CHEMICAL DURABILITY STUDIES OF IONOMERS AND MODEL COMPOUNDS FOR FUEL CELL APPLICATIONS

Transcription

1 CHEMICAL DURABILITY STUDIES OF IONOMERS AND MODEL COMPOUNDS FOR FUEL CELL APPLICATIONS by CHUN ZHOU Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy Dissertation advisors: Dr. David A. Schiraldi Dr. Thomas A. Zawodzinski, Jr Department of Macromolecular Science and Engineering Case Western Reserve University May, 2008

2 CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES We hereby approve the thesis/dissertation of candidate for the degree *. (signed) (chair of the committee) (date) *We also certify that written approval has been obtained for any proprietary material contained therein.

3 To my parents, to my wife, Ning, to my brother, for all your love and support along my journey

4 Table of Contents Table of Contents... List of Tables... List of Schemes... List of Figures... Acknowledgements... Abstract... i viii x xi xx xxi Chapter 1 Introduction Fuel Cell Basics Polymer Electrolyte Membrane Fuel Cells Catalyst, Electrodes, Membrane Materials, and Complete Fuel Cell Stacks Catalyst and Electrodes Other PEMFC Components and Fuel Cell Stacks Polymer Electrolyte Membranes Nafion - Synthesis, Solid State Structure, and Proton Conductivity Alternative Membrane Candidates a Other Perfluorinated Sulfonic Acids (PFSAs) b Styrene Based Ionomers c Post-Sulfonated Aromatic Polymers d Direct Copolymerization of Sulfonated Comonomers i

5 e Sulfonated Polyimides (SPI) Composite Membranes a Polymer Blends b Membranes Doped with Organics / Potential Anhydrous Membranes c Ionomer/Inorganic Particle Composites References Chapter 2 Literature Review and Research Overview Fuel Cell Challenges and Durability Durability Studies of Various Components End Plates and Current Collectors Sealing Gaskets Bipolar Plates Gas Diffusion Layer (GDL) Electrodes and Catalyst Literature Review of Durability Studies of Membranes Mechanical Degradation Chemical Degradation Chemical Degradation of Nafion PFSAs Objectives and Significance of Current Research References ii

6 Chapter 3 Degradation of Model Compounds Under Mimic Fuel Cell Conditions Introduction Materials Experiments Fluoride Concentration Measurement Fenton s Degradation Tests Degradation Test Procedures a Mild Condition Fenton s Degradation Test and Procedure b Harsh Condition Fenton s Degradation Test and Procedure in situ CO 2 Detection in Degradation Test UV Photolysis Degradation Test Intermediate Trapping by Stable Radicals F Nuclear Magnetic Resonance (NMR) Liquid Chromatography-Mass Spectrometry (LC-MS) Results and Discussion Fluoride Concentration Measurement Accuracy and Interferences a Sample ph b Ferrous Ion Concentration c Ferric Ion Concentration iii

7 3.4.2 Fluoride Generation from Fenton s Degradation Test Fluoride Generation from UV Photolysis Degradation Test Degradation Product Analysis by LC-MS a Degradation Product Analysis of MC b Degradation Product Analysis of MC c Degradation Product Analysis of MC d Degradation Product Analysis of MC Degradation Intermediate Trapping Experiments for MC7 and MC a Brief Review of Degradation Studies of Fluoroethers b Trapping Experiment Results and Proposed Mechanism Other Control Experiments of Degradation a Acid Catalysis Effect and Ether Hydrolysis b UV Photolysis Degradation of MC c MC Degradation as a Function of Hydrogen Peroxide Concentration Conclusions References Chapter 4 Chemical Degradation and Structure-and-Property Change of PFSA Ionomers Introduction Experiments iv

8 4.2.1 Fenton s Degradation Procedure for Membrane Samples a Materials and Membrane Pretreatment Protocol b Mild Fenton s Degradation Test c Modified Fenton s Degradation Test Fragments Collections and Extraction from Membranes LC-MS Experiment Fourier Transform Infrared Spectroscopy (FT-IR) Characterization Membrane Weight Loss, Water Up-take, and Equivalent Weight (EW) Measurements Membrane Proton Conductivity Measurement Scanning Electron Microscope (SEM) Characterization Dynamic Mechanical Analysis (DMA) Characterization Tensile Testing Differential Scanning Calorimetry (DSC) Characterization Wide Angle X-Ray Diffraction (XRD) Characterization Results and Discussion Fluoride Generation Fragments Identification Major Changes of Critical Membrane Properties a Conductivity b Weight Loss and Thickness Change c EW Change v

9 4.3.4 FT-IR (ATR) Analysis of Degraded Membrane Samples Other Comparison of Solid-State Properties a DSC Results b DMA Test and Tensile Test Results c Morphology Comparison and XRD Results Conclusions References Chapter 5 Overall Summary and Suggested Future Work Overall Summary Suggested Future Directions a MC Systems b Ionomer Systems c New Membrane Development Appendix 1 XRD and SEM Results for Pristine and Degraded Membranes A.1.1 XRD Results A Nafion A M-G1-NM A M-G1-PM A M-G1-BM vi

10 A.1.2 SEM Micrographs A Surface SEM Images a Nafion b 3M-G1-NM c 3M-G1-PM d 3M-G1-BM A Cross-section SEM Images a-1 Nafion Original a-2 Nafion Degraded b-1 3M-G1-NM Original b-2 3M-G1-NM Original c-1 3M-G1-PM Original c-2 3M-G1-PM Original d-1 3M-G1-BM Original d-2 3M-G1-BM Original Biblography vii

11 List of Tables Chapter 1 Table 1-1 Comparison of Various Fuel Cell Technologies... 3 Table 1-2 Methods of Catalyst Layer Preparation and Application... 6 Chapter 3 Table 3-1 Concentrations of Reagents Used in Mild Degradation Test Table 3-2 Concentrations of Reagents Used in Harsh Degradation Test Table 3-3 Tabulated Parent and Adduct Ions of TFA and PFPA Table 3-4 Proposed Degradation Products of MC8 with Expect Parent and Adduct Ions Table 3-5 Proposed Degradation Products of MC7 Based on LC-MS Analysis Table 3-6 Table 3-7 Tabulated Results of 4-hydroxy-TEMPO-trapped Radical Adduct from Various MC7 Degradation Aliquots... Tabulated Results of 4-hydroxy-TEMPO-trapped Radical Adduct from Various MC8 Degradation Aliquots Chapter 4 Table 4-1 Sample Designations and Characteristics of Various Ionomers Examined Table 4-2 Concentrations of Reagents Used in Mild Membrane Degradation Tests. 146 Table 4-3 Parent Ion and Derivative Ions of the Expected Product of 3M-G1-NM Table 4-4 Conductivity Change for Nafion and 3M Ionomers viii

12 Table 4-5 Weight Loss and Thickness for Nafion and 3M Ionomers Table 4-6 EW Change after Degradation for Various Membrane Samples Table 4-7 Common IR Peaks of Nafion and 3M Membranes ix

13 List of Schemes Chapter 3 Scheme 3-1 Diagram Showing the Unzipping Degradation From -COOH Chain End Scheme 3-2 Proposed Degradation Mechanism for MC Scheme 3-3 Schematic Summary of Degradation Product Identified from MC7 and MC Scheme 3-4 Proposed Degradation Route-1a for Ether Cleavage of - CF 2 -CF 2 -O- CF 2 - CF 2 - Structure Scheme 3-5 Proposed Degradation Route-1b for Ether Cleavage of - CF 2 -CF 2 -O- CF 2 - CF 2 - Structure Scheme 3-6 Proposed Degradation Route-2a for Ether Cleavage of - CF 2 -C(F)CF 3 - O-CF 2 - CF 2 - Structure Scheme 3-7 Proposed Degradation Route-2b for Ether Cleavage of - CF 2 -C(F)CF 3 - O-CF 2 - CF 2 - Structure Scheme 3-8 Proposed Degradation Route-2c for Ether Cleavage of - CF 2 -C(F)CF 3 - O-CF 2 - CF 2 - Structure Scheme 3-9 Proposed Degradation Route-2d for Ether Cleavage of - CF 2 -C(F)CF 3 - O-CF 2 - CF 2 - Structure x

14 List of Figures Chapter 1 Figure 1-1 Summary of the Applications and Advantages of Various Fuel Cell Technologies and Their Applications... 4 Figure 1-2 Schematic Diagram of a Membrane/Electrode Assembly (MEA)... 5 Figure 1-3 Schematic Diagram of Idealized Electrode Structure... 7 Figure 1-4 Diagram of a Single MEA Fuel Cell with Enlarged Cross-section Showing Structural Details... 8 Figure 1-5 Diagram of a Three Cell Fuel Cell Stack with Two Bipolar Plates and Two End Plates... 9 Figure 1-6 Chemical Structure of Nafion and the Synthetic Route of the Vinyl Ether monomer Figure 1-7 Cluster-network Model for the Morphology of Hydrated Nafion Figure 1-8 Evolution of a Smaller Number of Larger Clusters with Increased Hydration of Nafion Figure 1-9 Schematic Illustration of Membrane Hydration Level of Nafion Figure 1-10 Chemical Structures of Various Commercial and Development PFSAs 18 Figure 1-11 Radical Polymerization and Processing Conditions of Commercial PFSAs Figure 1-12 Chemical Structures of Various Styrenic Sulfonic Acid Ionomers Figure 1-13 Synthetic Route of BAM Type PSSAs Figure 1-14 Chemical Structures of Various Post-sulfonated Ionomers Figure 1-15 Schematic Representation of the Microstructures of Nafion and a xi

15 sulfonated PEEK Figure 1-16 Reactivity and Stability from the Placement of Sulfonic Acid Group in Postsulfonation versus Direct Copolymerization Figure 1-17 AFM Micrographs of BPSH-40 and Nafion Figure 1-18 Synthetic Route of Model Imide Compounds A and B Figure 1-19 Various Proton Transfer Pathways in PBI Doped with H 3 PO Figure 1-20 Chemical Structures of Typical Nitrogen-containing Aromatic Heterocycles Figure 1-21 Current-voltage Response of a Recast Nafion film (open circle) versus a Titania Composite Film (solid square) under Low Humidity at 115 o C Operating Temperature Chapter 2 Figure 2-1 Scheme of a Single PEM Fuel Cell Showing the Location of the Components Figure 2-2 Back Scattered Images by Electron Probe Micro Analyzer for MEA before and after 1800 hour of operation. (a) Anode, before; (b) cathode, before; (c) anode, after; (d) cathode, after Figure 2-3 TEM Micrographs from (a) Pristine Pt/Vulcan Sample; (b) Powders Scraped from the Cathode Surface of the Cycled MEA Sample Figure 2-4 Carbon Corrosion in the Absence of Pt Figure 2-5 Possible Sites for Radical Attack on Several Styrenic PFSAs Figure 2-6 Mechanisms (a-c) for Radical Attack and Degradation of Membrane xii

16 Containing Aromatic Groups Figure 2-7 Chain End un-zipping Degradation Mechanism Figure 2-8 The Formation of Carboxylic Acid End Groups in Fluoropolymer from the Hydrolysis of the Residue of Persulfate Initiator Figure 2-9 Plot showing relative fluoride emission rate (FER) from Fenton s test as a function of concentration of reactive end-groups (recreated from original plot in reference 46) Chapter 3 Figure 3-1 Possible Pathways for the Formation of Perfluoroalkene from Perfluorodecalin by a Single Electron Transfer (SET) Process Figure 3-2 Chemical Structures of MCs and Ionomers Studied in Current Research Figure 3-3 Scheme Showing General Sample Handling Procedure In Harsh Degradation Test Figure 3-4 A Typical Calibration Curve of Fluoride ISE Figure 3-5 Plot Showing Interference Check of Sample ph Values at [F - ]=50ppm 78 Figure 3-6 Plot Showing Interference Check of the Presence of Fe 2+ Ions of Various Concentrations at [F - ]=50ppm Figure 3-7 Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=50ppm Figure 3-8 Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=5ppm xiii

17 Figure 3-9 Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=0.5ppm Figure 3-10 Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=0.05ppm Figure 3-11 Fluoride Evolution from MCs as a Function of Mild Degradation Test Time (MC structures shown in top portion) Figure 3-12 Plot Showing Low Concentration Range of Figure Figure 3-13 Fluoride Evolution from MCs as a Function of Harsh Degradation Test Time Figure 3-14 Fluoride Evolution from MCs After 1 hour UV Exposure (Unfilled columns represent fluoride generated from UV exposure without H 2 O 2 added into the solution; filled columns show the fluoride generated from UV exposure with the presence of H 2 O 2 ) Figure 3-15 Plot Showing Low Concentration Range of Figure Figure 3-16 LC Chromatographic Traces of Degraded MC4 Reaction Product Mixture Figure 3-17 MS Spectra of LC Trace of a Degraded MC4 Reaction Product Mixture at Various Elution Times Figure 3-18 LC Chromatographic Trace of Degraded MC1 Reaction Product Mixture (top) and LC traces (relative abundance) of Selective Ions from TFA (bottom five traces) Figure 3-19 MC1 LC Trace (top) and Corresponding MS Spectrum (bottom ) at RT=1.9 min xiv

18 Figure 3-20 LC Chromatographic Trace of Degraded MC1 Reaction Product Mixture (top) and LC traces (relative abundance) of Selective Ions from PFPA (bottom four traces) Figure 3-21 MC1 LC Trace (top) and Corresponding MS Spectra (bottom ) at RT=1.9 min and 2.9 min Figure 3-22 LC Chromatographic Trace of Degraded MC8 Reaction Product Mixture (top), LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom four traces) Figure 3-23 MS Spectra Marked as MS-1 to MS-4 in Figure 3-22: (top-left) RT=3.9 min, (top-right) RT=20.1 min, (bottom-left) RT=19.3 min, (bottom-right) RT=18.2 min Figure 3-24 Ether Cleavage Decomposition Mechanism of PFPEs by Lewis Acid Sites Figure 3-25 (top) Possible Electron Beam Induced Decomposition of a Branched Fluoroether, and (bottom) the Formation and Dissociation of a Molecular Cation Intermediate of Perfluoroether Structure Studied by Time of Flight-Secondary Ion Mass Spectrometry (TOF-SIMS) in the Positive Ion Mode Figure 3-26 Possible γ Irradiation Induced Decomposition of a Branched Fluoroether Figure 3-27 Postulated Structure of Various Possible Radical Intermediates from MC7 Trapped by 4-hydroxy-TEMPO Radical (Molecular Weight is Shown in Parenthesis) xv

19 Figure 3-28 Postulated Structure of Various Possible Radical Intermediates from MC8Trapped by 4-hydroxy-TEMPO Radical (Molecular Weight is Shown in Parenthesis) Figure 3-29 LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC7 Solution (MC7-4-2Hr) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom two traces) Figure 3-30 MS Spectra Marked as MS-1 and MS-2 in Figure 3-29: (top) RT=4.7 min, (bottom) RT=20.4 min Figure 3-31 LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC7 Solution (MC7-4-2Hr) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom two traces) Figure 3-32 MS Spectra Marked as MS-1 and MS-2 in Figure 3-31: (top) RT=22.4 min, (bottom) RT=21.3 min Figure 3-33 Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC8 Solution (MC8-4-24Hr) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom three traces) Figure 3-34 MS Spectra Marked as MS-1 to MS-3 in Figure 3-33: (top) RT=22.4 min, (center) RT=16.9 min, and (bottom) RT=21.3 min Figure 3-35 LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC8 Solution (MC8-4-90Min) (top), and LC xvi

20 Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom two traces) Figure 3-36 MS Spectra Marked as MS-1 and MS-2 in Figure 3-35: (top) RT=23.2 min, (center) RT=16.9 min, and (bottom) RT=22.6 min Figure 3-37 Generalization of the Trapping Experiment Results for MC7 and MC8 as Shown in Table 3-6 and Table Figure 3-38 Resonance Stabilization of the Fluorocarbon-centered Radical from Adjacent Fluorine (left), and the Structure of Scherer s Radical (right). 129 Figure 3-39 Control Experiments to Check Acid Catalysis Effect for MC Figure 3-40 Control Experiments to Check Acid Catalysis Effect for MC Figure 3-41 Control Experiments to Check Acid Catalysis Effect for MC Figure 3-42 Fluoride Generation from MC1 (0.5mM) Solution at Various H 2 O 2 Concentration upon UV Exposure Figure 3-43 Fluoride Generation from MC1 (5mM) Solution at Various H 2 O 2 Concentration upon UV Exposure Figure 3-44 Fluoride Generation from MC1 (50mM) Solution at Various H 2 O 2 Concentration upon UV Exposure Figure 3-45 Summary Plot for Figure 3-42 to Figure Figure 3-46 Fluoride Generation from MC4 Solution as a Function of H 2 O 2 Concentration) Figure 3-47 Fluoride Generation from MC8 Solution as a Function of H 2 O 2 Concentration) xvii

21 Chapter 4 Figure 4-1 Modified Fenton s Degradation Test Procedure for Membrane Samples Figure 4-2 Cell Used for Determination of Membrane Conductivity. (1) Kel-F block; (2) thumbscrew; (3) open area to allow equilibrium; (4) membrane sample: (5) blackened Pt foil; (6) Pt ribbon lead Figure 4-3 Fluoride Generation as a Function of Degradation Time in Mild Fenton s Degradation Test Figure 4-4 Fluoride Generation as a Function of Degradation Time in Modified Fenton s Degradation Test Figure 4-5 Nafion Degradation Product LC Trace (top three, full and extracted chromatographs) and Corresponding MS Spectrum (bottom) at RT=7.7 min Figure F NMR of Nafion Degradation Major Product from Fenton s Degradation Test Solution Figure 4-7 Scheme Showing the Major Product Observed as a Result of Ether Cleavage of Nafion Figure 4-8 Expected Fragment as a Result of Ether Cleavage of 3M-G1-NM Figure 4-9 LC Chromatographic Trace of the Aliquot Solution from 3M Membrane Degradation Test Medium (top), LC Traces (relative abundance) of Selective Ions From Expected Products in Table 4-3 (bottom two traces) Figure 4-10 MS Spectrum Marked as MS-1 in Figure 4-9 (also shown as the top xviii

22 LC trace) at RT=4.4 min (bottom) Figure 4-11 ATR of Pristine and Degraded Nafion Samples Figure 4-12 ATR of Pristine and Degraded Nafion Samples Figure 4-13 ATR of Pristine and Degraded 3M-G1-NM Samples Figure 4-14 ATR of Pristine and Degraded 3M-G1-NM Samples Figure 4-15 A typical DSC Curve of Nafion With Two Heating Cycles Shown Figure 4-16 DSC Curve of Nafion Samples with Only First Heating Trace Shown Figure 4-17 DSC Curve of 3M-G1-NM Samples with Only First Heating Trace Shown Figure 4-18 DSC Curve of 3M-G1-PM Samples with Only First Heating Trace Shown Figure 4-19 DSC Curve of 3M-G1-BM Samples with Only First Heating Trace Shown Figure 4-20 DMA Results of Pristine and Degraded Nafion Samples, (left) Storage Modulus vs Temperature, (right) Tan δ vs Temperature Figure 4-21 Tensile Test Results (Measured by DMA) of Pristine and Degraded Nafion Samples Figure 4-22 DMA Results of Pristine and Degraded 3M-G1-NM Samples, (left) Storage Modulus vs Temperature, (right) Tan δ vs Temperature xix

23 Acknowledgements I would like to take this opportunity to express my sincere gratitude to my research advisors, Dr. David A. Schiraldi and Dr. Thomas A. Zawodzinski, Jr., for their encouragement, patience, and guidance throughout my entire study. I have greatly benefited from their exceptional knowledge and personality for my educational and professional development. I would also like to thank the members of my advisory committee, Dr. Stuart J. Rowan, for many of his suggestions and discussions, and Dr. Gary E. Wnek, for his insights and professional reference. I am grateful for the opportunity to use the LC-MS equipment in the Department of Chemistry at Case Western Reserve University. I am also greatly indebted to Mr. Jim Faulk for his training and assistance on the LC-MS system. I would like to thank all the fellow students from the Schiraldi group for all the good time we spent together and for your encouragement and support whenever I needed. I thank Dr. Hossein Ghassemi for his help with the conductivity measurement. I also thank our industrial collaborators for all their help and discussions, Dr. Mike Hicks, Dr. Mike Yandrasits, Dr. Mike Guerra, Dr. Tom Kestner, Dr. Qiu Zai-Ming, and Dr. Joel Miller from 3M, and Dr. Zhigang Qi from PlugPower. Lastly, my wife, Ning, deserves the most recognition. I owe her my entire life with love for all the sacrifice she made. xx

24 Chemical Durability Studies of Ionomers and Model Compounds for Fuel Cell Applications Abstract By CHUN ZHOU In this dissertation, a systematic investigation of the chemical durability study of perfluorinated sulfonic acid (PFSA) ionomers for polymer electrolyte membrane fuel cell (PEMFC) was conducted. Low molecular weight model compounds with various structural characteristics were employed as analogs to different moieties that are present in the ionomers. Model compounds and ionomers were degraded by hydroxyl radicals, which are the attacking species present in a running fuel cell. The hydroxyl radicals were created by Fenton s reagents, ferrous ion and hydrogen peroxide, or by direct UV photolysis of hydrogen peroxide. Fluoride release was measured and considered to be the measurement of chemical degradation of model compounds and ionomers. Degradation products from model compound systems and ionomer systems were identified by liquid chromatography-mass spectrometry (LC-MS) and 19 F nuclear magnetic resonance (NMR) experiments. Certain intermediate radical species that are present during the degradation were trapped by using stable radical solutions of 4-hydroxy-2,2,6,6- tetramethyl-piperidinooxy (4-hydroxy-TEMPO), followed by identification using LC- MS. The results from model compound systems revealed that: carboxylic acid groups are extremely labile toward the reaction with hydroxyl radicals, fluoroethers moieties can be cleaved by hydroxyl radicals, and the mechanistic steps involved in the ether cleavage xxi

25 reaction are proposed. The results from ionomer systems, i.e. commercial Nafion ionomer and 3M ionomers, showed good agreement with the conclusions reached from model compound systems. After degradation, low molecular weight fragments formed through the side chain cleavage from the fluoroether branching points of ionomers were identified by LC-MS. The proton conductivity of various ionomers also decreased after degradation. xxii

26 Chapter 1. Introduction 1.1 Fuel Cell Basics Fuel cells are currently being explored as energy conversion devices which can meet a range of societal needs for transportation, stationary, and portable power sources. Fuel cells offer the promise of high energy efficiency and low pollution by directly converting chemical energy into electrical energy. The two major electrochemical processes are the oxidation of fuel (typically hydrogen) at the anode side to release electrons which are transferred to the cathode side, and the reduction of oxidant (usually oxygen) at the cathode. The electron flow in these two processes via external circuit gives rise to current that can be used to drive an electrical load. The differences between fuel cells and batteries are not very straight-forward. The similarities are: they are both electrochemical devices; both devices rely on similar components, electrodes and electrolytes, to produce electric energy directly from the electrochemical reaction of fuel and oxidant. There are however major differences between fuel cells and batteries: batteries are both energy storage and energy conversion devices that provide energy through the conversion of internally stored reactants (both fuel and oxidant), while fuel cells are merely energy conversion devices that operate by externally supplied reactants. Fuel cells can continue to function as long as external fuels are supplied continuously, while batteries will cease to produce electric energy when the internally stored reactants are consumed. The basic principles of fuel cells were discovered in 1839 by Sir William Grove. 1 The applications of fuel cells as practical power source, however, was demonstrated 120 1

27 years later when NASA utilized hydrogen fuel cells to power space flights for the Gemini space missions. 2 The commercial potential of fuel cells was widely recognized by various industries since then, but the high cost and short operation time prevented the large-scale commercialization of such devices. Since the beginning of the Industrial Revolution, the over-reliance on heat engines employing combustion of fossil fuels has raised issues such as severe air pollution, the green house effect, and most importantly the steady depletion of the world s limited fossil fuel reserves. A rich variety of alternative power approaches, such as wind, wave, solar, geothermal energy, and so on, has been implemented for power generation. However promising they are, the majority of these alternative power sources suffers from the reliance on geographical location, seasonal fluctuation, and the difficulty in direct utilization in transportation applications (which accounts for a significant portion of air-polluting emissions). To address the automobile emission pollution issue, the U.S. Department of Energy (DOE) has funded fuel cell and battery powered vehicle research activities since the 1980 s. 2 Fuel cell powered vehicles hold great promise for zero emission of greenhouse gases if hydrogen is directly used as fuel because there is no combustion involved. However it should be pointed out that there are always CO 2 emissions involved in the production of hydrogen at present. The most promising yet distant approach to produce hydrogen would be the electrolysis of water by massive arrays of solar cells. Fuel cells also have the following advantages over batteries for automobile applications: smaller size, lighter weight, quick refueling, and longer range. 2 Fuel cells can be classified by the nature of the electrolyte and the operating temperature. The abbreviations, key features, and electrochemical reactions of major 2

28 types of fuel cells are outlined in Table Comparison of these different types of fuel cells in various applications, from portable electronics equipment and vehicular application to distributed power generation, is summarized in Figure Table 1-1. Comparison of Various Fuel Cell Technologies Fuel Cell Electrolyte Operating Temperature ( o C) Electrochemical Reactions Anode: CH 3 OH + H 2 O 6H + + 6e - + Direct Methanol (DMFC) Solid organic polymer (normally) 80~90 CO 2 Cathode: 1½ O 2 + 6H + + 6e - 3H 2 O Cell: CH 3 OH + 1½ O 2 2H 2 O + CO 2 Polymer Electrolyte Membrane (PEMFC) Solid organic polymer 30~80 Anode: H 2 2H + + 2e - Cathode: ½ O 2 + 2H + + 2e - H 2 O Cell: H 2 + ½ O 2 H 2 O Alkaline (AFC) Aqueous solution of potassium hydroxide soaked in a matrix 90~100 Anode: H 2 + 2OH - 2H 2 O + 2e - Cathode: ½ O 2 + H 2 O + 2e - 2OH - Cell: H 2 + ½ O 2 H 2 O Phosphoric Acid (PAFC) Phosphoric acid soaked in a matrix 175~200 Anode: H 2 2H + + 2e - Cathode: ½ O 2 + 2H + + 2e - H 2 O Cell: H 2 + ½ O 2 H 2 O Molten Carbonate (MCFC) Solution of lithium, sodium, and/or potassium carbonates soaked in a matrix 600~1000 Anode: H 2 + CO 2-3 H 2 O + CO 2 + 2e - Cathode: ½ O 2 + CO 2 + 2e - 2- CO 3 Cell: H 2 + ½ O 2 + CO 2 H 2 O + CO 2 Solid Oxide (SOFC) Solid zirconium oxide with a small amount of yttria 600~1000 Anode: H 2 + O 2-2H + + 2e - Cathode: ½ O 2 + 2e - O 2- Cell: H 2 + ½ O 2 H 2 O 1.2 Polymer Electrolyte Membrane Fuel Cells This thesis research will focus on polymer electrolyte membrane fuel cells (PEMFC), also called solid polymer fuel cell (SPFC). This class of fuel cells has attracted great and increasing interest from both industry and academia because they 3

29 seem to be the most suitable candidate for transportation applications, thanks to their moderate temperature of operation (30-80 o C) to render the possibility of quick start, good CO 2 tolerance by the electrolyte, and a combination of high power density and high energy conversion efficiency. 3 Typical applications Power / Watts Main advantages Range of application of the different types of fuel cell Cars, boats, and domestic Distributed power Portable electronic combined heat and power generation, CHP, and equipment systems (CHP) buses k 10k 100k 1M 10M Higher energy density Potential for zero Higher efficiency, less than batteries, faster emissions, high efficiency pollution, quiet recharging DMFC AFC MCFC SOFC PAFC PEMFC Figure 1-1. Summary of the Applications and Advantages of Various Fuel Cell Technologies and Their Applications The basic construction of a membrane/electrode assembly (MEA) of PEMFC is schematically shown in Figure 1-2. There are two major structural components, the composite electrodes and the polymer electrolyte sandwiched between the two electrodes. The operation mechanism of such a set-up is: the fuel (hydrogen gas) is oxidized and split into protons and electrons at the anode; the polymer electrolyte membrane (largely impermeable to the reactant gases but highly permeable to protons) transfers the protons which carry the ionic charge from anode to cathode; finally the protons transferred across the membrane will recombine with the electrons generated at the anode and transported 4

