep the compartment saturated with CO 2 .S ince M. extorquens is capable of extracellulare lectron transfer through their pili, it is able to take up the electrons directly from the electrode. [51] Consequently,i t can be inoculated electrochemically on ac arbon-based electrode. After af ew days of inoculation, the growth of the biofilm on the electrode could be observed and increased gradually by the inoculation time. After one month of inoculation, the CF electrode was covered with the biofilm as shown in Figure 1B.A dditionally,t he electrodes were dried under ambient conditions (pictures of electrodes are shown in Figure S1) overnight for optical microscope and SEM imaging as shown in   [62,63] During this period,h eadspace samples were collected daily and analyzed by gas chromatography (GC). The GC results showedt hat H 2 was the only product detected in this phase. The amount of H 2 produced is presenteda sab lack line in Figure 3. The arrows indicate the period when the cathodic chamber was purged withC O 2 .T he results revealt hat H 2 productionwas observed continuously and the highest H 2 production was observed in the first day,i mmediately after CO 2 purging with ap roduction rate of 0.86-1.04 mmol/day. However, the lowering in H 2 amount in the third to fifth day of the cycle might be because the microorganisms consumed H 2 as ar educing agent in their metabolisms as described in the Wood-Ljungdahl pathway. [64] Furthermore, the electrical charges( Q) consumed during this period were plotted together in Figure 3 (blue data points). These results showed that the electrons were consumedc onstantly during this period, not only in the region where H 2 was produced. The red dashed lines were plottedi no rder to present the linear fitting of charged ata points. The fitting provides the rates of electrical charges consumed in each purging cycle which weref ound to be in the rage of 170.5-418.8 C/day.T he %FE of H 2 production in the growingp hase was calculated for one day immediately after CO 2 purging. The %FEs were in the range of 35-98 %.
As the experiments were done in aqueous solution, H 2 generation from electrochemical water-splitting has to be taken into consideration. However,w eb elieve that the H 2 produced during the inoculation periodi sp roduced by microorganisms themselves, as part of their metabolism. To confirmt his, control experiments were carried out by constantly applyingt he same potentialo fÀ0.75 Vi nt he media by using ab are CF workinge lectrode (withoutt he biocatalyst). Thee xperiments were done under N 2 -a nd CO 2 -saturated conditions, and daily headspace samples were collected for ap eriod of 1week. The GC analysisr evealed that no H 2 was produced during the electrolysis of bare CF electrode in N 2 -o rC O 2 -saturated conditions (as shown in Figure S3 and S4). Considering the resultso btained for thesec ontrol experiments together with our observation that the productionr ate was increased after 2weeks of inoculation related to the more biofilm formed on the electrode.T hus, we believe that the H 2 constantly produced duringt he growth might be the product of microbial fermentation processes. [65] In this work, we aimed to study the long-term performance of the biocathodes resulting from the inoculation of M. extorquens and the effect of neutral red as ar edox mediatorb oth when the redox mediator is dissolved in the solution and when it was coated on the electrode. Three different electrosynthetic systemsw ere designed as MEC 1, 2a nd 3. MEC 1i s defineda st he electrosynthesis system equipped with the biofilm CF working electrode. In the second cell (MEC 2), we investigatedt he effects of neutral red as aredox mediatorinthe solution.T herefore, the cell was equipped with the biofilm CF as aw orking electrode andb uffer solution containingn eutral red as the electrolyte solution. In addition, to study the effects of the redox mediatorw hen it was directly attachedo nt he electrode,M EC 3w as developed by using the biofilm coated on PNR/CFa saworkinge lectrode in mediator-free buffer solution.

