MASTER 2 - PROFESSIONAL University of Toulouse III - Paul Sabatier Processes Engineering - Specialization of Electrochemical Processes 2011-2012

Evaluation of Unified Numerical and Experimental Methods for Improving Microbial Electrochemical Technologies (MXCs)

E. EKİN DALAK

Academic Advisors: Peter Winterton (UPS) Theo Tzedakis (UPS)

Promotors: Xochitl Dominguez (VITO) Karolien Vanbroekhoven (VITO) ()5°(VITO)

TABLE OF CONTENT TABLE OF CONTENT ........................................................................................................... 2 LIST OF FIGURES .................................................................................................................. 5 LIST OF TABLES .................................................................................................................... 8 GLOSSARY .............................................................................................................................. 9 ACKNOWLEDGEMENTS ................................................................................................... 11 ABSTRACT ............................................................................................................................ 12 OBJECTIVES......................................................................................................................... 13 CHAPTER 1: INTRODUCTION ......................................................................................... 14 CHAPTER 2: DESCRIPTION OF THE INSTITUTE ....................................................... 15 2.1 PROFILE ............................................................................................................................. 15 2.2 ACTIVITIES ........................................................................................................................ 15 2.3 RESEARCH FIELDS ............................................................................................................ 16 2.4 ELECTROCHEMISTRY AT VITO ...................................................................................... 17 CHAPTER 3: MICROBIAL ELECTROCHEMICAL SYSTEMS (MXCS) ..................... 18 3.1 TYPES OF MXCS ............................................................................................................... 19 3.1.1 MICROBIAL FUEL CELL (MFC) .................................................................................. 19 3.1.2 MICROBIAL ELECTROLYSIS CELL (MEC) ................................................................. 20 3.1.3 MICROBIAL ELECTROSYNTHESIS (MES) ................................................................... 21 3.2 ELECTRON TRANSFER MECHANISMS ............................................................................. 21 3.2.1 DIRECT ELECTRON TRANSFER (DET) ........................................................................ 21 3.2.2 MEDIATED ELECTRON TRANSFER (MET) .................................................................. 22 3.3 PERFORMANCE PARAMETERS ......................................................................................... 23 3.3.1 ENERGY GENERATION .................................................................................................. 23 3.3.2 TREATMENT EFFICIENCY ............................................................................................. 25 3.4 MXC DESIGNS .................................................................................................................. 26 3.4.1 REACTOR CONFIGURATIONS ........................................................................................ 26 2

3.4.3 FUEL TYPES ................................................................................................................... 29 3.4.4 MICROBE TYPES ............................................................................................................ 30 3.4.5 OPERATIONAL CONDITIONS ......................................................................................... 30 3.5 ELECTROCHEMICAL CHARACTERIZATION TECHNIQUES ............................................ 31 3.5.1 OPEN CIRCUIT VOLTAGE ............................................................................................. 32 3.5.2 CYCLIC VOLTAMMETRY (CV) ..................................................................................... 32 3.5.3 CHRONOAMPEROMETRY (CA) ..................................................................................... 33 3.5.4 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY (EIS) .......................................... 33 CHAPTER 4: DESIGN AND OPTIMIZATION OF MFCS VIA MODELING ............... 35 4.1 MODELING CURRENT AND POTENTIAL DISTRIBUTIONS IN MFC .................................... 36 4.1.1 OVERPOTENTIALS ........................................................................................................... 36 4.1.2 TYPES OF CURRENT AND POTENTIAL DISTRIBUTIONS .................................................. 38 4.2 NUMERICAL MODELING OF MFC VIA COMSOL MULTIPHYSICS .................................. 46 CHAPTER 5: MODELING WORK ..................................................................................... 47 5.1. COMSOL MODELING PROCEDURE ................................................................................ 47 5.2. INITIAL MODEL GEOMETRY, DEFINITIONS AND RESPONSE ........................................... 49 5.3 PRIMARY CURRENT DISTRIBUTION MODELS ................................................................... 51 5.3.1 INFLUENCE OF CURRENT COLLECTOR DESIGN ............................................................. 51 5.3.2 INFLUENCE OF LUG DIMENSIONING AND DESIGN ......................................................... 52 5.3.3 HALF CELL CONFIGURATIONS ....................................................................................... 54 5.3.4 INFLUENCE OF LUG POSITIONING ON DIFFERENT CURRENT COLLECTOR DESIGNS ... 55 5.3.5 INFLUENCE OF DIFFERENT PARAMETERS IN SELECTED GEOMETRY ........................... 57 5.3.6 INFLUENCE OF GRID SIZE ............................................................................................... 60 5.4 SECONDARY CURRENT DISTRIBUTION MODELS .............................................................. 62 5.5 SUMMARY AND PERSPECTIVES OF THE MODELING WORK .............................................. 63 CHAPTER 6: EXPERIMENTAL WORK ........................................................................... 66 6.1 MATERIALS AND METHODS ............................................................................................... 66 3

6.1.1 MICROBIAL GROWTH ..................................................................................................... 66 6.1.2 ELECTROCHEMICAL CELL COMPONENTS ..................................................................... 67 6.1.3 EXPERIMENTAL SET-UP AND OPERATION ...................................................................... 67 6.1.4 ELECTROCHEMICAL METHODS...................................................................................... 68 6.2 RESULTS AND DISCUSSION ................................................................................................. 69 6.2.1 HALF CELL EXPERIMENTS WITH ACETATE................................................................... 69 6.2.2 HALF CELL EXPERIMENTS WITH FUMARATE ................................................................ 72 6.2.2.1 GLUCOSE ADDITION AND CHANGE IN POLARIZATION POTENTIAL (50 MV) ............... 72 6.2.2.2 FUMARATE-GLUCOSE COMBINATION AS SUBSTRATE FROM THE INITIAL TIME ......... 74 6.2.2.3 INFLUENCE OF THE BACTERIAL GROWTH IN THE HALF CELL .................................... 75 6.2.2.4 GLUCOSE ADDITION AND CHANGE IN POLARIZATION POTENTIAL (250 MV) ............. 77 6.2.2.5 SEPARATION OF THE MEDIUM CIRCULATION IN COUNTER ELECTRODE AND WORKING ELECTRODE COMPARTMENTS.................................................................................................. 80

6.2.3 BIOELECTROCHEMICAL KINETICS

FOR

SECONDARY CURRENT DISTRIBUTION

MODELS ................................................................................................................................... 82 6.4 SUMMARY AND PERSPECTIVES OF THE EXPERIMENTAL WORK ...................................... 83 CHAPTER 7: CONCLUSION AND PERSPECTIVES ...................................................... 85 REFERENCES ....................................................................................................................... 86 APPENDIX 1: ELECTRODE KINETICS ................................................................................... 89 APPENDIX 2: NORMALIZATION............................................................................................ 91

4

LIST OF FIGURES

Figure 2.1: Organization chart of VITO Figure 3.1: Schematic diagram of a typical MFC Figure 3.2: Schematic diagram of a typical MEC Figure 3.3: Schematic view of DET via (A) membrane-bound cytochromes and (B) conducting nanowire Figure 3.4: Schematic view of MET via artificial and self-produced mediators Figure 3.5: Polarization curve (solid line) and power curve (dashed line) of MXCs Figure 3.6: Schematics of a two-compartment MFC in rectangular shape (left), miniature shape (right) Figure 3.7: One-compartment MFC Figure 3.8: Schematic of cylindrical shape MFC with open-air cathode Figure 3.9: Schematic of stacked MFC Figure 3.10: Schematic representation of potentiostatic regulation for three-electrode setup Figure 3.11: Typical CV for an MXC Figure 3.12: Typical CA response for an MXC Figure 4.1: Potential losses (overpotentials) over the polarization curve of MFC Figure 4.2: Two parallel plate electrodes opposite to each other in the walls of an insulating flow channel (solid curves: current lines, dashed lines: equipotential surfaces) Figure 4.3: Primary potential distribution over parallel plate electrode (A: ∆V=Eeq and B: ∆V>Eeq) Figure 4.4: Secondary distribution over parallel plate electrodes Figure 4.5: Concentration profile at electrode –electrolyte interface Figure 5.1: Initial model geometry (full cell configuration with porous electrodes and grid current collector) Figure 5.2: Primary current distribution profile over the initial model geometry as output image Figure 5.3: Primary current distribution profiles with plate (left) and grid (right) current collector designs Figure 5.4: Primary current distribution profiles with different lug widths (W_lug) 5

Figure 5.5: Primary current distribution profiles with different lug heights (H_lug) Figure 5.6: Primary current distribution profiles with plate (left) and grid (right) lug design Figure 5.7: Primary current distribution profiles for half-cell configurations with plate (left) and (grid) current collector Figure 5.8: Four geometry configurations with different lug and cc designs Figure 5.9: Primary current distribution profiles over the four geometry Figure 5.10: Normalized value of jmax vs normalized value of parameter Figure 5.11: Image of a mesh current collector made of welded wires Figure 5.12: Primary current distribution profiles over the reference model and decreased W_peld Figure 5.13: Primary current distribution profiles over the reference model and increased H_peld Figure 5.14: Secondary current distribution profiles over the outer boundary (left) and inner surface (right) of the electrode Figure 6.1: Schematic view of the MXC half cell in recycled flow batch mode Figure 6.1: Schematic view of the MXC half cell in recycled flow batch mode Figure 6.2: Shematic view of the determination of Eanapp and Ecatapp for acetate oxidation and fumarate reduction during CA Figure 6.3: CA – half cell experiment with acetate Figure 6.4: CV after inoculation with acetate at initial time (t=0) of CA Figure 6.5: CV after glucose addition at t=12 d of CA Figure 6.6: CA – 1st set of half cell experiments with fumarate Figure 6.7: CV after inoculation with fumarate at initial time (t=0) of CA Figure 6.8: CV after glucose addition at t=16 d of CA Figure 6.9: CA - 2nd set of half cell experiments with fumarate Figure 6.10: CV after inoculation with fumarate and glucose at t=7 d (left) and t=11 d (right) of CA Figure 6.11: CA - 3rd set of half cell experiments with fumarate Figure 6.12: CV after inoculation with fumarate and glucose at initial time (t=0) of CA Figure 6.13: CV after inoculation with fumarate and glucose at t=4 d of CA Figure 6.14: CA - 4th set of half cell experiments with fumarate 6

Figure 6.15: CV after fumarate inoculation at initial time (t=0) of CA Figure 6.16: CV after glucose inoculation at t=16 d of CA Figure 6.17: CV at lowest scan rate (1 mV/s) after glucose inoculation at t=16 d of CA Figure 6.18: CA -5th set of half cell experiments with fumarate Figure 6.19: CV after glucose inoculation at t=14 d of CA Figure 6.20: CV at lowest scan rate (1 mV/s) after glucose inoculation at t=14 d of CA Figure 6.21: Bioelectrochemical kinetics for secondary current distribution model: I vs E after achieving the jmax=7.32 a/m² at the 5th set of experiment

