The Pennsylvania State University The Graduate School College of Engineering ELECTRICITY FROM COMPLEX BIOMASS USING MICROBIAL FUEL CELLS

Transcription

1 The Pennsylvania State University The Graduate School College of Engineering ELECTRICITY FROM COMPLEX BIOMASS USING MICROBIAL FUEL CELLS A Dissertation in Agricultural and Biological Engineering by Farzaneh Rezaei 2008 Farzaneh Rezaei Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy December 2008

2 ii The dissertation of Farzaneh Rezaei was reviewed and approved* by the following: Tom L. Richard Associate Professor of Agricultural and Biological Engineering Dissertation Co-adviser Co-chair of Committee Bruce E. Logan Kappe Professor of Environmental Engineering Dissertation Co-adviser Co-chair of Committee Ali Demirci Associate Professor of Agricultural and Biological Engineering Ming Tien Professor of Biochemistry Roy Young Professor of Agricultural and Biological Engineering Head of the Department of Agricultural and Biological Engineering *Signatures are on file in the Graduate School.

3 iii Abstract Presently, scientists and engineers are working diligently to make it possible to gain energy from any possible renewable source. The urgency of these efforts is apparent when we think about the limited amount of oil and fossil fuel combined with the rapid increase of the world s population, and even more rapid increase in energy demand. An additional advantage of using renewable sources for energy production is the associated reduction in greenhouse gas emissions. With biomass, the oxidation of organic matter will produce carbon dioxide, a natural product that can be fed back into the natural carbon cycle and converted back into more biomass through photosynthesis. Electricity can be directly generated by bacteria in microbial fuel cells (MFC) from a variety of biodegradable substrates, including cellulose. Biomass that contains complex and simple carbohydrates is widely available, so the use of MFCs to produce electricity from these sources is potentially very promising. The limited study on using complex biomass in an MFC is mainly due to the low degradation rate of these materials. A sediment microbial fuel cell (SMFC) produces electricity through the bacterial oxidation of organic matter contained in the sediment. The power density is limited, however, due in part to the low organic matter content of most marine sediments. To increase power generation from these devices particulate substrates were added to the anode compartment. Three materials were tested: two commercially available chitin products differing in particle size and biodegradability (Chitin 20 and Chitin 80), and cellulose powder. Maximum power densities using chitin in this substrate enhanced sediment MFC (SEM) were 76 ± 25 mw/m 2 and 84 ± 10 mw/m 2 (normalized to cathode projected surface area) for Chitin 20 and Chitin 80, respectively, versus less than 2 mw/m 2 for an unamended control. Power generation over a 10-day period

4 iv averaged 64 ± 27 mw/m 2 (Chitin 20) and 76 ± 15 mw/m 2 (Chitin 80). With cellulose, a similar maximum power was initially generated (83 ± 3 mw/m 2 ), but power rapidly decreased after only 20 hours. Maximum power densities over the next 5 days varied substantially among replicate cellulose-fed reactors, ranging from 29 ± 12 to 62 ± 23 mw/m 2. These results suggest a new approach to power generation in remote areas based on the use of particulate substrates. While the longevity of the SEM was relatively short in these studies, it will be possible to increase operation times by controlling particle size, mass, and type of material needed to achieve desired power levels that could theoretically be sustained over periods of years or even decades. I examined the effect of three different chitin particle sizes on MFC power generation and longevity. The results demonstrated that an increase in particle diameter from the smallest (0.28 mm) to largest (0.78 mm) size tested resulted in an increase in longevity from 9 to 35 days. Coulombic efficiency based on removal of chitin was also increased from 18% for the smallest particles to 56% for the largest ones. However, the maximum power generation was lower for large particles (201 mw/m 2 ) as compared to the small and medium size (0.46 mm) (301 and 285 mw/m 2 ). Measured MFC longevity was explained by the fractal particle degradation model a mass transfer equation. Cellulases are used to achieve rapid conversion of cellulose to sugar for ethanol production, but these enzymes have not been previously tested for their effectiveness in MFCs. It was not known if cellulases would remain active in an MFC in the presence of exoelectrogenic bacteria or if enzymes might hinder power production by adversely affecting the bacteria. Electricity generation from cellulose was therefore examined in two-chamber MFCs in the presence and absence of cellulases. The maximum power density with enzymes and cellulose was 100 ± 7 mw/m 2 (0.6 ± 0.04 W/m 3 ), compared to only 12 ± 0.6 mw/m 2 (0.06 ± W/m 3 )

5 v in the absence of the enzymes. This power density was comparable to that achieved in the same system using glucose (102 ± 7 mw/m 2, 0.56 ± W/m 3 ) suggesting that the enzyme successfully hydrolyzed cellulose and did not otherwise inhibit electricity production by the bacteria. The addition of the enzyme doubled the Coulombic efficiency (CE) to CE=51% and increased COD removal to 73%, likely as a result of rapid hydrolysis of cellulose in the reactor and biodegradation of the enzyme. These results demonstrate that cellulases do not adversely affect exoelectrogenic bacteria that produce power in an MFC, and that the use of these enzymes can increase power densities and reactor performance. When cellulose is used as the substrate, electricity generation requires a microbial community with both cellulolytic and exoelectrogenic activity. This was not previously demonstrated in pure culture without the provision of an exogenous mediator. Using a specially designed U-tube MFC, I enriched a consortium of exoelectrogenic bacteria capable of using cellulose as the sole electron donor. After 19 dilution-to-extinction serial transfers of the consortium, 16S rrna gene-based community analysis using denaturing gradient gel electrophoresis and band sequencing revealed that the dominant bacterium was Enterobacter cloacae. An isolate designated E. cloacae FR from this enrichment was found to be 100% identical to the type strain Enterobacter cloacae based on partial 16S rrna sequence. In polarization tests using the U-tube MFC and cellulose as a substrate, strain FR produced 4.9 ± 0.01 mw/m 2 compared to 5.4 ± 0.3 mw/m 2 for strain These results demonstrate for the first time that it is possible to generate electricity from cellulose using a single bacterial strain without the need for exogenous mediators. In order to use cellulosic feedstocks like switchgrass that will be annually harvested, there needs to be a low-cost method for their storage year round. Ensilage was previously shown

6 vi to have potential as a low-cost method for corn stover storage as well as simple pretreatment prior to ethanol production. Before using ensiled feedstock for ethanol fermentation with ph neutral microorganisms, the organic acids that were generated during ensilage will need to be removed or neutralized. Production of organic acid during ensilage was previously modeled for different crops. In this research, the model was used to predict the organic acid generation during ensilage of switchgrass. Observed data was highly correlated with predicted values (r>0.85). While these organic acids are problematic for fermentation, they are a potentially valuable byproduct. Conversion of these acids to electricity in a microbial fuel cell was analyzed as one such alternative.

7 vii Table of Contents List of Figures..xi List of Tables...xiii Acknowledgments....xiv Chapter 1 Introduction Overview Summary of the MFC research in this dissertation Literature cited... 6 Chapter 2 Literature Review Overview Bioelectricity Generation Using Microbial Fuel Cell (MFC) Power output from MFCs Exoelectrogenic Microorganisms Improving Power Generation from MFCs Applications of Microbial Fuel Cells Complex Biomass Cellulose Chitin Hydrolysis Storage of complex biomass (Ensilage) Modeling Organic Acid Generation during Ensilage Challenges to use Complex Carbohydrates as Substrate in Microbial Fuel Cells Literature Cited Chapter 3 Substrate-enhanced Microbial Fuel Cells for Improved Remote Power Generation from Sediment-Based Systems Abstract Introduction Materials and Methods Substrates Electrodes Medium... 42

8 viii SEM Construction and Operation Two-Chamber MFC Tests Analytics and Calculations Results Power Generation with Chitin Power Generation with Cellulose Polarization and Power Density Curves Volatile Fatty Acids (VFAs) Power Generation in a Two-Chambered MFC Discussion Acknowledgements Literature Cited Chapter 4 Effect the Particle Size on MFC Maximum Power Generation, Power Longevity, and Coulombic Efficiency Abstract: Introduction: Materials and Methods: MFC Construction and Operation: Substrate: Community Analysis: Analytics and Calculations: Modeling Particle degradation: Results: Effect of Particle Size on MFC Performance: Community Analysis of Dominant Bacteria in MFC: Modeling Particle Longevity: Discussion: Acknowledgments: Literature Cited: Chapter 5 Enzymatic hydrolysis of cellulose coupled with electricity generation in a microbial fuel cell

9 ix 5.1 Abstract Introduction Materials and Methods Medium: MFC Construction and Operation: Abiotic Rate of Cellulose Hydrolysis: Analytics and Calculations: Results Power Generation with Different Substrates: Maximum Power Density: Enzymatic Hydrolysis Rate of Cellulose in the Presence and Absence of Bacteria: Cellulase Activity, Cellulase Consumption, and Cellulose Degradation: Coulombic Efficiency: COD Removal: Discussion Acknowledgements Literature Cited: Chapter 6 Simultaneous cellulose degradation and electricity production by Enterobacter cloacae in an MFC Abstract Introduction Methods MFC Construction and Operation Enrichment Procedure DNA Extraction, PCR, and DGGE Cloning and Sequence Analysis Bacteria Isolation and Characterization Electricity Generation and Analyses Results Exoelectrogenic/Cellulolytic Enrichment Phylogenetic Analysis

10 x Biochemical, Physiological, and Electrochemical Characteristics of E. cloacae T and E. cloacae FR Discussion Acknowledgment Literature Cited: Chapter 7 Modeling Organic Acid Production from Ensilage to be used for Electricity Generation in an MFC Abstract Introduction Material and Methods Ensilage Analytical method: Model Assumptions Results: Parameter Estimation: Change of ph, Sugars and Organic Acid Concentration during Ensilage Process: Model Validation: Sensitivity Analysis Current and Energy Generation Conclusion and Suggestions Literature Cited: Chapter 8 Conclusion and Future Research Appendixes

11 xi List of Figures Figure 2-1 Schematic illustration of a microbial fuel cell modified from (Logan et al., 2006) Figure 2-2 Small part of Cellulose structure. The two chains can rotate relative to each other (Nelson and Cox, 2004) Figure 2-3 A short segment of chitin. (Nelson and Cox, 2004) Figure 3-1Chitin 80, cellulose, and Chitin 20 as substrates. Also shown is the carbon cloth anode filled with Chitin 80 before being sewn Figure 3-2 The small scale sediment fuel cell with substrate (SEM), anode is embedded inside the marine sediment at the bottom (anaerobic phase) and cathode is suspended in the overlaid seawater (aerobic phase) Figure 3-3 Power generation from SEM with Chitin 20 and Chitin 80 as substrate as well as a control (without substrate). Each line represents the average of triplicate reactors (error bars ±SD). Symbols are shown for each 10 data point, and standard deviations are shown for each 50 data point Average power density from six anodes embedded into the sediment layered into a glass tank. The data is the average of six replications (error bars ±SD) Figure 3-5 Power generation from three replications of SEMs with cellulose as substrate Figure 3-6 (A) Power density based on different resistance for all substrates. (Error bars are ±S.D. based on duplicate measurements). (B) Voltage based on different resistance for all substrates. (Error bars are ±S.D. based on duplicate measurements) Figure 3-7Power generation in a two-chamber MFC with Chitin 80 or cellulose (1 g/l) Figure 4-1 Voltage generation and longevity of MFC reactors fed with different particle sizes of Chitin Figure 4-2 Polarization curve for MFC run on different Chitin 20 particle sizes. (Error bars are ±S.D. based on duplicate measurements) Figure 4-3 Coulombic efficiency based on removed chitin measured at the completion of the batch cycle (CE rem ) and added chitin assuming all chitin was consumed (CE add ) for three different particle sizes Figure 4-4 DGGE analysis on the anode solution of the reactors fed with different particle sizes. Arrow shows the bands that were exerted from the gel for sequencing analysis Figure 4-5 Measured MFC longevity from the reactor s operational time and predicted MFC longevity from the mass transfer model with particle size as input Figure 5-1 Power density (normalized to cathode surface area, constant 1000Ω load) as a function of time from 2-chamber reactor fed with cellulose + enzyme, glucose, cellulose alone, and enzyme alone Figure 5-2 (A) Power and Voltage (B) as a function of current for three treatments, cellulose + enzyme, Enzyme, and Glucose (error bars are ± S.D based on duplicate measurements) obtained by varying external resistance Figure 5-3 (A) Enzymatic hydrolysis of cellulose without microbes. (B) Cellulose concentration remaining inside the reactor during enzymatic hydrolysis without microbes (Abiotic) versus

12 simultaneous saccharification and oxidation in an MFC. Data are cellulose concentrations inside the reactor relative to the initial cellulose concentration Figure 5-4 Enzyme activity in MFC reactors with and without cellulose Figure 6-1 Power generation from (A) the first cycle of U-tube using four different dilutions and (B) the last (19th) cycle of U-tube using four different dilutions Figure 6-2 DGGE bands of the 19 cycles of the most diluted U-tube that produced electricity. Bands 1 to 5 was extracted from the gel for sequencing Figure 6-3 Phylogenetic tree of extracted bands from last cycle of DGGE and closely related species of 16S rrna gene. The tree was constructed using the neighbor-joining method. Bootstrap values at nodes were calculated using 1000 replicates (only values >50% are indicated). The scale bar represents 2 % divergence Figure 6-4 Phylogenetic tree of U-tube dominant bacterium in clone library and closely related species of 16S rrna gene. The tree was constructed using the neighbor-joining method. Bootstrap values at nodes were calculated using 1000 replicates (only values >50% are indicated). The scale bar represents 0.5 % divergence Figure 6-5 Polarization curve to measure maximum power density generated in reactors inoculated with pure Enterobacter cloacae ATCC 13047T, isolate Enterobacter cloacae FR and mixed culture from last cycle Figure 6-6 Current density produced with Enterobacter cloacae ATCC 13047T in a U-tube with different carbon sources. Error bars are SD based on triplicate data collection over three batch cycles Figure Stella modeling to predict Clostridia concentration per gram of silage using environmental parameters Figure Stella modeling to predict LAB concentration per gram of silage using environmental parameters Figure Concentration of Lactic Acid (LA) generated during ensilage. Upper line is predicted by model and lower was observed from the lab scale ensiled switchgrass using 2007 results Figure Predicted Lactic Acid generation using the ensilage model and observed LA generation from the results of 2007 harvest Figure Concentration of Acetic Acid (AA) generated during ensilage. The line with symbols was observed from the lab scale ensiled switchgrass using 2007 results and the line without symbols was predicted by the model Figure Predicted Acetic Acid (AA) generation using the ensilage model and observed AA generation from the results of 2007 harvest xii

13 xiii List of Tables Table 4-1 Concentration of initial added chitin, and end product (fermentation product and biomass)* Table 4-2 BLAST results of the closest strain to sequenced band Table 5-1 Four different treatments with their total initial substrate concentrations Table 5-2 Summary table of percent substrate consumption, COD removal, enzyme activity lost, and coulombic efficiency Table 6-1 Closest reported strains to the sequence of the last cycle s bands from GenBank using BLAST program Table 6-2. Biochemical characteristics of Enterobacter Cloacae ATCC13047T and strain Enterobacter Cloacae FR Table Phase, Input and Output and Environmental Factors Affecting Rate Process from (Pitt et al., 1985) Table Input parameter estimations for the ensilage model Table Summary of ph, weight loss (W loss), hemicellulose, cellulose, and water soluble carbohydrates (WSC), Acetic Acid (AA), Lactic Acid (LA) for lab scale samples (2007 data) (db means dry basis) Table 7-4. Sensitivity Analysis

14 xiv Acknowledgments It is difficult to overstate my gratitude to my Ph.D. co-advisors, Drs. Tom Richard and Bruce Logan, for their enthusiasm, support, and great efforts to explain things clearly and simply. This dissertation would not have been possible without the kind support, the probing questions, and the remarkable patience of my committee members Drs. Ali Demirci and Ming Tien. I would like to thank all former and current graduate and undergraduate students of Dr. Logan and Dr. Richard s laboratories for their support and friendship. I wish to thank my parents Jalal Rezaei and Iran Rezaei. They bore me, raised me, supported me, taught me, and loved me. To them I dedicate this thesis. Lastly, and most importantly, I would like to thank my husband Hamid Ashrafi for his love, support, encouragement and patience during my entire work.

15 1 Chapter 1 Introduction 1.1 Overview Oil, coal, and natural gas provide more than 85% of the US energy needs. Among those, two thirds of electricity is generated using coal or natural gas and almost all transportation fuels are oil and natural gas (USDA, 2008). On top of that, global warming, a consequence of releasing CO 2 that was stored in fossil fuels for millions of years into the environment, is changing the mean surface temperatures and resulting in melting of the glaciers. Therefore, considering the energy crises, rising prices of fossil fuel, limited fossil fuel resources, and global warming due to use of fossil fuels, it should be a top priority to switch to alternative fuels that are renewable and minimize net CO 2 emissions. A wastewater treatment plant is one of the places that we can save and harvest energy, instead of using energy as we do today. As reported by Logan (2007), in US alone, 4-5% of the electricity production is used for the water infrastructure, and about 1.5% goes for wastewater treatment. The Microbial Fuel Cell (MFC) was recently introduced to harvest the energy content of the wastewater in the form of electricity, while simultaneously treating wastewater by removing chemical oxygen demand (COD). An MFC is a device to produce electricity from the direct oxidation of organic matter by microorganisms (Allen and Bennetto, 1993). In an MFC, microorganisms live in the anoxic environment of the anode side of the fuel cell, where they send the electrons, products of organic matter oxidation, to the anode electrode. These electrons then flow through a circuit to the cathode side where they combine with oxygen to produce water. For each electron released, an equivalent proton goes to the cathode side of the fuel cell through an Ion Exchange Membrane

16 2 (IEM). Logan (2004) calculated the economic potential of MFC electricity generation from a food processing plant that treats m 3 / year wastewater. If all the organic matter of that waste could be directly converted to electricity, and assuming $0.04 /KWh and 15 KJ/g BOD, the electricity generated would be worth $460,000/year (Logan, 2004). An MFC can be operated using a mixed bacterial culture or a pure culture, and simple or complex carbohydrates as substrates. In most cases, mixed cultures combined with simple substrates produce substantially higher power output as compared to pure cultures (Rabaey et al., 2004). However, it has recently been shown that pure culture Rhodopseudomonas palustris DX-1 can produce electricity at an even higher rate than that produced by a mixed culture in the same MFC (Xing et al., 2008). Furthermore, even for those strains thought to produce less power than mixed cultures (i.e. Geobacter sulfurreducens), it has recently been shown that selecting the right MFC architecture can result in electricity production comparable to mixed cultures (Nevin et al., 2008). There have been few studies focused on comparing pure cultures versus mixed culture inocula, as well as looking at complex carbohydrates and optimizing the power generation from these abundant carbohydrates in an MFC. 1.2 Summary of the MFC research in this dissertation The primary goal of this Ph.D research was to develop ways to use complex biomass resources in microbial fuel cells. The substrates that were investigated were cellulose and chitin. Two broad strategies for practical implementation of MFCs were recognized early in the development of this technology. While most investigators focused on systems that transport the organic substrate into the MFC (e.g. wastewater treatment), others decided to embed the MFC in a stationary organic matter source. An example of this second approach has been termed the

17 3 sediment microbial fuel cell (SMFC) (Logan, 2008; Reimers et al., 2001). In an SMFC device, the organic matter of the marine sediment is the source of electron donors. Most of the sediment accumulated in the seafloor contains % organic carbon by weight (Yen, 1977). However, much of this organic matter is tied up in recalcitrant compounds, and the readily available organic matter is not sufficient to generate high levels of power output from SMFCs. Although it was shown that adding particulate substrates improved power generation compared to unamended controls, the longevity of the SEMs (based on high power output) was only 30 days under laboratory conditions. However, MFC reactor longevity should be able to be obtained by formulating particles of specific sizes and composition to extend degradation times. When dealing with particulate substrates in microbial reactors, the particle size and its available surface area are among important factors that determined the hydrolysis rate and accessibility of the substrates to the microorganisms. The substrate dissolution rate can be modeled using an approach for particle dissolution assuming spherical particles (Logan, 1999; Thibodeaux, 1996). Any biodegradable materials should be able to be used in an MFC for electricity production. However, it is hard for microorganisms to degrade more complex biomass resources such as lignocellulosic material. The majority of MFCs are running based on simple carbohydrates such as glucose or acetate. Availability of complex biomass and the low price of these raw materials as compared to simple sugars suggest that researchers should investigate different methods to use these materials as feedstocks in MFCs. The potential energy production from corn stover, which is primarily composed of complex carbohydrates, was estimated by Zuo et al. (2005) at about kwh/year, with a value of $6.9 billion per year based on $0.15/kWh (Zuo et al., 2005). However, previous reports have suggested that power generation

18 4 in MFCs from insoluble cellulose can be limited by a low hydrolysis rate (Ren et al., 2007; Rezaei et al., 2007). Cellulose is the most abundant carbohydrate in the world and there have been several attempts to generate electricity in an MFC with cellulose is the sole electron donor. In all previous research, at least two types of incoula were used so that one or more bacterial species could hydrolyze cellulose to simpler carbohydrates, and other species generated electricity using the hydrolysis products. There is no known cellulolytic-exoelectrogenic bacterium. The lack of such a microbe could be due to the conventional method of isolating exoelectrogenic-cellulolytic bacteria. Conventional methods of isolating exoelectrogenic bacteria are based on reducing metal oxides (Logan and Regan, 2006; Lovley, 2006; Logan, 2008). However, not all exoelectrogenes are capable of reducing iron and not all iron-reducing bacteria can produce electricity in an MFC (Bretschger et al., 2007; Richter et al., 2007). A new method of isolating electrochemically active bacteria using U-tube MFCs was recently introduced (Zuo et al., 2008). In considering the use of lignocellulosic material, two issues should be addressed: degradation of the lignocellulose and storage. Hydrolysis of cellulose combined with electricity generation has been already covered in previous sections; therefore, storage of switchgrass was the main focus of this part of the research investigation. Agricultural bioenergy crops and residues such as switchgrass or corn stover are sources of lignocellulosic materials that are usually harvested once a year. However, to use them as a consistent source of energy they need to be available throughout the year. This necessitates storage of the lignocellulosic material after each harvest season. Ensilage is a solid state fermentation process that has long been used to preserve lignocellulosic materials as forage for year-round production of livestock. Furthermore, ensilage appears to assist with the hydrolysis needed to produce lignocellulosic ethanol (Richard

19 5 et al., 2001). There are several byproducts from the ensilage process including organic acids, solvents, extracellular enzymes, amino acids, and antibiotics. Recovery of these products will increase the efficiency of ensilage by producing value-added products. During the ensilage process, organic acid production results in a ph drop and prevents the growth of microorganisms that would otherwise use the feedstock as a source of organic carbon. However, organic acids should be removed from the system before feedstock being used for ethanol production. Extracted organic acids could have a great value in industry if the mixture could be separated to its individual acids, but it is difficult and expensive to separate these organic acids. It will be economically beneficial and reduce waste production if the mixed organic acids could be used as substrate in an MFC for electricity generation.

