The Pennsylvania State University. The Graduate School. College of Engineering

Transcription

1 The Pennsylvania State University The Graduate School College of Engineering ENGINEERED INTERFACIAL AND STRUCTURAL POROUS MEDIA ARCHITECTURE FOR POLYMER ELECTROLYTE FUEL CELLS A Dissertation in Mechanical Engineering by Michael P. Manahan, Jr Michael P. Manahan, Jr. Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy May 2013

2 The dissertation of Michael P. Manahan, Jr. was reviewed and approved* by the following: Matthew M. Mench Adjunct Professor of Mechanical Engineering Dissertation Advisor and Co-Chair of Committee Jack Brenizer J. Lee Everett Professor of Mechanical and Nuclear Engineering Co-Chair of Committee Aman Haque Professor of Mechanical Engineering Michael Hickner Assistant Professor of Materials Science and Engineering Karen A. Thole Professor of Mechanical Engineering Head of the Department of Mechanical and Nuclear Engineering *Signatures are on file in the Graduate School ii

3 ABSTRACT The purpose of this work is to explore engineered interfacial and structural architecture in fuel cell diffusion media. Emerging electrochemical storage and conversion devices have been under significant development in the past two decades, and the optimization of the interfacial and structural properties of their components presents a complex, multidisciplinary challenge. Interfaces in electrochemical devices must transport heat, mass, and current in an optimized manor to help sustain the electrochemical reaction and efficient operation. Engineering the interfacial and structural architecture of components in electrochemical devices such as fuel cells can yield significant performance breakthroughs that will make them increasingly economically viable for commercial implementation. This dissertation investigates engineered interfacial and structural architecture in fuel cell materials, namely the catalyst layer (CL) and the diffusion media (DM). Cracks were introduced in the CL in order to modify the CL DM interface. The gas-phase transport and performance of cracked CL was compared to a standard CL. Based on insights gained from the cracked CL investigation, the diffusion media were modified using laser treatment. Perforations were introduced via lasers on samples of virgin DM that contained hydrophobic content. Depending on laser choice, some laser-cut samples displayed a heat affected zone (HAZ) at the catalyst layer microporous layer interface, characterized by a region surrounding each perforation where hydrophobic content was removed. Experimental techniques such as polarization testing, electrochemical impedance spectroscopy, neutron radiography, Tafel slope analysis, and others were used to characterize the impact of the laser-modified DM and give fundamental insight into the iii

4 modified multi-phase transport. A key result of this dissertation proved a 25% increase in power density by optimizing the perforation diameter and the diameter of the heat affected zone. It was shown that the perforation and the heat affected zone uniquely contribute to the performance changes. Namely, the perforation allows for membrane rehydration while the cell is operated with 50% relative humidity inlet gas streams and the heat affected zones redistribute and retain liquid water, leading to improved gas-phase transport. Experiments were conducted to understand the fundamental transport characteristics. iv

5 TABLE OF CONTENTS Abstract... iii Table of Contents... v List of Tables... x List of Figures... xi Acknowledgements... xix 1. Introduction Background on Polymer Electrolyte Membrane Fuel Cells Polymer Electrolyte Fuel Cell Overview PEFC Components PEFC Electrochemical Description Performance Characteristics General Diagnostics Interfaces and Structures in Electrochemical Devices Importance of Interfaces and Structures Literature Survey Motivation and Objective Diagnostic Techniques in PEFCs Importance of Methodology and Error Analysis Polarization Testing Performance Characterization v

6 2.2.2 Tafel Slope Analysis Limiting Current Analysis Gas Modification Neutron Radiography Electrochemical Impedance Spectroscopy Electrochemical Active Area Measurement with Cyclic Voltammetry Double Tafel Plot Analysis References Effects of Surface Irregularities and Interfacial Cracks on Polymer Electrolyte Fuel Cell Performance Introduction Experimental Results and discussion Effect of Crack on Gas Transport Cracked and Negligible-Cracked CL Performance in Wet and Dry Conditions Cracked versus Negligible-Cracked CL Performance Conclusions References Laser Perforated Fuel Cell Diffusion Media: Related Changes in Performance and Water Content Introduction Experimental Methods and Setup vi

7 4.2.1 Laser Perforation In-situ Polarization Testing Neutron Radiography Results and Discussion Polarization Testing at Low Humidity Conditions Neutron Radiography Testing Virgin versus Perforated DM Neutron Radiography Testing High versus Low Current at Low Humidity Operations Polarization Testing at High Humidity Conditions Neutron Radiography Testing Virgin vs. Perforated DM Neutron Radiography Testing High vs. Low Current at High Humidity Condition Conclusions References Laser Perforated Fuel Cell Diffusion Media: Engineered Interfaces for Improved Ionic and Oxygen Transport Introduction Experimental Approach Materials and Laser Modification Polarization Testing Transport Limitation Analysis Electrochemical Impedance Spectroscopy Neutron radiography Results and Discussion vii

8 5.3.1 Testing at 50% Relative Humidity Testing at 120% Relative Humidity Conclusions References Laser Modified Fuel Cell Diffusion Media: Role of the Microporous Layer (MPL) and Heat Affected Zones Introduction Experimental and Modeling Approach Experiment and Setup Laser modification Spectroscopy Electrochemical Testing Net Water Drag Model Development Results and Discussion Spectroscopy Polarization Results at 50% Relative Humidity Electrochemical Impedance Spectroscopy Results at 50% Relative Humidity Double Tafel Slope Results at 50% Relative Humidity Net Water Drag at 50% Relative Humidity Over-humidified Testing Results Model Results Conclusions viii

9 References Concluding Remarks and Future Work Role of Laser-Induced Heat Affected Zones Role of Perforation Future study ix

10 LIST OF TABLES Table 3-1. The experimentally calculated parameters (Eq. (3.3)) divided by the numerical results (Eq. (3.1)) for dry and wet conditions shows the percentage of Fickian diffusion observed Table 3-2. A summary of possible factors involved to explain the behavior of the dry and wet condition performance (Figure 3.5 and Figure 3.6) Table 3-3. A quantitative summary of the performance of 5%-O 2 nitrox dry condition compared to the wet condition in the three regions for both cracked and negligiblecracked CLs Table 4-1. Parametric settings for laser perforations Table 5-1. Laser parameters used to perforate diffusion media Table 5-2. Diffusion media properties used for electrochemical characterization Table 6-1. Material nomenclature and physical effects Table 6-2. Physical material properties Table 7-1. Laser-modified DM with laser type and physical features x

11 LIST OF FIGURES Figure 1.1. An example of the test setup of Sir William Robert Grove s demonstration of the first fuel cell in 1839 [3] Figure 1.2. A general PEFC schematic showing the major components needed for efficient operation [2] Figure 1.3. (a) Vehicular mechanism (diffusion) for proton transport when water content in the membrane is low, and (b) Hopping or Grotthuss mechanism for proton transport when the water content is high in PEFC membranes. Adapted from Ref. [5] Figure 1.4. Scanning electron micrographs of several DM types, including (a) carbon paper, (b) carbon felt, and (c) carbon cloth. Adapted from Ref. [10] Figure 1.5. Typical land-channel design with both in- and through-plane transport required. This design necessitates the need for an intermediate layer between the CL and the flow channel Figure 1.6. Characteristic polarization curve for a PEFC with kinetic, ohmic, and mass transport regions labeled (I, II, III, respectively). Concept adapted from Ref. [2] Figure 1.7. General schematic showing activation energy in chemical/electrochemical reactions. Figure also depicts effect of a catalyst. Adapted from [16] Figure 1.8. Schematic of PEFC components, emphasizing the number of interfaces and structural components xi

12 Figure 1.9. A schematic of the link between structural and interfacial properties and their effect on transport within PEFCs Figure Suggested function of an MPL. a.) The MPL blocks liquid water from condensing near or traveling toward the CL MPL interface. b.) The MPL provides enough thickness for in-plane diffusion around potentially condensed water. From Ref. [42] Figure 2.1. Characteristic polarization curve for a PEFC with kinetic, ohmic, and mass transport regions labeled (I, II, III, respectively) Figure 2.2. Schematics adapted from (a) Ref. [14] and (b) Ref. [35] showing fuel cells in the path of a neutron beam. In these examples, the beam is parallel to the in-plane direction. Other references show the beam parallel to the through-plane direction as well Figure 2.3. Relaxation times of physical processes in PEFCs and the diagnostic techniques plausible for certain time ranges. Adapted from Figure 1 of Ref. [36] Figure 2.4. Diagram from Figure in Ref. [37] indicating the relationship between input AC current at frequency,, and the voltage response. The phase angle,, is the amount that the two responses are separated Figure 2.5. (a) Bode and (b) Nyquist plots of experimental data xii

13 Figure 2.6. (a) A simple equivalent circuit of an electrochemical cell, such as one of the electrodes in a PEFC, and (b) two possibilities for the equivalent of Z f, from Figure 3.1 of Ref. [47] Figure 2.7. Impedance spectra showing R el, R ct, and R d components at high, medium, and low frequency, respectively. Figure 1 from Ref. [44] Figure 2.8. Cyclic voltammograms of platinum in acid, adapted from Ref. [49]. Shaded area is the area used to calculate the ECSA Figure 3.1. SEM images of the surface of the a) cracked CL having crack widths of ~30μm, b) negligible-cracked CL having crack widths of ~5μm Figure 3.2. A schematic of the cross section of the experimental setup Figure 3.3. A comparison of the experimentally determined limiting current densities of a cracked CL (CRK) and a negligible-cracked CL (NCRK) at dry conditions for 5%-O 2 nitrox ( 5% N2 in legend) to 5%-O 2 heliox ( 5% He ) Figure 3.4. A comparison of the experimentally determined limiting current densities of a cracked CL (CRK) and a negligible-cracked CL (NCRK) at wet conditions Figure 3.5. A comparison of the polarization curves of wet and dry conditions for negligible-cracked CL (NCRK) cases. The performance was observed to have three distinct regions in which the dry or wet condition outperformed the other. The onset of Region II was observed began at approximately 0.26 A/cm xiii

14 Figure 3.6. A comparison of the polarization curves of wet and dry conditions for cracked CL (CRK) cases. The performance was observed to have three distinct regions in which the dry or wet condition outperformed the other. The onset of Region II began at approximately 0.2 A/cm Figure 3.7. A polarization curve comparing a cracked (CRK) and negligible-cracked (NCRK) CL under wet conditions Figure 4.1. (a) SEM image of the MPL surface with laser perforations. A heat-affected zone and 300-μm perforations are highlighted as key modifications due to the laser treatment, whereas the MPL cracks occur on both laser-treated and virgin samples; (b) Schematic of laser perforations, where x = 0.67 mm for polarization testing and 0.97 mm for NR testing; (c) An EDS spectrum of MPL surface within the heat affected zone. There is no fluorine peak, indicating the absence of PTFE; (d) An EDS spectrum of the virgin MPL surface, showing dominant carbon and fluorine peaks Figure 4.2. Rendering of the neutron radiography test cell with expanded view of perforated DM schematic and SEM image Figure 4.3. Performance data for 75 o C, 50% relative humidity, nitrox conditions with virgin DM ( ), 15% 300-μm perforated cathode-side DM ( ), and corresponding HFR values for the virgin ( ) and perforated ( ) cases. Perforated DM shows on average 38 mv higher potential in current densities less than 1.4 A cm -2 but failed to achieve steady state at current densities greater than 1.4 A cm xiv

15 Figure 4.4. Neutron data for water mass per volume distributions in the in-plane direction at low inlet relative humidity condition. L represents land, and C represents channel. (a) Water distributions for virgin DM ( ) and perforated DM ( ) under low current density (0.2 A cm -2 ) testing operation; (b) Water distributions for virgin DM ( ) and perforated DM ( ) under high current density (1.7 A cm -2 ) testing operation. A reduced number of data points are shown to improve clarity Figure 4.5. Neutron data for water mass per volume distributions in the through-plane direction at low inlet relative humidity condition. Virgin case at high current ( ), virgin case at low current ( ), perforated DM case at high current ( ), perforated DM case at low current ( ). A reduced number of data points are shown to improve clarity Figure 4.6. Performance data of 75 o C, 120% relative humidity, nitrox conditions for virgin DM case ( ), 15% 300-μm perforated cathode DM case ( ), and corresponding HFR values for the virgin ( ) and perforated ( ) cases. The DM perforations cause drastic performance losses (55 and 183 mv at 0.05 and 0.1 A cm -2, respectively), indicating poor water management in high humidity conditions Figure 4.7. Neutron images from high humidity (100% inlet relative humidity anode and cathode) testing, showing: (a) high current (1.2 A cm -2 ) for virgin DM case; (b) high current for perforated DM case; (c) low current (0.2 A cm -2 ) for virgin DM case; and (d) low current for perforated DM case. In each image, the right-hand-side represents the cathode xv

16 Figure 4.8. Neutron data for water mass per volume distributions in the in-plane direction at high inlet relative humidity condition. L represents land, and C represents channel. (a) Water distributions for virgin DM ( ) and perforated DM ( ) under low current density (0.2 A cm -2 ) testing operation; (b) Water distributions for virgin DM ( ) and perforated DM ( ) under high current density (1.7A cm -2 ) testing operation. A reduced number of data points are shown to improve clarity Figure 4.9. Neutron data for water mass per volume distributions in the through-plane direction at high inlet relative humidity condition. Virgin case at high current ( ), virgin case at low current ( ), perforated DM case at high current ( ), perforated DM case at low current ( ). A reduced number of data points are shown to improve clarity Figure 5.1. Schematic of the cathode side of the (a) virgin DM and (b) laser-treated, perforated DM (not to scale) Figure 5.2. Characteristic energy dispersive spectra for (a) virgin and 100-μm nohaz DM and (b) 300-μm w/haz and portions of 100-μm w/haz DM. The absence of a fluorine peak in DM w/haz samples indicates removal of hydrophobic content by Ybfiber laser treatment. (c) SEM image with HAZ around perforation Figure 5.3. (a) Polarization data at 50% RH. Solid symbols indicate DM with nohaz, open symbols indicate DM w/haz. (b) Corresponding high frequency resistance (HFR) obtained during polarization curves. Selected lines added to show lower HFR values for 100 μm DM (w/haz and nohaz) xvi

17 Figure 5.4. Tafel plots of (a) 100-μm w/haz (b) 100-μm nohaz, both with cathode-side absolute air pressure at 1 atm and 2 atm to vary oxygen concentration. Also included are logarithmic best fit lines in the single-, double-, and high-tafel slope regimes. Tafel slopes are indicated as b near each best fit, and units are in mv dec Figure 5.5. Electrochemical impedance spectra at (a) 0.4 A cm -2, (b) 1.0 A cm -2, (c) 1.4 A cm -2 at 50% RH. Selected frequencies are shown Figure 5.6. Through-plane water content distribution at 50% RH at constant current of 1.7 A cm -2 under steady-state operating conditions Figure 5.7. In-plane water content of a 17.2-cm 2 cell with 300-μm w/haz DM at 50% RH at constant current of 1.7 A cm -2. Periodic peaks (regardless of land-channel configuration) indicate water accumulation, and are the exact spacing of perforations (slightly less than 1 mm center-to-center spacing) Figure 5.8. (a) Polarization data and (b) impedance spectra at 120% RH. Solid symbols indicate DM with nohaz, open symbols indicate DM w/haz Figure 5.9. Schematic of liquid water transport through perforations w/haz and nohaz based on high levels of water condensation in the flow channels due to over-humidified inlet gas streams Figure 6.1. (Color online) (a) Drawing of perforation configuration, indicating the possibility of a HAZ depending on laser properties. (b-f) SEM images of each DM with EDS color mapping. Red (dark) represents carbon, and blue (light) represents PTFE. (b) xvii

18 MPL of a Virgin DM and 0 µm w/haz, (c) macro-dm of 0 µm w/haz, (d) MPL of 100 µm w/haz, (e) MPL of 100 µm without HAZ, and (f) MPL of 300 µm w/haz Figure 6.2. (a) Polarization results and (b) corresponding high frequency resistance plot from cells with inlet relative humidity of 50% on both the anode and cathode side. Both plots have the same legend Figure 6.3. Comparison of voltage differences between virgin DM and the modified DM cases at 50% RH Figure 6.4. Electrochemical impedance spectra at selected current densities Figure 6.5. Tafel plots for (a) 0 µm w/haz, (b) 100 µm w/haz, and (c) 100 µm no- HAZ Figure 6.6. Net water drag coefficient plotted at several current densities at 50% RH conditions Figure 6.7. Over-humidified polarization and HFR results Figure 6.8. Net water drag coefficient plotted against current density at 100% RH conditions Figure 6.9. Maximum temperature at CL vs. perforation diameter Figure 7.1. Scanning electron micrograph with energy dispersive spectroscopy color mapping showing perforations surrounded by heat affected zones (PTFE removed) with unaffected material (carbon and PTFE interspersed) everywhere else xviii

19 ACKNOWLEDGEMENTS I owe thanks to countless people who have helped me, walked with me, carried me, and supported me throughout this process. I have been humbled by the Ph.D. process, seeing that I am so incapable to accomplish all that this work encompasses without the support of those around me. First, I dedicate this dissertation to my Lord and Savior, Jesus Christ. The verse that was a constant theme during my time as a Ph.D. student was Psalm 46:10, Cease striving and know that I am God; I will be exalted among the nations, I will be exalted in the earth. This verse encouraged me so many times that You are over and above and with us through all our struggles. When the experiments were not going well and I was struggling with classwork and other life events, this verse reminded me that I must stop striving and simply recognize that You are sovereign over everything. My job is not to make everything go my way, but rather to see how You are working in every high and every low that I went through and to learn to trust You more. You have created this amazingly complex world, and You have given it to people like me to discover. I pray that the more I research and understand the world you made, the more I will see your masterful design in it all. To my advisor, Matthew Mench: You told me in my initial interview that picking a Ph.D. advisor is like marrying someone because of all the time you would spend with the person. You were right. But for all the times we ve shared together, I am forever grateful. Thank you for your tireless support and help through this process. It has been an absolute pleasure. I am so honored to be your student, and I look forward to years of collaboration in the future. xix

20 To my wife, Lauren: thank you for coming along my side during this process. You have been an invaluable support. Thank you for loving me as I am and for being my best friend. To our soon-to-be-born son, Emerson Peter, know that your dad already loves you so much that he is writing about you before you are born. I cannot wait to meet you. To my parents, Mike and Georgina: you two have been such a support to me. Thank you for being such a great support through the thick and the thin. And to my sisters, Jen and Rikki, thank you for all your continued love and support. You all have been such a support! To my labmates at Penn State: AK, Lijuan we went the long haul together, and I am so glad for that. Marta, Hemant, Tushar, Erinc you guys made the entire journey a lot of fun. To my new labmates at the University of Tennessee: Jason, Alan, Jake, Subhy, Grace. Joining forces with you has been a wonderful experience and you have been a great crowd to work with for the past two years. To friends in ThirdPlace, The Grove, and Fellowship Church: you were a lifeline to my spiritual grown and to keeping me sane. Thank you for your deep friendships, for your prayers, and for your constant encouragement. To all my teachers and professors who instilled in me a desire to learn and a desire to be curious. You will never know the impact you ve had on my life. xx

21 1. INTRODUCTION 1.1. BACKGROUND ON POLYMER ELECTROLYTE MEMBRANE FUEL CELLS Polymer electrolyte fuel cells (PEFCs), also referred to as polymer electrolyte membrane fuel cells (PEMFCs), are electrochemical devices that convert the chemical energy of a fuel and oxidizer into electrical energy through electrochemical reactions. The first known experiment of the hydrogen fuel cell was done in 1839 by Sir William Grove [1]. It was with a simple experiment like that shown in Figure 1.1 that he demonstrated the possibility of utilizing hydrogen and oxygen to generate current, essentially the reverse of the electrolysis of water. This experiment consisted of separate platinum electrodes submerged in a dilute sulfuric acid electrolyte solution with hydrogen and oxygen gas in separate compartments. However, his experiments yielded extremely low currents due to the low active surface area and high electrolyte resistance, and this discouraged significant development for the next century. In the late 1950s, the space race between the Soviet Union and the United States spurred significant development of fuel cells for auxiliary power on the space crafts. Low temperature PEFCs were first invented at General Electric in 1955 and were used for NASA s Gemini space program [2]. Still, high cost, low power output, and short operational lifetime inhibited its pervasive use for these applications. 1

22 Figure 1.1. An example of the test setup of Sir William Robert Grove s demonstration of the first fuel cell in 1839 [3]. In the past half-century, fuel cells of all types have been under significant development for auxiliary power. United Technologies Company (UTC) built the first commercially available phosphoric acid fuel cell (PAFC) that came about amid rising energy costs [4]. A development at Los Alamos National Laboratory in the mid-1980s allowed for an order-of-magnitude reduction in noble metal catalyst loading, greatly reducing the cost of the fuel cell [2, 3]. This advance, coupled with an increasing concern for environmental sustainability and national energy independence, has led to the continuing interest in developing fuel cells to power the future of our world. PEFCs, in particular, currently offer the most promising source of portable energy, especially for the automotive industry. 2

23 In modern PEFCs, two compartments the anode and the cathode are separated by a thin, polymer membrane that is electronically insulating (preventing an immediate short circuit) yet ionically conductive (allowing protons to be transferred from the anode to the cathode). Ideally, the membrane is impermeable to crossover of reactant species. If there is any specie crossover, then would reduce the concentration gradient and reduce the open circuit potential of the cell according to the Nernst equation. More on the components of PEFCs will be discussed in Section POLYMER ELECTROLYTE FUEL CELL OVERVIEW PEFC COMPONENTS A schematic of typical PEFC components is shown in Figure 1.2, and each component will be briefly discussed here. The polymer electrolyte membrane is at the center of each unit cell, and it separates the anode from the cathode. The membrane is typically between μm thick, and has multiple functions, including: i) a physical barrier to prevent reactant crossover, ii) conducting protons from the anode to the cathode, and iii) an electronic insulator to prevent a short circuit. Most modern membranes have a general structure like Nafion (DuPont), which is a copolymer between polytetrafluoroethylene (PTFE) and polysulfonyl fluoride vinyl ether [5]. So, the membrane consists of both a hydrophilic and ionically conductive phase (sulfonic acid groups) and a hydrophobic and mostly inert polymer (PTFE) that provides support and chemical stability. In PEM fuel cells, the hydrogen ions are transported across the membrane via hopping and vehicle mechanisms, as described by Kornyshev et al. [6] and Weber and Newman [7]. Membrane hydration is required for high proton conductivity 3

24 through the membrane. When water content is low, the hydrophilic regions of the membrane are fairly isolated, and the vehicular mechanism (diffusion-based) dominates the proton transport from the anode to the cathode. When water content is higher, however, the hydrophilic ionomer structure is significantly more connected, and the protons hop from one H 2 O molecule to another. This is often referred to as the hopping mechanism, or the Grotthuss mechanism. A schematic of these two mechanisms is shown in Figure 1.3, adapted from Ref. [5]. Figure 1.2. A general PEFC schematic showing the major components needed for efficient operation [2]. 4

25 (a) (b) Figure 1.3. (a) Vehicular mechanism (diffusion) for proton transport when water content in the membrane is low, and (b) Hopping or Grotthuss mechanism for proton transport when the water content is high in PEFC membranes. Adapted from Ref. [5]. Adjacent to the membrane is the catalyst layer (CL), which consists of platinum nanoparticles (typically 2-4 nm diameter) that are supported by carbon particles (~40 nm diameter). There is ionomer, typically of the same material of the membrane, that is mixed within this layer such that both electrons and protons can be effectively transported through this layer. The CL itself is typically from 5-30 μm thick. Often the combination of the membrane and the catalyst layer is referred to as the membrane electrode assembly (MEA) and is commercially available. Since hydrogen is diffuse and because the hydrogen oxidation reaction (HOR) is facile, the platinum loading (typically 5

26 expressed in mg Pt cm -2 ) is often lower on the anode than on the cathode in commercial applications. A porous substrate, often referred to as the diffusion medium (DM) or gas diffusion layer (GDL), is adjacent to the CL. It serves multiple functions [2, 8], including: i) inand thru-plane gas-phase reactant access from the flow plate (conventionally a landchannel design) to the CL; ii) multi-phase transport of water to and/or from the CL; iii) electrical contact between the flow field/bipolar plate and the CL; iv) rigid support of the membrane and CL, especially in conventional land-channel designs where local stresses can be high; v) heat transport to and from the CL, depending on the existing temperature gradients in the cell. A very common addition to the DM is a micro-porous layer (MPL). The inclusion of an MPL between the macro-dm and CL in PEFCs was likely first reported by Wilson et al. [9] in 1992 but is now widely used and studied. State-of-the-art bi-layered diffusion media consist of a macro-dm with pore diameters on the order of μm and an MPL with pore diameters in the μm range. In general, the MPL provides a smoother contact between DM and CL, reducing the electrical resistance between these layers, while serving as a protective layer to prevent the coarse DM fibers from puncturing through the CL and membrane. It serves several other functions which will be discussed at length in future sections. Scanning electron micrographs of several macro-dm types are shown in Figure 1.4, adapted from Ref. [10]. 6

27 (a) (b) (c) Figure 1.4. Scanning electron micrographs of several DM types, including (a) carbon paper, (b) carbon felt, and (c) carbon cloth. Adapted from Ref. [10]. The next component is the flow field, which, depending on the cell/stack design, can also act as the bipolar plate and current collector. In the schematic shown (Figure 1.2) and in most small-scale experimental setups, the flow field is not used for anything more than a flow field. In conventional designs, the flow field is in a land/channel configuration, where the gases flow through the channel and the current is conducted through the lands. While the lands are needed to conduct electricity, they create dead zones for the reactant gas. This is one reason the DM is required, so that in-plane 7

28 Through-plane direction diffusion can occur to create a more uniform reactant distribution at the DM CL interface. New designs have been emerging in the past decade that use metallic foam, which is essentially a porous metal that the reactant gases flow through [11-15]. This effectively eliminates the need for in-plane diffusion (no land/channel design), but other factors such as increased propensity for dryout can lead to performance issues with this design. In-plane direction Land Reactant channel Land DM Catalyst Layer Membrane MPL Land-Channel Virgin DM Design Figure 1.5. Typical land-channel design with both in- and through-plane transport required. This design necessitates the need for an intermediate layer between the CL and the flow channel. Additional hardware is required, such as manifolds, lead lines, humidifiers, coolant/heaters, blowers, etc., but discussion of these components will not be a focus here. 8

29 PEFC ELECTROCHEMICAL DESCRIPTION Knowing the configuration described in Section allows for fundamental understanding of the basic operating characteristics. Hydrogen is supplied to the anode side through the gas channels, and an oxidant (e.g., air, pure oxygen, or other oxygen mixtures) is supplied to the cathode side. Typically, the gases are humidified to provide the required moisture to the membrane and CL to allow for efficient proton transport. Oxidation occurs on a platinum site on the anode, splitting the diatomic hydrogen molecule into hydrogen ions (protons), H +, and electrons, e - ( (1.1). The protons are conducted through the ionomer in the CL and then through the membrane toward the cathode side. The electrons are conducted through the carbon of the CL and DM to an external circuit (where useful work can be used) and to the cathode side. Then, on the cathode side, the oxygen, proton and electron meet at a triple phase boundary where ionomer, carbon, and platinum all exist to form water (Eq. (1.2). The overall cell reaction is listed in Eq. (1.3). Anode: H - 2 2H 2e (1.1) Cathode: 1 - O2 2H 2e 2H 2 2 O (1.2) Overall: 1 O2 H O (1.3) 2 H PERFORMANCE CHARACTERISTICS The quintessential plot to determine the overall performance characteristics of PEFCs is the polarization curve, abbreviated the pol curve. A typical pol curve is shown in 9

