Ao

Ya

b

a

ARAA

KMPPE

1

ipt(pvtToobDst

(i

0d

Biosensors and Bioelectronics 28 (2011) 71– 76

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

j our na l ho me page: www.elsev ier .com/ locate /b ios

daptation to high current using low external resistances eliminates powervershoot in microbial fuel cells

iying Honga, Douglas F. Call a, Craig M. Wernerb, Bruce E. Logana,∗

Department of Civil and Environmental Engineering, Pennsylvania State University, University Park, PA 16802, USAWater Desalination and Reuse Center, King Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabia

r t i c l e i n f o

rticle history:eceived 26 May 2011ccepted 30 June 2011vailable online 23 July 2011

eywords:icrobial fuel cell

ower overshootolarizationxternal resistance

a b s t r a c t

One form of power overshoot commonly observed with mixed culture microbial fuel cells (MFCs) isdoubling back of the power density curve at higher current densities, but the reasons for this type ofovershoot have not been well explored. To investigate this, MFCs were acclimated to different externalresistances, producing a range of anode potentials and current densities. Power overshoot was observedfor reactors acclimated to higher (500 and 5000 �) but not lower (5 and 50 �) resistances. Acclimation ofthe high external resistance reactors for a few cycles to low external resistance (5 �), and therefore highercurrent densities, eliminated power overshoot. MFCs initially acclimated to low external resistancesexhibited both higher current in cyclic voltammograms (CVs) and higher levels of redox activity over abroader range of anode potentials (−0.4 to 0 V; vs. a Ag/AgCl electrode) based on first derivative cyclic

voltammetry (DCV) plots. Reactors acclimated to higher external resistances produced lower current inCVs, exhibited lower redox activity over a narrower anode potential range (−0.4 to −0.2 V vs. Ag/AgCl),and failed to produce higher currents above ∼−0.3 V (vs. Ag/AgCl). After the higher resistance reactorswere acclimated to the lowest resistance they also exhibited similar CV and DCV profiles. Our findingsshow that to avoid overshoot, prior to the polarization and power density tests the anode biofilm mustadapt to low external resistances to be capable of higher currents.

. Introduction

Two types of power overshoot are commonly observed in polar-zation measurements of microbial fuel cells (MFCs). The first typeroduces maximum power densities that are much higher thanhose that can be sustained at a fixed resistance over long timesVelasquez-Orta et al., 2009). This overestimation of maximumower (Type M) results from either a scan rate using linear sweepoltammetry (LSV) that is too fast, or a lack of sufficient time forhe biofilm to acclimate when resistances are changed. As a result,ype M overshoot is easily controlled by lowering the scan rater by increasing the time at a fixed resistance. The second typef power overshoot occurs when the power density curve doublesack (Type D) unexpectedly toward lower current densities. Type

power overshoot results in an underestimation of the best pos-ible performance of the reactors at higher current densities, andhe reason for this overshoot remains a concern in MFC studies.

Although power overshoot is widely encountered in MFCsIeropoulos et al., 2010), Type D overshoot has only been discussedn a limited number of mixed culture studies (Ieropoulos et al.,

∗ Corresponding author. Tel.: +1 814 863 7908; fax: +1 814 863 7908.E-mail address: blogan@psu.edu (B.E. Logan).

956-5663/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2011.06.045

© 2011 Elsevier B.V. All rights reserved.

2010; Nien et al., 2011), and in many studies no data are providedin the region where overshoot might occur in the power densitycurve (Cheng and Logan, 2011; Katuri et al., 2011; Zhang et al.,2010). Power overshoot has not previously been reported for purecultures, such as Geobacter sulfurreducens or Shewanella oneiden-sis, although in some studies data is also not reported beyond themaximum power density (Wei et al., 2010). Thus, the problem ofpower overshoot appears to be associated with studies using mixedcultures. It is known that both types of power overshoot are associ-ated with the anode and not due to cathode performance (Watsonand Logan, 2011; Winfield et al., 2011). The reasons for overshoothave been attributed to a variety of mechanisms, including masstransport limitations (Aelterman et al., 2006), electrical and ionicdepletion on the anode at low external resistances (Ieropoulos et al.,2010), substrate utilization (Nien et al., 2011), age of the anodebiofilm, extreme operating conditions (i.e. low pH) (Winfield et al.,2011), and insufficient acclimation time for microbes to adjustto a new resistance or a rapid change of voltage in an LSV scan(Watson and Logan, 2011). Mass transport limitations have beenruled out by others because stirring the anode chamber did not

prevent overshoot (Min et al., 2008; Nien et al., 2011). Electricaland ionic depletion is unlikely to be a factor as power overshootcan be eliminated without changes in solution chemistry. Type Dpower overshoot cannot be eliminated by long biofilm enrichment