20 6 1.3 Literature cited Allen, Robin M, and H Peter Bennetto Microbial fuel cells: electricity production from carbohydrates. Applied Biochemistry and Biotechnology 39 (2): Bretschger, O, A Obraztsova, C A Sturm, I S Chang, Y A Gorby, S B Reed, D E Culley, C L Reardon, S Barua, M F Romine, J Zhou, A S Beliaev, R Bouhenni, D Saffarini, F Mansfeld, B Kim, J K Fredrickson, and K H Nealson Current production and metal oxide reduction by Shewanella oneidensis MR-1 wild type and mutants. Applied and Environmental Microbiology 73: Leibensperger, R Y, and R E Pitt A model of clostridial dominance in ensilage. Grass and Forage Science 42: Levin, D.B., L. Pitt, and M. Love Biohydrogen production: prospects and limitations to practical applications. International Journal of Hydrogen Energy 29: Logan, B E Environmental transport process. New Jersey, NY: John Wiley & Sons, Inc. Logan, B E Feature Article: Extracting hydrogen and electricity from renewable resources. Environ. Sci. Technol 38 (9):160A-167A. Logan, B E Microbial Fuel Cells. Edited by J. W. Sons. New Jersey, NY: John Wiley & Sons, Inc Logan, B E, and J M Regan Electricity-producing bacterial communities in microbial fuel cells. Trend Microbiol 14: Lovley, D R Bug juice: harvesting electricity with microorganisms. Nature Reviews Microbiology 4 ( ). Nevin, K P, H Richter, S F Covalla, J P Johnson, T L Woodard, A L Orloff, H Jia, M Zhang, and D R Lovley Power output and columbic efficiencies from biofilms of Geobacter sulfurreducens comparable to mixed community microbial fuel cells. Environmental Microbiology doi: /j x. Rabaey, Korneel, Nico Boon, Steven D Siciliano, Marc Verhaege, and Willy Verstraete Biofuel cells select for microbial consortia that self-mediate electron transfer. Applied and Environmental Microbiology 70 (9): Reimers, C E, L M Tender, S Fertig, and W Wang Harvesting Energy from the Marine Sediment-Water Interface. Environ. Sci. Technol. 35: Ren, Z, T E Ward, and J M Regan Electricity production from cellulose in a microbial fuel cell using a defined binary culture. Environ. Sci. Technol. 14 (13): Rezaei, Fazaneh, Tom L. Richard, Rachel Brennan, and Bruce E. Logan Substrateenhanced microbial fuel cells for improved remote power generation from sedimentbased systems. Environmental Science & Technology 41 (11): Richard, T L, S Proulx, K J Moore, and S Shouse Ensilage technology for biomass pretreatment and storage. In ASAE Paper no ASAE. Richter, H, M Lanthier, K P Nevin, and D R Lovley Lack of electricity production by Pelobacter carbinolicus indicates that the capacity for Fe(III) oxide reduction does not necessarily confer electron transfer ability to fuel cell anodes. Applied and Environmental Microbiology 165: Rismani-Yazdi, Hamid, Ann D. Christy, Burk A. Dehority, Mark Morrison, Zhongtang Yu, and Olli H. Tuovinen Electricity generation from cellulose by rumen microorganisms in microbial fuel cells. Biotechnology and Bioengineering:doi bit

21 Thibodeaux, L J ENVIRONMENTAL CHEMODYNAMICS - Movement of Chemicals In Air, Water and Soil,. Vol. 2nd Edition. N.Y: J. Wiley. USDA [cited 2008]. Available from Xing, D, Y Zuo, S Cheng, J M Regan, and B E Logan Electricity generation by Rhodopseudomonas palustris DX-1. Environ. Sci. Technol 42 (11): Yen, T F Chemical aspects of marine sediments. In Chemistry of Marine Sediment, edited by T. F. Yen. MI: Ann Arbor Science Publisher. Zuo, Y, P-C Maness, and B E Logan Electricity production from steam-exploded corn stover biomass. Energy & Fuels 20 (4): Zuo, Y, D Xing, J M Regan, and B E Logan Isolation of the exoelectrogenic bacterium Ochrobactrum anthropi YZ-1 by using a U-tube microbial fuel cell. Applied and Environmental Microbiology 74 (10):

22 8 Chapter 2 Literature Review 2.1 Overview Today s world is facing a real energy crisis more than ever because of the high energy demand by developed and developing countries and increase in human populations in these countries and elsewhere. The supply of fossil fuels will not last forever and sooner or later it will be exhausted due to a very long time needed for recovery of these resources. Even if we assume they recover or new fossil fuel resources are discovered that meet the world s fuel needs for a long time, still with the conventional technologies y we are going to add more CO 2 to the environment that exacerbate global warming. Therefore, during the last decade, scientists and engineers have devoted their time and efforts to discover renewable energy resources. There is a hope that one day these renewable resources act as possible substitutes for fossil fuel. There are several advantages associated with renewable sources including reducing our dependency to fossil fuels (especially oil) and decreasing global warming by producing less greenhouse gases. Wind, solar energy, and biomass energies are examples of renewable resources that might ultimately be the long term solution to our energy demands. In this dissertation, complex but abundant biomass resources, such as cellulosic materials and chitin, was used as fuels to generate electricity using a microbial fuel cell (MFC) and it was tried to increase the power using these materials. In this endeavor, not only it is possible to use these biomass resources directly, but it is also possible to use them indirectly through their byproducts. These byproducts including the organic acid and other derivatives from the ensiled feedstock can be used in an MFC for electricity generation and to increase the efficiency of the

23 9 system by making value added products. For this reason, I modeled the generation of mixed organic acids from the ensiled storage of switchgrass and calculated the amount of electricity that potentially could be generated if these products were used in an MFCs. This literature review covers the following topics: the general principles and concept of MFCs, design, mode of function, types, and recent improvements in the MFC technology. The first section is followed by a short review of complex biomass, their storage and a review of hydrolysis, ensilage, and longevity of particulate particles. At the end, the present state-of-the-art is summarized for the entire chapter. 2.2 Bioelectricity Generation Using Microbial Fuel Cell (MFC) A fuel cell is a device that converts chemical energy, hydrogen, and oxygen to electrical energy, water, and heat with a high operational efficiency. In 1911, the idea of generating electricity by organisms was first proposed by Potter (1911), however, it was in the early 1990s that fuel cells became more interesting and more research was conducted in this subject (Allen and Bennetto, 1993). Fuel cells that directly convert chemical energy to electricity by microorganisms or enzymes are called biological fuel cells or bio-fuel cells. These two approaches to bioconversion define the two types of bio-fuel cell: microbial fuel cells and enzymatic fuel cells (Shukla et al., 2004). A Microbial Fuel Cell (MFC) is a device to produce electricity from the direct oxidation of organic matter by microorganisms (Allen and Bennetto, 1993). In order to generate electricity, microorganisms need to send electrons, a product of the oxidation of organic/inorganic matter, to an electron acceptor through an external circuit with a defined resistance. In an MFC the

24 10 oxidation process happens at the anode (negative electrode) and reduction happens at the cathode (positive electrode) side of the cell (Figure 2-1). Figure 2-1 Schematic illustration of a microbial fuel cell modified from (Logan et al., 2006). During this process, electrons move from the anode to the cathode through an external circuit where they will be captured by any available electron acceptor. Two commonly used electron acceptors in the cathode side of MFCs are oxygen (Bond and Lovley, 2003; Dubois et al., 1956; Chaudhuri and Lovley, 2003; Bond et al., 2002; Liu et al., 2004b; Liu and Logan, 2004b) and ferrocyanide (Rabaey et al., 2003; Schröder et al., 2003; Rabaey et al., 2004b). The end product in the cathode chamber is water (in the case where oxygen is the electron acceptor) or ferrocyanide (if ferricyanide is available). The maximum power generation achieved by using ferrocyanide as oxidant in cathode is about % more than is achieved using oxygen as an oxidant (Oh et al., 2004b). However, ferricyanide must be replaced as soon as it is consumed whereas oxygen is passively and continuously supplied by diffusion from air if air cathode is being used.

25 11 MFCs can be designed by either placing the anode and the cathode in two different chambers (2-chamber MFC) separated by a proton exchange membrane (PEM), or by having both electrodes in one chamber with or without any PEM (single chamber). Although MFC power output has improved significantly during the past decade, still most of the designs and operational methods are not proper for large scale MFCs (Logan and Regan, 2006; Rabaey and Verstraete, 2005). In scaling up the MFC, the price of the cathode should be reduced significantly to make MFC economically favorable. For instance, replacing platinum catalysts on cathode by non-precious metals such as Cobalt tetra-methyl phenylporphyrin (CoTMPP) reduced the cost while power generation was comparable to that of Pt catalysts (Cheng et al., 2006c). Furthermore, using tubular cathode or brush anodes are promising, because they can be scaled up easier than other electrode materials and designs (Zuo et al., 2007; Logan et al., 2007) Power output from MFCs Primary electricity generation from early MFC designs was so low, however, it has increased rapidly in just a few years and it increased 100,000 times using air cathode MFC (Logan, 2008). The maximum power generation varies based on both type of substrates and MFC architectures. In general, power generation from simple carbohydrates is higher than that produced by complex biomass or wastewater. In a single chamber air cathode MFC, the maximum reported power density is 1970 mw/m 2 from acetate (Cheng and Logan, 2007), whereas it is only 150 mw/m 2 with wastewater as electron donor (Logan, 2004). Protein, propionate, and butyrate also was used as substrate in MFC and all produced power less than that produced by pure carbohydrates (i.e., glucose,

26 12 acetate) when used in the same devices. Furthermore, the maximum power density varies based on the wastewater composition. The reported power generated (under different conditions) from meat packing wastewater is 80 mw/m 2 (Heilmann and Logan, 2006), beer brewery wastewater 206 mw/m 2 (Feng et al., 2008), animal wastewater 261 mw/m 2 (Min et al., 2005b), and paper recycling wastewater 672 mw/m 2 (in 100 mm PBS) (Huang and Logan, 2008a) Exoelectrogenic Microorganisms Exoelectrogens are group of bacteria that have ability to produce electricity in an anaerobic environment in MFCs by direct transfer of electrons outside the cell. Our information about different species of electrochemically active microorganisms is in developmental stage and the complete mechanisms for electricity generation have not fully been understood yet. In early studies of mechanisms of electricity generation using exoelectrogenic microorganisms, chemical mediators or electron shuttles thought to be needed to transfer the electron to the electrode. However, it was shown that adding mediators is not needed for all exoelectrogens as some can produce their own mediators (Rabaey et al., 2005; Rabaey et al., 2004b). So far there are three known mechanisms to send the electron to the available electron acceptor (electrode in MFC): using nanowires (Bretschger et al., 2007; Reguera et al., 2005), direct contact through membrane-bound electron transfer (Kim et al., 2002), and mediators produced by the bacteria (Rabaey et al., 2004b; Rabaey et al., 2005). Under anaerobic condition Shewanella purtefaciens is transferring the electrons to an insoluble electron acceptor by membrane-bound cytoplasm cytochromes available in its outer membrane (Kim et al., 2002). Nanowires are conductive appendages that are in iron reducing bacteria such as Geobacter and Shewanella species (Bretschger et al., 2007) as well as photosynthetic (non iron reducing) microorganisms (Bretschger et al., 2007). Pseudomonas aeruginosa has been shown to produce

27 13 its own mediator (mainly pyocyanin) that shuttles the electron to the electrodes (Rabaey et al., 2004b) MFC Electricity Generation with Mixed and Pure Culture There are a few known exoelectrogenic bacteria isolated directly from MFCs. Conventional methods of exoelectrogenic isolation are based on ability of these bacteria to respire using solid metal oxides on agar plates (Logan and Regan, 2006; Lovley, 2006; Logan, 2008). However, studies showed that not all exoelectrogenes can reduce iron and not all iron reducing bacteria can produce electricity in MFC (Bretschger et al., 2007; Richter et al., 2007). Electricity can be generated from a mixed culture of bacteria as well as pure cultures. Although there are few studies to compare MFC performance using pure or mixed cultures, the power produced from mixed cultures is thought to be higher than that produced by a pure culture (Zuo et al., 2008; Rabaey et al., 2004b; Rabaey et al., 2005). However, recent studies showed that pure cultures can produce electricity comparable or higher than that produced by a mixed culture (Nevin et al., 2008; Xing et al., 2008) by improving the MFC architecture. For instance, Geobacter metallireducens produced power (36-40 mw/m 2 ) comparable to the mixed culture bacteria (38 ± 1 mw/m 2 ) from wastewater; however, the internal resistance was extremely high (1286 Ω ) and was the main limiting factor not the bacteria (Min et al., 2005a). Using reactor with very low internal resistance (3 Ω) a maximum power density of 4310 mw/m 2 was produced in a two-chamber MFC using ferricyanide whereas using the same system P. aeruginosa strain KRA3 produced only 28 mw/m 2 (Rabaey et al., 2004b). An MFC reactor inoculated with Shewanella purtefaciens in a Mn 4+ -graphite anode and an air cathode produced 10.2 mw/m 2 (Park and Zeikus, 2002b), which was six fold lower than that produced with a sewage sludge (mixed culture) as inoculum in the same system (Park and Zeikus, 2003). Ochrobactrum

28 14 anthropi YZ-1 was isolated directly from U-tube MFC and it produced maximum power density of 89 mw/m 2 which was 17% lower than that produced with mixed culture in the same system (Zuo et al., 2008). There are also few exoelectrogenic strains shown to produce power higher than mixed cultures. For instance, in a recent study by Xing et al. (2008) Rhodopseudomonas strain DX-1 produced higher power density (2720 ± 60 mw/m 2 ) than mixed culture in the same device (Xing et al., 2008). Geobacter sulfurreducens produced a high power densities (1.8 W/m2) comparable to a mixed culture (Nevin et al., 2008) Improving Power Generation from MFCs There are several known factors to that can affect the MFC performance such as rate of substrate degradation, charge transfer from bacteria to the electrode on the anode side, proton mass transfer toward the cathode side, circuit resistance, MFC architecture (i.e. decreasing electrode spacing, advective flow through the anode) (Cheng et al., 2006b), anode surface area (Logan et al., 2007) and cathode materials (Cheng et al., 2006c; Cheng et al., 2006a; Zhao et al., 2005), solution conductivity (Liu et al., 2005), and presence or absence of an ion exchange membrane (Liu and Logan, 2004a; Oh and Logan, 2006). Therefore, the performance of each of these parameters influences the output power generation from MFCs. For instance, decreasing the electrode spacing from 4 cm to 2 cm in a single chamber MFC without PEM has resulted in an increase in the power generation by 67% (Liu et al., 2005). Effect of cathode performance, anode performance, and internal resistance on power generation is described in the following sections.

29 Improving Anode Performance Anode performance is improved by increasing the positive charge of the anode surface to adhere more negatively charged exoelectrogenes. This would reduce the acclimation time needed to generate maximum power. Addition of chemicals to the anode surface improved the anode performance. Applying chemicals (i.e. Mn +4, anthraquinone-1, 6-disulfonic acid (AQDS), Fe 3 O 4 or Fe 3 O 4 and Ni 2+ ) to the anode increased bacterial adhesion to the anode, decreased acclimation time, and increased the electrode transport to the anode (Park and Zeikus, 2002a; Lowy et al., 2006). Recently, it was shown that treating anode material by ammonia gas at a high temperature (700 C) decreased the acclimation time by 50% and increased the maximum power density 20%. This was probably due to the increase of the positive charge of the anode surface that likely increased the adhesion of the negatively charged bacteria to the anode (Cheng and Logan, 2007). Furthermore, using graphite fiber brush (ammonia treated) to increase the anode surface area concerted with other previously found factors to increase the power, was resulted in 2.4 W/m 2 power density, normalized to the cathode surface area (Logan et al., 2007) Improving Cathode Performance Cathode performance has been shown to be the main limiting factor in MFC power generation (Rismani-Yazdi et al., 2008; Rabaey and Keller, 2008). In conventional MFCs cathodes are placed in the aqueous phase where they are sparged with air in order to provide oxygen for the electrode. More effective catholytes, such as ferricyanide, can be used to increase the power density. Using ferricyanide power was increased 50-80% versus than that produced by dissolved oxygen (Oh et al., 2004). The highest power density achieved using ferricyanide in two-chamber MFC was 4310 mw/m 2 (normalized to the anode surface area) (Rabaey et al., 2004a). There are other catholytes as well such as ferric iron (Heijne et al., 2006), manganese

30 16 (Rhoads et al., 2005) and permanganate (You et al., 2006), however, the MFC must have a proton exchange membrane (PEM) or chemical exchange membrane (CEM) in order to prevent any diffusion of aqueous catholyte to the anode side, and the catholyte should be replaced as soon as it is reduced resulting in an increase the operational cost of the MFC. Therefore, oxygen is considered to be the most sustainable electron donor in MFC (Zhao et al., 2006). To increase MFC performance while keeping the cost of the device low, Liu and Logan (2004) showed that an MFC can be operated using an air cathode. In their initial results, they produced 262 mw/m 2. Moreover, they showed the power could be further increased to 494 mw/m 2 if the PEM was omitted from the system (Liu and Logan, 2004). Columbic efficiency and power density can be significantly increased by preventing oxygen flux to the anode when using an air cathode, and by minimizing water evaporation in aqueous cathode systems which creates a gas phase in the anode chamber. The columbic efficiency calculated for an air cathode MFC with the PEM was 40-55% while that without the PEM was 9-12%. The low columbic efficiency for a MFC without the PEM indicated oxygen was diffusing to the anode chamber when the PEM was not in place. To better isolate the anode from oxygen, a hydrophobic diffusion layer was implemented at the cathode. This hydrophobic diffusion layer decreased water flooding of the catalyst and increased the maximum power and columbic efficiency. The results showed that the addition of 4 diffusion layers increased the power generation by 42% (766 mw/m 2 ), which was the optimum improvement among other numbers of diffusion layers (Cheng et al., 2006). Using two layers of cloth on cathode surface further isolated the anode from oxygen while allowing to have smaller electrode spacing and resulted in a coulombic efficiency of 71% with maximum volumetric power of 672 W/m 3 (1120 mw/m 2 ) when operated in fed batch mode and 1010 W/m 3 (1800 mw/m 2 ) in a continuous flow mode (Fan et al., 2007).

31 17 To eliminate use of Pt catalyst while keeping the cathode performance the same, transition metal carbon cathodes such as iron(iii) phthalocyanine (FePc) and cobalt tetra-methyl phenylporphyrin (CoTMPP) were used and produced the power comparable or higher than that produced using Pt catalysts at. At current density higher than 0.2 ma/cm 2 CoTMPP performed better than FePc (Zhao et al., 2005). CoTMPP was also used as catalysts in an air cathode and it performed better than apply 0.1 mg-pt/cm 2 and it was only 12% lower in power generation than applying 0.5 mg-pt/cm 2. Cathode surface area affects the cathode performance especially when plain carbon cloth without catalyst is being used. Replacing the carbon cathode with catalyst with plain carbon cloth reduced the current and power at least by a factor of 10, however, increasing cathode surface area substantially could improve the power (Logan, 2008). Increasing the cathode surface area from 22.5 cm 2 to 67.5 cm 2 resulted in an increase in power of 24% (Oh et al., 2004). In sediment fuel cells, using long carbon brush cathode improved the cathode performance and longevity (Reimers et al., 2006). Using a tubular cathode also increased the cathode surface area and produced about two times (18 W/m 3 ) power when using two tubular cathodes and a brush anode, as compared to the same reactor with carbon paper cathode (9.9 W/m 3 ) (Zuo et al., 2007) Decreasing Internal Resistance Another factor that affects the maximum power density is internal resistance. In order to have high power generation internal resistance should be decreased. There are two primary parameters that will change the internal resistance of an MFC: the electrode spacing; and the solution ionic strength.

