Enhanced methane producing microbial electrolysis cells for wastewater treatment using poly(neutral red) and chitosan modified electrodes

Linz Institute for Organic Solar Cells (LIOS) Kepler University Linz, Altenberger Stra hathaichanok.seelajaroen@jku.at K1-MET GmbH, Stahlstrasse 14, 4020 Linz com acib GmbH (Austrian Centre of Industrial Graz, Austria Department of Agrobiotechnology, Insti University of Natural Resources and Life Sc 3430 Tulln an der Donau, Austria † Electronic supplementary informa 10.1039/d0se00770f Cite this: Sustainable Energy Fuels, 2020, 4, 4238


Introduction
The rise in anthropogenic carbon dioxide (CO 2 ) content, mainly due to fossil fuel burning, leads to an increase in global temperature and further to climate change. This has been rst recognized and manually calculated by Svante Arrhenius in his paper back in 1896. 1 Not only the environmental effects, but also the depletion of such fuels is critical. Therefore, cleaner and more sustainable sources of energy are urgently needed. [2][3][4] During the last few decades, bio-electrochemical systems (BESs) have gained a lot of interest. 3 The BES is, by denition, an electrochemical system where at least one of the redox reactions is microbially assisted and the electrode at which such a reaction occurs is called a bioelectrode (bioanode and biocathode). 5 BESs can be implemented in broad applications including electricity generation, 5 wastewater treatment, 6-8 bioremediation 9,10 and production of energy storage chemicals e.g. methane (CH 4 ), 2 formate, 11 and ethanol. 12 Such BESs are operated normally under mild working conditions, show high selectivity towards the desired product, and offer great availability and self-regeneration properties. 5 The rst evidence of electrical interaction between microorganisms, which were later named electroactive microorganisms, and an electrode was reported by Potter et al. in 1911. 13 This extracellular electron transfer 3,4,13 includes two mechanisms: direct and indirect/ mediated electron transfer. 14,15 Direct electron transfer is the inherent interaction between a bacterium, for example Geobacter sulfurreducens, and an electrode through redox proteins such as cytochromes or conductive pili, mentioned as nanowires. [14][15][16][17][18][19] On the other hand, indirect electron transfer describes an interaction between a microorganism and an electrode with the aid of electron shuttles, either endogenous moecules 20 (generated by microorganisms themselves as their secondary metabolites, like avin, 20 phenanzine 21 and quinones 18 ) or exogenous molecules 22,23 (externally added like methylene blue 23 and methyl viologen 24 ).
Bioanodes have been investigated for their capabilities of oxidizing organic matter, like glucose, 25 acetate 26 and more complex substrates like landll leachate 27 and wastewater from various sources. 6 Such oxidation generates electrons, which are transferred to the anode. For example, 24 electrons could be harvested from the oxidation of glucose, as shown in eqn (1) (E 0 0 ¼ À0.41 V vs. standard hydrogen electrode (SHE)). 28 A system, which contains a bioanode and an abiotic cathode, is called a microbial fuel cell (MFC). 4 This technology is considered as a promising wastewater treatment system due to its capability of organic oxidation concomitant with energy harvesting directly in electricity form. 6 C 6 H 12 O 6 + 6H 2 O / 6CO 2 + 24H + + 24e À In conventional MFCs, protons are generated from the oxidation of organic matters and coupled with oxygen, an electron acceptor, generating water at the cathode. To improve the efficiency of the system, introduction of prompted catalysts (like platinum) for useful chemical generation in the cathode chamber has been investigated. 29 Instead of using expensive materials, microbial biocathodes are an alternative potential option. 17,[30][31][32][33] The direct electron uptake from the cathode through microbes has been adopted for several applications, for example, treatment of heavy-metals (bioremediation), 34 reduction of nitrate (denitrication) 35 and reduction of CO 2 to valueadded fuels such as formate 11 and CH 4 (eqn (2), E 0 0 ¼ À0.24 V vs. SHE). 2 Such BESs, where energy is invested to enhance reaction kinetics or overcome thermodynamic energy barriers, are called microbial electrolysis cells (MECs). 5 The electrochemical CO 2 reduction to CH 4 usually suffers from large overpotential (for example, an overpotential of 1.22 V from using Cu electrodes 36 ). Our group has shown the utilization of such microbial cathodes in a MEC for CO 2 reduction to CH 4 with a relatively low overpotential of 0.25 V 37 and it has been reported that CH 4 production was observed in a broad range of production rates. 38 The combination of a bioanode and a biocathode now shall accomplish two tasks simultaneously and utilize the advantages of introducing microorganisms to both electrodes. [39][40][41][42] The coupling of CO 2 reduction with the aforementioned microbial oxidation of organic substances is, in principle, feasible by using electrons, protons and CO 2 which are generated during oxidation on the bioanode. 42 The required voltage of such systems could theoretically be eliminated due to a positive electrochemical cell voltage for CH 4 production with glucose oxidation (E cell ¼ 0.17 V).