30 via the external circuit to reduce the oxidant, oxygen gas, supplied at the cathode side to complete the circuit. The reactions that occur at the electrodes are shown below. Anode: Cathode: Overall: H 2 2H + + 2e ½ O 2 + 2H + + 2e H 2 O H 2 + ½ O 2 H 2 O + Electrical Energy Anode e A e Cathode H + H + O 2 H 2 H+ H + H + H 2 O H 2 2H + + 2e H + 1/2O 2 + 2H + + 2e H 2 O Catalytic Layer Polymer Electrolyte Membrane (PEM) Figure 1-2. Schematic Diagram of a Membrane/Electrode Assembly (MEA) 1.3 Catalyst, Electrodes, Membrane Materials, and Complete Fuel Cell Stacks PEMFC electrodes, MEA, and fuel cell stacks will be discussed in detail in the following section prior to the detailed discussion of the chemistry, structure and physical properties of various PEMFC polymer electrolyte candidates. 5

31 1.3.1 Catalyst and Electrodes The electrochemical reactions of hydrogen PEMFC consist of two separate reactions: 1) the oxidation half-reaction at anode: hydrogen gas is oxidized to produce electrons and protons; 2) the reduction half-reaction at cathode: the supplied oxygen gas recombines with the transferred protons and electrons. These two half reactions would normally occur very slowly at the moderate operating temperature of PEMFC, typically o C. For practical application, the rate of these reactions (especially the performance limiting oxygen reduction half-reaction, which is more than 100 times slower than the hydrogen oxidation reaction) must be increased, which can be achieved by employing various catalysts. Highly dispersed platinum nano-particles with the size of 2-3 nm have been shown to be the excellent catalyst choice, because of the high catalytic property and enormous increase of surface area at relatively low catalyst loading ( mg/cm 2 ). 2-4 Table 1-2. Methods of Catalyst Layer Preparation and Application (A) Bonding to membrane (B) Bonding to carbon cloth/paper Mode Application Mode Application A1 Hot-pressed Pt black/ptfe layers B1 Membrane - impregnated Pt/C//PTFE A2 Electroless deposition of Pt on membrane B2 B1 + Sputtered Pt layer A3 Hot-pressed Pt black/c//membrane layers B3 Pt catalyst electrodeposited at carbon/membrane interface As discussed previously, the electrodes in PEMFC are normally composite electrodes with high surface area of platinum catalyst particles dispersed in the porous electrode materials, typically carbon paper or carbon cloth. A suitable catalyst layer has to be designed for positioning in between the polymer electrolyte membrane and the 6

32 gas/fuel distributor, with the catalyst layers bonded to the membrane or bonding to carbon cloth/paper. The details of various approaches of these two fabrication methods are summarized in Table Out of all the fabrication routes available, the most efficient approach is to introduce electrolyte materials in the catalyst layer as a binder and good contact to the membrane through the establishment of a three-phase contact, yielding the good contact between reactant gas, electrolyte, and electrode catalyst, as schematically shown in Figure , 3 Thin layer of electrolyte Catalyst support Catalyst Bulk electrolyte Figure 1-3. Schematic Diagram of Idealized Electrode Structure Other PEMFC Components and Fuel Cell Stacks A complete single cell PEMFC diagram with the schematic depiction of MEA with backing/gas diffusion layers on both sides are Figure 1-4. In addition to the components already discussed above, the highly porous and conductive backing layers 7

33 (usually microns thick carbon paper/cloth) are designed to maximize the current collection from a MEA. These layers not only assist in the efficient diffusion of each reactant gas to catalyst sites, but also in the water management during the fuel cell operation by allowing the water produced at cathode to effectively leave the system. Figure 1-4. Diagram of a Single MEA Fuel Cell with Enlarged Cross-section Showing Structural Details 2 8

34 A three-cell PEMFC stack connected by bipolar plates is shown in Figure 1-5. Why are fuel cell stacks necessary? It is because of the fact that the ideal fuel cell voltage at about 80 o C is only about 1.18 volt 1, 2, and the actually operating voltage decreases when current density is increased. Effective commercial electric motors that operate at a few hundred volts would demand the connection of individual fuel cells in series to form a fuel cell stack. In between two individual fuel cells, bipolar plates are highly gas-impermeable to serve as the reactant gas flow guide, and electronically conductive to function as connector for the anode and cathode of two adjacent cells. Figure 1-5. Diagram of a Three Cell Fuel Cell Stack with Two Bipolar Plates and Two End Plates Polymer Electrolyte Membranes 9

35 An important component of a PEMFC is the polymer electrolyte membrane, which is sometimes called an ionomer. Ionomers are polymers that contain a small fraction of ionizable repeat units, compared with polyelectrolytes. The mole percent of the ionizable repeat units relative to total repeat units of ionomers is conventionally defined to be less than 15 mole percent, although in many cases this value can be and has been exceeded. 5, 6 To effectively function as proton conducting media in PEMFC, a good membrane candidate has to simultaneously meet many requirements listed below. High proton conductivity Poor electronic conductivity Low gas and vapor permeability Good mechanical strength as structural component and ease of MEA fabrication Chemical and electrochemical stability at operating conditions Nafion - Synthesis, Solid State Structure, and Proton Conductivity Nafion (Nafion is a registered trademark of DuPont, to be referred as Nafion in the following text) is the current benchmark ionomer for PEMFC applications. Nafion is a perfluorinated copolymer that consists of a hydrophobic poly(tetrafluoroethylene) (PTFE) backbone with pendant side chains of perfluorinated vinyl ethers terminated by sulfonic acid groups, as shown in Figure 1-6. The reported synthetic route of the comonomer, a perfluorinated vinyl ether, is also shown in Figure The synthesis starts with the reaction of TFE with SO 3 to form a sultone, which can rearrange to form sulphonyl fluorides by the treatment of NR 3. The sulphonyl fluoride then reacts with 10

36 hexafluoropropylene epoxide to produce a sulphonyl fluoride adduct, which is heated with sodium carbonate to form a vinyl ether with a pending sulphonyl fluoride. Nafion is radically copolymerized from TFE and this sulphonyl fluoride vinyl ether, followed by hydrolysis to produce the pendant sulfonic acid groups on the side chains The radical polymerization of fluoroolefins is highly exothermic and often occurs with great force The strong C-F bond in perfluoro compounds prevents extensive chain transfer reactions that are commonly observed in the radical polymerization of hydrocarbon olefins. PTFE and perfluorinated ethylene propylene (FEP) copolymer are well-known examples that can be polymerized to extremely high molecular weights with close to zero branching. 12 The solid state structure of Nafion has been extensively examined and reviewed. 14 A rich collection of literature is available for the interpretation of Nafion solid state morphology. The chemical structure of Nafion promotes a phase-separated morphology via segregation of hydrophilic (sulfonic acid rich) and hydrophobic (fluorocarbon rich) domains. Upon hydration, the hydrophobic domain confers upon the membrane mechanical sturdiness and prevents the ionomers from dissolving, while the sulfonic acid rich hydrates considerably to render the exceptionally high water transport and proton conductivity. The existence of clustered regions of hydrophilic domains is widely accepted and energetic consideration for the formation of such polar ionic clusters was proposed, 15 although the details of their arrangement and connectivity are still topics of debate. 11

37 Chemical structure of Nafion CF 2 CF 2 CF 2 x CF y O CF 2 CF O CF 2 2 SO 3 H CF 3 Typical synthesis of perfluoro- vinyl ether comonomer O CF CF SO 3 CF 2 CF 2 NR 3 O SO 2 F C CF 2 CF 2 SO 2 F CF CF 3 O O F C CF O CF 2 CF O CF 2 CF 2 SO 2 F CF 3 CF 3 Δ -COF 2 CF 2 CF O CF 2 CF O CF 2 CF 2 SO 2 F CF 3 Figure 1-6. Chemical Structure of Nafion and the Synthetic Route of the Vinyl Ether monomer 12

38 Figure 1-7. Cluster-network Model for the Morphology of Hydrated Nafion 14 The widely-accepted cluster-network model, originally developed by Gierke et al., has endured as basis for correlating the structure-property relationship of Nafion. The interpretation of interconnected large spherical ionic clusters (formed by polar ionic groups through an inverted micelle-like fashion) is schematically shown in Figure , The calculated cluster size has the range of 3-5 nm for a polymer with the equivalent weight of 1200, and each cluster contains 70 ion exchange sites and 1000 water molecules on average. The sizes of the ionic clusters strongly depend on the water content in the membrane. The reorganization and dimension changes of clusters upon alternation of hydration are shown in Figure : the increase in hydration leads to the coalescence of clusters by forming smaller number of larger clusters. Other less popular morphology models include: 1) the three-phase model that consists of the fluorocarbon domain, the ion cluster, and a third transitional interphase region 19 ; 2) the core-shell model 20, 21 ; 3) the local order model ; 4) the lamellar model 26 ; 5) the sandwich-like model 27 ; 6) the rod-like model 28. Among such extensive 13

39 research efforts to understand the morphology of Nafion, one feature that all of the models have agreed on is the ionic group aggregation in the perfluorinated polymer matrix to form a continuous network of ion clusters to allow for significant swelling of polar solvents and efficient transport through this nano-scale domains. Figure 1-8. Evolution of a Smaller Number of Larger Clusters with Increased Hydration of Nafion 16 The protonic conductivity of PEMFC candidates is strongly dependent on membrane water content, which in turn is also a strong function of operating temperature. A major research theme in PEMFC is to achieve the highest protonic conductivity by the investigation of water uptake characteristics, so as to match the hydration level needed for transport via proper design of membrane structure and cell/stack designs. Similar to 14

40 any other conducting medium, the magnitude of the specific conductivity is determined by the product of charge carrier mobility and charge carrier density. For Nafion, with the equivalent weight of 1100, the charge (proton) carrier density is similar to that of 1 M aqueous sulfuric acid solution, and the proton mobility in such a fully hydrated solid membrane is only one order of magnitude lower than the proton mobility in the aqueous solution. 3 The typical specific conductivity of fully hydrated Nafion type perfluorinated sulphonic acid (PFSA) membranes is about 0.1 S/cm at room temperature, and about 0.15 S/cm at 80 o C. λ ~ 2-3 : λ ~ 4-14 : λ > 14 : SO - 3 H 3 O + H 2 O Figure 1-9. Schematic Illustration of Membrane Hydration Level of Nafion 15

41 Two major proton transport mechanisms in hydrated PEMFC membranes are proposed. There are the hopping mechanism (or Grotthus mechanism), 29 and the vehicle mechanism, 30 although in reality the transport may well be the combination of both. In the vehicle mechanism, the proton is transported by the movement of the proton/water complex which is formed by hydrogen bonding of proton with one or several water molecules. The proton conductivity therefore is strongly dependent on the diffusion of water across the membrane. For the Grotthus mechanism, the water in the membrane forms a continuous network when the hydration level is above the percolation threshold. The proton initially forms hydrogen bonding to a water molecule, then this bonded pair dissociates and hydrogen bonds to another adjacent water molecules. Proton transport can therefore be visualized as hopping across the membrane via the Grotthus mechanism. The number of water molecules per sulfonic acid groups (λ) is an important parameter to quantitatively describe the water content and proton transport in hydrated membrane. The overall proton transfer process also involves the dissociation of the acid groups in the solid ionomer, and then the dissociated protons become mobile enough to transport in the water medium within the membrane via various transport mechanisms described above. The dissociation of proton from the -SO 3 H groups is a also function of λ. Paddison 31 showed that at least two or three water molecules per sulfonic acid group are required to facilitate the dissociation. Zawodzinski 30 proposed a simple qualitative description of the transport in Nafion, as schematically described in Figure 1-9. At low level of hydration, i.e. λ ~ 2-3, hydronium ions moves as in vehicle mechanism. The conductivity is low because there are few water molecules to help the dissociation and 16

42 solvation of sulfonic acids, and water molecules are strongly bound in the polymer matrix, resulting in significant resistance to the transport of hydronium ions. When hydration level increases to λ ~ 4-14, there is enough loosely bound water. Not only does the transport of hydronium via vehicle mechanism transport greatly increase, but the proton transport via hopping mechanism also increases when the population of water molecules increases. When the hydration level further increases to λ > 14, the water contained in the membrane becomes more bulk-like, which aids proton hopping. However, bulk-like water is more easily transported or dragged across the membrane by the movement of protons as they experience strong force from the migrating protons, namely the electro-osmotic drag effect Alternative Membrane Candidates Nafion type perfluorinated fluoro-carbon ionomers are attractive for their exceptional chemical and electrochemical stability in a operating fuel cell, although the biggest disadvantage is the cost. Nafion is priced at $ per square meter due to the expensive fluorination process and lack of competition in the market place. Additionally, running PEMFC at temperature above 100 o C has practical performance improvement because of the expected increase of the current performance-limiting oxygen reduction reaction at cathode, the improved catalyst tolerance against deactivation (poisoning) by CO (common impurity in the hydrogen fuel), and finally the effective system heat management since the heat rejection increases when the temperature difference between the system and the environment is larger. The high temperature performance of Nafion is less than satisfactory because of its severe dehydration. Active research efforts from both 17

43 industry and academia are being dedicated to the new membrane materials development with an emphasis on cost reduction, and overall PEMFC performance improvement. Major alternative commercial developmental membrane materials are summarized below. a. Other Perfluorinated Sulfonic Acids (PFSAs) CF 2 CF 2 CF 2 CF x y O CF 2 CF O m CF 2 CF 2 CF 3 n SO 3 H Nafion 3M Development Membrane Dow Flemion Aciplex m 1; n=1; y=1; x=7~20 m=0; n=2; y=1; x=7~20 m=0; n=1; y=1; x=3~10 m=0,3; n=2~3; y=1; x=3~10 m=0,3; n=1~3; y=1, x=1.5~14 Figure Chemical Structures of Various Commercial and Development PFSAs The general structures of commercial and development PFSAs are shown in Figure 1-10 to represent various formulas of development membrane from 3M, and commercial PFSAs from Dow (Dow ), Asahi Glass (Flemion ), and Asahi Chemicals (Aciplex ). The radical polymerization chemistries are similar to that of Nafion described in the above section. The general polymerization condition is radical emulsion 12, 32 polymerization using water soluble persulfate initiator, as depicted in Figure The majority of PFSA membranes are extruded from the sulfonyl fluoride precursor to 18

44 form films, followed by hydrolysis in basic conditions to convert the sulfonyl fluoride to sulfonate functionalities. Due to the chemical structure similarity, these commercial PFSAs exhibit similar morphology, chemical stability, and competitive PEMFC performance to Nafion when proper thickness and equivalent weights are selected. CF 2 CF 2 + CF 2 CF O CF 2 CF O CF 2 CF 2 m n SO 2 F Emulsion CF 3 Perfulfate Initiator CF 2 CF 2 CF 2 CF x y O CF 2 CF O CF 2 CF 2 m n SO 2 F NaOH CF 3 Ion Exchange CF 2 CF 2 CF 2 CF x y O CF 2 CF O m CF 2 CF 2 CF 3 n SO 3 H Figure Radical Polymerization and Processing Conditions of Commercial PFSAs b. Styrene Based Ionomers Alternative styrene based ionomers are attractive because of the ready commercial source of monomers and the relatively simple polymerization techniques. The common 19

45 styrene based ionomers are listed in Figure The first example of such class is polystyrene sulfonic acid ionomers (PSSA), originally developed by GE for NASA for the application as an on-board power source in the Gemini space program in 1960s. However, PSSAs showed very poor electrochemical stability, due to the oxidation of the benzylic C-H bond that eventually leads to the degradation of PSSAs, and very low power density (less than 50 mw/cm 2 ). To address the stability issue, Ballard Advanced Materials introduced a styrenic membrane, registered as BAM TM Membranes, based on a rich family of sulfonated copolymers containing α,β,β,-trifluoro-styrene comonomers. 33 These copolymers represent an important family of post-sulfonated styrenic ionomers targeting to improve the chemical stability. The general synthesis route is shown in Figure Typically, the unsulfonated precursor copolymers are synthesized by emulsion polymerization with dodecylamine hydrochloride as emulsifier and potassium persulfate as the initiator. The post-sulfonation is achieved by treating the polymer with a complex of sulfur trioxide complex. These fluorinated PSSAs showed better PEMFC performance than Nafion at current densities greater than 600 ma/cm But the largest advantages are very limited choice of solvent and high cost. The non-fluorinated styrenic block copolymers are produced by Dais Analytic under the trade name of Kraton membranes, 35 as shown in Figure Kraton membranes are very attractive candidates because of similar conductivity to Nafion ( S/cm when fully hydrated), 36, 37 and very rich and controllable morphology when compositions of the blocks are adjusted. However, the presence of oxidizable C-H bond similar to those of PSSAs is the major cause of limited performance life. 20

46 PSSA CH 2 CH 2 CH 2 CH a b SO 3 H BAM TM CF 2 CF CF 2 CF CF 2 CF CF 2 CF a b c d R 1 R2 R3 SO 3 H R 1, R 2, R 3 = alkyl, halogen, OR, CF=CF 2, CN, NO 2 Kraton CH 2 CH CH 2 CH a CH 2 b CH 2 CH 2 CH CH 2 CH CH 2 CH c d e f CH 2 CH 3 SO 3 H SO 3 H Figure Chemical Structures of Various Styrenic Sulfonic Acid Ionomers a CF 2 CF + b CF 2 CF Emulsifier Aq. initiator CF 2 CF CF 2 CF a b R R SO 3 :P(O)(OEt) 3 CF 2 CF CF 2 CF a b Figure Synthetic Route of BAM Type PSSAs SO 3 H R 21

47 c. Post-Sulfonated Aromatic Polymers An incomplete list of popular post sulfonated polymers is shown in Figure 1-14, including poly(ether sulfone)(pes) 38, 39, poly(oxy-1,4-phenyleneoxy-1,4- phenylenecarbonyl-1,4-phenylene)(peek), 40, 41 and other aromatic based structures Such sulfonated polyaromatics competes with Nafion on the cost, proton conductivity, and the high temperature performance as well as potential advantage of reducing methanol cross-over when methanol is used as fuel. A few reviews are available in the literature, summarizing the performance of sulfonated hydrocarbon polymers for PEMFC The morphology, water uptake, and water dynamic within the membrane were postulated to be dramatically different from that of Nafion. The most marked difference is the sulfonic acid dissociation. The pka of sulfonic acid of PEEK type polymer is around -1, while the pka of sulfonic acid for Nafion is around Less negative pka implies less acidic nature (less polar) and hence less proton dissociation for protonic conduction. Secondly, the hydrophilicity and hydrophobicity difference of polyaromatics is less dramatic than the counterparts of the perfluorinated nature of Nafion with extremely hydrophobic backbones and hydrophilic sulfonic acids. In addition to this difference, the rigidity imposed by bulky benzene backbone structures also contributes to the less pronounced nano-scale phase separation, resulting in the dramatic different phase separation shown in Figure 1-15, 41 based on the analysis of small angle X-ray scattering (SAXS) data. 41 The water channels in PEEK-based systems are narrower and less separated than those of Nafion, and highly branched with many dead-end channels

48 O O O C SO 3 H Sulfonated PEEK SO 3 H O n O O S O Sulfonated Poly(arylene ether sulfone) n N N N N n n O S SO 3 H n SO 3 H SO 3 H O SO 3 H Sulfoarylated PBI Sulfonated PPBP Sulfonated Poly(phenylene sulfide) SO 3 H O N N N N Sulfonated Poly(phenylquinoxaline) n Figure Chemical Structures of Various Post-sulfonated Ionomers Many sulfonation methods are available for the post-sulfonation reaction: concentrated sulfuric acid, acetyl sulfate, or chlorosulfonic acid. These methods are harsh in nature and can lead to possible chain cleavage/degradation and crosslinking. Additionally, the chemistry of the polymer imposes specific sites on which the sulfonation reaction can happen due to the electrophilic nature of the sulfonation reaction. 23

49 The location of the sulfonic acid sites was found to affect the hydrolysis and desulfonation of certain polymers. 47 Other critical issues with post-sulfonation are the inhomogeneity of the sulfonic acid distribution, and the difficult to control the degree of sulfonation (to avoid a final water-soluble polymer). Direct copolymerization of sulfonated monomers, on the contrary, is able to overcome majority of the disadvantages and to offer better design/control of the final chemistry of the polymers. 24

50 Figure Schematic Representation of the Microstructures of Nafion and a sulfonated PEEK 41 d. Direct Copolymerization of Sulfonated Comonomers The strategy of direct copolymerization of sulfonated polymers for PEMFC application is widely adapted by the McGrath and other research groups, 47, especially 25

51 poly(arylene ether)s, poly(arylene ether disulfonated sulfone), and their random copolymers. The stability enhancement is illustrated in Figure 1-16, showing the reactivity difference between post-sulfonation and direct copolymerization of sulfonated 47, 50 monomers when the backbone structure is identical. For structures obtained from post-sulfonation, the sulfonate resides on the electron-rich activated rings adjacent to the oxygen, resulting in the decrease of the electron density of these rings and making the aromatic ether bond very susceptible toward hydrolysis. On the contrary, the sulfonic acids can be placed on the electron-poor deactivated rings that next to electron withdrawing ketone or sulfone groups to enhance the hydrolytic stability, and furthermore to increase the acidity of sulfonic acid groups to achieve higher proton conductivity. Electron-rich activated ring SO 3 H Electron-poor deactivated ring O O O O S O n CH 3 O O C O S CH 3 SO 3 H O n SO 3 H Electron-rich activated ring Electron-poor deactivated ring Figure Reactivity and Stability from the Placement of Sulfonic Acid Group in Postsulfonation versus Direct Copolymerization 26

52 The morphology observed for the copolymers of poly(arylene ether sulfone)-copoly(arylene ether disulfonated sulfone) (also named as Biphenyl Sulfone H-form / BPSH copolymers) is very similar to that found for Nafion when probed by atomic force 47, 53 microscope (AFM), see Figure The proton conductivity can be fined tuned by the composition ratio of the disulfonated comonomer to obtain a good combination of high proton conductivity and suitable morphology. Phase image of BPSH-40 dark/soft region: sulfonic acid + water Phase image of Nafion 117 Figure AFM Micrographs of BPSH-40 and Nafion 47 e. Sulfonated Polyimides (SPI) Both phthalic (five-membered) and naphthalic (six-membered) dianhydrides and appropriate wholly aromatic diamines and heterocyclic analogues have been used to 27

53 synthesize various phthalic and naphthalic SPIs, 45, where the levels of sulfonation in the polymer backbones were controlled by varying the mole ratio of sulfonated diamine to unsulfonated diamines, as shown by the typical synthetic route in Figure The proton conductivity of such SPIs is less than Nafion, typically (2~7) x 10-3 S/cm, although the water uptake can reach as high as 40%, which possibly implies loosely organized ionic domains in solid state morphology. In terms of hydrolytic stability and long term PEMFC applications, six-membered SPIs were found to be much more stable 74, 75 than the five-membered SPIs. Figure Synthetic Route of Model Imide Compounds A and B Composite Membranes Considerable effort has been put into composite membranes designed to address the challenges of high temperature fuel cell applications low methanol cross-over while maintaining high proton conductivity. 28

54 a. Polymer Blends Perhaps the best example in the PEM polymer blends catagory is the composite membranes, from W.L. Gore with the trade names of Gore-PRIMEA and Gore-Select, fabricated by blending Nafion with Gore-tex polymer (porous PTFE) as support. This unique blend offers excellent mechanical strength of very thin membranes that is highly desirable for PEMFC applications due to less cell areal resistance/ohmic loss. However, this kind of composite membrane is still not suitable for high temperature application, since the blend still relies on Nafion. Sulfonated and unsulfonated polymers have also been paired up, such as sulfonated PEEK and polysulfone, to show that the membrane properties may be dramatically changed by the addition of very small amount of sulfonated PEEK. 76 Kerres et al. successfully developed blends of acidic polymers, such as sulfonated polysulfone and PEEK, and basic polymers, such as poly(benzimidazole) (PBI), poly(ethyleneimine) (PEI), and poly(4-vinylpyridine) or (PVP). 77 The performance of there blends is excellent: very low swelling and methanol cross-over, yet comparable proton conductivity relative to Nafion. The results can be explained as follows: good conductivity offered by acidic polymers usually accompanies high water swelling, and the addition of compatible and less swellable basic polymers improves the dimensional stability while maintaining good proton conductivity. b. Membranes Doped with Organics / Potential Anhydrous Membranes One of the potential advantages of running a fuel cell at temperatures higher than 100 o C is to improve the performance-limiting slow oxygen reduction reaction. Higher 29

55 operating temperature, however, poses challenging issues such as water content management across the membrane, which is critical for proton conduction. Efforts have been put into develop anhydrous PEMFC membranes by replacing water with high boiling point molecules as proton solvents in the membranes. For hydrated PEMFC membranes, water can be supplied by humidifying the fuel while these high temperature alternative proton conducting solvents, such as phosphoric acid, sulfamide, and heterocycle compounds, have to be immobilized in the membrane during the operation. The immobilization in turn must yield a continuous network that can facilitate the transport of protons, primarily by Grotthuss mechanism. In such polymer/acid systems, the polymer functions as the solvent and the matrix for the acids: it has to be basic enough to interact and dissolve the acid, help the dissociation of the acids, and it must be chemically stable against hydrolysis at such a high operating temperature and excess acid environment. Polybenzimidazole (PBI) has been doped with H 3 PO 4 to target water free PEMFC operated at temperature higher than 130 o 78, 79 C with good conductivity. The proton conduction ability of H 3 PO 4 is believed to be the result of its self-ionization and the ability to form effective 3-D hydrogen bonding network. 80 Various pathways for proton transfer have been proposed to explain the protonic conductivity dependency on H 3 PO 4 doping level, relative humidity, and temperature, as shown in Figure 1-19, 78 which shows that: 1) the Grotthuss type transport is realized through the hydrogen bonding network; and 2) the involvement of water molecules in the conduction process yields higher conductivity at high relative humidity. 30

56 Figure Various Proton Transfer Pathways in PBI Doped with H 3 PO 4 78 Nafion was also doped with H 3 PO 4 to show reasonably high conductivity (>0.05S/cm) at elevated temperatures (100 ~ 180 o C). 81 Extensive proton exchange between the sulfonic acid groups, phosphoric acid, and the residual water was observed based on 1 H NMR analysis. 82 These two systems, just like many other systems such as poly(vinylalcohol)/h 3 PO 4 83, 84, suffer from the gradual acid depletion and the subsequent accumulation/flooding on the electrodes. 31

57 HN N N HN N NH Imidazole Pyrazole Benzimidazole Figure Chemical Structures of Typical Nitrogen-containing Aromatic Heterocycles Pioneered by Kreuer et al., heterocyclic species have been shown to be effective proton solvents. Typical structures, imidazole, pyrazole, and benzimidazole, are shown in Figure The basic nitrogen sites act as strong proton acceptors by forming proton charge carrier complex, such as (C 3 H 3 NH 2 ) +. Such heterocyclic structures offer great potential in high temperature PEMFC applications because of their high boiling points (188 ~ 360 o C), and the possibility of switching between proton donor and acceptor via protonation/deprotonation. Sulfonated PEK and PEEKK (structures shown in Figure 1-14) and imidazole were the first reported systems to show that imidazole could be used as proton solvent in polymer matrix at high temperature without the need of water. 85 After that, a wealth of literature based on similar concept has appeared, including Nafion blend with ionic liquids (such as 1-butyl, 3-methyl imidazolium triflate), 86 Nafion/solid imidazole, 87 Nafion/imidazole and imidazole-imidazolium salt, 88 poly(acrylic acid)/imidazole. 89 A novel idea to control the interaction between the nitrogen sites and proton to render controlled proton conductivity, via the introduction of various electron-withdrawing and 32

58 electron-donating groups to the ring structure, was recently proposed by the Zawodzinski group. 90 c. Ionomer/Inorganic Particle Composites To tackle the water management issue at high temperature, many research groups have attempted the polymer and inorganic particle blends, especially by using water absorbing particles such as zeolites, P 2 O 5, SiO 2, ZrO 2, and TiO 2. Porous silica gel particles with c.a. 20 nm size were dispersed in Nafion to show an increase of conductivity of 1.6 times higher than Nafion, while non-porous fillers had marginal effect on the conductivity. 91 Zeolites were also added into Nafion, but the conductivity was low at low relative humidity conditions although gas permeability/fuel cross-over was greatly improve upon the addition of filler. 92 Furthermore, zeolites were also added into PTFE to display a linear increase of conductivity with zeolite loading. 93 However, the high zeolite loading required to impart acceptable conductivity degrades the mechanical properties, especially tensile strength. The manipulation of the organization of ionic clusters by changing the physical and chemical properties of the filler is of great interest. To date, TiO 2 is the one of the few systems believed to have bonding interaction with sulfonic acid groups, which was probed by thermogravimetry study coupled with mass spectroscopy. 94 The exceptional electrochemical performance is shown in Figure 1-21, 94 where the polarization curves of Nafion with and without the addition of TiO 2 suggest that TiO 2 greatly improve the cell performance at 115 o C and 65% relative humidity. Other work by Arimura et al. showed that some polymers with flexible long alkyl laurate or stearate moities can be blended to 33

