Bio-electrochemical characterization of air-cathode Microbial Fuel Cells with microporous polyethylene/silica membrane as separator

Bioelectrochemistry 106 (2015) 115–124

Contents lists available at ScienceDirect

Bioelectrochemistry journal homepage: www.elsevier.com/locate/bioelechem

Bio-electrochemical characterization of air-cathode microbial fuel cells with microporous polyethylene/silica membrane as separator Nina Kircheva a,1, Jonathan Outin a, Gérard Perrier a, Julien Ramousse a, Gérard Merlin a,⁎, Emilie Lyautey b a b

LOCIE UMR CNRS 5271, Université de Savoie, 73376 Le Bourget du Lac, France CARRTEL, Université de Savoie, F-73000 Chambéry, France

a r t i c l e

i n f o

Article history: Received 22 October 2014 Received in revised form 20 May 2015 Accepted 22 May 2015 Available online 3 June 2015 Keywords: Electrochemical impedance spectroscopy Reticulated carbon foam Polyethylene membrane Nonionic membrane

a b s t r a c t The aim of this work was to study the behavior over time of a separator made of a low-cost and non-selective microporous polyethylene membrane (RhinoHide®) in an air-cathode microbial fuel cell with a reticulated vitreous carbon foam bioanode. Performances of the microporous polyethylene membrane (RhinoHide®) were compared with Nafion®-117 as a cationic exchange membrane. A non-parametric test (Mann–Whitney) done on the different sets of coulombic or energy efficiency data showed no significant difference between the two types of tested membrane (p b 0.05). Volumetric power densities were ranging from 30 to 90 W·m−3 of RVC foam for both membranes. Similar amounts of biomass were observed on both sides of the polyethylene membrane illustrating bacterial permeability of this type of separator. A monospecific denitrifying population on cathodic side of RhinoHide® membrane has been identified. Electrochemical impedance spectroscopy (EIS) was used at OCV conditions to characterize electrochemical behavior of MFCs by equivalent electrical circuit fitted on both Nyquist and Bode plots. Resistances and pseudo-capacitances from EIS analyses do not differ in such a way that the nature of the membrane could be considered as responsible. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The research and development of microbial fuel cells (MFCs) have become important parts of the "green" energy domain, due to their double functions — water purification and energy production. MFCs are devices that use bacteria as catalysts to oxidize organic and inorganic matter and to generate electrical current. The main components of the MFCs are the anode and the cathode with a separator as an optional component to prevent oxygen and substrate crossover [1]. At the cathode, electrons and protons react with oxygen with the help of catalysts such as platinum, and form water. In a MFC bacteria are the anodic catalysts for oxidation of an electron donor, often glucose or acetate in laboratory scale studies. A lot of designs and electrode materials have been tested in order to understand the involved phenomena and to improve efficiency of both organic charge removal in wastewater and electricity generation [1,2]. The generation of electrical energy by MFC systems is significantly limited by the properties of individual components such as electrodes, electrolyte and membrane. A membrane acts as a proton conductor with a certain electrical resistance; therefore, the membrane should be analyzed carefully [3]. In order to achieve high voltage and power density; it is essential to optimize factors ⁎ Corresponding author. E-mail address: [email protected] (G. Merlin). 1 Present address: LEM SA, 8 Chemin des Aulx, PO Box 35, CH 1228 Plan Les Ouates, Geneve, Switzerland.

http://dx.doi.org/10.1016/j.bioelechem.2015.05.016 1567-5394/© 2015 Elsevier B.V. All rights reserved.

influencing the internal resistance of MFCs. For example electrodes with high surface area and low resistance cation-exchange membranes could greatly improve the power density [4]. Separating membranes, that are necessary for preventing oxygen to reach the anode as well as substrate crossover, can be the source for power limitations coming from pH gradients [5,6], from biological and chemical fouling [6] or from mechanical deformations [7]. Cation (CEM) or Anion (AEM) Exchange Membranes are the most used in MFCs, with better efficiencies claimed for CEMs [7–9]. But electrochemical reactions on the cathode can be significantly limited due to high concentrations of cations that inhibit the migration of protons through the sulfonated membranes [5,10]. Moreover Ion Exchange Membranes (IEMs) are expensive [10]. To overcome these problems, various materials have been studied, including glass fiber, J-cloth and various polymers [11–16]. Such separators generally showed a higher proton migration capability and a better applicability than IEMs. However, these materials should be considered in terms of sustainability and longevity before being used for field applications [17]. Large-scale production polyethylene separators are of common use in lead–acid batteries. Such membranes can be mechanically reinforced by a glass fiber mat and are especially surface-designed for acid stratification [18]. Burkitt and Yu [19] recently included such a separator in their comparative study of AEM and CEM in MFCs. The aim of this work was to study the behavior over time of a separator made of a non-selective microporous membrane in an air-cathode MFC with a bioanode in reticulated vitreous carbon (RVC) foam. The

116

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

behavior was characterized by monitoring the degradation of the substrate (acetate), by chronoamperometry with calculations of coulombic and energy yields and by electrochemical impedance spectrometry (EIS). Performances of the microporous polyethylene (PE) membrane were compared with Nafion®-117 as a cationic exchange membrane in a similar MFC. EIS coupled with other electrochemical and biochemical measurements helps to provide a better understanding of the different limiting factors in MFCs and allows optimizing design and operation of MFCs for power production [20]. In order to better understand biofouling on membranes, bacterial and archeal community analyses were carried out. 2. Materials and methods

layer exposed to air. A stainless steel mesh maintained the assembly and served as a current collector, connected by a crocodile clip to the external circuit. A rubber seal ensured waterproofness. Two different membranes have been studied, a 0.18 mm-Nafion® N117 (Ion Power) and a 0.60 mm RhinoHide® polyethylene (ENTEK), both 40 × 40 mm2. Both membranes were treated before use, aiming at eliminating any substance that could interfere and impair ion transport, as well as to saturate it with positive ions. For this purpose the Nafion® cationic membrane requires an immersion in a H2SO4 1 M solution for 1 h at 110 °C, followed by a thorough rinsing out and immersion in a H2O2 3 wt.% solution for 1 h at 110 °C. Finally, the membrane was rinsed again and stored in distilled water. For its part, the microporous polyethylene membrane was immersed in absolute ethanol for 16 h at ambient temperature followed by rinsing in distilled water.