A schematic of an MEC system where a bioanode and a biocathode are combined, is demonstrated in Fig. 1a. Such systems have been investigated for various applications e.g. oxygen reduction to hydrogen peroxide, 30 denitrication 35 and CO 2 reduction to CH 4 . 42,43 The MEC approach for CH 4 production offers several advantages, for example the tunability of the degradation/generation rate. 41 Therefore, the MEC could be an effective addition to the existing anaerobic digestion systems by improving biogas quality. 5,44,45 The performance of BESs largely depends on the quality of electroactive biolms on electrodes and also on the electron transfer process. 46,47 Therefore, several attempts have been made to investigate the effects of different electrode materials and modications. 48 Various electrodes including carbon-based materials (e.g. graphite, carbon felt, carbon cloth and carbon mesh) and other inert materials (e.g. stainless steel) are commonly used in BESs. 14,29,46 Further, the surfaces of these materials were modied, aiming to improve microbial growth and electron transfer processes. One example is the modication of carbon-based electrodes with positively charged materials like ammonia, 49 chitosan, 50 or conductive polymers, for example, polyaniline 51 and polypyrrole. 52 Chitosan is a bio-polymer and a derivative of chitin, a component of arthropod exoskeletons. It has been applied in many applications due to its low cost, biocompatibility, non-toxicity and high chemical and thermal stability. 53 These modications resulted in improved bacterial colonization and enhanced electron transfer, which further improved the overall performance of BESs. 54,55 Another interesting material is poly(neutral red). It is a conductive polymer which could be prepared electrochemically in a slightly acidic to neutral pH solution 56 and its monomer, neutral red, is a staining dye which was used in various bio-electrochemical systems as a redox mediator to facilitate electron transfer between the microbe and electrode. [57][58][59] Recently, we reported on the contribution of poly(neutral red) modied carbon felt cathodes, for the enhancement of microbial electrochemical reduction of CO 2 to formate by Methylobacterium extorquens. The study showed an improvement in electron transfer processes leading to higher formate production rates, as compared to non-treated electrodes. 60 However, to the best of our knowledge, each of the chitosan and poly(neutral red) electrode modications has not been investigated for the anode and cathode at the same time in a MEC. Therefore, this idea has attracted our interest in order to investigate the effect of modied electrodes not only on one redox reaction but also on both redox processes simultaneously.
In this study, we report a full assembly of an MEC (Fig. 1b), equipped with bioelectrodes, for two applications: wastewater treatment at the bioanode and CO 2 reduction to CH 4 at the biocathode. The long-term performance of the methaneproducing MEC for synthetic wastewater treatment was monitored. The system was evolved by using bioelectrodes (bioanode and biocathode) inoculated with the same mixed culture microorganisms. The experiments were performed at a controlled anode potential of 0.40 V vs. Ag/AgCl (3 M NaCl) and the oxidation of synthetic wastewater was observed by monitoring the change of the anolyte chemical oxygen demand (COD) value, which reected directly the efficiency of wastewater treatment. Moreover, electromethanogenesis at the biocathode was tracked by CH 4 generation in the headspace. Furthermore, carbon felt electrodes were modied with chitosan and poly(neutral red), serving as supports for microorganisms for both the cathode and anode. These two materials were chosen due to the previously reported results, low material costs, good biocompatibility and facile synthesis. 52,55 The performance of all three MECs namely MEC 1 (equipped with non-modied carbon felt electrodes), MEC 2 (equipped with poly(neutral red) modied carbon felt electrodes) and MEC 3 (equipped with chitosan modied carbon felt electrodes) were compared, reecting the effect of electrode modications.