59 align and alter the hydrophilic regions of the solid state structure of Nafion, which causes the great increase of conductivity in hydrated state. 91 Further systematic investigation however is still of importance to fully understand such observations. Additionally, phosphotungstic acid, phosphatoantimonic acid, and heteropolyacids have been attempted with limited success in improving the conductivity 62, of the matrix polymer. H 2 Air, 115 o C TiO 2 Plain Figure Current-voltage Response of a Recast Nafion film (open circle) versus a Titania Composite Film (solid square) under Low Humidity at 115 o C Operating Temperature 94 34

60 References 1. Larminie, J.; Dicks, A., Fuel Cell Systems Explained. John Wiley & Sons Ltd, England Thomas, S.; Zalbowitz, M., Fuel Cells - Green Power. Los Alamos National Laboratory Publication for U.S. Department of Energy Gottesfeld, S.; Zawodzinski, T. A., Polymer Electrolyte Fuel Cells. Advances in Electrochemical Science and Engineering, Edited by Alkire, R.C. et al., WILEY- VCH. 1997, Vol.5, Li, X., Principles of Fuel Cells. Taylor & Francis, New York, NY Tant, M. R.; Mauritz, K. A.; Wilkes, G. L., Ionomers, Synthesis, Structure, Properties and Applications. Blackie Academic and Professional, London, UK Schlick, S. E., Ionomers: characterization, theory, and applications. CRC Press, Inc Steck, A. E.; Stone, C., Development of the BAM membrane for fuel cell applications. Proceedings of the Second International Symposium on Materials for Fuel Cell Systems. Montreal, Quebec, Canada 1997, Connollym, D. J.; Gresham, W. F., US Patent 3,282, Ezzell, B. R.; Carl, W. P.; Mod, W. A., US Patent 4,358, Souzy, R.; Ameduri, B., Functional fluoropolymers for fuel cell membranes. Progress in Polymer Science 2005, 30, (6), Wall, L. A. E., Fluoropolymers. Wiley-Interscience Banks, R. E.; Smart, B. E.; Tatlow, J. C., Organofluorine Chemistry, Principles and Commercial Applications. Plenum Press, New York Hougham, G.; Cassidy, P. E.; Johns, K.; Davidson, T., Fluoropolymers 1. Synthesis. Klumer Academic / Plenum Publishers, New York Mauritz, K. A.; Moore, R. B., State of understanding of Nafion. Chemical Reviews 2004, 104, (10), Mauritz, K. A.; Rogers, C. E., A Water Sorption Isotherm Model for Ionomer Membranes with Cluster Morphologies. Macromolecules 1985, 18, (3),

61 16. Gierke, T. D.; Munn, G. E.; Wilson, F. C., The Morphology in Nafion Perfluorinated Membrane Products, as Determined by Wide-Angle and Small- Angle X-Ray Studies. Journal of Polymer Science Part B-Polymer Physics 1981, 19, (11), Gierke, T. D.; Hsu, W. Y., Perfluorinated Ionomer Membranes; Eisenberg, A., Yeager, H.L., Eds. ACS Symposium series No. 180, American Chemical Society, Washington, DC Hsu, W. Y.; Gierke, T. D., Ion-Transport and Clustering in Nafion Perfluorinated Membranes. Journal of Membrane Science 1983, 13, (3), Yeager, H. L.; Steck, A., Cation and Water Diffusion in Nafion Ion-Exchange Membranes - Influence of Polymer Structure. Journal of the Electrochemical Society 1981, 128, (9), Fujimura, M.; Hashimoto, T.; Kawai, H., Small-Angle X-Ray-Scattering Study of Perfluorinated Ionomer Membranes.2. Models for Ionic Scattering Maximum. Macromolecules 1982, 15, (1), Fujimura, M.; Hashimoto, T.; Kawai, H., Small-Angle X-Ray-Scattering Study of Perfluorinated Ionomer Membranes.1. Origin of 2 Scattering Maxima. Macromolecules 1981, 14, (5), Dreyfus, B.; Gebel, G.; Aldebert, P.; Pineri, M.; Escoubes, M.; Thomas, M., Distribution of the Micelles in Hydrated Perfluorinated Ionomer Membranes from Sans Experiments. Journal De Physique 1990, 51, (12), Gebel, G.; Lambard, J., Small-angle scattering study of water-swollen perfluorinated ionomer membranes. Macromolecules 1997, 30, (25), Gebel, G.; Moore, R. B., Small-angle scattering study of short pendant chain perfuorosulfonated ionomer membranes. Macromolecules 2000, 33, (13), Rollet, A. L.; Gebel, G.; Simonin, J. P.; Turq, P., A SANS determination of the influence of external conditions on the nanostructure of nafion membrane. Journal of Polymer Science Part B-Polymer Physics 2001, 39, (5), Litt, M. H., Polymer Preprints 1997, 38,

62 27. Haubold, H. G.; Vad, T.; Jungbluth, H.; Hiller, P., Nano structure of NAFION: a SAXS study. Electrochimica Acta 2001, 46, (10-11), Rubatat, L.; Rollet, A. L.; Gebel, G.; Diat, O., Evidence of elongated polymeric aggregates in Nafion. Macromolecules 2002, 35, (10), Agmon, N., The Grotthuss Mechanism. Chemical Physics Letters 1995, 244, (5-6), Zawodzinski, T. A.; Davey, J.; Valerio, J.; Gottesfeld, S., The Water-Content Dependence of Electroosmotic Drag in Proton-Conducting Polymer Electrolytes. Electrochimica Acta 1995, 40, (3), Paddison, S. J., Proton conduction mechanisms at low degrees of hydration in sulfonic acid-based polymer electrolyte membranes. Annual Review of Materials Research 2003, 33, Souzy, R.; Ameduri, B.; Boutevin, B.; Gebel, G.; Capron, P., Functional fluoropolymers for fuel cell membranes. Solid State Ionics 2005, 176, (39-40), Wei, J.; Stone, C.; Steck, A. E., US Patent 5,422, Ding, J. F.; Chuy, C.; Holdcroft, S., Enhanced conductivity in morphologically controlled proton exchange membranes: Synthesis of macromonomers by SFRP and their incorporation into graft polymers. Macromolecules 2002, 35, (4), Serpico, J. M.; Ehrenberg, S. G.; Fontanella, J. J.; Jiao, X.; Perahia, D.; McGrady, K. A.; Sanders, E. H.; Kellogg, G. E.; Wnek, G. E., Transport and structural studies of sulfonated styrene-ethylene copolymer membranes. Macromolecules 2002, 35, (15), Devaux, J.; Delimoy, D.; Daoust, D.; Legras, R.; Mercier, J. P.; Strazielle, C.; Nield, E., On the Molecular-Weight Determination of a Poly(Aryl-Ether-Ether- Ketone) (Peek). Polymer 1985, 26, (13), Wnek, G. E.; Serpico, J. M.; Ehrenberg, S. G.; Zador, E., Hydrocarbon proton exchange membranes. ABSTRACTS OF PAPERS OF THE AMERICAN CHEMICAL SOCIETY 1999, 217(2), U488-U488, 340-PMSE 37

63 38. Gebel, G., Structural evolution of water swollen perfluorosulfonated ionomers from dry membrane to solution. Polymer 2000, 41, (15), Gao, Y.; Robertson, G. P.; Guiver, M. D.; Jian, X. G., Synthesis and characterization of sulfonated poly(phthalazinone ether ketone) for proton exchange membrane materials. Journal of Polymer Science Part a-polymer Chemistry 2003, 41, (4), Fanta, P. E., The Ullmann Synthesis of Biaryls, In 1964; Vol. 64, pp Kreuer, K. D., On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells. Journal of Membrane Science 2001, 185, (1), Figuly, G. D.; Martin, J. C., Ortho Lithiations of Arenesulfonic Acids - New Methodology for Electrophilic Aromatic Substitutions. Journal of Organic Chemistry 1980, 45, (18), Frost, C. G.; Hartley, J. P.; Whittle, A. J., Indium-catalysed aryl and alkyl sulfonylation of aromatics. Synlett 2001, (6), Fujimoto, C. H.; Hicknert, M. A.; Cornelius, C. J.; Loy, D. A., Ionomeric poly(phenylene) prepared by diels-alder polymerization: Synthesis and physical properties of a novel polyelectrolyte. Macromolecules 2005, 38, (12), Genies, C.; Mercier, R.; Sillion, B.; Petiaud, R.; Cornet, N.; Gebel, G.; Pineri, M., Stability study of sulfonated phthalic and naphthalenic polyimide structures in aqueous medium. Polymer 2001, 42, (12), Gieselman, M. B.; Reynolds, J. R., Water-Soluble Polybenzimidazole-Based Polyelectrolytes. Macromolecules 1992, 25, (18), Hickner, M. A.; Ghassemi, H.; Kim, Y. S.; Einsla, B. R.; McGrath, J. E., Alternative polymer systems for proton exchange membranes (PEMs). Chemical Reviews 2004, 104, (10), Rikukawa, M.; Sanui, K., Proton-conducting polymer electrolyte membranes based on hydrocarbon polymers. Progress in Polymer Science 2000, 25, (10),

64 49. Rusanov, A. L.; Likhatchev, D. Y.; Mullen, K., Pronton-conducting electrolyte membranes based on aromatic condensation polymers. Russian Chemical Review 2002, 71, (9), Kim, Y. S.; Wang, F.; Hickner, M.; McCartney, S.; Hong, Y. T.; Harrison, W.; Zawodzinski, T. A.; McGrath, J. E., Effect of acidification treatment and morphological stability of sulfonated poly(arylene ether sulfone) copolymer proton-exchange membranes for fuel-cell use above 100 degrees C. Journal of Polymer Science Part B-Polymer Physics 2003, 41, (22), Robeson, L. M.; Matzner, M., US Patent 4,380, Ueda, M.; Toyota, H.; Ouchi, T.; Sugiyama, J. I.; Yonetake, K.; Masuko, T.; Teramoto, T., Synthesis and Characterization of Aromatic Poly(Ether Sulfone)S Containing Pendant Sodium-Sulfonate Groups. Journal of Polymer Science Part a-polymer Chemistry 1993, 31, (4), Wang, F.; Hickner, M.; Kim, Y. S.; Zawodzinski, T. A.; McGrath, J. E., Direct polymerization of sulfonated poly(arylene ether sulfone) random (statistical) copolymers: candidates for new proton exchange membranes. Journal of Membrane Science 2002, 197, (1-2), Harrison, W. L.; Wang, F.; Mecham, J. B.; Bhanu, V. A.; Hill, M.; Kim, Y. S.; McGrath, J. E., Influence of the bisphenol structure on the direct synthesis of sulfonated poly(arylene ether) copolymers. I. Journal of Polymer Science Part a- Polymer Chemistry 2003, 41, (14), Wiles, K. B.; Bhanu, V. A.; Wang, F.; McGrath, J. E., Synthesis and characterization of sulfonated poly(arylene sulfide sulfone) copolymers as candidates for proton exchange membranes. Abstracts of Papers of the American Chemical Society 2002, 224, U359-U Kim, Y. S.; Sumner, M. J.; Harrison, W. L.; Riffle, J. S.; McGrath, J. E.; Pivovar, B. S., Direct methanol fuel cell performance of disulfonated poly-arylene ether benzonitrile) copolymers. Journal of the Electrochemical Society 2004, 151, (12), A2150-A

65 57. Wang, F.; Li, J. K.; Chen, T. L.; Xu, J. P., Synthesis of poly(ether ether ketone) with high content of sodium sulfonate groups and its membrane characteristics. Polymer 1999, 40, (3), Wang, F.; Chen, T. L.; Xu, J. P., Sodium sulfonate-functionalized poly(ether ether ketone)s. Macromolecular Chemistry and Physics 1998, 199, (7), Xing, P. X.; Robertson, G. P.; Guiver, M. D.; Mikhailenko, S. D.; Kaliaguine, S., Sulfonated poly(aryl ether ketone)s containing naphthalene moieties obtained by direct copolymerization as novel polymers for proton exchange membranes. Journal of Polymer Science Part a-polymer Chemistry 2004, 42, (12), Gao, Y.; Robertson, G. P.; Guiver, M. D.; Jian, X. G.; Mikhailenko, S. D.; Wang, K. P.; Kaliaguine, S., Direct copolymerization of sulfonated poly(phthalazinone arylene ether)s for proton-exchange-membrane materials. Journal of Polymer Science Part a-polymer Chemistry 2003, 41, (17), Summer, M. J.; Harrison, W. L.; Weyers, R. M.; Kim, Y. S.; McGrath, J. E.; Riffle, J. S.; Brink, A.; Brink, M. H., Novel proton conducting sulfonated poly(arylene ether) copolymers containing aromatic nitriles. Journal of Membrane Science 2004, 239, (2), Kim, Y. S.; Wang, F.; Hickner, M.; Zawodzinski, T. A.; McGrath, J. E., Fabrication and characterization of heteropolyacid (H3PW12O40)/directly polymerized sulfonated poly(arylene ether sulfone) copolymer composite membranes for higher temperature fuel cell applications. Journal of Membrane Science 2003, 212, (1-2), Genies, C.; Mercier, R.; Sillion, B.; Cornet, N.; Gebel, G.; Pineri, M., Soluble sulfonated naphthalenic polyimides as materials for proton exchange membranes. Polymer 2001, 42, (2), Cornet, N.; Diat, O.; Gebel, G.; Jousse, F.; Marsacq, D.; Mercier, R.; Pineri, M., Sulfonated polyimide membranes: a new type of ion-conducting membrane for electrochemical applications. Journal of New Materials for Electrochemical Systems 2000, 3, (1), Vallejo, E.; Pourcelly, G.; Gavach, C.; Mercier, R.; Pineri, M., Sulfonated polyimides as proton conductor exchange membranes. Physicochemical 40

66 properties and separation H+/Mz+ by electrodialysis comparison with a perfluorosulfonic membrane. Journal of Membrane Science 1999, 160, (1), Zhang, Y.; Litt, M. H.; Savinell, R. F.; Wainright, J. S.; Vendramini, J., Molecular design of polyimides toward high proton conducting materials. Abstracts of Papers of the American Chemical Society 2000, 220, U307-U Litt, M. H.; Zhang, Y.; Savinell, R. F.; Wainright, J. S., Molecular design considerations in the synthesis of high-conductivity PEMs for fuel cells. Abstracts of Papers of the American Chemical Society 1999, 218, U560-U Guo, X. X.; Fang, J. H.; Watari, T.; Tanaka, K.; Kita, H.; Okamoto, K. I., Novel sulfonated polyimides as polyelectrolytes for fuel cell application. 2. Synthesis and proton conductivity, of polyimides from 9,9-bis(4-aminophenyl)fluorene-2,7- disulfonic acid. Macromolecules 2002, 35, (17), Fang, J. H.; Guo, X. X.; Harada, S.; Watari, T.; Tanaka, K.; Kita, H.; Okamoto, K., Novel sulfonated polyimides as polyelectrolytes for fuel cell application. 1. Synthesis, proton conductivity, and water stability of polyimides from 4,4 '- diaminodiphenyl ether-2,2 '-disulfonic acid. Macromolecules 2002, 35, (24), Yin, Y.; Fang, J. H.; Cui, Y. F.; Tanaka, K.; Kita, H.; Okamoto, K., Synthesis, proton conductivity and methanol permeability of a novel sulfonated polyimide from 3-(2 ',4 '-diaminophenoxy)propane sulfonic acid. Polymer 2003, 44, (16), Einsla, B. R.; Kim, Y. S.; Hickner, M. A.; Hong, Y. T.; Hill, M. L.; Pivovar, B. S.; McGrath, J. E., Sulfonated naphthalene dianhydride based polyimide copolymers for proton-exchange-membrane fuel cells II. Membrane properties and fuel cell performance. Journal of Membrane Science 2005, 255, (1-2), Einsla, B. R.; Hong, Y. T.; Kim, Y. S.; Wang, F.; Gunduz, N.; McGrath, J. E., Sulfonated naphthalene dianhydride based polyimide copolymers for protonexchange-membrane fuel cells. I. Monomer and copolymer synthesis. Journal of Polymer Science Part a-polymer Chemistry 2004, 42, (4),

67 73. Shobha, H. K.; Sankarapandian, M.; Glass, T. E.; McGrath, J. E., Sulfonated aromatic diamines as precursors for polyimides for proton exchange membranes. Abstracts of Papers of the American Chemical Society 2000, 220, U278-U Kim, Y. S.; Hickner, M. A.; Dong, L. M.; Pivovar, B. S.; McGrath, J. E., Sulfonated poly(arylene ether sulfone) copolymer proton exchange membranes: composition and morphology effects on the methanol permeability. Journal of Membrane Science 2004, 243, (1-2), Kaucher, M. S.; Lam, Y. F.; Pieraccini, S.; Gottarelli, G.; Davis, J. T., Using diffusion NMR to characterize guanosine self-association: Insights into structure and mechanism. Chemistry-a European Journal 2004, 11, (1), Bowen, W. R.; Doneva, T. A.; Yin, H. B., Polysulfone-sulfonated poly(ether ether) ketone blend membranes: systematic synthesis and characterisation. Journal of Membrane Science 2001, 181, (2), Kerres, J.; Ullrich, A.; Meier, F.; Haring, T., Synthesis and characterization of novel acid-base polymer blends for application in membrane fuel cells. Solid State Ionics 1999, 125, (1-4), Ma, Y. L.; Wainright, J. S.; Litt, M. H.; Savinell, R. F., Conductivity of PBI membranes for high-temperature polymer electrolyte fuel cells. Journal of the Electrochemical Society 2004, 151, (1), A8-A Wang, J. T.; Savinell, R. F.; Wainright, J.; Litt, M.; Yu, H., A H-2/O-2 fuel cell using acid doped polybenzimidazole as polymer electrolyte. Electrochimica Acta 1996, 41, (2), Munson, R. A., Self-Dissociative Equilibria in Molten Phosphoric Acid. In 1964; Vol. 68, pp Savinell, R.; Yeager, E.; Tryk, D.; Landau, U.; Wainright, J.; Weng, D.; Lux, K.; Litt, M.; Rogers, C., A Polymer Electrolyte for Operation at Temperatures up to 200-Degrees-C. Journal of the Electrochemical Society 1994, 141, (4), L46-L Wasmus, S.; Valeriu, A.; Mateescu, G. D.; Tryk, D. A.; Savinell, R. F., Characterization of H3po4-Equilibrated Nafion(R)-117 Membranes Using H-1 and P-31 Nmr-Spectroscopy. Solid State Ionics 1995, 80, (1-2),

68 83. Vargas, M. A.; Vargas, R. A.; Mellander, B. E., More studies on the PVAl+H3PO2+H2O proton conductor gels. Electrochimica Acta 2000, 45, (8-9), Vargas, M. A.; Vargas, R. A.; Mellander, B. E., New proton conducting membranes based on PVAL/H3PO2/H2O. Electrochimica Acta 1999, 44, (24), Kreuer, K. D.; Fuchs, A.; Ise, M.; Spaeth, M.; Maier, J., Imidazole and pyrazolebased proton conducting polymers and liquids. Electrochimica Acta 1998, 43, (10-11), Doyle, M.; Choi, S. K.; Proulx, G., High-temperature proton conducting membranes based on perfluorinated ionomer membrane-ionic liquid composites. Journal of the Electrochemical Society 2000, 147, (1), Yang, C.; Costamagna, P.; Srinivasan, S.; Benziger, J.; Bocarsly, A. B., Approaches and technical challenges to high temperature operation of proton exchange membrane fuel cells. Journal of Power Sources 2001, 103, (1), Sun, J. Z.; Jordan, L. R.; Forsyth, M.; MacFarlane, D. R., Acid-organic base swollen polymer membranes. Electrochimica Acta 2001, 46, (10-11), Bozkurt, A.; Meyer, W. H.; Wegner, G., PAA/imidazol-based proton conducting polymer electrolytes. Journal of Power Sources 2003, 123, (2), Subbaraman, R.; Ghassemi, H.; Zawodzinski, T. A., 4,5-dicyano-1H-[1,2,3]- triazole as a proton transport facilitator for polymer electrolyte membrane fuel cells. Journal of the American Chemical Society 2007, 129, (8), Arimura, T.; Ostrovskii, D.; Okada, T.; Xie, G., The effect of additives on the ionic conductivity performances of perfluoroalkyl sulfonated ionomer membranes. Solid State Ionics 1999, 118, (1-2), Kwak, S. H.; Yang, T. H.; Kim, C. S.; Yoon, K. H., Polymer composite membrane incorporated with a hygroscopic material for high-temperature PEMFC. Electrochimica Acta 2004, 50, (2-3), Poltarzewski, Z.; Wieczorek, W.; Przyluski, J.; Antonucci, V., Novel proton conducting composite electrolytes for application in methanol fuel cells. Solid State Ionics 1999, 119, (1-4),

69 94. Bocarsly, A. B.; Krishnan, L.; Zhang, T.; Mann, J., Organic-inorganic composite membrane for elevated temperature operation of PEM fuel cells in the o C range. Fourth International Symposium on Proton Conducting Membrane Fuel Cells, The Electrochemical Society: Honolulu, HI Zaidi, S. M. J.; Mikhailenko, S. D.; Robertson, G. P.; Guiver, M. D.; Kaliaguine, S., Proton conducting composite membranes from polyether ether ketone and heteropolyacids for fuel cell applications. Journal of Membrane Science 2000, 173, (1), Mecheri, B.; D'Epifanio, A.; Di Vona, M. L.; Traversa, E.; Licoccia, S.; Miyayama, M., Sulfonated polyether ether ketone-based composite membranes doped with a tungsten-based inorganic proton conductor for fuel cell applications. Journal of the Electrochemical Society 2006, 153, (3), A463-A Dimitrova, P.; Friedrich, K. A.; Stimming, U.; Vogt, B., Modified Nafion((R))- based membranes for use in direct methanol fuel cells. Solid State Ionics 2002, 150, (1-2), Sweikart, M. A.; Herring, A. M.; Turner, J. A.; Williamson, D. L.; McCloskey, B. D.; Boonrueng, S. R.; Sanchez, M., 12-tungstophosphoric acid composites with sulfonated or unsulfonated epoxies for high-temperature PEMFCs. Journal of the Electrochemical Society 2005, 152, (1), A98-A

70 Chapter 2. Literature Review and Research Overview 2.1 Fuel Cell Challenges and Durability PEMFCs are promising energy conversion technologies with the potential to convert chemical energy to electrical energy upon demand. PEMFCs are however facing substantial challenges before large-scale commercialization can be realized. The infrastructure for the transportation and storage of the fuel, the high cost of stack with the contributions from membranes, expensive precious metal catalysts, and large quantities of bipolar plates that usually made from graphite or its composites, and finally the durability of the system, especially for the automobile applications, must be addressed. From the DOE Hydrogen Program 2006 annual report, 5,000 hours or 150,000 miles by 2010 has been the target life time for PEMFC system to be used in transportation applications. 1, 2 Life time testing can be dated back to the system evaluation of highly pressurized PEMFC stacks developed by GE for the space program (stationary system), where the best result was thousands of hours without failure. 3 Ballard has demonstrated an operating life time of more than 2,000 hours using real drive cycle testing under regular driving conditions. 4 The large gap between the target and state-of-the-art life times demands critical and extensive research efforts to improve the system durability. Durability issues can be studied by two drastically different approaches: one is through theoretical modeling based on the available life time data, and the other is through actual experimental testing. The real experimental observation can be a very expensive process if the cost of fuels, the test station, and costs of labor, electricity, and 45

71 system maintenance are all factored in. The prediction of life time, through proper statistical modeling of the in situ (in-field) and ex situ (accelerated) testing data, seems to be more feasible for continuous system fine-tuning and mitigation to improve the performance. A: End plates B: Current collector/gaskets C: Graphite flow field plates D: MEA + Gasket Figure 2-1. Scheme of a Single PEM Fuel Cell Showing the Location of the Components 5 Regardless of whether in situ or ex situ testing data is used for life time evaluation, from the literature it is unclear whether there is a single, dominant mode of failure for PEMFC systems, with one or more of the following factors cited as likely 46

72 contributors: the failure of catalysts, mechanical failure of membrane, and chemical degradation of membrane. Most likely, all contribute under specific conditions. The durability study of various components of a PEMFC system will be reviewed below. The various components are schematically shown in Figure Durability Studies of Various Components End Plates and Current Collectors The end plates and current collectors shown in Figure 2-1 are normally made by good metal conductors such as copper and aluminum. 5, 6 The exposure of copper and aluminum to acidic water in PEMFC system under operating conditions may cause the oxidation and dissolution of metal ions, which can be carried over and cycled back into the stack. 7 The metal ion contamination from end plates causes fuel cell performance decay and membrane degradation (to be discussed in detail in the following sections). 5 Additionally, the exposure of these plates to reactant gases and water vapor can lead to the formation of metal oxide films that may eventually lead to increase in the throughthickness conductivity of the current collector and the interfacial resistance between these plates and the adjacent bipolar plates Sealing Gaskets Sealing gaskets are very important components for PEMFC system because they are necessary to separate the gas/fuel compartments and prevent leaks to outside world. Silicon rubbers are often used as the sealing materials. These sealing gaskets are compressed between bipolar plates in the acidic environment that is rich in hot humid 47

73 vapors and reactive fuels. The long-term durability of sealing gaskets is critical, because severe fuel leakage and alternation of the stoichiometric ratio of reactant gases can result in the decrease of stack power efficiency if they degrade and malfunction. 8 Out of the limited literature, a few reports show the following degradation modes and their corresponding consequences: 1) primarily by surface probing techniques, such as XPS and EDX, residues of the silicone fragments were detected on the electrode surface, and significant enrichment of silicone residues were also found on the platinum catalyst surface. 9 Both phenomena may result in the poisoning of the catalyst and alter the hydrophilic/hydrophobic characteristics of the electrodes. 2) The surface chemistry of the silicone rubber was found to undergo de-crosslinking and chain-scission by XPS analysis, and silicon-based leachants were detected by atomic adsorption spectroscopy. 10 It is therefore necessary to also improve the chemical sturdiness of the sealing materials Bipolar Plates Bipolar plates are often made of graphite based polymer composites such as epoxy resins filled with graphite fibers. Fowler suggested that the degradation of bi-polar plates can happen by corrosion, fouling, cracking, fuel leaking, and the deformation and leaching out of contaminants to the cell. 8 The polymer binders in the bipolar plates may be degraded when the cell repeatedly goes through thermal and hydration cycles. The degradation can cause loss of adhesion, loss of integrity (which increases contact resistance of decrease of electronic conductivity), and loss of mechanical strength Gas Diffusion Layer (GDL) 48

74 Figure 2-2. Back Scattered Images by Electron Probe Micro Analyzer for MEA before and after 1800 hour of operation. (a) Anode, before; (b) cathode, before; (c) anode, after; (d) cathode, after. 11 The performance decay of GDLs can be characterized by changes in hydrophilicity and by dimensional changes such as porosity and tortuosity. A common but lethal mode of the GDL degradation is the delamination of GDL and catalyst layers, 49

75 which was observed to cause the failure of a 5-kW PEMFC stack after 8,000 hours of operation. The delamination at both electrodes before and after the cell stack failure is shown in Figure The cause for such a delamination process was postulated to be the tensile stresses as a result of MEA hydration and dehydration cycle. 6 Additionally, the extended exposure of MEA to water can lead to a permanent loss of GDL performance because of deterioration involving changes of porosity, changes of the carbon surface functional groups, and the accumulation of hydrophilic impurities as well as the loss of PTFE coatings. 8 The detailed mechanistic explanation was not discussed Electrodes and Catalyst In addition to membrane degradation, the degradation of electrode and catalyst is critical for PEMFC performance decay,. A number of mechanisms contribute to the overall catalyst degradation: 1) catalyst agglomeration over the operation course; 12 2) catalyst dissolution and migration; 13 3) the corrosion of the carbon based catalyst 14, 15 support. The combination of these degradation modes leads to the loss of utilizable catalyst surface area and a decrease of catalytic efficiency due to the alteration of the effective three-phase contact catalytic interface described in the first chapter. Platinum dissolution was reported in literature to occur via two different possible pathways 16, 17 : 1) Electrochemical: Pt Pt e - 2) Chemical: First formation of platinum oxide: Pt + H 2 O PtO + 2H + + 2e - Followed by dissolution: PtO + 2H + Pt 2+ + H 2 O 50

