UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Novel microporous layers with improved interfacial characteristics for PEM fuel cells Janse van Vuuren, Magrieta Jeanette Leeuwner 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_february_jansevanvuuren_magrietajeanetteleeuwner.pdf [ 14.47MB ]
Metadata
JSON: 24-1.0360782.json
JSON-LD: 24-1.0360782-ld.json
RDF/XML (Pretty): 24-1.0360782-rdf.xml
RDF/JSON: 24-1.0360782-rdf.json
Turtle: 24-1.0360782-turtle.txt
N-Triples: 24-1.0360782-rdf-ntriples.txt
Original Record: 24-1.0360782-source.json
Full Text
24-1.0360782-fulltext.txt
Citation
24-1.0360782.ris

Full Text

NOVEL MICROPOROUS LAYERS WITH IMPROVED INTERFACIAL CHARACTERISTICS FOR PEM FUEL CELLS  by  Magrieta Jeanette Leeuwner Janse van Vuuren  B.Eng., Stellenbosch University, 2007 M.Sc., Stellenbosch University, 2011   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Chemical and Biological Engineering)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  November 2017   © Magrieta Jeanette Leeuwner Janse van Vuuren, 2017  ii  Abstract  High efficiencies and reduced greenhouse gas emissions promote proton exchange membrane fuel cells (PEMFCs) as a promising energy conversion technology. However, its widespread commercial application is hampered by certain cost, performance and durability limitations. The interface between the MPL (microporous layer) and the cathode CL (catalyst layer) plays an important role in a PEMFC’s overall performance, since it houses the reaction sites for the oxygen reduction reaction. The interface may furthermore significantly affect mass transport behavior, ohmic contributions and the hydration state of the membrane at different humidities. The main objective of the study was therefore to advance PEMFC research through the development of alternative MPLs offering dual-functional improvements: enhanced interfacial characteristics and improved operational flexibility via suitability for low cathode humidity applications.   Alternative MPLs were evaluated based on extensive material characterization and single cell performance testing. Graphene was demonstrated to be a promising alternative. The material displays beneficial interfacial characteristics (a stacked planar morphology, superior conductivity, adhesive behaviour, and improved electrical connectivity with the CL) and furthermore results in improvements in the kinetic and ohmic polarization regions, compared to the conventional CB (carbon black) MPL. Although graphene MPLs also suffer from mass transport limitations, the problem can be addressed through the addition of CB. The addition increases the MPL’s water permeability, which helps to establish a balance between water removal (for the prevention of flooding) and water retention (for membrane hydration) at high and low RH (relative humidity). For graphene tested under one-dimensional control, this results in synergistic performance enhancements, showing a 30% and 80% increase in the maximum power density at 100% and 20% cathode RH. In addition to increased water permeability, other common effects resulting from the creation of CB composites (also observed for reduced graphene oxide and graphite) include decreased surface wettability and through-plane resistance.   iii  For the application to low loaded CCMs (0.1 mg cm-2), the potential to improve performance with graphene-based MPLs appears restricted by the catalyst loading itself. Nevertheless, graphene helps to improve performance preservation of low loaded CCMs at low humidity conditions, as also demonstrated for conventionally loaded CCMs (0.4 mg cm-2).     iv  Lay summary  A hydrogen fuel cell is a device that generates power from hydrogen and oxygen gas. Even though fuel cells are considered environmentally friendly, they have to become less expensive and more efficient to promote widespread commercialization. Fuel cells consist of different layers that are sandwiched tightly together. One of these layers, the microporous layer, and the way in which it ‘interfaces’ or contacts another layer, can have a big effect on the device’s performance. By using graphene (a 2-D layer of carbon atoms) to make the microporous layer, the contact with the other layer is improved and the fuel cell becomes more conductive to electricity. These properties help to improve the overall performance and durability of the device. Graphene also enables fuel cells to be more flexible to operate over a wider range of conditions, including both very humid and dry conditions.   v  Preface  The work presented in this thesis, including the literature review, experimental design and execution, data analysis and interpretation, and thesis preparation, was completed by Magrieta Jeanette Leeuwner Janse van Vuuren under the direct supervision of Professors Előd Gyenge and David Wilkinson at the Department of Chemical & Biological Engineering, the University of British Columbia. Three undergraduate student interns assisted with certain experimental aspects, under direct supervision of the author:  Arghya Patra performed certain fuel cell tests relating to Chapter 4;  Alex Wong prepared the microporous layers used in Chapter 5;  Rui Du performed certain characterization tests relating to Chapters 4 and 5.   The publications arising from this work are listed:  A version of Chapter 4 has been published (M. J. Leeuwner, D. P. Wilkinson, E. L. Gyenge, Fuel Cells 2015, 15, 790–801.). The author is solely responsible for the preparation of the publication and its content.  Part of Chapter 5 has been published (A. T. Najafabadi, M. J. Leeuwner, D. P. Wilkinson, E. L. Gyenge, ChemSusChem 2016, 1–10). The publication was prepared together with Amin Taheri  Najafabadi. He was furhtermore responsible for all experimental aspects relating to the preparation of the electrochemically exfoliated graphene and its characterization in raw material form. The author performed all fuel cell tests and characterization relating to the microporous layers. The author is listed as co-first author for this particular publication.   All other work presented in this thesis (remainder of Chapter 5, and Chapter 6) has not yet been published and two publications are currently in preparation for submission: o M.J. Leeuwner, A. Patra, E.L. Gyenge, D.P. Wilkinson, “An experimental investigation into alternative microporous layers: A study of novel and composite materials”, In preparation;  vi  o M.J. Leeuwner, E.L. Gyenge, D.P. Wilkinson “Application of novel microporous layer to high- and low-loaded catalyst coated membranes”, In preparation.  The work has also been presented at the following conferences by the author:  M.J. Leeuwner, E.L Gyenge, D.P Wilkinson, CaRPE-FC 6th Annual Technical Meeting, Vancouver, 2017.  M.J. Leeuwner, E.L Gyenge, D.P Wilkinson, CaRPE-FC 5th Annual Technical Meeting, Vancouver, 2016.  M.J. Leeuwner, E.L. Gyenge, D.P. Wilkinson, CaRPE-FC 4th Annual Technical Meeting, Vancouver, 2015.  M.J. Leeuwner, I. Martens, E.L. Gyenge, D.P. Wilkinson, D. Bizzotto, 97th Canadian Chemistry Conference and Exhibition, Vancouver, 2014.  M.J. Leeuwner, I. Martens, D. Bizzotto, E.L. Gyenge, D.P. Wilkinson, CaRPE-FC 3rd Annual Technical Meeting, Vancouver, 2014.  M.J. Leeuwner, I. Martens, D. Bizzotto, E.L. Gyenge, D.P. Wilkinson, Hydrogen and Fuel Cell Conference, Vancouver, 2013.  M.J. Leeuwner, I. Martens, D. Bizzotto, E.L. Gyenge, D.P. Wilkinson, CaRPE-FC 2nd Annual Technical Meeting, Vancouver, 2013.    vii  Table of Contents  Abstract ................................................................................................................................ ii Lay summary .........................................................................................................................iv Preface .................................................................................................................................. v Table of Contents ................................................................................................................. vii List of Tables ........................................................................................................................ xii List of Figures ...................................................................................................................... xiv List of Symbols .................................................................................................................... xxi List of Abbreviations .......................................................................................................... xxiv Acknowledgements ........................................................................................................... xxvi Dedication ....................................................................................................................... xxviii Chapter 1: Introduction ......................................................................................................... 1 1.1 Motivation ....................................................................................................................... 1 1.2 Objectives........................................................................................................................ 4 1.3 Thesis layout ................................................................................................................... 5 Chapter 2: Literature review .................................................................................................. 8 2.1 Basic operation and structure of a PEMFC ..................................................................... 8 2.2 Thermodynamic and electrochemical aspects ............................................................. 11 2.2.1 The thermodynamic equilibrium .............................................................................. 11 2.2.2 Polarization or performance losses .......................................................................... 13 2.2.2.1 OCV loss ............................................................................................................ 13 2.2.2.2 Kinetic loss or activation overpotential ............................................................ 15 2.2.2.3 Ohmic loss ......................................................................................................... 16 2.2.2.4 Mass transport loss ........................................................................................... 18 2.2.3 Cathode and anode loss contributions ..................................................................... 19 2.2.4 Power output ............................................................................................................ 20 2.3 Operating modes .......................................................................................................... 21 2.4 Transport mechanisms.................................................................................................. 22  viii  2.4.1 Overall transport ....................................................................................................... 22 2.4.2 MPL water management mechanisms ..................................................................... 24 2.5 Requirements of the ideal MPL .................................................................................... 27 2.6 MPL configurations ....................................................................................................... 27 2.7 MPL material and design modifications ....................................................................... 29 2.7.1 Thickness or carbon loading ..................................................................................... 29 2.7.2 Wettability ................................................................................................................ 29 2.7.3 Porous and microstructural characteristics .............................................................. 30 2.7.4 MPL materials (conductive component) .................................................................. 30 2.8 Introduction to graphene ............................................................................................. 33 2.9 Investigation of CL interfacial characteristics ............................................................... 34 Chapter 3: Experimental methods ........................................................................................ 38 3.1 Ex situ characterization methods.................................................................................. 38 3.1.1 Structure and morphology ........................................................................................ 40 3.1.2 Surface roughness ..................................................................................................... 41 3.1.3 Layer thickness .......................................................................................................... 42 3.1.4 Wettability ................................................................................................................ 43 3.1.5 Through-plane resistance ......................................................................................... 45 3.1.6 In-plane (volume) resistivity ..................................................................................... 45 3.1.7 Interfacial contact resistance .................................................................................... 46 3.1.8 Interfacial contact area ............................................................................................. 47 3.1.9 Gas permeability ....................................................................................................... 48 3.1.10 Water permeability ............................................................................................... 49 3.1.11 Porosity ................................................................................................................. 49 3.1.12 Active electrical connectivity ................................................................................ 50 3.1.13 Assessment of compressive effects on MPL properties ....................................... 51 3.2 MPL preparation ........................................................................................................... 51 3.3 MEA assembly ............................................................................................................... 53 3.4 Fuel cell hardware ......................................................................................................... 53  ix  3.5 Fuel cell operation and protocols ................................................................................. 55 3.5.1 Leak testing ............................................................................................................... 55 3.5.2 Start-up (and shut-down) ......................................................................................... 56 3.5.3 Conditioning .............................................................................................................. 57 3.5.4 Performance testing - Polarization ........................................................................... 57 3.5.5 Operating modes ...................................................................................................... 58 Chapter 4: Freestanding MPLs ............................................................................................. 59 4.1 Introduction .................................................................................................................. 59 4.2 MPL materials ............................................................................................................... 61 4.3 Specifics of characterization methods .......................................................................... 63 4.4 Specifics of fuel cell operation and testing ................................................................... 63 4.5 MPL characterization results ........................................................................................ 64 4.6 Performance results ...................................................................................................... 70 4.7 Further evaluation of GN foam MPL for fuel cell application....................................... 76 4.8 Conclusions ................................................................................................................... 79 Chapter 5: GDL-based MPLS and their composite effects ...................................................... 82 5.1 Introduction .................................................................................................................. 82 5.2 MPL materials ............................................................................................................... 84 5.3 Specifics of characterization methods .......................................................................... 87 5.4 Specifics of fuel cell operation and testing ................................................................... 87 5.5 Experimental data and analysis .................................................................................... 90 5.6 Case study 1: Base MPL materials ................................................................................ 95 5.6.1 Characterization results ............................................................................................ 95 5.6.2 Performance results at 100% cathode RH .............................................................. 102 5.6.3 Performance results at 20% cathode RH ................................................................ 105 5.7 Case study 2: Composite GR+CB ................................................................................. 108 5.7.1 Characterization results .......................................................................................... 109 5.7.2 Performance results at 100% cathode RH .............................................................. 111 5.7.3 Performance results at 20% cathode RH ................................................................ 113  x  5.8 Case study 3: Composite RGO+CB .............................................................................. 115 5.8.1 Characterization results .......................................................................................... 115 5.8.2 Performance results at 100% cathode RH .............................................................. 118 5.8.3 Performance results at 20% cathode RH ................................................................ 120 5.9 Case study 4: Composite GN+CB ................................................................................ 122 5.9.1 Characterization results .......................................................................................... 122 5.9.2 Performance results at 100% cathode RH .............................................................. 125 5.9.3 Performance results at 20% cathode RH ................................................................ 127 5.10 Global trends and other considerations ..................................................................... 129 5.10.1 Effect of CB addition on composite properties .................................................. 129 5.10.2 Dominant components of composite properties ............................................... 130 5.10.3 Compression ....................................................................................................... 134 5.10.4 Kinetic behavior .................................................................................................. 135 5.10.5 Ohmic behavior ................................................................................................... 137 5.10.6 Interfacial characteristics .................................................................................... 140 5.10.7 Experimental uncertainty ................................................................................... 143 5.11 Conclusions ................................................................................................................. 144 Chapter 6: Application of graphene-based MPLs to low loaded CCMs .................................. 147 6.1 Introduction ................................................................................................................ 147 6.2 MPL materials ............................................................................................................. 149 6.3 Specifics of characterization methods ........................................................................ 151 6.4 Specifics of fuel cell operation and testing ................................................................. 151 6.5 MPL characterization results ...................................................................................... 152 6.6 Evaluation of CGN (in TP-5 under one-dimensional control) ..................................... 158 6.7 Evaluation of CCM loading (in TP-50 under stoichiometric control) .......................... 161 6.8 Evaluation of longer-term performance (in TP-50 under stoichiometric control) ..... 167 6.9 Evaluation of CCM type (in TP-5 under one-dimensional control) ............................. 170 6.10 Conclusions ................................................................................................................. 175 Chapter 7: Conclusions and future considerations ............................................................... 178  xi  7.1 Significance and contributions .................................................................................... 178 7.2 Conclusions ................................................................................................................. 181 7.3 Recommendations and future considerations ........................................................... 190 References ......................................................................................................................... 194 Appendices ........................................................................................................................ 204 Appendix A: Chapter 3 ............................................................................................................ 204 Appendix B: Chapter 4 ............................................................................................................ 215 Appendix C:  Chapter 5 ........................................................................................................... 216 Appendix D: Chapter 6 ............................................................................................................ 221   xii  List of Tables  Table 2.1: Summary of the literature on CL interfacial characteristics. ....................................... 36 Table 3.1: Ex situ characterization methods used throughout various phases of the study. ...... 38 Table 3.2: General operating conditions of single cell performance tests. .................................. 57 Table 4.1: Typical material properties of the commercial CB and the freestanding MPLs. ......... 62 Table 4.2: Operating conditions of single cell performance tests in TP-5 cell hardware under constant flow control. ................................................................................................................... 63 Table 4.3: Roughness parameters of the commercial CB and the freestanding MPLs. ............... 66 Table 4.4: Kinetic parameters of the commercial CB and the freestanding MPLs at 100% cathode RH and low flow rates (as specified in Table 4.2). ........................................................................ 72 Table 4.5: OCV resistance and range in ohmic drop of the commercial CB and the freestanding MPLs at 100% cathode RH and low flow rates (as specified in Table 4.2). .................................. 73 Table 5.1: Material properties of the base MPL materials. .......................................................... 86 Table 5.2: Operating conditions of single cell performance tests performed in TP-5 cell hardware under one-dimensional control.................................................................................... 88 Table 5.3: Surface properties and thickness of the GDL-based MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets)............................................................................................................................ 92 Table 5.4: Electrical properties of the GDL-based MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets)........................................................................................................................................................ 93 Table 5.5: Transport and structural properties of the GDL-based MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets)............................................................................................................................ 94 Table 5.6: Summary of dominant components on surface properties and thickness, based on average values (the schematic representations are not to scale). ............................................. 132 Table 5.7: Summary of dominant components on electrical properties, based on average values (the schematic representations are not to scale). ...................................................................... 133  xiii  Table 5.8: Summary of dominant components on structural and transport properties, based on average values (the schematic representations are not to scale). ............................................. 134 Table 5.9: Kinetic parameters associated with the different GDL-based MPLs at 100% and 20% cathode RH. ................................................................................................................................. 136 Table 5.10: OCV resistances of the GDL-based MPLs at 100% cathode RH. .............................. 138 Table 6.1: Material properties of CB and CGN in dry powder form (details of GN used in Chapter 5 are also provided for comparison purposes). .......................................................................... 150 Table 6.2: Operating conditions of single cell performance tests performed in TP-50 cell hardware under stoichiometric control. .................................................................................... 151 Table 6.3: Surface properties and thickness of the CB, CGN and CGN+CB MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). ..................................................................................................... 152 Table 6.4: Electrical properties of the CB, CGN and CGN+CB MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets).......................................................................................................................... 153 Table 6.5: Transport and structural properties of the CB, CGN and CGN+CB MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). ..................................................................................................... 153 Table 6.6: Main differences between TP-5 and TP-50 operation. .............................................. 157 Table 6.7: Kinetic parameters associated with the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs at λair = 2 and 100% cathode RH. ..................................... 163 Table 6.8: OCV resistances of the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs. ...................................................................................................................... 165 Table 6.9: Physical properties of the employed commercial CCMs. .......................................... 172 Table 6.10: OCV resistances of the CB and CGN MPLs in combination with the different types of commercial CCMs. ...................................................................................................................... 173   xiv  List of Figures  Figure 1.1: (a) Performance improvement resulting from the incorporation of a MPL; (b) cathode GDL without and (c) with a MPL. A Gore CCM with 0.4 mg Pt cm-2 on both sides was used and Sigracet 25BC was employed as the combined anode GDL+MPL................................... 2 Figure 1.2: Schematic representation of the reported spatial disconnect between a conventional carbon black MPL and electrolessly deposited CCM; SEM image credit – Isaac Martens. ............ 3 Figure 1.3: Schematic representation of the thesis layout, also demonstrating the progression of the work. ......................................................................................................................................... 7 Figure 2.1: (a) Basic PEMFC structure, (b) cathodic components and (c) a triple phase boundary site. .................................................................................................................................................. 9 Figure 2.2: Schematic representation of a typical polarization curve obtained by subtracting voltage losses from the theoretical equilibrium voltage; Ee is the theoretical equilibrium cell voltage, EOCV is the open circuit voltage, and ilim is the limiting current density. ........................ 14 Figure 2.3: Generation of the iR-corrected curve through addition of ohmic losses to the uncorrected polarization curve..................................................................................................... 17 Figure 2.4: Oxygen and water concentration profiles at the PEMFC cathode (not based on real data, only for illustration purposes). ............................................................................................ 19 Figure 2.5: Voltage losses by individual components in a PEMFC obtained using a fuel cell with reference electrodes. .................................................................................................................... 20 Figure 2.6: Generation of a power density curve through multiplication of the voltage and current density. ............................................................................................................................. 21 Figure 2.7: The main water transport processes involved in a PEMFC. ....................................... 23 Figure 2.8: Environmental SEM micrographs showing vapor condensation and liquid water breakthrough from (a) CL, (b) MPL, and (c) GDL. For each, micrographs at three different elapsed times are shown. ............................................................................................................. 26 Figure 2.9: Expanded view of the three different MPL assembly modes employed in the literature. ...................................................................................................................................... 28  xv  Figure 2.10: Timeline of reviewed journal publications on alternative MPL materials spanning from 2001 to 2016 (author’s own publications excluded during investigation period). ............. 32 Figure 2.11: Differences in molecular structure of graphite, reduced graphene oxide and graphene. ...................................................................................................................................... 34 Figure 3.1: Schematic representation of the ex situ characterization methods employed during the study. ...................................................................................................................................... 39 Figure 3.2: Schematic representation of the ex situ characterization methods employed during the study (continued). .................................................................................................................. 40 Figure 3.3: Contact angles and sample images for (a) a hydrophilic surface (reduced graphene oxide MPL), (b) a hydrophobic surface (Johnson Matthey CCM), and (c) a superhydrophobic surface (CB MPL); Young’s relation between the contact angle and surface energy is also illustrated in (a); γsv is the solid surface free energy, γsl is the solid|liquid interfacial free energy, γlv is the liquid surface free energy, and θ is the contact angle. .................................................. 44 Figure 3.4: Illustrations of (a) Wenzel and (b) Cassie-Baxter wetting behaviour on rough surfaces. ........................................................................................................................................ 44 Figure 3.5: (a) Experimental setup for MPL spray deposition illustrating GDL rotation and the raster-like spray pattern; (b) GDL surface before and after spray deposition. ............................ 52 Figure 3.6: (a) Components of an assembled Tandem TP-5 fuel cell and (b) the assembly jig in the fuel cell test station. ............................................................................................................... 54 Figure 3.7: (a) Components of an assembled Tandem TP-50 fuel cell and (b) the assembly jig in the fuel cell test station. ............................................................................................................... 55 Figure 3.8: Simplified process flow diagram of the 2kW Hydrogenics fuel cell test station. ....... 56 Figure 4.1: Images of the baseline commercial CB MPL and the alternative freestanding MPLs; Higher magnification images provided in inserts. ........................................................................ 61 Figure 4.2: The employed protocol for start-up/shut-down stress tests. .................................... 64 Figure 4.3: Optical profilometry images illustrating the surface structure and morphology of the commercial CB and the freestanding MPLs. ................................................................................. 65  xvi  Figure 4.4: Characterization properties of the commercial CB and the freestanding MPLs: (a) wettability, (b) in-plane resistivity, (c) interfacial contact resistance, (d) through-plane resistance, and (e) the percentage of overall contact area between 650 and 827 kPa(g). ......... 67 Figure 4.5: Cross-sectional SEM images illustrating the extent of MPL adhesion to the CCM after fuel cell testing. ............................................................................................................................. 70 Figure 4.6: Fuel cell performance of the commercial CB and the freestanding MPLs at 100% cathode RH and low flow rates (as specified in Table 4.2). .......................................................... 71 Figure 4.7: Association between OCV resistance and MPL thickness. ......................................... 73 Figure 4.8: Effect of compression on kinetic, ohmic and mass transport performance of (a) the commercial CB, (b) perforated SS, (c) perforated GR MPLs, all with 5% PTFE in the GDL, and (d) the GN foam MPL with 0% PTFE in the GDL; Operating conditions of Table 4.2 and low flow rates were applied. ....................................................................................................................... 75 Figure 4.9: Evaluation of the GN foam MPL: (a) effect of PTFE load in the GDL, (b) effect of air flow rate, (c) power density curves, (d) effect of humidity, (e) stress tests  with performance after 150 hours  in insert, and (f) reproducibility. ........................................................................ 77 Figure 4.10: (a) Schematic representation of water management in the conventional CB MPL and (b) the GN foam MPL; (c) Conceptual representation of water transport in the open space between graphene flakes. ............................................................................................................ 78 Figure 5.1: Images of the four base MPL materials in dry powder form; SEM image credits – Blaise Pinaud. ................................................................................................................................ 86 Figure 5.2: Difference in powder volume density of the base MPL materials (illustrated for 200 mg in each vial). ............................................................................................................................ 86 Figure 5.3: Differences in reactant distribution and polarization behaviour for one-dimensional and stoichiometric control. ........................................................................................................... 89 Figure 5.4: Surface images of the base MPLs in (a) uncompressed and (b) compressed states; Higher magnification images are provided in the inserts. ............................................................ 95 Figure 5.5: Radar charts of material properties for the base MPLs in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the base MPLs. .............................................................................................................................. 96  xvii  Figure 5.6: High contrast and low magnification image of the RGO MPL’s surface, illustrating clusters encircled in red. ............................................................................................................... 97 Figure 5.7: Cross-sectional images (compressed state) of the base MPLs, with inserts at higher magnification. ............................................................................................................................... 98 Figure 5.8: Power law dependencies between the through-plane resistance and compression of the base MPLs. .............................................................................................................................. 99 Figure 5.9: Association between in-plane resistivity and thickness for the base MPLs. ............ 100 Figure 5.10: Performance results for the base MPLs at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance. ..... 103 Figure 5.11: Association between ohmic loss and in-plane resistivity (compressed state) for the base MPLs (at 100% cathode RH). .............................................................................................. 104 Figure 5.12: Performance results for the base MPLs at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. ..................................... 107 Figure 5.13: Surface images of the GR+CB MPL in (a) uncompressed and (b) compressed state; (c) Cross-sectional image (compressed state) of the GR+CB MPL; Higher magnification images are provided in the inserts. ......................................................................................................... 109 Figure 5.14: Radar charts of material properties for the GR+CB case study in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the GR+CB case study. ............................................................................................... 110 Figure 5.15: Performance results for the GR+CB case study at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance. ..... 112 Figure 5.16: Performance results for the GR+CB case study at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. ..................................... 114 Figure 5.17: Surface images of the RGO+CB MPL in (a) uncompressed and (b) compressed state; (c) Cross-sectional image (compressed state) of the RGO+CB MPL; Higher magnification images are provided in the inserts. ......................................................................................................... 116  xviii  Figure 5.18: Radar charts of material properties for the RGO+CB case study in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the RGO+CB case study. ............................................................................................ 117 Figure 5.19: Performance results for the RGO+CB case study at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance. .............................................................................................................................. 119 Figure 5.20: Performance results for the RGO+CB case study at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. ..................................... 121 Figure 5.21: Surface images of the GN+CB MPL in (a) uncompressed and (b) compressed state; (c) Cross-sectional image (compressed state) of the GN+CB MPL; Higher magnification images are provided in inserts. ............................................................................................................... 123 Figure 5.22: Radar charts of material properties for the GN+CB case study in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the GN+CB case study. .............................................................................................. 124 Figure 5.23: Performance results for the GN+CB case study at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance. ..... 126 Figure 5.24: Performance results for the GN+CB case study at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. ..................................... 128 Figure 5.25: General effects on composite properties resulting from CB addition in a 1:1 weight ratio (the schematic representations are not to scale). ............................................................. 130 Figure 5.26: Association between apparent Tafel slope and water permeability of the GDL-based MPLs at 100% and 20% cathode RH. ................................................................................ 137 Figure 5.27: (a) Association between OCV resistance at 100% cathode RH and in-plane resistivity (compressed state) for the GDL-based MPLs; (b) Association between OCV resistance at 100% cathode RH and thickness (compressed state) for the GDL-based MPLs. ................................. 138 Figure 5.28: Association between in-plane resistivity (compressed state) and ohmic loss at current densities of 500 mA cm-2, 760 mA cm-2 and 1000 mA cm-2 (at 100% cathode RH)........ 139  xix  Figure 5.29: Schematic illustrating the transfer of a GN MPL from the GDL to the CCM after fuel cell testing. .................................................................................................................................. 142 Figure 5.30: Cross-sectional SEM images demonstrating the extent of adhesion of the GDL-based MPLs to the CCM after fuel cell testing. ........................................................................... 142 Figure 6.1: Breakdown of the  projected system stack cost at 50 000 systems per year, based on 2016 technology[168]. ................................................................................................................... 148 Figure 6.2: Surface images of the CGN and CGN+CB MPLs in (a) uncompressed and (b) compressed states; Higher magnification images are provided in the inserts. ......................... 154 Figure 6.3 Radar charts of material properties for the CB, CGN and CGN+CB MPLs in (a) uncompressed and (b) compressed state; (c) Porosity and (d) water permeability results for the CB, CGN and CGN+CB MPLs. ....................................................................................................... 155 Figure 6.4: Cross-sectional images (compressed state) of the CGN and CGN+CB MPLs; Higher magnification images are provided in the inserts. ..................................................................... 156 Figure 6.5: Association between in-plane resistivity and thickness for the CB, CGN and CGN+CB MPLs. ........................................................................................................................................... 157 Figure 6.6: Performance results of CGN in the TP-5 cell and under one-dimensional flow control: (a) Polarization performance comparison at 100% cathode RH, effect of cathode humidity on (b) polarization performance, (c) ohmic loss and (d) iR-corrected performance; (e) Voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2 and 20% RH. ............................ 159 Figure 6.7: Performance results with JM High and JM Low CCMs in TP-50 cell and λair = 2: (a, b) Polarization performance and (c, d) ohmic loss. ........................................................................ 162 Figure 6.8: (a) Association between OCV resistance and in-plane resistivity (compressed state) for the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs; (b) Association between OCV resistance and thickness (compressed state) for the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs. ........................................ 165 Figure 6.9: Performance results with JM High and JM Low CCMs in TP-50 cell, λair = 4 and 100% cathode RH: (a, b) Polarization performance and (c, d) ohmic loss. .......................................... 167 Figure 6.10: Longer-terms test at 1000 mA cm-2 and 100% and 20% cathode RH for (a, b)  JM High and (c,d) JM Low CCM. ....................................................................................................... 169  xx  Figure 6.11: Sample cross-sectional images demonstrating adhesion of CGN MPL to JM High and JM Low CCMs. ............................................................................................................................. 169 Figure 6.12: Comparative CB and CGN MPL polarization performance and ohmic loss for (a, b) Gore CCM, (c, d) JM High CCM and (e, f) JM Low CCM in TP-5 cell under one-dimensional flow control and at 100% cathode RH. ............................................................................................... 171 Figure 6.13: Interfacial resistances between the CB, CGN and CGN+CB MPLs and the different types of commercial CCM. .......................................................................................................... 174 Figure 7.1: Schematic overview of the thesis, highlighting the areas of novelty and contributions of the research. ........................................................................................................................... 180   xxi  List of Symbols  Greek symbols Symbol   Description       Unit 𝛼    Kinetic transfer coefficient Γ     Charge required to reduce a monolayer of protons  C cm-2Pt 𝛾     Surface tension (in capillary)     N m-1 𝛾𝑙𝑣     Liquid surface free energy     N m-1 𝛾𝑠𝑣     Solid surface free energy     N m-1 𝛾𝑠𝑙     Solid|liquid interfacial free energy    N m-1 𝜀     Porosity       % 𝜂𝑘𝑖𝑛    Kinetic loss / activation overpotential   V 𝜂𝑀𝑇    Mass transport loss / Concentration overpotential V 𝜂𝑜ℎ𝑚    Ohmic loss      V 𝜃     Contact angle       ° 𝜆𝑗   Stoichiometric ratio of species j  𝜇𝑓     Fluid viscosity       Pa s 𝜌   In-plane or volume resistivity     mΩ cm 𝜌𝑙𝑖𝑞   Density of wetting liquid     kg m-3 σ Standard deviation of property value 𝜏     Sample thickness      cm 𝜏𝑃𝑀     Tortuosity of porous media       𝜐𝑗    Stoichiometric coefficient of species  ∅     Sample diameter      m  Roman symbols Symbol   Description      Unit 𝐴      Geometric electrode area     cm2 𝑎 (subscript)   Property relating to anode 𝑎𝑗    Activity of species j 𝑏    Tafel slope      V dec-1  xxii  𝑐 (subscript)   Property relating to cathode 𝐸°    Equilibrium cell potential at standard conditions  V 𝐸𝑒    Equilibrium cell potential, non-standard conditions V Ecorrected  iR-corrected cell potential     V 𝐸𝐶𝑆𝐴      Electrochemically active surface area    cm2Pt gPt-1 𝐹    Faraday’s constant     C mol-1 ∆𝐺    Change in Gibbs free energy, non-standard conditions J mol-1 ∆𝐺°    Change in Gibbs free energy, standard conditions J mol-1 𝑖    Total current density     A cm-2 𝑖𝑙𝑖𝑚    Limiting current density     A cm-2 𝑖0    Exchange current density    A cm-2 𝑘     Geometrical correction factor (in-plane resistivity) 𝑘 (subscript)   Index in x direction 𝑘𝑓     Fluid permeability      m2 or darcy 𝑙 (subscript)   Index in y direction 𝑀      Number of data points in x direction ∆𝑚     Weight difference resulting from liquid saturation  kg 𝑁     Number of data points in y direction 𝑁𝑗     Moles of specie j     moles 𝑛    Number of electrons transferred ∆𝑃𝑐𝑎𝑝    Capillary pressure across fluid interface   Pa ∆𝑃     Pressure drop across sample (permeability)   Pa Pcell     Power output of fuel cell    W cm-2 𝐿      Pt loading       gPt cm-2electrode  𝑅    Universal gas constant     J mol-1 K-1 𝑅𝑎    Average surface roughness    μm 𝑅𝑝𝑜𝑟𝑒    Radius of pore/capillary tube     m 𝑅𝑞    Root mean square surface roughness   μm 𝑅𝑠     Theoretical sheet resistance     mΩ square-1 𝑅𝑥     Bulk resistance of component x    Ω 𝑅𝑥|𝑦     Interfacial resistance of components x and y  Ω  xxiii  ∆𝑆°    Change in entropy, standard conditions   J mol-1 K-1 𝑇    Temperature      K 𝑡    Time       s 𝑞𝑃𝑡     Charge density       C cm-2electrode ?̇?      