32 18 The maximum power output in a single chamber MFC was increased from 720 mw/m 2 to 1330 mw/m 2 when the solution strength was increased from 100 to 400 mm (Liu et al., 2005). Furthermore, increasing phosphate buffer concentration from 50 mm to 200 mm that caused the solution conductivity increase from 7.5 ms/cm to 20 ms/cm improved the power from MFC (Cheng and Logan, 2007; Logan et al., 2007; Feng et al., 2008). However, increasing the conductivity above 40 ms/cm showed to have adverse effect on power generation due to increase of working potential for anode and decrease the voltage (Oh and Logan, 2006). Decreasing the electrode spacing also can decrease the internal resistance. Reducing the electrode spacing from 4 to 2 cm increased the power from 720 to 1210 mw/m 2. Further decrease in the electrode spacing from 2 to 1 cm adversely affected power generation probably due to the diffusion of oxygen from the cathode to the anode (Liu et al., 2005). To solve this problem the same reactor with 1 cm spacing between electrodes was operated in continuous mode with conducting flow directly from the anode to the cathode and resulted to reach maximum power of 1540 mw/m 2 (Cheng et al., 2006b). Another solution was to use a cloth to keep the electrodes separated, and to reduce the oxygen diffusion to the anode. The maximum power density achieved in that study when two cloth-electrode-assembly (CEA) was used was 1120 mw/m 2 for fed batch and 1800 mw/m 2 for continuous flow (Fan et al., 2007) Applications of Microbial Fuel Cells Microbial fuel cells are promising technology for simultaneous wastewater treatment and electricity generation. Liu et al (2004) demonstrated that MFC can be used to treat wastewater to the acceptable level. Combing wastewater treatment with electricity generation from MFC can capture the energy, product of biodegradation of organic/inorganic matter in wastewater, and turn an energy consuming facility to an energy producing one. Using an MFC, energy will be

33 19 captured from a wastewater treatment plant that will be sufficient to power the facility (Logan, 2008). Another practical application of an MFC is to power a small oceanography device in a remote marine location through sediment microbial fuel cell. There are also a few studies to examining the application of MFCs for bioremediation by reduction of chemical at the cathode or oxidation of chemical at the anode. For instance, the idea of in-situ nitrate bioremediation using biocathode was explored using mixed culture or pure culture of Geobacter metallireducens, and an applied potential ~0.7 V. MFC also was suggested to be used for bioremediation of petroleum contaminated groundwater by oxidation of the chemical at the anode (Jin and Morris, 2007) Wastewater Treatment Although the energy generated by glucose or acetate, per unit COD, is much higher than that produced by domestic wastewater, the fact that wastewater treatment is accomplished at the same time as electricity generation makes this an attractive application. There are several advantages in using MFCs for wastewater treatment including producing energy while treating wastewater, no need for aeration, lower sludge production as MFC is an anaerobic process that produced less solid compared to aerobic process, and removal of odor (Kim et al., 2008) by reducing chemical associated with odor (Logan, 2008). So far, application of MFCs has been examined in wastewater treatment from domestic wastewater (Liu et al., 2004a), swine wastewater (Min et al., 2005b), meat packing wastewater (Heilmann and Logan, 2006), food processing wastewater (Kim et al., 2004), brewery wastewater (Feng et al., 2008), and paper recycling wastewater (Huang and Logan, 2008b). In early MFC studies, 80% COD was removed from primary clarifier of wastewater treatment plant while producing 26 mw/m 2 (Liu et al., 2004a). Using swine wastewater 86 ± 6% soluble COD was removed combined with 83 ± 4%

34 20 ammonia removal (Min et al., 2005b). Brewery wastewater was also treated by MFC by maximum COD removal of 98% while producing power (155 mw/m 2 ) (Feng et al., 2008). In another study using paper mill recycling wastewater 73 ± 1% soluble COD was removed while producing 501 ± 20 mw/m 2 power (in 50 mm PBS) (Huang and Logan, 2008a) Sediment Microbial Fuel Cell (SMFC) Different chemicals can be reduced at different depths of marine sediment and leave the system; however, the remaining byproducts will cause a voltage gradient of up to 0.75 V that could be captured in a device named the sediment microbial fuel cell (SMFC) (Reimers et al., 2001). In this device, a simple anode such as graphite electrode is inserted into the anoxic condition of the sediment while the cathode is placed into the seawater to capture the electron coming from the anode using an external circuit. The microbial communities of marine sediment, primarily Fe (III) reducing microorganism such as Geobacteraceae, are capable of oxidizing organic matter to generate electricity in SMFC (Bond et al., 2002). Most of the sediment accumulated in the seafloor contains % organic carbon by weight (Yen, 1977). A sediment area with only 2% organic carbon has an energy density of J/L if complete oxidation occurs (Cai and Reimers, 1995). The organic matter of the marine sediment is the source of electron donors in an SMFC device. Decay of these organic materials will slowly release soluble electron donors into the sediment over time. SMFCs in real world applications would be designed based on this slow release of electron donors, with initial power production typically even lower as the microbial community is established, with a ramp-up period that can last for weeks or even months.

35 21 The first generation of SMFCs provided stable power output at around mw/m 2 of anode surface area (Reimers et al., 2001; Tender et al., 2002). Although, the power output was so low, these initial researchers recognized that the SMFC output might be enhanced by improving sediment composition, electrode design, and temperature (Reimers et al., 2001). For instance, they showed that by decreasing the temperature the power density was decreased as well. Also, as shown by Reimers et al. (2001), the power density might be different by using different sediment compositions. For instance, power density with 3-3.5% organic matter content was stabled at 5 mw/m 2 while with 5-6% organic matter content power was 10 mw/m 2, which indicated higher organic matter resulted to have higher the power generation. Lowy et al. (2006) modified the anode of SMFCs, experimenting with different anode compositions to achieve higher power density. The three different anode modifications used in their experiments were: graphite modified by adsorption of anthraquinone-1, 6-disulfonic acid (AQDS) or 1,4-naphthoqinone (NQ), graphite-ceramic anode composite containing Mn 2+ and Ni 2+, and a graphite paste containing Fe 3 O 4 or Fe 3 O 4 and Ni 2+ on graphite anode. The maximum power generated when using AQDS was approximately 98 mw/m 2 with a cell voltage of 0.24 V, and 105 mw/m 2 when the system was operated with Mn 2+ and Ni 2 anode at 0.35 V. However, over time the voltage from these modified anodes dropped to the level of the unmodified graphite anode (Lowy et al., 2006). Modification of temperature to enhance performance of SMFCs does not appear to have been investigated, but increasing temperature to the optimum for the SMFC microbial community would likely improve kinetics as it does in other microbial processes. Tender et al. (2002) demonstrated the effect of sediment composition, investigating two different marine sediments collected from different places. The sediment from the Yaquina Bay Estuary near Newport generated the higher power density, about 28 mw/m 2, while sediment

36 22 from the Tuckerton, New Jersey generated maximum power densities of 26.6 mw/m 2 (Tender et al., 2002). 2.3 Complex Biomass Complex biomass such as chitin and cellulose are among the most abundant biopolymer materials in the world. However, linkages between the monomers of these materials make them hard to degrader to simpler, less recalcitrant carbohydrates. For the same reason there is not enough study in MFCs examining the power generation using complex carbohydrates as fuel. Therefore, the main goal of this dissertation was to use complex carbohydrates, especially chitin and cellulose, inside MFCs Cellulose Polysaccharides containing only one kind of monosaccharide are called homopolysaccharides. Cellulose, a homopolysaccharide of glucose units, is the most abundant carbohydrate in the world. Cellulose is the main structure of the plant cell wall, consisting of cellobiose which is two glucose molecules linked by β (1-4) bonds (figure 2-2). Cellulose has a tendency to have intra- and intermolecular hydrogen bonding. Based on the density of these bonds, cellulose can be crystalline, a result of highly packed structure, or amorphous which is a low packing structure (Brown, 2003). Plant material can have both crystalline and amorphous cellulose. The crystalline cellulose is the most difficult part to degrade and greatly effects cellulose degradation rate. Therefore, a higher crystalline fraction, results in a lower cellulose degradation rate. Furthermore, the crystalline cellulose is insoluble in most solvents.

37 23 Figure 2-2 Small part of Cellulose structure. The two chains can rotate relative to each other (Nelson and Cox, 2004) Chitin Chitin is a homopolysaccharide with a linear structure (Figure 2-3). Chitin is the second most abundant biomaterial in the world after cellulose/lignocellulose (Dew, 2004). It is a slowly degradable material which can be used as an electron donor in biological processes. For instance, chitin has been used to treat sediment and groundwater contaminated by toxic cleaning solvent tetrachloroethene (PCE) (Brennan et al., 2006). It is also widely available as a seafood industry byproduct in coastal areas. Chitin is a homopolysaccharide with a linear structure. The monosaccharide component of chitin is N-Acetyl-β-D-glucosamine. N-Acetyl-β-D-glucosamine has the chemical formula C 8 H 13 NO 5 and a molecular weight of g. Chitin can be dissolved in strong acids such as hydrochloric acid, sulfuric acid, acetic acid, and 78% to 97% phosphoric acid (Yue-ping et al., 2003), but it cannot be dissolved in water, weak acid, alkali, ethyl alcohol, or other organic solvents. It is biodegradable by lysozyme and the products have no toxicity. This complex biomass varies in size and has both rapid and slowly degradable components. Depending on the particle size and its pretreatment, the ratio of rapid to slowly degradation fractions may vary for different chitin feedstocks. For instance, chitin SC 20 has a larger particle

38 24 size than SC 80, and is more rapidly biodegradable since it contains more protein and volatile acids than SC 80. Figure 2-3 A short segment of chitin. (Nelson and Cox, 2004) Hydrolysis Hydrolysis occurs when molecules are broken apart through reactions with water, and in biological processes will convert polymers like carbohydrates and proteins to their subunit monomers. These reactions are used to degrade the complex structure of lignocellulosic material through biomass pretreatment and saccharification. Three of the hydrolysis strategies used for biomass conversion is acid hydrolysis, alkaline hydrolysis, and enzymatic hydrolysis. Acid Hydrolysis Sulfuric acid (H 2 SO 4 ) and hydrochloridic acid (HCl) are two relatively low cost acids used to degrade lignocellulosic material. These acids can be used in either concentrated or diluted forms. However, there are several problems associated with concentrated acid hydrolysis. These acids are toxic, corrosive and hazardous, and the large quantities used in concentrated acid hydrolysis need to be recovered to achieve an economically favorable process (Sun and Cheng,

39 ). These problems are less acute with dilute acid hydrolysis (where acid concentrations are typically less than 1%), which can directly hydrolyze hemicellulose and facilitate subsequent enzymatic hydrolysis of cellulose. Alkaline Hydrolysis Alkaline hydrolysis hydrolyzes the lignocellulosic material by saponification. In this process, the ester bonds between lignin and hemicellulose are broken which increase the porosity of lignocellulosic material (Fan et al., 1987). Diluted NaOH and ammonia have been used as alkaline solutes to separate the structural linkage between lignin and carbohydrates in lignocellulosic material. Using diluted NaOH in hardwood decreased the lignin content from 24-55% to 20%. However, this treatment was not as effective for softwood with lignin content more that 26%, presumably because lignin is not easily accessible in softwoods (Millet et al., 1976). Delignification efficiency was 60-80% for corn cobs and 65-85% for switchgrass when using ammonia to break the lignin linkages (Iyer et al., 1996). Enzymatic Hydrolysis Enzymatic hydrolysis is another alternative for biomass hydrolysis. Cellulases are the enzymes responsible for cellulose cleavage, while hemicellulases are responsible for hydrolysis of hemicellulose. The cost associated with enzymatic hydrolysis is low compared to other alternatives, since this process can be successfully done under mild condition (ph 4.8 and temperature C) without having corrosion problem (Duff and Murry, 1996). Enzymatic hydrolysis is more effective if it is used after another hydrolysis pretreatment strategy. For instance, after alkaline hydrolysis pretreatment, the porosity of lignocellulosic material will increase; so the hemicellulose and cellulose fibers will be more accessible to the relatively large

40 26 enzymes required for enzymatic hydrolysis. Many pretreatment strategies hydrolyzed much of the hemicellulose chemically, so subsequent enzymatic hydrolysis is focused on the cellulose fraction. There are a wide variety of bacteria and fungi capable of producing cellulases. These microorganisms can be aerobic or anaerobic, and either mesophilic or thermophilic. Cellulase actually refers to a combination of three enzymes types: endoglucanase, cellobiohydrolase, and β-glucosidase, all of which work in sequence to hydrolyze cellulose to glucose (Reese, 1975). First an endoglucanase like endo 1, 4-β-D-glucane will cleave one of the internal glycosidic bonds from an unbroken glucan chain in a cellulose microfiber. Second, an exoglucanase like 1, 4-β-D-glucancellobiohydrolase will attach to the newly exposed ends of the glucan chain to cleave and release cellobiose dimmers. Finally, β-glucosidase will hydrolyze the cellobiose dimmers to glucose. In 1981 White and Brown visualized, monitoring by microscopy, the cellulose cleavage by this cellulase system and showed that each individual enzyme could not hydrolyze cellulose if the others were missing (White and Brown, 1981). Sun and Chang (2002) listed the factors that affect enzymatic hydrolysis as follows: substrate, cellulase activity, and reaction conditions such as temperature and ph. 2.4 Storage of complex biomass (Ensilage) Ensilage is one of several solid state fermentations used for preservation or conversion of biomass. Solid State Fermentation (SSF) refers to a microbial process that occurs on solid material with sufficient moisture for microbial growth, but in the absence or near-absence of free water (Pandey et al., 1999). Other SSF processes include composting and several food fermentations that preserve or produce sauerkraut, kimchee, soy sauce, and other products.

41 27 These SSFs can be classified as natural SSF or pure culture SSF depending on the microbial diversity allowed. There are several applications in which each of these strategies has been applied. Ensilage, kimchee, sauerkraut, and composting processes are couple of examples of natural SSF, where mixed wild cultures live on biomass substrates that have not been sterilized or pastuerized. Producing high-value end products from industrial SSF is usually best attained with pure cultures (Pandey, 1992), but these high value products are difficult to recover from the process (Pandey et al., 2000). In order to use biomass feedstocks for a manufacturing process that will operate year round, there is a need to preserve them after the harvesting season. Ensilage, an example of natural SSF, is an ancient biomass preservation process with more than 3000 years of history and proposed to be a low cost method for the storage of biomass (Richard et al., 2001). Ensilage can be used to preserve green plant materials, crop residues, or domestic wastes for animal feed or other purposes. Effective preservation requires appropriate SSF conditions including optimum moisture, ph, temperature, and nutrition for the microbial ecosystem. Sealing the silage material from air can provide anaerobic conditions, and lactic acid fermentation can reduce the ph to about 4, which provides long-term preservation (Ren, 2006). Lactic acid bacteria are naturally occurring in biomass feedstocks and the resulting ensiled material. These bacteria ferment the water soluble carbohydrates (WSCs) in the feedstock into organic acids. Organic acids are very important in the ensilage process since they keep the ph low enough to inhibit the growth of undesired microorganisms such as clostridia and enterobacteria. The reason that these microorganisms are not desirable for the process is that they can consume lactic acid, therefore increasing the ph, and thus providing a suitable environment for growth of other microorganisms that can further decrease the biomass dry matter and quality.

42 28 After ensilage, if feedstocks are being used for ethanol production, organic acids should be removed from the material. If organic acids remain in the system, they will keep the ph low and inhibit the growth of some of the organisms that might be used for ethanol production. The generated organic acid can be used inside MFC for electricity generation Modeling Organic Acid Generation during Ensilage A predictive simulation model to predict the organic acid generation from ensilaged feedstock was developed by Pitt and coworkers (Pitt et al., 1985; Leibensperger and Pitt, 1987). These models simulate the majority of microbial and biochemical process during ensilage and it was designed to operate on different type of feedstock including legume, mixed grasses, or whole plant corn. The models could predict the changes in ph, hemicellulose degradation, soluble sugars and organic acids, growth and death of clostridia and lactic acid bacteria, of different to explain the changes in ph, were based on biodegradation of feedstocks. 2.5 Challenges to use Complex Carbohydrates as Substrate in Microbial Fuel Cells Any biodegradable material can be used in an MFC as fuel to produce electricity. These materials might be complex or simple carbohydrates, or other bio-molecules such as proteins or lipids. To date, most of the studies that have been conducted in MFCs focus on simple sugars such as glucose. Relatively little MFC research has focused on complex carbohydrates, which comprise the major fraction of biodegradable materials in the world. While it is more difficult to work with complex carbohydrates as the fuel in MFCs, the rewards are potentially great, since so much of the earth s biomass is in these complex carbohydrate forms. There is some promising initial evidence that biomass crops and residues can also be used as feedstock for MFCs. Zuo at el. (2005) demonstrated that electricity generated from corn

43 29 stover hydrolysates, which primarily contain hemicellulose, is comparable to the electricity produced with glucose. These researchers found that the maximum power densities achieved with neutral hydrolysates and acid hydrolysates were equal to 475 and 422 mw/m 2, respectively. These values were slightly lower than those observed for the same type of MFC operated with glucose, which had a maximum power generation of 494 mw/m 2. Zuo et al (2005) were able to increase the power output to 810±3 mw/m 2 and 861±37 mw/m 2 for neutral and acid hydrolysates, respectively, by using a diffusion layer to reduce oxygen transfer from the cathode. The potential energy production from corn stover was estimated at about kwh/year which equals $6.9 billion per year based on $0.15/kWh ( they assumed all the carbohydrates of corn stover were converted to glucose, and that the energy recovery from the MFC is 10%) (Zuo et al., 2005). Maximum power densities produced in MFCs using cellulose have been lower than those achieved using soluble substrates. For example, a co-culture of cellulolytic fermentor Clostridium cellulolyticum and the exoelectrogen Geobacter sulfurreducens generated electricity, and the power generated using soluble cellulose (carboxymethyl cellulose; CMC), was comparable (143 mw/m 2, R= 1000 Ω) to that obtained with glucose. However, the maximum power density generated using insoluble cellulose (MN301) was 58% less (59 mw/m 2 ) than that with CMC. The removal of insoluble cellulose based on chemical oxygen demand (COD) was <30% and cellulose degradation was <50% (Ren et al., 2007). Power generation and cellulose degradation were lower when a mixed culture inoculum (wastewater sludge) was used instead of a co-culture. A consortium of rumen bacteria was found to produce electricity using cellulose (Rismani-Yazdi et al., 2007) at a power density (55 mw/m 2 ) comparable to that obtained using mixed soluble carbohydrates in the same reactor set up.

44 30 Although, the cellulose degradation rate was not measured in this study, it was claimed that power output was not limited by cellulose hydrolysis rate. It should be noted that in both of these previous studies by Rismani-Yazdi et al. (2007) and Ren et al. (2007) that ferricyanide was used as electron acceptor at the cathode. The absence of any oxygen being used in the system can be important, as oxygen diffusion into the anode chamber has been found to affect the enrichment and operation of MFCs for some substrates (Kim et al., 2007).

45 Literature Cited Allen, R M, and H P Bennetto Microbial Fuel-Cells: electricity production from carbohydrates. Appl Biochem Biotech 39: Bond, D R, and D R Lovley Electricity production by Geobacter sulfurreducens attached to electrodes. Applied Environmental and Microbiology 69: Bond, Daniel R, Dawn E Holmes, Leonard M Tender, and Derek R Lovley Electrodereducing microorganisms that harvest energy from marine sediments. Science 295 (5554): Bretschger, O, A Obraztsova, C A Sturm, I S Chang, Y A Gorby, S B Reed, D E Culley, C L Reardon, S Barua, M F Romine, J Zhou, A S Beliaev, R Bouhenni, D Saffarini, F Mansfeld, B Kim, J K Fredrickson, and K H Nealson Current production and metal oxide reduction by Shewanella oneidensis MR-1 wild type and mutants. Applied and Environmental Microbiology 73: Chaudhuri, S K, and D R Lovley Electricity generation by direct oxidation of glucose in mediatorless microbial fuel cells. Nat. Biotechnol. 21: Cheng, S, H Liu, and B E Logan. 2006a. Increased power and coulombic efficiency of singlechamber microbial fuel cells through an improved cathode structure. Electrochem Commun 8: Cheng, S, H Liu, and B E Logan 2006b. Increased power generation in a continuous flow MFC with advective flow through the porous anode and reduced electrode spacing. Environ Sci Technol 40 (7): Cheng, S, H Liu, and B E Logan. 2006c. Power densities using different cathode catalysts (Pt and CoTMPP) and polymer binders (Nafion and PTFE) in single chamber microbial fuel cells. Environ. Sci. Technol. 40 (1): Cheng, S, and B E Logan Ammonia treatment of carbon cloth anodes to enhace power generation of microbial fuel cells. Electrochem. Comm. 9: Dubois, M, K A Gilles, J K Hamilton, P A Rebers, and F Smith Colorimetric Method for determination of Sugars and Related Substances. Analytical Chemistry 28 (3): Duff, S J B, and W D Murry Bioconversion of Forest Products Industry Waste Cellulosics to Fuel Ethanol: A Review. Bioresource Technology 55:1-33. Fan, L T, M M Gharpuray, and Y-H Lee Cellulose Hydrolysis Biotechnology Monographs.