76 Platinum surface area loss was also proposed to be the result of three fundamentally different processes: 1) platinum dissolution and redeposition via the Oswald ripening mechanism, which results in coarsening of catalyst particles 18, 19 ; 2) coalescence of platinum nanoparticles through the migration of platinum nanocrystals on the carbon supports, also called Pt sintering 20 ; 3) agglomeration directly induced by the 14, 15 corrosion and disappearance of carbon support. Figure 2-3 shows the size change of platinum catalysts. 19 Figure 2-3. TEM Micrographs from (a) Pristine Pt/Vulcan Sample; (b) Powders Scraped from the Cathode Surface of the Cycled MEA Sample 19 PEMFC voltage decay is well recognized as one of the most common failure modes responsible for automotive fuel cell application Carbon corrosion in the electrodes is believed to be the major cause for such gradual voltage decay during start- 51

77 21, 23 up/shut-down cycles. Two possible routes are attributed as the cause of carbon support dissolution: 1) the carbon corrosion due to local fuel starvation as a consequence of the blockage of hydrogen access to anode catalyst layer 14, 15, 24 ; 2) the electrochemical oxidation of the carbon surface, initially changing to oxide functionalities followed by the eventual formation of CO 2 at the cathode. 25 As illustrated in Figure , carbon can react with oxygen radicals, such as hydroxyl and hydroperoxyl radicals, generated electrochemically in a running fuel cell (the formation of such radical species will be explained in details in the next section). Figure 2-4. Carbon Corrosion in the Absence of Pt 20 52

78 In addition to the above two major degradation modes that adversely affect the fuel cell performance, the catalyst poisoning and contamination can also cause the performance loss in a PEMFC. One common poison for Pt catalysts is CO introduced as an impurity from hydrogen gas production. CO poisons/deactivates the catalytic activity of platinum catalysts by adsorption of CO molecules on the platinum surface, effectively blocking the available catalytic sites for cell reactions. 3 Other contaminants, such as those discussed above, i.e. sealing material fragments, membrane fragments, and the rusty corrosion product of various metal plates, may also have similar blocking effect to decrease the cell performance. 2.3 Literature Review of Durability Studies of Membranes One of the major barriers to commercialization of PEMFC systems at the present time is the durability of the polymer electrolyte membranes themselves, hence limiting the functional life times of the fuel cell systems. 26 An area of intense research interest is the understanding of the degradation mechanisms of the membrane so as to suggest approaches to mitigate degradation or develop new material designs that can provide future membranes with adequate robustness to meet the durability requirements. Membrane degradation can be broadly categorized into two major types, mechanical degradation and the chemical degradation. A PEMFC membrane may experience these two mechanisms in parallel in a running fuel cell. Possible contributors to the membrane degradation will be discussed first, followed a brief review of chemical degradation studies of various PEMFC membranes. 53

79 2.3.1 Mechanical Degradation Mechanical failure of PEMFC membranes is often found to be the cause of early failure of a stack. For example, the development of a pin-hole across the membrane eventually leads to the leakage of reactant gases Mechanical degradation of membranes can take other forms, such as cracks, tearing, and punctures. 29 Thin membranes are potentially advantageous to achieve higher proton conductance. For such thin membranes, extra caution must be taken in handling and assembling the stack, as any foreign particle or over-tightening can easily lead to perforations which may trigger sudden mechanical failures. For automotive applications, it is hypothesized that the repeated rapid changes in power output from fuel cell stack may cause large changes in the local relative humidity for the membrane. This frequent large change of humidity translates to the change of hydration level of membrane as a response, which subsequently causes membranes to repeatedly swell and shrink. In the cell stack, membranes are highly constrained. Such dimensional changes therefore induce tensile and compressive loads - a fatigue type mechanical process that the membranes may encounter. Experiments have been designed to mimic such a fatigue process by exposing the membrane to repeated cycles of humidity changes. 30 The fatigue strength or the safety limit of Nafion was found to be ~ 1.5 MPa, which is 1/10 of the tensile strength of the membrane. The cyclic stress and dimensional change in certain extreme conditions can be substantial to cause the mechanical failure of the membrane Chemical Degradation 54

80 Membrane chemical degradation was observed to be a life-limiting factor in 1960 s Gemini space program. The PSSA type membrane had only 500~1,000 hours of life time. The indications of chemical degradation include large quantities of low molecular weight polystyrene sulfonic acids identified in the product water, and sulfobenzoic acid and p-benzaldehyde sulfonic acid functionalities found on the postmortem membrane. 31 After performing a series of control experiments, such as controlling the gaseous environment of being pure oxygen, pure hydrogen, or both, with or without the presence of the platinum catalyst particles, the degradation mechanism was proposed to be the hydroxyl and hydroperoxyl radical species attacks via the following steps 20, : 1) H 2 2H. (via Pt catalyst) 2) H.. + O 2 (diffused through membrane to anode) HO 2 3) HO 2. + H. H 2 O 2 4) H 2 O 2 + M 2+ M 3+ + HO. + OH - (M 2+ and M 3+ are the contaminant metal ions) 5) HO. + H 2 O 2 H 2 O + HO 2. From other studies, it was proposed and electrochemically confirmed that hydrogen peroxide can also be formed from the direct incomplete reduction of oxygen on either the platinum surface 5 or at the anode 34 via the scheme: O 2 + 2H + + 2e - H 2 O 2. The presence of H 2 O 2 was also detected in the product water, 35 and the hydroxyl radicals were also found on the cathode side by spin trapping method detected by electron paramagnetic resonance (EPR). 35 Even more convincingly, Panchenko et al developed an in situ EPR technique that can directly monitor the generation of radical species in the 55

81 fuel cell that was put in the spectrometer. 36 Also, by spin trapping method, various free radicals were detected: organic radical round on membrane surface near the cathode, and free radical intermediates of the oxidation reaction near the anode. 35 Figure 2-5. Possible Sites for Radical Attack on Several Styrenic PSAs 35 The formation of H 2 O 2 and highly reactive radical species can potentially account for the majority of the degradation products found for membranes with oxidizable C-H bonds. Recent reports reconfirmed the degradation of PSSA by hydroxyl radicals. 37 Free 56

82 radical addition to the aromatic rings, the structure of interest of various sulfonated PEEK, PSU, and polystyrene based membranes, was observed and shown in Figure Sulfonated model compound studies under hydroxyl radical environment shows that within a few hours, the chain scission of styrenic-type structure, the loss of sulfonic acid group from the aromatic ring, and the degradation of aromatic rings (opening) all occurred. The chemistries involved are depicted in Figure b) 57

83 c) Figure 2-6. Mechanisms (a-c) for Radical Attack and Degradation of Membrane Containing Aromatic Groups Chemical Degradation of Nafion PFSAs It is well known within the field that the effluent water from operating fuel cells contain significant and relatively constant levels of fluoride ions. This is coupled to 58

84 gradual thinning of the membrane, suggesting chemical degradation to be a consequence of their operation, and perhaps a major contributing factor to their ultimate failure. The current state of knowledge describing possible chemical mechanisms of 26, 33 degradation for PEMFC membranes was recently reviewed. Using X-ray photoelectron spectroscopy (XPS) to analyze the MEA before and after fuel cell operation, Huang found that certain amount of -CF2- groups of Nafion are degraded into -HCF- and -CCF- groups. 39 Schlick et al. reported the observation of a polymeric radical where the unpaired electron is located on the tertiary backbone carbon atom in Nafion PFSA that is linked to the pendant side chain. 40 Chain end radicals with structures like R f -O-CF 2 -CF 2 were also identified on the side chain radical by electron spin resonance (ESR), when Nafion membranes, saturated with metal counter ions, were exposed to UV radiation with the presence of H 2 O 2. The reaction between the Fe(III) counter ions and sulfonic acid groups on the side chains was proposed to produce such chain end radicals: R-O-CF 2 -CF 2 -SO Fe(III) R-O-CF 2 -CF 2 -SO 3 + Fe(II), followed by rearrangement via elimination of SO 2 and O 2. Direct soaking of Nafion in 3% (v/v) aqueous H 2 O 2 solution for up to 30 days leads to the formation of S-O-S bond as determined by Fourier transform infrared spectroscopy (FTIR). The S-O-S bond formation was thought to be the result of crosslinking of sulfonic acid groups on the side chains, which subsequently reduces the ductility and proton conductivity. In another study 41, FTIR studies revealed trace amount of R-SO 2 F or S-O-S formation when Nafion was degraded by H 2 O 2 /Fe(II) solutions; such reagents are commonly known as Fenton s reagent, and are widely used to generate hydroxyl and hydroperoxyl radicals. 42 The authors also commented that the side chains were decomposed more easily than the main chain, based on the 19 F nuclear 59

85 magnetic resonance (NMR) integral ratio changes of Nafion repeat units. FTIR, 13 C NMR, 19 F NMR, and mass spectroscopic (MS) analysis of the degradation test solution exhibited fluorinated fragments with the structure largely resembling the derivated 30, 43 Nafion side chain structure. O R f CF 2 C OH OH -CO 2 OH R f CF 2 R f CF 2 OH O R f C OH H 2 O -HF O R f C F - HF Figure 2-7. Chain End un-zipping Degradation Mechanism CF 2 CF 2 O SO H 2 O + HSO - CF 2 CF 2 OH 4 CF 2 CF 2 OH + H 2 O CF 2 COOH + 2 HF Figure 2-8. The Formation of Carboxylic Acid End Groups in Fluoropolymer from the Hydrolysis of the Residue of Persulfate Initiator 44 Despite valuable information revealed from the literature cited above, the detailed degradation mechanism(s) leading to the observed chemical structural changes of Nafion and other PFSAs is still poorly understood. One important mechanism was proposed by Curtin et al. to explain the fluoride generation pathway. 45 As shown in Figure 2-7, the proposed degradation process starts from the carboxylic acid end groups (-COOH) that may be present in small concentrations in PFSAs. These end groups are unintentionally 60

86 introduced from the manufacturing process of Nafion and other ionomers via the hydrolysis of the persulfate initiators used in the polymerization of Nafion, as shown in Figure The degradation is proposed to proceed by a main chain unzipping mechanism: hydroxyl radicals abstract hydrogen atoms from terminal -COOH, followed by decarboxylation to form primary perfluorinated radicals. These primary radicals then react with available hydroxyl radicals to form primary fluorinated alcohols, which are highly unstable and rapidly decompose to acyl fluorides with elmination of HF. Subsequent hydrolysis of acyl fluorides yields carboxylic acid ends to re-enter the degradation cycle, shortening the chain by one net carbon unit. Fluorination of end groups leads to the reduction of reactive end group contents, but the fluoride evolution was not eliminated even when the reactive end groups are reduced to be close to zero, as shown in Figure A second degradation mechanism is therefore necessary to account for the significant, non-zero fluoride evolution observed when carboxylic acid end groups are eliminated, and potentially to explain the structures of PFSA degradation products identified as discussed above. 2.4 Objectives and Significance of Current Research The present study addresses the need for the construction of a coherent model of chemical degradation mechanisms PEMFC PFSA type ionomers that mimics the fuel cell reactive environment. In order to gain the benefit of standard chemical methods generally not easily deployed when studying the intractable ionomers, a family of low molecular weight model compounds (MCs) with structural characteristics similar to moieties found in PFSAs was examined. Additionally, the degradation test and product 61

87 analysis of MCs allow for the comparison of reactivities of different moieties toward degradation. In parallel, the degradations of the benchmark PFSA membranes, Nafion and 3M ionomers, were also investigated under the same condition used with the MCs. Thorough post-degradation structural properties analysis was also carried out in an attempt to correlate the chemical degradation and polymer property deteriorations. Relative kinetics of fluoride generation, as well as characterization of degradation products were considered as mechanistic probes. Fenton Test Relative Fluoride Emission Rate (FER) Normalized End-group Count Figure 2-9. Plot showing relative fluoride emission rate (FER) from Fenton s test as a function of concentration of reactive end-groups (recreated from original plot in reference 46) 62

88 References 1. DOE, Hydrogen Program Technical Targets Lightner, V., DOE Hydrogen program record, record #5036, date April 20, DOE Hydrogen program record Gottesfeld, S.; Zawodzinski, T. A., Polymer Electrolyte Fuel Cells. Advances in Electrochemical Science and Engineering, Edited by Alkire, R.C. et al., WILEY- VCH. 1997, Vol.5, Pozio, A.; Silva, R. F.; De Francesco, M.; Giorgi, L., Nafion degradation in PEFCs from end plate iron contamination. Electrochimica Acta 2003, 48, (11), Wiklinson, D. P.; St-Pierre, J., Chapter 47: Durability. Handbook of Fuel Cells - Fundamentals, Technology, Applications. Wielstich, W.; Gasteiger, H.A.; Lamm, A.(Eds.), John Wiley & Sons Liu, D., Durability study of proton exchange membrane fuel cells via experimental investigations and mathematical modeling. PhD Dissertation, Virginia Polytechnic Institute and State University Fowler, M.; Mann, R. F.; Amplett, J. C.; Peppley, B. A.; Roberge, P. R., Chapter 50: Reliability Issues and Voltage Degradation. Handbook of Fuel Cells - Fundamentals, Technology, Applications. Wielstich, W.; Gasteiger, H.A.; Lamm, A.(Eds.), John Wiley & Sons Schulze, M.; Knori, T.; Schneider, A.; Gulzow, E., Degradation of sealings for PEFC test cells during fuel cell operation. Journal of Power Sources 2004, 127, (1-2), Tan, J.; Chao, Y. J.; Van Zee, J. W.; Lee, W. K., Degradation of elastomeric gasket materials in PEM fuel cells. Materials Science & Engineering, A: Structural Materials: Properties, Microstructure and Processing 2007, A445- A446,

89 11. Ahn, S. Y.; Shin, S. J.; Ha, H. Y.; Hong, S. A.; Lee, Y. C.; Lim, T. W.; Oh, I. H., Performance and lifetime analysis of the kw-class PEMFC stack. Journal of Power Sources 2002, 106, (1-2), Luo, Z.; Li, D.; Tang, H.; Pan, M.; Ruan, R., Degradation behavior of membraneelectrode-assembly materials in 10-cell PEMFC stack. International Journal of Hydrogen Energy 2006, 31, (13), Bi, W.; Gray, G. E.; Fuller, T. F., PEM Fuel Cell Pt/C Dissolution and Deposition in Nafion Electrolyte. Electrochemical and Solid-State Letters 2007, 10, (5), B101-B Patterson, T. W.; Darling, R. M., Damage to the cathode catalyst of a PEM fuel cell caused by localized fuel starvation. Electrochemical and Solid State Letters 2006, 9, (4), A183-A Reiser, C. A.; Bregoli, L.; Patterson, T. W.; Yi, J. S.; Yang, J. D. L.; Perry, M. L.; Jarvi, T. D., A reverse-current decay mechanism for fuel cells. Electrochemical and Solid State Letters 2005, 8, (6), A273-A Darling, R. M.; Meyers, J. P., Kinetic model of platinum dissolution in PEMFCs. Journal of the Electrochemical Society 2003, 150, (11), A1523-A Ball, S.; Hudson, S.; Theobald, B.; Thompsett, D., The effect of dynamic and steady state voltage excursion on the stability of carbon supported Pt and PtCo catalysts. ECS Transactions 2006, 3, Yu, P.; Pemberton, M.; Plasse, P., PtCo/C cathode catalyst for improved durability in PEMFCs. Journal of Power Sources 2005, 144, (1), Ferreira, P. J.; la O, G. J.; Shao-Horn, Y.; Morgan, D.; Makharia, R.; Kocha, S.; Gasteiger, H. A., Instability of Pt/C electrocatalysts in proton exchange membrane fuel cells - A mechanistic investigation. Journal of the Electrochemical Society 2005, 152, (11), A2256-A Cai, M.; Ruthkosky, M. S.; Merzougui, B.; Swathirajan, S.; Balogh, M. P.; Oh, S. H., Investigation of thermal and electrochemical degradation of fuel cell catalysts. Journal of Power Sources 2006, 160, (2), Kawahara, S.; Mitsushima, S.; Ota, K. I.; Kamiya, N., Deterioration of Pt catalyst under potential cycling. ECS Transactions 2006, 3,

90 22. Ye, S.; Hall, M.; Cao, H.; He, P., Degradation resistant cathodes in polymer electrolyte membrane fuel cells. ECS Transactions 2006, 3, Yu, P. T.; Gu, W. B.; Makharia, R.; Wagner, F. T.; Gasteiger, H. A., The impact of carbon stability on PEM fuel cell startup and shutdown voltage degradation. ECS Transactions 2006, 3, Chizawa, H.; Ogami, Y.; Naka, H.; Matsunaga, A.; Aoki, N.; Aoki, T., Study of accelerated test protocol for PEFC focusing on carbon corrosion. ECS Transactions 2006, 3, Waje, M. M.; Li, W. Z.; Chen, Z. W.; Yan, Y. S., Durability investigation of cupstacked carbon nanotubes supported Pt as PEMFC catalyst. ECS Transactions 2006, 3, Schiraldi, D. A., Perfluorinated polymer electrolyte membrane durability. Polymer Reviews 2006, 46, (3), Liu, W.; Ruth, K.; Rusch, G., Membrane Durability in PEM Fuel Cells. Journal of New Materials for Electrochemical Systems 2001, 4, (4), Liu, W.; Crum, M., Effective testing matrix for studying membrane durability in PEM fuel cells: Part I. Chemical durability. ECS Transactions 2006, 3, Crum, M.; Liu, W., Effective testing matrix for studying membrane durability in PEM fuel cells: Part II. Mechanical durability and combined mechanical and chemical durability. ECS Transactions 2006, 3, Tang, H.; Peikang, S.; Jiang, S. P.; Wang, F.; Pan, M., A degradation study of Nafion proton exchange membrane of PEM fuel cells. Journal of Power Sources 2007, 170, (1), LaConti, A. B.; Hamdan, M.; McDonald, R. C., Handbook of fuel cells: fundamentals, technology, and applications, Eds., Vielstich, W., Lamm, A., Gasteiger, H. 2003, 3, Xie, J.; Wood, D. L.; Wayne, D. M.; Zawodzinski, T. A.; Atanassov, P.; Borup, R. L., Durability of PEFCs at high humidity conditions. Journal of the Electrochemical Society 2005, 152, (1), A104-A

91 33. Collier, A.; Wang, H.; Yuan, X. Z.; Zhang, J.; Wilkinson, D. P., Degradation of polymer electrolyte membranes. International Journal of Hydrogen Energy 2006, 31, (13), Liu, W.; Zuckerbrod, D., In situ detection of hydrogen peroxide in PEM fuel cells. Journal of the Electrochemical Society 2005, 152, (6), A1165-A Panchenko, A.; Dilger, H.; Moller, E.; Sixt, T.; Roduner, E., In situ EPR investigation of polymer electrolyte membrane degradation in fuel cell applications. Journal of Power Sources 2004, 127, (1-2), Panchenko, A.; Dilger, H.; Kerres, J.; Hein, M.; Ullrich, A.; Kaz, T.; Roduner, E., In-situ spin trap electron paramagnetic resonance study of fuel cell processes. Physical Chemistry Chemical Physics 2004, 6, (11), Yu, J.; Yi, B.; Xing, D.; Liu, F.; Shao, Z.; Fu, Y.; Zhang, H., Degradation mechanism of polystyrene sulfonic acid membrane and application of its composite membranes in fuel cells. Physical Chemistry Chemical Physics 2003, 5, (3), Hubner, G.; Roduner, E., EPR investigation of HO. radical initiated degradation reactions of sulfonated aromatics as model compounds for fuel cell proton conducting membranes. Journal of Materials Chemistry 1999, 9, (2), Huang, C.; Tan, K. S.; Lin, J.; Tan, K. L., XRD and XPS analysis of the degradation of the polymer electrolyte in H2-O2 fuel cell. Chemical Physics Letters 2003, 371, (1,2), Kadirov, M. K.; Bosnjakovic, A.; Schlick, S., Membrane-Derived Fluorinated Radicals Detected by Electron Spin Resonance in UV-Irradiated Nafion and Dow Ionomers: Effect of Counterions and H2O2. Journal of Physical Chemistry B 2005, 109, (16), Chen, C.; Levitin, G.; Hess, D. W.; Fuller, T. F., XPS investigation of Nafion membrane degradation. Journal of Power Sources 2007, 169, (2), Walling, C., Fenton's reagent revisited. Accounts of Chemical Research 1975, 8, (4),

92 43. Healy, J.; Hayden, C.; Xie, T.; Olson, K.; Waldo, R.; Brundage, M.; Gasteiger, H.; Abbott, J., Aspects of the chemical degradation of PFSA ionomers used in PEM fuel cells. Fuel Cells (Weinheim, Germany) 2005, 5, (2), Pianca, M.; Barchiesi, E.; Esposto, G.; Radice, S., End groups in fluoropolymers. Journal of Fluorine Chemistry 1999, 95, (1-2), Curtin, D. E.; Lousenberg, R. D.; Henry, T. J.; Tangeman, P. C.; Tisack, M. E., Advanced materials for improved PEMFC performance and life. Journal of Power Sources 2004, 131, (1-2), Escobedo, G., Strategies to improve the durability of perfluorosulfonic acid membranes for PEM fuel cells. KFTCA International Symposium 2005, December 8-9, (Presentation data). 67

93 Chapter 3. Degradation of Model Compounds Under Mimic Fuel Cell Conditions 3.1 Introduction The requirement to withstand the chemically aggressive operating environment of PEM systems, i.e. oxidizing fuel, electrochemically formed reactive species of fuel (such as radicals), and high temperature, made perfluorinated ionomers (such as Nafion) and structurally-related products from 3M and Dow as early choices for use in PEM systems. These leading PEM candidates have been selected for their high proton conductivities and high mechanical, thermal and chemical stabilities. The perfluorinated nature of these polymers offers considerably higher stability and system life than are generally attainable with non-fluorinated polymers. Fluorination of organic molecules imparts extremely good thermal stability and relative chemical inertness due to the high strength of the resultant carbon-fluorine bonds. 1-8 Reaction of saturated perfluorinated compounds often requires extreme conditions. However, there are occasions where rather unexpected reactivity is observed for certain perfluorinated compounds at relatively mild conditions, such as the reaction of perfluorodecalin with certain sulfur and oxygen based nucleophiles in dipolar aprotic solvent at only o C to convert to a fully substituted thioether derivative with a yield 9, 10 of about 65% in 10 days along with the detection of fluoride. The tertiary carbon center in perfluorodecalin was believed to be critical in the proposed reaction mechanism as molecules like perfluorocyclohexane and n-perfluorohexane failed to react under equivalent conditions. A radical process, with the formation of a relatively stable tertiary 68

94 radical as the key to the subsequent defluorination process, was proposed to explain the mechanism as shown in Figure Na SAr + SET F F 2 - -F SET F F F F F - F -F F F -F F F SET Figure 3-1. Possible Pathways for the Formation of Perfluoroalkene from Perfluorodecalin by a Single Electron Transfer (SET) Process For Nafion-based PEMFC systems, numerous papers have reported continuous generation of fluoride during fuel cell operation These observations raise the question of how and where the fluoride is evolved, and its relationship with the cell performance. The structural similarities of Nafion, 3M and Dow membranes shown previously in Figure 1-10, offer the possibility to generalize the contributing sources leading to the degradation: the end-group chemistries, the fluoro-carbon moieties, and the ether linkages. In this chapter, a family of low molecular weight model compounds (MCs), bearing different structural characteristics similar to those found in PFSAs discussed above, will be used to closely examine the reactivity in degradation tests. The benefits of using MCs are: standard chemical analytical methods generally not available when studying the intractable fluorinated ionomers; and the possibility of examining the 69

95 reactivities of individual and/or combination of moieties through proper selection of MC structures. This work is potentially of great interest in both the fuel cell applications and general fluorine chemistry. Further mitigation of polymers with desired durability can be designed if the leading sources for degradation can be identified. Nafion CF 2 CF 2 CF 2 x CF y O CF 2 CF O CF 2 2 SO 3 H CF 3 MC1 MC2 O O 3M Ionomer CF 2 CF 2 CF 2 CF x y O CF 2 CF 2 2 SO 3H MC3 O HO C CF O CF 2 CF 2 CF 3 HO C CF O CF 2 CF 2 SO 3 H 2 HO C CF 2 CF 2 CF 2 SO 3 H CF 3 CF 3 MC4 MC5 MC6 O HO C CF 2 6 CF 3 F 3 C CF 2 CF 2 CF 2 SO 3 H F 3 C CF 2 CF 2 H 6 MC7 MC8 CF 3 CF 2 O CF 2 CF 2 SO 3 H 2 F 3 C CF 2 O CF 2 CF O CF 2 SO 3 H 2 CF 3 Figure 3-2. Chemical Structures of MCs and Ionomers Studied in Current Research 3.2 Materials The structures of model compounds (hereafter MC) considered in this work, along Nafion and 3M ionomers are given in Figure 3-2. MC1, perfluoro(2-methyl-3- oxahexanoic) acid, 97%, was purchased from Lancaster Synthesis. MC2, perfluoro(2- methyl-3-oxa-7-sulfonic heptanoic) acid, 96%, MC3, perfluoro(4-sulfonic butanoic) acid, 70

96 were provided by 3M. MC4, perfluoro-n-octanoic acid, 98%, MC5, nonafluorobutanesulfonic acid, and MC6, 1H,-perfluorooctane, were purchased from SynQuest Labs. MC7, perfluoro(3-oxahexanoic sulfonic) acid, and MC8, perfluoro(4- methyl-3-oxaoctanoic sulfonic) acid, were provided by 3M. All the MCs were used as received. Ferrous sulfate heptahydrate, 99%, and hydrogen peroxide solution, 30% (w/v), were obtained from Fisher. Total Ionic Strength Adjustment Buffer (TISAB II and TISAB III, with CDTA) solutions were purchased from Thermal Orion. Acetonitrile and ammonium acetate, both HPLC grade, were purchased from VWR. 4-Hydroxy-2,2,6,6- tetramethyl-piperidinooxy (4-hydroxy-TEMPO) was ordered from Acros and used as received. 3.3 Experiments Fluoride Concentration Measurement Fluoride ion concentration in aqueous solutions was measured using an ion selective electrode (ISE) (Mettler-Toledo, ISE part # , meter model number MX300), which was calibrated over the range ppm fluoride using NaF aqueous solutions. The detection accuracy limit is at least 0.1 ppm (5.26 x 10-6 M), which still gives a satisfactory calibration curve fit when compared to the theoretical value using the Nernst equation. All of the fluoride concentration data reported here were obtained by a direct measurement method against the calibration curve: the electrode was immersed into a solution containing 2 ml sample and 2 ml TISAB II solution (the solution was constantly stirred) and an potential reading of the meter was recorded after equilibrium was reached, typically 5-10 min. The electrode was checked 71

97 daily by a solution of known fluoride concentration to ensure accuracy and was recalibrated whenever deviation was observed Fenton s Degradation Tests Fenton s reagent, a combination of H 2 O 2 and a ferrous salt, is a very effective method to generate hydroxyl and peroxyl radicals. 21 These radicals are the most commonly attributed attacking species for the PEMFC as introduced in the previous chapters. The decomposition of H 2 O 2 is a very complicated system, but the major and generally-accepted reactions are shown below H 2 O 2 + Fe 2+ Fe 3+ + HO - + HO. HO. + H 2 O 2 HO. + H 2 O Many factors, such as stoichiometry and order of addition of reagent, can easily 21, 28 alter the products. In this study, known amount of H 2 O 2 was slowly added to a wellstirred solution containing locally excess Fe 2+ to minimize the consumption of hydroxyl radicals via reacting with hydrogen peroxide, i.e. the second reaction above, so as to maximize the generation of the highly reactive hydroxyl radicals to mimic the attacking species in the FC operation Degradation Test Procedures The exact concentration of H 2 O 2 formed in a real fuel cell is very difficult to measure, and is a function of many factors such as membrane thickness and location relative to the catalyst layers. A typical concentration measured by one group was found to be ppm, 14 which translates to approximately 0.5 mm. In this study, the ex-situ 72

98 accelerated degradation tests have been carried out in two extremes, namely the mild condition degradation test where low concentrations of H 2 O 2 and Ferrous ions are used, and the harsh condition degradation test with high concentrations of H 2 O 2 and Ferrous ions. a. Mild Condition Fenton s Degradation Test and Procedure Table 3-1. Concentrations of Reagents Used in Mild Degradation Test Fe(II) H 2 O 2 MC DI Water Reagents Concentrations 1.25 mm (ca 70 ppm) 11 mm 100 mm 50 ml (total) The concentrations of MCs and Fenton s reagents are tabulated in Table 3-1. Each MC was first mixed with a 40 ml aqueous solution containing 1.25 mm ferrous ions by dissolving ferrous sulfate heptahydrate in water (all the concentrations herein are calculated based on the total volume, i.e. the final volume after all the reagents are introduced into the reactor), then the solution was bubbled with nitrogen dry gas for at least 10 minutes to remove the oxygen that might quench radicals. The solution was subsequently heated to 70 ± 2º C, and hydrogen peroxide was added through an addition funnel at a slow dropping rate, typically 10 to 20 drops per minute. The reaction mixture was held at approximately 70ºC under nitrogen purge for 24 hours, followed by removal of a 2 ml aliquot from the reactor for fluoride concentration measurement. The tests were continued by adding fresh ferrous ions and hydrogen peroxide to react by the same 73