7 d Bioe

taialkgc

buuMbtdpaAiwcstdw(ptt

etbcacSbdectnel

sttdMr2sc0iuhtceeds

2 Y. Hong et al. / Biosensors an

imes (up to 100 days), using commonly reported sampling times at fixed resistance (typically 15–30 min), or slow scan rates (1 mV/s),ndicating that maturity of the biofilm is not the main issue (Watsonnd Logan, 2011). Type D overshoot can be avoided by using veryong times at fixed resistances (a day or more). This suggests thatey changes occur in the capabilities of the biofilm for currenteneration, but it has not previously been understood what thesehanges might be or how to identify them.

We hypothesized that Type D overshoot could be eliminatedy first adapting an anode biofilm to high current densities bysing a low external resistance. MFCs are typically started upsing a high external resistance (e.g. 1000 �) (Luo et al., 2011;ichaelidou et al., 2011; Zhang et al., 2009), with the doubling

ack of Type D overshoot never occurring at resistances higherhan those used to acclimate the reactors. When resistances areecreased in a polarization test, two changes occur: the anodeotential becomes more positive, and the current increases. Thenode potential rises from values typically around −0.4 V (vs.g/AgCl) at 1000 � to more positive values, with an erratic decrease

n current density often observed at the same anode potentialhere Type D overshoot occurs. Overshoot may therefore be a

onsequence of the increase in anode potential to values out-ide those that can be used by the terminal electron donors ofhe microbes. The optimum mid-point anode potential of a pre-ominant high-current producing bacterium (G. sulfurreducens)as found to be −0.15 V vs. a normal hydrogen electrode (NHE)

Armstrong, 2005; Marsili et al., 2008; Srikanth et al., 2008). Aolarization test can drive the anode potential to be much higherhan this optimum, likely resulting in sub-optimal current produc-ion.

Overshoot might also be due to exoelectrogens lacking redoxnzymes that are only turned on during high current produc-ion. At lower current densities (0.1–0.4 A/m2-anode), pure cultureiofilms of G. sulfurreducens express different outer membraneytochromes compared to higher current density (2.0–4.5 A/m2-node) biofilms. Deleting the gene coding of the cytochrome for lowurrent production in this bacterium (outer membrane cytochrome; omcS) does not inhibit current production in high current densityiofilms. Conversely, deleting the gene for the high current pro-ucing cytochrome (omcZ) prevents current production (Holmest al., 2006; Nevin et al., 2009). Pre-acclimating bacteria to lowurrent using high external resistances and then decreasing resis-ances during MFC polarization over a period of a few minutes mayot allow enough time for bacteria, such as G. sulfurreducens, toxpress the redox enzymes needed for high current production atower resistances.

To test the hypothesis that pre-acclimation to high current den-ities could eliminate Type D overshoot, we acclimated MFC anodeso a range of external resistances and monitored current densi-ies and anode potentials. Previous studies have shown that settingifferent external resistances affects the polarization behavior ofFCs, but the use of different resistances does not necessarily

esult in power overshoot (Jung and Regan, 2011; Lyon et al.,010; Rismani-Yazdi et al., 2011). However, the MFCs used in thesetudies had high internal resistances and produced relatively lowurrent densities (0.04–0.28 A/m2 in Jung and Regan, 2011; up to.6 A/m2 in Rismani-Yazdi et al., 2011) compared to those obtained

n air-cathode MFCs using high surface area anodes. Therefore, wesed high surface area carbon brush anodes in order to obtain theighest current densities possible in our system. It is shown herehat Type D power overshoot can occur in MFCs acclimated to lowerurrent densities (higher external resistances), and that it can be

liminated by acclimating these reactors for a brief time to a lowerxternal resistance and thus higher current densities. We furtheremonstrate that first derivative cyclic voltammetry (DCV) analy-is can be used to understand how the redox activity of the biofilm

lectronics 28 (2011) 71– 76

improves at a broader range of anode potentials once acclimatedto lower resistances.