2.1. MFC construction 2.2. MFC operation The cell's external dimensions were 60 × 60 × 20 mm3, giving a working liquid volume of 25 cm3. The anode was built using a small stainless steel nut and carbon cloth pieces which sandwich the Carbon Foam (CF) piece (20 × 20 × 5 mm3, 3750 m−2 m−3, 24 pores per cm, Goodfellow). The cathode was a specific multi-layered air-cathode based on a carbon-supported catalyst, supplied by Paxitech (Grenoble, France), (Scheme 1). The electrolyte-exposed active face (40 × 40 mm2) was a 0.5 mg·cm−2 Pt-catalyzed carbon powder bound by PTFE in proportion 70/30 w/w. The other side was a 70 μm-PTFE gas diffusion

Two experiments were conducted in time with similar conditions except for acetate concentration and number of replicates. In the first experiment two reactors were operated in parallel, each with one type of membrane; 50 mM acetate and a duration of 123 days. For the second experiment two additional reactors were run in parallel giving two reactors for each type of membrane; 10 mM acetate and a duration of 60 days. The values of the state variables of the bioelectrochemical system (pressure, temperature, conductivity, pH) and inoculum corresponded to stable

Scheme 1. Schematic representation of the air-cathode microbial fuel cell used in this study.

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

117

operating conditions [21]. The MFCs were placed in a controlled climate chamber at 25 °C. The anode chamber was inoculated with a suspension of electroactive microorganisms from a working MFC initially enriched with activated sludge from a local wastewater treatment plant (Le Bourget du Lac, France). Ten milliliters of the inoculum was diluted in 15 mL of a buffered nutrient solution. The buffered nutrient solution was composed of Na2HPO4 4.05 g, K2HPO4 2.9 g, and NH4Cl 0.1 g per liter of deionized water, with two acetate concentrations of 50 mM and 10 mM. Conductivity of buffered nutrient solution was in the range of 3–6 mS·cm−1. Yeast extract (0.05 g/L), vitamins and traces of minerals were added from stock solution. The medium was fully renewed upon the appearance of a voltage drop as a sign of the full consumption of the substrate. The period between two feedings corresponds to a batch cycle and was ranging from three to fifteen days depending on acetate concentration. The reactors were operated with a fixed 1 kΩ external load.

Table 1 Membranes physical properties. Data from ion-power.com; http://static.entek.com/SEP-DESIGN-CONSIDERATIONS-FOREFB-MICRO-HEV-APPS.pdf; [3].

2.3. Electrochemical characterization and chemical analysis

Bacterial and archeal community diversity analyses were carried out on biofilms collected from both MFC membranes. For the Nafion-MFC, the biofilm from the anodic side (Na) was collected and analyzed. No sample was collected from the cathodic side, as no biofilm development was observed. For the RhinoHide® MFC, two biofilm sub-samples were characterized: one from the cathodic side of the membrane (RHc) and one from the anodic side of the membrane (RHa). For each sample, DNA extraction was carried out using DNeasy Plant Mini Kit (Qiagen) according to the manufacturer's protocol. Bacterial diversity was assessed by cloning and sequencing 16S rRNA gene libraries. A ~ 1450 pb fragment of the 16S rRNA gene was amplified as follows; the final reaction mix (25 μL) consisted of 1 × PCR buffer (Promega), 1.5 mM MgCl2, 0.3 mg mL−1 bovine serum albumin, a 200 μM concentration of deoxynucleoside triphosphate (Eurogentec), a 0.5 μM concentration (each) of 27F and 1492R primers (Eurogentec), 1.25 U of Taq polymerase (Promega), and 5 μL of DNA extract as the template. Amplification was performed with a Veriti thermal cycler instrument (Applied Biosystems) under the following conditions: after an initial denaturation at 95 °C for 4 min, 35 cycles of denaturation (94 °C, 1 min), annealing (55 °C, 1 min), and extension (72 °C, 2 min) were performed, followed by a final extension (72 °C, 15 min). For Archaea, a ~ 660 pb fragment of the 16S rRNA gene was amplified as follows; the final reaction mix (25 μL) consisted of 1 × PCR buffer (Promega), 1.5 mM MgCl2, 0.4 mg mL− 1 bovine serum albumin, a 800 μM concentration of deoxynucleoside triphosphate (Eurogentec), a 0.5 μM concentration (each) of 340F and 1000R primers (Eurogentec), 0.8 U of Taq polymerase (Promega), and 5 μL of DNA extract as the template. Amplification was performed with a Veriti thermal cycler instrument (Applied Biosystems) under the following conditions: after an initial denaturation at 98 °C for 2 min, 30 cycles of denaturation (95 °C, 30 s), annealing (60 °C, 30 s), and extension (72 °C, 1.5 min) were performed, followed by a final extension (72 °C, 7 min) Sequencing of the cloned bacterial and archeal 16S rRNA gene products was carried out by Macrogen (The Netherlands), using primer SP6. Following sequence alignment using MEGA-6 software [23], sequences sharing more than 97% of similarities were considered as being members of the same OUT (operational taxonomic unit). Sequence analysis and phylogenetic tree construction were carried out using the Ribosomal Database Project, release 11, update 3 [24,25]. Sequences were aligned using the RDP aligner, and the phylogenetic tree was constructed using the Tree Builder tool and imported into the online UniFrac interface [26,27] to specifically test for differences in diversity between the two MFCs based on phylogenetic relationships. Clone abundances were used to calculate the Shannon–Wiener diversity index and similarity between the different samples was calculated using the Jaccard similarity index. A total of thirteen partial bacterial and three partial archaeal 16S rRNA gene sequences have been deposited in the GenBank sequence database under accession numbers KP003904 to KP003919.