Electropolymerization of neutral red
The PNR/CF electrode was prepared by the oxidative electrochemicalp olymerization and used as working electrode in MEC 3. The electropolymerization was performed in 0.1 m phosphate buffers olution pH 6.0c ontaining 1mm neutralr ed at potentials between À1.0 to 1.0 Vw ith the scan rate of 50 mV s À1 over 20 cycles as shown in Figure 4A,y ielding a poly(neutral red)-coating onto the CF electrode (PNR/CF). In the first cycle, three oxidation peaks at À0.39, 0.20 and around 0.8 Vw ere observed. The first two redox couples indicated electrochemical features of protonated neutralr ed and its polymer,r espectively.A nother irreversible oxidation peak at around0 .8 Vi sr eferred to the formation of cation radicals pecies that can initiate polymerization. These results are in agreement to the results reported previously. [55,57] Upon polymerization, the current of the first two redox couples increased as more redox active polymer is formed on the electrode. Furthermore, ap ositive shift in the potentialp ositions and broader features were observed. These phenomena could be related to the changes in the electrode surface and the branching of the polymer.After removal of the remaining monomer solution by rinsing with water,t he resulting PNR/CF electrode was electrochemically characterized by cyclic voltammetry.C Vs were   Figure S5. Furthermore, the SEM images of PNR/CF and bare CF were shown in Figure 4B, revealing the formation of polymericfilm.
With the similarm anner to PNR/CFp reparation, neutral red was electropolymerizedo nto Cr/Au-coated glass electrodes for FTIR and UV/Vism easurements. Figure 5s hows FTIR spectra of neutralr ed (red line) andp oly(neutral red) (blue line) together with the proposeds tructure of poly(neutral red). The FTIR spectrumo fp oly(neutral red) shows characteristic absorption peaks at 1609, 1185 and 812 cm À1 ,w hich are attributed to vibrationo fC =Co rC =N, in-plane bending of CÀHa nd deforma-tion of aromatic ring, respectively. [66] The absorption peaks of poly(neutral red) were shifted 10 cm À1 for n C=Co rC =N ,7cm À1 for d CÀH and 9cm À1 for aromatic ring deformation towards negative wavenumber,c ompared to those of neutral red monomer. The shifts referred to increase in p-conjugation of the polymer. Additionally,UV/Vis absorption of the polymericfilm was investigated, the spectra for which are shown in Figure S6. The absorptions pectra exhibit maxima at wavelengths of 531, 500 and 455 nm for neutral red aqueous solution, drop-casted neutral red monomer, and poly(neutral red) film, respectively.T he absorption spectrum of poly(neutral red) film was consistent with its of monomer with broader features, which is attributed to highera ggregation of aromatic rings in the polymeric layers.

Long-term microbial electrosynthesis
After one-month inoculation, the electrolyte solutions in both compartments were replaced completely with 0.2 m phosphate buffers olution pH 7.0. Ac onstant potentiala tÀ0.75 Vw as applied under CO 2 -saturated condition. In the cathode compartment,t he clearb uffer solutionb ecame partly cloudy because of the remaining medium absorbed in the CF electrode. Thus, the solution was replaced again with the buffer solution.
In all MECs, the experiments were performed in the twocompartment three-electrode set-up contained 0.2 m phosphate buffer solution equipped with the biofilm-coated CF (or biofilm-coated PNR/CF in MEC 3) workinge lectrode, aA g/AgCl   Figure 6i nb lue and red lines, respectively.I na ll MECs, there is an increase in reductive current starting from À300 mV observed for CO 2 -saturated systems as compared to N 2 -saturated conditions indicating CO 2 reduction ast he predominant reaction. In the CVs of MEC 1, the reductivec urrents are lower compared to MEC 2a nd MEC 3, showing enhancement in the electron-transfer processes in the presence of neutral red as ar edox mediator. The reductive peaks at around À300 mV in the CVs that were recorded in MEC 2a nd 3s ystemsb elonged to the reductive peaks of neutralred monomer.The observation of neutral red monomer peak in MEC 3i sassumed that it come from the degradation of poly(neutral red).
