UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Platinum ionomer composite films Martens, Isaac 2018

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2019_february_martens_isaac.pdf [ 70.67MB ]
JSON: 24-1.0375790.json
JSON-LD: 24-1.0375790-ld.json
RDF/XML (Pretty): 24-1.0375790-rdf.xml
RDF/JSON: 24-1.0375790-rdf.json
Turtle: 24-1.0375790-turtle.txt
N-Triples: 24-1.0375790-rdf-ntriples.txt
Original Record: 24-1.0375790-source.json
Full Text

Full Text

Platinum Ionomer Composite FilmsbyIsaac MartensB.Sc., University of Calgary, 2011A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Chemistry)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)December 2018© Isaac Martens 2018The following individuals certify that they have read, and recommend to the Faculty of Graduate andPostdoctoral Studies for acceptance, the dissertation entitled: Platinum Ionomer Composite Films, sub-mitted by Isaac Martens, in partial fulfillment of the requirements for the degree of Doctor of Philosophyin Chemistry.Examining Committee:Dan Bizzotto Co-supervisorDavid P. Wilkinson Co-supervisorChadwick Sinclair Supervisory Committee MemberPierre Kennepohl University ExaminerEdouard Asselin University ExaminerAicheng Chen External ExamineriiAbstractThe complex interface between polymer electrolytes and nanostructured electrodes is key for the oper-ation of many electrochemical devices. While the surface science of electrocatalysts have been exten-sively studied using simplified model systems, such as Pt single crystals, the extent to which lessonsfrom these models apply to nanostructured interfaces remains poorly understood. In this work, the in-terface between Pt nanoparticles and solid polymer electrolytes is explored in terms of catalytic activity,morphology and durability. The nucleation and growth of nanoparticles is controlled to produce contigu-ous ultrathin catalyst layers in a solution processable fashion using electroless deposition. The reactivityand structure of the Pt-polymer surface was probed with advanced characterization techniques usingsynchrotron radiation. Direct imaging of the solid-electrolyte-interphase was accomplished with X-rayspectromicroscopy. Degradation of this interface was visible after a simulated aging protocol. Finally,the mechanism of surface oxidation and reduction on Pt nanoparticles was explored with diffraction,moving towards an atomistic understanding of fuel cell electrochemistry inside functional devices underrelevant operational conditions.iiiLay SummaryThe heart of most electrical energy storage devices such as batteries and hydrogen fuel cells are typi-cally composed of blended, porous, wet solids. The chemical reactions of water molecules, electrons,dissolved salts, and plastic binders inside these complex materials are poorly understood, but are tech-nologically important. A better understanding of corrosion inside these materials is necessary to con-struct longer-lived devices. In this thesis, a structural model of the hydrogen fuel cell was fabricated.The corrosion of plastic inside this model was detected by high-magnification X-ray imaging. The cor-rosion chemistry of the materials were altered when blended together in a composite. The corrosion ofplatinum metal was also studied, which is a major problem for low cost fuel cells. X-rays were used tolook through the side of a fuel cell during operation, providing a detailed picture of the rust formed onthe surface of the platinum.ivPrefaceIn accordance with UBC ethics guidelines, this dissertation is an original intellectual product of the author,Isaac Martens. I was the lead investigator for this material, where I was responsible for all major areasof concept formation, data collection and analysis, as well as the majority of manuscript composition.In the course of my work, I have had the good fortune to benefit enormously from many researchers ofexcellent caliber. The price for performing multinational, interdisciplinary and highly collaborative work isa certain dilution of shares. Many ideas and key pieces of experimental data presented here were eitherdirectly contributed by colleagues, or arrived at through a process best described as cross-pollination.Their contributions have been explicitly acknowledged in figure captions whenever appropriate, andconsent for their inclusion has been obtained.Funding and in-kind contributions for this work are gratefully acknowledged from NSERC programs(CGSM, CGSD, MSFSS, Strategic Partnership, and Discovery), UBC (4YF), the Catalysis Research forPolymer Electrolyte Fuel Cells Network, Johnson-Matthey Fuel Cells, McMaster University, the CanadianCentre for Electron Microscopy, the Canadian Light Source, and the European Synchrotron RadiationFacility.Material included in the first experimental chapter has been published in two full papers:• Martens, I.; Pinaud, B.A.; Baxter, L.; Wilkinson, D.P; Bizzotto, D.; “Controlling Nanoparticle Inter-connectivity in Thin-Film PlatinumCatalyst Layers,” J. Phys. Chem. C. 2016, 120(38):21364–21372I.M. was responsible for literature survey, conducting all experiments, and manuscript preparation.B.A.P. and L.B. assisted with the electrochemistry and synthesis, respectively. D.P.W. and D.B. su-pervised the work.• Ingle, N.; Sode, A.; Martens, I.; Gyenge, E.; Wilkinson, D.P.; Bizzotto, D. “Synthesis and Charac-terization of Diverse Pt Nanostructures in Nafion”. Langmuir. 2014, 30(7):1871–1879.N.I. and A.S. performed the experiments. I.M. assisted with data analysis. All authors contributed tomanuscript composition.vPrefaceMaterial included in the second experimental chapter will be published under the authors Martens,I.; Melo, L.G.A.; Wilkinson, D.P.; Bizzotto, D.; Hitchcock, A.P. “Understanding X-ray induced beam dam-age in perfluorosulfonic acid using correlative microscopy” In preparationI.M., L.M. and A.P.H. conducted the literature survey, data acquisition and data analysis. A.P.H., D.B.and D.P.W. supervised the work. All authors contributed to the manuscript, with I.M. contributing >80%.Material included in third experimental chapter will be published under the authors• Martens, I.; Drnec, J.; Chattot, R.; Rasola, M.; Blanco, M.V.; Honkimäki, V.; Wilkinson, D.P.; Biz-zotto, D. “Probing the Dynamics of Platinum Surface Oxides in Fuel Cell Catalyst Layers using InSitu X-ray Diffraction” submitted to JACS (Manuscript ID: ja-2018-078487)I.M., J.D. and M.B. performed the X-ray experiments. R.C. contributed the electron microscopy. M.R.assisted with the cell design. I.M. wrote the manuscript. V.H., D.B. and D.P.W supervised the work.• Martens, I.; Pusa, J.; Rasola, M.; Honkimäki, V.; Drnec J.; Wilkinson, D.P.; Bizzotto, D. “X-rayTransparent PEM Fuel Cell Design for In Situ WAXS/SAXS Tomography” In preparationI.M. collected and analyzed the data, and prepared the manuscript. J.P. assisted with test station oper-ation. M.R. contributed to the cell design. V.H., J.D., D.P.W. and D.B supervised the work.viContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vContents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiAbbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxivAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxviiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxx1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 PEM fuel cell construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Ionomer membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Oxygen reduction and Pt electrochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.4 Thesis objectives and scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Experimental Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.1 Reagents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.2 Electroless deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.2.1 Platinum (II) tetramine nitrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.2.2 Membrane electrode fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.3 Half-cell electrochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13viiContents2.4 Electron microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.4.1 Scanning electron microscopy (SEM) . . . . . . . . . . . . . . . . . . . . . . . . . 142.4.2 Energy dispersive X-ray spectroscopy (EDX) . . . . . . . . . . . . . . . . . . . . . 152.4.3 Transmission electron microscopy (TEM) . . . . . . . . . . . . . . . . . . . . . . . 152.4.4 Electron energy loss spectrometry (EELS) . . . . . . . . . . . . . . . . . . . . . . 162.5 X-ray photoelectron spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172.6 Inductively coupled plasma optical emission spectroscopy (ICP-OES) . . . . . . . . . . 172.7 UV-Visible spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182.8 In plane conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192.9 195Pt Nuclear magnetic resonance (NMR) spectroscopy . . . . . . . . . . . . . . . . . . 192.10 Confocal laser scanning profilometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.11 Atomic force microscopy (AFM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.11.1 Current sensing AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222.12 Surface enhanced Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.13 Fluorescence microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.14 Infrared spectromicroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.15 X-ray microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.15.1 Near Edge X-ray Absorption Spectroscopy (NEXAFS) . . . . . . . . . . . . . . . 292.15.2 Synchrotron Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.15.3 Scanning transmission X-ray microscopy (STXM) . . . . . . . . . . . . . . . . . . 302.16 X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 Electrolessly Deposited Pt Catalyst Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.1 Electroless deposition of Pt films on Nafion membranes . . . . . . . . . . . . . . . . . . . 363.1.1 Overview of electroless deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.1.2 Literature review and objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.1.3 Platinum and borohydride solution equilibria . . . . . . . . . . . . . . . . . . . . . 413.1.4 Electroless deposition of Pt on Nafion . . . . . . . . . . . . . . . . . . . . . . . . . 443.2 Controlling nucleation and growth during the deposition . . . . . . . . . . . . . . . . . . . 443.3 Following electroless deposition with in situ X-ray diffraction . . . . . . . . . . . . . . . . 523.4 Controlling nanoparticle interconnectivity in electroless films . . . . . . . . . . . . . . . . 573.4.1 Topography & in-plane conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . 58viiiContents3.4.2 Conductive AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.4.3 XPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653.4.4 Electrochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683.5 Optical profilometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703.6 Ionomer adsorption effects on electroless Pt . . . . . . . . . . . . . . . . . . . . . . . . . . 723.7 Electrochemical stability of electroless Pt . . . . . . . . . . . . . . . . . . . . . . . . . . . . 763.8 Fuel cell testing of electroless Pt layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773.9 Electroless deposition of Pt nanowires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803.10 Electrolessly deposited Pt alloys and their hydrogen oxidation activity . . . . . . . . . . 843.11 Alternative applications: optical reflectivity and sensing . . . . . . . . . . . . . . . . . . . 943.12 Patterned electroless deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1013.13 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044 Reactivity of the Pt-PFSA Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1064.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1064.2 STXM imaging of electrolessly deposited films . . . . . . . . . . . . . . . . . . . . . . . . 1074.3 Understanding X-ray induced beam damage in perfluorosulfonic acid using correlativemicroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1114.4 STXM Imaging of Pt nanoparticles inside PFSA . . . . . . . . . . . . . . . . . . . . . . . . 1244.4.1 AFM and FTIR of Pt-PFSA dispersions . . . . . . . . . . . . . . . . . . . . . . . . 1274.5 STXM imaging of a stress-tested electroless catalyst film . . . . . . . . . . . . . . . . . . 1284.6 SERS of Pt-PFSA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1314.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1345 In situ High Energy X-ray Diffraction of Pt Catalyst Layers . . . . . . . . . . . . . . . . . . 1355.1 Introduction to Pt oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1355.2 Interpreting XRD patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1385.2.1 The synchrotron advantage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1395.2.2 Background subtraction strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . 1405.2.2.1 Polynomial background correction . . . . . . . . . . . . . . . . . . . . . . 1415.2.2.2 Differential Subtraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1415.2.2.3 Resonant diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1425.2.3 Cell design and measurement geometry . . . . . . . . . . . . . . . . . . . . . . . . 145ixContents5.2.4 Instrumental broadening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1495.2.5 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.2.5.1 XRD MEA sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1535.3 In situ diffraction of Pt oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1565.3.0.1 Rietveld refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1585.3.1 Equilibrium conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1605.3.2 Cyclic voltammetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1655.3.3 Pt oxidation/reduction kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1675.3.4 Pair distribution function analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1735.3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1756 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1776.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1776.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181xList of Tables3.1 Selected properties of electrolessly deposited Pt catalyst layers for selected literaturereferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.2 Optimization parameters used in developing electroless Pt catalyst thin films. . . . . . . 483.3 In-plane four-point probe electrical resistivity of Pt-Nafion nanoparticle thin films recordedin the wet and dry states. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653.4 Comparison of electrolessly deposited Pt ORR activity with competing catalyst layer tech-nologies (dispersed on carbon rotating disc electrodes, O2-saturated 0.1-0.5 M HClO4,0.9 V vs RHE). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685.1 Structural parameters refined from XRD scans collected at 0.6V vs RHE. . . . . . . . . 1605.2 First order rate constants (s-1) determined for Pt place-exchange. . . . . . . . . . . . . . 169xiList of Figures1.1 Diagram of a conventional hydrogen PEM FC.1 . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Membrane electrode assembly construction. Exploded perspective diagram (left) andscanning electron microscopy image of cross section (right). . . . . . . . . . . . . . . . . 41.3 Chemical structure of Nafion, a well-known PFSA (left). Proposed cluster-morphologyproposed for PFSA (right). The blue portion represents the hydrated fraction, while thepolymer backbone fills the yellow areas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.4 Cyclic voltammogram of Pt catalyst layer (black, top trace) and Pt bead electrode (red,bottom trace). CV collected at 200 mV/s, 0.2M H2SO4 for catalyst layer, 50mV/s, 3 MH2SO4 for bead electrode, Ar purged. Area integrated for surface area calculations ishighlighted in blue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.5 Linear sweep voltammogram of Pt in the presence of 1 atm molecular oxygen, recordedin a half cell. Dashed lines highlight the ORR current at 0.9 V for this catalyst layer. Curverecorded at 20 mV/s using 0.5 M HClO4 at 294 K and iR corrected. . . . . . . . . . . . 82.1 Electrochemical half-cell design used for testing ORR activity of electroless films. . . . 142.2 Conceptual diagram of atomic force microscope. Adapted from reference2 . . . . . . . 212.3 Current sensing AFM schematic. Conductive coating (black) on cantilever and tip allowscurrent to flow between substrate and AFM tip with the application of a suitable bias. . 222.4 Typical configuration of FTIRmicroscope. Adapted fromBruker Hyperion 2000 instrumentmanual. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252.5 Infrared flux intensity at the detector using a globar source as a function of microscopecondensing aperture3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.6 Transmission FTIR absorbance spectra of SiN window sample holder (2 cm-1 resolution).Inset: higher magnification of the baseline region . . . . . . . . . . . . . . . . . . . . . . . 28xiiList of Figures2.7 FTIR spectra of X-ray damaged Nafion sample obtained using synchrotron and conven-tional sources. Synchrotron dataset collected by Scott Rosendahl (CLS). . . . . . . . . 282.8 Linear attenuation of X-ray photons in water. Calculated from tabulated data4 . . . . . 302.9 Block diagram of STXM instrumentation showing configuration of source and beamline(a) and detail of ZP and sample stage (b). Taken from Ade and Hitchcock, reprinted withpermission.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312.10 Bragg’s law, as applied to XRD (a) Diagram of a transmission mode powder XRD exper-iment (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333.1 Cartoon schematic of homogeneous and heterogeneous electron transfer mechanismsfor electroless deposition. M/M+ and RA/RA+ represent the reduced and oxidized formsof metal and reducing agent species. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.2 Free energy diagram of nanoparticle nucleation and growth predicted by classical nucle-ation theory (a). The classical LaMer model of nucleation and growth of nanoparticlesover time, illustrating different concentrations of depositing species. Taken from6, pub-lished by Royal Society of Chemistry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373.3 Acrylic cells used for electroless deposition showing open face of solution chambers (left).With Nafion sealed between the two halves, movement of bulk solution between the twois restricted (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443.4 Idealized target structure for electroless deposition of Pt onto ionomer membranes . . . 453.5 Pt concentration in each solution during electroless deposition. . . . . . . . . . . . . . . 453.6 Cross sectional BSE-SEM micrographs of Pt catalyst layers showing progress duringoptimization. Early attempts (left) have diffuse, disconnected Pt particles of large sizethroughout the cross section of the polymer. After refining the deposition process, densePt films are reliably formed on the Nafion surface (right). . . . . . . . . . . . . . . . . . . . 493.7 Appearance of the optimized electroless Pt deposition. Variation in brightness is causedby the mirror finish of the Pt surface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493.8 High resolution BSE-SEM of the Pt-Nafion near-monolayer thin-film in cross-section (top),from top-down perspective (middle), and a brightfield TEM cross-section microtomed ata slightly inclined angle (bottom). TEM image courtesy of Gianluigi Botton and CarmenAndrei. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50xiiiList of Figures3.9 Particle size distribution of the three Pt films. 50-100 particles were counted for eachsample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513.10 High resolution TEM images of the Pt nanoparticles show both extensive aggregationand coalescence of very small Pt grains (left), as well as larger single crystal particles.Images collected by Carmen Andrei. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523.11 Brightfield TEM (top row) and BSE-SEM (bottom row) cross-sections of the sub-monolayer,near-monolayer and multilayer samples. TEM images collected by Carmen Andrei. . . 533.12 Schematic of electroless deposition cell during XRD reaction monitoring. Red dots in-dicate where the diffraction signal was monitored in cross sectional view (a). Swellingof the membrane changes the position of the Pt coated surface during the reaction (b).The coloured bars (numbered one through four) show the different positions of the beam.These drawings are not to scale, the curvature of the membrane and the width of theX-ray beams have been exaggerated here. . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.13 XRD of the 311 and 222 reflections collected during the electroless deposition (a). Fullpowder patterns (b). These transients were recorded in the magenta and red regionslabelled in Fig. 3.12B, respectively. Red arrows indicate the onset of reflections. Bluearrows track the evolution in apparent lattice parameter over the course of the reaction. 553.14 Beam damage from X-ray exposureduring electroless deposition. SE-SEM image of thedamaged film (a) in top-down perspective. Pt coated material is on the right hand side.Photograph of the Pt film on Nafion membrane after removal from cell, showing X-raydamage.(b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563.15 Contact mode AFM of the film surface discerns small clusters of Pt particles. Bare Nafionfilm on one of the bubble defects (top row) has smooth rolling surface features. On thePt film (bottom row), small clusters of particles and significant roughening are seen. Leftcolumn: tip deflection. Right column: topography. Scale bar is 400nm. . . . . . . . . . . 593.16 AFM phase images of Pt-Nafion during dehydration.. Each frame 500 by 500 nm. Eachframe was sequentially acquired over five minutes, for a total of 30 minutes of drying time.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603.17 Typical CS-AFM images of two different Pt-Nafion films (top and bottom) show large prob-lems with tip contamination, visible as large dark regions in the conductivity map. To-pography (left), tapping amplitude (middle), and CS conductivity (right) images recordedsimultaneously. Scales are identical to those in Figure 3.15. . . . . . . . . . . . . . . . . 62xivList of Figures3.18 Schematic of CS-AFM circuit optimized for imaging Pt-Nafion catalyst layer. . . . . . . . 623.19 Optical micrograph of CS-AFM tip positioned near the edge of the sputtered gold coun-terelectrode contact (a). The AFM tip is positioned at the end of the triangular cantiliver.Red box indicates region scanned by AFM. Column averaged conductivity (b) of imageover Au-Pt interface. The actual conductivity map is shown in (c). . . . . . . . . . . . . 633.20 Current-sensing AFM of the Pt-Nafion surface. Topography channel shows a bubble fea-ture on the film surface (A). Conductivity map of same area (B) shows no conductivityinside the bubble region. Imaging the Pt at higher magnification and current sensitivityshows most Pt particles are interconnected, forming dense, long-range conductive net-works (C). (A) & (B) are 1x1 μm field of view, and (C) is 400x400 nm. . . . . . . . . . . . 643.21 Survey X-ray photoelectron spectrum of a Pt-Nafion film showing peak assignments. In-set: Fermi level region (bottom left) and Pt 4f peaks (top right). . . . . . . . . . . . . . . . 663.22 XPS of the Pt 4f doublet for electroless films of the standard recipe Pt (black) and of athicker PtRu layer (red) after calibration vs Mo signal. . . . . . . . . . . . . . . . . . . . . 673.23 Cyclic voltammogram of the near-monolayer Pt-Nafion film in Ar atmosphere, 50 mV/s(left). Oxygen reduction reaction (ORR) Tafel plot measured inO2, 1 atm, 20 mV/s (right).Both curves were recorded using 0.5 M HClO4 at 294 K and IR corrected. . . . . . . . 693.24 Profilometry image of an electrolessly deposited Pt film. The dark region in the upper lefthas been hot pressed against carbon and delaminated, while the bright region has not.White light full color brightfield (left) total reflected laser intensity (middle) and topographymap (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713.25 Profilometry image of an electrolessly deposited Pt film after hot-pressing. Top row:brightfield finder frames at 5x and 10x objectives. Bottom row: True-color brightfield(left) total reflected intensity (middle) and topography (right).Red dashed lines indicatethe magnified regions of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 723.26 Profilometry image of an electrolessly deposited Pt film. Total intensity map (left) andtopography (right) for 50x (top) and 100x (bottom) objectives. Data collected over centreof region shown in Figure 3.25. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.27 Extracted optical profilometry linescan from a crack in electrolessly deposited Pt film ob-served in Figure 3.26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73xvList of Figures3.28 Topography of electroless deposited film after delamination fromMPL, away from crackedregions in Figure 3.26. Image was flattened with straight line in X and Y dimensions toenhance contrast. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 743.29 CV of electroless catalyst layer after sputtering 50 nm Au film onto the Pt surface (left).SEM image of sputtered gold thin film on top of electroless Pt layer collected at obliqueangle (middle). Top-down optical microscope image of Au film after equilibrating sampleat ambient humidity (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753.30 ORR activity of electrolessly deposited catalyst layer over the course of catalyst supportAST showing stable performance after initial cleaning. ORR activity measured at 0.9 V,after cycling at 500 mV/s between 1-1.5 V. . . . . . . . . . . . . . . . . . . . . . . . . . . . 773.31 CV of the electroless layers change slowly over 2000 cycles of the catalyst AST (0.6-1.0V, 50 mV/s) in half cell, with black arrows indicating changes due to adsorption of theionomer over time. Each color represents 200 CV cycles. Measured in half cell with 18mm diameter working electrode in 1 M HClO4. . . . . . . . . . . . . . . . . . . . . . . . . . 783.32 Fuel cell testing of electroless catalyst layer. 5 cm2, 100% RH, 80ºC, H2/air. . . . . . . 793.33 Polarization curve for optimized electroless film in 5 cm2 fuel cell. 100% RH, 80ºC, H2/air.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 793.34 Brightfield TEM cross-section of Pt nanowires electrolessly deposited in Nafion mem-brane. (a) A bundle of nanowires buried beneath the surface. (b) and (c) higher mag-nification images showing the lattice fringes characteristic of a Pt single crystal. Imagescollected by Brad Ross. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 813.35 Representative brightfield TEM cross section of electroless deposited membrane con-taining nanowires. Low magnification image showing the homogeneity of the layer (left).Medium magnification image showing the morphology near the membrane surface (mid-dle). High magnification image showing the morphology of the individual nanorod clusters(right). Images courtesy of Gianluigi Botton. . . . . . . . . . . . . . . . . . . . . . . . . . . 823.36 Electron diffraction image of the nanorod sample (left). The lookup table is logarithmic toenhance visual contrast of the rings. Integrated radial profile of the diffractogram (right).Image courtesy of Gianluigi Botton. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 823.37 CV of nanorod sample in half cell, purged with Ar gas, at variable sweep rates (left). Theintegral of the H-UPD desorption feature was plotted as a function of scan rate (right).298K, 0.5M HClO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83xviList of Figures3.38 Cartoon schematic of bifunctional mechanism for CO oxidation on a PtRu surface. . . . 853.39 Brightfield TEM of electrolessly deposited PtRu nanoparticles (left) and EDX spectrumof PtRu film showing peak fits for the Pt M edge and Ru L edge (right). Red filled re-gions correspond to the experimental data, with dark and light blue lines representing thebackground and modelled elemental transitions respectively. . . . . . . . . . . . . . . . 863.40 Low resolution HAADF (bottom) and extracted EELS intensity maps (top row) of a flakefrom the surface of the PtRu catalyst film. Data collected by Andrea Korinek. . . . . . . 873.41 High resolution HAADF (bottom) and EELS images (top row) of a flake from the surface ofthe PtRu catalyst film. Arrows indicate identical locations as a guide to the eye. The boxindicates the region of interest mapped with EELS. Data collected by Andrea Korinek. . 883.42 High resolution HAADF (bottom) and EELS images (top row) of a flake from the surfaceof the PtRu catalyst film. Data collected by Andrea Korinek. . . . . . . . . . . . . . . . . . 893.43 CV of electroless deposited Pt and PtRu films under nitrogen atmosphere. 250 mV/s, 295K, 0.5 M HClO4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 903.44 CO stripping voltammetry for the electrolessly deposited Pt and PtRu catalyst layers. 20mV/s sweep rate, 295 K, 0.5 M HClO4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913.45 HOR CV of Pt and PtRu films in the presence of H2 gas with 100 ppm CO. 295 K, 0.5 MHClO4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 923.46 Steady state HOR chronoamperometry of the Pt and PtRu films after introduction of 100ppm CO into H2 stream. The potential of both electrodes was held at 0.3 V. . . . . . . . 933.47 Images of platinum thin films under ambient conditions after depositing sub-monolayer,near-monolayer, and multilayer Pt nanoparticle films, demonstrating the evolution of themirrored surface. ICP-OES and electrochemical surface area measurements from eachfilm were performed in triplicate using a half cell, bars represent average +/- range. . . 953.48 Transmission UV-Visible spectra of hydrated and dehydrated near-monolayer Pt nanopar-ticle film demonstrating environmentally switchable optical density. . . . . . . . . . . . . 963.49 Specular angular reflectance curves of the high, medium and low loading Pt films usings-polarized (left) and p-polarized (right) light at 650 nm. Data courtesy of David Troianiand Agilent Technologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 973.50 S-polarized absolute reflectivity spectra at a 65 degree angle of incidence. The Pt stan-dard was calculated from literature data. Data courtesy of David Troiani and Agilent Tech-nologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98xviiList of Figures3.51 Actuation of electroless Pt as an IPMC, showing cartoon schematic of mechanism, deviceconstruction, and actuation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1003.52 Photolithography of electroless layers. Impregnation of membrane with Ag+ and exposureto shortwave UV (A). Ag seed nuclei are left exclusively where the mask did not coversample (B). Rinsing away the unreacted Ag+ ions (C). Galvanic displacement of silvermetal with Pt2+ ions (D). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1013.53 Optical spectroscopy of Nafion films before and after photonucleation of Ag seeds, andafter conversion to Pt shell, Ag core nanoparticles. . . . . . . . . . . . . . . . . . . . . . . 1023.54 Low resolution photopatterning of Ag nanoparticles inmembrane leaves yellowAg nanopar-ticles (left). Brightfield TEM image of the Ag nanoparticles (right). . . . . . . . . . . . . . 1023.55 Electroless Pt film onmembrane embossed with DVD corrugation showing structural color(A). SEM image of the corrugation showed Pt particles embedded into the surface (B).AFM of the Pt film highlights the corrugation banding (C). Linescan from AFM image (D).. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1033.56 Different analytical techniques allowed characterization of electronic conductivity in thePt film over a large range of length scales. . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044.1 Multicomponent fit of STXM C 1s stack of embedded electroless catalyst layer cross sec-tion. Images collected by Adam Hitchcock. . . . . . . . . . . . . . . . . . . . . . . . . . . . 1094.2 STXM images of the electroless catalyst layer at the C 1s and F 1s edges using threecomponent fit, as well as C 1s spectra. Images collected by Adam Hitchcock. . . . . . . 1104.3 Normalized C 1s and F 1s spectra of the damaged and undamaged PFSA as measuredin Figure 4.2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1114.4 PFSA spin coated film damaged by 320 eV soft X-rays in three areas: 4, 5 and 40 MGyimaged by (A) fluorescence microscopy (B) STXM OD image at 292.4 eV, energy sen-sitive to the C 1s *C-F bonds and (C) AFM amplitude map. All three techniques clearlydistinguish the damaged and undamaged areas. STXM image collected by Lis Melo. . 113xviiiList of Figures4.5 High resolution images of the PFSA damaged areas of 40 (top row) and 4 MGy (bottomrow) (A, B, C) show STXM OD image at 292.4 eV, AFM topography and phase contrastimages, respectively, of the 40 MGy damaged area, (D, E, F) show STXM OD imageat 292.4 eV, AFM topography and phase contrast images, respectively, of the 4 MGydamaged area. Blue and green dashed lines correspond to linescans shown in Figure4.6. STXM images collected by Lis Melo. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1144.6 Extracted linescans of the: 40 MGy damaged area using STXM 292.4 eV image shownin Fig.4.5A and AFM topography image of the same region shown in Fig. 4.5B (A and B,respectively), 4 MGy damaged area using STXM 292.4 eV image shown in Fig.4.5D andAFM topography image of the same region shown in Fig. 4.5E (C and D, respectively). 1144.7 NEXAFS spectra of undamaged and X-ray damaged PFSA. (A) C 1s spectra, (B) O 1sand (C) F 1s spectra, for the undamaged and 5, 4, and 40 MGy dosed areas (Fig. 1). (A)also contains an expanded plot of the 283-288 eV region. Data collected by Lis Melo. . 1164.8 FTIR Absorbance spectra of reference and X-ray exposed PFSA (A). Differential ab-sorbance FTIR spectra of dosed vs undosed PFSA (B). Magnification of FTIR plot inC=O region (C). Magnification of FTIR plot in C-O region, with each spectra offset forclarity. (D) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1174.9 Quantitation of the STXM (A and B) and FTIR (C) spectra. (A) shows the quantitationof the change for the whole spectra at the 1s edge while (B) shows the difference in ODof each specified peak. The error bars are the calculated standard error attributed tosystematic instrumentation errors. (c) peak integrals of FTIR absorbance spectra. Alldata was normalized to the undamaged area to plot fractional differences. . . . . . . . . 1194.10 Comparison of the dose-damage response with C 1s and F 1s STXM NEXAFS spectraSTXM and FTIR peak integral fractional changes. . . . . . . . . . . . . . . . . . . . . . . . 1204.11 Reference Nafion FTIR absorbance spectra showing hydration dependence (top, adaptedfrom reference7). FTIR spectra of undamaged Nafion ultrathin film after STXM measure-ment and storage under ambient conditions (bottom, 2 cm-1). . . . . . . . . . . . . . . . 1234.12 Backscatter SEM image of small (A) and large Pt nanoparticles (B) dispersed inside aPFSA membrane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124xixList of Figures4.13 STXM and AFM of the large nanoparticle STXM sample shows divots and raised bumps,indicating the surface distorted near Pt particles during microtomy. STXM frame C 1s292.6 eV, 500 nm scale bar (A). AFM tapping amplitude map (B). AFM topography map,with extracted linescans over different features (C). AFM frame is 5 μm x 5 μm x 30 nm.The STXM and AFM images were collected over nearby, but not identical regions. . . . 1254.14 Deconvoluting PFSA damage on Pt nanoparticles using STXM. In the overlay image, theundamaged PFSA spectra is green, and the damaged PFSA is red. . . . . . . . . . . . . 1254.15 Damage check frame collected at 292.6 eV after the stack in Fig. 4.14 shows extremedamage to the sample, visible as a dark square where the images were collected. Lines-cans over the square show the damage to the spectral features. . . . . . . . . . . . . . . 1264.16 AFM topography of the Nafion film (left), and Pt-Nafion (middle) over regions given a doseof 10 MGy. Extracted line profiles (right) demonstrating ablation of the sample after X-rayexposure. Both AFM frames are 5x5 μm and have the same topographic scale. Blue linesindicate the regions used for linescan analysis. . . . . . . . . . . . . . . . . . . . . . . . . 1274.17 Differential absorbance FTIR spectra on platinized membrane for three different doseswith 320 eV X-rays. Comparative integrations of the Pt-loaded and Pt-free membrane forthe three different regions previously measured. . . . . . . . . . . . . . . . . . . . . . . . 1294.18 STXM images of stress tested electroless catalyst layer. (A) C 1s stack map of the PFSA,OD292-278 eV. (B) Enlarged region of (A), same image. (C) C 1s stack map of C=C bonds,OD284.9-280 eV. (D) F 1s stack map of C-F bonds, OD689-680 eV. . . . . . . . . . . . . . . 1304.19 Reaction schematic of SERS substrate fabrication . . . . . . . . . . . . . . . . . . . . . . 1314.20 Visually tracking the deposition of gold onto Pt film. Initial electroless Pt film (A). The samefilm after five minutes of immersion in the Au bath (B) After 60 minutes of Au deposition(C). All images are the same scale, and of the same sample, but at different locations. 1314.21 Gold to platinum ratio of the SERS substrates during the deposition. (A) SEM image ofthe optimized SERS film after 30 minutes showing highly roughened surface (B). . . . . 1324.22 Raman spectra of regular electrolessly deposited Pt (left) and after surface enhancementfrom Au coating (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1335.1 Pt oxide formation on 111 surface. CV (left) of Pt/C (left), with coloured bars indicatingthe surface chemistry of the Pt at different potentials (right). . . . . . . . . . . . . . . . . . 136xxList of Figures5.2 Raw powder diffractogram showing the subtracted polynomial background (red). Datawas analyzed over the region bounded by the vertical red lines. The weighted residualfor the fitted curve is shown below. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1415.3 Raw (red) and differentially subtracted (blue) diffractograms showing contribution of vari-ous non-Pt cell components to overall signal (A). Signal-to-noise and background is mas-sively improved by correcting for unwanted scattering. Pairwise differential diffractogramsfrom four spots (0 to 3) on the GDE sample (B). Peaks at scattering vectors less than 2.8correspond to the cell wall and graphite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1425.4 Calculated total atomic scattering factors for Pt4 (a). Energy scan over the Pt K-edgeshowing intensity oscillation of the 220 reflection (b). The peak closest to the Pt K-edgeis coloured black as a guide for the eye. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1435.5 Normalized powder diffractograms for Pt catalyst measured on (red) and off (black) theK-edge, showing the change in relative intensities for the 111, 220 and 200 reflectionsresulting from anomalous contrast variation. Note that the red curve has been offset inthe horizontal direction for clarity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1445.6 X-ray transmission liquid cell design, exploded view. CAD model drawn by Miika Rasola. 1455.7 Cell design configured for through-plane and in-plane geometries. The black disc is theGDE sample. The white connectors visible on the right side are ports for electrolytecirculation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1485.8 Scanning the beam position vertically through the catalyst layer for alignment of a catalystlayer sample. The Pt 220 reflection is shown here, with a black line indicating optimalalignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505.9 Real-space geometry and inset diffractogram of highly crystalline sample, free of broad-ening artifacts (a) and with instrumental broadening from sample thickness (b). . . . . 1515.10 Self-absorption inside the sample can alter peak shape of diffracting rays . . . . . . . . 1525.11 The large area XRD detector can be either inserted or moved out of the beam path tocollect WAXS (left) or SAXS (right) data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1525.12 Raw diffraction detector image of the Pt electrode (A) and azimuthal texture map of thePt 220 reflection showing homogeneous intensity distribution (B). . . . . . . . . . . . . . 1545.13 Cross-section SEM of the GDE showing MPL and GDL regions. . . . . . . . . . . . . . . 1545.14 Cross-section SEM images of the Pt/C catalyst layer sprayed on the surface of the MPL.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155xxiList of Figures5.15 High resolution SEM image of the GDE catalyst film showing Pt nanoparticles dispersedon carbon. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1555.16 TEM images of Pt catalyst from before (A) and after (B) cycling. Images collected byRaphael Chattot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1565.17 Powder diffractogram of Pt catalyst layer with Rietveld fit (A). Blue points, green line,and red line correspond to experimental data, fitted Pt pattern, and residual respectively.Black ticks indicate the calculated positions of Pt reflections. Detail of the Pt 220 reflectioncorresponding to oxidized and reduced catalyst surface is in (B). Oxidation of the Pt sur-face decreases the diffraction intensity, broadens the peak, and shifts the peak positiontowards lower angle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1575.18 Differential signal showing powder diffractogram of oxidized Pt catalyst subtracted fromreduced material. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1585.19 Structural parameters extracted from diffractograms collected at each potential step. Lat-tice parameter (A), particle size (B), peak area (C) of the modeled Pt phase. Arrows in-dicate the data corresponding to forward and reverse steps. Overlaid is the CV collectedat 5 mV/s for reference. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1615.20 Comparison of the diffractograms measured in-situ after the CV experiment shown in Fig.3, with several as-received HiSPEC catalysts of varying particle size. The intensity ofeach diffractogram has been normalized. . . . . . . . . . . . . . . . . . . . . . . . . . . . 1625.21 Particle size distributions of Pt catalyst from TEM images before (black) and after (red)electrochemical treatment, with normal distribution fits. . . . . . . . . . . . . . . . . . . . 1635.22 Peak area collected during cyclic voltammetry at various scan rates (A). The ’step’ scancorresponds to the potential step experiment shown in Fig. 5.19. Lattice parameter (redpoints), peak area (black line) and current (blue line) from 5 mV/s CV (B). The latticeparameter has been smoothed with a three point moving average (red line). . . . . . . 1655.23 Structural evolution of Pt catalyst during potential steps. Potential step profile over time(A). Lattice parameter (B), Particle size (C), Peak area (D), as obtained from Rietveldrefinement. The potential step was made at the time point of zero from a resting potentialof 0.48 V. (Black 0.98 V, red 1.18 V, blue 1.28 V). The error bars represent the refinementcovariance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1685.24 Curve fitting on particle size and scale factor for Pt oxidation steps to 1.18 (A) and 1.28V (B), as shown in Figure 5.23. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169xxiiList of Figures5.25 RC time constant determination from current decay after potential step . . . . . . . . . . 1705.26 Curve fitting of lattice parameter and peak area for reduction of Pt oxide after conditioningat 1.28 V, as shown in Figure 5.23. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1715.27 Transient response of lattice parameter and peak area for the reduction of Pt oxide duringa potential step from 1.3 to 0.5 V. Solid lines show first order exponential fits. . . . . . . 1725.28 Pair distribution function schematic showing radial scattering function of coordination en-vironment around Pt atom (A). G(r) for oxidized Pt nanoparticles. (B) . . . . . . . . . . . 1745.29 PDF G(r)s for the oxidized and reduced Pt catalyst (red and black, respectively) as wellsimulated patterns from single crystal structures of PtO2 and Pt metal (green and blue).The simulated PDFs have been vertically offset for visual clarity. . . . . . . . . . . . . . 174xxiiiAbbreviationsALS Advanced Light SourceAFM Atomic Force MicroscopyALD Atomic Layer DepositionAST Accelerated Stress TestATR Attenuated Total ReflectanceBSE Backscattered ElectronCLS Canadian Light SourceCCM Catalyst Coated MembraneCL Catalyst LayerCNC Computer Numerical ControlCLSM Confocal Scanning Laser MicroscopyCE Counter ElectrodeCS-AFM Current Sensing Atomic Force MicroscopyCV Cyclic VoltammetryDOE Department of EnergyECSA ElectroChemical Surface AreaEMF Electromagnetic FieldEELS Electron Energy Loss SpectroscopyEDX Energy Dispersive X-ray spectroscopyxxivAbbreviationsESRF European Synchrotron Radiation FacilityEXAFS Extended X-ray Absorption Fine StructureFFT Fast Fourier TransformFTIR Fourier Transform InfraRedFC Fuel CellFWHM Full Width Half MaximumGDE Gas Diffusion ElectrodeGDL Gas Diffusion LayerHAADF High Angle Annular Dark Field MicroscopyHOR Hydrogen Oxidation ReactionICP-OES Inductively Coupled Plasma Optical Emission SpectroscopyIPMC Ionic Polymer Metal CompositesLSV Linear Sweep VoltammetryLUMO Lowest Unoccupied Molecular OrbitalMEA Membrane Electrode AssemblyMCT Mercury Cadmium TellurideMEMS Microelectromechanical SystemsMPL MicroPorous LayerMAD Multiwavelength Anomalous DispersionNIST National Institute of Standards and TechnologyNEXAFS Near Edge X-ray Absorption Fine StructureNMR Nuclear Magnetic ResonanceORR Oxygen Reduction ReactionxxvAbbreviationsPDF Pair Distribution FunctionPFSA Perfluorosulfonic AcidPE Place ExchangePGM Platinum Group MetalPEEK Polyether Ether KetonePMMA PolyMethylMethAcrylatePTFE PolyTetraFluoroEthylenePEM Proton Exchange MembranePITM Pt In The MembraneROS Reactive Oxygen SpeciesRE Reference ElectrodeRC Resistor CapacitorRHE Reversible Hydrogen ElectrodeRDE Rotating Disc ElectrodeSEM Scanning Electron MicroscopySTEM Scanning Transmission Electron MicroscopySTXM Scanning Transmission X-ray MicroscopySTM Scanning Tunneling MicroscopySE Secondary ElectronSAED Selected Area Electron DiffractionSHINERS Shell Isolated Nanoparticle Enhanced Raman SpectroscopySNR Signal to Noise RatioSAXS Small Angle X-ray ScatteringxxviAbbreviationsSEI Solid Electrolyte InterphaseSHE Standard Hydrogen ElectrodeSERS Surface Enhanced Raman SpectroscopyTEM Transmission Electron MicroscopyTUNA Tunneling AFMUHP UltraHigh PurityUHV UltraHigh VacuumUV-Vis Ultra-Violet Visible SpectroscopyUPD Underpotential DepositionWAXS Wide Angle X-ray ScatteringWE Working ElectrodeXAFS X-ray Absorption Fine StructureXANES X-ray Absorption Near Edge StructureXAS X-ray Absorption SpectroscopyXRD X-Ray DiffractionXPS X-ray Photoelectron SpectroscopyZP Zone platexxviiAcknowledgmentsFirst, I want to thank David Wilkinson and Dan Bizzotto for providing me with a tall ship, and a starto steer her by. I am especially grateful for their patience and in granting me considerable latitude todevelop my own research. Their commitment to supporting independent scholarship, even at personalcost, is in keeping with the highest traditions of academic service.I raise a glass to Jakub Drnec for lending me a dream and some photons. Placing a career-sizedbet on our work required no small amount of trust in each other. My time in France on the beamlinerefreshed a sense of achievement and interest in science that grad school had long since bled dry.The McMaster folks have been first-rate collaborators. I thank Adam Hitchcock for both his excellentcharacter and undogged vision as one of the finest experimentalists I have ever met. Lis Melo is deeplyacknowledged for her patience in manuscript preparation and wading through a bog of scientific incon-sistencies. I am indebted to mes conspirateurs intrépides Frédéric, Laetitia, Raphaël et Tristan for theirefforts and trust.Sydney Brenner tells us that “progress in science depends on new techniques, new discoveries, andnew ideas, probably in that order”. I am indebted to the synchrotron labs responsible for facilitatingmuch of my work, and all those who tend to the machine. The ESRF Structure of Materials group arerecognized for their generous and continued support.Gethin Owen deserves recognition for olympian efforts maintaining instrumentation to impeccablestandards. Robin Stoodley receives my thanks and apologies for tolerating my weakness with deadlines.The Bizzotto group members past and present have been excellent labmates. Tony, Jannu, Landis,Kaylyn, and others, it’s been a treat. Kamil, Laurie, and Jonas, you were enthusiastic about followingmy ideas even when we were flying blind.Any science contained in this thesis reflects great credit upon the quality of Walter’s espresso ma-chine and the craft beer industry of the lower mainland, responsible for fueling my PhD.I want to thank those who inspired me to make this journey without knowing it: George Shimizu, MikeMislan, Derek Lowe, and Tomas Vojkovsky. You offered a look behind the curtain of this industry. ThosexxviiiAcknowledgmentsearly lessons have consistently proven the most valuable.Friends and family need to be recognized, absent whose support this work would not have beenpossible. I am deeply grateful for the friendship of Nick McGregor, whose common perspective andcountenance was critical in staying the course these past years. I hope our paths meet again. I thankall members of the cranky chemists club, and hope to remain close with you in the future. AmandaMusgrove, you taught me about AFM, the evils that await postdocs, and the importance of napping. LizFisher, you put up with my disarray and delivered when it counted the most. Marie, you were endlesslypatient during my writing. Your support and love let me look past the horizon of grad school. Jess, Ben,Andrea, Kevin, and others, you kept me in good spirits as the clock ticked on.Finally, I would like to thank my parents, for everything.xxixDedicationFor my fatherxxxChapter 1IntroductionFuel cells are one of the oldest, but most promising technologies for energy storage and power genera-tion.8 The basic configuration of a fuel cell (FC) is similar to that of a battery, and consists of an anodeand cathode, separated by an electrolyte. The differentiating feature of the fuel cell is that the physicalcathode and anode are not electrochemically consumed during operation, and instead draw on a supplyof fuel which may be externally replenished. The hydrogen fuel cell, and more specifically the polymerelectrolyte membrane (PEM) fuel cell is the best known and commercialized design (Fig. 1.1). In thisdevice, a thin (20-200 μm) polymer electrolyte film is used to prevent mixing of cathode and anode gasstreams, which has the unusual property of proton conductivity.Hydrogen is dissociated into protons and electrons at the anode. The electrons are forced to flowthrough the external circuit, while the protons only diffuse through the polymer film towards the cathode.At the cathode, oxygen is reduced in the presence of protons to produce water.H2 −−*)− 2H+ + 2e− Eo = 0Vvs.SHE (1.1)O2+ 4e− + 4H+ −−*)− 2H2O Eo = 1.23Vvs.SHE (1.2)Both the hydrogen oxidation reaction (HOR) and the oxygen reduction reaction (ORR) are greatlyfacilitated by Pt catalysts. HOR on Pt surfaces in acidic media is extremely fast, while ORR is compara-tively sluggish.9–11 Suitable ORR performance requires the addition of relatively high catalyst loadingsto the cathode (presently around 30 g for an average automobile).12 The total cost of Pt for a typicalautomotive fuel cell at today’s prices represents a small fraction of the total system cost (30 g at $30-50USD/g = $900-1500). However, the global market for platinum group metals (PGMs) is relatively smalland volatile, so the widespread adoption of PEM FC technology raises questions about the supply andprice stability of these catalyst materials at high production volumes. The high cost and rarity of PGMshas created great interest towards improving the efficiency of their use, whether by increasing the intrinsic11.1. PEM fuel cell constructionFigure 1.1: Diagram of a conventional hydrogen PEM FC.1electrocatalyst activity for ORR, recycling/recovery, or replacement with cheaper alternatives. Reducingthe PGM loading necessary for performance targets requires efficient electrode designs, and controlover the morphology of the catalyst interface. Using the lowest possible loading leads to further issuesover the lifecycle of the device, when slow degradation of the Pt surface creates performance loss.13These three themes of intrinsic activity, electrode morphology, and catalyst degradation are explored inchapters 3, 4, and 5 respectively.1.1 PEM fuel cell constructionPEM FCs are composed of a number of layers which together form a membrane electrode assembly(MEA). A cross-section of a commercial MEA is visible in Figure 1.2. Beginning in the centre, the PEMitself is formed out of perfluorosulfonic acid (PFSA) or ’ionomer’. Modern high performance membranesare thin enough (15-20 μm) that mechanical reinforcement or internal layering is required. Attached di-rectly to both sides of the membrane is the catalyst layer (CL). The catalyst is typically Pt/C, Pt nanopar-ticles (~2 nm) dispersed on carbon nanoparticles (50 nm diameter). Ionomer is dispersed throughoutthe catalyst layer, serving as a mechanical binder and to provide proton conductivity to catalyst sites.The anode and cathode CLs are normally fabricated by spraying inks of the ionomer and catalyst directlyonto the PEM to produce a catalyst-coated membrane (CCM). Inks generally include large fractions of21.2. Ionomer membranesvolatile solvents, such that the aerosolized material dries “in-flight”, yielding a porous CL. High levels ofporosity allow for better mass transport of gases into and out of the film. The CLs are typically 10 μmthick on both sides. Many other methods of producing catalyst layers have been explored, from simplepainting to magnetron sputtering.14–16 Note that in the SEM cross section of the commercial PEM, thethickness of the ionomer and anode/cathode layers vary by up to 100% in different locations along thesample (Fig. 1.2). This heterogeneity highlights the level in difficulty of establishing process control overMEA fabrication, even in a commercial setting.Above the CL is the microporous layer (MPL). The morphology of the MPL resembles the CL, but theionomer and Pt/C are replaced by polytetrafluoroethylene (PTFE) and Pt-free carbon nanoparticles. Therole of the MPL is to provide a contiguous, intermediate interface with properties between the nanoscalefeatures of the CL and the layers above it containing micron scale features.17,18 It provides electricalcontact with the CL, while wicking away liquid water, distributing gases homogeneously, and reduc-ing mechanical strain on the CCM. The MPL follows the counterintuitive naming convention for porousmaterials, where micro, meso, and macro pores refer to the <2 nm, 2-50 nm and >50 nm domains re-spectively.19 The CL and MPL contain a pore-size distribution that bridges from large macropores (1μm) down to micropores. Above the MPL is the gas diffusion layer (GDL). The GDL consists of wovenor nonwoven carbon fibre paper, made hydrophobic with PTFE coating. The large carbon fibres in theGDL provide a high level of mechanical rigidity and electrical conductivity relative to the carbon below.This ensures minimum resistive losses and mechanical distortion, while maintaining sufficient porosityto allow for gas diffusion and humidification of the device. Around the edges of the GDL, a gasket orsubgasket liner provides a tight seal against gas leaks.The MEA is held between flow field plates. The plates, normally graphite, contain channels whichdirect the flow of reactant gases in a tortuous path over the MEA from the gas inlet towards the outlet.The MEA and flow field complete one electrochemical cell. As visible from the electrochemical reactionsabove, each cell produces a maximum of about 1V. For practical power generation, many cells are linkedtogether in series, forming a “stack”.20 The size of each cell (and the number of cells in the stack) aretailored to the power requirements for a particular application.1.2 Ionomer membranesThe polymer electrolyte membrane lies at the heart of a PEM FC. The operating conditions and per-formance envelope of cells are governed by the idiosyncratic needs of the ionomer membrane. An31.2. Ionomer membranesFigure 1.2: Membrane electrode assembly construction. Exploded perspective diagram (left) and scan-ning electron microscopy image of cross section (right).Figure 1.3: Chemical structure of Nafion, a well-known PFSA (left). Proposed cluster-morphology pro-posed for PFSA (right). The blue portion represents the hydrated fraction, while the polymer backbonefills the yellow areas.understanding of the ionomer chemistry is necessary to explain how PEM FCs are engineered.Virtually all ionomers used in commercial hydrogen fuel cells are composed of a perfluorosulfonicacid copolymer. The polymer consists of a PTFE backbone, periodically branched by side chains ter-minating in a sulfonic acid (Fig. 1.3, left). Protons can hop between neighbouring side chains, creatingionically conductive pathways through the polymer. Importantly, the proton mobility through a monolithicPFSA membrane is only 3-4 times slower than in bulk aqueous electrolyte.21 The perfluorination of thepolymer prevents radical damage by reactive oxygen species (ROS) and degradation under highly acidicconditions. The electronic and anionic conductivity of the PFSA are extremely low, so no short circuitingoccurs. PFSA can be extruded or cast into membranes which are impermeable to gases.41.3. Oxygen reduction and Pt electrochemistryHydration of the PFSA is a critical aspect of its performance. Despite consistent efforts to developanhydrous proton conduction22, membranesmust remain fully hydrated in order to conduct protons. Thisplaces several limits on PEM operation: the temperature must not exceed 100℃, and the gas streamsmust be continuously humidified. If bulk liquid water condenses on or inside either electrode, the masstransport of gases is locally disrupted.23 Fuel or oxygen starvation greatly degrades cell performanceand lifetime. Of course, water is actively produced inside the cathode as a product of theORR. Therefore,operating a fuel cell requires dynamically balancing multiple heated, humidified gas streams to keep themembrane hydrated, while preventing “flooding”.The need for membrane hydration is driven by the unique nanomorphology of the PFSA.24–27 ThePTFE fraction of the polymer is hydrophobic, while the acidic side chains are hydrophilic. This differ-ence leads to microscopic phase segregation of the two fractions (Fig. 1.3, right). Water can be insertedinto the PFSA structure in a continuously variable fashion, producing a rich phase diagram of polymerpacking including dispersed, micellar, inverse micellar, and compacted phases. The membrane ad-sorbs approximately 20-40% of its volume in water during fuel cell operation, and swells accordingly.The interaction between water and hydrophilic domains tunes the proton conductivity over six ordersof magnitude. The side chains and high mobility of the PTFE backbone prevent strong packing inter-actions and reptation, leading to a dynamic microstructure which is partially amorphous, and partiallycrystalline. The structure of PFSA has been extensively studied for decades, but no real consensus hasemerged towards the morphology which creates its unique material properties.1.3 Oxygen reduction and Pt electrochemistryAlthough the Nernstian Eo listed for ORR is 1.23 V, the actual voltage established across a cell underopen circuit conditions is much lower, about 0.95-1 V.28 This voltage drop at high potentials is ascribedto activation overpotential for ORR, and the presence of mixed potentials29 from surface oxidation sidereactions. More specifically, the ORR on Pt surfaces is slow enough that reaction kinetics strongly influ-ence the observed voltage. The mechanism for the ORR on Pt surfaces remains poorly understood, inpart because more than one pathway exists.9,30 Under most conditions, adsorbed oxygen is reduced towater in a four electron process. An alternative two-electron path, producing hydrogen peroxide as eitheran intermediate or final product is known. The activation barrier for ORR is attributed to the breaking ofthe double bond in diatomic oxygen, which is generally regarded as the mechanism’s rate determiningstep.31 This has been proposed as either the first electron transfer to an adsorbed O2 molecule32, or51.3. Oxygen reduction and Pt electrochemistryFigure 1.4: Cyclic voltammogram of Pt catalyst layer (black, top trace) and Pt bead electrode (red,bottom trace). CV collected at 200 mV/s, 0.2M H2SO4 for catalyst layer, 50mV/s, 3 M H2SO4 for beadelectrode, Ar purged. Area integrated for surface area calculations is highlighted in bluethe formal dissociation of oxygen on the surface.Depending on the electrochemical potential, electrode surface chemistry and experimental condi-tions, multiple mechanisms or partial mechanisms can be simultaneously active.33 The kinetics of ORRdeviate from a simple Butler-Volmer framework, although elements of this relationship are still commonlyused (i.e.Tafel plots).34 A typical ORR Tafel plot for Pt shows two slopes, depending on the potential.At high potential, where the surface is partially oxidized, a slope of 60 mV/dec is observed. At low po-tentials (high current density) the slope is approximately 120 mV/dec. For these reasons, fundamentalparameters such as exchange current density at zero overpotential are rarely reported for ORR. A cata-lyst’s specific activity for the ORR is normally reported as the current at 0.9 V vs the reversible hydrogenelectrode (RHE), normalized to the real surface area. This number (A/cm2 Pt) provides informationregarding the intrinsic activity of the catalyst surface chemistry. The mass activity (A/ mg Pt) is alsocommonly used, which normalizes the activity to the amount of Pt used, and accounts for factors suchas particle size.The standard cyclic voltammogram (CV) for polished and nanoparticle Pt surfaces in acidic mediumunder inert atmosphere is given in Figure 1.4.This CV shows three main regions. The double-layer region for Pt exists between 0.4-0.8 V, where61.3. Oxygen reduction and Pt electrochemistrythe metallic surface is free of adsorbates and behaves as an ideally polarizable electrode. At lowerpotentials (0-0.4 V), the adsorption and desorption of underpotentially deposited hydrogen (UPD) isvisible. This monolayer of H atoms is analytically useful, since it allows a precise measurement of thereal Pt surface area (blue region). Integrating the charge passed in this region of the CV, divided by theknown specific pseudocapacitance of Pt (210 μC/cm2) yields the electrochemical surface area (ECSA)of the Pt. At higher potentials (0.8-1.6 V) the formation and reduction of Pt oxide is seen. Differencesin the CV between a polished Pt bead and the nanoparticle Pt are apparent. Both the H-UPD and Ptoxidation/reduction peaks are much better defined on the bead. This is because the polished metalsurface has grains on the order of hundreds of nm in size. Each grain exhibits a well defined surfacecrystallography, with the peaks at 0.12 and 0.27 V representing H-adsorption to the 110 and 100 surfaceorientations against a broader 111 feature.35 The nanoparticles do not have a well-defined, long rangeordering of surface atoms, and possess many edge, corner, and defect sites which blur the peaks. BothCVs show a slight tilt from background ORR, which occurs when the electrolyte is not completely purgedfree of O2.It is well understood that Pt electrochemistry is quite sensitive to a wide range of variables: the natureof the supporting electrolyte36, the surface crystallography37,38, the size of the exposed crystal facets39,pH, trace impurities40, surface defects41, history42, etc. Furthermore, these effects are strongly cou-pled, and cannot be treated as isolated variables.43 Unless otherwise specified, this work refers to Ptor Pt/C in the presence of hydrated Nafion ionomer. The behaviour of this interface certainly divergesfrom the well-defined model system of a Pt(111) surface in non-adsorbing electrolyte (i.e.HClO4).35The CV in the presence of oxygen gas is very different, and O2 is reduced at all potentials less thanthe Nernstian equilibrium (Fig. 1.5). Starting at approximately 1 V, a strong reductive current is detected.Even though sluggish, the ORR of oxygen dissolved in aqueous electrolyte quickly reaches masstransport limited rates when measured in a conventional cell. Two general solutions exist for improvingmass transport of oxygen to the electrode surface. First, a rotating disc electrode (RDE) can be usedto enhance the transport of oxygen bearing liquid electrolyte towards the electrode surface throughconvection. This is the most common configuration found in the literature.12 Second, a half-cell designmay be used. In this configuration, part of the catalyst surface is in contact with electrolyte, while anotherpart is in contact with air or oxygen gas. The mass transport in the gas phase is greatly enhanced, sincethe concentration of oxygen is not limited by solubility.In contrast with the ORR on the cathode, the anodic HOR is a simpler, better understood reaction.10Themechanism is believed to involve at least two out of three proposed steps, known as the Tafel/Volmer71.3. Oxygen reduction and Pt electrochemistryFigure 1.5: Linear sweep voltammogram of Pt in the presence of 1 atm molecular oxygen, recorded ina half cell. Dashed lines highlight the ORR current at 0.9 V for this catalyst layer. Curve recorded at 20mV/s using 0.5 M HClO4 at 294 K and iR corrected.or Heyrovsky/Volmer pathways.H2 −−*)− 2Hds (Tfel step) (1.3)H2 −−*)− Hds +H++ e− (Heyrovskystep) (1.4)Hds −−*)− H+ + e− (Volmerstep) (1.5)Characterizing the HORonPGMsurfaces is challenging because of its extremely high speed in acidicmedium. This nearly ideal nonpolarizable behaviour is responsible for the good reference electrodecharacteristics of an RHE. Unlike most reference electrodes (SHE, Ag|AgCl, etc), the RHE does not havea well-defined, fixed potential, and simply reflects activity of the H+(q) species. Because the ORR,HOR, and (some) Pt oxidation reactions include the same number of protons per electron transferred,the electrochemistry can be conveniently referenced against a hydrogen electrode outside standardstate conditions. In a two-electrode PEM FC, the fast HOR kinetics cause the anode to serve as a81.3. Oxygen reduction and Pt electrochemistrypseudoreference electrode. The anode adopts a relatively constant potential, with most of the changesin polarization occurring across the cathode.The oxidation of Pt is an especially complex issue with several overlapping redox processes. Stan-dard reduction potentials for several relevant processes are given below.31,44–47Pt2+ + 2e− −−*)− Pt Eo = 1.19Vvs.SHE (1.6)PtO2+ 2H+ + 2e− −−*)− PtO+H2O Eo = 1.05Vvs.SHE (1.7)PtO+ 2H+ + 2e− −−*)− Pt+H2O Eo = 0.88− 0.98Vvs.SHE (1.8)PtO2+ 4H+ + 2e− −−*)− Pt2+ + 2H2O Eo = 0.84Vvs.SHE (1.9)[PtCl4]2− + 2e− −−*)− Pt+ 4Cl− Eo = 0.73Vvs.SHE (1.10)Starting with the lowest potential entry, it is obvious that the corrosion of Pt is greatly facilitated by thepresence of chloride ion contaminants, shifting the onset of dissolution approximately 0.5 V earlier. Theoxidation of Pt under inert conditions begins at about 0.8 V. It is useful to consider the initial oxidation asthe formation of a PtO or PtOH monolayer. At higher potentials (1.1-1.8 V), this initial adlayer is furtheroxidized, forming a monolayer of poorly defined oxide, nominally α-PtO2. Further oxidation at potentialshigher than 1.8 V yields new, poorly understood phases of the Pt oxide, but these are generally notconsidered to be formed inside fuel cell electrodes.48 In PEMFCs, the maximum potential window isapproximately 0-1.5 V, with potentials greater than 1 V only transiently established during startup andshutdown sequences. Note that in aqueous electrolyte, the pure dissolution reaction at 1.2 V does notreadily proceed, since no bare metal sites exist, and the surface is protected by an oxide layer.At 1.1 V (on a 111 surface), the adsorbed layer of O atoms reaches complete coverage, and anelectrochemical transformation of the oxide called place exchange occurs.49 This reaction involves theadsorbed layer of oxygen atoms slipping underneath the uppermost layer of Pt atoms. New oxygenatoms adsorb onto the fresh Pt sites generated. Although these processes have been studied in detailover several decades using well-defined model systems, the transferability of the surface chemistry91.4. Thesis objectives and scopetowards practical catalysts is poorly known. The Pt oxidation and oxide reduction reactions are notviewed as reversible electrochemical couples, and the mechanism in the oxidation direction should notbe considered the same as the reduction direction.1.4 Thesis objectives and scopeThe surface chemistry of Pt inside fuel cell electrodes remains poorly characterized. There exist largegaps in understanding between the behaviour of well defined single crystals and nanoparticle surfaces.The adsorption phenomena, identity, and structure of various species at the interface under environ-mental conditions relevant for device operation is largely unresolved. The stability and evolution ofthese surfaces over the course of a device’s life cycle are also unknown. The path towards a better un-derstanding of these issues is blocked by three key experimental problems. First, there is a lack of goodsamples which capture elements of the Pt-PFSA surface chemistry, while remaining structurally simpleenough to interrogate. Second, there is a pressing need for new techniques which can directly probethe surface structure of these materials in a detailed but unambiguous fashion. Third, the dynamic na-ture of the polarized interface makes techniques capable of studying buried nanoparticle surfaces underpotential control especially desirable.In this thesis, aspects of these three related problems are addressed. Chapter three involves the de-velopment of a (relatively) well-defined interface between Pt nanoparticles and PFSA as a model systemfor PEM fuel cell catalyst layers. The controllable assembly, structure, and electrochemical character-istics of this system are explored, with an emphasis on dynamic hydration effects. In chapter four, thismodel system is then used to directly probe the stability of the PFSA side of the interface, using ad-vanced characterization techniques. The limitations of different analytical tools for studying ultrathinPFSA films on Pt surfaces are addressed by combining multiple approaches, such as AFM, STXM,and FTIR, which converge on a quantitative understanding of structural and chemical changes at theinterface. Finally, chapter five addresses the development of new synchrotron spectroelectrochemicaltechniques for studying the structure and stability of nanoparticles in situ. Altogether, this work aimsto fill gaps between fundamental single crystal surface science and device level performance charac-terization of PEM fuel cells by fabricating model systems of the Pt-PFSA interface, and enabling thecharacterization of these complex materials in greater detail than previously available.10Chapter 2Experimental Methodology2.1 ReagentsThe use of ultrapure water free of chloride and trace organic contamination is essential for reproducibleelectrochemistry on Pt surfaces. Therefore, type I water (18.2 MΩcm) produced by a Milli-Q Integral 5system was used for all experiments. Total organic carbon was monitored, and was less than 4 ppb inall cases. All electrolyte was prepared from acid of the highest quality available (double-distilled fromquartz, ultratrace metals grade). All reagents were weighed out to a minimum accuracy of +/-1 mg, or+/-1%, whichever was more precise.2.2 Electroless depositionThe electroless deposition followed the procedure published previously50,51, with several small modifi-cations. A piece of Nafion is held clamped between two chambers, allowing different solutions to accesseach side. One side is filled with dilute sulfuric acid (pH 1 with 13 mM potassium sulfate), while the otheris filled with a cationic Pt salt in dilute KOH (pH 13). After allowing the Pt cations to diffuse into the Nafionfor one hour, the Pt solution is then replaced by a solution of sodium borohydride after rinsing the celltwice with water. The clamped cell reactor used a 9 x 9 cm opening over which Pt was deposited. Theconcentration of Pt was 2 mM, while the borohydride solution was 0.2 M, and the water bath was heldat 318 K. The borohydride solution was prepared immediately before the deposition by adjusting waterto pH 11 with KOH, then adding the NaBH4 powder and stirring until dissolved. The reactions werequenched by pouring the solutions out of the reactor followed by rinsing the Pt layer with dilute nitricacid. The submonolayer, near-monolayer, and multilayer films were reacted with borohydride solutionfor 4, 8, and 25 min, respectively. Nafion 115 (Ionpower, Inc.) was cleaned by boiling in 1 M sulfuric acid(1 h), and then water (1 h), then dehydrated inside an oven at 350 K under dynamic vacuum for severalhours before use. Sudden temperature changes to the membrane are avoided, and boiling solutions112.2. Electroless depositionare allowed to cool briefly before water is slowly added to bring them to room temperature before themembranes are removed from the liquid. The Pt layer is more heterogeneous near the edges of thefilm, so samples were taken from near the middle of the membrane for all measurements. The preciselocations on each Pt film sampled for each analytical technique are not identical, but are representative.After deposition, the Pt films were soaked in 0.5 M perchloric acid to exchange salts and fully hydratethe Nafion.The PtRu alloy was prepared by spiking the Pt solution with an aliquot of Ru(NH3)6Cl3 to a finalconcentration of 8 mM. To avoid precipitation, the Ru was added only in the last five minutes of Ptsoaking.2.2.1 Platinum (II) tetramine nitrateReferred to as ’Pt-amine’ throughout the text. Several batches of this Pt(NH3)4(NO3)2 salt were pur-chased (Strem, 78-2010) and used without further purification.It has been appreciated since the pre-’nano’ literature that, as phrased by Bradley, “perhaps the mostimportant irritant in colloid synthesis is irreproducibility”.52–54 The cause of these issues often arises fromsolid or soluble impurities in the starting materials, which act as nucleation sites or specifically adsorbedligands on nascent surfaces, respectively.While quoted as >99% purity on trace metals basis, several differences between batches of the Pt-amine reagent were noticed. The appearance of this compound varied between clear, colourless, largecrystals and a dark, brown powder. The color of the compound was likely derived from reduction of thePt(II) ions to Pt nanoparticles. It was noticed that older samples had turned darker over time. This saltwas stored at room temperature in a dark cupboard in tightly sealed amber glass bottles to minimize anyphotoreactions. Once dissolved into solution, photosensitivity was no longer observed over two yearsof aging. The UV-Vis spectra of these aged Pt solutions was checked periodically to monitor for anydisproportionation or ligand exchange, but no changes were observed.It was noticed that all solid samples and solutions of this salt possess a quite noticeable sweet fishyodour, distinct from ammonia. The volatility of the Pt complex should be very low. Commercial samplesof the chloride salt from a different manufacturer (Pt(NH3)4Cl2, Sigma) were odourless, as were sam-ples of Pt tetramine nitrate prepared from the chloride salt by repeated evaporation from concentratedHNO3. Taken together, this indicates a trace alkylamine impurity probably left over from synthesis andpurification. Industrially, this complex is likely produced from the common precursor K2PtCl4 using am-monia followed by nitrate ion exchange.55 The direct addition of ammonia to PtCl4 2+ ions precipitates122.3. Half-cell electrochemistrythe virtually insoluble impurity known as Magnus’ green salt, PtCl4Pt(NH3)4.56 The addition of alky-lamines at low concentration has been reported to disrupt the crystallization of Magnus green, improvingsolubility.57 Alkylamines are known to be potent adsorbates and structure directing ligands for platinumsurfaces and nanoparticles (eg oleylamine).58–60 Organic contaminants