The chamber frame was connected to the earth. Cell voltage was collected every 1 h through an Agilent 34970A Data Logger connected to a computer. EIS and cyclic voltammetry measurements were performed according the method described in previous studies [21] with a PGSAT 128N potentiostat coupled to a frequency response analyzer FRA2 using a NOVA 1.7 software (Metrohm-Autolab). 50 Hz and harmonics were skipped to avoid disturbance in EIS measurements. Impedance spectra were fitted with ZView2 software (Scribner Associates). Acetate concentrations and other Volatile Fatty Acids (VFA) metabolites were determined using a gas phase chromatograph (Perkin Elmer Autosystem XL) connected to a FID detector. The different acids were separated in a Perkin Elmer specific column (Elite FFAP, length 15 m, inner diameter 0.53 mm, Nitroterephthalic Acid Modified Polyethylene Glycol film thickness 1 μm). During the analysis, the detector and injector were put at respectively 275 °C and 250 °C. The splitless injection mode is undertaken with a carrier gas flow of 7.5 mL·min−1. The column heating sequence was a 2 min-stay at 80 °C, a 8 °C·min−1-ramp up to 120 °C followed by a 10 min-stay, a cleaning 45 °C·min−1-ramp up to 170 °C followed by a final 2 min-stay. For biomass determination on membranes or RVC, the total protein content was determined by the colorimetric bicinchoninic acid (BCA) assay [22] (Bicinchoninic Acid Kit for Protein Determination, Sigma, France) using bovine serum albumin (BSA) as standard and expressed in μg·cm−2 or g−1 media BSA equivalent. 2.4. Membrane properties The low-cost uncharged membrane tested was a microporous polyethylene (PE)/silica material currently used as a separator in lead–acid batteries (RhinoHide®). This separator was composed of ultra-high molecular weight polyethylene fibrils plus silica aggregates with a SiO2/PE ratio ranging from 2.1 to 2.9. The thickness was 0.150 mm. On one side a 0.4 mm fiberglass retainer mat was present giving an overall thickness of 0.6 mm. The charged and non-porous membrane used as reference was a proton exchange membrane (Nafion®-117). Main physicochemical properties are summarized in Table 1. Pore sizes are very different between the two types of membrane. A membrane having a pore size less than 1 nm is typically considered as a non-porous membrane. For RhinoHide®, about 10 μL/g (1.7% of pore volume) pores have a diameter higher than 1 μm (diameter of bacteria). For water captivity (water content) corresponding to the interaction of water with the sulfonate groups of the polymer chains [3], the value is only 16.3% for Nafion®-117. With RhinoHide®, water captivity is only the result of water filling pores and the value corresponding to the porosity is 55%. Electrical resistivity is lower for the uncharged membrane than for Nafion®-117. Measured values for ion exchange capacity and transport number show cation selectivity of Nafion®-117 (Table 1).

Nafion®-117

RhinoHide®

Thickness Pore-size

0.183 mm 1–4 × 10−9 m

kO2 Water captivity Electrical resistance Electrical resistivity Transport number Ion exchange capacity

0.49 × 10−3 cm/s 16.3% 940 mΩ·cm2 12,048 mΩ·cm 0.99 0.93 meq/g

0.150 mm [2 × 10−6 m−2–10−8 m] (d50 ~ 1 × 10−7 m) 0.38–0.48 × 10−3 cm/s 55% 55–65 mΩ·cm2 3727–4849 mΩ·cm NA NA

2.5. Bacterial community characterization

118

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

2.6. Statistical analysis Monitoring of the reactors was periodically done: six times for the first experiment (with two reactors) and nine times for the second experiment (with four reactors). The repeatability of measurements gives a statistical view of the behavior and fate of microbial cells over time. This type of system is complex and therefore not strictly reproducible, so the series values were compared over time. Statistical analysis was performed using non-parametric tests and differences between data sets were expressed as significant at a significance level of p b 0.05 (less than one in twenty chances of being wrong). The p value is the estimated probability of rejecting the null hypothesis (hypothesis of “no difference” of the test when that hypothesis is true). Statistical analyses have been performed using the StatEL software (adSciences, France). 3. Results and discussion 3.1. Coulombic and overall energy efficiencies Air-cathode MFCs needed about two to three weeks after inoculation to reach stable potential levels under a constant external load (R = 1 kΩ). The maximal obtained potential was around 380 mV (Fig. 1). Regular potential drops were observed with time, indicating that the acetate has been entirely consumed, and the solution should be replaced with fresh one (Fig. 1). The anodic chamber was then completely emptied and re-filled with new acetate buffer solution. These total refreshments were needed every 10–15 days for 50 mM acetate concentrations and every 3–4 days for 10 mM ones, showing acetate consumption rates of about 4 mM·d−1. These results illustrate the biofilm's high robustness relative to starvation, as cells quickly reached potentials close to last stable potentials. MFCs generate electricity by degradation of organic substrates (acetate in this study). The coulombic efficiency is the ratio of the number of coulombs obtained to the theoretical number of coulombs if all substrate were bio-electrochemically oxidized (Eq. (1)): Zt M C E¼

I  dt 0

F  b  V MFC  C

ð1Þ

where F is the Faraday's constant (96,485 C/mol), VMFC is the effective volume of the MFC (0.025 L), C is the substrate concentration (g/L), M is the molecular weight of the substrate, and b is the oxidation number of substrate (for acetate b = 8) [1,28]. The other important MFC parameter that should be evaluated is the energy recovery of the system is the overall energy efficiency, calculated as the ratio of the power produced by the cell over a time interval t to the heat of combustion of the organic substrate added in that time frame (Eq. (2)): Zt U cell  I  dt EE ¼

0

ð2Þ

ΔH  madded

where ΔH is the heat of combustion of acetate (ΔH =−874160 J·mol−1) and madded is the number of added acetate moles. Coulombic and energy efficiencies for both types of MFCs were varying according to time. With an acetate concentration of 50 mM, coulombic efficiency was ranging from 9% to 51% for Nafion-MFC and from 2% to 71% for PE-MFC. For energy efficiency, values were ranging from 0.5% to 21% for Nafion-MFC and from 0.3% to 14% for PE-MFC. Lower values were observed at the beginning of the experiment, because the biofilm was in a development phase within the first weeks. There was no significant difference (p b 0.05) between results achieved with both acetate concentrations (10–50 mM). This brings us to the conclusion that substrate concentration did not influence the biofilm's activity in the concentration region 10–50 mM [28,29]. Dependence of power production (Ps) on carbon source concentration can be described by the Andrew kinetic model (Eq. (3)) [30]: Ps ¼

P max  C KS;P þ C þ C

2

.

ð3Þ KI;P

where Pmax is the maximal power production expressed as 100%, and KS,P and KI,P are respectively the half-saturation and half-inhibition constants (g/L), with KS,P = 0.06156 g/L and KI,P = 2.422 g/L established from data of Perrin et al. [30] and from this experiment. The calculated electricity productions were approximately 65% and 75% of maximal production (Pmax) for respectively 10 mM and 50 mM acetate concentrations.