99 procedure described above, and another fluoride measurement was carried out after an interval of 24 hours. The sample process was repeated for 5-6 cycles, for a total of ca. 130 hours to complete the test. The data were presented by plotting the amount of fluoride detected as a function as accumulated degradation test time/intervals. Error bars signify the standard deviation of two to four replicas for various MCs. b. Harsh Condition Fenton s Degradation Test and Procedure Table 3-2. Concentrations of Reagents Used in Harsh Degradation Test Fe(II) H 2 O 2 MC DI Water Reagents Concentrations 400 mm 400 mm 100 mm 50 ml (total) The concentrations of MCs and Fenton s reagents are tabulated in Table 3-2. Solutions of Fe 2+ and MC were firstly mixed and stirred, with the N 2 bubbling for about 30 min. H 2 O 2 was then slowly added into the well-stirred solution by addition funnel. The reaction temperature was adjusted to about 70 o C, and the reactor throughout the whole course is purged by dry N 2 gas. The average reaction time was controlled to be around 24 hours. Once the reaction was finished, the precipitates generated will settle down to the bottom after about 30 min without stirring. F - measurement was done on 2 ml aliquots from the solution. Subsequently, the solution was filtered through a glass filter, and the precipitates were washed by copious amounts of water and acetonitrile. Finally, water and actonitrile were taken out by a rotary evaporator, leaving the treated 74

100 MC in the flask. This treated MC was used again to do the subsequent tests. A detailed scheme showing the harsh condition test procedure is shown in Figure 3-3. MC + Aging bath addition of H 2 O 2 N 2 protection T 68 o C Ca. 24 Hrs Precipitation Filtration & Extraction Oil Phase Water Phase F- Measurement Washing by EDTA Extraction by 1-Butanol F- Measurement check whether F- was carried over Oil Phase Rotavaporation to take out 1-Butanol 19 F NMR Figure 3-3. Scheme Showing General Sample Handling Procedure In Harsh Degradation Test in situ CO 2 Detection in Degradation Test In some degradation tests of carboxylic acid-containing MCs, the purging nitrogen gas was vented to a 50mM sodium hydroxide solution in a test tube containing phenolphthalein as indicator (pink when basic, colorless when acidic) to detect the generation of carbon dioxide. Control experiments were run without hydrogen peroxide, and without carboxylic acid-containing MCs to demonstrate that no false position CO 2 reading would be obtained. 75

101 3.3.5 UV Photolysis Degradation Test In parallel to the Fenton s reaction, UV photolysis of hydrogen peroxide was exploited as a metal-free source of radical for MC degradation. An advantage of this approach is that it eliminates the iron ions present in Fenton s testing, potentially not present in such high concentrations under actual fuel cell operation. It is well known in the literature that hydroxyl and peroxyl radicals are generated when hydrogen peroxide is exposed to UV radiation (ref). The light source used was an Oriel standard 100W Mercury lamp with a wavelength range of nm. MCs were mixed with DI water (100 mm, total volume of testing samples 3 ml in a quartz crucible placed at about 20 cms away from light source) and then further exposed to the UV radiation for 1 hour at room temperature, with and without the presence of hydrogen peroxide (400 mm) Intermediate Trapping by Stable Radicals 0.11 M of 4-hydroxy-TEMPO aqueous solution was prepared by dissolving 4- hydroxy-tempo in de-ionized (DI) water. The resulting 4-hydroxy-TEMPO solution was kept in the refrigerator after preparation and during the course of each degradation test cycle (typically 24 hours). The trapping of possible radical intermediates was done by quickly removing 1 ml of reaction medium from the degradation flask to a vial containing 9 ml prepared 4-hydroxy-TEMPO solution at different times after the initiation of each degradation cycle. Typical time variables per degradation cycles are: 30 minutes, 1.5 hours, 3 hours, 6 hours, and 12 hours. Subsequently, the vial containing 4-hydroxy-TEMPO and the reaction medium was hand-shaken for 15 seconds to mix thoroughly, followed by storing in the refrigerator prior to the LC-MS and NMR analysis. 76

102 F Nuclear Magnetic Resonance (NMR) NMR spectra were obtained using a Varian AS MHz spectrometer. Acetonitrile-D 3 (Fisher) was used as the solvent for MC treatment experiments. All the chemical shifts are referenced to CFCl 3 (defined as 0 ppm) as standard Liquid Chromatography-Mass Spectrometry (LC-MS) The LC-MS analysis was carried in a Thermo LC-MS system equipped with an HP/Agilent Zorbax column (Eclipse XDB-C18, 2.1mm X 15cm). HPLC grade ultra pure water was used to prepare mobile phases. Solvent A: Aqueous 6 mm ammonium acetate. Solvent B: 95/5 Acetonitrile/water containing 6 mm ammonium acetate. The solvent gradient started with a constant 5%B for 5 minutes, then ramped from 5% B to 100% B in 25 minutes, followed by holding at 100%B for 5 minutes. The sample injection volume was 2 μl, and the mobile phase flow rate was 0.25 ml/min. The ionization method employed was electrospray, with m/z MS negative ion scanning range. 3.4 Results and Discussion Fluoride Concentration Measurement Accuracy and Interferences A typical calibration curve of a fluoride ISE is shown in Figure 3-4, where the fluoride concentration can be back calculated from the potential reading of the meter by the fitting equation of the calibration curve. Caution has been taken to ensure the accuracy of the fluoride concentration measurement by examining the effect of sample ph, ferrous ion concentration, and ferric ion concentration on measured readings. 77

103 y = x R 2 = mv Log [F-] / M Figure 3-4. A Typical Calibration Curve of Fluoride ISE a. Sample ph 1.0E-02 Expected value from calibration curve at [F - ] = 50 ppm [F-] / M 1.0E-03 [F - ] measured 1.0E PH Figure 3-5. Plot Showing Interference Check of Sample ph Values at [F - ]=50ppm 78

104 Sample ph value related interference can occur when ph > 7 (OH - will interfere the electrode response to fluoride), and when ph < 5 (the proton can complex a portion of fluoride in solution by forming the undissociated acid HF molecules and the (HF 2 ) -1 ions). 29 The control experiments for a sample with ph range of 1 ~ 7 showed that the interference from sample ph can be eliminated by the addition of TISAB II or TISAB III buffer solutions during the measurement, as seen in Figure 3-5. b. Ferrous Ion Concentration Control experiments on a sample spiked with ferrous ion concentration in the range of 50~500 ppm were measured to show that Fe 2+ does not infere with the measurement when TISAB II is used, as shown in Figure E-02 Expected value from calibration curve at [F - ] = 50 ppm [F-] / M 1.0E-03 [F - ] measured 1.0E [Fe2+] / mg/l Figure 3-6. Plot Showing Interference Check of the Presence of Fe 2+ Ions of Various Concentrations at [F - ]=50ppm 79

105 c. Ferric Ion Concentration Significant interference with the fluoride concentration measurement was found in the presence of ferric ions, when the concentration is greater than 0.05 M, as shown in Figure 3-7 to Figure This Fe 3+ interference could be eliminated by a serial sample dilution method, in which an accurate fluoride concentration was obtained by diluting the sample to the point (typically 100-fold dilution) where the concentration of Fe 3+ was below 25mM, and therefore does not interfere while keeping the fluoride concentration well above the detection limit of the electrode. Therefore, if an ISE is used to measure the concentration of fluoride, caution has to be taken to correct for the interference of Fe 3+ ions, which may have been the cause for several under-reported values and contradictions in the fuel cell literature. Expected value from calibration curve at [F - ] = 50 ppm 1.0E E E-03 [F-] / M 1.0E E-05 [F - ] measured 1.0E E [Fe3+] / M Figure 3-7. Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=50ppm 80

106 Expected value from calibration curve at [F - ] = 5 ppm 1.0E E E-04 [F-] / M 1.0E E-06 [F - ] measured 1.0E E [Fe3+] / M Figure 3-8. Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=5ppm Expected value from calibration curve at [F - ] = 0.5 ppm 1.0E E E-05 [F-] / M 1.0E E-07 [F - ] measured 1.0E E [Fe3+] / M Figure 3-9. Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=0.5ppm 81

107 Expected value from calibration curve at [F - ] = 0.05 ppm 1.0E-05 [F-] / M [F - ] measured 1.0E [Fe3+] / M Figure Interference Check of the Presence of Fe 3+ Ions of Various Concentrations at [F - ]=0.05ppm 82

108 3.4.2 Fluoride Generation from Fenton s Degradation Test MC1 O MC2 O MC3 O HO C CF O CF 2 CF 2 CF 3 HO C CF O CF 2 CF 2 SO 3 H 2 HO C CF 2 CF 2 CF 2 SO 3 H CF 3 CF 3 MC4 MC5 MC6 O HO C CF 2 CF 6 3 F 3 C CF 2 CF 2 CF 2 SO 3 H F 3 C CF 2 CF 2 H 6 MC7 MC8 CF 3 CF 2 O CF 2 CF 2 SO 3 H 2 F 3 C CF 2 O CF 2 CF O CF 2 SO 3 H 2 CF 3 F - released / % of total F atoms in MC MC1 MC2 MC3 MC4 MC5 MC6 MC7 MC Accumulated Aging Time / Hrs Figure Fluoride Evolution from MCs as a Function of Mild Degradation Test Time (MC structures shown in top portion) 83

109 F - released / % of total F atoms in MC MC5 MC6 MC7 MC Accumulated Aging Time / Hrs Figure Enlarged Plot Showing Low Concentration Range of Figure 3-11 The fluoride evolution from MCs treated with Fenton s reagents under mild degradation condition is plotted in Figure 3-11 and Figure The fluoride concentration is presented as the atomic percentage ratio of fluoride released relative to the total fluorine atoms from each model compound. MCs containing carboxylic acid groups showed significantly higher rates of fluoride generation than those without that functional group. The fluoride evolution of MCs without carboxylic acid groups, MC5 (four-carbon perfluorinated sulfonic acid molecule), MC6 (eight-carbon fluorocarbon molecule containing one hydrogen atom), MC7 (perfluorinated sulfonic acid molecule containing one linear ether linkage) and MC8 (perfluorinated sulfonic acid molecule containing both linear and branched ether linkages), shows the order: MC5 > MC7 ~ MC8 > MC6. The lowest fluoride generation rate exhibited by MC6 is expected due to 84

110 its low solubility in aqueous solution (phase separation was observed). The relatively higher fluoride release rate observed for MC5 may be due to the fact that MC5 contains fewer fluorine atoms (shorter chain than MC7 and MC8), which may skew the fluoride release calculation values used in the plot. It is important to note that even for these least reactive model compounds (MC6, MC7, and MC8), non-zero fluoride generation rates were observed. This result implies that fluoride release pathways from PFSAs other than the carboxylic acid end group degradation exist and are kinetically competitive (a comparison of the slopes of MC4 and MC8 in terms of fluoride generation/degradation rates suggests a factor of 500 difference between carboxylate and non-carboxylate PFSA analogs). This point is relevant since MC8 is structurally a close analogue of Nafion, while MC7 is the analogue of other popular commercial PFSAs (non-branched PTFE and perfluorovinylether sulfonic acid copolymers) manufactured by 3M and Dow. Carboxylic acid-containing MCs, MC1 (branched perfluorinated ether with one terminal carboxylic acid), MC2 (branched perfluorinated ether with both terminal carboxylic acid and sulfonic acid), MC3 (four-carbon fluorocarbon with terminal carboxylic acid and sulfonic acid), and MC4 (eight-carbon fluorocarbon with one terminal carboxylic acid), exhibit the following fluoride generation rate order: MC3 > MC2 ~ MC4 > MC1. The overall trend is monotonic and relatively linear, accounting for as much as 2% of the total fluorine content of the starting materials over 120 hours incubation time. The fluoride generation rate of such -COOH containing MCs appeared to be at least one order of magnitude higher than MCs without -COOH groups, as shown in Figure Shorter chain MC3 again shows higher fluoride release ratio. Control experiments in the absence 85

111 of hydrogen peroxide, with or without Fe 2+ catalyst, failed to generate detectable concentration of fluoride with the MC s Accumulated F- detected / % of total F atoms in MC MC-1 MC-2 MC-4 MC-5 MC-6 MC-7 MC Accumulated treated time / hrs Figure Fluoride Evolution from MCs as a Function of Harsh Degradation Test Time For the harsh degradation test, the fluoride evolution from the MCs is plotted in Figure The results obtained under the two sets of Fenton s conditions are similar, but not identical. The harsh Fenton s conditions literally do not result in greater model compound decomposition than was observed under mild conditions. It may be the case that under the harsh conditions, much of the generated radical species are lost to self- 21, 28 recombination or other non-productive reactions. In each case, the model compounds possessing carboxylic acid groups degrade most rapidly Fluoride Generation from UV Photolysis Degradation Test 86

112 F - released / % of total F atoms in MC MC1 MC2 MC3 MC4 MC5 MC6 MC7 MC8 Figure Fluoride Evolution from MCs After 1 hour UV Exposure (Unfilled columns represent fluoride generated from UV exposure without H 2 O 2 added into the solution; filled columns show the fluoride generated from UV exposure with the presence of H 2 O 2 ) F - released / % of total F atoms in MC MC4 MC5 MC6 MC7 MC8 Figure Enlarged Plot Showing Low Concentration Range of Figure

113 MC aqueous solutions were exposed to UV irradiation to degrade at room temperature, and the fluoride generation was measured. UV irradiation was carried out on MC solutions with and without H 2 O 2, and the resultant fluoride generations are shown in Figure 3-14 and Figure For MCs containing carboxylic acid groups (MC1 - MC4), UV irradiations led to fluoride generation even without the presence of added H 2 O 2, probably due to UV-facilitated decarboxylations that may further trigger structural changes Higher concentrations of fluorides were generated when H 2 O 2 was added to the MC solutions subject to UV irradiation. MC5, MC6, and MC7 and MC8, were found to generate much less fluoride than the -COOH containing MCs, i.e. MC1, MC2, MC3 and MC4, even when UV irradiation was carried with H 2 O 2 added. Overall, similar degradation trends were observed between the Fenton s degradation test and UV degradation tests. For the lower concentration range shown in Figure 3-15, the presence of H 2 O 2 upon UV exposure does not lead to greater release of fluoride. With only UV exposure, generation of fluoride from MCs that do not contain labile -COOH group may be explained by the direct UV degradation of such fluoroethers structures that potentially can lead to subsequent elimination of fluoride ions Degradation Product Analysis by LC-MS a. Degradation Product Analysis of MC4 MC4 O HO C CF 2 6 CF 3 While the chain end unzipping mechanism is widely accepted in the literature, the structures of the resultant degradation products have not been confirmed. Chain end 88

114 unzipping mechanism products were verified by the degradation product analysis of MC4, a molecule that contains only a carboxylic acid group on a linear perfluorinated linear aliphatic chain (R f -COOH). After the Fenton s test, LC-MS analyses were carried out on the reaction mixture of MC4, shown in Figure 3-16 and Figure Full Chromatograph Relative Abundance MS 6 MS 5 MS 1 MS 2 MS 3 MS 4 Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Traces of Degraded MC4 Reaction Product Mixture In Figure 3-16, the full chromatograph is shown on the top where it is obvious that there are a series of peaks at different elution times. The full chromatography trace can be extracted by a specific ion molecular weight to show the relative ion intensity of that specific ion at different elution times. The resulting chromatography trace is called an extracted chromatograph hereafter. This has been carried out on the full 89

115 chromatograph of the MC4 reaction mixture. Six ion molecular weight ranges were used to extract the full chromatograph: Da, Da, Da, Da, Da, Da, and the resultant extracted chromatographs are shown in Figure The full chromatograph is accurately deconvoluted into six peaks at six elution times, and the MS spectrum of each peak at these six elution times were recorded and shown in Figure 3-17 with the designation of MS-1 to MS-6. C f 7-COO - MS 1 C f 6-COO - MS 2 Relative Abundance C f 5-COO - C f 4-COO - MS 3 MS 4 C f 3-COO - MS 5 C f 2-COO - MS 6 m/z Figure MS Spectra of LC Trace of a Degraded MC4 Reaction Product Mixture at Various Elution Times 90

116 MS-1 in Figure 3-17 showes that the peak at this elution time is the intact MC4, where the 413 Da ion is assigned to be the parent ion by losing a proton and form a negative anion, and the 827 Da ion is assigned to an adduct ion of a molecule of MC4 and a parent ion of MC4. These two ions should serve as good signature ions for solving the MS spectrum with the structure similar to MC4. A close examination of the rest of MS spectra (MS-2 to MS-6) reveals that at those five earlier elution times, the molecular weights of various parent ions differ by 50 Da, and the molecular weights of adduct ions differ by 100 Da, and therefore MS-2 is assigned to be the C f 6-COOH because the difference of this molecule and MC4 (C f 7-COOH) is a -CF 2 - unit (50 Da). The difference of 100 Da (a net decrease of two -CF 2 - units) for the adduct ions and a shorter elution time (a shorter elution time is expected for a molecule with shorter hydrophobic tail for the reverse phase C18 column used) both confirm this assignment. Similar analogues have also been observed for MS-3 to MS-6. The stepwise loss of CF 2 units, reforming another terminal carboxylic acid groups, is completely consistent with the unzipping degradation mechanism for degradation of molecules like MC4 under the testing conditions. This LC-MS analysis of MC4 clearly shows the evolution of the perfluorinated eight-carbon acid into its seven through three carbon analogs, as shown in Scheme 3-1. Another degradation product, CO 2 gas, was trapped by the method described in the experimental section. It was observed that the color of the test tube turned from pink to colorless within 1 hour upon the addition of H 2 O 2 into the flask. 19 F NMR indicated that there were no fluorinated organic compounds in the trapping test tube to cause the 91

117 observed color change, and the detection of CO 2 again supports the unzipping mechanism. C f 7-COOH C f 6-COOH C f 5-COOH HO. Radical C f 2-COOH C f 3-COOH C f 4-COOH Scheme 3-1. Diagram Showing the Unzipping Degradation From -COOH Chain End b. Degradation Product Analysis of MC1 MC1 O HO C CF O CF 2 CF 2 CF 3 CF 3 From fluoride evolution data and the degradation product analysis of MC4, it is clear that the terminal carboxylic acid groups are very reactive toward hydroxyl radical attack, and the products of degradation can be readily explained using the chain end unzipping mechanism (Scheme 3-1). The fate of more complex structures, containing the ether links and branched structures common in PFSAs also need to be examined. MC1 and MC2 are suitable for such a comparison, since both contain carboxylic acid, ether, and tertiary carbons. In addition, the degradation of MC1 and MC2 will potentially reveal the subsequent degradation fate of the side chains of Nafion, should they be cleaved from the polymer main chain. The LC-MS product analysis results of MC1 are shown in Figure 3-18 to Figure LC-MS analysis of degraded MC1 identified intact starting material MC1, trifluoroacetic acid (TFA) and pentafluoropropionic acid (PFPA). 92

118 The parent ions and other corresponding adduct ions for these products are tabulated in Table 3-3 (the symbol * is used to indicate the formation of adduct ions). Table 3-3. Tabulated Parent and Adduct Ions of TFA and PFPA TFA MW of Ions PFPA MW of Ions CF 3 -COOH 114 CF 3 -CF 2 -COOH 164 (TFA-H + ) 113 (PFPA-H + ) 163 TFA * (TFA-H + ) 227 PFPA * (PFPA-H + ) 327 (TFA-H + +Na + ) * (TFA-H + ) 249 (PFPA-H + +Na + ) * (PFPA-H + ) 349 2(TFA-H + +Na + ) * (TFA-H + ) 385 2(PFPA-H + +Na + ) * (PFPA-H + ) 535 3(TFA-H + +Na + ) * (TFA-H + ) 521 3(PFPA-H + +Na + ) * (PFPA-H + ) 721 4(TFA-H + +Na + ) * (TFA-H + ) 657 MS 1 Full Chromatograph Extracted Chromatograph MW: Da Relative Abundance Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of Degraded MC1 Reaction Product Mixture (top) and LC traces (relative abundance) of Selective Ions from TFA (bottom five traces) 93

119 Relative Abundance MS 1 Relative Abundance Time (Minute) MS 1 RT=1.9 Min m/z Figure MC1 LC Trace (top) and Corresponding MS Spectrum (bottom ) at RT=1.9 min 94

120 Full Chromatograph Relative Abundance Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of Degraded MC1 Reaction Product Mixture (top) and LC traces (relative abundance) of Selective Ions from PFPA (bottom four traces) 95

121 Relative Abundance MS 1 MS 2 Relative Abundance Relative Abundance Time (Minute) m/z m/z MS 1 RT=2.4 Min MS 2 RT=2.9 Min Figure MC1 LC Trace (top) and Corresponding MS Spectra (bottom ) at RT=1.9 min and 2.9 min From the full LC trace of the MC1 reaction mixture (see the top portion of Figure 3-18), the large peak at retention time ca. 9.0 min is the intact MC1 reagent. TFA is observed at 1.9 min, and PFPA peak is a broader peak from mins. It should be mentioned that the other peaks at 1.7 minute, 2.1 minute, and 6.8 minute are identified to be the contaminants present in the LC-MS system background, and thus are excluded from the degradation products analysis. Mobile phase empty checks in between data 96

122 acquisition found similar ion patterns from those peaks, and the expected TFA and PFTA peaks were found from at least 3 independently degraded samples. The chromatographs extracted by expected ions from TFA are shown in Figure 3-18, in which the peaks of different ion molecular weights appear at very similar elution times. A similar trend is observed for Figure 3-20, although the peaks are considerably broader than those of TFA. MS spectra shown in Figure 3-19 and Figure 3-21 clearly exhibit the expected ions from TFA and PFTA tabulated in Table 3-3. O HO. HO C CF O CF 2 CF 2 CF 3 CF 3. CO 2 + H 2 O + F 3 C CF O CF 2 CF 2 CF 3 F 3 C. OH CF O CF 2 CF 2 CF 3 F 3 C SET.. CF O CF 2 CF 2 CF 3 OH - HF H+ F 3 C C O CF 2 CF 2 CF 3 F 3 C CF O CF 2 CF 2 CF 3 O H hydrolysis F 3 C C OH + HO CF 2 CF 2 CF 3 - HF F C CF 2 CF 3 - HF HO C CF 2 CF 3 O O O Scheme 3-2. Proposed Degradation Mechanism for MC1 Along with the result of degradation product determination for MC1 by 19 F NMR (details and data will be presented in Section 3.4.5), the degradation mechanism of MC1 97

123 is proposed as shown in Scheme 3-2. In this mechanism, radical abstraction of a carboxylic acid hydrogen atom initiates decarboxylation of the model compound. The decarboxylated radical intermediate can then undergo single electron transfer (SET) or atom abstraction to produce the observed ether. Capture of a second hydroxyl radical would be expected to produce perfluorinated propyl acetate; we have independently demonstrated that such perfluorinated alkyl acetates will rapidly hydrolyze to the perfluorinated acetic and propionic acids which were identified as MC1 reaction products. Alternatively, the formation of CF 3 -COOH and HO-CF 2 -CF 2 -CF 3 can be resulted from the direct acid-catalyzed hydrolysis of the postulated hemi-acetal intermediate (the structure which leads to the formation of the perfluorinated ester via elimination of HF, as shown in Scheme 3-2). c. Degradation Product Analysis of MC8 MC8 F 3 C CF 2 O CF 2 CF O CF 2 SO 3 H 2 CF 3 MC8 was chosen as a small molecule analog to the Nafion polymer itself, substituting a perfluoroethyl group for the polymer backbone. LC-MS analysis of MC8 degradation products identified four significant species in addition to the starting material. LC traces and MS spectra (the expected parent and adduct ions of products 1-4 as tabulated in Table 3-4 are highlighted with circles) are presented in Figure 3-22 and Figure The major product appeared to be a fluorinated carboxylic acid compound (MC8, Product-1), which could be expected to result from cleavage of the ether group 98

124 near the methyl end. Three other structures (Products-2 to 4) were also identified as being degradation products from MC8 or MC8 degradation product-1. One of these reaction products (product-3) corresponds to a MC1 degradation product as well. What can be clearly stated is that all of the major degradation products of MC8 involve the cleavage of the ether link, analogous to loss of side chain of Nafion. No evidence for loss of sulfonic acid groups was observed as well. Table 3-4. Proposed Degradation Products of MC8 with Expect Parent and Adduct Ions MC8, Product-1 (MC8-PRDT1) O HO C CF O CF 2 CF 2 SO 3 H MC8, Product-2 (MC8-PRDT2) CF 3 CF O CF 2 CF 2 SO 3 H CF 3 CF 3 MC8-PRDT1 m/z MC8-PRDT2 m/z (MC8-PRDT1-H + ) 341 (MC8-PRDT2-H + ) 365 (MC8-PRDT1-H + )-CO MC8-PRDT2 * (MC8-PRDT2-H + ) 731 (MC8-PRDT1-2H + ) 170 (MC8-PRDT1-2H + )-CO 2 F 277 MC8, Product-3 (MC8-PRDT3) CF 3 CH O CF 2 CF 2 SO 3 H MC8, Product-4 (MC8-PRDT4) H CF 2 O CF 2 CF 2 SO 3 H F MC8-PRDT3 m/z MC8-PRDT4 m/z (MC8-PRDT3-H + ) 297 (MC8-PRDT4-H + ) 247 MC8-PRDT3 * (MC8-PRDT3-H + ) 595 (MC8-PRDT4-H + +Na + ) * (MC8-PRDT4-H + ) 517 (MC8-PRDT3-H + +Na + ) * (MC8-PRDT3-H + )

125 Full Chromatograph Relative Abundance MS 1 MS 2 MS 3 Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da MS 4 Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of Degraded MC8 Reaction Product Mixture (top), LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom four traces) 100

126 MS 1 MS 2 Relative Abundance m/z MS 3 MS 4 Relative Abundance Relative Abundance Relative Abundance m/z m/z m/z Figure MS Spectra Marked as MS-1 to MS-4 in Figure 3-22: (top-left) RT=3.9 min, (top-right) RT=20.1 min, (bottom-left) RT=19.3 min, (bottom-right) RT=18.2 min d. Degradation Product Analysis of MC7 MC7 CF 3 CF 2 O CF 2 CF 2 SO 3 H 2 MC7 was selected as structural analogue to PFSAs with one ether linkage on the side chain, such as 3M and Dow products. LC-MS was also carried out to analyze the 101

127 degradation products. The identified products and impurity compound present are listed in Table 3-5 with expected ions. Again, the major degradation products of this MC require ether cleavage, similar to the loss of side chains in polymer itself. It is clear that degradation mechanisms other than the chain end unzipping mechanism are possible in MCs which are structurally analogous to PMSEs. Specifically for MC7 and MC8, structures without carboxylic acid groups also underwent degradation, likely through an ether cleavage reaction. Experiment design and data for probing the detailed cleavage mechanism will be discussed in the next section, and the degradation of polymers for comparison and correlation will be discussed in the next chapter. Scheme 3-3 is presented to summarize the key products identified before the radical intermediate trapping experiment in next section. Table 3-5. Proposed Degradation Products of MC7 Based on LC-MS Analysis Designation, Molecular Weight Structure MC7, 416 Da F 3 C CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H Impurity, 398 Da HF 2 C CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H Product A, 276 Da HOOC CF 2 CF 2 CF 2 SO 3 H Product B (TFA), 114 Da F 3 C COOH Product C, 392 Da HOOC CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H Product D, 300 Da CF 3 CF 2 CF 2 CF 2 SO 3 H 102

128 MC7, Product D MC7, Product B F 3 C COOH CF 3 CF 2 CF 2 CF 2 SO 3 H HOOC MC7, Product A CF 2 SO 3 H CF 2 CF 2 CF 3 CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H MC7 HF 2 C CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H Impurity MC7, Product C HOOC CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H O MC8, Product-1 (MC8-PRDT1) MC8, Product-2 (MC8-PRDT2) HO C CF O CF 2 CF 2 SO 3 H CF 3 CF O CF 2 CF 2 SO 3 H CF 3 CF 3 F 3 C CF 2 O CF 2 CF O CF 2 SO 3 H CF 3 2 MC8 CF 3 CH O CF 2 CF 2 SO 3 H H CF 2 O CF 2 CF 2 SO 3 H F MC8, Product-3 (MC8-PRDT3) MC8, Product-4 (MC8-PRDT4) Scheme 3-3. Schematic Summary of Degradation Product Identified from MC7 and MC8 103