30 Figure 1.6. The generalized regions of the pol curve are indicated on Figure 1.6, namely: Region I activation polarization, Region II ohmic polarization, and Region III concentration polarization. Fuel cell text books such as in Refs. [2, 3] go into useful detail, but here an overview will be presented. Figure 1.6. Characteristic polarization curve for a PEFC with kinetic, ohmic, and mass transport regions labeled (I, II, III, respectively). Concept adapted from Ref. [2]. Region I in Figure 1.6 is referred to as the activation polarization region because the most dominant factor in the performance characteristics is a result of the kinetic polarization at each electrode. That is, activation polarization is the voltage overpotential required to overcome the activation energy of the electrochemical reaction at each electrode. Figure 1.7 depicts a generic chemical/electrochemical reaction process, and indicates the activation energy required to take reactants and turn them into products. 10

31 Each electrode has an activation energy associated with it. Activation losses are influenced by many factors, including the reaction mechanism, catalyst type, CL morphology, operating conditions, impurities or poisons, species concentrations, and more. Activation polarization can be described by the Butler-Volmer equation for electrodes, and is given by: i i o af cf exp act exp act (1.4) RuT RuT where i represents the current density (A cm -2 ), i o is the exchange current density (A cm -2 ), a and c are the respective anodic and cathodic charge transfer coefficients, F is Faraday s constant, R u is the universal gas constant, T is the temperature, and act is the activation overpotential. As was previously mentioned the hydrogen oxidation reaction (HOR) has much lower activation polarization compared to the oxygen reduction reaction (ORR). Typically, simplifications can be made for the HOR assuming linearized kinetics, and to the ORR, assuming the anodic and cathodic charge transfer coefficients are equal ( a ), as follows: c i RuT act, a (1.5) i nf o act, c RuT i sinh 1 F 2i o (1.6) 11

32 act Figure 1.7. General schematic showing activation energy in chemical/electrochemical reactions. Figure also depicts effect of a catalyst. Adapted from [16]. Region II in Figure 1.6, known as the ohmic polarization region, is characterized by a linear voltage drop with increasing current. Simply, ohmic iar ohmic (1.7) where A is the cell active planform area, and R ohmic is the ohmic resistance within the cell. Typically, R ohmic is measured using high frequency methods, and includes bulk, contact, and proton transport resistances. High frequency resistance (HFR) will be discussed in detail in Section In many applications, the ohmic resistance is dominated by proton transfer through the membrane. In certain cases, experiments require different relative humidity of the inlet gases for different experiments. If one 12

33 wishes to understand the difference in performance observed between difference relative humidity cases, but wishes to eliminate the contribution of the ohmic resistance (especially due to membrane hydration variation), the voltage can be compensated for the ohmic losses by adding ohmic to the measured voltage. Any different in the IRcompensated polarization curves will be mostly due to activation or concentration polarization difference. Finally, Region III in Figure 1.6 is the concentration polarization region, also known as the mass-transport region. Region III is characterized by a relatively rapid drop in voltage for small increases in current at high current values. This sharp drop in performance is due to concentration polarization, which is the reduction of reactant surface concentration at the triple phase boundary, thus reducing the sustainability of the electrochemical reaction. The onset of the mass-transport region can be caused by numerous factors, and some of the most common are: i) Gas-phase diffusion. Gas-phase diffusion limitations may exist in any component of the PEFC, but is most often limited in the cathode CL. The CL has a submicron pore size which results in Knudsen diffusion, and the oxygen often must diffuse through a thin film of ionomer to arrive at the triple phase boundary. Concentration limitations can also exist in the DM or flow channels depending on the operating conditions. Additionally, when oxygen is consumed on the cathode side, the inert gas (e.g., nitrogen when air is used) remains. When oxygen consumption increases, this may tend to leave a blanket of nitrogen, which will 13

34 inhibit the transport of oxygen, and therefore reduce the performance of the fuel cell. ii) Reactant blockage due to flooding. Flooding is a term used to describe the condition when liquid water blocks a significant portion of the available transport pathways for reactant gas to arrive at the reaction sites. Flooding can happen in the flow channels, DM, CL, or at the interfaces of these layers. Flooding usually occurs at high current when water generation in the cell is high. iii) Impurity-occupied reaction site. Every platinum particle has a certain number of available reaction sites. When impurities (e.g., carbon monoxide, sulfur) occupy these sites, this limits the number of reactants that can be reduced or oxidized on the surface, and therefore causes a decrease in performance. iv) Non-gas-phase diffusion effect. As mentioned, oxygen often must be diffused through a thin film of ionomer. Under standard operating conditions, this is the most common concentration polarization. The concentration polarization is commonly expressed by Eq. (1.8), conc R ut ln 1 nf i il (1.8) Both the anode and the cathode have polarization losses and their own limiting current. Experimentally, it is impossible to run the fuel cell to a true limiting current due to the onset of hydrogen evolution on the cathode side below about 100 mv [17]. However, 14

35 as suggested in Figure 1.6, the limiting current can be reasonably extrapolated from a measurement made between 100 and 200 mv. In summary, the pol curve is a summation of multiple sources of polarizations, and the following equation can be used to easily see each of these components: E cell o E ( T, P) act, a act, c ohmic conc, a conc, c x (1.9) where x is the crossover loss due to the small gas-phase permeability of the membrane or imperfect gasket materials GENERAL DIAGNOSTICS The pol curve is a very useful plot to gain an overall understanding of a cell s performance characteristics. It is mostly comparative in nature, however. For example, comparing the pol curves of two cells with different CLs, for example, may lend insight into the performance characteristics of the CLs. Most commonly, the pol curve can be used to tell the researcher where to investigate observed phenomena using other diagnostic techniques. Beyond the pol curve, there are a plethora of experimental techniques to delve into the fundamentals of each region of the pol curve. These will be discussed in detail in Chapter 2 and their application to the present thesis proposal will be shown in subsequent chapters as well. 15

36 1.3. INTERFACES AND STRUCTURES IN ELECTROCHEMICAL DEVICES IMPORTANCE OF INTERFACES AND STRUCTURES Every electrochemical device has interfaces. Figure 1.8 shows a schematic layout of a typical PEFC, with the current collectors (CC), flow plate (FP), GDL, MPL, CL, and PEM. One can see numerous interfaces, including the CC FP, FP GDL, GDL MPL, MPL CL, and CL PEM on both the anode and cathode sides. The GDL, MPL, and CL are all porous materials with tunable properties to optimize performance and interaction with adjacent layers. With multi-phase transport occurring in each of these components, their design cannot be underestimated. A literature survey will be given in Section as interfaces and structure applies to the proposed thesis. Figure 1.8. Schematic of PEFC components, emphasizing the number of interfaces and structural components. Interfaces and structure are closely linked. Figure 1.9 shows a schematic of how interfaces can affect PEFC performance. It also implies that the underlying structure of the two mating surface has a direct impact on the interfacial behavior. In terms of specific PEFC components, for example, the structure of the MPL and the CL has a direct impact on how multi-phase reactants and products will be transported across the interface 16

37 between them. With this close link, it will be important to have an understanding of the properties of these layers, and apply this knowledge to their interfacial interaction Flat Flat Interface Mating Rough Rough Interface Mismatching Interface Different Pore Size / Roughness Interface Worst Case Scenario: Gas cannot diffuse through interfacial area due to water-filled gap. Heat and current is disrupted. Best Case Scenario: Mass, heat, and current can easily transport across porous interface. Figure 1.9. A schematic of the link between structural and interfacial properties and their effect on transport within PEFCs LITERATURE SURVEY In this section, a review on the pertinent literature related to the interfaces and structures of PEFC components is given. For PEFCs, the DM (which includes the MPL in PEFCs) and the CL are of particular interest since they are two layers that significantly affect the electronic, heat, and multi-phase mass transport. In PEFCs, the DM is a porous material that is compressed between the flow field and the CL in conventional cell designs. Diffusion media serve numerous, well-established functions, discussed in Section The design of the DM is highly dependent on the operating conditions, application requirements, and cell design. For example, the need for 17

38 in-plane diffusivity changes drastically if there is a land/channel flow plate or a metal foam flow plate. While commercial DM widely available from multiple companies (e.g., SGL Carbon Group, Toray, MRC, Gore), these are off-the-shelf products that may be insufficiently optimized for certain applications. Before discussing more on the effect of the structure and interface of DM on PEFC transport characteristics, two more important aspects must be addressed, i.e., hydrophobic content and the MPL. Most PEFC DM are loaded with hydrophobic content, typically polytetrafluoroethylene (PTFE), commonly to a value between 5-30 wt% [8, 18-20]. The primary role of the PTFE content is to prevent flooding, which is the accumulation of liquid water in the fuel cell to the extent of blocking reactant gas from the reaction sites. The PTFE can aid in the removal of liquid water from the DM and flow channels, and can prevent excessive saturation in the bulk DM. PTFE in the CL, MPL, and DM has been shown to have positive effects on fuel cell performance, and several optimization studies have been conducted [19, 21-23]. The addition of PTFE was shown to create a mixture of hydrophilic and hydrophobic pores [23-27]. Studies analyzed the effect of PTFE loading in the fuel cell DM [19, 20, 22, 24, 28-30] and methods to tailor the ratio of hydrophilic and hydrophobic pores were suggested [31]. Hydrophilic pores are theoretically desirable to aid in wicking water away from the CL reaction sites, thus minimizing flooding and reducing thin film generation; however, they can also act as water condensation and storage locations that promote high DM saturation, which could undesirably foster flooding conditions. Hydrophobic pores, on the other hand, have been shown to reduce the DM saturation [32-34]; too many hydrophobic pores, however, can create a strongly hydrophobic layer, impeding necessary liquid permeation [18, 22] or 18

39 causing pore electrical contact along the MPL CL interface. Additionally, excessive PTFE content in the DM can impede evaporative removal of liquid water at shut down [10, 35]. Therefore, it can be qualitatively surmised that a mixture of hydrophilic and hydrophobic pores is necessary for optimal water transport characteristics with the fuel cell. The micro-porous layer (MPL), which was adapted from phosphoric acid fuel cells, is placed between the macro-dm and CL in PEFCs and was likely first reported by Wilson et al. [9] in 1992 but is now widely used and studied. State-of-the-art bi-layered diffusion media consist of a macro-dm with pore diameters on the order of μm and an MPL with pore diameters in the μm range. The addition of a MPL between the CL DM interface has been shown to significantly improve cell performance [8, 24, 33, 34, 36-38]. The MPL increases the surface area contact for improved electrical transfer between the DM and the CL and provides a protective layer to prevent the coarse DM fibers from puncturing through the membrane. Some studies have indicated that the MPL can act as a capillary barrier to inhibit liquid water from entering the cathode-side DM from the CL and consequently forces water to move toward the anode side, referred to as back diffusion [21, 39-41]. Other studies suggest that the cathode MPL prevents the condensed water, which has aggregated in the larger pores of the macro-porous DM, from moving toward the membrane and forming a thin film at the CL, causing the undesirable flooding effect [33, 38, 40-43]. A schematic of the latter-suggested function of the MPL is show in Figure 1.10 from Ref. [42]. The inclusion of interfacial and thermal effects is also critical to consider when analyzing these effects [43-60]. Electroosmotic drag moves water from the anode to the cathode side of the membrane [61-63]. Water concentration can 19

40 also move water back to the anode compartment via back diffusion. Pressure gradients and thermal gradients can also affect water movement across the membrane in either direction [42, 52, 64-67]. The net direction of water transport across the membrane is referred to as the net water drag (NWD), defined as the net amount of water molecules transferred across the membrane per proton transferred from anode to cathode. The NWD is therefore sensitive to operating conditions such as gas flow rates and humidification and material properties like membrane thickness, porosity of the DM and MPL, and PTFE content [68-71]. Research that used NWD measurement to understand the effect of the MPL on water transport concluded that the NWD for cells with an MPL at the cathode compared to cells without an MPL were not statistically significant, although they noted that the presence of MPL significantly improved fuel cell performance [36, 72]. Indeed, operating conditions and material properties significantly affect the results of this topic, and some of these studies suggest a combination of these mechanisms, which is more likely. While this capillary barrier may exist, recent studies revealed the existence of deep cracks and other surface defects on the MPL surface [33, 49, 50, 73], which are argued to serve as preferential pathways for water removal from the cathode CL towards the MPL DM interface [33, 73]. The interface between the MPL and CL is critical to avoid delamination and degradation during freeze/thaw testing [54, 55]. 20

41 Figure Suggested function of an MPL. a.) The MPL blocks liquid water from condensing near or traveling toward the CL MPL interface. b.) The MPL provides enough thickness for in-plane diffusion around potentially condensed water. From Ref. [42]. Numerous studies have investigated interfacial and structural effects of the DM on PEFC performance. Kim et al. [52-55] investigated the effects of DM stiffness, thickness, and presence/absence of the MPL. Even structural properties like stiffness and thickness have interfacial impacts. They conclude that a stiffer DM results in more uniform compression pressure transmitted to the CL DM interface, which aids in mitigating freeze 21

42 damage. They also determined the MPL assists in limiting residual water at shutdown. Several groups (Turhan et al. [44, 45], Chen et al. [56], Sinha and Wang [57]) studied the interface between the flow field and the DM (FF DM interface). They found a strong relationship between the channel hydrophobicity and the through-plane liquid water distribution, transport, and removal. In particular, a hydrophilic channel was shown to remove liquid water from under the lands, which aided in mitigating flooding behavior and residual water after shutdown. Bajpai et al. [49] investigated the effects of the CL MPL interface on the cell performance via a modeling study based on ex-situ, experimental morphology data from Swamy et al. [50]; they found a 57 mv drop at 1 A cm -2 as a result of the interfacial effects. Khandelwal et al. [46, 47] also noted the importance of interfacial effects in their model development. Several recent studies have aimed to apply the fundamental knowledge obtained from the works previously mentioned. Manahan et al. [43, 74, 75], Gerteisen et al. [76, 77], Markötter et al. [78], Alink et al. [79], and Kimball et al. [80] have investigated the introduction of laser perforations onto the cathode-side DM. By laser perforating the cathode-side DM, Manahan et al. argued they had altered the interface between the CL and the MPL, as well as the structure of the DM, by implementing 100-μm and 300-μm diameter perforations equally spaced across the DM. They investigated the effect on performance and water content and found that the perforations increase the cell voltage by 6% at moderate current densities and increased the maximum power density by 25% compared to unaltered materials when operated at 50% relative humidity. When operated with over-humidified inlet gases (120% relative humidity), the perforations resulted in mixed performance, improving performance at low to moderate current densities but 22

43 showing increased mass transport losses at high current densities. Neutron radiography (water content) results indicated the perforations acted as water pooling locations, which may have led to excessive flooding in the high humidity conditions. Studies by Gerteisen et al. show that perforations may increase the stability of stack performance and increase the limiting current density; Markötter et al. and Alink et al. used synchrotron x-ray radiography to confirm the finding that Manahan et al. showed, namely that perforations can act as water pooling and/or transport channels and improve the water uptake of the membrane. Kitahara et al. [59, 81, 82] and Blanco et al. [60] have displayed improved performance with low-humidity inlet streams by adding layers to the DM which optimize liquid water retention in the cathode in order to maintain membrane hydration and improve durability. Kitahara et al. introduced a hydrophilic layer on top of an alreadyexisting hydrophobic MPL, and they observed an increase in performance in low humidity conditions. Blanco et al. have shown that the use of a perforated metallic layer on the cathode side (added to traditional DM) led to greater water retention, which boosts performance and durability in dry conditions. Other studies are investigating other forms of DM modifications, including metallic DM [83], and metallic foam [14, 15, 84]. In summary, there are a growing number of publications suggesting that the bulk diffusion media, MPL, and/or the CL MPL interfacial structure is lacking significant optimization and understanding. The performance benefits shown in the previously mentioned publications are clear evidence that further research in structural and interfacial modification can result in further significant improvements. 23

44 MOTIVATION AND OBJECTIVE Section gives the background and importance to interfacial and structural modifications on the overall performance in electrochemical systems. It has become evident that the commercially available materials for the CL and DM in PEFCs are not optimized for particular operating conditions, and significantly more research should be dedicated to their development in order to have PEFCs penetrate the market. Current state-of-the-art materials for PEFCs are not yet optimized, and their interfacial and structural properties should be investigated. In this study, interfacial and structural properties of PEFC DM will be studied to determine specific design considerations and underlying phenomena for future designs of these system components. 24

45 REFERENCES [1] W. R. Grove, "On Voltaic Series and Combination of Gases by Platinum," Philosophical Magazine, vol. 14, pp , [2] M. M. Mench, Fuel Cell Engines. Hoboken: John Wiley & Sons, Inc., [3] J. Larminie and A. Dicks, Fuel Cell Systems Explained, Second ed. West Sussex: Wiley, [4] M. L. Perry and T. F. Fuller, "A Historical Perspective of Fuel Cell Technology in the 20th Century," Journal of The Electrochemical Society, vol. 149, pp. S59-S67, [5] A. Z. Weber and J. Newman, "Transport in Polymer-Electrolyte Membranes, I. Physical Model," Journal of the Electrochemical Society, vol. 150, pp. A1008- A1015, [6] A. A. Kornyshev, A. M. Kuznetsov, E. Spohr, and J. Ulstrup, "Kinetics of Proton Transport in Water," The Journal of Physical Chemistry B, vol. 107, pp , 03/26/ [7] A. Z. Weber and J. Newman, "Transport in Polymer-Electrolyte Membranes, II. Mathematical Model," Journal of The Electrochemical Society, vol. 151, pp. A311-A325, [8] M. Mathias, J. Roth, J. Fleming, and W. Lehnert, "Diffusion Media Materials and Characterisation," in Handbook of Fuel Cells: Fundamentals, Technology and Applications. vol. Vol. 3, W. Vielstich, A. Lamm, and H. A. Gasteiger, Eds., ed Hoboken: John Wiley & Sons, 2003, pp

46 [9] M. S. Wilson and S. Gottesfeld, "Thin-Film Catalyst Layers for Polymer Electrolyte Fuel Cell Electrodes," Journal of Applied Electrochemistry, vol. 22, pp. 1-7, [10] K. T. Cho and M. M. Mench, "Effect of Material Properties on Evaporative Water Removal from Polymer Electrolyte Fuel Cell Diffusion Media," Journal of Power Sources, vol. 195, pp , [11] A. Kumar and R. G. Reddy, "Modeling of Polymer Electrolyte Membrane Fuel Cell with Metal Foam in the Flow-Field of the Bipolar/End Plates," Journal of Power Sources, vol. 114, pp , [12] A. Kumar and R. G. Reddy, "Materials and Design Development for Bipolar/End Plates in Fuel Cells," Journal of Power Sources, vol. 129, pp , [13] O. J. Murphy, A. Cisar, and E. Clarke, "Low-Cost Light Weight High Power Density PEM Fuel Cell Stack," Electrochimica Acta, vol. 43, pp , [14] A. K. Srouji, L. J. Zheng, R. Dross, A. Turhan, and M. M. Mench, "Performance and Mass Transport in Open Metallic Element Architecture Fuel Cells at Ultra- High Current Density," Journal of Power Sources, vol. 218, pp , [15] L. J. Zheng, A. K. Srouji, R. Dross, A. Turhan, and M. M. Mench, "Computational Engineering of Porous Flow Field PEFCs to Enable High Temperature High Power Density Operation," Journal of the Electrochemical Society, vol. 160, pp. F119-F128, January 1, [16] J. Brooks. (2011). Activation Energy Plot. Available: 26

47 [17] D. R. Baker, D. A. Caulk, K. C. Neyerlin, and M. W. Murphy, "Measurement of Oxygen Transport Resistance in PEM Fuel Cells by Limiting Current Methods," Journal of The Electrochemical Society, vol. 156, pp. B991-B1003, [18] G.-G. Park, Y.-J. Sohn, T.-H. Yang, Y.-G. Yoon, W.-Y. Lee, and C.-S. Kim, "Effect of PTFE Contents in the Gas Diffusion Media on the Performance of PEMFC," Journal of Power Sources, vol. 131, pp , [19] S. Park, J.-W. Lee, and B. N. Popov, "Effect of PTFE Content in Microporous Layer on Water Management in PEM Fuel Cells," Journal of Power Sources, vol. 177, pp , [20] L. Giorgi, E. Antolini, A. Pozio, and E. Passalacqua, "Influence of the PTFE Content in the Diffusion Layer of Low-Pt Loading Electrodes for Polymer Electrolyte Fuel Cells," Electrochimica Acta, vol. 43, pp , [21] W. Dai, H. Wang, X.-Z. Yuan, J. Martin, J. Shen, M. Pan, and Z. Luo, "Measurement of Water Transport Rates across the Gas Diffusion Layer in a Proton Exchange Membrane Fuel Cell, and the Influence of Polytetrafluoroethylene Content and Micro-porous Layer," Journal of Power Sources, vol. 188, pp , [22] S. Park and B. N. Popov, "Effect of Cathode GDL Characteristics on Mass Transport in PEM Fuel Cells," Fuel, vol. 88, pp , [23] E. C. Kumbur, K. V. Sharp, and M. M. Mench, "Validated Leverett Approach for Multiphase Flow in PEFC Diffusion Media. Part I: Hydrophobicity Effect," Journal of The Electrochemical Society, vol. 154, pp. B1295-B1304,

48 [24] M. Mathias, J. Roth, J. Fleming, and W. Lehnert, "Beginning-of-Life MEA Performance--Efficiency Loss Contributions," in Handbook of Fuel Cells: Fundamentals, Technology and Application. vol. Vol. 3, W. Vielstich, A. Lamm, and H. A. Gasteiger, Eds., ed: John Wiley & Sons, 2003, pp [25] E. C. Kumbur, K. V. Sharp, and M. M. Mench, "Validated Leverett Approach for Multiphase Flow in PEFC Diffusion Media. Part II: Compression Effect," Journal of The Electrochemical Society, vol. 154, pp. B1305-B1314, [26] E. C. Kumbur, K. V. Sharp, and M. M. Mench, "Validated Leverett Approach for Multiphase Flow in PEFC Diffusion Media. Part III: Temperature Effect and Unified Approach," Journal of The Electrochemical Society, vol. 154, pp. B1315- B1324, [27] J. T. Gostick, M. W. Fowler, M. A. Ioannidis, M. D. Pritzker, Y. M. Volfkovich, and A. Sakars, "Capillary Pressure and Hydrophilic Porosity in Gas Diffusion Layers for Polymer Electrolyte Fuel Cells," Journal of Power Sources, vol. 156, pp , [28] Z. Qi and A. Kaufman, "Improvement of Water Management by a Microporous Sublayer for PEM Fuel Cells," Journal of Power Sources, vol. 109, pp , [29] F. Lufrano, E. Passalacqua, G. Squadrito, A. Patti, and L. Giorgi, "Improvement in the Diffusion Characteristics of Low Pt-loaded Electrodes for PEFCs," Journal of Applied Electrochemistry, vol. 29, pp ,

49 [30] G. Velayutham, J. Kaushik, N. Rajalakshmi, and K. S. Dhathathreyan, "Effect of PTFE Content in Gas Diffusion Media and Microlayer on the Performance of PEMFC Tested under Ambient Pressure," Fuel Cells, vol. 7, pp , [31] R. R. Fredley, "Porous Support Layer for an Electrochemical Cell," USA Patent, [32] J. Benziger, J. Nehlsen, D. Blackwell, T. Brennan, and J. Itescu, "Water Flow in the Gas Diffusion Layer of PEM Fuel Cells," Journal of Membrane Science, vol. 261, pp , [33] J. T. Gostick, M. A. Ioannidis, M. W. Fowler, and M. D. Pritzker, "On the Role of the Microporous Layer in PEMFC Operation," Electrochemistry Communications, vol. 11, pp , [34] J. H. Nam and M. Kaviany, "Effective Diffusivity and Water-saturation Distribution in Single- and Two-layer PEMFC Diffusion Medium," International Journal of Heat and Mass Transfer, vol. 46, pp , [35] K. T. Cho and M. M. Mench, "Fundamental Characterization of Evaporative Water Removal from Fuel Cell Diffusion Media," Journal of Power Sources, vol. 195, pp , [36] H. K. Atiyeh, K. Karan, B. Peppley, A. Phoenix, E. Halliop, and J. Pharoah, "Experimental Investigation of the Role of a Microporous Layer on the Water Transport and Performance of a PEM Fuel Cell," Journal of Power Sources, vol. 170, pp ,

50 [37] S. Park, J.-W. Lee, and B. N. Popov, "Effect of Carbon Loading in Microporous Layer on PEM Fuel Cell Performance," Journal of Power Sources, vol. 163, pp , [38] R. P. Ramasamy, E. C. Kumbur, M. M. Mench, W. Liu, D. Moore, and M. Murthy, "Investigation of Macro- and Micro-porous Layer Interaction in Polymer Electrolyte Fuel Cells," International Journal of Hydrogen Energy, vol. 33, pp , [39] A. Z. Weber and J. Newman, "Effects of Microporous Layers in Polymer Electrolyte Fuel Cells," Journal of The Electrochemical Society, vol. 152, pp. A677-A688, [40] J. H. Nam, K.-J. Lee, G.-S. Hwang, C.-J. Kim, and M. Kaviany, "Microporous Layer for Water Morphology Control in PEMFC," International Journal of Heat and Mass Transfer, vol. 52, pp , [41] H. Meng, "Multi-dimensional Liquid Water Transport in the Cathode of a PEM Fuel Cell with Consideration of the Micro-porous layer (MPL)," International Journal of Hydrogen Energy, vol. 34, pp , [42] J. P. Owejan, J. E. Owejan, W. Gu, T. A. Trabold, T. W. Tighe, and M. F. Mathias, "Water Transport Mechanisms in PEMFC Gas Diffusion Layers," Journal of The Electrochemical Society, vol. 157, pp. B1456-B1464, [43] M. P. Manahan, M. C. Hatzell, E. C. Kumbur, and M. M. Mench, "Laser Perforated Fuel Cell Diffusion Media Part I: Related Changes in Performance and Water Content," Journal of Power Sources, vol. 196, pp ,

51 [44] A. Turhan, K. Heller, J. S. Brenizer, and M. M. Mench, "Passive Control of Liquid Water Storage and Distribution in a PEFC through Flow-field Design," Journal of Power Sources, vol. 180, pp , [45] A. Turhan, S. Kim, M. Hatzell, and M. M. Mench, "Impact of Channel Wall Hydrophobicity on Through-plane Water Distribution and Flooding Behavior in a Polymer Electrolyte Fuel Cell," Electrochimica Acta, vol. 55, pp , [46] M. Khandelwal, S. Lee, and M. M. Mench, "Model to Predict Temperature and Capillary Pressure Driven Water Transport in PEFCs After Shutdown," Journal of the Electrochemical Society, vol. 156, pp. B703-B715, [47] M. Khandelwal and M. M. Mench, "An Integrated Modeling Approach for Temperature Driven Water Transport in a Polymer Electrolyte Fuel Cell Stack after Shutdown," Journal of Power Sources, vol. 195, pp , [48] M. C. Hatzell, A. Turhan, S. Kim, D. S. Hussey, D. L. Jacobson, and M. M. Mench, "Quantification of Temperature Driven Flow in a Polymer Electrolyte Fuel Cell Using High-Resolution Neutron Radiography," Journal of the Electrochemical Society, vol. 158, pp. B717-B726, [49] H. Bajpai, M. Khandelwal, E. C. Kumbur, and M. M. Mench, "A Computational Model for Assessing Impact of Interfacial Morphology on Polymer Electrolyte Fuel Cell Performance," Journal of Power Sources, vol. 195, pp ,