Fig. 1. Illustration of cell potential evolution in steady-state conditions (example of a given period). ⋄ Nafion®-117; • PE.

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

119

Fig. 2. Phylogenetic tree for the 13 bacterial 16S rRNA gene OTUs obtained from Nafion®-117 (Na) and RhinoHide® (RHa and RHc) membranes. Bootstrap values are based on 1000 runs. The tree was obtained using neighbor-joining with Methanocaldococcus jannaschii as the outgroup.

This theoretical difference is only 10% for both concentrations and thus confirms the observed results. A non-parametric test (Mann–Whitney) done on the different sets of coulombic or energy efficiency data showed no significant difference between the two types of membrane (p b 0.05).

3.2. Membrane resistance and biofouling estimation The use of membranes or other types of separators can decrease power generation and produce pH gradients between anode and cathode [31], and many of these membranes are expensive. Experiments with single air-cathode MFCs showed that using a separator (plastic mesh, polypropylene membrane) standing against the cathode increases charge transfer and diffusion resistance [32], due to the presence of trapped water between air-cathode and separator and due to biofilm growth. Pore size and porosity are determining parameters

in proton transport (related to diffusion, which is dependent on the proton concentration difference between anodic and cathodic chambers) through the water channel. The uncharged polyethylene membrane more fully separates the anodic and cathodic chambers in terms of oxygen concentration than does the Nafion® membrane, while protons can migrate between chambers through the uncharged polyethylene membrane's water channel by Fickian diffusion [3]. The common reason cited for cathodic limitation is the difficulty in providing protons to catalyst sites, but some authors have demonstrated that this is not the availability of protons but the transport of OH− anions from the catalyst layer to the bulk liquid that largely governs potential losses [33]. According to resistivity properties of membranes (Table 1) the theoretical membrane resistance (Rm) can be determined (Eq. (4)):.

Rm ¼ ρ 

L r ¼ A A

ð4Þ

120

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

Table 2 Average values for power and current densities and internal resistances with Nafion® and polyethylene membranes for initial acetate concentrations of 50 and 10 mM. Values of Pmax and Jmax were expressed by specific area or volume of carbon foam. [Acetate] 50 mM Type membrane

Nafion® Mean ± SDa RhinoHide® Mean ± SDb

[Acetate] 10 mM

Pmax mW·m−2

Pmax W·m−3

Jmax mA/m2

Rmax Ω

Rmax kΩ·cm2

Pmax mW·m−2

Pmax W·m−33

Jmax mA/m2

Rmax Ω

Rmax kΩ·cm2

17.2 ± 8.1

64.5 ± 30.4

44.7 ± 8.1

621 ± 193

24.9 ± 7.7

15.4 ± 3.7

57.8 ± 13.9

138.2 ± 53.1

507.1 ± 254.5

20.3 ± 10.2

11.4 ± 3.8

42.8 ± 14.3

55.9 ± 22.8

807 ± 191

32.3 ± 7.6

14.4 ± 9.1

54 ± 34.1

67.9 ± 27.1

860.7 ± 209.0

34.4 ± 8.4

SD: standard deviation. a n data points = 6. b n data points = 9.

where ρ: electrical resistivity in Ω·m; L: thickness in m; A: section area in m2, and r: electrical resistance in Ω·m2. Resistance is low for PE/silica membrane and close to 3.5–4.5 mΩ. These values are lower than for Nafion®-117, showing a value of 14 mΩ. In both cases contribution of the membrane to the experimental overall resistance seems negligible. After 8 months of functioning the amount of biomass on membrane anodic side was 309 μgBSA/cm2 for PE/silica membrane and 246 μgBSA/cm2 for Nafion®-117 one. On cathodic side the amount was 303 μgBSA/cm2 for PE/silica membrane and was zero for Nafion®-117. Similar amount of biomass was observed on both sides of PE/silica membrane illustrating bacterial permeability of this type of separator. Assuming a density for biofilm of 106 g m−3, and a protein content of 1–2% of the biofilm matrix [34], the thickness of biofilm on the membrane can be estimated to be 150–300 × 10−6 m, in the same order of membrane thickness without biofilm. Assuming a specific area resistance for mixed biofilm (Rb) of 0.5 to 1.5 kΩ·cm2 [35] the theoretical resistance of biofilm developed on membrane could be 31–91 Ω if one side is fouled with a biofilm, and twice for two sides fouled such as observed for PE/silica membrane. Contribution to the overall resistance becomes highly significant for the biofouled membrane in this case. This observation leads to conclude that charge transfer within the bare membrane was not a limiting factor. However, the bacterial permeability of the membrane may result in biofouling on both sides of the separator, leading to significant electric resistances.

3.3. Bacterial community diversity on membranes A total of 77 partial bacterial 16S rRNA gene sequences (~1100 bases) from the 3 clone libraries made from biofilms covering both air-cathode MFC membranes were recovered. Phylogenetic analyses were performed on all sequences, and applying a similarity cut-off of 97% allowed detecting 13 distinct OTUs (Fig. 2). Shannon diversity index was 0 for the cathodic side of RhinoHide® membrane (RHc) with only one OTU detected, 1.330 for the anodic side of RhinoHide® membrane (RHa) (richness of 6 OTUs) and 2.032 for Nafion®-117 (richness of 9 OTUs) exhibiting the more diversified community. Two OTUs were related to the Bacteroidetes phylum (Sphingobacterium and Dysgonomonas genera, OTUs 10 and 13, respectively); 1 OTU (OTU 3) was related to the Firmicutes phylum (Flavonifractor genus); and 10 OTUs were related to the Proteobacteria phylum. Amongst Proteobacteria, OTU 11 was related to the genus Brevundimonas (α-Proteobacteria); 5 OTUs were related to the β-Proteobacteria's genera Cupriavidus (OTU 2), Zoogloea (OTU 7), Acidovorax (OTUs 8 and 12) and Delftia (OTU 9); and 4 OTUs were related to the γ-Proteobacteria's genera Pseudomonas (OTUs 1 and 4), Citrobacter (OTU 5), and Stenotrophomonas (OTU 6). The Nafion®117 membrane community was dominated by clones affiliated to OTU 3 (31% of the clones) and OTU 3 (15% of the clones), and RHa community was dominated by clones affiliated to OTU 2 (53% of the clones), OTU 1 (19% of the clones) and OTU 11 (16% of the clones). The only OTU detected in RHc community was OTU 1.