129 3.4.5 Degradation Intermediate Trapping Experiments for MC7 and MC8 MC7 CF3 CF 2 O CF 2 CF 2 SO 3 H 2 MC8 F 3 C CF 2 O CF 2 CF O CF 2 SO 3 H 2 CF 3 The degradation product analysis of MC7 and MC8 suggests that ether cleavage reactions lead to the formation of the various degradation products detected. There are few literature reports on such ether cleavage reactions under Fenton degradation conditions. The details of the reaction mechanism are of great importance for both general fluorine chemistry and current research efforts with the target of achieving the durability needed by further development and mitigation of structure of current polymer candidates. The knowledge of chemical degradation pathways of perfluoroethers in the presence of hydroxyl radicals reported in this study may also be beneficial to the postservice handling of such robust and therefore environmentally challenging compounds in the advanced oxidation processes (AOP), where chemical treatment procedures are designed to degrade organic and inorganic compounds by hydroxyl radicals. Fluoroether functionality represents an important family of structural units not only for current commercial PFSA ionomers in this research, but also for many high performance polymers such as fluoroelastomers, 2 TFE-co-(perfluoro vinyl ether) thermoplastic copolymers, 2, 5 and poly(perfluoro ether) as excellent lubricant of choice for magnetic data storage systems and extraterrestrial applications Additionally, there is an increasing popularity of fluoroether containing polymers to serve as protecting 104

130 films for various components, such as photomasks, under high energy lithography 31, 33 processes (e.g., 157 nm photons to achieve detail resolution lower than 45 nm). a. Brief Review of Degradation Studies of Fluoroethers O CF 2 F O C F O CF 2 O O CF 2 O F C O + CF 3 O Lewis Acid Figure Ether Cleavage Decomposition Mechanism of PFPEs by Lewis Acid Sites 37 Although various types of perfluoropolyethers have intrinsically high chemical stability due to the strength of C-F bond after fluorination, chemical degradation of these compounds in various environments have been observed. The chemical degradation can be the result of thermal degradation, tribomechanical shearing, electron beams and irradiation, or by exposure to surfaces that contain Lewis acid catalysts Thermal stability studies have revealed thermal decomposition temperatures for perfluoropolyethers (PFPE) of around o C. 45 In the typical pyrolysis degradation study, the degradation products of a series of PFPEs identified by gas chromatograph coupled with mass spectrometer (GC-MS) include: shortened PFPEs resulted from C-O and C-C bond fissions, and various unsaturated perfluorinated structures with double bonds. The degradation was proposed to be a radical process: radicals are formed at elevated temperatures, and the generated radicals can undergo recombination or 105

131 elimination of F.. or CF 3 to form the observed fluorinated olefins. Alumina surfaces (Al 2 O 3, a Lewis acid), which are present and in direct contact with the fluoroether lubricants in various magnetic data storage devices such as hard drives, can effectively catalyze and lower their thermal decomposition temperatures to less than 200 o C. The degradation observed on such Lewis acid catalyst-rich surfaces was assumed to be caused primarily by an ether cleavage process. The mechanism describing such a mechanism is shown in Figure 3-24, 37 where a difluoroacetal group of PFPE can form a bidentate form interaction with a Lewis acid site through the lone pair electron on ether oxygen atoms. 48 The formation of such bidentate structure was postulated to be effective in lowering the bond strength of C-O bond. As a result, C-O scission can occur through a disproportionation reaction by intramolecular 1,3-fluorine transfer to form the products terminated with acyl fluoride and trifluoromethyl groups. This degradation process was proposed to be occur more readily at acetal moieties, such as -O-CF 2 -O- or -O-CF(CF 3 )- O-, than at longer polyether segments, such as -O-CF 2 -CF 2 -CF 2 -O-. Due to the stringent chemical stability requirement for applications in high energy lithography processes and space applications where components are exposed to ionizing radiation, extensive degradation studies under electron beam, UV, and γ irradiation exposures have been carried out. The proposed degradation mechanisms in these studies are summarized in Figure and Figure Radical intermediate formations are key steps after the preferential scission of C-O and C-C bonds in both mechanisms, due to the relative lower bonding energy of these two bonds in perfluorinated compounds: C- O, c.a. 440 KJ/Mol; C-C, c.a. 400~415 KJ/Mol; and C-F, 520 ~540 KJ/Mol. 7, 8, 45, 49 The energetic parameters of the C-O bond scission via radical such processes have been 106

132 calculated by performing ab initio calculations on linear PFPEs with the typical repeat unit structure of -(CF 2 ) n -O- (n=1, 2, or 3): bond scission energy, c.a. 300 KJ/Mol; activation energy without Lewis acid catalyst, 300~400 KJ/Mol; and lowered activation energy with Lewis acid catalyst, c.a. 210 KJ/Mol. 49 OCF 3 e - CF 2 CF 2 CF CF 2 CF 2. CF 2 CF 2 CF CF 2 CF 2 +. OCF 3 e - Ȯ F. + COF 2 CF 2 CF 2 CF CF 2 CF 2 +. CF3 O CF 3 R f CF 2 CF 2 CF +. CF 2 CF 2 Decomposition R O CF CF 2 O R' R O CF CF 2 O R' + CF 3 CF 3 Radical Cation Formation Figure (top) Possible Electron Beam Induced Decomposition of a Branched Fluoroether, 41 and (bottom) the Formation and Dissociation of a Molecular Cation Intermediate of Perfluoroether Structure Studied by Time of Flight-Secondary Ion Mass Spectrometry (TOF-SIMS) in the Positive Ion Mode

133 Figure Possible γ Irradiation Induced Decomposition of a Branched Fluoroether 32 b. Trapping Experiment Results and Proposed Mechanism As described in the experiment section, trapping experiments were carried out by removing aliquots of solution from the degradation test medium, followed by quenching aqueous solutions of a stable water soluble radical, 4-hydroxy-TEMPO, at different reaction times. Figure 3-27 and Figure 3-28 list all postulated adduct structures and molecular weights of 4-hydroxy-TEMPO and the expected radical intermediates, should the degradation of ether bond proceed via (or involve) radical processes. 108

134 MC7 CF 3 CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H MC7-Impurity HF 2 C CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H MC7-ADT-1 CF 3 CF 2 O N OH (291) MC7-ADT-2 HO N O O CF 2 CF 2 CF 2 CF 2 SO 3 H (469) MC7-ADT-3 CF 3 CF 2 O O N OH (307) MC7-ADT-4 HO N O CF 2 CF 2 CF 2 CF 2 SO 3 H (453) MC7-ADT-5 HO N O CF 2 CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H (569) MC7-ADT-6 HO N O CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H (519) Figure Postulated Structure of Various Possible Radical Intermediates from MC7 Trapped by 4-hydroxy-TEMPO Radical (Molecular Weight is Shown in Parenthesis) 109

135 MC8 CF 3 CF 2 O CF 2 CF O CF 2 CF 2 SO 3 H CF3 MC8-ADT-1 CF 3 CF 2 O N OH (291) MC8-ADT-2 HO N O O CF 2 CF O CF 2 CF 2 SO 3 H (535) CF 3 MC8-ADT-3 CF 3 CF 2 O O N OH (307) MC8-ADT-4 HO N O CF 2 CF O CF 2 CF 2 SO 3 H (519) CF 3 MC8-ADT-5 CF 3 CF 2 O CF 2 CF O N OH (457) CF3 MC8-ADT-6 HO N O O CF 2 CF 2 SO 3 H (369) MC8-ADT-7 CF 3 CF 2 O CF 2 CF O O N OH (473) CF3 MC8-ADT-8 HO N O CF 2 CF 2 SO 3 H (353) Figure Postulated Structure of Various Possible Radical Intermediates from MC8 Trapped by 4-hydroxy-TEMPO Radical (Molecular Weight is Shown in Parenthesis) 110

136 Table 3-6. Tabulated Results of 4-hydroxy-TEMPO-trapped Radical Adduct from Various MC7 Degradation Aliquots Parent Ion MW (Da) MC7-Original 1-40 Min 1-90 Min 1-3 Hr Hr 2-4 Hr 4-30min 4-2 Hr 4-4 Hr 4-24 Hr MC7-ADT-1 MC7-ADT-2 MC7-ADT-3 MC7-ADT-4 MC7-ADT-5 MC7-ADT , (3 Min) +, (20 Min) +, (22 Min) +, (22 Min) - - (Low) (Low) +, (2.6 Min) +, (20 Min) , (2.8 Min) +, (20 Min) - - +, (22 Min) (Low) +, (21 Min) (Medium) +, (2.7 Min) +, (20 Min) - +, (22 Min) - +, (21 Min) (Low) or (Too Low) (Medium) +, (5.2 Min) +, (20 Min) - - +, (18 Min) +, (21 Min) (Low) (Medium) +, (2.6 Min) +, (20 Min) - - +, (22 Min) +, (21 Min) (Low) (Medium) +, (4.7 Min) +, (20 Min) - - +, (22 Min) +, (21 Min) (Low) (Medium) +, (4.2 Min) +, (20 Min) - - +, (22 Min) +, (21 Min) (Low) (Low) (Medium) +, (5.8 Min) - +, (22 Min) - - +, (21 Min) (Low) (Medium) The major trapped adducts identified by LC-MS are tabulated in Table 3-6 (MC7) and Table 3-7 (MC-8) for twenty samples analyzed during trapping experiments. Designations of samples in these two tables are as follows: degradation test cycle-time of the removal of aliquot since the initiation of the degradation test cycle, for example, 1-40Min is used for the sample removed at 40 minutes after the initiation of 1 st degradation test cycle of the specific MC. The + and - are used to indicate the presence or absence of the postulated adduct with the parent ion molecular weight listed on the top of each column. The elution time of the detected adduct structure is listed in a parenthesis after the + or - signs. Conclusions that can be reached from these data 111

137 are: 1) the trapping experiments are quite effective in terms of reproducibility, most adducts have well reproduced mass spectra and elution times; 2) there are a number of specific adducts (MC7-ADT-1, MC7-ADT-2, and MC8-ADT-5) are observed reproducibly. Table 3-7. Tabulated Results of 4-hydroxy-TEMPO-trapped Radical Adduct from Various MC8 Degradation Aliquots Parent Ion MW (Da) MC8-Original 1-40 Min 1-90 Min 1-3 Hr Hr 2-2 Hr 3-2 Hr 4-30 Min 4-90 Min 4-4 Hr 4-24 Hr MC8-ADT-1 MC8-ADT-2 MC8-ADT-3 MC8-ADT-4 MC8-ADT-5 MC8-ADT-6 MC8-ADT-7 MC8-ADT , (24 Min) - - +, (16.6 Min) - or (Too (Low) , (23 Min) (Low) , (23 Min) +, (16 Min) +, (23 Min) (Low) (Low) - - +, (19 Min) +, (23 Min) +, (16 Min) - (Low) (Low) - +, (23 Min) - +, (23 Min) +, (23 Min) +, (23 Min) (Low) (Low) (Low) (Low, slight , (23 Min) +, (16 Min) +, (23 Min) (Low) (Low) , (16 Min) - (High) - +, (23 Min) +, (23 Min) +, (23 Min) +, (16 Min) +, (23 Min) (Low) (High) (Low) (Low) - +, (24 Min) - +, (24 Min) - +, (5 Min) (High) (Low) +, (23 Min) (Low) +, (22 Min) (Low) (Low) +, (22 Min) (Low) +, (16 Min) (High) +, (16 Min) (High) +, (21 Min) (High) +, (23 Min) - (Low) - +, (23 Min) (Low) - +, (23 Min) , (23 Min) +, (23 Min) (Low, slight +, (24 Min) - (Low) +, (23 Min) +, (22 Min) (Low) (High) - - +, (23 Min) (Medium) - 112

138 Relative Abundance MS 1 MC7-ADT-1 Full Chromatograph Extracted Chromatograph MW: Da MC7-ADT-2 MS 2 Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC7 Solution (MC7-4-2Hr) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom two traces) Relative Abundance CF 3 CF 2 O N OH m/z MC7-ADT-1 MS 1 RT=4.7 Min Relative Abundance HO N O O CF 2 CF 2 CF 2 CF 2 SO 3 H m/z MC7-ADT-2 MS 2 RT=20.4 Min Figure MS Spectra Marked as MS-1 and MS-2 in Figure 3-29: (top) RT=4.7 min, (bottom) RT=20.4 min 113

139 Relative Abundance MS 1, MC7-ADT-5 Full Chromatograph Extracted Chromatograph MW: Da MS 2, MC7-ADT-6 Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC7 Solution (MC7-4-2Hr) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom two traces) Relative Abundance HO N O MC7-ADT-5 CF 2 CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H MS 1 RT=22.4 Min m/z Relative Abundance MC7-ADT-6 HO N O CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H MS 2 RT=21.3 Min m/z Figure MS Spectra Marked as MS-1 and MS-2 in Figure 3-31: (top) RT=22.4 min, (bottom) RT=21.3 min 114

140 Typical LC-MS analyses for the MC7 trapping experiments is shown in Figure 3-29 to Figure Although readily detectable by LC-MS, the concentrations of the detected adducts are quite low compared with the highest peak in LC trace for the intact MCs, and the magnitude of the ion abundance is comparable to the degradation products described in the earlier section of this chapter. The LC trace of adducts MC7-ADT-1 and MC7-ADT-2 are shown in Figure 3-29, and their structures are shown in the corresponding MS spectra in Figure 3-30, and the corresponding MS spectra are shown in Figure 3-30, with the chemical structures shown and the parent ions circled. Note that the other predominant ions appeared in the MS spectra have been identified to be either the fragmented parent ions or background ions of the LC-MS system. MC7-ADT-5 and MC7-ADT-6 are shown in Figure 3-31, and the corresponding MS spectra are shown in Figure The LC peaks for MC7-ADT-5 (which shows up under the broad LC peak for intact MC7, possibly due to similar interaction with the column imposed by the structural resemblance to MC7) and MC7-ADT-6 (which shows up at earlier elution than MC7-ADT-5) appear to be noisier than those for MC7-ADT-1 and MC7-ADT-2. Similarly, a typical LC-MS analysis for the MC8 trapping experiments is shown in Figure 3-33 to Figure 3-36 for the MC8 adducts: MC8-ADT-1, MC8-ADT-5, and MC8-ADT-6 (shown in Figure 3-33 and with MS spectra in Figure 3-34); MC8-ADT-3 and MC8-ADT-8 (shown in Figure 3-35 and with MS spectra in Figure 3-36). 115

141 Full Chromatograph Relative Abundance MS 1 MC8-ADT-5 MC8-ADT-1 MS 2 Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da MC8-ADT-6 MS 3 Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC8 Solution (MC8-4-24Hr) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom three traces) 116

142 Relative Abundance Relative Abundance MC8-ADT-1 CF 3 CF 2 O N OH m/z MC8-ADT-5 CF 3 CF 2 O CF 2 CF O CF 3 m/z N OH MS 1 RT=5.0 Min MS 2 RT=16.9 Min Relative Abundance HO N O O CF 2 CF 2 SO 3 H MC8-ADT-6 MS 3 RT=21.3 Min m/z Figure MS Spectra Marked as MS-1 to MS-3 in Figure 3-33: (top) RT=22.4 min, (center) RT=16.9 min, and (bottom) RT=21.3 min 117

143 Full Chromatograph Relative Abundance MC8-ADT-3 MS 1 Extracted Chromatograph MW: Da MC8-ADT-8 MS 2 Extracted Chromatograph MW: Da Time (Minute) Figure LC Chromatographic Trace of 4-hydroxy-TEMPO-trapped Aliquot Solution of a Degraded MC8 Solution (MC8-4-90Min) (top), and LC Traces (Relative Abundance) of Selective Ions from Proposed Products (bottom two traces) 118

144 Relative Abundance CF 3 CF 2 O O N OH MC8-ADT-3 MS 1 RT=23.2 Min m/z Relative Abundance HO N O CF 2 CF 2 SO 3 H MC8-ADT-8 MS 2 RT=22.6 Min m/z Figure MS Spectra Marked as MS-1 and MS-2 in Figure 3-35: (top) RT=23.2 min, (center) RT=16.9 min, and (bottom) RT=22.6 min 119

145 Figure 3-37 summarizes the key products identified before the discussion of proposed mechanisms to explain their formation in the following section. MC7 CF 3 CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H CF 3 CF 2.. O CF 2 CF 2 CF 2 CF 2 SO 3 H Abundant.. Abundant CF 3 CF 2 O CF 2 CF 2 CF 2 CF 2 SO 3 H MC8 CF 3 CF 2 O CF 2 CF O CF 2 CF 2 SO 3 H.. CF3 CF 3 CF 2 O CF 2 CF O CF 2 CF 2 SO 3 H CF 3 CF 3 CF 2 O.. CF 2 CF O CF 2 CF 2 SO 3 H CF 3 CF 3 CF 3 CF 2 O CF 2 CF Abundant.. CF 3 O CF 2 CF 2 SO 3 H. CF 2 O CF 2 CF O. CF 2 CF 2 SO 3 H CF 3 Figure Generalization of the Trapping Experiment Results for MC7 and MC8 as Shown in Table 3-6 and Table

146 Degradation Route-1a F F CF 2 C O C CF 2 F F - e. - OH F F 1. + CF 2 C O C CF OH F F 2 CF F C... F.. CF 2 C F.. O C CF 2 F F 3 F 8 HO F - CF 2 CF 3 - OH CF 2 CF 2 OH HF O C CF 2 O - HF F C CF H 2 2 O 7 6 -HF. OH F H O O C CF - O 2 2. CF 2 CF 2. F 9 10 OH. OH HO CF 2 CF 2 11 Scheme 3-4. Proposed Degradation Route-1a for Ether Cleavage of - CF 2 -CF 2 -O-CF 2 - CF 2 - Structure 121

147 Degradation Route-1b F F CF 2 C O C CF 2 F F - e. - OH F F 1. + CF 2 C O C CF OH F F 2 CF 2 F C. Destabilized O C CF 2 12 F F 13 F Scheme 3-5. Proposed Degradation Route-1b for Ether Cleavage of - CF 2 -CF 2 -O-CF 2 - CF 2 - Structure 122

148 Degradation Route-2a F F CF 2 C O C CF 2 CF 3. - e - OH F F F 1. + CF 2 C O C CF OH CF 3 F 2 CF 2 F C OH - OH... F.. CF 2 C F.. O C CF 2 CF 3 15 CF 3 14 F 8 16 O - HF CF 2 C CF 3 F - CF 2 CF 2 CF OH F H O O C CF - O 2 2. CF 2 CF 2. F 9 10 OH HO O C CF HF H 2 O F O C CF HF HO CF 2 CF OH Scheme 3-6. Proposed Degradation Route-2a for Ether Cleavage of - CF 2 -C(F)CF 3 -O- CF 2 - CF 2 - Structure 123

149 Degradation Route-2b F F CF 2 C O C CF 2 CF 3. - e - OH F F F 1. + CF 2 C O C CF OH CF 2 F C OH CF 2... F C... CF 3 F O C CF 2 F F - OH H O O C CF 2 15 CF 3 18 CF 3. OH F 13. OH F 9 - HF -O 2 O CF 2 C CF 3 16 HO O C CF HF H 2 O F 6 O C CF 2 -HF HO CF 2 CF OH. CF2 CF 2 10 Scheme 3-7. Proposed Degradation Route-2b for Ether Cleavage of - CF 2 -C(F)CF 3 -O- CF 2 - CF 2 - Structure 124

150 Degradation Route-2c F F CF 2 C O C CF 2 CF 3. - e - OH F F F 1. + CF 2 C O C CF OH CF 3 F 2 CF 2 F C.. O F C CF 2 CF 3 19 F 12 Destabilized Scheme 3-8. Proposed Degradation Route-2c for Ether Cleavage of - CF 2 -C(F)CF 3 -O- CF 2 - CF 2 - Structure 125

151 Degradation Route-2d F F CF 2 C O C CF 2 CF 3. - e - OH F F F 1. + CF 2 C O C CF OH CF 2 21 F C O CF 3. OH OH. OH CF 2 F.. C.. O. CF 3 20 CF 3 F.. + C CF 2 F F - F OH.. + F.. C CF 2 F CF 2 18 F C - O 2. CF 3. OH CF 2 F C CF 3 CF 2 CF 3 4 O OH -HF CF 2 C CF HO CF 2 CF 2 - HF O F C CF 2 H 2 O - HF 6 11 HO O C CF 2 7 Scheme 3-9. Proposed Degradation Route-2d for Ether Cleavage of - CF 2 -C(F)CF 3 -O- CF 2 - CF 2 - Structure 126

152 The generalization of the trapping experiment results for MC7 and MC8 shown in Table 3-6 and Table 3-7 suggests the most abundant 4-hydroxy-TEMPO-trapped adduct structures, as depicted in Figure The ion abundance of such adducts are two 2~5 times higher than the rest of all postulated adducts for MC7 and MC8 Combining the results of trapping experiments and the degradation product analysis for MC7 and MC8, the possible degradation routes are proposed and shown in Scheme 3-4 and Scheme 3-5 (MC7 type), and Scheme 3-6 to Scheme 3-9 (MC8 type). The initiation step among all these proposed routes is the formation of a possible radical cation (structure 2), which is hypothesized to be the result of oxidation by hydroxyl radical via a single electron transfer (SET) manner, due to the extremely high standard reduction potential of hydroxyl radical, c.a V (significantly higher than that of H 2 O 2 for 1.78V ) Similar cationic molecular radical species for perfluoroethers have been proposed and reported as seen in earlier discussion and Figure 3-25, 34, 54 although there may be differences in the energetic characteristics involved in the formation of such intermediate, i.e. under ionizing irradiation conditions versus electron transfer to hydroxyl radical in this work. The degradation of linear perfluoro ethers (MC7 type) via radical ether bond cleavage is shown in Scheme 3-4 (Route-1a) and Scheme 3-5 (Route-1b). The cleavage of the C-O bond in structure 2 to form structure 3 and 8 can potentially be driven by the formation of resonance stabilized structures shown. 7, 8 The fluorinated carbocation, structure 3, has been shown to be stabilized by the alpha fluorines via the mesomeric interaction of an unshared pair with the empty orbital of the carbon cation center, although the competing inductive effect from adjacent fluorine atoms tends to destabilize 127

153 2, 7, 8 it. Subsequent degradation of structure 8 can account for various adducts with terminal -COOH groups that are directly observed by LC-MS or trapped by 4-hydroxy- TEMPO for MC7. Further degradation of structure 3 can lead to the formation of -CF 3 and also degradation products containing -COOH. Route-1b is less likely than Route-1a because primary perfluoro- radical (structure 12) is destabilized. 55 The degradation routes for a branched perfluoroether (MC8 type) are more complicated. All possible degradation routes are proposed in Scheme 3-6 to Scheme 3-9, while the mechanism depicted in Scheme 3-7 (Route-2b) is thought to be the most likely mechanism, based on the discussion below. In Route-2a (Scheme 3-6), the formation of structure 14 may not be very likely due to the destabilizing inductive effect of the neighboring -CF 3 to compete with the stabilization of alpha fluorine resonance as discussed for structure 3 above. The final structure 16 was not observed in current experiments, possibly because of poor solubility in the aqueous degradation test. In Route-2b (Scheme 3-7), the secondary carbon radical in structure 18 may be stabilized. The stabilization ability of the tertiary carbon center for perfluorinated 9, 10 compound was reported in literature as discussed in the introduction section of Chapter 2. This argument agrees with the most abundant structure trapped by 4-hydroxy- TEMPO radical for MC8 as shown in Figure The stability of the radical structure via resonance of the adjacent alpha fluorine electron, was also well accepted (depicted in Figure 3-38). 7 Also shown in Figure 3-38 is the structure of a kinetically stable tertiary carbon center perfluorinated radical, known as Scherer s radical which persists at room 55, 56 temperature even in the presence of molecular oxygen. The final degradation 128

154 product from structure 18, i.e. structure 16, was however not be observed. The good reproducibility of MC8-ADT-5 in the trapping experiments also reinforces the higher probability of structure 18. The possible cleavage of C-O bond from the difluoromethylene side is shown in Scheme 3-8 and Scheme 3-9. Route-2c in Scheme 3-8 is unlikely for the same reason that structure 12 is destabilized as discussed in Scheme 3-5. Among all degradation routes for MC8 type ether, degradation route 2-b seems to be the most probable route, or at least occurs in higher probability than other proposed route, it is therefore reasonable to propose that the ether cleavage of a branched perfluoroether may primarily occur through the breaking of the C-O on the branched carbon side. This hypothesis will be further examined in Chapter 4 through the comparison of the degradation results of the polymer systems with the MC systems discussed so far C F C F CF 3 CF 3 CF CF CF 3. CF CF 3 CF 3 CF 3 Scherer's Radical Figure Resonance Stabilization of the Fluorocarbon-centered Radical from Adjacent Fluorine (left), and the Structure of Scherer s Radical (right) 129

155 3.4.6 Other Control Experiments of Degradation In addition to the experiments to probe the MC degradation mechanisms above, various control experiments for the purpose of clarification were also carried out. a. Acid Catalysis Effect and Ether Hydrolysis Given the consideration that hydrocarbon ether hydrolyzes easily in the presence of an acid catalyst, control experiments were carried out to verify whether such an effect applies to current fluoro-ether. Additionally, some MCs do not contain any strong acid functionality, e.g. MC1. Thus, mild-condition degradation experiments were also conducted in the presence of 1 equivalent (1 eq.) of strong sulfuric acid purposely added to the reaction mixture. The data are shown in Figure 3-39 to Figure The designations of the data in these Figures are as follows: MC# refers to the mild Fenton s degradation test of the corresponding MC; and MC#-R refers to the replica of the mild Fenton s degradation test; MC# + H 2 SO 4 (1eq.) refers to the mild Fenton s degradation test with the addition of 1 eq. sulfuric acid; and MC#-Acid Hydrolysis refers to the control experiment where fluoride concentrations were measured from the mixture of MC# and sulfurc acid (without the presence of Fenton s reagents). For MC2 and MC3, seen from Figure 3-40 and Figure 3-41, the added H 2 SO 4 does not show any catalytic enhancement of the degradation of MC2 and MC3, i.e. no increase of fluoride generation (or MC2/MC3 degradation) was observed. For MC1, there are no detectable fluorides when only MC1 and 1 eq. sulfuric acid are mixed together with the addition of Fenton s reagents. The degradation test with 1 eq. 130

156 purposely added sulfuric acid also seemed to suggest marginal catalytic effect, although the error bars seen in Figure 3-39 is somehow large. The product analysis primarily by 19 F NMR does not exhibit noticeable spectral changes Accumulated F- detected / % of total F atoms in MC MC1 MC1-R MC1+H2SO4(1eq) MC1-Acid Hydrolysis MC1 + H2SO4 (1eq)-R Accumulated Testing Time / Hrs Figure Control Experiments to Check Acid Catalysis Effect for MC1 131

157 Accumulated F- detected / % of total F atoms in MC MC2-R2 MC2-R3 MC2 Acid Catalysis Accumulated Testing Time / Hrs Figure Control Experiments to Check Acid Catalysis Effect for MC Accumulated F- detected / % of total F atoms in MC MC3 MC3-R MC3 Acid Catalysis Accumulated Testing Time / Hrs Figure Control Experiments to Check Acid Catalysis Effect for MC3 132

158 b. UV Photolysis Degradation of MC1 Extensive control experiments for the UV photolysis degradation of MC1 were carried out with the concentration of MC1 being varied as: 0.5, 5, and 50 mm. At each MC concentration, four concentrations of hydrogen peroxide (0.5, 5, 50, and 500 mm) are used to probe the effect of concentration of H 2 O 2 in the degradation test. The results are plotted in Figure 3-42 to Figure F - % of total F atom in MC F - % of total F atom in MC 3.0 [MC1] = 0.5 mm [MC1] = 0.5 mm [H 2 O 2 ] / M [H 2 O 2 ] / M Figure Fluoride Generation from MC1 (0.5mM) Solution at Various H 2 O 2 Concentration upon UV Exposure 133

159 F - % of total F atom in MC F - % of total F atom in MC 4.0 [MC1] = 5 mm [MC1] = 5 mm [H 2 O 2 ] / M [H 2 O 2 ] / M Figure Fluoride Generation from MC1 (5mM) Solution at Various H 2 O 2 Concentration upon UV Exposure F - % of total F atom in MC F - % of total F atom in MC [MC1] = 50 mm [MC1] = 50 mm [H 2 O 2 ] / M [H 2 O 2 ] / M Figure Fluoride Generation from MC1 (50mM) Solution at Various H 2 O 2 Concentration upon UV Exposure 134