The long-term performance studies of all MECs for the re-ductionofCO 2 to formate were studied by applying aconstant potentiala tÀ0.75 Vu nder CO 2 -saturated condition. Thee lectrolysis was monitored by investigating the amount of formate produced in the solution using ion chromatography (IC) and the component of headspace gas using GC. The cathodic chamberw as purged with CO 2 for 2h every 2days to keep the system saturated with CO 2 and the electrolyte solution was replaced every 3-4 weeks. MEC 1w as monitored over 8weeks. In MEC 2, the experiment was performed in the presence of dissolvedn eutralr ed with ac oncentration of 50 mm in the cathodic electrolyte solution and was monitored over 12 weeks. Additionally,M EC 3w as developed and studied for 17 weeks. The running time of each MEC was limited to practical issues. According to the chromatography results, formate was the only product observed in all MECs. During the study, the electrolyte solution was replaced with an ew solution every 3-4 weeks. Each electrolyte replacement was labelled sequentially,n amely cycles 1-4. Figure 7s hows the plot of accumulative formate concentration and electricalc harges consumed during each cycle and the resultsa re summarized in Ta ble 1. Moreover,t he plots were linearly fitted (shown as dashedl ines), resulting to the rates of formate production and charge consumption.I nc ase of the overall running period,t he estimated production rate was calculated from the total amount of produced formate and charges. In MEC 1, formate production was found to be moderately stable and reached 2.0 and 1.1 mm in cycles 1a nd 2, with formate production rates of 66.7 and 71.7 mm/day,r espectively.T he corresponding %FEs were found to be 3a nd 2%.I nM EC 2, the produced formate concentration reached 2.9, 2.4 and 1.0 mm in cycles 1, 2  and 3, respectively.T he production rate increased significantly after the 16th running day in cycle 1, with the productionr ate of 96.9 mm day À1 for the first cycle. After that, the rate went down to 99.8 and 40.7 mm day À1 in cycles 2a nd 3, respectively. While the currentd ensity of the system increased in every cycle. Consequently,c alculated %FEs were substantially lower from 8% in the first cycle, to 5a nd 1% in the second and third cycle, respectively.I nc ase of MEC 3, formate was produced slowly in the first cycle with the rate of 42.1 mm day À1 andt hen reachedt he highest production rate of 133.5 mm day À1 in the second cycle. The rate decreased gradually to 88.8 and 34.4 mm day À1 in the third and fourth cycle, respectively.T he current density remained stable and %FEs were found to be 2, 8, 4a nd 2%,i nc ycles1,2 ,3and 4, respectively.F or comparison, another set of calculations was done for a5 7-dayp eriod. The results revealed that with the contribution of neutral red, 36 and 20 %m ore formate was accumulatedi nM EC 2a nd MEC 3, respectively.H owever,w efound out that after the second cycle, the systemsc ontaining neutralr ed were broken down, which might indicate that neutral red and poly(neutral red) degrade during the experiment. Compared to the previous report of the MEC-containing methylobacterium, in which methyl viologen was used as an electron mediator,f ormate accumulated to reachaconcentration of about 10 mm in 20 h. [51] However,t he cell reactiond uration was only up to 80 hd ue to the unstable character of methylv iologen.A dditionally,m ethyl viologen (also known as paraquat, ah erbicide) is toxic to mammals therefore using methyl viologen is not desirable. Regardingt hese practical issues, neutralr ed systems are more promising.
The mechanism of the electron transfer via neutralred mediator is not clear.I tc an be ar edox mediator and/orab iological mediator. One of the proposed mechanismso fn eutralr ed as a redox mediator in the homogeneous approach is presented in Scheme 1. When dissolved in neutralp Hw ater,n eutralr ed is in its acidic form (NRH + ). While applying potentiala tÀ0.750 V versus Ag/AgCl (3 m KCl), NRH + will be electrochemically reduced (NRH 2 ; E 0 = À0.535 Vv s. Ag/AgCl( 3 m KCl) for NRH 2 / NRH + ). [54] The resultingN RH 2 diffuses into the inner cytoplasmic membrane and then is oxidized into NRH + with the transfer of two electrons and one protonf or the reduction of NAD + to NADH (E 0 = À0.530 Vv s. Ag/AgCl (3 m KCl) for NAD + /NADH). [54] NADH is knownt ob ean essential coenzyme in the enzymatic reduction of CO 2 to formate,c atalyzed by formate dehydrogenase, andt he reaction takes place inside the microbial cells. Consequently,w ith the presence of neutral red in the system,t he electron transferfrom the electrode to the reaction sites is greatly improved, thus higherp roduction rates could be achieved.I na ddition, using immobilizedr edox mediators provides ap ractical system because the mediator is attached onto the electrode, thus, the problemso fm ediator contamination and product separation could be excluded. Additionally, PNR/CFw as found out to be as uitable support for the microorganismss inceb etter coverage of biofilm was observed on the electrode as shown in Figure 2B and C, S1 and S2. It is attributed to higher nitrogen to carbon ratio from amine groups of poly(neutral red) coated onto the electrode which promote the bacteria adhesion to the electrode. [67][68][69] Another set of control experimentsw ere made in order to confirm that withoutb acteria, even when using neutral red, no formate production could be observed. The experiments were performedi nasimilar set-up containing 0.2 m phosphate buffer solutionp H7.0 equipped with aC Fe lectrode without microorganisms inoculated on the electrode. The first experiment was done by applying ac onstant potential of À0.75 V under N 2 -saturated conditions for aw eek. After that, the cath- ode chamber was purged with CO 2 for 2h and the cell was monitored for an additional week. Then, neutral red solution was added into the cathodics olution and the experiment continued for another week. In all cases no formate production could be observed, confirming again that the CO 2 reduction can only occur when the bio-catalyst is presento nt he CF electrode (as shown in Figure S7).