Every OTU except OTU 2 and OTU 3 was specifically recovered from communities associated to one membrane type. OTUs 1, 4, 6 and 10 were only detected in the RhinoHide® membrane covering biofilms, and OTUs 3, 5, 7, 8, 9, 12, and 13 were only detected in the Nafion®-117 membrane covering biofilm. Similarity between both membrane bacterial communities was 15.4%, whereas between both sides of the RhinoHide® membrane, similarity was 16.67%. Altogether, this demonstrated the strong selective effect of the membrane on bacterial community. Cathode being the limiting compartment in the designed MCFs, this could have affected the fuel cell performance, whereas no effect of the microbial community structure on performance is generally observed when anode is the limiting compartment [36]. The biofilm covering the cathodic side of the RhinoHide® membrane appeared to be monospecific, the only detected OTU being related to Pseudomonas stutzeri. Some P. stutzeri strains were demonstrated to synthetize redox mediators involved in electron extracellular transfers [37]. P. stutzeri strain TR2 is able to use nitrite or nitrous oxide as an electron acceptor [38]. It is likely that nitrogen-compound from the MCF anodic side is used by P. stutzeri for respiration using electrons and protons coming from anodic chamber in the cathodic side where O 2 concentration is limiting for the development of other bacterial species. The UniFrac significance test probability for Nafion®-117 and RhinoHide® communities was 0.58, indicating that there was no significant difference (p N 0.1) between the two bacterial communities. The UniFrac test infers for phylogenetic differences between two communities. Here, at the phylum level, most phyla were detected from both Nafion®117 and RhinoHide® communities. At the genera/species level, the Nafion®-117 and RhinoHide® associated communities were different, indicating a selective effect of the membrane type, as opposed to what previously observed [39]. Archaeal related 16S rRNA gene sequences were only detected for the Nafion®-117 membrane covering biofilm. The 32 partial sequences (~ 660 bases) corresponded only to 3 distinct OTUs, all related to methanogenic Archaea. The first OTU shared 100% of similarity with Methanobacterium palustre 16S rRNA gene sequence, the second shared 100% of similarity with Methanobrevibacter arboriphilus sequence, and the last OTU was affiliated to Methanomassiliicoccus luminyensis (similarity of 99%). A recent study comparing the effect of cationic (Ultrex CMI-7000 and Ultrex AMI-7000) and anionic (Nafion®-117) membranes on the anodic chamber microbial diversity also detected the development of methanogenic Archaea and suggested that the Nafion®-117 membrane could favor the selective enrichment of methanogenic populations [38].

3.4. Linear sweep voltammetry Polarization curves have been realized just few hours after renewal of acetate buffer solution when cell potential was stabilized. These curves were conducted in chronoamperometric mode after 5 min of quiet time at open circuit potential (I = 0).

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

Power curves were extracted from cell polarization data. From the maximum power, the Ohm law allows the determination of the maximum internal resistance (Rmax). Table 2 lists the surface and volume power densities, current densities and internal resistances of the MFCs with Nafion® and polyethylene membranes for initial acetate concentrations of 50 and 10 mM. Values for Rmax did not differ so much, the most resistive component being the electrolyte, with low ionic concentrations. Although those current densities were weak (from 8 to 25 mW·m−2) due to high specific area of CF (3750 m−2 m−3), values of volumetric power density were in same order of magnitude than those obtained with similar kinds of MFC [30,40] i.e. a few Watts per cubic meter of anode volume or a few tens of Watts per cubic meter of carbon foam (Table 2). Using non-parametric test (Mann–Whitney), differences between Nafion®-MFC and PE-MFC and between 50 mM and 10 mM concentrations were not statistically significant. The null-hypothesis (Ho) cannot be rejected at a confident level of 5% (p b 0.05) even if average values were not the same (Table 2). Great variability was observed for most parameters (standard deviation accounted for up to 50% of the average value). This variability did not seem totally time dependent as shown in Fig. 3 and corresponds to some system instability over time. Looking at polarization curves, it can be seen for both membranes that cell voltage declined versus current density (Fig. 3A and C) . The first drop usually observed is connected to activation losses, which are related to the limitation of biological metabolism at the anode and to oxygen reduction at the cathode. The linear part is connected to global ohmic resistance of the system (charge transfer) and the last part at higher current densities is related to concentration polarization. In some cases the curve had a shape of swelling for higher current densities as observed for example at day 192 (10 mM acetate) with the

121

Nafion® membrane. This steep drop in voltage at higher current densities resulted in power overshoot occurrences in power density curves. Power overshoot [41] was systematically observed for scan rates higher than 0.5 mV s−1. Two types of power overshoots are commonly observed in polarization measurements with microbial fuel cells. The first type produces maximum power densities that are much higher than those corresponding to the long-term experiments with fixed external loads [42]. This overestimation of maximum power (type M) is a result of a too rapid scan rate, which does not provide enough time for acclimatization of the biofilm [43]. The second type of power overshoot occurs when the power density curves doubles back (type D) unexpectedly towards lower current densities. It is known that both types of overshoots are related to cathode performance [44]. The reasons for this kind of overshoot can been attributed to a variety of mechanisms, including mass transport limitations [45], electrical and ionic depletion on the anode at low external resistance [43], nature of substrate [46], age of the anode biofilm or extreme operating conditions like very low pH [47]. To overpass these experimental limitations, the MFCs were tested under a scan rate of 0.2 mV·s−1. The twist in the power curves was still visible in some cases for the Nafion® membrane (Fig. 3A and B) and was absent or barely visible for the polyethylene membrane (Fig. 3C and D). It is possible that the reduction in scan rate was not large enough to eliminate this phenomenon. 3.5. EIS measurements The EIS measurements were carried out at Open Circuit Voltage (OCV) corresponding to a potential difference of about 0.54 V between anode and cathode. Applying a weak perturbation such as sinusoidal potential to a stable system (OCV conditions), in the absence of flowing internal electrons, will allow its characterization in terms of in-phase

Fig. 3. Typical cell potential and cell power curves for MFCs using Nafion® (A and B) or polyethylene membrane (C and D), scan rate 0.2 mv/s and acetate concentration 10 mM. Panels A and B: ___ 201 days after inoculation; — 208 days after inoculation; … 220 days after inoculation; -...- 233 days after inoculation. Panels C and D: ___ 192 days after inoculation; — 213 days after inoculation; … 215 days after inoculation; -…- 221 days after inoculation.

