This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

Transcription

1 This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier s archiving and manuscript policies are encouraged to visit:

2 Bioresource Technology 99 (2008) Contents lists available at ScienceDirect Bioresource Technology journal homepage: Loading rate and external resistance control the electricity generation of microbial fuel cells with different three-dimensional anodes Peter Aelterman, Mathias Versichele, Massimo Marzorati, Nico Boon, Willy Verstraete *,1 Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, B-9000 Ghent, Belgium article info abstract Article history: Received 27 February 2008 Received in revised form 18 April 2008 Accepted 21 April 2008 Available online 3 June 2008 Keywords: Polarization curve Biocatalyst Biological fuel cell Cathode Bio electrochemical system (BES) The electricity generation, electrochemical and microbial characteristics of five microbial fuel cells (MFCs) with different three-dimensional electrodes (graphite and carbon felt, 2 mm and 5 mm graphite granules and graphite wool) was examined in relation to the applied loading rate and the external resistance. The graphite felt electrode yielded the highest maximum power output amounting up to 386 W m 3 total anode compartment (TAC). However, based on the continuous current generation, limited differences between the materials were registered. Doubling the loading rate to 3.3 g COD L 1 TAC d 1 resulted only in an increased current generation when the external resistance was low ( X) or during polarization. Conversely, lowering the external resistance resulted in a steady increase of both the kinetic capacities of the biocatalyst and the continuous current generation from 77 (50 X) up to 253 (10.5 X)Am 3 TAC. Operating a MFC at an external resistance close to its internal resistance, allows to increase the current generation from enhanced loading rates while maximizing the power generation. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Microbial fuel cells (MFCs) use the catalytic activity of microorganisms for the conversion of various organic and inorganic components into electricity (Allen and Bennetto, 1993). Due to improvements of the reactor design (Rabaey et al., 2005a; Zuo et al., 2007), the application of new electrode materials (Logan et al., 2007; Park and Zeikus, 2002), improved cathode reactions (Clauwaert et al., 2007; Zhao et al., 2005) and the enrichment of highly specialized microbial communities (Logan and Regan, 2006), the performance and durability of MFCs has increased steadily over the last years. The reported power densities vary from 55 W m 3 net anode compartment (NAC) for the liter scale systems (Ter Heijne et al., 2007) to 1010 W m 3 NAC for the milliliter scale (Fan et al., 2007), but a continuous search to improve the power density is ongoing. The anode electrode, which contains a matrix for the attachment of the microorganisms, is an important factor of the MFC design. Modifying the electrode surface by the binding of mediators such as neutral red and metals such as iron and manganese has resulted in an increased current generation (Park and Zeikus, 2002). Foam structured anodes with carbonized polymers with ferrocene catalyst result in a higher biocompatibility with Escherichia coli (Morozan et al., 2007). By using carbon electrodes modified with quinone/quinoid groups (Scott et al., 2007) and by combining a * Corresponding author. Tel.: +32 (0) ; fax: +32 (0) address: Willy.Verstraete@UGent.be (W. Verstraete). 1 highly conductive electrolyte and an ammonium electrode treatment (Cheng and Logan, 2007), a doubling of the maximum power density could be achieved compared to unmodified graphite and carbon felt. Specifically, however, the use of a three-dimensional electrode has resulted in the increase of the volumetric power densities (Rabaey et al., 2005a). Sell et al. (1989) were the first to report the use of a packed bed of granular graphite as a threedimensional electrode. Meanwhile, other three-dimensional electrodes have been tested, including reticulated vitreous carbon (He et al., 2005), granular activated carbon (He et al., 2006) and carbon brushes (Logan et al., 2007). In comparison to the use of flat or two-dimensional electrodes, three-dimensional electrodes have higher surface to volume ratios of up to m 2 m 3 (Freguia et al., 2007) which can support an increased attachment of bacteria. It also results in a lower surface based current density for a given flow of electrons. As most losses are related to the current density, this can result in an increased power output (Hoogers, 2003). The application of three-dimensional structures is inevitable for the generation of high currents in future larger MFC systems. Next to the design, the current generation of MFCs is influenced by the operational parameters applied. The electrode potential of the anode has been shown to regulate the activity of the electron-transferring microorganisms (Aelterman et al., 2008a). A step-wise increase of the loading rate, by changing the substrate concentration or the flow rate, resulted successively in an increase, saturation and decrease of the current generation during polarization (Moon et al., 2006; Rabaey et al., 2003). The external resistance controls the ratio between the current generation and the /$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi: /j.biortech