160 F - % of total F atom in MC [MC1] = 0.5 mm [MC1] = 5 mm [MC1] = 50 mm [H 2 O 2 ] / M Figure Summary Plot for Figure 3-42 to Figure The overall fluoride generation plot in Figure 3-45 suggests that the magnitude of fluoride generation is independent of MC1 concentration. For each identical MC1 concentration, higher concentration of H 2 O 2 surprisingly leads to lower measured fluoride concentration. It is possible to explain such observation by the reaction between hydroxyl radical and H 2 O 2 to form less reactive hydroperoxyl radicals, as shown by: HO. + H 2 O 2 HOO. + H 2 O, i.e. any newly formed hydroxyl radical will likely be consumed by surrounding hydrogen peroxide than by MC molecules. Therefore, the increase of H 2 O 2 concentration may actually decrease the UV degradation efficiency due to such a cage effect of H 2 O

161 c. MC Degradation as a Function of Hydrogen Peroxide Concentration F - released / % of total F atoms in MC mM H2O2 10mM H2O2 1mM H2O Accumulated Degradation Time / Hrs Figure Fluoride Generation from MC4 Solution as a Function of H 2 O 2 Concentration) F - released / % of total F atoms in MC 100mM H2O2 10mM H2O2 1mM H2O Accumulated Degradation Time / Hrs Figure Fluoride Generation from MC8 Solution as a Function of H 2 O 2 Concentration) 136

162 Figure 3-46 and Figure 3-47 show two sets of control experiment carried out to examine the effect of hydrogen peroxide concentration on the degradation, where the concentration of H 2 O 2 was varied from 1mM to 100mM. The concentrations of MC and Fe 2+ were respectively controlled to be 100 mm and 1.25 mm (identical to those used in mild condition degradation tests). This control experiment also has very important implication to the interpretation of the degradation in operating PEMFC condition, where the concentration of H 2 O 2 can vary constantly. For MC8, the effect of H 2 O 2 concentration on fluoride generation is marginal, if any at all. For MC4, the increase of H 2 O 2 concentration has a drastic effect on the fluoride generation. The origin of this difference is not clear at this moment, but it can possibly be explained as follows. In these control experiments, H 2 O 2 is always in excess with respect to ferrous ion (1.25 mm). Other than being rapidly consumed by ferrous ions, a small population of H 2 O 2 can therefore homolytically split into hydroxyl radicals at elevated temperature, and such newly formed radicals have two routes: they can either react with the MC substrate or recombine and further decompose in the Fenton s cycle without reacting with MCs. As shown previously, the reactivity of -COOH functionality (as in MC4) with hydroxyl radical is significantly higher than the cleavable ether functionality (as in MC8), therefore the probability of MC4 to catch the hydroxyl radicals formed purely from thermal homolysis of H 2 O 2 is higher, leading to the higher concentration of fluoride generated. The increase of fluoride release for MC4 does not scale with the increase of H 2 O 2 concentration, suggesting that not all the extra H 2 O 2 are utilized in reacting and degrading MC4. 137

163 3.5 Conclusions Combining the product analyses with the relative fluoride generation rates in this study, a viable model for PFSA radical degradation presents itself: 1. To the extent that backbone carboxylic acid groups exist in a PEM membrane, those groups will serve as the preferred sites of attack. 2. Ether linkages, which connect the ionomeric side chains groups to PTFE backbones are also viable points of attack for peroxide radicals, and can lead to side chain cleavage. 3. The mechanism of ether cleavage in the presence of hydroxyl radicals is proposed. The ether cleavage of a branched perfluoroether primarily occurs by oxidation of the perfluoether through a single electron transfer to form a radical cation, followed by C-O bond cleavage. Thermodynamic stability analyses and the evidence from trapping experiments suggest that the most likely route of cleavage is the one that generates a carbon-centered radical and a oxygencentered cation, i.e. Route-2b. 138

164 References 1. Wall, L. A. E., Fluoropolymers. Wiley-Interscience Banks, R. E.; Smart, B. E.; Tatlow, J. C., Organofluorine Chemistry, Principles and Commercial Applications. Plenum Press, New York Scheirs, J. E., Modern Fluoropolymers, High Performance Polymers for Diverse Applications. John Wiley & Sons Ltd, England Hougham, G.; Cassidy, P. E.; Johns, K.; Davidson, T., Fluoropolymers 1. Synthesis. Klumer Academic / Plenum Publishers, New York Hougham, G.; Cassidy, P. E.; Johns, K.; Davidson, T., Fluoropolymers 2. Properties. Klumer Academic / Plenum Publishers, New York Drobny, J. G., Technology of Fluoropolymers. CRC Press, Inc Chambers, R. D., Fluorine in Organic Chemistry. CRC Press, Inc Uneyama, K., Organofluorine Chemistry. Blackwell Publishing Rowan, S. J., Supramolecular crystal engineering: design and synthesis of novel persubstituted aromatic host molecules. PhD Dissertation, University of Glasgow MacNicol, D. D.; Robertson, C. D., New and unexpected reactivity of saturated fluorocarbons. Nature 1988, 332, Teranishi, K.; Kawata, K.; Tsushima, S.; Hirai, S., Degradation mechanism of PEMFC under open circuit operation. Electrochemical and Solid-State Letters 2006, 9, (10), A475-A Tang, H.; Peikang, S.; Jiang, S. P.; Wang, F.; Pan, M., A degradation study of Nafion proton exchange membrane of PEM fuel cells. Journal of Power Sources 2007, 170, (1), Pozio, A.; Silva, R. F.; De Francesco, M.; Giorgi, L., Nafion degradation in PEFCs from end plate iron contamination. Electrochimica Acta 2003, 48, (11), Liu, W.; Zuckerbrod, D., In situ detection of hydrogen peroxide in PEM fuel cells. Journal of the Electrochemical Society 2005, 152, (6), A1165-A Liu, W.; Ruth, K.; Rusch, G., Membrane Durability in PEM Fuel Cells. Journal of New Materials for Electrochemical Systems 2001, 4, (4),

165 16. Liu, W.; Crum, M., Effective testing matrix for studying membrane durability in PEM fuel cells: part I. Chemical durability. ECS Transactions 2006, 3, (1, Proton Exchange Membrane Fuel Cells 6), Knights, S. D.; Colbow, K. M.; St-Pierre, J.; Wilkinson, D. P., Aging mechanisms and lifetime of PEFC and DMFC. Journal of Power Sources 2004, 127, (1-2), Kinumoto, T.; Inaba, M.; Nakayama, Y.; Ogata, K.; Umebayashi, R.; Tasaka, A.; Iriyama, Y.; Abe, T.; Ogumi, Z., Durability of perfluorinated ionomer membrane against hydrogen peroxide. Journal of Power Sources 2006, 158, (2), Chen, C.; Levitin, G.; Hess, D. W.; Fuller, T. F., XPS investigation of Nafion membrane degradation. Journal of Power Sources 2007, 169, (2), Aoki, M.; Uchida, H.; Watanabe, M., Decomposition mechanism of perfluorosulfonic acid electrolyte in polymer electrolyte fuel cells. Electrochemistry Communications 2006, 8, (9), Walling, C., Fenton's reagent revisited. Accounts of Chemical Research 1975, 8, (4), Collier, A.; Wang, H.; Yuan, X. Z.; Zhang, J.; Wilkinson, D. P., Degradation of polymer electrolyte membranes. International Journal of Hydrogen Energy 2006, 31, (13), Bosnjakovic, A.; Schlick, S., Naflon perfluorinated membranes treated in Fenton media: radical species detected by ESR spectroscopy. Journal of Physical Chemistry B 2004, 108, (14), Bosnjakovic, A.; Schlick, S., Spin Trapping by 5,5-Dimethylpyrroline-N-oxide in Fenton Media in the Presence of Nafion Perfluorinated Membranes: Limitations and Potential. Journal of Physical Chemistry B 2006, 110, (22), Kadirov, M. K.; Bosnjakovic, A.; Schlick, S., Membrane-Derived Fluorinated Radicals Detected by Electron Spin Resonance in UV-Irradiated Nafion and Dow Ionomers: Effect of Counterions and H2O2. Journal of Physical Chemistry B 2005, 109, (16),

166 26. Panchenko, A.; Dilger, H.; Moller, E.; Sixt, T.; Roduner, E., In situ EPR investigation of polymer electrolyte membrane degradation in fuel cell applications. Journal of Power Sources 2004, 127, (1-2), Schiraldi, D. A., Perfluorinated polymer electrolyte membrane durability. Polymer Reviews 2006, 46, (3), Walling, C., Intermediates in the reactions of Fenton type reagents. Accounts of Chemical Research 1998, 31, (4), ThermoOrion, Fluoride ISE Product Specification and Manual Chen, J.; Zhang, P.-y.; Liu, J., Photodegradation of perfluorooctanoic acid by 185 nm vacuum ultraviolet light. Journal of Environmental Sciences (Beijing, China) 2007, 19, (4), Blakey, I.; George, G. A.; Hill, D. J. T.; Liu, H. P.; Rasoul, F.; Whittaker, A. K.; Zimmerman, P., XPS and F-19 NMR study of the photodegradation at 157 nm of photolithographic-grade teflon AF thin films. Macromolecules 2005, 38, (10), Forsythe, J. S.; Hill, D. J. T., The radiation chemistry of fluoropolymers. Progress in Polymer Science 2000, 25, (1), Lee, K.; Jockusch, S.; Turro, N. J.; French, R. H.; Wheland, R. C.; Lemon, M. F.; Braun, A. M.; Widerschpan, T.; Dixon, D. A.; Li, J.; Ivan, M.; Zimmerman, P., 157 nm pellicles (thin films) for photolithography: Mechanistic investigation of the VUV and UV-C photolysis of fluorocarbons. Journal of the American Chemical Society 2005, 127, (23), Spool, A. M.; Kasai, P. H., Perfluoropolyethers: Analysis by TOF-SIMS. Macromolecules 1996, 29, (5), Zhao, X.; Bhushan, B.; Kajdas, C., Lubrication studies of head-disk interfaces in a controlled environment, Part 2: degradation mechanisms of perfluoropolyether lubricants. Proc Instn Mech Engrs, Part J 2000, 214, ( ). 36. Kasai, P. H., Z-dol and carbon overcoat: the bonding mechanism. Tribology Letters 2007, 26, (2),

167 37. Liu, J. W.; Stirniman, M. J.; Gui, J., Catalytic decomposition of perfluoropolyether lubricants. Ieee Transactions on Magnetics 2003, 39, (2), Denkenberger, K. A.; Bowers, R. A.; Jones, A. D.; Mueller, K. T., NMR studies of the thermal degradation of a perfluoropolyether on the surfaces of gammaalumina and Kaolinite. Langmuir 2007, 23, (17), Pacansky, J.; Waltman, R. J., The effect of Lewis acid catalysis on the decomposition of CF3OCF3 to COF2 and CF4. Journal of Fluorine Chemistry 1997, 83, (1), Pacansky, J.; Waltman, R. J., The decomposition of CF3OCF3 to COF2 and CF4. Journal of Fluorine Chemistry 1997, 82, (1), Pacansky, J.; Waltman, R. J.; Jebens, D., Electron beam irradiation of a perfluoroalkoxy fluorocarbon resin: Tetrafluoroethylene/perfluoromethyl vinyl ether copolymer. Macromolecules 1996, 29, (24), Kasai, P. H.; Wheeler, P., Degradation of Perfluoropolyethers Catalyzed by Aluminum-Chloride. Applied Surface Science 1991, 52, (1-2), Kasai, P. H.; Spool, A. M., Perfluoro-Poly(ethers) with Hydroxyl End Groups: Analysis by TOF-SIMS. Journal of Physical Chemistry B 1998, 102, (38), Kasai, P. H., Perfluoropolyethers with Acid End Groups: An ESR Study of Decarboxylation. Chemistry of Materials 1994, 6, (9), Eapen, K. C.; Chen, L. S.; Chen, G. J., Thermal decomposition of low-molecular weight perfluoroalkylethers. Journal of Fluorine Chemistry 1997, 81, (2), Forsythe, J. S.; Hill, D. J. T.; Logothetis, A. L.; Seguchi, T.; Whittaker, A. K., Thermal and mechanical properties of radiation crosslinked poly(tetrafluoroethylene-co-perfluoromethyl vinyl ether). Radiation Physics and Chemistry 1998, 53, (6), Marchionni, G.; Petricci, S.; Guarda, P. A.; Spataro, G.; Pezzin, G., The comparison of thermal stability of some hydrofluoroethers and hydrofluoropolyethers. Journal of Fluorine Chemistry 2004, 125, (7),

168 48. Kasai, P. H., Perfluoropolyethers - Intramolecular Disproportionation. Macromolecules 1992, 25, (25), Waltman, R. J., A computer modelling study on the interaction of -(CF2CF2O)- polyperfluorinated ethers with Lewis acid sites: perfluorodiethyl ether. Journal of Fluorine Chemistry 1998, 90, (1), Buxton, G. V.; Greenstock, C. L.; Helman, W. P.; Ross, A. B., Critical-Review of Rate Constants for Reactions of Hydrated Electrons, Hydrogen-Atoms and Hydroxyl Radicals (.Oh/.O-) in Aqueous-Solution. Journal of Physical and Chemical Reference Data 1988, 17, (2), Koppenol, W. H.; Liebman, J. F., The Oxidizing Nature of the Hydroxyl Radical - a Comparison with the Ferryl Ion (Feo2+). Journal of Physical Chemistry 1984, 88, (1), Klaning, U. K.; Sehested, K.; Holcman, J., Standard Gibbs Energy of Formation of the Hydroxyl Radical in Aqueous-Solution - Rate Constants for the Reaction Clo2- + O3 Revirsible O3- + Clo2. Journal of Physical Chemistry 1985, 89, (5), Zumdahl, S. S., Chemistry. D.C.Heath and Company, 3rd Ed.,Lexington, MA 1993, Pg Faucitano, A.; Buttafava, A.; Karolczak, S.; Guarda, P. A.; Marchionni, G., The chemical effects of ionizing radiations on fluorinated ethers. Journal of Fluorine Chemistry 2004, 125, (2), Dolbier, W. R., Structure, Reactivity, and Chemistry of Fluoroalkyl Radicals. In 1996; Vol. 96, pp Scherer, K. V.; Ono, T.; Yamanouchi, K.; Fernandez, R.; Henderson, P.; Goldwhite, H., F-2,4-Dimethyl-3-Ethyl-3-Pentyl and "F-2,4-Dimethyl-3- Isopropyl-3-Pentyl - Stable Tert-Perfluoroalkyl Radicals Prepared by Addition of Fluorine or Trifluoromethyl to a Perfluoroalkene. Journal of the American Chemical Society 1985, 107, (3),

169 Chapter 4. Chemical Degradation and Structure-and-Property Change of PFSA Ionomers 4.1 Introduction The results of MC degradation experiments, product analyses, and trapping of intermediates presented in Chapter 3 suggest that the major degradation routes of PFSA type ionomers include the cleavage of ether bonds, which can subsequently lead to the degradation and loss of side chain moieties, and the eventual decrease of proton conductivity. The degradation of Nafion and 3M ionomers (containing fluoroethers/side chain sub-units) is now examined and compared to the results of MC systems. The organization of Chapter 4 is as follows: 1) experimental section; 2) results directly observed from degradation experiments in terms of fluoride generation, degradation product identification, thickness and weight loss; and 3) other results of spectroscopic and solid-state property changes. 4.2 Experiments Fenton s Degradation Procedure for Membrane Samples a. Materials and Membrane Pretreatment Protocol Table 4-1 lists five membrane samples used in this study with the sample designations and explanation of major differences. Nafion (Nafion 117 membrane) was purchased from Aldrich. The rest of membranes were provided by 3M. The stabilization of reactive end groups (carboxylic acid or acyl fluoride) in fluoropolymers is 144

170 routinely achieved by heating the polymer in the presence of steam, or reaction with methanol, ammonia, or amines to convert the end groups to more stable ester, amide, or difluoromethyl groups, 1-3 or treating with elemental fluorine to provide an extremely stable CF 3 end group and very low level of ionic contamination. 4, 5 The methods employed to manufacture the polymers used in this study were not disclosed due to proprietary consideration from 3M. Information about relative degree of stabilization of such polymers was only provided to be three main categories: not modified (designated NM), partially modified (designated PM), and best modified (designated BM). Also in Table 4-1, Nafion is used for the commercially available Nafion 117 sample (EW=1100, thickness is about 175µm). 3M-Nafion is used to described the Nafion membrane recast by 3M (EW=1100, thickness is about 40µm). Table 4-1. Sample Designations and Characteristics of Various Ionomers Examined Ionomers 3M-G1-NM 3M-G1-PM 3M-G1-BM 3M-Nafion Nafion Characteristics NM = Not end group modified PM = Partially end group modified BM = Best end group modifed Nafion recast by 3M Commercial Nafion sample Membrane samples were converted to the acid form by the following protocol prior to the degradation test: membranes were cleaned by heating in a 1.5% v/v hydrogen 145

171 peroxide solution at 70 o C for 1 hour, followed by washing the membrane in a hot DI water bath (70 o C) for 1 hour. The membrane was then boiled in 1M sulfuric acid for 1 hour to convert to acid form. The whole process was completed by subsequent washing with boiling DI water for 1 hour. b. Mild Fenton s Degradation Test Solutions containing 1.25 mm Fe 2+ ions were prepared by dissolving ferrous sulfate in DI water. Membrane samples were then added to this solution. A three-neck flask was sealed by two septum caps and one addition funnel, also capped with septum cap. The flask was then bubbled with Nitrogen for 1 hour, followed by increasing the temperature to 70 o C in an oil bath. H 2 O 2 (the amount is calculated based on concentration listed in Table 4-2) was added in by addition funnel when the temperature was constant. The system was bubbled with argon throughout the course of experiment in order to keep oxygen from reacting with the radicals. Final concentrations of reagents are listed in Table 4-2. Table 4-2. Concentrations of Reagents Used in Mild Membrane Degradation Tests Fe(II) H 2 O 2 Ionomers DI Water Reagents Concentrations 1.25 mm (ca 70 ppm) 11 mm 20 ~ 50 mg 50 ml (total) 146

172 c. Modified Fenton s Degradation Test Ionomers/H + (acid form) Ion exchange with FeSO 4 solution (0.1M) 70 o C, 2 Hr Ionomers/Fe 2+ F - Measurement NMR, LC-MS 70 o C H 2 O 2 Degraded Ionomers/H + Convert back to H + form 70 o C, 2 Hr, 0.5M H 2 SO 4 Degraded Ionomers Ionomers/Fe 3+ Structural Characterization Thermal & Mechanical Properties Figure 4-1. Modified Fenton s Degradation Test Procedure for Membrane Samples The degradation test procedure for membrane samples is depicted in Figure 4-1. The degradation test started with the acid form of Nafion. The first step was to convert Nafion to its Fe(II)-saturated form by immersing the sample in a 0.1M FeSO 4 solution for 2 hours at 70 C. After the ion exchange, the sample was removed and rinsed with deionized water to remove the residual ion exchange solution from the membrane, and further blotted dry with paper towels. The samples appeared to be light yellow in color prior to the degradation test. The targeted Fe(II) loading was quantitative relative to the concentration of sulfonic acids, which was calculated from the equivalent weight of Nafion. The second step was to put the Fe(II) saturated Nafion into a water bath (which did not contain ferrous ions like in previous Fenton s degradation tests). Then the solution was degassed with dry nitrogen gas for at least 10 minutes prior to the addition 147

173 of H 2 O 2. The solution was then heated to 70 ± 2 C and hydrogen peroxide (concentration 0.1M based on the total volume of the reaction media, identical to the concentration used in the harsh Fenton s experiment) was slowly introduced into the flask to react through an addition funnel. The reaction mixture was held at 70 C under a purge of nitrogen for a certain period of reaction time, typically ca. 35 hours. The fluoride concentration measurement was subsequently carried out by removal of 2ml aliquots from the reactor. The reaction media was not discarded but used in the following test (same amount of water 2 ml was added back to the reactor to balance the total volume along with the addition of hydrogen peroxide for the next round of degradation experiment). The third step was to convert the degraded sample (dark brown color) back to acid form (colorless) by immersing in 0.5M H 2 SO 4 at 70 C for 1 hour. The next round of degradation test was then carried out by following the procedure described above Fragments Collections and Extraction from Membranes Low molecular weight degradation product fragments were obtained by direct sampling of the degradation bath or room temperature solvent extraction from membrane samples. For the direct sampling method, aliquots of 2ml reaction medium were removed at different reaction times and transferred into Pyrex 15ml conical centrifuge tubes, followed by centrifuging at 2000~3000 rpm for 5 to 10 minutes (Centrifuge model: Centrific TM from Fisher Scientific). After centrifuging, samples were filtered and collected by using 0.2 micron Iso-Disc TM filters (from Sulpelco). For the room temperature solvent extraction method, membrane samples with the weight range of 50 ~ 250 mg were soaked in about 5 ml DI water, methanol, or acetonitrile at room 148

174 temperature for 1~3 days in capped glass vials. The solutions were decanted into vials for further analysis by NMR and LC-MS LC-MS Experiment The degradation product identification method using LC-MS was identical to that used for MC systems described in Chapter Fourier Transform Infrared Spectroscopy (FT-IR) Characterization Different modes of FT-IR spectroscopy characterization were carried out on the original and degraded ionomer samples by a Perkin-Elmer spectrometer. For transmission mode FT-IR, cm -1 wave number range was scanned with 32 acquisitions. For attenuated total reflectance (ATR) mode, the wave number range was cm -1 with an acquisition of 16 scans. For the ATR-microscope mode, the scanning wave number was from cm -1, and acquisition was 16 scans Membrane Weight Loss, Water Up-take, and Equivalent Weight (EW) Measurements Protonation and Drying Procedure. Membrane samples (fresh or aged) were converted to protonated form by soaking in H 2 SO 4 (0.5~1 M) for one hour, followed by immersion in boiling DI water bath to remove excess acids. Before transferring to vacuum drying oven, all membrane samples were washed with copious amount of DI water and blotted dry using Kim-wipe TM tissues. Samples were dried at 75 o C for

175 hours under vacuum. After drying and between various tests, samples were kept in either clean plastic bottles or plastic sample bags in a desiccator cabinet. Membrane Weight Loss. Clean glass weighing bottles with sealing caps were used as containers. The weighing bottles were first cleaned in an acid bath for 3 hours followed by thorough washing with detergent and DI water before drying in an oven at about 150 o C. Caution was taken to ensure that the matching cap provides sufficient sealing by a leak test with DI water. After drying, the sealing caps were put on the weighing bottles before quickly being transferred into a desiccator to minimize water condensation on glass walls while cooling down. Fresh and degraded membrane samples were dried by the protocol described above after protonation. After drying, the oven temperature was gradually cooled down to about 35 o C while maintaining vacuum. The membrane samples were then quickly transferred into the weighing bottle and capped for weight measurement using an analytical balance. Water Up-take Measurement. After drying, membrane samples were immersed in DI water at room temperature (the DI water was boiled prior to use to remove dissolved carbon dioxide to avoid interference with titration). Control experiments and the literature showed that the water-uptake of membrane can reach equilibrium in less than 5 minutes. 6 After equilibrium, samples were quickly blotted dry to remove surface water and transferred to weighing bottles for weight measurements. 150

176 EW Measurement: The dried samples were soaked in NaCl (1-3M) overnight to exchange Na + for H + from membranes. The exchanged protons were titrated with NaOH using phenolphthalein as an indicator. The concentration of NaOH used was standardized by oxalic acid solution of known concentration, and was found to be about 3 mm. The equivalent weight data were typically averaged from three independent titrations. The formula used to calculate EW is: EW = (amount of polymer in grams) / (titrated amount of H + in moles) Membrane Proton Conductivity Measurement Figure 4-2. Cell Used for Determination of Membrane Conductivity. (1) Kel-F block; (2) thumbscrew; (3) open area to allow equilibrium; (4) membrane sample: (5) blackened Pt foil; (6) Pt ribbon lead. 6 Following a reported method, proton conductivity was measured by using the cell sketched in Figure The window slot shape was used to ensure membrane 151

177 equilibrium and direct contact to liquid water. The resistivity of the membranes at 5kHz was measured by using a pair of pressure-attached high surface area Pt electrodes. All electrodes were cleaned and kept in DI water in a plastic beaker throughout the course of measurement. Conductivity measurements were carried out in a liquid water bath at around 20 o C. The cell constant was calculated from the spacing of the electrodes, the thickness, and the width of the sample. The thickness and width of samples were averaged from three different locations of the sample, the center and both edges inside the window slot. The formula used to calculate the proton conductivity (S/cm) is: σ = l / (A. R), where l is known from the cell as 1.8 cm, and A (cm 2 ) is the area calculated from width and thickness, and R (Ω) is the resistivity obtained from the AC impedance measurement (frequency range: 1~10,000 Hz, amplitude: V, quiet time: 2 s) Scanning Electron Microscope (SEM) Characterization Surface and cross-sectional micrographs were obtained by a Phillips model XL-30 ESEM scanning electron microscope. Prior to gold sputtering, membrane samples were converted to acid form by the protonation and drying procedure described above. Samples used for cross-sectional imaging were obtained by fracturing the sample with tweezers after immersing the samples in liquid nitrogen for about 5 minutes Dynamic Mechanical Analysis (DMA) Characterization Rectangular sample specimens (width: 5-6 mm and length: c.a. 10 mm) were cut from the fresh and degraded samples and subjected to DMA test by using a TA 152

178 Instruments Q800 DMA analyzer. The test specimen was clamped in a tensile fixture and the sample chamber flushed with liquid nitrogen to cool the sample down to -100 C. Samples were first equilibrated after cooling for about 5 minutes, and then the temperature was ramped at 3 C/min from -100 to 130 C with an oscillatory stress applied at a frequency of 1 Hz. Thermo-mechanical properties of the sample such as storage modulus, loss modulus and loss tangent-tan δ, were measured as a function of temperature. The glass transition temperature, T g, was obtained as the maxima of the Tan δ plot against temperature Tensile Testing Room temperature (25 C) tensile measurements were perfomed on the membrane samples using a DMA (TA Instruments Q800) utilized as a tensile tester. The measurements were made in the controlled force mode (sample dimensions: width: 5-6 mm; length: c.a. 10 mm). The samples were initially equilibrated at 25 C for 5 minutes. The force applied was ramped from an initial pre-load value of N to 18N at a rate of 0.1 N/min. The data generated was utilized to plot stress-strain curves for the samples Differential Scanning Calorimetry (DSC) Characterization DSC measurements were carried out on a Mettler-Toledo DSC822e/700 DSC analyzer in two heating-cooling cycles between -50 and 250 C with a ramping rate of 10 C/min for 4-10 mg samples Wide Angle X-Ray Diffraction (XRD) Characterization 153

179 X-ray diffractograms (XRD) were recorded on fresh and degraded membrane samples using a Scintag X-1 Advanced X-Ray Diffraction system. Cu-Ka radiation was employed (k = Ǻ; tube current 39.5 ma, tube voltage 44.5 kv). 2θ angular regions between 1 and 50 were scanned with a resolution of 0.05 degrees. 4.3 Results and Discussion Fluoride Generation F - released, mmol / gram of membrane (Normalized by Nafion weight) M-G1-NM 3M-G1-BM 3M-Nafion Accumulated Aging Time / Hrs Figure 4-3. Fluoride Generation as a Function of Degradation Time in Mild Fenton s Degradation Test Membrane samples were initially degraded under the mild Fenton s degradation test condition, wherein hydroxyl radicals are generated in the degradation bath to attack 154

180 the membrane. The fluoride generation results are normalized by membrane sample weights and plotted in Figure 4-3. The fluoride release rates were found to be extremely low and many properties of membrane samples were found unchanged after 5-6 repetitions of degradation test cycle. The modified Fenton s degradation test condition, wherein radicals are generated inside the membrane to induce more effective degradation, was therefore employed to degrade the membranes. The fluoride generation from 3M-G1 and Nafion using the modified Fenton s degradation test condition is shown in Figure 4-4. Acc. [F - ] Measured (normalized by sample weight) / mmol*g M-G1-NM 3M-G1-PM 3M-G1-BM 3M-Nafion Accumulated Aging Time / Hrs Figure 4-4. Fluoride Generation as a Function of Degradation Time in Modified Fenton s Degradation Test Both Figures 4-3 and 4-4 showed a gradual increase of fluoride generation over the testing time. 3M-Nafion samples seemed to be more resistant toward chemical attack 155