52 [50] T. Swamy, E. C. Kumbur, and M. M. Mench, "Characterization of Interfacial Structure in PEFCs: Water Storage and Contact Resistance Model," Journal of The Electrochemical Society, vol. 157, pp. B77-B85, [51] S. Kim and M. M. Mench, "Investigation of Temperature-Driven Water Transport in Polymer Electrolyte Fuel Cell: Phase-Change-Induced Flow," Journal of The Electrochemical Society, vol. 156, pp. B353-B362, [52] S. Kim and M. M. Mench, "Investigation of Temperature-Driven Water Transport in Polymer Electrolyte Fuel Cell: Thermo-Osmosis in Membranes," Journal of Membrane Science, vol. 328, pp , [53] S. Kim, M. Khandelwal, C. Chacko, and M. M. Mench, "Investigation of the Impact of Interfacial Delamination on Polymer Electrolyte Fuel Cell Performance," Journal of The Electrochemical Society, vol. 156, pp. B99-B108, [54] S. Kim, B. K. Ahn, and M. M. Mench, "Physical Degradation of Membrane Electrode Assemblies Undergoing Freeze/Thaw Cycling: Diffusion Media Effects," Journal of Power Sources, vol. 179, pp , [55] S. Kim and M. M. Mench, "Physical Degradation of Membrane Electrode Assemblies Undergoing Freeze/Thaw Cycling: Micro-Structure Effects," Journal of Power Sources, vol. 174, pp , [56] K. S. Chen, M. A. Hickner, and D. R. Noble, "Simplified Models for Predicting the Onset of Liquid Water Droplet Instability at the Gas Diffusion Layer/Gas Flow Channel Interface," International Journal of Energy Research, vol. 29, pp ,

53 [57] P. K. Sinha and C.-Y. Wang, "Liquid Water Transport in a Mixed-Wet Gas Diffusion Layer of a Polymer Electrolyte Fuel Cell," Chemical Engineering Science, vol. 63, pp , [58] B. Avasarala and P. Haldar, "Effect of Surface Roughness of Composite Bipolar Plates on the Contact Resistance of a Proton Exchange Membrane Fuel Cell," Journal of Power Sources, vol. 188, pp , [59] T. Kitahara, H. Nakajima, and K. Mori, "Hydrophilic and Hydrophobic Double Microporous Layer Coated Gas Diffusion Layer for Enhancing Performance of Polymer Electrolyte Fuel Cells under No-Humidification at the Cathode," Journal of Power Sources, vol. 199, pp , [60] M. Blanco, D. P. Wilkinson, and H. Wang, "Application of Water Barrier Layers in a Proton Exchange Membrane Fuel Cell for Improved Water Management at Low Humidity Conditions," International Journal of Hydrogen Energy, vol. 36, pp , [61] M. Eikerling, Y. I. Kharkats, A. A. Kornyshev, and Y. M. Volfkovich, "Phenomenological Theory of Electro-Osmotic Effect and Water Management in Polymer Electrolyte Proton Conducting Membranes," Journal of the Electrochemical Society, vol. 145, pp , January 1, [62] B. S. Pivovar, "An Overview of Electro-Osmosis in Fuel Cell Polymer Electrolytes," Polymer, vol. 47, pp , [63] T. A. Zawodzinski, C. Derouin, S. Radzinski, R. J. Sherman, V. T. Smith, T. E. Springer, and S. Gottesfeld, "Water Uptake by and Transport Through Nafion

54 Membranes," Journal of the Electrochemical Society, vol. 140, pp , January 1, [64] M. Tasaka, T. Hirai, R. Kiyono, and Y. Aki, "Solvent Transoport across Cation- Exchange Membranes under a Temperature Difference and under an Osmotic Pressure Difference," Journal of Membrane Science, vol. 71, pp , [65] J. P. G. Villaluenga, B. Seoane, V. M. Barragán, and C. Ruiz-Bauzá, "Thermo- Osmosis of Mixtures of Water and Methanol through a Nafion Membrane," Journal of Membrane Science, vol. 274, pp , [66] R. Zaffou, J. S. Yi, H. R. Kunz, and J. M. Fenton, "Temperature-Driven Water Transport Through Membrane Electrode Assembly of Proton Exchange Membrane Fuel Cells," Electrochemical and Solid-State Letters, vol. 9, pp. A418- A422, January 1, [67] R. Bradean, H. Haas, K. Eggen, C. Richards, and T. Vrba, "Stack Models and Designs for Improving Fuel Cell Startup From Freezing Temperatures," ECS Transactions, vol. 3, pp , October 20, [68] G. J. M. Janssen and M. L. J. Overvelde, "Water Transport in the Proton- Exchange-Membrane Fuel Cell: Measurements of the Effective Drag Coefficient," Journal of Power Sources, vol. 101, pp , [69] K.-H. Choi, D.-H. Peck, C. S. Kim, D.-R. Shin, and T.-H. Lee, "Water Transport in Polymer Membranes for PEMFC," Journal of Power Sources, vol. 86, pp ,

55 [70] Q. Yan, H. Toghiani, and J. Wu, "Investigation of Water Transport through Membrane in a PEM Fuel Cell by Water Balance Experiments," Journal of Power Sources, vol. 158, pp , [71] Y. Cai, J. Hu, H. Ma, B. Yi, and H. Zhang, "Effect of Water Transport Properties on a PEM Fuel Cell Operating with Dry Hydrogen," Electrochimica Acta, vol. 51, pp , [72] K. Karan, H. Atiyeh, A. Phoenix, E. Halliop, J. Pharoah, and B. Peppley, "An Experimental Investigation of Water Transport in PEMFCs: The Role of Microporous Layers," Electrochemical and Solid-State Letters, vol. 10, pp. B34- B38, January 1, [73] F. E. Hizir, S. O. Ural, E. C. Kumbur, and M. M. Mench, "Characterization of Interfacial Morphology in Polymer Electrolyte Fuel Cells: Micro-porous Layer and Catalyst Layer Surfaces," Journal of Power Sources, vol. 195, pp , [74] M. P. Manahan and M. M. Mench, "Increased Performance of PEFCs with Engineered Mass-Transport Pathways," ECS Transactions, vol. 16, pp , [75] M. P. Manahan and M. M. Mench, "Laser Perforated Fuel Cell Diffusion Media: Engineered Interfaces for Improved Ionic and Oxygen Transport," Journal of the Electrochemical Society, vol. 159, pp. F322-F330, January 1, [76] D. Gerteisen, T. Heilmann, and C. Ziegler, "Enhancing Liquid Water Transport by Laser Perforation of a GDL in a PEM Fuel Cell," Journal of Power Sources, vol. 177, pp ,

56 [77] D. Gerteisen and C. Sadeler, "Stability and Performance Improvement of a Polymer Electrolyte Membrane Fuel Cell Stack by Laser Perforation of Gas Diffusion Layers," Journal of Power Sources, vol. 195, pp , [78] H. Markötter, R. Alink, J. Hau mann, K. Dittmann, T. Arlt, F. Wieder, C. Tötzke, M. Klages, C. Reiter, H. Riesemeier, J. Scholta, D. Gerteisen, J. Banhart, and I. Manke, "Visualization of the Water Distribution in Perforated Gas Diffusion Media Layers by Means of Synchrotron X-ray Radiography," International Journal of Hydrogen Energy, vol. in press, [79] R. Alink, J. Haußmann, H. Markötter, M. Schwager, I. Manke, and D. Gerteisen, "The Influence of Porous Transport Layer Modifications on the Water Management in Polymer Electrolyte Membrane Fuel Cells," Journal of Power Sources, vol. 233, pp , [80] E. E. Kimball, J. B. Benziger, and Y. G. Kevrekidis, "Effects of GDL Structure with an Efficient Approach to the Management of Liquid water in PEM Fuel Cells," Fuel Cells, vol. 10, pp , [81] T. Kitahara, H. Nakajima, M. Inamoto, and M. Morishita, "Novel Hydrophilic and Hydrophobic Double Microporous Layer Coated Gas Diffusion Layer to Enhance Performance of Polymer Electrolyte Fuel Cells under both Low and High Humidity," Journal of Power Sources, vol. 234, pp , [82] T. Kitahara, H. Nakajima, and K. Mori, "Hydrophilic and Hydrophobic Double MPL Coated GDL to Enhance PEFC Performance under Low and High Humidity Conditions," ECS Transactions, vol. 41, pp ,

57 [83] F.-Y. Zhang, S. G. Advani, and A. K. Prasad, "Performance of a Metallic Gas Diffusion Layer for PEM Fuel Cells," Journal of Power Sources, vol. 176, pp , [84] J. Kim and N. Cunningham, "Development of Porous Carbon Foam Polymer Electrolyte Membrane Fuel Cell," Journal of Power Sources, vol. 195, pp ,

58 2. DIAGNOSTIC TECHNIQUES IN PEFCS Many diagnostic techniques for PEFCs have been well established, and a plethora of pertinent literature is available for determining various properties of the system. The discussion in this chapter will provide a broad range of in-situ and ex-situ techniques based on physical and electrochemical processes. 2.1 IMPORTANCE OF METHODOLOGY AND ERROR ANALYSIS Throughout the work presented herein, rigorous experimental care was given to produce repeatable data. In all experimental work, especially that with electrochemical systems, slight variations in experimental conditions can invalidate the data. Dedication to producing repeatable and reliable data was used in all experiments conducted. When possible and where appropriate, care was taken to include error bars on data points so any uncertainty/variation in results is known by all who analyze the plots. In many of these results, these error bars are key to have confidence in stating performance results. Diagnosing a polarization curve is critical (Section 2.2); however, the methodology used to obtain the pol curve is just as important. All polarization curves in this thesis were obtained by stepping the current or voltage, starting from open circuit voltage (OCV), toward higher current/lower voltages. Many references have shown the effect of hysteresis the effect of previous test conditions on the performance to be a strong factor in experimental uncertainty. To combat this hysteresis effect during polarization testing, the cell was maintained at a constant condition (e.g., constant voltage or constant current) for at least 10 minutes prior to conducting any pol curve. Additionally, any 38

59 polarization curve shown in the experimental sections are an average of three or more polarization curves, and the error bars are added to indicate the maximum and minimum averaged value for the individual pol curves. In Section 2.4, neutron radiography results are presented. This work was conducted at the National Institute for Standards and Technology (NIST) in Gaithersburg MD. Limited neutron beam time is available, so test times are limited. Nonetheless, experimental consistency was kept as a priority. Any results presented represent steady-state values. At any given current density, the cell was run for 20 minutes prior to collecting neutron data. Since neutron radiography reveals liquid water content in a fuel cell, it is best to use constant current conditions since this controls the water generation at the cathode catalyst layer. Then, when results between two cells or two operating conditions are compared, the experimentalist can be ensured that a known amount of water was produced in the cell. The methodology of electrochemical impedance spectroscopy (Section 2.5) is also very important. In all experiments, the cell was run at constant current for 20 minutes prior to probing the system with alternating current. Then, 10 consecutive impedance sweeps were made with the AC signal amplitude of 5% of the current set value. 2.2 POLARIZATION TESTING PERFORMANCE CHARACTERIZATION Polarization testing is the quintessential experiment for electrochemical devices. For batteries, they often take the form of charge/discharge cycles, and for fuel cells, they are 39

60 called polarization curves, or pol curves. A sample curve is shown in Figure 2.1. A detailed description has already been given in Chapter 1, and only a brief review of the three regions shown in Figure 2.1 will be discussed here. - Region I is referred to as the activation polarization region because the most dominant factor in the performance characteristics is a result of the kinetic polarization at each electrode. That is, activation polarization is the voltage overpotential required to overcome the activation energy of the electrochemical reaction at each electrode. - Region II is known as the ohmic polarization region, is characterized by a linear voltage drop with increasing current governed by Ohms Law. - Region III is the concentration polarization region, also known as the masstransport region. Region III is characterized by a relatively rapid drop in voltage for small increases in current at high current values. This sharp drop in performance is due to concentration polarization, which is the reduction of reactant surface concentration at the triple phase boundary, thus reducing the sustainability of the electrochemical reaction. 40

61 Figure 2.1. Characteristic polarization curve for a PEFC with kinetic, ohmic, and mass transport regions labeled (I, II, III, respectively). These three regions can be effectively used to characterize the general performance of a PEFC, however, they are not mutually exclusive. That is, for example, even in Region III (concentration polarization), ohmic and activation losses are still present and have an effect on the performance. These regions are used as general guidelines, indicating which polarization source is most dominant at any given current TAFEL SLOPE ANALYSIS The Tafel slope, a measure of the electrode kinetics, can be measured in a variety of ways (including from a simple pol curve), but in general is the slope of the semi-log plot 41

62 of the voltage versus the log of the current. A common expression for the Tafel kinetics that is typically valid for the oxygen reduction reaction (ORR) is shown in Eq. (2.1) [1]: act, c i i blog io x (2.1) where b is the Tafel slope, i the current density, i x the current density due to hydrogen crossover, and i o the exchange current density. When plotting a Tafel plot and calculating the Tafel slope, the cell voltage should always be corrected for ohmic losses, and the current corrected for crossover current [1, 2]. Experimentally, pure oxygen is typically used on the cathode side in order to remove any diffusion-related limitations, which isolates the kinetics [1], but air or helox typically do not alter results significantly at low to moderate currents. Typical Tafel slopes are between 50 and 80 mv dec -1*, with some reaching upwards of 120 mv dec -1 [1-4] depending on the electrode composition and structure. Ref. [5] suggested the reaction mechanism that yields approximately 60 mv dec -1. So, one use of a Tafel slope analysis is to determine kinetic parameters (such as the exchange current density and transfer coefficient [3-5]) LIMITING CURRENT ANALYSIS One result obtainable from the pol curve is the approximation of the limiting current. However, if high concentrations of oxygen are used as oxidant (even air-levels), flooding and gradients (e.g., temperature [6] and reactant concentration) become significant at higher current. Several references indicate that the oxygen transport resistance can be * Note that a decade is not an SI unit, however, the standard method to report a Tafel slope is in the given units. 42

63 experimentally obtained by operating a fuel cell to the limiting current under different conditions [7-10]. However, if using air, the limiting current may produce significant amounts of liquid water in the cell, skewing pure gas-phase diffusion results. Therefore, it may be desirable to operate in diffusion-limited conditions without being at high currents where other effects may skew results. In this case, low oxygen concentrations can be used (e.g., 1-5% O 2 ) with the balance in nitrogen (or other inert gases). Several studies have done this when investigating the diffusion-limited limiting current, where negligible liquid water effects were desirable [7, 8]. 2.3 GAS MODIFICATION Any in-situ experiment can be conducted with hydrogen on the anode and any form of oxidant on the cathode. Often, air is used since it is readily available and inexpensive for experimental tests (typically compressed breathing grade air is used). However, pure oxygen can be used, or any percentage of oxygen desired. For example, when air-based results are desired but oxygen partial pressure variation is also required, Perry et al. [11] suggests using half air, air, and double air to vary oxygen partial pressure instead of using air and pure oxygen. They suggest this so the oxygen concentration is not changed from air (~21% O 2 ) to five times the value with pure oxygen; rather, it is changed by a factor of two (half air, 10.5%, double air, 42% O 2 ) so properties close to air can be determined. In the same vein, the inert gas can be modified. While air is the typical oxidant (containing 21% O 2 and ~79% N 2 ), a mixture called helox can be substituted, which replaces the 79% nitrogen with 79% helium. O 2 -He has a binary diffusion coefficient of 43

64 0.822 cm 2 sec -1, compared to that of O 2 -N 2 of cm 2 sec -1 at 317 K [12]. That is, oxygen is approximately 3.5 times more diffuse in helium as it is in nitrogen. Therefore, helox can be used to greatly reduce mass-transport limitations. One consideration to make is that water vapor is also significantly more diffuse in helium than in nitrogen [12], so evaporative water removal changes may be present in addition to pure gas-phase diffusion effects. The same principles mentioned here can be applied to the limiting current analysis technique mentioned in Section where low oxygen concentrations may be desirable to maintain sufficiently low currents to avoid large gradients and flooding effects. That is, it may be desirable to operate in diffusion-limited conditions without being at high currents where other effects may skew results. In this case, low oxygen concentrations can be used (e.g., 1-5% O 2 ) with the balance in nitrogen (or other inert gases). 2.4 NEUTRON RADIOGRAPHY Water management in PEFCs is a significant area of research in PEFCs since many properties depend on water content (e.g., membrane hydration, DM permeability, flooding effects, etc.). Therefore, knowing the liquid water content and profiles within the cell is important. Many in-situ techniques have been developed to visualize liquid water in the cell, but many of these involve modifying the cell components in order to be compatible with the technique. For example, magnetic resonance imaging (MRI) requires that no metal components be present, which changes the thermal and compression behavior of typical PEFC components. Other visual techniques require transparent 44

65 components that also have similar effects. These changes will inevitably cause a change in the water profiles. Neutron radiography (NR) also known as neutron imaging however, still permits the use of typical PEFC components. The fundamental working principle of NR is based on neutrons passing through an object (e.g., a fuel cell) that is design to be optically thin, and the attenuation of the original intensity of the neutron source is measured [13]. The ratio of the transmitted neutron intensity ( I ) to the initial neutron intensity ( I o ) a material is given in Ref. [14]: I I o exp( NL) (2.2) eff where eff the effective neutron attenuation cross section per molecule for that material, N is the density of the molecule, and L is the optical path length in the material. Figure 2.2 shows a schematic of the fuel cell in the path of a neutron source. Typically, the beam-direction material thickness (both hardware and soft goods) should be kept as thin as is reasonably possible and realistic in order to prevent to maximize the effect of the water attenuation whilst minimizing any attenuation from other components. Once through the fuel cell components, the neutrons that were not attenuated by liquid water hit a scintillator plate, which releases photons. These photons, then, hit a charge coupled device (CCD) camera, allowing for the resulting image to be viewed digitally [15, 16]. A recent update on specific information on the facilities at the National Institute of Standards and Technology (NIST) in Gaithersburg MD can be found in Ref. [17]. 45

66 At this point, numerous groups have used NR in water content studies, for example Refs. [18-31]. This technology has proven itself to be insightful and reliable. Other in-situ visualization techniques have been developed, such as synchrotron X-ray radiography [32-34]. Instead of interacting with the nucleus of hydrogen atoms like neutron radiography, X-ray techniques interact with the electron cloud surrounding the nucleus. This means that neutrons have a higher interaction probability with small elements like hydrogen compared to X-rays. While there are spatial resolution advantages to using X- ray methods, current NR developments are nearing those of X-ray techniques. (a) (b) Figure 2.2. Schematics adapted from (a) Ref. [14] and (b) Ref. [35] showing fuel cells in the path of a neutron beam. In these examples, the beam is parallel to the in-plane direction. Other references show the beam parallel to the through-plane direction as well. The hydrogen atom has a large attenuation for neutrons, which means it scatters the neutron path, rather than absorbing neutrons (like boron) or letting them pass through (like many metals). Since hydrogen is dense in liquid water (compared to hydrogen gas or water vapor), significantly high attenuation is achieved with liquid water in the cell and visualization of liquid water in the cell is made possible. Typically, a dry image is 46

67 taken in which no liquid water is present in the cell. Then, during experimentation, any water that is in the cell will be attenuated, and images can be effectively compared for different operating conditions. 2.5 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY Electrochemical impedance spectroscopy (EIS), also known as just impedance spectroscopy, is arguably the single most powerful electrochemical technique available. Ref. [36] says, The power of the method lies in the fact that by small signal perturbation it reveals both the relaxation times and relaxation amplitudes of various processes present in a dynamic system over a wide range of frequencies. An excellent resource on the fundamentals of EIS can be found in Ref. [37], which covers EIS as a general electrochemical technique (not applied to PEFCs). PEFCs represent a complex, dynamic system that warrants the use of EIS. Charge carrier transport, electronic and ionic conduction, adsorption, dissociation, ionization, gas diffusion Electrode formation, adsorption of gas impurities, microstructural changes Corrosion, interdiffusion, agglomeration Cell operation Start-up Degradation µs 1 min 10 h h Electrochemical impedance spectroscopy Long term measurements I/V-characteristics log(t) (s) Figure 2.3. Relaxation times of physical processes in PEFCs and the diagnostic techniques plausible for certain time ranges. Adapted from Figure 1 of Ref. [36]. 47

68 As was previously mentioned, when direct current (DC), I, passes through an electrochemical device, the device voltage, V, deviates from the previous equilibrium value (V o ). So, when both I and V are known, the polarization resistance can be determined for the device. In general, and certainly in PEFCs, the I-V relationship is nonlinear. This is DC method is the same as that discussed in Chapter 1 and Section 2.2, called polarization testing. EIS, on the other hand, utilizes alternating current (AC) measurements. The device is perturbed by an AC input current of certain frequency and amplitude, and the AC voltage output is monitored. Figure 2.4 shows the concept of AC circuit responses. The left side shows the current, I, with a amplitude signified as the radius oscillating at frequency. The voltage response, E, to the AC input signal differs in amplitude (less than that of I, as shown) and phase angle,. Note that this concept can be reversed, i.e., the voltage is varied in a sinusoidal manor and the current response is monitored. (This latter scenario is the condition presented in the right-hand-side plot in Figure 2.4.) The current-voltage signal will be different at each frequency, which is the aim of EIS to exploit the frequency dependence nature of the response to extract information from the electrochemical system. 48

69 Figure 2.4. Diagram from Figure in Ref. [37] indicating the relationship between input AC current at frequency,, and the voltage response. The phase angle,, is the amount that the two responses are separated. Two simple cases can be considered. First, consider a system of a pure resistance. If the input sinusoidal voltage is represented as, e Esin( t) (2.3) the current response would simply be governed by Ohm s law, or, i e R E sin( t) R (2.4) where the phase angle is zero. Au contraire, consider a system with a pure capacitance, C. Then, the fundamental relation that governs the response is, de i C dt (2.5) Therefore, 49

70 i CE cos( t) E X C sin( t ) 2 (2.6) where X C is the capacitive resistance, or 1 / C. Since a phase angle has now been introduced, common notation is to define components along the ordinate axis as imaginary components. When a resistor (R) and a capacitor (C) are in series with a voltage applied across them, there exists both real and imaginary components. In equation form, Esin( t) I sin( t)( R jx C ) I sin( t)z (2.7) where Z=(R - jx C ) is referred to as the impedance. So, then, the way in which the voltage varies with the current is governed by the impedance, which consists of a real component and an imaginary component. These values are typically plotted on characteristic plots, i.e., a Bode plot ( log Z and vs. log ) and/or a Nyquist plot (Z Im vs. Z Re over a given frequency range). In the Nyquist plot, the frequency,, is implied and often selected frequencies are explicitly called out to aid in interpretation. Sample experimental data of Bode and Nyquist plots are shown in Figure 2.5a and Figure 2.5b, respectively. 50

71 Phase Angle, Phase Angle Log Z log frequency (a) Log Z -Z'' Imaginary Impedance (Ohm cm 2 ) Hz 100 Hz > 3000 Hz < 1 Hz Z' Real Impedance (Ohm cm 2 ) (b) Figure 2.5. (a) Bode and (b) Nyquist plots of experimental data. Numerous studies have employed EIS as a diagnostic tool in PEFC research, e.g., Refs. [36, 38-46]. A number of application- and instruction-oriented publications are particularly suited for understanding and applying EIS techniques to PEFC research [36, 37, 41, 42, 47]. These references (and many others) establish that any electrochemical device can be described in terms of an equivalent circuit. That is, the physical components and processes (e.g., porous electrode with distributed catalyst sites supported by carbon, electrically resistive components, charge transfer at the electrode/electrolyte interface, etc.) can be generally modeled as a combination of resistances, capacitances, and complex impedances (and inductances at high frequencies) [47]. Figure 2.6 shows a sample equivalent circuit that is often used to describe an electrochemical cell, such as the anode or cathode of a PEFC. R el is the electrolyte resistance between the working electrode (Pt surface in PEFCs) and the reference electrode (often the anode side in PEFCs). C d is a capacitor which represents the physical double layer present at the electrode/electrolyte interface. Z f is the Faradaic impedance, 51

72 which correlates to the charge transfer process at the electrode/electrolyte interface. Figure 2.6b refers to the possible breakdown of the Faradaic impedance component. Figure 2.6. (a) A simple equivalent circuit of an electrochemical cell, such as one of the electrodes in a PEFC, and (b) two possibilities for the equivalent of Z f, from Figure 3.1 of Ref. [47]. Finally, the Nyquist plot can be reasonably understood as follows and is schematically shown in Figure 2.7. At high frequency (which corresponds with the lowest impedance values), capacitive effects act like a short circuit, leaving only the electrolyte resistance (R el ). At high frequency, we are probing the extremely fast processes, which include electronic and ionic transport through the cell. (This is the same principle that is used with high frequency resistance (HFR) values previously discussed. HFR measures R el only at one frequency, whereas EIS covers a broad range of frequencies.) As the sinusoidal frequency decreases to moderate values (e.g., between Hz), a mid- 52

73 frequency arc is observed, which is associated with charge transfer processes. These processes are comparatively slower than electron and ion transfer, but are still very fast compared to diffusion related processes. As denoted in Figure 2.7, the diameter of the charge transfer arc is R ct, but its (approximate) position on the real-impedance axis is the sum of R el + R ct. Finally, at low frequency perturbations, we are probing the slow processes involved in PEFC operation, namely, diffusion-related processes. This can be any combination of diffusion processes in the fuel cell, including diffusion through ionomer thin films in the catalyst layer to flooding effects caused by DM pore blockages. Figure 2.7. Impedance spectra showing R el, R ct, and R d components at high, medium, and low frequency, respectively. Figure 1 from Ref. [44]. 53

74 It should be noted that three individual, well-defined arcs cannot always be observed. Since the processes can have significant overlap, it can sometimes be difficult to determine the fundamental polarization occurring. For this reason, EIS has two uses: 1) it can be used to compare the same PEFC under different conditions; that way, the arcs and different frequencies will be notably different, and comparison of individual polarizations may be appropriately identified. 2) It is generally helpful (if not required) to combine the interpretation of EIS results with other electrochemical experiments, such as those presented in this chapter. The combination of EIS with other experimental methods can prove itself to be an extremely powerful tool in diagnosing cell performance limitations. 2.6 ELECTROCHEMICAL ACTIVE AREA MEASUREMENT WITH CYCLIC VOLTAMMETRY Cyclic voltammetry (CV) is a transient electrochemical technique in which the voltage is cycled at a constant rate (often called the slew rate) from a lower voltage value to an upper voltage value and back to the lower voltage value to complete one cycle. The slew rate and the number of cycles tested depend on the experiment being conducted, and the voltage range depends on the chemistry of the system. CV is a very powerful technique when analyzing redox systems. It can be used for i) to delineate the reactions at an electrode as a function of applied voltage, ii) determine the electrochemical active area of an electrode, iii) determine the presence of adsorbed surface poisons, iv) to study performance dependence on crystal lattice structure and particle size of electrodes, and v) to study catalyst degradation [2]. Here, however, the discussion will be limited to the determination of electrochemical active surface area (ECSA or ESA) in PEFCs. 54

75 General CV principles require three electrodes: a reference electrode (RE), a counter electrode (CE), and a working electrode (WE). The working electrode is the electrode whose ECSA is to be determined. Humidified nitrogen is flowed through the working electrode side of the PEFC. In hydrogen fuel cells, use of a true reference electrode is difficult due to the configuration, and therefore the hydrogen side typically doubles as the RE and the CE. This is called the dynamic hydrogen electrode (DHE). The three electrodes are connected to a potentiostat, and the CV experiment can be conducted. Typical ranges of slew rate are from mv s -1, and the voltage range is cycled from 50 or 100 mv to approximately 1.0 V. It is recommended that several sweeps be cycled so a steady state value can be obtained. The method to determine ECSA, three assumptions must be made: i) each platinum site can adsorb only one hydrogen proton, ii) every electrochemically active platinum site will be occupied with hydrogen during the transition from hydrogen adsorption to hydrogen evolution, and iii) the charge density for a hydrogen atom on the platinum surface is μc cm Pt as found in literature [48]. Then, the determination of the ECSA is straight forward, A Pt 210x10 6 I(t)dt 1 (2.8) 2 C cm (L )(A ) 10 Pt ca g where A pt is the catalyst loading in m 2-1 Pt g Pt, the integral I(t)dt is the integrated area of the hydrogen oxidation peaks as shown in Figure 2.8 (discussed later), loading in units of mg cm -2 Pt L ca is the catalyst, A g is the geometric (planform) area. The one-tenth factor 55