Fluid volumetric flow rate     m3 s-1  𝑉𝑝𝑜𝑟𝑒     Pore volume of layer      m3 𝑉𝑡𝑜𝑡𝑎𝑙     Total volume of layer      m3 𝑣𝑓      Fluid velocity       m s-1 𝑊    Electrical work      J mol-1 𝑧     Surface height relative to reference mean plane  μm  𝑧𝑚𝑎𝑥     Maximum average property value in dataset 𝑧𝑚𝑖𝑛     Minimum average property value in dataset 𝑧𝑛𝑜𝑟𝑚     Normalized value (in the 0 to 1 range)     xxiv  List of Abbreviations  BET  Brunauer-Emmett-Teller CB  Carbon black CCM  Catalyst coated membrane CE  Counter electrode CFP  Carbon fiber paper CGN  Company manufactured graphene CL  Catalyst layer CV  Cyclic voltammetry CVD  Chemical vapor deposition ECSA  Electrochemically active surface area EIS  Electrochemical impedance spectroscopy FF  Flow field FFP  Flow field plate GDE  Gas diffusion electrode GDL  Gas diffusion layer GR  Graphite GN  Graphene Hupd  Hydrogen underpotential deposition HOR  Hydrogen oxidation reaction HFR  High frequency resistance MEA  Membrane electrode assembly MPL  Microporous layer OCV  Open circuit voltage ORR  Oxygen reduction reaction PEM  Proton exchange membrane PEMFC  Proton exchange membrane fuel cell PMMA  Polymethyl methacrylate  xxv  PTFE  Polytetrafluoroethylene RE  Reference electrode RGO  Reduced graphene oxide RH  Relative humidity SEM  Scanning electron microscopy SS  Stainless steel TPB  Triple phase boundary WE  Working electrode XPS  X-ray photoelectron spectroscopy   xxvi  Acknowledgements  I would like to express my sincere gratitude towards the following persons for their contributions: To my supervisors, Professors Előd Gyenge and David Wilkinson, for affording me the opportunity to pursue my PhD and for your valuable technical guidance and continuous support throughout. To Professors Fariborz Taghipour and Dan Bizzotto, for serving as my committee members and providing valuable technical input. To the Catalysis Research for Polymer Electrolyte Fuel Cells Network, for providing the funding for this research. To Professor Boris Stoeber and Iman Mansoor for providing access to an optical profilometer; to Professor Dan Bizzotto and Nidal Alswareh for providing access to the Jandel four-point probe setup; to Professors Peter Englezos and Savvas Hatzikiriakos and Mehr Negar Mirvakili for providing access to the contact angle measurement system; to Johnson Matthey for providing catalyst coated membranes; to Amin Taheri Najafabadi for producing the electrochemically exfoliated graphene; and to NanoXplore for also providing graphene. The equipment and materials were invaluable to the research and are greatly appreciated. To those that provided technical assistance, Greenlight Innovation, Winton Li, Greg Afonso, Isaac Martens, Saad Dara, Arman Bonakdapour, Macarena Cataldo, Blaise Pinaud, Gethin Owen and Guobin Sun, for sharing your experience and insight. To my helping hands, Arghya, Alex, Rui and Kimia, for your hard work, assisting with countless measurements and tests, and your wonderful company in the laboratory. To all my great friends in CHBE and Labs 618, 608 and 149, for your camaraderie and friendship. Melissa joon, thank you especially for all the laughs and wonderful times in our office!  To the Van Wyks and the Fantastic Five, for opening your home to us, introducing us to ‘Canadian life’, for pizza-nights, and your support.  xxvii  To Nola, Jantjie, Kim and Zara, the Van Vuuren clan, and the rest of my extended family and friends back in South Africa, for your love, and supporting me in this endeavour. To my most remarkable parents, Johan and Magrieta, for your unconditional love, and for your constant cheering and support that can be felt over 16 000 km away. I love you dearly. To my wonderful husband, Frankie, whom I love so very much. This accomplishment is as much yours as it is mine (you can have the ‘Ph’ and I will take the ‘D’). Thank you for your joy and humour, and that you always dance with me throughout life, no matter what tune is playing. To my Saviour, for Your love, grace and enduring hope. And all the amazing wonders that I continue to discover in life, through others, science and nature.  xxviii  Dedication           Aan Frankie, Ma en Pa, en al my geliefdes To Frankie, Mom and Dad, and all my loved ones  Chapter 1: Introduction     1  Chapter 1:  Introduction  1.1 Motivation  Under the ever-increasing demand for clean and sustainable energy resources, fuel cells are considered a promising option for energy conversion due to their high efficiencies and environmentally friendly characteristics. In particular, proton exchange membrane fuel cells (PEMFCs) prove very advantageous because of their associated low or zero harmful emissions, high power density, low operating temperature, quick start-up and low noise levels. In spite of these favorable features though, PEMFCs have not found widespread commercial application yet, since some costs and performance limitations still need to be optimized[1–3].  In recent years, the microporous layer (sometimes also referred to sub-layer or intermediate layer) has been established as a key component in PEMFCs due to the beneficial influence on performance. Improvements are attributed to different factors and usually depend on the MEA (membrane electrode assembly) design and operating conditions. As shown in Figure 1.1, MPLs (microporous layers) typically help to reduce mass transport limitations through improved water removal and an accompanying reduction in flooding (blocking of reactive sites through liquid water). Ohmic losses are further reduced through improved interfacial contact with the CL (catalyst layer) and/or increased membrane hydration. Finally, the MPL also plays a vital role in maintaining the mechanical integrity of the MEA by acting as a buffer region which provides support and prevents damage of the GDL (gas diffusion layer) to the CL[4–6].  Although the MPL presents many benefits, its interface with the CL may still serve as a source of ohmic and mass transport losses, due to imperfect contact and differences in layer properties[7,8]. This is of particular concern for the cathodic catalyst interface which houses the reaction sites for the oxygen reduction reaction, which is also the limiting reaction in PEMFCs. A better understanding of the factors influencing the MPL|CL interfacial characteristics, could therefore allow tailoring of the MPL to enhance performance even further. The concept also applies to low loaded CCMs (catalyst coated membranes), where the trend is to decrease the Pt Chapter 1: Introduction     2  loading to 0.1 mg cm-2 and even lower[9–12]. Other CCMs with novel structures, such as the electrolessly deposited Pt of Sode et al.[13], reportedly also suffered from insufficient electrical connectivity, due to the spatial disconnect between nano-sized Pt particles and the micro-scaled features of conventional MPLs (schematic representation provided in Figure 1.2). To ensure maximum utilization of CCMs with low Pt loading and/or novel structures, it therefore becomes even more important to establish as many electrical connections to active Pt sites as possible.    Figure 1.1: (a) Performance improvement resulting from the incorporation of a MPL; (b) cathode GDL without and (c) with a MPL. A Gore CCM with 0.4 mg Pt cm-2 on both sides was used and Sigracet 25BC was employed as the combined anode GDL+MPL.  Chapter 1: Introduction     3   Figure 1.2: Schematic representation of the reported spatial disconnect between a conventional carbon black MPL and electrolessly deposited CCM; SEM image credit – Isaac Martens.  While numerous studies have focused on the incorporation of different MPL materials, composition and structures[6], the general emphasis lies on the MPL’s overall effect on mass transport and performance, with very little work focusing on the MPL|CL interface in more detail. Of the limited studies related to the interface, the majority of the investigations are based on modelling approaches[7,8,14]. These interfacial studies are also typically limited to conventional CB (carbon black) MPLs without any consideration of alternative materials and/or morphologies. Therefore, there remains a need to pay closer attention to MPL|CL interfacial aspects during the experimental investigation of alternative MPLs.  In addition to minimizing performance losses, an equally strong driving force in PEMFC research is to increase operational flexibility and durability.  In this context, operation at low humidity conditions, especially on the cathode side, is of particular interest[15–17]. PEMFC operation requires that the PEM (proton exchange membrane) and ionomer in the CL is well hydrated, to enable the transport of protons as reactive species. To ensure proper hydration, the reactant feed streams are therefore typically humidified. The use of humidifiers, does however incur additional costs and parasitic losses[16,18,19]. Various approaches have been taken to reduce or eliminate the use of humidifiers. Amongst others, these methods include employing alternative self-humidifying membranes[20–22], including hygroscopic (water absorbing) agents to the CL[23–25] or adding separate hygroscopic/hydrophilic layers[18,26]. However, this problem may Chapter 1: Introduction     4  potentially also be addressed by employing alternative MPL materials[17,27], given the component’s significant influence on the PEMFC’s water management. The creation of an alternative cathode MPL offering dual functional enhancements (improved interfacial characteristic and suitability for low cathode humidity conditions), is furthermore very attractive from a design and manufacturing perspective.  1.2 Objectives  The main objective of this study was to enhance fuel cell performance by integration of alternative cathode MPLs with improved MPL|CL interfacial characteristics. This necessitates an enhanced understanding of interfacial factors. Furthermore, special attention was paid to how electrical connectivity might be improved. In conjunction to improved interfacial characteristics, the alternative MPL should increase operational flexibility by facilitating sufficient or enhanced mass transport at both high and low cathode humidity conditions. In support of the aforementioned, the following secondary objectives were carried out:    A critical literature review of PEMFC fundamentals, MPL design modifications and CL interfacial studies;  Identification of key factors that influence the MPL|CL interface;  Exploration of novel and promising materials as alternatives to conventional MPLs with improved interfacial characteristics and enhanced operational flexibility;  Development and assessment of characterization techniques for the investigation of MPL properties;  Evaluation of the effect of alternative MPLs on overall performance and kinetic, ohmic and mass transport losses;  Exploration of the feasibility of implementing alternative MPLs at larger scale;  Investigation of the application of alternative MPLs to low loaded CCMs.   Chapter 1: Introduction     5  1.3 Thesis layout  The various objectives are addressed throughout the thesis as described in the following section. The general layout and progression in the work are depicted in Figure 1.3.   Chapter 2 presents the fundamental theory relating to the thesis content and focuses on aspects such as PEMFC operation and typical performance behaviour. The MPL is also discussed in more detail with regards to its primary functions, transport mechanisms and different configurations. Reviews of previous work on MPL modifications and CL interfacial studies are also provided.    Chapter 3 describes the experimental aspects involved with MPL preparation, characterization methods, cell assembly and hardware, and common testing protocols employed throughout the study.  The initial investigation, detailed in Chapter 4, focuses on gaining an improved understanding of MPL|CL interfacial factors. The freestanding MPL configuration is employed, and incorporates three commercial alternatives: perforated stainless steel; perforated graphitic sheet and graphene foam. An initial suite of characterization techniques, used to measure various MPL bulk and interfacial properties, is also employed. Due to material constraints, performance testing is limited to a small scale fuel cell (5 cm2 active area) operated under constant flow control. A paper on this work has been published in Fuel Cells (2015).  Chapter 5 presents a more fundamental assessment of alternative materials prepared as GDL-based MPLs, to enable precise engineering of the MPL load and wetproofing content. Three alternatives are considered: commercial graphite, commercial reduced graphene oxide and electrochemically exfoliated graphene (produced at lab-scale). The creation of CB (carbon black) composites, is also assessed as another means to engineer MPLs. The chapter details the implementation of more robust methods to analyze performance differences, such as: including characterization methods for improved interpretation of mass transport behaviour, evaluating the effect of compression on Chapter 1: Introduction     6  MPL properties in more detail, employing one-dimensional flow control (using a 5 cm2 active area) and incorporating longer-term testing at low cathode humidity. The work in this chapter has resulted in one publication in ChemSusChem (2016) thus far, while another paper is currently being prepared for submission.  Chapter 6 assesses a company manufactured graphene, also produced through an electrochemical method, and specifically evaluates its application at larger scale (49 cm2 active area) and under different operating conditions (stoichiometric flow control). At the same time, the application of the MPLs to low loaded CCMs is also evaluated. A paper on this work is currently being prepared for submission.   The significance of the research, main conclusions and recommendations of the thesis are summarized in Chapter 7.  Appendix A, B, C and D provide additional experimental information for Chapters 3, 4, 5 and 6 respectively.  Chapter 1: Introduction     7    Figure 1.3: Schematic representation of the thesis layout, also demonstrating the progression of the work.Chapter 2: Literature review     8  Chapter 2: Literature review  The fundamental theory relating to PEMFCs is presented and includes an overview of the basic operation and structure, thermodynamic and electrochemical aspects, transport mechanisms and MPL particulars. Previous studies on MPL modifications are also reported. A review of alternative conductive MPL materials, in particular, show considerable consideration of carbon blacks and carbon nanotubes, while the absence of graphene-based materials is highlighted. Reference is also made to instances where alternative materials (typically more hydrophilic) improved performance preservation at low humidity. A review of CL interfacial studies furthermore reiterates the need to consider interfacial factors such as compression, interfacial resistance and layer morphology, during the investigation of alternative MPLs.  2.1 Basic operation and structure of a PEMFC  A fuel cell is an electrochemical device which converts the chemical energy of a fuel and oxidant to electrical energy, water and heat. The energy is utilized through physical separation of the redox reactions, thereby resulting in two half-cell reactions each taking place at a different electrode: the reduction reaction (electron consumption) at the cathode and the oxidation reaction (release of electrons) at the anode. In a PEMFC, hydrogen is commonly used as fuel and oxygen as oxidant. The two species react to yield the following reactions:  Hydrogen oxidation half-cell reaction at anode H2 →2H+ + 2e-, E0 = 0.00 V  Equation 2.1 Oxygen reduction half-cell reaction at cathode ½O2 + 2H+ + 2e- → H2O, E0 = 1.23 V Equation 2.2 Overall redox reaction ½O2 + H2 → H2O, E0 = 1.23 V Equation 2.3  Chapter 2: Literature review     9  The basic operation and structure of a PEMFC is shown in Figure 2.1 (a). As illustrated, the two electrodes, a cathode and an anode, are physically separated by a PEM (proton exchange membrane). Typically Nafioni membranes are employed as PEMs. At the anode, the fuel (hydrogen) is fed through the FFP (flow field plate) to a GDL. This GDL generally consists of woven carbon cloth or non-woven CFP (carbon fiber paper) and helps to evenly diffuse the hydrogen towards the CL. Usually the GDL is also treated with PTFE to introduce hydrophobicity. As mentioned in Section 1.1, it has become common practice to also include a MPL, due to the beneficial features it introduces. The MPL typically consists of a porous CB layer, containing PTFE, deposited directly onto the GDL. The CL itself consists of a combination of catalyst particles (usually Pt), supported on carbon particles and ionomer (usually Nafion) which acts as a binding agent. The hydrogen diffuses into the void spaces and/or dissolves into ionomer to reach the catalyst particles where it reacts according to the HOR (hydrogen oxidation reaction) depicted in Figure 2.1 (a).   Figure 2.1: (a) Basic PEMFC structure, (b) cathodic components and (c) a triple phase boundary site.                                                       i Nafion is the commercial name of a perfluorosulfonic acid membrane developed by DuPont. It has a polytetrafluoroethylene (PTFE or Teflon) backbone structure, terminating in sulfonic acid functional groups which provide the charge sites for proton transport. In the MEA it serves as both the electrolyte and separator (since it also prevents electron transport). Its adoption as a PEMFC component from 1970 onwards has led to significant technological improvements[3]. Nafion is also classified as an ionomer – a polymer which contains electrically neutral repeat units and a small amount of ionic-containing repeat units. Chapter 2: Literature review     10  The electrons released through the HOR are transported along electronically conductive pathways from the catalyst particle (and its support), through the anode MPL and GDL, to the anode FFP. From here, the electrons flow through an external circuit to the cathode, performing electrical work in the process.  Similarly, the electrons on the cathode side are transported from the cathode FFP to the CL. The protons are transported from the ionomer surrounding the anodic reaction sites, through the electrically insulating PEM, to the ionomer in the cathode CL. During this process, water is transported through electro-osmotic drag. The transfer of protons is consequently highly dependent on the water content of the PEM. To prevent membrane dehydration, the hydrogen and oxygen streams are most often humidified. However, excessive amounts of water can also block (or flood) the reaction sites from the reactant gases. A fine balance therefore exists in the water management of the PEMFC[28–30].  On the cathode side, oxygen (usually in the form of air) is fed into the gas flow channels in the FFP and then diffuses through the GDL and MPL to the CL. Here the oxygen diffuses into the void spaces and/or dissolves into the ionomer (or water) to reach the catalyst particles and react with the transported electrons and protons according to the ORR (oxygen reduction reaction) (Figure 2.1 (a) and (b)). Both the ORR and HOR are therefore constrained to spatial sites, referred to as the TPB (triple phase boundary), where protons, reactant and electrically connected catalysts particles contact (shown in Figure 2.1 (c) for ORR). Since the ORR is sluggish and acts as the limiting reaction, the maximization of its TPB sites is a major focus in PEMFC development[31].   The water resulting from the ORR is absorbed into the ionomer next to the catalyst particle. Upon reaching a void space in the cathode CL, the water is desorbed into either water vapor or liquid water, depending on the relative humidity[32]. Both the GDL and MPL should therefore be able to effectively manage this two-phase flow. To support this role, hydrophobic treatment is typically applied to the GDL and MPL, introducing pore spaces with a dual wetting structure: hydrophobic regions act as a pathway for gas transport and the remaining hydrophilic regions facilitate liquid water transport[33]. The MPL is further believed to help prevent flooding by Chapter 2: Literature review     11  wicking liquid water away from the CL through capillary action[34,35]. Water exits through the gas flow channels where it is removed as part of the oxygen (or air) stream. The literature also proposes other mechanisms by which the MPL facilitates mass transport and improves overall performance. The main water transport processes and proposed mechanisms are briefly summarized in Section 2.4.  Together, the assembly of the GDLs, MPLs, CLs and PEM are referred to as the MEA (membrane electrode assembly). There are two methods by which the CLs are typically prepared: the first is by coating the catalyst onto the GDL or MPL and is typically referred to as a GDE (gas diffusion electrode); the second is by depositing the catalyst directly onto the membrane to form what is commonly referred to as a CCM. Currently, the CCM method is the most widely used due to the advantages it holds over the former method. It offers the desired features of lower Pt loadings, higher utilization efficiency and an improved CL|PEM interface[36,37]. The CCM configuration will therefore also be employed in this study.  2.2 Thermodynamic and electrochemical aspects  2.2.1 The thermodynamic equilibrium  For an electrochemical power source, such as a PEMFC, the thermodynamic maximum electrical work that can be produced is equal to the change in Gibbs free energy of the reaction (at constant pressure and temperature) as presented in Equation 2.4. The Gibbs free energy can furthermore be related to the PEMFC’s equilibrium potential, taking into account the number of electrons transferred during the electrochemical reaction (Equation 2.5).  In Equation 2.5, the standard values for the cell potential and Gibbs free energy are represented, and designated by ‘˚’. Standard conditions are specified at a pressure of 101.325 kPa (or 1 atm) and species activities of 1. Measurements are generally taken at a temperature of 298 K (or 25°C).  For the formation of water (Equation 2.3) in liquid phase, ∆𝐺°= 237.1 kJ mol-1, while 2 electrons are transferred. Chapter 2: Literature review     12  𝑊 =  −∆𝐺 Equation 2.4  𝐸° =−∆𝐺°𝑛𝐹 =−(−237100)2 ×  96485 = 1.23 𝑉 Equation 2.5   where  𝑊 = electrical work (J mol-1)  ∆𝐺° = change in Gibbs free energy, at standard conditions (J mol-1)  𝐸° = equilibrium cell potential at standard conditions (V)  𝑛 = number of electrons transferred  𝐹 = Faraday’s constant = 96485 C mol-1  The standard equilibrium potential can furthermore be expressed as the difference between the cathode and anode half-cell potential (specified in Section 2.1). The subscripts ‘𝑐’ and ‘𝑎’ are used to refer to cathodic an anodic terms throughout the text.  𝐸° =  𝐸𝑐°  − 𝐸𝑎°  =  1.23 −  0.00 =  1.23 𝑉 Equation 2.6  If the reaction takes place at non-standard conditions, some corrections need to be made. The temperature is corrected by adjusting the Gibbs free energy according to Equation 2.7. To correct for non-ideal specie activitiesii (𝑎𝑗 ≠  1), the Nernst equation is used (Equation 2.8).   ∆𝐺 =  ∆𝐺° − ∫ ∆𝑆°𝑑𝑇 Equation 2.7                                                      ii Activity is essentially the ‘effective’ concentration of a species. The distinction between activity and other measures of composition (such as concentration) is made based on the fact that molecules in non-ideal gases (at high pressure) and non-ideal solutions (at high concentrations) interact with each other through repulsion or attraction, thereby changing the ‘effective’ concentration. Chapter 2: Literature review     13  𝐸𝑒 =−∆𝐺𝑛𝐹−𝑅𝑇𝑛𝐹𝑙𝑛 (∏ 𝑎𝑝𝑟𝑜𝑑𝑢𝑐𝑡,𝑗𝜐𝑗𝑗∏ 𝑎𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡,𝑗𝜐𝑗𝑗) Equation 2.8   where  ∆𝐺 = change in Gibbs energy, non-standard conditions (J mol-1)  ∆𝑆° = change in entropyiii, at standard conditions (J mol-1 K-1)  𝐸𝑒 = equilibrium cell potential, non-standard conditions (V)  𝑅 = universal gas constant = 8.314 J mol-1 K-1  𝑇 = temperature (K)  𝑎𝑗 = activity of species j  𝜐𝑗 = stoichiometric coefficient of species j   2.2.2 Polarization or performance losses  Electrochemical polarization refers to the decrease in the electrode potential/voltage from the ideal equilibrium value. These losses (also referred to as overpotentials) are captured in polarization or performance curves which plot the change in voltage as a function of current (or current density). Polarization curves also serve as the most common tool to evaluate fuel cells’ performance. The four losses associated with fuel cell performance[38] are presented below and visually depicted in Figure 2.2:  2.2.2.1 OCV loss  The voltage associated with a zero current discharge is termed the OCV (open circuit voltage) and represented as 𝐸𝑂𝐶𝑉. Theoretically, the thermodynamically determined equilibrium voltage can be achieved as the current approaches zero. In practice, the OCV is always lower than the                                                      iii Entropy is a quantitative measure of the amount of thermal energy in a closed thermodynamic system that is unavailable for conversion to useful work.  Chapter 2: Literature review     14  equilibrium voltage, though. This decrease is primarily attributed to a phenomena referred to as fuel crossover. Since the PEM is not necessarily perfectly impermeable to H2, some of the gas can ‘cross over’ to the cathode side to react with O2 directly. The lower potential resulting from fuel crossover is also commonly referred to as a mixed potential. Other factors that can contribute to a lower OCV are impurities, leakage and internal ‘short-circuit’ currents (when small currents cross the PEM).   Figure 2.2: Schematic representation of a typical polarization curve obtained by subtracting voltage losses from the theoretical equilibrium voltage; Ee is the theoretical equilibrium cell voltage, EOCV is the open circuit voltage, and ilim is the limiting current density. Chapter 2: Literature review     15  2.2.2.2 Kinetic loss or activation overpotential  Performance in the low current density region of a polarization curve (also referred to as the kinetic region) is dominated by kinetic loss. This loss refers to the energetic penalty that is incurred to ‘drive’ the chemical reactions and is estimated by the Butler-Volmer equation:  𝑖 = 𝑖0 [𝑒(𝛼𝑎𝐹𝑅𝑇 𝜂𝑘𝑖𝑛) − 𝑒(− 𝛼𝑐𝐹𝑅𝑇 𝜂𝑘𝑖𝑛)] Equation 2.9    where  𝑖 = total current density (A cm-2)  𝑖0 = exchange current density (A cm-2)   𝛼 = kinetic transfer coefficient  𝜂𝑘𝑖𝑛 = kinetic loss or activation overpotential (V)  The kinetic loss is also sometimes referred to as the kinetic/surface/activation overpotential. The exchange current density represents the rate of the reaction at equilibrium. In turn, the kinetic transfer coefficient (𝛼) is a dimensionless quantity that signifies the fraction of the potential at the electrode|electrolyte interface used to lower the free energy barrier so that the electrochemical reaction can take place[39]. As illustrated in Equation 2.9, it is also common practice to normalize the total current (𝐼) based on the geometric active area (𝐴) and consequently express it as current density (𝑖 =𝐼𝐴). For the case where kinetic loss is high, and either the cathodic or anodic exponential term of Equation 2.9 dominates, the simplified Tafel equations can be employed (Equation 2.10 and Equation 2.11). Per the convention followed in this text, cathodic currents are negative, anodic currents positive and the net total current is positive.     Chapter 2: Literature review     16  𝜂𝑘𝑖𝑛,𝑎 =  𝑏𝑎 log (𝑖𝑎𝑖0,𝑎)  Equation 2.10  𝜂𝑘𝑖𝑛,𝑐 =  −𝑏𝑐 log (𝑖𝑐𝑖0,𝑐) Equation 2.11    where  𝑏 =2.3𝑅𝑇𝛼𝐹 = Tafel slope (V decade-1)  2.2.2.3 Ohmic loss  The intermediate, linear region of the polarization curve is dominated by ohmic loss and is aptly referred to as the ohmic region. The ohmic loss is associated with cell resistance and includes the bulk resistances of the electronically conducting components, the PEM and CL ionomer (ionic-conducting components), as well as all interfacial resistances. The ohmic loss is represented by Ohm’s law (Equation 2.12). The main contributor to the cell resistance is associated with the ionic resistance of the PEM and CL ionomer. This resistance furthermore increases as the membrane and ionomer becomes dehydrated. This combined membrane and CL ionomer resistance is designated as RPEM in Equation 2.12. With regards to resistive behavior and hydration state, the word ‘membrane’ will also be used in the remainder of the text to refer to the combined membrane and CL ionomer effects.   𝜂𝑜ℎ𝑚 = 𝑖𝑅𝑐𝑒𝑙𝑙            =  𝑖 ⌈(𝑅𝐹𝐹𝑃 + 𝑅(𝐹𝐹𝑃|𝐺𝐷𝐿) + 𝑅𝐺𝐷𝐿 + 𝑅(𝐺𝐷𝐿|𝑀𝑃𝐿) + 𝑅𝑀𝑃𝐿 + 𝑅(𝑀𝑃𝐿|𝐶𝐿) + 𝑅(𝐶𝐿|𝑃𝐸𝑀))𝑐 +                    (𝑅𝐹𝐹𝑃 + 𝑅(𝐹𝐹𝑃|𝐺𝐷𝐿) + 𝑅𝐺𝐷𝐿 + 𝑅(𝐺𝐷𝐿|𝑀𝑃𝐿) + 𝑅𝑀𝑃𝐿 + 𝑅(𝑀𝑃𝐿|𝐶𝐿) + 𝑅(𝐶𝐿|𝑃𝐸𝑀))𝑎 +                                                                                𝑅𝑃𝐸𝑀⌉  Equation 2.12   Chapter 2: Literature review     17    where   𝜂𝑜ℎ𝑚= ohmic loss (V)   𝑅𝑥 = bulk resistance of component x (Ω)   𝑅𝑥|𝑦 = interfacial resistance of components x and y (Ω)  Typically, resistance values are normalized to the MEA’s geometric area, and expressed as the area-specific resistance in Ω cm2. To obtain such area-independent resistance, which allows fuel cells of different sizes to be compared, the resistance needs to be multiplied by the geometric MEA area. Sometimes it is also desirable to consider fuel cell performance in the absence of ohmic losses. This is typically done by constructing an ‘iR-corrected’ curve (Figure 2.3), where the effect of ohmic loss is omitted by adding it to the original cell voltage.  Ecorrected = Ecell + iRcell Equation 2.13   where   Ecorrected = iR-corrected cell potential (V)   Figure 2.3: Generation of the iR-corrected curve through addition of ohmic losses to the uncorrected polarization curve.   Chapter 2: Literature review     18  2.2.2.4 Mass transport loss  Species concentration gradients exist within the MEA due to the consumption of reactants and the formation of products at the CL (refer to example in Figure 2.4). At some point in the higher current density region, the consumption and formation rates exceed the speed at which the reactant and product can be transported. The phenomenon is especially prevalent at the cathode side where water is also produced: if liquid water is removed at an insufficient rate, the CL floods, hindering oxygen diffusion to the active Pt sites. The loss associated with the depletion/accumulation of species is termed the mass transport loss. The corresponding section of the polarization curve is also designated as the mass transport region (Figure 2.2). In literature, the mass transport loss is sometimes referred to as the diffusion or concentration overpotential. The limiting current density, 𝑖𝑙𝑖𝑚, corresponds to the maximum current density that is achieved when the concentration of reactant at the CL reaches zero. Equation 2.14 illustrates the relationship between the mass transport loss and the limiting current density.  𝜂𝑀𝑇 = −𝑅𝑇𝑛𝐹∑ 𝜐𝑗𝑗𝑙𝑛 (1 −𝑖𝑖𝑙𝑖𝑚,𝑗) Equation 2.14    where  𝜂𝑀𝑇  = mass transport loss (V)   𝑖𝑙𝑖𝑚,𝑗  = limiting current density for species j (A cm-2)  Chapter 2: Literature review     19   Figure 2.4: Oxygen and water concentration profiles at the PEMFC cathode (not based on real data, only for illustration purposes).  2.2.3 Cathode and anode loss contributions  Polarization curves and performance losses are typically represented for the entire fuel cell, encompassing both the cathode and anode contributions. Isolating the electrodes’ individual contributions requires the incorporation of a reference electrode, which is difficult to do. However, given that the ORR is the limiting reaction, it is commonly understood that the kinetic loss is dominated by the cathode while the anodic contribution is typically assumed to be negligible. Similarly, mass transport loss is also dominated by the cathode side due to the complexities involved with water production and transport. This dominating contribution of the cathode towards performance loss is illustrated in Figure 2.5.  Chapter 2: Literature review     20   Figure 2.5: Voltage losses by individual components in a PEMFC obtained using a fuel cell with reference electrodes.iv  2.2.4 Power output  Another useful metric of fuel cell performance is the power output. The metric is typically represented through a plot of the power vs. current density (Figure 2.6) and is mathematically represented as:  Pcell = iEcell Equation 2.15  where  Pcell = power density (W cm-2)                                                       iv Reprinted from International Journal of American Institute of Chemical Engineers, Vol. 49, Issue 12, W. He, G. Lin and T. Van Nguyen, Diagnostic tool to detect electrode flooding in proton-exchange-membrane fuel cells, 3221-3228., Copyright (2003), with permission from John Wiley & Sons, Inc. Chapter 2: Literature review     21   Figure 2.6: Generation of a power density curve through multiplication of the voltage and current density.  2.3 Operating modes  There are two main modes in which PEMFCs can be operated during polarization with regards to the reactant flow rates[40]. For the first, constant flow control, the reactant flow rates remain fixed throughout the entire polarization. This mode of operation typically ensures sufficient reactant feed, even at high current densities. In the second mode, stoichiometric control, the flow rate is adjusted stoichiometrically for each specific current according to Faraday’s law:  𝑖𝑛𝐹= −1𝜐𝑗∙𝑑𝑁𝑗𝐴𝑑𝑡 Equation 2.16   where  𝑁𝑗 = moles of species j (moles)  𝑡 = time (s)                𝑑𝑁𝑗𝑑𝑡 = theoretical rate of electrochemical reaction (mol s-1)                𝐴 = geometric electrode area (cm2)  Chapter 2: Literature review     22  In Equation 2.16, 𝜐𝑗 is taken as negative for reactants species and positive for product species. During stoichiometric control, a fixed ratio between reactant supply and consumption is maintained for the entire polarization curve:  𝜆𝑗 =  (𝑑𝑁𝑗𝑑𝑡 )𝑠𝑢𝑝𝑝𝑙𝑦 (𝑑𝑁𝑗𝑑𝑡 )𝑡ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙𝑙𝑦 𝑐𝑜𝑛𝑠𝑢𝑚𝑒𝑑 ⁄  Equation 2.17     where  𝜆𝑗 = stoichiometric ratio of species j   Since the ORR is very slow and not 100% efficient, the oxidant is usually supplied in excess. For air, typical ratios are between 2 and 4. Even though the HOR is very facile, hydrogen is also usually fed in slight excess, usually with 𝜆𝐻2 =  1.5. In practice, the choice of operating mode is determined by various factors such as the cell design, desired application and reactant availability.  2.4 Transport mechanisms  2.4.1 Overall transport  The main transport processes are reviewed in this section, to provide the context for the subsequent discussion on the MPL transport mechanisms.   Reactant gases are transported via forced convection through the flow field channels.  Gases move from the flow field channels to the CL by means of molecular diffusion, due to established Chapter 2: Literature review     23  concentration gradients. Due to the smaller pore sizes in the CL and MPL, Knudsen diffusionv can also play a significant role[41,42]. With regards to water transport, the main processes are depicted in Figure 2.7 and described below[29,30]:    Figure 2.7: The main water transport processes involved in a PEMFC.   Firstly, water is produced through the ORR at the cathode CL.  Typically, water is also supplied through the humidified air stream.  Similarly, water can also be supplied through the humidified hydrogen stream.  On the cathode side, water is removed through the flow of air exiting the FFPs.  Likewise, water is removed through the flow of hydrogen through the FFPs on the anode side. Water transport in the FFPs is established by convection due to the flow of feed                                                      v Knudsen diffusion occurs when the system character length is similar to or smaller than the mean free path of the molecules. Collisions between molecules and the system walls therefore become more prominent than collisions between molecules themselves. Chapter 2: Literature review     24  gases. In the porous GDL, MPL and CL, vapor phase water transport is dominated by diffusion, while water liquid transport is governed by capillary pressure.  One of the main water transport mechanisms inside the PEM is electro-osmotic drag. This mechanism relates to water molecules being ‘dragged’ by protons as they move from the anode to the cathode side to partake in the ORR reaction.  At the cathode side, water is typically introduced as product of the ORR, humidified feed streams and electro-osmotic drag. This may result in a concentration gradient between the cathode and anode sides, causing water to diffuse through the PEM. This process is typically referred to as back diffusion.  Water can also be pushed via convective forces (due to pressure gradients across the PEM). In comparison to electro-osmotic drag and back-diffusion, this effect is typically considered negligible due to the low membrane hydraulic permeability and relatively low pressure differences[29,30].  Thermal-osmotic drag refers to temperature-driven flow of water through the membrane. Until recently this transport process has typically been ignored and its role is not yet well understood[29]. It has been shown though, that the temperature gradient across the membrane pushes water towards the colder side[43].  2.4.2 MPL water management mechanisms  Literature proposes a variety of mechanisms by which the MPL can improve water management and performance. The main mechanisms presented in literature are briefly summarized below:  Enhancement of water drainage from the cathode CL to GDL through increased capillary action The MPL enhances water removal via capillary action from the fine CL pores, to the microporous pores of the MPL and finally to the macroporous pores of the GDL [44–46]. Water removal from the CL is thereby improved to reduce the occurrence of flooding.   Chapter 2: Literature review     25   Improvement of water back diffusion/hydraulic permeation to the anode The MPL helps to form a pressure barrier on the cathode side that subsequently increases the water flux through the membrane towards the anode [4,47–49]. Lin et al. [50] proposed that the pressure barrier is created by the presence of small hydrophobic pores in the MPL which reduce liquid water permeability towards the cathode GDL. Less water is therefore removed through the GDL, allowing improved oxygen transport towards the CL.   Formation of a liquid water barrier The MPL prevents larger water droplets, formed within the cathode GDL, from diffusing back to the CL[51]. This prevents the formation of liquid water films on the CL.   Control of  water morphology/pathways Compared to the bare GDL, the MPL limits the number of water entry and breakthrough points, thereby reducing water saturation in the GDL and stabilizing water pathways[52–56]. Nam et al.[57] also showed that the CB MPL limits the water droplet size at the CL|MPL interface (illustrated in Figure 2.8).   Neither enhancement of back diffusion nor promotion of water removal from the CL to the GDL Karan et al.[34] and Atiyeh et al.[58] suggested that the performance and durability improvements, obtained through the incorporation of the MPL, are not associated with the net water drag at all. The exact mechanism responsible for the improvement remained uncertain, however, and an alternative was not proposed.        Chapter 2: Literature review     26   Figure 2.8: Environmental SEM micrographs showing vapor condensation and liquid water breakthrough from (a) CL, (b) MPL, and (c) GDL. For each, micrographs at three different elapsed times are shown.vi  As indicated by the above summary, there is no general consensus in the literature on the dominating MPL mechanism, with some mechanisms gaining more support than others and some even in direct contradiction to each other. Such differences may possibly be attributed to the fact that mechanisms have been formulated under a variety of circumstances (different MEA components, hardware and operating conditions) and through various different methods (through experimental water balances[34], visualization techniques[47,56,57] and modelling[4,49], for example). All the proposed theories also do not appear mutually exclusive, however, and it                                                      vi Reprinted from International Journal of Heat and Mass Transfer, Vol. 52, Issues 11 -12, J.H. Nam, K. Lee, G. Hwang, C. Kim and M. Kaviany, Microporous layer for morphology control in PEMFC, 22779-2991., Copyright (2009), with permission from Elsevier Chapter 2: Literature review     27  seems plausible that a combination of mechanisms may present or interact at once. For example, it seems feasible that the MPL can enhance capillary effects and also serve to control water morphology at the same time.  2.5 Requirements of the ideal MPL   As described in Section 1.1, the MPL helps to reduce ohmic losses through improved interfacial contact with the CL, as well as mass transport losses through improved water management. To fully realize these benefits and enhance overall performance, it should therefore have the following properties:    sufficient gas permeability to enable transport of reactant and product gases;   suitable water transport properties to promote a balance between water removal (for the prevention of flooding) and water supply (for the prevention of membrane dehydration);   suitable thermal conductivity to ensure temperature uniformity and proper thermal management;   high mechanical integrity;  chemical stability for enhanced durability at various operating conditions;  excellent electrical conductivity and low interfacial contact resistance to allow electron transport and;  a morphology and structure promoting interfacial contact for maximization of continuous electrical pathways.  2.6 MPL configurations  Three MPL configurations have been presented in the literature (Figure 2.9): Freestanding MPLs, GDL-based MPLs and CCM-based MPLs. GDL-based MPLs are most commonly used and Chapter 2: Literature review     28  are applied through well-established methods such as spraying, painting, screen printing or rolling[38].   The use of freestanding MPLs is less common, but films consisting of CB and PTFE have previously been used in this capacity[51,59–62]. Freestanding MPLs offer the potential benefit of ease of manufacturing and assembly, since they may be treated independently from the GDL and applied simply by insertion into the MEA.   CCM-based MPLs, containing CB, have previously been investigated by Cipollini[63] and Park et al[64].  Cipollini reported a 15 – 30 % decrease in the resistive voltage drop due to improved interfacial contact. The manufacturing of CCM-based MPLs does present some practical challenges though. Firstly, the sintering temperature of the wetproofing agent becomes crucial, since Nafion PEM is only thermally stable up to 280°C[65]. Alternative wet-proofing agents, such as polyvinylidene fluoride were therefore employed, due to its lower sintering temperature compared to PTFE (177°C vs. 335°C)[64]. Furthermore, the use of conventional MPL spray deposition is complicated by the fact that the PEM easily wrinkles/contracts upon contact with the ink.   Figure 2.9: Expanded view of the three different MPL assembly modes employed in the literature.    Chapter 2: Literature review     29  2.7 MPL material and design modifications  Various studies have focused on the enhancement of transport processes and overall performance in the MEA through modification of the MPL’s physical and chemical properties. In accordance with Park et al. [6], these efforts are broadly categorized into the modification of the following parameters: thickness or carbon loading, wettability, porous structure and microstructural characteristics, and MPL materials.  It should be noted that this review excludes work on alternative catalyst support materials. The MPL and catalyst support are often confused (especially since CB has been used as material for both). However, the MPL and catalyst support essentially refer to separate fuel cell components which, from a research perspective, are fine-tuned with different purposes in mind.  2.7.1 Thickness or carbon loading  The thickness of the MPL is related to the carbon loading applied to the layer. Studies, based on conventional CB MPLs, have shown that there exists an optimum thickness (or loading)[66–70]: if the layer is too thin, the GDL’s surface roughness (and ohmic resistance) is not reduced sufficiently; if the MPL is too thick, it may limit mass transfer due to increased diffusion paths. Park et al.[66], suggests an optimum loading of 0.5 mg/cm2 to achieve the maximum limiting current density.  2.7.2 Wettability  Although hydrophobic agents can improve water management, excessive treatment can also enhance flooding by reducing hydrophilic pathways[50,71]. Previous studies have therefore investigated the optimum loading of PTFE required in the MPL, and reported a value of 20 wt%[72]. Hydrophilic treatment has also been incorporated in the MPL to further prevent dehydration of the membrane at low relative humidity[73–75]. Some of the different agents incorporated to induce hydrophilic character include silica nanoparticles[75,76], cylindrical aluminosilicate fibers[71] and Nafion solution[77]. Generally the results indicate that a Chapter 2: Literature review     30  combination of hydrophobic and hydrophilic treatment has a positive effect on water management and performance.  2.7.3 Porous and microstructural characteristics  Attempts to improve the porous characteristics of the MPL include the use of different carbon-based materials in the MPL (refer to Section 2.7.4). Other methods include the incorporation of pore-forming agents such as lithium carbonate[78] and an ammonium salt[79]. The addition of pore-formers generally proves beneficial and some studies also report the advantageous effects of a bimodal porosity distribution to facilitate gas diffusion and water transport through different pore sizes[46,78,79]. Another method used to alter the physical structure of MPLs is perforation through laser-cut holes[80]. Manahan et al.[80] reports a decreased performance for a MPL+GDL combination due to excessive water pooling at the perforation sites, while similar studies for stand-alone GDLs showed improved performance due to holes acting as direct pathways to the gas flow channels[81,82].  2.7.4 MPL materials (conductive component)  CB, which is the conventional MPL material, is produced through the incomplete combustion or thermal decomposition of heavy petroleum products[83]. The most common varieties used in MPLs are Vulcan XC72 or Vulcan XC72R (the latter is a more feathery version which is easy to disperse[84]). In the investigation of alternative MPL materials, numerous other CB types have also been investigated, such as  acetylene black[69,85–88], Mogul L[86], Black Pearls 2000[87], Ketjenblack[88,89], Hicon Black[90],  paraffin wax carbon[91] and RGN CB[90]. The materials were applied to varying degrees of success with common cited reasons for performance improvements over Vulcan CB being the creation of average smaller pore sizes[69,86,91] and hydrophobic wetting behaviour[91,92]. Graphite was also investigated by Passalacqua et al.[86] showing performance comparable to that of Vulcan XC72.   Chapter 2: Literature review     31  Other MPL alternatives that have been considered include more recently discovered carbon allotropes and nanostructures. Carbon nanotubes, in particular, have received a lot of attention and have been investigated as either single component MPLs[93–95] or as part of CB composite MPLs[27,96–100]. Performance improvements over CB materials were reported in certain cases and were generally attributed to lower resistance[97–99] and larger pores/increased permeability leading to improved mass transport[94–96]. Some other nanostructured carbons that have been studied also include: carbon nanofibers[92,101–103], nanospheres[103,104] and a CNT-composite Buckypaper[100].   Non-carbonaceous materials, such as antimony doped tin oxide[105] and silicon carbide[106], have also been studied. Generally, the materials resulted in poorer performance which was attributed to higher resistance (this was partly addressed in the antimony doped tin oxide study by coating particles with a thin carbon layer).   As previously mentioned, materials were also incorporated as composite MPLs, with CB+CNT and CB+CB (different types of CB)[17,46] combinations being the most common. In certain cases, a synergism in performance was observed and was generally attributed to the creation of a favorable pore structure. For example, Wang et al.[46] detail how the combination of two types of CB (acetylene black and Black Pearls 2000) creates a bi-functional pore structure that enables sufficient transport of both reactant gases and liquid water. With regards to low humidity application, several studies also reported improved performance preservation through the use of alternative MPL materials[17,27,91,96,97]. One reason provided for these improvements was the introduction of more hydrophilic materials[27,96].  