3 8896 P. Aelterman et al. / Bioresource Technology 99 (2008) cell voltage (Logan et al., 2006). A high external resistance results in a high cell voltages and low current; the opposite is true in the case of a low external resistance. Theoretically, the optimal resistance for maximum power generation approximates the internal resistance (Logan et al., 2006). However, the prolonged effect of the external resistance and the loading rate on the activity of the biocatalyst in MFCs has not yet been investigated. From a construction point of view, a three-dimensional electrode allowing the highest power output for a given volume is preferred. The aim of this study was to examine the performance of five three-dimensional anode materials (graphite felt, carbon felt, 5 mm graphite granules, 2 mm graphite granules and graphite wool) in continuously operated MFCs. Moreover, the external resistance and volumetric loading rate were periodically changed to study their effect on the continuous electricity generation, biomass quantity and microbial community. 2. Materials 2.1. Reactor setup Five reactors were constructed from perspex frames as described previously (Aelterman et al., 2006). The total empty volume of one frame was 156 ml (TAC, total anode compartment). Five anode materials were tested: graphite felt (Sigratherm GFA, SGL Carbon group, Germany), carbon felt (Sigratherm KFA, SGL Carbon group, Germany), graphite wool (graphitwolle, SGL Carbon group, Germany), 5 mm graphite granules (Le Carbone, Belgium) and 2 mm graphite granules (Le Carbone, Belgium). The five anode frames were completely filled with a three-dimensional electrode material. A graphite rod (5 mm diameter, Morgan, Belgium) was used to collect the electrons. The total volume of each electrode material was 156 ml. To obtain a homogenous liquid upflow through the anode compartment, a perforated tube was installed in the anode compartment inlet. Each anode compartment had a hydraulic circuit containing an individual influent tubing, outlet tubing and a recirculation loop. The cathode frames contained an inlet and outlet to which an individual cathode recirculation loop and common buffer vessel (2 L) were connected. A cation exchange membrane (Ultrex TM CMI7000, Membranes International Inc., USA) was used between the anode and cathode of each MFC unit. The five MFCs were separated by four rubber sheets (5 mm). The cathode electrodes consisted of a packed bed of graphite granules, (5 mm, Le Carbone, Belgium) and a graphite rod (5 mm diameter, Morgan, Belgium) to collect the electrons. Prior to use, the granules were washed three times with water Operational conditions and inoculation The three MFCs were operated in a continuous mode at an external resistance of 50, 25 and 10.5 X. The MFCs were inoculated with the effluent of an active acetate-fed MFC. During the start-up period, the external resistance was gradually lowered from 500 to 50 X. The duration of the periods and the corresponding external resistances are outlined in Table 1. A sterile synthetic influent containing 1.0 g L 1 sodium acetate prepared as previously described (Rabaey et al., 2005b) was continuously fed to the individual MFCs by a peristaltic pump (Watson Marlow, Belgium) at 13.3 or 26.6 ml h 1 corresponding to a volumetric loading rate of 1.6 or 3.3 g COD L 1 TAC d 1. Each MFC anode had a recirculation loop with a flow rate of 100 ml h 1. The outline of the operational conditions is summarized in Table 1. The catholyte was prepared according to Park and Zeikus (2003) and consisted of 2 L 100 mm K 3 Fe(CN) 6 aqueous solution in a 3gL 1 KH 2 PO 4 and a 6 g L 1 Na 2 HPO 4 buffer (Merck, Belgium) which was recirculated from the buffer vessel through the cathode matrix at a flow rate of 2.5 L h 1. The oxidation/reduction potential of the catholyte was controlled by a periodic renewal of the catholyte solution when the solution was decolorized. The MFCs were operated at a room temperature of 23 ± 2 C Data acquisition, electrochemical and statistical analysis The cell voltage was continuously measured every 60 s using a datalogger (HP 34970, Agilent, The Netherlands). The current was calculated from the cell voltage and the external resistance using Ohms law. The currents were averaged over a period of 12 h after the catholyte had been replenished to assure equal cathode performance for all tests. Polarization curves were obtained by imposing a linear potential decrease of 1 mv s 1 from the open circuit voltage (OCV) potential to a cell potential of 0 mv followed by a linear voltage increase of 1 mv s 1 of the cell potential to the original OCV by a multichannel potentiostat (BiStat, Bio-Logic, France). The polarization curves of the reactors were measured in triplicate with a preceding OCV period of approximately 1 h. The slope of the linear ohmic part of the polarization curves was determined as the R slope. Using the ratio of the cell voltage and the current during the maximum power generation, the optimal resistance (R opt ) was calculated. Subsequently, the ohmic losses (R ci ) were determined using the current interrupt method (Aelterman et al., 2006). The latter was also used to verify the electrical contact between the rod and the electrode materials. The coulombic efficiency was calculated according to Rabaey et al. (2005a) and was based on the applied loading rate. Table 1 Overview of the operational parameters (B v : volumetric loading rate; R ext : external resistance), the coulombic efficiency (CE), continuous current density (I cont ), continuous power density (P cont ), the maximum current densitiy (I max ) and the maximum power density (P max ) during the various experimental periods Period Day a b B v R ext (X) CE (%) I c cont (A m 3 TAC) I c max (A m 3 TAC) P d cont (W m 3 TAC) P d max (W m 3 TAC) * ± 3 74 ± ± ± ± ± 4 66 ± ± ± ± ± ± ± ± ± ± ± ± ± ± 72 Break 58 < >26 ± 6 59 ± ± ± ± ± 2 77 ± ± ± ± ± 2 77 ± ± ± ± ± ± ± ± ± ± ± ± ± ± 46 * The resistance was periodically lowered from 500 to 100 to 50 X. a Days counted from inoculation. b The volumetric loading rate (B v ), expressed as g COD L 1 TAC d 1, was adjusted by changing the flow rate from 13.3 to 26.6 ml h 1. c Based on the average continuous values of all materials excluding the 2 mm graphite granules during periods of 12 h after changing the catholyte solution. d Based on the average values derived from the polarization curves recorded in the respective periods for all materials excluding the 2 mm graphite granules.