2. Materials and methods

2.1. MFC reactor construction and operation

Cube-shaped MFCs with a cylindrical chamber (28 mL, 7 cm2

cross section) were constructed without a membrane as previouslydescribed (Liu and Logan, 2004). The brush anode was constructedfrom carbon fibers (PANEX®33 160 K, ZOLTE) wound into a titaniumwire core (2.5 cm long × 2.5 cm diameter, 0.22 m2 surface area),heat treated at 450 ◦C for 30 min (Feng et al., 2010), and placedhorizontally in the center of the chamber. Air cathodes (projectedsurface area of 7 cm2) were made from carbon cloth (30 wt.% wetproof, Fuelcellearth, #CC640WP30) with four PTFE diffusion layersand 0.5 mg-Pt cm−2 (Cheng et al., 2006). The electrode spacing was2.5 cm (center of the anode to the surface of the cathode).

Eight MFCs were inoculated using primary clarifier effluentcollected from the Pennsylvania State University waste watertreatment plant, covered with foil to exclude light, and oper-ated in a temperature controlled room (30 ◦C). The medium wasa 50 mM phosphate buffer solution (PBS) containing (g L−1): 4.28Na2HPO4, 2.45 NaH2PO4·H2O, 0.31 NH4Cl, 0.13 KCl (pH = 7, con-ductivity = 8.8 mS/cm), vitamins and minerals (Cheng et al., 2009)and 1 g L−1 sodium acetate. The reactors were connected to 5, 50,500 or 5000 � resistors, and all tests were conducted in duplicate.

2.2. Analysis

The voltages across the resistors were recorded every 20 minusing a multimeter data acquisition system (model 2700 Keith-ley Instruments, Cleveland, OH). Polarization and power densitycurves were obtained using linear sweep voltammetry (LSV) andvariable resistance (VR) methods. Before the analysis, the reactorswere completely emptied, refilled with fresh growth medium con-taining 1 g L−1 sodium acetate and left at open circuit conditionsfor 1 h. LSVs were conducted from open circuit voltage (OCV) andterminated at 0 V vs. the cathode at a scan rate of 0.2 mV/s (anexample is shown in Fig. S1 in Supplementary data). This scan ratewas selected in order to avoid Type M overshoot typically observedusing higher scan rates (>1 mV/s) (Velasquez-Orta et al., 2009). TheVR polarization data were obtained by changing the external resis-tance (Rex) in the following order: OCV, 5000, 1000, 500, 300, 200,150, 100, 50, 30, 5 � at 20 min intervals, with the voltage recordedusing a digital multimeter (Model 83 III, Fluke) or a multimeter dataacquisition system (model 2700 Keithley Instruments, Cleveland,OH) over a single batch cycle. The VR method required 200 mincompared to 80 min using LSV. Current density normalized to thecathode surface area (IC; mA/cm2) was calculated from IC = E/RAC,where E the measured voltage (V), R is the external resistance (�),and AC is projected cathode surface area (7 cm2). Power densities(PC; mW/cm2) were calculated using PC = IE/AC.

Half cell cyclic voltammetry (CV) was conducted on the anode ofthe MFCs with the cathode as the counter electrode, and an Ag/AgClelectrode (BASi) as the reference located between the anode and thecathode (1 cm to cathode and ∼0.5 cm to anode). The brush anodeand the air cathode were in the same chamber, with the catalystcoating on the air cathode facing the solution. Scans (1 mV/s) werestarted at an initial anode potential (EAN) of −0.7 V and ended at afinal potential of +0.1 V for 3 cycles, with the third cycle used for

CV plots. First derivative CV (DCV) was derived from the CV data byplotting the slope of each CV data point against the voltage (dI/dE).The slope across a data point was calculated using central differencequotient. The DCV data points were fitted by a polynomial of degree

Y. Hong et al. / Biosensors and Bioelectronics 28 (2011) 71– 76 73

FaR

1h

3

3

aaasdt−(

tofwm2pfA5−(5a

twc

Fig. 2. Anode potentials vs. Ag/AgCl recorded over the course of one cycle for each

ig. 1. Current generation of MFCs operated at (A) high external resistance (5000nd 500 �) and (B) low external resistance (50 and 5 �) during the start-up phase.eplicates are shown with dashed and solid lines.