76 is to convert final units of A pt. Two notes should be made when using this equation: first, units are important. As in all calculations, units should be checked for consistency. The units listed should be used as described and should not be converted to SI units. Secondly, the integral term is obtained in one of two ways: either the x-axis of Figure 2.8 can be converted from V to seconds using the slew rate of the experiment conducted, or the integral of the data as-is can be calculated, and the final area can be converted to the proper units with the slew rate. The final units of the integral term in Eq. (2.8) should be coulombs. Typical values of ECSA for state-of-the-art PEFC electrodes are typically 2-1 between 40 and 70 m. Pt g Pt 56

77 Figure 2.8. Cyclic voltammograms of platinum in acid, adapted from Ref. [49]. Shaded area is the area used to calculate the ECSA. 2.7 DOUBLE TAFEL PLOT ANALYSIS Tafel plots are constructed with the log of the current (corrected for hydrogen crossover) vs. voltage (corrected for ohmic losses). Typically, pure Tafel slope plots are run using humidified hydrogen on the anode, and humidified pure oxygen on the cathode. As was previously mentioned in Section 2.2.2, typical Tafel slopes are around 60 mv 57

78 dec -1, some reaching upwards of 120 mv dec -1 [1-4] depending on the electrode composition and structure. Another variation of a Tafel slope analysis has been used. A method of delineating between oxygen- and ionic-transport limited scenarios was first suggested by Perry and coworkers [50] and further work was conducted several years later [11, 43, 51, 52]. These works present a thorough description of their model and experiments, and only a brief summary will be provided here. The goal of the suggested experimental technique is to determine the source of the well-established double Tafel slope region. Losses that contribute to the double Tafel slope region are a combined effect of the kinetics of the electrode (which yields the single Tafel slope) plus another source of loss, namely, either oxygen transport or ionic transport. That is, when a double Tafel slope appears, it signifies the ORR is limited by either Tafel kinetics and oxygen diffusion or by Tafel kinetics and proton migration in the CL. The current is first-order in oxygen concentration if the performance is dominated by kinetics and oxygen transport since both the ORR kinetics and oxygen transport are first order processes. However, if the ORR kinetics and ionic transport dominate the losses, it has been derived and shown experimentally that the current is half-order in oxygen concentration [53]. Experimentally, then, we can vary the total pressure or the oxygen partial pressure and analyze the response of the current to such changes, and thus determine the source of limitation. Should the experimentalist be interested in the ORR limitation source when using air, it is recommended that cathode stream should be kept close to the air condition when varying the oxygen partial pressure. For example, instead of using air and pure 58

79 oxygen to vary the partial pressure (a five-fold change), it is better to use air (21% O 2, balance N 2 ), half-air (10.5% O 2, balance N 2 ), and double-air (42% O 2, balance N 2 ), as suggested in Ref. [48]. 59

80 REFERENCES [1] H.A. Gasteiger, W. Gu, R. Makharia, M.F. Mathias, B. Sompalli, Begnning-of-life MEA Performance--Efficiency Loss Contributions, in: H.A.G. W. Vielstich, A. Lamm (Ed.) Handbook of Fuel Cells: Fundamentals, Technology, and Applications, John Wiley & Sons, 2003, pp [2] M.M. Mench, Fuel Cell Engines, John Wiley & Sons, Inc., Hoboken, [3] L. Giorgi, E. Antolini, A. Pozio, E. Passalacqua, Electrochim. Acta, 43 (1998) [4] A. Parthasarathy, S. Srinivasan, A.J. Appleby, C.R. Martin, J. Electrochem. Soc., 139 (1992) [5] A. Parthasarathy, C.R. Martin, S. Srinivasan, J. Electrochem. Soc., 138 (1991) [6] H. Bajpai, M. Khandelwal, E.C. Kumbur, M.M. Mench, J. Power Sources, 195 (2010) [7] D.R. Baker, C. Wieser, K.C. Neyerlin, M.W. Murphy, ECS Transactions, 3 (2006) [8] D.R. Baker, D.A. Caulk, K.C. Neyerlin, M.W. Murphy, J. Electrochem. Soc., 156 (2009) B991-B1003. [9] M.P. Manahan, S. Kim, E.C. Kumbur, M.M. Mench, ECS Transactions, 25 (2009) [10] U. Beuscher, J. Electrochem. Soc., 153 (2006) A1788-A

81 [11] M.L. Perry, R. Balliet, R.M. Darling, Experimental Diagnostics and Durability Testing Protocols, in: M.M. Mench, E.C. Kumbur, T.N. Veziroglu (Eds.) Polymer Electrolyte Fuel Cell Degradation, Elsevier, Boston, 2012, pp [12] E.L. Cussler, Diffusion: Mass Transfer in Fluid Systems, 3rd Edition ed., Cambridge University Press, New York, [13] J.C. Domanus, G. Bayon, Practical Neutron Radiography, Kluwer Academic Publishers, Boston, [14] R.J. Bellows, M.Y. Lin, M. Arif, A.K. Thompson, D. Jacobson, J. Electrochem. Soc., 146 (1999) [15] A.K. Heller, in: Nuclear Engineering, The Pennsylvania State University, University Park, [16] A. Turhan, in: Mech. Eng., The Pennsylvania State University, University Park, [17] D.S. Hussey, D.L. Jacobson, M. Arif, K.J. Coakley, D.F. Vecchia, Journal of Fuel Cell Science and Technology, 7 (2010) [18] A.B. Geiger, A. Tsukada, E. Lehmann, P. Vontobel, A. Wokaun, G.G. Scherer, Fuel Cells, 2 (2002) [19] R. Satija, D.L. Jacobson, M. Arif, S.A. Werner, J. Power Sources, 129 (2004) [20] T.A. Trabold, J.P. Owejan, D.L. Jacobson, M. Arif, P.R. Huffman, Int. J. Heat Mass Transfer, 49 (2006)

82 [21] N. Pekula, K. Heller, P.A. Chuang, A. Turhan, M.M. Mench, J.S. Brenizer, K. Ünlü, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 542 (2005) [22] A. Turhan, K. Heller, J.S. Brenizer, M.M. Mench, J. Power Sources, 160 (2006) [23] A. Turhan, K. Heller, J.S. Brenizer, M.M. Mench, J. Power Sources, 180 (2008) [24] J.P. Owejan, T.A. Trabold, J.J. Gagliardo, D.L. Jacobson, R.N. Carter, D.S. Hussey, M. Arif, J. Power Sources, 171 (2007) [25] D. Kramer, J. Zhang, R. Shimoi, E. Lehmann, A. Wokaun, K. Shinohara, G.G. Scherer, Electrochim. Acta, 50 (2005) [26] J.P. Owejan, T.A. Trabold, D.L. Jacobson, M. Arif, S.G. Kandlikar, Int. J. Hydrogen Energy, 32 (2007) [27] D. Kramer, E. Lehmann, G. Frei, P. Vontobel, A. Wokaun, G.G. Scherer, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 542 (2005) [28] M.A. Hickner, N.P. Siegel, K.S. Chen, D.N. McBrayer, D.S. Hussey, D.L. Jacobson, M. Arif, J. Electrochem. Soc., 153 (2006) A902-A908. [29] D.S. Hussey, D.L. Jacobson, M. Arif, J.P. Owejan, J.J. Gagliardo, T.A. Trabold, J. Power Sources, 172 (2007) [30] M.A. Hickner, N.P. Siegel, K.S. Chen, D.S. Hussey, D.L. Jacobson, M. Arif, J. Electrochem. Soc., 155 (2008) B427-B434. [31] A.Z. Weber, M.A. Hickner, Electrochim. Acta, 53 (2008)

83 [32] C. Hartnig, I. Manke, J. Schloesser, P. Krüger, R. Kuhn, H. Riesemeier, K. Wippermann, J. Banhart, Electrochem. Commun., 11 (2009) [33] I. Manke, C. Hartnig, M. Grunerbel, W. Lehnert, N. Kardjilov, A. Haibel, A. Hilger, J. Banhart, H. Riesemeier, Appl. Phys. Lett., 90 (2007) [34] P.K. Sinha, P. Halleck, C.-Y. Wang, Electrochem. Solid-State Lett., 9 (2006) A344- A348. [35] M.P. Manahan, M.C. Hatzell, E.C. Kumbur, M.M. Mench, J. Power Sources, 196 (2011) [36] E. Ivers-Tiffee, A. Weber, H. Schichlein, Electrochemical Impedance Spectroscopy, in: W. Vielstich, A. Lamm, H.A. Gasteiger (Eds.) Handbook of Fuel Cells: Fundamentals, Technology and Application, John Wiley & Sons, Hoboken, 2003, pp [37] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Applications, 2nd ed. ed., John Wiley & Sons, Hoboken, [38] T.E. Springer, T.A. Zawodzinski, M.S. Wilson, S. Gottesfeld, J. Electrochem. Soc., 143 (1996) [39] N. Wagner, J. Appl. Electrochem., 32 (2002) [40] D. Malevich, E. Halliop, M.S. Saha, J. Suryana, B. Peppley, J. Pharoah, K. Karan, ECS Meeting Abstracts, 902 (2009) [41] D. Malevich, E. Halliop, B.A. Peppley, J.G. Pharoah, K. Karan, J. Electrochem. Soc., 156 (2009) B216-B224. [42] R. Makharia, M.F. Mathias, D.R. Baker, J. Electrochem. Soc., 152 (2005) A970- A

84 [43] M. Eikerling, A.A. Kornyshev, J. Electroanal. Chem., 475 (1999) [44] M. Ciureanu, R. Roberge, The Journal of Physical Chemistry B, 105 (2001) [45] Y. Bultel, K. Wiezell, F. Jaouen, P. Ozil, G. Lindbergh, Electrochim. Acta, 51 (2005) [46] M.P. Manahan, M.M. Mench, ECS Transactions, 16 (2011) [47] X.-Z. Yuan, C. Song, H. Wang, J. Zhang, Electrochemical Impedance Spectroscopy: Fundamentals and Applications, Springer, New York, [48] T.R. Ralph, G.A. Hards, J.E. Keating, S.A. Campbell, D.P. Wilkinson, M. Davis, J. St-Pierre, M.C. Johnson, J. Electrochem. Soc., 144 (1997) [49] C.H. Hamann, A. Hamnett, W. Vielstich, Electrochemistry, Wiley-VCH, Weinheim, [50] M.L. Perry, J. Newman, E.J. Cairns, J. Electrochem. Soc., 145 (1998) [51] F. Jaouen, G. Lindbergh, G. Sundholm, J. Electrochem. Soc., 149 (2002) A437- A447. [52] J. Ihonen, F. Jaouen, G. Lindbergh, A. Lundblad, G. Sundholm, J. Electrochem. Soc., 149 (2002) A448-A454. [53] K.A. Striebel, F.R. McLarnon, E.J. Cairns, Industrial & Engineering Chemistry Research, 34 (1995)

85 3. EFFECTS OF SURFACE IRREGULARITIES AND INTERFACIAL CRACKS ON POLYMER ELECTROLYTE FUEL CELL PERFORMANCE The present study seeks to investigate the impact of surface irregularities and cracks at the catalyst layer (CL) and micro-porous layer (MPL) interface on the mass and electronic transport of polymer electrolyte fuel cells (PEFCs). Two different CLs were compared, i.e. one with negligible cracking and the other with high cracking (~6% planform reduction), under a combination of various operating conditions, including high/low relative humidity, and the presence of nitrogen/helium inert gases in the cathode inlet stream. A limiting current density analysis indicated that the cracked CL demonstrated a small increase (0.5%) in the Fickian diffusion of the reactants across the cathode electrode compared to the negligible-cracked CL case. Furthermore, the results from a relative humidity analysis showed that the protonic resistance in the CL might dominate the moderate current density region (between ~0.2 and 0.45 A/cm 2 ). Finally, comparison of the cell performance for cracked and negligible-cracked CL cases suggests that the cracks may act as water pooling sites, which in turn, may enhance the performance in the high current density region (over 0.45 A/cm 2 ) due to the decreased water surface coverage and/or enhanced water removal. 3.1 INTRODUCTION Polymer electrolyte fuel cells (PEFCs) maintain the most promising alternative portable energy source, especially for automotive applications. Nonetheless, there still remain substantial areas of research to better understand the dynamics of this complex 65

86 device and to optimize its components for improved efficiency. Of particular interest is the optimization of the interfaces between the components of a PEFC. Changes in interfacial morphology of PEFC components can have a significant effect on the cell performance. Several studies [1-6] have investigated the contact resistances and related losses at the bipolar plate and gas diffusion layer interface. In addition, numerous studies have shown that cycling the load and the temperature under normal operating conditions can produce and cause the growth of cracks in the membrane electrode assembly (MEA) [7-9], which may directly affect the morphology of the catalyst layer (CL) micro-porous layer (MPL) interface. In most fuel cell studies, the interface between the CL and the MPL surfaces is assumed to be perfectly smooth [10, 11]; however, in reality, the CL and MPL surfaces are highly rough [12], and therefore the contact between these two layers is not perfect. Recently, several studies have investigated the contact resistance at this interface [13, 14]. Based on this motivation, this study seeks to understand the impact of surface irregularities and cracks at the CL MPL interface on PEFC performance by investigating the mass and electronic transport under different operating conditions via a detailed in-situ experimental analysis. 3.2 EXPERIMENTAL SIGRACET Gas Diffusion Layer (SGL 10BB series coated with MPL) and commercially available MEA were used in this study. The CL was modified by introducing cracks, effectively reducing the CL contact surface area by approximately 6% (Figure 3.1a). This affected the CL MPL interface and the results were compared to the negligible-cracked CL case (Figure 3.1b). The inlet gases used were 5% O 2 with the balance in nitrogen (5%-O 2 nitrox) and 5% O 2 with the balance of helium (5%-O 2 66

87 heliox). Lower oxygen content gas was used in order to keep thermal gradients to a minimum. Two relative humidity conditions were tested, i.e. 100%/100% anode/cathode relative humidity ( wet conditions) and 70%/50% anode/cathode relative humidity ( dry conditions), to investigate the effect of liquid water for the cracked and negligiblecracked CL. The inert gases were varied in order to examine the effect of reactant diffusion rate in the cracked versus negligible-cracked CL. (a) (b) Figure 3.1. SEM images of the surface of the a) cracked CL having crack widths of ~30μm, b) negligible-cracked CL having crack widths of ~5μm. Typically, variability in cell assembly, component properties, and operating conditions may yield inconsistent/unrepeatable performance data, which may inhibit capturing the effects of surface cracks on the performance. Meticulous care was taken in this investigation, however, to ensure repeatable data collection in order to reduce experimental uncertainty. Several key tools were employed to eliminate experimental uncertainty, including: i) implementation of metallic gaskets to control the overall compression of the components [15, 16]; ii) use of subgaskets to obtain the exact MEA 67

88 active area (area exposed to reactant gases) and double the sealing surface for the silicone gasket (refer to Figure 3.3); iii) use of silicone gaskets to eliminate the external gas leakage; iv) use of coolant to achieve constant temperature boundary conditions and enhance temperature control of cell components [17-19]; and v) vertical orientation of gas flow channel design and cell orientation to reduce the amount of buildup of liquid water and minimize the performance variation [20]. Figure 3.2 illustrates several of the experimental improvements, such as the use of a metallic/silicone gaskets, and precision of the component size/placement. Graphite Plate Diffusion Media 37 mm 29 mm 26 mm 22.5 mm 35 mm MEA Silicone Gasket Metallic gasket Subgasket Figure 3.2. A schematic of the cross section of the experimental setup. Experiments were performed by an in-house fuel cell testing station and an Arbin E- load was used to control the current/voltage cycles. Polarization data for each test were obtained once the cell reached the steady-state condition. In each test, the cell was cycled between 0.4 V and OCV conditions for 20 minutes. This step was then followed by 10 minutes of constant 0.6 V operation to ensure that the cell possessed nearly the same water content and temperature distribution regardless of previous testing conditions. 68

89 Constant current steps were applied and the associated voltage response was monitored. Each constant current step was performed for two minutes in order to achieve a quasisteady-state condition. Fluctuations in the current density were observed especially in the high current density region. This can be attributed to the flooding conditions and/or oxygen starvation. When the fuel cell could not meet the current draw, a final set of data was taken at 0.1 V constant voltage for 15 seconds in order to measure the limiting current density as accurately as possible without overheating the cell. After one cycle (from zero current to the approximate limiting current), the cell was run at constant 0.6 V for 10 minutes, and then the cycle was repeated two more times. 3.3 RESULTS AND DISCUSSION EFFECT OF CRACK ON GAS TRANSPORT The first phase of this study was focused on examining the effects of CL surface cracks on the gas transport and evaluating the mode of diffusion (Fickian or other) across the cathode electrode. A detailed comparison was performed between the limiting current densities of a cracked and a negligible-cracked CL when 5%-O 2 nitrox and 5%-O 2 heliox were used separately in the cathode stream. The ratio of the Fickian (molecular) diffusion coefficient of heliox to nitrox is given [21] as: The limiting current density is defined [22] as, D O2 Heliox (3.1) 2.8 DO 2 Nitrox numerical i l nfd eff C (3.2) 69

90 where n is the equivalent electrons per mole of reactant, F is the charge carried on one equivalent mole, C is the concentration of the reactant at the boundary of the flow channel, and is the distance to the electrode surface from the flow channel boundary. It was assumed that and C are constant because of the efforts made to eliminate the experimental uncertainty in the data collection. Therefore, the ratio of the limiting current density of the 5%-O 2 heliox case and 5%-O 2 nitrox case can be expressed as an experimental ratio of the diffusion coefficients: i l, O2 Heliox DO 2 Heliox (3.3) i l, O Nitrox 2 D O Nitrox experimental The ratio of the limiting current densities noted in Eq. (3.3) has been determined experimentally and compared to the theoretical Fickian diffusion coefficient ratios given in Eq. (3.1). Figure 3.3 and Figure 3.4 show the polarization behavior of the cell for these two cases, in which the arrow at the high current density region highlights the difference between the limiting current of the 5%-O 2 nitrox and 5%-O 2 heliox. The change in limiting current densities is due to the change in the inert gas from 5%-O 2 nitrox to 5%- O 2 heliox. Table 3-1 tabulates the experimentally calculated parameters (Eq. (3.3)) divided by the numerical results (Eq. (3.1)) for dry and wet conditions. These results in Table 3-1 indicate that regardless of moisture content, the percentage of molecular diffusion in the cracked CL was 0.5% higher than that of the negligible-cracked CL, likely resulting from the comparatively large voids of the cracks. Furthermore, the percent molecular diffusion of the dry condition cases (both cracked and negligiblecracked CL) was 0.6% higher than the respective wet conditions, indicating that water fills the larger voids in the CL, forcing more non-fickian diffusion to occur. In the 70 2

91 IR Compensated Voltage (V) IR Compensated Voltage (V) negligible-crack CL case, the diffusion will tend slightly toward non-fickian diffusion because of the absence of the larger cracks in the CL Cell Temp: 75 o C Dry Condition: 70%/50% Inlet RH Pressure: 1 atm CRK Dry 5% He CRK Dry 5% N2 NCRK Dry 5% He NCRK Dry 5% N Current Density (A/cm 2 ) Figure 3.3. A comparison of the experimentally determined limiting current densities of a cracked CL (CRK) and a negligible-cracked CL (NCRK) at dry conditions for 5%-O 2 nitrox ( 5% N2 in legend) to 5%-O 2 heliox ( 5% He ) Cell Temp: 75 o C Wet Condition: 100%/100% Inlet RH Pressure: 1 atm CRK Wet 5% He CRK Wet 5% N2 NCRK Wet 5% He NCRK Wet 5% N Current Density (A/cm 2 ) Figure 3.4. A comparison of the experimentally determined limiting current densities of a cracked CL (CRK) and a negligible-cracked CL (NCRK) at wet conditions. 71

92 Table 3-1. The experimentally calculated parameters (Eq. (3.3)) divided by the numerical results (Eq. (3.1)) for dry and wet conditions shows the percentage of Fickian diffusion observed. i l, O2 Heliox l, O2 Nitrox Dry Condition D O Heliox 2 i D O Nitrox 2 Molecular Diffusion Wet Condition Molecular Diffusion Cracked CL 42.8% 42.2% Negligible-Cracked CL 42.3% 41.7% CRACKED AND NEGLIGIBLE-CRACKED CL PERFORMANCE IN WET AND DRY CONDITIONS The second phase of this study was aimed at investigating the effects of CL surface cracks on the cell performance under wet and dry conditions. Figure 3.5 and Figure 3.6 show a comparison of wet and dry conditions for cracked and negligible-cracked CL cases, respectively. Based on the experimental data, three distinct regions, as summarized below (also graphically displayed in Figure 3.5 and Figure 3.6), can be identified for comparing the wet and dry conditions. Region I: Dry performance > wet performance Region II: Dry performance < wet performance Region III: Dry performance > wet performance Several factors listed in Table 3-2 can be the cause of these three regions. Depending on the current density, these factors may have varying degrees of relative dominance in the overall performance of the fuel cell. It is important to note that IR-compensated voltage does not account significant portions of the ionic resistance in the CL, and therefore is not accounted for in Figure 3.5 and Figure

93 IR Compensated Voltage (V) IR Compensated Voltage (V) Cell Temp: 75 o C Wet Condition: 100%/100% Inlet RH Dry Condition: 70%/50% Inlet RH Pressure: 1 atm NCRK Wet 5% N2 NCRK Dry 5% N I II II III Current Density (A/cm 2 ) Figure 3.5. A comparison of the polarization curves of wet and dry conditions for negligible-cracked CL (NCRK) cases. The performance was observed to have three distinct regions in which the dry or wet condition outperformed the other. The onset of Region II was observed began at approximately 0.26 A/cm Cell Temp: 75 o C Wet Condition: 100%/100% Inlet RH Dry Condition: 70%/50% Inlet RH Pressure: 1 atm CRK Wet 5% N2 CRK Dry 5% N I II II III Current Density (A/cm 2 ) Figure 3.6. A comparison of the polarization curves of wet and dry conditions for cracked CL (CRK) cases. The performance was observed to have three distinct regions in which the dry or wet condition outperformed the other. The onset of Region II began at approximately 0.2 A/cm 2. 73

94 Table 3-2. A summary of possible factors involved to explain the behavior of the dry and wet condition performance (Figure 3.5 and Figure 3.6). Factors affecting cell performance Dry condition Comparison Wet condition Oxygen diffusion Oxygen concentration Proton resistance D O2,dry (enhanced performance) y O2,dry (enhanced performance) R H+,dry (worse performance) > D O2,wet > y O2,wet > R H+,wet Reaction location Approach membrane -- Most of CL At low current, the oxygen reduction reaction tends to occur close to the membrane [22], giving importance to how readily oxygen can pass through the porous CL. Region I (low current density) of Figure 3.5 and Figure 3.6 show increased voltages for the dry conditions, both with the cracked and negligible-cracked CL. In dry conditions, both the oxygen diffusion coefficient and the oxygen concentration will be greater, which could lead to the observed performance increase in Region I. In Region II, the higher current draw may have forced more of the CL thickness to be used, which will increase the relative significance of the protonic resistance compared to the oxygen diffusion or concentration effect. This would yield a reduced performance in dry conditions due to the increased need for proton conduction through the CL. Finally, Region III shows that the dry condition displayed enhanced performance over the wet condition. Since oxygen transport may be dominant through the diffusion media and CL, liquid water obstructing the oxygen s path to the reaction site may have caused a decrease in the wet condition performance. 74

95 Quantitatively, Table 3-3 depicts the percentage by which the dry condition outperformed the wet condition at selected current densities. Negative values (Region II) reflect the wet outperforming the dry. Inspecting Figure 3.7, it is clear that at high current density, both the cracked and the negligible-cracked CL perform nearly identically. Therefore, when the Region III results in Table 3-3 are analyzed, the wet condition can serve as a reference. It is concluded, therefore, that cracks on the CL surface cause a 16% increase in voltage (38.2% minus 22.1%) in dry conditions compared to negligiblecracked CL. Thus, it might be deduced that the cracks in the CL augment the availability of oxygen transport to the reaction sites in dry conditions. Table 3-3. A quantitative summary of the performance of 5%-O 2 nitrox dry condition compared to the wet condition in the three regions for both cracked and negligiblecracked CLs. Region Performance Summary Cracked CL Negligible-Cracked CL Region I (0.025 A/cm 2 ) Dry > Wet 1.6% 2.5% Region II (0.40 A/cm 2 ) Dry < Wet -2.5% -4.7% Region III (0.48 A/cm 2 ) Dry > Wet 38.2% 22.1% CRACKED VERSUS NEGLIGIBLE-CRACKED CL PERFORMANCE The main emphasis in the final phase of this study was placed on the effects that CL surface cracks have on the overall cell performance. This was achieved by comparing the polarization curves of cracked and negligible-cracked CLs at various conditions. Each condition tested showed the same trend as in Figure 3.7, where the cracked CL underperformed the negligible-cracked CL in the lower current density regions by 75

96 IR Compensated Voltage (V) approximately 3%, but the cracked CL performed nearly identically to the negligiblecracked CL in the high current density region (over ~0.4 A/cm 2 ). The lower performance in the low current density region can be attributed to the reduction in active area due to the cracks. The equal or enhanced performance of the cracked CL to the negligiblecracked CL in the high current density region is possibly due to enhanced water management with a cracked CL, i.e. by reducing a thin film that blocks reactants or by enhancing water removal from the reaction sites Cell Temp: 75 o C Wet Condition: 100%/100% Inlet RH Dry Condition: 70%/50% Inlet RH Pressure: 1 atm CRK Wet 5% N2 NCRK Wet 5% N Current Density (A/cm 2 ) Figure 3.7. A polarization curve comparing a cracked (CRK) and negligible-cracked (NCRK) CL under wet conditions. 3.4 CONCLUSIONS The effects of surface irregularities and CL cracks on the mass and electronic transport of PEFCs were examined by comparing a CL with surface cracks and one with negligible cracking at various operating conditions. It was observed that the presence of cracks on the CL surface yields a slight increase (0.5%) in Fickian diffusion, and that the 76

97 increase in Fickian diffusion was the same for both wet (100%/100% anode/cathode relative humidity) and dry conditions (70%/50% anode/cathode relative humidity). Secondly, the performance curves of cracked and negligible-cracked CLs in wet versus dry conditions yielded three distinct regions: Region I: low current density regions (between 0 and 0.2 A/cm 2 ) showed the dry condition outperformed the wet condition by approximately 2% in the cracked CL case; Region II: the moderate current density region (between ~0.2 and 0.45 A/cm 2 ) showed the wet condition had up to 2.5% higher voltage than the dry condition for the same cracked CL case; and Region III: the high current density region (greater than 0.45 A/cm 2 ) showed the dry condition outperformed the wet condition by 38.2% for the cracked CL. From these results, it is concluded that the cracked CL displayed significant improvement over the negligible-cracked CL in dry conditions at high current density. Through this study, it became clear that the presence of surface irregularities may in fact enhance the overall performance at certain operating conditions. This result inspired further research into modified interfaces and structures, which will be discussed at length in the following chapters. 77