7

LIST OF TABLES

Table 4.1: Hypotheses and parameters for each type of current and potential distribution Table 5.1: Geometry and material parameters of the reference model Table 5.2: Variation of the parameters and jmax Table 6.1: Components of 1 L NBAF medium

8

GLOSSARY *Listed in appearance order *Abbreviations BES SCT MXC MFC MEC MES PEM DET MET OCV BOD COD TOD CE EE CA CV EIS WE RE CE FRA cc EET NBAF SHE AC SS *Notations and Symbols S E I R Pd A Pv V Yx/s t η Φ j x L K k j0 α

Bioelectrochemical Systems Separation and Conversion Technology Microbial Electrochemical System Microbial Fuel Cell Microbial Electrolysis Cell Microbial Electrosynthesis Proton exchange membrane Direct electron transfer Mediated electron transfer Open Circuit Voltage Biologic oxygen demand Chemical oxygen demand Total oxygen demand Coulombic efficiency Energy efficiency Chronoamperometry Cyclic voltammetry Electrochemical impedance spectroscopy Working electrode Reference electrode Counter electrode Frequency response analyzer Current collector Extracellular electron transfer Nutrient Broth Acetate Fumarate Standard Hydrogen Electrode Activated carbon Stainless steel Substrate Potential Current Resistance Power density Area Volumetric power Volume Growth yield Time Overpotential potential gradient Current density Distance from the center of the electrode Length of the electrode Complete elliptic integral Conductivity of the electrolyte Exchange current density Charge transfer coefficient

9

n R T F y N z c D A M G F *Indices cell ext max int eq ohm act conc preced n avg M S an, A cat, C ox red i min app

Number of electron exchanged Universal gas constant Temperature Faraday’s constant y-Coordinate Flux Oxidation state Concentration Diffusion coefficient Acetate Mediator Glucose Fumarate Electrochemical cell External Maximum Internal Thermodynamic equilibrium Ohmic Activation Concentration Preceding Normal component Average Metal Surface Anode Cathode Oxidation Reduction Specie of i Minimum Applied

10

ACKNOWLEDGEMENTS

I would like to express my heartfelt gratitude to my promoter, Xochitl Dominguez-Benetton for her continuous support and guidance to make possible this work, but especially for her great kindness and always having her office door open whenever I need. Thank you for enforcing me to do my best in the most professional way, thus showing me how to do research. I also would like to acknowledge my co-promoters, Deepak Pant and Karolien Vanbroekhoven for making me part of their team since the first moment of this internship and for their insightful comments on my work. I am thankful to all for giving me the opportunity of working in such a great environment but also continuing to my future studies in their group. I am thankful to my academic advisor, Peter Winterton for his motivation, precious help, and valuable contribution to this study. I also would like to express my deepest gratitude to my professor Theo Tzedakis for holding his students at high educational level and letting me part of this program. They are an outstanding example of how to be a successful professional but most importantly, how to be a kind and caring person. I thank to my co-workers whom I share the office, laboratory and life in this small town for their dear friendships throughout my stay in Belgium. I also thank to all my classmates, for their kindness, especially, for their effort to understand my French. I would also like to thank to the people who are in Turkey; my family and friends, for their support and encouragement in every step of this challenging period my life where I change three countries in two years. My special thanks to Omar, for being such an amazing person and being in my life.

.

11

ABSTRACT

Microbial Electrochemical Systems (MXCs) were evaluated through numerical computational modeling by using COMSOL Multiphysics. 3-D models based on primary current distribution displayed that usage of grid current collector in Microbial Fuel Cells (MFCs) increases the distribution profile at the electrolyte-electrode interface. The key parameters affecting the system performance were determined lug positioning, grid size in this study. The secondary current distribution models with bioelectrochemical kinetics obtained from experimental work showed a more homogenous distribution profile for the same than the primary current distributions models. The maximal current density solutions of the secondary current distribution models showed accuracy with the experimental results. While bioelectrochemical kinetics was achieved, the electrochemical performances of X strains were monitored. X strains proved their ability of mediated electron transfer (MET) in a high conductive medium (145 mS/cm) by achieving 7.32 A/m² of maximal current density in the half-cell MFC containing fumarate and glucose mixture as substrate and AQDS as mediator.

12

OBJECTIVES

The present document is a product of five months of research work that has been conducting at VITO (Mol, Belgium) and it was written in order to obtain the degree of Masters in Process Engineering – Electrochemical Processes of University of Toulouse III – Paul Sabatier. The overall aim of this study is to investigate the microbial electrochemical systems (MXCs) in order to gain the knowledge needed for the improvement of these systems, through a unified approach which combines computational modeling and experimental methods. This principal target is translated in the following sub-goals; 1. To evaluate current distributions at the electrode-electrolyte interfaces in fuel cells, through developing numerical computational models in COMSOL 2. To evaluate current distributions at the electrode-electrolyte interfaces in microbial fuels cells, through experimental electrochemically-active microbial kinetics and numerical computational models in COMSOL. 3. To construct models for defined electrochemical reactor geometries to solve the primary distributions at half cell and full cell level. 4. To construct models to solve the secondary distributions at half cell and full cell level, for the system previously defined in primary current distribution cases. 5. To evaluate different operational parameters in the constructed models in order to improve electrochemical cell components and assemblies in microbial fuel cells. 6. To analyze the experimental electrochemical response of electrochemically active biofilms in order to obtain bioelectrochemical kinetics for constructed models. Furthermore, examination of the electrochemical responses of the chosen bacteria in the context of microbial electrosynthesis (MES) is considered as a secondary target of this study.

13

CHAPTER 1: INTRODUCTION

Recent petroleum shortage and global energy crisis have triggered researches towards sustainable production. Microbial electrochemical systems (MXCs) have emerged as a sustainable and innovative way of energy production and chemical synthesis with the ability of electron transfer between microbes and electrodes, while decreasing the energy needs for the process. Today, such technologies have gained great world-wide attention due to their high

conversion, selectivity, efficiency and promising potential in wide range of applicability. Many studies have focused on the field of MXCs and the primary target thereof has become to increase the performances of MXCs towards scale-up and developing cost-efficient designs. Despite the great advances, MXCs are still not well understood, since they are rather complex systems involving many disciplines; such as microbiology, engineering, electrochemistry. Therefore, efforts should be given from different aspects and collective approaches should be considered, such as combining numerical methods with experimental studies for a better understanding in terms of increasing the performances of such complex systems. This is why this study approached in a versatile way to MXCs. After a brief description of brief description of institute where the current study takes place (Chapter 2); Chapter 3 and 4 are dedicated to provide the state-of-the-art of MXCs and designing and optimization of these systems through modeling approaches respectively. Chapter 5 introduces the modeling work performed, whereas Chapter 6 explains the experimental work conducted in parallel. Chapter 7 presents the overall conclusion and the perspectives of the current work.

14

CHAPTER 2: DESCRIPTION OF THE INSTITUTE

2.1 Profile VITO - Flemish Institute for Technological Research is an autonomous public research company. This independent and customer-oriented research institute was founded in 1991 with the purpose of providing innovative technological solutions as well as scientifically based advice and support in order to stimulate the sustainable development of Flanders society. VITO is located in Mol, Antwerp and today, it is one of the largest Belgian research institutes with approximately 700 researchers of 15 different nationalities plus supporting staff. 2.2 Activities VITO works with companies, governments, universities and other research institutions, both in Belgium and abroad. Therefore, VITO activities can be summarized as being based on 3 main objectives: i.

Innovation for industry

Companies tend to combine environmental and company profits to find their way to ecoefficiency by re-evaluating their product design, optimizing their processes, and re-using their waste products. VITO supports them in all these domains by developing important innovations, and follows up on all international innovations with customer-driven basic projects. ii.

Technological pillar for governments

As a research partner for local, Flemish, and European governments; VITO delivers the necessary scientific input for government decision takers to build new policies in an efficient and goal-oriented way. VITO also develops barometers that measure the results of a policy and that quickly reveal where adjustments are necessary. iii.

International scientific research

VITO works with universities and other research institutions that lead to various common international research programs, publications and communications at international conferences and symposia. It also plays an active part in Europe such as in the Framework 15

Programs of the European Union. Besides all this, it carries out its own strategic research in different technological fields. Presently VITO is working strategically in China, India and Vietnam. 2.3 Research Fields Research at VITO is centered on three main societal challenges: transition towards a society less dependent on fossil fuels, transition towards a more sustainable industry in Flanders, and improved quality of life by better use of the environment. With the aim of answering the needs of the development of these challenges, different research topics that are currently investigated at VITO can be listed as; o Earth observation o Environmental modeling o Transition energy and environment o Energy technology o Materials technology o Environmental risk and health o Separation and conversion technology In Figure 2.1, the departments dedicated to conducting these research fields can be seen together with the main organizational structure of VITO.

Figure 2.1: Organization chart of VITO 16

2.4 Electrochemistry at VITO Electrochemical engineering and electrochemistry studies have a very important role at VITO and they are currently carried out for various types of material developments (separators, electrodes, and membrane-electrode assemblies), electrochemical cell designs as well as their diverse applications (wastewater treatment in microbial fuel cells, bio-electrosynthesis, bioelectrolysis processes). The current report is a product of a performed work in a research group which investigates Bioelectrochemical Sytems (BES) in a versatile way, from different educational backgrounds and scientific experiences such as electrochemistry, materials science and engineering, biotechnology, microbiology and process engineering. This BES-research group is located under the ‘Separation and Conversion Technology (SCT) Unit’ that belongs to the ‘Industrial Innovation Department’ as can be seen from organization chart of VITO above. The following chapters (Chapter 3 and 4) will help to provide a better understanding of some of the studies of this BES-research group by presenting the state of the art.