Results and discussion
Two different carbon felt electrode modications were investigated by using poly(neutral red) and chitosan. The poly(neutral red) modied carbon felts were prepared by electropolymerization of neutral red, as reported in our previous work. 60 The chitosan modied electrodes were prepared chemically through carbodiimide chemistry (Scheme 1). Firstly, the carbon felt surface was oxidized under strong acid conditions of concentrated HNO 3 . The resulting carboxylic acid groups were later activated through coupling reactions of N-hydroxy-3-(3dimethylaminopropyl)-N 0 -ethylcarbodiimide (EDC) and Nhydroxysuccinimide (NHS) prior to reacting with primary amine groups of chitosan.
To investigate the long-term performance of MECs, three MECs namely MEC 1, 2 and 3 were built. In MEC 1, non-modied carbon felts were used as the anode and cathode, while MEC 2 and 3 were equipped with poly(neutral red) and chitosan modied electrodes, respectively.
All bioelectrodes in each MEC were inoculated with a mixed culture taken from sewage sludge in a two-compartment electrochemical cell with a controlled potential of 0.40 V vs. Ag/AgCl (3 M NaCl). During the adaptation phase, the bioanode was fed once a week with synthetic wastewater consisting of organic substances like glucose and acetate, while the biocathode was fed with glucose and CO 2 to ensure enough biomass formation. Aer a 4 week adaptation, the faradaic currents were observed at around 2 mA, indicating successfully developed bioelectrodes. 61 Aer the adaptation period, the cathodic solution was replaced with a glucose-free medium. All three developed MECs showed capabilities of organic oxidation in the anodic compartment and simultaneous CH 4 generation in the cathodic compartment. To investigate the systems' performance, longterm electrolyses of MEC 1, 2 and 3 were carried out, by applying a constant potential of 0.40 V vs. Ag/AgCl (3 M NaCl) at the anode in a batch operation mode. During the operation of around 90 days, half of the anodic solution was replaced twice a week with a freshly prepared medium, consisting of synthetic wastewater with an average COD concentration of 600 mg L À1 , while cathodic chambers were purged with CO 2 two times a week. The systems were monitored for 3 running cycles, through analysis of organic degradation by COD determination of anodic solutions and headspace analysis of CH 4 production in cathodic chambers.
The accumulated COD removal in each running cycle (red circle data point) was plotted together with total electrical charge (Q) (blue triangle data point) over the entire running time, as shown in Fig. 2a-c for MEC 1, 2 and 3, respectively. The data were tted linearly as presented in dashed lines, showing reaction rates in each cycle. The total COD removal, COD removal rate, average COD removal efficiency, total Q and Q rate are summarized in Table 1.
The results revealed that accumulated COD removal in MEC 1 increased from 1.0 g L À1 in the rst cycle to 1.8 and 2.2 g L À1 in the second and third cycles, respectively, showing an enhancement of the oxidation process over the running cycles. Signicantly higher COD removal values were observed at 1.6 and 1.9 g L À1 in the rst cycle of MEC 2 and 3. In the second cycle, the accumulated COD values found in MEC 2 and 3 were enhanced by 1.0 and 0.5 g L À1 from the rst cycle, respectively, while a slight drop was detected in the third cycle. The removal rates obtained from the slope of tting lines, were found to be in the range of 40-85 mg per L per day. The results revealed that the removal rate increased largely from 40 mg per L per day in the rst cycle to 70 mg per L per day in the third cycle of MEC 1, while the highest removal rates were observed in MEC 2 and 3 with 85 and 80 mg per L per day in cycle 2, respectively. The large increase in COD removal from the rst to the second cycle in all MECs might be related to biolm growth in the rst cycle. Over the whole experiment, average COD removal rates were observed at 57, 73 and 72 mg per L per day in MEC 1, 2 and 3, respectively. Additionally, average COD removal efficiencies were calculated giving the overall efficiencies of 44%, 64% and 67% in MEC 1, 2 and 3, respectively. These results reected that enhanced organic degradation strongly related to both modications.