To clarify the source of the carbon,a nother control experiment was done in MEC 2b ya pplying constant potential at À0.75 Vu nder N 2 -saturated atmosphere.T he electrolysisw as monitored by investigating the formate concentration and headspace sample analyzed with ion and gas chromatography, respectively.T he results show that no formate product or any gas products was observed duringt he control experiments. This information confirmed that the observed formate was produced by M. extorquens inoculated on the CF electrode with CO 2 as the carbonsource in the system.

Characterizationb ye lectrochemical impedance spectroscopy
The electrochemical impedance experimentsw ere performed in order to characterize the microbial electrochemical set-up. The Bode plot for the two-electrode systems is showni nF igure 8A and all data is summarized in Ta ble 2. Measured with the two-electrode system (PtÀPt), the resistanceo ft he 0.2 m phosphate buffer solution pH 7.0 (R sol )w as 85.0 W cm À2 .A dditionally,t he resistance of the Nafion membrane (R NF )w as determined as 0.87 kW cm À2 .I nt he obtained electrical circuit, the resistance of the electrolyte and the Nafion membrane are connected in series with each other.T he resistance of the workinge lectrodes (R WE ), which are the bare CF electrode and the one coated with the biofilm, were observed as 60.5 and 462.0 kW cm À2 ,r espectively.T he constantp hase element (CPE) was used for the descriptiono fn on-ideal capacity of the sponge-like CF electrode. Hence, the resistance of the biofilm was found to be 65.6 kW cm À2 and the capacitance of the CF coated with the biofilm of M. extorquens is calculated as 0.058 Fcm À2 .B ased on the electrochemical impedance spectroscopyr esults, the two-compartment configuration was characterized in detail and indicated negligible losses of the system.T oc haracterize the whole electrochemical set-up,a n Ag/AgCl reference electrode was introduced into the cell. The Bode plot of the three-electrode set-up is shown in Figure 8B and the corresponding data is summarized in Ta ble 2. For the complete analysis of the electrochemical impedance spectroscopy,t he IviumSoft developed by Ivium Te chnologies was appliedasthe evaluation program.Based on the fit of the electrochemical impedance spectra, the values listed in Ta ble 2 were extracted. The electrochemical circuits used therefore, are provided in the Figure S8 in combination with the fitting plot. The correspondingN yquist plots are also provided in the Supporting Information ( Figure S9).