3. All potentials are reported vs. Ag/AgCl (+210 mV vs. a standardydrogen electrode).

. Results

.1. MFC performance

MFCs that were acclimated to higher external resistances (500nd 5000 �) demonstrated reproducible cycles of current gener-tion after 10 days compared to almost 30 days for the reactorst the lower resistances (5 and 50 �) (Fig. 1; potentials over timehown in Fig. S2 in Supplementary data). Anode potentials recordeduring the course of a typical cycle ranged from −0.42 V (5000 �)o −0.38 V (500 �) for the high external resistance reactors and0.30 V (50 �) to −0.24 V (5 �) for low external resistance reactors.

Fig. 2).After operation for two months at a fixed resistance, polariza-

ion curves were obtained using LSV. Type D power overshoot wasbserved for MFCs originally acclimated to 500 and 5000 �, but notor reactors acclimated to 5 and 50 � (Fig. 3A). This is consistentith previous results of power overshoot occurring for MFCs accli-ated to high resistances (Lyon et al., 2010; Rismani-Yazdi et al.,

011; Watson and Logan, 2011). Overshoot was also observed whenolarization data were obtained using the VR method over a singleed-batch cycle at 20 min intervals (Fig. S3 in Supplementary data).t current densities of ca. 0.3 mA/cm2 the anode potential of the000 � acclimated reactors exhibited a sharp increase from ca.0.35 V to −0.25 V with no further increases in current obtained

Fig. 4). This same effect was not observed for MFCs acclimated to �, as current densities of ca. 0.75 mA/cm2 were obtained at annode potential of ca. −0.35 V.

When all reactors were switched to a 5 � external resis-ance and operated for 6 cycles, power overshoot was eliminatedhen polarization curves were performed using LSV (Fig. 3B). The

urrent continued to increase as the anode potential increased

reactor at varied external resistance. Replicates are shown with dashed and solidlines.

above −0.3 V for MFCs originally acclimated to the higher resis-tances. All MFCs produced equivalent maximum power densitiesof 1140 ± 40 mW/m2 (Fig. 3B). This shows that acclimating reac-tors to the low external resistances eliminated power overshootand allowed the biofilm to achieve increased current densities asthe anode potential increased. There is some variation in the powerdensity curves for these reactors due to the complexity of the bac-terial communities, but the variation does not correspond to theoriginal resistance used. The similarity of all polarization curvesimplies that the initial external resistance did not impact the sub-sequent performance of the reactors if the anode biofilms wereadapted to a low resistance. A subsequent acclimation of all MFCsto 5000 � for 1 cycle (2.5 day reaction time) once again producedpower overshoot in all reactors that was similar to that previouslyobserved with reactors originally acclimated to 5000 � (Fig. 3C).

3.2. CV analysis

CV analysis of the anodes showed that the response of the anodebiofilms was significantly altered through acclimation of all reac-tors to the 5 � resistance (Fig. 5A). Lower peak current production(0.5–3 mA) was observed for anodes acclimated to higher resis-tances whereas higher peak currents (6–14 mA) were obtainedwith anodes acclimated to lower resistances. Switching all reac-tors to 5 � resulted in shifting the peak current of the reactorsoriginally acclimated to high resistances to higher values of ca.5–11 mA (Fig. 5B). This increase in current is consistent with polar-ization data that acclimation of the biofilm to lower resistancesimproved performance. CV scans (Fig. 5B) of the MFCs after accli-mation to the 5 � resistance were not all identical, but variationsamong replicates were larger than differences based on the originalresistance.