46 32 Fan, Y, H Hu, and H Liu Enhanced coulombic efficiency and power density of aircathode microbial fuel cells with an improved cell configuration. Journal of Power Sources 171 (2): Feng, Y, X Wang, and B E Logan Brewery wastewater treatment using air-cathode microbial fuel cells. Applied microbial biotechnology 78 (5): Heijne, A T, H V M Hamelers, V D Wilde, R A Rozendal, and C J N Buisman A bipolar membrane combined with Ferric Iron reduction as efficient cathode system in microbial fuel cells. Environ. Sci. Technol 40: Heilmann, J, and B E Logan Production of electricity from proteins using a single chamber microbial fuel cell. Water Environ. Res. 78 (5): Huang, L, and B E Logan. 2008a. Electricity generation and treatment of paper recycling wastewater using a microbial fuel cell. Applied Microbial Biotechnology 80 (2): Huang, L, and B E Logan. 2008b. Electricity generation and treatment of paper recycling wastewater using microbial fuel cell. Applied Microbial Biotechnology 80 (2): Iyer, P V, Z -W Wu, S B Kim, and Y Y Lee Ammonia recycled percolation process for pretreatment of herbaceous biomass. Appl Biochem Biotech 57/58: Jin, Song, and Jeff Morris The feasibility of using microbial fuel cell technology in bioremediation of hydrocarbons in groundwater. Environmental Science and Health, Part A:In press. Kim, B H, H S Park, H J Kim, G T Kim, I S Chang, J Lee, and N T Phung Enrichment of microbial community generating electricity using a fuel cell type electrochemical cell. Applied and Environmental Biotechnology 63 (6): Kim, H J, H S Park, M S Hyun, I S Chang, M Kim, and B H Kim A mediator-less microbial fuel cell using a metal reducing bacterium, Shewanella putrefaciens. Enzyme Microb. Technol. 30: Kim, J -R, J Dec, M A Bruns, and B E Logan Removal of odors from swine wastewater by using microbial fuel cells. Applied Environmental and Microbiology 74 (8): Kim, J R, S H Jung, B E Logan, and J M Regan Electricity generation and microbial community analysis of ethanol powered microbial fuel cells. Bioresour Technol 98 (13): Leibensperger, R Y, and R E Pitt A model of clostridial dominance in ensilage. Grass and Forage Science 42:

47 33 Liu, H, S Cheng, and B E Logan Power Generation in Fed-Batch Microbial Fuel Cells as a Function of Ionic Strength, Temperature, and Reactor Configuration. Environ. Sci. Technol. 39: Liu, H, and B E Logan. 2004a. Electricity generation using an air-cathode single chamber microbial fuel cell in the presence and absence of a proton exchange membrane. Environ. Sci. Technol. 38 (14): Liu, H, R Ramnarayanan, and B E Logan. 2004a. Production of electricity during wastewater treatment using a single chamber microbial fuel cell. Environ. Sci. Technol 38: Liu, Hong, and Bruce E Logan. 2004b. Electricity generation using an air-cathode single chamber microbial fuel cell in the presence and absence of a proton exchange membrane. Environmental Science & Technology 38 (14): Liu, Hong, Ramanathan Ramnarayanan, and Bruce E Logan. 2004b. Production of electricity during wastewater treatment using a single chamber microbial fuel cell. Environmental Science & Technology 38 (7): Logan, B E Microbial Fuel Cells. Edited by J. W. Sons. New Jersey, NY: John Wiley & Sons, Inc Logan, B E, S Cheng, V Watson, and G Estadt Graphite fiber brush anodes for increased power production in Air-Cathode Microbial Fuel Cells. Environmental Science & Technology 41: Logan, B E, and J M Regan Electricity-producing bacterial communities in microbial fuel cells. Trend Microbiol 14: Lovley, D R Bug juice: harvesting electricity with microorganisms. Nature Reviews Microbiology 4 ( ). Lowy, D A, L M Tender, J G Zeikus, D H Park, and D R Lovley Harvesting energy from the marine sediment water interface II Kinetic activity of anode materials. Biosens. Bioelectron. Millet, M A, A J Baker, and L D Scatter Physical and chemical pretreatment for enhancing cellulose saccharification. Biotech.Bioeng Symp. 6: Min, B, S Cheng, and B E Logan. 2005a. Electricity generation using membrane and salt bridge microbial fuel cells. Water Research 39 (9): Min, B, J -R Kim, S -E Oh, J M Regan, and B E Logan. 2005b. Electricity generation from animal wastewater using microbial fuel cells. Water Research 39 (20):

48 34 Nelson, D L, M M Cox Lehninger Principles of Biochemistry. Fourth ed. Nevin, K P, H Richter, S F Covalla, J P Johnson, T L Woodard, A L Orloff, H Jia, M Zhang, and D R Lovley Power output and columbic efficiencies from biofilms of Geobacter sulfurreducens comparable to mixed community microbial fuel cells. Environmental Microbiology doi: /j x. Oh, S -E, and B E Logan Proton exchange membrane and electrode surface areas as factors that affect power generation in microbial fuel cells. Appl Microbiol Biotechnol 70: Oh, S -E, B Min, and B E Logan Cathode Performance as a Factor in Electricity Generation in Microbial Fuel Cells. Environ. Sci. Technol. 38: Pandey, A Recent process developments in solid-state fermentation. Process Biochemistry 27: Pandey, A, W Azmi, J Singh, and UC Banerjee Types of fermentation and factors affecting it. In Biotechnology: Food Fermentaion, edited by V. K. Joshi and A. Pandey. New Delhi: Educational Publishers. Pandey, A, C R Soccol, and D Mitchell New development in solid state fermentation: I- bioprocess and products. Process Biochemistry 35: Park, D H, and J G Zeikus. 2002a. Improved fuel cell and electrode designs for producing electricity from microbial degradation. Biotech. Bioengin. 81 (3): Park, D H, and J. G. Zeikus. 2002b. Impact of electrode composition on electricity generation in a single-compartment fuel cell using Shewanella putrefacians. Appl. Microbiol. Biotechnol. 59: Park, D. H., and J. G. Zeikus Improved fuel cell and electrode designs for producing electricity from microbial degradation. Biotechnology and Bioengineering 81 (3): Pitt, R E, R E Muck, and R Y Leibensperger A quantitative model of the ensilage process in lactate silage. Grass and Forage Science 40: Potter, M.C Electrical effects accompanying the decomposition of organic compounds. Proc. Roy. Soc. London Ser. B 84: Rabaey, K, N Boon, S D Siciliano, M Verhaege, and W Verstraete. 2004a. Biofuel cells select for microbial consortia that self-mediate electron transfer. Appl. Environ. Microbiol 70:

49 35 Rabaey, K, and J Keller Microbial fuel cell cathodes: from bottleneck to prime opportunity? Water Sci Technol 57 (5): Rabaey, K, and W Verstraete Microbial fuel cells: novel biotechnology for energy generation. Trend Biotechnol 23: Rabaey, Korneel, Nico Boon, Monica Hofte, and Willy Verstraete Microbial phenazine production enhances electron transfer in biofuel cells. Environmental Science & Technology 39 (9): Rabaey, Korneel, Nico Boon, Steven D Siciliano, Marc Verhaege, and Willy Verstraete. 2004b. Biofuel cells select for microbial consortia that self-mediate electron transfer. Applied and Environmental Microbiology 70 (9): Rabaey, Korneel, Geert Lissens, Steven D Siciliano, and Willy Verstraete A microbial fuel cell capable of converting glucose to electricity at high rate and efficiency. Biotechnology Letters 25 (18): Reese, E T Biological Transformation of Wood,, edited by W. Liese. Reguera, Gemma, Kevin D McCarthy, Teena Mehta, Julie S Nicoll, Mark T Tuominen, and Derek R Lovley Extracellular electron transfer via microbial nanowires. Nature 435: Reimers, C E, P Gigruis, H A Stecher III, L M Tender, N Ryckelynck, and P Whaling Microbial fuel cell energy from an ocean cold seep. Geobiology 4: Reimers, C E, L M Tender, S Fertig, and W Wang Harvesting Energy from the Marine Sediment-Water Interface. Environ. Sci. Technol. 35: Ren, H Effect of Cell Wall Degrading on Enzymes and Chemicals on Corn Stover Preservation and Pretreatment During Ensilage Processing. Phd diss, Agricultural and Biological Engineering, Pennsylvania State University, University Park. Ren, Z, T E Ward, and J M Regan Electricity production from cellulose in a microbial fuel cell using a defined binary culture. Environ. Sci. Technol. 14 (13): Rhoads, A, H Beyenal, and Z Lewandowski Microbial fuel cell using anaerobic respiration as an anodic reaction and biomineralized manganese as catholic reactant. Environ Sci Technol 39: Richard, T L, S Proulx, K J Moore, and S Shouse Ensilage technology for biomass pretreatment and storage. In ASAE Paper no ASAE.

50 36 Richter, H, M Lanthier, K P Nevin, and D R Lovley Lack of electricity production by Pelobacter carbinolicus indicates that the capacity for Fe(III) oxide reduction does not necessarily confer electron transfer ability to fuel cell anodes.. Applied and Environmental Microbiology 73: Rismani-Yazdi, H, S M Carver, A D Christy, and oo H Tuovinen Cathodic limitations in microbial fuel cells: An overview. Journal of Power Sources 180: Rismani-Yazdi, H, A D Christy, B A Dehority, M Morrison, Z Yu, and O H Tuovinen Electricity generation from cellulose by rumen microorganisms in Microbial Fuel Cells. Biotech Bioeng 97 (6): Schröder, U., J. Niessen, and F. Scholz A generation of microbial fuel cells with current outputs boosted by more than one order of magnitude. Angew. Chem. Int. Ed. 42 (25): Sun, Y, and J Cheng Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresource Technology 83:1-11. White, A R, and R M Brown Enzymatic hydrolysis of cellulose: Visual characterization of the process. Cell Biology 78 (2): Xing, D, Y Zuo, S Cheng, J M Regan, and B E Logan Electricity generation by Rhodopseudomonas palustris DX-1. Environ. Sci. Technol 42 (11): You, S, Q Zhao, J Jiang, J Zhang, and S Zhao A microbial fuel cell using permanganate as the catholic electron acceptor. Power Sources 162: Zhao, F, F Harnisch, U Schroder, F Scholz, P Bogdanoff, and I Herrmann Application of pyrolysed iron(ii) phthalocyanine and CoTMPP based oxygen reduction catalysts as cathode materials in microbial fuel cells. Electrochem. Commun. 7: Zhao, F, F Harnisch, U Schroder, F Scholz, P Bogdanoff, and I Herrmann Challenges and constraints of using oxygen cathodes in microbial fuel cells. Environ Sci Technol 40: Zuo, Y, S Cheng, D Call, and B E Logan Tubular membrane cathodes for scalable power generation in microbial fuel cells. Environ. Sci. Technol 41 (9): Zuo, Y, P-C Maness, and B E Logan Electricity production from steam-exploded corn stover biomass. Energy & Fuels 20 (4): Zuo, Y, D Xing, J M Regan, and B E Logan Isolation of the exoelectrogenic bacterium Ochrobactrum anthropi YZ-1 by using a U-tube microbial fuel cell. Applied and Environmental Microbiology 74 (10):

51 37 Chapter 3 Substrate-enhanced Microbial Fuel Cells for Improved Remote Power Generation from Sediment-Based Systems Abstract A sediment microbial fuel cell (SMFC) produces electricity through the bacterial oxidation of organic matter contained in the sediment. The power density is limited, however, due in part to the low organic matter content of most marine sediments. To increase power generation from these devices particulate substrates were added to the anode compartment. Three materials were tested: two commercially available chitin products differing in particle size and biodegradability (Chitin 20 and Chitin 80), and cellulose powder. Maximum power densities using chitin in this substrate enhanced sediment MFC (SEM) were 76 ± 25 mw/m 2 and 84 ± 10 mw/m 2 (normalized to cathode projected surface area) for Chitin 20 and Chitin 80, respectively, versus less than 2 mw/m 2 for an unamended control. Power generation over a 10-day period averaged 64 ± 27 mw/m 2 (Chitin 20) and 76 ± 15 mw/m 2 (Chitin 80). With cellulose, a similar maximum power was initially generated (83 ± 3 mw/m 2 ), but power rapidly decreased after only 20 hours. Maximum power densities over the next 5 days varied substantially among replicate cellulose-fed reactors, ranging from 29 ± 12 to 62 ± 23 mw/m 2. These results suggest a new approach to power generation in remote areas based on the use of particulate substrates. While the longevity of the SEM was relatively short in these studies, it will be possible to increase operation times by controlling particle size, mass, and type of material needed to achieve desired power levels that could theoretically be sustained over periods of years or even decades. 1 Material presented in this chapter was published in the following paper: Rezaei, F., T.L. Richard, R. Brennan, and B.E. Logan Substrate-enhanced microbial fuel cells for improved remote power generation from sedimentbased systems. Environ. Sci. Technol. 41(11):

52 Introduction A microbial fuel cell (MFC) is a device that can directly produce electricity from the bacterial oxidation of organic matter such as glucose or acetate (Allen and Bennetto, 1993), or inorganic species such as sulfides (Rabaey et al., 2006). Reimers et al., (2001) first demonstrated that microbial communities naturally present in marine sediments could produce electricity (Reimers et al., 2001). Sediment MFCs consist of a non-corrosive but conductive anode such as a graphite rod embedded in an anaerobic marine sediment, and a cathode made of a graphite plate or carbon fibers (Hasvold et al., 1997) suspended in the overlying oxygenated seawater. The electrons released by the bacterial degradation of the organic matter flow from the anode to the cathode through an external circuit, while protons diffuse through the water between the electrodes. The electrons and protons then react at the cathode with oxygen, forming water. The microbial communities that produce power in sediment MFCs have primarily been classified as Fe(III) reducing bacteria in the Geobacteraceae family (Tender et al., 2002). Desulfurumonas spp. were found to be dominant on electrodes placing in marine sediments, while Geobacter species predominated on electrodes placed in freshwater sediments (Lovley, 2006). The first reported sediment MFCs provided stable power output of mw/m 2 (normalized to anode projected surface area) (Reimers et al., 2001; Tender et al., 2002). Although, the power out put was low, it was realized that power output could be improved by using sediments with higher organic matter contents, improving the electrode design, and selecting optimal temperatures (Reimers et al., 2001). In the first sediment MFC tests the organic matter content was 2-6% (Tender et al., 2002), a value which is already quite high relative to typical marine sediments. Using sediments with an organic matter content of 4-6%, Lowy et al. (Lowy et al., 2006a) showed that power densities could be increased up to 2.5 times by

53 39 modifying the anode with different metals or known mediators. The maximum power generated using AQDS (9, 10-anthraquinone-2, 6-disulfonic acid) bound to the anode was ~ 98 mw/m 2 (normalized to anode surface area), and ~105 mw/m 2 was produced using a ceramic-graphite composite anodes containing Mn 2+ and Ni 2+. However, the maximum voltage decreased rapidly over time (as the square root of time over several days). For example, while the AQDS-modified anode initially produced five times greater power than plain graphite, power decreased to that of the unmodified graphite anode within a few days (Lowy et al., 2006b). Power output was also improved by modifying the cathode. For example, higher voltages were sustained using brush cathodes containing graphite carbon fibers compared to spinal coated stainless steel wool cathodes due to a 50% reduction in the internal resistance (Hasvold et al., 1997). One obstacle to widespread distribution of sediment MFCs as remote power sources is the low organic matter content of many sediments. While sediments can range in organic carbon content from 0.1 to 10% by weight (Yen, 1977), many sediments have organic matter contents of 0.4 to 2.2% (Chiou and kile, 2000). Thus, a low concentration of organic matter could prohibit sufficient power generation in some locations. Cold seeps have been tested as potential sites for increased power generation due to their higher organic matter content (Reimers et al., 2006). Besides the limited numbers of such sites, the lifetime of sediment MFCs in these locations was reduced by anode passivation due to the build up of sulfide oxidation products (Reimers et al., 2006). A new approach for increased power generation by MFCs in sediments is proposed here based on including a particulate organic substrate within the anode matrix. Bacteria needed to act as the biocatalysts are already present in the sediment and seawater, but their growth rate can be limited by substrate availability and reducing power. We reasoned that the use of a particulate

54 40 substrate could provide a source of sustained fuel for a sediment MFC as the rate of degradation of particulate substrates is slower than that of soluble materials as breakdown is limited by particle surface area. Finkelstein et al. (Finkelstein et al., 2006) conducted studies where acetate was added to the reactor. However, while adding a soluble substrate could increase power over the short-term in a contained laboratory reactor, soluble substrates would be quickly lost by diffusion into the surrounding water in a system in the field. Power generation from particulate substrates has not been previously been examined in a sediment MFC. To examine the idea of this new type of a substrate enhanced microbial fuel cell (SEM) we examined power output using two different types of particulate substrates: chitin, an easily degraded material in marine environments; and cellulose. 3.3 Materials and Methods Substrates. Chitin is a polysaccharide of N-acetyl-β-D-glucosamine (C 8 H 13 NO 5 ), and it is the second most abundant material in the world after lignocellulose (Dew, 2004). It is widely available as a seafood industry byproduct in coastal areas and is readily degraded by marine bacteria (Svitil and Kirchman, 1998). Chitin has previously been used as a slowly degradable material for in situ bioremediation of tetrachloroethene (PCE) (Brennan et al., 2006). Depending on the particle size, pretreatment method, and exact chemical composition (ratio of rapidly to slowly degradable fractions), different forms of chitin can be used. In this study ChitoRem TM SC-20 (Chitin 20) and ChitoRem TM SC-80 (Chitin 80) were used as received (JRW Bioremediation, LLC, from Lenexa, KS, USA). Chitin 20 consisted of crushed crab shells containing approximately 20% chitin, 25% protein, 40% calcium carbonate, and 15% water. Chitin 80 was composed of deproteinized and demineralized crab shells containing approximately 95% chitin which were sieved to a particle

55 41 size less than 20 mesh (< 841 μm). Chitin 20 is more easily degraded than Chitin 80 due to its higher protein content. Cellulose particles were tested as received, with a diameter of 50 μm as specified by the manufacturer (Sigmacell cellulose, type 50 particle size, Sigma Aldrich) Electrodes. The anode was made of non-wet proofed carbon cloth (Type A, E-TEK, Somerset, NJ, USA) connected to an external circuit using a titanium wire (0.81 mm diameter, 99.7%, Sigma- Aldrich, St. Louis, MO, USA ). Carbon cloth (5 cm by 9 cm, total projected surface area of A an = m 2 ) was sewn together in the shape of a pillow (3 cm by 5 cm final size) and filled with 2 g of a particulate substrate (Figure 3-1), or left empty (control). The pillow-shaped anode was sewn closed and wrapped with titanium wire to connect it to the external circuit. The cathode was made of carbon paper (2 cm 5 cm, projected surface area of A cat = m 2 ) containing 0.35 mg/cm 2 of Pt on one side (10% of Pt/C catalyst, 30% wet-proofing) (E-TEK, Somerset, NJ, USA).

56 42 Cellulose Chitin 80 Chitin 20 Carbon cloth filled with chitin 80 Figure 3-1Chitin 80, cellulose, and Chitin 20 as substrates. Also shown is the carbon cloth anode filled with Chitin 80 before being sewn Medium. Natural seawater (Real Ocean Pure Seawater, PETCO.com) was amended with a phosphate buffer and nutrients (NH 4 Cl, 0.31 g/l; NaH 2 PO 4 H 2 O, 0.75 g/l) in order to ensure that nutrient limitations or ph changes did not affect power generation. Other nutrients and trace elements were assumed be available in the sediment or seawater. An anaerobic sediment from the Delaware Bay (kindly provided by David Kirchman, University of Delaware) with an organic matter content of 4.8 ± 0.42 % was combined with standard sand, sieved to a uniform size (~ 0.5 millimeters), in a 1:1 ratio SEM Construction and Operation. SEM Reactors (Kimax * GL 45 Media/Storage Bottle; 500mL capacity) were filled with 250 ml of a 50:50 mixture of sediment and sand mixture, and 250 ml of seawater (Figure 3-2).

57 43 The anode was placed at the bottom of the bottle and connected to the cathode using a copper wire sealed with plastic tubes. The circuit was completed using a 1000 Ω resistor as a load, with the voltage monitored across the resistor every 30 minutes using a data logger (ADC-16, Pico Technology Ltd). Tests were run in triplicate with a single, non-amended control reactor. Deionized water was added to the reactors to replace water lost to evaporation. A Cathode Seawater B Marine Sediment Anode Figure 3-2 The small scale sediment fuel cell with substrate (SEM), anode is embedded inside the marine sediment at the bottom (anaerobic phase) and cathode is suspended in the overlaid seawater (aerobic phase) Two-Chamber MFC Tests. Additional tests were performed using two-chambered MFCs to examine performance of these particulate substrates under conditions typically used for conventional (soluble) substrates. Particulate substrates (1 g/l) were added to media containing the following (g/l): NH 4 Cl, 0.31;

58 44 KCl, 0.13; NaH 2 PO 4 H 2 O, 4.97; Na 2 HPO 4 H 2 O, 2.75; and a mineral (12.5 ml) and vitamin (12.5 ml) solution as reported previously (Lovley and Philips, 1988). Anaerobic sludge (secondary clarifier) obtained from the Pennsylvania State University Wastewater Treatment Plant was used to inoculate these reactors. The two-chamber reactors were constructed as previously described using two media bottles (200 ml capacity) with side arms containing a Nafion membrane (Nafion 117, Dupont Co., Delaware; projected surface area of m 2 ) clamped between the tube ends (Oh and Logan, 2006). A graphite brush was used as the anode (25 mm diameter x 25 mm length, A an =0.22 m 2 ) (PANEX K, Zoltek, St. Louis, MO, USA) (Logan et al., 2006a). The cathode was 1 cm 9.5 cm carbon paper containing 0.35 mg/cm 2 Pt (10% of Pt/C catalyst, 30% wet-proofing; E-TEK, Somerset, NJ, USA) coated on one side (A cat = m 2 ) Analytics and Calculations. Volatile fatty acids (VFAs) were measured using a gas chromatograph (GC) (Agilent, 6890) as described previously (Oh et al., 2004). The concentration of organic matter in the marine sediment was measured as volatile solids (VS), based on differences in dry weight (110 C for 48 hr) and combusted weight (550 C for 8 h) for ten samples (10 g each). Solution conductivity was measured using a conductivity meter (OAKTON, CON6, Acron series), and kept between ms/cm (except as noted) as solution conductivity affects power density (Liu et al., 2005). Cell voltages (V) were measured using a data acquisition system (Pico- ADC 16, Alison Technology Corporation, Kingsville, TX) connected to a computer. Anode/cathode potentials were measured using a multimeter (83 III, Fluke, USA) using an Ag/AgCl reference electrode (RE-5B, Bioanalytical systems, USA). Current (i) was calculated as i=v/r, where R is the external circuit resistance. Power (P) was calculated as P= iv, and normalized by the cathode

59 45 projected cathode area (A cat ), as done in previous studies (Cheng et al., 2006; Logan et al., 2007). The maximum power density was measured by varying the external resistance between 100 Ω and 200 kω, and waiting until voltage was stable (~30 minutes). The total remaining substrate at the end of a test could not be directly measured due to the organic matter content of the inoculum. Therefore, coulombic efficiencies (CEs) were estimated by assuming that all substrate was completely degraded when the voltage was reduced to a low value (around 10 mv). CEs were calculated as previously described (Oh et al., 2004) assuming 32 moles of electrons produced per mole of chitin, and 24 moles of electrons produced per mole of cellulose. 3.4 Results Power Generation with Chitin. Power generation was observed with little lag with both chitin substrates, reaching a maximum stable power density within 80 hours (Figure 3-3). Average power densities during maximum power production (from 54 to 290 hours) in three reactors with Chitin 80 were 90 ± 4, 75 ± 7, and 60 ± 5 mw/m 2 (n= 470 for each reactor). Power densities from the three reactors were significantly different (p = 0.001, 95% CI, t-test) during maximum power production, for reasons not well understood. While distances between electrodes were set to be the same in all reactors, the motion of the water due to aeration may have moved the cathode or affected localized dissolved oxygen concentrations near the cathode. Replacing the cathodes with new cathodes, or switching them between reactors in this case did not affect power. Repeated tests with additional reactors also resulted in differences between duplicate reactors, suggesting that bacterial distribution and decomposition rates of the substrate were in fact controlling maximum

60 46 power densities and producing different results. In order to provide an assessment of the range of data in the different reactors, averages are shown with associated standard deviations based on 3 points for each symbol (Figure 3-3). For the combined reactors, the average maximum power density produced using Chitin 80 was 76 ± 15 mw/m 2. Power densities from the control reactors lacking substrate amendments were below 2 mw/m 2. Note that these power densities are normalized to the cathode projected surface area (one side, which contains the Pt catalyst). If the anode surface area is used, multiply these power densities by 0.66 (one side, folded anode), 0.33 (two sides, folded), or 0.11 (both sides, unfolded anode). No matter which area is used for normalization, in all cases power production with the particulate substrate was larger than that of the control. To further investigate the reasons of having difference among replications, six anodes were embedded into the sediment layered into a glass tank to make a uniform environment for the anodes. Six cathodes were suspended in the overlaying seawater. The results from this experiment showed similar power density with no significant differences among replications (Figure 3-4), indicating that the differences between replications in individual SEM reactors were due to the differences in sediment environment and bacterial distribution. With Chitin 20 as the substrate, average power densities for the three reactors were 94 ± 6, 56 ± 4, and 41 ± 6 mw/m 2 (n = 513, from 43 hr to 290 hr). Taken together, the power generation for these three reactors averaged 64 ± 27 mw/m 2. Again, there were significant differences in power production among the reactors (p = 0.001, 95% CI). However, the higher power produced by one of the reactors (94± 6 mw/m 2 ) this time was identified to be due to cathode performance. When the cathode from this reactor (at that time producing 110 mw/m 2 ) was moved to the reactor with the lowest power (41 mw/m 2 on average, but producing 71 mw/m 2 at that time), the power instantly increased from 71 to 107 mw/m 2. Thus, this cathode

61 47 was more effective than the other two cathodes, resulting in the observed differences in power generation. The reasons why the cathodes differed in performance are not clear as they were all made from the same commercially prepared piece of carbon paper. Power Density (mw/m 2 ) Chitin80 Chitin20 Control Time (hr) Figure 3-3 Power generation from SEM with Chitin 20 and Chitin 80 as substrate as well as a control (without substrate). Each line represents the average of triplicate reactors (error bars ±SD). Symbols are shown for each 10 data point, and standard deviations are shown for each 50 data point. 60 Power Density (mw/m 2 ) Time (hours) 3-4. Average power density from six anodes embedded into the sediment layered into a glass tank. The data is the average of six replications (error bars ±SD)

62 Power Generation with Cellulose. When cellulose was used as substrate, power was produced but levels were more erratic than observed with chitin. For two of the reactors, power generation increased over time following inoculation, but for the third reactor the power increased more slowly (Figure 3-5). The maximum power generation for the first reactor reached 98 ± 2 mw/m 2 (n=46, 50 to 73 hr), and was significantly greater (p = 0.004, 95% CI) than power produced by the other two reactors of 73 ± 4 mw/m 2 (n=140, 100 to 170 hr) and 78 ± 3 mw/m 2 (n=52, 50 to 76 hr). The power output for one reactor was again attributed to the efficiency of the catalyst. When the cathode from the reactor that had produced a maximum of 98 mw/m 2 was hooked to the reactor that had produced a maximum of 73 mw/m 2, power immediately increased (in this case from 9 to 42 mw/m 2 ). A comparison of the results from the two chitin substrates and cellulose substrate showed that maximum power densities using the different substrates was significantly different (p = 0.001, 95% CI). Power Density (mw/m 2 ) Rep 1 Rep 2 Rep Time (hr) Figure 3-5 Power generation from three replications of SEMs with cellulose as substrate.