4 P. Aelterman et al. / Bioresource Technology 99 (2008) To asses the differences between the different operational conditions and the performance of the materials, the average continuous current generation (see above) and maximum current and power generation (obtained during polarization) were analyzed using a two-sided Student T-test with at a significance level of 0.95 (p-value of 0.05). The equality of variances was tested with a F-test at a significance level of 0.95 (p-value of 0.05). The linear Pearson correlation (r) and the R-square were calculated for all combinations of the maximum power (P max ), the maximum current (I max ), the inverse of the optimal resistance (1/R opt ), the slope of ohmic part of the polarization curve (R slope ) as obtained during the polarization and the ohmic losses (R ci ) from current interrupt measurements during period 5 8 for all reactors Phospholipids analysis for determining microbial biomass The procedure used in this study to determine microbial phospholipids was a modification of that found in Findlay et al. (1989) and is described in Aelterman et al. (2008a). The reactors were opened and samples of the mixed biomass-loaded electrodes (1 cm 3 ) were randomly withdrawn in triplicate from the reactor on day 113, 127, 141 and 157 after inoculation. The samples were replaced by the same amount of fresh electrode material to maintain the same electrode volume. The biomass concentration was calculated based on the conversion factor of lg of biomass- C per 100 mmol of phospholipid (Findlay et al., 1989) and a conversion factor of 2 g biomass-vss per g of biomass-c based on an empirical formula of sludge organisms (C 5 H 7 NO 2 )(Porges et al., 1953) Microbial community analysis Samples to analyse the microbial composition on the anode materials were obtained by opening the reactor and withdrawing at random 1 cm 3 of moist anode material at day 113 and 157 after inoculation. The samples were replaced by the same amount of fresh electrode material to maintain the same electrode volume. The total DNA and RNA extraction and 16S rrna gene analysis for all bacteria by polymerase chain reaction denaturant gradient gel electrophoresis (PCR DGGE) was done as described previously (Boon et al., 2003). The normalization and analysis of DGGE gel patterns was done with the BioNumerics software 2.0 (Applied Maths, Kortrijk, Belgium). The calculation of the matrix of similarities is based on the Pearson product-moment correlation coefficient. The clustering algorithm of Ward was used to calculate the dendrograms. Focussing on the active microorganisms (RNA DGGE), some ecological parameters were calculated using a recently developed molecular analysis toolbox (Marzorati et al., 2008). The range-weighted richness (Rr) is a value to indicate the carrying capacity of a given environment, the rate of change is indicative for the dynamics of the system and the Pareto Lorenz curves are used to describe the structure of the microbial community associated with the different reactors. 16S rdna gene fragments (Band 1 and 2) were cut out of the DGGE gel with a clean scalpel and added to 50 ll of PCR water. After 12 h of incubation at 4 C, 1 ll of the PCR water was reamplified with primer set P338F and P518r. Then 5 ll of the PCR product was loaded onto a DGGE gel, and if the DGGE pattern showed only one band, it was sent out for sequencing. DNA sequencing of the ca. 180-bp fragments was carried out by ITT Biotech-Bioservice (Bielefeld, Germany). Analysis of DNA sequences and homology searches were completed with standard DNA sequencing programs and the BLAST server of the National Center for Biotechnology Information (NCBI) using the BLAST algorithm (Altschul et al., 1997). 3. Results 3.1. Electricity generation using three-dimensional anode materials A set of five microbial fuel cells (MFCs), each with a different anode electrode material, was evaluated for more than 140 days under various operational conditions. All reactors, except the reactor with the 2 mm graphite granules, needed 6 days before a significant current and power increase was noted. The reactor with the 2 mm graphite granules electrodes needed another 4 days before a comparable power generation at 50 X was generated. The continuous current generation of the reactors with graphite felt, carbon felt, 5 mm graphite granules and graphite wool electrodes was not significantly different. However, the power and current output of the reactor with the 2 mm graphite granules was, except at an external resistance of 50 X, significantly lower. The ohmic loss (R ci ) of the MFC with the 2 mm granules was almost double (3.8 ± 0.4 X) compared to the average ohmic losses of the other MFCs (2.0 ± 0.4 X). The internal resistance (R opt ) decreased from 13.5 ± 4.2 to 3.3 ± 0.3 X, for all reactors excluding the 2 mm graphite granules reactor. The internal resistance of the latter remained stable at 8.2 ± 1.2 X. The reactor with the 2 mm granules had a less stable electricity generation. Only when the MFCs were operated at the lowest external resistance (10.5 X) and at the highest loading rate (3.3 g COD L 1 d 1 ), differences of the maximum power generation amongst the other reactors were noted. The graphite felt and carbon felt electrodes generated the highest maximum power outputs of respectively 386 and 356 W m 3 TAC. The graphite wool electrode and the 5 mm graphite