Improved current and power density with a micro-scale microbial fuelcell due to a small characteristic length

Hao Ren a,n, César I. Torres b,c, Prathap Parameswaran b, Bruce E. Rittmann b, Junseok Chae a

a School of Electrical, Computer and Energy Engineering, Arizona State University, Tempe, AZ, USAb Swette Center for Environmental Biotechnology, Biodesign Institute at Arizona State University, Tempe, AZ, USAc School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ, USA

a r t i c l e i n f o

Article history:Received 15 March 2014Received in revised form8 May 2014Accepted 15 May 2014Available online 5 June 2014

Keywords:Micro-scale microbial fuel cellMass transferScaling effectPower density

a b s t r a c t

A microbial fuel cell (MFC) is a bio-electrochemical converter that can extract electricity from biomass bythe catabolic reaction of microorganisms. This work demonstrates the impact of a small characteristiclength in a Geobacteraceae-enriched, micro-scale microbial fuel cell (MFC) that achieved a high powerdensity. The small characteristic length increased the surface-area-to-volume ratio (SAV) and the masstransfer coefficient. Together, these factors made it possible for the 100-mL MFC to achieve among thehighest areal and volumetric power densities – 83 μW/cm2 and 3300 μW/cm3, respectively – among allmicro-scale MFCs to date. Furthermore, the measured Coulombic efficiency (CE) was at least 79%, whichis 2.5-fold greater than the previously reported maximum CE in micro-scale MFCs. The ability to improvethese performance metrics may make micro-scale MFCs attractive for supplying power in sub-100 mWapplications, especially in remote or hazardous conditions, where conventional powering units are hardto establish.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

A microbial fuel cell (MFC) is an electrochemical device toconvert the electron equivalents of organic compounds intoelectrical current. Key to MFCs is that they employ anode-respiring bacteria (ARB) as living catalysts at the anode (Bond etal., 2002; Chaudhuri and Lovley 2003). Unlike other techniques toconvert organic material to electricity, the MFC is unique becauseit allows direct electricity generation (Rabaey and Verstraete2005). The organic fuel for MFCs most often comes from waste-water, making the MFC a waste-to-energy technology that haveled efforts to scale up MFCs (Logan 2010, 2006; Rabaey andVerstraete 2005). In addition, MFCs have promise for bioremedia-tion of recalcitrant compounds (Morris and Jin 2008) and supply-ing power for remote sensors in hazardous or hard-to-accessconditions (Ren et al., 2012).

For remote applications, an interesting direction is miniaturiz-ing MFCs: decreasing the characteristic length to the micrometerrange, resulting in chamber volumes of 100s of μL (Ren et al.,2012). Miniaturizing MFCs takes advantages of microfabrication(Choi and Chae, 2009; Lee et al., 2001; Welch et al., 2005; Yangand Chae, 2008), adopted from standard integrated-circuits fabri-cation, which allows small-sized chip-scale devices for powering

ultra-low-power electronics, especially in remote conditions (Choiand Chae, 2009, 2012, 2013; Choi et al., 2011; Ren et al., 2012). Ourprior work (Choi and Chae, 2012) marks the highest areal powerdensity, up to 33 μW/cm2, in a micro-scale MFC, but it is still far lowerthan that of macro-/meso-scale counterparts (up to 680 μW/cm2)(Fan et al., 2008).

We utilize the ARB Geobacteraceae because of their provenability to deliver a high power density. Geobacteraceae oxidizeacetate at the anode to release electrons and protons:

CH3COO� þ2H2O-2CO2þ8e� þ7Hþ ð1aÞ

Acetate first must be transported from the bulk liquid to thebiofilm, and then it diffuses into the biofilm, where it is oxidized.Electrons are conducted to the anode and then pass through anexternal circuit to arrive at the cathode, where they reduce anoxidant. In our case, they reduced ferricyanide to ferrocyanide.