in the starting material wereanalyzed by GC-MS after extraction of the Pt salt into LC-MS grade acetonitrile/chloroform. Only tracequantities of xylenes and trimethylbenzenes were detected in the organic fraction.Different batches of Pt-amine were compared head-to-head using electroless deposition to assessreproducibility. It was determined that the electroless deposition from older Pt salt samples was slower.Since the Pt-amine solid is somewhat hygroscopic, old batches probably yield apparent lower activitysimply from hydrate formation. Solutions of Pt-amine did not lose activity over time. The variability inelectroless layers using different batches of Pt-amine could not be discerned against the natural variationof replicates performed using the same batch. The relatively high cost of this starting material and desireto develop scalable industrially relevant process are further incentives for using technical grade reagents.2.2.2 Membrane electrode fabricationMEAs of the electroless film were fabricated by hot pressing the various layers together. Hydrated,platinized membranes were laminated against a Sigracet 25-BC MPL-GDL carbon electrode (Fuel CellStore) in a hydraulic press at 125ºC and 1200 psi for 90 s. The MEA was sandwiched between 0.5 cmthick PTFE plates to distribute the mechanical pressure. After pressing, the MEAs were left to cool toroom temperature under light mechanical pressure (<1 psi) to produce a well-bonded electrode.2.3 Half-cell electrochemistryA three electrode cell was used, featuring a Pt mesh counter electrode and Hg/HgSO4( 0.5 M H2SO4)reference electrode (0.696 V vs RHE). For the electrolessly deposited catalyst layers, the working elec-trode was encased in a PTFE compartment, where the bare Nafion side of the membrane was in contactwith electrolyte, while the Pt and carbon coated side of the membrane was sealed against a gold ringheld inside a liquid-tight, gas-purged chamber (Fig. 2.1).Polarization curves were measured in 0.5 M HClO4 (GFS Chemicals) purging with UHP grade argonor oxygen (Praxair) at 294 K and 1 atm, after 200 cleaning cycles at 500 mV/s to remove adsorbed con-taminants. Reported values are adjusted versus the reversible hydrogen electrode (RHE). Polarizationcurves were measured on the first positive going scan.132.4. Electron microscopyFigure 2.1: Electrochemical half-cell design used for testing ORR activity of electroless films.2.4 Electron microscopyElectron microscopy uses accelerated electrons to image samples with high spatial resolution. The veryshort De Broglie wavelengths of electrons and their high scattering cross section with matter allows formagnifications much higher than with optical microscopy, where diffraction limits the performance of thesystem. In this thesis, two different optical geometries were used: the scanning electron microscope andthe transmission electron microscope. In addition to simple imaging, electron microscopy was coupledwith two different electron spectrometric techniques used for localized elemental analysis.2.4.1 Scanning electron microscopy (SEM)An SEM rasters an electron beam across a sample surface, recording the electrons emitted by thesample at each point. The accelerating voltage of the primary incident beam in all SEM images was1 keV. This allows for imaging non-conductive samples such as polymers with high surface sensitivity,and reduced buildup of electrostatic charging.61 The microscope was operated in either secondaryelectron (SE) or backscatter electron (BSE) mode. Each incident primary electron produces a showerof “secondary” electrons from the sample. The kinetic energy of SEs is very low, averaging 20 eV. SEmode produces images with high signal to noise and edge definition, but the contrast is nontrivial tointerpret.62 In backscatter mode, a modest negative voltage (-100 V) is applied to the detector during142.4. Electron microscopyimaging, such that secondary electrons are repelled and not detected. Only higher energy electrons fromthe primary beamwhich backscatter off the sample are counted. The contrast in these images correlateswith electron density, where bright features are locally electron dense. This allows Pt nanoparticlesinside a polymer matrix to be easily discriminated.Several precautions are required for imaging delicate samples. A 50 pA probe current and shortdwell times were used to minimize optical aberration and minimize beam damage. Image processingincluded linear histogram contrast adjustment and despeckling in ImageJ.63 Selected images were alsoprocessed to reduce high frequency noise with fast Fourier transform filtering, as indicated in figure cap-tions. Fracturing membranes under liquid nitrogen produces a clean exposed edge for cross sectionalanalysis. Samples were dried under dynamic vacuum at 350 K for 1-2 hours before transfer to the SEMchamber. SEM images were collected on an FEI Helios 650 microscope equipped with an immersionlens and an EDAX TEAM Pegasus system at the Centre for High Throughput Phenogenomics at theUniversity of British Columbia.2.4.2 Energy dispersive X-ray spectroscopy (EDX)Energy dispersive X-ray spectroscopy (EDX) measures the energy of characteristic X-rays following ex-citation by the incident electron beam. The spectra is produced through single photon counting, wherethe energy of each X-ray is resolved by integrating the number of secondary electrons generated in-side a cooled silicon drift detector.64 EDX is capable of detecting elements from Li-U with a detectionlimit of around 0.1 wt%, although the accuracy of relative elemental abundance is normally consideredsemiquantitative because of matrix effects. Since the electron beam is controlled by the microscope,the analysis can be performed in a spatially resolved manner. The spatial resolution of EDX is governedby the scattering of the primary electron beam. X-rays generated up to several microns deep insidethe sample can escape, and the electron beam at typical energies spreads out at least microns widelaterally. 20 kV was used as the accelerating voltage, to produce optimal excitation efficiency. Matrixescape depth correction factors were applied using EDAX software, and manually double checked.652.4.3 Transmission electron microscopy (TEM)In TEM, a high energy electron beam (100-300 kV) is focused onto a very thin transparent sample.Electrons passing through the sample are directly detected. Two different configurations were used. In’normal’, brightfield imaging, electrons scattered by the sample are not detected. The unscattered beam152.4. Electron microscopyis projected onto an image plate. Dark regions in the image correspond to electron dense locations onthe sample. At very high spatial resolution, the contrast of brightfield images includes a large diffractioncomponent. This is typically visible as lattice fringes, where planes of atoms in crystalline samples serveas diffraction gratings.The other TEM configuration was high angle annular dark field imaging (HAADF). In this configu-ration, the beam is tightly focused to a single spot on the sample, which is then rastered across thesample, much like in SEM. This is known as Scanning Transmission Electron Microscopy (STEM). InHAADF imaging, the scattered and unscattered beam are separated inside the microscope. An annu-lar detector records scattered electrons which leave the sample at “high” angle, typically 2-10 degrees.The intensity of the deflected beam is positively correlated to the local electron density of the sample,approximately Z2. Diffraction contrast also disappears, due to the high angles employed.66 Therefore,electron dense materials such as heavy metal catalysts can be imaged at atomic resolution. The beamexiting the sample deflected at lower angles is passed into the separate EELS detector, detailed below.A large number of different TEM instruments were used to provide a variety of analytical capabilities(SAED, EDX, EELS, HAADF, STEM, etc). Samples of electroless films were prepared by ultramicro-toming thin sections (50-100 nm thickness) of each sample embedded in Spurr’s resin. For the in-situXRD sample, catalyst was recovered from the electrode with a micropipette, by gently rolling a drop ofisopropyl alcohol over the surface. These catalyst-rich isopropyl alcohol drops were then transferredonto copper TEM grids with lacey carbon film, and dried under ambient conditions.2.4.4 Electron energy loss spectrometry (EELS)The forward scattered beam from the STEM can be directed into amagnetic sector analyzer most familiarto chemists for their use in mass spectrometers.67 Electrons passing through a weak magnetic field drifttube or ’prism’ experience the Lorentz force, and are dispersed according to their kinetic energy. Ionoptics project the dispersed beam onto a detector.Electrons which scatter inelastically off atoms lose energy corresponding to characteristic energylevels of orbitals. By observing different ionization edges in the EELS spectrum, elements can be quan-titatively mapped in STEM with much higher sensitivity and resolution than with EDX.In this work, the Pt M4-5 edge and Ru L3 were measured and extracted using a Gatan GIF detector.The probe diameter was about 0.1 nm.162.5. X-ray photoelectron spectroscopy2.5 X-ray photoelectron spectroscopyX-ray photoelectron spectroscopy (XPS) measures the binding energies of core electrons from a solidsample surface. X-rays with sufficient energy impinge upon the sample with sufficient energy to ionizecore-shell electrons. Low energy photoelectrons generated near the sample surface can escape andtravel into a spectrometer where their residual kinetic energy is measured. The binding energy of eachelectronic orbital is characteristic for different elements. The difference in energy between the incidentphotons and the kinetic energy of photoelectrons (KE) is directly connected to the binding energy throughEinstein’s relation:Bndngenergy= h− KE−Φ (2.1)where Φ is the work function, a sample dependent calibration factor. The low energy of the photo-electrons provides an extremely short inelastic mean free path, so the peaks only yield signal from thetop few nm of the sample.X-ray photoelectron spectra were collected on a custom-built Omicron UHV apparatus at ESRF ID03.The incident beam was operated at 20 mA and 14 kV using an Al target. No monochromator was used,and the entire sample holder was exposed. Angle resolved XPS was performed in order to optimize thesignal to noise from the low brightness X-ray source. By observing the photoelectrons closer to grazingincidence, the surface sensitivity of the spectra are improved by a factor of cos(Θ). Takeoff angle wasoptimized at approximately 12 degrees normal to the surface. Freestanding electroless catalyst layerswere baked at 85ºC overnight at 1 mbar in a vacuum oven to remove bulk hydration prior to transferto the UHV system. Samples were fixed to the sample holder using Mo screws in contact with the Ptfilm. No charging compensation was used, although the sample holder was grounded. Spectra werecollected overnight, in a single high resolution pass. Chamber pressure during data collection remainedstable at approximately 5 x 10-9 mbar .2.6 Inductively coupled plasma optical emission spectroscopy(ICP-OES)ICP-OES is a common technique for trace analysis of dissolved metals. An acidic solution is nebulizedinto a stream of argon gas, and this aerosol is passed through a plasma torch at approximately 104172.7. UV-Visible spectroscopyK. This heat converts compounds into atoms with high efficiency and low sensitivity to sample matrix.The atoms emit light and this emission is detected in a UV-Visible spectrophotometer. This technique isespecially useful because the linear detection range is large. The limit of detection for platinum is <50ppb.For ICP analysis, all Pt samples were dissolved in aqua regia. These solutions were boiled down re-peatedly withHNO3to removeHCl, then diluted to volumewith enough acid for a finalHNO3concentrationof 2%. This procedure is necessary to remove the presence of dissolved gaseous NOX species, theactive components of aqua regia. Volatile matrix components change the nebulization droplet size dis-tribution, desolvation, and detection efficiency. Experiments were performed to check matrix sensitivity.The Pt signal was not very sensitive (<3% rmsd) to a range of HNO3 concentrations (0.5-5%).HCl wasalso tested as a matrix with similar results. HNO3solutions have a more stable vapour pressure, andare less corrosive towards instrument components than HCl, so were used whenever possible. Ex-periments comparing the use of an internal standard, Pt standard addition, and external standard alsoshowed similar results in the >100 ppb range. An external standard approach was used to minimizepreparation time and sample volume required. The plasma power, viewing height, and flow rates wereoptimized for the detection of the Pt 306 nm line. This line is slightly less sensitive than others in thefar UV (203 nm, 219 nm, etc) but has higher day to day reproducibility in sensitivity curves, and is lessprone to scattering interference.1 cm2 circles were punched from the deposited membranes, and digested in 2 mL of aqua regia ina vial. An Agilent 725 spectrometer operating in radial mode was used for all reported measurements,which represent the average of three measurements each for three different sub-samples. Calibrationcurves showed excellent linearity (R2 =0.9999) over a 0.05-500 ppm range using certified Pt standards(Delta Scientific).2.7 UV-Visible spectroscopyTransmission mode UV-Vis spectra were acquired using an Agilent Cary 5000. The spectrometer wasequipped with a home built quartz environmental sample holder to control humidity. Reflectance spectrawere measured using a Universal Measurement Accessory. Collimating slits (three degrees) were in-stalled on the source and detector producing a spot size of 0.5 cm on the sample to average any spatialheterogeneity of the electroless Pt film.182.8. In plane conductivity2.8 In plane conductivityIn-plane resistance was measured with a homemade Jansen 4-probe device using a single point 1 kHzmeasurement and geometric correction factors, and reported as the average of 4 replicates.68,692.9 195Pt Nuclear magnetic resonance (NMR) spectroscopyNuclear magnetic resonance spectroscopy is a versatile tool for studying the chemical structure andhydrodynamic properties of spin-active atomic nuclei in solution and solid-state. The magnetic momentof nuclei with spin precess along an axis in an external magnetic field. The precession rate around thedirection for a given magnetic field is known as the Larmor frequency, a unique characteristic property ofevery nucleus. Nuclei will resonantly couple to radiation tuned at this specific frequency. Chemical bondschange the magnetic environment of the nucleus, shifting the exact resonant frequency, and providinginformation on its local environment. NMR is most popular for light elements in which the frequency isquite high. However, spectra of many heavier elements can provide otherwise unattainable structuralinformation. The 195Pt isotope is especially convenient. The combination of a 1/2 nuclear spin, highnatural abundance (34%), high gyromagnetic ratio, and wide chemical shift window (6700 ppm) makeit an excellent technique for following Pt coordination chemistry.70 These factors combine to yield anunusually high sensitivity, with 0.4% the receptivity of 1H, and 21x better than unenriched 13C. As a ruleof thumb, Pt species with concentrations of low mM to high μM may be readily observed on a typicalmodern instrument in a 1 minute scan. As usual for heteronuclei, polarization transfer experiments canbe used to enhance this sensitivity.NMR has been used to analyze the composition of electrodeposition baths, most commonly trackingthe fate of additives by 31P or 1H.71 Pt, Al, Sn, Pb, and Cd are the commonly electroplated metals thatmeet the general criteria for direct NMR detection (low spin isotopes and high receptivity). This techniqueis probably underrated within the electrochemistry community because high electrolyte concentrations(>0.5 M) strongly absorb radio waves and cause problems with automatic tuning and matching on thespectrometer probe.72,73 These problems can be addressed by manually tuning the probe and usingthin sample tubes. NMR spectra were recorded in 3mm tubes using a Bruker Avance 400 with a directbroadband observe probe. 195Pt spectra are unusually sensitive to acquisition parameters. Sampleswere equilibrated inside the magnet to provide good temperature control. A 1 s ring down with 75 kHzsweep width, and 1 Hz FID resolution were used. 10% D2O was added to all samples to provide a192.10. Confocal laser scanning profilometrystrong frequency lock.2.10 Confocal laser scanning profilometryIn this technique, light is rastered across a reflective sample while the scattered intensity is recorded.The optical microscopy system is constructed in a confocal fashion: only light scattered from a verynarrow focal depth is captured, while the rest is rejected using a spatial filter.74 While classical opticallimits are placed on the lateral resolution of this system, the position of the surface can be inferred byinterpolation of the confocal scans. Therefore, very fine demarcations are resolvable in the depth of thesample, down to approximately 10 nm.75 By using a high power laser as a source, 3D scans can bequickly acquired. Ultraviolet lasers not only increase the lateral spatial resolution, but are much moresensitive due to increased light scattering. An Olympus LEXT OLS3100 equipped with a 408 nm laserwas used.2.11 Atomic force microscopy (AFM)Atomic Force Microscopy is the most common technique for imaging and measuring the topography ofnanoscale materials and surfaces. A conceptual diagram of the instrument is show in Figure 2.2.A laser is focused on the reflective back surface of a long, flexible cantilever. On the opposite sideof this cantilever is a sharp tip. When this tip is brought into the proximity of a surface, attractive andrepulsive interactions bend and twist the cantilever. This motion changes the position of the reflectedlaser spot, which is recorded by a simple position sensitive detector. The cantilever is attached to apiezoelectric stage capable of moving the assembly in the x, y, and z directions. By scanning the piezoin the x and y directions while implementing a feedback loop between the position of the laser and theposition of the cantilever, it is possible to map the sample surface.In this thesis, several imaging modes were utilized. Simple contact mode imaging was used to mapsurfaces. In contact mode mode, the tip maintains constant contact and is dragged across the surfacewhile the piezo compensates for any topographic changes. The tip deflection is held at a constantnonzero value over time through feedback. Contact mode has the major advantage that it is simple touse in wet, damp, and viscous media. This means that images can be recorded in liquid water andambient humidity, which is important for our hydration sensitive polymer electrolytes. The disadvantageis in contact mode, the tip exerts strong lateral forces on the sample when scanning quickly. These202.11. Atomic force microscopy (AFM)Figure 2.2: Conceptual diagram of atomic force microscope. Adapted from reference2forces can easily destroy the surface of soft, delicate samples that are not uniformly flat. Tapping modeimaging was also utilized, also known as AC, intermittent contact, or non-contact AFM. In this mode,an piezo oscillation is applied to the cantilever at a vibrational resonance frequency(50-350 kHz). Thisvibration causes the tip to tap against the surface and dampens the oscillation, which is kept at a constantamplitude through feedback. The amplitude feedback of the measured vibration is then used measurethe tip-sample interaction and sample topography. This mode has the advantage of exerting very lowlateral forces on a sample compared to contact mode. Unfortunately, tapping mode is difficult to apply onhydrated materials for three reasons. First, liquids mechanically dampen and broaden a sharp cantileverresonance into a lower quality series of coupled oscillations.76 Secondly, depending on the cantilever,the oscillation is on the order of 20 nm. If the surface holds more than a monolayer or two of water,a tip must drill through this unstable interface while acquiring a pixel. Typical imaging conditions use10-1000 taps per pixel, and the resulting image can be a convolution of water overlayer and substratedepending on settings. Third, wet surfaces create strong adhesive forces through the formation of aliquid meniscus with an AFM tip, and soft cantilevers can easily become ’pinned’ against it regardlessof the applied tapping amplitude.Atomic force microscopy (AFM) images were collected on an Agilent 5500 microscope equipped witha 90 μm scanner using AC and CSAFM nosecones. SiN 30kHz cantilevers with <5 nm on-scan-angle212.11. Atomic force microscopy (AFM)Figure 2.3: Current sensing AFM schematic. Conductive coating (black) on cantilever and tip allowscurrent to flow between substrate and AFM tip with the application of a suitable bias.pyramidal tips (Vistaprobes) were used for the contact mode imaging. Pt-Ir coated SiN 30 kHz probes(nominally 0.35N/m) were used in current-sensing experiments. For current-sensing, an environmen-tal chamber was purged with 5 SCFH dry nitrogen for at least 1hr prior to data collection. Sampleswere dried under dynamic vacuum in a desiccator for 1hr. All images were collected at a scan rate <1μm/s. Images were flattened with a 2D 2nd order polynomial and scar correction using Gwyddion 2.37software.772.11.1 Current sensing AFMCurrent sensing AFM (CS-AFM) measures different types of conductivity in samples.78 An AFM can-tilever with a conductive coating sputtered over the tip is used. A DC bias is applied between the tipand an electrode placed in contact with the sample. The current passing through the tip and sampleis integrated, and recorded simultaneously with other AFM imaging parameters (topography, etc). Thiscurrent mapping works only when operating the AFM in contact mode, and is analogous to conventionalscanning tunnelling microscopy. CS-AFM is useful for probing low to medium conductivity samples,such as semiconductors or thin films. The difficulty and unpopularity of this analytical technique meanthat very little guidance is available in the literature and from instrument manufacturers on best practicesor standard operating procedures.The related electrochemical imaging mode Tunnelling AFM (TUNA) is much more common than CS,and is the tapping mode equivalent, where the biased tip provides only transient contact. Compared withTUNA, the CS mode is much less sensitive (uA vs fA), and creates much more damage to the sample asa result of the contact imaging mode conditions. The key advantage to using the current sensing modeis control over electrochemical potential of the tip. If transient (<50 kHz) contact is made with the tip, theelectronic signal measured is largely a capacitive current, as the tip is polarized and depolarized. Thisis useless for measuring faradaic responses in a sample or the structure of polymer electrolytes. For222.12. Surface enhanced Raman SpectroscopyCS mode, the tip-sample bias can be set to a wide range of values, from mV to approximately 10 V. Inour configuration the bias is applied to the sample while the tip floats at virtual ground. Before imaging,the tip is held in contact with a grounded conductive substrate to correct for any voltage offsets.Piezo elements inside an AFM are driven by high voltage scan generators that quickly ramp up anddown, creating local EMF interference. The naturally low amplitude of the DC signal passing through thetip necessitates amplification as close to the source as possible to reduce noise. A micro-preamplifieris therefore connected directly to the conductive cantilever. Space and EMF requirements limit theperformance of this device, which ultimately govern the capabilities of CS-AFM. The dynamic rangeof the current channel is very low (approximately 0.1-10 nA). The electronic conductivity of materialsextends over an exceptionally broad range, from graphene (108 S/m) to Teflon (10-25 S/m). The resultis that the CS-AFM images recording electronic conductivity typically have binary contrast.The choice of cantilever conductive coating is also critical. Gold, PtIr alloy, PtCr alloy, diamond-likecarbon, and other proprietary conductive coatings are commercially available for AFM tips. Gold coatedtips are exceedingly soft, and can be destroyed over the course of a single image. Diamond coatingsare the most durable option, but have relatively low conductivity, and tips cannot be sharpened as finelyas metallic probes. PtCr tips are cheaper than PtIr but passivate, oxidize or spontaneously dealloy.PtIr coatings with a Cr adhesion layer are the best choice for silicon cantilevers on most samples, andwere used for all experiments presented in this work. The Pt/Ir coating is brittle and has a tendency toflake off the tip, after which conductivity is instantly lost. On soft samples such as Nafion, this problemis mitigated. Metal coatings must be sputtered on top of a regular tip, which means the tip radius isincreased from the usual <10 nm to 25-50 nm. This limits the spatial resolution available with through-the-tip electrochemical AFM measurements.2.12 Surface enhanced Raman SpectroscopyRaman spectroscopy measures the inelastic scattering of photons from a sample. This analytical tech-nique is typically used to probe low frequency modes, such as vibrational and rotational excitations. Anincident photon excites molecules into a virtual excited state, which re-emits a photon with slightly differ-ent wavelength. The difference in photon energy corresponds to the vibrationally excited mode. Ramanspectroscopy faces three main challenges. First, the inelastic scattering cross section is very small,so ordinary Raman experiments are weak and insensitive. Lasers are used to provide intense, highlymonochromatic light and the spectrophotometer must be quite sensitive. Second, the elastic scattering232.13. Fluorescence microscopycross section is quite high, and interference notch filters must be used to strip out a large quantity of lightat the original incident wavelength. Third, any fluorescence will mask small inelastic scattering signals.Surface enhanced Raman spectroscopy (SERS) is a variant on the traditional measurement whichattempts to overcome the sensitivity problem noted above. On rough metal surfaces, high electric fieldgradients can be established over subnanometer length scales. Photons of the appropriate wavelengthcan directly excite metallic surface plasmons, which couple with the vibrational bands of adsorbedmolecules. The signal from SERS can be enhanced by extraordinary factors, providing up to singlemolecule sensitivity. Unfortunately, the nature of the enhancement is not well understood. On nanopar-ticle coated surfaces, the enhancement is known to be highly spatially heterogeneous, dominated by theirreproducible arrangements of particle-particle interactions which create “hotspots” of SERS intensity.A Renishaw InVia Raman microscope equipped with a 695 nm laser was used to record SERSspectra. The laser was focused through a 20x objective into a 30 μm line focus, with a 100 s integrationat 5 mW intensity.2.13 Fluorescence microscopyFluorescence images were collected using anOlympus IX70microscope in transmissionmode equippedwith an EM-CCD Evolve digital camera. LED excitation at 375 nm was combined with a 515-550 nmemission filter.2.14 Infrared spectromicroscopyInfrared (IR) spectroscopy is useful for probing fundamental vibrational excitations of organic molecules.Broadband infrared light sources, such as the globar offer much lower brightness and flux compared withvisible-spectrum lamps. Also, commonly available detectors offer poor sensitivity to infrared light. Thetwo factors lead to poor signal-to-noise in infrared measurements. Virtually all infrared instruments areconfigured as Michelson interferometers to benefit from a number of optical advantages that maximizesensitivity and improve resolution. Interferograms are recorded, then converted into Fourier-transforminfrared (FTIR) spectra. An FTIR microscope follows a design analogous to conventional optical mi-croscopes, with only a few modifications (Fig. 2.4). The interferometer serves as a monochromator.The magnification limits of FTIR microscopy are governed by the diffraction limit of light, and the242.14. Infrared spectromicroscopyFigure 2.4: Typical configuration of FTIR microscope. Adapted from Bruker Hyperion 2000 instrumentmanual.252.14. Infrared spectromicroscopyFigure 2.5: Infrared flux intensity at the detector using a globar source as a function of microscopecondensing aperture3optical quality of the light source. The maximum resolution of a microscope is normally quoted at theRayleigh criterionR=1.22λ2NA(2.2)where R is the diffraction-limited spatial resolution, λ is the wavelength of light (10 μm@ 1000 cm-1),and NA is the numerical aperture of the objective lens and condensing aperture. This yields about 8 μmresolution on our instrument. Unfortunately, the low brightness of the light source often prevents workfrom being performed near this maximum.The intensity of the light focused into the detector is physically limited by the conservation of etendue.79For practical reasons of signal-to-noise, the microscope requires an aperture of roughly 15-20 μm toacquire high-quality spectra in a reasonable amount of time (Fig. 2.5). The fine-structure of Nafionspectra require high spectral resolution to accurately measure. Per the standard “trading rules” of FTIRspectroscopy80, increasing spectral resolution proportionally decreases signal to noise ratio (SNR), andincreases measurement time. High spectral resolution also requires configuring the optical bench witha smaller source aperture, which further reduces the signal intensity, degrading spectral quality. Unlessotherwise stated, spectra were collected at 8 cm-1 spectral resolution to balance resolution with SNR .262.14. Infrared spectromicroscopyA liquid nitrogen cooled HgCdTe detector was used to maximize sensitivity.The FTIR spectra of Nafion is extremely complicated and band assignment remains controversial.7Fluorinated materials have complex vibrational spectra in comparison to standard hydrocarbon poly-mers, where carbon atoms dominate the centre of mass in chemical bonds. The extra mass fromfluorine moves bond vibrations to lower wavenumbers, into the complex ’fingerprint region’ of the IRspectrum. The larger Van der Waals radius for fluorine vs hydrogen (135 vs 120 pm) also increasessteric repulsions between tightly packed neighbouring atoms. This leads to increased coupling (closelyspaced vibrational modes) and as well as broadening which contribute to spectral overlap.81 Unfortu-nately, commercial PFSA including Nafion does not have a well-defined molecular formula or structure.This means the spectra of nominally identical materials, even from the samemanufacturer, might changeover time or show batch-to-batch variation. The sidechain and backbone conformations depend exten-sively on the degree of hydration, which alter the intensity of accessible vibrational modes. Finally, underambient conditions Nafion is significantly above its glass transition temperature. Because the polymeris partially molten, the polytetafluoroethylene (PTFE) backbone and sidechains are far more dynamicthan many polymers and respond to their chemical environment.82The sample holder used for all experiments was a SiXNY window. To produce stress-free ultrathinfilms over a large area with homogeneous thickness, amorphous window material is grown by low-pressure chemical vapour deposition. This process results in significant deviation in stoichiometry fromordinary Si3N4.83 The FTIR spectra is relatively sensitive to these heterogeneous defects and impurities.These species are likely damaged by radiation, as they are sensitive to post-synthetic annealing andambient hydration.84–86Luckily, the major bands for these impurities (N-H 3400cm-1, Si-H 2200 cm-1) occur far away fromthose in the Nafion spectra (Fig. 2.6). The strong Si-N stretch (830 cm-1) is significant, but the material isfairly resistant to radiation damage itself.87 Sharp peaks representing residual water vapour (1750-1250cm-1) and the CO2 asymmetric stretch (2400 cm-1) are also visible in the baseline.Synchrotron FTIR (SR-FTIR) spectroscopy was attempted to improve the spectral SNR at high spatialresolution (Fig. 2.7). Spectra were collected using the Mid-IR beamline at the Canadian Light Source,using similar downstream optics as for the conventional benchtop FTIRmicroscope. The high brightnessof synchrotron radiation allows for higher flux to be focused onto the sample.88 The flux at 15 μm spotsize is improved up to 30x over the conventional globar.The flux and the precision of the SR-FTIR microscope are superior to conventional benchtop spec-trometers, especially at high spectral resolution (<4 cm-1and below). However, the transfer pipe optics272.14. Infrared spectromicroscopyFigure 2.6: Transmission FTIR absorbance spectra of SiN window sample holder (2 cm-1 resolution).Inset: higher magnification of the baseline regionFigure 2.7: FTIR spectra of X-ray damaged Nafion sample obtained using synchrotron and conventionalsources. Synchrotron dataset collected by Scott Rosendahl (CLS).282.15. X-ray microscopyof the beamline which carry the IR light into the instrument are much more complex. High intensityX-rays create heating and mechanical noise that must be actively cancelled using adaptive optics. Thisproduces flicker in the source intensity on the seconds to minutes time scale, leading to more challeng-ing background and offset correction. Because quantitative spectroscopy with very small signals wasdesired, the more stable conventional microscope was a better tool in our application. If it were possibleto acquire signal from the background and the sample locations simultaneously, such as with a 2D areafocal-plane-array detector, our samples would enormously benefit from SR-FTIR.89 The narrow accep-tance of Schwarzschild-Cassegrain geometry objective lenses and short detector sample distance in thecurrent microscope design prevent the use of an area detector in conjunction with synchrotron radiation.Transmission mode spectra were obtained using an Equinox 55 spectrometer and Hyperion micro-scope (Bruker). The microscope was equipped with an MCT detector and 36x objective (Coherent Inc.,NA 0.5). To improve signal-to-noise while minimizing drift, background and sample point spectra weremeasured in an interleaved fashion. 