17

CHAPTER 3: MICROBIAL ELECTROCHEMICAL SYSTEMs (MXCs)

Bioelectrochemical Systems (BESs) consist of devices in which electrochemically active biological components are used as catalysts for electrochemical reactions occurring at electrodes (Pant et al., 2012). BESs are generally classified depending on the type of biocatalyst which can be enzymes or microbes; however, focus here is on Microbial Electrochemical Systems (MXCs) since enzymatic systems are out of the scope of the present work. The MXC concept was first proposed by Potter in 1910 based on the idea that microbial catalytic activities and conversions could generate electrical current (Potter, 1910). This is the first description of the Microbial Fuel Cell (MFC); however, MFCs did not gain much attention until the 1980s when it was found that energy generation can be enhanced by the addition of electrochemically active mediators that accelerate electron transfer in the system. Although mediated electron transfer highly increased the attractiveness of MFCs, toxicity and instability of most of the mediator species limited their applications and the real breakthrough was made at the beginning of 2000 when certain particular bacteria were found to be capable of direct (mediator-less) electron transfer which opens new pathways for MFC technologies (Du et al., 2007). Since then, MFCs have received world-wide attention for alternative energy production, wastewater treatment and fuel recovery from organic waste (Pant et al., 2012). In 2005, a new technology was developed for hydrogen generation; named the Microbial Electrolysis Cell (MEC). The same potential benefits of using MFCs to generate electricity from wastewater treatment were applied for electricity-driven hydrogen generation within MECs (Liu et al., 2005b). More recently, in 2010, Microbial Electrosynthesis (MES) was described for bioproduction of valuable chemicals, a principle that was proven when electricity-driven acetate production was achieved from CO2 reduction (Nevin et al., 2010). These three systems now are considered as presenting highly sustainable and innovative approaches for energy production, fuel recovery and chemical synthesis (Dominguez et al.,

18

2012). Nowadays, numerous studies are performed on MXC technologies and innovative designs have emerged along with newer concepts in MXC applications (Pant et al., 2012); (Watanabe, 2008). In this way, Chapter 3 introduces the fundamentals of Microbial Electrochemical Systems (MXCs) in order to provide a broad perspective on these novel technologies. 3.1 Types of MXCs 3.1.1 Microbial Fuel Cell (MFC) Microbial Fuel Cells (MFCs) are devices that use bacteria as catalysts to oxidize organic substrates and generate electrical current (Logan et al., 2006). For instance, bacteria in the anodic chamber oxidize the added substrates (S) and release electrons (e -) towards an anode, as well as protons (H +). Carbon dioxide (CO 2) is produced as an oxidation product. The electrons released are transported from the anode (A) to the cathode (C) through an external circuit which generates electricity. After passing the proton exchange membrane (PEM), the protons enter the cathodic chamber where oxygen (O 2) reduction takes place and water (H 2O) is formed (See Figure 3.1).

Figure 3.1: Schematic diagram of a typical MFC Typical electrode reactions of MFC are shown below using acetate (CH 3COO-) as a model substrate:

19

 Anodic reaction: Acetate oxidation CH3COO- + 2H2O ⟶ 2CO2 + 7H+ + 8e Cathodic reaction: Oxygen reduction O2 + 4e- + 4H+ ⟶ 2H2O MFC operates as a galvanic cell in which the anodic potential (E an) is lower than the cathodic potential (E cat). Cell reactions occur spontaneously and as a result; electrical current is generated. 3.1.2 Microbial Electrolysis Cell (MEC) Microbial Electrolysis Cells (MECs) function with almost the same mechanism as MFCs; however, instead of gaining energy from the system, additional energy needs to be applied in order to drive the electrochemical reactions. When separate reactions occurring at the anode (A) and the cathode (C) are analyzed in a MXC, it is observed that the anodic potential (E an) is higher than the cathodic potential (E cat) implying that the MFC operates as an electrolytic cell (Dominguez et al., 2012).

Figure 3.2: Schematic diagram of a typical MEC In case of hydrogen (H 2) production in a MEC, an organic substrate (S), e.g. acetate, is oxidized by electrochemically-active (EA) bacteria in the anodic chamber and carbon dioxide (CO2) is produced as an oxidation product. The electrons released (e-) are transferred to the cathodic chamber. The protons enter the cathodic chamber after passing

20

the proton exchange membrane (PEM). On the cathode, proton (H +) reduction takes place and hydrogen (H2) is produced (See Figure 3.2). 3.1.3 Microbial Electrosynthesis (MES) Microbial Electrosynthesis (MES) has evolved from the concept of MEC. In MES, bacteria still oxidize the added substrates at the anode and released electrons are transported from the anode to the cathode where oxidized species are reduced into value-added products. For such production, MESs can follow at least three different approaches: reduction of CO2 to an organic product (e.g. acetate) reduction of an organic substrate to a desired product (e.g. fumarate to succinate or glucose to butanol) oxidation of an organic substrate to a desired product (e.g. glycerol to ethanol) (Dominguez et al., 2012) Today, the idea of producing fuels and chemicals from CO2 or waste organics is one of the strongest driving forces for MXC studies. With these empirical studies as well as the increased knowledge about electron transfer mechanism gained over the past few years MES has a great potential to become a key process in future bioproduction (Rabaey and Rozendal, 2010). 3.2 Electron Transfer Mechanisms Electron transfer from microbes to electrodes (or vice versa) is the process that links microbiology and electrochemistry in MXCs; this is why determination of the mechanisms of electron transfer is regarded as the key issue for elucidating overall system behavior. So far, two main possible mechanisms have been described: direct electron transfer (DET) and mediated electron transfer (MET). 3.2.1 Direct Electron Transfer (DET) Direct electron transfer mechanism (DET) occurs via physical contact between the electrode and bacterial cell components or membranes without involvement of dissolved redox species. DET requires the bacteria to possess organelle- or membrane-bound redox proteins; e.g. cytochromes to carry out the electron transport between the bacterial cell and the electrode. 21

Metal reducing bacteria often need solid terminal electron acceptors like Fe(III) oxides in their natural environment and in the case of MXCs, the anode plays the role of the solid electron acceptor. Most research concerning direct electron transfer has focused on these metal reducing bacteria, specifically, Geobacter and Shewanella (Schröder, 2007). Recently, it has been reported that Geobacter sulfurreducens and Shewanella oneidensis can evolve electrochemically conducting molecular pili (nanowires) which maintain a conductive path between the cytochromes present in their outer cell membrane and the electrode. This allows microorganisms to reach and utilize more distant solid electron acceptors (Figure 3.3). The formation of such nanowires may allow the development of thick biofilms and thus higher anode performances (Schröder, 2007).

Figure 3.3: Schematic view of DET via (A) membrane-bound cytochromes and (B) conducting nanowire (Schröder, 2007) 3.2.2 Mediated Electron Transfer (MET) Microbes can use redox active molecules that ‘shuttle’ electrons from bacteria to the electrode surfaces. These conductive molecules, so called mediators, can be artificial or self–produced by the bacteria (Figure 3.4). The most common artificial mediators are neutral red, thionin, methyl viologen or anthraquinones (Rabaey and Rozendal, 2010). The addition of the mediators increases the performance in terms of current generation since they facilitate electron transfer, but when it comes to selectivity, toxicity and limited stability DET can be considered as a great advantage over MET (Rabaey and Rozendal, 2010); Schröder, 2007).

22

Figure 3.4: Schematic view of MET via artificial and self-produced mediators (Schröder, 2007) 3.3 Performance Parameters MXC performance can be evaluated in many ways, but principally through energy generation and treatment efficiency. Although the newest application areas have emerged for MXCs, the analysis of the parameters associated with these two current purposes are regarded to be well-established approaches in terms of evaluating the system performance. 3.3.1 Energy Generation Theoretical and Actual Cell Potential The theoretical cell potential or electromotive force (emf) of the overall reactions occurring in an MXC is defined as the potential difference between the cathode and the anode. The emf refers to the best possible cell potential which is the maximum cell potential that can be attained in a MFC and the minimum potential required to drive it. However, the actual cell potential (Ecell) is lower than this theoretical value because of irreversible energy losses, the so called overpotential (η). Ecell = emf – η Efficient MXC designs therefore need to focus on reducing overpotentials as much as possible in order to optimize system performance. Different types of overpotentials occurring in a MXC due to different phenomena are largely described in Chapter 4.

23

Power Output Power can be regarded as the most significant output to be evaluated in terms of electricity generation. Direct electrical power measurement is done through measuring the cell potential (Ecell; V) and the current (I; A) at a fixed external resistance (R ext; Ω). Thus, power (P; W) is calculated as; P =IEcell Power output can be attained in different forms. For an objective evaluation, direct power measurement is often normalized to the electrode surface area (A; m2) which makes possible comparison between different systems. This normalized value is called the power density (Pd; W.m–2) and can be calculated as: Pd = Ecell2 /ARext When power output is needed to be normalized to the electrode volume (V; m 3), generally with the purpose of facilitating the calculations of reactor size or costing, it is called volumetric power (Pv; W.m–3) and it is calculated as: Pv = Ecell2 / VRext Polarization Curve Polarization and power curves are functional methods to calculate the maximum power (Pmax) that can be attained in MXCs as well as Rint (Ω) and OCV (V) magnitudes. Rint (Ω) is the internal resistance of the system and OCV (V) is the open circuit voltage. OCV is defined as the measured cell potential after some time in the absence of current. A polarization curve illustrates the cell potential (Ecell) as a function of current (I); it is plotted by measuring currents at different potentials. Polarization curves provide power curves, which plot power (P) versus current (I) (Figure 3.5).

24

Figure 3.5: Polarization curve (solid line) and power curve (dashed line) of MXCs (Watanabe, 2008) As seen in the figure above, the relationship between Ecell and I is expressed as: Ecell = OCV – IRint 3.3.2 Treatment Efficiency When MXCs are applied for waste treatment; they are evaluated in terms of treatment efficiency (%) which can be calculated through removal of the biological oxygen demand (BOD; kg), chemical oxygen demand (COD; kg) or total organic carbon (TOC; kg). COD is the most common measurement for treatment efficiency which is referred to as COD-removal efficiency, in most cases. It is an indicator of fuel conversion either into current or biomass, by showing the ratio between the removed and influent organics. In addition to COD-removal efficiency, treatment efficiency can be evaluated trough other parameters which can be listed as: Coulombic efficiency (CE, %) is the ratio of electrons recovered as current to the maximum number of electrons contained in the fuel. Loading rate (kg.m–3.d–1) indicates the rate at which COD that is loaded into a MXC. It is measured by normalizing the amount of COD loaded into the electrode volume (m3) and time (d). Growth yield (Y x/s) is an index that shows substrate utilization as the electrons are converted into biomass. It is found by normalizing the amount of COD produced to time (d) (Logan et al., 2006).

25

Energy efficiency (EE %) is calculated as the ratio of power produced to the heat energy obtained by combustion of the substrates added. It is the most significant parameter to evaluate the MXCs in terms of energy recovery processes (Watanabe, 2008). All together, these parameters provide an accurate characterization of the waste or wastewater treatment that can be correlated to the electrochemically active biomass at the electrode surface, as well as to the active biomass with no electrochemical contributions to the process (Dominguez et al., 2012). Nevertheless, comparisons of these performance data in terms of both energy generation and treatment efficiency are difficult since every study refers to a specific combination of reactor volume, membrane, organic load, and bacteria. In addition to energy generation and treatment efficiency, the relative amount of product formation from a substrate with the specific microbial community is started to be regarded as a key performance parameter along with the concept of electricity-driven biosynthesis, particularly for MES studies. 3.4 MXC Designs In order to increase the performance parameters described above, efficient MXC designs are required. In recent years, various studies have been carried out in this direction by using different reactor, material, microbe and fuel configurations under different operating conditions. Even though the studies reported offer valuable knowledge, it is important to note that the microbiological or electrochemical optimizations for one type of MXC are not always optimal for another type. This is why MXCs should be considered as complex systems, and when evaluating a single part, the effect of all other parts of the system should also be taken into account (Dominguez et al., 2012); (Watanabe, 2008). 3.4.1 Reactor Configurations Reactors have been constructed with various configurations in order to obtain better MXC performances, in other words, to minimize the potential losses. This effort involves a number of modifications, mostly concerning the system geometry. For example, placing the electrodes within a shorter distance from each other is one of key factors that minimize the ohmic drop

26

in electrochemical systems and consequently MXCs, since the resistance is proportional to distance (Logan et al., 2006). A typical double-chamber (two-compartment) reactor consists of an anodic and a cathodic chamber separated by a proton exchange membrane (PEM) as they are illustrated in the Figure 3.1 and 3.2 before. They typically run in batch mode, mostly, in laboratories. They can be in various shapes such as rectangular or miniature as it can be seen in Figure 3.6.