63 Polarization and Power Density Curves. Polarization data were obtained by varying the circuit external resistance (Figure 3-6A). Data were obtained during start up when power in the three reactors was 86 mw/m 2 (Chitin 20), 91 mw/m 2 (Chitin 80) and 50 mw/m 2 (cellulose) with a 1000 Ω resistor. Power density curves where then calculated based on the voltages and current. Maximum power densities reflected these general differences in power production, with 87 ±10 mw/m 2 and 80 ±19 mw/m 2 for Chitin 80 and Chitin 20, and 45 ± 21 mw/m 2 for cellulose (Figure 3-6B). The internal resistance calculated based on using the slope of polarization curve indicates that power generation was limited by internal resistance (Cheng et al., 2006; Logan et al., 2006b) Over a range of to 0.6 ma (Figure 3-6B), the internal resistances from two reactors of each treatment were calculated as 646 ± 134 Ω and 1297 ± 442 Ω for the Chitin 80 and Chitin 20 substrates, and 1762 ± 901 Ω for cellulose substrate. Differences in cathode potentials in the reactors containing Chitin 20 and cellulose likely account for the large standard deviation among reactors in internal resistance. The average power production in these reactors was in general inversely correlated with internal resistance (p=0.01 the slope, n=5; data not shown), showing that internal resistance variations among reactors produced differences in power densities. Except for the situations noted above for the cathode, where power was increased when cathodes were switched, these differences were a consequence of the biological development of power generation with the particulate substrates.

64 50 (A) Power (mw) Chitin20 Chitin80 Cellulose Current (ma) (B) Voltage(mV) Chitin20 Chitin80 cellulose Current (ma) Figure 3-6 (A) Power density based on different resistance for all substrates. (Error bars are ±S.D. based on duplicate measurements). (B) Voltage based on different resistance for all substrates. (Error bars are ±S.D. based on duplicate measurements) Volatile Fatty Acids (VFAs). The degradation of chitin and cellulose requires hydrolysis of the particulate substrate, and this degradation step could result in diffusion of the substrate out of the sediment and its

65 51 accumulation in the overlying water. However, VFAs measured in the overlying water of the anode chamber in the SEM reactors were all less than 10 ppm (data not shown), indicating little accumulation of these components in the overlying water Power Generation in a Two-Chambered MFC. The power generated using two of the particulate substrates (Chitin 80 and cellulose) was further examined in a two-chamber MFC that has previously been tested with soluble substrates. The time to peak power production for the Chitin 80 and cellulose required substantially longer time for the two chamber system (320 or 577 hours) than with the SEMs (57 or 60 hours). Power production with a 1000 Ω resistor reached a maximum of ~35 mw/m 2 for both substrates (Figure 3-7). These maximum power densities are slightly lower than that achieved in this system using acetate (45 mw/m 2 ) (Min et al., 2005), but similar to that obtained with glucose (37 mw/m 2 ) (unpublished data), suggesting the rate of power generation with these substrates was limited more by internal resistance of the MFC than by substrate degradation kinetics. The CE for the two substrates was 10% for cellulose and 13% for Chitin 80. However, we cannot be certain that all the particulate substrate was fully degraded. These CEs are substantially lower than those measured using soluble substrates of 22% for glucose (unpublished data) and 53% using acetate (Oh and Logan, 2006).

66 52 Power Density (mw/m2) Chitin 80 Cellulose Time(hr) Figure 3-7Power generation in a two-chamber MFC with Chitin 80 or cellulose (1 g/l). 3.5 Discussion Power generation by sediment MFCs was substantially increased by the addition of particulate substrates into the anode material. Maximum power densities for SEMs ranged from 54 to 112 mw/m 2 for Chitin 20 and 80, while it varied from 78 to 101 mw/m 2 for cellulose. In contrast, the unamended control (4.8 % organic matter) produced <2 mw/m 2. In MFC tests with soluble substrates power production at a fixed external resistance is reproducible between reactors, but only after several repeated feeding cycles when reactors are operated in a fed batch mode (Cheng and Logan, 2007). Here we observed significant variability in reactor performance that was likely due to a combination of several different factors including: variations between reactors typical of start up conditions such as bacterial access to the substrate; substrate particle size distributions between reactors; and orientation of the cathode relative to the anode with water sparging and local differences in dissolved oxygen concentrations. Differences in power

67 53 output by the cellulose fed reactors were more apparent than with chitin-fed reactors. Part of this difference could be due to the nature of the cellulose versus that of the chitin. Cellulose powder used here contained both amorphous (8%) and crystalline (92%) regions, making it possible that initial and higher power production was associated with the more easily degradable amorphous portions (Brown, 2003), followed by lower power production associated with the more recalcitrant crystalline regions. In contrast, chitin is a polysaccharide that exhibits rapid and sustained degradation and therefore is associated with more constant power production. Although it was shown that adding particulate substrates improved power generation compared to unamended controls, the longevity of the SEMs was only 30 days under laboratory conditions. However, reactor longevity should be able to be controlled by formulating particles of specific sizes and composition to extend degradation times. Experience with non-aqueous phase liquids in subsurface environments, and globular immiscible liquids in streams and rivers has demonstrated that the lifetime of materials is dependent on mass transfer controlled by limited surface area (Logan, 1999; Thibodeaux, 1996). Thus, it seems likely that we should be able to engineer a particulate substrate for an SEM through control of particle size. For example, if we assume spherical particles, we could model the time for complete dissolution as (Logan, 1999): t s 1/ 3 3m0 4 c 4 K c 3 s 2 / 3 where ρ c is density of the particle (g/cm 3 ), K the mass transfer coefficient (cm/s), m 0 the initial particle mass (mg) calculated as m= ρ c πd 3 /6, where d is diameter of the particles, and c s the surface concentration. For a particle in a stagnant flow field, K=D/d, where D is the chemical diffusivity in water, so the lifetime of the particle is proportional to d 2. This analysis assumes

68 54 complete exposure of the surface for dissolution. The analogy between dissolution and microbial degradation is not perfect because the bacteria will consume the substrate as opposed to releasing it. However, we can think of this surface flux resulting from dissolution as being analogous to reduction of mass due to bacterial consumption. Using a lifetime for Chitin 80 of t s =30 d for d= 0.05 cm, (average Chitin 80 particle size) we estimate from t s ~d 2 that by increasing particle size to d=0.17 cm we could extend the lifetime of the particle to 1 year, or using d=0.78 cm we could extend the lifetime particle to 20 years. While further research is in progress to test this model, it provides a conceptual model for testing the scale up of the SEM to reactor sizes previously used in field tests. These SEM reactors do not overcome a major limitation of many MFCs which is high internal resistance. Our results indicated internal resistances ranging from 646 to 1762 Ω, which are a consequence of the electrode spacing (9 cm) and conductivity of the porous media and water. The effect of electrode spacing on power generation in MFCs is well known (Oh et al., 2004; Cheng et al., 2006), and thus it is a major concern for SEM design. There does not appear to be a simple solution here for decreasing electrode spacing of a sediment fuel cell as the anode must be located in the anaerobic bottom layer, while the cathode needs to be sufficiently distant that dissolved oxygen concentrations are high. There are no other reports of internal resistances for sediment MFCs developed by others, so we do not know to what extent this may be affecting power generation in other systems. However, decreases in internal resistance for sediment MFCs can be expected to produce commensurate increases in power as demonstrated in many different MFC tests. Although we have designed and tested the SEM with the anode imbedded in the sediment, there is no reason why a properly designed system could not be used with the SEM

69 55 suspended in water (on a buoy) or situated on top of the sediment. When placed in the sediment, it might also be possible to achieve long term power generation by seeding the sediment with particulate substrates rather than placing them in the anode material as done here. With proper design the anode chamber could allow for maintaining anaerobic conditions in the anode, with the water flow enhanced past the cathode if it is off the ocean floor. In addition, it might be possible under these circumstances to achieve a lower internal resistance by decreasing the electrode spacing. A combination of reduced internal resistance, and careful design of particulate substrate size and loading, will make SEMs effective as a widely distributable power source for both freshwater and marine studies. 3.6 Acknowledgements Support was provided in-part by the Pennsylvania Experiment Station. The authors thank D. W. Jones for his help in analytical measurements, and S. Cheng and J.-R. Kim for their helpful suggestions and comments. 3.7 Literature Cited Allen, R M, and H P Bennetto Microbial Fuel-Cells: electricity production from carbohydrates. Appl Biochem Biotech 39: Brennan, Rachel A, Robert A Sanford, and Charles J werth Chitin and corncobs as electron donor sources for the reductive dechlorination of tetrachloroethene. Water Research 40: Brown, R C Biorenewable Resources. Ames, Iowa: Blackwell publishing company. Cheng, S, H Liu, and B E Logan Increased power generation in a continuous flow MFC with advective flow through the porous anode and reduced electrode spacing. Environ Sci Technol 40 (7): Cheng, S, and B E Logan Ammonia treatment of carbon cloth anodes to enhace power generation of microbial fuel cells. Electrochem. Comm. 9:

70 56 Chiou, C T, and D E kile Contaminant Sorption by Soil and Bed Sediment: U.S. Department of the Interior, U.S. Geological Survey. Dew, S E Polysaccharides. In Chemistry: Foundations and Applications, edited by J. J. Lagowski. New York: Macmillan Reference. Finkelstein, D. A., L. M. Tender, and J. G. Zeikus Effect of electrode potential on electrode-reducing microbiota. Environmental Science & Technology 30 (22): Hasvold, f, H Henriksen, E Mevaer, G Citi, B f Johansen, T Kjfnigsen, and R Galetti Sea- Water battery for subsea control systems. Journal of Power Sources 65: Liu, H, S Cheng, and B E Logan Power Generation in Fed-Batch Microbial Fuel Cells as a Function of Ionic Strength, Temperature, and Reactor Configuration. Environ. Sci. Technol. 39: Logan, B E Environmental transport process. New Jersey, NY: John Wiley & Sons, Inc. Logan, B E, S Cheng, V Watson, and G Estadt Graphite fiber brush anodes for increased power production in Air-Cathode Microbial Fuel Cells. Environmental Science & Technology 41: Logan, B L, B Hamelers, R Rozendal, U Schroder, J Keller, W Verstraete, and K Rabaey. 2006b. Microbial Fuel Cells: Methodology and technology. Environ. Sci. Technol 40 (17): Lovley, D R microbial energizers:fuel cells that keep on going. Microbe 1 (7): Lovley, D R, and E J P Philips Novel mode of microbial energy metabolism: Organism carbon oxidation coupled to dissililatory reduction of iron and manganese. Appl. Environ. Microbiol 54 (6): Lowy, D A, L M Tender, J G Zeikus, D H Park, and D R Lovley Harvesting energy from the marine sediment water interface II Kinetic activity of anode materials. Biosens. Bioelectron. Min, B, S Cheng, and B E Logan Electricity generation using membrane and salt bridge microbial fuel cells. Water Research 39: Oh, S -E, and B E Logan Proton exchange membrane and electrode surface areas as factors that affect power generation in microbial fuel cells. Appl Microbiol Biotechnol 70: Oh, S -E, B Min, and B E Logan Cathode Performance as a Factor in Electricity Generation in Microbial Fuel Cells. Environ. Sci. Technol. 38:

71 57 Rabaey, K, K V D Sompel, L Maigien, N Boon, P Aelterman, P Clauwaert, L D Schamphelaire, H T Pham, J Vermeulen, M Verhaege, P Lens, and W Verstraete Microbial fuel cells for sulfide removal. Environ. Sci. Technol 40: Reimers, C E, P Gigruis, H A Stecher III, L M Tender, N Ryckelynck, and P Whaling Microbial fuel cell energy from an ocean cold seep. Geobiology 4: Reimers, C E, L M Tender, S Fertig, and W Wang Harvesting Energy from the Marine Sediment-Water Interface. Environ. Sci. Technol. 35: Svitil, A L, and D L Kirchman A chitin-binding domain in a marine bacterial chitinase and other microbial chitinases: implications for ecology and evolution of 1,4-bglycanases. Microbiology 144: Tender, L M, C E Reimers, H A StecherIII, D E Holmes, D R Bond, D A Lowy, K Pilobello, S J Fertig, and D R Lovley Harnessing microbially generated power on the seafloor. Nat. Biotechnol. 20 (8): Thibodeaux, L J ENVIRONMENTAL CHEMODYNAMICS - Movement of Chemicals In Air, Water and Soil,. Vol. 2nd Edition. N.Y: J. Wiley. Yen, T F Chemical aspects of marine sediments. In Chemistry of Marine Sediment, edited by T. F. Yen. MI: Ann Arbor Science Publisher.

72 58 Chapter 4 Effect the Particle Size on MFC Maximum Power Generation, Power Longevity, and Coulombic Efficiency 4.1 Abstract: Microbial fuel cells are promising technology for bioelectricity generation from organic substrates, sediments, and wastewaters. However, this technology still needs to be improved significantly to be commercially viable and make a significant contribution to energy demand. Complex carbohydrates such as cellulose and chitin are the most abundant biomaterials in the world. These substrates can degrade slowly over time to provide long term fuel for the reactor, replacing simple carbohydrates that are more expensive, less available and consumed faster. However, little is known about the fundamentals of power generation and MFC longevity using these materials as fuel. In this research, effect of three different chitin particle sizes was examined for MFC power generation and longevity. The results demonstrated that an increase in particle diameter from smallest size (0.28 mm) to largest size (0.78 mm) resulted in an increase in longevity from 9 to 35 days. Coulombic efficiency based on removal of chitin was also increased from 18% for the smallest particles to 56% for the largest ones. However, the maximum power generation was lower for large particle (201 mw/m 2 ) as compared to the small and medium size (301 and 285 mw/m 2 ). The longevity of the particles followed a fractal degradation model with a three dimension fractal dimension (D 3 ) of 2 D

73 Introduction: There have been significant improvements during the last few years in increasing the power generation from microbial fuel cells (MFC). An MFC is a device that generates electricity through microbial oxidation of organic or inorganic substrates. Each MFC consists of two electrodes (anode and cathode) connected to each other via an external circuit with a known resistance. These electrodes might also be separated from each other by an ion exchange membrane. In general, electricity is generated by electrons flowing from the oxidation of substrates in the anode across a circuit to the cathode side of the MFC. The abundance and degradation rate of the fuel being used inside an MFC should be among the most important factors in scaling up the system and making the MFC marketable. However, these factors are neglected in a majority of MFC studies where simple carbohydrates have been used frequently. Complex carbohydrates such as cellulose and chitin are the most abundant biopolymeric materials in the world and compose a large portion of wastewaters from many industrial and municipal sources, and hence could be an ideal fuel to be used in MFCs. However, there are few studies in which particulate substrates (i.e. cellulose) have been used inside MFCs (Rezaei et al., 2008a; Ren et al., 2007b; Rismani-Yazdi et al., 2007; Rezaei et al., 2007). Addition of particulate substrates might be useful in practical applications of MFCs, especially in remote areas, where a long term fuel that degrades slowly over time could power sensors or communications equipment. For example, power output from sediment fuel cells was increased by addition of particulate substrates (chitin or cellulose) in substrate enhanced microbial fuel cell (SEM) (Rezaei et al., 2007). While most researchers have been focused on improving maximum achievable power from the MFC reactor and none tried to increase the

74 60 power longevity, Rezaei,et.al (2007) proposed to increase the MFC longevity by increasing the particle size. Effects of substrate physical properties (i.e. bulk density or permeability) that are directly or indirectly related to particle size have been investigated and modeled for solid state fermentation system such as composting (Richard et al., 2004; Ahn et al., 2004), but it has not been previously examined in MFC. Therefore, in this research we added an important factor to improve the reactor performance by investigating the effect of particle size on MFC longevity and ability to model the longevity based on mass transfer model for dissolution of particles. Although it is challenging to predict the maximum power generation from particulate substrates (Logan, 2008), in the current study we develop an approach to predict the longevity of the MFC reactor by modeling the lifetime of substrate. The effect of particle size on MFC longevity, maximum power generation, coulombic efficiency, microbial community, type and concentration of intermediate product was monitored and analyzed. 4.3 Materials and Methods: MFC Construction and Operation: A single-chamber bottle MFC (320 ml capacity, Corning Inc., NY, USA) with brush anode and air cathode was assembled as described previously (Cheng and Logan, 2007). The anode was made of large carbon fiber brush 5 cm diameter, 7 cm length and total surface area A An = 1.06 m 2 (Logan et al., 2007) which was treated by ammonia gas under high temperature (700 C) (Cheng and Logan, 2007). The air-cathode (3 cm diameter) was made of 30 wt% wet proofed carbon cloth (type B-1B, E-TEK, NJ, USA ), coated with platinum catalysts and four diffusion layers (Cheng et al., 2006a).