granules generated a lower maximum power output of 321 and 257 W m 3 TAC, respectively. While, the 2 mm granules only generated 143 W m 3 TAC The influence of the loading rate on the electricity generation The loading rate determines the maximum amount of charge or the average current which can be generated during a defined timeframe. The influence of increasing the loading rate on the continuous electricity generation at different external resistances was investigated. The various regimes, their duration, the coulombic efficiency and the average continuous and maximum current and power generation, excluding the results of the 2 mm graphite granules, are summarized in Table 1. At an external resistance of 50 X, the continuous current generation improved significantly when the loading rate was increased from an irregular feeding (B v < 1.6 g COD L 1 TAC d 1 ) to a continuous feeding of 1.6 g COD L 1 TAC d 1. A subsequent doubling of the loading rate to 3.3 g COD L 1 TAC d 1, by doubling the flow rate of the influent, did not result in a significant increase of the continuous current generation at 50 X (Table 1). At an external resistance of 10.5 X however (Periods 1 and 3), the continuous current significantly increased by doubling the loading rate from 1.6 to 3.3 g COD L 1 TAC d 1. Remarkably, in all cases the maximum current generation during polarization increased significantly when the loading rate was increased (Table 1). This indicates that only at low resistances or at near maximum currents, an increase of the loading rate resulted in an increased electricity generation The current generation in relation to a lowering of the external resistance During subsequent periods of at least 7 days at a loading rate of 3.3 g COD L 1 d 1, the external resistance of all reactors was lowered from 50 to 25 and finally to 10.5 X which resulted in a significant increase of the continuous current generation (Fig. 1) and power generation (Periods 5 8, Table 1). Although an increase of

5 8898 P. Aelterman et al. / Bioresource Technology 99 (2008) can potentially deal with changing environmental conditions and preserve their functionally Correlations between the electrical parameters Fig. 1. Overview of the continuous current generation (A m 3 average current density during a period of 12 h) for the five electrode materials during subsequent changes of the volumetric loading rate and the external resistance. the continuous current was noted, the maximum current generation as recorded during polarization did not change significantly. However, when a qualitative assessment of the polarization curves was made (Fig. 2), a clear shift of the polarization curves at cell voltages below 200 mv was noted. When the MFCs were operated at lower resistances, the phenomena associated with mass transfer or kinetic limitations observed during polarization lowered, resulting in a less steep descent of the current density, starting at cell voltages of 200 mv. The 2 mm and 5 mm granular graphite reactors, appeared to suffer less from this leveling off of the current. As the hexacyanoferrate catholyte enabled an equal cathode performance of the five MFCs, these different characteristics can be mainly attributed to the changes of the anode performance Biomass concentrations and microbial composition The phospholipids concentrations associated with the electrode materials were monitored over time to estimate the biomass concentration. Overall, a doubling of the biomass concentration occurred during the operational period from day 113 until day 157 (Table 2). The microbial composition was analyzed with PCR DGGE at two points in time: at the beginning of period 4 and at the end of period 8. Moreover, using respectively a RNA and DNA extraction, the active and present species within the microbial community could be discerned. Fig. 3 gives an overview of the band pattern of both the DNA and RNA DGGEs. Cluster analysis revealed that, based on the DNA composition of the microbial community, there was 80% similarity between the microbial communities in all reactors, both at period 1 and period 4, with the presence of one dominant species in all reactors. The DGGE pattern of the RNA extraction showed that in contrast to the DNA extraction, two species were highly active in all reactors. Band 1 and 2 have been sequenced and showed homology with respectively Geobacter and Desulfomonile limimaris. It was found that the range-weighted richness (Rr) of the RNA DGGE for all reactors, except the 2 mm graphite granules, was 30 ± 4, the latter was 50. The rate of change of a temporal frame of 6 weeks was 22 ± 5% for all reactors. Finally, the Pareto Lorenz curves were characteristic for microbial communities in which the most fitting species are dominant and present in high numbers, while the majority of species are present in decreasing lower amounts. Due to the elevated concentration of some species and the availability of many others, the communities The parameters obtained from the polarization curves and the ohmic losses as determined during the current interrupt method (R ci ) were analyzed using a statistical pearson correlation analysis. Fig. 4 gives an overview of the scatter plot matrix and the respective Pearson correlation (r) and R-square numbers. The maximum current (I max ) and maximum power output (P max ) were positively correlated (r = 0.90 and R 2 = 0.80). The