Furthermore, the electrical charge plots revealed that during all three cycles, electrons were consumed continuously in all MECs. In MEC 1, accumulated charges increased from 2.0 Â 10 3 C in the rst cycle to 3.4 Â 10 3 and 3.7 Â 10 3 C in the second and third cycles, respectively. The accumulated charges collected in the rst cycle of MEC 2 and 3 were found to be almost two times higher than that of MEC 1, suggesting an enhanced electron transfer process at the rst cycle in the modied electrode containing systems. This observation might relate to better coverage of biolms on modied carbon felts. The accumulated charges were further improved in the second cycle and dropped slightly in the third cycle of MEC 2. While those of MEC 3 were of 3.5 Â 10 3 and 4.2 Â 10 3 C in cycle 2 and 3, respectively. Average rates for electron ux were observed for MEC 1, 2 and 3 at 100, 133 and 127 C per day, respectively. The highest electron ux was observed in MEC 2 with 150 C per day in cycle 2, followed by MEC 3 with 140 C per day in cycle 3. Although MEC 1 showed a much lower Q rate in the rst cycle relative to MEC 2 and 3, it reached the rate of around 100 C per  day which is in the same range as MEC 3 in the second cycle. This observation suggested that biolm formation on bare carbon felts took longer, as compared to treated carbon felts.
In the cathodic compartment, the CH 4 production was quantied twice a week. Fig. 2d-f show the plots of accumulative CH 4 concentration (black square data point) and accumulative Q (blue triangle data point) during each running cycle of MEC 1, 2 and 3. The CH 4 production rate and the average faradaic efficiencies are summarized in Table 1. The produced CH 4 observed in MEC 1 rose continuously from 0.8 mmol in the rst cycle to 1.3 mmol in the second cycle to nally 1.8 mmol in the third cycle. In MEC 2, the amount of produced CH 4 in cycles 1 and 2 was relatively stable at 2.6 and 2.7 mmol, respectively. However, the production declined to 2.2 mmol in the third cycle, while produced CH 4 in MEC 3 increased from 2.8 mmol in cycle 1 to 3.3 mmol in cycle 2 and dropped signicantly to 2.0 mmol in cycle 3. The decline observed in MEC 2 and 3 in the third cycle, might result from the instability of the modied electrodes and/or detachment of biolms. The CH 4 production rates in the three MECs were found to be in the range of 25-110 mmol per day and the average production rates over the entire experiments in MEC 1, 2 and 3 were 42, 80 and 87 mmol per day, respectively. Together with the obtained charges, faradaic efficiencies towards the CO 2 reduction to CH 4 were calculated according to the equation given in the experimental part. The corresponding faradaic efficiencies were averaged within each running cycle. MEC 1 showed efficiencies of 39% in cycle 1, 29% in cycle 2 and 38% in cycle 3, and an overall average efficiency of 35% while the highest faradaic efficiency in MEC 3 was reached at 72% in cycle 2 and then decreased to 36% in cycle 3. The overall average efficiencies of MEC 2 and 3 were reported to be 52 and 55%, respectively.
Considering CH 4 production relative to the removed COD (CH 4 yield), MEC 1 revealed a relatively stable CH 4 yield at 0.09 L CH 4 /g COD while the yield observed from MEC 2 and 3 decreased from the rst cycle to the third cycle and the average yields were of 0.13 and 0.14 L CH 4 /g COD , respectively.
Concerning the performance in both cathodic and anodic chambers, MEC 2 and MEC 3 showed signicantly higher organic degradation efficiencies, faradaic efficiencies toward the conversion of CO 2 to CH 4 and CH 4 yield, as compared to those of MEC 1. Moreover, higher numbers of electrons were delivered in the modied electrode containing systems. This might be due to biolm coverage and/or facilitated electron transfer by the coating. As suggested in the previous report, a positively charged electrode modication with chitosan coating, could enhance the interaction between the electrode and Gram-negative microorganisms like Sporomusa ovata, resulting in improved microbial electrolysis rates. 54 However, the instability of the modied systems for CO 2 reduction to CH 4 was observed during the long-term running.
Compared to the reported studies on methane-producing MECs using non-treated carbon felt electrodes (Table 2), our results (MEC 1) showed lower COD removal efficiency but signicantly higher CH 4 yield. While, with the electrode modi-cations (MEC 2 and 3), the higher efficiencies were now comparable to those of the continuous-operating MECs equipped with graphite granules and nickel mesh. Furthermore, using carbon felt offers several advantages as it is highly exible, has high active surface area, high conductivity and is less corrosive. 62 However, it is important to be noted that apart from electrode material, other parameters like cell conguration, inoculation, substrate and operation parameters are of importance. 63 Therefore, this comparison was aimed to show the general view of this eld and parameters which could be further optimized.