75 61 Each reactor was filled with 300 ml of aqueous medium, contained 50 mm phosphate buffer (PBS; 2.45 g/l NaH 2 PO 4 H 2 O and g/l Na 2 HPO 4 ), 12.5 ml/l minerals and 5 ml/l vitamins (Lovley and Philips, 1988). Reactors were inoculated (10% v/v) once by wastewater from a secondary clarifier collected from Penn State Wastewater Treatment Plant Substrate: ChitoRem TM SC-20 (Chitin 20) was sieved and used as substrates in all reactors (JRW Bioremediation, LLC, from Lenexa, KS, USA). To separate different particle sizes, Chitin 20 was ground and sieved by US standard sieve series (Fisher Scientific Co., TX, USA), sieve numbers from 20 to 70 (S#20, 25, 30, 35, 40, 45, 50, 60, and S#70). Particles remaining on each sieve were collected and stored in separate bags. The particles retained on sieve sizes S# 25, 40 and 60 were selected for testing, representing large, medium, and small particles within the overall range. Based on a number distribution assumption (described herein), the average selected particle sizes (d, diameter) were calculated as 0.78 mm, 0.46 mm, and 0.28 mm respectively, using the following equation (Logan, 1999): d nd n (1) where d is average particle size (diameter), n is the number of particles of size d. Since the available surface area of particles affects the biodegradability of the substrate, Chitin 20 was added to the MFC based on equal surface area, calculated from the average particle size. Based on that, the concentration of Chitin 20 added to the reactors was 0.5, 0.82, and 1.4 g/l for small, medium, and large particles, respectively (Table 4-1) instead of equal mass. Duplicate reactors for each size were used. In all tests, the anode solution was changed

76 62 when the reactor was producing voltage lower than 150 mv. Longevity of the reactor then was defined as the operation time for a batch cycle. Before running the tests, all reactors were checked to insure similar anode and cathode performance by using 0.5 and 1 g/l of fine Chitin 20 particle (particles that passed through sieve S#70) and monitoring the maximum power generation and reactor longevity. In preliminary tests there was no significant difference between different concentrations (0.5 and 1 g/l) of chitin 20 using fine particle sizes (data not shown); therefore, the mass concentration was not a factor in power production Community Analysis: Samples were collected from the brush anode and also the bulk solution of one of each different particle size reactors. Bacterial DNA was extracted using PowerSoil DNA isolation kit (MO BIO Laboratories, CA, USA) according to the manufacturer s instructions. Polymerase chain reaction (PCR) was used to amplify the V6-V8 region of the 16S rdna gene using an icycler iq TM thermocycler (Bio-Rad Laboratories, US). Forward (GC968F) and reverse (1401R) primers (Watanabe et al., 2001), GC968F (5'- CGCCCGCCGCGCCCCGCGCCCGGCCCGCCGCCCCCGCCCCAACGCGAAGAACCTTA C-3') and 1401R (5'-CGGTGTGTACAAGACCC-3'), were used in the PCR as described previously (Zuo et al., 2008). PCR products were then separated using denaturing gradient gel electrophoresis (DGGE), with DCode universal mutation detection system (Bio-Rad Laboratories, US), the DGGE then was used to monitor the bacterial diversity in each samples (Zuo et al., 2008; Ren et al., 2007a). Intense bands in each sample were excised from the gel and the DNA then extracted by suspending the excised gel fragment in DI water overnight at 4 C. DNA samples were re-amplified by PCR using the same primer except for the forward primer

77 63 that lacked the GC clamp, as described elsewhere (Rezaei et al., 2008b). PCR products were purified using QIAquick PCR purification kit (QIAGEN, CA, USA) according to manufacturer s instructions and sequenced by an ABI 3730XL DNA sequencing machine (Applied Biosystems, US). Sequenced fragments were then identified using the BLAST program to search for the closest reported strains Analytics and Calculations: Voltage was measured across an external resistor (1000 Ω, unless mentioned otherwise) using a data logger (Keithley Instruments, OH, USA). Current was calculated as I=V/R and power was calculated as P=IV. Power was normalized to the cathode projected surface area (A Cat = 7.06 cm 2 ) as the cathode has been shown to be limiting factor in most studies (Rismani- Yazdi et al., 2008; Rabaey and Keller, 2008). To test the maximum achievable power from the reactor and its corresponding resistance, a polarization curve was obtained by varying the external resistance (4 Ω to 400K Ω) recording data after an hour allowing each voltage to be stable. Reactors were running as an open circuit for two hours prior to obtaining the polarization curve. At present there is no standard method to measure the concentration of chitin remaining in the solution. One approach to approximate the amount of chitin is to measure the dry matter bacterial cell content of the substrate containing chitin within each MFC. However, the mass of bacteria of the anode solution that is approximately twice the protein content (Guedon et al., 1999) should be subtracted from the result. In order to take this approach, at the end of each batch, the anode solution was homogenized using Branson Sonifier 450 (Branson ultrasonic, CT, USA), 1 ml of the solution was filtered (triplicate for each sample) using pre-weighed 0.2 µm polycarbonate filters and bacterial cell dry weight was measured based on standard procedures

78 64 (Reddy, 2007). A total of 1 ml of homogenized solution was used to measure protein content by Bradford Quick Start Bradford Protein (Bradford, 1976) assay at 595 nm absorbance (BioRad Laboratories, CA, USA). Coulombic efficiency was calculated for a batch cycle as (Logan et al., 2006; Cheng et al., 2006b): C E s M s Idt 0 (2) Fb v C es t an where M s is the molecular weight of substrate, Δc is the change of substrate concentration over the batch cycle over time=t b, F=Faraday s constant, V An is the volume of liquid in the anode, and b es =32 is the moles of electrons for chitin, according to the reaction C 8 H 13 NO H 2 O 8 CO 2 + NH H e - (3) The coulombic efficiency was calculated two ways, either from the mass of chitin removed at the end of the batch cycle, or Δc= C i - C f (CE rem ), or by assuming all of the added substrate was removed (CE add ). The concentrations of volatile fatty acids (acetate, butyrate, formate, propionate) and solvents (acetone, methanol, ethanol, n-propanol and butanol) were determined by gas chromatography (Liu and Logan, 2004) Modeling Particle degradation: The lifetime for complete degradation of particles by dissolution is (Logan, 1999; Thibodeaux, 1996): t s K dm 1 m0 c 0 wccs A( mc ) (4)

79 65 where K is the mass transfer coefficient (cm/s), m c the particle mass (mg), A the particle surface area, and c s the surface concentration. Assuming a spherical particle mass of particle is m = ρ c πd 3 /6, where d is diameter of the particles and A = πd 2,, we have that A(m c ) = π(6m/ρ π) 2/3. Assuming stagnant flow, the mass transfer coefficient K=D/d where D is diffusion coefficient. Therefore, the longevity of the sphere particle is: t s 1 K c wc eq c 6 2 / 3 m0 0 dm m c 2 3 c (5) Integrating equation (4) results in: t s 3m K c 1/ 3 0 s c 6 2 / 3 (6) based on particle mass, or t s 2 cd 2Dc s (7) based on particle diameter. Thus, we see that lifetime of spherical particles should scale as t s ~d 2. To predict the MFC longevity based on this model, the following assumptions were used: chitin density (ρ c ) of 1.12 g/cm 3, c s the surface concentration equal to g/cm 3 (estimated from Rezaei et al, (2007) results) and D = 10-5 cm 2 /s. Correlation between predicted and measured longevity (longevity of one complete batch cycle) was then calculated using Minitab14. Many particles are known to be fractal, and therefore a second model was developed base upon non-euclidean relationships. For fractal particles, the surface area of the particle was D2 assumed to be A = al where l is a characteristic length, and D 2 a two-dimensional fractal D3 dimension. The volume of the fractal was assumed to be V= bl, where D 3 is a three-

80 66 D3 dimensional fractal dimension. The mass of the particle is therefore m = bl. Using these relationships in eq. 3, the lifetime for a fractal particle is: t s D ( D D2 D D 3 3 D2 3( b) D3 m0 3 D2 ) KCs (8) Assuming stagnant flow K is equal to D/l, we have: t s D ( D D 3 3( b) D3 D2 1 2 ) DC s l (9) Collapsing all constants into a single value of ξ, the scaling relationship for longevity of a single particle based on fractal geometry is: D3 D2 1 ts l. (10) Notice that this scaling relationship is the same as derived above for a spherical particle when D 3 =3 and D 2 = Results: Effect of Particle Size on MFC Performance: The MFC loading was normalized based on particle surface area as the particle surface area affects biodegradability of substrate and power generation. Normalizing MFC loading based on particle surface area resulted in similar but not identical peak power production. The maximum voltage obtained from the MFC reactors fed with small particle size (d = 0.28 mm) with a 1000 Ω fixed external resistance was 461 mv (301 mw/m 2 ), slightly higher than that produced by the medium (d = 0.46 mm) size (448 mv; 285 mw/m 2 ). The largest chitin particle (d = 0.78 mm) produced less power (376 mv; 201 mw/m 2 ) than the other two sizes (Figure 4-1).

81 67 Voltage (V) Time (day) d = 0.28 mm d= 0.46 mm d = 0.78 mm Figure 4-1 Voltage generation and longevity of MFC reactors fed with different particle sizes of Chitin 20. Results from the polarization curves show that the maximum achievable power was ± 0.04 mw for small, ± 0.03 mw for medium, and ± 0.02 mw for large particle sizes (Figure 4-2). In all tests maximum power density was achieved when reactors were connected to a 500 Ω resistor (Figure 4-2).

82 Power (mw) R = 0.14 mm R = 0.23 mm R = 0.39 mm Voltage (mv) Current (ma) Figure 4-2 Polarization curve for MFC run on different Chitin 20 particle sizes. (Error bars are ±S.D. based on duplicate measurements). Longevity of the MFC reactors increased as the particle sizes increased (Figure 4-1) as expected from equation 5. The reactors fed with the smallest selected particles produced power for the least amount of time (9 days), followed by the reactors fed with medium size particles (15 days). The reactors treated with the largest particles lasted for 35 days before power generation

83 69 declined (Figure 4-1). At the end of the batch cycle, 0.3, 0.34 and 0.37 g/l substrate was consumed when the reactor was fed with small, middle, and large particles, respectively (Table 4-1). In all tests, there were no detectable VFAs measured at the completion of batch cycles. Methanol and small amount of acetone (less than 5 mg/l) were the only solvents detected (Table 4-1). The MFC fed with small particles accumulated the most methanol (120.7 ± 29 ppm), followed by reactors with medium (86.7 ± 3 ppm) and large particle sizes (77.5 ±3 ppm) (Table 4-1). Biomass accumulation was the highest in reactors fed with medium size chitin (127 ± 25 mg/l), slightly higher than that in reactors with small (115 ± 12 mg/l) and large particle sizes (112 ± 14 mg/l) (Table 4-1). Table 4-1 Concentration of initial added chitin, and end product (fermentation product and biomass)* Average Initial Chitin Particle size Substrate Biomass Methanol Acetone Consumed (diameter, cm) concentration *Units for all data are mg/l except as noted. Coulombic efficiency was calculated based on chitin removed (CE rem ) at the end of the batch cycle and the initial chitin added (CE add ) to the reactor, as well as by assuming there was complete degradation of the substrate. CE rem increased as the particle size increased (Figure 4-3). As indicated in Table 4-1, the maximum CE rem (56%) was achieved when largest selected chitin particle (d = 0.78 mm) was used. CE rem was 18% and 30% for particle sizes d = 0.28 mm and d = 0.46 mm, respectively (Table 4-1). CE add followed the exact same trend as CE rem but with lower values as indicated in Table 4-1 and shown in Figure 4-3.

84 70 60 Coloumbic Efficiency (%) CErem CEadd Particle Size (d, mm) Figure 4-3 Coulombic efficiency based on removed chitin measured at the completion of the batch cycle (CE rem ) and added chitin assuming all chitin was consumed (CE add ) for three different particle sizes Community Analysis of Dominant Bacteria in MFC: Results from DGGE bands showed that particle size did not affect the dominant bacteria of the solution (Figure 4-4). Bands number 1 to 5 were excised from the gel and analyzed further by sequencing. Bands number 1, 2, and 5 were common among three sizes whereas bands 3 and 4 were only common between small and medium particle size (Figure 4-4). Sequencing the bands and entering the result in BLAST program for band number 1 to 5 showed that their closest strains with the percent identity to each strain are: Uncultured bacterium clone AE1_aaa03e10 16s rrna (87%), Clostridium sticklandii (96%), Enterobacter cloacae strain E717 16S rrna gene (100%), Fusibacter paucivorans 16S rrna gene (93%), and Bacillus sp. R partial 16S rrna gene (91%) (Table 4-2).

85 71 Band 1 Band 3 Band 4 Band 2 Band 5 d=0.28 mm d=0.46mm d=0.78mm Figure 4-4 DGGE analysis on the anode solution of the reactors fed with different particle sizes. Arrow shows the bands that were exerted from the gel for sequencing analysis.

86 72 Table 4-2 BLAST results of the closest strain to sequenced band Band# BLAST Results Identity (%) 1 Uncultured bacterium clone AE1_aaa03e10 16s rrna 87 2 Clostridium sticklandii, small subunit rrna 96 3 Enterobacter cloacae strain E717 16S rrna gene, partial sequence Fusibacter paucivorans 16S rrna gene, partial sequence Bacillus sp. R partial 16S rrna gene Modeling Particle Longevity: Using equation 7 with the mentioned assumptions, the longevity for spherical particles was predicted to be 11, 27, and 69 days for average particle sizes of 0.28, 0.46, and 0.78 mm, respectively (Figure 4-5). MFC longevity was measured as the time for completion of one batch cycle and was 9, 15, and 35 days for particle size 0.28, 0.46, and 0.78 mm, respectively (Figure 4-5). The correlation of the observed and predicted longevity assuming sphere particles was R 2 = (p value < 0.05). A non-linear regression based on the measured lifetime and the particle size (d, mm) yields: 1.3 t s 117d (11) Thus we see that particle longevity is not proportional to d 2, as predicted by the spherical model. Based on the fractal lifetime model (eq. 10), we calculate from the regression result that D 3 - D 2 +1=1.3, or that D 3 -D 2 =0.3. We do not know the value of either D 2 or D 3. However, if D 3 was less than 2, then D 3 =D 2 and the regression analysis would indicate t~d 1. Thus, we can conclude that the particles are fractal and that D 3 has a value 2 D

87 73 80 Longevity (day) Measured Predicted (A) Predicted (B) Particle size (d, cm) Figure 4-5 Measured MFC longevity from the reactor s operational time and predicted MFC longevity from the mass transfer model with particle size as input. 4.5 Discussion: MFC longevity was increased as the particle size increased. This was tested by using different particle sizes of the same material (Chitin 20) to eliminate the effect of particle chemistry on power longevity. Furthermore, particles were added to the MFCs based on equal surface area based on the assumption that surface area would limit power production. Power and voltage output from MFCs usually increases as the substrate concentration increases following the saturation kinetics with different half saturation constant (K s ) (Liu et al., 2005; Liu and Logan, 2004). However, the same power and voltage is expected to be achieved when reactor is operated at substrate concentrations higher than the saturation concentration. In these experiments, the reactors were not expected to be limited by chitin concentration as preliminary

88 74 tests showed similar power generation and longevity using 0.5 or 1 g/l chitin of the same size particles. Particle size did not affect the type of intermediate product (methanol and acetone) accumulated at the end of batch, but it did affect the concentrations of those products. Methanol degradation rates have been found to be slower than for other alcohols such as ethanol under iron reducing condition (Suflita and Mormile, 1993; Lovley and Philips, 1986). Therefore, it is not surprising that methanol accumulated the most in reactors fed with smaller particles that were quickly consumed and had shorter longevity. Coulombic efficiency of the reactor was increased as the particle size increased. Increasing particle size was parallel to decreasing the accumulated methanol concentration and the ratio of biomass to degraded substrate. Accumulation of methanol which is not a favorable material for electricity generation (Kim et al., 2007), was likely to be one of the main factors of coulombic efficiency decrease for small particle. DGGE analysis of anode solution of different particle sizes showed close similarity of the dominant bacteria. MFCs fed with the largest selected particle were not operated as many batch cycles as the ones fed with smaller particle size due to their longer longevity. Therefore, the slight difference in DGGE bands for the largest particles as compared to other sizes was likely due to the less stable bacterial community in those reactors. Among sequenced bands, Enterobacter cloacae, was recently shown to have exoelectrogenic activity and it was able to produce electricity using N-Acetyl-D-glucosamine, a monomer of chitin (Rezaei et al., 2008b). Enterobacter species have been shown to have chitinase activity (Dahiya et al., 2005; Tang et al., 2001). Therefore, it is likely that the absence of this bacterium in the anode solution for the

89 75 largest particle size resulted to have the lower power generation in those reactors compared to smaller sizes. MFC longevity was modeled based on a mass transfer model and assuming complete dissolution of spherical particles. However, the model over predicted the longevity of larger particles. This difference between theory and observation is likely due to the fact that the particles were fractal objects. A non-linear regression analysis of the lifetime of the particles based on power generation produced a relationship between longevity and particle size to the power 1.3. If the particles were spheres, this value would have been 2, and if they were highly amorphous with D 3 <2, then this power would have been 1. Thus, we concluded that D 3 was in the range of 2 to 2.3. Clearly, additional research is needed to develop a better model of chitin particle degradation. It has been shown in this study that larger particles can improve MFC longevity. However, still more research is needed to improve the maximum power generation and coulombic efficiency by eliminating limiting factors such as oxygen penetration into the reactor. Addition of particulate substrates to sediment fuel cells, where there is a better sealing to prevent oxygen penetration to the anode might allow for longevity closer to the predicted one. This statement will be stronger when comparing the longevity of the MFC reactors tested here to the previously reported longevity in SEM where the reactor ran for 30 days using d=0.5 mm (Rezaei et al., 2007). Our results here can further assist scientists and engineers to better understand and model the SEM longevity for practical applications and long term operation. Moreover, particle size can be used as a factor for improving MFC performance for large scale reactors. Also it is needed to test a wider range of particles and find a threshold for particle size in which reactor

90 76 performed the best, and more closely examine the diversity of bacteria acclimating in those reactors. Acknowledgments: Support was provided in-part by the Pennsylvania Experiment Station and the National Science Foundation Grant CBET The authors thank Dr. Rachel Brennan for her valuable suggestions and comments. 4.6 Literature Cited: Ahn, H K, T L Richard, T D Glanville, J D Hamron, and D L Reynolds Laboratory determination of compost physical modeling parameters. ASAE paper no Bradford, M M A rapid and sensitive method for quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72: Cheng, S, H Liu, and B E Logan. 2006a. Increased performance of single-chamber microbial fuel cells using an improved cathode structure. Electrochem Commun 8: Cheng, S, H Liu, and B E Logan. 2006b. Increased power generation in a continuous flow MFC with advective flow through the porous anode and reduced electrode spacing. Environ Sci Technol 40 (7): Cheng, S, and B E Logan Ammonia treatment of carbon cloth anodes to enhace power generation of microbial fuel cells. Electrochem. Comm. 9: Dahiya, N, R Tewari, R P Tiwari, and G S Hoondal Chitinase from Enterobacter sp. NRG4: Its purification, characterization and reaction pattern. Journal Bacteriology 8 (2): Guedon, E, S Payot, M Desvaux, and H Petitdemange Carbon and electron flow in Clostridium cellulolyticum grown in chemostat culture on synthetic medium. Journal Bacteriology 181: Kim, J R, S H Jung, B E Logan, and J M Regan Electricity generation and microbial community analysis of ethanol powered microbial fuel cells. Bioresour Technol 98 (13): Liu, H, S Cheng, and B E Logan Production of electricity from acetate or butyrate in a single chamber microbial fuel cell. Environ. Sci. Technol. 39 (2):

91 77 Liu, H, and B E Logan Electricity generation using an air-cathode single chamber microbial fuel cell in the presence and absence of a proton exchange membrane. Environ. Sci. Technol. 38 (14): Logan, B E Environmental transport process. New Jersey, NY: John Wiley & Sons, Inc. Logan, B E Microbial Fuel Cells. Edited by J. W. Sons. New Jersey, NY: John Wiley & Sons, Inc. Logan, B E, S Cheng, V Watson, and G Estadt Graphite fiber brush anodes for increased power production in Air-Cathode Microbial Fuel Cells. Environmental Science & Technology 41: Logan, B L, B Hamelers, R Rozendal, U Schroder, J Keller, W Verstraete, and K Rabaey Microbial Fuel Cells: Methodology and technology. Environ. Sci. Technol 40 (17): Lovley, D R, and E J P Philips Organic Matter Mineralization with Reduction of Ferric Iron in Anaerobic Sediments. Appl. Environ. Microbiol 51: Lovley, D R, and E J P Philips Novel mode of microbial energy metabolism: Organism carbon oxidation coupled to dissililatory reduction of iron and manganese. Appl. Environ. Microbiol 54 (6): Rabaey, K, and J Keller Microbial fuel cell cathodes: from bottleneck to prime opportunity? Water Sci Technol 57 (5): Ren, N, D Xing, B E Rittmann, L Zhao, T Xie, and X Zhao. 2007a. Microbial community structure of ethanol type fermentation in bio-hydrogen production. Environ Microbiol 9 (5): Ren, Z, T E Ward, and J M Regan. 2007b. Electricity production from cellulose in a microbial fuel cell using a defined binary culture. Environ. Sci. Technol. 14 (13): Rezaei, F, T L Richard, R A Brennan, and B E Logan Substrate-Enhanced microbial fuel cells for improve remote power generation from sediment-based system. Environ Sci Technol 41: Rezaei, F, T L Richard, and B E Logan. 2008a. Enzymatic hydrolysis of cellulose coupled with electricity generation in a microbial fuel cell. Biotech Bioeng. Rezaei, F, D Xing, R Wagner, J M Regan, T L Richard, and B E Logan. 2008b. Simultaneous cellulose degradation and electricity production by Enterobacter cloacae in an MFC. In Appl Environ Microbiol. Richard, T L, A Veeken, V de-wilde, and H V M Hamelers Air-filled porosity and permeability relationships during solid-state fermentation. Biotechnology Progress 20:

92 78 Rismani-Yazdi, H, S M Carver, A D Christy, and oo H Tuovinen Cathodic limitations in microbial fuel cells: An overview. Journal of Power Sources 180: Rismani-Yazdi, H, A D Christy, B A Dehority, M Morrison, Z Yu, and O H Tuovinen Electricity generation from cellulose by rumen microorganisms in Microbial Fuel Cells. Biotech Bioeng 97 (6): Suflita, J M, and M R Mormile Anaeronic biodegradation of known and potential gasoline oxygenates in the terrestrail subsurface. Environ. Sci. Technol 27 (5): Tang, Y, J Zhao, S Ding, S Liu, and Z Yang Purification and properties of chitinase from Enterobacter aerogenes. Acta Microbiologica Sinica 41 (1): Thibodeaux, L J ENVIRONMENTAL CHEMODYNAMICS - Movement of Chemicals In Air, Water and Soil,. Vol. 2nd Edition. N.Y: J. Wiley. Watanabe, K, Y Kodama, and S Harayama Design and evaluation of PCR primers to amplify bacterial 16S ribosomal DNA fragments used for community fingerprinting. J Microbiol. Methods 44: Zuo, Y, D Xing, J M Regan, and B E Logan Isolation of the exoelectrogenic bacterium Ochrobactrum anthropi YZ-1 by using a U-tube microbial fuel cell. Applied and Environmental Microbiology 74 (10):

93 79 Chapter 5 Enzymatic hydrolysis of cellulose coupled with electricity generation in a microbial fuel cell Abstract Electricity can be directly generated by bacteria in microbial fuel cells (MFC) from a variety of biodegradable substrates, including cellulose. Particulate materials have not been extensively examined for power generation in MFCs, but in general power densities are lower than those produced with soluble substrates under similar conditions likely as a result of slow hydrolysis rates of the particles. Cellulases are used to achieve rapid conversion of cellulose to sugar for ethanol production, but these enzymes have not been previously tested for their effectiveness in MFCs. It was not known if cellulases would remain active in an MFC in the presence of exoelectrogenic bacteria or if enzymes might hinder power production by adversely affecting the bacteria. Electricity generation from cellulose was therefore examined in two-chamber MFCs in the presence and absence of cellulases. The maximum power density with enzymes and cellulose was 100 ± 7 mw/m 2 (0.6 ± 0.04 W/m 3 ), compared to only 12 ± 0.6 mw/m 2 (0.06 ± W/m 3 ) in the absence of the enzymes. This power density was comparable to that achieved in the same system using glucose (102 ± 7 mw/m 2, 0.56 ± W/m 3 ) suggesting that the enzyme successfully hydrolyzed cellulose and did not otherwise inhibit electricity production by the bacteria. The addition of the enzyme doubled the Coulombic efficiency (CE) to CE=51% and increased COD removal to 73%, likely as a result of rapid hydrolysis of cellulose in the reactor and biodegradation of the enzyme. These results demonstrate that cellulases do not adversely affect exoelectrogenic bacteria that produce power in an MFC, and that the use of these enzymes can increase power densities and reactor performance.