½FeðCNÞ6�3� þe�-½FeðCNÞ6�4� ð1bÞ

Ions also move from the anode to the cathode via the liquidmedium and a proton-exchange membrane (PEM) is to establishcharge neutrality. Among the ions, the most critical one is theproton (Hþ). Eq. (1) shows that 7 mol Hþ are produced per moleacetate; these Hþ must be transported out of the biofilm to avoidacidifying the biofilm, which retards the rate of biological reaction(Franks et al., 2009; Torres et al., 2008). The protons combine withthe basic member of an acid/base buffer, which diffuse out of the

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/bios

Biosensors and Bioelectronics

http://dx.doi.org/10.1016/j.bios.2014.05.0370956-5663/& 2014 Elsevier B.V. All rights reserved.

n Corresponding author.E-mail address: hren12@asu.edu (H. Ren).

Biosensors and Bioelectronics 61 (2014) 587–592

biofilm. Thus, increasing the mass-transport rate of the Hþ-carrying buffer should improve MFC performance.

Here, we present a micro-scale, flat-plate MFC with a highsurface area to volume ratio, which we postulate will enhancemass transfer of acetate and Hþ-carrying buffer. Fig. 1 gives aschematic of the MFC. It has a PEM and two silicone gasketssandwiched between two glass slides pre-fabricated with goldelectrodes. One rectangular hole is patterned in each gasket, andthey define the anode and cathode chambers with identicaldimensions. Two nanoports provide microfluidic pathways foranolyte and catholyte that flows through the correspondingchambers, as shown in Fig. 1 (b). W and L are the lateral dimensionand height of the cross-section of chambers, respectively. L isdesigned to be much smaller than W, resulting in a very highsurface-area-to-volume ratio for each chamber. Furthermore, L isvery small to make the distance between the anode and thecathode small.

Scaling down the characteristic length of a micro-scale MFCought to result in increased mass transport, consequently enhan-cing the MFC's current and power density. When mass transport islimiting, the maximum current density is proportional to the fluxdefined by the diffusion of substrate or products to the electrode,

as defined by

jL ¼ nFDΔCλ

ð2Þ

where n is the number of electrons equivalents corresponding tothe limiting compound, F is the Faraday constant (96,485 C/mole�), D is the normalized diffusivity of the limiting compound [m2/s], ΔC is the concentration gradient of the limiting compound[mol/m3], and λ is the diffusion layer thickness [m]. jL can increaseif λ is minimized in the MFC, thus increasing the possible powerdensity obtained.

A miniaturized MFC may benefit from faster mass transfer dueto its inherently smaller L (Gerlach and Dotzel 2008; Qian andMorse 2011). In a laminar flow, likely in microfluidic environments(Stone et al., 2004), the Reynolds number, Re, and the masstransfer coefficient, kc [m/s], can be defined as (Sherwood et al.,1975):

Re¼ ρvdh=μ and kc ¼ 0:664ðρvdh=μÞ1=2ðμ=ρDÞ1=3ðD=LÞ ð3Þwhere ρ, v, dh, L, μ, and D are the specific density of the fluid[kg/m3], the linear velocity of the fluid [m/s], the hydraulicdiameter of the anode chamber [m], the characteristic length

Fig. 1. Schematic and images of the micro-scale MFC: (a) a lateral view of the MFC after assembly; (b) cross-sectional of (a) via plane A–A0 , the height of the chambers is L andthe width of the chambers is W, blue arrows indicate the direction of anolyte and catholyte; (c) anode chamber defined by the gasket; (d) the different parts of the MFCbefore assembly; (e) Cr/Au electrode on top of a glass slide; 6 holes (one for inlet, one for outlet, and 4 for screws) mechanically drilled through the glass; (f) silicon gasketmounted on top of the Cr/Au electrode to define the anode/cathode chamber; and (g) assembled MFC with microfluidic ports for inlet/outlet for anode/cathode chambers.