98 REFERENCES [1] B. Avasarala, P. Haldar, J. Power Sources, 188 (2009) [2] Y. Zhou, G. Lin, A.J. Shih, S.J. Hu, J. Power Sources, 163 (2007) [3] Z. Wu, Y. Zhou, G. Lin, S. Wang, S.J. Hu, J. Power Sources, 182 (2008) [4] P. Zhou, C.W. Wu, G.J. Ma, J. Power Sources, 159 (2006) [5] R.H.J. Blunk, D.J. Lisi, Y.-E. Yoo, I. Charles L. Tucker, AlChE J., 49 (2003) [6] A. Kraytsberg, M. Auinat, Y. Ein-Eli, J. Power Sources, 164 (2007) [7] W. Liu, K. Ruth, G. Rusch, J. New Mater. Electrochem. Syst., 4 (2001) [8] Y. Li, J.K. Quincy, S.W. Case, M.W. Ellis, D.A. Dillard, Y.-H. Lai, M.K. Budinski, C.S. Gittleman, J. Power Sources, 185 (2008) [9] Y.-H. Lai, C.K. Mittelsteadt, C.S. Gittleman, D.A. Dillard, ASME Conference Proceedings, 2005 (2005) [10] P. Zhou, C.W. Wu, J. Power Sources, 170 (2007) [11] T. Hottinen, O. Himanen, S. Karvonen, I. Nitta, J. Power Sources, 171 (2007) [12] F.E. Hizir, S.O. Ural, E.C. Kumbur, M.M. Mench, J. Power Sources, 195 (2010) [13] I. Nitta, O. Himanen, M. Mikkola, Electrochem. Commun., 10 (2008) [14] R. Makharia, M.F. Mathias, D.R. Baker, J. Electrochem. Soc., 152 (2005) A970- A977. [15] J. Ge, A. Higier, H. Liu, J. Power Sources, 159 (2006) [16] I. Nitta, T. Hottinen, O. Himanen, M. Mikkola, J. Power Sources, 171 (2007) [17] A.Z. Weber, J. Newman, J. Electrochem. Soc., 153 (2006) A2205-A2214. [18] M. Coppo, N.P. Siegel, M.R.v. Spakovsky, J. Power Sources, 159 (2006)

99 [19] L. Wang, A. Husar, T. Zhou, H. Liu, Int. J. Hydrogen Energy, 28 (2003) [20] A. Mughal, X. Li, Int. J. Environ. Stud., 63 (2006) [21] T.M. Company, in, Toyota Motor Company, [22] M.M. Mench, Fuel Cell Engines, John Wiley & Sons, Inc., Hoboken,

100 4. LASER PERFORATED FUEL CELL DIFFUSION MEDIA: RELATED CHANGES IN PERFORMANCE AND WATER CONTENT Chapter 3 investigated catalyst layer cracking and the role of interfacial changes on performance and transport. The results of that study namely the increase in performance when a cracked CL was introduced in low humidity conditions encouraged the investigation of intentionally modifying the diffusion media as another form of interfacial modification. In this study, cathode-side, bi-layered diffusion media (DM) samples with micro-porous layer were perforated with 300 µm laser-cut holes (covering 15% of the surface area in a homogenous pattern) using a ytterbium fiber laser to investigate the effect of structural changes on the gas and water transport. Under reduced humidity conditions (50% inlet relative humidity on the anode and cathode), the perforated DM were observed to increase the potential by an average of 6% for current densities ranging from 0.2 to 1.4 A cm -2. However, the perforated DM showed reduced performance for current densities greater than 1.4 A cm -2 and at all currents under high-humidity conditions. Neutron radiography experiments were also performed to understand the changes in liquid water retention characteristics of DM due to the laser perforations. Significant water accumulation and water redistribution was observed in the perforated DM, which helps explain the observed performance behavior. The results indicate that the perforations act as water pooling and possible channeling locations, which significantly alter the water condensation, storage, and transport scheme within the fuel cell. These observations suggest that proper tailoring of fuel cell DM possesses significant potential to enable fuel cell operations with reduce liquid overhead and high performance. 80

101 4.1 INTRODUCTION Water management is a deterministic factor in performance, stability, and durability of polymer electrolyte fuel cells (PEFCs). The major limiting aspect of the water management arises from multi-phase flow effects, which represent a major bottleneck, not only for efficient fuel cell operation, but also for operational stability and durability under prolonged operations. From a water management perspective, the porous diffusion medium (DM) is a critical component, as it functions to deliver reactant to, and products away from, the catalyst layer (CL) [1,2]. Since the transport in PEFCs involves both liquid and gas-phases, the fuel cell DM serves a critical role by enabling capillary transport of liquid away from the electrodes to avoid severe performance losses caused by flooding. Therefore, proper tailoring of fuel cell DM is critical to establish an optimal water management during PEFC operation. The addition of a microporous layer (MPL) between the DM and CL has been shown to significantly improve cell performance [e.g., 2-8]. In general, the MPL provides more smooth contact between DM and CL, reducing the electrical resistance between these layers, while serving as a protective layer to prevent the coarse DM fibers from puncturing through the CL and membrane. Some studies suggest that due to the small pore size, the MPL acts as a capillary barrier, directing the liquid water flow towards anode side, and helping membrane to maintain high hydration level [9, 10]. However, recent experimental studies have been inconclusive on this topic [3-11]. For instance, recent studies revealed the existence of deep cracks and other surface defects on the MPL surface [4, 12-14], which are argued to serve as preferential pathways for water removal 81

102 from the cathode CL towards the MPL DM interface [4, 12]. A recent study by Owejan et al. [15] from GM also confirms this finding, suggesting that the primary role of MPL is to prevent condensed water from pooling at the CL rather than enhancing the capillarydriven transport towards anode side. The nature of the interface between the MPL and CL is also shown to be critically important to avoid interfacial delamination during freeze/thaw cycling [16, 17]. Polytetrafluoroethylene (PTFE) in the CL, MPL, and DM has been shown to have a positive impact on fuel cell performance, and several optimization studies have been conducted (e.g., [18-21]). The addition of PTFE was shown to create a mixture of hydrophilic and hydrophobic pores [2, 21-24]. A majority of the efforts in this area were performed to analyze the effects of PTFE loading and developing methods to tailor and optimize the ratio of hydrophilic and hydrophobic pores in the fuel cell DM [2,19,20,25-29]. From a transport perspective, hydrophilic pores are theoretically desirable to aid in wicking water away from reaction sites in order to minimize flooding and reduce liquid film generation; however, they can also act as water pooling locations that promote high DM saturation, which could foster flooding related losses and durability concerns. Hydrophobic pores, on the other hand, have been shown to reduce the DM saturation [4, 6, 30]. Too many hydrophobic pores, however, can create a strong hydrophobic layer with high resistance to liquid water, promoting local water buildups at the MPL CL interface, which promote poor electrical contact along the MPL CL interface [19,31] and flooding related performance losses. Additionally, excessive PTFE content in the DM can impede evaporative removal of liquid water at shut down [32, 33], causing retention of significant water inside the cell. Clearly, a proper mixture of hydrophilic and 82

103 hydrophobic pores is necessary for optimal water transport with the fuel cell. It should be noted that beyond a simple balance between hydrophobic and hydrophilic pores, the connectivity and tortuosity is also critical to provide desirable transport of liquid and gas phases of product and reactant. It is worthwhile to briefly discuss modes of liquid transport within DM. Especially at higher current, where heat generation and temperature gradients are more significant, phase-change-induced (PCI) flow is a key mode of water transport from the electrodes. In PCI flow, liquid water near the warmer CL evaporates, moves through the DM by diffusion, and then condenses in a colder location near or in the channel. PCI flow has been shown to be a significant mode of liquid water and heat transport, especially when bi-layered, PTFE-containing DM is present [34-36]. Another mode of transport within the porous DM is capillary flow [1, 21-23, 37]. Due to the small pore sizes, gravitational and viscous forces are relatively small compared to capillary forces. Depending on the operating conditions and materials used, capillary and PCI water transport mechanisms have varying degrees of dominance. Another important factor affecting the water distribution inside the cell is the interfacial regions, such as that between MPL and CL. The impact of the MPL CL interface morphology on water management of the fuel cell is not yet well-understood. Recent visualization studies have indicated that interfacial effects may be responsible for altering the saturation profile and water storage near the CL MPL and MPL DM interfaces [12-14, 38, 39]. To date, most fuel cell models assume the interfacial regions to be infinitely thin, having a perfect contact and homogenous in-plane properties [36,37]. 83

104 As a result, they predict a sharp discontinuity at these interfaces due to the change in pore structure and hydrophobicity. However, recent experimental evidences obtained using x- ray tomography [38] and neutron imaging [39] have shown that this assumption can be in error. In a study by Hartnig et al. [38], significant water accumulation near the MPL surface is observed by using synchrotron x-ray. Similarly, Turhan et al. used neutron radiography and observed an inverted liquid-saturation jump distribution around these interfaces, which was attributed to water storage in MPL cracks and along the interface region [39]. Other studies [12-15] have also shown that the interfacial morphology of the CL and MPL (i.e., surface cracking and irregularities) can lead to significant liquid water storage, altering the saturation values observed with neutron imaging. Obtaining through-plane water distribution profiles has become increasingly important for understanding the water distribution in DM and improving the fidelity of models. More accurate measurement of through-plane liquid distribution profiles has recently become possible due to enhancements in high-resolution neutron radiography (NR) [40]. With the recent advancements in NR, many researchers have been able to better understand both liquid- and vapor-phase water transport in the DM. Hickner et al. [41] was able to investigate MEA hydration under different gas feed flow rates and current densities. They found that evaporative driven flow is critical for effective removal water from the cell. The results were further used to develop a model that rationalized observed water transport trends [42]. Other studies investigated passive methods for altering water distributions by customizing the channel wall hydrophobicity and manipulating PCI flow through controlled temperature gradients [39, 43]. 84

105 To date, there are contradictory hypotheses in the pertinent literature that portray the water transport mechanisms differently. Some suggest that capillary transport occurs in a tree-like structure, while an opposing concept proposes a fingering and channeling mechanism. In a study by Benziger and coworkers [30], it was proposed that the ideal DM should consist of a few large pores that transport and/or store liquid water (depending on hydrophobicity of the pores), and numerous small pores (ca. 40 µm) that are relatively free of liquid water and serve to transport the reactants. Gostick et al. [4] also suggested that introducing large holes for water passage may improve the effectiveness of the MPL. The general concept is to create a DM with ideal proportions of hydrophilic and hydrophobic pores in order to avoid excess water while preventing membrane dryout and oversaturation of the DM. Two groups [44-46] have published work studying the predictions of Refs. [4,30]. Studies have investigated the effect of introducing large-diameter holes into the DM, both along the channel path [44,46], and under a channel [46]. Both [45,46] have investigated cell stability changes with holes. In this chapter, in-situ measurements of unaltered and laser-perforated fuel cell DM were performed to examine the effects of tailored internal structure and modified pore characteristics of the DM on the performance and transport characteristics of PEFCs. Commercial DM samples coated with MPL were perforated with 300-µm laser-cut holes using a ytterbium fiber laser, and then subjected to extensive experimental evaluation. Similar studies in the literature [4, 30] suggest that perforations or enhanced pore distribution could improve the effectiveness of the DM. In continuing along this trajectory, this study offers benchmark experimental data to help understand the dominance and role of PCI and capillary transport modes on the cell performance, and 85

106 indicates that optimization of the DM structure is possible for reduced liquid overhead, but requires a detailed understanding of the complex water and heat transport phenomena inside the cell. 4.2 EXPERIMENTAL METHODS AND SETUP LASER PERFORATION A ytterbium fiber laser was used to introduce evenly-spaced through-holes, each with a diameter of 300 μm, to the cathode-side DM and MPL of a PEFC, covering approximately 15% of the total geometric surface area (see Figure 4.1). Laser power level, pulse duration, and the distance between the last plano-convex lens and the DM surface were identified as the key parameters that affect the diameter and depth of the laser-cut holes [47]. Therefore, these key perforation parameters were optimized in order to achieve the desired perforation dimensions, and are listed in Table 4-1. Table 4-1. Parametric settings for laser perforations. Cell Power (Watt) Duration (µs) Lens to DM Distance (mm) D hole (mm) N hole X (mm) Percent Area with Holes (%) 5 cm 2 PSU 17.2 cm 2 NIST Laser perforation has the advantage of physically burning carbon/binder material away, instead of simply translating material to the surrounding areas, which may happen with micromachining techniques [44]. The use of a laser, however, causes a heat-affected 86

107 zone (HAZ), which is the dark region surrounding the perforation shown in Figure 4.1a [47]. The diameter of the HAZ is primarily dependent on the key laser perforation parameters previously described. The heat-affected zone was analyzed using an energy dispersive x-ray spectroscopy (EDS) and an environmental scanning electron microscope (ESEM) by Hizir [47]. The EDS showed only a dominant carbon peak in the heataffected zone (Figure 4.1c), indicating the absence of PTFE. The regions beyond the HAZ displayed carbon and fluorine peaks (Figure 4.1d), indicating the presence of PTFE. The ESEM also showed a marked decrease in contact angle of water droplets in the heataffected zone, indicating that the HAZ is more hydrophilic than the non-affected portion of the MPL surface. Therefore, the 300-μm perforations are expected to yield both a low capillary pressure zone (due to the comparatively large pore size of the perforation) and a hydrophilic region surrounding the perforation (due to the removal of PTFE), both of which are expected to promote water accumulation in, and near, these areas [47]. It is worthwhile to note that the laser perforation dimensions and pattern were chosen in order to mimic a pattern that could be easily implemented into industrial design. 87

108 x x x x 300 μm Carbon (a) (b) Carbon Fluorine kev (c) (d) kev Figure 4.1. (a) SEM image of the MPL surface with laser perforations. A heat-affected zone and 300-μm perforations are highlighted as key modifications due to the laser treatment, whereas the MPL cracks occur on both laser-treated and virgin samples; (b) Schematic of laser perforations, where x = 0.67 mm for polarization testing and 0.97 mm for NR testing; (c) An EDS spectrum of MPL surface within the heat affected zone. There is no fluorine peak, indicating the absence of PTFE; (d) An EDS spectrum of the virgin MPL surface, showing dominant carbon and fluorine peaks IN-SITU POLARIZATION TESTING In this chapter, SGL 10BB series DM samples coated with MPL (SIGRACET Gas Diffusion Layer ) and commercially available membrane electrode assemblies were 88

109 used. The anode-side DM in each experiment remained unaltered, while the cathode DM used was varied as described. In-situ testing was performed with a custom designed 5 cm 2 double serpentine fuel cell using a Scriber Associates 850C Fuel Cell Test System. After careful assembly of the fuel cell components, each cell underwent a common breakin procedure involving 100% inlet relative humidity flow at the anode and cathode. The cell was cycled from open circuit voltage to 0.6 V and then 0.4 V at 1 minute hold intervals. The cell was considered ready for experimentation when the current at 0.6 and 0.4 V no longer changes significantly with time. Regular experiments were conducted with over-humidified (120%/120% inlet RH anode/cathode) and low-humidity (50%/50% A/C) inlet reactant gases both at high (75 o C) and low (50 o C) temperatures. Nitrox (dry air) and heliox cathode gases, which contain N 2 and He as the inert cathode gas mixed with 21% O 2 in a dry state, respectively, were used. Flow rates were held constant at 139 and 332 sccm for the anode and the cathode at the atmospheric exit pressure, respectively. Typically, at elevated temperatures, the rate of liquid water evaporation from the PEFC is higher, and gas-phase transport is more dominant. Conversely, lower temperature operation is expected to reduce evaporation rates, causing liquid water effects to be more important. The ratio of the limiting current of a cell with nitrox and heliox cathode gases can be compared to the ratio of the nitrogen and helium diffusivities in oxygen. This comparison can be used to quantify the relative dominance of molecular diffusion over other forms of diffusion, such as Knudsen diffusion. This approach is described in Ref. [48, 49] and employed in this study to elucidate the gas- and liquid- 89

110 phase transport phenomena associated with the introduction of laser perforations in the DM. To minimize the effects of hysteresis and ensure repeatability among tests, a preconditioning cycle, which includes 10 minutes of load cycling followed by 10 minutes of constant voltage, was conducted before each test. Furthermore, each cell was operated under galvanostatic mode for the duration of the polarization testing to ensure constant water generation in these tests. Each galvanostatic step was conducted for 3 minutes to achieve quasi-steady state conditions. To approach the limiting current under each condition, a 1-minute potentiostatic step at 0.2 V was added to each polarization test. In some cases, especially those with perforated cathode-side DM, severe flooding occurred at the final step, and the data could not be obtained. For comparison purposes, the cell voltage was compensated with the high frequency resistance (HFR) measured during testing in order to eliminate the losses due to membrane hydration differences. Finally, it should be noted that the error bars were added to the polarization data to indicate the level of repeatability from test to test. For all cases, two to four polarization sequences were conducted for the measurements being recorded. The upper and lower error bars given in the figures represent the highest and lowest average voltage response measured for any given current density, respectively NEUTRON RADIOGRAPHY Neutron radiography (NR) testing was conducted to observe and quantify the changes in water management characteristics of virgin and perforated DM samples. Imaging was performed at the National Institute of Standards and Technology (NIST) in Gaithersburg, 90

111 Maryland. The high-resolution imaging system at NIST has a resolution of ca. 13 μm, and the captured images have a nominal pixel pitch of 10 μm. The fuel cell was imaged in the through-plane direction, and a schematic of the cell and perforated DM is shown in Figure 4.2. In the NR experiments, the same cell components (SGL 10 BB series DM with MPL and commercially available membrane electrode assemblies) were used; however, the NR cell had an active area of 17.2 cm 2 with an in-plane to through-plane length ratio of ca. 0.3 for improved resolution. Perforated DM samples were prepared using the same technique described earlier. The samples used in NR testing had 300-μm diameter perforations with a center-to-center distance of 0.97 mm, which constituted ca. 7% of the geometric surface area of the DM samples. After appropriate break-in procedures, each cell (virgin and perforated DM) was operated at a constant current for 30 minutes at 65 o C under an inlet relative humidity of 100%/100% and 50%/50% (anode/cathode) at a stoichiometry of 2/2 hydrogen/nitrox (air). In high humidity cases, the current densities of 0.2 A cm -2 and 1.2 A cm -2 were selected, whereas the current densities of 0.2 A cm -2 and 1.7 A cm -2 were chosen for the low humidity case. During this 30-minute interval, neutron images and performance data were collected under low and high humidity conditions to characterize the performance and quantify the water content differences between the virgin and perforated DM cases. 91

112 Beam direction 1 mm Figure 4.2. Rendering of the neutron radiography test cell with expanded view of perforated DM schematic and SEM image. 4.3 RESULTS AND DISCUSSION The following sections describe the results of low and high humidity testing of perforated and unaltered DM cases. The polarization testing was conducted at the Pennsylvania State University, and the neutron imaging tests were performed at NIST (Gaithersburg, MD). A fuel cell fixture with DM that has perforations corresponding to 15% geometric area of the DM was used for the polarization testing, while a fuel cell with DM that has perforations corresponding to 7% geometric area of the DM was utilized for the neutron imaging testing. It was observed that the measured performance trends of the 15% and 7% perforated DM show very similar behavior, indicating that the similar physical phenomena exist in the 15% and 7% perforated DM cases POLARIZATION TESTING AT LOW HUMIDITY CONDITIONS In these tests, cell fixtures with a perforated cathode-side DM and with a virgin (unaltered) cathode-side DM were exposed to low humidity conditions (50% relative 92

113 humidity) with nitrox (21% O 2 and 79% N 2 ), which was used as an inlet cathode gas. Figure 4.3 shows the IR-compensated polarization data of these two cases obtained at low humidity condition. As shown in Figure 4.3, the perforated DM case is observed to exhibit an average of 38 mv higher potential for the current densities ranging from 0.1 to 1.4 A cm -2. Above 1.4 A cm -2, however, the losses are observed to rapidly increase, and the cell could not sustain quasi-steady performance beyond 1.6 A cm -2. A maximum current density of 1.79 A cm -2 was recorded at 0.2 V for the perforated DM case, whereas the cell with unaltered DM was observed to draw a 20% higher maximum current density of 2.15 A cm -2 at 0.2 V. IR Compensated Voltage (V) higher performance for perforated DM mass-transport dominated performance for perforated DM HFR (mohm) 75 o C 50% RH A/C Current Density (A cm -2 ) Figure 4.3. Performance data for 75 o C, 50% relative humidity, nitrox conditions with virgin DM ( ), 15% 300-μm perforated cathode-side DM ( ), and corresponding 93

114 HFR values for the virgin ( ) and perforated ( ) cases. Perforated DM shows on average 38 mv higher potential in current densities less than 1.4 A cm -2 but failed to achieve steady state at current densities greater than 1.4 A cm -2. While the perforated DM shows an increase in performance in the lower current regions, the significant performance drop in the higher current regions (i.e., greater than 1.4 A cm -2 ) indicates the existence of increased mass transport losses due to the perforations in the DM (Figure 4.3). Beyond 1.6 A cm -2, the performance of the cell with perforated DM is highly unstable. While comparable error bars between the two cells are seen at 1.6 A cm -2, the virgin cell was observed to quickly recover the mass transport losses and regain stability as it effectively removes liquid water build-ups from reaction sites. The observed difference in performance of these two cases at low humidity conditions can be attributed to the altered mass transport caused by the perforations. At low humidity, the redirection of water into the perforated regions appears to enhance oxygen transport to the catalyst, while at higher current, flooding occurs as a result of the additional perforations. In terms of HFR measurements, the two cases perforated and virgin DM interestingly did not show any particular correlation at low humidity conditions; that is, neither case showed consistently higher or lower resistance values over each other within the entire current density range. This indicates that the increase in electron path around the perforations does not cause a substantial increase in the overall measured cell resistance, and the bulk ohmic resistance in the membrane is not significantly affected by the presence/absence of DM perforations for the tested conditions. It should be noted that 94

115 the HFR will mostly not record changes in the CL ionic conductivity, however, so dryout of the CL is not indicated by the HFR response. The similar cell resistances observed for these two cases can be explained by the relatively high in-plane electronic conductivity of the DM and CL materials [50], which appears to compensate the additional path of resistance introduced by perforations. In terms of membrane resistance, similar values can be explained through analyzing the changes in the water transport patterns inside the cell due to the addition of perforations. In general, the MPL serves to provide smooth contact between the DM and the CL, and can act as a capillary barrier to avoid excess water accumulation to the pores of the DM [8,10,37]. Since the perforations effectively eliminate a certain portion of this barrier, one can argue that water would be less readily retained in the perforated case, leading to membrane dehydration, especially under low humidity conditions. This was not observed in this study, however, may indeed occur if even lower inlet humidity conditions are used. However, the hydrophilic and low-capillary pressure nature of the perforated regions may enable a significant amount of water storage, especially at high current conditions. Once fully saturated, the water stored in the perforated regions can feed the CL and membrane, enabling it to retain similar hydration level as compared to the unaltered DM case, which would yield comparable ohmic resistance values, as observed in the HFR testing (Figure 4.3) NEUTRON RADIOGRAPHY TESTING VIRGIN VERSUS PERFORATED DM To further analyze the potential impact of perforations on the cell performance, neutron imaging tests were performed under low humidity conditions. Interestingly, for 95

116 low current densities, neutron images indicate negligible differences in total water amount between the virgin and perforated DM, as shown in Figure 4.4a. While the total water amount in the cell is approximately the same, the perforations redistribute the water (discussed later), which may be the cause for the increased performance. At lower current densities, the perforations may a) act as water pooling locations, thus opening previously filled pores for gas transport, and/or b) serve as large-diameter gas conduits, depending on the amount of liquid water present. Both options could increase the gas access to the CL, which would increase the species concentration at the electrode, hence reducing the polarization and increasing performance. At higher current, however, the perforated DM is observed to have significantly higher liquid water mass in all locations, most likely due to the increased water pooling in the perforations sites. The measured water mass values at high current densities also show distinct water distribution trends, which are plotted in Figure 4.4 and Figure 4.5. These results were obtained by extracting water thickness measurements from pixels in the membrane/dm region of the cell in both the in-plane and through-plane direction of the fuel cell. In Figure 4.4, x = 0 mm corresponds to a row of pixels normal to a centrally-located land channel interface, and x = 7 mm corresponds to the length of four lands and three channels (each 1 mm wide). The secondary x-axis is included to clarify the locations from which water mass values were extracted. In both low and high current density conditions, the periodic peaks in the perforated DM are observed at locations corresponding to x = 0.56, 1.53, 2.48, 3.37, 4.22, 5.16, 6.07, 6.89 mm (Figure 4.4). These periodic peaks, which are not observed in the virgin DM case, indicate the possible pooling locations induced by the perforations, which had a center-to-center distance of 96

117 ca mm. In a recent study by Gerteisen et al. [45], it has been suggested that perforations may create paths of least resistance for water to flow through the DM towards the channels. These preferential pathways would correspondingly impede the capillary fingering transport mode that typically takes place in the hydrophilic pore network of DM [51]. The present neutron imaging results namely the periodic peaks observed at perforation locations experimentally verify this hypothesis. Furthermore, the local minima on either side of the peaks in the perforated DM show lower water mass values than the ones at the corresponding location in the virgin DM case, indicating that the water has translocated from the bulk DM into the hydrophilic perforation regions. Therefore, it can be hypothesized that under low-humidity operating conditions, the perforated DM exhibits the potential benefit of dictating the preferred path of liquid water in the water management process. To the best of the authors knowledge, this is the first reported experimental validation of this phenomenon. 97

118 50 L C L C L C L 65 o C 50% RH A/C Water mass (mg cm -3 ) X1 X Distance (mm) (a) 50 L C L C L C L 65 o C 50% RH A/C 40 X1 X2 Water Mass (mg cm -3 ) Distance (mm) (b) Figure 4.4. Neutron data for water mass per volume distributions in the in-plane direction at low inlet relative humidity condition. L represents land, and C represents channel. (a) Water distributions for virgin DM ( ) and perforated DM ( ) under low current density (0.2 A cm -2 ) testing operation; (b) Water distributions for virgin DM ( ) and perforated DM ( ) under high current density (1.7 A cm -2 ) testing operation. A reduced number of data points are shown to improve clarity. 98

119 When the data in Figure 4.4b are integrated across the x-axis, the results show that the perforated DM contains 46% more water mass than the virgin DM at 1.7 A cm -2. From Figure 4.5, it is clear that this additional 46% water mass is a result of the accumulated water in the perforated regions at the cathode. The water accumulation into perforated regions can be explained by analyzing the water transport modes inside the cell. It is hypothesized that the prime modes of liquid water removal can be altered by the presence of the perforations. For instance, phase-change-induced (PCI) flow is considered as a prominent mode of water transport within the PEFCs. It was shown [34] that the PCI flows tend to increase with hydrophobic content in the DM regardless of the temperature gradient, since PCI flow is dependent on the presence of a vapor zone. Since the laser pyrolyzes the PTFE coatings at the vicinity of the perforated regions (i.e. making these regions more hydrophilic), it is very likely that the water removal by PCI flow would become less favorable in the perforated DM. In addition, it is very likely that the capillary flow can be inhibited by the perforations, such that introducing large perforations (300- μm, as in this study) significantly decreases the capillary pressure due to the increased pore size, impeding the removal of liquid water via capillary-driven transport within the cell. As a result, water accumulation is more likely to dominate the performance of the perforated DM cell at higher current operations. 99

120 50 Anode DM MPL / MEA Cathode DM 65 o C 50% RH A/C Water mass (mg cm -3 ) Distance (mm) Figure 4.5. Neutron data for water mass per volume distributions in the through-plane direction at low inlet relative humidity condition. Virgin case at high current ( ), virgin case at low current ( ), perforated DM case at high current ( ), perforated DM case at low current ( ). A reduced number of data points are shown to improve clarity NEUTRON RADIOGRAPHY TESTING HIGH VERSUS LOW CURRENT AT LOW HUMIDITY OPERATIONS In addition to the side-by-side comparison of the virgin and perforated DM, it is noteworthy to compare how each diffusion medium variety responds to the increase in water production from low to high current. At low current (Figure 4.4a and Figure 4.5), 100