A timeline of journal publications on alternative MPL materials (as reviewed above) is presented in Figure 2.10. The activity in the field over the past decade and a half emphasizes the need for alternative MPLs to help meet the increasing demands in performance and operational flexibility. The timeline also indicates that the bulk of the investigations focused on Chapter 2: Literature review     32  CB, with an increased gradual focus on CNTs and other nanostructures as their development improved.    Figure 2.10: Timeline of reviewed journal publications on alternative MPL materials spanning from 2001 to 2016 (author’s own publications excluded during investigation period).  Graphene is another carbon allotrope which has received a lot of attention in the scientific community since its discovery in 2004. It has found widespread application in electrochemical conversion and storage devices, due to its favorable properties, such as its high electronic conductivity, thermal conductivity, mechanical strength and flexibility[107]. Graphene has also shown promise in PEMFCs as an alternative to CB as an oxygen reduction catalyst support, generating improved electrochemical performance and Pt utilization[108,109]. Other studies have also investigated doped graphene structures as alternative metal-free catalysts for the ORR (this application, however, has proved to be more effective in alkaline media[110]). Despite these promising features, however, graphene has not previously been investigated as an alternative MPL. This absence can likely be attributed, to the higher cost of the material, its fairly recent discovery, and the fact that there typically is a stronger research focus on CL development Chapter 2: Literature review     33  (rather than MPL development). A brief introduction to the material’s structure and main synthesis methods is provided in the following section.  2.8 Introduction to graphene  Graphene consists of a single layer of hexagonally arranged carbon atoms, and can be produced through two general methods: the bottom-up approach, by which carbonaceous molecules are used as precursors to synthesize graphene ‘atom-by-atom’; and the top-down approach, by which graphene is synthesized through the exfoliation or separation of graphite or its derivatives[111]. Synthesis via chemical vapor deposition (CVD) is an example of the former method. While the method typically produces high quality graphene, it is very costly.   A commonly employed top-down method, is the synthesis of RGO (reduced graphene oxide) through the sequential chemical oxidation and reduction of graphite. Due to the use of harsh chemicals, the synthesis method typically introduces oxygen-containing functional groups and defects at the material’s surface[112]. Such methods are appealing from a manufacturing perspective, however, since the synthesis technique is more suitable for mass production and involves lower costs[113]. In an attempt to minimize lattice defects and eliminate harsh chemicals associated with the production of RGO, top-down electrochemical techniques have also been developed for graphene synthesis. These methods usually involve an ionic conductive solution (electrolyte) paired with graphite electrodes. An electric power source is then used to render structural changes within the graphite precursor upon ionic intercalation. Such electrochemical exfoliation methods have proved successful in yielding high quality few-layer graphene with low oxygen content[114,115]. Figure 2.11 illustrates the main differences between the two-dimensional structures of RGO and graphene and the three-dimensional structure of graphite (the original building block during top-down synthesis).  It should be noted that graphene-based materials include a broad variety of materials with different particle sizes, shapes, number of layers and oxygen content. The exact classification of Chapter 2: Literature review     34  such graphene materials, is furthermore marred by confusion and inconsistency. In this study, materials are designated as ‘graphene’, based on Bianco et al.’s general description of graphene as a “2D material whose thickness is always much less than 100 nm”[116].   Figure 2.11: Differences in molecular structure of graphite, reduced graphene oxide and graphene.  2.9 Investigation of CL interfacial characteristics  Previously, the influence of the CL and MPL on performance losses has been modelled without including any distinguishable interfacial region or simply assuming perfect contact [4,44,117]. Some modelling studies have incorporated the GDL|CL interface (Table 2.1), but neglected the presence of the MPL and the surface structures were also not taken into account [2,118]. The GDL|CL contact resistance was also investigated experimentally [119–121].   Table 2.1 indicates that the vast majority of MPL|CL interfacial studies are based on modelling approaches. Swamy et al. [7,14], Kalidindi et al.[122], Bajpai et al. [8] and Zenyuk et al.[123]  all modelled the MPL|CL interface by incorporating experimental surface profile and roughness information (as collected by Hizir et al.[124], for example). The key factors influencing interfacial contact were estimated as the local compression pressure, elasticity of the GDL and the surface morphology of the mating interfaces. The work also showed that the MPL dominates surface structure and contact resistance at the MPL|CL interface. It was further postulated that interfacial voids promote water retention/pooling, thereby influencing mass transport losses. The prominence of interfacial voids was also investigated by Tabe et al.[125] (by freezing liquid Chapter 2: Literature review     35  water at the interface), Prass et al.[126] (using X-ray micro-computed tomography) and Kim et al.[127] (specifically focusing on interfacial delamination).   Three experimental accounts of MPL|CL contact resistance measurements were reviewed. Ye et al.[128] employed a four-point resistance measurement cell. The MPL|CL contact resistance was not measured in isolation however, and reported together with the GDL|FFP contact resistance. Prass[129] also used a custom four-point resistance cell with a gold-plated copper disk as a substitute for the PEM. Kleeman et al.[130] estimated the MPL|CL contact resistance through a combined through-plane in-plane measurement. For all these experimental studies there was no clear focus on the linkage with interfacial effects such as morphology though. It is also important to note that all the studies on the MPL|CL interface were only limited to conventional CB MPLs. Table 2.1 summarizes the approaches and findings of the major interfacial studies. When considering quantitative estimations of the CL interfacial contact resistance, it is interesting to note that the values for high degrees of compression (typically above 800 – 1000 kPa or 116 – 145 psi) appear very small (or are even deemed negligible). Chapter 2: Literature review     36  Table 2.1: Summary of the literature on CL interfacial characteristics. Interface Investigated  interfacial properties Approach Materials Main findings Reference GDL|CL  GDL|CL contact resistance  EIS (electrochemical impedance spectroscopy) of H2/O2 and H2/N2 cells fitted to transmission line model; Contact resistance estimated based on varying layer thickness  GDL: Toray CFP  CL: Pt catalyst with CB support coated on Nafion 112 PEM  RGDL|CL estimated as 3.4 mΩ cm2 Makharia et al.[119] GDL|CL  Cell compression  GDL|CL contact resistance Modelling  GDL: Generic  CL: Generic  Current density distribution is very sensitive to homogeneity of compression & GDL|CL contact resistance  Increasing compression increases the total current Hottinen et al. [2,118] GDL|CL  Cell compression  GDL|CL contact resistance ElS on H2/H2 cell  GDL: Sigracet 10BA  CL: Gore Primea 5510 Series CCM  GDL|CL contact resistance decreases with increasing compression  RGDL|CL: 44 – 8 mΩ cm2 (660 – 4710 kPa) Nitta et al.[120] GDL|CL  Contact pressure Qualitative assessments using pressure sensitive film  GDL: Roll-good materials and Toray 060  CL unspecified  Channel center compression significantly lower than landing areas Kleeman et al.[130] GDL|CL  Cell compression  Contact pressure & resistance (exerted by GDL on CCM-side in absence of CCM) Through-plane resistance measurements   GDL: continuous paper, woven GDL and non-woven GDLs  CL: Gore Primea 54 Series CCM  Channel center has significantly lower compression and higher contact resistance than landing areas Butsch et al.[121]  MPL|CL  Cell compression  Combined MPL bulk & MPL|CL contact resistance Combined in-plane & through-plane resistance measurements coupled with modelling  MPL: Unspecified  CL: Unspecified   Contact resistance decreases with increase in compression  RMPL|CL: 100 – 0 mΩ cm2 (0 - 2000 kPa)   Kleeman et al.[130] MPL|CL  MPL|CL interfacial voids (delamination) Modelling, including HFR measurement on cell setup  MPL: Sigracet 10BB  CL: Gore 5710 CCM  Interfacial delamination increase total cell resistance significantly Kim et al.[127] MPL|CL  Cell compression  GDL|CL & MPL|CL contact resistance (reported in combination with GDL|FFP contact resistance) Four-point resistance cell with Cu electrodes  MPL: Vulcan XC72 CB on CFP (Toray 120) & on wetproofed carbon cloth  CL: Carbon coated Nafion 115 PEM Contact resistance decreases with:  Use of carbon paper vs. carbon cloth  Compression of GDL  Less PTFE  Addition of MPL Ye et al.[128] Chapter 2: Literature review     37  Table 2.1: Summary of the literature on CL interfacial characteristics (continued). Interface Investigated  interfacial properties Approach Materials Main findings Reference MPL|CL  MPL|CL interfacial morphology Optical profilometry  MPL: Sigracet 10BB  CL: Cracked CL, unspecified  MPL & CL  surface roughness could lead to imperfect contact and promote water pooling  MPL surface has a higher roughness and may dominate in interfacial contact Hizir et al.[124] MPL|CL  Cell compression  MPL|CL interfacial morphology  MPL|CL contact resistance Modelling, including experimentally determined surface profiles  MPL: Sigracet 10BB  CL: Cracked CL, unspecified  Interfacial voids promote water retention and increase contact resistance  MPL dominates surface structure & contact resistance at MPL|CL interface  MPL|CL interfacial contact is mainly controlled through local compression pressure, elasticity of the diffusion media & surface morphology of MPL & CL  RMPL|CL: 4 – 0.5 mΩ cm2 (1000 – 3000 kPa) Swamy et al.[7,14] MPL|CL  MPL|CL interfacial morphology Modelling, including experimentally determined surface profiles  MPL: Sigracet 10BB  CL: Cracked CL, unspecified  Interfacial voids increase ohmic losses  Ohmic loss is highly influenced by in-plane resistivity of MPL and CL  Bajpai et al.[8] MPL|CL  Cell compression  MPL|CL interfacial morphology Modelling, including experimentally determined surface profiles  MPL: Sigracet 10BB  CL: Cracked CL, unspecified  Interfacial voids leads to water pooling and reduction in limiting current  Extent of this effect is highly dependent on the geometry and size of voids Kalidindi et al.[122] MPL|CL  Cell compression  MPL|CL interfacial morphology  MPL|CL contact resistance Modelling, including experimentally determined surface profiles Case study 1:  MPL: Sigracet 10BC  CL: Ion Power Case study 2:  MPL: MRC U105  CL: Gore Primea CCM  Contact resistance is similar for cracked and non-cracked interfaces (cracks occupy very little area)  Cracked interface has higher water retention capability  RMPL|CL: 100 – 0 mΩ cm2 (0 – 0.12 normalized compression pressure) Zenyuk et al.[123] MPL|CL  MPL|CL interfacial morphology  Liquid water at interface immobilized through freezing and processed through image analysis  MPL: Sigracet 25BC  (also considered Sigracet 25BA not containing a MPL)  CL: Gore Primea 5570 CCM  MPL ensures finer contact with CL and suppresses water retention  Coating CL directly onto MPL (as a GDE) results in finer interfacial contact and decreases flooding Tabe et al.[125] MPL|CL  Cell compression  MPL|CL interfacial morphology  X-ray micro-computed tomography  MPL: Sigracet 25BC  CL: Carbon coated Nafion 117 membrane  The void fraction decreases with an increase in compression and increased similarity in layer roughness  Large interfacial voids are found close to MPL cracks Prass et al.[126] MPL|CL  Cell compression  MPL|CL contact resistance  CL composition  Four-point resistance cell with CL deposited onto gold disks  MPL: Sigracet 25BC  CL: Deposited on gold-plated copper disks  Contact resistance decreases with increase in compression  Contact resistance increases with CL ionomer content and RH  RMPL|CL: 62 – 2.5 mΩ cm2 (340 - 4060 kPa) Prass[129] Chapter 3: Experimental methods     38  Chapter 3: Experimental methods  The chapter presents the general experimental methods employed in the study. The theory and procedures of all characterization methods are described. The methods used to prepare MPLs and to assemble MEAs are also detailed. With regards to fuel cell testing, the two sets of fuel cell hardware that were used, are presented: the TP-5 cell with an active area of 5 cm2 and the TP-50 cell with an active area of 49 cm2. Lastly, the general fuel cell operating procedures and protocols are provided.    3.1 Ex situ characterization methods  Various ex situ MPL characterization methods were employed throughout the different phases of the study as shown in Table 3.1.   Table 3.1: Ex situ characterization methods used throughout various phases of the study. Characterization method Chapter 4 Freestanding MPLs Chapter 5 GDL-based MPLs and their composite effects Chapter 6 Application of graphene-based MPLs to low loaded CCMs Structure and morphology   Change in equipment   Surface roughness   Change in equipment   Layer thickness    Wettability    Through-plane resistance    In-plane (volume) resistivity   Change in equipment   Interfacial contact resistance   Change in equipment   Interfacial contact area  x Excluded from further assessments x Gas permeability x   Water permeability x   Porosity x   Extensive assessment of compressive effects on MPL properties x   Active electrical connectivity x  Investigated, but not included in further assessments x Chapter 3: Experimental methods     39  The suite of techniques employed for each phase of the study was adapted and fine-tuned as research progressed. The motivation for these changes is provided in the respective chapters. For certain measurements, new equipment were also incorporated as more suited and/or more easily accessible equipment became available. A schematic representation of each characterization method is furthermore provided in Figure 3.1 and Figure 3.2.   Figure 3.1: Schematic representation of the ex situ characterization methods employed during the study.   Chapter 3: Experimental methods     40   Figure 3.2: Schematic representation of the ex situ characterization methods employed during the study (continued).   3.1.1 Structure and morphology   In Chapter 4, MPLs’ surface morphologies and structures were investigated by use of SEM (scanning electron microscopy), performed on a Helios Nanolab 650 Focused Ion Beam SEM. For Chapter 5 and Chapter 6, the majority of images were collected using a Hitachi SU3500 SEM. Samples for cross-sectional imaging were prepared by freeze-fracturing using liquid nitrogen. Cross-sectional images were only taken for compressed samples since the uncompressed samples incurred significant damage during sample preparation and freeze-fracturing.    Chapter 3: Experimental methods     41  3.1.2 Surface roughness  In Chapter 4, a Wyko NT100 optical profilometer was employed to investigate each MPL’s surface structure and roughness in more detail (refer to an example surface profile provided in Figure 3.1 (a)). Optical profilometry provides the means to obtain a high resolution image of a surface at a relatively fast speed (compared to atomic force microscopy) and without deforming fragile surfaces (unlike stylus profilometry)[124]. The method employs the interference of white light and relies on surface reflectiveness to generate images. Since the commercial MPL used in Chapter 4 had a non-reflective surface, it was sputter-coated with a thin layer of gold (~ 40 nm) using a Leica EM MED020 coating system (following the example of Hizir et al.[124]). All other freestanding MPLs had sufficient reflectivity. Eighteen different sample images were collected at a magnification of 40x for each MPL. The total sampling area covered was 111 μm x 148 μm with a sampling interval of 232.28 nm and a resolution of 3 nm[124] in the vertical direction.   The average and root mean square three-dimensional surface roughnesses were calculated from the image data using the equations presented below. The root mean square roughness was included to gain a better sense of the actual feature size.  𝑅𝑎 =1𝑀𝑁∑ ∑|𝑧𝑘𝑙|𝑁𝑙=1𝑀𝑘=1 Equation 3.1  𝑅𝑞 = √1𝑀𝑁∑ ∑|𝑧𝑘𝑙2 |𝑁𝑙=1𝑀𝑘=1 Equation 3.2      Chapter 3: Experimental methods     42      where     𝑅𝑎 = average roughness (μm)     𝑅𝑞 = root mean square roughness (μm)     𝑀 = number of data points in x direction     𝑁 = number of data points in y direction     𝑘 = index in x direction     𝑙 = index in y direction     𝑧 = surface height relative to reference mean plane (μm)  Surface roughness measurements in Chapter 5 and Chapter 6, were performed using a Olympus LEXT OLS3100 laser scanning confocal microscope (with a 408 nm laser). Compared to the previously used optical profilometer, the confocal microscope offered various cost and time saving benefits, in addition to sufficient resolution. Since this technique relies on a different operating mechanism than optical profilometry, reflective sample surfaces were also not required. It was therefore not necessary to coat the non-reflective MPLs with gold. Surface roughness parameters could furthermore be extracted for an entire plane, and not only a single line, thus providing a truer sample representation. Data was generated for a plane of 640 μm x 480 μm at lateral and axial resolutions of 120 nm and 10 nm[131] respectively. Five different locations were analyzed on each MPL’s surface at a magnification of 20x. Surface roughness parameters were again calculated based on Equation 3.1 and Equation 3.2.   3.1.3 Layer thickness  Layer thickness was determined using a micrometer gauge. To minimize MPL damage during measurements, samples were placed between protective shims of known thickness. These measurements were subsequently also used in the calculation of in-plane resistivity, and gas and water permeability. In Chapter 5 and Chapter 6, where GDL-based MPLs were employed, the total GDL+MPL thickness is typically reported. The thickness of the MPL itself was also estimated by subtracting the GDL thickness from the combined GDL+MPL thickness. This Chapter 3: Experimental methods     43  estimation is based on the assumption that the MPL does not penetrate the GDL at all. The estimated MPL thickness was incorporated to determine the MPL apparent porosity via the liquid saturation method (Section 3.1.11).  3.1.4 Wettability  It is common practice to employ contact angle measurements to gain insight into MPL surface wettability (or hydrophobic/hydrophilic character)[132]. Surface wetting is accomplished when the surface energy of the solid is larger than that of the liquid [133]. Surface energy refers to the energy required for a species to increase its surface area. The relation between the contact angle and surface energy for a flat surface is defined by Young’s equation (also refer to Figure 3.3 (a)):   𝛾𝑠𝑣 = 𝛾𝑠𝑙  +  𝛾𝑙𝑣 cos 𝜃 Equation 3.3  where  𝛾𝑠𝑣 = solid surface free energy (N m-1) 𝛾𝑠𝑙 = solid|liquid interfacial free energy (N m-1) 𝛾𝑙𝑣 = liquid surface free energy (N m-1)   𝜃 = contact angle (°)  The contact angle is the angle between the solid|liquid and liquid|vapor interfaces. Geometrically this is represented by drawing a tangent on a droplet profile at the junction of the three phases.  The following broad classification is used to categorize a surface according to contact angles: < 90° corresponds to a hydrophilic surface; > 90° corresponds to a hydrophobic surface and > 150° is considered super-hydrophobic (refer to Figure 3.3).  The above description for Young’s equation is for an ideal or flat surface, while rough surfaces are more accurately described by Wenzel and Cassie-Baxter behaviour[134]. As illustrated in Chapter 3: Experimental methods     44  Figure 3.4, the underlying assumption of the Wenzel state is homogenous wetting, i.e., the liquid-solid interface follows the solid surface. The Wenzel model furthermore implies that hydrophilic surfaces become more hydrophilic after roughening, while hydrophobic surfaces become more hydrophobic. For the Cassie-Baxter state it is assumed that the surface consists of different patches, resulting in heterogeneous wetting. One such example is when the liquid-solid interface does not follow the solid surface due to air-pocket formation.   Figure 3.3: Contact angles and sample images for (a) a hydrophilic surface (reduced graphene oxide MPL), (b) a hydrophobic surface (Johnson Matthey CCM), and (c) a superhydrophobic surface (CB MPL); Young’s relation between the contact angle and surface energy is also illustrated in (a); γsv is the solid surface free energy, γsl is the solid|liquid interfacial free energy, γlv is the liquid surface free energy, and θ is the contact angle.   Figure 3.4: Illustrations of (a) Wenzel and (b) Cassie-Baxter wetting behaviour on rough surfaces.  For contact angle measurements, water droplets of approximately 2-4 μL were dispensed onto the MPL surfaces and images were captured with a high resolution digital camera (Nikon D90) under a light source (illustrated in Figure 3.1 (b)). Droplets were dispensed on approximately 6 different locations on each surface to determine the average contact angle. Subsequent image analysis was performed with the software package FTA32, version 2.0 (First Ten Ångstroms) to extract the contact angles. The same method was used to measure the contact angle throughout the entire study. Chapter 3: Experimental methods     45  3.1.5 Through-plane resistance  For GDLs, the through-plane resistance is the most commonly measured electrical property[132]. Experimental setups, similar to the one employed in this study, are typically used. The combined MPL+GDL (sample surface area of 5 cm2) was positioned between two gold-plated brass plates. A LCR-meter (GW-Instek LCR-819), employing the four-wire Kelvin method, was connected to the plates and used for resistance measurements.  A small current was applied between the plates through the outer two probes and the corresponding voltage was measured through the inner two probes as a function of compression. Compression was regulated via a pneumatic piston assembly (refer to equipment drawing in Appendix A.1). The resistance contributions of the wires and gold-plated brass plates were excluded during each measurement. The resultant overall resistance could therefore be expressed as the sum of all the bulk and contact resistances (Figure 3.1 (c)).   3.1.6 In-plane (volume) resistivity   In-plane resistivity is most commonly measured using the four-point probe method[135] and is sometimes also referred to as the volume resistivity. For low resistivity measurements, such as for the materials investigated in this study, the four-point probe method is furthermore preferred since the contact resistance between the sample and probe is negligible[132]. If the material thickness is less than the probe spacing (which is generally the case in this study), the samples qualify as thin films with the resistivity expression as presented in Equation 3.4. The correction factor, 𝑘, is used to account for the fact that the samples have a finite geometry (with sample edges affecting the current distribution). Smits[136] defined thin film correction factors for various geometries which depend on the film thickness, shape, dimension and probe spacing. Square samples with an area of 5 cm2 were investigated, corresponding to a correction factor of 𝑘 = 0.9825. Sheet resistance, in the equation below, represents a theoretical resistance which ‘hides’ the fact that layers have different thicknesses, thereby not taking the true bulk of material into account. Chapter 3: Experimental methods     46  𝜌 = 𝑅𝑠𝜏 = (𝑉𝐼) 𝑘𝜏 Equation 3.4  where  𝜌 = in-plane or volume resistivity (mΩ cm) 𝑅𝑠 = theoretical sheet resistance (mΩ square-1) 𝑘 = geometrical correction factor     𝜏 = sample thickness (cm) 𝑉 = voltage measured across inner probes (mV)  𝐼 = current passed through outer probes (mA)  In Chapter 4, the in-plane resistivity was measured with a four-point probe setup (Jandel Engineering) consisting of tungsten pins arranged linearly with a constant tip spacing of 1 mm (Figure 3.1 (d)). These tips were blunt and spring loaded to prevent sample penetration. The probes were placed on each sample while current was passed between the two outer probes and the voltage measured between the two inner probes (using a lock-in amplifier). To account for sample variations, measurements were repeated in different orientations (minimum of 3).   In Chapter 5 and Chapter 6, a Signatone four-point probe (S-302-4) with blunt tungsten-carbide tips was used for in-plane resistivity measurements. The tip spacing and radius was 1.016 mm and 0.254 mm respectively. A Bio-Logic potentiostat (VPM3) was used for data acquisition. As control measure, the commercial CB MPL (Sigracet 25BC) was found to yield similar in-plane resistivity results on this and the previous four-point probe setup (32.2 mΩ cm ± 0.6 vs 33.6 ± 1.5 mΩ cm).  3.1.7 Interfacial contact resistance  A slight modification of the in-plane resistivity setup enabled assessment of the MPL|CL contact resistance, which is also used to gauge the electrical connectivity (the lower the contact Chapter 3: Experimental methods     47  resistance, the higher the degree of electrical connectivity achieved between the MPL and CL). By placing two MPL pieces separated by a small gap (0.5 mm) on top of the CCM, the current was forced to flow across the MPL|CL interface and to run in the lateral direction before crossing the MPL|CL interface again (shown if Figure 3.1 (e)). The measured voltage drop therefore relates to the GDL+MPL bulk resistance and interfacial contact resistance. It also includes CL in-plane resistance which is considered to remain the same for a specific CCM. This setup is a simplified version of that used by Kleeman et al.[130] which also included a simulation component and compressive effects for the determination of a voltage offset. In the absence of these features, experimental values are not adjusted through an offset and are reported as scaled or normalized values. As a special measure in Chapter 5 and Chapter 6, samples were compressed before assembly to avoid frayed edges and damage to the GDL-based MPLs.  3.1.8 Interfacial contact area  Pressure film is typically used in fuel cells to determine whether the compression pressure exerted on the FFPs is uniform or not[137]. In Chapter 4, however, pressure film was used to assess the interfacial contact between the MPL and CL, and how the different MPLs respond to the same compressive force. A color-calibrated pressure sensitive film, FUJIFILM Prescale (Sensor Product, Inc.), was sandwiched between each MPL and the CCM in an actual fuel cell setup with FFPs, gaskets and the remainder of the MEA components (Figure 3.1 (f)). The film is rated for pressures between 483 – 2413 kPa (70 – 350 psi) and has a spatial resolution of 5 – 15 μm. Image analysis (performed in ImageJ) was used to enable a quantitative comparison of the overall contact area over the flow field channels within a given pressure range eg. 690 – 827 kPa(g)  (100 – 120 psi(g)). This was done by converting the resultant imprint to a binary image and applying thresholding to extract the number of contact points within the given range.     Chapter 3: Experimental methods     48  3.1.9 Gas permeability  To help understand how the different MPLs affect mass transport behaviour, gas permeability, water permeability and porosity measurements were performed in Chapter 5 and Chapter 6. Permeability represents the ability of a porous material to allow fluid flow through its available open volume[138]. The property is typically determined by use of Darcy’s law (Equation 3.5) with the fluid’s superficial velocity calculated from its volumetric flow rate (Equation 3.6). The superficial velocity is defined based on the sample’s geometric area and neglects the actual interstitial area of the sample. Darcy’s law furthermore only applies to laminar flow, implying that its application is restricted to the linear section of the pressure gradient (∆𝑃𝜏) versus velocity plot. For convenience, permeability is often expressed in ‘darcy’, instead of m2 with 1 darcy = 0.9869 x 10-12 m2.  𝑘𝑓 =𝑣𝑓𝜇𝑓𝜏∆𝑃 Equation 3.5  𝑣𝑓 =?̇?𝜋 ∅2 4⁄ Equation 3.6      where          𝑘𝑓 = fluid permeability (m2 or darcy)     𝑣𝑓 = superficial fluid velocity (m s-1)     𝜇𝑓 = fluid viscosity (Pa s)     𝜏 = travel path of fluid / sample thickness (m)     ∆𝑃 = pressure drop across sample (Pa)     ?̇? = fluid volumetric flow rate (m3 s-1)     ∅ = sample diameter (m)  Chapter 3: Experimental methods     49  Gas permeability measurements were performed by use of specially designed equipment which facilitated gas flow in the through-plane direction (depicted in Appendix A.2). To prevent leakage, the samples were sealed in the chamber with two sets of o-rings and tightly clasped in the jig by bolts. Inlet flow rates were controlled using a rotameter and the outlet flow was left open to the atmosphere (Figure 3.2 (a)).  Handheld digital manometers (HHP-2020 and HHP-2023, Omega) were used to measure the pressure drop across the sample chamber. Oxygen was used as the gas and the viscosity was taken as 2.02x10-5 Pa s. Samples were cut to a size of 3 cm diameter, with the total flow area corresponding to 5.07 cm2 (diameter of  2.54 cm).  3.1.10 Water permeability  Water permeability measurements were performed by Porous Materials Inc. (Ithaca, USA) and also calculated by used of Darcy’s law (Equation 3.5). The samples were sealed in a chamber underneath a reservoir filled with water. Pressure was applied to the water above the sample by incrementing a regulator, gradually forcing water flow in the through-plane direction. After holding the pressure at each point for a certain time, the corresponding change in water volume/mass was measured (refer to Figure 3.2 (b) for basic setup). For samples with very low permeabilities, epoxy was used to provide additional sealing. The viscosity of water was taken as 1.00x10-3 Pa s. To mimic actual fuel cell operation, the MPLs were placed to face the water inlet. Only compressed samples were employed for water permeability measurements. Similar sized samples, as used for the gas permeability measurements, were employed.   3.1.11 Porosity  Following the example of Williams et al.[139] and Hussaini and Wang[140],  the MPLs’ apparent porosities were estimated by use of the liquid saturation method (Figure 3.2 (c)). ‘Apparent’ refers to the fact that sealed pores (considered unavailable for mass transport) are not included in the porosity calculation. The method involves submerging the sample in a completely wetting liquid and determining the porosity based on the consequent mass difference (as shown in Chapter 3: Experimental methods     50  Equation 3.7). Hexane, confirmed to be completely wetting for all materials, was used in this case and its density at room temperature was taken as 659 kg m-3.   𝜀 =𝑉𝑝𝑜𝑟𝑒𝑉𝑡𝑜𝑡𝑎𝑙× 100% =∆𝑚 𝜌𝑙𝑖𝑞⁄𝐴𝜏× 100% Equation 3.7      where     𝜀 = porosity (%)     𝑉𝑝𝑜𝑟𝑒 = pore volume of layer (m3)     𝑉𝑡𝑜𝑡𝑎𝑙 = total volume of layer (m3)     ∆𝑚 = weight difference resulting from liquid saturation (kg)     𝜌𝑙𝑖𝑞 = density of wetting liquid (kg m-3) 𝐴 = layer area (m2) 𝜏= thickness of layer (m)  The porosities of the individual MPLs were calculated based on the GDL porosity, combined GDL+MPL porosity and estimated MPL thickness. The measured GDL porosity of 77% ± 2%, was in close agreement with reports from the literature (78.3%[141]). Porosity measurements were furthermore only performed in the compressed state to prevent MPL flaking during submersion in hexane.   3.1.12 Active electrical connectivity  A technique was also developed to try and quantify the degree of ‘active electrical connectivity’ between the MPL and CL. In this context, active electrical connectivity is defined as another interfacial parameter representing the establishment of electrical connections between the MPL and active Pt sites in the CL. Active electrical connectivity can be distinguished from electrical connectivity (or essentially interfacial resistance), in that the latter considers all electrical connections between the MPL and CL, and the former only considers electrical Chapter 3: Experimental methods     51  connections to active sites in the CL (i.e. TPB points where a Pt particle has access to protons, oxygen, electron-conducting support/MPL). The premise of the technique was to use ECSA (electrochemically active surface area) measurements, via the Hupd (hydrogen underpotential deposition) technique, to quantify the degree of active electrical connectivity (Figure 3.2 (d)). While the technique unfortunately did not produce consistent results, more detail on its design, application and the challenges involved are presented in Appendix A.3, to serve as reference for any future development.  3.1.13 Assessment of compressive effects on MPL properties  The majority of MPL properties presented in Chapter 5 and Chapter 6 were assessed in both the uncompressed and compressed states. The exceptions are cross-sectional imaging, water permeability and porosity measurements which were performed in the compressed state only, due to significant sample damage incurred during measurement in the uncompressed state. For uncompressed state measurements, MPL samples were used as prepared through spray deposition, without any alteration. For compressed state measurements, MPLs were first compressed at 827 kPa(g) (or 120 psi(g)) between the two plates of the through-plane resistance device. This compression pressure corresponds to the conditions used during fuel cell testing. The MPLs were then removed and subjected to the various ex situ characterization techniques. The MPLs did not show any degree of rebound after removal, due to the fragile and compressible nature of the layers.  3.2 MPL preparation   The spray deposition method described in the following section only applies to GDL-based MPLs used in Chapter 5 and Chapter 6 (the freestanding MPLs used in Chapter 4 did not require any preparation and were simply inserted into the MEA). MPL materials were made into inks by mixing with isopropanol and water in an 80:20 volume ratio. PTFE (DISP 30, Fuel Cell Earth) was also added to constitute 20% of the total MPL weight and was used to act as binder and impart Chapter 3: Experimental methods     52  hydrophobicity to the layers. The amount of PTFE was based on the optimum value suggested by literature[72], and is also very similar to that of the commercial CB MPLs used in Chapter 4 (Sigracet 25BC contains 23 wt% PTFE). All the ink components were thoroughly mixed by sonication in an ultrasonic bath for approximately 2 hours. The inks were then deposited onto a GDL, Toray 060 CFP containing 5 wt% PTFE (TGPH060-4005, Fuel Cell Earth), through spray deposition (illustrated in Figure 3.5).   Figure 3.5: (a) Experimental setup for MPL spray deposition illustrating GDL rotation and the raster-like spray pattern; (b) GDL surface before and after spray deposition.  During spray deposition the GDL was placed on a hot plate (heated to 80 °C) to accelerate drying of the ink. Care should be taken to not significantly overheat the hotplate, since cracks tend to form when the MPL dries too rapidly. A steel frame was used to hold the GDL in place during spraying. An airbrush (TG-3F gravity feed airbrush, Paasche Airbrush Company) was used to deposit the MPL ink and was held perpendicular to, and approximately 15 – 20 cm above, the GDL surface. The airbrush was sequentially moved in a raster-like fashion over the GDL and rotated by 45 degrees after each complete pass to ensure uniformity. At regular intervals, the Chapter 3: Experimental methods     53  GDL+MPL was allowed to dry completely and weighed to determine the difference from the original GDL weight. This process was continued until a total MPL material load of 1.5 mg cm-2 was achieved. Thereafter, the MPL was heated in a furnace at 350 °C for 30 minutes to allow the PTFE to distribute evenly throughout the layer (327 °C is the crystalline melting point of PTFE[142]).   With regards to MPL loading, there is no universally accepted standard. Literature reports values in the range of 0.2[66] to 3.5 mg cm-2[93], while 0.5 mg cm-2 is also suggested as an optimum for CB MPLs[66]. In this study, lower loadings (< 1 mg cm-2) were avoided to ensure increased coverage of the GDL, while 1.5 mg cm-2 was set as an upper limit, due to material restrictions.  3.3 MEA assembly  For each phase of the study, the same commercial GDL, Toray-060 CFP, containing a 5% PTFE loading was used as the cathode GDL. In Chapter 4 and 5, the CCM consisted of the commercial material Gore Primea Series 5510 (Gore Fuel Cell Technologies) with a Pt loading of 0.4 mg cm–2 on both sides. In Chapter 6, Johnson Matthey CCMs with Pt loadings of 0.4 mg cm–2 and of 0.1 mg cm–2 were used. Sigracet 25BC was employed as the combined anode GDL+MPL in all cases. During assembly the CCM was sealed with Kapton polymide film (C2345-1R-10000, Matrix Technology Ltd.), to prevent peripheral reactant cross-over. In an effort to reduce the amount of variables to consider, hot-pressing was also not incorporated.  3.4 Fuel cell hardware  Two sets of fuel cell hardware were employed in the study: TP-5 cell hardware with an active area of 5 cm2, and TP-50 fuel cell hardware with an active area of 49 cm2 (both from Tandem Technologies). The smaller TP-5 cell hardware was employed in Chapter 4 and Chapter 5 (and also in Chapter 6 to a lesser extent), primarily due to constraints on MPL materials. Its FFPs, Chapter 3: Experimental methods     54  with serpentine flow channels and an exposed channel area of 38.9%, were set up in a cross-flow configuration. Additional MEA sealing was introduced through silicon gaskets embedded in the FFPs. Figure 3.6 shows all the components of the assembled TP-5 fuel cell. The components were sandwiched together and compressed via a pneumatic piston in the assembly jig.   The TP-50 fuel cell hardware was incorporated in Chapter 6 for assessment at larger scale. Serpentine flow channels with a partial co-flow configurationvii were employed. Ge and Yi[143] found that co-flow mode, in particular, promotes dehydration at low humidity conditions. The channel width was 1.27 mm with a total exposed channel area of approximately 46%. The MEA components were sandwiched together and compressed via a pneumatic pressure bladder in the assembly jig (refer to Figure 3.7).    Figure 3.6: (a) Components of an assembled Tandem TP-5 fuel cell and (b) the assembly jig in the fuel cell test station.                                                        vii Flow on the cathode and anode sides move in the same general direction (e.g. from top to bottom), but gas inlets are located on opposite ends of the fuel cell jig (e.g. cathode inlet at left and anode inlet at right). This results in counter-flow at the channel level and co-flow on the flow field level. Chapter 3: Experimental methods     55   Figure 3.7: (a) Components of an assembled Tandem TP-50 fuel cell and (b) the assembly jig in the fuel cell test station.   3.5 Fuel cell operation and protocols  Performance tests were executed by use of the 2kW Hydrogenics fuel cell test station (G100, Greenlight Innovation). A simplified process flow diagram of the test station is presented in Figure 3.8. As shown, the station enables control of the reactant gases’ temperatures, pressures, RHs (relative humidities) and flow rates. Cell temperature is regulated by using water as a heating/cooling medium. Fuel cell operation involved the following protocols: leak testing, start-up (and shut-down), conditioning and performance testing.   3.5.1 Leak testing  As a safety measure, the MEA and hardware were leak tested before the onset of fuel cell operation. For this purpose, all the lines and the fuel cell hardware were filled with nitrogen gas and pressurized to the operating pressure of 202.6 kPa(g). Soapy water was then applied to all connections and interfaces between the fuel cell plates. The presence of bubbles indicated a gas leak which necessitated remedial action.   Chapter 3: Experimental methods     56   Figure 3.8: Simplified process flow diagram of the 2kW Hydrogenics fuel cell test station.  3.5.2 Start-up (and shut-down)  The start-up and conditioning procedures that were employed were based on that of Blanco[144]. The control systems of the test station are quite complex, given that a multitude of interacting variables are controlled simultaneously. A sequential procedure was therefore applied to simplify variable control during the start-up protocol. Firstly, the coolant/heater flow, which regulates the cell temperature, was switched on. Thereafter, the reactant gas flows were set to 0.1 L min-1. The temperature and pressure of the reactant gases were increased gradually in steps of 20 °C and 50 kPa respectively (step increases were only applied once the previous setpoint has been reached and stabilized for 5 minutes). By setting the dewpoint Chapter 3: Experimental methods     57  temperatures equal to the gas temperatures, the reactants were further humidified fully (100% RH). When the final operating temperatures and pressures were reached, the gas flow rates were also increased gradually to the final values (refer to Table 3.2 for the general operating conditions). The gas flow rates used for each phase of the study was unique, and is therefore presented and discussed in more detail within the respective chapters. For shut-down, the start-up procedure was applied in reverse order.  Table 3.2: General operating conditions of single cell performance tests. Operating parameter Hydrogen Air Pressure  202.6 kPa(g) 202.6 kPa(g) Relative humidity 100% 100% Temperature 75°C 75°C Cell compression 120 psi(g) or 827 kPa(g)  Coolant flow rate  0.4 – 0.5 L min-1  3.5.3 Conditioning  After the desired operating conditions were achieved, the cell was ‘conditioned’. The primary reason for conditioning the MEA is to ensure that the membrane becomes sufficiently hydrated for proton transport[145]. During this procedure the load was gradually increased until the voltage dropped to approximately 0.5 V. The fuel cell was then left to operate at these conditions until the voltage behaviour plateaued (with no significant change in the overall trend observed for at least an hour). It typically took between 8 – 16 hours to fully condition a MEA containing a Gore CCM.  3.5.4 Performance testing - Polarization  Polarization tests were performed in galvanostatic mode in the forward scan direction, from OCV to the limiting current density. Polarization data were collected for a minimum of 60s at each current density. Different times were used for each phase of the study, depending on the Chapter 3: Experimental methods     58  employed flow rates and/or operating mode (the exact times are provided in each respective chapter). Each polarization curve was ultimately calculated as the average of three repetitions, at minimum. Short purges were employed before the onset of each polarization, to remove accumulated water from previous tests. HFR (high frequency resistance) measurements were also employed to determine the ohmic resistance and were performed by use of a LCR meter (GW-Instek LCR-819) at a frequency of 1 kHz. Specific details relating to other tests performed during each phase of the study, are presented within the respective chapters.  3.5.5 Operating modes  Both constant flow control and stoichiometric control modes were utilized during the study. Constant flow control was used with the TP-5 cell, since the testing equipment could not realize the small flow rates required for stoichiometric control at low current densities. Further justification for constant flow control with the smaller sized cell is provided in Appendix A.4. Stoichiometric control was therefore also only employed with the larger-sized TP-50 cell. The particular flow rates, used during each phase of the study, are presented in the different chapters.   Chapter 4: Freestanding MPLs     59  Chapter 4: Freestanding MPLs  In this chapter, the conventional CB MPL is comparatively studied with respect to three freestanding MPL alternatives: graphene foam, perforated graphitic sheet and perforated stainless steel. The graphene foam shows beneficial interfacial properties that contribute to improved performance in the kinetic and ohmic polarization regions. These improvements are attributed to the graphene’s ability to conform at local length scales, an ability to intimately adhere to the CL and the layer’s superior conductivity. The graphene furthermore shows promise as an MPL for low humidity applications. The results also highlight the interplay of various factors that influence the MPL|CL interface and ultimately the polarization performance, such as: morphology, conductivity, compression and adhesive effects between layers.  Please note: Even though all the alternative layers considered in this thesis do not necessarily classify as microporous, the term ‘‘MPL’’ was retained as common terminology and used throughout to refer to the layer situated between the cathode GDL and CL.  4.1 Introduction  The focus of this initial phase of the study was to gain an improved understanding of the factors influencing the MPL|CL interface. For this purpose, the freestanding MPL configuration was employed. Freestanding layers consisting of conventional CB and PTFE have previously been used as MPLs[51,59,60] and showed improved water management capability, similar to GDL-based MPLs. As mentioned before, this particular MPL configuration offers easier and accelerated assembly compared to conventional GDL-based MPLs, since the layers can be applied through simple insertion into the MEA.   Three commercial, freestanding layers were chosen as alternative MPLs for the cathode side: a perforated stainless steel (SS) sheet, a perforated graphitic (GR) sheet, and a graphene foam (designated as GN foam). These chosen layers fulfilled the basic requirements for electron and mass transport, i.e., anticipated suitable electrical conductivity and a porous/perforated structure. Perforated SS sheets have also previously been used as GDLs and water barrier layers Chapter 4: Freestanding MPLs     60  with some success in PEMFCs, particularly for low humidity applications[146]. Perforated GR sheets have similarly been used as GDLs and showed advantages over commercial material for room temperature application[147]. As previously mentioned (Section 2.7.4), graphene is another carbon allotrope that is of particular interest, due to its widespread application in electrochemical storage and conversion, and reported superior electrical conductivity[107]. To the author’s best knowledge, this work details the first investigation of all the freestanding layers as MPLs.  For comparison purposes, single cell performance tests were employed, since they are regarded as the ultimate evaluation of fuel cell components. However, their performance behaviour is not always easy to interpret, because it can be influenced by a myriad of factors. This chapter therefore also incorporated the first suite of characterization methods developed as additional comparative assessment tools, used to better understand differences in material and interfacial properties, and to improve the understanding of performance behaviour.  Based on the aforementioned and the literature review (Chapter 2), the following hypotheses were specified for this phase of the study:  The freestanding GN foam will create a MPL with comparatively low resistance and improved interfacial characteristics;  The GN foam MPL will result in performance improvements compared to the baseline commercial CB MPL;  The employed characterization methods (imaging, optical profilometry, contact angle measurements, through-plane resistance measurements, in-plane resistivity measurements, interfacial contact resistance measurements and interfacial contact area measurements) will distinguish differences in the material properties of the investigated MPLs.    