Conclusions
We studied the electrocatalysis of biofilm consisting of M. extorquens for the electrochemical reduction of CO 2 to formate.  Table 2. List of impedance data for the two-electrode and three-electrode systems. The formation of the biofilm on the sponge-like carbon-based electrode togetherw ith the continuous H 2 production was observed during the biofilm growth period. As M. extorquens is known as am icroorganism capable of reducing CO 2 to formate as ap art of its metabolism, we studied the utilization of this biofilm for the reductiono fC O 2 to formate by bio-electrochemicalc atalysis, purging the system with CO 2 .T he three microbial electrosynthesis cells were set up as described using CO 2 as the only carbon source.B ya pplying ac onstantp otential of À0.75 Vv ersusA g/AgCl, the chromatographic analysis showedt hat formate was produced from the direct electron injectionp rocess from the carbon-based electrode to the microorganisms. Additionally,t he effects of neutralr ed as ar edox mediator, were also investigated in the performance of the systems in homogeneous as well as heterogeneous way using an electropolymerized neutralr ed film on the electrode prior to the growing of the biofilm. The increasei nt he formate productionr ate and the Faradaic efficiencyo ft he cells having neutralr ed is attributed to the improvement of the electron transfer from the electrode onto the biosystems. The heterogeneous approach using poly(neutral red) providesasimplification of the system as it reduces the contamination and product separationi nt he reactionc hamber.L ong-term studies of up to 17 weeks demonstrated continuous formate formation, showingt he stabilitya nd sustainability of the bio-electrosynthesis systems. Further investigations are underway and we strongly propose these microbial electrosynthesiss ystems for renewable energy storage (chemicale nergy storageu sing CO 2 conversion), CO 2 reduction in general (CCU) and bio(electro)catalysis for other biotechnological conversions.

Experimental Section
Chemicals, materials and methods: Carbon felt electrodes were purchased from SGL Carbon GmbH and aP tw ire was used for the electrical contact. Nafion perfluorinated membrane (Nafion 324) was purchased from Sigma-Aldrich. All chemicals were of analytical grade, purchased from commercial sources and used without further purification. The phosphate buffer (PB) solutions were prepared from appropriate amounts of K 2 HPO 4 and KH 2 PO 4 solution to reach desired pH value.
The glass substrates were cleaning via sonication for 15 mins each in acetone, 2% Hellmanex solution, deionized water and 2-propanol at the last step. Then, the glass substrates were treated in an evaporation chamber for thermal evaporation of 5nmc hromium followed by 80 nm gold as being the Cr/Au substrate for FTIR measurement. At ransparent Cr/Au-coated glass electrode was prepared in as imilar manner with thickness of 3nmc hromium and 15 nm gold for UV/Vis measurement.,t he Cr/Au electrode was used with the thickness of 5nmo fc hromium and 80 nm of gold. For UV/Vis measurement, the transparent Cr/Au electrode was used with the thickness of 3nmo fc hromium and 15 nm of gold.
The absorption spectra were recorded on aP erkinElmer Lambda 1050 UV/Vis/NIR spectrophotometer.
ATR-FTIR spectra of neutral red and poly(neutral red) were recorded on aB ruker Vertex 80-ATR machine over 128 scans. The absorption spectra of neutral red aqueous solution and films were measured at room temperature on aP erkinElmer Lambda 1050 UV/Vis/ NIR spectrophotometer.T he monomer film was prepared by dropcasting aqueous solution of neutral red at the concentration of 0.5 mg mL À1 on atransparent Cr/Au electrode.
SEM images were obtained using JEOL JSM-6360 LV scanning electron microscope at the accelerating voltage of 7.0 kV. Electrochemical polymerization of neutral red: Following the previous report, [55] before use, the CF electrode was electrochemically pre-treated in 0.1 m KNO 3 aqueous solution at potentials between 0a nd 1.0 Vw ith the scan rate of 50 mV s À1 over 20 cycles. After that, the electropolymerization of neutral red was carried out at potentials between À1.0 and 1.0 Vw ith the scan rate of 50 mV s À1 over 20 cycles in 0.1 m phosphate buffer solution pH 6.0 containing 1mm neutral red and 0.1 m KNO 3 ,y ielding ap oly(neutral red) coated CF (PNR/CF). The PNR/CF electrode was electrochemically characterized at potentials between À1.0 and 1.0 Va t different scan rates of 5, 10, 25, 50 and 100 mV s À1 .F or FTIR and UV/Vis measurements, the poly(neutral red) was electropolymerized onto Cr/Au-coated glass electrodes with the similar manner to the CF electrode.