Aer the long-term operation of around 90 days, samples of bioanodes and biocathodes of all MECs were dried under ambient conditions overnight for SEM measurements. Fig. 3 presents SEM images of bioanodes and biocathodes of MEC 1, 2 and 3. The bioanodes obtained from modied carbon felt electrodes showed better coverage and thicker biolms (Fig. 3b and c). High magnication images (Fig. 3d-f), from both modied and non-modied carbon felt electrodes are colonized from different cell shape bacteria, revealing mixed microbial strains. In the SEM images of all biocathodes, rod-shaped cells of bacteria were found to be located on carbon bers (Fig. 3j-l). This characteristic shape refers to some species from phylum Firmicutes like S. ovata 64 and the Euryarchaeota Methanobacterium palustre, 2 which are capable of growing autotrophically by using a cathode as the sole electron donor and CO 2 as the carbon source.
The utilization of electrochemical impedance spectroscopy (EIS) in BESs is highly signicant since extensive information of the systems can be extracted, such as the charge transfer resistances and the mechanism of the electron transfer, among many others. In this present study, the electrical loss in the system was evaluated. All impedance spectra were recorded in the frequency range of 10 À1 to 10 5 Hz with a perturbation amplitude of 50 mV. Firstly, the resistance of the medium as an electrolyte solution (R sol ) was determined in a one-compartment cell, having platinum electrodes as the working (WE) and counter (CE) electrodes, showing 8.9 U. The platinum electrodes were then transferred to a two-compartment electrochemical cell containing a medium to determine the resistance of a Naon membrane (R NF ). The electrical circuit used for data tting is shown in Fig. 4. In this conguration, R NF was found to be 530 U and connected in series with R sol , the resistance of carbon felt (R CF ) and the resistance of biolm (R biolm ). R CF is in parallel with the constant phase element of carbon felt (CPE CF ) and R biolm is in parallel with the capacitance of the biolm (C biolm ), representing the bioanode.
Then, the platinum electrodes were replaced with bare carbon felt electrodes and the impedance spectrum was measured in a similar manner. Aer that, these carbon felt electrodes were used as electrodes for biolm formation in MEC 1. With these two bioelectrodes, impedance spectra were recorded to investigate changes of the system in the presence of biolms. Fig. 5 presents Bode plots for the two-electrode setup of platinum, bare carbon felt and carbon felt with the biolm, and the data evaluated from the measurements are summarized in Table 3.
The constant phase element was used for the description of the non-ideal capacity of spongy-like carbon felt electrodes. Thus, the resistances of electrodes (R CF ) as bare carbon felt and carbon felt with a biolm were found to be 1.1 and 0.4 kU, respectively. The calculated resistance and capacitance of the biolm (R biolm , C biolm ) were 0.4 kU and 0.1 F. Further, the results reveal that, with the biolm on electrodes, a decrease in capacitive current was observed in a low frequency regime due to enhanced electron transport, indicating the establishment of a conductive biolm and a larger surface area due to biolm formation. 61 Moreover, EIS results indicated negligible losses of this MEC.
Apart from the observed enhanced performance of the modied electrodes containing MECs, the economic aspects should be addressed as they impact signicantly on up-scaling. However, only a few studies have been reported on the cost of electrode modication for this type of MEC. We performed cost studies for such modications from the price of chemicals which were used in each modication process (see details in the ESI †). The calculation showed that poly(neutral red) modication ($100 V per m 2 ) was much cheaper, as compared to chitosan modication ($5400 V per m 2 ) due to the costly coupling reagents (NHS and EDC). Compared to the reported metal based electrodes like Ni mesh 66 ($8600 V per m 2 ), carbon felt ($170 V per m 2 ) modied with poly(neutral red) (in total $270 V per m 2 ) are much cheaper and showed similar results with respect to the CH 4 yield (Table 2). Moreover, this poly(neutral red) modied carbon felt is more cost effective than the conventional platinum based electrodes (e.g. Pt (60%) carbon cloth is $4100 V per m 2 , 62 ) which are normally used and, being the main cost of MFC technology made it less feasible for upscaling reactors.