Chapter 4: Freestanding MPLs     61  In answer to these hypotheses, the following objectives were specified for this phase of the study:  Gaining an improved understanding of key interfacial factors;  Identification of a promising MPL alternative through a preliminary assessment with freestanding MPLs;  Development of initial suite of characterization methods and testing protocols for the comparative evaluation of alternative MPLs.  4.2 MPL materials  For comparative purposes, a commercial material (Sigracet 25BC) was employed as a baseline conventional GDL+MPL. The MPL consists of a porous carbon layer deposited onto a CFP GDL. PTFE (polytetrafluoroethylene) is also incorporated in the MPL to act as binding agent and to introduce hydrophobicity. Examination of the MPL’s surface shows that it is prone to forming small cracks, as illustrated in Figure 4.1.   aReprinted with permission from ACS Materials[148] bSEM image credit - Isaac Martens  Figure 4.1: Images of the baseline commercial CB MPL and the alternative freestanding MPLs; Higher magnification images provided in inserts.  Typical properties for each investigated cathode MPL are provided in Table 4.1 while the morphology is shown in Figure 4.1. SS perforations were produced by photoetching 316L-type sheets at a density of 134 holes per cm2 (VACCO Industries). The perforated GR sheets were manufactured from expanded graphite particles which have been reinforced with resin and Chapter 4: Freestanding MPLs     62  compressed into sheets (Graftech International). Perforations were created through mechanical impact at a density of 1550 per cm2.   The GN foam was produced by ACS Material[148] according to the following procedure detailed in Lin et al.[149]: CVD (chemical vapor deposition) was used to deposit carbon atoms, by means of CH4 decomposition (at 1000°C and ambient pressure), onto a nickel foam scaffold.  The resultant graphene was covered with a protective layer of PMMA (polymethyl methacrylate), followed by etching of the nickel foam by use of hydrochloric acid. Acetone was then used to remove the PMMA to yield the final graphene foam.  Lin et al.[149] also provides the elemental composition of the GN foam, reporting a very high carbon content: 96.72 wt.% carbon, 1.28 wt.% oxygen and 0.34 wt.% phosphorus. According to the classification system presented by Bianco et al.[116], the GN foam can be classified as multi-layer graphene, based on its average thickness of 8 monolayers. It should further be noted that the GN foam’s structure changes significantly upon compression (at approximately 827 kPa(g)) in the fuel cell hardware: the initial three-dimensional honeycomb structure reduces to a layered flake-like structure (Figure 4.1). Henceforth, all references made to the GN foam MPL (and all analyses and interpretation thereof), refers to it in this compressed flake-like state.  Table 4.1: Typical material properties of the commercial CB and the freestanding MPLs. MPL Thickness (μm) Open volume / porosity  (%) Perforation / pore size  (μm) Perforation density (hole cm-2) Area specific density / loading  (mg cm-2) MPL PTFE load  (%) Commercial CB 50 46[150] 0.024[151] N/A Proprietary 23 Perforated SS 500 21.23a 500a 134 9.84 0 Perforated GR sheet 131 26.78a 170b 1550 30.08 0 Compressed GN foam 10 - 40 20 – 40a 30 – 200b N/A 0.92 0 aBased on manufacturer’s specifications bEstimated from imaging  Chapter 4: Freestanding MPLs     63  4.3 Specifics of characterization methods  An initial suite of characterization methods was developed to assess the following MPL properties: Morphology, surface roughness, wettability, through-plane resistance, in-plane resistivity, interfacial contact resistance and interfacial contact area. The theory and procedures of all characterization methods are described in Section 3.1.  4.4 Specifics of fuel cell operation and testing  MEAs were assembled with Gore CCM with 0.4 mg cm-2 Pt loading on both sides and Sigracet 25BC as the combined anode GDL+MPL. The freestanding MPLs were simply inserted in between a Toray 060 CFP (5 wt% PTFE) GDL and the CCM on the cathode side. The TP-5 cell hardware, with an active area of 5 cm2, was used due to MPL material limitations. The cell was operated in constant flow control mode and two sets of flow rates, ‘high flow’ and ‘low flow’, were employed during testing (Table 4.2).  Table 4.2: Operating conditions of single cell performance tests in TP-5 cell hardware under constant flow control. Operating parameter Hydrogen Air Pressure  202.6 kPa(g) 202.6 kPa(g) Relative humidity 100% 100% Temperature 75°C 75°C Flow rate – ‘Low’ 0.12 L min-1 0.39 L min-1 Flow rate – ‘High’ 0.20 L min-1 0.65 L min-1 Cell compression 827 kPa(g) or 120 psi(g) Coolant flow rate  0.4 L min-1  Compression tests were also performed to investigate the relationship between cell compression and performance. The investigated compressions ranged from 483 kPa(g) or 70 psi(g), the lowest compression at which most MEAs were leak tight, to 1034 kPa(g) or 150 psi(g), the upper limit on the compression line. For each compression pressure, three Chapter 4: Freestanding MPLs     64  polarization tests were performed, after which the compression was increased by 69 kPa (10 psi). The cell was left for 1-2 hours to stabilize at the increased compression and then the process was repeated. A newly made MEA was used for each MPL’s compression test to exclude any structural damages incurred by previous testing.  For humidity tests, the cathode humidity was decreased by adjusting the dewpoint temperature according to a psychrometric chart. For each humidity setting, three polarization tests were performed, after which the humidity was adjusted to a new lower value. The cell was left for one hour to stabilize at the lower humidity after which the process was repeated. Purging between each test was considered crucial to remove any excess water produced at higher humidity conditions. Tests were performed from higher to lower humidities to prevent the early onset of irreversible membrane damage (caused by drying).  A type of ‘stress-testing’ was also incorporated to assess a MEA’s response to dynamic conditions. A protocol consisting of a repetitive start-up and shut-down sequence (as depicted in Figure 4.2) was employed.    Figure 4.2: The employed protocol for start-up/shut-down stress tests.  4.5 MPL characterization results  The profilometry images (Figure 4.3) emphasize prominent features and the diversity in morphological structures amongst the different MPLs. Furthermore, Table 4.3 provides the three-dimensional average surface roughness and root mean square roughness determined for each layer. Overall, the commercial MPL displays an intermediate roughness (Ra = 1.01 ± 0.18 Chapter 4: Freestanding MPLs     65  μm) and shows a porous carbon layer with prominent macroscopic features such as cracks and protruding carbon fibers. On the other hand, the perforated SS sheet exhibits fine linear protrusions and possesses the smoothest and most uniform surface (Ra = 0.13 ± 0.02 μm). The perforated GR sheet displays a mesh-like structure with a very high surface roughness (Ra = 1.72 ± 0.66 μm) which can be attributed to the presence of the terraced regions (also shown in Figure 4.1). Similarly, the compressed GN foam displays a very rough and non-uniform surface (Ra = 1.54 ± 0.68 μm), due to its transformation to distinct and somewhat disconnected flakes. Although values for the root mean square roughness (Rq) are higher, the general trend remains similar (Table 4.3).   Figure 4.3: Optical profilometry images illustrating the surface structure and morphology of the commercial CB and the freestanding MPLs.  As expected, the wettability results show that the alternative MPLs are less hydrophobic compared to the commercial MPL, which contains additional wetproofing and has a contact Chapter 4: Freestanding MPLs     66  angle of 149° ± 2 (Figure 4.4 (a)). The GN foam MPL furthermore demonstrates a large variance in contact angles (87° ± 17). This variance may be attributed to the fairly rough and irregular surface of the compressed graphene. Different wetting behaviors may therefore arise, depending on the position of the droplet. As described in Section 3.1.4, the Wenzel model accounts for droplets exposed to the inherent surface roughness of a material. If the droplet is furthermore exposed to trapped air, it may display behaviour that is typically described by the Cassie-Baxter model for heterogeneous surfaces. It is uncertain which wetting behaviour dominates at this point, however.   From Figure 4.4 (b) it is found that the commercial CB MPL has the highest in-plane resistivity of all the alternative MPLs at 7.23 ± 0.31 mΩ cm. The most likely reason for the high in-plane resistivity is the presence of non-conductive PTFE and the particulate nature of the carbon. In turn, the perforated SS sheet displays the lowest in-plane resistivity at 0.40 ± 0.03 mΩ cm. This is because, unlike the other MPLs, the layer is not inherently porous and thereby provides a continuous pathway for in-plane conduction. Based on the results for the MP|CL interfacial contact resistance, the following relationship is obtained: GN foam < perforated GR sheet < commercial CB < perforated SS sheet (Figure 4.4 (c)). This trend indicates that the GN foam establishes the highest degree of electrical connectivity with the CL.  Table 4.3: Roughness parameters of the commercial CB and the freestanding MPLs. MPL Average surface roughness, Ra (μm) Root mean square roughness, Rq (μm) Commercial CB 1.01 ± 0.18 1.55 ± 0.48 Perforated SS  0.13 ± 0.02 0.2 ± 0.14 Perforated GR  1.72 ± 0.66 2.40 ± 0.88 GN foam  1.54 ± 0.68 1.98 ± 0.96   Chapter 4: Freestanding MPLs     67   Figure 4.4: Characterization properties of the commercial CB and the freestanding MPLs: (a) wettability, (b) in-plane resistivity, (c) interfacial contact resistance, (d) through-plane resistance, and (e) the percentage of overall contact area between 650 and 827 kPa(g). Chapter 4: Freestanding MPLs     68  In accordance with previous work[130,150], Figure 4.4 (d) indicates that the overall through-plane resistance strongly depends on compression. Although compressions between 483–1034 kPa(g) were investigated during fuel cell testing, the resistances in the lower compression regions are also of interest for cases where insufficient or non-uniform compression may occur under the channels. The results indicate that the GN foam configuration has the lowest overall resistance in all compression regions, followed by the perforated GR sheet. The configuration with the commercial MPL has a slightly higher through-plane resistance (2 mΩ cm2 and 4 mΩ cm2 higher than the GR sheet and GN foam at the final compression pressure). The reason for the higher resistance of the CB MPLs is because it contains additional wetproofing in the MPL itself. This trend is in agreement with the results of El-kharouf et al.[150] showing an increase of the resistance with PTFE content. Finally, the significantly larger overall resistance of the perforated SS sheet (almost 80 times higher than that of the GN foam MPL at the highest compression) might be attributed to the extremely high contact resistance of the material in the fuel cell setup, similar to the contact resistance between SS flow field plates and carbon fiber GDLs reported by Andre´ et al.[152]. Except for the perforated SS, the general trends in the in-plane resistivity and through-plane resistance are in agreement.   The contact pressure imprint between the CCM and MPL as obtained with the pressure sensitive films, show that the contact pressure is the greatest in the landing areas between the flow channels (Figure 3.1 (f)). The image analysis results further indicate that, for a compression pressure of 827 kPa(g), the perforated SS and perforated GR sheets result in the largest percentage contact area, at 22% and 30% in the investigated region (Figure 4.4 (e)). On the macroscopic scale, both these MPLs have well-structured and fairly uniform surfaces, and therefore exert more uniform pressure distributions. On the other hand, the GN foam and commercial MPLs generate lower contact areas (17% and 12%, respectively). Lower contact areas would suggest a larger void fraction at the MPL|CL interface for the latter two MPLs.  The aforementioned interfacial contact results were somewhat contradicted by examining the MEAs post fuel cell testing though. The compressed GN foam MPL displayed a unique Chapter 4: Freestanding MPLs     69  adherence to the CL and could not be peeled off or removed from the CCM as a single unit, while clusters of flakes could only be detached by scratching the surface. The behavior is in stark contrast with all the other MPLs which show limited adherence (Figure 4.5) or none at all (perforated SS sheet). This observation may, in part, be explained by the GN foam’s unique morphological structure: as the GN foam is compressed over time, it is reduced to distinct flakes which are no longer fixed in a framework, increasing the layer’s ability to locally conform to the CL. It is believed that this transformation also benefits from the fact that the GN foam MPL is not fixed to the GDL. However, for MPLs such as the commercial one, carbon particles are bound to the GDL and the ability of the MPL to conform to the CL is therefore somewhat restricted by the GDL’s own movement. These results may point towards another factor that is at play, namely an adhesive effect brought on by chemical affinity. From literature it is known that there exists a chemical affinity between graphene materials and Nafion, a key component of the CL. The interaction is attributed to a strong hydrophobic interaction between the fluoro-backbones of Nafion and graphene’s surface[153,154]. Such an affinity may therefore assist in promoting interfacial adhesion. It is important to point out that the use of GN foam essentially also created a CCM-based MPL, establishing intimate contact between the MPL and CCM. The intimate contact is also in agreement with the very low interfacial contact resistance measured for the GN foam (Figure 4.4). The creation of a self-made CCM-based MPL, by use of a freestanding layer, deems very promising from a manufacturing perspective. This is because difficulties involved with the manufacturing of CCM-based MPLs (refer to Section 2.6), can be avoided.  These aforementioned results however, also suggest that determining the interfacial contact area by means of pressure sensitive film, is not of real value, since it does not reflect true interfacial conditions during fuel cell operation. Pressure films just represent the initial interfacial contact and therefore do not reflect changes that occur during long-term operation and compression. Nor does it reflect any direct effect (such as chemical adhesion) that the MPL and CL may have on each other.   Chapter 4: Freestanding MPLs     70   Figure 4.5: Cross-sectional SEM images illustrating the extent of MPL adhesion to the CCM after fuel cell testing.  4.6 Performance results  Initial polarization tests were performed at the operating conditions described in Table 4.2 at the specified low flow rates and a compression of 827 kPa(g) (Figure 4.6). Compared to the commercial CB MPL, the perforated GR sheet and the GN foam show performance improvements in the kinetic region. The GN foam fairs best overall at 200 mA cm–2, with a cell voltage of 0.907 V,   compared to 0.772 V for the commercial MPL. The perforated SS, on the other hand, performs the worst, with very low cell voltages of 0.703 V and 0.562 V at 100 and 200  mA cm–2, respectively. The apparent Tafel slope of the ORR was furthermore also calculated via Tafel analysis (Equation 2.11) based on the polarization curves up to a current density of 200 mA cm-2. Following the explanation in Section 2.2.3, it was assumed that the anodic surface potential was negligible and polarization points were iR-corrected. The resultant Tafel plots are provided in Appendix B.1, while the Tafel slopes are presented in Table 4.4. As another kinetic metric, the current density, at a voltage of 0.9 V (for the iR-corrected curve), is also provided.   Chapter 4: Freestanding MPLs     71  Results from the Tafel analysis show the following trend in the apparent Tafel slopes (Table 4.4): GN foam < perforated GR < commercial CB < perforated SS. Lower Tafel slopes are indicative of higher kinetic transfer coefficients, which favor the ORR kinetics. The values of the Tafel slopes are considered fairly high, though, especially in the case of the perforated SS (b = 264 mV dec-1). Typically, experimentally measured Tafel slopes fall between 30 – 300 mV dec-1, with higher values (> 120 mV dec-1), as in this case, usually indicating a contribution of other effects such as changes in adsorption behaviour, mass transport and ionic conductivity[155]. Ideally, such analysis would furthermore be performed at higher resolution with more points collected at very low current densities. For the TP-5 cell (with an active area of 5 cm2), data could not be collected at current densities below 40 mA cm-2, however, because of limitations set by the test station’s load bank. For iE = 0.9V, the differences between the GN foam, perforated GR and commercial CB MPLs are not distinct, ranging between 51 and 54 mA cm-2 (with a standard deviation of  7 mV generally associated with data points in this region). The perforated SS, however, shows a much lower performance and achieves a current density less than half of the other MPLs (21 mA cm-2).   Figure 4.6: Fuel cell performance of the commercial CB and the freestanding MPLs at 100% cathode RH and low flow rates (as specified in Table 4.2).  Chapter 4: Freestanding MPLs     72  Table 4.4: Kinetic parameters of the commercial CB and the freestanding MPLs at 100% cathode RH and low flow rates (as specified in Table 4.2). MPL Apparent Tafel slope (mV dec-1) iE = 0.9 V, iR-corrected (mA cm-2) Commercial CB 164 51 Perforated SS sheet 264 21 Perforated GR sheet 143 54 GN foam 111 51  Performance in the ohmic region displays the following trend: GN foam > perforated GR > commercial CB > perforated SS. Performance enhancements with the GN foam MPL over the commercial CB furthermore extend to 50 mV within the ohmic region. The trend in ohmic performance is furthermore reflected in the OCV resistance and ohmic loss (Table 4.5), showing a decrease in resistance/loss with an increase in ohmic performance. The ohmic loss is expressed as the range spanning from the minimum to maximum voltage drop experienced in the ohmic region of each MPL (with the ohmic region identified as the linear region of the polarization curve). It is also found that a thicker MPL results in a higher OCV resistance. In fact, the linear correlation between MPL thickness and OCV resistance suggests a strong association with a correlation coefficient (or goodness of fit) of 0.96 (Figure 4.7). In the context of this study, the differences in ohmic loss can be attributed to differences in the MPL bulk resistance, MPL|CL interfacial resistance and membrane resistance, including the CL ionomer resistance (refer to Equation 2.12). It remains very difficult to determine the exact contribution of different resistive components to the observed ohmic loss, however (although the membrane resistance will always account for the bulk of the resistance). For the GN foam MPL, for example, its much lower ohmic loss (and higher ohmic performance) can be attributed to a lower bulk resistance, interfacial resistance, improved membrane hydration or any combination of these factors. Given that the reactant gases were fully humidified, and that the cell design inherently suffered from excessive flooding at low flow rates (also refer to Appendix A.4), all MEAs’ membranes were thought to be very well hydrated. The membrane resistances would Chapter 4: Freestanding MPLs     73  therefore likely also have been very similar. The reduction in the GN foam MPL’s ohmic loss is therefore primarily attributed to its lower bulk and interfacial resistances.  Table 4.5: OCV resistance and range in ohmic drop of the commercial CB and the freestanding MPLs at 100% cathode RH and low flow rates (as specified in Table 4.2). MPL OCV resistance (mΩ cm2) Ohmic loss (ΔmV) Commercial CB 225 87 - 170 Perforated SS sheet 648 112 - 211 Perforated GR sheet 208 50 - 150 GN foam 101 39 - 138    Figure 4.7: Association between OCV resistance and MPL thickness.  Furthermore, all of the alternative MPLs display greater mass transport losses than the commercial CB MPL (Figure 4.6). This might be attributed, in part, to their less hydrophobic nature (Figure 4.4 (a)) leading to water retention and consequent hampering of O2 mass Chapter 4: Freestanding MPLs     74  transferviii. Note that the commercial CB MPL contains 23 wt% PTFE, whereas the alternative MPLs themselves are non-wetproofed (Table 4.1). The CB MPL also presents as the most porous layer, with a porosity of 46% (Table 4.1). Perforated SS and perforated GR sheets, both have smaller open volumes (21.23% and 26.78% respectively), while there also exists a lack of smaller interconnecting porous networks, which would normally help to facilitate effective mass transport and promote water removal. For GN foam MPL, on the other hand, the pores between the stack of flakes will be connected (albeit in a somewhat tortuous fashion), while the total porosity can be as high as 40%.   Fuel cell polarization curves were also recorded as a function of compression. Figure 4.8 shows the general trends in the different regions of the polarization curve as a function of compression. Increasing the compression has virtually no effect on the electrode kinetic region of the polarization curves for any of the MPLs. In the ohmic controlled region of the polarization curve, cell potential increases by 100 to 130 mV with compression for the commercial MPL and perforated SS sheet (Figure 4.8 (a) and (b)). As shown by Figure 4.4 (d), these are the two materials with the highest through-plane resistance. Therefore, reducing the through-plane resistance by increased compression has a significant benefit on the cell potential in the ohmic region. The compression effect is comparatively much smaller for GN foam and the perforated GR sheet (Figure 4.8 (c) and (d)), due to their inherently lower through-plane resistances (Figure 4.4 (d)).                                                        viii Although water transport itself can occur in both the vapor and liquid phase, reference to water induced mass transport limitations refer mostly to water in the liquid phase. This is because of the dominance of liquid water flooding at high current densities in the TP-5 cell hardware (also refer to the explanation presented in Section A.4).  Chapter 4: Freestanding MPLs     75   Figure 4.8: Effect of compression on kinetic, ohmic and mass transport performance of (a) the commercial CB, (b) perforated SS, (c) perforated GR MPLs, all with 5% PTFE in the GDL, and (d) the GN foam MPL with 0% PTFE in the GDL; Operating conditions of Table 4.2 and low flow rates were applied.  For the commercial MPL, a beneficial effect of compression can again be observed in the mass transport controlled region as the cell potential increases by as much as 230 mV (Figure 4.8 (a)). Such performance enhancement may be attributed to the highly hydrophobic pores of the MPL that experience a reduction in porosity when compressed from 483 to 1034 kPa(g). The reduced pore sizes enhance the water wicking effect by increasing the capillary suction pressure. The phenomenon is also mathematically illustrated by considering the Young-LaPlace Chapter 4: Freestanding MPLs     76  equation: ∆𝑃𝑐𝑎𝑝 =2𝛾𝑐𝑜𝑠𝜃𝑅, where ∆𝑃𝑐𝑎𝑝 is the capillary pressure across the fluid interface (Pa);  𝛾 is the surface tension (J m-2), 𝜃 is the contact angle, and 𝑅𝑝𝑜𝑟𝑒 is the radius of the pore/capillary tube (m). As the pore radius decreases, the capillary pressure therefore increases, promoting water wicking through the pore. None of the alternative MPLs demonstrated the same type of enhancement because they have lower porosity, much larger pores and also more hydrophilic pores to start with (refer to Table 4.1). The mass transport region for the GN foam configuration shows a moderate improvement in the cell potential (~ 60 mV) when the compression is increased from 483 kPa(g) to 620 kPa(g). However, between 758 kPa(g) and 1034 kPa(g),  a voltage loss of 160 mV is incurred. The highly compressed GN foam undergoes a significant reduction in pore volume, thereby diminishing its capacity for effective countercurrent two-phase (liquid/gas) flow.  4.7 Further evaluation of GN foam MPL for fuel cell application  To further evaluate the graphene MPL, a series of fuel cell experiments were performed varying the PTFE loading in the GDL (Figure 4.9 (a)), the air humidity (Figure 4.9 (b)) and the air flow rate (Figure 4.9 (c) and (d)). Furthermore, start-up/shut-down stress tests and the reproducibility of the polarization curves are also presented (Figure 4.9 (e) and (f), respectively). Upon reduction of the PTFE load in the cathode GDL from 5 to 0%, the performance of the fuel cell containing the GN foam MPL was found to increase even further in the ohmic region, but as expected, the mass transfer limited current density decreased (Figure 4.9 (a)). In the absence of PTFE, the GN foam MPL leads to higher water retention on the cathode side. This, in turn, increases the ionic conductivity in the CL and provides a higher driving force for water back-diffusion which further enhances membrane hydration. Thus, at a current density of 760 mA cm–2 (in the ohmic region), the GN foam MPL provides a 100 mV increase in the cell potential compared to the commercial CB MPL.    Chapter 4: Freestanding MPLs     77   Figure 4.9: Evaluation of the GN foam MPL: (a) effect of PTFE load in the GDL, (b) effect of air flow rate, (c) power density curves, (d) effect of humidity, (e) stress tests  with performance after 150 hours  in insert, and (f) reproducibility.  Chapter 4: Freestanding MPLs     78  It is thought that the greater tendency for cathode side water accumulation at low flow rates for the GN foam MPL (compared to the commercial CB) can likely be attributed to the less hydrophobic nature of the material, but may also be due to its unconventional pore structure under compression. Typically, for a conventional CB MPL, the pores available for mass transport are considered to form fairly open channels that are well interconnected throughout the entirety of the MPL (Figure 4.10 (a)). For the graphene MPL however, the pore spaces available for water transport mostly lie between stacked flakes, in the lateral direction as pictured in Figure 4.10 (b) and (c). Water (in either vapor or liquid phase) therefore has a more tortuous path to be expelled from the MPL.   The enhanced membrane humidification and cathode CL ionic conductivity with the GN foam MPL and non-wetproofed GDL, is beneficial when the cell is operated below saturation levels of humidity, as shown by Figure 4.9 (b). Improvements of up to 50 and 70 mV are gained in the ohmic and mass transport limiting region respectively, by decreasing the cathode humidity to 20%. These results are in contrast with that of the commercial MPL, indicating a decrease in performance at the same conditions and suggesting that the membrane is less hydrated at lower humidity.    Figure 4.10: (a) Schematic representation of water management in the conventional CB MPL and (b) the GN foam MPL; (c) Conceptual representation of water transport in the open space between graphene flakes.  At 100% RH the onset of flooding can be delayed, however, by operating at higher flow rates (Figure 4.9 (c)). A far more significant improvement is observed for the GN foam MPL compared to the commercial counterpart. At these conditions, a maximum power density of 870 mW cm–2 Chapter 4: Freestanding MPLs     79  is achieved, which is a significant improvement over the commercial MPL at 730 mW cm–2 (Figure 4.9 (d)). The start-up/shut-down stress tests indicate that the MEA maintains its overall performance very well (Figure 4.9 (e)). Although minor variations occur due to flooding, no consistent loss is observed over the sequential tests. The insert also shows the MEA’s response after 10 hours and 150 hours of operation (during which time over 80 polarization tests were performed and several start-up/shut-down sequences incorporated).   This aforementioned assessment shows that the GN foam MPL endures well and indicates a promising response for dynamic conditions and longer term operation. The reproducibility of the configuration was also investigated by testing three, separately made GN foam MEAs. Despite the compressed GN foam’s irregular morphology, very good reproducibility was obtained (Figure 4.9 (f)). The variance only increased slightly in the mass transport limiting region, due to the onset of flooding.  4.8 Conclusions   The following aspects are identified as key factors influencing MPL|CL interfacial characteristics: interfacial contact with the CL, MPL surface morphology, local surface conformability, and adhesive effects between layers. The comparative approach, used in this study to investigate four widely different MPLs, shows that there exists a unique interplay between these effects that determines, in large measure, the impact on the fuel cell performance. For example, based on the fact that the perforated SS has the smoothest surface (0.9 μm lower surface roughness than the commercial CB MPL) and a high degree of initial interfacial contact (10% higher than the commercial CB MPL), one might expect improved interfacial characteristics. However, its application does not show any performance improvement and is primarily attributed to its through-pane resistance being 80 times greater than that of the other MPLs. On the other hand, the GN foam MPL has an average surface roughness 0.5 μm higher than that of the commercial CB MPL. One would therefore expect a lower degree of interfacial contact with the CL. However, due to the layer’s distinct flake-like Chapter 4: Freestanding MPLs     80  structure and adhesive effects with the CCM, it can locally conform to the CL under pressure. In addition, the GN foam shows superior conductivity and presents the lowest interfacial resistance (27% lower than the commercial CB MPL on a scaled basis), lowest through-plane resistance (77% lower than the commercial CB MPL at the final compression) and a very low in-plane resistivity (75% lower than the commercial CB MPL). Together, the high conductivity, CL adhesion and the enhanced conformability of the graphene flakes, present improved CL interfacial characteristics.  The GN foam also results in performance enhancements of up to 50 mV in the ohmic polarization region, compared to the commercial CB MPL. This improvement is attributed to the layer’s lower interfacial and bulk resistance. Results also show that the compressed GN foam MPL favors water retention, compared to the commercial CB MPL, due to a more tortuous structure and a more hydrophilic surface (as evidenced through a 62˚ smaller contact angle). Increased water retention proves beneficial for conditions that would normally lead to drying (such as low air humidity). For example, improvements of up to 50 and 70 mV are gained in the ohmic and mass transport limiting regions, by decreasing the cathode RH to 20%.    Fuel cell compression is confirmed to affect MPL material properties, as illustrated by the extreme morphological transformation of the GN foam MPL upon compression. Such transformations also prompts the inclusion of compressed state characterization for all MPLs in the subsequent phases of the study. Compression further leads to increased performance in the ohmic region as layers compact and interfacial and bulk resistances reduce. For the commercial CB MPL, performance increases by up to 100 mV as compression is increased from 483 kPa(g) to 1034 kPa(g). Mass transport performance also generally starts to decrease beyond a certain point as compression increases. This happens when porosity is reduced to the extent that mass transport becomes compromised, as displayed by the 160 mV loss between 758 kPa(g) and 1034 kPa(g) for the GN foam MPL (at 1700 mA cm-2).      Chapter 4: Freestanding MPLs     81  While this initial investigation provides valuable insight into interfacial aspects and the benefits of the GN foam as a MPL, the employed characterization methods do not provide adequate insight into performance behavior. For instance, the interpretation of mass transport effects remains somewhat based on educated speculation. This supports the development of more in-depth and robust assessment techniques, as presented in the following chapter. The interfacial contact area measurements (via pressure sensitive film) furthermore prove ineffective because it does not account for the true interfacial behavior experienced in situ.  Portions of this chapter are taken from: Fuel Cells: From Fundamentals to Systems, Vol. 15, Issue 6, M.J. Leeuwner, D.P. Wilkinson and E.L. Gyenge, Novel graphene foam microporous layers for PEM fuel cells: Interfacial characteristics and comparative performance, 790-801, Copyright (2015), with permission from John Wiley & Sons, Inc. Chapter 5: GDL-based MPLs and their composite effects     82  Chapter 5: GDL-based MPLS and their composite effects  This phase of the study involved the preparation of GDL-based MPLs, to enable precise engineering of the material and wetproofing load. The materials that were considered, included: graphite, reduced graphene oxide, electrochemically exfoliated GN (produced at lab-scale) and their CB-composites (prepared in a 1:1 weight ratio). Common effects resulting from the introduction of CB in composites were identified: decreased surface wettability, through-plane resistance, porosity, gas permeability (with the exception of graphite in the latter two cases) and increased water permeability. In the case of the GN+CB MPL, these property alterations helped to establish a balance between water removal and water retention at the CL interface at both 100% and 20% cathode RH. This behavior did not only help enhance performance preservation at low humidity, but also manifested in synergistic performance enhancements, showing a 30% and 80% increase in the maximum power density at 100% and 20% cathode RH, compared to the CB MPL.  5.1 Introduction  There are several approaches to engineering alternative MPLs, such as: material substitution, variation in loading, introduction of hydrophobic or hydrophilic components, and changes to the porous structure (as detailed in Section 2.7). In the previous chapter, three freestanding layers were considered as novel substitute/alternative MPL materials. While the study helped to identify key interfacial factors and highlighted GN foam as a promising MPL alternative, all the investigated MPLs were limited to commercial materials that were simply used as received. Each layer was therefore associated with its own loading and none contained any wetproofing. To gain a more in-depth understanding during the comparative study of MPL alternatives, however, it becomes essential to carefully control parameters such as these. To properly engineer MPLs for this purpose, conventional GDL-based MPLs were therefore investigated and prepared for this phase of the study.  Another method by which MPLs are engineered includes the preparation of composite MPLs (consisting of two or more conductive components). Most often, composites are created with Chapter 5: GDL-based MPLs and their composite effects     83  CB, owing to its established use as the conventional MPL material. From a cost perspective, it is also appealing to combine novel MPL materials, which are often scarce and expensive, with CB. Even though composite MPLs have demonstrated benefits in different capacities[16,97,98,156], their application has largely been determined on a more ad hoc basis. To the best of the author’s knowledge, however, there is no study, which attempts to understand the general interactions and effects of different composite MPLs more comprehensively.    In Chapter 4 it was also shown that cell compression has a significant impact on interfacial effects and fuel cell performance. An extreme response to compression was furthermore illustrated by the GN foam, by undergoing a complete transformation from a three-dimensional honeycomb structure to a flat stack of flakes. Such deformation evidently affects material properties. Despite the extreme effects that compression can have though, the majority of MPL characterization studies only report results for MPLs in their uncompressed state; while studies incorporating compressive effects are generally more common for bare GDLs. It is therefore important to assess MPLs in their compressed state, and to consider how materials may respond differently to the same degree of compression.  The hypotheses relating to this phase of the study are:  A graphene GDL-based MPL, with similar material and PTFE loadings as a CB MPL, will result in improved interfacial characteristics and performance enhancement;  The creation of a graphene and CB composite will alleviate mass transport limitations associated with the more tortuous structure of a graphene MPL;  The creation of CB-composite MPLs (containing a similar CB weight %) will alter material properties and display similar interactions (general trends) amongst the different composite MPLs;  MPL compression (at 827 kPa(g) or 120 psi(g)) will induce changes in all material properties, but to different degrees for the different materials. Chapter 5: GDL-based MPLs and their composite effects     84  In light of the aforementioned, the following goals were furthermore specified for this phase of the study (ultimately also supporting the main objective to develop MPLs with improved interfacial characteristics and enhanced operational flexibility):  Investigation of MPL alternatives through precise engineering of parameters such as the MPL load and PTFE content;   Evaluation of the creation of composite MPLs as another means to fine-tune MPL properties and performance behaviour;  Gaining a more realistic view of MPL properties experienced in situ and the assessment of different materials’ response to compression.  5.2 MPL materials  Based on the promising results obtained with the GN foam MPL in Chapter 4, graphene-based materials were incorporated again in this phase of the study. Two material options, produced through top-down methods, were considered: commercial RGO (reduced graphene oxide), and electrochemically exfoliated graphene produced by a previous PhD student in the Gyenge research group at University of British Columbia, Amin Taheri Najafabadi[114].   As the original building block of graphene-based materials during top-down synthesis, GR (graphite) was also considered as MPL alternative. The material, in the form of Ashbury 850 GR, was previously investigated by Passalacqua et al.[86] and showed promise by yielding a performance comparable to that of Vulcan XC72 CB. GR was also considered in Chapter 4, but in the form of an un-altered perforated sheet. For comparison purposes, conventional CB was again included in the study. The following designations will be used to identify the four ‘base’ MPL materials in this chapter:   CB – conventional carbon black (Vulcan XC72R, Cabot Corporation);  GR – natural graphite (14736, Alfa Aesar);  RGO – reduced graphene oxide (HP-RGO-05G, Graphene Supermarket); Chapter 5: GDL-based MPLs and their composite effects     85   GN – electrochemically exfoliated graphene (produced by Amin Taheri Najafabadi from the Gyenge research group).  These four base materials presented a large variety in particle structures, as illustrated in Figure 5.1 and Table 5.1: CB particles are spherical and significantly smaller in size than all the other particles; GR particles consist of numerous layers that are stacked tightly on top of each other to form larger particles with smooth surfaces and fairly well-defined edges; RGO consists of large flakes with highly ragged/irregular surfaces and edges; GN consists of small flat flakes that appear very smooth and easily assemble to form larger planar structures. As example of the latter, Figure 5.1 (d) depicts a larger GN flake with the edges of the original smaller flakes still recognizable.   In bulk powder form, these materials also exhibit great differences. RGO and CB are extremely light and feathery, while the GR and GN powders are more compact. These differences are also reflected in the approximate powder densities provided in Table 5.1, and visually illustrated in Figure 5.2. In general, the BET (Brunauer-Emmett-Teller) surface areas of all the materials are also very high, with the exception of GR. The measured value for GR is in the same range as published data[157,158], and is typical due to the material’s non-porous, crystalline structure. As expected, elemental XPS (X-ray photoelectron spectroscopy) analyses furthermore show that all the materials basically consist of carbon. The RGO does, however, also contain some nitrogen, which is typically indicative of functionalization induced through the synthesis of RGO from graphene oxide[159] (the exact synthesis path of this commercial product remains unknown, however). According to the classification system presented by Bianco et al.[116], RGO is aptly referred to as reduced graphene oxide, while the GN can also be classified as graphene microsheets.     Chapter 5: GDL-based MPLs and their composite effects     86   Figure 5.1: Images of the four base MPL materials in dry powder form; SEM image credits – Blaise Pinaud.   Figure 5.2: Difference in powder volume density of the base MPL materials (illustrated for 200 mg in each vial).  Table 5.1: Material properties of the base MPL materials. Material CB GR RGO GN Original particle size (μm) ~ 0.05 2 – 15a 3 – 10a 0.5 – 0.8 Dry powder volumetric density (kg m-3) ~ 60 ~ 130 ~ 20 ~ 420 BET surface area (m2 g-1) 222[160] ~ 2b ~ 400a ~ 250[161],b Avg. number of monolayers N/A Very large numberc 1a <5[162] Color Black Grey Black Silver-grey Elemental composition (%)d C: 98 O: 2 C: 99 O: 1 C: 89 O: 3 N: 8 C: 97 O: 3 aManufacturer’s specification b Specific surface areas were measured with a Micromeritics surface area analyzer (ASAP2020) using nitrogen adsorption cGeneralized description of  graphite as consisting of thousands of graphene layers[163] d XPS analyses were performed with a Leybold Max200 and five channeltrons from an unmonochromated Mg Ka X-ray source (1253.6 eV)  Chapter 5: GDL-based MPLs and their composite effects     87  MPLs of all the base materials were deposited on a Toray 060 CFP (5 wt% PTFE) GDL, using the methods described in Section 3.2. A material loading of 1.5 mg cm-2 was applied, with 20 wt% PTFE content. Using the base materials, three CB composite MPLs were also created: GR+CB, RGO+CB and GN+CB. The materials were combined in a 1:1 weight ratio, while keeping the total carbon loading constant at 1.5 mg cm-2 (i.e., RGO+CB composite MPL contained 0.75 mg cm-2 RGO and 0.75 mg cm-2 CB).   5.3 Specifics of characterization methods   Based on the work with freestanding MPLs (Chapter 4), the suite of characterization techniques was adapted to enable a more robust assessment of material properties and accompanying performance differences. In addition to the methods employed in Chapter 4, gas permeability, water permeability and porosity were included to gain a more in-depth understanding of mass transport behaviour (also refer to Table 3.1). The majority of properties were furthermore assessed in both the compressed and uncompressed state. Samples in the fully compressed state were prepared at a compression pressure of 827 kPa(g) or  120 psi(g), similar to the conditions used during fuel cell testing.  However, certain material properties (cross-sectional structure, apparent porosity and water permeability) could only be determined in the compressed state, due to practical limitations (refer to Section 3.1). All characterization methods were furthermore applied to combined GDL+MPLs as a whole. This is because GDL-based MPLs, in contrast to freestanding MPLs, are completely bonded to the supporting GDL and therefore greatly influenced by its underlying structure, while also being inseparable during measurements. To ensure fair comparison, however, the exact same GDL (Toray 060 CFP, 5 wt% PTFE) was used in each case.  5.4 Specifics of fuel cell operation and testing  All MEAs were assembled according to the method described in Section 3.3, with Gore CCM (0.4 mg cm-2 Pt loading on both sides), and Sigracet 25BC as the combined anode GDL+MPL. Chapter 5: GDL-based MPLs and their composite effects     88  Due to material restrictions, the TP-5 cell hardware, with an active area of 5 cm2, was used. To allow for more rigorous assessment at both high and low cathode humidity conditions, the operating conditions and testing protocols were furthermore adapted for this phase of the study. The cell was operated in constant flow control mode, specifically one-dimensional flow control, and the operating conditions displayed in Table 5.2 were employed.  Table 5.2: Operating conditions of single cell performance tests performed in TP-5 cell hardware under one-dimensional control. Operating parameter Hydrogen Air Pressure  202.6 kPa(g) 202.6 kPa(g) Relative humidity 100% 100% or 20% Temperature 75°C 75°C Flow rate 0.12 L min-1 2 L min-1 Cell compression 120 psi(g) or 827 kPa(g)  Coolant flow rate 0.4 L min-1  The specified flow rates were used to achieve one-dimensional operation on the cathode side. One-dimensional control implies that an effectively uniform reactant distribution (concentration gradient between the inlet and outlet smaller than 10%) was established over the cathode active area for the current density range studied. One-dimensional control allows for a more rigorous assessment by providing a better representation of polarization behaviour over the whole active area. However, for the case where large concentration gradients do exist (typical of stoichiometric control), polarization is presented as an average representing behaviour over all the different concentration ranges (supporting graphics provided in Figure 5.3). While the one-dimensional control mode also reduces the severity of mass transport effects and delays its onset, it does not eliminate it completely and simply provides a more fundamental platform for interpretation in the absence of significant concentration gradients. It is furthermore important to note that this one-dimensional approach is just intended for research purposes and not for practical implementation.   Chapter 5: GDL-based MPLs and their composite effects     89   Figure 5.3: Differences in reactant distribution and polarization behaviour for one-dimensional and stoichiometric control.  Polarization tests were performed by collecting data for 60 s at each current density (due to very stable behaviour under one-dimensional control). For low humidity testing, the MEA was prepared by allowing it to operate at 20% cathode RH for thirty minutes before polarization. This was done to ensure that reaction conditions have stabilized. During this time, the cathode flow rate was also lowered to 0.39 L min-1 to prevent premature membrane drying or damage. To highlight performance differences at low humidity conditions, longer-term tests were also performed at a constant current density of 1000 mA cm2. This current density was chosen because it usually falls in the region associated with the maximum power output, and hence also the typical operating range, of most cells (ohmic region of the polarization curve corresponding to a performance between 0.6 – 0.7 V). During longer-term testing the corresponding voltage drop and rise in resistance was measured over the course of 5 – 6 hours. The tests may also be considered as a form of accelerated durability testing, considering the ‘harsher-than-normal’ conditions associated with low humidity and the high flow rates of the one-dimensional approach.  Chapter 5: GDL-based MPLs and their composite effects     90  5.5 Experimental data and analysis  As stated in Chapter 3, most MPL properties were determined by performing multiple measurements. The exceptions are the through-plane resistanceix and water permeability measurements which were performed once for each sample, due to material limitations and the destructive nature of these tests. For properties determined from multiple measurements, the values are reported as the average and the standard deviation (Z ± σ), with the latter reflecting the range in the measured data. Certain properties are furthermore dependent on other measured properties containing uncertainty. For example, in-plane resistivity, apparent MPL porosity, gas permeability and water permeability, are dependent on the sample thickness. For these cases, propagation of uncertainty rules were applied in the calculation of the final uncertainty[164].   