512 scans were collected at each position immediately after oneanother. Absorbance spectra from each sequential pair of background and sample measurements werenormalized at the baseline to correct offsets, then averaged with 6-10 replicates. Error bars on integra-tion plots represent the standard deviation between integrals of replicates for each peak. Alignmentbetween the optical and IR light paths was checked using an electron microscopy grid. Scanner velocitywas 20 kHz, with double-sided acquisition, and a 16 kHz low pass filter. Samples were mounted andleft to equilibrate in a dry air stream inside the microscope for several days before measurements. Theabsorbance of each region was normalized to the local thickness of the film determined from STXM.2.15 X-ray microscopy2.15.1 Near Edge X-ray Absorption Spectroscopy (NEXAFS)A basic understanding of X-ray absorption spectroscopy is essential to usefully apply more sophisticatedvariations such as STXM. As with XPS, an incident X-ray excites core electrons. Unlike traditional XPS,the excitation X-rays are tuned to the binding energy of an orbital. Near this edge, the electron will nothave enough energy to enter an entirely unbound state that can ballistically escape the sample surface.The electron can instead transition to bound, unfilled states, such as antibonding orbitals as they doin UV-Visible spectroscopy. A number of detection methods can be implemented, such as collectingAuger electrons, transmitted X-rays, or fluorescent X-rays. For the light elements treated here, XANES,292.15. X-ray microscopyFigure 2.8: Linear attenuation of X-ray photons in water. Calculated from tabulated data4NEXAFS, EXAFS, XAFS, and XAS can be approximated as a single absorption phenomenon, since theintensity of the sharp edge signal and unbound scattering are low.2.15.2 Synchrotron RadiationSeveral experiments were conducted at synchrotrons, and a basic understanding of X-ray - matter in-teractions is useful for interpreting these results. A synchrotron accelerates electrons to relativistic ve-locities. This electron beam generates photons when magnetically deflected by the Lorentz force. Theenergy of these photons can be tuned over a wide range, from microwaves to the hard X-ray regime.Both the brightness (photons/radian/mm2at the source) and the flux (photons per second) of the beamare extremely high. The cross-section for photon-electron interactions is strongly dependent on energy(Fig.2.8). The shape of this function is roughly independent of sample composition, and is known as the’universal curve’. Low energy (soft) X-rays are strongly absorbed by matter, while higher energy (hard)photons penetrate further before being scattered.2.15.3 Scanning transmission X-ray microscopy (STXM)Scanning transmission X-ray microscopy is a useful tool for probing the structure of low-Z materials withhigh spatial and chemical resolution. Soft, monochromated X-rays are passed through a diffraction lenscalled a zone plate (ZP) which focuses them to a point. This spot can be rastered across a thin-filmsample which absorbs some of the light. Unattenuated X-rays are detected behind the sample with a302.15. X-ray microscopyFigure 2.9: Block diagram of STXM instrumentation showing configuration of source and beamline (a)and detail of ZP and sample stage (b). Taken from Ade and Hitchcock, reprinted with permission.5simple photomultiplier tube. The energy of the X-rays can be tuned into order to provide spectromicro-scopic images. These images, where each pixel represents a spectra, are called stacks. The energyof the X-rays is normally selected to fall near the core-LUMO transitions for particular atoms inside thesample. Operated in this mode, STXM provides spatially resolved NEXAFS spectroscopy.There are several advantages to using STXM instead of EELS or EDX in a TEM. Spectral resolutionof X-rays is equal or better to state of the art electron optics. STXM allows ambient pressure operation,without degradation in spatial resolution. Depending on the photon energy, STXM samples can be upto several μm thick, while TEM samples must be <300 nm for EDX, or <50 nm for EELS. The damagefrom X-rays is approximately 500 times less than electron imaging techniques90. However, irradiationdamage is still substantial for sensitive materials such as fluorinated polymers.Two microscopes at two synchrotrons were used in the STXM experiments reported here: the STXM beamline at the Advanced Light Source (ALS), and the 10ID-1 Spectromicroscopy beam-line at the Canadian Light Source (CLS).91 These instruments use the same basic design, although theCLS STXM uses an undulator in place of a bendingmagnet source, allowing for higher flux andmagneto-optical experiments not relevant for the work performed here.One possibility considered as a X-ray damage mechanism was sample heating. The absorbed doserate at the ALS STXM for the C 1s edge approaches 2.2 GGy/s. Assuming a reasonable heat capacityfor Nafion, if all the absorbed photon energy is thermally converted, this corresponds to local heatingrate of(2,200,000,000 J/kg/s)/(1,000 J/kg/K) = 2,220,000K/sExposures of 1-100 ms per pixel then correspond to a local temperature increase of 104-105 K underthe beam in the absence of cooling mechanisms. However, recent in situ measurements of melting312.15. X-ray microscopypoints performed under irradiation show an offset of <1 K92. This means that most of the absorbedphoton’s energy is converted into breaking chemical bonds, at least for organic samples. Energeticrelease from K shell knockout processes in low Z materials, such as Auger electron emission, cannotmeaningfully escape the sample and are reabsorbed. This calculation also offers some insight towardsthe extraordinary flux of synchrotron radiation, even using an old bending magnet source 106 lowerpower than modern undulators.The dispersed Pt ionomer samples STXM were fabricated by direct reduction with H2 gas. Mem-branes were incubated with a 0.5 mg/mL cisplatin solution at 90ºC for 1 hour to ionize and exchangeprotons for Pt (II). The samples were placed in a tube furnace actively purged with 5% H2/95% Ar gas ,and heated at 5ºC/min to 85ºC. The oven dwelled at 85ºC for 10 minutes, then cooled to room temper-ature. For the large particle sample, the film was exposed to a UV hand lamp (dual band, 254 and 360nm) for 5 minutes before the reduction treatment, in order to seed the membrane with Pt nanoparticles.Both membrane samples appeared jet-black after reduction.STXM samples were prepared by ultramicrotomy. The samples were embedded in a special epoxyresin consisting of trimethylolpropane triglycidyl ether and 4,4’-methylenebis (2-methylcyclohexylamine).This embedding media has excellent radiation resistance and an easily extractable NEXAFS spectrawhile preserving good mechanical properties.93 The samples were then cut to between 50-100 nmthickness using a diamond knife with standard block facing technique on an ultramicrotome.94 Theseultrathin sections were floated onto a 30 nm thick SiN window (Norcada Inc., Canada). Windows weremounted to kinetic pin plates91 and imaged. The spectra were calibrated against the C 1s-C 3s Rydbergpeak of CO2 gas.95 For C 1s measurements, a filter of low pressure nitrogen gas or 50 nm Ti was usedto suppress second order diffracted light from the monochromator. Optical alignment of the zone plate(Centre for X-ray Optics, 25 nm outer zone, 240 μm diameter) and order sorting aperture was performedat the start of each sample dataset. Focal depth scans were performed before every high resolutionimage.For the work using spin-cast PFSA, a Dupont D521 ionomer dispersion containing 5 wt% ionomersolution (Ion Power Inc.) was diluted with isopropanol (99.5%, Caledon Laboratory Chemicals) to 2wt%. The solution was spin-coated onto a cleaved mica substrate using a (6708D, Specialty CoatingSystems) spin coater, at 4500 rpm for 30 secs under ambient temperature and pressure. After spin-coating, the mica was cut into small ~1 mm2 squares using a clean scalpel blade. The PFSA film wasthen floated from the mica onto the surface of distilled water, previously cleaned by passing lens paperover the surface. Floating pieces of film were transferred onto SiN windows.322.16. X-ray diffractionFigure 2.10: Bragg’s law, as applied to XRD (a) Diagram of a transmission mode powder XRD experi-ment (b)2.16 X-ray diffractionX-ray diffraction permits characterization of crystalline samples at the atomic scale. Photons interactwith planes of atoms in crystals according to Bragg’s lawnλ= 2dsn(Θ) (2.3)where n is the order of the diffraction, λ is the photon wavelength, d is the lattice spacing, and Θ isthe scattering vector.When the change in pathlength for the photons hitting different parallel planes of atoms shifts thephase of the photon by 180º, the phases of the scattered photons are synchronized (Fig. 2.10). Thismeans the X-rays scatter coherently through constructive interference at specific crystallographic orien-tations. Our samples consist of many randomly oriented Pt nanoparticles (>1011crystallites), of whichsome small fraction are in the Bragg condition for each possible reflection. The result is rotationallysymmetric ’powder’ diffractograms containing rings instead of spots. Many useful pieces of otherwiseinaccessible atomic scale information can be gleaned from XRD. All materials with “long-range order” intheir electron density, including many liquids, diffract X-rays at specific angles, which depend on the size,shape, and symmetry of the crystal unit cell. These peaks produce a characteristic pattern. Amorphousmaterials, such as glasses, do not produce sharp peaks. A key difference between X-ray diffraction andvirtually every other form of spectroscopy is that the precise patterns are trivial to calculate for a knownstructure. This exceptionally strong link between theory and experiment reliably allows for the directdetermination of the crystal structure from the observed diffraction pattern.96 While this work focuses332.16. X-ray diffractionon Pt metal, for which the crystal structure is already well known, the fine structure of the diffractogramcontains rich structural information useful for characterizing nanomaterials.97In this work, the x-axis in diffractograms is often presented in units of momentum transfer (q) whenconvenient. The powder diffraction community usually reports scattering angle, denoted as 2Θ. Momen-tum transfer is a normalized parameter for elastic scattering and is independent of the incident radiationwavelength. q and 2Θ are interchangeable through the relationq=4pisn(Θ)λ(2.4)A commercially prepared gas diffusion electrode was used (Fuel Cells Etc), consisting of Pt/C (HiS-PEC 4000, 40 wt% Pt on Vulcan XC72R, Johnson Matthey) sprayed onto a carbon fibre gas diffusionmedia with microporous layer (Sigracet 29BC). The Pt loading was 0.3 mg/cm2. The electrochemicalsurface area was determined to be 25 m2/g by integrating the hydrogen desorption region of the cyclicvoltammogram.98 A BioLogic SP-250 potentiostat was used for all experiments. A large, flame-cleanedPt wire was used as a counter electrode. The leakless Ag|AgCl reference electrode (eDAQ, ET072-1)was calibrated externally against a reversible hydrogen electrode (RHE). Potential values have beenconverted with respect to the RHE. The electrolyte was ~0.25M HClO4 (70%, UHP Fisher) in ultrapuredeionized water (Elga Purelab). Electrolyte was continuously de-aerated with ultrapure argon in a sep-arate sparging flask, and recirculated through the X-ray cell compartment with a peristaltic pump. Thesolution resistance was determined by impedance spectroscopy (5.9 Ω), and IR drop was compensatedat 85% during the measurements. The cell, and all material contacting the electrolyte was cleaned bysoaking in a bath of concentrated sulfuric acid (98%, Fisher) and hydrogen peroxide (30%, Fisher) be-fore thorough rinsing with water. All electrochemical experiments were performed at 296K. Orienting theelectrolyte exit at the top of the cell helped to minimize buildup of gas generated at the counterelectrode.The long tubing (2-3 m) and thin Mylar films on the endcaps of the cell did not perfectly exclude O2 fromthe electrolyte, which caused slight ORR current even with vigorous sparging. We cannot consider thesurface of the Pt nanoparticles absolutely free of oxygen, but the quality of the electrochemistry is in linewith previous half-cell designs.78 keV photons were used, just below the Pt K-α edge in order to minimize X-ray absorption dam-age to the sample. XRD was measured using a Perkin Elmer 1621 detector and 1 s exposure times.All experiments were performed on multiple sample locations to ensure reproducibility. Whole patternrefinement was performed in GSAS-II99. For the fast time resolved measurements, the data was col-342.16. X-ray diffractionlected using 100 ms exposures using a Pilatus3 X CdTe 2M detector. The electrolyte concentrationwas increased to 1 M, and the electrode area was reduced five-fold to reduce the electrochemical timeconstant.35Chapter 3Electrolessly Deposited Pt CatalystLayers3.1 Electroless deposition of Pt films on Nafion membranes3.1.1 Overview of electroless depositionElectroless deposition is a technique for producing metal thin films. The deposition of metal films thoughgalvanic replacement electrochemistry has a long history, dating as early as 4th century BC for coinageapplications.100 In the more common electrodeposition/electroplating process, metals from a solutionare deposited onto a conductive substrate, driven by a voltage applied by an external power supply. Inan electroless process, a reducing agent reacts with a metal complex in solution to produce metallicfilms (Fig. 3.1). The reducing agent can either react directly with the metal complex (primary nucle-ation/precipitation), or the electron transfer can occur through a conductive surface (heterogeneous nu-cleation and growth).101,102Electroless deposition does not require the application of an external voltageor a conductive substrate, only the colocalization of metal ions and reducing agent.Electroless deposition requires careful attention to nucleation and growth. Primary nucleation ofFigure 3.1: Cartoon schematic of homogeneous and heterogeneous electron transfer mechanisms forelectroless deposition. M/M+ and RA/RA+ represent the reduced and oxidized forms of metal and re-ducing agent species.363.1. Electroless deposition of Pt films on Nafion membranesFigure 3.2: Free energy diagram of nanoparticle nucleation and growth predicted by classical nucleationtheory (a). The classical LaMer model of nucleation and growth of nanoparticles over time, illustratingdifferent concentrations of depositing species. Taken from6, published by Royal Society of Chemistry.metal particles in solution has a substantial activation barrier, due of the high free surface energy of verysmall particles. The redox couple for the electroless reaction is generally selected such that the reactionis spontaneous, but only slightly, such that the free energy is greater than the barrier for secondarynucleation and growth, but less than that required for rapid primary nucleation (Fig. 3.2).103 The largesurface area to volume for very small particles reduces their stability. Above the critical radius (rc),particle growth is spontaneous and largely irreversible. By manipulating the effective concentration ofPt formed in solution, nanoparticle nucleation and growth can be manipulated.This normally limits the metal deposition to secondary nucleation, and growth on pre-existing sur-faces, without simply precipitating bulk metal from solution. Electroless processes have two key ad-vantages: the substrate may be nonconductive, and the coating is extremely conformal. To producecoatings on nonconductive samples, either a trace catalytic layer of conductive material is initially ap-plied, or a small amount of primary nucleation from the bath can be allowed. Disadvantages can includea limited lifetime of the plating bath, which slowly decomposes after the reducing agent andmetal salt aremixed, as well as relatively poor control over the morphology and thickness of the deposited metal.104This chapter focuses on the electroless reduction of Pt to produce fuel cell catalyst layers. Litera-ture for electroless deposition of Pt is fairly sparse and of limited scope compared with other metals(Ni, Sn, Cu, Ag, Au, Pd). Pt and Pt alloy films were developed and used for their extreme resistanceto mechanical wear and chemical attack. No major developments for electroless processes on noblemetals have been reported in the literature since the 1970’s.105,106 Electroless Pt deposition has beensuperseded by more efficient electrodeposition techniques, and advancements in metallurgy to replacePt. The poor shelf stability of electroless plating bath solutions containing highly concentrated salts andtheir sensitivity towards contamination represent a significant expense and has limited adoption of elec-373.1. Electroless deposition of Pt films on Nafion membranestroless techniques for precious metals.107 Rising prices and the development of Ni superalloys for highperformance mechanical applications has confined Pt films to catalytic use outside the most aggressivechemical environments.108–110 The limited variety of water soluble Pt complexes commercially availableand their generally poor hydrolytic stability additionally limits the diversity of plating chemistries in theacademic and patent literature.3.1.2 Literature review and objectivesThere is extensive literature precedent of electrolessly depositing platinum group metals onto ionomermembranes.111 The first reports date from the early 1980’s in Japan, led by Torokai and Takenaka.112In the eponymous ’T-T process’, solution containing metal ions are held against one side of a mem-brane inside a clamped chamber, while the reducing agent is circulated over the opposite side of themembrane. The cationic species (either metal or reductant) diffuses through the membrane, reachingthe opposing solution, and reacting immediately at the interface. This process has been used not onlyfor fuel cell catalysts113, but also for a variety of PEM technologies including water electrolysis114, re-versible hydrogen PEMs115, direct methanol fuel cells116, methanol electrolysis, ethylene reduction117,and chlor-alkali processes118.A major disadvantage of the T-T process is the lack of control over film morphology, since the gen-erally incompatible chemistries of the metal and reducing agent must be simultaneously managed ina single step. The electroless process can be better controlled by separating the deposition into twostages: impregnation and reduction.119 This technique involves loading the PFSA membrane with acationic metal or reductant, then exposing the ion-exchanged membrane to a solution of the oppositereagent. This approach has been more successful in controlling the nucleation and growth of electrolesscatalyst films.While serving as a proof of concept, platinized membrane catalyst layers have consistently shownthree key deficiencies (Table 3.1). First, the Pt particles are of extremely large size for catalyst appli-cations, around 100 nm. Secondly, these particles are deposited without control over layer thickness,producing films several microns thick on the polymer surface. Pt below the surface is virtually inac-tive for electrocatalysis, because it cannot efficiently access gaseous reactants such as O2. Third, thenanoparticles are distributed diffusely and not electrically interconnected. Particles not connected to anexternal circuit cannot contribute to the cell current. Poor connectivity and large particle size translateinto a very low electrochemically active surface area per gram of Pt deposited. These problems arereflected in the poor catalyst layer performance that has been reported. Additionally, Pt loadings in the383.1. Electroless deposition of Pt films on Nafion membranesReference Pt deposit thickness Particle sizeSheppard et al.120 30 μm not reportedHawut and Pruksathorn121 not measured, heterogeneous 2-3 particlesKucernak and Muir122 11 μm not reportedBessarabov and Michaels123 1.5 μm Pt “islands” 300 nm longShen et al.124 unclear, delaminated duringassembly200 nm particlesMillet et al.125 10 μm not reportedMillet126 >10 μm 300 nm particles at surfaceVerbrugge127 10 μm not reportedThompson et al.128 dispersed through membrane 0.5 μm particlesTable 3.1: Selected properties of electrolessly deposited Pt catalyst layers for selected literature refer-ences1-10 mg/cm2 range are typical, comparing very unfavourably against the typical 0.5-0.05 loadings usedfor state of the art fuel cell membranes.The outlook for platinized ionomer membranes is clear. Electrolessly deposited catalyst films withsubstantially lower performance compared with Pt/C catalyst films must justify their additional cost inorder to become technologically viable. Three paths towards viability will be considered here:1. Applications less sensitive to the cost of Pt than automotive PEMFCs are excellent opportunitiesfor advanced high value materials.2. The catalytic activity or stability of the electroless Pt could be massively improved, directly con-tributing towards cell performance.3. The electroless film could offer ancillary benefits not directly associated with catalyst activity, butbeneficial for fabrication, system operation, etcFor the first objective, several applications for electrolessly deposited Pt-Nafion composite materialswere identified. Towards the second, an attempt was made to improve the catalytic activity of the elec-troless layers by shrinking nanoparticle size, improving interconnectivity of the particles, and growingcrystallographically oriented structures.The third opportunity was approached indirectly. An electrolessly deposited Pt catalyst film has manypotential advantages over conventional spray-painted or printed films. Electroless films are extremelymechanically stable and flexible, with metal particles firmly anchored into the surface. This allows cata-lyst layers to resist delamination and ablation induced by swelling, heat, pressure and impinging gasesinside electrochemical cells. These properties are especially beneficial for unconventional electrode ge-ometries, beyond the usual stacked bipolar plate configuration.129,130 Electroless films can in theory be393.1. Electroless deposition of Pt films on Nafion membranesmade extremely thin, down to atomic layers on well-defined substrates.131 In PEMFCs, this reduces thedistance that reactants and products must diffuse, potentially increasing activity and ionic conductivity.This should also allow faster water transport to and from the membrane, which stabilizes/improves fuelcell performance under high or low humidity conditions.An electroless catalyst film does not require physical support on carbon nanoparticles, which slowlyoxidize to CO2 under electrochemical potential cycling. This carbon corrosion has been identified asa major deactivation mechanism for conventional fuel cell catalysts, where Pt particles detached fromcorroded carbon quickly aggregate and ripen or become electrically disconnected.132–134 An electrolesscatalyst would be immune to such deactivation, potentially increasing catalyst stability.Because electroless catalysts are grown in the solution phase, the chemistry is far more tunablethan thermally grown Pt/C catalysts. An electroless Pt film would be solution processable, in contrast toconventional catalyst layers produced from sprayed inks. Solution processability generally offers muchmore scalable production. Finally, an electroless Pt film is thin enough that additional catalyst films canbe layered directly on top, allowing for composite catalyst layers with the potential for optimal perfor-mance under a variety of conditions. Most chemical or physical thin film deposition techniques beinginvestigated, for example ALD or magnetron sputtering, require processing under vacuum or high tem-perature. Hydrogen fuel cell membranes are structurally dynamic and possess a delicate intrinsicallyhydrated nanostructure. This mismatch of environmental compatibility means that nanostructured cata-lyst films must be fabricated separately, then carefully transferred onto membranes, creating a need forthe development of deposition techniques that are natively compatible with ambient conditions and thepolymer electrolyte chemistry.The use of membrane-bound Pt for self-humidification, or reduction in crossover-induced radicaldamage have been suggested.135 During long term PEMFC operation, dissolved Pt from the cathodediffuses into the membrane, and is reduced by crossover hydrogen, forming a band of nanoparticlesembedded inside the cross section of the PEM. The influence of this platinum-in-the-membrane (PITM)on the performance and stability of the cell has been repeatedly studied, with very little consensus.136,137The presence of the PITM has been suggested to dramatically improve the stability of the ionomerin real-world automotive environments, one of the major technological discoveries of the Whistler busprogram.138–140 Conversely, the presence of artificially implanted PITM has also been seen to seriouslydegrade the stability of the membrane.141,142The usually proposed mechanism for all effects is that the PITM serves as a recombination catalyst,lowering the quantity of H2 and O2 crossover by reacting to water it in the middle of the membrane.403.1. Electroless deposition of Pt films on Nafion membranesThe generation of H2O2 and radical species inside each catalyst layer is therefore reduced. However,these radicals may simply end up instead being trapped in the middle of the membrane where they aregenerated, causing more damage. The generation of H2O2 and radical species is strongly potentialdependent, and the mixed solution potential on the PITM surface depends on the operating conditionsof the fuel cell.137 This hypothesis accounts for the inconsistent results observed in the literature. Self-humidifying membranes would operate in a similar fashion.The push to improve performance in automotive applications has led to the use of extremely thinmembranes, in order to minimize electrolyte resistance. Current state of the art membranes are usually15-20 μm thick, down from 150-180 μm commonly used in older systems. Thin membranes are moremechanically delicate, suffer from increased gas crossover, and dry out faster when exposed to gaseswith low humidification. By adding a PITM layer, the crossover fuel gas can be put to good use, keepingthe membrane hydrated under dry operating conditions. Hydrated membranes not only have higherconductivity, improving performance, but are also far less susceptible to mechanical failure modes suchas cracking.Unfortunately, membrane chemical stability, mechanical failure, and in situ hydration are difficult tostudy, because they are very dependent on cell geometry and scaling. The thermal and mass transportphenomena observed in small, 1-5 W cells do not meaningfully recreate the environmental conditionsinside large kW scale stacks. Evidence exists that PITM could improve the performance envelope andlifetime of PEM devices, although the precise nature of the improvement remains poorly characterized.3.1.3 Platinum and borohydride solution equilibriaThe redox behaviour of Pt and borohydride species are complex, and depart from simple Nernstian be-haviour. Both the oxidation and reduction reactions are kinetically slow, irreversible, and involve specieswhich specifically adsorb on surfaces. Both participate in hydrolysis reactions which modify their reac-tivity.The Pt(NH3)4+2 ion has been proposed to undergo ligand exchange with water in solution.109,143,144Pt(NH3)42++H2O−−→←−− Pt(NH3)3H2O2+ +NH3It has been inferred by electrochemical kinetics that the tetramine species is not redox active in solu-tion.145 The reduction potential for the tetramine complex is sufficiently negative that it cannot be quicklyreduced on a carbon electrode within the electrochemical window of water, even at pH 13. However,413.1. Electroless deposition of Pt films on Nafion membranesthe reduction potential for aquo complexes is significantly higher, and therefore more reactive.146 Theseobservations are in agreement with reduction potentials calculated from thermodynamic values.47Pt(NH3)2+4 + 2e− −−→←−− Pt+ 4NH3 Eo = 0.62Vvs.SHE (3.1)Pt2+ + 2e− −−→←−− Pt Eo = 1.19Vvs.SHE (3.2)This hydrolysis equilibrium seems to establish a buffered quantity of highly reactive mono, di andtriaquo Pt species in solution. The equilibrium constant for the ligand exchange reaction is quite low,and only trace quantities of the aquo complex appear to be generated. Frustratingly, the active specieshave never been directly observed or characterized, even though Pt tetramine salts are critical for use incommercial electroplating baths. The formation constant for the coordination of Pt2+ with four ammonialigands has been reported as between 1031-1035.109,144 No evidence or justification was given to sug-gest how this number was calculated or measured. If accurate, this puts the concentration of the mostreactive free Pt2+species between 2-200 nM in our solutions, with higher concentrations of the partiallyhydrolyzed species.The electrochemical kinetics of Pt-amine reduction are complex, because the deposition of Pt onPt is autocatalytic, and can grow dendritically. This prevents a simple determination of the Keq for theligand exchange equilibrium from Levich analysis, and no mass transport limited currents can be de-tected prior to the onset of hydrogen evolution. The solution chemistry of amine substituted squareplanar Pt complexes have been studied extensively as the active species in cisplatin cancer treatment.Partially aquated Pt-amine species are deprotonated above pH 9.147 Above trace concentrations, thesecomplexes oligomerize in solution through hydroxo bridges, forming at least dimers and trimers. Swap-ping an amine for a hydroxide reduces the overall positive charge on the complex, which decreases thesolubility by several orders of magnitude.148In order to understand what species were present during the deposition, Pt-amine stock solutionswere analyzed by 195Pt NMR. NMR spectra of Pt-amine electroplating solutions have been previouslystudied, although in the presence of different electrolyte and pH.71 The Pt(NH3)4+2 ion in our solutionsappears as a peak at about -2567 ppm (referenced to PtCl62-). The precise chemical shift of this peakis sensitive to pH, Pt concentration and temperature, but is in good agreement with literature values. Noother peaks were detected in the 195Pt spectra after one hour of acquisition. The triamine monoaquocomplex should appear at -2065±5 ppm if present.149423.1. Electroless deposition of Pt films on Nafion membranesJ1 coupling to four 14N nuclei (spin = 1) produces a broad multiplet at least 20 ppm wide. In com-parison, the PtCl62- external standard yields a sharp peak approximately 1ppm wide. This broadeningdramatically lowers the sensitivity of the measurement for N-coordinated Pt species, but the chemi-cal shift window for Pt is so large that even broadened peaks will not overlap. Quadrupolar broadeningfrom 14N decreases at high temperature, so the samples were heated to 315 K. On standard commercialprobes it is not possible to observe Pt while decoupling low gamma nuclei such as 14N.By using polarization transfer it is possible to observe the Pt spectra indirectly using the more sensi-tive proton channel while irradiating the platinum nucleus (1H-{195Pt}). The theoretical signal enhance-ment from this heteronuclear single quantum coherence experiment is approximately 100 fold (γH/γPt +quenched Overhauser effect). The large chemical shift anisotropy of Pt at high fields (<7 T) substantiallybroadens this J2 H-N-Pt coupling and cancels the enhancement.150,151 For this reason, a lower fieldmagnet (200-300 MHz) produces better spectra but of course decreases sensitivity. Proton exchangebetween water and the amine moieties means water suppression is required, which further adds to theoverall complexity of the gradient pulse sequence. Further scans using the indirect pulse sequenceexperiment 1H-{195Pt} also failed to detect new signals.Based on the SNR of this spectra and previous work, it is possible to estimate an upper boundon the concentration of Pt(NH3)4-x(H2O)x2+ species to at least the low micromolar. Electrospray massspectrometry of dilute aqueous solutions in positive ion mode detected Pt(NH3)42+ , as well as largeamounts of Pt(NH3)32+ , Pt(NH3)41+ , and Pt(NH3)31+cations. The relative quantities of these ions weresensitive to the applied declustering potential, indicating that they were likely formed in the gas phaseduring desolvation, and not in solution.The aqueous solution chemistry of borohydride in the electroless process is also complex. BH4- is asufficiently powerful reducing agent such that it decomposes through water splitting in aqueous solutionsless basic than concentrated KOH. Much evidence suggests that the full eight electron oxidation ofborohydride does not participate in Pt deposition.152 Instead, transient, partially oxidized intermediates,such as BH3OH-, or BH2(OH)22- probably serve as the active species for reducing metal ions duringdeposition.153–155 Not only are these intermediate species more reactive than BH4- itself, but they areknown to specifically adsorb on noble metal surfaces strongly enough to block metal deposition.156 Theirtransient nature, tendency to form complex equilibria, and irreversible redox properties make analysisof the active species in borohydride reactions challenging.157The complex solution equilibria of both Pt-amine and borohydride greatly complicate an understand-ing of the chemical reactions occurring during electroless deposition.433.2. Controlling nucleation and growth during the depositionFigure 3.3: Acrylic cells used for electroless deposition showing open face of solution chambers (left).With Nafion sealed between the two halves, movement of bulk solution between the two is restricted(right).3.1.4 Electroless deposition of Pt on NafionPt has been deposited onto ionomermembranes by several previousmembers of the Bizzotto group.50,51The depositions described in this thesis used these previously identified reaction conditions as a startingpoint for the optimization of the catalyst layer morphology.A piece of ionomer membrane is tightly clamped between two acrylic chambers (Fig. 3.3). A solutionof Pt-amine in pH 13 KOH is added to one side of the cell, while the other is filled with a solution of pH1 H2SO4buffered with dilute K2SO4. The Pt solution is allowed to impregnate the Pt membrane for 1hr,and is then emptied and rinsed. The Pt solution is replaced with a solution of NaBH4, which reacts toform Pt nanoparticles on the surface.3.2 Controlling nucleation and growth during the depositionWhile the objective was to maximize the ORR activity and ECSA of the catalyst layer, directly measuringthese properties is not the best screening approach, because many variables contribute to the observedactivity. This would also require expensive ICP-MS measurements and time intensive electrode fabrica-tion for each sample. Instead, SEMwas used to inspect the cross section of the Pt film to simultaneouslymeasure the particle size and interconnectivity. The reaction conditions were additionally optimized forreproducibility and homogeneous coverage of Pt over the membrane, measured by eye and XRF map-ping. The optimal structure of catalyst layer was idealized as a monolayer of 2-3 nm particles embeddedhalfway into the surface of the ionomer film (Fig.3.4)To control the nucleation and growth of the nanoparticles, it is necessary to understand where the443.2. Controlling nucleation and growth during the depositionFigure 3.4: Idealized target structure for electroless deposition of Pt onto ionomer membranesFigure 3.5: Pt concentration in each solution during electroless deposition.Pt ions are during each phase of the reaction. The Pt concentration in each solution was monitoredthroughout a deposition. Previous work on the mass balance of Pt from the deposition chamber showedthe Pt species could be quantitatively recovered from the solutions and membrane. Small aliquots wereperiodically removed from each solution during the Pt soaking and deposition phase. These aliquotswere measured with ICP-OES, and the concentration of Pt in each solution was determined over time.The concentration of the Pt in solution quickly decreases during the soaking phase (Fig. 3.5). How-ever, only trace quantities of Pt were detected in the buffer solution on the other side of the membrane,regardless of soaking time. Therefore, a large quantity of Pt ions must be drawn into the membrane,but do not break through to the other side. After removing the Pt-amine solution and replacing withborohydride, the Pt concentration in this solution was quite low, similar to the acidic buffer. The lowconcentrations of the Pt in the buffer and borohydride solutions (20-100 μM) are still significantly above453.2. Controlling nucleation and growth during the depositionthe detection limit determined from matrix-matched blank solutions. Once the Pt solution is exchangedfor borohydride, the diffusion gradient for Pt ions is reversed. The Pt ions diffuse back towards theborohydride solution, where they quickly react at the interface.It is useful to review aspects of nanoparticle crystal growth and design at this stage. To producenanoparticles with the minimum possible size, the relative rate of particle nucleation versus growthshould be maximized. Unfortunately, it is nearly impossible to engineer conditions where nucleationoccurs with the exclusion of growth and ripening.158 The ideal deposition conditions should therefore bea burst of nucleation events, which rapidly deplete the surrounding diffusion media of reactants, followedby termination of the reaction. This type of process is known as the LaMer model in classical nucleationtheory.159 It is possible to bias the conditions during electroless deposition to favour this mechanism,based on first principles. LaMer type nanoparticle nucleation and growth is favoured when:1. The viscosity of the membrane should be maximized, to slow the rate of diffusion for Pt species.2. The concentration of the Pt species is kept as low as possible.3. The temperature is kept as low as possible, to minimize the rate of Pt diffusion.4. The effective diffusion rate of the borohydride solution towards Pt nuclei should be maximized,through temperature, viscosity, and concentration.5. The diffusion of borohydride into and through the membrane should be kept to an absolute mini-mum.6. The reducing agent should have the lowest reduction potential possible, to maximize the overpo-tential of the Pt reduction.7. The length of the reaction is kept as short as practicable, to minimize growth.Based on these principles, several conclusions can be drawn about which experimental parametersshould be screened. The effective viscosity of the membrane can be minimized by dehydrating it underheat and vacuum, which dramatically decreases the ionic conductivity. The concentration of active Ptspecies can be minimized by establishing a complexation equilibrium where most of the Pt is trapped inan unreducable form. Electroless Pt deposition using impregnation-reduction has been performed usingeither a cationic Pt species paired with an anionic reducing agent (Pt(NH3)42+ with BH4-), or vice versa(PtCl62-with N2H5+). The rate of diffusion in the solution is much faster than in the membrane. Therefore,the cationic Pt and anionic reductant is strongly preferred. The osmotic balance of the membrane should463.2. Controlling nucleation and growth during the depositionbe controlled such that the ionic strength of the borohydride solution is relatively equal to that of theopposing buffer. This allows water and Pt species to diffuse into the membrane, but prevents drawingborohydride deep into the membrane. The temperature of the reaction should be minimized, as hightemperatures quickly rehydrate the ionomer film.Borohydride is the strongest water-soluble reducing agent with stability longer than a few seconds.44,103Conveniently, the reaction products are fairly innocuous and soluble borates, unlike the low valent ni-trogen, phosphorous, and sulfur compounds produced by other reducing agents. This ensures thePt nanoparticles are active for electrocatalysis and not poisoned. Because borohydride decomposesthrough hydrogen evolution, it has pH dependent stability. At neutral or lower pH, borohydride ions areconsumed in milliseconds or less.160 It is necessary to prevent any diffusion of borohydride throughthe membrane by keeping a sufficiently concentrated acidic buffer on the opposite side of the mem-brane. While the diffusion of anions through the membrane is normally prohibited by charge repulsioneffects, the unusually large permeability to borohydride has been explained through the formation ofneutral contact ion pairs.161 Keeping the pH low inside the membrane also increases the local strengthof the reducing agent, by actively generating partially hydrolysed reactive intermediates at the mem-brane surface. However, if the flux of protons through the membrane is too high, then the pH of the bulkborohydride solution will be negatively effected, and the half-life of active species very short. The useof borohydride also allows for a simple reaction quench with acidic solution once the deposition is com-plete. Based on these guidelines, the reaction conditions for the electroless deposition were screened.All these factors largely influence the kinetics of Pt nucleation and growth. Selectively tuning the thermo-dynamics of nanoparticle growth is challenging. The thermodynamic parameters are contained in theNernst equation: the redox couple itself, the concentration of each species, and the temperature. Thetemperature affects both kinetics and thermodynamics. The concentration of the species in the bathsolution is easily manipulated, but this does not reflect the concentration inside the ionomer surfacewhere the reaction occurs.These variables were optimized in several rounds, with intermediate measurements of the ORR andECSA on occasional samples. Dozens of reactions were run in an iterative fashion which determined thevariable sensitivity, as well as the best synthesis conditions. This optimization was very successful, anddramatically improved the quality of the catalyst layers. It was possible to tune the deposition conditionsto yield structures ranging from large, isolated particles buried deep inside the membrane, to tightlypacked Pt nanoparticles as small as the critical radius (Fig. 3.6). For the optimized film, high resolutionSEM showed dense clusters of roughly spherical Pt nanoparticles in a band at the Nafion surface, with473.2. Controlling nucleation and growth during the depositionVariable Influence on deposit morphology Range(preferred)Temperature This controls the rate of Pt deposition, lifetime ofborohydride, diffusion, and Nafion rehydration.20-60ºC (45ºC)ReducingagentControls the rate of Pt reduction. A powerful reducingagent maximizes rate of Pt nucleation/growth, yieldingsmall particles.BorohydrideReducingagentconcentrationMust stably buffer without diffusing into membrane 0.05-0.5M(0.2M)NafionhydrationHydrating Nafion increases ionic permeability 0 - 100% RHSurfactants &SolventsAntifoaming agents increase the homogeneity of thedeposition by stopping evolved hydrogen bubbles fromadhering to the membrane.BRIJ, Triton,(PEG)PtconcentrationHigher Pt concentration produces more particles, butalso increases the layer thickness and the particle size.0.5-5 mM (2mM)Buffer pH Buffer must be strongly acidic to prevent borohydridefrom penetrating the membrane surface.0.85-1.25 (1.0)BuffercompositionBuffer cations must be composed of salts with lowequilibrium coefficient for NafionAny acid(H2SO4)DepositiontimeThis balances the rates of reactions against the lifetimesof transiently stable species.5 minutes - 1hr(12-15 min)MembranethicknessMediates diffusion and osmotic gradient, serves as poolfor Pt cations.Any thicknessTable 3.2: Optimization parameters used in developing electroless Pt catalyst thin films.only a light scattering of diffuse particles deeper into the membrane.After optimizing the deposition conditions, material very similar to the target structure could be reliablyobtained. By eye, this material appears as a dark, reflective film on the surface of the transparentionomer (Fig 3.7).The adhesion of the film is extremely good, and cannot be delaminated using the adhesive tapetest162, boiling acid, or even scratching with a blade. This mechanical durability is critical for use insidefuel cell environments. High resolution BSE-SEM images of the film cross section and from top-downperspective reveal the morphology of the deposited film (Fig. 3.8). Pt nanoparticles are embedded inthe surface of the ionomer, but only a very small fraction of the particles are buried below. From the topdown, the particles are tightly clustered into a dense network. Embedding the film in epoxy, followed byultramicrotomy and brightfield TEM allows an even higher resolution understanding of the nanoparticleshape, and particle-particle contacts. A high fraction of the nanoparticles appear to be interconnected,but the depth penetration and projection view in both BSE-mode SEM and TEM images do not allow fora decisive analysis. EDX spectroscopy and electron diffraction data are consistent with pure Pt metal(data not shown).483.2. Controlling nucleation and growth during the depositionFigure 3.6: Cross sectional BSE-SEM micrographs of Pt catalyst layers showing progress during opti-mization. Early attempts (left) have diffuse, disconnected Pt particles of large size throughout the crosssection of the polymer. After refining the deposition process, dense Pt films are reliably formed on theNafion surface (right).Figure 3.7: Appearance of the optimized electroless Pt deposition. Variation in brightness is caused bythe mirror finish of the Pt surface.493.2. Controlling nucleation and growth during the depositionFigure 3.8: High resolution BSE-SEM of the Pt-Nafion near-monolayer thin-film in cross-section (top),from top-down perspective (middle), and a brightfield TEM cross-sectionmicrotomed at a slightly inclinedangle (bottom). TEM image courtesy of Gianluigi Botton and Carmen Andrei.503.2. Controlling nucleation and growth during the depositionFigure 3.9: Particle size distribution of the three Pt films. 