122

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

CPE1

CPE2

Rs

R1

R2

CPE3

Scheme 2. Equivalent electrical circuit used for EIS modeling. Rs represents series resistances accounting for wires, contacts, membrane, electrolyte, …; R1 and R2 and CPE1 and CPE2 are for resistances and pseudo-capacitances for the two time constants identified by the fitting. CPE3 is accounting for diffusion.

and out-of-phase electrical elements (resistances and capacitances), here for the whole cell including the separator. These conditions differ from the long-running cell that was operated at maximum power conditions thanks to an adjustable external resistance. Working at OCV warrants stable conditions in order to compare various systems. The MFCs were kept at OCV for a period of 2–3 h before conducting the measurements. Applied frequency range was 105 to 10− 2 Hz. Amplitude of the AC signal was 10 mV, which is small enough not to perturb the MFC operation. The EIS measurements were performed using anode as a working electrode and air-cathode as both counter and reference electrodes. The high frequency part of the spectrum is reflecting the cell internal resistance (separator, electrolyte, contacts and wires). The medium frequency domain can feature one or several semi-circuits corresponding to various interface and charge transfer components. The low frequency domain is known to feature a straight line in the Nyquist plot, typically related to diffusion phenomena [48–51]. Impedance spectra for the entire cell were analyzed and modeled by the equivalent circuit model illustrated by Scheme 2. This model was the best fitting circuit for all experiments. According to this model, Rs accounts for ohmic contribution of contacts, wires, electrolyte and membrane. It is widely admitted [51] that R1 can account for charge transfer resistance of the RVC electrode and R2 for charge transfer resistance of the air-cathode. Constant Phase Elements CPE1 and CPE2 are distributed (non-ideal) capacitors, and CPE3 corresponds to diffusion contribution to the measured impedance. The use of CPE for modeling of frequency-depending capacitors is justified by the inhomogeneous properties of electrodes like surface roughness, variable coating thickness and spatial distribution of reaction rates.

Table 3 Fitting parameters for Nafion® and polyethylene membrane MFCs, for initial acetate concentrations of 50 mM; mean of three values for period from day 15 to day 45 (50 mM) and for period from day 192 to day 233 (10 mM). Rs represents series resistances accounting for wires, contacts, membrane, electrolyte, …; R1 and R2 and CPE1 and CPE2 are for resistances and pseudo-capacitances for the two time constants identified by the fitting. CPE3 is accounting for diffusion. p is a parameter reflecting non-ideality in capacitances. Concentration

50 mM

Membrane

Nafion®

PE

Nafion®

10 mM PE

Membrane Rs (Ω) CPE1 (μF·sp.−1·cm−2) p1 R1 (Ω) CPE2 (μF·sp.−1·cm−2) p2 R2 (Ω) CPE3 (μF·sp.−1·cm−2) p3 Weight sum of squares

Nafion® 16.2 12.3 0.61 33.1 6.8 0.79 5.3 19.6 0.79 0.005

PE 19.7 20.8 0.5 0.7 2.2 0.6 27.9 50.8 0.9 0.0041

Nafion® 53.5 2.0 0.5 35.4 2.2 0.9 32.3 116.4 0.65 0.036

PE 37 4.0 0.64 440 4.3 0.51 50.4 10.1 1 0.25

In fact, for an ideal capacitor of capacitance C, ω being the applied pulsation, impedance ZC is written as. Z C ¼ 1=iωC

ð5Þ

A constant phase element CPE is rather characterized by its impedance ZCPE as Z CPE ¼ 1=CPEðiωÞP :

ð6Þ

The pseudo-capacitance CPE is expressed in F·sp.−1. p is a parameter reflecting non-ideality. If p equals 1, then Eq. (6) is identical to Eq. (5). When a CPE is placed in parallel to a resistor, a depressed semi-circle is produced in the complex Nyquist plan, and a maximum is seen in Bode diagrams (Fig. 4). Fitting parameters can be used to tentatively characterize the electrochemical behavior of the MFC. Such kind of analysis can provide interesting information about electrochemical reaction kinetics and bacterial metabolisms [51]. The obtained values for Rs for all MFCs were in the range 16–20 Ω for the 50 mM acetate concentration and 37–53 Ω for the 10 mM one (Table 3). These values are inside usual range for various microbial fuel cells with carbon electrodes [48,52] obtained at OCV conditions.

Fig. 4. A typical set of impedance spectra, here for a global polyethylene membrane MFC (50 mM acetate, day 58), presented as the Nyquist (A) and Bode (B) plots. The resulting fit is also shown on the Nyquist diagram. • Log Z; ⋄ phase angle.

