Microfluidic Microbial Fuel Cells for Microstructure Interrogations by Erika Andrea Parra A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Engineering - Mechanical Engineering in the Graduate Division of the University of California, Berkeley Committee in charge: Professor Liwei Lin, Chair Professor Carlos Fendandez-Pello Professor John D. Coates Fall 2010 Microfluidic Microbial Fuel Cells for Microstructure Interrogations Copyright 2010 by Erika Andrea Parra 1 Abstract Microfluidic Microbial Fuel Cells for Microstructure Interrogations by Erika Andrea Parra Doctor of Philosophy in Engineering - Mechanical Engineering University of California, Berkeley Professor Liwei Lin, Chair Thebreakdownoforganicsubstancestoretrieveenergyisanaturallyoccurringprocessin nature. Catabolicmicroorganismscontainenzymescapableofacceleratingthedisintegration of simple sugars and alcohols to produce separated charge in the form of electrons and protons as byproducts that can be harvested extracellularly through an electrochemical cell to produce electrical energy directly. Bioelectrochemical energy is then an appealing green alternative to other power sources. However, a number of fundamental questions must be addressedifthetechnologyistobecomeeconomicallyfeasible. Powerdensitiesarelow, hence the electron flow through the system: bacteria-electrode connectivity, the volumetric limit of catalyst loading, and the rate-limiting step in the system must be understood and optimized. This project investigated the miniaturization of microbial fuel cells to explore the scaling of the biocatalysis and generate a platform to study fundamental microstructure effects. Ultra- micro-electrodes for single cell studies were developed within a microfluidic configuration to quantify these issues and provide insight on the output capacity of microbial fuel cells as well as commercial feasibility as power sources for electronic devices. Several devices were investigated in this work. The first prototype consisted of a gold array anode on a SiO passivation layer that intended to imitate yet simplify the complexity 2 ofa3Dcarbonstructureona2Dplane. UsingGeobactersulfurreducens,anorganismbelieved toutilizedirectelectrontransfertoelectrodes,the1mm2 electrodedemonstratedamaximum current density of 1.4 µA and 120 nW of power after 10 days. In addition, the transient current-voltage responses were analyzed over the bacterial colonization period. The results indicatedthatovera6-dayperiod, thebacteriaincreasedthecapacitanceofthecell5-orders- of-magnitude and decreased the resistance by 3X over the bare electrode. Furthermore, over short experimental scales (hours), the RC constant was maintained but capacitance and resistance were inversely related. As the capacitance result coincides with expected biomass increase over the incubation period, it may be possible for an electrical spectroscopy (impedance) non-invasive technique to be developed to estimate biomass on the electrode. Similarly, the R and C relationship over short experimental scales could be explored further 2 to provide insight on biofilm morphology. Lastly, fluorescence and SEM microscopy were used to observe the biofilm development and demonstrated that, rather than growing at even density, the bacteria nucleated at points on the electrode, and dendritically divided, until joining to form the “dense” biofilm. In addition, viable microorganisms undergoing cell divisionwerefounddozensofmicronsfromelectrodesurfaceswithoutvisiblepiliconnections. To investigate single-cell catalysis or microstructure effects, a sub-micro-liter microflu- idic single-channel MFC with an embedded reference electrode and solid-state final electron acceptor was developed. The system allowed for parallel (16) working ultra-micro-electrodes and was microscopy compatible. With Geobacter sulfurreducens, the semiconducting ITO electrodes demonstrated forward bias behavior and suitability for anodic characterization. The first prototype demonstrated, with 179 cells on the electrode, a per cell contribution of 223 fA at +400 mV (vs. SHE). The second prototype with a 7 µm diameter electrode produced a current density of 3.9 pA/µm2 (3.9 A/m2) at +200 mV (vs. SHE) and a signal- to-noise ratio (SNR) of 4.9 when inoculated at a seeding density of 109 cells/mL. However, diluting the sample by 10x produced an SNR of 0.5, suggesting that obtaining single cell electrontransferratestoanelectrodeovershortexperimentaltimescalesmaynotbepossible with the system as tested. Nevertheless, the platform allows microstructure characterization and multiplexing within a single microfluidic chamber. i Acknowledgments An old proverb states that “it takes a village to raise a child”. Similarly, the completion of a dissertation is the culmination of the work of a personal and professional community. First and foremost, I must thank Jorge and Maria Isabel, my parents, and my sister Monica for their continuous support throughout the years and their