First derivative analysis of the CV data provided evidence ofchanges in the biofilm redox activity before and after acclimationof all reactors to the 5 � resistance. MFCs originally acclimated tothe higher resistances (500 and 5000 �) exhibited redox activityover a relatively narrow anode potential range of −0.4 to −0.2 V(Fig. 6A). However, MFCs acclimated to the lower resistances (5and 50 �) exhibited higher levels of redox activity over a broaderrange of anode potentials (−0.4 to 0 V) (Fig. 6B). This broader rangeof activity for the low resistance adapted reactors explains why

the anodes did not exhibit power overshoot as the anode poten-tials became more positive due to the capacity for redox activity athigher potentials. In contrast, the reactors acclimated to high resis-

74 Y. Hong et al. / Biosensors and Bioelectronics 28 (2011) 71– 76

Fig. 3. MFC power density curves. (A) Reactors originally acclimated to differenteaa

tw

o(catag

4

pDr

Fig. 4. Effect of external resistance on the polarization property of MFCs after

study (Katuri et al., 2010). However, when the MFCs were subse-quently acclimated to a lower resistance of 5 �, both the redoxintensity and the range over which the anode was active became

xternal resistances for two months. (B) All reactors switched to 5 � for six cycles,fter which (C) the reactors were all switched to 5000 � for one cycle. Replicatesre shown with dashed and solid lines.

ances had a lack of redox activity at more positive anode potentialshere higher current production was needed.

After switching all reactors to 5 �, redox activities shown by theriginal low external resistance reactors were observed in all MFCsFig. 6C). This result explains our findings that current densitiesontinued to increase as the anode potential increased in reactorscclimated to the lower resistances (resulting in no overshoot), andhat polarization data was similar for these MFCs (Fig. 3B). Oncecclimated to the low resistances, the biofilms could sustain currenteneration even at higher anode potentials.

. Discussion

Acclimating MFCs to a low external resistance (5 �) prior toerforming polarization curves eliminated power overshoot. TheCVs obtained here for mixed cultures acclimated to high external

esistances (500 or 5000 �) showed relatively low redox activity

two months of operation. Open (anode) and closed (cathode) symbols representelectrode potentials recorded vs. Ag/AgCl respectively. Replicates are shown withdashed and solid lines.

in a narrow anode potential range of −0.4 to −0.2 V, and theseMFCs exhibited power overshoot above an anode potential of about−0.3 V. These potentials are similar to the average peak potentialsidentified for G. sulfurreducens of ∼−0.36 V (−0.15 V vs. NHE) usingDCV (Marsili et al., 2008; Srikanth et al., 2008), ∼−0.32 V (−0.105 Vvs. NHE) using differential pulse voltammetry (Marsili et al., 2008),and peaks of −0.45 and −0.39 V using DCV reported in another

Fig. 5. Cyclic voltammetry of the MFCs. (A) Reactors originally acclimated to dif-ferent external resistances for two months. (B) All reactors switched to 5 � for sixcycles. Anode potential reported vs. Ag/AgCl. Replicates are shown with dashed andsolid lines.

Y. Hong et al. / Biosensors and Bioe

Fig. 6. First derivative cyclic voltammetry of the MFCs at 1 mV/s. (A and B) Reactorsota

masfodhep

dctecstce

riginally acclimated to different external resistances for two months. (C) All reac-ors switched to 5 � for six cycles. Anode potential reported vs. Ag/AgCl. Replicatesre shown with dashed and solid lines.

uch broader, spanning a wider potential range of −0.4 to 0 V. Thiscclimation procedure resulted in the elimination of power over-hoot originally observed in polarization and power density curvesor the high resistance adapted reactors obtained using either LSVr variable resistance methods. We interpret these results as evi-ence that the exoelectrogenic biofilm improved its capability forigher current production by either synthesizing more redox activenzymes or activating new enzymes that could function at moreositive anode potentials.

It is not yet clear from these electrochemical studies whether theevelopment of broad peaks in the DCVs is a result of a change in theapabilities of specific bacteria such as G. sulfurreducens, the con-ributions of other exoelectrogenic bacteria in the biofilm, or themergence of other dominant bacteria on the anode that allowsurrent generation over a broader range of potentials. Previous

tudies have shown that acclimation to different external resis-ances will result in differences in the composition of the bacterialommunity (Jung and Regan, 2011; Lyon et al., 2010; Rismani-Yazdit al., 2011). The composition of the microbial communities in our

lectronics 28 (2011) 71– 76 75

reactors was not a focus of this study, but the extent to which com-munity structure may change once resistances are changed needsto be further investigated. However, it may be difficult even usingrDNA based techniques to directly associate changes in the biofilmwith the exoelectrogenic activity of specific microorganisms, espe-cially if the ability of specific bacteria to use different potentialschanges with conditions. Cytochrome profiling of pure cultures,such as G. sulfurreducens, would be helpful to examine whetherdifferential expression of redox enzymes at varied external resis-tances is a contributing factor to low and high current production.It is known that biofilms of G. sulfurreducens up-regulate particularcytochromes when producing higher current (Holmes et al., 2006;Nevin et al., 2009), but this analysis has only been done using setpotential anodes, and it is unknown if the same effect occurs usingvariable resistance MFCs.