94 2 Material presented in this chapter was published in the following paper: Rezaei, F., T.L. Richard, and B.E. Logan Enzymatic hydrolysis of cellulose coupled with electricity generation in a microbial fuel cell. Biotechnol. Bioengin. ASPA 80

95 Introduction Cellulose is the most abundant biopolymer on the earth, and is the focus of considerable interest as a renewable energy resource. Cellulosic materials are desirable feedstocks for alternative fuels and energy carriers such as ethanol, biodiesel, or hydrogen since they are renewable and abundant (Mielenz, 2001; Ni et al., 2006; Powlson et al., 2005). Cellulose can also be used as a fuel in a microbial fuel cell (MFC) for direct electricity production (Rismani-Yazdi et al., 2007, Ren et al., 2007, Rezaei et al., 2007) using exoelectrogenic bacteria. The complex structure and very low hydrolysis rate of cellulose presents challenges for using this material in an MFC (Ren et al., 2007, Rezaei et al., 2007). Both cellulolytic and exoelectrogenic microorganisms are needed in an MFC as no single strain has yet been shown to be able to generate electricity directly from cellulose. Maximum power densities produced in MFCs using cellulose have been lower than those achieved using soluble substrates. For example, a co-culture of cellulolytic fermentor Clostridium cellulolyticum and the exoelectrogen Geobacter sulfurreducens generated electricity, and the power generated using soluble cellulose (carboxymethyl cellulose; CMC), was comparable (143 mw/m 2, R= 1000 Ω) to that obtained with glucose. However, the maximum power density generated using insoluble cellulose (MN301) was 58% less (59 mw/m 2 ) than that with CMC. The removal of insoluble cellulose based on chemical oxygen demand (COD) was <30% and cellulose degradation was <50% (Ren et al., 2007). Power generation and cellulose degradation were lower when a mixed culture (sludge) was used instead of a co-culture. A consortium of rumen bacteria was found to produce electricity using cellulose (Rismani-Yazdi et al., 2007) at a power density (55 mw/m 2 ) comparable to that obtained using

96 82 mixed soluble carbohydrates in the same reactor set up. Although, the cellulose degradation rate was not measured in this study, it was claimed that power output was not limited by cellulose hydrolysis rate. It should be noted that in both of these previous studies by Rismani-Yazdi et al. (2007) and Ren et al. (2007) that ferricyanide was used as electron acceptor at the cathode. The absence of any oxygen being used in the system can be important, as oxygen diffusion into the anode chamber has been found to affect the enrichment and operation of MFCs for some substrates (Kim et al., 2007). One way to improve power generation is to use enzymes to increase the hydrolysis rate of cellulose. Enzymatic hydrolysis is widely used in bioethanol production to achieve higher rate of cellulose to sugar conversion. Cellulase usually refers to a mixture of different enzymes involved in cellulose hydrolysis to glucose, which can include endoglucanase, cellobiohydrolase, and β- glucosidase (Reese, 1975). One limitation with using cellulase is that there is a reduction in rates due to end product (cellobiose and glucose) inhibition. Simultaneous saccharification and fermentation (SSF) overcomes this problem by hydrolyzing cellulose and fermenting the hydrolysis product at the same time (Sun and Cheng, 2002). In this research it was hypothesized that enzymatic hydrolysis of cellulose could improve electricity generation rate in MFC s using cellulosic substrates. It has not been previously shown, however, whether the presence of the exoelectrogenic bacteria in the MFC would inhibit enzyme activity. Also, it is not known if the enzyme might adversely affect the exoelectrogenic bacteria. The main objective of this research was therefore to show that adding cellulases would increase the hydrolysis rate of cellulose without interfering with the activity of electrochemically active bacteria and consequently, that adding these enzymes would increase power output. To show the effect of enzyme on power output, power generation with both the cellulose and the enzyme

97 83 (cellulose + enzyme) was compared to that with only cellulose, and with reactors fed either only enzyme or glucose. Furthermore, to monitor the effect of microorganisms on enzymatic hydrolysis of cellulose, the hydrolysis rate in an active MFC was compared to the hydrolysis rate in an abiotic control. 5.3 Materials and Methods Medium: The medium was a phosphate buffered solution (PBS; ph= 7) containing (g/l): NH 4 Cl, 0.31; KCl, 0.13; NaH 2 PO 4 H 2 O, 2.45; Na 2 HPO 4, 4.576; and mineral (12.5 ml/l) and vitamin (12.5 ml/l) solutions (Lovley and Philips, 1988). Cellulose, cellulase, cellulose + cellulase (50% each), and glucose were used as substrates, with each total substrate added at a concentration at 1.3 ± 0.03 g-cod/l (Table 5-1). Cellulose was microcrystalline insoluble cellulose, 15% amorphous and 85% crystalline (Fan et al. 1980), of type 50-50, cotton linters, having a 50-μm particle size (Sigmacell, Sigma-Aldrich Co, St. Louis. MO). The term cellulase as used here refers to a combination of Novozyme 188 ( -glucosidase) and Celloclast 1.5L (synonymous with a Novozyme cellulase from Trichoderma reesei ATCC 26921) (Sigma- Aldrich Co, St. Louis, MO) (Ramos et al. 1993). Cellulase was added to the cellulose with the loading rate equal to 30 cellobiose units (CBU) of Novozyme 188 and 15 filter paper units (FPU) of Celloclast 1.5L per gram of cellulose (Ramos et al. 1993).

98 84 Table 5-1 Four different treatments with their total initial substrate concentrations. Parameter Cellulose Enzyme Cellulose + Enzyme Glucose Particulate COD concentration (g/l) 1.3 ± ± Soluble COD concentration (g/l) ± ± ± 0.05 Cellulose (g/l) Enzyme (μl) C a and 138 C and 10.6 N N b Glucose (g/l) a C for Cellucast, and b N for Novozyme MFC Construction and Operation: A two-chamber MFC was used that consisted of two media bottles (200 ml capacity, Penn State glass workshop, University Park, PA, USA), joined together via a glass tube on either side of a cation exchange membrane (Nafion 117, Dupont Co., Delaware) as previously described (Oh and Logan, 2006). Oxygen was used as the electron acceptor in the cathode by air sparging. The cathode was 1 cm 10 cm (A cat = m 2 ) carbon paper coated on one side with a 10% Pt/C catalyst applied at a rate of 0.35 mg Pt/cm 2 (10% of Pt/C catalyst, 30% wet-proofing; E-TEK, Somerset, NJ, USA). A graphite brush electrode (25 mm diameter 25 mm length, A an =0.22 m 2 ) was used as the anode (Logan et al., 2006). Each reactor was inoculated once with sludge (20%) from secondary clarifier obtained from the Pennsylvania State University Wastewater Treatment Plant and 50 mm PBS medium. Reactors were then run in fed-batch mode for several cycles using only media to achieve at least

99 85 two consecutive batches with the same maximum power density. The medium was then changed to a 200 mm PBS medium for the rest of experiments. Medium was replaced when the voltage dropped to <50 mv Abiotic Rate of Cellulose Hydrolysis: A control experiment was conducted in the absence of bacteria (abiotic control) to determine the effect of bacteria on enzymatic hydrolysis rates. The amount of glucose produced and the amount of cellulose consumed were measured in a flask supplemented with cellulose and cellulase (same loading rate as above) under aseptic conditions. Samples (3 ml) taken over 168 hours were analyzed for cellulose, glucose, and cellubiose concentrations. At the end of experiment, a sample was placed on an YPD (Yeast, Peptone, Dextrose) agar plate to monitor bacterial contamination. The ph of the last sample was also monitored to ensure constant ph conditions and the lack of ph changes due to microbial activity Analytics and Calculations: COD was measured using a colorimetric assay (HACH, DR2010 spectrophotometer) based on a standard method (APHA et al. 1992). Soluble COD (SCOD) was measured after centrifugation (Eppendorf centrifuge, 5403) and filtration of the supernatant using a 0.2 μm porediameter syringe filter. Particulate COD was measured after removing bacterial biomass by centrifuging and washing the samples with 10 mm PBS (10 mm Na 2 HPO 4,130 mm NaCl, ph 7.2) followed by 0.2 N sodium hydroxide (NaOH) as previously described (Ren et al. 2007). Samples were taken from the reactors at 0, 3, 24, and 168 h as well as at the end of batch. Cellulose concentrations in the samples were measured by removing bacterial biomass as described above. Cellulose pellets were solubilized in 67% (v/v) sulfuric acid (Updegraff 1969),

100 86 and the concentration measured using by the phenol-sulfuric acid calorimetric method (Dubois et al. 1956). Glucose and cellobiose concentration were measured using ion chromatography, with the PA20 column and 10 mm NaOH as the mobile phase (Dionex model ICS-3000, Sunnyvale, CA). Enzyme activities of Celloclast 1.5L and the mixed solution (mixture of the two enzymes, Celloclast 1.5L and Novozyme 188) were measured based on the filter paper unit (FPU) using standard methods (Ghose 1987; Mandels et al. 1981). Initial activity of Novozyme 188 was measured based on the cellobiose assay (CBU) (Sternberg et al. 1977). Samples taken from MFC were filtered (0.2 μm pore-diameter) to remove microorganisms prior to measurement of enzyme activity and concentration. The total enzyme concentration was estimated based on measurement of soluble protein using the Bradford protein assay (Bradford 1976). Activity of the enzyme was measured based on FPU for each sample taken from cellulose + enzyme reactor and enzyme alone reactor. Cell voltages (V) were measured across the external circuit containing a resistor (1000 Ω, except as noted below) using a data acquisition system (2700, Keithly, USA) connected to a computer. Voltages were recorded every 20 minutes. Current (i) (A) was calculated as i=v/r, where R is the external circuit resistance (Ω). Power (P) (W) was calculated as P= iv, and normalized by the cathode area (A cat ) (m 2 ). The maximum power was calculated from the voltage measured over n points at the beginning of the cycle when the highest voltage was stable for the indicated period of time. Power density (mw/m 2 ) was calculated by dividing the power (1000 iv for the given units) by the one side of the cathode surface area (containing the catalyst) as the cathode is the main factor limiting power production in an MFC (Rismani-Yazdi et al., 2008; Rabaey and Keller, 2008). The volumetric power density (W/m 3 ) was calculated based the

101 87 volume of the anode chamber ( m 3 ). Polarization curves were obtained by varying the circuit load by changing the external resistor from 4 Ω to 200 kω after leaving the reactor under open circuit potential (OCP) for an hour, with each measurement taken under maximum power output conditions (with fresh substrate) after the voltage had stabilized (~20 minutes). The internal resistance calculated as the slope of the voltage-current graph obtained by varying the external resistance. Coulombic efficiency (number of electron recovered as current in the reactor to the total electron stored in biomass) was calculated based on COD removal as previously described (Cheng et al., 2006). 5.4 Results Power Generation with Different Substrates: When cellulose was added to MFCs with cellulase, the average maximum power density was 100 ± 7 mw/m 2 (0.6 ± 0.04 W/m 3 ) (n=1377, based on voltages measured over 489 hours). This power density was comparable to that achieved using the same system with glucose (102 ± 7 mw/m 2, 0.56 ± W/m 3; n=1447, 482 hours). Power generation by the reactor fed cellulose + enzyme was not limited by the concentration of cellulose. Increasing the initial cellulose concentration from 0.5 g/l to 1 g/l did not increase the maximum power density (102 ± 8 mw/m 2, 0.56 ± 0.04 W/m 3 ; data not shown). Enzymes were used as a substrate for power generation, as shown power production of 103 ± 7 mw/m 2, 0.57 ± W/m 3, (n=536, 178 hours) which is similar to that obtained in the other two reactors (Figure 5-1). The longevity of the reactor with only enzyme, however, was 10 days less than that of the cellulose + enzyme reactor. These results demonstrate that cellulases not only increased power generation to levels similar to that with glucose, but that they were subsequently used as an additional substrate for power generation.

102 Power Density (mw/m 2 ) Cellulose + Enzyme Cellulose Glucose Enzyme Time (day) Figure 5-1 Power density (normalized to cathode surface area, constant 1000Ω load) as a function of time from 2-chamber reactor fed with cellulose + enzyme, glucose, cellulose alone, and enzyme alone. The voltage curves for all treatments changed with the first two cycles, and were then constant for the third and fourth cycles except for the reactor fed only cellulose (control). Repeated attempts to re-start the reactor using cellulose and no enzyme did not result in stable power generation after multiple batch cycles. Typically, the reactor fed only cellulose would generate powering the first cycle, and then it would fail to produce power when refilled. The reactor could be re-started by adding fresh inoculum, but power in the second cycle was not stable. Power generated during this first cycle was 48% less (52 ± 2 mw/m 2 (0.26 ± 0.01 W/m 3 ), n= 451, 150 hours) than that produced in repeatable cycles using the cellulose + enzyme or glucose MFCs, but part of this power generation could be due to the organic matter in the wastewater inoculum. Others (Ren et al., 2007; Rismani-Yazdi et al., 2007) have found power

103 89 generation using in two-chamber reactors, but oxygen was not used in the second chamber and thus there was no potential for oxygen contamination of the system. It was hypothesized that the lack of power generation on subsequent feeding cycles was due to: (1) oxygen leakage into the anode chamber; (2) the removal after each batch cycle of microorganisms needed to rapidly scavenge oxygen; and (3) removal of microorganisms needed for cellulose hydrolysis that were likely attached to the cellulose. Therefore, we modified our procedure to leave ml of solution from each batch cycle in the reactor for the next batch cycle. With this modification, we achieved stable and reproducible power generation over successive batch cycles, with a maximum power density of 12 ± 0.6 mw/m 2 (0.06 ± W/m 3 ) (n=16) (Figure 5-1). This is substantially less power than that generated using enzymes Maximum Power Density: Polarization curves were obtained to determine if the different systems could achieve higher power densities using lower external resistances. The maximum power densities for the MFCs using cellulose + enzyme (98 ± 0.05 mw/m 2, 0.54 ± W/m 3 ) or glucose (103 ± 3 mw/m 2, 0.57 ± W/m 3 ) were essentially unchanged relative to the constant resistance results tests. Maximum power density obtained using enzyme alone was about the same level (114 ± 2 mw/m 2, 0.68 ± W/m 3 ) as glucose and cellulose + enzyme, where the slight difference in power could be a result of cathode performance (Figure 5-2 A). The internal resistance was 1035 ± 57 Ω (95% confidence interval) for the cellulose + enzyme reactor, with a lower value found for the reactor fed only enzyme (819 ± 29 Ω) (Figure 5-2 B).

104 90 Power Density (mw/m 2 ) Voltage (mv) A B Cellulose + Enzyme Enzyme Glucose Current (ma) Figure 5-2 (A) Power and Voltage (B) as a function of current for three treatments, cellulose + enzyme, Enzyme, and Glucose (error bars are ± S.D based on duplicate measurements) obtained by varying external resistance Enzymatic Hydrolysis Rate of Cellulose in the Presence and Absence of Bacteria: In absence of microorganisms approximately 20% of the cellulose was hydrolyzed during the first 24 h (Figure 5-3 A). After 168 h, 51% of the cellulose was degraded. The cellobiose concentration was initially low (<0.1 g/l) and was not measurable (<0.002 g/l) after 72 h (data not shown). The final and initial phs of the reactor were the same (6.8), suggesting no substantial activity by fermentative microorganisms. Moreover, there were no colonies on the YPD agar plates, confirming that the conditions remained free of microorganisms throughout the experiments.

105 91 In the presence of bacteria, using samples obtained directly from the MFC, 50% of the initial cellulose was degraded during the first 24 h (Figure 5-3 B). This is 1.5 times higher than that achieved in the abiotic environment. After 168 h, 60% of the initial cellulose was degraded showing that the hydrolysis rates decreased after 24 h (Figure 5-3 B). Comparing the cellulose degradation inside the MFC with the abiotic reactors showed that cellulose degraded faster inside the MFC. Although the concentration of end product that is inhibitory to the enzyme varies with different substrates and enzymes (Gruno et al., 2004), there is a possibility that the higher degradation rate inside the MFC was due to removal of chemicals by microorganisms that may have contributed to end-product inhibition of enzyme activity. The concentration of cellobiose or glucose was not detectable (<0.002 g/l) in the samples taken from MFC reactor.

106 92 Concentration (g/l) Cellulose concentration over its initial concentration Cellulose Glucose (B) (A) Time (hr) Abiotic MFC Figure 5-3 (A) Enzymatic hydrolysis of cellulose without microbes. (B) Cellulose concentration remaining inside the reactor during enzymatic hydrolysis without microbes (Abiotic) versus simultaneous saccharification and oxidation in an MFC. Data are cellulose concentrations inside the reactor relative to the initial cellulose concentration Cellulase Activity, Cellulase Consumption, and Cellulose Degradation: The activity of the remaining enzyme in the cellulose + enzyme MFC declined by 25% during the first 24 hours of the experiment, whereas the reactor containing only enzyme lost more than 60% of its activity during the same period (Figure 5-4). At the end of a batch cycle, the activity of the remaining enzyme inside the cellulose + enzyme reactor was 66% of the initial activity, whereas in the reactor containing only enzyme it was 26% of the initial activity (Table 5-2).

107 93 Enzyme activity/initial activity Cellulose + Enzyme Enzyme Time (hr) Figure 5-4 Enzyme activity in MFC reactors with and without cellulose. Table 5-2 Summary table of percent substrate consumption, COD removal, enzyme activity lost, and coulombic efficiency. Parameter (%) Cellulose Enzyme Cellulose + Enzyme Glucose Soluble COD removal N/A Particulate COD removal 14 N/A a 73 N/A Cellulose degraded 15 N/A 74 N/A Coulombic efficiency b Enzyme consumed N/A N/A Enzyme activity lost N/A N/A a N/A : Not Applicable l. b. Based on COD removal (soluble COD removal for enzyme alone and glucose, particulate COD removal for cellulose and total (particulate + soluble) COD removal for cellulose + enzyme).

108 94 The enzyme concentration based on soluble protein content was compared in the MFC containing cellulose + enzyme and MFC containing only enzyme. The percent concentration of remaining enzyme in solution was lower in reactor containing only enzyme (15%) as compared to cellulose + enzyme reactor (36%) (Table 5-2). A lower enzyme concentration shows that more enzyme was consumed inside the reactor containing only enzyme than in the cellulose + enzyme reactor. In the cellulose + enzyme MFC, 74% of the initial cellulose was degraded at the end of cycle. In the reactor fed only cellulose 15% of the initial cellulose was degraded at the end of cycle, suggesting that adding enzyme enhanced cellulose degradation and that it provided more soluble substrate to be oxidized by the exoelectrogenic microorganisms Coulombic Efficiency: Coulombic efficiency measured for cellulose + enzyme was slightly higher (51%) than the CE from the reactor containing only enzyme (44%) (Table 5-2). Both values were comparable to that obtained using glucose (48%). Coulombic efficiency from the cellulose + enzyme MFC was higher than that using only cellulose (23%) in a two-chamber MFC COD Removal: Soluble COD removal was 69% from the cellulose + enzyme reactor and it was 86% in the reactor fed only enzyme (Table 5-2). This amount of COD removal indicated that the cellulose + enzyme reactor did not generate power from only cellulose or enzyme. Cellulosic COD removal was 73% in the cellulose + enzyme reactor (Table 5-2). The COD removal for an MFC containing only cellulose was 14%, which is 421% less than that achieved here when using enzymes with the cellulose.