50-100 particles were counted for each sample.The film is composed of nanoparticles with a wide particle size distribution (PSD), approximately2-7 nm in diameter (Fig. 3.9). Particles inside the surface layer could not be measured accurately dueto overlap, so the PSD was collected using a region just below the film surface. The particles at theactual surface may not be precisely the same size, and only non-overlapping, individual particles werecounted.Unlike the highly oriented nanowire structures previously detected50, these particles do not exhibitpreferential growth. Instead, high resolution images show defects, twinning, and extensive aggregationof Pt grains, in addition to single crystal particles (Fig. 3.10).This particle size distribution and mixture of appearances is characteristic of two separate growthprocesses.163 Larger, well-spaced crystallites indicate growth through Ostwald ripening, at the expenseof smaller, metastable embryonic Pt particles. The highly aggregated structures with grains below thecritical size for bare Pt (about 2 nm) are characteristic of both secondary nucleation, or dynamic coales-cence of individual particles. Both mechanisms are well-known to occur for colloidal Pt.Isolated particles below the Nafion surface are spheroid, showing evidence of coalescence andtwinning with no clear preferential growth observed. Near the Nafion surface where particle density ishigher, a small number of the Pt nanocrystals have begun to impinge on each other and can cluster intoagglomerations of particles (Fig. 3.10).As expected, the particle size increases slightly with longer film deposition times. Images of films513.3. Following electroless deposition with in situ X-ray diffractionFigure 3.10: High resolution TEM images of the Pt nanoparticles show both extensive aggregation andcoalescence of very small Pt grains (left), as well as larger single crystal particles. Images collected byCarmen Andrei.grown for 5,8, and 25 minutes are shown in Figure 3.11. Previous results indicate that the subsurfacePt particles are slightly smaller than those at the surface, so the distribution should not be taken asperfectly reflecting the entire film. In the TEM cross section images, the Nafion membrane is in theupper right-hand corner, while the embedding epoxy is on the bottom left.The submonolayer sample exhibits an incompletely populated surface coverage of nanoparticles.The near-monolayer shows tightly packed nanoparticles forming approximately a single layer. Continueddeposition results in a “multilayer” morphology, where the Pt nanoparticle film grows deeper into thepolymer membrane, while still remaining fairly close packed.3.3 Following electroless deposition with in situ X-ray diffractionThe electroless deposition of Pt was tracked during synthesis with the use of in situ high energy XRD.The standard deposition chambers were clamped, such that the incident beam was aligned with theplane of the Nafion sheet. Diffractograms were collected continuously over the course of the reaction.In addition to monitoring the signal over time, the diffraction was collected from four locations on themembrane (Fig. 3.12). Based on some preliminary ex-situ testing, it was determined that the absolutemaximum X-ray flux must be used in order to detect Pt at the ultralow loadings (<5 ug/cm2) presentduring the initial stages of the reaction. It was therefore necessary to use a fully focused beam, yieldinga 20 μm spot size horizontally. During the deposition, the membrane swells and bulges inside the cavity.This mechanical flexing means the alignment of the beam with the membrane surface drifts over the523.3. Following electroless deposition with in situ X-ray diffractionFigure 3.11: Brightfield TEM (top row) and BSE-SEM (bottom row) cross-sections of the sub-monolayer,near-monolayer and multilayer samples. TEM images collected by Carmen Andrei.533.3. Following electroless deposition with in situ X-ray diffractionFigure 3.12: Schematic of electroless deposition cell during XRD reaction monitoring. Red dots indi-cate where the diffraction signal was monitored in cross sectional view (a). Swelling of the membranechanges the position of the Pt coated surface during the reaction (b). The coloured bars (numbered onethrough four) show the different positions of the beam. These drawings are not to scale, the curvatureof the membrane and the width of the X-ray beams have been exaggerated here.course of the reaction. To counter this, the X-ray beam was scanned repeatedly through the plane ofthe membrane.Compared with the half-cell experiments, the low Pt loading and significant background of the PMMAdeposition cells create a background signal that is challenging to subtract. Bubbles created during thereaction alter the attenuation of the X-rays, also creating instabilities in the measurement. To correct forthe background, the signal from the first time point was subtracted from each subsequent diffractogram.Then, the diffractograms were offset such that the intensity was normalized at 10º 2Θ.The diffraction signal from each of the locations are markedly different. Two positions (bars two andthree, blue and green) have very little Pt signal and are masked by changes in the background. Theregion indicated by a red bar (Fig. 3.12B, bar 1) has the highest Pt signal, but does not appear to bestable over time as the membrane fluctuates. The signal corresponding to the magenta region (bar 4)is less intense, but more stable. This is likely because the membrane drifts less near the edges whereit is clamped into position, although the beam passes through a smaller cross section of the Pt film.After an induction period with no detectable Pt, crystalline nanoparticles grow over time (Fig. 3.13).For the magenta region (bar 4), the particles are visible after approximately 10 minutes. In the unstableregion (red, bar 1), traces of the Pt signal are distinctly visible starting between 3 and 5 minutes after543.3. Following electroless deposition with in situ X-ray diffractionFigure 3.13: XRD of the 311 and 222 reflections collected during the electroless deposition (a). Fullpowder patterns (b). These transients were recorded in the magenta and red regions labelled in Fig.3.12B, respectively. Red arrows indicate the onset of reflections. Blue arrows track the evolution inapparent lattice parameter over the course of the reaction.initiation, indicated by red arrows (Fig. 3.13b). The unusually low scattering angles of the Pt reflectionsare a result of the very short wavelength of X-rays used here (0.016 pm)The surface structure and ordering of clusters or embryonic nanoparticles is unclear. The appar-ent lattice parameter of the Pt nanoparticles increases over time towards the value for the bulk metal,indicating structural relaxation during growth (blue arrows). This shift stops once the particle size ap-proaches the critical radius. Unfortunately the large pathlength of the sample combined with the driftand flexure of the membrane makes it difficult to determine if this effect is a real phenomena or opticalartifact. Ostwald-ripening models of particle growth predict large quantities of lattice strain on very smallnanoparticles, associated with high levels of surface tension.164 Previous studies using in situ wideangle X-ray scattering (WAXS) during particle nucleation and growth show mixed evidence for the exis-tence of such shifts in the lattice parameter for sub-critical nanoparticles.165–170 In situ XRD studies inwhich no lattice parameter shifts are detected have a common factor, in that they use strongly adsorbingcapping ligands, such as alkanethiols, to slow down and direct the particle growth. These ligands arelikely to alter the native surface strain of nanoparticles.The patterns in Figure 3.13 are of too low quality for Rietveld refinement. The particle size wasestimated from the peak FWHM using the Scherrer equation and comparison with simulated datasets.Particles down to approximately 0.75-1 nm in size are detected during the early stages of the reaction.Below about 2.5 nm, estimates of particle size are unreliable because the link between the size of actualparticles and coherently scattering domains are obscured by surface reconstruction, and variation in the553.3. Following electroless deposition with in situ X-ray diffractionFigure 3.14: Beam damage from X-ray exposureduring electroless deposition. SE-SEM image of thedamaged film (a) in top-down perspective. Pt coated material is on the right hand side. Photograph ofthe Pt film on Nafion membrane after removal from cell, showing X-ray damage.(b)Scherrer constant.171,172 Recalling that the Pt lattice parameter is 0.39 nm, this means that the observedparticles are on the order of 2-3 unit cells, which is a fundamental lower limit for analytical techniquessuch as diffraction which probe long range order.For radiation safety reasons, the electroless deposition could not be manually stirred during thereaction in the same way as offline experiments. Instead, the solution was stirred for an initial 30 s, andthen left to react. Due to the very precise alignment of the X-ray beam, automated mechanical/magneticstirring was impossible, as this vibrates the membrane inside the cell. The X-ray beam was turned offduring the initial 90 s of the reaction. During this time the hutch safety lockout and computer controlwas initialized. Minimizing X-ray dose before and during the initial nucleation stages of the reaction,greatly reduces the influence of the X-ray beam.173 The scattering signal from the Pt metal at the startof the reaction is also virtually zero, so there is no benefit in collecting data during this phase. Thedeposition was conducted at room temperature, which slows down the reaction and eliminated the needfor a heating bath in the beampath.It was noticed that continuous beam exposure damaged the ionomer material. After an hour of expo-sure, the membrane was noticeably thinned in the exposed region. SEM analysis showed no changesin the morphology of the Pt film from the dosed regions, although the membrane appeared corroded(Fig. 3.14a).The X-ray exposed region is visible as a horizontal line in the middle of the image. The darker, leftside of the SEM sample corresponds to membrane that was held under the edges of the cell, and no563.4. Controlling nanoparticle interconnectivity in electroless filmsPt was deposited in this location. The damage in this region appears very similar to the damage overthe platinized fraction of the sample. The damage was severe enough that the membrane was partlytorn during removal from the deposition chamber. This indicates that about one hour of continuous, fullyfocused, unattenuated X-ray beam (1013 photons per second at 73 keV) is enough to seriously destroyionomer materials, and the effects of this should be taken into consideration for future experiments,particularly during accelerated stress tests. One way to evaluate whether the X-rays influence the Ptchemistry, or merely degrade the organic polymer, is to selectively alter the Pt photoadsorption crosssection. Repeating the experiment using photons slightly above the Pt K-edge (75 vs 73 keV) producedthe same damage to the polymer, but introduced large differences in the Pt deposit versus the unexposedregions. Pt nanoparticles were rapidly and preferentially nucleated along the path of the beam, whichwere visible as a black line on the sample. These effects can be attributed to Pt K-shell photoelectronswhich are quickly readsorbed and reduce Pt ions.174 This result indicates that beam-induced chemistryis probably not responsible for the nucleation and growth seen in Figure 3.13. Ultimately, the influenceof X-ray generated radicals on Pt nucleation and growth cannot be eliminated, and care should be takenwhen interpreting in situ WAXS data.3.4 Controlling nanoparticle interconnectivity in electroless filmsA distinguishing feature of many thin film catalysts is reliance on electrical connectivity directly betweennoble metal particles. It is then essential that Pt grains form dense networks, since electrically isolatedcatalyst particles are inactive in terms of cell current. Once loaded into an electrode assembly, theionomer electrolyte can distort (swell, stretch and bend) under the heat, hydration, and pressure of anactive fuel cell. A useful thin-film catalyst layer must accommodate these mechanical stresses andmaintain good contact with both the ionomer and the electrode.175,176 A better understanding of howcatalyst particles form electrical interconnections is useful in extracting robust and reliable performancefrom thin film systems inside electrochemical cells.By carefully controlling the conditions, it is possible to assemble nanoparticle layers in the submono-layer, nearmonolayer, andmultilayer regime. This ultrathin, ultralow Pt loading catalyst is an ideal systemfor studying interparticle electrical connectivity problems and their structural influence on performance.By growing the Pt thin film directly on the polymer electrolyte, good interfacial contact can be maintainedwith both the ionomer and the electrode independent of the dynamic hydrated morphology. The localand bulk electronic or optical behaviour of the film can be directly linked to the nanostructure, which is573.4. Controlling nanoparticle interconnectivity in electroless filmsdynamic under environmental conditions relevant to electrochemical cells.3.4.1 Topography & in-plane conductivityDuring the electroless reaction, borohydride ions from the solution on one side of the membrane de-compose in the presence of acid diffusing through the membrane from the opposing solution, yieldinghydrogen bubbles in solution as a side product. These small hydrogen bubbles stick to the surface ofthe Nafion membrane if the deposition solution is not stirred. The adherent hydrogen bubbles blockthe membrane’s access to borohydride solution, and a transparent vacancy in the Pt film several mi-crons wide is generated at these locations. These features are valuable for surface analysis, becausethey represent an excellent control reference point for an undeposited Nafion film subjected to other-wise identical conditions. The environmental sensitivity of Nafion and its dependence on thermal andhydration history make straightforward comparisons of two independent samples challenging.177,178 Byperforming microanalysis on the contiguous Pt film and an adjacent bubble spot, it is possible to almostcompletely discount variability in sampling and instrument performance that routinely hinder surfaceanalysis.Backscatter SEM images do not accurately reflect sample topography, only electron density, and socannot differentiate between exposed nanoparticles and those buried just below the polymer surface.Monte Carlo simulations179 suggest that the SEM electron beam can penetrate through several nmof polymer using the lowest accelerating voltage feasible. Contact mode AFM was used to probe thecatalyst surface (Fig. 3.15). The topography of the bare Nafion (top right) is comparatively smooth, withrolling features tens of nm in size. The tip deflection data channel (top left) similarly shows no sharpfeatures. In contrast, the Pt-Nafion catalyst film shows a significantly roughened texture, with smallerclusters of particles on the surface (bottom row). These experiments were run under ambient conditionsin air.It was observed that the Pt-Nafion surface was much more hydrophilic than bare Nafion. Manypublications refer to a ’hydrophobic skin’ formed by Nafion membranes when exposed to air. Thiscan be safely inferred from an anomalously large hysteresis in water contact angle180, AFM measure-ments181,182, and observations of grazing incidence X-ray/neutron scattering measurements82. Modelspropose a change in the water channel orientation and sidechain structure within the first few nm of themembrane.A growing body of papers have used AFM to probe the surface of Nafion. AFM data extends beyondtraditional topographic imaging, and several other imaging modes are popular. Phase contrast involves583.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.15: Contact mode AFM of the film surface discerns small clusters of Pt particles. Bare Nafionfilm on one of the bubble defects (top row) has smooth rolling surface features. On the Pt film (bottomrow), small clusters of particles and significant roughening are seen. Left column: tip deflection. Rightcolumn: topography. Scale bar is 400nm.tracking the lag or shift between the input AC modulation and the output tapping of the tip. Increasingadhesion increases the phase shift, and this mode is extremely sensitive to changes in energy dissi-pation and sample stiffness. Unfortunately, the phase, measured in degrees, cannot be meaningfullycalibrated. It should be considered qualitative, and is often quite irreproducible, as polymer contami-nation on the tip greatly modifies the contrast. Interpretation of the AFM data on Nafion is decidedlynontrivial.183 McLean et al. use phase contrast imaging and claim to directly visualize the ionic domainswithin Nafion membranes at high resolution.184 Their images change dramatically as the force of thetapping increases.Phase contrast mode AFM could not easily distinguish between Pt and Nafion at the hydrated sur-face. To better understand the signals arising from phase contrast imaging, a time course experimentwas performed. AFM imaging dry surfaces is much easier than wet ones. If a wet surface must beimaged it should be completely wet in a stable fashion (i.e. submerged). However, imaging underwa-ter complicates the experimental setup for soft and delicate samples. Initially, the Pt/Nafion membranewas fully hydrated by boiling, cooled, then gently patted mostly dry and fixed to the sample plate. Themicroscope alignment was preconfigured such that images could begin immediately after only a shortapproach time (< 30s). Tapping mode images were acquired, and a single region was imaged repeat-593.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.16: AFM phase images of Pt-Nafion during dehydration.. Each frame 500 by 500 nm. Eachframe was sequentially acquired over five minutes, for a total of 30 minutes of drying time.edly over time (Fig. 3.16). Each frame represents five minutes of scanning, over which the Pt/Nafionfilm dries and equilibrates to ambient conditions. The changes are most visible in the phase contrastchannel. The Pt film slowly emerges from the hydrated Nafion.By tracking features while the surface dries, the relative distance between points can be determined.This experiment confirms that the bulk morphology of the film does not change during the drying periodbetween a mostly wet surface and one equilibrated at ambient humidity. While microscopic reorganiza-tion may occur at the length scale of the individual water channels, no large distortion of the polymer isobserved. The imaging was continued past the sixth frame shown here, with no effects visible as thefilm dried further.3.4.2 Conductive AFMCurrent-sensing AFM (CSAFM) was used to measure local in-plane conductivity of the catalyst film.An AFM tip coated with Pt-Ir alloy is electrically biased against a counter electrode in contact with thesample. 100 nm of Au was sputtered onto a masked surface of the catalyst film serving as a counter603.4. Controlling nanoparticle interconnectivity in electroless filmselectrode to allow continuous imaging over the counter electrode-sample interface. This geometry allowsprecise positioning of the tip, a known distance from the Au current collector. Current flowing throughthe tip is integrated at each pixel operating in normal contact mode. The current is quite sensitive totip-sample interactions. For example, water disrupts electron transfer when imaging in the attractivetip-sample force regime.The signal derived from CS-AFM is typically quite noisy, especially so on rough, organic, or partiallyhydrated samples. Even monolayers of ambiently adsorbed water on surfaces is sufficient to perturb thecurrent signal, just as in STM. This sensitivity to tip-sample interaction influences the design and datacollection strategy of CS-AFM experiments distinct from regular AFM. Several options for addressingthis noise exist. The simplest way to improve the quality of the current signal is to simply press the tipharder into the sample and provide better electrical contact. This obviously degrades the quality of thetopographic data and damages the sample. The second option is to reduce hydration at the surface.Imaging samples inside an environmental chamber purged with dry nitrogen was quite effective forsurfaces with no internal hydration. This passive dehydration strategy is not effective for Nafion and otherhydrated materials. A third way is to progressively collect multiple image frames. The first image cansweep the surface clean of nonconductive contamination and water, while the second scan producescleaner data. This strategy is slow, damages the sample, and works best on dirty, highly conductivesamples.Typical results from CS-AFM imaging are seen in Figure 3.17, where contamination of the tip is highlydetrimental to the conductive image quality. In both of these images sets, the tip becomes contaminated,is cleaned during imaging, and is re-contaminated again. Each time the tip is contaminated, presum-ably by polymer, the electrical conductivity is disrupted, and the current current immediately drops tozero current. Interestingly, the topography and amplitude signal are of low quality and full of horizontalstreaking artifacts where the conductivity channel is well-behaved, and vice versa. This indicates thatPt nanoparticles off the surface are likely sticking to the tip, and making a bridging electrical connectionbetween the surface and the conductive tip coating. The effects of tip contamination were mitigated byusing much higher setpoints than for regular contact mode imaging. Pressing harder into the sampleproduced nicer conductivity images, at the cost of correlative topographic measurements.The configuration of the counterelectrode-sample contact is as important as the working electrodetip. Manufacturers suggest a wire pressed into contact with the sample surface, but this presents severalissues. Contact resistance or mechanical noise between the CE wire and the sample directly compro-mise the CS-AFM signal. Secondly, a CE wire cannot be positioned closer than perhaps 1.5 cm from613.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.17: Typical CS-AFM images of two different Pt-Nafion films (top and bottom) show large prob-lems with tip contamination, visible as large dark regions in the conductivity map. Topography (left),tapping amplitude (middle), and CS conductivity (right) images recorded simultaneously. Scales areidentical to those in Figure 3.15.the tip without crashing into the nosecone, chip, or cantilever. Materials with low in plane conductivitysuffer from iR drop over this relatively large distance. Samples with heterogeneous conductivity (exactlywhen mapping is useful!) that have junctions or disconnects on this length scale would be electronicallyisolated. A low profile counterelectrode contact was developed to overcome these problems (Fig. 3.18).Contacting this counterelectrode with a ’pogo-stick’ style gold or copper wire allows excellent contactwith no electrical or mechanical noise at the junction, without interfering with the AFM measurement.Lithographically masking off part of a sample and sputtering a thick layer of gold (25-50 nm) is aFigure 3.18: Schematic of CS-AFM circuit optimized for imaging Pt-Nafion catalyst layer.623.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.19: Optical micrograph of CS-AFM tip positioned near the edge of the sputtered gold counter-electrode contact (a). The AFM tip is positioned at the end of the triangular cantiliver. Red box indicatesregion scanned by AFM. Column averaged conductivity (b) of image over Au-Pt interface. The actualconductivity map is shown in (c).convenient way to produce a well-bonded counter electrode with large surface area. After lifting off themask, this thin-film contact can then be directly scanned over, yielding a sharp interface and allowingconductivity measurements to be performed at precise distances from the current collector (Fig. 3.19A).If the gold contact is included inside the frame of the AFM image, it offers a highly conductive referencematerial to ensure the CS measurement is functioning correctly line by line during the data collection(Fig. 3.19B and C). 50 nm gold films are also starkly visible under the AFM optical microscope, so thetip can be positioned carefully at the beginning of the measurement. This avoids the time-consumingsearch required if the contact is placed on the bottom of sample, if the sputtered film too thin, or is thesame color as the sample.CSAFM imaging of the catalyst film shows a conductive surface, confirming the Pt is exposed onthe membrane surface boundary, and accessible for electrocatalysis (Fig. 3.20). The radius of thePt-Ir AFM tip is nominally 30 nm, so individual particles cannot be resolved. The large difference inconductivity between Nafion and Pt metal produces an image with almost binary contrast. Again thebubble defects serve as a reference spot, and no conductivity is seen inside the bubble regions, whiletheir periphery shows electrical contact. Local metallic electrical conductivity is recorded when usingrelatively small tip-sample biases (<50 mV). Larger tip-sample biases up to 10 V are customarily applied633.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.20: Current-sensing AFM of the Pt-Nafion surface. Topography channel shows a bubble fea-ture on the film surface (A). Conductivity map of same area (B) shows no conductivity inside the bubbleregion. Imaging the Pt at higher magnification and current sensitivity shows most Pt particles are inter-connected, forming dense, long-range conductive networks (C). (A) & (B) are 1x1 μm field of view, and(C) is 400x400 nm.in CS-AFM to improve sensitivity of the current channel. However, this leads to Pt-catalyzed faradaicreactions including water electrolysis and the ORR, which decrease image contrast through the ionicallyconductive Nafion substrate.185 The non-faradaic nature of the channel can be verified by reversing thebias during imaging.By placing the counter electrode microscopically close to the tip, it is possible to measure the lengthscale of in-plane conductivity. Thin-films can form conductive domains that may be locally connected,but globally disconnected. Fig. 3.20 was collected at 50-100 μm from the sputtered contact. Thesignal attained demonstrates that the membrane forms in-plane interparticle electrical connectivity onthis length scale, which is more than adequate for a fuel cell using standard microporous carbon films tointerface with the catalyst layer. While the precise spatial distribution of electrical contacts at this interfaceinside an operational fuel cell is difficult to probe, the carbon film exhibits surface roughness and porosityin the 0.8-1.3 μm range, with hydration dependent cracks several times larger.186,187 Unfortunately, CS-AFM analysis operates best on completely dehydrated samples. Figure 3.20A and 3.20B were collectedin a slightly hydrated environment, yielding intermittent electrical contact, while the sample was fully driedin 3.20C. Under dry conditions, at least 78% of the membrane surface registers metallic conductivity,which argues for a high degree of interconnected, exposed Pt on the membrane surface. Subsequentexperiments using higher setpoints (stronger tip-sample interactions) attempting to push through surfacewater layers on hydrated samples resulted in damage to the Nafion and abrasion of the tip’s delicateconductive coating. This imaging mode is not suitable for measuring absolute in-plane conductivity, dueto limited dynamic range, contact resistance, and amplifier noise.To place these measurements on dry films in the context of hydrated counterparts as found in func-643.4. Controlling nanoparticle interconnectivity in electroless filmsSample Resistivity, fullyhydrated (Ω/~) dry(Ω/~) Percent changeSub-monolayer Pt/Nafion 775 710 10%Near-monolayer Pt/Nafion 608 116 410%Multilayer Pt/Nafion 349 63 270%Table 3.3: In-plane four-point probe electrical resistivity of Pt-Nafion nanoparticle thin films recorded inthe wet and dry states.tional fuel cell membranes, four-probe thin-film resistivity was performed (Table 3.3). The in-plane con-duction of the samples is clearly hydration dependent, with dry films showing lower resistivity, implyinga higher degree of particle electronic interconnectivity. Because the four probes are 1 mm apart, thebulk impedance of the film is accurately represented. The conductivity of the Nafion substrate is severalorders of magnitude lower than the Pt films.3.4.3 XPSXPS was used to characterize the electronic structure of the catalyst layer surface. While normally XPSprovides elemental analysis with a slight amount of chemical information, the photoelectron spectra alsocontains information regarding the conductivity of the interface.Sample preparation for hydrated Nafion for UHV compatibility is nearly impossible without an in situcryostage. Heating Nafion samples to 85ºC under vacuum was able to remove enough water suchthat sample outgassing in the large UHV analysis chamber did not raise the pressure above 1x 10-8mbar. Unfortunately, heating Nafion to these temperatures overnight produces strong yellow-browndiscolouring, indicative of polymer decomposition. The baked out sample was also strongly adhesivecompared to the original sample. Baking out the sample for 3-4 hours allowed similar levels of UHVcompatibility with less discolouring. In all cases, several hours of intermediate high vacuum pumpingwas required to reach a vacuum of 1 x 10-6 mbar. Evenwith a defocused beam, the X-ray exposure visiblydiscoloured the samples during measurement due to the long length of the scan. Unmonochromatizedsources characteristically heat up the sample surface, and can reach >100ºC in the top layers of thesample. Decoupling incidental beam damage from induced thermal damage is beyond the scope ofthis analysis, but the X-ray source was positioned as far as possible away from the sample to minimizethermal effects.The full spectra of a Pt/Nafion sample is seen in Figure 3.21. At least 10 peaks can be assignedin the spectra. Some of these peaks (B, N, Mo, Cr) reflect the contribution of the sample plate madeof boron nitride ceramic, Mo screws, and Cr grounding wires. Because these samples were baked out653.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.21: Survey X-ray photoelectron spectrum of a Pt-Nafion film showing peak assignments. Inset:Fermi level region (bottom left) and Pt 4f peaks (top right).and prepared under normal atmospheric conditions, they also contain substantial adventitious carbonand oxygen adsorbed impurities. These elements are also present in the Nafion so the interpretationof these peaks is complex. Pt has several strong peaks visible. Many publications have deconvolutedseveral species from the Pt 4f feature of catalysts at 70-80 eV, including PtO, PtOH, and Pt(OH)2.188,189Our data does not have sufficiently high S/N or spectral resolution to discern these surface species.Shown below are two spectra of electrolessly deposited films (Fig. 3.22). The first sample (black)is the standard Pt-Nafion catalyst layer. The second sample was the same PtRu film as that used inthe TEM-EELS experiment (discussed later), which shows higher loading and a contiguous metallic filmacross the surface.Strong shifts in peaks are visible, because not all of the sample is equally conductive. When noncon-ductive samples are placed into the X-ray beam, photoelectron emission causes an excess of electronsat the surface, and a depletion of electrons in the subsurface of the sample. The work function of thesample is locally altered by this charging phenomena, changing the observed kinetic energy by a feweV. The C 1s signal is normally used for XPS energy calibration. In this case, the carbon may either beinsulating or conductive, and is not a good choice for referencing. The spectra can be aligned using thesignal from the grounded Mo screws, which should have a well defined work function for both samples.663.