ohmic losses (R ci ), as determined by the current interrupt method, and the ohmic losses (R slope ), as determined by the slope of the polarization curve, were also positively correlated (r = 0.89 and R 2 = 0.79) but the R slope values were systematically higher compared to R ci. The type of losses which correlated best with the maximum current and power density was the inverse of the optimal resistance (1/R opt ) with a Pearson correlation of respectively, r = 0.94 (R 2 = 0.89) and r = 0.92 (R 2 = 0.84). The correlations between R opt at the one hand and R slope and R ci at the other hand, were less apparent (data not shown). Moreover, excluding the results of the 2 mm graphite granules, the values of R ci only dropped with a factor of 1.4 over time, while the values of R slope and especially R opt notably decreased over time with a factor of 2.4 and 4.2, respectively. In case of the 2 mm graphite granules, the value of these three parameters was similar and remained the same over time. 4. Discussion 4.1. Three-dimensional electrode materials enabled high power output The use of three-dimensional electrode materials has represented an important development for the electricity generation of microbial fuel cells (Rabaey et al., 2005a; Sell et al., 1989). This study showed that various three-dimensional electrode materials enabled a high maximum power output with values amounting up to 386 W m 3 TAC. Although a non-sustainable hexacyanoferrate cathode was used, these values indicate that the activity of the biocatalyst in the anodes was not limiting to generate these current and power outputs. Moreover, the use of a hexacyanoferrate cathode enabled an equal cathode performance. Compared to previous studies which have used three-dimensional electrodes in combination with a hexacyanoferrate cathode (Aelterman et al., 2006; He et al., 2006; Rabaey et al., 2005a), the results in this study are amongst the highest reported. Interestingly, the differences between the continuous electricity generation of the various materials were low, except for the 2 mm graphite granules which had a lower performance. This was not only reflected in a lower power and current generation, but also in ohmic loss (R ci ) and internal resistance (R opt ) values which were double as high compared to the other reactors. The reason for the higher ohmic loss might be due to the higher amount of contact points between the small granules which imposed a higher resistance for the electron transport. Also, clogging of the tubing was observed due to a wash out of these small granules. This imposed additional stress on the biocatalyst which could result in a lower activity of the biocatalyst. These results confirm that both carbon and graphite matrices are well suited for bacterial adhesion and electron transfer Attuning the external resistance and loading rate This study showed that increasing the loading rate needs to be accompanied by a decrease of the external resistance in order to increase the continuous current generation. The applied external resistance controls the ratio between the cell voltage, determined

6 P. Aelterman et al. / Bioresource Technology 99 (2008) Fig. 2. Forward sweep of the polarization curves of the five reactors with graphite felt, carbon felt, 2 mm graphite granules, 5 mm graphite granules and graphite wool electrodes recorded during five periods with subsequent changes of the volumetric loading rate and external resistance (see legend and Table 1 for details about the operational conditions during the different periods). Table 2 Overview of the biomass concentrations (g VSS m 3 TAC) in function of time Period Sample time (day) Graphite felt Carbon felt 2 mm Graphite granules 5 mm Graphite granules Graphite wool ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 37 The experimental parameters corresponding to the different periods are summarized in Table 1. by the difference between the cathode potential and the anode potential, and the amount of electrons per unit of time which flow through the systems. The Gibbs free energy which is available for the microorganisms during substrate oxidation is proportional to the number of electrons transferred to the electrode and the potential difference between the anode potential and the redox potential of the substrate (Thauer et al., 1977). As a result, the higher the anode potential and the higher the current, the more energy the bacteria theoretically gain per unit of time (Aelterman et al., 2008a; Schroder, 2007). Interestingly, this drive would result in an increase of the current generation and a lowering of the cell voltage by the bacteria, both influencing the power generation. At the

7 8900 P. Aelterman et al. / Bioresource Technology 99 (2008) Fig. 3. DGGE patterns of the RNA and DNA sequences from the microbial communities of the five electrode materials during period 5 and period 8 (see Table 1 for details about the operational conditions). Fig. 4. The scatter plot matrix showing the correlation between the maximum current density (I max ), the maximum power density (P max ), the ohmic losses (R ci ), as determined by the current interrupt method, and the ohmic losses (R slope ), as determined by the slope of the polarization curve, and the inverse of the internal resistance (1/R opt ). The respective Pearson correlation (r) and R-square numbers are mentioned in the scatter plots.