passion for education and science. Papito, tu dedicacion y talento por la ingenieria a sido una inspiracion para mi. Mamita, siempre te agredecere por haber creido en mi cuanto mas lo necesitaba. Hermanita, gracias por darme alas y escusa para conocer el mundo! Tambien debo agradecerle a Douglas M., mi esposo, por ser mi raiz y darme perspectiva durante los momentos intensos con su ejemplo y serenidad. Moredirectly, IwanttothankProf. LiweiLinandProf. JohnCoatesfortheirmentorship and support both with academic and personal matters. In addition, Prof. Carlos Fernandez- Pello and Prof. Alice Agogino I would like to acknowledge for taking a personal interest in my success from the beginning of my graduate career. Within the Lin Lab, I would like to thank everyone for their comradeship and support! Heather, Adrienne, and Shogo - thank you for being you! Brian, Qin, Armon, Yingqi, Ryan, Ryan, Kevin, Jiyoung, Kosuke, and more recently Alina and Rich - you guys have been instrumental in getting me through these years! Similarly, I would like to acknowledge the Coates lab members, and particularly Kelly, for taking me under their biological wing. Suz, Sarah, Esther, Fabian, you guys are family! From speed soccer to 4am airport rides - thank you! Lastly, I would like to acknowledge the CUES office and Beatriz Lopez-Flores for provid- ing the scaffold through which Berkeley became my home. ii List of Figures 1.1 Motivation of the project is to learn about organic energy conversion. In nature, photosynthesis process in green plants converts carbon dioxide and water into glucose with the assistance of light. Next, microorganisms ferment glucose into acetate, and iron-reducing bacteria breakdown acetate to provide electrons that can be captured by machinery. The goal of this project is to learn from nature and apply engineering approaches to generate and extract energy from the natural processes through artificial systems. . . . . . . . . . 2 1.2 Cartoon of microbial mechanism that is exploited to scavenge energy from microorganisms. As shown, the bacteria G. sulfurreducens utilize acetate (vinegar) as a nutrient and generate carbon dioxide, protons, and electrons. Theelectronsaretransferredfromthecytoplasmthroughthemembranetothe extracellular environment using mechanisms intrinsic to the microorganism (inset). The microorganisms also produce pili or “organic nanowires” believed toactasconduitsfortheelectrons. Whenmicroorganismsareneartometallic substrates, electrons can transferred to produce direct current. . . . . . . . . 3 1.3 Electrical energy is harvested from microorganisms. The initial energy comes from acetate (redox of -0.3V vs. SHE) that serves as the bacteria’s elec- tron and carbon source. Next, electrons generated from their metabolism as waste are collected by an anode. The electrical potential of these electrons is dependent on many environmental factors but has empirically demonstrated between -0.2 to +0.2 V (vs.SHE). Lastly, the cathode interactions with oxi- dants (ie. dioxygen, ferricyanide) creates a voltage potential gradient across the electrodes that causes electrons to drift across an electrical circuit (load) and into the catholyte, completing the circuit. . . . . . . . . . . . . . . . . . 6 iii 1.4 Schematic of typical microbial fuel cell. Microorganisms catalyze anodic reac- tions and “strip” electrons and protons from carbohydrates, such as acetate, to generate energy for their own metabolic processes. Carbon dioxide is also released in the process. The electrons from metabolism are transferred to an anode. The system utilizes two electrodes separated by a semipermeable polymer exchange membrane (PEM) that keeps anolyte and catholyte from mixing. Voltage potential is generated across the electrodes, and power is produced by the fuel cell. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.5 Sediment microbial fuel cells demonstrating marine (left) and soil (right) en- vironment scavenging [1]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.6 SchematicofMEMSelectrodemicrobialfuelcelldevelopedpresentedinChap- ter 3. The micro-electrode increased spacial and temporal resolution of fun- damental studies. However, the system did not allow real-time monitoring of biomass and large size (1 mm2) provided a largely heterogeneous signal. . . . 12 1.7 Side-viewschematicofthesecondgenerationsingle-channelultra-micro-electrode microfluidic microbial fuel cell presented in Chapter 4. All electrodes lay within the same plane. The cathode and reference electrodes are solid state andburiedbeneathacastedNafionmembrane. Thesystempermittedbiomass loading estimation, and the 50 µm electrode area mitigated heterogeneity. . . 12 2.1 Relationship between power density and anode surface area in MFCs. Taken from Dewan and Lewandowski (2008). . . . . . . . . . . . . . . . . . . . . . 