Chemicals
Low molecular weight chitosan and 3-amino-7-dimethylamino-2-methylphenazine hydrochloride (neutral red) were purchased from Sigma-Aldrich. Carbon felt electrodes were received from Alfa Aesar. Other chemicals were of analytical grade and were used as received. Phosphate buffer solutions were prepared from KH 2 PO 4 and K 2 HPO 4 to achieve the desired pH value. A Naon peruorinated membrane (Naon 117) was received from Chemours. With a slight modication to the reported study, 69 the Naon membrane was pre-treated prior to use by boiling in H 2 O 2 (30% v/v), deionized water, 0.5 M H 2 SO 4 and   deionized water, sequentially. Each step was carried out for 1 h at 80 C.

Setup
All microbial electrocatalytic experiments were performed in two-compartment electrochemical cells with the same geometrical design (Fig. 1b). Each chamber had a volume of around 250 mL. The cathode and anode chambers were separated by using a pre-treated Naon 117 membrane allowing proton transport between the two chambers. Carbon felt electrodes (

Electrode modication
Preparation of the poly(neutral red) modied carbon felt. As described in the previous study, 60 the modication was done in a one-compartment cell. A carbon felt, a Ag/AgCl (3 M NaCl) and a Pt plate were used as working, reference and counter electrodes, respectively. Before electropolymerization, the carbon felt electrode was pre-treated by sweeping the potentials between 0 and 1.0 V with a scan rate of 50 mV s À1 over 20 cycles in a 0.1 M KNO 3 aqueous solution. Aer rinsing with deionized water, electropolymerization was carried out on the treated carbon felt electrode at potentials between À1.0 and 1.0 V with a scan rate of 50 mV s À1 for 20 cycles in 0.1 M phosphate buffer solution (pH 6.0), containing 1 mM neutral red and 0.1 M KNO 3 .
Preparation of the chitosan modied carbon felt. With slight modication from the reported study, 54 a carbon felt electrode was pre-treated with concentrated HNO 3 at room temperature for 14 h. Aer that, the electrode was washed with deionized water until the rinsed water was neutral and dried under vacuum for 14 h. To covalently modify with chitosan, the oxidized carbon felt was immersed in a coupling reagent of ethanol and water mixture (4 : 1 v/v) containing 50 mM Nhydroxy-3-(3-dimethylaminopropyl)-N 0 -ethylcarbodiimide (EDC) and 50 mM of N-hydroxysuccinimide (NHS) at room temperature for 2 h. The electrode was transferred into a 2% acetic acid solution containing chitosan (1% w/v) of which the pH was adjusted to 6.5 with a 1 M NaOH solution. The modi-cation took place at room temperature for 14 h, then the modied carbon felt electrode was carefully washed with ethanol and dried under vacuum.

Adaptation phase
For all MECs, anodes and cathodes were inoculated with the same inoculum prepared by centrifugation (4000 rpm for 10 min) of sewage sludge, collected from a communal wastewater treatment plant (Austria). The bioelectrodes were developed in the aforementioned two-compartment cell in which each chamber contained 20 mL of prepared inoculum. In cathodic chambers, 200 mL of the previously described medium and glucose (0.22 g) were added. While anodic chambers were lled with 200 mL of medium, containing additional carbon sources, to simulate municipal wastewater which included the following composition (per L): peptone (0.138 g), yeast extract (0.075 g), sodium acetate (0.088 g) and glucose (0.336 g). 43 The adaption phase was carried out at room temperature by applying a constant potential of 0.40 V vs. Ag/AgCl (3 M NaCl) at the anode for 4 weeks. In this phase, carbon sources were supplied once a week. For the anode compartment, half of the solution was replaced with a freshly prepared medium containing synthetic wastewater, while glucose (0.22 g) was added into the cathodic compartment without replacing the solution. Both chambers were kept saturated with CO 2 to ensure anaerobic conditions. Aer 4 weeks, 200 mL of the cathodic solution was replaced with a freshly prepared medium without glucose addition.