Figure 3.6: Schematics of a two-compartment MFC in rectangular shape (left), miniature shape (right)(Du et al., 2007) Due to their complex designs, double-chamber MFCs are difficult to scale-up even though they can be operated in either batch or continuous mode. Single-chamber (one-compartment) reactors (Figure 3.4) offer simpler designs and cost savings. They typically possess only an anodic chamber without the requirement of aeration in a cathodic chamber (Du et al., 2007).

Figure 3.7: One-compartment MFC (Watanabe, 2008) Park and Zeikus designed a one-compartment MFC consisting of an anode in a rectangular anode chamber coupled with a porous air-cathode that is exposed directly to the air as shown in Figure 3.7 (Park and Zeikus, 2003). 27

Liu and Logan also designed an MFC consisting of an anode placed inside a plastic cylindrical chamber and a cathode placed outside which can be seen in Figure 3.8 (Liu and Logan, 2004).

Figure 3.8: Schematic of cylindrical shape MFC with open-air cathode (Du et al., 2007) A stacked MFC is shown in Figure 3.9 for the investigation of performances of several MFCs connected in series and in parallel. Enhanced voltage or current output can be achieved by connecting several MFCs in series or in parallel (Aelterman et al., 2006).

Figure 3.9: Schematic of stacked MFC (Du et al., 2007) 3.4.2 Materials The material of the electrodes has a direct influence on cell performance since different activation polarization losses are observed with different types of materials due to their intrinsic characteristics. Undesired high activation polarization losses can be avoided by increasing the electrode quality. Carbon-based materials (e.g., activated charcoal, carbon cloth or graphite felt) are generally used to construct electrodes owing to their large surface areas. Platinum (Pt) and Pt-black 28

electrodes generally perform better than carbon-based electrodes and good catalysts when it comes to oxygen reduction. However, their high costs make practical applications prohibitive. Several studies have proved that modifications of carbon-based substrates result in better performance. Schröder et al. showed that higher currents can be gained with platinized-carbon cloth, compared to unmodified carbon cloth under the same operating conditions (Schröder et al., 2003). Park and Zeikus reported that the incorporation of the metal ions (Mn4+ and Fe3+) and neutral red as mediators at the anode level enhances electron transfer and results in greater power generation (Park and Zeikus, 2000). An efficient ion transfer system, and more particularly proton-exchanger, also increases the performance in MXCs by reducing the internal resistance and concentration polarization losses. Proton-transfer efficiency depends on the type of the proton exchange membrane (PEM). The most common PEM is Nafion with its high proton selectivity. However, owing to the transport of other cation species (Na+, K+, NH4+, Ca2+, Mg2+) within Nafion the need arises for a PEM which has better selectivity for the protons and not for the other cations (Rozendal et al., 2006). Besides, Nafion membranes are costly; Ultrex membranes and Zirfon have been provided as useful alternatives for achieving suitable ion-exchange at more affordable prices (Dominguez et al., 2012). Oh and Logan showed that the ratio of PEM surface area to system volume is another limiting factor, as well as type of the material for power generation. They reported that internal resistance decreases with increases of PEM surface area in MFCs (Oh and Logan, 2006). In another study, Liu and Logan replaced the PEM with an air-cathode membrane as a separator, in order to boost gas transfer between the compartments of the system. They showed this approach increases power generation; however, one drawback of this method is that it can lead to reduced electron recovery (Liu and Logan, 2004); (Rozendal et al., 2008); (Watanabe, 2008). 3.4.3 Fuel Types Fuel type and concentration influence MXC performance by changing the cell power density. The power density differs widely with the different type of fuels with a specific microbial consortium. Besides a higher fuel concentration gives a higher power density output (Du et al., 2007).

29

A great variety of substrates can be used in MFCs for electricity production, ranging from pure compounds to complex mixtures of organic matter present in wastewater. It is difficult, from the literature; to compare MFC performances due to different operating conditions, surface areas and types of electrodes and different microorganisms used (Pant et al., 2010b). 3.4.4 Microbe Types In MXCs electrons are transferred from the organic substrate to the anode through the microbial respiratory chain that depends on the type of microbes involved. A microbial community consists of a mixed culture usually shows better performance than a pure culture, due to their broader substrate specificity which allows wider substrate utilization (Rabaey and Verstraete, 2005). The selection of a suitable microbial consortium for a given MXC performance is extremely important but difficult to achieve, since microbial structure and activity depend on the operating conditions selected for a particular situation. This is another reason for preferring mixed cultures over pure cultures, due to the different adaptation behaviors of microbes. However, microbial succession in MXCs is still under early studies and its effects on the performance of these systems is still unclear, especially for non-short-term batch experiments (Watanabe, 2008). 3.4.5 Operational Conditions As mentioned above, operational conditions modify microbial activities in MXCs. In addition to that, they also play a significant role on the electrochemical kinetics and transport phenomena in the cell. Numerous operating conditions such as pH, oxidation/reduction potential, ionic strength, and temperature can be counted as factors strongly affecting performance (Liu et al., 2005a). pH differences between the anodic and cathodic chambers have critical impacts on the driving force of proton exchange through diffusion. A different example concerns ionic strength. Liu et al. found that addition of NaCl to MFC improved the power generation by increasing the conductivity. The use of buffers or weak acids has also shown to improve MXC performance under particular operational conditions (Dominguez et al., 2012; (Liu et al., 2005a).

30

3.5 Electrochemical Characterization Techniques Characterization is a very important issue to elucidate the performance and properties of complex systems like MXC; thereby, the system efficiency can improve for existing and developing designs. Various types of characterization techniques can be applied from different disciplines for this purpose. For example, microbial characterization can identify the microbial composition at the electrode and the ones suspended in the electrolyte while microscopic characterization can be applied for morphological determination of electrode surfaces, and more importantly biofilm structures. Electron transfer between microbes and electrodes allows characterizing MXCs by potentiostatic techniques which perturb the system with potential and measure the current as output; such as open circuit voltage (OCV) measurements, chronoamperometry (CA) , and cyclic voltammetry (CV) ; more recently electrochemical impedance spectroscopy (EIS). In order to apply these techniques a potentiostat is required which operates typically in threeelectrode-setups, consisting of a working electrode (WE, anode or cathode), a reference electrode (RE), and a counter electrode (CE) (Figure 3.10).

Figure 3.10: Schematic representation of potentiostatic regulation for three-electrode setup More advanced measurements can be done when the potentiostat is equipped with a frequency response analyzer (FRA), allowing electrochemical impedance spectroscopy measurements (EIS).

31

3.5.1 Open Circuit Voltage Open circuit voltage (OCV) is measured by means of the potential difference between the working electrode and a reference electrode in the absence of current. The application of this technique is simple. The disadvantage of OCV measurements is the possible overinterpretation of results. The technique is recommended to be used along with othe r electrochemical techniques to determine the cathodic and anodic influences of the electrochemical processes since with OCV the separate contributions are monitored together. 3.5.2 Cyclic Voltammetry (CV) Cyclic voltammetry (CV) is an electrochemical technique in which current (I) is recorded through a working electrode, while the applied potential (E) to the electrode is controlled as a linear function of time (t). In CV experiments, the potential is applied reversibly at a certain scan rate (Figure 3.11).

Figure 3.11: Typical CV for an MXC This is a widely applied technique, particularly in the systems where complex electrode reactions take place (Bard and Faulkner, 2001). In MXCs, the ability to determine the standard redox potential of electrochemically active elements of the system gives a broad understanding of the microbial activities and electrode performances (Logan et al., 2006).

32

3.5.3 Chronoamperometry (CA) Chronoamperometry (CA) is an electrochemical technique in which the potential of the working electrode is stepped and the resulting current occurring at the electrode is monitored as a function of time (Figure 3.12).

Figure 3.12: Typical CA response for an MXC CA generates high charging currents which decay exponentially with time. Since the current is integrated over a relatively long time, CA gives a better signal in comparison to other techniques for MXCs. CA also allows studying the microbial capabilities for electron transfer as well as determination of the optimal conditions for the system (Dominguez et al., in press). 3.5.4 Electrochemical Impedance Spectroscopy (EIS) In addition to the conventional techniques mentioned above, electrochemical impedance spectroscopy (EIS) has started to gain attention for profound MXC analyses (Strik et al., 2008). In EIS, the frequency response of an electrochemical system to an alternate signal is analyzed in a transfer function, between an input signal (e.g. voltage) and an output one (e.g. current) through a frequency-response analyzer (FRA). EIS can be used to measure the ohmic and internal resistance of MXCs as well as to provide additional insight into the operation of an MFC. The interpretation of E IS data can be rather complex. This rather sophisticated EIS method can provide superior and additional information about the system (Logan et al., 2006). 33

The four aforementioned methods are also applied during this research and their analyses are largely presented in Chapter 6. In addition to electrochemical techniques, modeling is considered as a powerful method for MXC characterizations. Chapter 4 and 5 are dedicated to introduce the state-of-the-art and practical applications of MXCs characterizations by modeling respectively.

34

CHAPTER 4: DESIGN AND OPTIMIZATION OF MFCs VIA MODELING

The interest in MFC development towards industrial scale has risen sharply in recent years due to their possibility to yield electricity from organic waste or biomass decomposition; however, many challenges in terms of increasing system performance still exist. This may be understood better considering the fact that multiple processes (physical, electrochemical, chemical, or biological) and phenomena (diffusion, adsorption, or ion migration) are simultaneously involved; thereby accurate optimizations for systems like MFCs are rather complex. Most of the time, experimental studies are useful but not satisfactory for explaining all the pertinent system parameters; the reason for this is that they focus on either microbiological or engineering aspects separately. Modeling, and more particularly multi-physics modeling, can be considered as an appropriate method in order to gather information from several disciplines for increasing the overall system performance through a multidisciplinary approach. The research in this direction for MFCs, coupling microbial with electrochemical dynamics and kinetics was successfully addressed by Picioreanu et al. in 2007, when they presented a computational model for biofilm-based MFCs. In their studies, they showed a heterogeneous current distribution over the electrode surface for young biofilms, but a uniform distribution in older and more homogeneous biofilms by two-dimensional (2D) and three-dimensional (3D) model simulations (Picioreanu et al., 2007);(Picioreanu et al.,2008). Furthermore, a newer modeling approach was developed by Picioreanu et al. to observe the influence of 2D and 3D biofilm and electrode geometry on the MFC performances by using very efficient combination of Matlab, Java, and COMSOL Multiphysics (Picioreanu (Picioreanu et al., 2010). However, such models have appeared from the perspective of understanding the fundamentals of electrochemically-active biofilm behavior over the electrodes, rather than from the engineering approach with the purpose of optimizing reactor geometries, membrane-electrode assemblies, construction and operational conditions.