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

Despite of the fact that Rs values for 50 mM concentration were close together for both cells, the values for 10 mM showed a trend towards lower series resistances with PE membranes. Concerning CPEs, the adjustable parameter p was comprised between 0.5 and 1, showing that considering capacitors, although non-ideal, besides resistors can be validated for an electric representation of interface phenomena. Global analyses through the equivalent electrical circuit of Scheme 2 were also validated by the low values reported for weight sum of squares. 4. Conclusion The non-selective microporous membrane tested in an air-cathode MFC with RVC bioanode did not result in a significantly different behavior compared to the classical proton exchange membrane (Nafion®). Volumetric power densities were in the range of few tens of Watts per cubic meter of RVC foam. Biomass was observed on both sides of PE/silica membrane illustrating bacterial permeability of this type of separator. Even there was no significant difference between Nafion®-117 and RhinoHide® bacterial communities, methanogenic populations have been detected on anodic side of Nafion®-117 membrane and a monospecific denitrifying population on cathodic side of RhinoHide® membrane. These results confirm that biofouling on membranes is an important factor acting on MFC behavior. Great variability was observed for most electrochemical parameters, corresponding to a given instability of the system over long periods of operation whatever membrane type in this study. Power overshoot observed in some case at higher current densities in power density curves can be an illustration of this instability. The parameters from electrochemical impedance spectroscopy fittings are similar to those of previously reported data for various microbial fuel cells with carbon electrodes at OCV conditions. The report of a lower series resistance with polyethylene membrane corresponds to lower specific resistance and better permeability for this type of material. Other resistances and pseudo-capacitances from EIS analyses do not differ in such a way that the nature of the membrane could be considered as responsible at OCV. The opportunity of using a non-selective microporous membrane as a separator must be confirmed by testing under different operating conditions. Acknowledgments This work has received a post-doctoral financial support from Assemblée des Pays de Savoie. The authors would like to thank Thierry Goldin and Cédric Poinard for the helpful discussions and technical support, together with Dr. Angel Kirchev from CEA-LITEN for the microporous polyethylene membrane samples. We thank Nathalie Tissot for the molecular biology assistance. References [1] B.E. Logan, B. Hamelers, R. Rozendal, U. Schröder, J. Keller, S. Freguia, P. Aelterman, W. Verstraete, K. Rabaey, Microbial fuel cells: methodology and technology, Environmental Science and Technology Vol. 40 (2006) 5181–5192. [2] Z. Du, H. Li, T. Gu, A state of the art review on microbial fuel cells: a promising technology for wastewater treatment and bioenergy, Biotechnology Advances Vol. 25 (2007) 464–482. [3] Y. Kim, S.-H. Shin, S. Chang, S.-H. Moon, Characterization of uncharged and sulfonated porous poly(vinylidene fluoride) membranes and their performance in microbial fuel cells, Journal of Membrane Science 463 (2014) (2014) 205–214. [4] B.E. Logan, Scaling up microbial fuel cells and other biolectrochemical systems, Applied Microbiology and Biotechnology Vol. 85 (2010) 1665–1671. [5] J.R. Kim, S. Cheng, S.-E. Oh, B.E. Logan, Power generation unsing different cation, anion, and ultrafiltation membranes in microbial fuel cells, Environmental Science and Technology Vol. 41 (2007) 1004–1009. [6] R. Rozendal, H.V.M. Hamelers, C.J.N. Buisman, Effects of membrane cation transport on pH and microbial fuel cell performance, Environmental Science and Technology Vol. 40 (2006) 5206–5211.

123

[7] X. Zhang, S. Cheng, X. Huang, B.E. Logan, Improved performance of single-chamber microbial fuel cells through control of membrane deformation, Biosensors and Bioelectronics Vol. 25 (2010) 1825–1828. [8] J.R. Kim, G.C. Premier, F.R. Hawkes, R.M. Dinsdale, A.J. Guwy, Development of a tubular microbial fuel cell (MFC) employing a membrane electrode assembly cathode, Journal of Power Sources Vol 187 (2009) 393–399. [9] Y. Mo, P. Liang, W.+.X. Huang, H. Wang, X. Cao, Enhancing the stability of power generation of single-chamber microbial fuel cell using an anion exchange membrane, Journal of Chemical Technology and Biotechnology Vol. 84 (2009) 1767–1772. [10] R.A. Rozendal, H.V.M. Hamelers, K. Rabaey, J. Keller, C.J.N. Buisman, Towards practical implementation of bioelectrochemical wastewater treatment, Trends Biotechnol. 26 (2008) 450–459. [11] Y. Fan, H. Hu, H. Liu, Enhanced coulombic efficiency and power density of aircathode microbial fuel cell with an improved cell configuration, Journal of Power Sources Vol. 171 (2007) 348–354. [12] X. Zhang, S. Cheng, X. Wang, X. Huang, B.E. Logan, Separator characteristics for inceasing performance of microbial fuel cells, Environmental Science and Technology Vol. 43 (2009) 8456–8461. [13] X. Zhang, S. Cheng, P. Liang, X. Huang, B.E. Logan, Scalable air cathode microbial fuel cell using glass fiber separator, plastic mesh supporter, and graphic fiber brush anodes, Bioresource Technology Vol. 102 (2011) 372–375. [14] J.X. Leong, W.R.W. Daud, M. Ghasemi, K.B. Liew, M. Ismail, Ion exchange membranes as separators in microbial fuel cells for bioenergy conversion: a comprehensive review, Renewable and Sustainable Energy Reviews Vol. 28 (2013) 575–587. [15] B.R. Dhar, H.-S. Lee, Membranes for bioelectrochemical systems: challenges and research advances, Environmental Technology Vol. 34 (2013) 1751–1764. [16] J. Yu, Y. Park, T. Lee, Effect of separator and inoculum type on electricity generation and microbial community in single-chamber microbial fuel cells, Bioprocess and Biosystems Engineering Vol. 37 (2014) 667–675. [17] S. Choi, J.R. Kim, J. Cha, Y. Kim, G.C. Premier, C. Kim, Enhanced power production of a membrane electrode assembly microbial fuel cell (MFC) using a cost effective poly [2,5-benzimidazole] (ABPBI) impregnated non-woven fabric filter, Bioresource Technology 128 (2013) (2013) 14–21. [18] R.W. Pekala, C. La, G. Fraser-Bell, M. Ulrich, S. Gerts, D. Trueba, R. Waterhouse, Separator design considerations for EFB micro-HEV applications, 13th European Lead Batteries Cionference, Paris, September 28th, 2012. [19] R. Burkitt, E. Yu, Application of anion exchange membranes in microbial fuel cells, Anion-Exchange Membranes for Energy Generation Technologies, IAS workshop, University of Surrey, June 25th, 2013, 2013.http://www.ias.surrey.ac.uk/workshops/ membranes/papers/Burkitt.pdf. [20] Z. He, F. Mansfeld, Exploring the use of electrochemical impedance spectroscopy (EIS) in microbial fuel cell studies, Energy Environmental Science Vol. 2 (2008) 215–219. [21] G. Lepage, G. Perrier, G. Merlin, X. Dominguez-Benetton, N. Aryal, Multifactorial evaluation of the electrochemical response of a microbial fuel cell, RSC Advances Vol. 4 (2014) 23815–23825. [22] K. Tamura, G. Stecher, D. Peterson, A. Filipski, S. Kumar, MEGA6: Molecular Evolutionary Genetics Analysis version 6.0, Molecular Biology and Evolution Vol. 30 (2013) 2725–2729. [23] Q. Wang, G.M. Garrity, J.M. Tiedje, J.R. Cole, Naïve Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy, Applied and Environmental Microbiology Vol. 73 (2007) 5261–5267. [24] J.R. Cole, Q. Wang, J.A. Fish, B. Chai, D.M. McGarrell, Y. Sun, C.T. Brown, A. PorrasAlfaro, C.R. Kuske, J.M. Tiedje, Ribosomal Database Project: data and tools for high throughput rRNA analysis, Nucleic Acids Research Vol. 41 (Database issue) (2014) D633–D642. [25] C. Lozupone, R. Knight, UniFrac: a new phylogenetic method for comparing microbial communities, Applied and Environmental Microbiology Vol. 71 (2005) 8228–8235. [26] C. Lozupone, M. Hamady, R. Knight, UniFrac — an online tool for comparing microbial community diversity in a phylogenetic context, BMC Bioinformatics Vol. 7 (2006) 371. [27] M. Bradford, A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein–dye binding, Analyt. Biochem. 72 (1976) (1976) 248–254. [28] M. Wang, Z. Yan, Electricity generation by microbial fuel cells fuelled with Enteromorpha prolifera hydrolysis, Int J. Electrochem. Sci. 8 (p) (2013) 2104–2111. [29] K. Rabaey, N. Boon, Hofte, Microbial phenazine production enhances electron transfer in biofuel cells, Environ. Sci. Technol. Vol.39 (2005) 3401–3408. [30] E. Martin, O. Savadogo, S.R. Guiot, B. Tartakovsky, The influence of operational conditions on the performance of a microbial fuel cell seeded with mesophilic anaerobic sludge, Biochemical Engineering Journal Vol. 51 (2010) 132–139. [31] R.A. Rozendal, H.V.M. Hamelers, R.J. Molenkamp, C.J.N. Buisman, Performance of single chamber biocatalyzed electrolysis with different types of ion exchange membranes, Water Research Vol. 41 (2007) 1984–1993. [32] B. Wei, J.C. Tokash, F. Zhang, Y. Kim, B.E. Logan, Electrochemical analysis of separators used in single-chamber, air-cathode microbial fuel cells, Electrochimica Acta Vol. 89 (2013) 45–51. [33] S.C. Popat, D. Ki, B.E. Rittmann, C.I. Torres, Importance of OH− transport from cathodes in microbial fuel cells, ChemSusChem 5 (2012) 1071–1079. [34] I.W. Sutherland, The biofilm matrix — an immobilized but dynamic microbial environment, Trends in Microbiology 9 (2001) 222–227. [35] N.S. Malvenkar, J. Lau, K.P. Nevin, A.E. Franks, M.T. Tuominen, D.R. Lovley, Electrical conductivity in a mixed biofilm, Appl. Environ. Microbiol. Vol 78 (2012) 5967–5971. [36] M.D. Yates, P.D. Kiely, D.F. Call, H. Rismani-Yazdi, K. Bibby, J. Peccia, J.M. Regan, B.E. Logan, Convergent development of anodic bacterial communities in microbial fuel cells, The ISME Journal, Vol. 6 (2012) 2000–2213.