The effect of anode potential on overshoot requires furtherinvestigation as this would provide additional insight into underly-ing reasons for power overshoot. Changing the external resistanceof an MFC, as done here, affects both the current and anode poten-tial but anode potentials can be controlled using a potentiostat.There are several studies that have tested G. sulfurreducens at dif-ferent set anode potentials, and taken together these results suggestthat changes in anode potential alter enzyme expression. Wei et al.(2010) found that current densities produced by G. sulfurreducensbiofilms increased, as a result of more biomass produced in propor-tion to the increased current, when an MFC was set at −0.2 V (0 V vs.NHE) compared to a more negative potential (−0.37 V, or −0.16 Vvs. NHE). When the potential was set above −0.2 V (0 V NHE) therewas no increase in current or biomass. Similarly, Marsili et al. (2008)found the respiration rate of G. sulfurreducens was unchanged atpotentials above ∼−0.2 V for cells initially grown at a potential of+0.04 V (0.24 V vs. NHE). These two results indicate that there wasno change in the capabilities of this bacterium to extract energyfrom different potentials or to increase the rate of respiration atpotentials more positive than −0.2 V. However, these results can-not be used to determine if the same or different enzymes wereused at the various potentials.

Busalmen et al. (2008) noticed a shift of formal redox poten-tial of G. sulfurreducens from about −0.08 V (when the anode waspolarized at 0.1 V or 0.4 V vs. NHE) to ca. 0.48 V (after polarizing at0.6 V for 18 h), as well as higher current. The original redox poten-tials were restored by lowering the anode potential back to 0.1 Vfor 18 h. The authors suggested this change in redox potential wasevidence for use of an alternative electron transport pathway at the0.6 V potential (Busalmen et al., 2010). In our experiments here wefound that switching all MFCs to 5000 � for one cycle (2.5 days) sub-sequently resulted in power overshoot in all polarization curves.Similarly, in a previous study it was shown that power overshootwas eliminated by acclimating the MFC to each specific resistancefor only one fed-batch cycle 1–2 days each (Watson and Logan,2011). This rapid response of the mixed cultures to changes inset resistance (as evidenced by the presence or absence of powerovershoot) makes it likely that the performance of the biofilm atdifferent anode potentials encountered during polarization tests isdue to microorganisms already present in the biofilm that adaptto these higher potentials, not the emergence of new microorgan-isms. Additional studies that focus specifically on anode potentialsand power overshoot with pure and mixed cultures will be helpfulto better clarify if these changes are due to microbial adaptation tochanges in potentials.

5. Conclusions

Power overshoot resulted from insufficient acclimation of anexoelectrogenic biofilm to either high current densities or more

7 d Bioe

pperibtnpcifiec

A

ta(

A

t

R

A

AB

B

6 Y. Hong et al. / Biosensors an

ositive anode potentials. Type D power overshoot, where theower density curve doubles back instead of reaching higher lev-ls, disappeared when reactors were adapted to a low externalesistance (5 �). Acclimation resulted in the biofilm exhibitingmproved redox activity over a broader range of anode potentialsased on first derivative analysis of CVs. Additional studies on bothhe role of high current production induced by using low exter-al resistances and the effect of anode potentials on eliminatingower overshoot will provide greater insight into the mechanismsontrolling overshoot. Further analysis of the community changesn the MFCs at different external resistances and cytochrome pro-ling of pure cultures operated under the same conditions may alsolucidate how changes that may occur in exoelectrogenic biofilmsontribute to overshoot prevention.

cknowledgements

This research was supported by Award KUS-I1-003-13 fromhe King Abdullah University of Science and Technology (KAUST)nd a National Science Foundation Graduate Research FellowshipD.F.C.).

ppendix A. Supplementary data

Supplementary data associated with this article can be found, inhe online version, at doi:10.1016/j.bios.2011.06.045.

eferences

elterman, P., Rabaey, K., Pham, T.H., Boon, N., Verstraete, W., 2006. EnvironmentalScience & Technology 40, 3388–3394.