H. Ren et al. / Biosensors and Bioelectronics 61 (2014) 587–592588

[m], the viscosity of the fluid [kg/m/s], and the normalizeddiffusivity of the limiting compound (m2/s), respectively. Thehydraulic diameter is calculated by:

dh ¼ 4A=p ð4Þwhere A¼WL and p¼2(WþL) are area [m2] and wetted perimeter[m] of the cross section of the anode chamber, respectively. W andL are the lateral dimension [m] and height [m] of the cross sectionof chamber. L is also the characteristic length of the anodechamber, and the hydraulic diameter (dh) becomes twice thecharacteristic length (L) when the lateral dimension of the anodechamber dominates the height (W4L), which prevails in micro-scale MFCs. Based on Eq. (3), kc should become larger as the linearvelocity (v) increases or the characteristic length (L) decreases.

The diffusion-layer thickness, λ [m], can be calculated by:

λ¼ Dkc

¼ L

0:664ðρvdh=μÞ1=2ðμ=ρDÞ1=3ð5Þ

Because dh becomes 2L when the lateral dimension of the anodechamber dominates the height (W4L), λ is proportional to L1/2;hence, λ decreases as the characteristic length scales down,resulting in smaller diffusion-layer resistance. Fig. 2 illustrateshow scaling down L, with all other parameters unchanged, causesthe mass transfer coefficient to increase and the diffusion layerthickness to decrease at 15 μL/min.

In summary, down-scaling the characteristic length enhancesthe surface area to volume ratio and mass transfer coefficient,consequently improving the power density of a micro-scale MFC.The next section addresses the implementation of a micro-scaleMFC to adopt the scaling effect on the characteristic length,including the fabrication of the micro-scale MFC and the experi-mental procedures. Then, the power density of the micro-scaleMFC is evaluated, demonstrating both scaling effects.

2. Material and methods

2.1. Micro-scale MFC fabrication

MFC fabrication began with mechanically drilling six throughholes on two glass slides (micro slides, 76�38�0.12 cm3, VWR):one inlet, one outlet, and four for assembly. Afterwards, Cr/Au

(20 nm/200 nm) was deposited by sputtering on the anode/cath-ode (Fig. 1(e)). Then, the nanoports (10–32 coned, IDEX Health &Science) and fluidic tubing (PEEK polymer, IDEX Health & Science)were aligned and glued to the inlets/outlets to supply anolyte andcatholyte. Silicone rubber gaskets were patterned to define thechamber and electrode (Fig. 1(f)), and the PEM (Nafion 117, SigmaAldrich) was patterned to have the same dimension as theelectrode. Finally the MFC was assembled with four screw boltsand nuts to minimize oxygen/electrolyte leakage (Fig. 1(g)). Thevolume of each chamber was 100 μL, and the size of electrode was4 cm2.

2.2. Inoculum

The inoculum for the micro-scale MFC was obtained from anacetate-fed microbial electrolysis cell (MEC) that had a Geobacter-aceae-enriched bacterial community originally enriched fromanaerobic-digested sludge from the Mesa Northwest Water Recla-mation Plant (MNWWRP, Mesa, AZ). Clone libraries of the 16S-rRNA gene showed that the biofilm anode was a mixed bacterialculture dominated by Geobacter sulfurreducens (Torres et al., 2009).The anolyte was 25-mM sodium acetate medium with 1680 mgKH2PO4, 12,400 mg Na2HPO4, 1600 mg NaCl, 380 mg NH4Cl, 5 mgEDTA, 30 mg MgSO4 �7H2O, 5 mg MnSO4 �H2O, 10 mg NaCl, 1 mgCo(NO3)2, 1 mg CaCl2, 0.001 mg ZnSO4 �7H2O, 0.001 mg ZnSO4 �7H2O, 0.1 mg CuSO4 �5H2O, 0.1 mg AlK(SO4)2, 0.1 mg H3BO3,0.1 mg Na2MoO4 �2H2O, 0.1 mg Na2SeO3, 0.1 mg Na2WO4 �2H2O,0.2 mg NiCl2 �6H2O, and 1 mg FeSO4 �7H2O (per liter of deionized(DI) water) (pH 7.870.2). For start-up, inoculum and anolyte weremixed with a volume ratio of 1:1. The catholyte was 100-mMpotassium ferricyanide in 100-mM phosphate buffer solution(pH 7.4). The anolyte and catholyte were supplied into the micro-scale MFC using a syringe pump. The MFC operated at 25 1C.