35

Regarding the fact that the geometry greatly affects system performance, current and potential distribution is possibly the most crucial point when modeling an electrochemical cell since there is direct association between them (Yoon et al., 2003). Hence, modeling approaches based on current and potential distribution can guide to a better industrial prospection for MFCs. Although few studies have introduced that approach, electrochemical engineering models of MFC systems (single-cell and stacks) concerning current and potential distributions are expected to be one of the significantly developing fields in the very near future, in parallel, of course, to the progress of MFC studies from the practical engineering point of view. 4.1 Modeling Current and Potential Distributions in MFC 4.1.1 Overpotentials Overpotentials occurring in an electrochemical system are key parameters to be understood for modeling based on current and potential distributions since they play a major part in defining the type of the distribution. As briefly mentioned in section 3.3.1, overpotential (η) is defined as the potential difference between the half reaction reduction potential (Eeq) and the potential at which the redox event is experimentally observed (E) (Bard and Faulkner, 2001). It is directly linked to the efficiency of any electrochemical system and in case of MFCs overpotential signifies recovery of less energy than the thermodynamics would predict; in other words, energy losses. η = E – Eeq For MFCs, four major overpotentials are described: i.

Ohmic overpotential (ηohm)

The ohmic overpotential in an MFC includes both the resistance to the flow of electrons through the electrodes and interconnections, and the resistance to the flow of ions through the membrane (if present) due to the geometry of the system. Ohmic overpotential can be reduced by minimizing electrode spacing, using a membrane with a low resistivity, (if practical) increasing solution conductivity to the maximum tolerated by the bacteria.

36

ii.

Activation overpotential (ηact)

Due to the activation energy needed for a redox reaction, activation overpotential occurs during the transfer of electrons by electrochemical reactions at the electrode surface. It can be reduced by increasing the electrode surface area, improving electrode catalysis, increasing the operating temperature, and through the establishment of an enriched biofilm on the electrode. iii.

Concentration overpotential (ηconc)

The concentration overpotential occurs when the rate of mass transport of species to or from the electrode limits current production. It mainly occur at high current densities due to diffusion. It also considers bubble formation due to the evolution of gas at the electrode; it comprises all phenomena that stimulate concentration differences of the charge-carriers between the bulk solution and the electrode surface (Bard and Faulkner, 2001; Picioreanu et al., 2007) iv.

Overpotential associated to preceding chemical or biochemical reactions (ηpreced)

Although important, the overpotential associated with preceding chemical or biochemical reactions is frequently ignored from overpotential considerations. However, this should not be neglected in the case of microbially-mediated systems, since the involvement of sensitively regulated metabolic chains will always and inevitably precede or succeed the purelyelectrochemically mediated phenomena. Such overpotentials may be masked by the ohmic and concentration losses, since the metabolic influence of bacteria can occur at both the bulk electrolyte or at the electrochemical interface adjacent to the electrode when a microbial biofilm is formed (Dominguez et al. in press). Figure 4.1 displays the overpotentials (or potential losses) occurring in an MFC over a polarization curve.

37

Figure 4.1: Potential losses (overpotentials) over the polarization curve of MFC A polarization curve analysis of a MXC can indicate to what extent the various losses listed contribute to the overall potential drop. This can point to possible measures to minimize them in order to approach the ideal potential. Thus, for an MFC total overpotential can be expressed as; η = ηact + ηconc + ηohm + ηpreced Since both the electrolyte and the electrodes obey Ohm's law, ηohm can be expressed as IRint, in which I is the current flowing through the MFC and Rint is the total cell internal resistance of the MFC. η = ηact + ηconc + IRint + ηpreced 4.1.2 Types of Current and Potential Distributions The distribution of current and potential is highly important in electrochemical systems since the output and the performance of the system can be strongly affected by them (Orazem and Tribollet, 2008). Current and ions flow through the paths that are subject to less resistance, which leads to a certain distribution in the electrochemical systems. This distribution can be due to many factors such as geometry, conductivity of the materials, and the different contributions to overpotential.

38

Typically, a classification is made based on some general rules and assumptions in order to determine such distribution for a macroprofile: i.

Primary current and potential distributions

In the case of primary distribution, the passage of the current through the system is controlled by the ohmic resistance. Therefore, primary distribution applies when the ohmic resistance dominates and surface overpotentials can be neglected (Newman and Thomas-Alyea, 2004; Orazem and Tribollet, 2008). Primary current distribution is independent of flow rate, since it is considered that convection is great enough to eliminate concentration variations, and consequently the distribution is considered symmetric. The electrolyte that is adjacent to the electrode is taken to be an equipotential surface, under the assumption that the concentrations are uniform within the electrolyte. The current density is infinite at the end of the electrodes since the current can flow through the solution beyond the ends of the electrodes (Newman and Thomas-Alyea, 2004). The potential distribution at the electrode surface (ΦS) is a solution of the Laplace’s equation. An example on this solution in case of two parallel plate electrode configurations can be seen in Figure 4.2.

Figure 4.2: Two parallel plate electrodes opposite to each other in the walls of an insulating flow channel (solid curves: current lines, dashed lines: equipotential surfaces) (Newman and Thomas-Alyea, 2004) Generally, the primary distribution shows that the more inaccessible parts of an electrode receive a lower current density. When the electrode and the insulator lie in the same plane, the primary current density is inversely proportional to the square root of the distance from the edge for positions sufficiently close to the edge which can be expressed as:

39

jn javg

cosh / K (tanh 2 ) sinh 2

sinh 2 (2 x / L)

where jn is the normal component of current density on the electrode (A/m2), javg is the average current density, K is the complete elliptic integral of the first kind, x is the distance measured from the centre of the electrode, L the length of the electrode and ε = ПL/2h (Newman and Thomas-Alyea, 2004). Based on the previous explanations, the following hypothesis can be defined for a primary potential distribution model: Activation and concentration overpotentials are neglected. The electrodes are considered as perfect conductors; therefore, the electrode potential (ΦM) is constant. The electrolyte potential over electrodes (ΦS) is constant. The outer surface of the electrodes is considered to be insulating:

0

S

The conductivity of the electrolyte (k) is constant. Limit conditions:  The electrodes are at equilibrium conditions: Ean = Ean,eq Ecat = Ecat,eq  The electrolyte at the electrode surface obeys Ohm’s Law:

j

k

S

Figure 4.3 displays the primary potential distribution model in an electrochemical cell consisting of two parallel-plate electrodes. At equilibrium conditions (∆V = Eeq), no potential gradient within the electrolyte exists (Φelectrolyte= ΦS,A=ΦS,C) and ηohm is therefore neglected (Figure 4.3-A). As a result, current density is equal to zero (j = 0 A/m2). When the potential difference is larger than the equilibrium potential (∆V > Eeq), a potential gradient is established within the electrolyte (ΦS,A ≠ ΦS,C). As a result, current density is no longer equal to zero (j ≠ 0 A/m2). For this case, the potential difference over the equilibrium potential is distributed within the electrolyte (Figure 4.3-B).

40

A

B

Figure 4.3: Primary potential distribution over parallel plate electrodes (A: ∆V=Eeq and B: ∆V>Eeq) (modified from Viaplana, 2010) ii.

Secondary current and potential distributions

The secondary distribution is considered when the reaction kinetics cannot be neglected. Activation overpotential that is associated to the electrochemical reactions at the electrode becomes relevant while concentration variations at the electrolyte are neglected. Therefore, the electrolyte that is adjacent to the electrode can no longer be considered as an equipotential surface (Newman and Thomas-Alyea, 2004; Orazem and Tribollet, 2008). The potential distribution at the electrode surface (ΦS) is a solution of the Laplace’s equation with a more complex boundary condition resulting from the polarization of the electrodes. The electrode kinetics is expressed by the following equation which describes how the electrical current on an electrode depends on the electrode potential, so called Butler-Volmer equation:

j

j0 exp

nF Ox E RT

EeqOx

exp

1

nF RT

E Re d

EeqRe d

At small overpotentials the Butler-Volmer equation can be linearized as:

jn

s

djn d s

s

0

j0 nF RT

s

y

y 0

This provides a linear boundary condition for the Laplace’s equation. y is the coordinate normal to the electrode surface. At sufficiently small overpotentials, the equation can be linearized as: 41

y

J

s

y 0

J (in/iavg) is a dimensionless parameter and equals to nj0nF/ RT. For J ∞ primary current distribution is obtained where the ohmic resistance dominates over the kinetics resistance at the interface. For any finite value of J, secondary distribution is obtained which is more uniform and finite at the edge of the electrode (Newman and Thomas-Alyea, 2004). The following hypothesis can be defined for a secondary potential distribution model based on the previous explanations: Activation overpotential exists: the overpotential is distributed as ohmic drop in the electrolyte and surface overpotential at the the electrode. Concentration overpotential is neglected. The electrodes are considered as perfect conductors; therefore, the electrode potential (ΦM) is constant. The electrolyte potential over electrodes (ΦS) is not constant. And it depends on the local current density. The outer surface of the electrodes is considered to be insulating:

S

0

The conductivity of the electrolyte (k) is constant. Limit conditions:  The electrode are not at equilibrium conditions: j

j0 exp

nF Ox E RT

EeqOx

exp

1

nF RT

E Re d

EeqRe d

 The electrolyte at the electrode surface obeys Ohm’s Law:

j

k

S

Figure 4.4 displays the secondary potential distribution model in an electrochemical cell consisting of two parallel-plate electrodes. The potential gradient within the electrolyte is established (ΦS,A≠ΦS,C) and therefore j ≠ 0 A/m2.

42

Figure 4.4: Secondary distribution over parallel-plate electrodes (modified from Viaplana, 2010) Far from equilibrium conditions, following equations are used : Anode: jan

j0 exp an

Cathode: jcat

nF Ean RT (1

j0 cat exp

Ean,eq )nF

RT

Ecat

Ecat,eq

When a potential difference close to the equilibrium potential j can be described as: j

j0

nF E Eeq RT

j0

nF RT

M

S

Eeq

Therefore, for each electrode: Anode: j an Cathode: jcat

j0 an

nF RT

j0cat

nF RT

M ,A

M ,C

S,A

S ,C

Ean Ecat

Detailed information about the electrode kinetic equations and their simplified forms based on different limiting conditions is provided in Appendix 1. iii.

Tertiary current and potential distributions

The tertiary distribution takes into account the concentration changes mostly due to diffusion; therefore, mass transfer phenomena (reflected as concentration overpotential) play an important role, as well as Ohmic resistance and kinetic limitations (Newman and ThomasAlyea, 2004; Orazem and Tribollet, 2008).