rmstrong, F.A., 2005. Current Opinion in Chemical Biology 9, 110–117.usalmen, J.P., Esteve-Nunez, A., Feliu, J.M., 2008. Environmental Science & Tech-

nology 42, 2445–2450.usalmen, J.P., Esteve-Nunez, A., Berná, A., Feliu, J.M., 2010. Bioelectrochemistry 78,

25–29.

lectronics 28 (2011) 71– 76

Cheng, S., Liu, H., Logan, B.E., 2006. Electrochemistry Communications 8, 489–494.Cheng, S., Xing, D., Call, D.F., Logan, B.E., 2009. Environmental Science & Technology

43, 3953–3958.Cheng, S., Logan, B.E., 2011. Bioresource Technology 102, 4468–4473.Feng, Y., Yang, Q., Wang, X., Logan, B.E., 2010. Journal of Power Sources 195,

1841–1844.Holmes, D.E., Chaudhuri, S.K., Nevin, K.P., Mehta, T., Methé, B.A., Liu, A., Ward, J.E.,

Woodard, T.L., Webster, J., Lovley, D.R., 2006. Environmental Microbiology 8,1805–1815.

Ieropoulos, I., Winfield, J., Greenman, J., 2010. Bioresource Technology 101,3520–3525.

Jung, S., Regan, J.M., 2011. Applied and Environmental Microbiology 77,564–571.

Katuri, K.P., Kavanagh, P., Rengaraj, S., Leech, D.A.n., 2010. Chemical Communications46, 4758–4760.

Katuri, K.P., Scott, K., Head, I.M., Picioreanu, C., Curtis, T.P., 2011. Bioresource Tech-nology 102, 2758–2766.

Liu, H., Logan, B.E., 2004. Environmental Science & Technology 38, 4040–4046.Luo, Y., Zhang, R., Liu, G., Li, J., Qin, B., Li, M., Chen, S., 2011. Bioresource Technology

102, 3827–3832.Lyon, D.Y., Buret, F., Vogel, T.M., Monier, J.-M., 2010. Bioelectrochemistry 78, 2–7.Marsili, E., Rollefson, J.B., Baron, D.B., Hozalski, R.M., Bond, D.R., 2008. Applied and

Environmental Microbiology 74, 7329–7337.Michaelidou, U., ter Heijne, A., Euverink, G.J.W., Hamelers, H.V.M., Stams, A.J.M.,

Geelhoed, J.S., 2011. Applied and Environmental Microbiology 77, 1069–1075.Min, B., Román, Ó., Angelidaki, I., 2008. Biotechnology Letters 30, 1213–1218.Nevin, K.P., Kim, B.-C., Glaven, R.H., Johnson, J.P., Woodard, T.L., Methé, B.A., DiDonato

Jr., R.J., Covalla, S.F., Franks, A.E., Liu, A., Lovley, D.R., 2009. PLoS One 4, e5628.Nien, P.-C., Lee, C.-Y., Ho, K.-C., Adav, S.S., Liu, L., Wang, A., Ren, N., Lee, D.-J., 2011.

Bioresource Technology 102 (7), 4742–4746.Rismani-Yazdi, H., Christy, A.D., Carver, S.M., Yu, Z., Dehority, B.A., Tuovinen, O.H.,

2011. Bioresource Technology 102, 278–283.Srikanth, S., Marsili, E., Flickinger, M.C., Bond, D.R., 2008. Biotechnology and Bioengi-

neering 99, 1065–1073.Velasquez-Orta, S.B., Curtis, T.P., Logan, B.E., 2009. Biotechnology and Bioengineering

103, 1068–1076.Watson, V.J., Logan, B.E., 2011. Electrochemistry Communications 13, 54–56.Wei, J., Liang, P., Cao, X., Huang, X., 2010. Environmental Science & Technology 44,

3187–3191.Winfield, J., Ieropoulos, I., Greenman, J., Dennis, J., 2011. Bioelectrochemistry 81,

22–27.Zhang, C., Li, M., Liu, G., Luo, H., Zhang, R., 2009. Journal of Hazardous Materials 172,

465–471.Zhang, X., Cheng, S., Huang, X., Logan, B.E., 2010. Biosensors and Bioelectronics 25,

1825–1828.