2.3. Operation and analysis

The current was monitored every minute by recording thevoltage drop across an external resistor connected between theanode and the cathode using a data acquisition system (DAQ/68,National Instrument). During start-up, a 148-Ω resistor was used,and the MFC operated at a flow rate of 1 mL/min through eachchamber. Once the start-up was completed, the flow rate, con-trolled by a syringe pump, was increased for a series of steadystates at which polarization measurements were performed. Togenerate polarization curves, we measured voltage using a seriesof resistors from 148 Ω to 932 kΩ. The current through the resistorwas calculated via Ohm's law (I¼U/R, where I, U, and R are current,voltage and resistance, respectively), and the resulting outputpower was calculated via Joule's law (P¼ I2R). The MFC's internalresistance was calculated by linearly fitting the Ohmic region i.e.,when the voltage drop (ΔU) was approximately linear with theincrease of current (ΔI) of each curve of areal current densityversus voltage across external resistor (Logan et al., 2006)): i.e.,R¼ΔU/ΔI.

The minimum Coulombic Efficiency (CE) of the micro-scale MFCwas computed by integrating current profiles over the time (Logan etal., 2006) and dividing this cumulative current (Cp) by the currentthat could have been produced if all the influent acetate had beenoxidized to produce current (CT¼V� b�NA� e� [molsubstrate]): CE¼CP/CT�100%, where V is the volume of anode chamber [m3], b is thenumber of moles of electrons produced by oxidation of substrate[8 mol e�/mol acetate], NA is Avogadro's number [6.023�1023

molecules/mol], e is electron charge [1.6�10–19 C/electron], andmolsubstrate is the moles of acetate oxidized.

Fig. 2. Scaling effect for micro-scale MFCs; moving from L¼1 cm to L¼0.1 μmincreases mass transfer coefficient from 9.1�10�7 to 2.9�10�4 m/s, and lowersdiffusion layer thickness from 9.7�10�4 to 3.1�10�6 m. The values plotted are fora flow rate of 15 mL/min.

H. Ren et al. / Biosensors and Bioelectronics 61 (2014) 587–592 589

3. Results and discussion

It took 6 days to complete the start-up process, at which timethe current density exceeded 20 μA/cm2. After start up, the flowrate was increased from 1 to 15 μL/min to characterize the powerdensity as a function of flow rate. Fig. 3 shows the current densityrecorded at different flow rates using a 148-Ω resistor (blue curve).The current density increased almost linearly with the flow rate upto 12 μL/min, where it saturated. The current density declinedwhen the flow rate was further increased to 17 μL/min due tocathode erosion from ferricyanide etching the gold cathode. SEMimages of the biofilm at the end of the experiments (Fig. S1)showed a range of aggregation patterns on the anode surface, withan average coverage ratio of 30% and a per-cell electro-generationrate of about 106 electrons/s.

A high flow rate at the MFC anode is not always synonymouswith a decreased number of ARB on the anode surface, unless it iscoupled with high shear rate and other advective flow patternsthat promote detachment from biofilm surfaces. Previous studieshave documented the ability of Geobacteraceae cells to accumulateto high cell numbers on the biofilm anode at high flow rates

(or low hydraulic retention times (HRTs)) compared to this study(Lee and Rittmann 2009; Parameswaran et al., 2012).

The voltages for a series of external resistors were recorded forfixed anolyte flow rates. Then, polarization curves (Fig. 4) wereobtained from these data. The curves of the output voltage versuscurrent density, Fig. 4(a), show a high OCV of 0.81 V. The internalresistances were calculated by linearly fitting the Ohmic region (i.e., where the voltage drop (ΔU) is approximately linear with thecurrent density (ΔI)) for each individual polarization curve (Fig. 4(a)). The curves show similar internal resistances of the micro-scale MFC, independent of the flow rate (2.170.6 kΩ). However,the limiting current, jL, changed significantly from 91 mA/cm2 at5 mL/min. to 240 mA/cm2 at 15 mL/min. These experiments confirmthat a transport limitation to or from the bulk liquid limited thetotal current density produced by the Geobacteraceae-enrichedbiofilm. Similar conclusions have been recently reached at usingnuclear magnetic resonance techniques and rotating disk electro-des (Renslow et al., 2013).