The characterization results of all materials are summarized in Table 5.3, Table 5.4 and Table 5.5. Where applicable, the relative change between the uncompressed and compressed states is also presented in square brackets. Additional experimental data are furthermore provided in the appendices, relating to surface roughness (Appendix C.1) and through-plane resistance (Appendix C.2). To aid in the interpretation of the extensive amount of data, characterization results (which were collected in both the compressed and uncompressed state) are visually presented in the form of radar charts. In the context of this study, the radar charts allow for the direct comparison of different materials, while also highlighting any trends that may exist between different properties themselves. For easier comparison and to indicate the relative position of a measurement, the characterization results were also normalized based on the entire dataset (including all compressed and uncompressed results):                                                         ix To confirm the general reproducibility of through-plane resistance data, duplicate measurements were performed for the CB MPL with two different samples. The measurement was found to be very reproducible, showing no noticeable variation in the higher compression region (refer to Appendix C.2). Chapter 5: GDL-based MPLs and their composite effects     91  𝑧𝑛𝑜𝑟𝑚 =  𝑧 −  𝑧𝑚𝑖𝑛𝑧𝑚𝑎𝑥  −  𝑧𝑚𝑖𝑛   Equation 5.1   where  𝑧𝑛𝑜𝑟𝑚 = normalized value (in the 0 to 1 range)   𝑧𝑚𝑖𝑛 = minimum average property value in dataset   𝑧𝑚𝑎𝑥= maximum average property value in dataset  In most cases, the trends in the uncompressed and compressed state results were fairly similar. For the sake of brevity, the discussion in the following sections therefore primarily centers on the compressed state results. The discussion does extend to include uncompressed state results, however, when the observed trends are significantly different or noteworthy transitions are displayed. The characterization and performance results are furthermore presented sequentially through the following four different case studies:   Base MPL materials case study, in which the three alternative base MPL materials are evaluated and compared to the conventional CB MPL;  GR+CB, RGO+CB and GN+CB composite MPL case studies, in which the effects of CB-composites are investigated.           Chapter 5: GDL-based MPLs and their composite effects     92       Table 5.3: Surface properties and thickness of the GDL-based MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). MPL Contact angle (°) Average surface roughness (μm) Total thickness, GDL+MPL (μm)   0 kPa(g) 827 kPa(g) 0 kPa(g) 827 kPa(g) 0 kPa(g) 827 kPa(g) CB 146 ± 4 149 ± 2 [2% ↑] 18.5 ± 2.1 12.7 ± 2.0 [31% ↓] 285 ± 9  248 ± 3 [13% ↓]  GR 148 ± 2 140 ± 3 [5% ↓] 9.5 ± 1.0 3.4 ± 0.5 [64% ↓] 259 ± 9 229 ± 4 [12% ↓]  RGO 32 ± 7 145 ± 2 [353% ↑] 19.5 ± 1.3 3.1 ± 0.5 [84% ↓] 376 ± 20  305 ± 7 [19% ↓]  GN 140 ± 1 112 ± 5 [20% ↓] 5.4 ± 0.5 1.5 ± 0.3 [72% ↓] 219 ± 2  206 ± 1 [6% ↓]  GR+CB 148 ± 2 147 ± 2 [1% ↓] 14.2 ± 1.4 9.2 ± 1.3 [35% ↓] 265 ± 9  236 ± 4 [11% ↓]  RGO+CB 148 ± 3 147 ± 2 [1% ↓] 7.6 ± 0.7 3.1 ± 0.8 [59% ↓] 327 ± 14  277 ± 4 [16% ↓]  GN+CB 150 ± 1 143 ± 2 [5% ↓] 7.1 ± 1.1 3.6 ± 0.4 [49% ↓] 232 ± 6  218 ± 5 [6% ↓]        Chapter 5: GDL-based MPLs and their composite effects     93      Table 5.4: Electrical properties of the GDL-based MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). MPL In-plane resistivity (mΩ cm) Through-plane resistance (mΩ cm2) Interfacial resistance (normalized) 0 kPa(g) 827 kPa(g) 34 kPa(g)a 827 kPa(g) 827 kPa(g)b CB  9.4 ± 0.5  8.1 ± 0.4 [14% ↓] 92.8 4.9 [95% ↓] 0.26 GR  7.9 ± 0.5  7.3 ± 0.6 [8% ↓] 82.3 5.9 [93% ↓] 0.30 RGO  12.1 ± 1.0  9.2 ± 0.5 [24% ↓] 741.1 50.3 [93% ↓] 1.00 GN  7.1 ± 0.4  6.6 ± 0.2 [7% ↓] 61.3 5.3 [91% ↓] 0.00 GR+CB  8.2 ± 0.5  7.2 ± 0.4 [12% ↓] 73.4 3.8 [95% ↓] 0.15 RGO+CB  10.2 ± 0.6  8.3 ± 0.4 [19% ↓] 276.9 18.3 [93% ↓] 0.09 GN+CB  7.1 ± 0.3  6.5 ± 0.2 [8% ↓] 68.1 3.8 [94% ↓] 0.15  aThrough-plane resistance data collected at very low compression is grouped with the ‘uncompressed’ data.   bSamples were compressed upfront, but no additional compression was exerted during measurements.       Chapter 5: GDL-based MPLs and their composite effects     94     Table 5.5: Transport and structural properties of the GDL-based MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). MPL Gas permeability (darcy) Water permeability (darcy) Apparent MPL porosity (%) 0 kPa(g) 827 kPa(g) 827 kPa(g) 827 kPa(g CB 3.65 ± 1.30x10-1 2.51 ± 4.85x10-2 [31% ↓]  7.30x10-1 ± 9.65x10-3  72 ± 4 GR 1.29 ± 5.27x10-2 1.09 ± 2.23x10-2 [16% ↓]  3.76x10-1 ± 6.10x10-3  73 ± 8 RGO 4.30x10-1 ± 2.28x10-2 1.31x10-1 ± 2.89x10-3 [70% ↓]  1.78x10-3 ± 3.91x10-5  43 ± 4 GN 6.68x10-3 ± 1.46x10-4 3.63x10-3 ± 3.26x10-5 [46% ↓]  4.57x10-5 ± 1.86x10-7  37 ± 5 GR+CB 1.30 ± 5.68x10-2 1.14 ± 1.58x10-2 [13% ↓]  6.35x10-1 ± 1.03x10-2  74 ± 8 RGO+CB 2.35x10-1 ± 1.04x10-2 4.36x10-2 ± 6.05x10-4 [81% ↓]  2.68x10-3 ± 3.67x10-5  32 ± 4 GN+CB 2.69x10-3 ± 8.80x10-5 2.15x10-3 ± 5.65x10-5 [20% ↓]  6.64x10-5 ± 1.53x10-6  24 ± 7    Chapter 5: GDL-based MPLs and their composite effects     95  5.6 Case study 1: Base MPL materials  5.6.1 Characterization results  Figure 5.4 shows how the base MPLs create surfaces of different uniformities. The resulting roughness differences are also depicted in Figure 5.5 (a) and (b). In the compressed state, CB particles create the roughest surface (12.7 ± 2 μm), since particle coverage is primarily concentrated along the GDL fibers, leaving some GDL pores exposed. The larger GR, RGO and GN particles, in turn, result in more uniform and smoother surface coverage.    Figure 5.4: Surface images of the base MPLs in (a) uncompressed and (b) compressed states; Higher magnification images are provided in the inserts.  RGO also has a very high surface roughness in its uncompressed state (19.5 ± 1.3 μm) as indicated in Figure 5.5 (a), due to the formation of protruding clusters (visually demonstrated in Figure 5.6). Upon compression, though, the surface is smoothened drastically (by 84%). The GR and GN MPLs also undergo high degrees of surface smoothening (64% and 71% respectively). The GN flakes ultimately form the smoothest layer with an average surface roughness of 1.5 ± Chapter 5: GDL-based MPLs and their composite effects     96  0.3 μm in its compressed state. These smoothening effects are also illustrated through comparison of Figure 5.4 (a) and (b) which show how surface features and pores are reduced upon compression (for three-dimensional renderings of the surfaces, also refer to Appendix C.1).   Figure 5.5: Radar charts of material properties for the base MPLs in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the base MPLs.  Figure 5.5 (b) shows that the CB MPL forms the most hydrophobic surface (149˚ ± 2˚), followed by RGO (145˚ ± 2˚) and the GR  MPLs (140˚ ± 3 ˚), in the compressed state. While the contact angle of GN technically also falls in the hydrophobic regime, its value is much lower at 112˚ ± 5˚. This more hydrophilic nature, compared to CB, is consistent with that observed for the compressed GN foam in Chapter 4 (87˚ ± 17˚ contact angle in the absence of wetproofing). It is furthermore interesting to observe how surface smoothening, induced by compression, Chapter 5: GDL-based MPLs and their composite effects     97  influences the surface wettability. For example, comparison of Figure 5.5 (a) and (b) indicate that the GR and GN MPLs become less hydrophobic upon compression. The behavior suggests that Wenzel wetting dominates (as described in Section 3.1.4) and implies homogenous wetting of the surface. However, it is possible that some degree of air pocket formation (Cassie-Baxter model) may also exist. In turn, the CB MPL does not show any significant change in surface wettability with compression, while that of the RGO MPL is extreme (from a contact angle of 32° ± 7˚ to 145° ± 2˚ upon compression). The exact reason for RGO’s shift from the hydrophilic to the near super-hydrophobic wetting regime, is unclear though. During wettability measurements, it seemed as though water penetrated into the uncompressed RGO MPL. It is theorized that this penetration is prevented upon compression, due to a significant decrease in surface roughness and porosity.    Figure 5.6: High contrast and low magnification image of the RGO MPL’s surface, illustrating clusters encircled in red.  The differences in the MPLs’ cross-sectional structures are furthermore illustrated in Figure 5.7. CB forms a layer with extremely fine pores atop the GDL. GR particles form larger-sized interstitial pores and accumulate in random orientations. The resultant MPL has higher volume density though, and manifests as a thinner layer than the CB MPL (total thickness of 229 ± 4 μm and 248 ± 3 μm for GR and CB, respectively, in the compressed state). Both RGO and GN MPLs show particles that compile as stacked flakes with pore spaces mostly orientated in the longitudinal direction. Although these MPLs show similar cross-sectional structures, they Chapter 5: GDL-based MPLs and their composite effects     98  display very different degrees of compactness: The RGO MPL has a compressibility of 19% (the largest amongst all MPLs) and is approximately 100 μm thicker than the GN MPL in the compressed state. Notably, the smallest degree of compressibility is also obtained with the GN MPL (6%). The trend observed in MPL thickness in Figure 5.5 (a) and (b) (RGO > CB > GR > GN), is inversely related to that of the dry powder density in Table 5.1 (GN > GR > CB > RGO).    Figure 5.7: Cross-sectional images (compressed state) of the base MPLs, with inserts at higher magnification.  At very low compression, the base MPLs show the following trend in through-plane resistance: RGO > CB > GR > GN (Figure 5.5 (a)). In the high compression region, the through-plane resistances become very similar though  (as shown in Figure 5.5 (b)), with the exception of RGO which still remains almost ten times higher at this point (50 mΩ cm2 for RGO versus 4 – 6 mΩ cm2 for the other MPLs). The lower conductivity for RGO, compared to GN and GR, is consistent with reports from literature which attribute the difference to the presence of functional groups and defects on RGO’s surface[112,165].  Interestingly, the through-plane resistances of all the MPLs show the same extreme response during compression, decreasing by 90 – 95%. Mathematically, this can be related to a power law representation of the resistance-compression relationship, with the relative change in resistance directly related to the exponent of the power law. These are all fairly similar, ranging between -0.73 and -0.94 for the different MPLs (refer to Figure 5.8). The pre-exponential factors, in turn, do not bear connection to the relative change, but only represent the scale/magnitude of the resistance measurements. Similar power law dependencies have also been illustrated between stress and strain of bare GDLs[166]. Chapter 5: GDL-based MPLs and their composite effects     99   Figure 5.8: Power law dependencies between the through-plane resistance and compression of the base MPLs.  The RGO MPL also has the highest in-plane resistivity at 9.2 ± 0.5 mΩ cm, in the compressed state (Figure 5.5 (b)). The overall trend in in-plane resistivity data (RGO > CB > GR > GN) furthermore shows a strong linear correlation with MPL thickness with correlation coefficients of 0.97 and 0.98 in the compressed an uncompressed states respectively (also shown in Figure 5.9). This implies that the inherent MPL compactness (as reflected in the MPL thickness) is a determining factor for the in-plane resistivity of all these carbon-based layers. The interfacial resistance of the base MPLs (Figure 5.5 (c)) furthermore show some agreement with in-plane and through-plane resistance measurements, with RGO having the highest interfacial resistance and GN the lowest, while that of GR and CB are fairly comparable. Consistent with observations of the freestanding GN foam (Chapter 4), the GN MPL is also associated with the highest degree of electrical connectivity by demonstrating the lowest interfacial resistance of all MPLs in the dataset, including composite MPLs.  Chapter 5: GDL-based MPLs and their composite effects     100   Figure 5.9: Association between in-plane resistivity and thickness for the base MPLs.  The compressed state apparent porosities (Figure 5.5 (d)) indicate that the base MPLs can be classified into two broad ranges: the high porosity MPLs, CB and GR, with porosities of 72% ± 4% and 74% ± 8%; and the low porosity MPLs, RGO and GN, with porosities of 43% ± 4% and 37% ± 5% respectively. The low porosity of the GN MPL is attributed to its flakes that stack very densely. This is further exacerbated by the flakes’ tendency to assemble into larger sized planar structures. In addition to being the least porous layer, the GN MPL also has the lowest gas permeability (3.63 x 10-3 ± 3.26 x 10-5 darcy in the compressed state),  which is approximately a 1000 times smaller than that of the most permeable MPLs, CB and GR (also refer to normalized comparisons in Figure 5.5). The extremely low gas permeability of GN is attributed to the fact that the pores lie primarily in the longitudinal direction between tightly stacked flakes. The resultant mass transport pathways are therefore highly disconnected and tortuous. The general inverse relationship between permeability and tortuosity are illustrated in relationships such as Chapter 5: GDL-based MPLs and their composite effects     101  the one defined by Kozeny-Carman for porous media[167]: 𝑘𝑓 ∝𝜀3𝜏𝑃𝑀2 , where 𝜏𝑃𝑀 and 𝜀 represents the tortuosity and porosity, respectivelyx.  Although both RGO and GN MPLs consist of stacked flakes in their compressed states, the RGO particles are larger and more irregularly shaped. As previously shown, RGO flakes also do not stack and agglomerate to the same extreme levels as GN, likely due to RGO’s functionalization and surface defects. These factors account for RGO’s gas permeability, which, while still in the low range, is 100 times higher than for the GN MPL. This also points toward the presence of larger and/or more connected pore spaces. It is also noteworthy to mention that RGO’s gas permeability (similar to its surface roughness) is significantly affected by compression and reduces by 70% between the uncompressed and compressed state.  While the CB and GR MPLs have similar porosities, the highest compressed state gas permeability is associated with the CB MPL at 2.51 ± 0.05 darcy (Figure 5.5 (b)). This suggests that the fine pores of the CB MPL are better connected than the larger pores observed between the irregularly shaped GR particles (refer to Figure 5.7). The compressed state water permeabilities also follow the same general trend as gas permeability, and suggest a similar dependence on overall layer structure (CB > GR > RGO > GN as depicted in Figure 5.5 (e)). The high water permeability of the CB MPL (0.730 ± 0.01 darcy) is furthermore believed to be promoted by the small diameter of its pore spaces which is understood to improve capillary action and water drainage (refer to discussion on Young-Laplace equation in Section 4.6). In the absence of this increased capillarity, and with a similar degree of porosity, the GR MPL’s water permeability is lower at 0.376 ± 0.01 darcy.                                                       x The Kozeny-Carman relation is provided as an example to show how (for a given porous structure) a decrease in permeability (and decrease in porosity) can generally be associated with an increase in tortuosity. Other factors, such as the interstitial area between particles and the channel geometry also play a role, however. Due to the complexity and variety of porous structures, there is no universal relation between these parameters, and the application of such relations is typically limited to specific particle/pore sizes and shapes. Chapter 5: GDL-based MPLs and their composite effects     102  Although the RGO and GN MPLs’ lower water permeability is primarily attributed to the layers’ extremely dense and tortuous structures, it is also expected that the materials’ wetting behaviour influences this parameter. Based on the contact angle measurements for the RGO and GN MPLs (which showed less hydrophobic behaviour in one of their respective compression states) one would expect that these materials also have a higher internal wettability than CB and GR. Increased internal wetting would definitely contribute to lower water permeability, but the exact effect cannot unfortunately be quantified based on surface contact angle measurements alone.  5.6.2 Performance results at 100% cathode RH  The base MPLs’ performance results at fully humidified conditions are presented in Figure 5.10. The performance in the kinetic region shows the following trend (close-up of kinetic region provided in insert of Figure 5.10 (a)): GN > RGO > GR > CB.  The improved performance of GN over CB is consistent with observations of the GN foam and commercial CB MPL, presented in Chapter 4. A more detailed discussion on the performance differences observed in the kinetic region is presented in Section 5.10.4.  The resistive loss associated with the ohmic region (identified as the linear region of the polarization curve) is also provided in Figure 5.10 (c). In the context of this study, the differences in ohmic loss can be related to differences in the MPL bulk resistance, MPL|CL interfacial resistance and membrane resistance, including the CL ionomer resistance (refer to Equation 2.12). Membrane resistance (RPEM) is furthermore assumed to be fairly similar for fully humidified membranes. Figure 5.10 (c) shows that the RGO MPL results in the highest ohmic loss at 100% RH, followed by CB, GR and GN (112, 101, 92 and 78 mV at 760 mA cm-2, respectively). This ohmic loss behaviour shows strong agreement with trends in the bulk resistance, also depicting the following decreasing order: RGO > CB > GR > GN. A strong linear correlation furthermore exists with the compressed state in-plane resistivity, depicting a correlation coefficient of 0.98 with ohmic loss at 500 mA cm-2 and 760 mA cm-2 (refer to Figure Chapter 5: GDL-based MPLs and their composite effects     103  5.11). It is furthermore insightful to see how the ohmic losses relate to actual polarization results. In this case, the polarization results show that the lowest ohmic region performance is partially attributed to the CB MPL, despite the RGO MPL having the highest ohmic loss. By correcting for ohmic losses (Figure 5.10 (d)), it can be seen that CB’s lower performance (up to 1400 mA cm-2), is a continuation of its lower kinetic region performance.   Figure 5.10: Performance results for the base MPLs at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance.  Chapter 5: GDL-based MPLs and their composite effects     104   Figure 5.11: Association between ohmic loss and in-plane resistivity (compressed state) for the base MPLs (at 100% cathode RH).  For the mass transport region, the most impressive performance is associated with the GR MPL with a maximum current density of 3800 mA cm-2 (Figure 5.10 (a)). This high performance is believed to be related to the layer’s high permeability and porosity. As demonstrated in Section 5.6.1, the CB MPL has even higher gas and water permeabilities than the GR MPL, though. Nonetheless, it achieves much lower current densities in the mass transport region (with ilim = 2800 mA cm-2). This suggests that the CB MPL may be ‘too’ permeable at these conditions and water is removed too fast from the CL interface to keep the membrane sufficiently hydrated. Dehydration induces additional performance losses through increased membrane resistance. It is, however, very difficult to confidently determine the extent of membrane dehydration at 100% RH. These effects are typically more exaggerated at lower cathode humidity, and will therefore be assessed in greater detail in the following section.  On the opposite end of the spectrum, GN achieves the lowest performance in the mass transport region with ilim = 1700 mA cm-2, as shown in Figure 5.10 (a). The reason for this is because the MPL has the lowest porosity, and more specifically, the lowest gas and water Chapter 5: GDL-based MPLs and their composite effects     105  permeabilities. The poor mass transport performancexi furthermore complements the notion that the GN flakes are stacked very densely, and that the longitudinal pore spaces between the flakes are very disconnected. Some pore spaces may furthermore be completely sealed off from mass transport, contributing to the layer’s low apparent porosity. The higher degree of tortuosity does not only impede oxygen diffusion to the CL, but also promotes water retention and subsequent flooding. The extreme degree of flooding (beyond 1400 mA cm-2) is also observed in the higher degree of performance variation (illustrated through larger error bars of standard deviation in Figure 5.10 (a)).  Even though the RGO MPL has an apparent porosity only slightly higher than the GN MPL, Figure 5.10 (a) shows that it has significantly improved mass transport behaviour (ilim = 3000 mA cm-2). This is because both its gas and water permeabilities were found to be two orders of magnitude higher than that of the GN MPL, due to larger and/or better connected mass transport pathways. The trend in maximum power densities (Figure 5.10 (b)) furthermore shows a strong dependency on mass transport behavior. GR also displays an exceptionally high maximum power density at 1277 mW cm-2, which is approximately 40%, 46% and 83% higher than that of CB, RGO and GN MPLs, respectively. The results indicate that the GR MPL may serve as a promising MPL alternative for applications at similar conditions.  5.6.3 Performance results at 20% cathode RH  Most MPLs show marked deterioration in performance at low cathode humidity (compare Figure 5.10 (a) and Figure 5.12 (a)). The exception is the GN MPL for which performance is maintained extremely well, even showing a 4% increase in the MPL’s maximum power density. This performance improvement is achieved through the reduction of flooding at these low humidity conditions. In contrast to the GN MPL, the CB MPL’s maximum power density reduces                                                      xi Although water transport itself can occur in both the vapor and liquid phase, reference to water induced mass transport limitations again refer mostly to water in the liquid phase. This is because of the dominance of liquid water flooding at high current densities in the TP-5 cell hardware (also refer to the explanation presented in Section A.4). Chapter 5: GDL-based MPLs and their composite effects     106  by 29% to 653 mW cm-2, the lowest of all the MPLs. The relative change in maximum power density for GR and RGO MPLs is similar (8% decrease).  Performance at low humidity is also closely related to the extent of membrane dehydration that occurs at 20% RH. This is also reflected in the ohmic loss’s relative change between 100% and 20% RH (compare Figure 5.10 (c) and Figure 5.12 (c)). The CB MPL, which undergoes the most significant performance loss, also shows the most extreme increase in resistance (47% at 760 mA cm-2). The GR and RGO MPLs again display similar increases in resistance (14% and 12% at 760 mA cm-2), while the GN MPL closely mimics its resistance at lower humidity (2% decrease at 760 mA cm-2). The implication is that the membrane in the GN configuration remains equally and sufficiently hydrated.   Long-term tests at low humidity also support the aforementioned findings (refer to Figure 5.12 (e) and (f)). Over the course of five hours, the GN MPL shows a unique ability to preserve performance with a voltage loss of only 1.4 mV h-1 and approximately 0.2 mΩ cm2 h-1 increase in resistance. For the CB MPL, the voltage loss and resistance increase is far more drastic, however (464 mV h-1 and 159 mΩ cm2 h-1). The CB MEA suffers damage after only 0.9 hours, yielding it incapable of producing any current thereafter. This confirms, beyond a doubt, that the CB MPL is the most prone to membrane dehydration of all the base MPLs. The extreme contrast in the CB and GN behaviour furthermore reinforces the understanding that membrane hydration is strongly related to the MPL’s water retention capabilities. This, in turn, is affected by the MPLs’ underlying structures and permeabilities.   Chapter 5: GDL-based MPLs and their composite effects     107   Figure 5.12: Performance results for the base MPLs at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. Chapter 5: GDL-based MPLs and their composite effects     108  As evidenced through characterization results, the CB MPL manifests as a high porosity layer with pores that are very well connected (supported by the fact that it has the highest permeabilities). It is also likely, as suggested by Nam et al.[57], that the small pores of the CB MPL limit the size of water droplets, thereby preventing water accumulation at the CL interface. At low humidity conditions (and possibly even high humidity conditions too) the faster water removal from the CL interface leaves an insufficient amount to hydrate the membrane. Insufficient hydration compromises proton transfer, increases membrane resistance and ultimately results in performance loss and/or membrane damage. In contrast, the GN MPL was shown to be the least porous and permeable of all layers. With very tortuous pathways created between the GN flakes, water is removed at a much slower rate. In fact, water is retained ‘too’ well – while enough water is retained to prevent losses from membrane dehydration, it also inhibits reactant access to the CL. Consequently, a significant amount of flooding still occurs, even at low humidity conditions where the current density is limited to a maximum of 1600 mA cm-2 (refer to Figure 5.12 (a)).    After the GN MPL, the RGO and GR MPLs consecutively maintain performance the best during long-term testing. The following overall trend is therefore established within the base MPL case study at low humidity: the lower the MPL permeability, the better its water retention capabilities, and the better its ability to keep the membrane hydrated and maintain performance.    5.7 Case study 2: Composite GR+CB  Three composite case studies were considered where CB was added to each of the alternative base materials to form GR+CB, RGO+CB and GN+CB MPLs, respectively. All the composite MPLs were prepared in 1:1 weight ratio and have a similar total loading (1.5 mg cm-2) as the base MPLs. In the following section, composite effects are also emphasized by evaluating the degree of relative change in material properties (or performance), with the addition of CB to the original base material. Chapter 5: GDL-based MPLs and their composite effects     109  5.7.1 Characterization results  At a glance, the surface of the composite MPL more closely resembles that of the CB MPL, showing similar partial coverage of the GDL (compare Figure 5.13 to Figure 5.4). This resemblance is also reflected in the surface roughness of the compressed MPLs, i.e. the composite’s surface roughness is more closely related to that of the CB MPL (see Figure 5.14 (b)). Wetting behaviour, particularly in the compressed state, also demonstrates a dominating influence by CB (Figure 5.14 (b)): while the GR MPL has a contact angle of 140° ± 3˚, the composite’s is similar to that of the original CB MPL (149° ± 2˚ and 147° ± 2˚ for CB and GR+CB MPLs respectively).    Figure 5.13: Surface images of the GR+CB MPL in (a) uncompressed and (b) compressed state; (c) Cross-sectional image (compressed state) of the GR+CB MPL; Higher magnification images are provided in the inserts.  The cross-sectional image of the GR+CB MPL indicates a structure very similar to that of the porous CB MPL, with slight irregularities introduced by the GR particles’ protruding edges (Figure 5.13 (c)). The composite layer has an intermediate thickness, which is 3% higher than that of the original GR MPL in the compressed state (Figure 5.14 (a) and (b)). In terms of Chapter 5: GDL-based MPLs and their composite effects     110  compressibility, though, the GR+CB MPL shows fairly similar behaviour to the single component MPLs (13%, 12% and 11% compressibility for the CB, GR and GR+CB MPLs, respectively).   Figure 5.14: Radar charts of material properties for the GR+CB case study in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the GR+CB case study.  The incorporation of CB furthermore results in a lower through-plane resistance, compared to the original GR MPL. In the compressed state, this corresponds to a relative reduction of 36% (note that the differences in Figure 5.14 appear fairly small as part of the normalized dataset). This effect is attributed to the fact that the extremely small CB particles fill the spaces between the larger GR particles. The space filling effect, in turn, helps to increase particle connectivity, creating shorter/more direct pathways for electron transport. The same effect is also illustrated Chapter 5: GDL-based MPLs and their composite effects     111  in the interfacial resistance (Figure 5.14 (c)), but is, however, not distinct when considering the in-plane resistivity (7.3 ± 0.6 mΩ cm and 7.2 ± 0.4 mΩ cm for the GR+CB and GR MPLs in the compressed state).   With regards to permeability, it is found that the composite MPL has higher gas and water permeabilities than the original GR MPL. The increase in gas permeability is very slight, however, with only a 4% relative increase in the compressed state (yielding a permeability of 1.14 ± 1.58 x 10-2 darcy for the GR+CB MPL). Water permeability changes induced by the addition of CB, are more distinct though (Figure 5.14 (e)), and show a 69% relative increase from that of the original GR MPL. This implies that the composite’s structure is uniquely tailored to increase water permeability. It therefore seems plausible that CB particles can also establish a network of small channels throughout the composite which help to accelerate water drainage. Combining CB and GR, furthermore does not seem to affect layer porosity significantly, likely due to the fact that the original MPLs also have average porosities between 72% and 73% (Figure 5.14 (d)).  5.7.2 Performance results at 100% cathode RH  Inspection of the GR+CB MPL’s polarization data (Figure 5.15 (a)) reveals that its performance lies between that of the CB and GR MPLs in all regions of the polarization curve. More specifically, its performance at fully humidified conditions is more closely related to that of the GR MPL. This trend is also reflected in the maximum power densities (Figure 5.15 (b)). Through the addition of CB, the maximum power density is found to decrease by 10% from the original GR MPL’s 1277 mW cm-2, to 1143 mW cm-2 for the composite MPL.  Chapter 5: GDL-based MPLs and their composite effects     112   Figure 5.15: Performance results for the GR+CB case study at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance.   Furthermore, despite the fact that the GR+CB MPL has the lowest in-plane, through-plane and interfacial resistance values for this case study, it demonstrates an intermediate ohmic loss, as shown in Figure 5.15 (c). The implication is that different extents of membrane hydration may exist amongst the three MEAs. Characterization results furthermore indicated slightly higher gas permeability for the GR+CB MPL, compared to the GR MPL. The trend also does not manifest in a mass transport improvement, as expected, however. Rather, in this case, mass Chapter 5: GDL-based MPLs and their composite effects     113  transport performance (GR > GR+CB > CB) shows a stronger association with the materials’ water permeabilities (GR < GR+CB < CB). This furthermore suggests that the GR+CB MPL (and the CB MPL) cannot retain enough water to keep the membrane sufficiently hydrated under the employed one-dimensional control. Upon dehydration, additional performance losses are therefore incurred with these MPLs, by increased membrane resistance and an accompanying decrease in proton concentration.   5.7.3 Performance results at 20% cathode RH  At low humidity, the maximum power density of the GR+CB MPL reduces to 767 mW cm-2. This is 35% lower than that of the GR-only MPL at the same conditions (Figure 5.16 (b)). The power density curve also illustrates how the composite’s performance now more closely resembles that of the CB MPL. The resemblance is reiterated by considering the composite MPL’s relative change between high and low humidity polarization (comparing Figure 5.15 (a) and Figure 5.16 (a)). For example, at 1400 mA cm-2, the GR+CB MPL’s performance decreases by 16%. This is only slightly less than the CB MPL’s 18% decrease at the same point. In turn, the GR MPL experiences a smaller decrease of only 7%. Similarly, when considering ohmic losses (Figure 5.15 (a) and Figure 5.16 (c)), the composite and CB MPLs show increases of 43% and 47% at 760 mA cm-2, compared to a smaller increase of 14% for the GR MPL. The greater performance loss and resistance increase for the composite MPL (compared to the original GR MPL) is attributed to its increased water removal capability, which accelerates membrane dehydration at low humidity.  This behavior is also reflected during longer-term testing, with the composite showing a performance loss of 59 mV h-1 (Figure 5.16 (e)).  This is a significant increase from the GR MPL’s 30 mV h-1 loss. It is interesting to note that the GR+CB MPL demonstrates a similar initial slope in voltage and resistance behaviour as the CB MPL during longer-term testing. However, at the point where the CB performance drops off completely (~ 0.8 h), the GR particles in the composite lend the additional water retention necessary to establish a more gradual decline.  Chapter 5: GDL-based MPLs and their composite effects     114   Figure 5.16: Performance results for the GR+CB case study at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. Chapter 5: GDL-based MPLs and their composite effects     115  Another interesting consideration from this case study, is the fact that the three investigated MPLs have porosities in the same high range (with average apparent porosities between 72% - 74%), but are associated with very different mass transport and long-term behaviour. This emphasizes the complexity of mass transport in PEMFCs and the fact that properties such as porosity cannot be considered in isolation. Rather, it remains important to include properties that reflect actual flow/transport behaviour and are influenced by the materials’ inherent porous structures and tortuosity. Ultimately, results of the GR+CB MPL case study indicate that there is no performance benefit gained from the addition of CB to GR, for either high or low humidity applications under one-dimensional control.  Rather, its presence introduces performance losses associated with membrane dehydration, which become particularly severe at low humidity conditions.  5.8 Case study 3: Composite RGO+CB  5.8.1 Characterization results  As with RGO, the GDL surface is completely covered by the RGO+CB composite material (refer to Figure 5.17 (a) and (b)). Close inspection of the composite’s surface in the uncompressed state (insert of Figure 5.17 (a)) furthermore reveals that CB particles primarily collect in the fissures between the highly irregular RGO particles. Surface roughness measurements, illustrated in Figure 5.18 (a), also bear evidence of this filling effect, with the roughness of the composite MPL significantly lower than that of the RGO-only MPL in the uncompressed state (7.6 ± 0.7 μm vs. 19.5 ± 1.0 μm). Upon compression, the CB particles seemingly become trapped in the fissures, though, and the surface roughness of the two layers appears similar (at approximately 3.1 μm). There is furthermore no discernable difference between the two MPLs’ wettability in the compressed state (145° ± 2° and 147° ± 2° contact angle for the RGO and RGO+CB MPL respectively).   Chapter 5: GDL-based MPLs and their composite effects     116  The cross-sectional structure of the composite MPL seems denser than that of the RGO-only MPL, with CB appearing to occupy the interstitial space between RGO flakes (Figure 5.17 (c)). Substituting half of the RGO with CB furthermore reduces the total thickness by 10% in the compressed state. Nevertheless, the composite MPL is still associated with a fairly high degree of compressibility (16%), similar to the original RGO MPL (19%).   Figure 5.17: Surface images of the RGO+CB MPL in (a) uncompressed and (b) compressed state; (c) Cross-sectional image (compressed state) of the RGO+CB MPL; Higher magnification images are provided in the inserts.  As with the GR+CB composite, the addition of CB to RGO helps to reduce through-plane resistance (Figure 5.18 (a) and (b)). Although this reduction is the most significant amongst the three composites (64% in the compressed state), the composite’s final through-plane resistance of 18.3 mΩ cm2 still remains very high, compared to most other MPLs at 4 – 6 mΩ cm2 (the trends in through-plane resistance are more clearly illustrated in Appendix C.2). A similar reducing effect is displayed for in-plane resistivity, but is less significant, yielding a 9% relative decrease only. The RGO+CB MPL furthermore has a lower interfacial resistance than the RGO MPL, and surprisingly, also the CB MPL (Figure 5.18 (c)). It is possible that the composite’s much Chapter 5: GDL-based MPLs and their composite effects     117  smoother surface (surface roughness 9.6 μm lower than for the CB MPL in the compressed state) contributes to this phenomenon.   Figure 5.18: Radar charts of material properties for the RGO+CB case study in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the RGO+CB case study.  Unlike with the GR+CB MPL, where the addition of CB increased gas permeability, it is found to cause a 67% relative decrease in the RGO+CB MPL’s case, in the compressed state (also refer to normalized positions in Figure 5.18 (a) and (b)). The behavior agrees with visual observations which show that the CB particles fill the pores between RGO particles, causing greater obstruction to gas flow. This concept is also supported by apparent porosity measurements (Figure 5.18 (d)) which indicate that the composite has slightly less available open space for gas transport (average porosity of 37% and 43% for RGO+CB and RGO MPLs respectively). The trend Chapter 5: GDL-based MPLs and their composite effects     118  in water permeability (Figure 5.18 (e)) is furthermore consistent with that of the GR+CB case study: the addition of CB results in higher water permeability for the composite, compared to the original RGO MPL. In this case, the relative increase in water permeability is 51%. This result is consistent with the notion that CB particles establish a network of pore spaces which serve as more effective pathways for water transport.  5.8.2 Performance results at 100% cathode RH  The performance results for the RGO+CB case study at 100% cathode RH are provided in Figure 5.19. The polarization curve of the composite MPL displays significant performance gains in the ohmic and mass transport region, compared to the RGO MPL (Figure 5.19 (a)). This result is different from that observed for the GR+CB case study, in which the composite induced performance losses, compared to the original GR MPL. As illustrated in Figure 5.19 (b), RGO+CB MPL’s synergisticxii effect also manifests in a 22% relative increase in the maximum power density (from 876 mW cm-2 for the RGO MPL to 1071 mW cm-2 for the composite MPL). For the most part, the kinetic performance of the RGO and RGO+CB MPLs appears to be very similar (refer to the close-up of the kinetic region in the insert of Figure 5.19 (a)).  The RGO+CB MPL’s improvement in ohmic performance corresponds to a 12% decrease in ohmic losses at 760 mA cm-2, relative to the RGO MPL (Figure 5.19 (c)). Smaller ohmic losses can be ascribed to the fact that the composite MPL has lower though-plane and in-plane resistance. The trend in ohmic loss (RGO > CB > RGO+CB), does however, not completely reflect the trend in material bulk resistance (RGO > RGO+CB > CB). The fact that a larger ohmic loss is associated                                                      xii Synergism in the context of this study relates to the phenomenon where the combined effect of two MPL materials is greater than what is expected. The rule of mixtures is used to provide an estimation of the materials’ combined effect in composite MPLs. For example, for the RGO+CB composite MPL, the expected maximum power density is calculated as the weighted average of the CB and RGO MPLs’ maximum power densities, with the weightings based on the 1:1 weight ratio ((0.5 x 842) + (0.5 x 876) = 859 mW cm-2).  In reality, the RGO+CB MPL’s maximum power density (1071 mV cm-2) is more than 200 mW cm-2 higher than the value estimated from the rule of mixtures. The composite effect is therefore described as synergistic. Chapter 5: GDL-based MPLs and their composite effects     119  with CB than expected (based on material resistance alone), reinforces the belief that the CB MEA experiences higher membrane resistance than the other MEAs.    Figure 5.19: Performance results for the RGO+CB case study at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance.  As shown in Figure 5.19 (a), the composite’s synergistic effect also extends into the mass transport region. This is illustrated by ilim increasing from 3000 mA cm-2 for the RGO-only MPL, to 3400 mA cm-2 for the composite MPL. When only considering gas permeability, this Chapter 5: GDL-based MPLs and their composite effects     120  performance gain seems counter-intuitive though, since RGO+CB has the lowest gas permeability of all the MPLs in this case study. Water permeability measurements, however, again provide greater insight into the performance behavior. For the composite MPL, in particular, the inclusion of CB was shown to accommodate improved water transport. CB again helps to minimize the amount of flooding via improved water drainage compared to the RGO-MPL. At the same time, gas transport also does not appear compromised though, despite the composite having the lowest gas permeability. This furthermore supports the notion of a dual porous structure within certain CB composites which specifically help to enhance water transport, while also allowing for sufficient gas transport (originally suggested by Wang et al.[156] for a composite consisting of acetylene black and Black Pearls 2000). This may be achieved with gas primarily transported through macropores (located between the larger RGO flakes in this case), whereas water is preferentially transported through the CB particles’ micropores. Ultimately, all of the aforementioned results indicate that the RGO+CB composite MPL offers numerous advantages for high humidity application.  5.8.3 Performance results at 20% cathode RH  A comparison of high and low humidity polarization shows that the composite MPL undergoes a more significant voltage loss than the RGO-only MPL (compare Figure 5.19 (a) and Figure 5.20 (a)). At 2800 mA cm-2, for example, the voltage of the RGO+CB and RGO MPLs decrease by 36% and 23%, respectively. The associated maximum power densities still rank in the following order though (as also shown in Figure 5.20 (b)): RGO+CB (932 mW cm-2) > RGO (808 mW cm-2) > CB (653 mW cm-2).   Chapter 5: GDL-based MPLs and their composite effects     121   Figure 5.20: Performance results for the RGO+CB case study at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. Chapter 5: GDL-based MPLs and their composite effects     122  Longer-term tests, depicted in Figure 5.20 (e), also indicate that the composite’s durability at low humidity is more compromised, displaying a voltage loss of 27 mV h-1 compared to the RGO MPL’s loss of 17 mV h-1. The addition of CB in a 1:1 weight ratio, while considered advantageous to enhance mass transport at 100%RH, therefore proves less favorable for low humidity conditions. This is because of the accelerated water removal in the vicinity of the membrane and its associated drying effect. Evidence of accelerated drying for the composite is also provided in Figure 5.20 (f), depicting a higher increase in resistance, compared to the RGO MPL (9 vs. 6 mΩ cm2 h-1).  It should be noted, however, that the composite’s durability still remains in extreme contrast to that of the CB MPL.  5.9 Case study 4: Composite GN+CB  5.9.1 Characterization results  While the GN MPL appears as the smoothest and most uniform surface, the GN+CB surface seems slightly more fragmented and uneven (compare images for GN in Figure 5.4 to Figure 5.21 (a) and (b)). This difference is also evidenced by surface roughness measurements exhibiting a 132% relative increase with the incorporation of CB, in contrast to the original GN MPL’s low surface roughness of 1.5 ± 0.3 μm (reference made to compressed state, normalized positions also depicted in Figure 5.22 (b)). The presence of CB furthermore leads to an increase in contact angle, and is most clearly demonstrated considering the GN and GN+CB MPLs in their compressed states (with the contact angle increasing from 112° ± 5° to 143° ± 2°). In addition, the composite MPL is found to mimic the Wenzel wetting behaviour of the original GN MPL, showing increased wettability upon compression. It is again possible that some degree of air pocket formation (Cassie-Baxter model) may also exist, however. The respective change in contact angle after compression is much smaller for the composite though, compared to the GN-only MPL (5% vs. 20% relative change).   Chapter 5: GDL-based MPLs and their composite effects     123  Figure 5.21 (c) illustrates that the GN+CB MPL has a similar cross-sectional structure than the RGO+CB MPL (although far more compact), with CB particles filling the interstitial space between the larger particles. The composite layer’s overall thickness is furthermore found to be slightly higher than that of the GN MPL as indicated in Figure 5.22 (a) and (b) (6% relative increase in both uncompressed and compressed state). Both the GN and GN+CB layers furthermore show a similar low degree of compressibility (also 6%).    Figure 5.21: Surface images of the GN+CB MPL in (a) uncompressed and (b) compressed state; (c) Cross-sectional image (compressed state) of the GN+CB MPL; Higher magnification images are provided in inserts.  The trend in resistance for CB-composites remains consistent in this case study: Both the through-plane resistance and in-plane resistivity are lower in the GN+CB MPL’s case, compared to the GN MPL (note that the differences in Figure 5.22 (a) and (b) once again appear fairly small as part of the normalized dataset). Interestingly, amongst all the investigated MPLs, the GN+CB MPL is also associated with the lowest through-plane resistance for majority of the compression range (as illustrated in Appendix C.2), as well as the lowest in-plane resistivity (6.5 ± 0.2 mΩ cm) in the compressed state. The interfacial resistance of the composite MPL is Chapter 5: GDL-based MPLs and their composite effects     124  slightly higher than for the GN MPL though (Figure 5.22 (c)), and may potentially be accounted for by its rougher surface.    Figure 5.22: Radar charts of material properties for the GN+CB case study in (a) uncompressed and (b) compressed state; (c) Interfacial resistance, (d) porosity and (e) water permeability results for the GN+CB case study.  A direct consequence of CB particles collecting between the GN flakes in the composite, is a 41% relative decrease in gas permeability, compared to the GN-only MPL. In fact, at 2.15 x 10-3 ± 5.65 x 10-5 darcy in the compressed state, the GN+CB MPL has the lowest gas permeability of all the investigated MPLs. The reduction in gas permeability is furthermore accompanied by a decrease in the average apparent porosity, from 37% for the GN-only MPL, to 24% for the composite MPL (Figure 5.22 (d)). The lower porosity is also indicative of the fact that less open space is available for gas transport in the GN+CB MPL.  Chapter 5: GDL-based MPLs and their composite effects     125  Similar to the other two composite case studies, the water permeability of the GN+CB MPL is also higher compared to its original constituent MPL (Figure 5.22 (e)). The degree to which water permeability is improved through the addition of CB (45%), is furthermore very similar to that observed for the RGO+CB case study (51%). This improvement in water permeability once again reinforces the notion that CB particles create unique pathways by which water flow/drainage is improved.  5.9.2 Performance results at 100% cathode RH  As shown in Figure 5.23 (a), adding CB to GN also has an extreme synergistic effect, with the most obvious change being the significant increase in the mass transport region. For instance, ilim increases from 1700 mA cm-2 for the GN-only MPL to 3200 mA cm-2 for the composite MPL.  