121 the perforated DM only contains 2% more water mass than the virgin DM. Since relatively little water is entering the cell via the partially-humidified gases, and relatively little water is produced at the CL at low current, the difference between the amount of water stored in the cell for the virgin and perforated case is expected to be minimal. In other words, the amount of liquid water in the DM is not significant at these conditions, thus the water management differences between virgin and perforated DM are otherwise insignificant. In terms of high current densities, the perforated DM contains 33% more water at high current than at low current, whereas the virgin DM only shows a 4% increase from low to high current. It is reported that at higher current, the PCI flow plays a critical role in removal of liquid water from the cell [42]. Thus, since the PCI flow dominance is severely reduced with the perforated DM (due to the loss of hydrophobic binder and the introduction of artificial flow conduits), it is very likely that there would be higher water mass accumulation in the perforated DM case as compared to the virgin DM under the high current densities. Figure 4.5 clearly supports this argument, showing higher water accumulation in the perforated DM than the virgin DM case POLARIZATION TESTING AT HIGH HUMIDITY CONDITIONS Figure 4.6 shows the high humidity condition (120% inlet relative humidity anode and cathode) performance data for the same setup previously described. As seen in Figure 4.6, the perforated DM case exhibits extremely poor performance under high humidity condition, showing a 183 mv lower potential than the virgin DM at a current density of 0.1 A cm -2. Furthermore, it was not possible to reach stable operations higher than 0.1 A 101

122 cm -2 for the perforated DM. The virgin DM, however, was capable of achieving limiting current densities ca. 2.3 A cm -2. This drastic performance difference can be directly attributed to the water management differences between the cells. The over humidification of the cell facilitates the condensation of water generated during operation. The increase in condensed liquid water enhances the water accumulation and promotes the formation of liquid water buildups, which prevent reactants to access the reaction sites. As a result, the mass transport losses dominate the polarization of the cell, even at low current. While the cell with perforated DM cannot handle the excess liquid water, the PCI flow and capillary-driven transport appear to properly manage the water distribution within the cell for the virgin DM case. 102

123 IR Compensated Voltage (V) HFR (mohm) 75 o C, 120% RH A/C Current Density (A cm -2 ) Figure 4.6. Performance data of 75 o C, 120% relative humidity, nitrox conditions for virgin DM case ( ), 15% 300-μm perforated cathode DM case ( ), and corresponding HFR values for the virgin ( ) and perforated ( ) cases. The DM perforations cause drastic performance losses (55 and 183 mv at 0.05 and 0.1 A cm -2, respectively), indicating poor water management in high humidity conditions. Before presenting neutron imaging data on these test conditions, it is beneficial to note the reasons of poor performance and instability of the perforated DM at high-current and/or high-humidity operation, as shown in Figure 4.3 and Figure 4.6. The decreased performance observed at high current density is in conflict with the data presented by 103

124 Gerteisen et al. [44], where it was shown that the implementation of DM perforations increased the limiting current density by 8% to 22%. Four key observations should be made to explain this difference. First, the polarization testing reported in [44] was performed comparatively at low relative humidity conditions, thus their results are similar to the observations at the low humidity conditions presented herein. Secondly, the diameter of the perforations used by [44] was approximately 100 μm, which is 1/3 of the size of the perforations (300-μm diameter) presented in the current study. The relatively smaller diameter of the holes in [44] significantly enhances the capillary-driven water transport, causing liquid water to be more readily wicked away from the perforated regions and CL. Furthermore, as [46] notes, larger diameter pores, especially those under the channels, may lead to increased film coverage in the channels, leading to blockage of the reactant gases. Due to the high humidity scenario presented here, thin film coverage is likely. Another possible reason for this observed discrepancy may stem from the fact that the perforations reported in [44] were performed only adjacent to the channel locations, which corresponds to less than 1% of the total area. However, in the present study, the perforations were uniformly disturbed across 7 or 15% of the geometric area of the DM, covering both land and channel locations. Increased coverage area of perforation might have caused the observed performance difference between these two studies. Finally, in [44], a scan rate of 10 mv s -1 was used to obtain polarization data, which yielded 8% to 22% increase in limiting current of the cell for the perforated case. This is distinctly different from the test procedure that was followed in the present study, where the quasi-steady state measurements of 3-minute dwell intervals were performed to obtain performance data for 104

125 the tested cases. It is anticipated that the quasi-steady-state measurements would allow for more accurate quantification of the impact of water accumulation inside the cell on the cell performance, whereas the use of a rapid voltage scan rate may display mostly the transient behavior of the perforations. The salient aspect of this comparison is to emphasize the evidence that an optimized perforated structure can yield desired performance metrics both under low and high humidity conditions, but this can only be achieved by careful consideration of various system parameters, which encourages further investigation NEUTRON RADIOGRAPHY TESTING VIRGIN VS. PERFORATED DM Figure 4.7 displays neutron images obtained under high humidity (100% inlet relative humidity) conditions. Figure 4.7a and Figure 4.7c show neutron data for the virgin DM at high and low current, respectively, whereas Figure 4.7b and Figure 4.7d are the perforated DM case at high and low current operation, respectively. Qualitatively, the perforated DM images show significantly more water accumulation than the virgin DM case. The most notable observation is the significant water accumulation (both under lands and channels) for the perforated DM case, which clearly indicates that the perforations enhance the liquid water storage in the cell regardless of humidity condition. Such a feature is beneficial for low relative humidity operation (for membrane hydration), but not for high relative humidity operation. Other studies [44,45] have shown that the perforated DM improves cell performance at high current operations, indicating that there is a significant potential to optimize both low and high humidity operation by altering the perforation size, location and cell geometry. 105

126 Figure 4.7. Neutron images from high humidity (100% inlet relative humidity anode and cathode) testing, showing: (a) high current (1.2 A cm -2 ) for virgin DM case; (b) high current for perforated DM case; (c) low current (0.2 A cm -2 ) for virgin DM case; and (d) low current for perforated DM case. In each image, the right-hand-side represents the cathode. Figure 4.8 and Figure 4.9 show the quantitative values of in-plane and through-plane water content of these cases obtained from the neutron images shown in Figure 4.7. For all the tested conditions, the total water mass per volume of the DM was observed to be higher in the perforated case than the virgin DM. At low current operation (Figure 4.8a), the total water amount in the perforated DM was measured to be 38% higher than the virgin DM, and similarly, the perforated DM was observed to contain 30% more water mass under high current/high humidity conditions (Figure 4.8b). As previously discussed, the EDS and ESEM analysis showed that the area surrounding the laser perforations was 106

127 more hydrophilic due to the removal of the PTFE during laser-perforation process. It is very likely that these highly hydrophilic regions may facilitate the condensation of water vapor and liquid storage in these regions, especially at high relative humidity condition. Previous studies [52] have shown that vapor phase diffusion preferably takes place through the hydrophobic DM, causing the vapor flux to condense under the landings and along the channel walls. With the introduction of the large hydrophilic perforations, the condensation locations can be altered in a way that excessive water accumulation would be encouraged, especially towards the perforated regions inside the DM, causing significant performance drop during operation. Figure 4.9 shows the water mass per volume distribution [in the through-plane direction] of unaltered and perforated DM cases at high humidity condition. As seen in Figure 4.9, the perforated-dm water profiles steadily increase through the cathode-side MPL region and through half of the macro-dm substrate. However, significantly lower water content is observed for the virgin case. A similar trend is also seen in Figure 4.5, where a sharp drop in the MPL region water mass is observed for virgin DM case. Perforations in the DM structure may also impede the function of the MPL to restrict water backflow from the DM into the CL, which could result in significant water coverage at the CL and consequently severe performance drop, as observed in Figure 4.6. The water mass peaks observed mid-way through the macro-dm substrate in Figure 4.9 support this argument, showing the existence of significant water diffusion gradient for the perforated DM case. 107

128 120 L C L C L C L 100 Water mass (mg cm -3 ) o C 100% RH A/C Distance (mm) (a) 120 L C L C L C L 100 Water mass (mg cm -3 ) o C 100% RH A/C Distance (mm) (b) Figure 4.8. Neutron data for water mass per volume distributions in the in-plane direction at high inlet relative humidity condition. L represents land, and C represents channel. (a) Water distributions for virgin DM ( ) and perforated DM ( ) under low current density (0.2 A cm -2 ) testing operation; (b) Water distributions for virgin DM ( ) and perforated DM ( ) under high current density (1.7A cm -2 ) testing operation. A reduced number of data points are shown to improve clarity. 108

129 200 Anode DM MPL / MEA Cathode DM Water mass (mg cm -3 ) o C 100% RH A/C Distance (mm) Figure 4.9. Neutron data for water mass per volume distributions in the through-plane direction at high inlet relative humidity condition. Virgin case at high current ( ), virgin case at low current ( ), perforated DM case at high current ( ), perforated DM case at low current ( ). A reduced number of data points are shown to improve clarity NEUTRON RADIOGRAPHY TESTING HIGH VS. LOW CURRENT AT HIGH HUMIDITY CONDITION Neutron imaging test were also performed for perforated and unaltered DM cases under high humidity conditions at different current densities. The perforated DM was observed to contain 14% less water at high current (Figure 4.8b) than at low current (Figure 4.8a). The same trend was also observed for the virgin DM case, where the virgin 109

130 DM is observed to retain 9% less water at high current. It should be noted that this behavior was the opposite of what was observed at low humidity operation. To explain this phenomenon observed at high humidity condition, several factors need to be carefully analyzed. First, PCI flow may be present at low and high current, but is more dominant at higher current densities due to the steeper temperature gradients [34]. Therefore, it is expected that the magnitude of water removal by PCI flow at elevated current would be higher, causing the low current operation (with lower temperature gradients) to retain higher water mass, as observed. Secondly, with the increased inlet relative humidity, it is expected that there will be an increase in water vapor condensation throughout the cell. The increased water saturation will further enhance the capillarydriven water transport, especially in the connected water pathways of DM [51,53]. Along with the high shear forces in the gas channel (due to high current operation), the increased PCI flow and enhanced capillary-driven water removal can explain the lower water content observed at high current density operation. 4.4 CONCLUSIONS This chapter presents investigations of the in-situ performance characteristics of virgin (unaltered) and laser-perforated DM (cathode-side only) by employing polarization testing and neutron radiography. DM samples were perforated with 300-µm laser-cut holes using a ytterbium fiber laser and compared to a virgin DM to investigate the effect of structural modifications on the water and gas transport characteristics of a PEFC. The perforated DM under low-humidity conditions (50% inlet relative humidity) showed on average a 6% voltage increase over the virgin DM for current densities ranging from

131 to 1.4 A cm -2. However, the cell assembled with perforated DM was observed to experience severe performance drop under low-humidity operation at high current (greater than 1.4 A cm -2 ), and at all currents under high-humidity operation (120% inlet relative humidity). The neutron radiography showed that the perforations act as water pooling locations, collecting and/or channeling water from the surrounding DM and CL. It was observed that large laser perforations alter the impact of phase-change-induced flow in removal of excess water from the DM. Furthermore, it is hypothesized that the removal of the PTFE binder due to the laser processing is crucial to the performance stability as the loss of hydrophobicity can further promote liquid water accumulation to the perforation regions, yielding significant water buildups that can cause local flooding. The results of this study, along with the other efforts in literature [44,45], suggest that proper tailoring of fuel cell DM (i.e. adjusting the perforation size, geometry and location) possesses significant potential to enable PEFC operations with reduced liquid overhead and higher performance under a wide range of operating conditions. Studies are underway to further optimize and analyze the effects of the perforations geometry on the PEFC performance. 111

132 REFERENCES [1] A. Bazylak, D. Sinton, and N. Djilali, Journal of Power Sources, vol. 176, pp , [2] M. Mathias, J. Roth, J. Fleming, and W. Lehnert, "Diffusion Media Materials and Characterisation," in Handbook of Fuel Cells: Fundamentals, Technology and Applications. vol. 3, W. Vielstich, A. Lamm, and H. A. Gasteiger, Eds., Hoboken: John Wiley & Sons, 2003, pp [3] H. K. Atiyeh, K. Karan, B. Peppley, A. Phoenix, E. Halliop, and J. Pharoah, Journal of Power Sources, vol. 170, pp , [4] J. T. Gostick, M. A. Ioannidis, M. W. Fowler, and M. D. Pritzker, Electrochemistry Communications, vol. 11, , [5] S. Park, J.-W. Lee, and B. N. Popov, Journal Power Sources, vol. 163, pp , [6] J. H. Nam and M. Kaviany, International Journal of Heat and Mass Transfer, vol. 46, pp , [7] H. A. Gasteiger, W. Gu, R. Makharia, M. F. Mathias, and B. Sompalli, "Begnningof-life MEA Performance--Efficiency Loss Contributions," in Handbook of Fuel Cells: Fundamentals, Technology, and Applications. vol. Vol. 3, H. A. G. W. Vielstich, A. Lamm, Ed., ed: John Wiley & Sons, 2003, pp [8] R. P. Ramasamy, E. C. Kumbur, M. M. Mench, W. Liu, D. Moore, and M. Murthy, International Journal of Hydrogen Energy, vol. 33, pp , [9] H. Meng, International Journal of Hydrogen Energy, vol. 34, pp ,

133 [10] A. Z. Weber and J. Newman, Journal of the Electrochemical Society, vol. 152, pp. A677-A688, [11] S. Ge and C.-Y. Wang, Journal of the Electrochemical Society, vol. 154, pp. B998- B1005, [12] F. E. Hizir, S. O. Ural, E. C. Kumbur, and M. M. Mench, Journal of Power Sources, vol. 195, pp , [13] T. Swamy, E. C. Kumbur, and M. M. Mench, Journal of the Electrochemical Society, vol. 157, pp. B77-B85, [14] H. Bajpai, M. Khandelwal, E. C. Kumbur, and M. M. Mench, Journal of Power Sources, vol. 195, pp , [15] J. P. Owejan, J. E. Owejan, W. Gu, T. A. Trabold, T. W. Tighe, and M. F. Mathias, Journal of the Electrochemical Society, vol. 157, pp. B1456-B1464, [16] S. Kim and M. M. Mench, Journal of Power Sources, vol. 174, pp , [17] S. Kim, B. K. Ahn, and M. M. Mench, Journal of Power Sources, vol. 179, pp , [18] W. Dai, H. Wang, X.-Z. Yuan, J. Martin, J. Shen, M. Pan, and Z. Luo, Journal of Power Sources, vol. 188, pp , [19] S. Park and B. N. Popov, Fuel, vol. 88, pp , [20] S. Park, J.-W. Lee, and B. N. Popov, Journal of Power Sources, vol. 177, pp , [21] E. C. Kumbur, K. V. Sharp, and M. M. Mench, Journal of the Electrochemical Society, vol. 154, pp. B1295-B1304,

134 [22] E. C. Kumbur, K. V. Sharp, and M. M. Mench, Journal of the Electrochemical Society, vol. 154, pp. B1305-B1314, [23] E. C. Kumbur, K. V. Sharp, and M. M. Mench, Journal of the Electrochemical Society, vol. 154, pp. B1315-B1324, [24] J. T. Gostick, M. W. Fowler, M. A. Ioannidis, M. D. Pritzker, Y. M. Volfkovich, and A. Sakars, Journal of Power Sources, vol. 156, pp , [25] Z. Qi and A. Kaufman, Journal of Power Sources, vol. 109, pp , [26] L. Giorgi, E. Antolini, A. Pozio, and E. Passalacqua, Electrochimica Acta, vol. 43, pp , [27] F. Lufrano, E. Passalacqua, G. Squadrito, A. Patti, and L. Giorgi, Journal of Applied Electrochemistry, vol. 29, pp , [28] G. Velayutham, J. Kaushik, N. Rajalakshmi, and K. S. Dhathathreyan, Fuel Cells, vol. 7, pp , [29] R. R. Fredley, Porous Support Layer for an Electrochemical Cell, US Patent # 5,998,058, [30] J. Benziger, J. Nehlsen, D. Blackwell, T. Brennan, and J. Itescu, Journal of Membrane Science, vol. 261, pp , [31] G.-G. Park, Y.-J. Sohn, T.-H. Yang, Y.-G. Yoon, W.-Y. Lee, and C.-S. Kim, Journal of Power Sources, vol. 131, pp , [32] K. T. Cho and M. M. Mench, Journal of Power Sources, vol. 195, pp , [33] K. T. Cho and M. M. Mench, Journal of Power Sources, vol. 195, pp ,

135 [34] S. Kim and M. M. Mench, Investigation of Temperature-Driven Water Transport in Polymer Electrolyte Fuel Cell: Phase-Change-Induced Flow, Journal of the Electrochemical Society, vol. 156, pp. B353-B362, [35] A. Z. Weber and J. Newman, Journal of the Electrochemical Society, vol. 153, pp. A2205-A2214, [36] Y. Wang and C.-Y. Wang, Journal of the Electrochemical Society, vol. 153, pp. A1193-A1200, [37] U. Pasaogullari and C.-Y. Wang, Electrochimica Acta, vol. 49, pp , [38] C. Hartnig, I. Manke, R. Kuhn, N. Kardjilov, J. Banhart, and W. Lehnert, Applied Physics Letters, vol. 92, pp , [39] A. Turhan, S. Kim, M. Hatzell, and M. M. Mench, Electrochimica Acta, vol. 55, pp , [40] D. S. Hussey, D. L. Jacobson, M. Arif, J. P. Owejan, J. J. Gagliardo, and T. A. Trabold, Journal of Power Sources, vol. 172, pp , [41] M. A. Hickner, N. P. Siegel, K. S. Chen, D. S. Hussey, D. L. Jacobson, and M. Arif, Journal of the Electrochemical Society, vol. 155, pp. B427-B434, [42] A. Z. Weber and M. A. Hickner, Electrochimica Acta, vol. 53, pp , [43] M. C. Hatzell, A. Turhan, S. Kim, D. Hussey, D. Jacobson, and M. M. Mench, Journal of the Electrochemical Society, vol. 158, pp. B717-B726, [44] D. Gerteisen, T. Heilmann, and C. Ziegler, Journal of Power Sources, vol. 177, pp ,

136 [45] D. Gerteisen and C. Sadeler, Journal of Power Sources, vol. 195, pp , [46] E. Kimball, J. Benziger, and Y. Kevrekidis, Fuel Cells, vol. 10, pp , [47] F. E. Hizir, Master of Science in Mechanical Engineering, The Pennsylvania State University, [48] M. P. Manahan, S. Kim, E. C. Kumbur, and M. M. Mench, ECS Transactions, vol. 25, pp , [49] Toyota Motor Company, Fickian Diffusion Coefficient Ratio, Internal Communication to Penn State University, [50] S. Kim, M. Khandelwal, C. Chacko, and M.M. Mench, Journal of the Electrochemical Society, vol. 156, pp. B99-B108, [51] S. Litster, D. Sinton, and N. Djilali, Journal of Power Sources, vol. 154, pp , [52] A. Turhan, K. Heller, J. S. Brenizer, and M. M. Mench, Journal of Power Sources, vol. 180, pp , [53] E. C. Kumbur, K. V. Sharp, and M. M. Mench, Journal of Power Sources, vol. 176, pp ,

137 5. LASER PERFORATED FUEL CELL DIFFUSION MEDIA: ENGINEERED INTERFACES FOR IMPROVED IONIC AND OXYGEN TRANSPORT The purpose of this chapter is to explore engineered interfacial and structural architecture in fuel cell diffusion media (DM). Perforations were introduced via lasers on samples of virgin DM that contained hydrophobic content. Depending on laser choice, some laser-cut samples displayed a heat affected zone (HAZ) at the catalyst layer microporous layer interface, characterized by a region surrounding each perforation where hydrophobic content was removed. At 50% inlet relative humidity, DM with homogeneously dispersed 100-μm perforations and a HAZ displayed a 25% power density increase compared to virgin DM. Analyzing the oxygen concentration dependence in the double-tafel region showed transport resistances were dominated by oxygen at moderate current values. Electrochemical impedance spectroscopy (EIS) and neutron radiography results indicated charge- and mass-transport impedances and liquid water redistribution play an important role, depending on the operating current density. Results suggested two mechanisms for the increased performance of the 100-μm DM with HAZ: i) liquid water storage and through-plane water redistribution led to rehydration of the catalyst layer and membrane, and ii) in-plane water redistribution led to improved oxygen transport through the DM. The results of this study shed light on the importance of interfacial and structural architecture of fuel cell DM. 5.1 INTRODUCTION Polymer electrolyte fuel cells (PEFCs), also known as polymer electrolyte membrane fuel cells (PEMFCs), have been the subject of intensive research over the past two 117

138 decades. Advances in the fundamental understanding of PEFCs have led to improvements in system performance and therefore have increased their marketability; yet the systems have only recently entered various commercial markets. The cathode side of the fuel cell is often a key limiting factor due to slow kinetics, complex multi-phase transport, and thermal management. These complexities have been well studied, but there is still a need to augment fundamental understanding and attain improvements in the cathode performance and cell durability. The diffusion medium (DM) is a porous material that is compressed between the flow field and the catalyst layer (CL) in conventional cell designs. Diffusion media serve numerous, well-established functions [1, 2], including: i) in- and through-plane gas-phase reactant access from the flow plate (conventionally a land-channel design) to the CL; ii) multi-phase transport of water to and/or from the CL; iii) electrical contact between the flow field/bipolar plate and the CL; iv) rigid support of the membrane and CL, especially in conventional land-channel designs where local stresses can be high; and v) heat transport to and from the CL, depending on the existing temperature gradients in the cell. Given its multi-functional role in PEFC operation, the DM structure has been a topic of intensive research with respect to each of these functions. Oxygen molecules have a tortuous path to travel: from the inlet channels/manifolds, through the channels, into the diffusion media, which typically includes a microporous layer (MPL), through the micro- and nano-pores of the catalyst layer, and likely through a water or ionomer thin film to a catalyst reaction site. The oxygen transport spans multiple length scales, from bulk convective diffusion in the manifold/channel region to Knudsen and film diffusion in the CL [3]. Beuscher [4] used experimental limiting-current 118

139 methods to estimate the oxygen-transport resistance through CARBEL type DM (230 μm compressed thickness) and estimated that the DM accounts for 25% of the total transport resistance. Beuscher also estimated that Knudsen and film diffusion (in MPL and CL) account for over 50% of the transport resistance when using Gore 5510 series catalyst layers. Manahan et al. [5] also found a similar result using the same experimental methods and materials. Baker et al. [6] estimated the oxygen-transport resistance through several Toray DM in operating conditions where, theoretically, no vapor condensation occurs. Their studies investigated pressure-dependent (Fickian) and pressure-independent (Knudsen and/or film) transport resistances, and they varied DM thickness to assess flow channel contributions. They concluded that the macro DM accounts for ~50% of the oxygen transport resistance twice that found in Beuscher s work in Toray 060 DM with polytetrafluoroethylene (PTFE) treatment and with a 25 μm MPL (~190 μm total compressed thickness). Baker et al. attributes the difference between their work and Beuscher s to liquid water generation and operating conditions adjustments. Perry et al. [7], and later Jaouen et al. [8] and Ihonen et al. [9], evaluated methods to delineate between oxygen-transport and ionic-transport limited performance when operating at voltages within the double Tafel slope region. They established that the cell current should scale linearly with oxygen concentration in the double-tafel region if the system is limited by kinetics and oxygen transport; however, it should scale to the onehalf power of oxygen concentration if limited by kinetics and ionic transport in the CL. O Neil et al. [10] modeled oxygen gain methods to determine if oxygen transport was limited externally (in DM) or internally (in CL). They determined that oxygen gain goes 119

140 to a theoretically infinite value if oxygen-transport resistance is externally limited, but oxygen gain is limited to a value of twice its original value if internally limited. The inclusion of an MPL between the macro-dm and CL in PEFCs was likely first reported by Wilson et al. [11] in 1992 but is now widely used and studied. State-of-theart bi-layered diffusion media consist of a macro-dm with pore diameters on the order of μm and an MPL with pore diameters in the μm range. Still, over two decades after the introduction of the MPL, the mechanisms that lead to improved performance when an MPL is introduced are still debated. Some studies suggest the MPL acts as a liquid water capillary barrier, forcing generated water at the cathode side toward the anode side, thus reducing cathode-side flooding effects and rehydrating the membrane [12-14]. Other studies propose it maintains the vapor phase of water produced in the cathode CL and prevents liquid water that has condensed in the cathode-side macro-dm from migrating back toward the anode side [13-18]. The inclusion of interfacial and thermal effects is also critical to consider when analyzing these effects [17, 19-28]. Indeed, operating conditions and material properties significantly affect the results of this topic, and some of these studies suggest a combination of these mechanisms, which is more likely. Due to the multi-phase nature of PEFC operation, the oxygen transport and overall cell performance are directly affected by the results of the liquid-phase water management. Each region of the fuel cell (flow channels, DM, CL, membrane) has a direct impact on the water management. It has been shown that channel design [e.g., 19, 20], DM structure [e.g., 17, 29-32], MPL [e.g., 12, 15, 16], and CL [e.g., 33, 34] all have strong impacts on multi-phase water management. Not only the components themselves, 120

141 but the interfaces between these components have shown to be critical [e.g., 5, 17, 19, 27, 28]. Each component mentioned has complexities to resolve in and of itself, and the integration of each component is essential to good cell and stack performance. A key component to the DM is its hydrophobic content. Numerous studies investigated PTFE content effects on performance, e.g., Refs. [35-38]. Mendoza et al. recently developed a technique for carbon and PTFE mapping of DM using Raman spectroscopy [39]. It has been established that an overall PTFE content between 5 and 20 wt% is optimal for common operating conditions [e.g., 35-37], although a fundamental, definitive, pore-level study of this question has not yet been published. Recently, studies conducted by the authors and other groups have shown promise for further DM and MPL structural modification for enhanced performance [17, 40-44]. In particular, Manahan et al. [17, 40] and Gerteisen et al. [41, 42] have shown performance enhancements resulting from laser perforations through the cathode-side DM. The present work follows on the work of Manahan et al. [17], who showed that 300-μm diameter perforations altered the hydrophobicity of the DM and served as liquid water storage and/or transport locations, leading to improved performance with low-humidity conditions. Recently, Markötter et al. [45] showed perforations often serve as liquidwater transport pathways. Kitahara et al. [43] and Blanco et al. [44] have displayed improved performance with low-humidity inlet streams by adding layers to the DM which optimize liquid water retention in the cathode in order to maintain membrane hydration and improve durability. For many applications, low-humidity operation is desirable to reduce parasitic losses and system complexity. 121

142 This chapter investigates fundamental transport and performance characteristics of cathode-side DM whose interfacial and structural properties have been modified via laser treatment. Namely, the source of an observed 25% power density increase with the introduction of 100-μm diameter perforated DM with a HAZ at the CL MPL interface is investigated. Previous results from Chapter 4 (studied 300-μm perforations) are built upon and integrated to elucidate fundamental enhancements in performance of certain laser perforations and surface engineering. Various experimental techniques are used, including polarization testing, Tafel-slope analysis, electrochemical impedance spectroscopy, and neutron radiography. The results for the perforated DM are compared against a cell with unaltered DM. Furthermore, two different lasers were used to create the 100-μm perforations, and the impact of surface modification via laser selection is studied. 5.2 EXPERIMENTAL APPROACH MATERIALS AND LASER MODIFICATION All experimental tests were conducted using a 5-cm 2 active area fuel cell, designed with liquid coolant channels, vertical-direction channel inlets and outlets (gravityassisted), and a six-bolt configuration to maximize test-to-test and cell-to-cell repeatability. A double serpentine flow channel design with a channel-land ratio of 1.88 was used. Cell components include a W. L. Gore & Associates (Newark, DE) membrane (18-μm thick, reinforced) with Gore PRIMEA 5510-series electrodes on both the anode and cathode, containing 0.4 mg Pt cm -2 loading on each side. All membrane electrode assemblies used in experimentation were cut from the same batch, ensuring the most consistent membrane and CL properties. As-received (herein, virgin ) SIGRACET 122