15 2.2 Generalized schemes for obtaining direct electron transfer between redox pro- teins and electrodes. The electrode in (a) has been modified by a promoter that modifies the surface to adsorb weakly basic pyridyl groups that interact with the positively charged interaction domains of proteins causing the redox centers to come close enough to the surface for electrons to tunnel to the electrode. The scheme for direct electron transfer shown in (b) indicates the possible role of multivalent cations in promoting stable attractive interactions between redox proteins with “negatively” charged interaction domains and the deprotonated surface of pyrolytic graphite [2]. . . . . . . . . . . . . . . . 18 2.3 Plant schematic showing the inputs and outputs of an electrochemical cell. . 22 2.4 Polarization schematic of typical hydrogen fuel cell performance depicting cell voltage drops due to common thermodynamic, kinetic, and mass transport losses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.5 Model of hydrogen fuel cell kinetic losses including activation, ohmic, and mass transfer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 2.6 Faradaic currents defined by electrode/electrolyte charge transfer from redox reaction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 2.7 Non-faradaic (capacitive) currents occur due to rearrangement of species at electrode/electrolyte interface but where no redox reactions occur. . . . . . . 30 iv 2.8 Geometrical relationship between bulk material resistivity, ρ, and intrinsic ohmic system resistance, R . . . . . . . . . . . . . . . . . . . . . . . . . . 30 ohmic 2.9 Scaling of the electrolyte resistance, electrode capacitance, and RC cell time constantasafunctionofmicroelectrodesize. Typicaldoublelayercapacitance and basal media conductivity values (C0 = 50 µF/cm2 and σ = 5 mS/cm) d are used in this simulation. Decreasing microelectrode dimensions decreases the experimentally accessible time scale limits. . . . . . . . . . . . . . . . . . 35 2.10 Models of electron externalization mechanisms in Shewanella and Geobac- ter species utilizing cytochrome network and pili “organic nanowires”. From K.Weber and J.D.Coates [3] . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.11 Simplified redox tower of electrons through G. sulfurreducens in bioelectro- chemical systems. Initially, acetic acid is internalized at a redox of -0.28V (vs. SHE), and electrons loose energy as they progress through metabolism and externalization. Assuming c-type cytochrome with a midpoint redox poten- tial of +0.25V (vs. SHE) as final protein in the electron transport chain, a fuel cell voltage of +0.57V can be predicted with oxygen as the final electron acceptor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 2.12 Anodicvoltagedependsonbiocatalystphysiology. Microorganisms’metabolic chain as well as electron externalization mechanism affects anode potential. Step A:Metabolism (dotted line) affects the electrical performance in three ways. 1) The terminal protein in the electron transfer path for externaliza- tion governs the redox at which the electrons may be harvested from the cell. 2) The density of this terminal protein (cytochrome) determines the ex- change current density that is intimately related with activation losses. 3) The metabolic rate of the cell must also be sufficient as to supply the termi- nal cytochromes with electrons and hence maintain the cells’ membrane at a negative potential. Step B:Bacteria-Electrode Transfer Mechanisms (double line) also affects the system and provide a characteristic “contact” resistance. In the case of direct bacteria-electrode contact, such as in G. sulfurreducens, where protein nanowires are believed to act as electrical conduits, the electron transport is governed by the nanowires’ material properties and the physical network density. In the species where metabolites transport the electrons to theelectrode,thetransferresistanceisaffectedintwoways. 1)Themetabolite acts a mediator that will provide a maximum (lower) redox for the transfer. 2) The current density may be diffusion limited particularly if insufficient metabolite concentration is present to mediate the electrons. . . . . . . . . . 41 v 2.13 An ohmic model of R = dV/di can be used to estimate a lumped metabolic resistance. Assuming a voltage drop of 0.53 V from the acetate fuel to c-type cytochromes in the membrane, and a metabolic rate of 200 fA/cell [4, 5], the resulting per cell faradaic resistance is 2.7 TΩ. Hence, the kinetics are estimated at 1000x slower than that of oxygen on platinum on a per micron area basis [6]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 2.14 Model depicting kinetic transport within anodic microbial fuel cells. In addi- tion, tothetypicalPEMfuelcellkinetics, thesystemalsoincludesametabolic overpotential, electron transfer resistance, and microorganism capacitance. . 43 3.1 Micro-electrode array chip demonstrating the dimensions of the fabricated electrode. The design was to mitigate diffusion effects and planarize 3D cloth structures to provide a known surface area. . . . . . . . . . . . . . . . . . . . 