The mass transport improvement is also complemented by a 68% increase in the maximum power density, from 697 mW cm-2 for the GN MPL, to 1173 mW cm-2 for the composite MPL (Figure 5.23 (b)). These remarkable enhancements are primarily attributed to the improved water permeability of the composite MPL. Bearing in mind that water permeability was improved to a similar extent in the RGO+CB case study with the addition of CB (51%), the performance enhancement in this particular case appears even more extreme due to the severe flooding experienced by the original GN MPL. Once again, the impression is created that gas transport is not compromised under these circumstances, despite the fact that the composite MPL also has the lowest gas permeability and porosity of all the investigated MPLs. These trends again support the concept of a dual porous structure which helps to effectively manage gas and water transport via macroporous and microporous pathways, respectively.   In the kinetic region, the performance of the GN and GN+CB MPLs appears fairly similar (a close-up of the kinetic region is provided in the insert of Figure 5.23 (a)). It is generally difficult to identify a noticeable difference between the ohmic performance for the GN+CB and GN MPLs, since the GN MPL’s ohmic region is very compressed due to early onset flooding. The differences in the layers’ in-plane and through-plane resistances also do not manifest in Chapter 5: GDL-based MPLs and their composite effects     126  significant differences in the ohmic loss, with that of the composite only being slightly higher (Figure 5.23 (c)). Of all the GDL-based MPLs considered in this phase of the study, GN and its composite MPL are associated with the lowest ohmic loss.    Figure 5.23: Performance results for the GN+CB case study at 100% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss and (d) iR-corrected performance.   Chapter 5: GDL-based MPLs and their composite effects     127  5.9.3 Performance results at 20% cathode RH  Comparison of high and low humidity polarizations (compare Figure 5.23 (a) and Figure 5.24 (a)), shows that the GN+CB MPL maintains performance exceptionally well, similar to the GN MPL. As consequence, the GN+CB composite also achieves the highest power density of all the MPLs at low humidity (1188 mW cm-2 as indicated in Figure 5.24  (b)). This maximum power density also represents a remarkable 81% improvement over the CB MPL at low humidity.  The composite furthermore shows virtually no change in ohmic loss during low humidity polarization (compare Figure 5.23 (c) and Figure 5.24 (c)).   The aforementioned results suggest that the outstanding performance of both the GN-based MPLs at low humidity can be attributed to better water retention at the CL interface, which ensures that the membrane remains sufficiently hydrated. This behavior is a consequence of the MPLs’ morphologies and porous structures: the dense stack of GN flakes close to the membrane creates more tortuous pathways, decreasing water permeability and making it more difficult to “lose” water through the MPL under drier conditions. As previously shown, for GN in isolation, the high water retention can cause unfavorable flooding. However, for the composite MPL, a balance is maintained between the water-retention capabilities of the GN flakes and the CB, which creates the additional pathways required for sufficient water removal. This further suggests that the GN:CB ratio can be tailored to balance the two competing phenomena of water removal and water retention, as required for specific applications and operating conditions. For longer-term testing, the composite MPL displays a performance loss and resistance increase of 13 mV h-1 and 3 mΩ cm2 h-1 (Figure 5.24 (e) and (f)), which are the second lowest of all the GDL-based MPLs, after GN.    Chapter 5: GDL-based MPLs and their composite effects     128   Figure 5.24: Performance results for the GN+CB case study at 20% cathode RH: (a) polarization performance, (b) power density curves, (c) ohmic loss, (d) iR-corrected performance, (e) voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2. Chapter 5: GDL-based MPLs and their composite effects     129  5.10 Global trends and other considerations  Throughout the discussion of the different case studies, additional commonalities and global trends (concerning all GDL-based MPLs), have come to light. These trends, and other important considerations, are summarized and discussed in greater detail in the following sections:   5.10.1 Effect of CB addition on composite properties  For the composite MPLs (prepared in a 1:1 weight ratio), the addition of CB to the alternative base materials (GR, RGO and GN), have demonstrated the following general effects:   Decreases surface wettability – CB generally introduces additional hydrophobicity to composite surfaces and results in larger contact angles (Figure 5.25 (a)).  Decreases through-plane resistance – CB helps to establish additional contact points between the larger particles of GR, RGO and GN, thereby creating more direct pathways for electron transport (Figure 5.25 (b)).  Decreases apparent porosity and gas permeability – CB fills the pore spaces between RGO and CB (mostly orientated in the longitudinal direction between stacked flakes), resulting in a reduction of the total and/or connected void volume (Figure 5.25 (c)). In the case of GR, CB addition does not induce significant changes in these properties, likely due to the fact that GR and CB have comparable porosities.  Increases water permeability – CB creates a network of small and well connected pore channels which introduces more effective water removal pathways (Figure 5.25 (d)).  Chapter 5: GDL-based MPLs and their composite effects     130   Figure 5.25: General effects on composite properties resulting from CB addition in a 1:1 weight ratio (the schematic representations are not to scale).  5.10.2 Dominant components of composite properties  In the previous section, the typical responses to the addition of CB were summarized. To better grasp the size of the induced changes, the dominant components of composite properties are also identified in the following section. In the context of this study, the dominant component is defined as the material that has the greatest influence on a composite’s property. The dominant components were determined based on the average values of material properties and the following general trends were identified from the summary, presented in Table 5.6, Table 5.7 and Table 5.8:  Chapter 5: GDL-based MPLs and their composite effects     131   Surface wettability – CB has a dominating influence on the measured contact angle, imparting its higher hydrophobicity.  Surface roughness – In the compressed state, composites containing planar shaped particles/flakes (RGO+CB and GN+CB), are dominated by their presence, also resulting in more uniform coverage of the GDL.  Thickness – Composite MPLs have layer thicknesses in-between that of the original components. The component with the highest volumetric density also typically dominates the property. For example, in the GN+CB MPL, GN has the highest volumetric density. The composite’s thickness is therefore more closely related to that of the original GN MPL.   In-plane resistivity – Following the trend in thickness, the composite MPLs’ in-plane resistances are also dominated by the component with the highest volumetric density (it should be noted that trends in thickness and in-plane resistivity are less distinct when measurement variance is taken into account).  Through-plane resistance – In both the uncompressed (or ‘slightly’ compressed) and fully compressed states, the through-plane resistance of composites are dominated by the least resistant material. In the high compression region, CB presents as the most conductive base MPL, therefore exerting a dominating influence on all the composite MPLs.   Interfacial resistance – Similar to the compressed state through-plane resistance, CB also dominates the interfacial resistance of the composite MPLs (this can be understood based on the fact that interfacial resistance essentially also presents as a through-plane resistance).  Gas permeability – Composite MPLs’ gas permeabilities are dominated by the component consisting of larger sized particles (GR, RGO and GN), all with lower inherent gas permeability than CB.  Water permeability – For materials in the low porosity range (RGO+CB and GN+CB), water permeability is dominated by the respective base material (RGO and GN). For GR in the high porosity range, water permeability is dominated by CB. Chapter 5: GDL-based MPLs and their composite effects     132   Porosity – Similar to gas permeability, the composite MPLs’ apparent porosities appear to be dominated by the larger sized particles (this is less clear for the GR+CB study though, due the close proximity of their average values and the larger variance associated with the measurements).   Table 5.6: Summary of dominant components on surface properties and thickness, based on average values (the schematic representations are not to scale).  *Dominant component could not be identified due to similar averages        Chapter 5: GDL-based MPLs and their composite effects     133       Table 5.7: Summary of dominant components on electrical properties, based on average values (the schematic representations are not to scale).         Chapter 5: GDL-based MPLs and their composite effects     134   Table 5.8: Summary of dominant components on structural and transport properties, based on average values (the schematic representations are not to scale).   5.10.3 Compression  While the important role of compression in fuel cell performance is well known (and also proven in Section 4.6), this study specifically highlights its significance during the evaluation of different materials properties. Through both SEM images and characterization results, it is shown that the MPLs all respond very differently to the same degree of compression. The RGO MPL displays the most extreme response (70% and 84% decrease in gas permeability and surface roughness, for example). In fact, changes for this material are so extreme that it seemingly presents completely different features in the uncompressed state. Initially RGO has a highly hydrophilic surface, compared to a nearly superhydrophobic surface in its compressed Chapter 5: GDL-based MPLs and their composite effects     135  state. Particles also appear to undergo a morphological transition, from highly irregular particles that assemble in clusters, to smoothened stacks of planar flakes.  Certain material properties are furthermore affected by compression more than others, including through-plane resistance (90 – 95% relative decrease), surface roughness (30 – 80% relative decrease) and gas permeability in certain cases (10 – 70% relative decrease)xiii. On the other hand, total thickness and in-plane resistivity generally only decrease by 5 – 20% upon compression. The inclusion of CB in composite MPLs furthermore results in smaller degrees of relative change upon compression.   Even though MPLs are not compressed uniformly (due to the presence of flow field channels), the compressed state properties are considered to dominate and provide a more accurate reflection of the environment experienced during actual fuel cell operation. Assessment of the uncompressed state does, however, remain important to gain a better sense of the potential spread in properties. During the investigation of these novel MPLs, the assessment of both compressions states also helped to understand the transition from the original raw material/powder to the final MPL in compressed form.  5.10.4 Kinetic behavior  The behavior in the kinetic region was investigated in more detail by performing a Tafel analysis based on the polarization curves up to a current density of 200 mA cm-2 (Equation 2.11). The anodic surface potential was assumed to be negligible and polarization points were iR-corrected. The resultant Tafel slopes are provided in Table 5.9, while the Tafel plots are presented in Appendix C.3. The current density, at a voltage of 0.9 V (for the iR-corrected curve), is also provided in Table 5.9.                                                      xiii Porosity and water permeability are likely also significantly affected by compression, but this could unfortunately not be evaluated due to the fact that the properties could only be measured in the compressed state. Back-calculation of the uncompressed state porosity (as a function of the change in thickness) was furthermore avoided, due to the large degree of variance associated with cumulative uncertainty effects. Chapter 5: GDL-based MPLs and their composite effects     136  Table 5.9: Kinetic parameters associated with the different GDL-based MPLs at 100% and 20% cathode RH. MPL 100% Cathode RH 20% Cathode RH Apparent Tafel slope (mV dec-1) iE = 0.9 V, iR-corrected (mA cm-2) Apparent Tafel slope (mV dec-1) iE = 0.9 V, iR-corrected (mA cm-2) CB 107 42 124 49 GR 100 60 116 54 RGO 97.6 73 103 72 GN 91.3 79 95 84 GR+CB 101 57 117 51 RGO+CB 98.0 70 104 66 GN+CB 91.2 79 98 76  Table 5.9 indicates that the apparent Tafel slopes fall within the typical expected range for the ORR with Pt catalysts (60 – 120 mV dec-1)[155]. In all instances, the GN-based MPLs are associated with the highest current density at 0.9 V, as well as the lowest Tafel slopes. In contrast, the CB MPL has the highest Tafel slopes and achieves the lowest current densities at 0.9 V. The following overall trend is furthermore observed in the Tafel slope, at both high and low humidity conditions: GN < GN+CB < RGO < RGO+CB < GR < GR+CB < CB. A similar trend is also displayed for the water permeability of the MPLs, suggesting an association between the parameter and the Tafel slope (as illustrated in Figure 5.26). The influence of water permeability likely manifests in the kinetic region due to its impact on ionic conductivity under one-dimensional control: as water permeability decreases and the water retention capability of a MPL increases, the water concentration near the CL also increases, resulting in a higher proton concentration through more effective membrane hydration. The association also appears more distinct under the harsher and drier conditions associated with low humidity operation.    Chapter 5: GDL-based MPLs and their composite effects     137   Figure 5.26: Association between apparent Tafel slope and water permeability of the GDL-based MPLs at 100% and 20% cathode RH.  5.10.5 Ohmic behavior  As previously mentioned, the differences in ohmic resistance (or ohmic loss) can be related to differences in the MPL bulk resistance, MPL|CL interfacial resistance and membrane resistance (refer to Equation 2.12). It remains very difficult, however, to determine the exact contribution of each resistive element. To provide some insight, the OCV resistance was also measured at fully humidified conditions using a lower cathode air flow rate of 0.39 L min-1 (to minimize the possibility of membrane dehydration). The OCV resistances (Table 5.10) were found to show a strong association with the in-plane resistivities (particularly in the compressed state) of the different MPLs, resulting in a linear correlation coefficient of 0.97 (as shown in Figure 5.27). Since in-plane resistivity and MPL thickness are also closely related, a similar linear dependence exists between the two variables, with a correlation coefficient of 0.91.   Chapter 5: GDL-based MPLs and their composite effects     138  Table 5.10: OCV resistances of the GDL-based MPLs at 100% cathode RH. MPL OCV resistance (mΩ cm-2) CB 141 GR 129 RGO 154 GN 115 GR+CB 132 RGO+CB 144 GN+CB 116   Figure 5.27: (a) Association between OCV resistance at 100% cathode RH and in-plane resistivity (compressed state) for the GDL-based MPLs; (b) Association between OCV resistance at 100% cathode RH and thickness (compressed state) for the GDL-based MPLs.  During polarization, the ohmic loss at fully humidified conditions again shows the strongest association with the in-plane resistance, in the compressed state (Figure 5.28). The observed association is not exact, though, since membrane dehydration may play a more significant role at this point. For example, during the evaluation of different case studies, it appeared that the CB and GR+CB MPLs experienced some membrane dehydration at fully humidified conditions Chapter 5: GDL-based MPLs and their composite effects     139  (refer to Section 5.7.2). Their membrane resistances (RPEM) would therefore be much higher than for the other MEAs, which likely all experience very high degrees of, if not complete, membrane hydration (due to lower water permeabilities). Incidentally, all the correlation coefficients displayed in Figure 5.28 increase to 0.94 in the absence of the CB and GR+CB data points.    Figure 5.28: Association between in-plane resistivity (compressed state) and ohmic loss at current densities of 500 mA cm-2, 760 mA cm-2 and 1000 mA cm-2 (at 100% cathode RH).  In general, the through-plane and interfacial resistance do not display a similar association with ohmic loss. For example, while RGO demonstrates significantly higher through-plane and interfacial resistances than all other MPLs, it does not result in equally high differences in ohmic losses (a case in point is how RGO’s through-plane and interfacial resistance is approximately 10 and 2 times higher than that of GR, but its ohmic loss is only 1.2 times higher than that of GR, which corresponds to its difference in in-plane resistivity). The apparent dominance of the MPL’s in-plane resistance on ohmic loss differences, ultimately also suggests that the contribution of the interfacial resistance might be very small. The notion also agrees with predictions from most modelling studies, as highlighted in Section 2.9. The dominating influence of the in-plane resistance (and also thickness) is not necessarily considered universal, Chapter 5: GDL-based MPLs and their composite effects     140  however, since its prominence in this study may only have manifested due to the fact that very similar layers (all porous and carbon-based) were employed.   During low humidity operation, differences in membrane resistance typically contribute more significantly to differences in ohmic loss. This was definitely the case in this study, and is best illustrated by considering the resistance increase during longer-term operation at 20% RH. Since bulk resistances remain the same at these conditions, the observed resistance increase can be directly related to an increase in membrane resistance (or membrane dehydration). Naturally, membrane dehydration is strongly influenced by the MPLs water retention capability, establishing the following general trend for the studied MPLs: the more water permeable an MPL (or the lower its water retention capability), the higher the degree of membrane dehydration and resistance increase at low humidity conditions (CB > GR+CB > GR > RGO+CB > RGO > GN+CB > GN).   5.10.6 Interfacial characteristics  Similar to the work with freestanding MPLs (Chapter 4), this study emphasizes several important interfacial characteristics, including: interfacial morphology, MPL wettability, adhesive effects, compression and interfacial resistance (discussed in the previous section). Interfacial morphology, in particular, is found to contribute significantly to overall performance. In this context, the interfacial morphology does not only refer to the MPL’s roughness profile, but also to the type of morphological structures presented at the interface. These furthermore relate to the ability of the MPL structure to retain water in the vicinity of the CL and are divided into two extreme groups: water draining structures (porous network CB, for example) and water retaining/pooling structures (such as the large planar and tortuous structures of RGO and GN). Naturally, employing materials that are inherently more hydrophilic would also push materials towards the water retention side of the scale. As demonstrated in the case studies, water retaining structures are typically not beneficial for mass transport, due to increased flooding. These structures do, however, prove very beneficial for low humidity performance Chapter 5: GDL-based MPLs and their composite effects     141  preservation under one-dimensional control. Introducing both water draining and retaining structures at the interface, can furthermore help to manage mass transport more effectively, as required by different applications or operating conditions. The concept that interfacial morphology can significantly influence performance by affecting water retention capabilities, has also been proposed by several modelling studies on the MPL|CL interface [7,8,14,122,124] (these studies focused on CB MPLs for which large cracks/voids were identified as water retaining structures).   It is also believed that water retention at the CL is furthermore promoted by preferential adhesion of the GN-based MPLs to the CL. Consistent with previous observations with the GN foam MPL, post-testing examination revealed that both the GN and RGO MPLs were transferred and adhered to the CCM. This transfer is illustrated more clearly for the GN MPL in Figure 5.29. It is important to note that, in this case, all the MPLs were bonded to the GDL and presented similar conformabilities (unlike with the GN foam MPL which was a freestanding layer). This implies that GN and RGO particles actually detached from the underlying GDL completely to adhere to the CCM. This adherence phenomenon highlights a very advantageous aspect of graphene-based MPLs, either freestanding or GDL-based: By using graphene MPLs, one essentially creates a self-made CCM-based MPL. In other words, the same intimate CL contact, that one would expect to achieve with a CCM-based MPL, can be established while avoiding the difficulties involved with the manufacturing of CCM-based MPLs (refer to Section 2.6).  The graphene-based composites (RGO+CB and GN+CB) also displayed adhesion, but to a slightly lesser extent, due to the presence of CB. Preferential adhesion of the GN-based MPLs is also illustrated by cross-sectional SEM images of the MPL|CL interface, post-testing (refer to Figure 5.30). As described before, the reason for the preferential adherence of GN-based materials to the CCM is attributed to the strong hydrophobic interaction between the fluoro-backbones of Nafion and graphene’s surface[153,154].   Chapter 5: GDL-based MPLs and their composite effects     142   Figure 5.29: Schematic illustrating the transfer of a GN MPL from the GDL to the CCM after fuel cell testing.    Figure 5.30: Cross-sectional SEM images demonstrating the extent of adhesion of the GDL-based MPLs to the CCM after fuel cell testing.      Chapter 5: GDL-based MPLs and their composite effects     143  5.10.7 Experimental uncertainty  Repeatability measurements were performed for all performance tests. For the most part, very little variance is displayed in the MPLs’ performance results at 100% cathode RH. The GN MPL displays more extreme variance in the mass transport limiting region of the polarization curve (beyond 1400 mA cm-2), due to severe flooding. At 20% RH, the variance associated with most MPLs typically increases, as membrane dehydration gradually increases with the repetition of performance tests at these conditions. The variance for the GN-based MPLs at high and low humidity remains fairly similar though, due to the improved ability to maintain performance at low cathode humidity.    Repeatability measurements were also performed for the majority of characterization methods. The exceptions to the latter are through-plane resistance, interfacial contact resistance and water permeability measurements, for which only one measurement was performed due to material restrictions and damage/structural changes incurred through the measurement technique itself. The variance is found to typically decrease upon compression. This trend is attributed to the fact that the MPLs become more compacted, also presenting smoother and more uniform surfaces in the compressed state. With the exception of contact angle measurements, higher values in measured properties are also generally associated with a higher degree in variance. For example, the RGO MPL which presents as the thickest layer in the compressed state, also has the highest degree of variation associated with the measurement (with a total thickness of 376 ± 20 μm). Sources of variance may include equipment variance, MPL material variance, and variance introduced through human involvement during sample preparation and the execution of measurements.       Chapter 5: GDL-based MPLs and their composite effects     144  5.11 Conclusions  The GN MPL illustrates improved interfacial characteristics compared to the conventional CB MPL: low interfacial resistance (26% lower on a normalized basis), low bulk resistance (19% lower in-plane resistivity in the compressed state), smooth planar morphology (88% lower surface roughness in the compressed state), a slightly more hydrophilic nature (37˚ lower contact angle in the compressed state) and CL adhesion. The latter three properties specifically also promote water retention at the CL interface, benefitting low humidity operation. As another graphene variant, the RGO MPL also presents some beneficial interfacial properties (a planar morphology, a more hydrophilic nature and CL adhesion), but due to a high degree of functionalization (with 8% nitrogen) and surface defects, the layer compacts less densely (99 μm thicker than the GN MPL) and has higher bulk resistance (40% higher in-plane resistivity than the GN MPL).   While the GN and RGO MPLs show kinetic and ohmic polarization improvements compared to the CB MPL at 100% RH, the layers also suffer from mass transport limitations. However, the addition of CB alleviates these limitations, and extends the limiting current density by 400 mA cm-2 for the RGO MPL and 1500 mA cm-2 for the GN MPL. CB particles introduce additional pathways for liquid water removal between the tortuous structures associated with the planar morphologies of RGO and GN. The creation of additional water removal pathways is confirmed through increased water permeabilities for the composite MPLs (45% and 51% for the GN and RGO MPLs, respectively). The improved water management furthermore results in synergistic performance improvements, which are most extreme in the case of GN+CB, as reflected by an increase of approximately 30% in the maximum power density at 100% RH. The GN+CB MPL also achieves similar kinetic and ohmic performances to those of the GN MPL, at only half of the original GN load. In the case of GR, the addition of CB does not introduce any performance benefit. Rather, performance losses are exaggerated due to increased gas and water permeability, resulting in a 10% decrease in the maximum power density.  Chapter 5: GDL-based MPLs and their composite effects     145  At 20% RH, the GN-based MPLs also show excellent performance preservation, resulting in a 80% improvement in the maximum power density of the GN+CB MPL over the CB MPL. This behaviour, and the beneficial interfacial characteristics of the GN-based MPLs, justify the selection of the graphene-based MPLs for the application to low loaded CCMs in the final phase of the study.  The creation of CB-composite MPLs with larger sized GR, RGO and GN particles, results in certain general property effects. These general trends include the following (the degree of relative change for the GN+CB MPL is provided as example): decreased surface wettability expressed as an increase in contact angle (27%), decreased through-plane resistance (28%), decreased porosity (13%), increased water permeability (45%), decreased gas permeability (41%) and decreased porosity (13%), with the exception of GR in the latter two cases. Even though this study only focused on composites prepared in a 1:1 weight ratio, it provides a more comprehensive understanding of how different morphologies interact with each other, thereby laying the groundwork for the exploration of different loadings, materials and compositional ratios. The work will furthermore serve useful for future studies which focus on the purposeful manipulation of MPL properties, with specific applications and/or operating conditions in mind.  As expected, the MPLs all respond very differently to the same degree of compression. For example, reductions in gas permeability range from 10% - 70% amongst the different investigated MPLs. The most extreme response to compression is furthermore displayed by the RGO MPL (70% and 84% decrease in gas permeability and surface roughness, for example). Such high degrees of change in property values highlight the importance of characterizing materials in the compressed state.   Finally, the use of one-dimensional control in this phase of the work, enabled a more fundamental assessment of MPLs. In essence, MPL structure and its role in water retention was put under a ‘magnifying glass’ and its subsequent influence on various performance aspects, could be delineated more clearly. It remains unclear, however, whether similar performance Chapter 5: GDL-based MPLs and their composite effects     146  enhancements can be expected in a different operating mode such as stoichiometric control. Thus far, investigations have also been limited to a small active area (5 cm2), due to material restrictions. These issues will be addressed in the following chapter.   Portions of this chapter are taken from: ChemSusChem, Vol. 9, Issue 13, A.T. Najafabadi, M.J. Leeuwner (co-first author), D.P. Wilkinson and E.L. Gyenge, Electrochemically Produced Graphene for Microporous Layers in Fuel Cells, 1689-1697, Copyright (2016), with permission from John Wiley & Sons, Inc.    Chapter 6: Application of graphene-based MPLs to low loaded CCMs     147  Chapter 6: Application of graphene-based MPLs to low loaded CCMs  A company manufactured graphene, CGN, was found to result in similar performance improvements in the kinetic and ohmic polarization regions, as previously obtained with GN in Chapter 5. The material also shows improved mass transport, due to its larger flakes covering the GDL in a less uniform fashion. CGN is also the only other MPL material which displays maintained and/or improved performance at low humidity conditions and under one-dimensional control. The important role of the CCM type and loading is further highlighted by the application of CGN MPL to JM (Johnson Matthey) CCMs. Comparison of the Gore and JM High CCMs (both with a cathode catalyst loading of 0.4 mg cm-2), indicates that structural and/or compositional differences in the CL can influence the MPL|CL interaction and the extent of the performance benefit gained with the CGN MPL. Comparison of the JM High and JM Low CCMs (with cathode catalyst loadings of 0.4 and 0.1 mg cm-2, but similar CL formulations), shows that the potential to improve performance with CGN, becomes lower the less catalyst there is. CGN still results in improved performance preservation at low humidity conditions with low loaded CCMs, however.  6.1 Introduction  As previously mentioned, the Pt catalyst is one of the main contributors to the high cost of PEMFCs. A recent report from the United State of America’s Department of Energy estimates that the catalyst can account for up to 43% of the total cost of a 80 kW PEM automotive system, depending on the market price of Pt and the total volume production[168] (refer to Figure 6.1 for total cost breakdown). Such high costs has prompted a strong research focus on reducing the total amount of Pt from conventional loadings such as 0.4 mg cm-2, to 0.1 mg cm-2 or even below[9–11]. Low loaded CLs have been created successfully through various different techniques, such as ink jet printing[12], direct membrane deposition[169]xiv, plasma sputtering[170], electro-spray[11]xv and electroless deposition[13], to name just a few. The development of low loaded CLs has also been accompanied by the introduction of novel catalyst structures, such as                                                      xiv Conventional membrane is replaced by two ionomer layers deposited directly onto cathode and anode GDEs. xv Process by which electricity is employed to disperse liquid into a fine aerosol. Chapter 6: Application of graphene-based MPLs to low loaded CCMs     148  3M’s nanostructured thin films[10]. Unfortunately, low loaded CLs do not necessarily maintain similar performance and durability as conventionally loaded CLs. Except for increased kinetic losses, the poorer performance of low loaded CLs has also been attributed to increased resistance[169,171] and flooding[172]. Low loaded CLs are furthermore typically coupled with conventional CB MPLs during practical investigations. This raises the question of whether alternative MPLs could also be used to achieve performance enhancements with low loaded CCMs.   Figure 6.1: Breakdown of the  projected system stack cost at 50 000 systems per year, based on 2016 technology[168].  In the previous chapters, GN foam and GN flakes demonstrated beneficial interfacial characteristics. The materials also showed excellent applicability for low humidity applications with the potential to fine-tune material properties and performance aspects (such as mass transport behavior) through the introduction of CB particles. A graphene-based material was therefore also applied to low loaded CCMs in this phase of the study. Building on the work in previous chapters, which centered on the investigation of a small active area (5 cm2) operated under constant flow control, a larger active area (49 cm2) and stoichiometric control conditions were also investigated. Since larger sized MPLs were required, a graphene, produced at larger scale, was also incorporated.  30%8%43%6%5%8%Bipolar platesMembranesCatalyst and applicationGDLsMEA frames and  gasketsBalance of stackChapter 6: Application of graphene-based MPLs to low loaded CCMs     149  The underlying suppositions for this phase of the study are captured in the following hypotheses:  Graphene-based MPLs will result in similar performance enhancement with low loaded CCMs as obtained with high loaded CCMs;  Performance enhancements obtained with graphene-based MPLs can be reproduced at a larger scale and under different operating modes;  Graphene produced through an electrochemical method at commercial scale will result in similar improved performance (compared to CB MPLs), as the GN produced through electrochemical exfoliation at lab-scale.   Based on the hypotheses, the following objectives were specified for this phase of the study (also supporting the main objective to develop MPLs with improved interfacial characteristics and enhanced operational flexibility):  Evaluation of the application of graphene-based MPLs to low loaded CCMs;  Assessment of graphene-based MPLs’ performance at larger scale and under a different operating mode (i.e., stoichiometric control);  Evaluation of a graphene produced at larger scale and comparison to the previously employed GN flakes (Chapter 5).  6.2 MPL materials   The requirement for larger sized MPLs necessitated the use of a graphene material produced at a much larger scale compared to the previous laboratory-scale production (Chapter 5). A product manufactured by the company NanoXplore[165] through an electrochemical method[173], was therefore employed.  The company manufactured graphene will be designated as CGN, to distinguish it from the GN used in Chapter 5. The most prominent difference between the two types of graphene is their particle size, with that of the CGN being much larger (Table 6.1). According to the graphene classification system presented by Bianco et al., CGN can also be Chapter 6: Application of graphene-based MPLs to low loaded CCMs     150  ascribed to the sub-category of ultrafine graphite. To serve as a baseline comparison, conventional CB (Vulcan XC72R, Cabot Corporation) was again incorporated in the investigation.  Table 6.1: Material properties of CB and CGN in dry powder form (details of GN used in Chapter 5 are also provided for comparison purposes). Material CB CGN GN (Chapter 5) Original particle size (μm) ~ 0.05 3 - 18a  0.5 – 0.8 Dry powder volumetric density (kg m-3) ~ 60 ~ 258 ~ 420 Avg. number of monolayers N/A ~ 12b < 5 Color Black Silver-grey Silver-grey Elemental composition (%) C: 98 O: 2 C > 93b C: 97 O: 3  aEstimated from SEM images  bManufacturer’s specifications  MPLs were prepared using the spray-deposition technique described in Section 3.2. Toray 060 CFP (containing 5 wt% PTFE) was again used as GDL material. In accordance with Chapter 5, a total MPL loading of 1.5 mg cm-2 and PTFE content of 20 wt% was applied. The following MPLs were prepared: CB, CGN and CGN+CB (composite prepared in 1:1 weight ratio).  To assess the MPLs’ application to low loaded CCMs, a comparison of high and low catalyst loaded CCMs of similar type was required. Two commercial CCMs from Johnson Matthey were employed for this purpose: ‘JM High’ with a cathode catalyst loading of 0.4 mg cm-2 and ‘JM Low’ with a cathode catalyst loading of 0.1 mg cm-2. Both CCMs have an anode catalyst loading of 0.01 mg cm-2. More details on these CCMs’ properties are presented in Section 6.9. All MEAs were furthermore assembled with Sigracet 25BC as the combined anode GDL+MPL.    Chapter 6: Application of graphene-based MPLs to low loaded CCMs     151  6.3 Specifics of characterization methods  The same characterization procedures as used in Chapter 5 were applied in this phase of the study. In addition, the following CCM properties were also characterized using the methods described in Section 3.1: structure and morphology, surface roughness and wettability.  6.4 Specifics of fuel cell operation and testing  For investigation at a larger scale, tests were performed on the TP-50 cell with an active area of 49 cm2. Stoichiometric flow control was furthermore employed under the operating conditions illustrated in Table 6.2.  Table 6.2: Operating conditions of single cell performance tests performed in TP-50 cell hardware under stoichiometric control. Operating parameter Hydrogen Air Pressure  202.6 kPa(g) 202.6 kPa(g) Relative humidity 100% 100% or 20% Temperature 75˚C 75˚C Stoichiometric ratio 1.5 2 and 4 Cell compression 120 psi(g) or 827 kPa(g)  Coolant flow rate 0.5 L min-1  Polarization tests were performed in the same fashion as described in Section 3.5.4. However, since the stoichiometric amounts required at a current density of 0 mA cm-2 are zero, the following reactant flow rates were employed at OCV conditions:  0.2 L min-1 for hydrogen and 1.5 L min-1 for air. Data was collected for 120 s at each current density. Similar to Chapter 5, longer-term tests were also performed at a constant current density of 1000 mA cm2 for 5 – 6 hours in the TP-50 cell. However, for a more detailed assessment, the tests were performed at both 100% and 20% cathode RH. To induce slightly ‘harsher’ conditions, these tests were executed at air:H2 stoichiometric ratios of 4:1.5 (even larger stoichiometric amounts could not be used due to the large pressure drops associated with the TP-50 cell).  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     152  6.5 MPL characterization results  The characterization results of the CB, CGN and CGN+CB MPLs are presented in Table 6.3, Table 6.4 and Table 6.5, with the relative change between their uncompressed and compressed states presented in square brackets. Characterization results are again visually presented in the form of radar charts, normalized based on the entire dataset (encompassing both the uncompressed and compressed state results). Additional experimental data are furthermore provided in the appendices, relating to surface roughness (Appendix D.1) and through-plane resistance (Appendix D.2).   Table 6.3: Surface properties and thickness of the CB, CGN and CGN+CB MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). MPL Contact angle (°) Average surface roughness (μm) Total thickness, GDL+MPL (μm)   0 kPa(g) 827 kPa(g) 0 kPa(g) 827 kPa(g) 0 kPa(g) 827 kPa(g) CB 146 ± 4 149 ± 2 [2% ↑] 18.5 ± 2.1 12.7 ± 2.0 [31% ↓] 285 ± 9  248 ± 3 [13% ↓]  CGN 144 ± 2 132 ± 1 [8% ↓] 14.5 ± 1.3 5.6 ± 1.4 [61% ↓] 235 ± 8 219 ± 3 [7% ↓] CGN+CB 146 ± 3 142 ± 3 [3% ↓] 13.3 ± 1.7 5.9 ± 1.0 [56% ↓] 259 ± 10 229 ± 5 [12% ↓]         Chapter 6: Application of graphene-based MPLs to low loaded CCMs     153  Table 6.4: Electrical properties of the CB, CGN and CGN+CB MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). MPL In-plane resistivity (mΩ cm) Through-plane resistance (mΩ cm2) 0 kPa(g) 827 kPa(g) 34 kPa(g)a 827 kPa(g) CB  9.4 ± 0.5  8.1 ± 0.4 [14% ↓] 92.8 4.9 [95% ↓] CGN 7.2 ± 0.6 6.5 ± 0.3 [10% ↓] 56.3 5.6 [90% ↓] CGN+CB 7.6 ± 0.5 7.2 ± 0.4 [5% ↓] 73.6 4.0 [95% ↓] aThe through-plane resistance data collected at very low compression is grouped with the ‘uncompressed’ data.   Table 6.5: Transport and structural properties of the CB, CGN and CGN+CB MPLs at indicated sample compressions (relative change between uncompressed and compressed states indicated in square brackets). MPL Gas permeability (darcy) Water permeability (darcy) Apparent MPL porosity (%) 0 kPa(g) 827 kPa(g) 827 kPa(g) 827 kPa(g) CB 3.65 ± 1.30x10-1 2.51 ± 4.85x10-2 [31% ↓]  7.30x10-1 ± 9.65x10-3  72 ± 4 CGN 9.46x10-1 ± 3.23x10-2 1.09x10-1 ± 1.50x10-3 [88% ↓] 1.70x10-1 ± 2.33x10-3 61 ± 7 CGN+CB 1.05 ± 4.05x10-2 6.44x10-1 ± 1.41x10-2 [38% ↓] 3.12x10-1 ± 6.82x10-3 54 ± 8  Surface images of the CGN-based MPLs (Figure 6.2) indicate that the MPL materials do not completely cover the GDLs, with some pores of the underlying substrate (originally 20 - 50 μm[175] in size)  remaining partly exposed. This is different from the GN MPL presented in Chapter 5 which demonstrated fairly uniform coverage (Figure 5.4). The main reason for Chapter 6: Application of graphene-based MPLs to low loaded CCMs     154  differences in the uniformity/coverage of the two graphene layers is the particles size: for the GN MPL from Chapter 5, the significantly smaller particles (0.5 – 0.8 μm) are able to securely lodge in the pores of the underlying GDL, while the larger CGN flakes (3 - 18 μm) cover the GDL fibers in a slightly more haphazard fashion. The addition of CB furthermore does not seem to induce significant differences in the extent of surface coverage, as also evidenced by surface roughness results (refer to Figure 6.3). For example, in the compressed state, the surface roughnesses of the CGN and CGN+CB MPLs are 5.6 ± 1.4 and 5.9 ± 1.0 μm, respectively. Both layers furthermore indicate similar degrees of smoothening upon compression (61 and 56% for the CGN and CGN+CB MPL).   Figure 6.2: Surface images of the CGN and CGN+CB MPLs in (a) uncompressed and (b) compressed states; Higher magnification images are provided in the inserts.  Similar to the other studied graphene-based materials, the CGN MPL presents a less hydrophobic surface than the conventional CB MPL, with a contact angle of 132˚ ± 1˚ in the compressed state. However, the addition of CB again decreases the wettability (as shown in Figure 6.3), resulting in an 8% increase in contact angle in the compressed state. The CGN-Chapter 6: Application of graphene-based MPLs to low loaded CCMs     155  based MPLs furthermore also demonstrate Wenzel-dominated wetting behavior, with the surfaces’ wettability increasing upon smoothening.    Figure 6.3 Radar charts of material properties for the CB, CGN and CGN+CB MPLs in (a) uncompressed and (b) compressed state; (c) Porosity and (d) water permeability results for the CB, CGN and CGN+CB MPLs.  Figure 6.4 shows that the CGN-based MPLs also primarily consist of layers of stacked flakes. In isolation, the CGN flakes stack fairly densely, resulting in a layer with a compressibility of 7% and an overall thickness of 219 ± 3 μm in the compressed state. The presence of CB results in a slightly thicker layer (229 ± 5  μm in compressed state) with a higher compressibility of 12%. As before, the trends in overall thickness are also reflected in the layers’ in-plane resistivities, displaying a linear correlation coefficient of 0.99 for the compressed state values (also refer to Figure 6.5). The compressed state in-plane resistivity of the CGN MPL is also comparable to that Chapter 6: Application of graphene-based MPLs to low loaded CCMs     156  of the GN MPL from Chapter 5 (6.5 ± 0.3 vs. 6.6 ± 0.2 mΩ cm). The previously observed trend in through-plane resistance also remains consistent, as the presence of CB results in a 29% decrease, from the CGN MPL’s original through-plane resistance of 5.6 mΩ cm2, to 4 mΩ cm2 for the composite MPL.  The higher degree of tortuosity which is introduced by large planar structures (such as RGO and GN), is also apparent in the case of the CGN MPL:  the gas and water permeability of the CGN MPL is 2.4 and 0.56 darcy lower than that of the conventional CB MPL. Due to the fact that some GDL pores remain partly exposed in CGN MPL, the layer’s permeabilities are not as low as the GN MPL in Chapter 5, however. For example, the CGN MPL’s gas permeability of 0.19 ± 1.5 x 10-3 darcy is 30 times higher than that of the GN MPL (referencing compressed state values).  The layer also has a higher average apparent porosity of 61%, indicating less dense stacking and/or larger pore sizes than in the case of the GN MPL (average apparent porosity of 37%). Similar to the GR+CB case study, the addition of CB to CGN results in an increase in both gas and water permeability, by 0.54 and 0.14 darcy, respectively (also illustrated in Figure 6.3 (b) and (d)).             Figure 6.4: Cross-sectional images (compressed state) of the CGN and CGN+CB MPLs; Higher magnification images are provided in the inserts.  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     157   Figure 6.5: Association between in-plane resistivity and thickness for the CB, CGN and CGN+CB MPLs.  An overview of the CGN+CB case study furthermore reveals that the MPLs demonstrate the same general trends regarding the addition of CB and dominant structures, as observed for the 1:1 composite MPLs in Chapter 5. The exception is the trend in gas permeability: CB does not lead to a decrease in gas permeability (like with GN+CB and RGO+CB) but rather an increase, similar to the GR+CB case study.  Please note: In the following sections, reference is made to tests performed in both the TP-5 and TP-50 cells. To help prevent confusion, the main differences between these two cells and their respective operating modes are provided in the table below.  Table 6.6: Main differences between TP-5 and TP-50 operation. Cell hardware Active area (cm2) Employed operating mode Practical implications TP-5 5 One-dimensional (constant) flow control Comparatively higher cathode flow rate  below 1200 mA cm-2 TP-50 49 Stoichiometric (variable) flow control Comparatively higher cathode flow rates  above 1200 mA cm-2 for λair =2    Chapter 6: Application of graphene-based MPLs to low loaded CCMs     158  6.6 Evaluation of CGN (in TP-5 under one-dimensional control)  To aid in the interpretation of the CGN MPL’s results, the material’s performance was first assessed in the TP-5 cell (5 cm2 active area) under one-dimensional flow control. A comparison was also made with the GN MPL’s performance (from Chapter 5). For consistency purposes, similar MEA components were used (Gore Primea Series 5510 CCM with 0.4 mg Pt cm-2 on both sides and Sigracet 25BC as the combined anode GDL+MPL).  The most noticeable difference in the performance of CGN MPL, compared to the GN MPL, is the significant improvement in mass transport (Figure 6.6 (a)). The improvement is attributed to the CGN MPL’s much higher apparent porosity, gas permeability and water permeability, as evidenced in Section 6.5. The MPL therefore presents as a less dense and tortuous structure, which helps to lower the degree of flooding and to achieve a much higher limiting current density than the GN MPL (3200 vs. 1700 mA cm-2). For the most part, the kinetic and ohmic performance of the two types of graphene appear fairly similar, with that of the CGN being only slightly lower than that  of the GN MPL from Chapter 5. For example, at 760 mA cm-2, the CGN and GN MPLs achieve 0.721 V and 0.728 V respectively. Compared to the conventional CB MPL, the CGN MPL shows improved performance throughout the entire polarization curve.  Figure 6.6 (b) and (d) illustrates how lowering the cathode humidity effects the CGN MPL’s performance. Compared to other GDL-based MPLs investigated under one-dimensional control, the CGN MPL’s response is very unique, showing regions of both performance deterioration and improvement. Below 1600 mA cm-2, the water in the electrode is removed effectively. At 20% RH it is furthermore removed at a fast enough rate that the membrane becomes slightly dehydrated, resulting in performance loss. At current densities of 2000 mA cm-2 and higher, however, where larger amounts of water are produced, some water is retained by the MPL. At fully humidified conditions, this is likely still associated with some degree of flooding. At lower cathode humidity, however, performance is enhanced due to the overall reduction of water and flooded conditions (e.g. at 3000 mA cm-2 performance is improved by 32%).  