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.22: XPS of the Pt 4f doublet for electroless films of the standard recipe Pt (black) and of athicker PtRu layer (red) after calibration vs Mo signal.Following this correction, the Pt bands from each sample do not align (Δ3.5 eV). The 4f7/2 peak appearsat the expected 71.1 eV for PtRu sample, indicating that a large portion of the metallic Pt surface is elec-trically conductive, while in the other sample, it is not. The small chemical shift from alloying is negligiblehere. In our samples, the differential charging is reflected in broad spectral features over the low-lossregion (Fig.3.21, inset). XPS peak fitting normally considers pseudo-Voight lineshapes, with Lorentzianand Gaussian profiles from exciton lifetime and instrumental broadening respectively. The Pt 4f peakswere acceptably fit using known values for the spin-orbit coupling, peak asymmetry, and instrumentalFWHM as constraints against a linear Shirley background. The peak widths of each feature are verysimilar in both samples. This implies that most of the Pt atoms per sample collectively experience thesame work function and that the charging is distributed homogeneously over the surface, as opposedto a model with islands of locally electronically connected and disconnected Pt. Multiple scans of thesame sample reproduced the same peak positions. Both the Pt 4d and 4f orbitals shift by exactly thesame amount, which argues that the charging reached quickly reached equilibrium relative to the timescale of the measurements (8hrs). The uncalibrated analyzer sensitivity prevented an assessment forthe ’coverage’ of Pt to Nafion on the surface of the film based on relative peak areas. Careful analysisof photoelectron spectra including differential charging provides information about the conductivity of673.4. Controlling nanoparticle interconnectivity in electroless filmsProperty Electroless Pt Pt black Pt/C 3M NSTFLoading (mgPt/cm2geo) 0.024 0.04 0.020 0.05-0.15Surface area (m2/g Pt) 15 5.9 80-100 5-17Specific Activity (mA/cm2Pt) 0.48 0.9 (60ºC) 0.2 2-5.5Mass Activity (mA/mg Pt) 72 50 (60ºC) 160 200-800Reference This work 193 194,195 14Table 3.4: Comparison of electrolessly deposited Pt ORR activity with competing catalyst layer tech-nologies (dispersed on carbon rotating disc electrodes, O2-saturated 0.1-0.5 M HClO4, 0.9 V vs RHE).ultrathin films in an element selective fashion.3.4.4 ElectrochemistryIn order to utilize these Pt films for fuel cell applications, they must be laminated against a current-collecting carbon microporous layer/gas diffusion layer. Hot pressing the hydrated Pt-Nafion against thecarbon at 1000 psi and 125ºCproduces a permanently bondedMEA. These relatively extreme conditionsmay destroy the long-range morphology of the pristine Pt film when pressed against a mesoscopicallyroughened porous carbon surface. The continuous (>100 μm) lateral Pt-Pt electronic contact of theas-prepared electroless layer is likely lost. However, any in-plane damage from mechanical shearing isunimportant given the additional Pt-carbon electrical contacts established. MEAs fabricated from bothhomemade and commercially available carbon layers were hot-pressed, with no substantial changes tothe electrochemistry described below.To prove that the fully hydratedMEAs possessed electrochemically addressable Pt, CVswere recordedin a half-cell MEA arrangement.190,191 Cell designs based on compressed gaskets for electrode con-tacts and gas-tight seals were found to give reproducible surface area, but introduced some inaccuracyin catalyst activity measurements attributed to gasket sealing. Instead, catalyst activity was measuredin a modified ’floating cell’.192 It was found that by using fully dehydrated Nafion as a substrate in theelectroless deposition process instead of material equilibrated under ambient conditions (80% relativehumidity), Pt films corresponding to near-monolayer nanoparticle coverage could be attained with halfthe Pt loading (20-25 vs 50 ug/cm2). These optimized films consequently have the highest catalyticactivity for the ORR, even once returned to a fully hydrated state.Purged with argon, a clean Pt CV is achieved. Table 3.4 shows a comparison of different Pt catalystfilms. The hydrogen desorption region is used to estimate the electrochemically active surface area ofthe near-monolayer sample yielding 15 m2/g Pt. While this specific surface area is lower than in tra-ditional Pt/C catalysts (80-100 m2/g), it is in line with other highly active thin-film catalyst systems.196683.4. Controlling nanoparticle interconnectivity in electroless filmsFigure 3.23: Cyclic voltammogram of the near-monolayer Pt-Nafion film in Ar atmosphere, 50 mV/s (left).Oxygen reduction reaction (ORR) Tafel plot measured in O2, 1 atm, 20 mV/s (right). Both curves wererecorded using 0.5 M HClO4 at 294 K and IR corrected.Measurements show controllable Pt loading in the range of 0.02 to 0.1 mg/cm2 by inductively coupledplasma optical emission spectroscopy (ICP-OES). Interestingly, the gravimetric Pt surface area is max-imized for the near-monolayer film, which suggests an optimum loading for the films that minimizesparticle size while maximizing interparticle electrical contacts.197 A monolayer of large spherical parti-cles possesses much higher Pt loading than an equivalent monolayer of smaller ones. The best surfaceareas are most reproducibly attained by allowing the electroless deposition to progress until the reflectivesurface appears, then immediately quenching the reaction with acid. Because of the short total reactiontime and the rapid nature of nucleation and growth, small changes in the length of the reaction’s induc-tion period (1-2 minutes) negatively impact the reproducibility of catalyst layers deposited merely for aset period of time.The oxygen reduction catalytic activity of the optimized Pt layers was measured in an oxygen atmo-sphere (Fig. 3.23). At 0.9 V, the mass activity of the near-monolayer film was measured at 72 mA/mgPt,while the specific activity was 0.48 mA/cm2Pt. Precise benchmarking against state-of-the-art literaturedata is challenging, understanding that Pt loading, cell geometry, and testing protocol influence mea-sured catalytic activity. The electrolessly deposited film outperforms a layer based on unsupported Ptblack on a mass activity basis, where nanoparticles are not organized into a coherent interconnectedfilm, and are simply dispersed against the Nafion surface. The measured ORR specific activity is supe-rior compared to Pt particles grown on conventional carbon nanoparticle supports (0.48 vs 0.2 mA/cm2),but the mass activity falls short (72 vs 160 mA/mg, Table 3.4)). One likely cause of this performance693.5. Optical profilometrydifference is the larger particle size of the electrolessly deposited nanoparticles (2-7 nm) versus Pt/Csystems (2-3 nm). The electronic interconnectivity of the electroless Pt nanoparticles inside the assem-bled device remains uncertain. It is possible a sizable fraction are catalytically inactive as a result ofunoptimized MPL/GDL interfacial contact with the thin-film. Standard techniques to estimate total Ptsurface area utilization rely on the availability of freestanding catalyst (i.e. CO adsorption isotherms)and are not compatible with our electroless coatings. Crude estimates of theoretical Pt surface areabased on available TEM cross-sections suggest that up to 40% of our catalyst is electronically isolated,which would more than account for the low observed mass activity.The performance of the electroless catalyst inside a full MEA fuel cell under realistic operating con-ditions is difficult to predict using only half-cell or rotating disc electrode measurements.193 Oxygen sol-ubility and permeability in bulk phase Nafion is necessarily low which limits crossover between reactantstreams. Pt active sites buried inside the membrane will experience lower effective O2 concentrationsthan in the gas phase. However, it is still possible to drive the ORR current densities to above 1 A/cm2 inthe half cell for short periods of time, before water floods the porous carbon electrode. Taken together,oxygen solubility and permeability in Nafion influence the ORR activity under mass transport limitedconditions, but not in the kinetic regime (0.9 V). It is well-known that Nafion adsorption onto Pt surfacesnegatively impacts the ORR activity.198–200 Since the electrolessly deposited nanoparticles are embed-ded in the Nafion surface, it seems reasonable these thin films would respond not only to differencesin ionomer composition, but also to reorganization of the ionomer surface structure. Such structuralevolution has been long-implicated in cell “conditioning” and the discussion of internal Nafion hydratednanostructure, but sensitive in-situ probes and model catalyst layers have been heretofore lacking.2013.5 Optical profilometryOur electroless layers make excellent substrates for laser scanning profilometry because they are quitereflective. Profilometry allows us to answer one of the remaining questions regarding the in-plane con-ductivity of the electrolessly deposited layers, under native hydrated environment. Once the Nafion hasbeen fully expanded, cracks appear in the Pt film. Again, discontinuities are problematic from an elec-trical perspective. Once the film is hot-pressed against an MPL/GDL, the contiguous nature of the filmis no longer well-defined. To measure the effects of hot-pressing, a Pt film was pressed under stan-dard conditions, then carefully delaminated and imaged top-down. The Pt-Nafion film was larger thanthe MPL-GDL piece, such that the morphology of both could be measured simultaneously on the same703.5. Optical profilometryFigure 3.24: Profilometry image of an electrolessly deposited Pt film. The dark region in the upper lefthas been hot pressed against carbon and delaminated, while the bright region has not. White light fullcolor brightfield (left) total reflected laser intensity (middle) and topography map (right).sample.It is useful to simultaneously monitor a conventional brightfield color image, the total reflectedlaser intensity summed across the stack, and the topography position determined by maximum reflectedintensity. A band of Pt film that was not hot-pressed can be seen below, with the hot pressed regionin the upper left corner. Traces of carbon residue from the MPL are visible as black features near theupper left.One interesting feature of optical profilometry is that the Nafion is effectively invisible, and does notscatter UV light. This is in contrast to SEM and AFM imaging, where only the surface is visible. Inthe total intensity map, several dark features are visible. These are bubble induced defects, wherehydrogen was trapped against the growing Pt film. Even at this magnification, the scratches present onthe film before hot-pressing are relatively shallow, rare, and the film looks quite flat and contiguous. Incomparison, the hot-pressed area in the upper left has dramatically lowered scattering intensity due tomuch higher levels of roughness (Fig. 3.24)A location near the middle of the hot pressed film was also imaged (Fig. 3.25). Lower magnificationbrightfield images were collected to show the homogeneity of surface features across 5 mm of electrodearea. Long cracks are visible running down across the surface. These are also visible in the MPL, whichis difficult to image directly because of low reflectivity (data not shown). Dark, circular features are piecesof carbon black still adhering to the Pt film. Topographic images show the cracks, as well as the mottledstructure of the surface. The films have been broken up into islands approximately 3-5 μm in diameter.Note that the cracks in these images are actually raised topographic features, not depressed. This isbecause the soft Nafion material has been pressed into features in the relatively rigid carbon film.Linescans extracted from Figure 3.26 show that the Nafion membrane penetrates at least 4 μm intothe MPL film (Fig. 3.27). It is clear that the morphology of the electrolessly deposited film is alteredonce hot pressed against an MPL film (Fig. 3.28). Any long-range electronic characteristics in the film713.6. Ionomer adsorption effects on electroless PtFigure 3.25: Profilometry image of an electrolessly deposited Pt film after hot-pressing. Top row: bright-field finder frames at 5x and 10x objectives. Bottom row: True-color brightfield (left) total reflectedintensity (middle) and topography (right).Red dashed lines indicate the magnified regions of interest.as-synthesized are not necessarily present in the Pt film when used as a catalyst layer.3.6 Ionomer adsorption effects on electroless PtThe specific activity of electroless catalysts is difficult to understand and benchmark against other sys-tems because of ionomer adsorption to the Pt surface. Ionomer adsorption is a key phenomena hand-icapping the use of electrolessly deposited Pt for ORR electrocatalysis. Relatively few studies existregarding the specific adsorption of ionomer on Pt surfaces.Most of the electrically active, addressable Pt appears to be exposed on the surface. TEM crosssections do not provide sufficient contrast to evaluate to what extent the Pt is embedded or buried. Toselectively measure the fraction of subsurface Pt in the ionomer vs that exposed on the surface, a thinAu film was sputtered onto the electroless Pt. Metallic bonding to any exposed Pt sites would blocktheir electrochemistry, while deeply embedded Pt would still be exposed to electrolyte. This strategyhas been previously used to map the electrically connectivity of the subsurface Pt.51 The Au-Pt surfacewas sandwiched against a carbon GDL/MPL current collector and tested electrochemically (Fig. 3.29,left). The dramatic decrease of UPD features in the CV seems to indicate that very little subsurfacePt is electrically connected. The anodic peak at 0.5 V likely indicates that sputtering damages thesurface of the ionomer, and produces oxidizable contaminations. The sputtered gold film shows the723.6. Ionomer adsorption effects on electroless PtFigure 3.26: Profilometry image of an electrolessly deposited Pt film. Total intensity map (left) andtopography (right) for 50x (top) and 100x (bottom) objectives. Data collected over centre of region shownin Figure 3.25.Figure 3.27: Extracted optical profilometry linescan from a crack in electrolessly deposited Pt film ob-served in Figure 3.26733.6. Ionomer adsorption effects on electroless PtFigure 3.28: Topography of electroless deposited film after delamination from MPL, away from crackedregions in Figure 3.26. Image was flattened with straight line in X and Y dimensions to enhance contrast.same hydration-dependent electrical connectivity as the Pt layer, but on a larger scale that was directlyvisible with microscopy. In the SEM vacuum chamber, each flake of gold film on the surface was tightlypacked and electrically connected. After equilibrating under ambient humidity overnight, the membraneswelled, and the flakes pulled apart., which could be detected under a light microscope due to the highreflectivity of the gold surface.Sputtering Au directly onto an ionomer film in a low pressure argon plasma has the potential todamage ionomer in the surface region of the film and contaminate the Pt surface, so this electrochemicalapproach is not a completely reliable indicator of the surface exposure of the Pt film. However, it doesindicate that the subsurface Pt does not show the same electrochemical response as clean Pt surfaces.It has been repeatedly shown that adsorption of ionomer onto the Pt surface competitively displacesoxygen, lowering the ORR specific activity. In addition, the ionomer film blocks mass transport of O2,further reducing the ORR at the interface. Ionomer-free catalyst layers consistently exhibit higher ac-tivities, and the addition of even small amounts of ionomer systematically reduces their activity in allknown instances.14 The effect of ionomer in catalyst inks has been studied comprehensively by Kocha,who showed the addition of ionomer to conventional Pt/C catalyst inks reduces observed the ORR spe-cific activity by about half.202–204 The nature of the ionomer/Pt electrified interface has been studied743.6. Ionomer adsorption effects on electroless PtFigure 3.29: CV of electroless catalyst layer after sputtering 50 nm Au film onto the Pt surface (left).SEM image of sputtered gold thin film on top of electroless Pt layer collected at oblique angle (middle).Top-down optical microscope image of Au film after equilibrating sample at ambient humidity (right).through a number of techniques, including scanning probe microscopy205, Raman,206 various geome-tries of IR207–210, sum-frequency generation211, surface diffraction212, quartz crystal microbalance213,competitive adsorption214, and single crystal electrochemistry215. Taken together, some key, consistentfeatures emerge, as well as several unanswered problems. The sulfonate side chains are preferentiallyadsorbed to the Pt surface at potentials above H-UPD adsorption. The saturating coverage of the ad-sorbed layer is lower than that of bisulfate or chloride, about 0.15 of Pt sites. The adsorption of thesulfonate groups is fully reversible. Peaks related to Pt oxidation shift in the presence of ionomer, in-dicating a competitive interaction, while the H-UPD region is not affected, because the sulfonate is notadsorbed at low potentials. The assembly of ionomer films does not produce a surface structure withlong range order. The presence of ionomer on the Pt surface also shifts the ORR mechanism towardstwo electron reduction.198The single largest complication is the lack of well-defined interfaces. The CVs for Pt in the presenceof ionomer are almost completely irreproducible, which indicates that different groups in the academicliterature are interrogating different surface structures. Self-assembly of Nafion films on the surface isknown to be extremely sensitive to the preparation conditions.216–218 The potential at which the adsorp-tion/desorption occurs is not consistent, appearing anywhere between 0.4-1.1 V.Whether or not ionomerspecies are “specifically” adsorbed to the surface is inconsistent, even among publications from individ-ual groups.219,220The large concentration of impurities in the ionomer compounds render the results ofmany studies difficult to interpret.221Several studies have reported enhanced electrocatalytic performance on ionomer-supported cata-lysts.222–225 The enhancement from ionomer dispersion probably derives from differences in the prop-753.7. Electrochemical stability of electroless Pterties of the catalyst ink, and not the Pt catalyst surface. These changes in ink formulation manipulateECSA, adhesion to the RDE or membrane surfaces, stability under the measurement conditions, andvarious experimental factors, and do not appear internally consistent, reproducible, or transferable tofunctional PEM CCMs.Combined with the 2-10x lower intrinsic ORR activity from surface poisoning, the mass transportresistance of ionomer coated catalyst is dramatically increased.226 The mass transport of O2 dissolvedin ionomer is limited by solubility and rate of diffusion. These factors greatly reduce the availabilityof reactant gases on Pt surfaces buried under ionomer versus those exposed to the gas phase. Themass transport on buried Pt surfaces is complicated by the increased diffusion coefficient for oxygenin wet ionomer.227 As O2 is reduced, the generated water serves to increase O2 permeability, but alsoincreases the liquid barrier distance against further O2 transport towards the catalyst. For electrolessfilms, the slow and potential dependent adsorption of ionomer is also coupled with the O2 transport.These factors inhibit a simple model for understanding the ORR reactivity at Pt/Nafion interfaces.3.7 Electrochemical stability of electroless PtThe electrochemical stability of electrolessly deposited catalysts were investigated using acceleratedstress tests (ASTs). Two different testing protocols recommended by the Department of Energy arecommonly used to grade MEA stability.98The “catalyst” protocol involves cycling between 1 to 0.6 V at 50 mV/s. The catalyst protocol sim-ulates the potential profile of the cathode of moving from open circuit, zero power draw to maximumpower load conditions. In automotive applications this is equivalent to stepping on the accelerator, thenstepping on the brake, so this protocol is also known as the start-stop test. At 1 V, the Pt surface ispartially oxidized, while at 0.6 V, the Pt surface is reduced. Cycling between the two conditions greatlyaccelerates the degradation of the Pt nanoparticles vs steady state polarization.228 The ’support’ pro-tocol cycles between 1 and 1.5 V, which simulates cell polarization of the cathode when air is purgedthrough a hydrogen filled anode. This occurs during fuel cell startup and shutdown events. The mainphenomena associated with the degradation of the Pt nanoparticles are ripening and dissolution. Bothof these electrochemical testing profiles were applied to electroless catalyst layers in a half cell envi-ronment. It should be noted that while the DOE tests call for a 25-50 cm2 PEMFC and very specificenvironmental conditions (80ºC, flow rates, 100% RH), these are almost universally disregarded in theacademic literature. The ECSA and ORR activity after different cycles was investigated for the elec-763.8. Fuel cell testing of electroless Pt layersFigure 3.30: ORR activity of electrolessly deposited catalyst layer over the course of catalyst supportAST showing stable performance after initial cleaning. ORR activity measured at 0.9 V, after cycling at500 mV/s between 1-1.5 V.troless Pt catalyst, and a commercial Pt/C MEA. Starting with the support protocol, it was found thatthe electroless layers are not degraded. The activity during the first few hundred cycles increases asthe catalyst surface is cleaned (Fig. 3.30). Interestingly, the electrochemical trace recorded during thecatalyst stress test shows the slow reorganization of the ionomer adsorption onto the Pt (Fig. 3.31).Over the course of 2000 cycles between 0.6 and 1 V, the CV of the electroless Pt slowly shifts from thered to the purple curve. The quantity of oxide formed decreases, as well as the corresponding quantityof Pt oxide reduced. These 2000 cycles at 50 mV/s correspond to almost nine hours of continuouscycling. Adsorption of the sulfonic acid chains to the Pt surface pushes the onset of Pt oxidation tohigher potentials, in a similar fashion to bisulfate. A single cycle to 1.25 V reverts the CV to its initialstate by desorbing the ionomer. A further 2000 cycles shows an identical profile to Figure. 3.31. Becausethe H-UPD area was not negatively effected on the first cathodic scan, it is reasonable to exclude thepossibility of simple hydrocarbon contamination being responsible for these effects instead of ionomeradsorption, although these effects probably occur simultaneously.3.8 Fuel cell testing of electroless Pt layersThe optimized electroless Pt layer was assembled into a full 5 cm2 MEA. The electroless filmwas used asthe cathode, with a commercial GDEwith high Pt loading as the anode, so as not to limit the performanceof the cell. Typically, fuel cells are conditioned at high humidity, slowly increasing the current over time.Conditioning the cell sweeps away impurities, cleans the catalyst surface, and hydrates the membrane.Even under open circuit conditions, the cell experiencedmassive fluctuations in voltage (Fig. 3.32). Very773.8. Fuel cell testing of electroless Pt layersFigure 3.31: CV of the electroless layers change slowly over 2000 cycles of the catalyst AST (0.6-1.0 V,50 mV/s) in half cell, with black arrows indicating changes due to adsorption of the ionomer over time.Each color represents 200 CV cycles. Measured in half cell with 18 mm diameter working electrode in1 M HClO4.small loads (1-10 mA/cm2) also produced large oscillations. These potential oscillations are problematicfor fuel cell operation, and do not occur in high quality commercial MEAs. The oscillations may berelated to liquid water condensing on the Pt surface, reducing the activity of O2 or H2. Leaking gasketsor contamination can also cause depolarization. Measuring a polarization curve shows low performancefor the electroless catalyst layer (Fig. 3.33). The voltage was averaged over several minutes of operationat each current density.Interestingly, the shape of the polarization curve is different from that of standard porous Pt/C catalystlayers. Activation losses at low current density are observed between 0.8-0.6 V, about 100-200mV lowerthan expected for clean Pt, attributed to adsorption of ionomer to the catalyst surface. A large ohmicloss region is observed, but no clear mass transport limit is observed. This kind of linear dependenceis unusual, and likely indicates poor electrical connectivity with the Pt catalyst. The test station resultsalso highlight a limitation of thin film catalyst layers. A monolayer of catalyst particles can only containa small amount of catalyst. Therefore the power density of such a system is seriously limited, even ifthe normalized catalytic activity was higher than conventional systems. This issue is not unique to theelectroless films, and low power density is a common problem among alternatives to Pt/C, from atomiclayer deposition catalysts to systems based on nonprecious metals.229–231783.8. Fuel cell testing of electroless Pt layersFigure 3.32: Fuel cell testing of electroless catalyst layer. 5 cm2, 100% RH, 80ºC, H2/air.Figure 3.33: Polarization curve for optimized electroless film in 5 cm2 fuel cell. 100% RH, 80ºC, H2/air.793.9. Electroless deposition of Pt nanowires3.9 Electroless deposition of Pt nanowiresIt has been shown that electroless deposition of Pt on ionomer membranes can grow anisotropic nanos-tructures, in addition to the relatively spherical nanoparticles as shown in Figure 3.34.50 The fuel cellliterature shows extensive evidence that thin-film and extended surface catalyst architectures often havedramatically enhanced specific activity for the ORR, compared with regular Pt nanoparticles.232–234 Be-cause these extended surfaces show lower ECSA by virtue of their larger effective particle size, themass activity can either be higher or lower than commercial Pt, depending on the degree of specificactivity enhancement. The outstanding catalytic properties of these materials is usually credited to amixture of strain, defects, and surface crystallography.235 Some reports also show evidence of higherdurability, although this is less common.236 While the specific surface area of anisotropic nanoparticlesis by definition lower than for smaller particles, the improvements in activity and stability can more thancompensate for the lower surface area to volume. The most common strategy for producing anisotropicnanoparticles is through directed growth. The growth along one crystallographic orientation is usuallydirected by preferentially adsorbing ligands on one low index surface. This slows the growth in this planerelative to others.The electroless deposition is conducted in the presence of adsorbing molecular species, but nonethat are typically invoked as structure directing agents. O2, H2, borohydride and/or hydrolysis intermedi-ates, Pt-amine, and the ionomer are all known to adsorb to the catalyst surface. It is suggested that theionomer membrane itself may template the growth of the Pt nanowires/rods. The length and bundlednature of these rods is reminiscent of several model structures for the hydrated morphology of ionomermembranes.26,82 In these models the internal phase segregation produces hydrophilic domains withbundles of rod-like inverted cylindrical micelles. From this perspective, the Pt appears to grow alongthe water channels present inside the ionomer. The structure directing force for this reaction could beinhibition due to adsorption to the walls of the channels, or preferential mass transport of Pt salts downthe channel towards the ends of the rods. Many of the rods are single crystals, indicated by contiguousatomic lattice fringes visible down the length of the structures.The electroless deposition of the nanowires was extremely inconsistent and irreproducible. Dozensof attempts to grow Pt nanowires were made, without converging on a reliable protocol. Replicate de-positions conducted on the same day, using the same reagents and glassware, and nominally the sameprocedure yielded nanowires in less than 10% of trials. Because of this inconsistency, it is not pos-sible to conclusively identify the factors responsible for their growth. However, several trends were803.9. Electroless deposition of Pt nanowiresFigure 3.34: Brightfield TEM cross-section of Pt nanowires electrolessly deposited in Nafion membrane.(a) A bundle of nanowires buried beneath the surface. (b) and (c) higher magnification images showingthe lattice fringes characteristic of a Pt single crystal. Images collected by Brad Ross.noticed which shed light on the mechanism. Depositions resulting in nanorods visibly appeared as adark film much more slowly than depositions producing spherical nanoparticles. This is consistent witha model of slower, extended nanoparticle growth, compared with the LaMer burst nucleation-depletionpreviously discussed. The nanowires grew more successfully using Nafion membrane which had beenallowed to adsorb ambient moisture from the laboratory environment for extended periods of time (i.e.six months). Unfortunately, the hydration of the membrane controls many variables related to the particlegrowth. The membranes which contained nanorods often contained few spherical nanoparticles, whichappeared only deeper into the membrane (Fig. 3.35). This observation could reflect changing transportphenomena of the reagent species depending on the depth into the membrane, or simply that the sur-face region of the membrane adopts a different morphology than the bulk because of factors such ashydration. The thickness of layers producing nanowires were always much larger than those producingsmall, round particles. The Pt nanorods were always seen in a highly clustered form, compared withthe round nanoparticles, which usually appeared dispersed. The sample shown in Figure 3.34 werefabricated using conditions described in reference50, while the rods shown in Figure 3.35 were grownusing the same conditions described in the methodology section.Electron diffraction of the nanorods showed that they likely grew preferentially along the 100 axis(Fig. 3.36). This was determined by analysis of the relative peak areas after radial integration. The200 reflection (the second peak at approximately 90 pixels), is over twice as large as expected forkinematic electron diffracton off an fcc lattice, while the ratios for the 111, 220 and 311 reflections areof the appropriate size. This requires the assumption that the particles are randomly oriented, which isreasonable based on the lack of any visible preferred orientation in the images.The Pt loading of the sample in Figure 3.35 was determined as approximately 0.102 mg/cm2. The813.9. Electroless deposition of Pt nanowiresFigure 3.35: Representative brightfield TEM cross section of electroless depositedmembrane containingnanowires. Low magnification image showing the homogeneity of the layer (left). Medium magnificationimage showing themorphology near themembrane surface (middle). Highmagnification image showingthe morphology of the individual nanorod clusters (right). Images courtesy of Gianluigi Botton.Figure 3.36: Electron diffraction image of the nanorod sample (left). The lookup table is logarithmic toenhance visual contrast of the rings. Integrated radial profile of the diffractogram (right). Image courtesyof Gianluigi Botton.823.9. Electroless deposition of Pt nanowiresFigure 3.37: CV of nanorod sample in half cell, purged with Ar gas, at variable sweep rates (left). Theintegral of the H-UPD desorption feature was plotted as a function of scan rate (right). 298K, 0.5MHClO4film was prepared for electrochemical testing as described previously. The surface area of the Pt filmwas measured by integrating the H-UPD feature (Fig.3.37). The precise integral depended on the scanrate, even using iR compensation, converging towards a maximum at slow scan rates. The ECSA of thenanorod sample was 18 m2/g, as measured in the half cell. The larger particle size and the particlesdistributed deeper into the membrane do not appear to reduce the ECSA of the nanorod sample relativethe regular electroless films (15mm2/g). This is attributed to the bundled morphology of the nanorods,which likely increases the local interconnectivity of particles near the surface.The half cell was purged with O2 and the ORR activity was measured. The specific activity wascalculated at approximately 0.7 mA/cm2. This is about 40% higher than for the regular electroless films.The gain in specific activity for this nanorod catalyst is attributed to surface crystallography effects.While it is not trivial to assess the precise distribution of exposed surface crystallographies, the H-UPDregion of the CV did not exhibit strong features indicative of 100 or 110 faceting.237 Together with theelectron diffraction results, it is reasonable to assume that the exposed nanorod surfaces are enrichedin Pt(111). The relative activities of different surface crystallographies for the ORR is complex, anddepends very much on the adsorption of electrolytes.33 Nevertheless, up to hundred-fold differences inORR activity have been observed between different crystal orientations using Pt single crystals, so themodest enhancement in specific activity for the nanorods is very reasonable.833.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activity3.10 Electrolessly deposited Pt alloys and their hydrogenoxidation activityThe current 10x lower Pt loading on the PEMFC anode reduces the drive to replace expensive PGMcatalysts on this electrode. Instead, the largest technical problem for the anode is tolerance towardspoisoning. At the high potentials experienced on the cathode, small quantities of gaseous organic im-purities are quickly oxidized. On the anode, impurities are not so easily consumed, and are more likelyto adsorb and persist on the Pt surface, poisoning the catalyst. Poisoning from hydrocarbons is espe-cially problematic for cell designs powered by impure hydrogen generated through steam reformation ofhydrocarbons, or from raw hydrocarbon fuels, such as in the direct methanol fuel cell. Reformate andmethanol fuels have been largely abandoned in the automotive sector . However, these alternative fuelsare extremely attractive for stationary PEMFC applications, such as backup power generation.The vast majority of hydrogen generated today is derived from gasification of hydrocarbons, and isco-produced with large quantities of carbon monoxide. Rigorous purification of the H2 stream is possiblebut adds towards the final cost of the fuel. For these reasons, catalysts which preserve the naturally highHOR activity of Pt while exhibiting “CO-tolerance” are highly desirable. Pt alloys have been the mostsuccessful strategy for CO tolerant anodes to date, especially with ruthenium. Engineering solutions forCO tolerance, such as bleeding air into the fuel stream are also practiceable. The surface chemistryresponsible for providing CO tolerance on PtRu surfaces has been extensively investigated over the lastfew decades. The electrochemical oxidation of CO is proposed to follow a classic Langmuir-Hinshelwood“bifunctional mechanism” (Fig. 3.38).Pt+CO −−→ Pt−CO (3.3)R+H2O −−→ ROH+H+ + e− (3.4)Pt−CO+ROH −−→ Pt+R+CO2+H+ + e− (3.5)CO and oxygenated species are both co-adsorbed on the surface, and react spontaneously at suffi-ciently oxidative potentials, generating CO2. The Pt/C catalyst is prepared as a random alloy with a lessnoble metal. The alloying metal forms surface oxides at lower potentials than bare Pt surfaces (< 0.85843.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.38: Cartoon schematic of bifunctional mechanism for CO oxidation on a PtRu surface.V vs RHE). For Ru, oxygenated species are electroadsorbed from acidic aqueous electrolyte as low asabout 0.4 V, while on pure Pt no OH species are adsorbed until approximately 0.7 V. The precise detailsof the oxidation are highly dependent on the microstructure of the surface, the CO coverage, and theelectrochemical history of the electrode.PtRu films were electrolessly deposited onto ionomer membranes to produce CO tolerant HOR films.The electroless deposition of PtRu has been demonstrated previously by several groups.238,239 Thedeposition mechanism of the PtRu films was somewhat different than with the pure Pt. The only modifi-cation to the procedure was the addition of small amount of R(NH3)6Cl3 to the Pt solution during theinitial soaking phase of the electroless deposition. Pt (II) and Ru (III) salts are not stable in solution atpH 13, and result in the slow precipitation of PtRu metal over several minutes.R(NH3)3+6 + e− −−*)− R(NH3) 2+6 Eo = 0.2Vvs.SHE (3.6)R2+ + 2e− −−*)− R Eo = 0.45Vvs.SHE (3.7)The oxidation half reactions for this process have not been determined at this time.240 Ru (III) speciesare not stable at high pH and display complex redox chemistry in solution, including disproportionationthrough the wide number of oxidation states accessible to Ru.241 The Pt and Ru ions impregnated inthe ionomer membrane are rapidly reduced by borohydride. Unlike the pure Pt deposition, the unstablenature of the PtRu bath generates metallic particles in solution, which adhere to the surface of the mem-brane. These surface-adhering particles act as seeds for the nucleation and growth of a nanoporousPtRu film. The top layer of the PtRu film is not mechanically durable, and can be scraped off the mem-brane relatively easily. TEM of these detached particles show how they grow into larger flakes of aggre-gated particles (Fig.3.39). The quantity of Ru in the film could be tuned simply by changing the ratio of853.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.39: Brightfield TEM of electrolessly deposited PtRu nanoparticles (left) and EDX spectrum ofPtRu film showing peak fits for the Pt M edge and Ru L edge (right). Red filled regions correspond to theexperimental data, with dark and light blue lines representing the background and modelled elementaltransitions respectively.Pt and Ru precursors used in the synthesis. SEM-EDX analysis of the optimized film shows Pt and Ruin a near 50/50 atomic ratio, in agreement with ICP-OES quantification. This ratio is optimal to facilitatethe bifunctional mechanism shown above.242Good alloying between the Pt and Ru is critical for the bifunctional mechanism to occur. Whenelectrodepositing metals with standard reduction potentials several hundred mV apart, segregation oftenoccurs. The result is a heterogeneous admixture of particles composed of metal A and particles ofmetal B, instead of alloyed particles with composition AB. These admixtures of two metals are virtuallyinactive for CO oxidation, because only neighbouring Pt and Ru atoms can participate in the cooperativeoxidation.243 Compositional variation in the metal film is commonly observed, even when alloys areproduced.244,245 This problem is especially pronounced with electroless deposition.238 Weak reducingagents and slow plating rates accentuate segregation, because the relative rates for the reduction ofthe two metals are most different at potentials closest to the Nernstian equilibrium, as predicted by theBulter-Volmer relationship. Alloys are commonly electroplated at large negative voltages for this reason,where the deposition rates are mass-transport limited. High-angle annular dark field imaging (HAADF)combined with electron energy loss spectroscopy (EELS) was performed in a scanning transmissionelectron microscope (STEM) to examine the dispersion of the two metals in the catalyst film (Fig. 3.40).In the low resolution dark field image, the nanoporous character of this catalyst is easily discerned.Various small pores 5-10 nm in diameter are distributed evenly throughout the material. In the simul-taneously acquired EELS images, the separate Pt and Ru images, as well as the merged image showa homogeneous distribution of Pt and Ru throughout the flake of catalyst film. Unlike EDX, which islimited to about 2-3 nm spatial resolution, electron energy loss spectroscopy has atomic resolution for863.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.40: Low resolution HAADF (bottom) and extracted EELS intensity maps (top row) of a flakefrom the surface of the PtRu catalyst film. Data collected by Andrea Korinek.873.