Long-term performance studies
Aer the 4 week adaptation, all three MECs were studied continuously by applying a constant potential of 0.40 V vs. Ag/ AgCl (3 M NaCl) at the anode in a batch operation mode. Headspace samples from cathode chambers and liquid samples from the anolyte were taken twice a week for headspace product analysis and COD measurements, respectively. Aer sampling, the cathode chambers were purged with CO 2 to keep the chamber saturated, while half of the anodic solution was replaced with fresh medium containing synthetic wastewater with an average COD concentration of 600 mg L À1 and the chamber was ushed with N 2 to achieve anaerobic conditions. Every 4 weeks, 200 mL of cathodic solution was replaced with fresh medium to provide microorganisms with sufficient trace elements and vitamins, labelled as running cycles 1-3.

Determination of COD
The COD of liquid samples taken from the anodic solution were analyzed photometrically. Firstly, 2 mL of the liquid sample was mixed with the reagent in a COD test tube (Nanocolor COD 160 and 1500, Macherey-Nagel). Then, the mixture was heated at 148 C for 2 h in a thermostat (Nanocolor Vario Mini, Macherey-Nagel). Aerwards, the tubes were allowed to cool down to room temperature. Then COD was determined by using a LED photometer (Compact Photometer PF-3, Machery-Nagel) at wavelength 620 nm. COD removal efficiency can be calculated using the following calculation: in which DCOD is calculated from the change of COD 1 (COD concentration of anodic solution before reaction) to COD 2 (COD concentration of anodic solution aer reaction).

Headspace product analysis
Headspace samples from cathode chambers were taken by using a gas-tight syringe for CH 4 analysis. The analysis was done by using a gas chromatograph (Thermo Scientic GC ultra) equipped with a thermal conductivity detector (TCD). The faradaic efficiency (FE) can be calculated using the following equation: where n CH 4 is the number of moles of CH 4 calculated from the amount of CH 4 produced in the system, 8 is the number of electrons needed for the reduction of CO 2 to CH 4 , F is the faradaic constant (96 485.33 C mol À1 ), I is current recorded during the reaction and t is time of the reaction.

Scanning electron microscopy measurements
The samples were taken by cutting the bottom edge of each bioelectrode (size of 1.5 Â 0.5 cm 2 ) and allowing them to dry under ambient conditions overnight. Scanning electron microscope (SEM) images of dried bioelectrodes were taken using the JEOL JSM-6360 LV scanning electron microscope at the accelerating voltage of an electron beam of 7.0 kV with a working distance of 15 mm.

Electrochemical impedance
The impedance experiments were carried out in a twocompartment electrochemical cell separated with a pre-treated Naon 117 membrane (as mentioned above for microbial electrocatalytic experiments) by using an IVIUM CompactStat instrument. The impedance spectra were recorded within the frequency range of 10 5 to 10 À1 Hz at the perturbation amplitude of 50 mV.

Conclusion
In this study, we investigated the long-term performance of CH 4 producing MECs equipped with bioanodes and biocathodes for wastewater treatment, in terms of COD removal and CH 4 production. By applying a constant potential of 0.40 V vs. Ag/ AgCl (3 M NaCl) at the bioanode, organic degradation and CH 4 generation were monitored through COD measurement of the anodic electrolyte solution and cathodic headspace analysis, respectively. In addition, carbon felt electrodes were modied with poly(neutral red) and chitosan, serving as electrodes and supports for microbial colonization in MECs. The results of all three MECs revealed the possibility of oxidation of organic substances at bioanodes and simultaneous CH 4 generation at biocathodes. Compared to non-modied MECs, higher COD removal rates and CH 4 production rates of modied electrode containing systems were observed, especially in the rst running cycle. These revealed that modied electrodes with poly(neutral red) and chitosan could enhance biolm formation and activity. Furthermore, the corresponding efficiencies from those were improved, suggesting efficient electron transfer processes between electrodes and microorganisms. Compared to the previously reported MECs, our MECs showed comparable results even though they were batch operated which limited the substrate input. Further, the cost studies on the electrode material and modication prices suggested that the poly(neutral red) modied electrode is more cost-effective, as compared to other metal based materials, which provides economic feasibility for large-scale operation.

Conflicts of interest
There are no conicts to declare.