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     159   Figure 6.6: Performance results of CGN in the TP-5 cell and under one-dimensional flow control: (a) Polarization performance comparison at 100% cathode RH, effect of cathode humidity on (b) polarization performance, (c) ohmic loss and (d) iR-corrected performance; (e) Voltage and (f) resistance traces during longer-term testing at 1000 mA cm-2 and 20% RH. Chapter 6: Application of graphene-based MPLs to low loaded CCMs     160  The described transitional behavior suggests that there are two phases of water removal. For the first, associated with lower current densities (polarization deterioration region), water is likely primarily removed through the exposed GDL pores. As the current density increases and more water is produced, these primary pathways gradually become more flooded. Larger amounts of water are consequently transported (and also accumulate) in the more tortuous pathways between the CGN flakes, leading to the overall increase in water retention (corresponding to the polarization improvement region). This behaviour is also reflected in the ohmic loss in Figure 6.6 (c). At 20% cathode RH, the ohmic loss below 1600 mA cm-2 is 10 – 20 mV higher, due to increased membrane resistance from dehydration. However, at 1800 mA cm-2, the ohmic loss becomes similar to that at 100% RH, as the membrane again becomes sufficiently hydrated.   Longer-term tests (Figure 6.6 (e) and (f)) were furthermore performed at different current densities within the two identified polarization regions. At 1000 mA cm-2 (which falls within the region of performance deterioration at 20% RH) the overall performance loss is approximately 16 mV h-1. At 2000 mA cm-2 (which corresponds to the region of polarization improvement at 20% RH) the performance shows a smaller initial decline, after which it stabilizes, and even improves slightly at one point (at approximately 4.5 h). The overall performance loss at these conditions is 6 mV h-1, while the resistance is also maintained exceptionally well and displays a very small increase of 0.5 mΩ cm2 h-1.   Other than the GN-based MPLs (from Chapter 5), the CGN MPL is the only MPL which shows maintained and/or improved performance at low humidity conditions and one-dimensional control (albeit not for the entire polarization curve). Based on this factor, and comparable improvements in the kinetic and ohmic regions to that of the GN MPL, CGN proves to be a promising graphene material. Under one-dimensional control, the material furthermore shows excellent applicability for both high and low humidity conditions (given that it is operated within the ‘polarization improvement’ region at lower humidity).  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     161  6.7 Evaluation of CCM loading (in TP-50 under stoichiometric control)  The polarization results for JM High and JM Low CCMs obtained at λair = 2, are shown in Figure 6.7 (a) and (b). Results obtained at 100% cathode RH are represented by solid lines, while the dotted lines indicate results collected at 20% cathode RH. Consistent with reports from literature[169,171,172], the overall performance is found to decrease with the employment of a lower CL loading. For example, for the CB MPL at 100% RH, the performance with the JM Low CCM is 35 mV lower in the kinetic region (at 200 mA cm-2), 78 mV lower in the ohmic region (at i = 1000 mA cm-2) and 160 mV lower in the mass transport region (at 1600 mA cm-2) than with the JM High CCM. Generally, the ohmic losses associated with the JM Low CCM are also higher. As illustration, the ohmic loss for the CB MPL at 1000 mA cm-2 is 85 mV and 93 mV with the JM High and JM Low CCMs, respectively.  Compared to the CB MPL, the CGN-based MPLs typically introduce performance enhancements up to a current density of 500 mA cm-2 (illustrated in the inserts of Figure 6.7 (a) and (b) for results obtained at 100% RH).  With the JM High CCM, these performance enhancements are in the range of 10 – 26 mV, while they appear less significant for the JM Low CCM, in the range of 1 – 8 mV. The marked decrease in performance enhancement for the JM Low CCM suggests that the CL loading itself may be acting as a limiting factor on potential improvements.   The United States of America’s Department of Energy also specifies a technical target for transportation applications within this region of the polarization curve: at 0.8 V a fuel cell should achieve 300 mA cm-2 (target for the year 2020)[176]. In Figure 6.7 (a) the JM High CCM results show close agreement to this target, with the CGN+CB MPL meeting it exactly. In the case of the JM Low CCM, (for which the total MEA Pt loading target of 0.123 mg cm2 is also met) the achieved current density at 0.8 V is approximately 150 mA cm-2 for all the MPLs (Figure 6.7 (b)). It should be noted though, that the targets are based on testing with higher concentrations of oxygen (pure oxygen fed to the cathode at λoxygen = 1.8) and the performance of the investigated MEAs are expected to increase further at similar conditions. Chapter 6: Application of graphene-based MPLs to low loaded CCMs     162   Figure 6.7: Performance results with JM High and JM Low CCMs in TP-50 cell and λair = 2xvi: (a, b) Polarization performance and (c, d) ohmic loss.                                                       xvi As can be seen in Figure 6.7, the final voltage point for TP-50 tests is typically around 0.4 V (or even slightly above). This is much higher than the voltage associated with the limiting current density in TP-5 tests (typically between 0.1 and 0.2 V). The reason for this discrepancy is because the TP-50 tests are limited by a higher stack voltage (the voltage drop across the complete fuel cell, including FFPs and current collectors), due to much thicker and larger sized hardware. Consequently, polarization tests can typically not proceed to very high current densities, before the stack voltage approaches zero and the load can no longer be increased. Depending on the employed MEA, the mass transport region might therefore also not be probed in its entirety.   Chapter 6: Application of graphene-based MPLs to low loaded CCMs     163  To investigate the kinetic trends in more detail, the ORR kinetic parameters were also determined by means of a Tafel analysis (Equation 2.11) based on the polarization curves up to a current density of 200 mA cm-2. It was again assumed that the anodic surface potential is negligible and polarization points were iR-corrected. The resultant kinetic parameters are provided in Table 6.7, while the Tafel plots are presented in Appendix D.3. The current density at a voltage of 0.9 V (for the iR-corrected curve), is again provided as another kinetic metric.  Table 6.7: Kinetic parameters associated with the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs at λair = 2 and 100% cathode RH. MPL JM High CCM JM Low CCM Apparent Tafel slope (mV dec-1) Apparent exchange current density (mA cm-2) iE = 0.9 V,  iR-corrected (mA cm-2) Apparent Tafel slope (mV dec-1) Apparent exchange current density (mA cm-2) iE = 0.9 V,  iR-corrected (mA cm-2) CB 64 1.1 x 10-4 23 73 2.0 x 10-4 14 CGN 71 7.8 x 10-4 32 79 7.7 x 10-4 15 CGN+CB 69 5.0 x 10-4 32 77 5.0 x 10-4 14  Compared to the CB MPL, the CGN-based MPLs show increased Tafel slopes and exchange current densities, with the latter likely accounting for observed improvements in kinetic performance. The trends in apparent Tafel slope and exchange current density again suggests an association with MPL structural effects and water permeability (which displays the following trend: CGN < CGN+CB < CB). It is believed that MPL structures can influence electrode kinetics by modifying the chemical environment in the catalyst layer (e.g., the oxygen and water activities in the vicinity of the CL). However, it is also possible that associations between MPL structure and aspects such as mass transport and ionic conductivity, can influence these parameters (as previously discussed in Chapter 4 and Chapter 5). It has furthermore been confirmed that the catalytic activity of GN is insignificant compared to that of Pt[162] and cannot account for any improvements observed in the kinetic polarization region (in the literature, the Chapter 6: Application of graphene-based MPLs to low loaded CCMs     164  catalytic activity of GN is also largely associated with doped variants and application in alkaline media[110]).  The JM Low CCM furthermore shows a general increase in Tafel slope, while the exchange current densities remain fairly similar to that obtained with the JM High CCM. The current density at 0.9 V is also much lower than with the JM High CCM. This indicates less favorable kinetics for the JM Low CCM, which is typically expected due to its lower Pt loading. Consistent with the GN MPL from Chapter 5, the CGN MPL also suffers from poorer mass transport performance, while the addition of CB in the composite helps to alleviate the problem. The improvement is again attributed to accelerated water drainage from the CL, as also demonstrated by the CGN+CB MPL’s higher water permeability. With respect to the JM Low CCM, the CGN+CB MPL’s mass transport performance is also slightly better than that of the conventional CB MPL (Figure 6.7 (b)).   In general, the MPLs also show very little or no differences in their respective ohmic losses. For the JM High CCM, the maximum difference between the MPLs’ ohmic losses is only 4 mV (Figure 6.7 (c)). Differences in ohmic loss become slightly more distinct with the JM Low CCM (refer to insert of 100% RH results in Figure 6.7 (d)) and display the following trend: CGN < CGN+CB < CB. The OCV resistances also follow this general trend (Table 6.8) and show a very strong association with the MPLs’ in-plane resistivities and thickness, with linear correlation coefficients of 0.99 and 1.00 (Figure 6.8). Again, the measured OCV resistances are also higher and indicate more distinct differences when the MPLs are combined with the JM Low CCM. The relative magnitude of these resistances will be addressed again in Section 6.9. Due to the similarities in the ohmic loss, trends do not change when polarization curves are plotted in the iR-corrected format. The original and iR-corrected form of Figure 6.7 (a) is provided in Appendix D.4 as example.    Chapter 6: Application of graphene-based MPLs to low loaded CCMs     165  Table 6.8: OCV resistances of the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs. MPL Resistance at OCV (mΩ cm2) JM High CCM JM Low CCM CB 129 146 CGN 126 140 CGN+CB 127 142   Figure 6.8: (a) Association between OCV resistance and in-plane resistivity (compressed state) for the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs; (b) Association between OCV resistance and thickness (compressed state) for the CB, CGN and CGN+CB MPLs in combination with the JM High and JM Low CCMs.  Comparison of the high and low humidity results indicates that all the MEAs’ performances, as well as their ohmic losses (displayed in Figure 6.7 (c) and (d)), remain fairly similar. This general resemblance in performance at different humidities is consistent with results obtained by Blanco[177] who performed comparable tests, using similar stoichiometric ratios, operating conditions and testing equipment. Slight performance improvements are, however, observed in the mass transport regions, and are most prominent for the CGN-based MPLs with the JM High CCM (Figure 6.7 (a)). This implies that the layers are likely flooding at high current densities under fully humidified conditions and an air stoichiometric ratio of 2. At low humidity, the overall amount of water in the system is reduced, thereby alleviating flooding and improving Chapter 6: Application of graphene-based MPLs to low loaded CCMs     166  mass transport performance. The effect of low humidity conditions are investigated in more detail through longer-terms tests performed at harsher conditions with λair = 4 (refer to Section 6.8).   It is noteworthy to mention that the performance enhancement obtained with CGN-based MPLs at 100% RH, are far less significant in the case where JM High CCMs were used, compared to  when Gore CCM was used (although both have a catalyst cathode loading of 0.4 mg cm-2). It is thought that the difference can be attributed to the different flow rates used in the respective tests (the JM High CCM was tested under stoichiometric control in the TP-50, while the Gore CCM was tested under one-dimensional control in the TP-5. To confirm this theory, a brief evaluation of the JM CCMs was also performed using a higher air stoichiometric ratio of 4 in the TP-50 (4 was the upper limit on the air stoichiometric ratio due to the larger pressure drop associated with the TP-50 FFPs).   The results collected at λair = 4, are displayed in Figure 6.9. In contrast to expectations, the performance enhancements obtained in the kinetic and ohmic region with the CGN-based MPLs, remain within a similar range as with λair = 2: a maximum improvement of 17 mV and 5 mV is associated with the CGN-based MPLs when combined with JM High and Low CCMs, respectively (Figure 6.9 (a) and (b)). The benefit of employing the alternative MPLs with the low loaded CCMs therefore still appears very small. The fact that a higher air stoichiometric ratio of 4 did not result in more significant kinetic and ohmic performance improvements, as suspected, furthermore suggests that another factor might be influencing the CGN MPL’s extent of performance enhancement. This suspicion prompted a comparative investigation into the different types of CCM, as presented in the Section 6.9.  The ohmic loss associated with the different MPLs, also remain fairly similar to the tests performed at λair = 2 (Figure 6.9 (c) and (d)). However, the increased air flow rates have a fairly significant effect on the mass transport region performance. One example is the CGN MPL, which, in combination with the JM High CCM, shows an increase of 193 mV at 1200 mA cm-2.  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     167   Figure 6.9: Performance results with JM High and JM Low CCMs in TP-50 cell, λair = 4 and 100% cathode RH: (a, b) Polarization performance and (c, d) ohmic loss.  6.8 Evaluation of longer-term performance (in TP-50 under stoichiometric control)  Longer-term tests at λair = 4 were performed to get a better sense of the different MEAs’ performance maintenance and durability at 20% cathode RH. For the JM High CCM at 100% cathode RH, the CB MPL shows a very stable performance for the duration of the test, with an Chapter 6: Application of graphene-based MPLs to low loaded CCMs     168  average voltage loss of 2.4 mV h-1 (Figure 6.10 (a)). The CGN-based MPLs on the other hand, display a gradual decrease, likely due to the fact that the layers became more saturated over time, with a decreased ability to remove sufficient amounts of water from the CL.   At low humidity conditions with the JM High CCM, the opposite trend is observed, however (Figure 6.10 (b)). The CB MPL’s performance decreases (by 8.8 mV h-1) and its resistance increases (by 0.5 mΩ cm2 h-1) as the membrane starts to dehydrate more over time. The CGN-based MPLs show better performance maintenance, with the CGN MPL even demonstrating performance improvement after approximately 2.5 hours (likely due to the establishment of new steady state conditions in the absence of severe flooding). The layer’s resultant overall improvement is 0.6 mV h-1. Interestingly, the composite MPL closely mimics the behavior of the CB MPL at fully humidified conditions, while it more closely resembles that of the CGN MPL at low humidity conditions.    In general, the trends obtained with JM High CCMs are mirrored by the JM Low CCMs (Figure 6.10 (c) and (d)). As result, the CGN-based MPLs continue to lend excellent performance preservation at low humidity conditions, despite less significant performance enhancements obtained during polarization with the JM Low CCMs. This is illustrated by the CGN MPL’s small performance loss of 2 mV h-1 compared to 9.6 mV h-1 for the CB MPL at 20% RH Figure 6.10 (d). The improved performance preservation of the CGN-based MPLs is again attributed to their planar flakes that decrease the layers’ water permeability, compared to CB. Ultimately, these flakes, that also show adhesion to the CL (as shown in Figure 6.11), act as water retaining structures, which help to keep the membrane sufficiently hydrated.   Chapter 6: Application of graphene-based MPLs to low loaded CCMs     169   Figure 6.10: Longer-terms test at 1000 mA cm-2 and 100% and 20% cathode RH for (a, b)  JM High and (c,d) JM Low CCM.   Figure 6.11: Sample cross-sectional images demonstrating adhesion of CGN MPL to JM High and JM Low CCMs.  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     170  6.9 Evaluation of CCM type (in TP-5 under one-dimensional control)  Due to material restrictions on the Gore CCM, the evaluation of the different types of CCM was again performed on the TP-5 cell under one-dimensional control. Tests incorporated each of the respective CCMs (Gore CCM, JM High CCM and JM Low CCM) and the CGN and conventional CB MPLs. The results are presented in Figure 6.12.  A comparison of 100% RH results in Figure 6.12 (a), (c) and (e), indicates that the CGN MPL is associated with the most significant performance difference in combination with the Gore CCM. As an example, the performance of the CGN MPL is 45 mV higher than that of the CB MPL at 500 mA cm-2. In contrast, we find that the performance enhancements for the JM High and Low CCM are fairly comparable to what was previously obtained in the TP-50 cell under stoichiometric control. Improvements of 15 mV and 9 mV are observed with the JM High and Low CCMs at 500 mA cm-2. This confirms that the type of CL (or CCM) itself plays a significant role in the potential performance improvements that can be obtained through the use of alternative MPLs. While the effect of changing from stoichiometric to one-dimensional control, was not clearly visible below 500 mA cm-2 for the JM CCMs, it does more drastically affect behavior above that value: in one-dimensional control, performance enhancements extend to 2200 and 1600 mA cm-2 for the JM High and Low CCMs (albeit still small). This is expected, since the severity of flooding is minimized under one-dimensional control. The polarization curves are also extended due to delayed mass transport, as well as the TP-5’s lower stack voltage (refer to footnote of Figure 6.7). With regards to ohmic losses, it is also interesting to note that more distinct differences between the CB and CGN MPLs, are observed with the Gore and JM Low CCMs (compare Figure 6.12 (b), (d) and (f)).  The discrepancies in performance enhancements may be attributed to the extremely different nature of the CCMs. As evidence, the physical properties of the CCMs are provided in Table 6.9.    Chapter 6: Application of graphene-based MPLs to low loaded CCMs     171   Figure 6.12: Comparative CB and CGN MPL polarization performance and ohmic loss for (a, b) Gore CCM, (c, d) JM High CCM and (e, f) JM Low CCM in TP-5 cell under one-dimensional flow control and at 100% cathode RH. Chapter 6: Application of graphene-based MPLs to low loaded CCMs     172  Table 6.9: Physical properties of the employed commercial CCMs. Property Gore CCM JM High CCM JM Low CCM Cathode Pt loading (mg cm-2) 0.4 0.4 0.1 Anode Pt loading (mg cm-2) 0.4 0.01 0.01 Membrane thickness (μm) ~ 25[177]  ~ 17a ~ 17a Cathode CL thickness (μm) ~ 12.5[177] ~ 8a ~ 4a Cathode contact angle (°) 146 ± 3 116 ± 4 99 ± 3 Cathode surface roughness 0.94 ± 0.06 0.56 ± 0.01 0.53 ± 0.01 Cross-sectional image    aEstimated from SEM images  The cross-sectional SEM images, all presented at a similar scale, clearly show significant variance in the overall CCM size and structure. Most noticeable, is the fact that the Gore CCM, which has a similar cathode catalyst loading than the JM High CCM, presents as a thicker layer with higher surface roughness and decreased wettability (higher contact angle). Naturally, the differences between Gore and JM CCMs may also extend to chemical composition (type of catalyst, type of ionomer, type of membrane reinforcement, type of catalyst support, total support content, total ionomer content, type of anode CL, etc.) and manufacturing process, which largely remains unknown and/or proprietary information for these commercial materials. Amongst the myriad of factors, it is difficult to confirm which play a role in the larger difference observed between the CB and CGN MPLs with the Gore CCM under one-dimensional control. It is suspected though, that the more hydrophobic surface of the Gore CCM, when combined with the porous and highly water permeable CB MPL, will promote faster water removal from the CL, which in turn may accelerate membrane dehydration. This theory is somewhat supported by considering the relative performance change undergone by CB MPLs from 100% to 20% cathode RH (Figure 6.12 (a) and (c)): in combination with the Gore CCM, the CB MPL shows a far more significant performance loss (210 mV at 1600 mA cm-2), than with the JM High CCM (81 mV at 1600 mA cm-2). With the decrease in humidity, the increase in ohmic loss is also more 10 μm Chapter 6: Application of graphene-based MPLs to low loaded CCMs     173  prominent for the Gore CCM than the JM High CCM, indicating more significant membrane dehydration (Figure 6.12 (b) and (d)).  Although JM High and JM Low CCMs have similar membrane structures, the latter has a thinner CL which is also associated with the smoothest and most wettable surface. The chemical formulation for the JM High and JM Low CCMs are understood to be similar though. Throughout Sections 6.7, it was found that the JM Low CCM presents higher ohmic losses, compared to the JM High CCM. Differences between the ohmic losses of CB and CGN MPLs were also typically more distinct with the JM Low CCMs. These trends are also reiterated by the ohmic loss measured in the TP-5 cell under one-dimensional control (Figure 6.12 (d) and (f)). OCV resistance (reprinted in Table 6.10 below) and interfacial contact resistance measurements (normalized according to Equation 5.1 and provided in Figure 6.13) also indicate that the JM Low CCM is associated with the highest resistances.   Table 6.10: OCV resistances of the CB and CGN MPLs in combination with the different types of commercial CCMs. MPL Resistance at OCV (mΩ cm2) Gore CCM JM High CCM JM Low CCM CB 141 129 146 CGN 125 126 140  The mentioned trends seem somewhat counterintuitive, as one would expect smaller resistances associated with a thinner CL (the JM Low CCM has the thinnest CL at 4 μm). However, increased resistance is commonly reported for low loaded CCMs in literature. Weber and Kusoglu[171] reviewed this phenomena and detailed how an observed ‘undefined’ resistance increases with substantial decreases in catalyst loadings. Although very difficult to measure and compare, this ‘undefined’ resistance has been related to the ionomer-film resistance. The latter is thought to increase at low catalyst loadings through exacerbated oxygen transport resistance and local heat and water production. The author theorizes that this ‘undefined’ resistance may possibly also act as a dominant factor for low loaded CCMs, thereby limiting the performance Chapter 6: Application of graphene-based MPLs to low loaded CCMs     174  enhancement that can be obtained with alternative materials, such as the CGN-based MPLs. It is very likely that low loaded CCMs may be more susceptible to manufacturing inconsistencies and defects, which can also contribute to their higher overall resistance. With regards to interfacial aspects, the interfacial resistances (Figure 6.13) do confirm, however, that the CGN MPL helps to improve electrical connectivity for all the CCMs used. Although, as previously discussed, this may not necessarily contribute to significantly smaller ohmic losses during polarization (refer to 5.10.5).   Figure 6.13: Interfacial resistances between the CB, CGN and CGN+CB MPLs and the different types of commercial CCM.  Regarding changes at lower cathode humidity, the JM Low CCM also undergoes significant performance loss in combination with the CB MPL under one-dimensional control (Figure 6.12 (e)). With fewer TPB points, the MEA is likely affected more severely upon membrane dehydration, compared to a higher catalyst loading. Under the harsher testing conditions of one-dimensional control, the JM Low CCM also shows a performance loss at lower humidity with the CGN MPL. The performance loss is less severe than for the CB MPL, however, indicating that the MPL can help to improve durability. By increasing the material loading of the CGN MPL, to result in more uniform GDL coverage, the performance maintenance at low humidity might possibly be improved even further.  Chapter 6: Application of graphene-based MPLs to low loaded CCMs     175  6.10 Conclusions  The employed CGN proved to be a promising alternative MPL material and a viable option for larger scale application (with an active area of 49 cm2,xvii). As an electrochemically produced graphene variant with a larger flake size (3 – 18 μm compared to 0.5 – 0.8 μm for the GN flakes of Chapter 5), it presents unique properties when prepared as a MPL. The larger flake size creates a MPL with less uniform stacking and greater exposure of GDL pores (also evidenced in its compressed state gas permeability being 30 times higher than that of the GN MPL). These factors result in significantly improved mass transport behavior compared to the GN used in Chapter 5 (with ilim increasing from 1700 mA cm-2 to 3200 mA cm-2). Even though the MPL is more permeable, the material’s planar flakes do, however, also introduce structures that promote water retention, thereby helping to maintain (and in some cases even enhance) performance at lower humidity conditions. For example, comparative humidity testing shows a performance improvement of 32% at 3000 mA cm-2 for low humidity conditions. Except for the GN MPL, the CGN MPL is also the only other MPL material which displays maintained and/or improved performance during polarization at 20% cathode RH and under one-dimensional control (in the TP-5 cell hardware).   The CGN-based MPLs again provide performance enhancements with high loaded CCMs containing 0.4 mg cm-2 Pt (in the range of 10 – 26 mV). However, smaller improvements (in the range of 1 – 8 mV) are achieved with the low loaded CCM containing 0.1 mg cm-2 Pt. This finding is in contradiction to what was hypothesized. The result suggests that the interfacial and performance benefits of graphene are restricted by the catalyst loading itself, or in other words, that the potential to improve performance becomes lower the less catalyst there is. This phenomenon is attributed to the fact that low loaded CCMs may suffer from manufacturing defects and additional resistive losses. Higher resistive losses for low loaded CCMs, compared                                                      xvii This point does not speak towards financial viability, but rather quantities available for implementation at larger scale. Although a formal cost analysis is not within the scope of this study, reference will be made to important cost considerations in Chapter 7. Chapter 6: Application of graphene-based MPLs to low loaded CCMs     176  to high loaded CCMs, were confirmed through experimental measurements. As example, with the CGN MPL, the OCV resistance is 14 mΩ cm2 higher, interfacial resistance is 85% higher (on a normalized basis), and ohmic loss at 1000 mA cm-2 is approximately 7 mV higher, when using a low loaded CCM. However, despite the fact that the CGN-based MPLs lead to less significant performance enhancements with the low loaded CCMs, they still allow the MEA to maintain its performance better at low humidity conditions, compared to a CB MPL. This is illustrated through performance losses of 2 mV h-1 and 9.6 mV h-1 for the CGN and CB MPLs, respectively, during longer-term tests at 20% cathode RH. The work also highlighted that, in addition to the cathode catalyst loading, a CCM’s structure and/or composition can also influence the extent of performance enhancement observed with graphene-based MPLs. When optimizing MEA components for a specific application, it is therefore important to remain aware of potential differences in the MPL|CL interactions.  Considering the application of alternative MPLs at different size scales, the CGN MPL shows comparable performance enhancements of 10 – 20 mV up to 500 mA cm-2 (when combined with JM CCMs), for both small (TP-5) and larger scale (TP-50) application at 100% RH. In this study, the application at different size scales is furthermore associated with the employment of different fuel cell operating modes. As also illustrated in this chapter, the major difference between one-dimensional (TP-5) and stoichiometric control (TP-50), is the minimization/delay of mass transport effects with the former. The delay of mass transport effects manifests through performance enhancements extending to 2200 and 1600 mA cm-2 for the JM High and Low CCMs in the TP-5 (and not only to 500 mA cm-2 as with the larger sized cell).  Minimized mass transport losses, in addition to the lower stack resistance of the smaller cell hardware, also allow higher maximum current densities to be reached in the TP-5 (up to 1000 mA cm-2 higher). One-dimensional control furthermore exaggerates performance differences which are more easily related to differences in MPL properties, (particularly through comparative humidity testing). However, one-dimensional control is typically also associated with harsher conditions. While this can potentially lead to earlier onset of membrane dehydration for certain Chapter 6: Application of graphene-based MPLs to low loaded CCMs     177  MPLs, it does also improve insight into durability aspects. Ultimately, both cell sizes and operating modes proved useful within this research context.  Chapter 7: Conclusions and future considerations     178  Chapter 7: Conclusions and future considerations  7.1 Significance and contributions   PEMFCs have been identified as a key part of the solution to the global energy crisis. However, to make effective use of this energy conversion source, it is important to ensure high performance and low cost of the technology. Through improved insight into alternative materials (such as graphene-based MPLs) and their interfacial interactions with the CL, MPLs may be tailored to improve overall performance and efficiency. Improving operational flexibility (through enhanced performance and durability at low humidity conditions) can furthermore support cost reductions which are considered critical for the commercialization of the technology. To the best of the author’s knowledge, the following novel materials, methods, approaches and insights are presented in this study (refer to schematic representation in Figure 7.1):   Application and investigation of various novel MPL materials, including: perforated SS, perforated GR sheet, graphene (in the form of GN foam, RGO, electrochemically exfoliated GN and CGN) and new CB composites.  Experimental comparison of alternative MPLs with a stronger emphasis on interfacial characteristics. This approach presents insights into novel interfacial aspects (such as CL adhesion and its role in the creation of CCM-based MPLs) and provides an enhanced understanding of key interfacial factors and their role in overall performance.   Extensive representation of MPL characteristics in the compressed state. This allows for a more realistic depiction of MPL material properties experienced in situ.  Investigation of novel interfacial contact measurements (through pressure sensitive film) and active electrical connectivity measurements (via ECSA) as additional methods to assess interfacial characteristics. Interfacial contact measurements were found to not represent in situ conditions well and the variance with the active electrical connectivity measurements was too large to make definitive conclusions. Chapter 7: Conclusions and future considerations     179   Insight into how different morphologies affect water retention. Compared to the conventional CB’s small, spherical particles, graphene’s unique planar morphology enhances water retention capabilities, leading to enhanced performance preservation at low humidity.  Investigation of CB composites as a tool to engineer MPL properties. This involves identification of dominant structures and insights into common effects resulting from the addition of CB in a 1:1 weight ratio (for example, CB increases the water permeability, through-plane conductivity and contact angle when added to all the base materials).  Evaluation of graphene-based MPLs’ application to low loaded CCMs. This work highlights that a lower CL loading (as well as structural and compositional differences, as displayed between JM and Gore CCM) can limit potential performance gains obtained with graphene-based MPLs.  Thus far, findings have been reported in two publications[162,178] with another two publications currently in preparation.    Chapter 7: Conclusions and future considerations     180    Figure 7.1: Schematic overview of the thesis, highlighting the areas of novelty and contributions of the research. Chapter 7: Conclusions and future considerations     181  7.2 Conclusions  The conclusions are presented in terms of the specified objectives of the study (secondary objectives in Chapter 1), which are restated in the following section:    A critical literature review of PEMFC fundamentals, MPL design modifications and CL interfacial studies. Conclusions from the literature review include the absence of graphene as a previously investigated MPL alternative, justifying its incorporation in this study. It was also found that there is a scarcity of experimental CL interfacial studies, while the majority of work (following modelling approaches) is also limited to conventional CB MPLs. This conclusion highlighted the need to consider interfacial factors, in addition to other MPL bulk properties, during the investigation of alternative MPL materials.     Identification of key factors that influence the MPL|CL interface. The study confirms several important interfacial factors, which have also previously been referred to in the literature. These factors include: compression, surface morphology, conformability, and interfacial and bulk resistance. Two other factors, which have not previously been considered in a similar context, were also presented, namely, wettability and CL layer adhesion. The role of these factors and their complex interdependencies, are briefly summarized in relation to overall PEMFC performance.  Compression Compression can induce significant changes in MPL material properties (which can also manifest as changes in interfacial properties). For example, the MPL’s surface morphology alters drastically by smoothening upon compression (e.g. 84% reduction in surface roughness for the RGO MPL), leading to improved MPL|CL interfacial contact. The bulk resistance of MPLs also reduces upon compression, with through-plane resistance showing the most extreme response (90% – 95% change). Through improved interfacial contact and Chapter 7: Conclusions and future considerations     182  reduced bulk resistance, the performance of a MPL can increase significantly in the ohmic region. This is illustrated by improvements of up to 100 mV in the ohmic region obtained for the commercial CB MPL when it is compressed from 483 kPa(g) to 1034 kPa(g).  However, over-compression can limit mass transport through reduced porosity and permeability of the MPL, as shown by the 160 mV loss between 758 kPa(g) and 1034 kPa(g) compression for the GN foam MPL (at 1700 mA cm-2). Surface smoothening can also alter the MPL’s surface wettability at the interface, which affects water removal from the CL (and membrane hydration in general). The most extreme example of this is also the RGO MPL, for which the contact angle increases by 113˚ between the uncompressed and compressed state.  Surface morphology The surface morphology of the MPL relates to its roughness parameters, as well as the type of structures presented at the interface. If the MPL surface features are dominant (rougher than the CCM, as is the case for this study), a smoother MPL will result in improved interfacial contact. Improved interfacial contact, in turn, influences the interfacial contact resistance. In Chapter 5, for example, the GN MPL presents the smoothest surface (average roughness of 1.5 ± 0.3 μm) and also has the lowest interfacial resistance (0 on a normalized basis). On the other hand, the GN+CB MPL (with similar bulk resistance), has a surface roughness over double that of the GN MPL and also shows a 15% increase in the interfacial resistance (on a normalized basis).  The trend between surface roughness and interfacial resistance does not scale linearly and is also not absolute though, since other factors can also influence the interfacial resistance.  The type of structure presented at the interface (water draining structures, such as the porous water-removal network of CB particles; or water retaining structures, such as planar and tortuous graphene structures), also plays a dominant role by affecting water retention at the CL interface. For instance, all the planar (water retaining) structures employed in this study resulted in greater performance preservation during longer-term testing at 20% cathode RH, compared to the porous network of the CB MPL. The extreme contrast in the Chapter 7: Conclusions and future considerations     183  water retention capabilities of these different structures are illustrated through the associated performance losses at these conditions: 464 mV h-1 for the porous CB network, which is much higher than 1.4 mV h-1 for the planar GN flakes, 17 mV h-1 for the planar RGO flakes, and 16 mV h-1 for the planar CGN flakes.    Interfacial and bulk resistance The interfacial resistance is a complex parameter which is understood to be influenced by a myriad of other interfacial factors itself: compression, surface morphology, conformability, and CL adhesion. Evidently, the MPL’s bulk resistance also influences the interfacial resistance (since the composition of the studied MPLs is considered similar throughout the bulk and at the interface of the MPLs). It is difficult to determine to what extent the mentioned factors (and their interactions) influence the interfacial resistance. However, the GN foam, GN and CGN MPLs, generally present as the least resistive layers within their respective datasets. These layers also continuously display the lowest interfacial resistance (0 on a normalized basis) or highest degrees of electrical connectivity. It furthermore remains difficult to determine the exact contribution of the different resistive elements to the total ohmic cell resistance. However, results from Chapter 5 indicate that ohmic resistive losses are strongly associated with the in-plane resistivities of the MPLs. By way of illustration, correlation coefficients of 0.97 and 0.87 are displayed with the resistance measured at OCV (i = 0 mA cm-2) and during operation (i = 760 mA cm-2), respectively.  Wettability MPL surface wettability is an important interfacial factor due to its impact on water removal from the CL interface. All the investigated graphene MPLs presented lower contact angles (more wettable surfaces) compared to that of the conventional CB MPL in the compressed state (149˚ ± 2˚): 87˚ ± 17˚ for the non-wetproofed GN foam MPL, 112˚ ± 5˚ for the GN MPL, 132˚ ± 1˚ for the CGN MPL and 32˚ ± 7˚ for the RGO MPL (the much lower contact angle for RGO is only observed in the uncompressed state).  A greater degree of surface wetting promotes water retention in the vicinity of the CL. This property is very beneficial for ‘dry’ Chapter 7: Conclusions and future considerations     184  (low humidity) or harsher operating conditions, but typically also promotes flooding under ‘wet’ (fully humidified) conditions.   Adhesion and conformability All graphene-based MPLs establish extreme degrees of CL adhesion (as confirmed through SEM imaging). Graphene flakes can furthermore transfer and adhere independently, promoting a high degree of conformability under pressure (this is in contrast to MPL materials, such as CB, which remain bonded to the underlying GDL, with layer conformability therefore primarily dictated by the GDL itself). Increased adhesion and conformability promote intimate contact with the CL, while also increasing water retention at the CL interface. CL adhesion furthermore facilitates the creation of self-made CCM-based MPLs. This behavior is advantageous, since it avoids challenges involved with the manufacturing of CCM-based MPLs (refer to Section 2.6).     Exploration of novel and promising materials as alternatives to conventional MPLs with improved interfacial characteristics and enhanced operational flexibility.   Alternative MPL materials In each phase of the study, a graphene MPL (GN foam in Chapter 4, GN in Chapter 5, and CGN in Chapter 6) demonstrates the most beneficial interfacial characteristics. These benefits include: the lowest interfacial resistance (0 on a normalized basis), very low bulk resistance (compressed state in-plane resistivities at least 20% lower than that of the CB MPL), CL adhesion, smooth surface morphology consisting of large planar flakes, and increased surface wettability (e.g. the compressed state contact angle for the GN MPL is almost 40˚ below that of the CB MPL). The latter three properties, in particular, were also found to promote performance preservation and durability at low humidity conditions. When comparing the performance of graphene MPLs to that of the conventional CB MPL, graphene always showed improvements in the kinetic and ohmic polarizations regions. For example, under one-dimensional control, improvements of up to 54 mV are achieved with Chapter 7: Conclusions and future considerations     185  the GN MPL in the ohmic polarization region. However, the graphene MPLs also suffer from more severe mass transport limitations, with the achieved limiting current typically being much lower than for the CB MPL (1000 mA cm-2 in the case of the GN MPL). This matter is successfully addressed through the addition of CB, however (refer to the subsequent discussion on CB composites).    CB composites The combination of CB with larger structures (such as GR, RGO, GN and CGN) in a 1:1 weight ratio, results in certain general effects in the properties of the composite MPLs. These general trends include the following (as example, the degree of relative change for the GN+CB MPL is provided): decreased surface wettability expressed as an increase in contact angle (27%), decreased through-plane resistance (28%), decreased porosity (13%), increased water permeability (45%), decreased gas permeability (41%) and decreased porosity (13%), with the exception of GR in the latter two cases.  The increased water permeabiltity is attributed to the fine pore network of CB particles which provide additional water removal pathways. The increased water permeabilities also have a significant influence on performance. In the case of RGO+CB, GN+CB, and CGN+CB MPL, mass transport is promoted compared to that of the original graphene constituent. For the GN+CB MPL, under one-dimensional control, the implications are significant:  the maximum power density of the composite MPL is 30% and 80% higher than that of the CB MPL at 100% and 20% RH.    Types of graphene The graphene-based materials used in this study (GN foam, RGO, electrochemically exfoliated GN and CGN) include a large variety in particle/flake size, carbon content, and the number of monolayers. When comparing the RGO to all the other graphene types, its higher degree of functionalization (with 8% nitrogen) manifests in its through-plane resistance being 10 times higher than the other graphene-based MPLs. The higher degree of functionalization and surface defects also causes less dense stacking/agglomeration of the RGO flakes, with the layer presenting as the thickest GDL-based MPL (99 μm and 86 μm Chapter 7: Conclusions and future considerations     186  thicker than that of the GN and CGN MPLs). The greater thickness of the RGO MPL also relates to the MPL having the highest in-plane resistivity of all the GDL-based MPLs (also 40% higher than that of the GN and CGN MPLs). Furthermore, the effect of particle size is observed by comparing GN and CGN particles, both produced through electrochemical methods and containing a carbon content above 93%: the larger CGN particles (3 – 18 μm) form MPLs that are less dense and with greater GDL pore exposure, compared to the smaller GN flakes (0.5 – 0.8 μm).  This causes the CGN MPL to have a gas permeability which is 30 times higher than that of GN MPL.  Despite all of the aforementioned differences, certain key features distinguish all graphene variants from the other materials in this study: planar flake-like structures which assemble into stacks (with pore spaces primarily situated in the longitudinal direction), and adhesion to the CL. These properties, in particular, improve interfacial contact (visually confirmed through SEM imaging) and water retention capabilities. GN, CGN and RGO MPLs consecutively maintain performance the best during longer-term testing at 20% cathode RH, while comparative humidity polarization also confirmed performance improvement for the GN foam MPL at low humidity. In the end, however, it is important to remain cognizant of the large variety in graphene materials, and to confirm a graphene variant’s performance before practical implementation.  The conclusions of the following two objectives are presented together.  Development and assessment of characterization techniques for the investigation of MPL properties.  Evaluation of the effect of alternative MPLs on overall performance and kinetic, ohmic and mass transport losses. The employed ex situ characterization techniques effectively distinguish property differences amongst all the investigated MPLs. By coupling the characterization results with extensive fuel cell testing at 100% and 20% cathode RH, performance differences are also interpreted with greater insight and understood in context of certain material properties. For instance, Chapter 7: Conclusions and future considerations     187  trends in ohmic loss show a strong association with the in-plane resistivities of MPLs, with a correlation coefficient of 0.87 obtained at 760 mA cm-2 for the GDL-based MPLs employed in Chapter 5. Mass transport behavior, in turn, appears heavily influenced by the water permeability of the MPLs. The influence of water permeability is also reflected during longer-term testing at 20% cathode RH: the more water permeable an MPL (or the lower its water retention capability), the higher the degree of membrane dehydration and resistance increase (CB > GR+CB > GR > RGO+CB > RGO > GN+CB > GN). With regards to performance in the kinetic region, the apparent Tafel slopes also show a dependence on water permeability for the GDL-based MPLs of Chapter 5. The relationship between the parameters is presented through a logarithmic function, with correlation coefficients of 0.85 and 0.96 obtained at 100% and 20% RH. The influence of water permeability likely manifests in the kinetic region due to its impact on ionic conductivity: as water permeability decreases and the water retention capability of a MPL increases, the water flux to the CL also increases, resulting in more effective ionomer hydration and higher proton concentrations.  This study furthermore shows that properties such as porosity cannot be considered in isolation, when investigating mass transport behavior. The inclusion of properties (such as permeability measurements) that reflect on flow behaviour, and which are also influenced by the material’s inherent porous structure and tortuosity, is therefore of utmost importance. In light of this, there are additional characterization methods that would provide valuable supporting evidence for the interpretation of results. These techniques are discussed further in Section 7.3. Methods were also developed to assess the degree of interfacial contact (through pressure sensitive film) and active electrical connectivity (via ECSA), but proved to be ineffective, due to inaccurate representation of in situ conditions and insufficient sensitivity, respectively.    