Microbial electrosynthesis studies
Set-up and electrolytes: The inoculation and microbial electrosynthesis experiments were carried out in the two-compartment electrochemical cell of which the anode and cathode compartments were separated by apretreated Nafion membrane, allowing proton transport in the system. The membrane was prepared by soaking a commercial Nafion 324 sheet in a3m HCl solution for 2hfollowed by boiling the membrane in deionized water for an additional 30 min. The three-electrode cell consists of CF or PNR/CF (2.5 6.5 0.6 cm 3 )e lectrode as aw orking electrode, aA g/AgCl ( Inoculation of microorganisms: The formation of biofilm on CF electrodes was done in the aforementioned electrochemical set-up in which the cathodic electrolyte solution contained 10 %( v/v)o f M. extorquens pure culture and afterwards purged with CO 2 for 2h. Ac onstant potential at À0.75 Vw as applied for am onth with weekly CO 2 purging. During this period, 2mLo fh eadspace samples were collected daily and injected into ag as chromatography (GC) for the analysis of headspace products (CH 4 ,C Oa nd H 2 ). After one-month inoculation of the M. extorquens,t he biofilm formation on the electrode could be observed. The medium was removed and replaced one or two times with 0.2 m phosphate buffer solution pH 7.0 in order to get rid of the remaining media that went inside the sponge-like CF electrodes.
Long-term microbial electrosynthesis studies: In this work, 3d ifferent microbial electrosynthesis cells (MEC 1, 2a nd 3) were investigated. For all MECs, the long-term electrolysis was conducted constantly at applied potential of À0.75 Vunder CO 2 -saturated condition. The cathodic chambers were purged weekly with CO 2 for 2h in order to keep the systems saturated with CO 2 ,a nd the electrolyte was replaced with new buffer solutions every 3-4 weeks reported as different running cycles. The experiments were continuously performed for 8, 12 and 17 weeks for MEC 1, MEC 2a nd MEC 3, respectively.I ncase of MEC 2, neutral red solution was added into the cathodic electrolyte solution to achieve af inal concentration of 50 mm.
Product analysis: The headspace samples were taken using ag astight syringe and the product analysis for H 2 and CO was done by using aT hermo Scientific Trace GC Ultra equipped with at hermal conductivity detector (TCD). For the formate production analysis, the liquid samples were diluted with deionized water and their concentrations were determined by aT hermo Scientific Dionex-5000 ion chromatography (IC) system equipped with an IonPac AG19 guard column (2 50 mm), aD ionex AS19 column (2 250 mm) and aD ionex suppressor-conductivity detector by using gradient concentration of KOH as eluent. The Faradaic efficiency (%FE) toward the product is calculated as: %FE ¼ moles of product 1 n Â moles of electron Â 100 in which moles of product are calculated from the amount of H 2 or formate produced in the systems, n is the number of electrons needed for reduction (in both H 2 and formate production cases, n = 2) and moles of electron are calculated by the dividing of number of charges during electrolysis by Faradaic constant as 96 485.33 C·mol À1 .

Electrochemical characterization
Characterization by cyclic voltammetry (CV): The biofilm electrodes were characterized electrochemically after inoculation period, by means of cyclic voltammetry under the N 2 -a nd CO 2 -saturated conditions for all MECs. CVs were recorded in 0.2 m phosphate buffer solution at pH 7.0 (with 50 mm neutral red for MEC 2) at potentials between 0a nd À850 mV vs. Ag/AgCl (3 m KCl) with the scan rate of 1mVs À1 by using aJ aissle Potentiostat-Galvanostat IMP 88 PC.
Characterization by electrochemical impedance spectroscopy (EIS): All impedance experiments were carried out in at wo-compartment electrochemical cell separated with apretreated Nafion membrane. The impedance spectra were recorded in 0.2 m potassium buffer solution pH 7.0 within the frequency range of 10 5 to 10 À2 Hz at the perturbation amplitude of 50 mV.I norder to determine the parameters of the cell, two Pt plates were used as electrodes in the abovementioned set-up as ac ontrol experiment. Next, one Pt electrode was replaced by aC Fa nd used as aw orking electrode. Later on, the CF was replaced by the CF coated with the biofilm. Additionally,aAg/AgCl (3 m KCl) was introduced to the set-up to achieve the three-electrode, in order to characterize the complete electrochemical cell by the impedance spectroscopy.T he vindication for this procedure is explained in more detail in as eparate publication. [61]