8 P. Aelterman et al. / Bioresource Technology 99 (2008) highest external resistance (50 X), feeding the biomass with an increased loading rate did not result in a rise of the continuous current generation. The amount of energy which was available for growth and maintenance, as determined by the external resistance, was probably too low to sustain a higher metabolic activity of the bacteria. Therefore, no increase of the current generation at higher loading rates could be noted. Only during polarization or at lower external resistances, when higher electron fluxes and lower anode potentials were allowed, enabling a higher energy gain for the bacteria, significant increases of the current generation and power outputs were noted. These results indicate that the loading rate needs to be accompanied by an attuned external resistance which enables the biocatalyst to take advantage of the higher substrate dosage and convert it into current. Moreover, lowering the external resistance will not necessarily result in a lower power output as these result show that both the continuous current generation, power generation and the coulombic efficiency increased at lower external resistances. In general, when the external resistance approaches the internal resistance of the MFC, the increased current will be accompanied with a higher power output Relating the parameters expressing the losses in MFCs Several methods exist to determine the losses occurring in MFCs. The current interrupt method has been described to assess the ohmic losses in a MFC (Liang et al., 2007; Logan et al., 2006). In addition, the slope of the linear ohmic part of a polarization curve should be proportional to the ohmic losses (Logan et al., 2006). The internal resistance can be assessed by taking the ratio of the voltage and current occurring at maximum power generation, this is called the optimal resistance (R opt ) and this value should be equal to the internal resistance (Logan et al., 2006). The use of electrochemical impedance spectroscopy (EIS), is rapidly evolving in MFC research as an alternative method to distinguish the different losses (He et al., 2006). Due to the large amount of polarization curves and current interruption measurements available in this study, correlations were made to further improve the knowledge and understanding of these different losses. These correlations showed that the representatives for the ohmic losses, R slope and R ci were linearly correlated. However the values of the R ci were typically lower compared to R slope values, probably because the latter also included the activation losses which are not measured during the current interrupt measurement. Moreover, the values of R slope and R opt, which in addition to the ohmic losses can also include some overpotential losses, decreased substantially more over time compared to the R ci values. The increase of the biomass concentration could have resulted in a better interaction of the biocatalyst with the anode thereby lowering the overpotential losses during the conversion of substrate and the transfer of electrons to the electrode. In case of the 2 mm graphite granules, only minor differences were noted between the different losses, indicating that the biocatalyst did not improve over time Microbial diversity of the mixed community A higher number of dense bands were observed in the RNAbased DGGE compared to the DNA-based DGGE. This indicates that the MFC environment was suited for the growth and survival of a large variety of microorganisms and did not evolve to a mono-culture. This is supported by Marzorati et al. (2008) who suggested that the range-weighted richness (Rr) is an indication of the carrying capacity of a given environment. Rr values higher than 30 represent an environment with a good carrying capacity and this was the case for all electrode materials examined in this work. The results of the two other ecological parameters (dynamics and Pareto Lorenz curves) are characteristic for microbial communities which are equipped to be sustainable. These levels ensure a slow to moderate succession from one functional community to a new functional one (Marzorati et al., 2008). Many dominant organisms have been found in mixed communities which were enriched in MFCs, but a typical electricity generating microbial community has not been established yet (Aelterman et al., 2008b). A species corresponding to Geobacter spp. was highly active in our systems. Geobacter spp. has been investigated intensively for its electron-transferring properties (Reguera et al., 2005; Reguera et al., 2007). It was found that both axenic Geobacter cultures