Exploration of the feasibility of implementing alternative MPLs at larger scale. Tests performed with a 5 cm2 and a 49 cm2 active area, show that graphene-based performance enhancements, up to a current density of 500 mA cm-2, are within a similar Chapter 7: Conclusions and future considerations     188  range (10 – 20 mV). However, some differences are also observed in the performance of the different cell sizes, and are primarily related to differences in operating modes and cell hardware. For example, for the smaller sized cell, performance enhancements extend to 2200 and 1600 mA cm-2 for the JM High and Low CCMs (albeit still small). This is because the severity of flooding is minimized under one-dimensional control. The maximum current density of the polarization curves are also extended by up to 1000 mA cm-2, due to delayed mass transport and the lower stack voltage of the smaller TP-5 cell hardware. Finally, greater differences are also observed during comparative humidity testing in the small sized cell, due to the harsher conditions associated with one-dimensional control. While both cell sizes and operating modes proved useful within this research context, practical implementation ultimately necessitates that MEA components be optimized with regards to the hardware and operating conditions required for a specific application.    Investigation of the application of alternative MPLs to low-loaded CCMs. It was found that performance enhancements obtained with graphene-based MPLs are also restricted by the CL loading itself. This is illustrated through the JM High CCM (0.4 mg cm-2 Pt loading) achieving performance enhancements in the range of 10 – 26 mV, compared to smaller enhancements of 1 – 8 mV obtained with the JM Low CCM (0.1 mg cm-2 Pt loading). This phenomenon is attributed to the fact that low loaded CCMs suffer from additional resistive losses. Higher resistive losses for the JM Low CCM, compared to the JM High CCM, are confirmed through experimental measurements (these trends are illustrated for all MPLs but values obtained with the CGN MPL are provided as examples): 14 mΩ cm2 higher OCV resistance, 85% higher interfacial resistance (normalized basis), and 7 mV higher ohmic loss at 1000 mA cm-2. Despite the fact that the performance enhancement of the graphene-based MPLs with the low loaded CCMs are less significant, the material still proves beneficial, by lending improved performance maintenance during longer-term testing at 20% cathode RH (performance loss of 2 mV h-1 with CGN MPL compared to 9.6 mV h-1 with CB MPL in the TP-50). The results indicate that the material can be used to extend the durability of low loaded CCMs at low humidity conditions. Chapter 7: Conclusions and future considerations     189  Ultimately, based on the aforementioned conclusions, graphene and its CB composite prove to be promising alternatives to conventional CB MPLs. Graphene helps to establish improved interfacial characteristics, lower overall resistance and enhances performance. Operational flexibility is also enhanced through applicability for low cathode humidity (20% RH).   The MPL can also be tailored through the introduction of CB, to balance the two competing phenomena of water removal and water retention, as required for specific applications and operating conditions. As with the commercial application of any novel/alternative material, the performance and operational benefits of including graphene need to be weighed against the material cost. As a fairly new material, only discovered in 2004[179], graphene is still considered to be in its research and prototyping phase. By the end of 2015, the market cost of graphene was $100 per gram[180]. It is therefore likely that the inclusion of graphene is not yet financially feasible for most PEMFC applications, despite the performance enhancements that the material could introduce. It is projected, however, that the cost of graphene will decrease significantly as the commercial demand increases and the manufacturing processes are refined further. A report by Deloitte[180] states that the cost of graphene is eventually expected to reduce to that of the raw material, which is graphite in the case of top-down synthesis. The cost of graphene-based MPLs may furthermore decrease based on the inclusion of CB in composite form. The inclusion of graphene-based MPLs could also result in other cost reductions relating to system costs (decreased dependence on humidifiers and a reduction of parasitic losses) and PEMFC durability (reducing the need for MEA replacement). The aforementioned factors will certainly support the commercial viability of graphene as an MPL material in the long-term.       Chapter 7: Conclusions and future considerations     190  7.3 Recommendations and future considerations  To build upon this work, several avenues may be pursued in future studies:   Additional/alternative characterization techniques This study highlights the important role of MPLs’ internal structures. Although imaging and a variety of other techniques were used to represent structural differences, the incorporation of certain techniques, (if feasible) may offer supporting or additional evidence in future studies. For example, since surface wettability measurements do not always correlate well with observed mass transport behavior and only represent conditions experienced at the CL interface, internal wettability measurements may serve more useful. The disadvantage of such measurements is that it proves very time-consuming, since it involves measurements with several wetting fluids (via the Washburn method), which, in combination with the Owen-Wendts theory, is then used to extrapolate the internal contact angle of water[132]. Quantifying the pore size distribution and tortuosity (as well as more accurate descriptions of porosity) may also help to understand general flow behaviour better. However, a major disadvantage of conventional porosimetric techniques (such a mercury porosimetry) are that they are typically very material intensive and destructive, and would therefore require large amounts of sample material. This would not only necessitate large amounts of the original bulk powders, but also extensive preparation time to manufacture sufficient MPLs. Accurate assessment of these properties may therefore remain very challenging until alternative MPLs are produced at a commercial scale.   To obtain a better representation of MPL cross-sectional and porous structure, MPLs can be cast in resin and cut by means of grinding. A focused ion beam can also be used to cut samples, since grinding methods can still incur some sample damage[181]. The latter method can be very time-intensive though. Other characterization approaches that may prove very insightful in the study of alternative MPLs, include in operando visualization Chapter 7: Conclusions and future considerations     191  techniques, such as Synchrotron X-ray radiography[182]. Such techniques are typically not easily accessible, though, since they involve highly specialized equipment.   Modelling and optimization of different MPL structures While the study provides valuable insight into novel and alternative MPL structures, the experimental approach remains subject to experimental variation and error. Furthermore, due to the fact that a multitude of variables change when a new material is investigated experimentally (e.g. particle size, porosity, conductivity, wettability etc.) the effect of a single variable on performance and interfacial effects could not be studied in isolation. Modelling of alternative MPL structures and geometries, and their subsequent effects on the interface and mass transport phenomena, may therefore offer the chance to delineate these factors more clearly. Modelling may also not only provide insight into macroscopic effects, but allow investigation on smaller size scales. Even though comprehensive modelling of alternative structures could be very expensive computationally (particularly when considering both gas and liquid phase flow), it continuously becomes more feasible as computer power expands. By use of carefully designed experiments, certain MPL properties may also be optimized. For example, certain MPL engineering parameters that can be easily controlled (such as the MPL load, composite ratio and PTFE content), can be adjusted and investigated through a 2k factorial design. For a specific application, these parameters can then be optimized with regards to response variables, such as the performance at a specific current density, the limiting current density or the maximum power density.   Influence of different composite ratios, loadings and graphene properties Although this work provides insight into how CB-composites can be used as a tool to engineer MPL properties and overall performance, the investigation is limited to composites prepared in a 1:1 weight ratio. For a more comprehensive understanding, other ratios and material structures (e.g. differently shaped particles) should also be investigated in future. Graphene-based MPLs, and their composites with CB, may further Chapter 7: Conclusions and future considerations     192  be optimized with regards to material ratios, loading and wetproofing content. Experimental studies can also focus on the effect of different graphene properties (e.g. flake size, carbon content and number of monolayers). This may add further value to the tailoring of MPL characteristics and aid in choosing a graphene best suited for a specific PEMFC application.   Different MPL ink preparation and deposition methods The use of different solvents and solvent:water ratios may have a significant effect on the consistency of MPL inks. Different deposition methods, such as rolling, screen-printing and painting, may also result in MPL structural and uniformity differences. The effects of these aspects on overall MPL properties are therefore worth studying in greater detail.   Alternative MPL and CL configurations To better understand the potential benefits and challenges associated with CCM-based MPLs, this particular MPL configuration should be investigated more thoroughly. Such investigations would involve direct comparison of conventional CB and alternative MPL materials (such as graphene), as CCM-based MPLs. This work can also extend to the specific manufacturing aspects relating to the application of different materials as CCM-based MPLs. The incorporation of graphene-based MPLs as part of GDEs (with the CL deposited directly on top of a GDL-based MPL), can also be evaluated and compared to the versions presented in this study where GDL-based MPLs were used in conjunction with CCMs. Based on graphene’s previous use as catalyst support and as a catalytic component itself (in doped form), multifunctional graphene layers can furthermore be developed. For example, a graphene layer can be used to serve as both the MPL and CL components. These layers could potentially marry the graphene MPL’s benefits for low humidity operation, with that of incorporating alternative CL components (e.g. reduced Pt loading).   Chapter 7: Conclusions and future considerations     193   Alternate approaches to improve water management In addition to optimizing MPL parameters (such as the amount of PTFE, material loading and composite ratios), operating variables can also be tuned to enhance performance with graphene-based MPLs. Strategies may include increasing water removal from the cathode CL to the anode, through back-diffusion. Back-diffusion can be enhanced by operating the cathode at a higher pressure than the anode, establishing a greater pressure gradient (i.e. driving force) between the two regions. Alternatively, the water concentration gradient can also be increased by operating the anode at a very low humidity compared to the cathode side. To specifically enhance water removal in the vapor phase, it is necessary to prevent/minimize condensation within components such as the MPL. MPLs with higher thermal conductivities will establish lower thermal gradients and will help to retain heat for the promotion of the vapor phase existence of water. To fully understand graphene’s role in this regard, the thermal conductivities of graphene-based MPLs should be investigated. References     194  References  [1] F. Migliardini, O. Veneri, P. Corbo, J. Ind. Eng. Chem. 2011, 17, 633–641. [2] I. Nitta, S. Karvonen, O. Himanen, M. Mikkola, Fuel Cells 2008, 8, 410–421. [3] Y. Wang, K. S. Chen, J. Mishler, S. C. Cho, X. C. Adroher, Appl. Energy 2011, 88, 981–1007. [4] A. Z. Weber, J. Newman, J. Electrochem. Soc. 2005, 152, A677. [5] L. Cindrella, A. M. Kannan, J. F. Lin, K. Saminathan, Y. Ho, C. W. Lin, J. Wertz, J. Power Sources 2009, 194, 146–160. [6] S. Park, J.-W. Lee, B. N. Popov, Int. J. Hydrogen Energy 2012, 37, 5850–5865. [7] T. Swamy, E. C. Kumbur, M. M. Mench, J. Electrochem. Soc. 2010, 157, B77. [8] H. Bajpai, M. Khandelwal, E. C. Kumbur, M. M. Mench, J. Power Sources 2010, 195, 4196–4205. [9] A. Bonnefont, P. Ruvinskiy, M. Rouhet, A. Orfanidi, S. Neophytides, E. Savinova, Wiley Interdiscip. Rev. Energy Environ. 2014, 3, 505–521. [10] M. K. Debe, ECS Trans. 2012, 45, 47–68. [11] K. Takahashi, K. Kakinuma, M. Uchida, J. Electrochem. Soc. 2016, 163, F1182–F1188. [12] S. Shukla, K. Domican, K. Karan, S. Bhattacharjee, M. Secanell, Electrochim. Acta 2015, 156, 289–300. [13] A. Sode, N. J. C. Ingle, M. McCormick, D. Bizzotto, E. Gyenge, S. Ye, S. Knights, D. P. Wilkinson, J. Memb. Sci. 2011, 376, 162–169. [14] T. Swamy, F. E. Hizir, M. Khandelwal, E. C. Kumbur, M. M. Mench, ECS Trans. 2009, 25, 15–27. [15] G. Huang, M. Chang, Internatinal J. Electrochem. Sci. 2014, 9, 7819–7831. [16] T. Kitahara, H. Nakajima, M. Inamoto, K. Shinto, J. Power Sources 2014, 248, 1256–1263. [17] H. H. Chen, M. H. Chang, J. Power Sources 2013, 232, 306–309. [18] S. Hou, H. Su, H. Zou, D. Dang, H. Song, X. Li, S. Liao, Int. J. Hydrogen Energy 2015, 40, 15613–15621. [19] M. Blanco, D. P. Wilkinson, H. Wang, Int. J. Hydrogen Energy 2011, 36, 3635–3648. [20] K.-S. Lee, J. S. Spendelow, Y.-K. Choe, C. Fujimoto, Y. S. Kim, Nat. Energy 2016, 1, 16120. References     195  [21] B. Yang, Y. Z. Fu, A. Manthiram, J. Power Sources 2005, 139, 170–175. [22] F. B. Weng, J. S. Wang, T. L. Yu, J. Chinese Inst. Chem. Eng. 2008, 39, 429–433. [23] C.-Y. Jung, J.-Y. Yi, S.-C. Yi, Energy 2014, 68, 794–800. [24] I. Choi, H. Lee, K. G. Lee, S. H. Ahn, S. J. Lee, H. J. Kim, H. N. Lee, O. J. Kwon, Appl. Catal. B Environ. 2015, 168–169, 220–227. [25] H. Liang, L. Zheng, S. Liao, Int. J. Hydrogen Energy 2012, 37, 12860–12867. [26] L. Cindrella, A. M. Kannan, J. Power Sources 2009, 193, 447–453. [27] T. Kitahara, H. Nakajima, K. Okamura, J. Power Sources 2015, 283, 115–124. [28] M. Ji, Z. Wei, Energies 2009, 2, 1057–1106. [29] W. Dai, H. Wang, X. Yuan, J. J. Martin, D. Yang, J. Qiao, J. Ma, J. Power Sources 2009, 9461–9478. [30] Q. Yan, H. Toghiani, J. Wu, J. Power Sources 2006, 158, 316–325. [31] R. O’Hayre, D. M. Barnett, F. B. Prinz, J. Electrochem. Soc. 2005, 152, A439. [32] N. Zamel, X. Li, Prog. Energy Combust. Sci. 2013, 39, 111–146. [33] G. Lin, T. Van Nguyen, J. Electrochem. Soc. 2005, 152, A1942–A1948. [34] K. Karan, H. Atiyeh, A. Phoenix, E. Halliop, J. Pharoah, B. Peppley, Electrochem. Solid-State Lett. 2007, 10, B34. [35] S. B. Park, S. Kim, Y. Park, M.-H. Oh, J. Phys. Conf. Ser. 2009, 165, 12046. [36] H. Tang, S. Wang, S. P. Jiang, M. Pan, J. Power Sources 2007, 170, 140–144. [37] M. Hu, S. Sui, X. Zhu, Q. Yu, G. Cao, X. Hong, H. Tu, Int. J. Hydrogen Energy 2006, 31, 1010–1018. [38] J. Zhang, J. Wu, H. Zhang, PEM Fuel Cell Testing and Diagnosis, Elsevier B.V, Amstserdam, 2013. [39] A. N. Ghadge, M. Sreemannarayana, N. Duteanu, M. . Ghangrekar, J. Electrochem. Sci. Eng. 2014, 4, 315–326. [40] R. P. O’Hayre, Fuel Cell Fundamentals, 2006. [41] K. J. Lange, P.-C. Sui, N. Djilali, J. Power Sources 2011, 196, 3195–3203. [42] X. Zhang, Y. Gao, H. Ostadi, K. Jiang, R. Chen, Int. J. Hydrogen Energy 2014, 39, 17222–17230. References     196  [43] R. Zaffou, J. S. Yi, H. R. Kunz, J. M. Fenton, Electrochem. Solid-State Lett. 2006, 9, A418–A422. [44] U. Pasaogullari, C.-Y. Wang, Electrochim. Acta 2004, 49, 4359–4369. [45] J. H. Nam, M. Kaviany, Int. J. Heat Mass Transf. 2003, 46, 4595–4611. [46] X. L. Wang, H. M. Zhang, J. L. Zhang, H. F. Xu, Z. Q. Tian, J. Chen, H. X. Zhong, Y. M. Liang, B. L. Yi, Electrochim. Acta 2006, 51, 4909–4915. [47] D. Spernjak, A. K. Prasad, S. G. Advani, J. Power Sources 2007, 170, 334–344. [48] T. Kim, S. Lee, H. Park, Int. J. Hydrogen Energy 2010, 35, 8631–8643. [49] U. Pasaogullari, C.-Y. Wang, K. S. Chen, J. Electrochem. Soc. 2005, 152, A1574. [50] G. Lin, T. Van Nguyen, J. Electrochem. Soc. 2005, 152, A1942. [51] J. P. Owejan, J. E. Owejan, W. Gu, T. A. Trabold, T. W. Tighe, M. F. Mathias, J. Electrochem. Soc. 2010, 157, B1456. [52] Z. Lu, M. M. Daino, C. Rath, S. G. Kandlikar, Int. J. Hydrogen Energy 2010, 35, 4222–4233. [53] S. G. Kandlikar, M. L. Garofalo, Z. Lu, Fuel Cells 2011, 11, 814–823. [54] A. Casalegno, L. Colombo, S. Galbiati, R. Marchesi, J. Power Sources 2010, 195, 4143–4148. [55] J. T. Gostick, M. A. Ioannidis, M. W. Fowler, M. D. Pritzker, Electrochem. commun. 2009, 11, 576–579. [56] P. Deevanhxay, T. Sasabe, S. Tsushima, S. Hirai, Electrochem. commun. 2013, 34, 239–241. [57] J. H. Nam, K.-J. Lee, G.-S. Hwang, C.-J. Kim, M. Kaviany, Int. J. Heat Mass Transf. 2009, 52, 2779–2791. [58] H. K. Atiyeh, K. Karan, B. Peppley, A. Phoenix, E. Halliop, J. Pharoah, J. Power Sources 2007, 170, 111–121. [59] J. Shi, J. Tian, C. Zhang, Z. Shan, J. Power Sources 2007, 164, 284–286. [60] K. A. Friedrich, A. Bauder, J. Haußmann, H. Markötter, I. Manke, R. Alink, J. Scholta, Meet. Abstr. 2013, MA2013-02, 1544. [61] V. Gurau, M. J. Bluemle, E. S. De Castro, Y.-M. Tsou, T. A. Zawodzinski, J. A. Mann, J. Power Sources 2007, 165, 793–802. References     197  [62] N. Furuya, N. Mineo, J. New Mater. Electrochem. Syst. 2007, 10, 205–208. [63] N. E. Cipollini, Polymer Coating of Pem Fuel Cell Catalyst Layers, 2009, WO 2009131554 A1. [64] J. Park, U. Pasaogullari, L. J. Bonville, ECS Trans. 2014, 64, 353–359. [65] A. J. Bard, T. Journal, S. R. Samms, S. Wasmus, R. F. Savineil, 1996, 143, 1498–1504. [66] S. Park, J.-W. Lee, B. N. Popov, J. Power Sources 2006, 163, 357–363. [67] H.-K. Lee, J.-H. Park, D.-Y. Kim, T.-H. Lee, J. Power Sources 2004, 131, 200–206. [68] Z. Qi, A. Kaufman, J. Power Sources 2002, 109, 38–46. [69] L. R. Jordan, A. K. Shukla, T. Behrsing, N. R. Avery, B. C. Muddle, M. Forsyth, J. Power Sources 2000, 86, 250–254. [70] S. Carlos, J. Appl. Electrochem. 1996, 26, 297–304. [71] R. Schweiss, M. Steeb, P. M. Wilde, Fuel Cells 2010, 10, 1176–1180. [72] F. Lufrano, E. Passalacqua, G. Squadrito, A. Patti, L. Giorgi, C. R. Casaccia, S. M. Galeria, J. Appl. Electrochem. 1999, 29, 445–448. [73] J. H. Chun, K. T. Park, D. H. Jo, J. Y. Lee, S. G. Kim, S. H. Park, E. S. Lee, J.-Y. Jyoung, S. H. Kim, Int. J. Hydrogen Energy 2011, 36, 8422–8428. [74] S. Park, J.-W. Lee, B. N. Popov, J. Power Sources 2008, 177, 457–463. [75] L. Cindrella, A. M. Kannan, R. Ahmad, M. Thommes, Int. J. Hydrogen Energy 2009, 34, 6377–6383. [76] Y. Wang, S. Al Shakhshir, X. Li, Appl. Energy 2011, 88, 2168–2175. [77] M. Ahn, Y.-H. Cho, Y.-H. Cho, J. Kim, N. Jung, Y.-E. Sung, Electrochim. Acta 2011, 56, 2450–2457. [78] C. S. Kong, D. Kim, H. Lee, Y. Shul, T. Lee, 2002, 108, 185–191. [79] J. H. Chun, K. T. Park, D. H. Jo, J. Y. Lee, S. G. Kim, E. S. Lee, J.-Y. Jyoung, S. H. Kim, Int. J. Hydrogen Energy 2010, 35, 11148–11153. [80] M. P. Manahan, M. C. Hatzell, E. C. Kumbur, M. M. Mench, J. Power Sources 2011, 196, 5573–5582. [81] D. Gerteisen, T. Heilmann, C. Ziegler, J. Power Sources 2008, 177, 348–354. [82] D. Gerteisen, C. Sadeler, J. Power Sources 2010, 195, 5252–5257. References     198  [83] International Carbon Black Association, “What is carbon black?,” can be found under http://www.carbon-black.org/index.php/what-is-carbon-black, 2016. [84] Cabot Incorporation, “Specialty carbon black - Vulcan XC72R,” can be found under www.cabotcorp.com/~/media/files/product-datasheets/datasheet-vulcan-xc72pdf.pdf, 2016. [85] E. Antolini, R. . Passos, E. . Ticianelli, J. Power Sources 2002, 109, 477–482. [86] E. Passalacqua, G. Squadrito, F. Lufrano, A. Patti, L. Giorgi, J. Appl. Electrochem. 2001, 31, 449–454. [87] W.-M. Yan, D.-K. Wu, X.-D. Wang, A.-L. Ong, D.-J. Lee, A. Su, J. Power Sources 2010, 195, 5731–5734. [88] J. Yu, Y. Yoshikawa, T. Matsuura, M. N. Islam, M. Hori, Electrochem. Solid-State Lett. 2005, 8, A152. [89] J. Yu, M. N. Islam, T. Matsuura, M. Tamano, Y. Hayashi, M. Hori, Electrochem. Solid-State Lett. 2005, 8, A320–A323. [90] M. Hunsom, P. Piumsomboon, K. Pruksathorn, N. Tantavichet, S. Endoo, K. Charutavai, K. Poochinda, Renew. Energy 2011, 36, 369–373. [91] L. Cindrella, A. M. Kannan, Fuel Cells 2010, 10, 563–566. [92] A. Kannan, A. Menghal, I. Barsukov, Electrochem. commun. 2006, 8, 887–891. [93] A. M. Kannan, V. P. Veedu, L. Munukutla, M. N. Ghasemi-Nejhad, Electrochem. Solid-State Lett. 2007, 10, B47. [94] Z. Xie, G. Chen, X. Yu, M. Hou, Z. Shao, S. Hong, C. Mu, Int. J. Hydrogen Energy 2015, 40, 8958–8965. [95] G.-B. Jung, W.-J. Tzeng, T.-C. Jao, Y.-H. Liu, C.-C. Yeh, Appl. Energy 2013, 101, 457–464. [96] A. M. Kannan, P. Kanagala, V. Veedu, J. Power Sources 2009, 192, 297–303. [97] P. Gallo Stampino, L. Omati, C. Cristiani, G. Dotelli, Fuel Cells 2010, 10, 270–277. [98] H. Gharibi, M. Javaheri, R. A. Mirzaie, Int. J. Hydrogen Energy 2010, 35, 9241–9251. [99] S. Y. Lin, M. H. Chang, Int. J. Hydrogen Energy 2015, 40, 7879–7885. [100] D. Ebenezer, K. Neelima, M. Jagannatham, P. Haridoss, Fuel Cells 2016, 16, 349–355. [101] A. M. Kannan, L. Munukutla, J. Power Sources 2007, 167, 330–335. References     199  [102] Q. Duan, B. Wang, J. Wang, H. Wang, Y. Lu, J. Power Sources 2010, 195, 8189–8193. [103] H. Zamora, P. Cañizares, M. A. Rodrigo, J. Lobato, Fuel Cells 2015, 15, 375–383. [104] H. Zamora, J. Plaza, P. Canizares, J. Lobato, M. A. Rodrigo, ChemSusChem 2016, 9, 1187–1193. [105] Y. Jiang, J. Hao, M. Hou, H. Zhang, X. Li, Z. Shao, B. Yi, RSC Adv. 2015, 5, 104095–104100. [106] J. Lobato, H. Zamora, P. Cañizares, J. Plaza, M. A. Rodrigo, J. Power Sources 2015, 288, 288–295. [107] S. Han, D. Wu, S. Li, F. Zhang, X. Feng, Adv. Mater. 2014, 26, 849–864. [108] S. Park, Y. Shao, H. Wan, P. C. Rieke, V. V. Viswanathan, S. a. Towne, L. V. Saraf, J. Liu, Y. Lin, Y. Wang, Electrochem. commun. 2011, 13, 258–261. [109] A. Marinkas, F. Arena, J. Mitzel, G. M. Prinz, A. Heinzel, V. Peinecke, H. Natter, Carbon N. Y. 2013, 58, 139–150. [110] X. Zhou, J. Qiao, L. Yang, J. Zhang, Adv. Energy Mater. 2014, 4, 1–25. [111] M. S. A. Bhuyan, M. N. Uddin, M. M. Islam, F. A. Bipasha, S. S. Hossain, Int. Nano Lett. 2016, 6, 65–83. [112] O. C. Compton, S. T. Nguyen, Small 2010, 6, 711–723. [113] Q. Zheng, J.-K. Kim, J. Chem. Inf. Model. 2013, 53, 1689–1699. [114] A. T. Najafabadi, E. Gyenge, Carbon N. Y. 2014, 71, 58–69. [115] A. Taheri Najafabadi, E. Gyenge, Carbon N. Y. 2015, 84, 449–459. [116] A. Bianco, H. M. Cheng, T. Enoki, Y. Gogotsi, R. H. Hurt, N. Koratkar, T. Kyotani, M. Monthioux, C. R. Park, J. M. D. Tascon, et al., Carbon N. Y. 2013, 65, 1–6. [117] H. Meng, Int. J. Hydrogen Energy 2009, 34, 5488–5497. [118] T. Hottinen, O. Himanen, S. Karvonen, I. Nitta, J. Power Sources 2007, 171, 113–121. [119] R. Makharia, M. F. Mathias, D. R. Baker, J. Electrochem. Soc. 2005, 152, A970. [120] I. Nitta, O. Himanen, M. Mikkola, Electrochem. commun. 2008, 10, 47–51. [121] H. Butsch, C. Roth, D. Ritzinger, G. Hoogers, A. Bock, J. Electrochem. Soc. 2012, 159, B709. [122] A. R. Kalidindi, R. Taspinar, S. Litster, E. C. Kumbur, Int. J. Hydrogen Energy 2013, 38, 9297–9309. [123] I. V. Zenyuk, E. C. Kumbur, S. Litster, J. Power Sources 2013, 241, 379–387. References     200  [124] F. E. Hizir, S. O. Ural, E. C. Kumbur, M. M. Mench, J. Power Sources 2010, 195, 3463–3471. [125] Y. Tabe, Y. Aoyama, K. Kadowaki, K. Suzuki, T. Chikahisa, J. Power Sources 2015, 287, 422–430. [126] S. Prass, S. Hasanpour, P. K. Sow, A. B. Phillion, W. M??rida, J. Power Sources 2016, 319, 82–89. [127] S. Kim, M. Khandelwal, C. Chacko, M. M. Mench, J. Electrochem. Soc. 2009, 156, B99–B108. [128] D. Ye, E. Gauthier, J. B. Benziger, M. Pan, J. Power Sources 2014, 256, 449–456. [129] S. Prass, Interfacial Morphology and Contact Resistance between the Catalyst and Microprous Layers in Proton Exchange Membrane Fuel Cells, University of British Columbia, 2016. [130] J. Kleemann, F. Finsterwalder, W. Tillmetz, J. Power Sources 2009, 190, 92–102. [131] The Centre for High-Throughput Phenogenomics at the University of British Columbia, “Olympus LEXT Confocal,” can be found under http://www.phenogenomics.dentistry.ubc.ca/equipment/OlympusLEXTConfocal/, 2016. [132] A. Arvay, E. Yli-rantala, C. Liu, X. Peng, P. Koski, L. Cindrella, P. Kauranen, P. M. Wilde, A. M. Kannan, J. Power Sources 2012, 213, 317–337. [133] R. J. Good, L. A. Girifalco, J. Phys. Chem. 1960, 64, 561–565. [134] P. A. Tran, T. J. Webster, Int. J. Nanomedicine 2013, 8, 2001–2009. [135] M. F. Mathias, J. Roth, J. Fleming, W. Lehnert, Diffusion Media Materials and Characterisation, 2010. [136] F. M. Smits, Bell Syst. Tech. J. 1958, 711–718. [137] C.-Y. Wen, Y.-S. Lin, C.-H. Lu, J. Power Sources 2009, 192, 475–485. [138] S. Litster, N. Djilali, in Transp. Phenom. Fuel Cells, 2005, pp. 175–213. [139] M. V. Williams, E. Begg, L. Bonville, H. R. Kunz, J. M. Fenton, J. Electrochem. Soc. 2004, 151, A1173-1180. [140] I. S. Hussaini, C. Y. Wang, J. Power Sources 2010, 195, 3830–3840. [141] S. Odaya, R. K. Phillips, Y. Sharma, J. Bellerive, A. B. Phillion, M. Hoorfar, Electrochim. Acta References     201  2015, 152, 464–472. [142] Fuel Cell Store, “DuPont TM Teflon ® PTFE DISP 30 Fluoroplastic Resin - Property Sheet,” can be found under http://fuelcellstore.com/fuel-cell-components/dispersions/teflon-disp30, 2016. [143] S. H. Ge, B. L. Yi, J. Power Sources 2003, 124, 1–11. [144] M. Blanco, Study of Selected Water Management Strategies for Proton Exchange Membrane Fuel Cell, University of British Columbia, 2011. [145] Fuel Cells Etc, “MEA activation procedures,” can be found under http://fuelcellsetc.com/2013/04/membrane-electrode-assembly-mea-activation-procedures/, 2014. [146] M. Blanco, D. P. Wilkinson, H. Wang, Electrochem. Solid-State Lett. 2012, 15, B20. [147] M. S. Yazici, J. Power Sources 2007, 166, 424–429. [148] ACS Material, “3D Freestanding graphene foam,” can be found under http://www.acsmaterial.com/3d-freestanding-graphene-foam-930.html, 2015. [149] H. Lin, S. Xu, X. Wang, N. Mei, Nanotechnology 2013, 24, 415706. [150] A. El-kharouf, T. J. Mason, D. J. L. Brett, B. G. Pollet, J. Power Sources 2012, 218, 393–404. [151] R. Mukundan, J. . Davey, J. . Fairweather, D. Spernjak, J. . Spendelow, D. S. Hussey, D. Jacobson, P. Wilde, R. Schweiss, R. L. Borup, ECS Trans. 2010, 33, 1109–1114. [152] J. André, L. Antoni, J.-P. Petit, E. De Vito, A. Montani, Int. J. Hydrogen Energy 2009, 34, 3125–3133. [153] Y. Liu, L. Gao, J. Sun, Y. Wang, J. Zhang, Nanotechnology 2009, 465605, 1–7. [154] J. H. Jung, J. H. Jeon, V. Sridhar, I. K. Oh, Carbon N. Y. 2011, 49, 1279–1289. [155] E. Gyenge, Fuel Cells and Electrochemical Engineering, CHBE 577 Class Notes, 2014. [156] X. Wang, H. Zhang, J. Zhang, H. Xu, X. Zhu, J. Chen, B. Yi, J. Power Sources 2006, 162, 474–479. [157] O. N. Shornikova, E. V Kogan, N. E. Sorokina, V. V Avdeev, 2009, 83, 1161–1164. [158] M. P. Trammell, P. J. Pappano, Analysis of Natural Graphite , Synthetic Graphite , and Thermosetting Resin Candidates for Use in Fuel Compact Matrix, 2011. [159] D. Chen, H. Feng, J. Li, Chem. Rev. 2012, 112, 6027–6053. References     202  [160] S. M. Senthil Kumar, J. Soler Herrero, S. Irusta, K. Scott, J. Electroanal. Chem. 2010, 647, 211–221. [161] A. T. Najafabadi, E. Gyenge, Carbon N. Y. 2014, 71, 58–69. [162] A. T. Najafabadi, M. J. Leeuwner, D. P. Wilkinson, E. L. Gyenge, ChemSusChem 2016, 1–10. [163] Carbon Canada, “Graphene,” can be found under https://www.canadacarbon.com/what-is-graphene, 2017. [164] Harvard Universtiy, “A Summary of Error Propagation, Department of Physics,” 2007. [165] X. Cao, Y. Shi, W. Shi, G. Lu, X. Huang, Q. Yan, Q. Zhang, H. Zhang, Small 2011, 7, 3163–3168. [166] D. Todd, M. Schwager, W. Merida, J. Electrochem. Soc. 2014, 162, F265–F272. [167] M. Matyka, A. Khalili, Z. Koza, Phys. Rev. E - Stat. Nonlinear, Soft Matter Phys. 2008, 78, 1–8. [168] A. Wilson, J. Marcinkoski, D. Papageorgopoulos, R. Ahluwalia, B. James, C. Houchins, J. Huya-Kouadio, DOE Hydrogen and Fuel Cells Program Record Title: Fuel Cell System Cost -2016 Originator, 2016. [169] M. Breitwieser, M. Klingele, B. Britton, S. Holdcroft, R. Zengerle, S. Thiele, Electrochem. commun. 2015, 60, 168–171. [170] S. Y. Cha, J. Electrochem. Soc. 1999, 146, 4055. [171] A. Z. Weber, A. Kusoglu, J. Mater. Chem. A 2014, 2, 17207–17211. [172] J. P. Owejan, J. E. Owejan, W. Gu, J. Electrochem. Soc. 2013, 160, F824–F833. [173] InvestorIntel, “Beyond the experimental phase: NanoXplore already developing a market for graphene,” can be found under https://investorintel.com/sectors/technology-metals/technology-metals-intel/beyond-experimental-phase-nanoxplore-already-developing-market-graphene/, 2014. [174] NanoXplore, “NanoXplore,” can be found under http://www.nanoxplore.ca/, n.d. [175] M. F. Mathias, J. Roth, J. Fleming, W. Lehnert, in Polym. Electrolyte Membr. Fuel Cells Syst., 2010. [176] “DOE Technical Targets for Polymer Electrolyte Membrane Fuel Cell Components,” can References     203  be found under https://www.energy.gov/eere/fuelcells/doe-technical-targets-polymer-electrolyte-membrane-fuel-cell-components, 2016. [177] M. Blanco, Study of Selected Water Management Strategies for Proton Exchange Membrane Fuel Cells, 2011. [178] M. J. Leeuwner, D. P. Wilkinson, E. L. Gyenge, Fuel Cells 2015, 15, 790–801. [179] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva, A. A. Firsov, Science (80-. ). 2004, 306, 666–669. [180] Deloitte Global, Graphene: Research Now, Reap next Decade, 2016. [181] S. Ghosh, H. Ohashi, H. Tabata, Y. Hashimasa, T. Yamaguchi, Int. J. Hydrogen Energy 2015, 40, 15663–15671. [182] R. Banerjee, N. Ge, J. Lee, M. G. George, S. Chevalier, H. Liu, P. Shrestha, D. Muirhead, A. Bazylak, J. Electrochem. Soc. 2017, 164, F154–F162. [183] K. Cooper, Fuel Cell Mag. 2009, 1–3. [184] H. A. Gasteiger, S. S. Kocha, B. Sompalli, F. T. Wagner, Appl. Catal. B Environ. 2005, 56, 9–35. [185] N. Limjeerajarus, T. Yanagimoto, T. Yamamoto, H. Ohashi, T. Ito, T. Yamaguchi, J. Chem. Eng. Japan 2009, 42, 39–46.   Appendices     204  Appendices  Appendix A: Chapter 3  Appendix A provides additional detail on the experimental methods presented in Chapter 3.  A.1  Through-plane resistance equipment   Figure A.1.1 shows a drawing of the through-plane resistance equipment with pneumatic compression piston (design and manufacturing by Greg Afonso).   Figure A.1.1: Drawing of the through-plane resistance equipment.  A.2  Gas permeability equipment   A drawing of the gas permeability equipment is provided in Figure A.2.1 (final design and manufacturing by Greg Afonso). Appendices     205   Figure A.2.1: Drawing of the gas permeability equipment with (a) assembled, (b) cross-sectional and (c) internal views.  A.3  Active electrical connectivity  As mentioned in Section 3.1.12, a technique was also developed to try and quantify the degree of ‘active electrical connectivity’ between the MPL and CL. The active electrical connectivity is defined as an interfacial parameter which represents the establishment of electrical connections between the MPL and active Pt sites in the CL (also known as TPB sites). The premise of the technique was to quantify the degree of active electrical connectivity using ECSA (electrochemically active surface area) measurements, via the Hupd (hydrogen underpotential deposition) technique.   For the determination of the ECSA via Hupd, CV (cyclic voltammetry) analysis was employed whereby the WE (working electrode) was cycled within a certain voltage range that enables adsorption-limited reactions (typically between 0.05 V and 1.2 V). This allows the Appendices     206  electrochemically active sites to be determined from the charge required for monolayer hydrogen adsorption/desorption[183]. Figure A.3.1 illustrates a typical Pt CV in acidic media. Region ① and ② indicate the hydrogen desorption and adsorption peaks respectively. The ECSA was determined using Equation A.3.1. The charge density, 𝑞𝑃𝑡, was calculated by integration of the hydrogen adsorption peak, illustrated by the grey region in Figure A.3.1 (although the desorption peak can typically also be used). Current resulting from charging or discharging of the electric double layer (indicated between the red lines in Figure A.3.1) was excluded during the integration calculation, to avoid inflation of 𝑞𝑃𝑡.  𝐸𝐶𝑆𝐴 =𝑞𝑃𝑡Γ. 𝐿  where  𝐸𝐶𝑆𝐴 = electrochemically active surface area (cm2Pt gPt-1) 𝑞𝑃𝑡 = charge density (C cm-2electrode) Γ = charge required to reduce a monolayer of protons = 210 μC cm-2Pt 𝐿 = Pt loading (gPt cm-2electrode)  Equation A.3.1   Figure A.3.1: Typical CV illustrating ECSA measurement through Hupd method (with the grey region representing the integration of the hydrogen adsorption peak).  Appendices     207  The MEAs were assembled with GDL-based MPLs used in Chapter 5 (with a total loading of 1.5 mg cm-2). The MEAs were further inserted into the TP-5 hardware used for fuel cell testing, and compressed via a pneumatic piston (Figure A.3.2). To enable the transport of protons, the PEM had to be hydrated. 0.5 M H2SO4 was therefore fed to the cathode side through syringe injection (deionized water was also used at a later point and comparative tests revealed no significant difference between the CV responses obtained with the different liquids). Hydrogen gas was fed to the anode side at 0.03 L min-1 and approximately 1 atm. The cathode of the MEA served as the WE, while the anode side acted as both the CE (counter electrode) and the RE (reference electrode), since it was exposed to hydrogen gas at approximately 1 atm and resembled a standard hydrogen RE. A Bio-Logic potentiostat (VPM3) was used to perform CVs.    Figure A.3.2: Experimental setup for active electrical connectivity measurements, using liquid for membrane hydration.   Figure A.3.3 illustrates the ‘cleaning’ and data collection protocols that were followed. The cleaning procedure, which was used to remove organic and inorganic impurities, was executed by cycling the WE between 0.05 V and 1.2 V, at a scan rate of 500 mV s-1 for a 100 cycles. For data collection, the WE was cycled between 0.05 V and 1 V. By reducing the upper scan limit to 1 V, the formation and reduction of Pt oxides were decreased (compare Figure A.3.3 (a) and (b)). This allowed the electric double layer region to become more distinct and easier to identify, without compromising the adsorption peak size. Data was collected at the following conditions:  Appendices     208   10 cycles at 500 mV s-1  4 cycles at 100 mV s-1  2 cycles at 50 mV s-1  2 cycles at 20 mV s-1   Figure A.3.3: The (a) cleaning and (b) data collection protocols used during active electrical connectivity measurements; Illustrated for CB MPL+GDL at the cathode, Gore CCM, Sigracet 25BC MPL+GDL at the anode, and a cell compression of 827 kPa(g).  The data was corrected for resistance losses through in situ iR-compensation via Nyquist impedance measurements. In the following discussion, the provided CVs represent data collected at a scan rate of 500 mV s-1, while all ECSA measurements were based on the data collected at 20 mV s-1 (averaged over 2 cycles).  To confirm that the technique could be used in the conventional sense, to distinguish different CL activities, different CCMs were initially tested. The results shown in Figure A.3.4, indicate that the CCMs with higher catalyst loading (Gore CCM and JM High CCM), have higher ECSAs than CCMs with lower catalyst loading (JM Low and JM Ultra Low CCM), as expected. ECSA values for the Gore CCM and JM High CCM (75 m2 g-1 and 67 m2 g-1), are also similar, or in close range, to the approximate value of 75 m2 g-1 reported by Gasteiger et al.[184] for a CCM with the same CL loading (0.4 mg cm-2). The resistance further generally appears lower as the electrode Appendices     209  thickness decreases (𝜏𝐺𝑜𝑟𝑒 > 𝜏𝐽𝑀 𝐻𝑖𝑔ℎ > 𝜏𝐽𝑀 𝐿𝑜𝑤 ≈ 𝜏𝐽𝑀 𝑈𝑙𝑡𝑟𝑎 𝐿𝑜𝑤 as indicated in Table A.3.1). As anticipated, the bare PEM (without a CL) shows no electrochemical activity and extremely high resistance.   Figure A.3.4: (a) CV, (b) ECSA and (b) resistance data for the different CCMs with CB MPL+GDL at the cathode, Sigracet 25BC MPL+GDL at the anode, and a cell compression of 827 kPa(g).  Table A.3.1: Thicknesses and Pt loadings of the different CLs investigated through active electrical connectivity measurements. CCM Cathode CL thickness   (μm) Cathode Pt loading (mg cm-2) Gore  ~ 12.5[177] 0.4 JM High ~ 8a 0.4 JM Low  ~ 4a 0.1 JM Ultra Low  (original anode side of JM High CCM) ~ 4a 0.04  aEstimated from SEM images  To determine whether the technique could distinguish between different degrees of MPL|CL interfacial contact, MEAs containing the same CB cathode MPL were evaluated at different compression pressures (Figure A.3.5). The results indicate that ECSA measurements between compressions of 827 kPa(g) and 414 kPa(g) are distinctly different. However, the difference in Appendices     210  ECSA measurements between 621 kPa(g) and 414 kPa(g) were less significant and did not follow the expected trend (ECSAlower compression < ECSAhigher compression). These results served as the first indication that the technique’s sensitivity may be insufficient to effectively distinguish different degrees of contact and active electrical connectivity between the MPL and CL. To investigate this aspect in more detail, reproducibility tests were performed.    Figure A.3.5: (a) CV, (b) ECSA and (b) resistance data for different compression pressures applied to CB MPL+GDL at the cathode, Gore CCM and Sigracet 25BC MPL+GDL at the anode.  Reproducibility tests were performed by preparing five different MEAs, consisting of the same components. CB was used as the MPL on the cathode side in each case. The results show that the ECSA varies with up to 11 m2 g-1, while the resistance varies with up to 6 mΩ (refer to Figure A.3.6).  Appendices     211   Figure A.3.6: (a) CV, (b) ECSA and (b) resistance data for reproducibility tests with CB MPL+GDL at the cathode, Gore CCM, Sigracet 25BC MPL+GDL at the anode, and a cell compression of 827 kPa(g).   After the reproducibility of the technique was established, tests were performed with different layers interfacing the cathode CL. Distinctly different interfacing layers were chosen, to increase the chances of measuring significantly different ECSAs. The interfacing layers included: no interfacing MPL or GDL (bare cathode CCM), no interfacing MPL (bare GDL), CB MPL, and GN MPL. Results in Figure A.3.7 show that the absence of a GDL and MPL, results in the lowest ECSA at 59 m2 g-1, and the highest resistance at 68 mΩ. This fits with the general understanding that a bare CCM would establish less contact and less electrical connections with the FFPs.   Even though the bare GDL, CB MPL and GN MPL are considered to have distinctly different interfaces, their associated differences in the ECSA were not very significant. In fact, the ECSA differences between these interfacing layers (which differ up to 5 m2 g-1)  are smaller than the variation associated with the reproducibility test (varying up to 11 m2 g-1). The same also applies to the resistance measurements which differ with up to 6 mΩ (similar to the variation of the reproducibility test). The tests were replicated and again yielded ECSA differences smaller than the variation associated with reproducibility, while the overall trend in ECSA values also changed. Despite the fact that the technique resulted in a lower ECSA when no interfacing layer was used, ECSA measurements could not be used to distinguish different interfacing layers with Appendices     212  confidence. At this point, it was suspected that the technique’s sensitivity was insufficient to accurately measure differences in contact and active electrical connectivity.   Figure A.3.7: (a) CV, (b) ECSA and (b) resistance data for different cathode CL interfacing layers, with Gore CCM, Sigracet 25BC MPL+GDL at the anode, and a cell compression of 827 kPa(g).  Several attempts were subsequently made to improve the sensitivity of the technique, including:   Employing lower compression for ECSA measurement;  Using thinner or no anode MPLs and GDLs to increase the spacing between compressed MEA components;   Pre-hydrating the PEM to increase proton transport;  Using vapor, instead of liquid, to hydrate the PEM. The motivation behind the latter method was to try and minimize the influence of the interfacing layers’ different liquid permeabilities. The PEM was therefore hydrated from the anode side, by means of humidified hydrogen (100% RH at 50˚C). The cathode side was exposed to nitrogen flowing at 0.3 L min-1 (refer to Figure A.3.8).   Appendices     213   Figure A.3.8: Experimental setup for active electrical connectivity measurements, using vapor, via humidified hydrogen, for membrane hydration.   Unfortunately, none of the abovementioned attempts yielded improved sensitivity or consistent results. Ultimately, the results suggest that the ECSA is a very complex parameter, which may be influenced by a multitude of variables. It is also entirely possible that the active electrical connectivity, itself, does not significantly contribute to the ESCAs of different interfacing layers, and cannot be isolated accurately in this context. The author furthermore remained cognizant of numerous sources of experimental variation that could also have interfered with the technique’s sensitivity:  Differences in sample uniformity of GDL and MPLs;  Variance in MEAs assembled by hand;  Uncontrolled wrinkling and folding of CCMs induced by wetting or hydration.   A.4  Fuel cell protocols: Justification for constant flow control with TP-5 cell hardware  As mentioned, the 2 kW Hydrogenics testing equipment could not realize the small flow rates required for stoichiometric control at low current densities (typically below 500 mA cm2 for λair = 2) with a 5 cm2 active area. The use of constant flow control was further motivated by inherent mass transport limitations experienced with the small cell design. At an intermediate current density (700 mA cm-2), it was found that excessive flooding occurs at low air stoichiometries (or flow rates). However, as the flow rate is increased, voltage behaviour Appendices     214  stabilizes since the additional air flow helps to remove excess liquid water from the cathode (Figure A.4.1). Constant flow rates, resulting in higher stoichiometries at low and intermediate current densities, were therefore specified to lessen flooding in these respective regions.     Figure A.4.1: Voltage stability vs. air stoichiometry for the GN foam MPL in TP-5 cell hardware (Gore CCM with 0.4 mg Pt cm-2 loading on both sides was used in conjunction with Sigracet 25BC as the combined anode MPL+GDL).   Appendices     215  Appendix B: Chapter 4  Appendix B provides additional results on the Tafel analysis presented in Chapter 4.  B.1  Tafel analysis   Kinetic parameters of the freestanding and commercial CB MPLs were calculated from the Tafel plots provided in Figure B.1. Although difficult to confirm based on the number of points used in the analysis, the GN foam MPL’s Tafel slope of 111 mV dec-1, appears as an apparent combined representation of double Tafel slope behavior (with b = 142 mV above 0.84 V and b = 69 mV dec-1 below 0.84 V). This phenomenon is often associated with Pt catalyst, and is typically attributed to changes in the adsorption behavior in different potential regions[185].   Figure B.1.1: Tafel plots determined from Figure 4.6.     Appendices     216  Appendix C:  Chapter 5  Appendix C provides additional experimental data pertaining to the results presented in Chapter 5.  C.1  Additional surface roughness data for MPLs  In addition to the average surface roughness, the root mean square surface roughness was also measured (Table C.1.1). Since the trends in the root mean square surface roughness for the presented case studies generally remained similar to that of the average surface roughness measurements, only the latter were presented in the main text. Three-dimensional images of all the MPLs’ surfaces, in uncompressed and compressed states, are furthermore provided in Figures C.1.1 and C.1.2 for reference purposes.  Table C.1.1: Average and root mean square surface roughness data of the GDL-based MPLs (relative change between uncompressed and compressed states indicated in square brackets).  MPL Average surface roughness, Ra (μm)  Root mean square surface roughness, Rq (μm)  0 kPa(g) 830 kPa(g) 0 kPa(g) 830 kPa(g) CB 18.5 ± 2.1 12.7 ± 2.0 [31% ↓] 25.8 ± 2.7 18.9 ± 2.6 [27% ↓] GR 9.5 ± 1.0 3.4 ± 0.5 [64% ↓] 12.6 ± 1.1 4.9 ± 0.7 [61% ↓] RGO 19.5 ± 1.3 3.1 ± 0.5 [84% ↓] 25.2 ± 1.9 4.9 ± 2.5 [80% ↓] GN 5.4 ± 0.5 1.5 ± 0.3 [71% ↓] 6.7 ± 0.7 2.1 ± 0.4 [69% ↓] GR+CB 14.2 ± 1.4 9.2 ± 1.3 [35% ↓] 19.6 ± 1.6 13.4 ± 1.8 [32% ↓] RGO+CB 7.6 ± 0.7 3.1 ± 0.8 [60% ↓] 9.8 ± 0.9 3.9 ± 1.0 [60% ↓] GN+CB 7.1 ± 1.1 3.6 ± 0.4 [49% ↓] 9.2 ± 1.8 4.7 ± 0.5 [49% ↓]   Appendices     217   Figure C.1.1: Three-dimensional surface profiles of the base MPLs in uncompressed (left) and compressed (right) states.   Appendices     218      Figure C.1.2: Three-dimensional surface profiles of the composite MPLs in uncompressed (left) and compressed (right) states.    Appendices     219  C.2  Experimental through-plane resistance data for MPLs  Figure C.2.1 (a) provides reproducibility data of a CB MPL’s through-plane resistance. The through-plane resistance data over the entire compression range is furthermore provided for all the GDL-based MPLs in Figure C.2.1 (b). The power law functions used to describe the data are provided in Table C.2.1.   Figure C.2.1: (a) Reproducibility in through-plane resistance for the CB MPL; (b) Through-plane resistance data for all the GDL-based MPLs over entire compression range.  Table C.2.1: Power law functions describing the relationship between through-plane resistance and compression for the GDL-based MPLs. MPL Resistance = f(compression) Correlation coefficient (goodness of fit), R2 Relative change upon compression (%) CB 2595x-0.94 1.00 95 GR 1316x-0.81 0.98 93 RGO 17087x-0.87 1.00 93 GN 676x-0.73 0.99 91 GR+CB 1766x-0.92 1.00 95 RGO+CB 9015x-0.92 0.99 93 GN+CB 1279x-0.87 1.00 94 Appendices     220  C.3  Tafel analysis of MPLs  The Tafel plots, constructed for the determination of kinetic parameters, are provided in Figure C.3.1.   Figure C.3.1: Tafel plots determined for the GDL-based MPLs at (a) 100% and (b) 20% cathode RH.              Appendices     221  Appendix D: Chapter 6  Appendix D provides additional experimental data pertaining to the results presented in Chapter 6.  D.1  Additional surface roughness data  The root mean square surface roughness was also measured, and is presented in Table D.1.1. The general trends were again found to be similar to that of the average surface roughness. Three-dimensional images of all the MPLs’ surfaces, in uncompressed and compressed states, are furthermore provided in Figure D.1.1 for reference purposes.   Table D.1.1: Average and root mean square surface roughness data of the CB, CGN and CGN+CB MPLs (relative change between uncompressed and compressed states indicated in square brackets).  MPL Average surface roughness, Ra (μm)  Root mean square surface roughness, Rq (μm)  0 kPa(g) 830 kPa(g) 0 kPa(g) 830 kPa(g) CB 18.5 ± 2.1 12.7 ± 2.0 [31% ↓] 25.8 ± 2.7 18.9 ± 2.6 [27% ↓] CGN 14.5 ± 1.3 5.6 ± 1.4 [61% ↓] 19.2 ± 2.0 7.8 ± 1.7 [59% ↓] CGN+CB 13.3 ± 1.8 5.9 ± 1.0 [56% ↓] 18.4 ± 2.4 8.2 ± 1.2 [55% ↓]          Appendices     222       Figure D.1.1: Three-dimensional surface profile of the CB, CGN and CGN+CB MPLs in uncompressed (left) and compressed (right) states.     Appendices     223  D.2  Experimental through-plane resistance data  The through-plane resistance data for the MPLs used in Chapter 6 are provided in Figure D.2.1.   Figure D.2.1: Through-plane resistance data for the CB, CGN and CGN+CB MPLs.              Appendices     224  D.3 Tafel analysis  The Tafel plots, used for the determination of kinetic parameters, are provided in Figure D.3.1.   Figure D.3.1: Tafel plots determined for JM High and JM Low CCMs with the CB, CGN and CGN+CB MPLs at 100% cathode RH.  D.4 iR-correction  Figure D.4.1 indicates the difference of a polarization curve in its original and iR-corrected form. As mentioned before, iR-correction does not affect the observed trends at all, since the ohmic loss associated with each MEA is very similar. The effect of mass transport limitations does become slightly easier to recognize, however.  JM High JM Low Appendices     225   Figure D.4.1: Polarization results of the CB, CGN and CGN+CB MPLs with JM High CCM at λair = 2 in (a) original and (b) iR-corrected form.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0360782/manifest

Comment

Related Items