109 Discussion Previous reports have suggested that power generation in MFCs from insoluble cellulose can be limited by a low hydrolysis rate (Ren et al., 2007; Rezaei et al., 2007). Our results demonstrated that using enzymes to hydrolyze cellulose improved the hydrolysis rate, overall cellulose and COD removal, and increased the power density and coulombic efficiency of an MFC. Power production using insoluble crystalline cellulose was 100 ± 7 mw/m 2, a level that was comparable with that generated using glucose (102 ± 7 mw/m 2 ) in the same system. This power density with cellulose + enzyme was almost an order of magnitude greater than that achieved in this experiment using cellulose alone. It was also two times greater than that previously found using cellulose by others (Ren et al., 2007; Rismani-Yazdi et al., 2007) even though ferricyanide was used in these previous tests. It is well known that ferricyanide increases power production in MFCs compared systems using dissolved oxygen (Oh and Logan, 2006; Logan, 2008). The presence of microorganisms did not inhibit the hydrolysis rate of cellulose. In fact, the hydrolysis rate in the MFC producing electricity was slightly greater than that in an abiotic reactor likely due to less end-product inhibition through reduced levels of cellobiose and glucose. These observations are consistent with the findings of others that simultaneous saccharification and fermentation increases the hydrolysis rate of cellulose by reducing end-product inhibition (Sun and Cheng, 2002). Although, recycling the enzyme can effectively increase the yield of cellulose conversion and reduce the enzyme cost (Mess-Hartree et al., 1987), in our reactors there is little possibility that the enzyme added to the reactor could be reused as power was generated in the reactor from the enzymes. This degradation of the enzyme is not surprising as it

110 96 has been shown that electricity can be generated in an MFC using proteins as substrates (Heilmann and Logan, 2006). There are additional possibilities for improving cellulolytic rates using enzymes, such as varying the temperature, optimizing the ph, using multiple reactors and continuous flow, or adding surfactants. In this research, all experiments were conducted at room temperature but the optimum temperature is 30 C for fermentation and C for hydrolysis (Philippidis and Smith, 1995). Running the reactor at a higher temperature should therefore improve the kinetics. The optimum ph for enzymes used here are in the range of (Eriksson et al., 2002; Saha and Cotta, 2007), but low ph can adversely affect bacteria. Rates could be improved by determining a ph that optimizes both enzyme activity and microbial activity. Alternatively, cellulose hydrolysis could be conducted in a tank separate from the one used for power generation, perhaps allowing the recovery of enzyme from that first system. Although the rate of hydrolysis might be slower due to end-product inhibition, this disadvantage might be outweighed by the economic benefits of enzyme recovery. Immobilizing the enzyme and running the system in a continuous mode, rather than fed-batch mode, could also improve process performance and perhaps reduce the need for enzymes and thus decrease costs. Improving the activity of the enzyme would also make the process more economical. One of the problems with using cellulase is that it can become deactivated by adsorption onto cellulose (Converse et al., 1988). One way to prevent the irreversible deactivation of enzyme is the addition of a surfactant that can modify the surface of the cellulose by increasing desorption of cellulase (Sun and Cheng, 2002). These results demonstrate the technical feasibility of enzymatic hydrolysis of cellulose for direct electricity generation. It has also recently been shown that cellulose can be used to generate hydrogen gas at yields higher than that possible via fermentation in a microbial

111 97 electrolysis cell (MEC) (Cheng and Logan, 2007). Thus, it is likely that enzymatic treatment of cellulose could also be used as a method for hydrogen production in a similar system. Whether the use of cellulases will be economical in comparison to direct cellulose utilization will depend on costs of the enzyme and further improvements in current densities. The cost of enzymes are being reduced and power densities in MFCs are being increased, which combined may allow for new methods of renewable electricity generation in MFCs or hydrogen production in MECs. 5.6 Acknowledgements Support was provided in-part by the Pennsylvania Experiment Station, and NREL contract RFH to BEL. The authors are grateful to Megan Marshall and Zhiyong Ren for their helpful advice. 5.7 Literature Cited: APHA, AWA, WPCF Standard methods for the examination of water and wastewater.18th ed. Washington DC: American Public Health Association. Bradford MM A rapid and sensitive method for quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72: Cheng S, Liu H, Logan BE Increased power generation in a continuous flow MFC with advective flow through the porous anode and reduced electrode spacing. Environ. Sci. Technol. 40(7): Cheng S, Logan BE Sustainable and efficient biohydrogen production via electrohydrogenesis. PNAS. 104 (47): Converse AO, Matsuno R, Tanaka M, Taniguchi M A model for enzyme adsorption and hydrolysis of microcrystalline cellulose with slow deactivation of the adsorbed enzyme. Biotech. and Bioengin. 32: Dubois M, Gilles KA, Hamilton JK, Rebers PA, Smith F Colorimetric Method for determination of Sugars and Related Substances. Anal. Chem. 28(3):

112 98 Eriksson T, Börjesson J, Tjerneld F Mechanism of surfactant effect in enzymatic hydrolysis of lignocellulose. Enzyme Microb. Technol. 31(3): Fan LT, Lee Y-H, Beardmore DH Mechanism and enzymatic hydrolysis of cellulose: effects of major structural features of cellulose on enzymatic hydrolysis. Biotech. and Bioengin. 4: Ghose TK Measurement of cellulase activities. Pure Appl Chem. 59: Gruno M, Valjamae P, Pettersson G, Johnsson G Inhibition of the Trichoderma reesei cellulases by cellobiose is strongly dependent on the nature of the substrate. Biotech and Bioengin. 86(5): Heilmann J, Logan BE Production of electricity from proteins using a single chamber microbial fuel cell. Water Environ. Res. 78(5): Kim JR, Jung SH, Regan JM, Logan BE Electricity generation and microbial community analysis of ethanol powered microbial fuel cells. Bioresource Technol. 98(13): Logan BE Microbial Fuel Cells. New York: John Wiley & Sons. 200 p. Logan BE, Cheng S, Watson V, Estadt G Graphite fiber brush anodes for increased power production in Air-Cathode Microbial Fuel Cells. Environ. Sci. Technol. 41(9): Lovley DR, Philips EJP Novel mode of microbial energy metabolism: Organism carbon oxidation coupled to dissililatory reduction of iron and manganese. Appl. Environ. Microbiol 54(6): Mandels M, Andreotti R, Roche C Enzymatic hydrolysis of cellulose: evaluation of culture filtrates under use conditions. Biotechnol. Bioeng. Symp 23: Mes-Hartree M, Hogan CM, Saddler JN Recycle of enzyme and substrate following enzymatic hydrolysis of steam pretreated aspenweed. Biotech. and Bioengin. 30: Mielenz JR Ethanol production from biomass: technology and commercialization status. Curr. Opi. Microbiol 4: Ni M, Leung DYC, Leung MKH, Sumathy K An overview of hydrogen production from biomass. Fuel Process. Technol. 87:

113 99 Oh S-E, Logan BE Proton exchange membrane and electrode surface areas as factors that affect power generation in microbial fuel cells. Appl Microbiol Biotechnol 70: Powlson DS, Riche AB, Shield I Biofuels and other approaches for decreasing fossil fuel emissions from agriculture. Annals of Appl. Biol. 146: Philippidis GP, Smith TK Limiting factors in the simultaneous saccharification and fermentation process for conversion of cellulosic biomass to fuel ethanol. Appl Biochem Biotech 51/52: Rabaey K, Keller J Microbial fuel cell cathodes: from bottleneck to prime opportunity? Water Sci. Tech. 57(5): Ramos LP, Breuil C, Saddler JN The use of enzyme recycling and the influence of sugar accumulation of cellulose hydrolysis by Trichoderma cellulases. Enzyme Microb. Technol. 15: Reese ET Biological Transformation of Wood, In: Liese W, editor. p Springer, Berlin. pp: Ren Z, Ward TE, Regan JM Electricity production from cellulose in a microbial fuel cell using a defined binary culture. Environ. Sci. Technol. 41 (13): Rezaei F, Richard TL, Brennan RA, Logan BE Substrate-Enhanced microbial fuel cells for improve remote power generation from sediment-based system. Environ. Sci. Technol. 41(11): Rismani-Yazdi H, Christy AD, Dehority BA, Morrison M, Yu Z, Tuovinen OH Electricity generation from cellulose by rumen microorganisms in Microbial Fuel Cells. Biotech. and Bioengin. 97 (6): Rismani-Yazdi H, Carver SM, Christy AD, Tuovinen OH Cathodic limitations in microbial fuel cells: An overview. Journal of Power Sources 180: Saha BC, Cotta MA Enzymatic hydrolysis and fermentation of lime pretreated wheat straw to ethanol. J. Chem. Technol. Biotechnol. 82 (10) Sternberg D, Kumar PV, Reese ET β-glucosidase: microbial production and effect on enzymatic hydrolysis of cellulose. Can J Microbiol. 23: Sun Y, Cheng J Hydrolysis of lignocellulosic materials for ethanol production: a review. Bioresour. Technol. 83:1-11.

114 Updegraff DM Semimicro Determination of Cellulose in Biological Materials. Anal. Biochem. 32:

115 101 Chapter 6 Simultaneous cellulose degradation and electricity production by Enterobacter cloacae in an MFC 6.1 Abstract Electricity can be directly generated by bacteria in microbial fuel cells (MFCs) from many different biodegradable substrates. When cellulose is used as the substrate, electricity generation requires a microbial community with both cellulolytic and exoelectrogenic activity. Cellulose degradation with electricity production by a pure culture has not been previously demonstrated without addition of an exogenous mediator. Using a specially designed U-tube MFC, we enriched a consortium of exoelectrogenic bacteria capable of using cellulose as the sole electron donor. After 19 dilution-to-extinction serial transfers of the consortium, 16S rrna gene-based community analysis using denaturing gradient gel electrophoresis and band sequencing revealed that the dominant bacterium was Enterobacter cloacae. An isolate designated E. cloacae FR from this enrichment was found to be 100% identical to the type strain Enterobacter cloacae based on a partial 16S rrna sequence. In polarization tests using the U-tube MFC and cellulose as a substrate, strain FR produced 4.9 ± 0.01 mw/m 2 compared to 5.4 ± 0.3 mw/m 2 for strain These results demonstrate for the first time that it is possible to generate electricity from cellulose using a single bacterial strain without the need for exogenous mediators. 6.2 Introduction Exoelectrogenic microorganisms can release electrons to electron acceptors outside the cell, such as iron oxides or carbon anodes in microbial fuel cells (MFCs). Many different bacteria have

116 102 been shown to produce electricity in an MFC, including Rhodoferax (Chaudhuri and Lovley, 2003), Shewanella (Kim et al., 1999; Kim et al., 2002), Pseudomonas (Rabaey et al., 2004), Aeromonas (Pham et al., 2003), Geobacter (Bond and Lovley, 2003), Geopsychrobacter (Holmes et al., 2004b), Desulfuromonas (Bond et al., 2002), Desulfobulbus (Holmes et al., 2004a), Clostridium (Park et al., 2001), Geothrix (Bond and Lovley, 2005), Ochrobactrum (Zuo et al., 2008), and Rhodopseudomonas (Xing et al., 2008). These bacteria have been grown on simple soluble substrates, such as glucose or acetate that can be directly taken into the cell and used for energy production. Cellulose is the most abundant biopolymer in the world and there is great interest in using this material as a substrate in an MFC. However, the use of a particulate substrate in an MFC has not been well investigated. Cellulose must first be hydrolyzed to a soluble substrate that can be taken into the cell. In previous MFC tests this has required the use of enzymes to first hydrolyze the cellulose into sugars or the use of co-cultures or mixed cultures (Rezaei et al., 2008; Ren et al., 2007b; Rismani-Yazdi et al., 2007). For example, Ren et al. (Ren et al., 2007b) used a coculture of the cellulose fermentor Clostridium cellulolyticum and the exoelectrogen Geobacter sulfurreducens to generate electricity in an MFC fed with cellulose. Analysis of the anode microbial communities in other cellulose-fed MFC studies found that Clostridium spp. (in biofilm) and Comamonadaceae (in suspension) were predominant when using rumen contents as an inoculum (Rismani-Yazdi et al., 2007), while a rice paddy soil inoculum (Ishii et al., 2008) converged to a Rhizobiales-dominated anode community (more than 30% of the population). To date, it has not been demonstrated that a single microbe could accomplish both cellulose degradation and current generation.

117 103 Conventional methods of isolating exoelectrogenic microorganisms are primarily based on identifying microorganisms that can respire using soluble or insoluble metal oxides in agar plates (Logan, 2007; Logan and Regan, 2006; Lovley, 2006). However, not all dissimilatory metal oxide reducing bacteria (DMRB) are capable of producing electricity in an MFC, and not all bacteria that produce current in an MFC can grow using metal oxides (Bretschger et al., 2007; Richter et al., 2007). Therefore, these methods may miss important electrochemically active strains of microorganisms. A new method to isolate exoelectrogenic microorganisms was recently developed (Zuo et al., 2008) based on dilution-to-extinction and a specially designed U- tube MFC that enriches exoelectrogenic bacteria on the anode through their growth. Using this method, a bacterium was isolated that could produce electricity in an MFC but not respire using iron (Zuo et al., 2008). The main objective of this study was to isolate a bacterium capable of producing current from particulate cellulose. A cellulose-degrading consortium was diluted and serially transferred into U-tube MFCs using cellulose as the sole electron donor. Community analysis demonstrated the predominance of a single bacterium that was then isolated and compared to a culture collection strain for current generation in the MFC. 6.3 Methods MFC Construction and Operation. U-tube MFCs had a 10 ml anode chamber and a 30 ml cathode chamber constructed from glass anaerobic culture tubes as described previously (Zuo et al., 2008). The two chambers were separated by a cation exchange membrane (CMI 7000, Membranes International Inc, USA; 1.77 cm 2 ) and held together by a C-type clamp. The anode was ammonia-treated carbon cloth (type A,

118 104 E-Tek, USA) with a total surface area of A an = 1.13 cm 2. The cathode was made of five tow strands of 15 cm-long carbon fiber (GRANOC, Nippon Graphite Fiber Corporation, Japan) that were joined together at the top end using titanium wire. The anode solution (9 ml) consisted of a 50 mm phosphate buffer (PBS; 2.45 g/l NaH 2 PO 4 H 2 O and g/l Na 2 HPO 4 ), 0.31 g/l NH 4 Cl, 0.13 g/l KCl, and mineral (12.5 ml/l) and vitamin (12.5 ml/l) solutions (Lovley and Philips, 1988). To provide better nutrition media for enrichment of cellulolytic bacteria, autoclaved rumen fluid (30%, v/v) was added to the anode solution for the first 15 cycles. Cellulose was the primary substrate (0.4 %), consisting of 15 % amorphous cellulose and 85% crystalline cellulose (Fan et al., 1980) (type 50-50, cotton linters, 50-μm particle size; Sigmacell, Sigma-Aldrich Co, USA). The catholyte solution (29 ml) was 100 mm potassium ferricyanide, K 3 Fe(CN) 6, in PBS (100 mm). After assembling the reactor, both chambers were sparged with N 2 gas, sealed with a rubber stopper and an aluminum crimp top, and autoclaved prior to use for each fed-batch cycle Enrichment Procedure. Wastewater used for the initial inoculum was obtained from a paper recycling plant (American Eagle Paper Company, Tyrone, Pennsylvania). A dilution-to-extinction method was used to enrich exoelectrogenic and cellulolytic bacteria. U-tubes were initially inoculated with wastewater diluted to 10-1, 10-2, 10-3, and 10-4 with medium. After each cycle, the anode chamber suspension and the anode were transferred from each MFC into a sterile 15 ml tube (FALCON, Becton Dickenson Labware, USA) containing sterilized glass beads, and vortexed. In each subsequent transfer, homogenized suspension from the most dilute reactor that generated electricity was used to inoculate new batches. Samples were diluted at 10-1, 10-2, 10-3, and 10-4 for transfers 2 through 15, and 10-2, 10-4, 10-6, and 10-8 for transfers 16 to 19. An additional

119 105 sterile reactor (no inoculation) was used to monitor for possible contamination of the growth medium during each transfer. The remaining suspension in each tube was preserved at -20 C for further analysis DNA Extraction, PCR, and DGGE. DNA was extracted from the preserved anode suspension of the most diluted reactor showing power from each cycle using the PowerSoil DNA isolation kit (MO BIO Laboratories, US) according to the manufacturer s instructions. DNA integrity was verified using a 1% agarose gel. PCR was then performed using an icycler iq TM thermocycler (Bio-Rad Laboratories, US) to amplify the V6-V8 region of the 16S rrna gene (rdna) using the following primers (Watanabe et al., 2001), which included a GC clamp on the forward primer for subsequent denaturing gradient gel electrophoresis (DGGE) analysis: GC968F (5'- CGCCCGCCGCGCCCCGCGCCCGGCCCGCCGCCCCCGCCCCAACGCGAAGAACCTTA C-3') and 1401R (5'-CGGTGTGTACAAGACCC-3'). The PCR conditions were as described previously (Zuo et al., 2008). PCR products then were separated by DGGE using a DCode universal mutation detection system (Bio-Rad Laboratories, US) as described previously (Zuo et al., 2008; Ren et al., 2007a). Serial transfers were performed until the DGGE gel showed five bands that were consistent for more than two transfers. Each of these bands was excised from the gel using a sterile pipette tip and transferred to a sterile microcentrifuge tube. DNA was eluted from the bands by adding 40 μl deionized water, crushing the gel against the tube side using a pipette tip, and then incubating the tubes at 4 C overnight. DNA integrity was verified using a 1% agarose gel. Two sets of PCR were performed with this eluted DNA, the first to check the purity of each band using the same PCR and DGGE conditions described above (Ren et al., 2007a). After

120 106 confirming there was only one band, a second PCR was performed to re-amplify the bands for subsequent sequencing using the same PCR primers, except the forward primer lacked the GC clamp (968F: 5'-AACGCGAAGAACCTTAC-3'), with the following conditions: 95 C for 5 min; 35 cycles of 95 C for 1 min, 60 C for 30 s, and 72 C for 1.5 min; and finally 72 C for 7 min. PCR products then were purified using a QIAquick PCR purification kit (QIAGEN, USA) according to the manufacturer s instructions and sequenced using an ABI 3730XL DNA sequencer (Applied Biosystems, US) Cloning and Sequence Analysis. In addition to the DGGE analysis, extracted DNA from the last cycle was amplified using the following PCR primers to amplify nearly complete 16S rdna: 27F (5'- AGAGTTTGATCCTGGCTCAG-3') and 1541R (5'-AAGGAGGTGATCCAGCC-3') as described previously (Zuo et al., 2008; Ren et al., 2007a). The PCR condition was 95 C for 5 min; 35 cycles of 95 C for 1 min, 60 C for 30 s, and 72 C for 1.5 min; and finally 72 C for 7 min. PCR products were then cloned using a TOPO TA cloning kit (Invitrogen, US) according to the manufacturer s instructions. The plasmid of clones was extracted and purified using a QIAprep Spin Miniprep Kit (QIAGEN, USA) and sequenced in both directions using an ABI 3730XL DNA sequencer (Applied Biosystems, US). Sequences were analyzed in GenBank using the BLAST program, and a neighbor-joining phylogenetic tree was constructed according to Kimura s two-parameter method using the Molecular Evolutionary Genetics Analysis package (MEGA version 3)(Kimura, 1980) Bacteria Isolation and Characterization. Once the dominant bacterium was determined based on DGGE band sequences and nearly complete cloned 16S rdna sequences, the corresponding type strain Enterobacter

121 107 cloacae T was purchased from the American Type Culture Collection (ATCC) and grown based on their instructions. At the same time, we tried to isolate the bacterium directly from mixed culture of the last cycle by plating it on the nutrition suggested by ATCC and growing it overnight. Six colonies with the same colony morphology as the culture collection strain were selected and grown on nutrition broth overnight. To confirm the purity and similarity of the selected colonies to the determined dominant bacterium, DNA from each overnight suspension was extracted and nearly complete 16S rdna was amplified and sequenced as described earlier. Carbon utilization characteristics of Enterobacter cloacae T and the isolated bacterium were determined using BIOLOG GN2 MicroPlates following the manufacturer s instructions. The ability of the isolated strain to reduce iron was determined using insoluble hydrous ferric oxide (HFO; 100 mm) (Fredrickson et al., 2003), in 1 g/l cellulose and 1 g/l glucose in anaerobic tubes over 7 days at 30 C (triplicate tests). Uninoculated tubes (duplicate) were run as controls for contamination. Reduction of Fe (III) was measured using a ferrozine colorimetric method as described previously (Lovley and Philips, 1986) Electricity Generation and Analyses. Current and power generation in the MFCs were determined by measuring the voltage (V) every 20 minutes across a fixed external resistance (R=1000 Ω, except as noted) using a data acquisition system (Keithley, 2700, USA). Current was calculated as I=V/R and power was calculated as P=IV. Power density and current density were normalized to the projected area of the anode. Polarization curves were obtained by using a single resistor for two complete batch cycles (250 Ω to open circuit). The cellulose concentration remaining at the end of each batch cycle was measured using a colorimetric method (Rezaei et al., 2008; Ren et al., 2007b). Volatile fatty acids (VFAs) were

122 108 measured at the end of each cycle using gas chromatography (Liu and Logan, 2004). Coulombic efficiency (ratio of the recovered electrons as current to the total available electrons from the substrate) was calculated at the end of a cycle based on cellulose removal as described previously (Rezaei et al., 2008). Enterobacter cloacae T was also examined for electricity generation using cellulose and the following carbon sources (4 g/l): glucose, lactate, N-acetyl-D-glucosamine, glycerol, and sucrose. Two controls were run, an uninoculated reactor (using the same medium and 0.4% cellulose) to ensure there was no contamination, and an inoculated reactor without substrate to monitor the possibility of electricity generation from the working medium. 6.4 Results Exoelectrogenic/Cellulolytic Enrichment. -tube reactors were run for 19 cycles after wastewater inoculation. For cycles 1 to 15, the 10-4 dilution (the most diluted solution) produced electricity each time, and therefore the MFC with this dilution was used to inoculate the next series of reactors (Figure 6-1A). When higher dilutions were used (cycles 16 to 19), the highest dilution (10-8 ) did not show any power generation, therefore the next most dilute solution (10-6 ) was used to inoculate the subsequent batch (Figure 6-1B). For all 19 cycles, no power was generated with reactors lacking an inoculum.

123 109 A 0.08 Voltage (V) Blank B Voltage (V) Time (Day) Time (day) Figure 6-1 Power generation from (A) the first cycle of U-tube using four different dilutions and (B) the last (19th) cycle of U-tube using four different dilutions Phylogenetic Analysis. After 16 cycles, the community composition as indicated by the number and intensity of bands in the DGGE gels was constant over the next three cycles (Figure 6-2). Analysis of the

124 110 sequences from each of the five bands from the last cycle indicated that the top three bands were derived from members of the family Enterobacteriaceae, while bands four and five were 100% identical to Stenotrophomonas sp. and Exiguobacterium sp., respectively (Table 6-1, Figure 6-3). The first band was 100% identical to Klebsiella pneumonia that was recently introduced to be exoelectrogenic (Zhang et al., 2008). Bands 2 and 3 were respectively 99% and 100% similar to Enterobacter cloacae (Figure 6-3). Sequences from the clone library of the last cycle were also analyzed to further identify the dominant bacterium. Phylogenic analysis of the clone library from cycle 19 showed that all the analyzed cloned fragments belonged to Enterobacter species, with Enterobacter cloacae ATCC T (100% identity) as the dominant bacterium (Figure 6-4). An isolate obtained from the mixed culture using suspension from the last cycle had a colony morphology similar to that observed for the culture collection strain E. cloacae T, and the nearly full length 16S rdna sequence was identical to that of E. cloacae T. The isolate was designated as E. cloacae FR, and the sequence was entered into the GenBank database (accession number EU849019). Table 6-1 Closest reported strains to the sequence of the last cycle s bands from GenBank using BLAST program. Band# BLAST Results Identity (%) 1 Klebsiella pneumoniae Uncultured Enterobacter sp. clone wofoc_r Enterobacter cloacae strain E Enterobacter cloacae partial, strain ATCC13047T Stenotrophomonas sp. SWCH Exiguabacterium sp. ZM-2 100

125 M Figure 6-2 DGGE bands of the 19 cycles of the most diluted U-tube that produced electricity. Bands 1 to 5 was extracted from the gel for sequencing.