43

Tertiary current and potential distributions apply when Laplace’s equation is replaced by a series of n equations of the form: ci t

Ni

Ri

Where ci is the concentration of species i, N i is the net flux of species i, and Ri is the rate of generation of species, coupled with electroneutrality:

z i ci

0

i

Where n represents the number of ionic species in the system. Thus, tertiary distributions implicate the assumption that concentrations are uniform. Ohmic, kinetic and mass-transfer resistances all play a role in the distribution (Figure 4.5). The distribution of local current density results of the resolution of a complex problem that takes into account the Laplace’s equation and Ohm’s Law, as well as the convective diffusion equation that controls the transport of species to the electrode (Newman and Thomas-Alyea, 2004).

Figure 4.5: Concentration profile at the electrode –electrolyte interface The previous described Butler–Volmer equation is valid when the electrode reaction is controlled by electrical charge transfer at the electrode (and not by the mass transfer to or from the electrode surface from or to the bulk electrolyte). In the region of the limiting current, when the electrode process is mass-transfer controlled, the value of the current density becomes concentration-dependent:

j

Electrode C Re nF d j0 exp E Eeq RT C Re d

Electrode COx exp COx

(1

)nF RT

E Eeq

44

The following hypothesis can be defined for the tertiary potential distribution model: Concentration overpotential exits. The electrodes are considered as perfect conductors; therefore, the electrode potential (ΦM) is constant. The electrolyte potential over electrodes (ΦS) is not constant. The outer surface of the electrodes is considered to be insulating:

S

0

The conductivity of the electrolyte is constant. Limit conditions:  The electrode are not at equilibrium conditions:

j

j0

Electrode C Re nF d exp E Eeq RT C Re d

 Mass balance : Di Ci

Electrode

Electrode COx exp COx

(1

)nF RT

E Eeq

j nF

 The electrolyte at the electrode surface: j

k

S

F

zi Di Ci i

Table 4.1 summarizes the hypotheses and system parameters associated with each type of distribution. Table 4.1: Hypotheses and parameters for each type of current and potential distribution Distribution Primary Secondary Tertiary

Hypothesis Ohmic resistance Ohmic resistance Kinetic resistance Ohmic resistance Kinetic resistance Mass-transport resistance

Parameters Geometry, material conductivity Geometry, material conductivity, activation overpotential Geometry, material conductivity, activation overpotential, concentration overpotential

For MFCs, uniform current distribution over the electrodes is desirable for efficient operation. However; even for a simple-cell configuration, the calculation of the current distribution is a very complex problem. Furthermore, difficulties intensify when increasing the complexity of the cell geometry, which is the main reason to prefer numerical solutions rather than analytical solutions for such calculations.

45

4.2 Numerical Modeling of MFC via COMSOL Multiphysics For the numerical modeling of current and potential distributions, the most appealing tool to deal with complex environments like MFCs is nowadays ‘COMSOL Multiphysics’. Even though no published work exists in this direction so far, since modeling MFCs is a new approach itself, some researchers (Picioreanu C., Delft University of Technology, Netherlands and Bergel A., ENSIACET, France) made efforts to stimulate progress from the fundamental perspective. Engineering-oriented efforts are still highly required. COMSOL Multiphysics is an engineering simulation software that facilitates all steps of a computational modeling process; such as defining the geometry, surface meshing, specifying the physics, solving, and then visualizing the results. COMSOL versions above 4.0 have an application, ‘The Batteries & Fuel Cells Module’, that provides easy-to-use tools for simulation of fundamental processes of fuel cells. With it, the impact on performance of different materials, geometric configurations, and operating conditions can be quickly and accurately investigated. More importantly, the Module features have options to study primary, secondary, and tertiary potentials and current density distributions in electrochemical systems. The electrode reactions, which are coupled to the transport phenomena, provide full descriptions of the electrode kinetics including activation and concentration overpotentials. The cell can contain solid or porous electrodes with dilute or concentrated electrolytes included in the COMSOL Multiphysics library. At this point, in order to have a deeper perception of COMSOL and its function over modeling MFCs based on current and potential distributions; the following chapter (Chapter 5) will be more specific by presenting the practical modeling applications using this efficient tool.

46

CHAPTER 5: MODELING WORK

The present modeling study is conducted with the aim of developing new cost-efficient MFC designs but also improving the system performances of the existing MFCs that are currently under experimental evaluation at VITO. Towards that aim, three dimensional (3D) models based on current and potential distributions are constructed for single and stacked MFCs using COMSOL Multiphysics 4.2 as modeling tool. The current distribution profiles over the electrodes are investigated in order to obtain high system efficiencies and determine the electrochemically active sites. In addition, high local current density magnitudes are aimed at the same time they are homogeneously distributed over the electrode surface. The importance of having homogenous current and potential distribution for MFCs and the advantages of using COMSOL Multiphysics were largely explained in the previous chapter (Chapter 4). Typically, these studies are initiated with the primary current distribution and continued with secondary and tertiary current distribution models respectively, due to the fact that the nonuniformity is reduced from primary to tertiary distribution, in other words, primary current distribution displays the worst case scenario. This is why if the primary current distribution is as uniform as possible, the secondary and tertiary will more likely be uniform as well. The present investigation only covers up to the secondary current distribution model, since the resolution of a tertiary distribution would take deeper understanding and time than those appointed to the present work. Future investigations on this direction are nonetheless suggested to follow-up such work. 5.1. COMSOL Modeling Procedure There are general simulation instructions that should be followed when modeling COMSOL Multiphysics 4.2 for any application, yet these instructions can vary according to the aim of the study. This section briefly introduces the procedure used for building MFC models based on potential and current distributions by explaining the following modeling steps: I.

Model Wizard

When COMSOL is opened the Model Wizard opens by default in order to select the basic elements of the models such as space dimension and physics interfaces. After selecting the 3D 47

as space dimension, the physics interface is chosen as primary (or secondary, tertiary) current density distributions physic interface. II.

Parameter Definitions

After the Model Wizard, parameters used throughout the modeling should be defined on the parameter table that is under the Global Definitions section. Parameters are scalar numbers that can be used for geometric dimensions (e.g. width, length or depth), mesh sizes, physics characteristics (e.g. current or potential), and etc… III.

Geometry

This section is where the model geometry, which is a collection of bounded geometric entities, is built. The geometric entities are dimensioned and positioned based on previously defined parameters and connected to each other with several operations to form a model geometry. Various geometric entities can be used at different shapes and phases; here, solid blocks are preferred as geometric entities to construct the desired MFC models (See Figure 5.1). IV.

Physics

This section demonstrates the features of the previously selected physic interface; primary (or secondary, tertiary) current distribution physic interface. It provides tools for building detailed models of the configuration of the electrodes and electrolyte in electrochemical cells. It also includes descriptions of the electrochemical reactions and the transport properties that influence the performance of batteries, fuel cells, and other electrochemical cells. After building the geometric entities, each of them is attributed to different electrochemical cell components. Material properties, boundary and interface conditions, equations and initial conditions are set in this section. V.

Mesh

This section enables the discretization of the model geometry into small units of simple shapes, referred to as mesh elements. Free Tetrahedral is chosen as mesh technique generating an unstructured mesh with tetrahedral elements for 3D models. The size and sequencing of the mesh elements are introduced. VI.

Study

Finally, the model is run in this section by using the previously created meshes.

48

VII.

Results

After the simulation is completed, final image of the model is seen here by selecting the desired output data (See Figure 5.2). It should be noted that the provided general COMSOL modeling procedure is detailed and further improved towards the needs of each constructed model throughout the modeling work. 5.2. Initial Model Geometry, Definitions and Response A defined geometry was constructed according to an existing prototype of single-cell MFC, with all components fully developed at VITO. Stack-MFC models could be in the future developed using this approach, as well taking into account other than already-existing geometrical designs. In this work, different MFC-component configurations were proposed as deviations from this original geometry in order to improve the individual components, assemblies and full-prototype. For the full cell configuration, the initial model was considered to consist of a rectangular electrolyte domain separating two parallel arrays of porous electrodes (cathode and anode) supported by metallic grid current collectors (without a separator or membrane between the electrodes) and lugs, which are placed on top of each current collector (Figure 5.1).

Figure 5.1: Initial model geometry (full cell configuration with porous electrodes and grid current collector) 49

The metallic current collector provides electronic conductivity to the electrode and increases electron recovery. The lug was considered to be made of the same material as the current collector and it is used for maintaining the external electrical connection between the cell and the potentiostat, power source or external load (resistance), accordingly to the conditions in use. Once the geometry is constructed, it is simulated with the procedure explained above, in order to investigate the primary current distribution at the interface. The electrolyte current density (A/m2), referring to the current density (j) over the porous electrode-electrolyte interface was selected as the most relevant output to be analyzed. As a result, the local current density magnitudes are displayed in different colors over the electrode surface, which provide practical visualization of the current distribution profile. A color scale also is shown next to the geometry which attributes the color range to the numerical solutions for the local current density magnitudes (Figure 5.2). The visualization of the color profile along the geometry and the color scale magnitudes are both highly important for proper interpretations of the output data. The desired output is to have the highest possible maximum current density magnitude (jmax) homogenously or welldistributed at the electrode-electrolyte interface.

Figure 5.2: Primary current distribution profile over the initial model geometry as output image

50

In the figure above, it is seen (on the color scale) that the maximum local current density (jmax) obtained is 12.401 A/m² while minimum (jmin) is equal to 0.0174 A/m². On the other hand, the current distribution profile increases from blue to red on the porous electrode (in this case simulating a cathode), starting at the edge opposed to the lug. This can be interpreted in terms of a highly heterogeneous current distribution at the electrochemical interface between the porous electrodes and the electrolyte, highly dependent on the location of the lug at the current collector. This example is provided with the purpose to introduce the reader to the models that were developed within the context of this research. However, details on the parameters used and the cases of study are described in the following sections. 5.3 Primary Current Distribution Models This section describes the investigations performed on the primary current distribution in a fuel cell configuration during 0.02 A of discharge, at open circuit conditions. At this point, no parameter related to the microbial dynamics or kinetics is considered, as they are not relevant for this type of distribution. In primary current distribution, the potential losses due to electrode kinetics and mass transport are assumed to be negligible, and ohmic losses govern the current distribution in the cell; thus, primary current distribution study focused on the optimum geometry investigations in order to find the most homogenous distribution profile. 5.3.1 Influence of Current Collector Design The design of the current collector is perhaps the most significant issue for MFC designs, since this highly electrically conductive material (e.g. stainless steel) directly affects the system performance. In addition, current collectors are the most expensive components of an electrochemical cell in terms of material and fabrication costs, in the case of non-precious metal-based electrodes. This is why, it is essential to find a cost-efficient and performanceeffective current collector design. For this purpose, the options of using plate or grid current collector were investigated. For an accurate comparison, all the geometry parameters are kept constant for the models except the current collector design.