143 diffusion media type SGL 10 BB (bi-layered with 5 wt% PTFE content, 415 μm uncompressed thickness) were used on both the anode and cathode as the control group and were compressed by approximately 25% of their nominal thickness. Experiments were also conducted on laser-perforated SGL 10 BB samples, which were introduced on the cathode side only, and results are compared to the virgin DM. Laser perforations on the cathode-side DM of 100-μm and 300-μm diameter were made with two different lasers, i.e., a ytterbium fiber laser (IPG Photonics, Oxford, MA) and a neodymium-doped yttrium vanadate (Nd:YVO 4 ) laser (Coherent, Santa Clara, CA). A schematic of the cathode-side configuration of a virgin DM and a perforated DM is shown in Figures 5.1a and 5.1b, respectively, and laser details are given in Table 5-1. Figure 5.1. Schematic of the cathode side of the (a) virgin DM and (b) laser-treated, perforated DM (not to scale). Table 5-1. Laser parameters used to perforate diffusion media. Laser Type Wavelength (nm) Pulse Duration Pulse Energy (μj) Average Power (W) Nd:YVO ns Yb fiber μs

144 After laser treatment of multiple samples, the surfaces of the DM were investigated using scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS). SEM images revealed a heat affected zone (HAZ) on the MPL of some samples, i.e., an area surrounding the perforation that was visibly affected as a result of the laser treatment. Other laser-cut samples, however, showed no signs of a HAZ. Specifically, in the virgin DM and in DM samples processed using the Nd:YVO 4 laser, EDS at multiple locations on the MPL surface revealed two dominant peaks corresponding to the carbon peak and the fluorine peak, the latter peak indicating the presence of PTFE (Figure 5.2a). These DM will be referred to as virgin DM and 100-μm nohaz since both samples did not have evidence of a HAZ. Contrarily, it was found that the use of the Yb-fiber laser resulted in a HAZ surrounding the perforation, evidenced by a lack of a fluorine peak in certain areas surrounding the perforation (Figure 5.2b). This lack of a fluorine peak indicates the removal of PTFE from the substrate in certain regions. These heataffected samples will be referred to as 100-μm w/haz and 300-μm w/haz since both samples showed some evidence of PTFE content changes as a result of the laser treatment. The size of the HAZ depended on the perforation and laser properties. Specifically, the 100-μm w/haz DM showed an approximately 200-μm HAZ radius from the center of the perforation, leaving a portion of the DM hydrophobic content unchanged. On the other hand, no fluorine peaks were observed anywhere on the MPL surface of the 300-μm w/haz DM, indicating a removal of PTFE content across the entire sample. It may be inferred, therefore, that each of the laser-treated DM has varying PTFE contents (both total amount and spatial variation), and the hydrophobic/hydrophilic interfacial behavior will be distinctly different between the each of the samples. An 124

145 electron micrograph of a visible HAZ is shown in Figure 5.2c, showing some areas affected by the HAZ around the perforation but other areas unaffected. Table 5-2 summarizes the key properties of the DM used in this study. It should be noted that the perforated DM used for neutron radiography have a center-to-center distance of 0.97 mm (different spacing than listed in Table 5-2) to better capture the perforations in the inplane direction (see Ref. [17]). (a) kev (b) kev Unaffected MPL Heat affected zone Perforation (c) 1 mm Figure 5.2. Characteristic energy dispersive spectra for (a) virgin and 100-μm nohaz DM and (b) 300-μm w/haz and portions of 100-μm w/haz DM. The absence of a fluorine peak in DM w/haz samples indicates removal of hydrophobic content by Ybfiber laser treatment. (c) SEM image with HAZ around perforation. 125

146 Table 5-2. Diffusion media properties used for electrochemical characterization. Diffusion Media Laser Type Percent Planform Area Removed Heat Affected Zone (HAZ) determination by EDS Center-to-Center Distance / Number of Holes Virgin None 0% Untreated None 100 µm, no HAZ Nd:YVO 4 1.7% No HAZ observed within 1 μm of perforation edge 0.67 mm / ~ μm w/ HAZ Yb fiber 1.7% Yes, ~200 μm radius 0.67 mm / ~ µm w/ HAZ Yb fiber 15% Yes (entire surface) 0.67 mm / ~ POLARIZATION TESTING Polarization tests were conducted at a coolant-controlled cell temperature of 75 o C at 50% inlet relative humidity (RH) conditions on both the anode and cathode, as well as at 120% inlet RH conditions, as described in Ref. [17]. Reactant flow rates were held constant at 139 sccm H2 (anode) and 332 sccm air (cathode), which corresponded to a 2/2 stoichiometry at 2 A cm -2. Quasi-steady state conditions were achieved by using galvanostatic steps for 3-minute intervals at each current value, starting from open circuit voltage and increasing the current each interval. To ensure repeatability and accuracy, each polarization curve presented is an average of at least 3 cycles, and error bars indicate the maximum and minimum values obtained over those averaged cycles. Furthermore, data presented in polarization curves are IR-compensated in order to separate ohmic losses from other losses contributing to the total polarization and aid in analysis of the system. Polarization curves and HFR data alone, however, often are not sufficient to decisively conclude the locale of any given loss, and therefore additional experiments were conducted as described. 126

147 5.2.3 TRANSPORT LIMITATION ANALYSIS A method of delineating between oxygen- and ionic-transport limited scenarios was first suggested by Perry and coworkers [7] and further work was conducted several years later [8, 9, 46]. These works present a thorough description of their model and experiments, and only a brief summary will be provided here. The goal of the suggested experimental technique is to determine the source of the well-established double Tafel slope region. Losses that contribute to the double Tafel slope region are a combined effect of the kinetics of the electrode (single Tafel slope) plus another source of loss, namely, either oxygen transport or ionic transport. That is, when a double Tafel slope appears, it signifies the oxygen reduction reaction (ORR) is limited by either Tafel kinetics and oxygen diffusion or by Tafel kinetics and proton migration in the CL. The current is first-order in oxygen concentration if the performance is dominated by kinetics and oxygen transport since both the ORR kinetics and oxygen transport are first order processes. However, if the ORR kinetics and ionic transport dominate the losses, it has been derived and shown experimentally that the current is half-order in oxygen concentration [47]. Experimentally, then, we can vary the total pressure or the oxygen partial pressure and analyze the response of the current to such changes, and thus determine the source of limitation. Therefore, Tafel experiments here attempted to mimic the conditions suggested by Refs. [7-9]. The cell was kept at 75 o C and the flow rates were set to a constant stoichiometry of 5/5 (with minimum flow rates of 5/5 stoichiometry at 1 A cm -2 ) on the anode and cathode in order to minimize oxygen gradients in the bulk DM. The voltage was scanned 10 times from OCV to 0.3 V at a scan rate of 2 mv s -1. After each sweep, 127

148 the cell was held at V for 10 minutes to avoid hysteresis. The total pressure of the cathode gas stream was tested at both 1 atm and 2 atm (absolute) to vary oxygen concentration, and the anode stream was left open at atmospheric. Tests were also conducted by changing the partial pressure of oxygen with air (21% O 2, balance N 2 ), half-air (10.5% O 2, balance N 2 ), and double-air (42% O 2, balance N 2 ) [48], and the results were the same as those run at a modified total pressure. Tests were conducted only on 100-μm w/haz and 100-μm nohaz cells at 50% RH in attempt to determine the cause for their performance differences observed in polarization testing ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY Electrochemical impedance spectroscopy (EIS) is a powerful tool to probe electrochemical devices and understand fundamentals such as charge- and oxygentransport in situ [49-54]. Despite its ability to extract time-scale specific information, EIS is still best-used in conjunction with other test results to identify or verify certain phenomena. Experiments were conducted at constant current on a Scriber 850C (Scribner Associates, Southern Pines, NC) test stand to control the flow rates (3/3 stoichiometry), humidifiers (both 50% and 100% RH conditions), and current, and an 880 Impedance Analyzer (Scribner Associates, Southern Pines, NC) was used to apply the AC signal frequencies at amplitudes of 5% of the DC current value from 10kHz to 0.1 Hz. The cell temperature was coolant-maintained at 75 o C. These conditions, similar to the polarization testing conditions, allowed for stable voltages that are needed for data collection at low perturbation frequencies. In all tests, the previously-mentioned 5-cm 2 cell was held at constant current for 20 minutes before overlaying the AC signal to ensure steady-state operation. Impedance spectra presented here are an average of five or more consecutive 128

149 frequency scans at any given current. The data are also shifted on the real-impedance axis to remove the high-frequency resistance and to better compare the other competing effects (i.e., charge- and oxygen-transport resistances) [53-55] NEUTRON RADIOGRAPHY Manahan et al. [17] described in detail the experimental setup of the neutron radiography (NR) testing. Testing was conducted at the National Institute for Standards and Technology (NIST) (Gaithersburg MD) on a 17.2-cm 2 cell held at steady-state current values of 0.2 A cm -2 (low current) and 1.7 A cm -2 (high current) for low (50%) and high (100%) RH values. The plots shown are in the through-plane direction (anode to cathode) and the in-plane direction (stream-wise, averaging both anode and cathode sides). 5.3 RESuLTS AND DISCUSSION TESTING AT 50% RELATIVE HUMIDITY Polarization Testing at 50% RH: Polarization testing was conducted on all samples in order to observe overall cell performance differences between PEFCs built with virgin cathode-side DM and perforated cathode-side DM, i.e., DM with different perforation diameters (100-μm and 300-μm diameter) and those with and without HAZ due to different laser interactions with the sample. Figure 5.3a shows the polarization curves at 50% RH for all four DM types. The 100-μm nohaz sample has very similar performance compared to the virgin DM, especially at current densities below 1.6 A cm -2. Beyond 1.6 A cm -2, a more severe mass-transport tail is observed. The data presented here are IR-compensated, which means that the virgin DM and 100-μm nohaz DM have very similar performance when ohmic losses are removed. Figure 5.3b, however, shows the 100-μm nohaz DM has HFR values that are about 10% lower (average of all values) 129

150 compared to the virgin DM. This indicates the perforations likely allow for rehydration of the membrane, since the membrane ionic resistance dominates the HFR values under normal operating conditions. Both the 100-μm and 300-μm w/haz DM show improved performance over the virgin and 100-μm nohaz DM for current densities less than about 1.6 A cm -2 (an average of 7% voltage increase at each current value). At 1.6 A cm -2, the 100-μm and 300-μm w/haz DM deviate from each other, with the 300-μm w/haz DM showing a sharp mass-transport tail. The 100-μm w/haz DM, however, continues to show improved performance and boasts a 7% increase in limiting current density compared to the virgin DM. Notably, the 100-μm w/haz DM has 25% higher peak power density (0.94 W cm -2 ) compared to the virgin DM (0.75 W cm -2 ) (Figure 5.3a). The 100-μm w/haz DM also shows an average of 23% lower HFR compared to the virgin DM (Figure 5.3b). Based on the HFR data in Figure 5.3b, it can be concluded that the 100-μm perforations (both w/haz and nohaz) reduce the HFR of the cell. Since the perforations penetrate through the macro-dm and though the MPL, liquid water condensed in the macro-dm has a straight conduit to rehydrate the membrane and thus lower the HFR values. This supports the findings of Refs. [13-18] who suggest the MPL acts to prevent liquid water that has condensed in the cathode-side DM from migrating back toward the anode side. It can be noted, however, that the 300-μm w/haz DM has higher fluctuations in HFR values, and the magnitude of the HFR is sometimes higher than the virgin DM. This behavior is attributed to the highly hydrophilic nature of the 300-μm w/haz DM causing significantly higher water retention in the DM (see Ref. [17]). Consistent and 130

151 sufficient hydration of the membrane was not always achieved since liquid water was more likely to be retained in the hydrophilic 300-μm w/haz DM. Therefore, hydration of the membrane may have been more sporadic, and the HFR values were more inconsistent and sometimes higher than that of the other DM. With the virgin and 100-μm DM, the DM properties promoted consistent hydration of the membrane. Therefore, it can be definitively shown that the addition of properly-sized perforations allows for increased membrane hydration which lowers the ohmic losses observed in performance at 50% RH. However, the reason for the performance increase of the 100- μm w/haz DM compared to the 100-μm nohaz or virgin DM cannot be ascertained using only polarization curves and HFR results. 131

152 IR Compensated Voltage (V) Virgin DM m w/ HAZ 100 m w/ HAZ m no HAZ Current Density (A cm -2 ) (a) Power Density (W cm -2 ) HFR (mohm) Cu / HAZ 0.2 / HAZ HAZ (A cm -2 ) Power Density (W cm -2 ) HFR (mohm) Current Density (A cm -2 ) (b) Figure 5.3. (a) Polarization data at 50% RH. Solid symbols indicate DM with nohaz, open symbols indicate DM w/haz. (b) Corresponding high frequency resistance (HFR) obtained during polarization curves. Selected lines added to show lower HFR values for 100 μm DM (w/haz and nohaz). 132

153 Since ohmic losses are accounted for (IR-compensated), it is inferred that ion and/or oxygen transport may be a key factor in the performance differences, and further investigation was conducted. While HFR analysis is a helpful and necessary diagnostic tool, the fundamental processes it utilizes must be understood when analyzing the data to ensure proper understanding and interpretation. In particular, it is a technique that forces charge transfer through the path of least resistance in all components. In PEFCs, the observed HFR includes the resistance of the current collectors, DM, membrane, interfaces, and CL (primarily the electronic portion of the CL). The HFR does not include a significant portion of the ionic resistance in the CL because it is considered to be mostly in parallel with the electronic resistance of the CL, which generally has a much lower resistance compared to the ionic transport through the ionomer in the CL. Therefore, the cause for the performance increases with 100-μm w/haz DM over the 100-μm nohaz DM in Figure 5.3a could be attributed to ion transport in the CL and/or to oxygen transport differences. Oxygen Transport vs. Ionic Transport Resistance at 50% RH: Perry et al. [7] suggested a method to experimentally differentiate between the oxygen transport and ionic transport losses. Through the use of Tafel plots, one can investigate the effect of changing the oxygen pressure (total or partial oxygen pressure) on the current in the double-tafel region. The double-tafel slope arises from the combined losses of ORR kinetics and one of the following: ionic transport or oxygen transport. If the current scales linearly with oxygen concentration, this indicates the double Tafel slope arises from kinetics and oxygen transport. If the current scales to the one-half power with oxygen concentration, however, the increased losses are due to kinetics and ionic transport in the 133

154 CL. Refs. [8, 9] give an updated model to Perry et al. and also experimentally verify the proposed behavior. In this study, the technique is used to understand the cause of the increased performance observed with the 100-μm w/haz DM over the 100-μm nohaz DM. Figure 5.4 shows the current (corrected for cross-over current) from the scanned voltage (2 mv s -1 ) from open circuit voltage to 0.5 V (IR-corrected) at both 1 atm and 2 atm air on the cathode. Data points are shown for 10 consecutive sweeps. The double- Tafel slope region (b 200 mv dec -1 for these electrodes), from approximately 0.5 A cm -2 to 1.0 A cm -2, is the region that is exploited to extract information regarding the limiting transport process as previously outlined. As can be observed (and was calculated but not shown), both the 100-μm w/haz (Figure 5.4a) and the 100-μm nohaz (Figure 5.4b) DM scale linearly with oxygen concentration. That is, with the total pressure doubled, the current is doubled in the double-tafel region. This indicates, then, that both 100-μm w/haz and nohaz DM are oxygen-transport limited in the specified current range. Thus, the cause for the observed performance difference (Figure 5.3a) in the double Tafel region is due to a change in oxygen transport characteristics. Note that this method is only investigating from about 0.5 A cm -2 to 1.0 A cm -2 and therefore does not give direct information regarding low or high current values (outside the double Tafel region). 134

155 IR-Compensated Voltage (V) b = m w/haz 1 atm 100 m w/haz 2 atm b = - 88 mv/dec b = b = b = b= IR-Compensated Voltage (V) b = m noha 100 m noha Corrected Current Density (A cm -2 ) (a) Correc 0.90 mv/dec b = b = b = b= IR-Compensated Voltage (V) b = - 99 mv/dec b = b = m nohaz 1 atm 100 m nohaz 2 atm b = b = b = nt Density (A cm -2 ) Corrected Current Density (A cm -2 ) (b) Figure 5.4. Tafel plots of (a) 100-μm w/haz (b) 100-μm nohaz, both with cathode-side absolute air pressure at 1 atm and 2 atm to vary oxygen concentration. Also included are 135

156 logarithmic best fit lines in the single-, double-, and high-tafel slope regimes. Tafel slopes are indicated as b near each best fit, and units are in mv dec -1. Impedance Spectra at 50% RH: Electrochemical impedance spectroscopy (EIS) results can lend fundamental, time-scale-specific insight into the underlying processes behind the performance observed in the polarization testing. With its wide range of probing frequencies (spanning six orders of magnitude), one can delineate oxygen diffusion processes (low frequency, less than ~10 Hz) from faster cathode charge transfer processes (mid frequency, ~ Hz) and extremely fast electron transfer processes (high frequency, greater than ~100 Hz ). Figure 5.5 shows three impedance spectra taken at different steady-state current values at 50% RH. From the plots shown in Figure 5.5, a clear differentiation can be made between the tested samples. At 0.4 A cm -2 (Figure 5.5a), the DM w/haz (open symbols) show a significantly smaller charge-transfer arc compared to the DM without HAZ (virgin and 100-μm nohaz, closed symbols). This indicates that, at steady-state low current, DM w/haz reduce the resistance processes involved with the charge transfer, giving rise to their increased performance. Makharia et al. [54] note that the CL ionic resistance is inversely proportional to the ionomer conductivity, which is a strong function of the relative humidity, and other CL physical properties such as thickness, volume fraction of ionomer, and tortuosity. In the present work, the physical properties of the CL should be identical. Therefore, if the DM w/haz maintain higher effective RH due to their increased hydrophilic areas and liquid water storage, this would lead to a lower chargetransfer resistance by increasing the conductivity of the ionomer in the CL. Recall that 136

157 this increase in conductivity would not be evident from the HFR data, since the ionic and electronic resistances in the CL are mostly in parallel with one another. Also of importance, the 300-μm w/haz DM shows a pronounced low-frequency arc (less than 10 Hz) even at such low current densities, which indicates the increase in oxygen-transport related resistances. This is an early warning sign, indicating the 300-μm w/haz DM will have significant losses in oxygen-transport at higher currents. The 100- μm w/haz DM, however, does not show any significant oxygen-transport arc, further showing the importance of proper perforation design. Figure 5.5b shows impedance spectra at 1.0 A cm -2. At higher current, when more water is being generated in the cathode CL, the charge-transfer arcs begin to take on a very similar diameter, indicating a more equivalent charge-transfer resistance among all the samples. It can be seen that the 300-μm w/haz has a slightly lower charge-transfer arc, likely resulting from a more saturated DM which allows for increased hydration of the CL, reducing charge-transfer resistances. In the lower-frequency regime (< 10 Hz), the 100-μm w/haz DM shows an oxygen-transport arc that is ~10% smaller than the DM samples without HAZ. This result agrees with the results shown in the Tafel-plot analysis (Figure 5.4), which indicated the performance differences were dominated by oxygen-transport processes at moderate current density. At 1.4 A cm -2 (Figure 5.5c), the 100-μm w/haz shows both a smaller charge-transfer arc as well as a smaller oxygen-transport arc. From Figure 5.4 and as addressed in Ref. [7], this current density is in a regime where quadruple (or higher) Tafel slopes are observed, arising from the compounding effects of kinetic, ionic-transport, and masstransport losses. Therefore, it can be ascertained from Figure 5.5c that the 100-μm 137

158 w/haz DM shows better ionic transport and oxygen diffusion at higher currents, resulting from the combination of water storage in the DM and improved oxygendiffusion through the DM. These improvements are further analyzed using neutron radiography in the following section. -Z'' Imaginary Impedance (Ohm cm 2 ) -Z'' Imaginary Impedance (Ohm cm 2 ) Hz > 3000 Hz Virgin DM 300 m w/haz 100 m w/haz 100 m nohaz 10 Hz 10 Hz Z' Real Impedance (Ohm cm 2 ) Virgin DM 300 m w/haz 100 m w/haz 100 m nohaz 100 Hz > 3000 Hz 10 Hz (a) 10 Hz < 1 Hz Z' Real Impedance (Ohm cm 2 ) < 1 Hz (c) 0.4 A cm -2 50% RH 1.4 A cm -2 50% RH -Z'' Imaginary Impedance (Ohm cm 2 ) A cm -2 Virgin DM 50% RH 300 m w/haz 100 m w/haz 100 m nohaz 10 Hz 100 Hz > 3000 Hz < 1 Hz Z' Real Impedance (Ohm cm 2 ) (b) Figure 5.5. Electrochemical impedance spectra at (a) 0.4 A cm -2, (b) 1.0 A cm -2, (c) 1.4 A cm -2 at 50% RH. Selected frequencies are shown. Neutron Radiography at 50% RH: It was shown in Ref. [17] that as steady-state current increased, the total amount of liquid water in the 300-μm w/haz DM increased more than with virgin DM. Figure 5.6 shows the through-plane water content in a virgin DM compared to the 100- and 300-μm w/haz DM. It is clear that both the DM w/haz 138

159 show increased steady-state water content, indicating an increased water storage capacity. In addition to the total cathode-side water content, the water content near the cathode-side CL MPL interface has been modified. The water content steeply drops for the virgin DM but is nearly constant for both of the DM w/haz. The decrease in liquid water content in the MPL of the virgin DM indicates the MPL pores are filled with either water vapor and/or reactant gas. The increased water content in the DM w/haz (especially in the MPL) can result in significant rehydration of the CL and membrane, which would be one key factor in reducing the charge-transfer arc observed in Figure 5.5. Water content profiles for the 100-μm nohaz were not obtainable due to limited beam time; however, from the impedance spectra and polarization curves, one can surmise that the profile would closely follow that of the virgin DM. The water content may be slightly higher due to the increase in void volume (i.e., perforations have porosity of unity), but since the hydrophobic content remains practically unchanged, the water storage amount should be minimally different. 139

160 Figure 5.6. Through-plane water content distribution at 50% RH at constant current of 1.7 A cm -2 under steady-state operating conditions. Not only is there higher water content and through-plane water redistribution as current increases in DM w/haz, but DM w/haz have also been shown to redistribute the in-plane liquid water, as shown in Figure 5.7 and in Ref. [17] for 300-μm w/haz DM, especially as current increases. Markötter et al. [45] recently concluded using a high-resolution x-ray synchrotron that the perforations indeed act as preferential liquid water pathways, drawing water from surrounding regions. Though Markötter et al. did not definitively show the presence of a HAZ, they suspect the presence of a more hydrophilic, HAZ, around their perforations. The additional liquid water storage and the redistribution of liquid water in both the in- and through-plane directions confirm the conclusion ascertained from the EIS results. 140

161 Namely, two mechanisms lead to enhanced performance of the 100-μm w/haz DM: i) liquid water storage and interfacial through-plane redistribution rehydrate the CL and membrane, and ii) in-plane liquid water redistribution allow for increased oxygen transport to the reaction sites. The first mechanism is dominant at low current density (Figure 5.5a), and the second is dominant at moderate current densities (Figure 5.5b). Both mechanisms contribute to increased performance at high current densities (Figure 5.5c), where a 25% power density increase and a 7% limiting current density increase were observed. 50 L C L C L C L x1 x2 x3 40 Water Mass (mg cm -3 ) Position (mm) Figure 5.7. In-plane water content of a 17.2-cm 2 cell with 300-μm w/haz DM at 50% RH at constant current of 1.7 A cm -2. Periodic peaks (regardless of land-channel configuration) indicate water accumulation, and are the exact spacing of perforations (slightly less than 1 mm center-to-center spacing). 141

162 In light of the results presented here at 50% relative humidity, it is important to note several independent groups are achieving significant performance improvements with a common thread. Recently, some groups have directly identified their performance improvements to be caused by water-retention optimization on the cathode side DM; other groups have not explicitly stated that, but their results may be attributed to this effect. For example, Gerteisen et al. [41, 42] observed performance enhancements with the introduction of perforations on the cathode-side DM, likely due to similar reasons shown in the present work. Kitahara et al. [43] introduced a hydrophilic layer on top of an already-existing hydrophobic MPL, and they observed an increase in performance in low humidity conditions. Blanco et al. [44] has shown that the use of a perforated metallic layer on the cathode side (added to traditional DM) led to greater water retention, which boosts performance and durability in dry conditions. In summary, there are a growing number of publications suggesting that the bulk diffusion media, MPL, and/or the CL MPL interfacial structure is lacking significant optimization and understanding. The 25% power density increase shown in the present work by modifying the DM and CL MPL interfacial structure is a clear piece of evidence that further research can result in further significant improvements TESTING AT 120% RELATIVE HUMIDITY Over-humidified inlet gas streams occur during cold start-up, and local RH can become over-humidified toward the exit of a stack when water generation is high. Therefore, inducing over-humidified conditions is a desirable test condition to understand performance characteristics under these transient conditions. 142

163 Polarization Testing at 120% RH: Figure 5.8a shows the IR-corrected polarization behavior and HFR values of the PEFCs in over-humidified conditions at 75 o C. Throughout the range of current densities, the HFR values for all the cells are comparable, indicating the membrane is fully hydrated in the over-humidified gas stream. The perforations do not lead to lower HFR as they did in the 50% RH conditions. It is evident that the over-humidified conditions have caused excessive performance loss for the 300-μm w/haz DM, showing a maximum current density of 0.1 A cm -2 (open squares). It was confirmed with EIS (Figure 5.8b) that the excessive losses are due to oxygen-transport resistances, namely due to flooding, evidenced by a comparatively large-diameter impedance arc between 0.1 and 10 Hz for the 300-μm w/haz DM. Due to the hydrophilic nature of the laser-treated 300-μm w/haz DM, high liquid water saturation and surface coverage exists under these conditions. 143

164 IR Compensated Voltage (V) Virgin DM 300 m w/haz 100 m w/haz 100 m no HAZ 75 o C, 120% RH A/C Current Density (A cm -2 ) HFR (mohm) - Z'' Imaginary Impedance (Ohm cm 2 ) Hz H <1 H > 3000 Hz (a) Virgin DM 300 m w/haz 100 m w/haz 100 m no HAZ 75 o C, 120% RH A/C A cm -2 ) HFR (mohm) - Z'' Imaginary Impedance (Ohm cm 2 ) 1.4 Virgin DM m w/haz 100 m w/haz m nohaz Hz 10 Hz <1 Hz A cm <1 Hz 100% RH > 3000 Hz Z' Real Impedance (Ohm cm 2 ) (b) Figure 5.8. (a) Polarization data and (b) impedance spectra at 120% RH. Solid symbols indicate DM with nohaz, open symbols indicate DM w/haz. 144