The mass-transfer rate is a product of the concentrationdifference and mass transfer coefficient. As a result, mass transferis enhanced when the concentration and linear velocity of electro-lyte are increased. Fig. 4(a) clearly shows current limitation for thelower flow rates, suggesting that the maximum current producedby Geobacteraceae-enriched biofilm was controlled by a diffusiveprocess. In a separate experiment, the total salts concentrations ofthe anolyte and the catholyte, with the exception of the anodebuffer, were increased stepwise from 1� to 1.5� and to 2� . Theconcentration increase did not improve the areal power density(data not shown). This finding confirms that the transport ofsubstrate was not limiting, and the only possible diffusion limita-tion was due to buffer transport at the anode biofilm. Protontransport limitations has been demonstrated previously due to thehigh proportion of Hþ produced by ARB (as shown in Eq. (1))(Franks et al., 2009; Torres et al., 2008); our studies providefurther evidence of such limitations. Our micro-scale MFC providesa design in which such limitations can be minimized by increasingthe flow rate conditions.

The curves of the areal power density as a function of currentdensity for different flow rates, Fig. 4(b), demonstrate that theareal power densities were almost independent of flow rates atlow current densities, but reached a higher jL as the flow rateincreased. The areal power density at 5 μL/min. reached its max-imum power density (44 μW/cm2) at a low current density of71 μA/cm2, whereas that of 15 μL/min. had a maximum power

Fig. 3. Maximum areal current density versus flow rate and mass-transfercoefficient of limiting compound (assuming normalized diffusivity of limitingcompound is 0.76 m2/day) (error bars indicate relative standard deviation (RSD));the maximum areal current density increases proportionally to the mass transfercoefficient. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

Fig. 4. Polarization curves for the MFC at different flow rates (error bars indicate relative standard deviation (RSD)): (a) output voltage versus areal current density; and(b) areal power density versus areal current density.

H. Ren et al. / Biosensors and Bioelectronics 61 (2014) 587–592590

density of 84 μW/cm2 at a high current density of 214 μA/cm2. Thistrend can be attributed to a higher rate of mass-transfer ofdeprotonated buffer (e.g., HPO2�

4 ) into the anode's biofilm, provid-ing more rapid transport of acetate to Geobacteraceae-enrichedbiofilm and Hþ-carrying buffer out of the biofilm (e.g., H2PO4

�)(Torres et al., 2008).

The theoretical laminar mass-transfer coefficient was calculatedfrom Eq. (3), and it is compared in Fig. 3 to the current densityobtained through the polarization curves in Fig. 4. The computedmass transfer coefficient of the limiting compound correlates wellwith the current density for the different flow rates. This furthersupports that the increased current density arose from enhancedmass transfer. We analyzed the computed mass transfer coefficientand found it to be consistent with the measured current andkinetics for acetate utilization; the analysis is shown in SI.

Coulombic efficiency (CE) is a measure of how efficiently anMFC harvests electrons, and low CE has been a challenge formicro-scale MFCs (Choi et al., 2011; Ren et al., 2012). The max-imum CE reported by micro-scale MFCs so far is 31% (Choi et al.,2011), significantly lower than that of macro- and meso-scaleMFCs, which often reach a CE higher than 80% (Rabaey et al.,2004). The low range of reported CE values for micro-MFCs (0.03–31% (Ren et al., 2012)) is believed to be primarily due to highoxygen leakage into the anode chamber (Choi et al., 2011).

Anolyte flow was stopped and current over time domain wasmonitored while the catholyte kept flowing (Fig. S-2) (Choi et al.,2011). The minimum CE was 79%, at least 2.5-fold greater thanpreviously reported CE for micro-scale MFCs; it is comparable tothat of macro-/meso-scale MFCs. Thus, the anode chamber in ourmicro-scale MFC was well isolated from oxygen in the air. Themicro-scale MFCs reported previously had PDMS for their gaskets,which is a common microfabrication material due to its lowexpense. However, PDMS is extremely oxygen permeable (52,53171313 cm3 �mm/m2 �day atm) (Ren et al., 2012). This work eliminatedthe use of PDMS and deployed glass and silicon gasket as alter-natives, which lowered oxygen permeability substantially.