51

Figure 5.3: Primary current distribution profiles with plate (left) and grid (right) current collector designs As can be observed in Figure 5.3, the primary current distributions of the two types of current collectors studied include a heterogeneous profile. It is immediately perceptible that the range of current densities rather broad at the porous electrode-electrolyte interface; however, the use of a grid current collector enlarges the region of the porous electrodes that is active at higher current densities. Besides the distribution, higher jmax (12.401 A/m²) is obtained with grid current collector than with plate current collector (3.8764 A/m²). Therefore, it can be safely assumed that utilizing a grid current collector for further models concerning the geometry of study would lead to better performing MFC configurations. 5.3.2 Influence of Lug Dimensioning and Design Lug maintains the external connection between electrodes, in other words, electrons will be transported away from or towards a particular electrode through the respective lugs. For this reason, the contact phase between the lug and the cell is considered to have an important influence on the current distribution profile.

52

Figure 5.4: Primary current distribution profiles with different lug widths (W_lug) When the width of the lug is increased (keeping constant all other system parameters) even though jmax doesn’t differ significantly from one case to the other, it is observed that the distribution over the electrode is highly influenced by this parameter. In Figure 5.4, it is observed that the high current density region is distributed to a larger area in case of study on the right when compared to the one on the left. On the other hand, it can be observed that changing the lug height has no influence on the current distribution over the electrodes as expected (Figure 5.5). This is considered to be due to the high conductivity of the lug material that prevents current to encounter any considerable resistance over the lug.

Figure 5.5: Primary current distribution profiles with different lug heights (H_lug)

53

Finally, the options of using plate or grid lug were also investigated. There was no significant difference in between the two designs (Figure 5.6), neither in jmax nor in current distribution. Although the usage of the grid lug design has no considerable benefit over the output, this configuration was considered in further models for practical concerns since the lug is usually fabricated in one piece with the current collector, and usage of grid current collector design was previously proven advantageous.

Figure 5.6: Primary current distribution profiles with plate (left) and grid (right) lug designs Although it is proven that increasing the W_lug gives better distribution, the usage of lug which has the same width as the grid current collector was not considered due to the possible practical difficulties of maintaining the external connections between the cell and the electrochemical apparatus with a lug that covers the cell entrance, especially, in case of stackcell. 5.3.3 Half Cell Configurations Half cell configurations were also constructed, in order to examine the current distribution profiles at the single-electrode level. For that configuration, one of the electrodes is simply eliminated from the initial geometry and electrolyte domain is considered to be adjacent to an insulating material. Figure 5.7 confirms that the half cell configuration with both current collector designs (plate and grid) have nearly the same distribution profile as in full-cell array, as well as close jmax values over the cathode (See Figure 5.3 and 5.7).

54

Figure 5.7: Primary current distribution profiles for half-cell configurations with plate (left) and (grid) current collector 5.3.4 Influence of Lug Positioning on Different Current Collector Designs In this section, for fuel cell models based on primary current distribution, four different geometries were constructed with the guidance of the output obtained from the previous models. The geometries considering different current collector (cc) and lug configurations are: 1. Plate cc with one lug on each cc 2. Plate cc with two lug on each cc 3. Grid cc with two lug on each cc 4. Grid cc with one lug on each cc It was previously proven advantageous to increase the width of the lug. Here it is indented to observe the effect of using two lugs from cross sides of the cc instead of only one wide lug in order to better distribute the current over the electrode surface. Although, it was proven that the grid cc is more efficient than the plate cc, the study was conducted for both plate and grid cc designs in order to see of the if the influence of lug positioning overcomes the influence of the cc design. The mentioned geometries can be examined in Figure 5.8. The aim of this geometry and configuration study is to select one of the four configurations and perform the further progress and optimizations of the variables directly associated to the physics (current distribution). 55

Figure 5.8: Four geometry configurations with different lug and cc designs After building the geometries, they were simulated once again with the primary current distribution interface. Apart from the cc and lug designs, all other system parameters were kept constant. The results can be examined in Figure 5.9. It is seen that placing two lugs from cross corner on the cc gives a better current distribution (2nd and 3rd configurations); this beneficial effect of two-lug usage is more apparent for the 3rd configuration since it is combined with grid cc design. Although jmax (55.251 A/m2) value obtained from the 4th configuration is nearly as twice as the jmax obtained from 3rd configuration (29.443 A/m2), the preferable case is to obtain the better distribution instead of observing a higher jmax in one corner. Ultimately, the 3rd geometry is regarded as the ideal option as it provides a more homogenous current distribution over the surface. Further optimizations and progress studies were decided to be performed over that geometry.

56

1

2

4

3

Figure 5.9: Primary current distribution profiles over the four geometry 5.3.5 Influence of Different Parameters in Selected Geometry After the selection of the model geometry in the previous section, the subsequent studies were initiated over this model (Figure 5.9-3), hereby called reference model. The geometry and the material parameters of the reference model are listed in the Table 5.1. Table 5.1: Geometry and material parameters of the reference model Parameter Width of the current collector Height of the current collector Depth of the current collector Width of the lug Height of the lug Depth of the lug Width of the electrolyte Height of the electrolyte Depth of the electrolyte Width of each porous electrode frame Height of each porous electrode frame Depth of each porous electrode Number of the porous electrodes in x direction Number of the porous electrodes in y direction Space between the porous electrodes

Symbol W_cc H_cc D_cc W_lug H_lug D_lug W_e H_e D_e W_peld H_peld D_peld N_x N_z s_grid

Value 11 cm 11 cm 0.05 cm 2.788 cm 1.564 cm 0.05 cm 11 cm 11cm 1 cm 1.319 cm 0.4714 cm 0.05 cm 8 21 0.05 cm

57

Conductivity of the current collector Conductivity of the porous electrode Conductivity of the electrolyte

σ_cc σ_peld σ_e

4.8E6 S/m 9500 S/m 1 S/m

The objective here was to investigate the effects of different model parameters and determine the more influencing factors on the output data. This is why; seven key parameters were individually varied, while the rest of the parameters remained constant at characteristic ranges of the actual physical reference model. jmax values obtained from that ranges were recorded (Table 5.2). Table 5.2: Variation of the parameters and jmax

The variation ranges for each parameter were chosen in respect to the possible practical or physical laboratory implementations. For example, the conductivity of the current collector (σ_cc) was varied from 4.8E6 S/m to 9500 S/m in order to investigate the primary current distribution models in case of using less conductive current collector materials. The minimum value of this variation range was selected as 9500 S/m which is equal to the practical conductivity of the porous electrode (σ_peld); in this way, the case of not using conductive current collector was also examined. In order to compare accurately the effect of each parameter, both the variation ranges and j max ranges were normalized between 0 and 1, since the parameters vary in different ranges. Figure 5.10 displays the change of the jmax with alteration in different parameters. The normalization calculations are explained in Appendix 2. 58

Figure 5.10: Normalized value of jmax vs normalized value of parameter The geometry parameters such as electrode spacing (the distance between the cathode and anode, in other words, depth of electrolyte, D_e), lug size (the width of the lug, W_lug) and wire thickness (space between the porous electrodes, s_grid) have significant linear influence on the jmax; with the increase of these parameters jmax linearly decreases. The lug size is the most influencing factor among others since it establishes the region where current starts to be distributed as it was explained before. The electrode spacing has relatively less effect on the performance. Wire thickness is an important characteristic of the grid current collector since, in practice; the grid current collectors are typically made by welding the metallic wires to form a mesh (Figure 5.11). Therefore; when current collector is used together with porous electrode, the mesh openings are filled with the electrode material. Thus, the thickness of the wires that composes the mesh is considered as the space between the porous electrodes and has more influence comparing with other geometry parameters. When wires thickness decreases jmax increases.

59

Figure 5.11: Image of a mesh current collector made of welded wires Among the material conductivity properties, the less influencing one is determined as the conductivity of the (σ_cc) as it seen from the graph above that jmax changes slightly with the large alternation in σ_cc. However, σ_cc becomes more relevant when it approaches to the conductivity of porous electrode (σ_peld). This also points out the importance of the σ_peld comparing to σ_cc. Here, the conductivity of electrolyte (σ_e) doesn’t have a remarkable effect since it has relatively low value than electrode and cc material. This is study is important to decide the effective parameters that should be paid attention for a cost-efficient design. It is discovered that the high conductive cc material usage does significantly not increase the performance, so the cc material can be shifted from stainless steel to a slightly less conductive but also less expensive material. It is also found that decreased wire thickness augments the performance as well as reduces the cost of the cc since less metallic wire is used to fabricate it. It should be noted that in this section, jmax values are recorded in order to compare; however, for individual evaluation of each parameter both jmax and the output image of distribution profile should be taken into account for a global and a more accurate conclusion. 5.3.6 Influence of Grid Size The grid cc was determined as more performing than the plate cc. However, further improvements on design of the grid cc were foreseen. With this objective, smaller and larger frames that are filled with porous electrodes are considered. This means that first W_peld is decreased, and then H_peld is increased starting from the reference model.

60

Figure 5.12: Primary current distribution profiles over the reference model and decreased W_peld Figure 5.12 displays that when the W_peld is decreased from 1.319 to 0.5 cm the current distribution profile does hardly differ. However it is observed in the Figure 5.13 that the increase of H_peld from 0.4714 to 1 cm improves the current distribution profile. Higher jmax magnitudes are obtained at the regions that the current could not reach on previous cases.

Figure 5.13: Primary current distribution profiles of over the reference model and increased H_peld Increased H_peld values can practically be maintained by reducing the cc material in order to extend the mesh openings when fabricating the grid cc. This may result in cost benefits as well because the less material is needed for the construction of a same sized-electrode. With the same approach, since it is seen that W_peld has no noteworthy effect on the performance, 61

larger openings can be considered to the horizontal direction which reduces the cc material usage which is beneficial for the cost without altering the performance. Nonetheless, diminishing the majority of the material would directly impact on the mechanical properties of the electrode. Therefore, an optimization between a higher and homogeneously distributed current as well as a mechanically solid electrode and costs must be addressed. 5.4 Secondary Current Distribution Models This section describes the investigations performed on the secondary current distribution in a microbial fuel cell during at closed circuit conditions. The parameters related to the microbial kinetics that are obtained from the experiments are inserted to the model. The mass transport phenomena is assumed to be negligible, ohmic losses and activation polarization losses govern the current distribution in the cell. The cell geometry that is used for the secondary current distribution models was determined with respect to the primary current distribution modeling results. This is why; the gird current collector and two lug placed on the current collector from cross corners were used. In addition the increased H_peld value was used (H_peld=1 cm) as it was determined to be more performing in terms of distribution profile in the Section 5.3.6. The half-cell configuration was chosen for the secondary distribution profiles as in the experimental case. The other system parameters were also tried to be determined from the experimental set-up (active surface area of the electrodes, porosity of the porous electrodes, etc…) in order to imitate the real cases as much as can be. The bioelectrochemical kinetics obtained from the Section 6.2.2.5 was inserted as it was explained in the Section 6.2.3. The polynomial equation govern from current-potential curve (I vs E) (See Figure 6.21): y = -0.043x4 – 0.001x3 + 0.0107 x² + 0.0048x + 0.001 was inserted to the COMSOL and ffter building the geometry, it was simulated with the secondary current distribution interface. Figure 5.14 displays the secondary current distribution at the outer and inner boundary of the electrode. This current density values of the inner and the outer surfaces are different from each other since the closed circuit condition was used for the secondary distribution models. The result of interest is the inner surface of

62

the electrode since it is facing to the electrolyte; thus it represents the electrode-electrolyte interface in this case.