165 At moderate current densities, the virgin DM and the 100-μm perforated DM display similar performance. It should be noted that in over-humidified conditions, the choice of perforation diameter is crucial. At 50% RH in moderate current densities, the key parameter for performance improvement was laser selection (compare DM w/haz and DM nohaz in Figure 5.3a). At 120% RH in moderate current densities, however, the key parameter is perforation diameter, evidenced by the difference between 300-μm and 100- μm DM performance. Beyond 1.1 A cm -2, the 100-μm DM (both w/haz and nohaz) display increased losses compared to the virgin DM, and the 100-μm nohaz underperforms the 100-μm w/haz DM by 10% (averaged from current densities 1.1 to 1.9 A cm -2 ). Through a basic calculation based on the inlet flow rates, the dew point of the inlet gas, and the cell temperature, it is estimated that liquid dropout occurs at a rate equivalent to 30 A cm -2 of water generation at the tested conditions. Given this high amount of liquid water dropout, especially in and near the flow channel, some estimates can be made to predict the polarization behavior in Figure 5.8a. Figure 5.9 shows a schematic of the suggested liquid water transport in the perforated DM. Due to the high liquid-phase saturation in and near the flow channels (toward the tops of the schematics shown), water may tend to aggregate in the laser-created perforation and flow toward the anode side, similar to the capillary mechanism shown in Ref. [30] (but in the opposite direction due to the inverted saturation curve suggested here). In DM w/haz, the hydrophilic zones around the perforation inhibit this capillary-driven flow, which leads to two results: i) increased DM saturation, especially in the perforation and HAZ, and ii) less liquid water being transported toward the CL and the CL MPL interface. Therefore, the 100-μm 145

166 w/haz DM will show higher mass-transport resistance compared to the virgin DM due to high level of DM saturation; however, interfacial and CL flooding will be mitigated since liquid water is more likely stored in the DM rather than transported toward the anode side. According to this suggested mechanism, then, the 100-μm nohaz DM will show increased losses because liquid water is more readily transported to flood the CL and the CL MPL interface compared to the 100-μm w/haz DM. It is evident that perforations incur more severe mass-transport losses compared to virgin DM when operated in over-humidified conditions. As previously mentioned, overhumidified conditions can exist in automotive cell stacks, even when low-humidity inlet gas streams are used (e.g., startup, high current density, etc.). Therefore, a customized gradient in perforation density (perforations per planform area) over the active area may be desirable for optimized performance. Also, compared to the other techniques previously discussed (e.g., adding layers to existing DM for improved performance [43, 44]), laser perforations seem to be a more customizable option. With additional layers, it will likely be an all or nothing addition, whereas perforation properties can easily be customized, as proposed, for optimal performance. 146

167 HAZ Boundary Figure 5.9. Schematic of liquid water transport through perforations w/haz and nohaz based on high levels of water condensation in the flow channels due to over-humidified inlet gas streams. 5.4 CONCLUSIONS The purpose of this work was to explore fundamental transport and performance characteristics of engineered interfacial and structure architecture in fuel cell diffusion 147

168 media. A variety of laser perforations were introduced onto the cathode-side diffusion media (DM), changing perforation diameter and laser type, in order to determine underlying processes that affect overall performance. Namely, 100-μm and 300-μm diameter perforations were respectively introduced on the cathode-side DM. It was found that, depending on the laser selection, PTFE content was selectively removed from an area around the perforations referred to as the heat affected zone (HAZ). It was found that the 100-μm w/haz DM boasted a 25% power density increase compared to a virgin DM when operated at 50% relative humidity; however, the 100-μm nohaz DM did not show any improvements compared to the virgin DM. Investigation of how the current varies with oxygen concentration in the double-tafel region showed that transport resistances were dominated by oxygen transport for both DM with and without a HAZ at moderate current values. Electrochemical impedance spectroscopy and neutron radiography results indicated that charge- and mass-transport impedances and liquid water redistribution are significantly different between DM with HAZ and DM without HAZ and are dependent on the operating current density. Based on the experimental results presented, two mechanisms are suggested that result in a 25% power density increase and a 7% limiting current density increase for the 100-μm w/haz DM compared to the virgin DM. The first mechanism, dominant in low-current operation, shows that perforations with a HAZ enable greater liquid water storage and through-plane water redistribution that rehydrate the CL and membrane. The second mechanism, more dominant at moderate currents, is an in-plane liquid water redistribution that allows for increased oxygen transport to the reaction sites. Both mechanisms contribute to increased performance at high current densities in reduced RH 148

169 environments, where a 25% power density increase and a 7% limiting current density increase were observed. Non-optimized perforations (nohaz or over-sized diameter) lead to high losses and emphasize the importance of optimization. Finally, over-humidified conditions were tested to test perforation performance in non-ideal conditions, such as toward the exit of a stack or in cold start-up conditions. Results suggested that virgin DM be used instead of perforated DM where overhumidified conditions are likely. Perforation properties (diameter, spatial density, etc.) can be easily customized to suit the needs of a variety of applications and designs, making it a very attractive solution for increased power density while keeping parasitic humidification losses to a minimum. 149

170 REFERENCES [1] M. M. Mench, Fuel Cell Engines, pp , John Wiley & Sons, Inc., Hoboken (2008). [2] M. Mathias, J. Roth, J. Fleming and W. Lehnert, in Handbook of Fuel Cells: Fundamentals, Technology and Applications, W. Vielstich, A. Lamm and H. A. Gasteiger (Eds.), Vol. 3, pp , John Wiley & Sons, Hoboken (2003). [3] E. L. Cussler, Diffusion: Mass Transfer in Fluid Systems, pp , Cambridge University Press, New York (2009). [4] U. Beuscher, J. Electrochem. Soc., 153, A1788 (2006). [5] M. P. Manahan, S. Kim, E. C. Kumbur and M. M. Mench, ECS Transactions, 25, 1745 (2009). [6] D. R. Baker, D. A. Caulk, K. C. Neyerlin and M. W. Murphy, J. Electrochem. Soc., 156, B991 (2009). [7] M. L. Perry, J. Newman and E. J. Cairns, J. Electrochem. Soc., 145, 5 (1998). [8] F. Jaouen, G. Lindbergh and G. Sundholm, J. Electrochem. Soc., 149, A437 (2002). [9] J. Ihonen, F. Jaouen, G. Lindbergh, A. Lundblad and G. Sundholm, J. Electrochem. Soc., 149, A448 (2002). [10] K. O Neil, J. P. Meyers, R. M. Darling and M. L. Perry, Int. J. Hydrogen Energy, 37, 373 (2012). [11] M. S. Wilson and S. Gottesfeld, J. Appl. Electrochem., 22, 1 (1992). [12] W. Dai, H. Wang, X.-Z. Yuan, J. Martin, J. Shen, M. Pan and Z. Luo, J. Power Sources, 188, 122 (2009). 150

171 [13] J. H. Nam, K.-J. Lee, G.-S. Hwang, C.-J. Kim and M. Kaviany, Int. J. Heat Mass Transfer, 52, 2779 (2009). [14] H. Meng, Int. J. Hydrogen Energy, 34, 5488 (2009). [15] J. P. Owejan, J. E. Owejan, W. Gu, T. A. Trabold, T. W. Tighe and M. F. Mathias, J. Electrochem. Soc., 157, B1456 (2010). [16] J. T. Gostick, M. A. Ioannidis, M. W. Fowler and M. D. Pritzker, Electrochem. Commun., 11, 576 (2009). [17] M. P. Manahan, M. C. Hatzell, E. C. Kumbur and M. M. Mench, J. Power Sources, 196, 5573 (2011). [18] R. P. Ramasamy, E. C. Kumbur, M. M. Mench, W. Liu, D. Moore and M. Murthy, Int. J. Hydrogen Energy, 33, 3351 (2008). [19] A. Turhan, K. Heller, J. S. Brenizer and M. M. Mench, J. Power Sources, 180, 773 (2008). [20] A. Turhan, S. Kim, M. Hatzell and M. M. Mench, Electrochim. Acta, 55, 2734 (2010). [21] M. Khandelwal, S. Lee and M. M. Mench, J. Electrochem. Soc., 156, B703 (2009). [22] S. Kim and M. M. Mench, J. Membr. Sci., 328, 113 (2009). [23] S. Kim and M. M. Mench, J. Electrochem. Soc., 156, B353 (2009). [24] S. Kim, M. Khandelwal, C. Chacko and M. M. Mench, J. Electrochem. Soc., 156, B99 (2009). [25] M. Khandelwal and M. M. Mench, J. Power Sources, 195, 6549 (2010). [26] M. C. Hatzell, A. Turhan, S. Kim, D. S. Hussey, D. L. Jacobson and M. M. Mench, J. Electrochem. Soc., 158, B717 (2011). 151

172 [27] H. Bajpai, M. Khandelwal, E. C. Kumbur and M. M. Mench, J. Power Sources, 195, 4196 (2010). [28] T. Swamy, E. C. Kumbur and M. M. Mench, J. Electrochem. Soc., 157, B77 (2009). [29] K. T. Cho and M. M. Mench, J. Power Sources, 195, 3858 (2010). [30] U. Pasaogullari and C. Y. Wang, J. Electrochem. Soc., 151, A399 (2004). [31] E. C. Kumbur, K. V. Sharp and M. M. Mench, J. Electrochem. Soc., 154, B1315 (2007). [32] S. Park and B. N. Popov, Fuel, 88, 2068 (2009). [33] M. Uchida, Y. Aoyama, N. Eda and A. Ohta, J. Electrochem. Soc., 142, 4143 (1995). [34] M. Eikerling, J. Electrochem. Soc., 153, E58 (2006). [35] G.-G. Park, Y.-J. Sohn, T.-H. Yang, Y.-G. Yoon, W.-Y. Lee and C.-S. Kim, J. Power Sources, 131, 182 (2004). [36] L. Giorgi, E. Antolini, A. Pozio and E. Passalacqua, Electrochim. Acta, 43, 3675 (1998). [37] S. Park, J.-W. Lee and B. N. Popov, J. Power Sources, 177, 457 (2008). [38] T. Tanuma and S. Kinoshita, J. Electrochem. Soc., 159, B150 (2012). [39] A. J. Mendoza, M. A. Hickner, J. Morgan, K. Rutter and C. Legzdins, Fuel Cells, 11, 248 (2011). [40] M. P. Manahan and M. M. Mench, ECS Transactions, 16 (2011). [41] D. Gerteisen, T. Heilmann and C. Ziegler, J. Power Sources, 177, 348 (2008). [42] D. Gerteisen and C. Sadeler, J. Power Sources, 195, 5252 (2010). 152

173 [43] T. Kitahara, H. Nakajima and K. Mori, J. Power Sources, 199, 29 (2012). [44] M. Blanco, D. P. Wilkinson and H. Wang, Int. J. Hydrogen Energy, 36, 3635 (2011). [45] H. Markötter, R. Alink, J. Hau mann, K. Dittmann, T. Arlt, F. Wieder, C. Tötzke, M. Klages, C. Reiter, H. Riesemeier, J. Scholta, D. Gerteisen, J. Banhart and I. Manke, Int. J. Hydrogen Energy, in press (2012). [46] M. Eikerling and A. A. Kornyshev, J. Electroanal. Chem., 475, 107 (1999). [47] K. A. Striebel, F. R. McLarnon and E. J. Cairns, Ind. Eng. Chem. Res., 34, 3632 (1995). [48] M. L. Perry, R. Balliet and R. M. Darling, in Polymer Electrolyte Fuel Cell Degradation, M. M. Mench, E. C. Kumbur and T. N. Veziroglu (Eds.), pp , Elsevier, Boston (2012). [49] A. J. Bard and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, pp , John Wiley & Sons, Hoboken (2001). [50] E. Ivers-Tiffee, A. Weber and H. Schichlein, in Handbook of Fuel Cells: Fundamentals, Technology and Application, W. Vielstich, A. Lamm and H. A. Gasteiger, Editors, Vol. 2, pp , John Wiley & Sons, Hoboken (2003). [51] D. Malevich, E. Halliop, B. A. Peppley, J. G. Pharoah and K. Karan, J. Electrochem. Soc., 156, B216 (2009). [52] Y. Bultel, K. Wiezell, F. Jaouen, P. Ozil and G. Lindbergh, Electrochim. Acta, 51, 474 (2005). [53] U. Krewer, M. Christov, T. Vidakovic and K. Sundmacher, J. Electroanal. Chem., 589, 148 (2006). 153

174 [54] R. Makharia, M. F. Mathias and D. R. Baker, J. Electrochem. Soc., 152, A970 (2005). [55] T. Navessin, M. Eikerling, Q. Wang, D. Song, Z. Liu, J. Horsfall, K. V. Lovell and S. Holdcroft, J. Electrochem. Soc., 152, A796 (2005). 154

175 6. LASER MODIFIED FUEL CELL DIFFUSION MEDIA: ROLE OF THE MICROPOROUS LAYER (MPL) AND HEAT AFFECTED ZONES The results of the previous chapters have led this work to an innovative concept of obtaining a more optimized diffusion medium (DM) structure while mitigating degradation modes. In particular, through collaboration with the Applied Research Laboratory Laser Processing Facility at The Pennsylvania State University, a novel DM modification technique was introduced. By engineering laser parameters, a diffusion medium was modified such that heat affected zones were introduced but no perforation was created. This case, referred to as 0 µm w/haz DM, was compared to the previously-obtained 100 µm w/haz DM a diffusion medium with 100-µm diameter perforations with heat affected zones surrounding the perforation. This comparison lends insight into the role of the MPL and the MPL interface since the 0 µm w/haz DM has an intact, unaltered MPL, whereas the 100 µm w/haz DM has perforations going through the entire DM and MPL. The data suggest the MPL has an impact that depends on operating conditions. In-situ net water drag experiments indicate that in low humidity and low-to-moderate current densities, the cathode-side MPL forces more liquid- and/or vapor-phase water to back diffuse from the cathode to the anode. However, when more water is produced or the inlet streams are close to saturation, the MPL acts as a barrier to back diffusion, meaning the MPL prevents liquid water in the cathode DM from moving toward the anode. Furthermore, a computational model was developed to investigate the thermal gradients introduced as a result of perforations. The present work adds understanding to the role of the microporous layer (MPL) and the laser-induced heat affected zone (HAZ) in polymer electrolyte fuel cell (PEFC) performance. 155

176 6.1 INTRODUCTION Polymer electrolyte fuel cells (PEFCs) continue to receive significant attention as a promising technology for clean energy applications. Water management remains a primary topic of discussion due to the delicate balance of having a well-humidified polymer electrolyte membrane for efficient proton conduction while concurrently maintaining open pathways for gas diffusion from the manifold/channels to the reaction site in the catalyst layer. Multi-phase transport between the inlet channels and the reaction site necessitates the introduction of an interfacial layer, often referred to as the diffusion medium (DM). In general, a DM is placed on both the anode and cathode to aid in the mass-transport of products and reactants, but it also functions as electrical and heat transport pathways and a physical support for the membrane [1, 2]. The DM typically consists of a fibrous macro-dm with pore diameters on the order of µm and a micro-porous layer (MPL) with pore diameters are on the order of 0.1 to 0.5 µm. These layers are often treated with polytetrafluoroethylene (PTFE) to prevent excessive water accumulation, which can lead to the phenomenon of flooding where gas reactants are blocked by water coverage from the catalyst site. Numerous studies investigated the effect and optimization of PTFE content in PEFCs [3-7]. The MPL continues to be a topic of interest in the fuel cell literature due to its complex and multi-faceted role in PEFC transport and performance. Authors have suggested that the small pore size of the MPL forces water toward the anode side of the cell (back diffusion) [8-11]. Other studies have shown the opposite, namely that the MPL 156

177 maintains the water in vapor phase from the catalyst layer (CL) to the macro-dm or flow channel, thus mitigating any liquid back diffusion [12-14]. Still, others suggest a combination of these effects, which the previous chapters of this dissertation have asserted as well. The behavior and role of the MPL is highly dependent on operating conditions and material properties. Electroosmotic drag moves water from the anode to the cathode side of the membrane [15-17]. A high enough water concentration on the cathode side can move water back to the anode compartment via back diffusion. Pressure gradients and thermal gradients can also affect water movement across the membrane in either direction [12, 18-22]. The net direction of water transport across the membrane is referred to as the net water drag (NWD). It is defined as the net amount of water molecules transferred across the membrane per proton transferred, from anode to cathode. The NWD is therefore sensitive to operating conditions such as gas flow rates and humidification and material properties like membrane thickness, porosity of the DM and MPL, and PTFE content [23-26]. Research that used NWD measurement to understand the effect of the MPL on water transport concluded that the NWD for cells with an MPL at the cathode compared to cells without an MPL were not statistically significant, although they noted that the presence of MPL significantly improved fuel cell performance [27, 28]. There is an ongoing effort to optimize the porous macro-dm and MPL to boost performance. Groups are attempting laser modifications [29-35], layer additions [36-38], metallic DM [39], and metallic foam [40-42]. Previously, Manahan et al. [29-31] have shown that laser-induced perforations can be beneficial to PEFC performance, noting a 25% increase in peak power density when 50% inlet relative humidity (RH) gas streams 157

178 are used with 100 µm perforations that have a locally hydrophilic heat affected zone (HAZ). Energy dispersive spectroscopy (EDS) revealed that the HAZ was an area around the perforation where PTFE was removed due to the laser processing. An appropriatelysized HAZ was shown to be critical in storing and redistributing the liquid water throughout the cell. This storage and redistribution was suggested as the leading cause for improved performance [29, 31]. However, in over-humidified conditions (120% RH), perforated DM showed increased mass transport at high current densities. Steady-state water content profiles using neutron radiography indicated the perforations acted as water pooling locations [29, 31], confirming previous suggestions from other authors [14, 33, 43] that perforations could be effective in collecting and transporting liquid water. Concurrent independent work by another group both confirmed these findings and added understanding to water transport changes due to laser-induced DM modifications [32, 33]. In their most recent work, Alink et al. [32] confirmed the findings of Manahan et al. [31], namely that the HAZ surrounding a laser perforation is a leading cause for improved performance under dry conditions. They also showed that creating perforations using milling machines (therefore removing any laser affects such as the HAZ) was more beneficial in high humidity conditions where liquid water transport becomes increasingly important but detrimental in low humidity conditions due to excessive dryout. Nonetheless, this high humidity result differs from a previous conclusion from Manahan et al. [31], who by using tailored laser treatment to create a perforation without a HAZ, still showed increased mass-transport losses in over humidified conditions. This difference between Manahan et al. and Alink et al. may be due to the different operating 158

179 conditions, different materials used, and/or different perforation patters, locations, and properties. This chapter investigates the use of laser-modified diffusion media not only as a practical and economical method to achieve enhanced PEFC performance, but also as a tool to further understand the role of the MPL and MPL interface in an operating cell. Specifically, insight into the role of the MPL can be gained from two of the modified DM, namely, the 0 µm w/haz and 100 µm w/haz DM. These DM were made such that both have a heat affected zone with approximately the same diameter. The main difference, however, is the 0 µm w/haz DM has an MPL that is completely intact, whereas the 100 µm w/haz DM has a perforation through the entire DM, including the MPL. The 100 µm perforations remove only 1.7% of the planform area, therefore keeping good interfacial contact at the MPL CL interface as the DM flow field interface. Nonetheless, the perforations create discrete breakthrough locations that pierce the MPL barrier and alter the multiphase transport and performance. This difference between the 0 µm and 100 µm w/haz DM permits comparison between these two cases. The experimental modifications lend insight into the functional roles of the DM, MPL, and MPL interface and are a potential pathway for further enhancement in performance. 6.2 EXPERIMENTAL AND MODELING APPROACH EXPERIMENT AND SETUP The experimental setup is the same as those listed in Chapters 4 and 5. Any additional information is provided here for clarity. 159

180 6.2.2 LASER MODIFICATION SGL 10 BB diffusion media samples were custom-perforated using two different lasers with a variety of different parameters to achieve specified results. Only the DM on the cathode side was modified. The modifications made to the DM are summarized in Table 6-1. A ytterbium fiber laser (IPG Photonics, Oxford, MA) and a neodymium-doped yttrium vanadate (Nd:YVO 4 ) laser (Coherent, Santa Clara, CA) were used. Previously, it was concluded that the perforation and HAZ combined to produce improved performance in low humidity operating condition [29-31]. However, due to recent evidence that a perforation can lead to increased degradation in the CL [32] and more likely DM damage in freeze/thaw conditions, it may be desirable to avoid a perforation but obtain a heat affected zone in order to avoid degradation and still achieve performance gains. Therefore, in this work, a special laser treatment was applied such that no perforation was introduced, however a HAZ was created. This sample is referred to as 0 µm w/haz since there is no perforation but there exists a HAZ. In addition to investigating the performance characteristics, this scenario is well-positioned to further investigate the role of the MPL and the MPL CL interface. That is, when comparing the 0 µm w/haz DM with the 100 µm w/haz DM, the principal difference between the two is the existence of an intact MPL. The 100 µm w/haz DM has a perforation through the DM and MPL, whereas the 0 µm w/haz DM exhibits only a HAZ, leaving the MPL undamaged. The other samples are described by their perforation diameter and the presence or absence of a HAZ as shown in Table 6-1. Further details describing the other samples properties are described in Chapters 4 and

181 Table 6-1. Material nomenclature and physical effects. Designation Laser Type Percent Area Heat Affected Zone Removed (HAZ) / Diameter of HAZ Virgin DM None 0% None 0 µm w/haz Yb Fiber 0% Yes / ~200 µm 100 µm w/haz Yb Fiber 1.7% Yes / ~200 µm 100 µm no-haz Nd:YVO 4 1.7% None Observed 300 µm w/haz Yb Fiber 15% Yes / Entire Surface SPECTROSCOPY A Hitachi TM-3000 tabletop microscope equipped with a Bruker Quantax 50 EDS system was used to determine the elemental composition of each DM sample, i.e., to determine the presence or absence of PTFE due to any laser treatment ELECTROCHEMICAL TESTING The conditions from Chapters 4 and 5 were replicated in this chapter for consistency among results. An abbreviated experimental description will be given here for clarity. Polarization tests were conducted at a coolant-controlled cell temperature of 75 o C in 50% relative humidity (RH) condition on both the anode and cathode, as well as in 120% RH conditions. Reactant flow rates were held constant at 139 standard cubic centimeters per minute (sccm) H 2 (anode) and 332 sccm air (cathode), which corresponded to a 2/2 stoichiometry at 2 A cm -2. Quasi-steady state conditions were achieved by using galvanostatic steps for 3-minute intervals at each current value. Tests were repeated at least three times to ensure repeatability, and error bars are shown on the polarization plots to indicate the maximum and minimum averaged value for each set of tests. 161

182 Electrochemical impedance spectroscopy (EIS) was employed to understand chargeand mass-transport resistances while operating the cell. Experiments were conducted at constant current on a Scriber 850C (Scribner Associates, Southern Pines, NC) test stand to control the flow rates (3/3 stoichiometry), humidifiers (both 50% and 100% RH conditions), and current, and an 880 Impedance Analyzer (Scribner Associates, Southern Pines, NC) was used to apply the AC signal frequencies from 10 khz to 0.1 Hz. The cell temperature was coolant-maintained at 75 o C. The data are also shifted on the realimpedance axis to remove the high-frequency resistance differences and to better compare the other competing effects (i.e., charge and mass transfer) [44-46]. Tafel experiments here attempted to mimic the conditions suggested by Refs. [47-49]. The cell was kept at 75 o C and the flow rates were set at a constant flow rate equivalent to 8/8 stoichiometry at 1 A cm -2 on the anode and cathode, respectively, in order to minimize oxygen gradients in the DM. The voltage was scanned from OCV to 0.3 V at a scan rate of 2 mv s -1. The total pressure of the cathode gas stream was tested at both 101 kpa and 202 kpa (absolute) to vary oxygen concentration, and the anode stream was left open at atmospheric. Tests were conducted on the 0 µm w/haz, 100 μm w/haz, and 100-μm no-haz cells at 50% RH NET WATER DRAG According to the net water drag definition, the NWD coefficient can be written as: (6.1) where is the molar flow rate of water in or out of the anode. The average NWD was measured by using the same apparatus described in [50], from which the most important 162

183 details are repeated here. The water at the anode outlet was collected using a desiccant bottle filled with Dryerite (anhydrous CaSO 4, W. A. Hammond Drierite Co.). The change in mass,, was recorded after operating each point for around 40 minutes. The mass of water coming in the cell from the humidifiers at the anode,, is known from the operating parameters. The net water drag coefficient is therefore rewritten as, ( ) (6.2) For improved confidence in the results, water out of the cathode was also measured by desiccant in order to verify water mass conservation according to, (6.3) where the mass of water generated during the measurement is computed from galvanostatic operation, ( ) (6.4) In all tests mass from Eq. (6.3) was conserved with a maximum deviation of 7%, indicating a good steady-state measurement MODEL DEVELOPMENT An in-house computational model was developed to investigate the effect of laser perforation on thermal transport characteristics of the DM. A finite volume discretization was implemented on a two-dimensional computational domain to evaluate the steadystate cell temperature distribution in the through-plane direction. Some key model assumptions include: 163

184 1. Anisotropic material properties; homogeneous along each axis, 2. Pure conductive transport within perforation with the thermal properties of water vapor, 3. Zero heat flux across CL membrane interface, 4. Constant-temperature at the DM-bipolar plate interface, 5. Negligible contact resistance between layers, and 6. Uniform heat generated within CL. A primary interest of this study was the temperature distribution and its possible effect on performance, transport, and durability. The present model and assumptions are constructed to create a worst case scenario where the highest changes in temperature should be seen (e.g., assumptions 2 and 3). The domain was defined to span in width from perforation axis to halfway between perforations, establishing symmetry conditions on both left and right boundaries. The domain extends from the CL to the flow field. This steady-state heat transfer process with energy generation is governed by the heat diffusion equation, (6.5) where and represent material thermal conductivities, as summarized in Table 6-2. Two source terms exist in PEFC systems ( ): generation from an exothermic chemical reaction and resistive heating due to electrical current, also referred to as joule heating. For the present formulation of the cathode side only, the exothermic reaction generation term exists only within the catalyst layer and is determined by cell operating conditions and the thermodynamic potential of the hydrogen-oxygen redox couple. Joule heating 164

185 occurs throughout the remaining solid-phase domain assumed to be only in the membrane and is dictated by cell current and material properties [51]. The generation term can be written as a lumped form of kinetic, mass, and entropy losses with joule heating as: ( ) (6.6) where and represent cell operating conditions, is the thermodynamic potential or thermal voltage, and is either or, the electrical conductivities in the throughand in-plane direction (Table 6-2). Note that only for the electronic transport in the CL under a perforation is the in-plane electronic conductivity used since that is the only conduction path. Otherwise, the through-plane values are used. Constant temperature boundary condition was applied at the flow field DM interface, and symmetry boundary conditions (zero flux) were implemented at the remaining three faces. Table 6-2. Physical material properties. Parameter Description DM MPL CL Perforation Unit t Component thickness μm In-plane thermal conductivity 4.2 [52] 4.2 [52] 2.7 [52] (m ) Through-plane thermal conductivity 0.42 [52] 0.42 [52] 0.27 [52] (m ) In-plane electronic conductivity 3000 [53] 300 [53] 200 [53] - m Through-plane electronic conductivity 300 [53] 300 [53] 200 [53] - m 165

186 6.3 RESULTS AND DISCUSSION SPECTROSCOPY As shown in Figure 6.1, laser properties can be tuned to obtain a custom range of DM parameters. In these images, the red color represents carbon, and the blue represents fluorine, which arises from the presence of PTFE. The heat affected zone refers to the area surrounding the actual perforation where the heat of the laser has removed PTFE but left the carbon. The term with HAZ will refer to the samples containing at least some region where PTFE was removed (see also Table 6-2). 166

187 a b carbon, fluorine 300 µm c d carbon, fluorine 300 µm carbon, fluorine 200 µm e f carbon, fluorine 300 µm carbon, fluorine 300 µm Figure 6.1. (Color online) (a) Drawing of perforation configuration, indicating the possibility of a HAZ depending on laser properties. (b-f) SEM images of each DM with EDS color mapping. Red (dark) represents carbon, and blue (light) represents PTFE. (b) MPL of a Virgin DM and 0 µm w/haz, (c) macro-dm of 0 µm w/haz, (d) MPL of 100 µm w/haz, (e) MPL of 100 µm without HAZ, and (f) MPL of 300 µm w/haz. 167