47 3.2 Electrode fabrication process. The process begins with a silicon wafer with insulating oxide on top. Photoresist is applied and patterned to define the electrodes with a lift off process. . . . . . . . . . . . . . . . . . . . . . . . . . 48 3.3 (a)ExplodedviewoftheMEMSMFC.Anolyteandcatholytecirculatethrough corresponding chambers separated by a Nafion membrane. (b) Fuel cell ex- perimental setup using potassium ferricyanide as electron sink at cathode. . 49 3.4 Live/dead representative fluorescence images of the biofilm as it develops over time. Imageswereacquiredat(a)1day, (b)3days, (c)6days, and(d)10days after inoculation. The bacteria nucleate on metal electrode to grow vertical elongated clusters that overtime merge onto a continuous biofilm. Anode was continuouslyacceptingelectrons, andnegligiblenon-viablecellswerecaptured during the 10 day experiment. Dashed lines indicate gold electrode location. 51 3.5 SEM images after 6-days of operation showing (a) greater growth of cells on goldelectrodethatinsulatingsurface, (b)whitecirclesillustratingcelldivision occurring on SiO (oxide), and (c) bacteria illustrating significant appendages 2 protruding into electrodes. Bacteria are roughly 300 nm in diameter and 2 µm long. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 3.6 Anodic open circuit showing anodic potential of cells when harvested at a charged state. Cells manifested a -0.25 V (vs. SHE) when incubated in anaer- obic conditions for two weeks prior to inoculation. When cells are harvested during exponential growth phase, however, anodic voltage typically ranges between +0.2 to +0.3 V (vs. SHE). . . . . . . . . . . . . . . . . . . . . . . 55 3.7 Polarization curves after 0.1, 1, 6, and 10 days, respectively, from the micro- patterned microbial fuel cell. Note that V is roughly 630 mV with potas- oc,max sium ferricyanide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 3.8 Power densities obtained at various loads during ten days of continuous bac- terial respiration and growth on the electrode. . . . . . . . . . . . . . . . . 57 vi 3.9 Microbial fuel cell electrical responses to different load resistors acquired from a fuel cell six days after inoculation that provided a V (V - V ) of +0.2 V, oc c a where the synthesized data is shown in Figs 3.7 and 3.8. Anodic and cathodic redox potentials for various loads were acquired against an Ag/AgCl reference electrode and are shown corrected to the SHE scale. The corresponding loads were (a) 1 MΩ, (b) 300 kΩ, (c) 200 kΩ, (d) 100 kΩ, (e) 22 kΩ, (f) 5 kΩ, (g) 940 Ω, and V of 108 Ω (internal resistance of the DAQ utilized). . . . . . . 58 oc 3.10 Transient electrical current obtained from 100 kΩ resistor data in Fig. 3.9d above. Behavior is analyzed to extract resistance and capacitance snapshots of the system at the time of testing. . . . . . . . . . . . . . . . . . . . . . . 60 3.11 RC constants obtained for the system over time. Abiotic system’s RC is estimated at 100 ms. However, addition of the cells increases the delay to steady state. The RC estimate for G. sulfurreducens 6 days after inoculation is 140.5 sec. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 3.12 System resistance development over time. The microorganisms increase the system’s resistance by 5x upon inoculation. However, the resistance decreases to18%ofthecontrollevelafter6dayssuggestingthatthebacteriaandbiofilm can enhance the catalytic process. . . . . . . . . . . . . . . . . . . . . . . . 63 3.13 Capacitance of the system increases several orders of magnitude after the addition of the microorganisms. After 1 day, the bacteria acclimate to the electron externalization and begin attachment. After 6 days, capacitance increases by 63.5x over the day 1 value, which is the equivalent to 6 division cycles (26 = 64) and an average 20.0 hour doubling time for that period. . . 64 3.14 MFC capacitance is greatly affected by the addition and growth of cells on the anode. The system’s average capacitance increased quadratically over 6 days and seems to correlate with biocatalyst division on the electrode. . . . . 65 3.15 Electrical current of 10 minute intervals from a 2.5 hour period (6 days after inoculation) of continuous discharge superimposed to demonstrate output de- crease over time. This behavior suggests that the system is discharging and steady state has not been reached. . . . . . . . . . . . . . . . . . . . . . . . . 67 3.16 Resistance, R, capacitance, C, and RC values obtained from continuously running an electrode after 2 hours of open circuit conditions. They demon- strate the variability that stems from the “state” of the system at the time of testing and the inverse relationship between R and C. The RC constant maintains stable during discharge, however, suggesting that it could provide a more stable marker for characterization of biomass on microbial fuel cell anodes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69