Fig. 5(a) compares the volumetric power density and CE of thiswork with those of reported macro-/meso- and micro-scale MFCs.Our micro-scale MFC obtained 3300 μW/cm3, the highest volu-metric power density for all MFCs, regardless of scale. Fig. 5(b) liststhe key performance parameters of this work compared withthose of prior art. Notable is the high volumetric power density,which is due to the very small characteristic length, resulting in

the high SAV of the micro-scale MFC and relatively fast masstransfer. The corresponding areal power density was 83 mW/cm2,also the highest among all micro-scale MFCs. As noted before, ourmicro-scale MFC achieved a CE of 79%, more than 2.5 times higherthan that of the micro-scale MFCs and is comparable to the CE ofmacro/meso-scale MFCs.

High internal resistance could still be a stumbling block forfurther enhancement of micro-scale MFCs. Our earlier work showedthat the cathode/PEM/electrolyte resistance is low (all less than100 Ω cm2) (Choi et al., 2011); thus, in our current micro-scaleMFC, the anode overpotential seems to dominate the overall poten-tial losses at high current densities due to the limiting transport ofbuffer that counteracts proton accumulation, as shown in Fig. 4.Increasing the linear velocity in micro-scale MFCs can be a simplealternative to maximize current and power densities in MFCs.

The high power density and CE of the micro-scale MFC may findits application in powering sub-100 mW electronics, such as passiveradio frequency identification (RFID) tags, ultra-low power wirelesssensor network and ultra-low power microcontroller unit (MCU),especially for applications in remote or hazardous conditions, whereconventional powering units are hard to establish. Further studiesneed to be performed on autonomous operation, possibly to removethe use of syringe pump. Micro-scale MFCs also may be attractive inspace exploration for power supply and waste treatment.

4. Conclusion

Down-scaling the characteristic length of the micro-scale MFCprofoundly improved current and power densities of the micro-scale MFC by enhancing mass transfer and increasing jL. Themicro-scale MFC demonstrated an areal power density of 83 μW/cm2, a volumetric power density 3300 μW/cm3, and a CE of at least79%. The CE and areal power densities are comparable to those ofmacro-/meso-scale MFCs, and the volumetric power density is therecord high of all MFCs. Future work will focus on implementingthe micro-scale MFC for powering sub-100 mW electronics inremote or hazardous conditions.

Appendix A. Supporting information

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.bios.2014.05.037.

Fig. 5. (a) Comparison of volumetric power density and CE of this work with existing macro-/meso- and micro-scale MFCs(Chiao et al., 2006, 2011; Fan et al., 2012, 2007; Liuet al., 2008; Liu and Logan 2004; Qian et al., 2009; Rabaey et al., 2004; Siu and Chiao 2008; Mink et al., 2012); the CE of this work is substantially higher than those of micro-scale MFCs and is comparable with those of macro-/meso-scale MFCs, and the volumetric power density of this work is highest ever reported, regardless of scale; and(b) specifications and performance of prior micro-scale MFCs and the MFC of this work (based on relative standard deviation (RSD))

H. Ren et al. / Biosensors and Bioelectronics 61 (2014) 587–592 591

References

Bond, D.R., Holmes, D.E., Tender, L.M., Lovley, D.R., 2002. Science 295 (5554),483–485.