124

N. Kircheva et al. / Bioelectrochemistry 106 (2015) 115–124

[37] K. Watanabe, M. Manefield, M. Lee, A. Kouzuma, Electron shuttles in biotechnology, Current Opinion in Biotechnology, Vol. 20 (2009) 633–641. [38] M. Miyahara, S.-W. Kim, S. Fushinobu, K. Takaki, T. Yamada, A. Watanabe, K. Miyauchi, G. Endo, T. Wakagi, H. Shoun, Potential of aerobic denitrification by Pseudomonas stutzeri TR2 to reduce nitrous oxide emissions from wastewater treatment plants, Appl. Environ. Microbiol. Vol. 76 (2010) 4619–4625. [39] A. Sotres, J. Diaz-Marcos, M. Guivernau, J. Illa, A. Magri, F.X. Prenafeta-Boldu, A. Bonmati, M. Vinas, Microbial community dynamics in two-chambered microbial fuel cells: effect of different ion exchange membranes, Chemical Technology and Biotechnology (2014), http://dx.doi.org/10.1002/jctb.4465. [40] V.B. Oliveira, M. Simõesb, L.F. Melob, A.M.F.R. Pintoa, Overview on the developments of microbial fuel cells, Biochemical Engineering Journal Vol. 73 (2013) 53–64. [41] Y. Hong, D.F. Call, Adaptation to high current using low external resistances eliminates power overshoot in microbial fuel cells, Biosensors and Bioelectronics Vol. 28 (2011) 71–76. [42] S.B. Velasquez-Ora, D.F. Call, Energy from algae using microbial fuel cells, Biotechnology and Bioengineering Vol. 103 (2009) 1068–1076. [43] I. Leropoulos, J. Winfield, Effects of flow-rate, inoculum and time on the internal resistance of microbial fuel cells, Bioresource Technology Vol. 101 (2010) 3520–3525. [44] V.J. Watson, B.E. Logan, Analysis of polarization methods for elimination of power overshoot in microbial fuel cells, Electrochemistry Communications Vol. 13 (2009) 54–56.

[45] P. Aelterman, K. Rabaey, Continuous electricity generation at high voltages and currents using stacked microbial fuel cells, Environmental Sci. & Technol. Vol. 40 (2006) 3388–3394. [46] P.-C. Nien, C.-Y. Lee, Power overshoot in two-chambered microbial fuel cell (MFC), Bioresource Technology Vol. 102 (2011) 4742–4746. [47] J. Winfield, I. Leropoulos, The overshoot phenomenon as a function of internal resistance in microbial fuel cells, Bioelectrochemistry Vol. 81 (2011) 22–27. [48] Z. He, S.D. Wagner, S.D. Minteer, L.T. Angenent, An upflow microbial fuel cell with an interior cathode: assessment of the internal resistance by impedance spectroscopy, Environ. Sci. Tech. Vol. 40 (2006) 5212–5217. [49] G. Lepage, F.O. Albernaz, G. Perrier, G. Merlin, Characterization of a microbial fuel cell with reticulated carbon foam electrodes, Bioresource Technology Vol. 124 (2012) 199–207. [50] E. Barsoukov, J. Ross Macdonald, Impedance Spectroscopy Theory, Experiment, and Applications, 2nd edition John Wiley & Sons, New York, 2005. [51] D. Aurbach, B. Markovsky, M.D. Levi, New insights into the interactions between electrode materials and electrolyte solutions for advanced nonaqueous batteries, Journal of Power Sources Vol. 81 (1999) 95–111. [52] J.R. Kim, G.C. Premier, F.R. Hawkes, R.M. Dindsdale, A.J. Guwy, Development of a tubular microbial fuel cell (MFC) employing a membrane electrode assembly cathode, Journal of Power Sources Vol. 187 (2009) 393–399.