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.41: High resolution HAADF (bottom) and EELS images (top row) of a flake from the surface ofthe PtRu catalyst film. Arrows indicate identical locations as a guide to the eye. The box indicates theregion of interest mapped with EELS. Data collected by Andrea Korinek.elemental mapping. This allows for precise spectromicroscopy on the exact spatial distribution of Pt andRu inside individual crystallite grains (Fig. 3.41).High resolution EELS images indicate that the PtRu is almost completely homogeneous on the atomicscale. The probe size used in this image is approximately 0.1 nm. Several grains can be detected whichinclude enrichment of the Ru, indicated by arrows. Quantitative comparisons are precluded by the lowsignal to noise of the Rumap, which has a broader, lower intensity edge in the EELS spectrum comparedwith Pt. If galvanic exchange were occurring, then the surface of the particle would be enriched in Pt,while the middle of the grains would be enriched in Ru. The greyscale intensity of HAADF STEM imagesis directly proportional to electron density of the sample, and the signal to noise is much better than forthe EELS image. Pt is much more electron dense than Ru (21 vs 12 g/cm3). Therefore, particleswith Pt enriched surfaces would exhibit bright outlines in HAADF frames. These structural features are883.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.42: High resolution HAADF (bottom) and EELS images (top row) of a flake from the surface ofthe PtRu catalyst film. Data collected by Andrea Korinek.893.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.43: CV of electroless deposited Pt and PtRu films under nitrogen atmosphere. 250 mV/s, 295K, 0.5 M HClO4.not detected, especially in the second image where individual, non-overlapping grains are visible (Fig.3.42).Electron diffraction detected only the fcc metallic phase with no bcc component consistent withrandom alloying (data not shown).The electrocatalytic properties of the PtRu films were characterized using CV, linear sweep voltam-metry, and chronoamperometry in a half cell environment. It is useful to begin with the baseline CVsof Pt and PtRu under inert N2 in acidic media (Fig. 3.43). Compared with the regular Pt CV, the PtRucatalyst has two major differences. First, the H-UPD region is suppressed, as this reaction does notoccur as readily on Ru surfaces. Second, the potential onset of PtRu oxidation begins much earlier thanin the pure Pt film. The potential was not swept as high for the PtRu film because leaching of the Ruoccurs quickly at high potentials.Next, CO stripping voltammetry was performed. In this experiment, the electrode is held at a lowpotential, around 0.1 V, for perhaps 15 seconds to reduce any oxides present on the surface, whilepurging the cell with N2 gas. Then, the surface is exposed to a CO atmosphere. In this case, 1% COin an argon diluent was used for safety and convenience. The cell was filled with the CO mixture forbetween 1-5 minutes, then the CO was purged out of the cell using pure N2for 2 minutes, all whilekeeping the electrode under continuous potential control. Then, a linear sweep voltammogram was903.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.44: CO stripping voltammetry for the electrolessly deposited Pt and PtRu catalyst layers. 20mV/s sweep rate, 295 K, 0.5 M HClO4.collected, scanning the potential from 0.3 V towards the upper potential limit used in the previous CV(0.81 V for PtRu, 1.25 V for Pt) (Fig. 3.44).In the CO stripping voltammogram, the effect of the Pt alloy is clear. The peak potential for COoxidation is shifted lower in potential from 0.77 V on the pure Pt, to 0.53 V on the PtRu (a change of-240mV). This sweep results in the oxidation of the CO in the form of one or more peaks. The CO strippingis an extremely sensitive probe towards the surface structure of nanoparticles, and a detailed review ofthe peak shape, peak potential, and repeatability of this reaction on even pure Pt surfaces fall beyondthe scope of this discussion.41,246–251This test confirms that the alloy surface has been successfully prepared, and that the early onset ofCO oxidation occurs. However, this property is not a direct measurement of CO tolerance in a fuel cell.Most studies conclude CO tolerant fuel cell operation simply from the peak potential of CO stripping.This approach ignores the complex, competitive adsorption of H2, OH, and CO at the surface.35,252 Forthis, actual HOR measurements in the presence of CO were performed (Fig. 3.45).A simple polarization curve was measured after purging the half cell with H2 gas, and after cyclingwith 100 ppm CO present in the gas stream. CO impurities in technical grade H2 typically lie in the913.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.45: HOR CV of Pt and PtRu films in the presence of H2 gas with 100 ppm CO. 295 K, 0.5 MHClO4.10-100 ppm range. The PtRu film shows a small decrease in the HOR current density after cycling inthe H2/CO stream for several minutes. Most of this change occurs in the large overpotential region,while near 0 V the HOR activity is nearly identical before and after CO treatment. The pure Pt sampleshows significant poisoning of the catalyst surface. on the third CV cycle, the HOR activity has alreadydropped an order of magnitude. While cycling in mixed H2/CO is useful for determining the CO toleranceof cells, the dynamic restructuring of the Pt surface, and the strong dependence on the upper potentiallimit complicate the analysis. In a real operating cell, the anode is polarized to an effectively constantpotential. Under these conditions, the CO adsorption/poisoning, CO oxidation, and hydrogen oxidationexist in dynamic equilibrium on the nanoparticle surface. These more realistic steady state conditionscan be simulated using chronoamperometry in a half cell. In this experiment, the cell is kept polarizedin the mass-transport limited range for HOR (0.3 V) in a stream of pure H2. Then, 100 ppm of CO isintroduced, and the current is recorded over time (Fig. 3.46).The Pt sample loses almost all measurable HOR activity as the CO adsorbs to the surface and blockscatalytic sites. This process is not instantaneous, and requires several minutes to reach saturatingcoverage on the nanoparticles. The PtRu sample still exhibits a large degree of poisoning becausethe concentration and rate of adsorption for OH species is low at 0.3 V. The HOR is suppressed to923.10. Electrolessly deposited Pt alloys and their hydrogen oxidation activityFigure 3.46: Steady state HOR chronoamperometry of the Pt and PtRu films after introduction of 100ppm CO into H2 stream. The potential of both electrodes was held at 0.3 V.933.11. Alternative applications: optical reflectivity and sensingapproximately 40% of its activity in pure H2.These experiments highlight perhaps the most substantial and hitherto unrecognized advantage ofhalf cell electrochemistry: the ease of using gas mixtures. In traditional liquid electrolyte cells, gaseousreactants are usually introduced as saturated solutions produced by gas sparging. Aside from a fewflow-cell designs with on-line blending capability,253 it is effectively impossible to introduce variable con-centrations of gaseous species to the electrode surface. The production of liquid electrolytes containingmultiple dissolved gaseous species (i.e. H2 and CO) are even more challenging. With a half cell, con-trol over flow rates and gas composition are trivial. Blends can be easily produced with high accuracyand arbitrary composition in a continuous fashion using mass flow controllers, or as in our case, usingpre-blended tanks from manufacturers. Half cells further benefit from the use of independent referenceelectrodes. In a full MEA configuration, contaminating species added to either cathode or anode streamscan easily influence both electrodes through crossover. This precludes most meaningful electroanalyt-ical techniques, since the potential of both electrodes is now poorly defined. In half cell environments,a properly isolated reference electrode (e.g. a double junction Hg|HgSO4electrode) prevents any ad-verse effects of the contaminating species on the potential control of the working electrode. A betterunderstanding of electrode behaviour in the presence of trace contaminants will be critical as fuel celltechnology transitions from high-purity laboratory environments towards low-cost, real world applica-tions.254–257The electroless deposition of Pt onto ionomer membranes has been extended to PGM alloys, in-cluding PtRu. The structure of these random alloys was seen to be extremely well dispersed, which isunusual for an electroless process. The PtRu alloy demonstrated early onset CO oxidation, consistentwith the electrochemistry of conventional, carbon supported catalysts which normally require annealingat >700ºC. The use of a half-cell allowed for convenient, steady state measurements of simultaneousCO and H2 oxidation.3.11 Alternative applications: optical reflectivity and sensingIn an effort to develop alternative applications beyond fuel cell catalysts, several properties of the com-posite films were explored. The optical properties of the catalyst layer are directly linked to the electronicproperties of the film. At the macroscopic scale, the Pt-deposited membrane appears as a dark brownto grey film when hydrated, and is metallically reflective in air (Fig. 3.47, top).As particles form locally conductive networks, electron delocalization over these domains increases943.11. Alternative applications: optical reflectivity and sensingFigure 3.47: Images of platinum thin films under ambient conditions after depositing sub-monolayer,near-monolayer, and multilayer Pt nanoparticle films, demonstrating the evolution of the mirrored sur-face. ICP-OES and electrochemical surface area measurements from each film were performed intriplicate using a half cell, bars represent average +/- range.953.11. Alternative applications: optical reflectivity and sensingFigure 3.48: Transmission UV-Visible spectra of hydrated and dehydrated near-monolayer Pt nanopar-ticle film demonstrating environmentally switchable optical density.optical reflectance.258 Films with sub-monolayer particle coverage are brown to black, and show highoptical density due to small Pt particle size and strong Rayleigh scattering. Once the Pt film has grownto near-monolayer coverage, a grey highly reflective surface evolves abruptly. Further Pt deposition andfilm growth yields an opaque and reflective coating. While the near and sub-monolayer film are quitesimilar in loading, the multilayer film was allowed to react for longer, yielding a thicker Pt film. The ECSAis lowest for the multilayer film because the particles are slightly larger.Films at the boundary of the particle/continuous-film percolation threshold show switchable reflectiv-ity with hydration (Fig. 3.48). Simple transmission mode UV-Vis spectroscopy is a convenient proxy forreflectance, which is challenging to measure directly on wet surfaces. Reflective contributions of wateror optical quartz from the environmental cell are minimized when measuring optical transmission nor-mal to the surface. The similar refractive index of water and Nafion (1.33 vs 1.34)259 maintains constantnanoparticle light scattering intensity as the film hydrates and dehydrates. These two factors ensurechanges in optical density (absorbance) are only due to changes in sample reflectance as the poly-mer absorbs water. Reversible reflectance requires that percolating particle networks are dynamicallyequilibrating while the Nafion polymer electrolyte adsorbs water and expands and contracts by roughly963.11. Alternative applications: optical reflectivity and sensingFigure 3.49: Specular angular reflectance curves of the high, medium and low loading Pt films usings-polarized (left) and p-polarized (right) light at 650 nm. Data courtesy of David Troiani and AgilentTechnologies.10%.260 This argues for a catalyst layer model wherein platinum forms strong anchoring interactionswith the Nafion support, featuring electronic contact between proximal independent particles rather thana film with continuous interparticle metallic bonding. While all the films exhibit switchable reflectancebehaviour, the effect is naturally the strongest in the near-monolayer samples closest to the nanopar-ticle percolation threshold. Absolute specular reflectance measurements show the expected increasein reflectance with increasing Pt deposition. P-polarized angular reflectance curves display a Brewsterangle tunable between a Nafion/air and Pt/air interface (Fig. 3.49).The optical reflectivity of the three Pt films was measured after allowing hydrated films to equilibrateunder ambient laboratory conditions for several weeks. The reflectivity of sputtered Pt films is 65-90%for s-polarized light at 650 nm. For a p-polarized incident light, Brewster angles are indicated with dottedlines. Broadening of the illuminated spot at high angle of incidence precluded the precise determinationof Brewster angle for the high loading (multilayer) film. The dashed black line represents the polynomialfit and extrapolated Brewster angle for this sample. The Brewster angle for a Pt-air interface is 78degrees, and approximately 52-55 degrees for Nafion-air under these conditions.Reflectivity spectra for the three electrolessly deposited films were collected using s-polarized lightat a 65 degree angle of incidence. Also shown is the calculated spectrum for a 10 nm monolithic Pt film,neglecting thin-film interference effects (Fig. 3.50). Samples were taped to a matte black holder andgently pressed flat after soaking in hot water to minimize curling while drying. The small step artifactsin the experimental curves at 720 and 320 nm correspond to filter changes in the spectrometer.973.11. Alternative applications: optical reflectivity and sensingFigure 3.50: S-polarized absolute reflectivity spectra at a 65 degree angle of incidence. The Pt standardwas calculated from literature data. Data courtesy of David Troiani and Agilent Technologies.The continuous change in Brewster angle suggests the film refractive index can be explained viaan effective medium approximation, where any localized particle network structure or ’islands’ in thehydratedmaterial must appear small and spatially averaged compared to the wavelength being reflected.Measuring specular reflection as a function of incident wavelength effectively changes the length scale ofthe electron delocalization phenomena. The near-monolayer sample shows significant departure froma bulk Pt film spectra261 with maximum reflected intensity at 400 nm, implying that the hydrated Pt-Nafion forms partially segregated electronic domains a few hundred nanometers in size. However, thehigher-loading, multilayer Pt film exhibits a similar reflectivity curve to bulk Pt, indicating longer-rangeelectronic delocalization characteristic of metallic surfaces. Unfortunately these spectral features arecomplicated by nanoparticle diffuse reflectance and the imperfect flatness of the mirror surface. Whendried under ambient conditions the Nafion films tend to slightly curl and wrinkle even when pressedflat. This produces deviation in the reflected beam angle (visible as small ’steps’ in the spectra at filterchanges), as well as uncertainty in the exact incidence angle and resulting reflectivity. The high fillfactor, small particle sizes with wide relative size distribution, and discontinuous nanostructure of theNafion preclude simple structural interpretation via standard Maxwell-Garnet and Bruggeman compositeeffective media models.262,263 Understanding the complex electronic structure of hydrated nanoparticlelayers is important not only for thin-film catalysts, but for ionic polymer metal composites in general.These metallized electrolyte composites find application in a variety of MEMS, sensing, and flexible983.11. Alternative applications: optical reflectivity and sensingelectronics platforms that exploit their unique optical, conductive, and mechanical properties.264–266The large change in optical density, combined with the broadband nature of the shift make thesefilms ideal optical reporters for humidity and hydrogen sensors. Humidity sensors based on Pt-Nafioninteractions can use either the change in optical density, the change of in-plane electronic conductivity, ora combination of the two. Electrolessly deposited Pt-Nafion films have been previously reported for thisapplication.267While a number of alternative technologies are available for on-line humidity sensing, thePt-Nafion platform has the advantage that the surfaces are chemically non-reactive and reversible. Thisallows, for example, humidity measurements of highly corrosive HF, HCl, or even Cl2 vapour streams,which destroy the sensing elements of other electronic or colorimetric sensors. Humidity sensing withPt-Nafion films exhibit a nonlinear sigmoidal response, due to the percolation threshold. By tuning thePt loading, the percolation threshold of the Pt film can be altered, allowing for high sensitivity to changesin membrane hydration over a small window, analogous to a pH indicator. Additionally, the nature ofthe internal Nafion nanostructure allows these sensors to perform very well at the high and low ends ofthe relative humidity range (>90% or <10%), where typical sensors are unreliable The largest problemwith previously reported humidity sensors based on electrolessly deposited Pt-Nafion films is the slowresponse time. In these reports, lack of control over the electroless deposition necessitated a very thick,deep Pt film. The entirety of the membrane needed to hydrate and dehydrate, producing a sluggishresponse. The ultrathin films developed here would allow for very fast equilibration of the interface, andfaster electrical response.A large number of designs exist for hydrogen sensing.268 Most are employed for leak detection, es-pecially in fuel cell systems. Operating temperature range, and stability against interferences are themost challenging aspects of sensor design. Nafion membranes with electrolessly deposited Pt on bothsides have been extensively used to detect H2 gas in a concentration cell geometry. As with the hu-midity sensors, poor control over the Pt deposition results in a sensor with unstable and irreproducibleresponse. The ability to produce high surface area Pt in a thin film with good control over particle inter-connectivity is specifically called for as a necessary improvement in previous designs.269,270 Althoughthere is some opportunity to develop the electroless deposition of Pt for hydrogen detection, currentlyexisting sensors already perform quite well at low cost.The most significant application for Pt-Nafion devices is within the MEMS field.271–273 Ionic polymermetal composites (IPMCs) are some of the most promising flexible electronic devices, and can be usedas actuators or robust mechanical sensors. As shown in Figure 3.51 when plated on both sides, theelectroless layer behaves as a supercapacitor. Application of a voltage across the membrane results993.11. Alternative applications: optical reflectivity and sensingFigure 3.51: Actuation of electroless Pt as an IPMC, showing cartoon schematic of mechanism, deviceconstruction, and actuation.in the bulk movement of ions and solvent towards the polarized interface, and generates mechanicalactuation.Sputtering a thin film of gold over top the electroless Pt layer improves the lateral conductivity andactuation capability, as well as protecting against dehydration. . While the characterization techniquesused here could easily be applied towards these devices, the optimal morphology of the electroless filmis quite different than for fuel cells. For sensors and IPMCs, the cost of the Pt itself is not a significantfactor in the cost of the device, only about 10 cents per square centimetre per μm thickness. The Ptparticles must have high surface area and interconnectivity, but need to be deeply embedded within themembrane, instead of exposed at the surface. Typical IPMC fabrication uses a short round of electrolessprimary nucleation and then secondary nucleation and growth to improve the electrochemical intercon-nectivity of the existing primary nanoparticles. For maximum mechanical stopping force, the Pt film andthe ionomer membrane must be as thick as possible, with typical nanoparticulate Pt films between 1-20μm thick.274 The use of ultrathin Pt films reduces the total surface area of subsurface Pt, which reduces1003.12. Patterned electroless depositionFigure 3.52: Photolithography of electroless layers. Impregnation of membrane with Ag+ and exposureto shortwave UV (A). Ag seed nuclei are left exclusively where the mask did not cover sample (B).Rinsing away the unreacted Ag+ ions (C). Galvanic displacement of silver metal with Pt2+ ions (D).actuation capability.3.12 Patterned electroless depositionWhile sensor and IPMC devices do not benefit from careful control over Pt nucleation and growth normalto the plane of themembrane, controlling the deposition in the plane of themembrane is extremely useful.Most useful IPMC devices require complex directional actuation and therefore some degree of surfacepatterning, where a monolithic membrane is unsuitable. A variety of techniques for patterning Pt over themembrane surface have been disclosed, ranging from clean room style photoresist nanolithography tosealing portions of themembrane under adhesive tape or even CNCmilling.275–277 All these approachesare problematic for manufacturing devices. Strictly top-down methods such as tape masking and CNCmilling have limited spatial resolution and are time-consuming. Higher throughput methods such asphotolithography use solvents for the resist films, which greatly swell and distort the membrane.278Here, a general method for photolithograpically patterning the Pt film is demonstrated, as shown inFigure 3.52. Impregnating the membrane with silver ions instead of Pt creates a photosensitive ionomermembrane. Subsequent exposure to UV light produces metallic Ag nuclei.Developing the silver ions using a mild reducing agent and metal complex, such as Pt (II) or Ag(I)with hydroxylamine or hydrazine produces secondary growth of the existing seed particles. Galvanicdisplacement of the Ag with Pt can be conveniently followed with optical spectroscopy, because of thestrong surface plasmon of silver nanoparticles.279 The strong yellow color of the Ag (band at 410 nm)1013.12. Patterned electroless depositionFigure 3.53: Optical spectroscopy of Nafion films before and after photonucleation of Ag seeds, andafter conversion to Pt shell, Ag core nanoparticles.Figure 3.54: Low resolution photopatterning of Ag nanoparticles in membrane leaves yellow Ag nanopar-ticles (left). Brightfield TEM image of the Ag nanoparticles (right).disappears after incubation with any easily reduced Pt salt such as cisplatin or tetrachloroplatinate,leaving behind black nanoparticles (Fig. 3.53).Without any optimization of the reaction conditions, the resolution of the lithography is adequate, butthe growth of the nanoparticles is uneven (Fig. 3.54, left). The diffusion of the metal ions and reducingagent towards the nanoparticles is not homogeneous, especially around edges of the exposed areas.However, it may be adequate for IPMC applications, and can certainly be optimized. TEM shows theseed particles are relatively small, in the 2-5 nm range (Fig. 3.54, right). Unlike spin coating photoresists,this process keeps the membrane fully hydrated under aqueous conditions at all times, avoiding anyproblems with delamination.The soft mechanical properties of the ionomer membrane create opportunities for patterning the1023.12. Patterned electroless depositionFigure 3.55: Electroless Pt film on membrane embossed with DVD corrugation showing structural color(A). SEM image of the corrugation showed Pt particles embedded into the surface (B). AFM of the Ptfilm highlights the corrugation banding (C). Linescan from AFM image (D).polymer film itself. Using corrugated ionomer films is a common strategy to increase the surface areaof fuel cell catalyst layers.129,280,281 This approach is useful for thin film catalyst systems, such as 3M’sNSTF system, where the extra surface area allows for higher catalyst loading in the same geometricarea without increasing the thickness of the film. Also, nanocorrugated films exhibit lower wettabilitythan flat films. Corrugating fuel cell catalyst layers provides higher tolerance against flooding by liquidwater, resulting in improved performance at high current densities. As a proof of concept, the membranewas hot embossed into the grooves of a standard DVD. Depositing Pt over this membrane created aconformal film that persisted even after boiling in water (Fig. 3.55A and B).The Pt serves as a highly reflective coating, turning the catalyst film into a diffraction grating. AFMof the surface indicated that the corrugation was stable, even after boiling in acid, although the featuresbecame slightly smoothed as the membrane swelled (Fig. 3.55C and D). The corrugation did not no-ticeably influence the electrochemistry of the film. The embossed gratings are most interesting whencombined with the IPMC technology, allowing for the creation of electrically modulated monochromators,and other optoelectronic devices.1033.13. SummaryFigure 3.56: Different analytical techniques allowed characterization of electronic conductivity in the Ptfilm over a large range of length scales.3.13 SummaryPt was electrolessly deposited onto Nafion membranes using a modified impregnation-reduction ap-proach. Controlling the nucleation and growth of Pt nanoparticles allowed the selective deposition ofultrathin, high surface area Pt localized on the surface of the membrane, in a substantial improvementover previous work in this field. Because the Pt was deposited onto a hygroscopic polymer, the en-vironmental humidity of the membrane could manipulate the properties of the Pt film. The electricalconductivity of the layer was probed using a number of different analytical techniques, from the bulkdown to the single particle level (Fig. 3.56). Very few literatures examples exist of interfaces with thedegree of control over nanoparticle-nanoparticle contacts reported here.The electrocatalytic performance of the electroless Pt films for theORRwas evaluated. The activity ofthe electroless catalyst is limited by the interconnectivity of the particles, and the poor mass transport ofoxygen to the particle surface. The electrocatalytic activity of the electroless films is not competitve withcommercial designs, and the adsorption of ionomer to the Pt surface represents a further fundamentalbarrier to their use as highly active catalysts. At this time, there does not appear to be a method forimproving the low solubility of oxygen or hydrogen in ionomer, or to reduce the poisoning of the surfacefrom the ionomer.The deposition chemistry of the films were extended to Pt alloys, demonstrating CO tolerant HOR.The electroless Pt was seen to be extremely electrochemically stable vs standard Pt/C catalysts whenconducting fuel cell accelerated stress tests. These properties are attributed to the elimination of carbonsupports, the testing geometry, and the initial particle size distribution. The films were evaluated asplatforms for sensing humidity and hydrogen, as well as actuation applications, which are less costsensitive than fuel cell catalyst layers. Photolithography and hot embossing allowed for 3D patterning1043.13. Summaryof the Pt film, extending the scope of these electrolessly deposited composites as advanced functionalmaterials. The use of Pt-ionomer composites for these applications appears much more promising.105Chapter 4Reactivity of the Pt-PFSA Interface4.1 IntroductionControl over the structure and stability of Pt-PFSA interface is critical in the development of high perfor-mance fuel cell devices. The addition of PFSA to catalyst films dramatically boosts the ionic conductivityand mass transport of protons out of the anode and into the cathode. The ionomer also serves as adispersant and binding agent, improving the quality of the initial catalyst ink, as well as preserving highporosity and mechanical durability within CCM layers after they are sprayed. However, as demonstratedin the previous chapter, adsorption of PFSA to Pt nanoparticle surfaces drastically reduces electrocat-alytic activity. Degradation of the dispersed PFSA inside catalyst layers, as well as the PEM itself remainlimiting factors in managing the life cycle of cells. The membrane suffers from mechanical degradation(cracking, crazing), while both the membrane and the dispersed PFSA experience chemical degrada-tion, which is generally attributed to radical species generated at the electrode surfaces. For thesereasons, it is clear that the solid electrolyte interphase (SEI) layer between the Pt and the PFSA must behighly optimized to control the catalyst surface chemistry, electrode nano-morphology and cell durability.Despite its ubiquitous use, the study of Pt surfaces in the presence of polymer electrolytes has beenonly sparingly investigated. It is challenging to build well-defined model systems of this interface forseveral reasons:• Ionomer dispersions do not assemble onto surfaces in a well-defined, ordered fashion, and theirassembly is sensitive to environmental conditions.• Ionomer dispersions are generally not available in high purity, and often contain a variety of stronglyadsorbing impurities.• It is extremely difficult to clean electrodes coated with ionomer films. Flame annealing leads todecomposition, while electropolishing is not effective at desorbing polymers.• Structural or analytical data on the various polymer dispersions from commercial manufacturers1064.2. STXM imaging of electrolessly deposited filmsis almost completely unavailable.• The surface adsorption phenomena of Pt remains incompletely understood even for much simpler,well-studied systems such as in H2SO4 electrolytesThe electrolessly deposited catalyst films serve as a useful system which evades most of the aboveproblems. By depositing Pt onto membranes, the PFSA does not need to be assembled from solution.The Nafion membranes can be cleaned of organics, surfactants, and contaminant species much moreeffectively than polymer dispersions. The electroless Pt surface can be cleaned by cycling (temporarily,before ionomer readsorbs). The Nafion 115 membranes used for deposition are undoubtedly the beststudied and understood PFSA, in contrast with proprietary dispersions. In conventional electrochemistryand surface science, organic adsorbate species are assembled onto a well-defined metallic substrate.Here, this model is inverted. The Nafion polymer electrolyte can be thought of as the well-definedsubstrate, against which the Pt is assembled.There are relatively few analytical techniques suitable for investigating mixtures of hard and softmatter. Generally, techniques developed to work on one type are unresponsive to the other. Of therelatively few surface sensitive options which respond well to both hard and soft matter (e.g. AFM),most are poorly suited for studying composites and embedded surfaces. Therefore, the development ofnew methodologies for studying buried interfaces is especially desirable. In this chapter, new tools aredeveloped to probe the structure and chemistry of PFSA on Pt nanoparticles.4.2 STXM imaging of electrolessly deposited filmsThe structure of the PFSA at the surface of the electrolessly deposited Pt film was investigated us-ing Scanning Transmission X-ray Microscopy (STXM) in collaboration with Adam Hitchcock (McMasterUniversity). The Pt-coated membranes were embedded into epoxy, ultramicrotomed, and imaged incross-section. Image sequences (stacks) were collected at the C 1s and F 1s edges. While the instru-ment collects image frames at different energies, direct interpretation of individual images is challenging.The observed spectra from any pixel can be considered a sum of the spectra from all the different car-bonaceous or fluorinated material in that location. If the spectra of all the pure components containedin the sample are known, then the relative composition of every pixel can be deconvoluted using linearalgebraic techniques such as single value decomposition (Fig. 4.1). In this case, the spectra of the em-bedding epoxy and bulk Nafion PFSA are drawn from the region far away from the surface, and above1074.2. STXM imaging of electrolessly deposited filmsthe surface of the film, and the spectra are in close agreement with previous measurements. The Ptmetal absorbs some X-rays, but does not have any absorption edges near the C 1s region, so it can bemodelled as a constant offset in image intensity. The NEXAFS spectra of the PFSA from the surfaceregion containing the Pt is clearly different, and was modelled as a separate component extracted fromthe region just below the Pt film.The result is a Z-dimensional image, where Z is the number of fitted components. These imagescan be overlaid and color coded to map the local variations in composition (Fig. 4.2). The small divot inthe Pt film is the same feature in both images. In both C 1s and F 1s, the undamaged Nafion is green.In the C 1s, the damaged Nafion is red, while the Pt signal is cyan. In the F 1s, the damaged Nafion isblue while the Pt is purple.The PFSA near the Pt surface appears to be chemically modified. Because the NEXAFS spectrareflects nearest-neighbour and covalent bonding interactions, a change in the spectra implies that thebonding of the PFSA has been modified in this region. Conformation and orientational changes shouldnot influence the spectroscopy of these samples. Samples with thicker Pt films exhibited thicker regionswith altered spectra, drawing a direct link between the Pt in the membrane and the NEXAFS spectra.These changes appear in both the C 1s and F 1s spectra, which supports the conclusion that PFSAhas been chemically modified. Control experiments cutting through one of the “bubble” regions used forCS-AFM experiments in the previous chapter confirm that the changes in spectra disappear in regionswhere no Pt has been deposited.The normalized C 1s and F 1s spectra of the damaged and bulk PFSA can be compared, but are toocomplicated to precisely assign, and do not point towards an obvious chemical modification (Fig. 4.3).The C 1s spectrum shows a subtle increase in the pre-edge region around 290 eV, and a decrease ofthe main feature at 298 eV. The F 1s spectra show slight differences in the pre-edge at 685 eV, and thepost-edge between 700-710 eV.Two hypotheses emerge: either the electroless deposition chemistry locally modifies the PFSA struc-ture, or the changes in NEXAFS spectra are an artifact of the embedding and STXM imaging process.During the electroless deposition, the catalyst particles are polarized with borohydride ions, diatomichydrogen from decomposing borohydride, and oxygen from the atmosphere. While any one of thesespecies is not expected to create damage to the membrane, the colocalization of oxygen and hydrogenon the surface of a Pt particle could create highly reactive conditions. These conditions might be capa-ble of directly electrochemically damaging the PFSA, or generating radical species which go on to dothe same. The artifact hypothesis mainly involves the idea that X-ray beam damage alters the PFSA1084.2. STXM imaging of electrolessly deposited filmsFigure 4.1: Multicomponent fit of STXM C 1s stack of embedded electroless catalyst layer cross section.Images collected by Adam Hitchcock.1094.2. STXM imaging of electrolessly deposited filmsFigure 4.2: STXM images of the electroless catalyst layer at the C 1s and F 1s edges using threecomponent fit, as well as C 1s spectra. Images collected by Adam Hitchcock.1104.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.3: Normalized C 1s and F 1s spectra of the damaged and undamaged PFSA as measured inFigure 4.2.structure, or that the Pt catalyzes reactions with the epoxide resin. X-ray induced radical chemistry couldbe influenced by the presence of Pt nanoparticles, and therefore locally alter their spectroscopy. Bothprocesses may be simultaneously occurring inside the films.4.3 Understanding X-ray induced beam damage inperfluorosulfonic acid using correlative microscopyThe beam-sample interaction of pure PFSA needs to be understood before the response of nanos-tructured Pt-PFSA composites can be determined. Synchrotron X-ray microscopy has been previouslyused to probe the morphology of PFSA dispersed as proton-conducting ionomer and binder inside Pt/Cnanoparticle catalyst films.282,283 STXM is especially useful for analysis of polymer composites.284,285 Itprovides good spatial resolution and elemental/chemical speciation capabilities, combined with a dam-age rate relative to analytical outcome that is several orders of magnitude better than analytical electronmicroscopy.286–290The push to characterize the nanoscale morphology of ionomer with increased detail and in threedimensions intrinsically exposes samples to higher doses of ionizing radiation. Although PFSA is chem-ically quite inert, it is extremely sensitive to X-ray and electron beam damage.291,292 Upwards of 70% ofthe ionomer volume in a sample can be ablated during high-resolution, multi-element, tomographic softX-ray imaging and an even larger fraction can be lost with analytical electron microscopy.293 The use ofcryo-microscopy conditions somewhat reduces, but does not prevent the damage.294,295 Mitigating the1114.