181 than 3M ionomers from the mild degradation test data. The level of the fluoride generation, however, was very low and the difference observed in fluoride generation may practically be viewed to be comparable. For the modified degradation test, the first noticeable differences are the relative stabilities of the 3M ionomers was: 3M-G1-BM > 3M-G1-PM~3M-G1-NM. The end group stability imparted by best modified sample obviously leads to lower fluoride generation, while the other four membrane samples showed comparable levels of fluoride release. The fluoride levels generated in the modified degradation procedure were observed to be about ten times higher than those obtained in the initial Fenton s degradation, where the degradation might primarily happen on membrane surfaces. The fluoride evolution rates in our accelerated degradation tests was on the order of 1.0 x 10-6 g of fluoride/hr-cm 2, which is two orders of magnitude higher than the fluoride release rates observed in normally operating fuel cells. 7 This result shows good validation of the accelerated degradation test method used in this work, i.e. Fe 2+ ions are loaded into the membrane, and further addition of H 2 O 2 can generate attacking radical species inside the membrane. The efficiency of the degradation test may allow expedited alternation of membrane chemical structure and properties within relatively short period of testing time Fragments Identification Both the direct collection of degradation fragments and the aqueous extracts from modified Fenton s degradation test of Nafion membrane sample were analyzed using the same LC/MS. The LC-MS result was shown in Figure 4-5 (procedure described in 156

182 Chapter 3), and 19 F NMR spectrum of the aliquot solution from the degradation test medium was shown in Figure 4-6. Relative Abundance Major Product 1 from Nafion MS 1 Full Chromatograph Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Time (Minute) Relative Abundance MS 1 m/z Figure 4-5. Nafion Degradation Product LC Trace (top three, full and extracted chromatographs) and Corresponding MS Spectrum (bottom) at RT=7.7 min The major reaction product from Nafion degradation was identical to Product-1 derived from MC8 (Table 3-4, and the structure is also shown in Figure 4-6). Other than the expected ions listed in Table 3-4, the m/z value of 682 was assigned to a adduct ion consisting of a deprotonated parent ion and another intact product molecule. The identical degradation product from Nafion running in a fuel cell testing or Fenton s 157

183 degradation test was also reported by two other research groups. 7, 8 The NMR peaks were assigned based on the reported chemical shifts in literature. 7 HOOC a c d F C O CF 2 CF 2 SO 3 H CF 3 b b d c c c a Figure F NMR of Nafion Degradation Major Product from Fenton s Degradation Test Solution Nafion CF 2 CF 2 CF 2 CF x y O CF 2 CF O CF 2 SO 3 H 2 HO O C CF O CF 2 CF 2 SO 3 H CF 3 Side Chain Scission CF 3 Figure 4-7. Scheme Showing the Major Product Observed as a Result of Ether Cleavage of Nafion 158

184 This result strongly suggests that side chain cleavage occurs in the Nafion membrane as depicted in Figure 4-7, just as is the case with its small molecule analogues. It can be argued that the chain end unzipping mechanism in Nafion-like PFSAs might eventually lead to sub-structures similar to MC1, which can cause the cleavage of side chains following mechanism of MC1 as described in Scheme 3-2. This scenario could explain the degradation products observed herein with Nafion. Such a mechanism, however, does not explain the degradation products observed with polymer analogues (MC7 and MC8), nor does it explain the fluoride ion generation rate versus carboxylic acid end group presented in Figure 2-9. Based on the intermediate trapping experiments discussed in Chapter 3, the ether cleavage mechanism proposed predicts the formation of molecules with terminal -COOH or ketone type structures. The major degradation product observed from Nafion membrane contains a terminal carboxylic acid group, which exhibits good agreement with the expected structures. The structural changes of Nafion membrane after degradation were however difficult to characterize due to the intractability of the polymer. The ether cleavage induced degradation of Nafion can indeed present a second mechanistic pathway besides hydroxyl radical attack on carboxylic acid end groups. Fluoride can be generated during the Fenton s degradation test of non-carboxylic acid containing small molecule analogues as discussed above, and this degradation pathway primarily proceeds through ether cleavage. 159

185 F 2 C F 2 C x F 2 C F C y 3M-G1 O O F 2 C F 2 C F 2 C F 2 C SO 3 H HO C CF 2 CF 2 CF 2 SO 3 H Side Chain Scission Figure 4-8. Expected Fragment as a Result of Ether Cleavage of 3M-G1-NM Table 4-3. Parent Ion and Derivative Ions of the Expected Product of 3M-G1-NM NM-Product m/z Parent Ion 275 (NM-Product-H + )-CO (NM-Product-2H + ) 137 A similar degradation product identification was also carried on 3M-G1-NM. Figure 4-8 shows the expected structure of fragment should the ether cleavage reaction happens. Table 4-3 lists the parent ion and two other derivative ions of such expected product of 3M-G1-NM. The major degradation product of 3M-G1-NM was identified to be the expected structure, as shown in Figure 4-9 and Figure 4-10 with LC-MS results. This product of 3M ionomer was also identified as one of the degradation products from its MC analog, MC7. 160

186 Full Chromatograph Relative Abundance MS 1 Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Extracted Chromatograph MW: Da Time (Minute) Figure 4-9. LC Chromatographic Trace of the Aliquot Solution from 3M Membrane Degradation Test Medium (top), LC Traces (relative abundance) of Selective Ions From Expected Products in Table 4-3 (bottom two traces) 161

187 Relative Abundance MS 1 Relative Abundance Time (Minute) m/z MS 1 RT=4.35 Min Figure MS Spectrum Marked as MS-1 in Figure 4-9 (also shown as the top LC trace) at RT=4.4 min (bottom) The above results show very good agreement between the conclusion reached in the MC systems and the observed products from the corresponding polymer analog. The following sections will attempt to correlate chemical degradation pathway to the changes of membrane properties. 162

188 4.3.3 Major Changes of Critical Membrane Properties a. Conductivity The conductivity values were calculated by the value of resistance and other cell constants, as described in the experiment section. Table 4-4 shows the comparison of the conductivity change for both Nafion and 3M ionomers. The conductivity data were averaged from at least 2~3 different samples, except 3M-Nafion-Degraded due to sample depletion. Additionally, the small standard deviation of the conductivity value observed for all other samples implies that this single data point may also be representative. The percentage decrease of conductivity for 3M-G1-BM is unexpectedly higher than that of 3M-G1-NM. All the degraded ionomer samples nonetheless showed 6~12% of decrease in proton conductivity as expected from the loss of membrane side chains as a result of ether cleavage reaction. Table 4-4. Conductivity Change for Nafion and 3M Ionomers Sample Conductivity (S/cm) Conductivity Drop (%) 3M-Nafion ± M-Nafion-Degraded (single data point) 12 3M-G1-NM ± < M-G1-NM-Degraded ± M-G1-BM ± M-G1-BM-Degraded ±

189 b. Weight Loss and Thickness Change Table 4-5. Weight Loss and Thickness for Nafion and 3M Ionomers Weight Loss / % Thickness Change / % 3M-G1-NM 3M-G1-PM 3M-G1-BM 3M-Nafion Weight loss and thickness change before and after degradation are listed in Table 4-5. All the samples showed noticeable weight loss and thinning that are in good 9, 10 agreement with the observations obtained the operating PEMFCs. Weight loss of all four samples were closely within 10~15%, but thickness changes were ranged from 3~22% with 3M-G1-PM being smallest and 3M-Nafion largest. Note that the thickness data were averaged from at least 5 points from a sample by micrometer. c. EW Change Based on the above observation and discussion, the EW values of degraded membrane samples are expected to increase because of loss of side chains hence the available protons per mole of sulfonic acids. Such change was observed for 3M-G1-BM where the EW changed to 1448 from 1142 g/mol. EW values of 3M-G1-NM and 3M- Nafion showed decrease instead of the expected increase, which was probably resulted from the experimental error because the titrated sample size was too small. The EW 164

190 values of pristine Nafion samples also showed large standard deviation. The EW measured by titration seemed to be less accurate than the conductivity measurement, and therefore larger sample sizes and more data points are needed here in order to draw reasonable conclusions from these data. The loss of sulfonic acid side chains is nonetheless rather evident, based on the above results of product identification, decrease of conductivity, and the EW in the case of 3M-G1-BM. Table 4-6. EW Change after Degradation for Various Membrane Samples Sample EW g/mol 3M-G1-NM 1099 ± 36 3M-G1-NM-Degraded 898 (single data point) 3M-G1-BM 1142 ± 79 3M-G1-BM-Degraded 1448 (single data point) 3M-Nafion 1013 ± 341 3M-Nafion-Degraded 967 (single data point) FT-IR (ATR) Analysis of Degraded Membrane Samples From the literature, major IR vibrational peaks are assigned in the following Table The ATR spectra of pristine and degraded Nafion samples are shown in Figure 4-11 and Figure The peak at 1057 cm -1 -, assigned to S=O stretching of -SO 3 bonded to water (H 2 O), is clearly seen. The peak at 1415 cm -1, S=O stretching of -SO 3 H, decreases rapidly as the water content (hydration) increases because less free -SO 3 H groups are around. 165

191 Table 4-7. Common IR Peaks of Nafion and 3M Membranes cm -1, O-H stretching of water 1415 cm -1, S=O stretching of -SO 3 H 1057 cm -1, S=O stretching of -SO - 3 bonded to water (H 2 O) 967 and 982 cm -1, C-O-C stretching Intensity M-Nafion 3M-Nafion-Degraded-6 Cycle 3M-Nafion-Degraded-Replica-4 Cycle , S=O stretch of -SO 3 H Wave Number (cm -1 ) Figure ATR of Pristine and Degraded Nafion Samples 166

192 Another small emerging peak at 1790 cm -1 can be potentially assigned to the carbonyl peak of -CF 2 -C(O)-CF Should this assignment be correct, it is likely to conclude that after ether cleavage, a ketone functionality can be formed, as predicted from the current proposed ether cleavage mechanism in Chapter 3. There is another new peak at 1453 cm -1, which may be assigned to the formation of -S-O-S- bond (the crosslinking of between sulfonic acid groups on side chains) according the previous reports, where Nafion membrane degraded by Fenton s test showed similar IR spectral 16, 17 changes. Intensity 0.8 3M-Nafion 3M-Nafion-Degraded-6 Cycle 3M-Nafion-Degraded-Replica-4 Cycle S=O stretch of -SO C-O-C 967, Wave Number (cm -1 ) Figure ATR of Pristine and Degraded Nafion Samples 167

193 The ATR spectra of pristine and degraded 3M-G1-NM samples are shown in Figure 4-13 and Figure The other 3M ionomer samples showed similar spectral changes, and are not shown here for the purpose of brevity. Again, the small new peak at and 1450 cm -1, is clearly seen and can be similarly assigned to the formation of -S-O-Sbond. As to the possible formation of fluoro-ketone functionality, the peak around 1775 cm -1 is somehow smaller than that observed in the case of degraded Nafion samples. Since the differences in chemical structure between 3M-G1 and Nafion ionomers are the lack of the second C-O-C functionality (no further branching or tertiary carbon) on the side chains, the resemblance of spectral changes after degradation observed here might suggest common attacking sites. Intensity 3M-G1-NM 3M-G1-NM-Degraded-3 Cycle Wave Number (cm -1 ) Figure ATR of Pristine and Degraded 3M-G1-NM Samples 168

194 0.06 Intensity 3M-G1-NM 3M-G1-NM-Degraded-3 Cycle 0.05 S=O stretch of -SO C-O-C Wave Number (cm -1 ) Figure ATR of Pristine and Degraded 3M-G1-NM Samples Other Comparison of Solid-State Properties a. DSC Results The DSC results of various membrane samples are shown in Figure 4-15 to Figure All of these samples contain certain degree of crystallinity or chain alignment (X-ray diffraction peaks were also observed in XRD experiments), potentially as a result of the recasting process. As seen in a typical DSC curve containing traces of two heating cycles (all of them behave similarly thus only one curve is shown here), this crystalline peak melts at around 170 o C, and no crystallization peak was observed when cooling down from the melt, suggesting that the crystallinity may well be from the sample processing history. 169

195 For Nafion, the melting temperature of the crystalline peak was shifted to slightly lower temperature, while all other three 3M-G1 samples showed obvious increased melting temperatures after degradation. This is an interesting observation, despite the current lack of further experimental data to conclusively account for such changes. ^exo Chun_3M-Nafion_Degraded_090407, :07:03 Chun_3M-Nafion_Degraded_090407, mg 5 mw 癈 Lab: Chun e STAR SW 8.10 Figure A typical DSC Curve of Nafion With Two Heating Cycles Shown 170

196 ^exo ]1[Chun_3M-Nafion_Degraded_ Chun_3M-Nafion_Degraded_090407, mg Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 ]1[ Chun_3M-N afion_ Chun_3M-Nafi on_071007, mg 10 mw Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 癈 e Lab: Chun STAR SW 8.10 Figure DSC Curve of Nafion Samples with Only First Heating Trace Shown ^exo ]1[ Chun_3M-G1-NM_Deg raded_ Ch un_3m -G1-NM _Deg r ad ed_ , m g Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 ]1[Chun_3M-G1- NM_ Chun_3M - G1-NM _071007, mg 10 mw Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 癈 e Lab: Chun STAR SW 8.10 Figure DSC Curve of 3M-G1-NM Samples with Only First Heating Trace Shown 171

197 ^exo ]1[Chun_3M-G1-PM_Degraded_ Chun_3M-G1-PM_Degraded_090507, mg Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 5 mw ]1[Chun_3M- G1-PM_ Chun_3M-G1-PM_071007, mg Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 癈 e Lab: Chun STAR SW 8.10 Figure DSC Curve of 3M-G1-PM Samples with Only First Heating Trace Shown ^exo ]1[Chun_3M-G 1- BM_Degraded_ Chun_3M- G1-BM_Deg raded_090507, mg Integral normalized Onset Peak Endset mj Jg^ 癈 癈 癈 ]1[ Chun_3M-G1-BM_ Ch un_ 3M - G1- B M _ , m g 10 mw Integral normalized Onset Peak Ends et mj Jg^ 癈 癈 癈 癈 e Lab: Chun STAR SW 8.10 Figure DSC Curve of 3M-G1-BM Samples with Only First Heating Trace Shown 172

198 b. DMA Test and Tensile Test Results The results for Nafion are shown in Figure 4-18 and Figure 4-19, and the results for 3M-G1-NM samples are shown in Figure From limited experimental data, a summary of a few observations can be made: 1) DMA results of Nafion and 3M-G1-NM showed increase of Tg for ionic clusters, also some changes in beta transition of 3M-G1-NM, and the storage modulus (on left below) from DMA also showed increase values for both ionomers after the degradation. 2) Tensile test result of Nafion showed increased stress and decreased strain at break, possibly in agreement with side chain end group crosslinking in ionic clusters Storage Modulus (MPa) Nafion Degraded Tan Delta Nafion Original 1000 Nafion Original Nafion Degraded Temperature ( 癈 ) Universal V4.1D TA Instruments Temperature ( 癈 ) Universal V4.1D TA Instruments Figure DMA Results of Pristine and Degraded Nafion Samples, (left) Storage Modulus vs Temperature, (right) Tan δ vs Temperature 173

199 25 Nafion Degraded 20 Nafion Original Stress (MPa) Strain (%) Universal V4.1D TA Instruments Figure Tensile Test Results (Measured by DMA) of Pristine and Degraded Nafion Samples Storage Modulus (MPa) M-NM Degraded Tan Delta M-NM Degraded M-NM Original Temperature ( 癈 ) Universal V4.1D TA Instruments Temperature ( 癈 ) Universal V4.1D TA Instruments 3M-NM Original Figure DMA Results of Pristine and Degraded 3M-G1-NM Samples, (left) Storage Modulus vs Temperature, (right) Tan δ vs Temperature 174

200 c. Morphology Comparison and XRD Results No obvious morphological changes were observed before and after degradation. Other than subtle changes at very low scattering angles in XRD experiments (which is very difficult to compare due to the instrument sensitivity at this low angle for current wide angle XRD settings). The data are enclosed in the Appendix 1 for brevity. 4.4 Conclusions The modified Fenton s degradation test developed in this study resulted in greater degree of membrane degradation for Nafion and 3M membranes, which is potentially beneficial to facilitate the ex situ durability assessment and candidate screening in the development of new membranes for PEMFC. The major result generated in this research work is that degradation of Nafion and 3M membranes can also occur via the cleavage of ether bonds on the pendant side chains. The degradation products after the ether cleavage were clearly identified by LC-MS and NMR analyses. This side chain degradation pathway is complementary to the unzipping degradation from the reactive chain ends, and can be exploited to explain the fluoride release from the membrane when the reactive chain ends are greatly minimized in the latest generation of commercial PFSA membranes. The 3M membranes with a higher degree of end group stabilization (BM) generate the least fluoride in the degradation tests, among all membranes studied. The recast Nafion commercial membranes (3M-Nafion) and 3M membranes with the least and a moderate degree of end group modifications (NM and PM), however, showed comparable degradation resistance in the degradation test. Similar magnitude of decrease 175

201 in proton conductivity (which is expected when the side chains are cleaved from the main chain) was observed for all membranes after degradation tests. IR study revealed two features of the degraded membranes: the formation of ketone functionality resulted from the degradation, which may be explained by the proposed ether cleavage mechanistic route of fluoro ethers based on MC systems, and the formation of -S-O-S- crosslinking of side chains. Limited data from DMA and tensile testing may support the proposition of such crosslinking in the ionic cluster. 176

202 References 1. Carlson, D. P., US Patent 3,674, Carlson, D. P.; Kerbow, D. L.; Leck, T. J.; Olson, A. H., US Patent 4,599, Schreyer, R. C., US Patent 3,085, Goodman, J.; Andrews, S., Fluoride Contamination from Fluoropolymers in Semiconductor Manufacture. Solid State Technology 1990, 33, (7), Banks, R. E.; Smart, B. E.; Tatlow, J. C., Organofluorine Chemistry, Principles and Commercial Applications. Plenum Press, New York Zawodzinski, T. A.; Neeman, M.; Sillerud, L. O.; Gottesfeld, S., Determination of Water Diffusion-Coefficients in Perfluorosulfonate Ionomeric Membranes. Journal of Physical Chemistry 1991, 95, (15), Healy, J.; Hayden, C.; Xie, T.; Olson, K.; Waldo, R.; Brundage, M.; Gasteiger, H.; Abbott, J., Aspects of the chemical degradation of PFSA ionomers used in PEM fuel cells. Fuel Cells (Weinheim, Germany) 2005, 5, (2), Tang, H.; Peikang, S.; Jiang, S. P.; Wang, F.; Pan, M., A degradation study of Nafion proton exchange membrane of PEM fuel cells. Journal of Power Sources 2007, 170, (1), Liu, W.; Crum, M., Effective testing matrix for studying membrane durability in PEM fuel cells: part I. Chemical durability. ECS Transactions 2006, 3, (1, Proton Exchange Membrane Fuel Cells 6), Liu, W.; Ruth, K.; Rusch, G., Membrane Durability in PEM Fuel Cells. Journal of New Materials for Electrochemical Systems 2001, 4, (4), Heitner-Wirguin, C., Infra-red spectra of perfluorinated cation-exchanged membranes. Polymer 1979, 20, (3), Zhang, H.; Rankin, A.; Ward, I. M., Determination of the end-group concentration and molecular weight of poly(ethylene naphthalene-2,6- dicarboxylate) using infra-red spectroscopy. Polymer 1996, 37, (7), Wang, Y. Q.; Kawano, Y.; Aubuchon, S. R.; Palmer, R. A., TGA and timedependent FTIR study of dehydrating Nafion-Na membrane. Macromolecules 2003, 36, (4),

203 14. Falk, M., An Infrared Study of Water in Perfluorosulfonate (Nafion) Membranes. Canadian Journal of Chemistry-Revue Canadienne De Chimie 1980, 58, (14), Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G., The handbook of infrared and raman characteristic frequencies of organic molecules. Academic Press, San Diego, CA 1991, Pg Qiao, J. L.; Saito, M.; Hayamizu, K.; Okada, T., Degradation of perfluorinated ionomer membranes for PEM fuel cells during processing with H2O2. Journal of the Electrochemical Society 2006, 153, (6), A967-A Chen, C.; Levitin, G.; Hess, D. W.; Fuller, T. F., XPS investigation of Nafion membrane degradation. Journal of Power Sources 2007, 169, (2),

204 Chapter 5. Overall Summary and Suggested Future Work 5.1 Overall Summary This dissertation research systematically investigated chemical degradation of PFSA ionomers under conditions that mimic the chemically aggressive fuel cell operating conditions. MC systems were selected to impart the analytical capability and determination of reactivity difference among various structural moieties that exist in membranes. The results of MC systems not only verified the proposed degradation route and high reactivity of reactive carboxylic acid end groups in PFSAs, but also revealed a secondary degradation route of PFSAs through the cleavage of the ether bonds. The degradation rate of ether cleavage under radical conditions is one order of magnitude lower then that of carboxylic acid end group. The greater abundance of the ether bonds versus chain ends suggests that this secondary degradation route may contribute significantly to the overall degradation of the polymers. This ether cleavage reaction under the degradation conditions directly answers the secondary degradation route needed to explain the unsolved puzzle in the literature, i.e. the gradual release of fluoride ions and deterioration of membrane properties when a majority of the highly reactive end groups are stabilized. Such an outcome provides important knowledge to future membrane development to meet the durability requirements. The degradation experiments of polymer systems are mutually complementary to those of the MC systems. The identification of degradation products from polymer systems confirmed the proposed ether cleavage reaction observed in MC systems. The 179

205 property changes after the degradation revealed the expected decrease of conductivity, the possible ketone functionality formation, and crosslinking of side chains. During the progress of this dissertation research, a bonus discovery is the identification of the concentration-dependant interference of Fe 3+ species on fluoride concentration measurement using fluoride ISE. This finding may account for some contradictory fluoride generation data in the literature where various concentrations of Fe 3+ were employed to carry out Fenton s degradation test. 5.2 Suggested Future Directions a. MC Systems 1) The fluoride release as a function of degradation time was primarily studied at specific concentrations of the Fenton s reagents (mild condition). Preliminary reaction kinetics of the reaction between MCs and hydroxyl radicals were assessed using a linear fit of fluoride release and degradation time. Slopes obtained were then used to estimate the difference in reactivity between MCs. Further work can be carried out to study fluoride release as a function of H 2 O 2 concentration, MC concentration, and Fe 2+ concentration to obtain useful kinetic data that may be directly fed into the mathematical modeling for the prediction of life time a running fuel cell. 2) The isolation and collection of the degradation products are desired for analysis by various analytical instruments. Attempts to isolate the products in this work were not successful due to extremely low concentration and solubility limitation of some degradation products. 180

206 3) Fluoride generation was observed when MC solutions were exposed to UV irradiation. Further UV degradation experiments of MCs are worth investigating because the knowledge may greatly benefit the post service treatment of fluorinated compounds. 4) The degradation of MCs using aggressive electrochemical conditions has not been pursued in this research. The degradation under these conditions may be very useful to correlate the chemical degradation as a function of specific fuel cell operation conditions, such as at open cell voltage (OCV) condition. 5) The further verification of various intermediates from ether cleavage is highly desired. The proposed structure of the radical cationic initial intermediate needs to be experimentally observed. The most promising techniques for the direct observation of this important intermediate are: electron spin resonance (ESR) spectroscopy and cyclic voltametry (CV) study. The ESR experiment may be used to monitor the formation of this radical cation specie in situ during the degradation test. Whether the proposed radical cationic intermediate may be long-life or not needs to be experimentally determined, possibly through the kinetic analysis of product formation as a result of concentration variation of reactants. If this intermediate is stable enough, CV experiment can provide extremely useful information regarding the electron transfer process of such species, which are critical to fully understand the degradation mechanism. b. Ionomer Systems 1) Solid state 19 F and 13 C NMR experiments for PFSAs have not been conducted. Future work should be directed to carry out these experiments to investigate the end 181

207 group concentration, the side chain population, and the resultant structure changes in more details after the degradation tests. 2) Detailed solid state structure changes after degradation should be carried out to examine the changes in ionic cluster size, shape, and population after the gradual decrease of pendant side chains. High resolution micrographic methods such as transmission electron microscopy (TEM) and atomic force microscopy (AFM) may be used to examine the morphology changes. Additionally, small angle x-ray scattering (SAXS) experiments can also be employed to study the ionic clusters after the degradation. 3) Additional replicates of thermal and mechanical experiments (DSC, DMA, and tensile test) are desired to correlate the property changes with the chemical structure after degradation. c. New Membrane Development This research revealed that fluoroether moieties can be cleaved by hydroxyl radical species that can be formed in the fuel cell operating conditions. A direction for membrane development may be focused primarily on lowering the oxygen permeability across the membrane to prevent the formation of the hydroxyl radical precursor, i.e. hydrogen peroxide, during the fuel cell operation. It is also desirable that the proton conductivity of membrane be maintained or increased, while the oxygen permeation property is optimized. Fluoroether functionality moieties should be minimized in the next generation of PFSA type ionomers. The following three major types of structures may potentially be 182

208 advantageous over current bench-mark Nafion ionomers to meet the durability requirements. PFSAs without ether branch points on the pendant side chains. One example structure is given below. CF 2 CF 2 x CF 2 CF CF 2 y CF 2 n SO 3 H n = 2-5 PFSAs with minimal ether branch points on reasonably long side chains so that they may be confined in highly hydrophobic fluorocarbon region to minimize the contact with hydroxyl radicals that are primarily abundant in the hydrophilic ionic clusters. The representative structure is given below. CF 2 CF 2 x CF 2 CF O y CF 2 CF 2 SO 3 H n n = 5-8 Alternative PFSAs with transition metal ions that are immobilized by additional side chains. These metal ions are incorporated for the following two reasons. They can be considered as hydroxyl radical scavengers because the electron transfer to hydroxyl radicals is much faster than the ether cleavage reactions. One example of the idealized structure is given below. 183

209 = N moieties = Metal ions = -SO 3 H Novel composite membranes with hydrogen peroxide decomposition catalysts, such as magnesium dihydroxide, dispersed and crosslinked in the membrane. Idealized structure is shown below. O Nano Particle Surface = Mg OH + H 3 C Br O Br Triethylamine THF, Ice bath Mg O Br CH 3 = -SO 3 H ATRP initiator 184

210 A.1.1 XRD Results A Nafion Appendix 1. XRD and SEM Results for Pristine and Degraded Membranes Nafion Original Nafion Degraded Intensity (a.u.) θ (Deg) 185

211 A M-G1-NM NM Original NM Degraded Intensity (a.u.) θ (Deg) A M-G1-PM PM Original PM Degraded Intensity (a.u.) θ (Deg) 186

212 A M-G1-BM BM Original BM Degraded Intensity (a.u.) θ (Deg) 187

213 A.1.2 SEM Micrographs A Surface SEM Images a. Nafion Nafion, Original (1000X) Nafion, Degraded (1000X) b. 3M-G1-NM 3M-G1-NM, Original (1000X) 3M-G1-NM, Degraded (1000X) 188

214 c. 3M-G1-PM 3M-G1-PM, Original (1000X) 3M-G1-PM, Degraded (1500X) d. 3M-G1-BM 3M-G1-BM, Original (1000X) 3M-G1-BM, Degraded (1200X) 189

215 A Cross-section SEM Images a-1. Nafion Original Nafion, Original (250X) Nafion, Original (2500X) Nafion, Original (5000X) 190

216 a-2. Nafion Degraded Nafion, Degraded (250X) Nafion, Degraded (2500X) Nafion, Degraded (5000X) 191

217 b-1. 3M-G1-NM Original 3M-G1-NM, Original (250X) 3M-G1-NM, Original (2500X) 3M-G1-NM, Original (5000X) 192

218 b-2. 3M-G1-NM Degraded 3M-G1-NM, Degraded (250X) 3M-G1-NM, Degraded (2500X) 3M-G1-NM, Degraded (5000X) 193

219 c-1. 3M-G1-PM Original 3M-G1-PM, Original (250X) 3M-G1-PM, Original (2500X) 3M-G1-PM, Original (5000X) 194

220 c-2. 3M-G1-PM Degraded 3M-G1-PM, Degraded (250X) 3M-G1-PM, Degraded (2000X) 3M-G1-PM, Degraded (5000X) 195

221 d-1. 3M-G1-BM Original 3M-G1-BM, Original (250X) 3M-G1-BM, Original (2000X) 3M-G1-BM, Original (5000X) 196

222 d-2. 3M-G1-BM Degraded 3M-G1-BM, Degraded (250X) 3M-G1-BM, Degraded (2000X) 3M-G1-BM, Degraded (5000X) 197