Figure 5.14: Secondary current distribution profiles over the outer boundary (left) and inner surface (right) of the electrode As can be seen, the secondary current distribution is found to be much more homogenous for the same geometry than the primary current distribution as the kinetics overcomes the geometry influence. This is why, it is highly important to study the primary distribution profiles in order to find the optimal geometry and material configurations before studying the secondary current distribution models in order to see the kinetics effects. The jmax was obtained as 13.25 A/m² (5.14), however; it should be noted that the current density range over the inner surface of the electrode is 6-10 A/m² as can be read from the color scale and visual observation. This is a value is in the same range with the experimental result that was obtained in the Section 6.2.2.5 (7.32 A/m²). 5.5 Summary and Perspectives of the Modeling Work The primary current distribution profiles proved the importance of the cell geometry and the material properties in a fuel cell or microbial fuel cell. This is why the first focus of this study was to the model the primary current distribution in MFCs since it displays the worst-case scenario for the current distribution at the electrode electrolyte interface. It is seen that with numerous variations over these parameters, many different distribution profiles can be obtained. With respect to the existing VITO prototypes, the optimal geometry 63

was investigated for experimental usages. As a result, the grid current collector usage woth increased vertical mesh openings (H=1 cm) was found to be more performing. It was also proven that increased the contact phase between the lug and the cell increases the performance; thus , two lug placed on the current collector from the cross corners can be a better option for MFC geometries. In addition to geometry, material conductivity was found to be an important factor alhtoug its influence is less comparing with geometry parameters. The bioelectrochemical kinetics, which was obtained from the most performing experimental case, was introduced to this optimal geometry and secondary current distribution profile was found for the microbial fuel cell. The selected experimental case was the fumarate-glucose oxidation on a half-cell. Secondary current distribution models showed more homogenous profiles since the electrochemical kinetics is involved over the electrode surface. With the bioelectrochemical kinetics obtained from experimental work, the maximal current density was found as the same range than the experimental value. jmax obtained as 13.25 A/m² and the current density values at the interface was observed in between 6-10 A/m² which are close to the experimental value (7.32 A/m²). It is highly important to remember that the target is for the secondary current distribution modeling is not to obtain the same result with the experimental case. Sometimes simulating an experimental case with computational modeling can be misguided since the environmental factors are not considered in the simulations. However computational modeling and 3-D simulations are highly significant in order to gain an insight opinion and conceptual knowledge. Therefore; the accurate evaluations of the output images and the comparisons between the models constructed with different parameters are more important than the value itself. The future studies will be continued with the tertiary current distribution profiles. Tertiary distribution profiles should be modeled, especially when the gas-diffusion electrode or membranes are involved since mass-transfer plays a massive role when these components are introduced to the models. The introduction of the additional components mentioned above are inevitable for the stackMFCs. Another important future target is to build the stack-MFCs in COMSOL and 64

investigate the optimal configuration for cost-efficient operations so that the path towards scale-up of MFCs can be opened together with stack-cell development. One of the most important characteristics of COMSOL is to combine the different physics interfaces; thus, when electrochemistry is combined with fluid dynamics and mathematics it can be much more effective for modeling tertiary distribution profiles. With the addition of Optimization Module, the obtained models can be optimized.

65

CHAPTER 6: EXPERIMENTAL WORK

The experimental work is performed with the main goal of providing electrochemical kinetics for the modeling work described in the previous chapter (Chapter 5). The data obtained from the experimental kinetics are introduced for the secondary current distribution models as input. Furthermore, the electrochemical activities of the selected bacteria are investigated as well in the context of microbial electrosynthesis (MES) while the experimental kinetics is monitored. MES targets to have value-added product formation in a microbial electrochemical system (MXC) either in electrolysis (MEC) or in fuel cell mode (MFC). The selected bacteria, hereby called X strain due to the confidentiality concerns, are renowned for their capacity for highvalue product synthesis; nonetheless, their ability of electron (e-) transfer between substrate and electrode has not been a subject of deep investigations. In addition, X strain is a type of gram-positive bacteria, which have not particularly shown strong performance for extracellular electron transfer (EET) but have been found to reduce or oxidize redox mediators such as AQDS. Therefore, this chapter examines the electrochemical performance in MXCs of the X strain—which has not been investigated up till now, while supplying information for the modeling study that has been conducted in parallel. 6.1 Materials and Methods 6.1.1 Microbial Growth The X strain was routinely cultured with NBAF medium (Nutrient Broth Acetate Fumarate) containing 35 mM of fumarate as e- acceptor and 14.7 mM of acetate as e- donor. The eacceptor and donor were added from the previously prepared stock solutions to the medium at the appropriate volumes. 1 L of medium that comprises the ingredient listed in Table 6.1, including vitamin, mineral and salt mixtures, was also supplemented with 100 g NaCl and adjusted to pH 9 in order to maintain optimum conditions particularly for X strain to grow.

66

Table 6.1: Components of 1 L of NBAF medium Ingredient 100X NB Salts NB Mineral Elixir DL Vitamins CaCl2.2H2O MgSO4 .7H2O NaHCO3 Na2CO3.H2O Na2SeO4

Amount 10 ml 10 ml 0.75 ml 0.04 g 0.1 g 1.8 g 0.5 g 1.0 ml

The serum bottles containing the medium were flushed with N2 to remove any trace of O2, sealed, and autoclaved. The cultures were incubated (10% inoculum) in triplicate at 30 °C for electrochemical experiments. 1 L of NBAF medium is prepared based on the procedure which is modified from the protocol developed by Derek Lovely for the Geobacter sulfurreducens (Coppi et al., 2001). 6.1.2 Electrochemical Cell Components For the experimentation, half cell configuration which was previously designed at VITO, was used where the as anode or cathode was the working electrode (WE) whereas Ag/AgCl – 3 M KCl (+199 mV vs. SHE) was reference electrode (RE) and a Pt plate served as counter electrode (CE). Activated carbon (AC) (30% porosity) was used as electrode material and supported with stainless steel (SS) grid current collector. The current collectors were chosen to be grid as their significance was previously proven in Chapter 5. Zirfon, an ion permeable separator, was placed in between the WE and CE in order to prevent the interference of gases evolutions (O2 or H2) which can occur at the CE during the electrochemical measurements mode (Pant et al., 2010a). 6.1.3 Experimental Set-up and Operation The half cells were single-chamber cylinder-shaped reactors that were assembled from the components described above. They were operated in a recycled flow batch mode (See Figure 6.1).

67

Figure 6.1: Schematic view of the MXC half cell in recycled flow batch mode (modified from Pant et al., 2010) 1 L of NBAF medium containing X strains (%10 v/v inoculation, 100 ml), that were previously grown as explained in Section 6.1.1, was prepared and circulated continuously from the feed bottle to cell during the process. When preparing NBAF feed solution, fumarate or acetate was not provided to the medium since they were replaced by an electrode as eacceptor (anode) or e- donor (cathode) respectively for X strains to maintain the MET between substrate and electrode. Anthraquinone-2,6-disulphonate (AQDS) was also added to NBAF feed solution to serve as redox mediator. During the operation, medium was fed with substrates (acetate, fumarate or glucose) at the critical moments determined by electrochemical measurement for microbes to continue to carry out the electrochemical activity. The samples were taken from the feed bottle for the pH, conductivity and optical density measurements in order to control the desired operational conditions (pH 9, σ= 145 mS/cm, λ > 0.3) as well as for further analytical measurements in order to determine the product formation. 6.1.4 Electrochemical Methods The electrochemical performance of X strain as biocatalyst oxidation of acetate and reduction of fumarate were investigated. For this reason current evolution in the MXCs was monitored by cronoamperometry (CA) technique. CA measurements were done at constant applied

68

anodic (Eanapp) and cathodic (Ecatapp) potential values that are favorable for the achievement of the oxidation or reduction reactions in the half cells. The standard reduction potentials (Eo) of substrates and mediator but also polarization losses (η) were taken into account for determining applied potentials (See Figure 6.2). CA measurements were initiated at -200 mV (vs Ag/AgCl) for acetate oxidation and -600 mV (vs Ag/AgCl) for fumarate reduction to be achieved.

Figure 6.2: Shematic view of the determination of Eanapp and Ecatapp for acetate oxidation and fumarate reduction during CA Cyclic voltammetry (CV) technique was applied at the specific moments of the operation based which were determined based on CA screening but also whenever the system was intervened, e.g. substrate addition, change in potential. This way, the system was characterized at that particular moment of process and microbial electrochemical kinetics were obtained. CV was done at 3 scan rates for each time (1 mV/s, 10 mV/s, 100 mV/s). For every scan rate, 3 cycles were run in between -700mV and 400 mV vs Ag/AgCl. 6.2 Results and Discussion 6.2.1 Half Cell Experiments with Acetate Anodic activities of X strain were monitored trough CA measurement during 13 days in a cell initially inoculated with acetate (A) and AQDS mediator (M). Until the 5th day of CA, reduction current density values were observed (j0); however; no significant oxidation current was obtained from X strain neither with acetate nor with acetate and glucose combination as substrates. The maximal current density (jmax) was achieved as 0.03 A/m² in 13 days of experimentation. In a conducted with pure X strain (reference cannot be cited due to the confidentiality), the maximal current density obtained was reported as 0.06 A/m2 using AQDS as mediator and only glucose as substrate in a medium containing 5 g/L NaCl. Therefore, the oxidation current obtained in this study with acetate is lower than the value of interest but not negligible when compared to the literature studies.

70

Figure 6.4: CV after inoculation with acetate at initial time (t=0) of CA The CV run at initial time after inoculation with acetate and AQDS (Figure 6.4) shows that the reduction and oxidation peaks are not symmetric which is typical for microbial systems since they are not reversible systems . When Figure 6.4 and 6.5 are compared, it can be distinguished that one of the peaks, that is observed at t=0, later disappears. This can be explained as electrochemical reduction of one electrochemically active species -probably AQDS- since the CA also demonstrates reduction current as well in the beginning.

.

Figure 6.5: CV after glucose addition at t=12 d of CA

71

6.2.2 Half Cell Experiments with Fumarate 6.2.2.1 Glucose addition and change in polarization potential (50 mV) Cathodic activities of X strain were desired to be monitored through CA measurement over 30 days in a cell initially inoculated with fumarate (F) and AQDS mediator (M). Until the 5 th day of CA, reduction current density values were observed (j> Eeq

EeqOx

nF Ean RT

exp

1

nF

E Re d

EeqRe d

nF

E Re d

EeqRe d

RT

Ean,eq

Cathode : E