Chaudhuri, S.K., Lovley, D.R., 2003. Nat. Biotechnol. 21 (10), 1229–1232.Chiao, M., Lam, K.B., Lin, L.W., 2006. J. Micromech. Microeng. 16 (12), 2547–2553.Choi, S., Chae, J., 2013. Sens. Actuators A: Phys. 195 (1), 206–212.Choi, S., Chae, J., 2009. Biosens. Bioelectron. 25 (2), 527–531.Choi, S., Chae, J., 2012. Sens. Actuators A: Phys. 177 (0), 10–15.Choi, S., Lee, H.S., Yang, Y., Parameswaran, P., Torres, C.I., Rittmann, B.E., Chae, J.,

2011. Lab Chip 11 (6), 1110–1117.Fan, Y., Han, S.-K., Liu, H., 2012. Energy Environ. Sci. 5 (8), 8273–8280.Fan, Y.Z., Hu, H.Q., Liu, H., 2007. J. Power Sources 171 (2), 348–354.Fan, Y.Z., Sharbrough, E., Liu, H., 2008. Environ. Sci. Technol. 42 (21), 8101–8107.Franks, A.E., Nevin, K.P., Jia, H., Izallalen, M., Woodard, T.L., Lovley, D.R., 2009.

Energy Environ. Sci. 2 (1), 113–119.Gerlach, G., Dotzel, W., 2008. Introduction to Microsystem Technology: A Guide for

Students. Wiley.Lee, H.-S., Rittmann, B.E., 2009. Environ. Sci. Technol. 44 (3), 948–954.Lee, S., Nam, G.-J., Chae, J., Kim, H., Drake, A.J., 2001. IEEE Proceedings of Design

Automation Conference, 2001, pp. 852–857.Liu, H., Cheng, S., Huang, L.P., Logan, B.E., 2008. J Power Sources 179 (1), 274–279.Liu, H., Logan, B.E., 2004. Environ. Sci. Technol. 38 (14), 4040–4046.Logan, B.E., 2010. Applied Microbiol. Biotechnol. 85 (6), 1665–1671.

Logan, B.E., Hamelers, B., Rozendal, R.A., Schrorder, U., Keller, J., Freguia, S.,Aelterman, P., Verstraete, W., Rabaey, K., 2006. Environ. Sci. Technol. 40 (17),5181–5192.

Mink, J.E., Rojas, J.P., Logan, B.E., Hussain, M.M., 2012. Nano Lett. 12 (2), 791–795.Morris, J.M., Jin, S., 2008. J. Environ. Sci Health A 43 (1), 18–23.Parameswaran, P., Torres, C.I., Kang, D.-W., Rittmann, B.E., Krajmalnik-Brown, R.,

2012. Water Sci. Technol. 65, 1.Qian, F., Baum, M., Gu, Q., Morse, D.E., 2009. Lab Chip 9 (21), 3076–3081.Qian, F., Morse, D.E., 2011. Trends Biotechnol. 29 (2), 62–69.Rabaey, K., Boon, N., Siciliano, S.D., Verhaege, M., Verstraete, W., 2004. Appl.

Environ. Microb. 70 (9), 5373–5382.Rabaey, K., Verstraete, W., 2005. Trends Biotechnol. 23 (6), 291–298.Ren, H., Lee, H.-S., Chae, J., 2012. Microfluid. Nanofluid. 13 (3), 353–381.Renslow, R., Babauta, J., Majors, P., Beyenal, H., 2013. Energy Environ. Sci. 6 (2),

595–607.Sherwood, T.K., Pigford, R.L., Wilke, C.R., 1975. Mass Transfer. McGraw-Hill, New

York.Siu, C.P.B., Chiao, M., 2008. J. Microelectromech. Syst. 17 (6), 1329–1341.Stone, H.A., Stroock, A.D., Ajdari, A., 2004. Annu. Rev. Fluid Mech. 36, 381–411.Torres, C.I., Kato Marcus, A., Rittmann, B.E., 2008. Biotechnol. Bioeng. 100 (5),

872–881.Torres, C..I., Krajmalnik-Brown, R., Parameswaran, P., Marcus, A.K., Wanger, G.,

Gorby, Y.A., Rittmann, B.E., 2009. Environ. Sci. Technol. 43 (24), 9519–9524.Welch, W.C., Chae, J., Najafi, K., 2005. IEEE Trans. Adv. Packag. 28 (4), 643–649.Yang, Y., Chae, J., 2008. J. Micromech. Microeng. 18, 12.

H. Ren et al. / Biosensors and Bioelectronics 61 (2014) 587–592592