and enriched MFC microbial communities with the presence of Geobacter spp. are able to generate electricity (Aelterman et al., 2006; Bond and Lovley, 2003; Logan et al., 2005). However, the dominance of Geobacter spp. has not been found to be obligatory to support high electrical power outputs using mixed microbial communities. Moreover, in our system it was accompanied by a second highly active dominant species. This species was related to D. limimaris and has been described for its halorespiring and dechlorinating properties (Sun et al., 2001) but has, to our knowledge, not yet been reported in MFCs Optimizing the biomass distribution to improve the electricity generation In a previous study, using the same MFC design and a packed bed of 5 mm graphite granules as anode, the highest power output during polarization was 106 W m 3 TAC (corresponding to 275 W m 3 NAC) (Aelterman et al., 2006). In this study the packed bed granules enabled to generate a maximum power output of 175 ± 16 W m 3 TAC, which is almost 2 times higher. Next to a possible further enrichment of the bacterial culture, the optimization of the flow distribution and the increased loading rate are believed to cause this increase. The latter might have resulted in the growth of bacteria on a larger part of the electrode thereby increasing the total catalytic activity of the anode. Still, the bacterial distribution was not believed to be highly uniform because the high variability of the phospholipids analysis performed on the different sections of the electrodes. Intensively mixing the electrode material before the analysis, lowered the variability of the analysis (Aelterman et al., 2008a). Considering this variability, determining the biomass concentrations on the surface of large electrode systems must be done carefully. 5. Conclusions We have shown that various three-dimensional electrodes yielded a high continuous electricity generation in MFCs up to 104 W m 3. No biomass or particles accumulation was observed, but graphite and carbon felts, not withstanding their excellent performance, might pose clogging problems when using wastewaters containing colloids or suspended particles. A steady increase of the current generation over time was noted from 59 to 253 A m 3 over a period of 90 days. However, an increase of the loading rate only resulted in a higher current generation when the external resistance was kept low (up to 10.5 X) or during polarization. Therefore, to effectively increase the current generation by increasing the loading rate, the external resistance should be as low as possible. At the lowest is can be equal to the internal resistance of the MFC reactor. Acknowledgments This research was funded by the Research Foundation Flanders by the FWO Project G The samples provided by

9 8902 P. Aelterman et al. / Bioresource Technology 99 (2008) Thorsten Bueschens, the practical help of Petra Vandamme and the useful comments of Peter Clauwaert and Liesje De Schamphelaire were highly appreciated. References Aelterman, P., Freguia, S., Keller, J., Verstraete, W., Rabaey, K., 2008a. The anode potential regulates bacterial activity in microbial fuel cells. Appl. Microbiol. Biotechnol. 78, Aelterman, P., Rabaey, K., De Schamphelaire, L., Clauwaert, P., Boon, N., Verstraete, W., 2008b. Microbial Fuel Cells as an Engineered Ecosystem. In: J.D. Wall, C.S. Harwood, A.L. Demain (Eds.), Bioenergy. ASM. Aelterman, P., Rabaey, K., Pham, H.T., Boon, N., Verstraete, W., Continuous electricity generation at high voltages and currents using stacked microbial fuel cells. Environ. Sci. Technol. 40, Allen, R.M., Bennetto, H.P., Microbial fuel-cells electricity production from carbohydrates. Appl. Biochem. Biotechnol. 39, Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J., Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, Bond, D.R., Lovley, D.R., Electricity production by Geobacter sulfurreducens attached to electrodes. Appl. Environ. Microbiol. 69, Boon, N., Top, E.M., Verstraete, W., Siciliano, S.D., Bioaugmentation as a tool to protect the structure and function of an activated-sludge microbial community against a 3-chloroaniline shock load. Appl. Environ. Microbiol. 69, Cheng, S., Logan, B.E., Ammonia treatment of carbon cloth anodes to enhance power generation of microbial fuel cells. Electrochem. Commun. 9, Clauwaert, P., Van der Ha, D., Boon, N., Verbeken, K., Verhaege, M., Rabaey, K., Verstraete, W., Open air biocathode enables effective electricity generation with microbial fuel cells. Environ. Sci. Technol. 41, Fan, Y.Z., Hu, H.Q., Liu, H., Enhanced Coulombic efficiency and power density of air-cathode microbial fuel cells with an improved cell configuration. J. Power Sources 171, Findlay, R.H., King, G.M., Watling, L., Efficacy of phospholipid analysis in determining microbial biomass in sediments. Appl. Environ. Microbiol. 55, Freguia, S., Rabaey, K., Yuan, Z., Keller, J., Non-catalyzed cathodic oxygen reduction at graphite granules in microbial fuel cells. Electrochim. Acta 53, He, Z., Minteer, S.D., Angenent, L.T., Electricity generation from artificial wastewater using an upflow microbial fuel cell. Environ. Sci. Technol. 39, He, Z., Wagner, N., Minteer, S.D., Angenent, L.T., An upflow microbial fuel cell with an interior cathode: Assessment of the internal resistance by impedance spectroscopy. Environ. Sci. Technol. 40, Hoogers, G., Fuel Cell Technology Handbook. CRC, Boca Raton. Liang, P., Huang, X., Fan, M.Z., Cao, X.X., Wang, C., Composition and distribution of internal resistance in three types of microbial fuel cells. Appl. Microbiol. Biotechnol. 77, Logan, B.E., Cheng, S., Watson, V., Estadt, G., Graphite fiber brush anodes for increased power production in air-cathode microbial fuel cells. Environ. Sci. Technol. 20, 567. Logan, B.E., Hamelers, B., Rozendal, R., Schrorder, U., Keller, J., Freguia, S., Aelterman, P., Verstraete, W., Rabaey, K., Microbial fuel cells: methodology and technology. Environ. Sci. Technol. 40, Logan, B.E., Murano, C., Scott, K., Gray, N.D., Head, I.M., Electricity generation from cysteine in a microbial fuel cell. Water Res. 39, Logan, B.E., Regan, J.M., Electricity-producing bacterial communities in microbial fuel cells. Trends Microbiol. 14, Marzorati, M., Wittebolle, L., Boon, N., Daffonchio, D., Verstraete, W., How to get more out of molecular fingerprints: practical tools for microbial ecology. Environ. Microbiol. doi: /j x. Moon, H., Chang, I.S., Kim, B.H., Continuous electricity production from artificial wastewater using a mediator-less microbial fuel cell. Bioresour. Technol. 97, Morozan, A., Stamatin, I., Stamatin, L., Dumitru, A., Scott, K., Carbon electrodes for microbial fuel cells. Journal of Optoelectronics and Advanced Materials 9, Park, D.H., Zeikus, J.G., Impact of electrode composition on electricity generation in a single-compartment fuel cell using Shewanella putrefaciens. Appl. Microbiol. Biotechnol. 59, Park, D.H., Zeikus, J.G., Improved fuel cell and electrode designs for producing electricity from microbial degradation. Biotechnol. Bioeng. 81, Porges, N., Jasewicz, L., Hoover, S.R., Aerobic treatment of dairy wastes. Appl. Microbiol. 1, Rabaey, K., Clauwaert, P., Aelterman, P., Verstraete, W., 2005a. Tubular microbial fuel cells for efficient electricity generation. Environ. Sci. Technol. 39, Rabaey, K., Lissens, G., Siciliano, S.D., Verstraete, W., A microbial fuel cell capable of converting glucose to electricity at high rate and efficiency. Biotechnol. Lett. 25, Rabaey, K., Ossieur, W., Verhaege, M., Verstraete, W., 2005b. Continuous microbial fuel cells convert carbohydrates to electricity. Water Sci. Technol. 52, Reguera, G., McCarthy, K.D., Mehta, T., Nicoll, J.S., Tuominen, M.T., Lovley, D.R., Extracellular electron transfer via microbial nanowires. Nature 435, Reguera, G., Pollina, R.B., Nicoll, J.S., Lovley, D.R., Possible nonconductive role of Geobacter sulfurreducens pilus nanowires in biofilm formation. J. Bacteriol. 189, Schroder, U., Anodic electron transfer mechanisms in microbial fuel cells and their energy efficiency. Phys. Chem. Chem. Phys. 9, Scott, K., Rimbu, G.A., Katuri, K.P., Prasad, K.K., Head, I.M., Application of modified carbon anodes in microbial fuel cells. Process Saf. Environ. Prot. 85, Sell, D., Kramer, P., Kreysa, G., Use of an oxygen gas-diffusion cathode and a 3- dimensional packed-bed anode in a bioelectrochemical fuel-cell. Appl. Microbiol. Biotechnol. 31, Sun, B.L., Cole, J.R., Tiedje, J.M., Desulfomonile limimaris sp. nov., an anaerobic dehalogenating bacterium from marine sediments. Int. J. Syst. Evol. Microbiol. 51, Ter Heijne, A., Hamers, H.V.M., Buisman, C.J.N., Microbial fuel cell operation with continuous biological ferrous iron oxidation of the catholyte. Environ. Sci. Technol. 41, Thauer, R.K., Jungermann, K., Decker, K., Energy-conservation in chemotropic anaerobic bacteria. Bacteriol. Rev. 41, Zhao, F., Harnisch, F., Schroder, U., Scholz, F., Bogdanoff, P., Herrmann, I., Application of pyrolysed iron(ii) phthalocyanine and CoTMPP based oxygen reduction catalysts as cathode materials in microbial fuel cells. Electrochem. Commun. 7, Zuo, Y., Cheng, S., Call, D., Logan, B.E., Tubular membrane cathodes for scalable power generation in microbial fuel cells. Environ. Sci. Technol. 41,