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyeffects of beam damage in fluoropolymers, while achieving useful analytical results, is one of the keyobjectives in the ongoing development of methods for mapping PFSA in PEM fuel cell samples.296It is critical that any damage induced by the probe beam is minimized when analyzing partially de-graded or end-of-life fuel cell samples, and that such effects can be distinguished from electrochemicalcorrosion and aging. Some aspects of the radiation damage processes may also shed light on theradical degradation mechanisms of the PFSA membrane during electrochemical testing.A more systematic understanding of beam-induced chemistry and the dose-damage relationship isnecessary to optimize data collection strategies and quantitatively interpret microscopy data. The dam-age mechanism of PFSA under irradiation is a poorly defined process. X-ray beam damage has beenshown to cause formation of unsaturated C and O species and scission of C-O and C-S bonds.292 De-composition of side chains is said to be favoured over destruction of the polytetrafluoroethylene (PTFE)backbone.291 Reduced proton conductivity of X-ray damaged PFSA also indicates a cleavage of thesulfonic acid or ether moieties.297 Both the chemistry and morphology of PFSA are modified by ionizingradiation, accompanied by significant fluorine loss. While previous reports on X-ray induced reactionshave qualitatively established degradation pathways of ionomer, the dose dependence of these reac-tions remains very crudely understood.298 It is necessary to be precise in describing and discussingradiation damage mechanisms using different techniques.The chemical and morphological changes in PFSA induced by soft X-rays are characterized by com-bining NEXAFS with non-destructive microanalysis. First, a STXM was used to expose PFSA films withcontrolled radiation doses at precise locations. Then, NEXAFS spectra were collected at these locationswithout removing the sample from the microscope. This in-situ approach has been used extensively instudying dose-damage relationships for a variety of polymeric materials.290,299–301 The dosed regionswere then examined ex-situ with Fourier transform infrared spectromicroscopy (FTIR) and AFM to in-vestigate the chemical and structural evolution of irradiated PFSA.Figure 4.4 shows images of the three X-ray dosed regions using fluorescence microscopy, STXM (at292.4 eV) and AFM. While the spatial resolution and operating principle of the three techniques are verydifferent, the three damaged regions that received 4, 5 and 40 MGy are clearly visible in all images.Acid-form PFSA samples exposed to laboratory environments spontaneously accumulate volatileorganic contaminants.302 Membranes stored in open air eventually turn dark brown from acid-catalyzedpolymerization, and become weakly fluorescent. X-ray exposure bleaches this fluorescence. Fluores-cence microscopy with near-ultraviolet excitation was used to image this trace contamination on ionomersamples after X-ray exposure (Fig. 4.4A). Regions exposed to even low doses of X-rays are dark, in-1124.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.4: PFSA spin coated film damaged by 320 eV soft X-rays in three areas: 4, 5 and 40 MGyimaged by (A) fluorescence microscopy (B) STXM OD image at 292.4 eV, energy sensitive to the C1s *C-F bonds and (C) AFM amplitude map. All three techniques clearly distinguish the damaged andundamaged areas. STXM image collected by Lis Melo.dicating photobleaching. The fluorescence signal did not recover after storage for one year after X-raydosing.The thicknesses of several areas of the undamaged PFSA films were determined by AFM topographymeasurements over the edge of the film. The AFM thickness results agreed (55±5) nm with thosedetermined by STXM (50±-5 nm). The slightly larger AFM result could be attributed to the adsorptionof ambient moisture since AFM was performed on samples that had been equilibrated with air for manydays, while the STXM data were acquired under dry He at reduced pressure. Thicknesses measuredby AFM over the edge of the film also assume the samples lie perfectly flat against the substrate, andthat no “coffee-ring” artifacts make pieces of film thicker near the edges.Higher resolution STXM and AFM images of the 40 MGy and 4 MGy regions are shown in Figure 4.5.Linescans quantifying the thickness decay using both techniques are shown in Figure 4.6. Instead ofuniformly exposed circular areas with 500 nm diameter and spaced 500 nm apart, ‘doughnut’-like areasare visible in both damaged areas with both techniques and are the result of the defocused X-ray spotused to damage these areas. The STXM images show that the 40 MGy dosing (Fig. 4.5a) producedholes of approximately 15 nm depth in the sample, decreasing the film thickness by more than 10% (Fig.4.5A). The 4 MGy (Fig. 4.5C) region shows shallower, 5 nm deep holes, with less edge definition.AFM also shows that soft X-ray damage causes substantial changes to the surface morphology ofthe PFSA film. Fig. 4.4C displays the texture of the surface on the areas that received 4 and 40 MGy.1134.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.5: High resolution images of the PFSA damaged areas of 40 (top row) and 4 MGy (bottom row)(A, B, C) show STXM OD image at 292.4 eV, AFM topography and phase contrast images, respectively,of the 40 MGy damaged area, (D, E, F) show STXM OD image at 292.4 eV, AFM topography and phasecontrast images, respectively, of the 4 MGy damaged area. Blue and green dashed lines correspond tolinescans shown in Figure 4.6. STXM images collected by Lis Melo.Figure 4.6: Extracted linescans of the: 40 MGy damaged area using STXM 292.4 eV image shown inFig.4.5A and AFM topography image of the same region shown in Fig. 4.5B (A and B, respectively), 4MGy damaged area using STXM 292.4 eV image shown in Fig.4.5D and AFM topography image of thesame region shown in Fig. 4.5E (C and D, respectively).1144.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyA higher resolution image shows that each STXM pixel (step size) of the 40 MGy damage leaves adoughnut shaped impression with sunken edges (Fig. 4.5B). Similarly shaped but shallower featuresare seen in the 4 MGy region (Fig. 4.5E). Higher doses ablate deeper holes into the PFSA, as seen inFigure 4.6B. In this extracted linescan, the roughness was smoothed over 40 nm width (one “doughnut”)to highlight the average ablation. Damage was also detected in regions over which the post-exposureNEXAFS spectra were acquired, visible as rectangles in the top-centre and top-right of Figure 4.4C Thedepth of polymer ablation (2-4 nm) in these regions which received approximately 3 MGy was less thanin the areas given doses of 4, 5, and 40 MGy (~5, 5, and 13 nm respectively). Ablation depth wasmeasured halfway between peak and trough. The average ablation observed by AFM is in agreementwith STXM NEXAFS (Fig. 4.6), but shows better sensitivity and spatial resolution.In addition, ablated ’ejecta’ material is seen redeposited onto nearby areas. Irradiative mass loss ofsmall fragments from PTFE has been demonstrated by mass spectrometry.303 The redeposited materialis not as clearly visible in STXM images even though some particles appear to be larger than the 30 nmlateral resolution of themicroscope. AFM phase contrast images show that the debris has a very differentviscoelastic response compared to the undamaged PFSA (Fig 4.5C,F). Raised bubble-like features arevisible, similar to the internal voids generated in electron beam damaged specimens.304The ablatedfeatures detected in the surface of the PFSA sample by AFM appear despite carefully defocusing theX-ray beam, such that the dose was as evenly distributed across the surface as possible. The ring shapeof each hole reflects the projected intensity profile of a defocused Fresnel zone plate. Assuming perfectoptics and a 500 nm defocus, a circular region 187 nm wide in the middle of the ring is not dosed withX-rays. Because the spot size is circular but the step size is square, the corners of each STXM “pixel”are also unexposed. Therefore, the total non-exposed area at each spot size is at least 55%.A region of the film was damaged using a fully focused X-ray beam (500 nm pixel step size and a 30nm beam spot size) displayed much deeper ablation artifacts with the AFM (Fig. 4.4, unmarked squarein top right corner). Under properly defocused conditions, high-quality multi-element NEXAFS spectracan be collected over areas of tens of square microns with only trace amounts of beam damage. Sincethe ejecta particles generated from ablation are not visible in STXM frames at either 292.4 or 320 eV(Fig. 4.5A,D), they are likely of similar composition and density to the bulk material.The C 1s, O 1s andF 1s NEXAFS spectra of the undamaged and damaged PFSA are shown in Figure 4.7.Low doses of 4-5 MGy cause minimal change to the C 1s, while some detectable differences areseen in the O 1s and F 1s spectra (Fig. 4.7A, B and C, respectively). The spectra of the 40 MGy areashows substantial changes in the NEXAFS spectra at all three edges. There is a strong decline in the1154.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.7: NEXAFS spectra of undamaged and X-ray damaged PFSA. (A) C 1s spectra, (B) O 1s and(C) F 1s spectra, for the undamaged and 5, 4, and 40 MGy dosed areas (Fig. 1). (A) also contains anexpanded plot of the 283-288 eV region. Data collected by Lis Melo.C 1s peaks at 292.6 eV and 296 eV (Fig. 4.7A) and in the F 1s peaks at 689 eV and 694 eV (Fig.4.7C). These features are attributed to the to σv* C-F states associated with –(CF2-CF2)– regions in thebackbone of PFSA.305 There is also a decay of the C 1s σv*C-C broad feature present at 307 eV.New peaks appear in the C 1s edge at 285.3 and 287 eV (Fig. 4.7A, inset). These peaks are mostlikely C 1s (C=C) → pi*C=C and C 1s (C=O) → pi*C=O transitions consistent with the formation of C=Cand C=O bonds.306 These signals have also been observed in irradiated PTFE and X-ray damagedPFSA. At the O 1s edge, the broad peak between 535 and 545 eV are related to two transitions: 540eV was assigned to the ether bond and transition O 1s (C-O) *C-O and 536 eV is related to O 1s (S-O) σv*S-O transitions. Even with low dose damage (i.e. 4 and 5 MGy), the intensity of these peaks issomewhat reduced (Fig. 4.7B). The mass loss of fluorine is much larger than the changes for carbon oroxygen. Because fluorine also contributes to the background underneath the carbon and oxygen edges,a precise measurement of the latter elements is challenging to extract.The FTIR spectra of the X-ray damaged PFSA (Fig.4.8) show that X-ray doses even as low as 4MGy make detectable changes in the chemistry of the material. The infrared spectral assignment ofPFSA has been presented previously.307,308 The FTIR spectra of undamaged, partially hydrated PFSAcontains two strong C-F stretches (1220, 1150 cm-1), with a sulfonate stretch and two ether stretchescorresponding to the terminal and linker sidechain ethers (1052, 982, 969 cm-1) are shown in Figure4.8A. Which of these peaks is assigned to the ethers and sulfonate remains controversial due to thepotential presence of mechanically coupled vibrations and complex phase separation of the sidechains.However, all features can be functionally understood as belonging to either the C-F backbone (1220 and1150 cm-1) or sidechain moieties (1058, 982, and 969 cm-1). The feature at 1058 cm-1 and 982 cm-1are referred to as CO-1 and CO-2 respectively.1164.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.8: FTIR Absorbance spectra of reference and X-ray exposed PFSA (A). Differential absorbanceFTIR spectra of dosed vs undosed PFSA (B). Magnification of FTIR plot in C=O region (C). Magnificationof FTIR plot in C-O region, with each spectra offset for clarity. (D)1174.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyA higher magnification of the side-chain related peaks in shown in Figure 4.8D. The limited signalwhen measuring small differences in ultrathin samples at high magnification makes the overlapping969 and 982 cm-1 bands difficult to distinguish. These have been treated together as a single feature.Vibrational spectroscopy reveals changes in the bonding of the radiation damagedmaterial that remains.To more sensitively examine the changes due to irradiation, differential absorbance spectra are plotted(Fig. 4.8B):ΔAbs= og(dmged/ ndmged) (4.1)where Idamaged and Iundamaged are the observed intensity of irradiated and reference PFSA, respec-tively. In these differential spectra, upward going peaks indicate damage of the structural motif respon-sible for the vibrational transition. Negative going peaks indicate that the radiation damage has createdreaction products giving rise to new spectral features. A broad, negative-going peak centred at 1680cm-1 is detected for all X-ray damaged areas (magnified in Fig. 4.8C), which is assigned to C=O andC=C bond stretches, in accordance with the NEXAFS spectra and previous work.309FTIR indicates that all the bonds initially present in the PFSA are damaged by X-rays, and appearas positive-going peaks in the differential absorbance spectra, even for the areas given 4 and 5 MGy.Loss of the broad signal attributed to C-C bonds or sulfonate dimers between 1250 and 1350 cm-1 wasobserved at all doses but because of strong overlap, it was treated together with the more intense C-Fstretch region.310The NEXAFS and FTIR spectral information both show that X-rays exposure to PFSA causes forma-tion of new species (C=C and C=O) and result in damage of the C-F, C-C and side-chain related oxygenbonds. The major advantage of FTIR is the sensitivity to the ether and sulfonate vibrational modes whichclearly indicate significant damage to the side-chain moieties of PFSA. For NEXAFS, these changes arenot as visible in the O 1s spectra (Fig. 4.7B) mostly due to the ultra-low thickness of the PFSA. Thequantification of the spectral changes to both STXM and FTIR are shown in Figure 4.9 and combined inFigure 4.10.The total changes at each edge using NEXAFS can be tracked by linearly extrapolating the pre-edge background and integrating the intensity over the whole spectra (e.g. 280-340 eV for C 1s). Thefractional changes in total areas are shown in Figure 4.9A. There are negligible changes to the carboncontent even for 40 MGy. While changes in the C-C bonds are visible in Figure 4.7A, it is likely thatcarbon ablation competes with beam-induced carbon deposition onto the surface of the PFSA derived1184.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.9: Quantitation of the STXM (A and B) and FTIR (C) spectra. (A) shows the quantitation ofthe change for the whole spectra at the 1s edge while (B) shows the difference in OD of each specifiedpeak. The error bars are the calculated standard error attributed to systematic instrumentation errors.(c) peak integrals of FTIR absorbance spectra. All data was normalized to the undamaged area to plotfractional differences.from cracked hydrocarbons in the sample chamber.311 The degradation at the O 1s edge is complicated.The low oxygen content of the PFSA, hygroscopic nature of the film, and poor peak definition at this edgemakes quantitative analysis challenging. Previous experiments indicate the O 1s spectra are not linearwith sample thickness. Changes in the fluorine content are much more easily discerned. 10±3 % and40±5 % of the F 1s intensity is degraded at 4 and 40 MGy, respectively.Quantitative changes to particular spectral features as a function of X-ray dose can be calculated bymeasuring changes in OD at the relevant energy (Fig. 4.9B). The intensity at 292.4 and 690 eV reflectthe contribution of the C-F bonds. Both decay approximately 10% for the low dose (4,5 MGy) and almost40% for 40 MGy dose.For FTIR, each peak was numerically integrated, using an extrapolated polynomial baseline. Theintegral of each band was compared with the undosed regions on the same sample to determine thepercentage of X-ray induced damage for each dose and is shown in Figure 4.9C. The C-F regions aremuch more accurately quantified because of their greater area. Approximately 10% and 35% of the C-Fbonds are depleted with low (4-5 MGy) and high dose (40 MGy) X-ray exposure, respectively. FTIRshows that the ether depletion is quite severe, 40-50% loss for the low dose, and 60-80% loss for thehigh dose.The dominant physical interaction of soft X-rays with matter is photoabsorption, which damagesbonds both directly and indirectly, through the inelastic scattering of secondary electrons (Auger andlower energy photoelectron and inelastically scattered electrons).312 In PFSA the deposited energytriggers many outcomes, including bond breakage, bond formation, rearrangements and mass loss.1194.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.10: Comparison of the dose-damage response with C 1s and F 1s STXM NEXAFS spectraSTXM and FTIR peak integral fractional changes.FTIR, AFM and NEXAFS show that X-ray damage in PFSA leads to:1. fluorine mass loss2. physical ablation of large quantities of the sample3. redeposition of ablated material4. severe damage to the sidechains5. formation of C=C and C=O bonds, concomitant with cross-linkingAfter absorbing an X-ray, the C-F moiety likely produces a highly reactive fluorine radical that attacksother bonds.313 It should be noted that the commercial PFSA used in this study is “chemically-stabilized”.This term usually implies PFSA in which the polymer chain ends are terminated with CF3 groups insteadof COOH.314 This modification enhances the chemical stability of the polymer versus radical attack,also modifying the oxygen content, and presumably the initial presence of any C=O features in the FTIRspectra from chain ends.Carbon contamination arising from deposition of cracked hydrocarbon species is an issue during ion-izing radiation experiments especially at the low vacuum used in the STXM chamber. No carbonaceousresidue was detected on the bare SiN substrate, and the low doses employed minimize this contribution.The changes in local viscoelasticity detected by AFM phase contrast imaging on the redeposited debriscould be explained by cross-linking or oxidation of the PFSA.315 Cross-linking is also consistent withthe formation of the C=C bonds as detected by NEXAFS and FTIR (Fig. 4.7 and Fig. 4.8). Previous1204.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopystudies have identified carboxylic acids and double bonds as the products of thermal and electrochemi-cal degradation of PFSA.316 The poorly defined nature of the product distribution prevents precise FTIRspectral analysis. The 1360 cm-1 feature detected in radiation damaged PFSA could correspond to newC-C features in PTFE-like domains clipped free of side chains. Chain scission from γ-irradiation in PTFEand PFSA increases polymer crystallinity, and PTFE single crystals are observed to form spontaneouslyinside PFSA electron microscopy samples.317The peak profile of the asymmetric C-F stretch (Fig. 4.8, 1210 cm-1) is substantially altered afterirradiation, and displays a more complex pattern reflecting multiple overlapping C-F and C-C vibrations.The change in the C-F peak profiles and relative intensities in the differential absorbance spectra as afunction of dose indicate that the PFSA evolves in a nonstoichiometric manner, and that different C-Fbonds have different sensitivities towards beam damage.It is well-known from previous studies that the ether moieties are the points in PFSAmost susceptibleto chemical attack by radicals and X-rays.318,319 The relative damage of the 1058 and 982 cm-1 bandswas six times higher than the C-F damage at 4 MGy dose (Fig. 4.9). Precise kinetic analysis is limitedby the signal to noise of the small intensity of the sidechain bands. No attempt was made to compensatefor any preferential orientation of polymer backbone or sidechains.320 For these reasons, FTIR integralsshould be treated as semiquantitative.Sidechains in PFSA are critical in shaping the morphology responsible for the unique properties ofthese polymers. Their extreme sensitivity to radiation demonstrates the importance of measuring beamdamage in high spatial resolution analysis. Because C-O vibrations associated with the sidechain aremore X-ray sensitive than C-F vibrations, fuel cell catalyst layers using ionomer with lower equivalentweight (higher sidechain content) are likely to exhibit enhanced damage kinetics with STXM imaging. Itis anticipated that advanced PFSA ionomers using ‘short side-chains’ and ether-free linkages may showdifferent behaviour under STXM or TEM imaging.321,322While not appropriate for direct structural determination, fluorescence imaging (Fig. 4.4A) is usefulas a non-damaging diagnostic technique. The catalytic decomposition of trace organic atmosphericspecies is apparently halted once sidechains in the PFSA are destroyed by radiolysis. Of the analyticaltechniques presented here, fluorescence microscopy was the cheapest, fastest and most sensitive atimaging X-ray damage. Regions receiving only the dose from a large-area, defocused STXM stackappear bleached, as well as the regions purposefully exposed to X-rays. Common techniques such asXPS, X-ray diffraction or small angle scattering often do not allow for a simple and precise measurementof beam position and spot size. A cheap and fast tool for mapping X-ray exposure is useful in checking1214.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyfor damage to ionomer by analytical methods using ionizing radiation.Polymer ablation does not follow simple first order kinetics, and represents a degradation pathwaythat depends strongly on sample geometry and environment. The polymer fragments and oligomersgenerated by radiolysis probably evaporate into the microscope vacuum chamber. However, fluorinemass loss was still detected in control experiments where the X-ray exposure was performed at ambi-ent pressure and when the PFSA film was trapped between two SiN windows. Together with the knownformation of internal voids in beam-damaged specimens, the rapid generation of trapped gaseous degra-dation products is a reasonable mechanism for the ablation and partial redeposition of the polymer.The spectroscopy of X-ray damaged PFSA resembles that of free radical degradation phenomenaduring fuel cell operation. Electrochemically stressed MEAs previously studied by XPS showed fluorinedepletion with increased graphitic carbon presence, which is similar to these observations of X-ray dam-aged PFSA using STXM and FTIR.318 PFSA membranes chemically stressed under conditions whichmimic fuel cell degradation exhibited side chain disintegration as studied with NMR.323,324 To the extentwhich the damage phenomena involve similar radical chemistry, it should be possible to use radiationsensitivity in an X-ray microscope as a convenient proxy for studying the durability of ionomer materi-als and their composites. While Fenton reagents have long been used for nonspecific radical attack,STXM offers a more controlled exposure and built-in analytical capabilities. It is important that the doseis carefully monitored when studying aged or stress-tested fuel cell materials by STXM, to minimize theconvolution of beam-damage with device-associated degradation.As shown with AFM, FTIR and NEXAFS, high quality STXM measurements should strive to remainbelow 5 MGy to cause minimal physical and chemical damage in PFSA materials. Remaining belowthis limit requires special attention from users towards microscope settings and efficient data collectionstrategies. This limit severely restricts the analytical information from a given area with high spatialresolution.Existing literature on Nafion ultrathin films indicates they have unusually high hydration comparedwith bulk membranes due to large changes in polymer morphology.325 If adsorbed water comprises asignificant mass fraction of the thin film material, thickness determination and radiation dose becomedifficult to accurately measure. It is known that fine structure of the FTIR spectra of Nafion sidechainsresponds strongly to water adsorption. The FTIR spectra of undamaged Nafion film was measuredafter STXM experiments and storage under ambient atmosphere for several weeks. The spectra wascollected immediately after introduction into the dry gas stream of the spectrometer. Three pseudo-Voight peaks were fitted over the 1040-920 cm-1region. The FTIR spectra of the STXM sample matched1224.3. Understanding X-ray induced beam damage in perfluorosulfonic acid using correlative microscopyFigure 4.11: Reference Nafion FTIR absorbance spectra showing hydration dependence (top, adaptedfrom reference7). FTIR spectra of undamaged Nafion ultrathin film after STXM measurement and stor-age under ambient conditions (bottom, 2 cm-1).that of a very dry Nafion membrane that was baked out under dynamic vacuum (10-3 mbar) and storedin a glovebox environment.7This result indicates that the water content in Nafion films is very low in samples that are even tran-siently exposed to rough vacuum. For films with less than two water molecules per sulfonic acid, themass fraction of water is <3%, and can be neglected for the purposes of dose calculation. This indicatesthat either ultrathin films adsorb less water than thick extruded membranes, or take an unexpectedly longtime to equilibrate with a humid atmosphere.50,182,325 Attempts to rehumidify ultrathin films cracked anddelaminated the samples.1234.4. STXM Imaging of Pt nanoparticles inside PFSAFigure 4.12: Backscatter SEM image of small (A) and large Pt nanoparticles (B) dispersed inside aPFSA membrane.4.4 STXM Imaging of Pt nanoparticles inside PFSAPt surfaces catalyze radical reactions and is present in large quantities in fuel cell electrodes. To de-termine its influence on X-ray damage to PFSA, two platinized PFSA samples were fabricated by ion-exchanging cationic Pt salts into the membrane before reduction with hydrogen gas.In the first sample, relatively small Pt particles were deposited throughout the membrane, less than10 nm in size (Fig. 4.12A). These nanoparticles are below the spatial resolution limit of the STXM (ap-proximately 30 nm) and X-ray images of these samples show homogeneous contrast. In the secondplatinized sample, the membrane was seeded with Pt nuclei prior to reduction by exploiting the photo-sensitivity of Pt compounds. A smaller number of larger nanoparticles (80 nm) were formed, embeddedinside the membrane (Fig. 4.12B). Imaging the sample with large particles showed two interesting fea-tures. First, STXM images at the C 1s peak absorbance (292.6 eV) show that the PFSA reorganizesits nanostructure at the Pt interface (Fig. 4.13A). Near the Pt (dark spots) the intensity of the PFSAsignal is larger than in the surrounding membrane, forming lighter halos over a hundred nm thick onsome particles. This indicates there is more ionomer in the immediate vicinity of the Pt particles. Thislocal increase in ionomer intensity was determined to be a transmission projection artifact, created bythe microtome blade when cutting a sample with 80 nm particles into a film 50 nm thick. AFM shows thePFSA near the particles is locally thicker. Similarly sized holes are also visible where Pt particles werepresumably incorporated into the preceding microtome slice.Closer analysis of the STXM C 1s data demonstrates that the spectra of the PFSA near and onthe Pt particles is damaged (Fig. 4.14). The spectra were deconvoluted in a similar fashion as inFigure 4.1. While there is large amounts of X-ray damage throughout the polymer due to the high1244.4. STXM Imaging of Pt nanoparticles inside PFSAFigure 4.13: STXM and AFM of the large nanoparticle STXM sample shows divots and raised bumps,indicating the surface distorted near Pt particles during microtomy. STXM frame C 1s 292.6 eV, 500nm scale bar (A). AFM tapping amplitude map (B). AFM topography map, with extracted linescans overdifferent features (C). AFM frame is 5 μm x 5 μm x 30 nm. The STXM and AFM images were collectedover nearby, but not identical regions.Figure 4.14: Deconvoluting PFSA damage on Pt nanoparticles using STXM. In the overlay image, theundamaged PFSA spectra is green, and the damaged PFSA is red.1254.4. STXM Imaging of Pt nanoparticles inside PFSAFigure 4.15: Damage check frame collected at 292.6 eV after the stack in Fig. 4.14 shows extremedamage to the sample, visible as a dark square where the images were collected. Linescans over thesquare show the damage to the spectral features.resolution imaging, the X-ray damage is also correlated and locally enhanced with the positions of thenanoparticles. The damage surrounding the nanoparticles appears the same as X-ray damage in thenon-platinized samples. The damage enhancement extends less than 50 nm away from the Pt surface,which is approximately the spatial resolution of the image. At least two non-exclusive hypotheses couldexplain this phenomena. The simplest explanation is that Pt is electron-dense, and absorbs X-raysmuch more than the polymer component. This absorbed energy must be locally distributed, and mightprovide nearby PFSA with a higher local dose in the form of thermal energy or secondary electrons. Thephotoelectron, Auger and secondary electrons generated by absorbing a photon at a Pt atom insidePFSA has an estimated range (ignoring any chain processes) of approximately 3 nm. An alternateexplanation is that Pt could serve as a catalyst for PFSA degradation in the presence of radical species.In this case, the damage was incurred during the collection of the stack, as opposed to a separateexposure at a fixed dose.Almost 60% of the C-F bonds were lost during the collection of this stack, which highlights the degra-dation induced by STXM mapping at maximum spatial resolution (Fig. 4.15). However, the damage iseven higher in the regions near the nanoparticles, as visible in Figure 4.14. In addition to the damage,data collection of the stack required approximately 15 minutes. During this interval, small sample driftshifted many of the frames out of focus, which is very tight at the maximum resolution of the micro-scope. The beam damage and sample drift problems can be solved simultaneously by improving thedata collection strategy. Instead of collecting full C 1s and F 1s stacks (approximately 175 frames), amuch smaller number of frames at pre-selected energies can be acquired, once the precise spectral1264.4. STXM Imaging of Pt nanoparticles inside PFSAFigure 4.16: AFM topography of the Nafion film (left), and Pt-Nafion (middle) over regions given a doseof 10 MGy. Extracted line profiles (right) demonstrating ablation of the sample after X-ray exposure.Both AFM frames are 5x5 μm and have the same topographic scale. Blue lines indicate the regionsused for linescan analysis.regions of interest are known. This “stack-map” approach allows for drastically higher speed measure-ment (using less beamtime), while improving the image quality and reducing sample damage. This isaccomplished by collecting one OD image at just below the edge (280 eV), while another is collected atthe energy of a peak in the spectra. Subtracting the two reveals the change in intensity of that feature.4.4.1 AFM and FTIR of Pt-PFSA dispersionsThe platinized membrane with very small nanoparticles was used to evaluate the X-ray sensitivity of Pt-PFSA interface using AFM and FTIR. AFM shows that for an equivalent dose, the Pt containing samplehas much lower ablation. The texture of the damaged surface is much different, and the Pt sample iscovered by nanorods (Fig. 4.16).FTIR spectroscopy indicates X-ray damage in Nafion is altered by the addition of Pt catalyst (Fig.4.17). The sample with small Pt nanoparticles shows 8-14% total increase in damage for the 1050 cm -1stretch relative to the pure Nafion. The 982 cm -1 feature shows a 7-11% decrease with Pt dependingon dose. Interestingly, the C-F damage in the presence of Pt is half as large at 10 MGy dose, anda quarter as large at 40 MGy. Together, the AFM and FTIR establish that the C-F bond and polymerablation is reduced in the presence of Pt. This has serious implications for the STXM analysis of fuel cellsamples containing high loadings of Pt nanoparticles. Preferential degradation inside a catalyst layercould create the appearance of ionomer segregation, or artificially alter the quantity of Nafion detected.The microtomed membrane had a much more heterogeneous thickness than the spin-cast layers, whichlimits the use of FTIR quantification. Because the thickness of each dosed region is not equivalent, thechanges in peak area cannot be properly compared as in the previous section with spin-cast PFSA,although the dose dependent changes are informative, and the relative peak areas of different features1274.5. STXM imaging of a stress-tested electroless catalyst filmremain accurate. AFM and FTIR indicate lower C-F X-ray damage in the presence of Pt, when thefraction of Pt is low and does not contribute substantially to the total absorbance of the film. For sampleswith enough Pt that the absorption of X-rays is (locally) increased, the film receives more energy andthe damage is increased (Fig. 4.14).4.5 STXM imaging of a stress-tested electroless catalyst filmAn electrolessly deposited catalyst MEA was subjected to accelerated stress testing. The MEA wascycled 200 times between hydrogen and oxygen evolution (0-2 V) at 100 mV/s in an argon atmosphere.The Pt catalyst layer was carefully delaminated and rinsed free of all carbon black particles. The crosssection of this film was imaged using STXM.STXM stack maps were collected at three energies at the C 1s edge, and two energies at the F 1sedge (Fig. 4.18). The pixel size and spot size of all these images was 40 nm. In the C 1s images(A,B,C), the PFSA is on the left, while the epoxy is on the right side. For D, the PFSA is on the bottom.In A and B, the image reflects the intensity of the largest C 1s peak in the spectra. A faint artifact fromultramicrotomy is visible as a broad white vertical line in A. The Pt appears as a dark, wrinkled line. ThePFSA has approximately constant thickness right up the edge of the Pt band. A slightly white trace ofthe black Pt line is visible in part B to the left and right of the Pt, which is attributed to a small increase inthickness created by distortion during sectioning. In part C, the image reflects intensity of C=C bonds inthe material. A clear increase in intensity is visible only on the PFSA side of the Pt layer, indicated by ayellow arrow. The damage penetrates 100-150 nm into the film. Part D shows the intensity of the mainF 1s peak corresponding to C-F bonds, and was acquired over the same area as C. In this image, aclear decrease of intensity is visible near the Pt film, indicating the loss of fluorine. The thickness of theC-F depleted band is approximately 80 nm wide. The spectroscopic signature of the C=C bonds allowsthe assignment of the damage to the electrochemical treatment, and not sample preparation artifacts orthinning. The very low dose used to capture these frames (1-2% of that shown in Figure 4.15) allowsfor confident differentiation of the stress testing damage from beam induced damage. Damage checkframes at the F 1s edge after the C 1s stack map revealed very little change in the imaged material.These images indicate that STXM can be used to directly image the SEI layer of Pt and PFSA.Damage to the PFSA penetrates tens to hundreds of nm away from the surface of the Pt.1284.5. STXM imaging of a stress-tested electroless catalyst filmFigure 4.17: Differential absorbance FTIR spectra on platinized membrane for three different doses with320 eV X-rays. Comparative integrations of the Pt-loaded and Pt-free membrane for the three differentregions previously measured.1294.5. STXM imaging of a stress-tested electroless catalyst filmFigure 4.18: STXM images of stress tested electroless catalyst layer. (A) C 1s stack map of the PFSA,OD292-278 eV. (B) Enlarged region of (A), same image. (C) C 1s stack map of C=C bonds, OD284.9-280eV. (D) F 1s stack map of C-F bonds, OD689-680 eV.1304.6. SERS of Pt-PFSAFigure 4.19: Reaction schematic of SERS substrate fabricationFigure 4.20: Visually tracking the deposition of gold onto Pt film. Initial electroless Pt film (A). The samefilm after five minutes of immersion in the Au bath (B) After 60 minutes of Au deposition (C). All imagesare the same scale, and of the same sample, but at different locations.4.6 SERS of Pt-PFSASurface enhanced Raman spectroscopy (SERS) was developed as a novel technique for investigat-ing the nearby surroundings of metallic nanoparticles inside catalyst layers. SERS is one of the mostsensitive spectroscopic techniques available, and works well in conjunction with in situ measurementsunder ambient conditions. The surface enhancement of Raman signals from pure Pt surfaces withinthe UV-visible region is quite poor. The amplification is dramatically enhanced on Au or Ag surfaces,which offer strong localized surface plasmons in a convenient optical range. By using Pt catalysts asseed particles, it is possible to deposit gold selectively on pre-existing Pt surfaces, without nucleatingnew nanoparticles. A cartoon schematic for this deposition reaction is shown in Figure 4.19.Au(CN2)- ions decompose into gold and HCN when protonated. Nafion coated with Pt is introducedinto a bath of potassium gold cyanide kept basic with dilute KOH. Mild heating allows for a very gentledeposition that selectively grows nanoparticles without allowing primary nucleation. The free Au ions aregenerated on the surface of the polymer, which is also where the Pt catalyst is. Optimizing the depositionof gold is much easier than silver films, because the reaction progress can be monitored optically.It was found that the gold deposition initially deposited quickly onto the Pt, then slowed down after1314.6. SERS of Pt-PFSAFigure 4.21: Gold to platinum ratio of the SERS substrates during the deposition. (A) SEM image of theoptimized SERS film after 30 minutes showing highly roughened surface (B).coating the Pt particles. The colour of the initial film is a bright, reflective metallic silver. After a shortperiod of Au deposition, the surface of the film turns purple, indicating the presence of gold nanoparticlesurfaces. Continuing the deposition for extended periods eventually turns the filmmetallic gold coloured,indicating long range metallic bridging between particles. It was possible to track the gold deposition bycutting off small slices from a sample during the deposition, dissolving the metal film off these individualpieces, and analyzing the gold and platinum content in the resulting solutions (Fig. 4.21). The quantity ofgold can then be easily understood using the Pt as an internal standard. SEM of the film after 30 minutesdeposition shows both highly dispersed Pt-Au nanoparticles, and the beginning of gold overgrowth intolarger continuous flakes (Fig. 4.21B).The ratio of Au:Pt indicates that the gold deposits on average more than one monolayer, but lessthan about 5 layers. For nanoparticles in the 3-8 nm size range, depositing an additional monolayershell corresponds to between 20-50% of the nanoparticle volume. The SERS signal was highest forthe roughest Au surfaces after about 5-10 minutes of deposition, in line with current understanding con-cerning the enhancement mechanisms.326,327 The purple areas visible in Figure 4.20B had the highestenhancement, while the SERS activity of the continuously metallic gold coloured film (Fig. 4.20C) wasonly slightly higher than the undeposited film.The Raman spectral assignment for PFSA and PFSA metal composites has been previously stud-ied.206 Zeng et al. compared the regular, unenhanced spectra of Nafion with SER spectra measuredat the Au@SiO2 and Pt coated Au@SiO2 interface. They observed that the spectra changed dramati-cally, with around 20 new peaks and a complete disappearance of the peaks present in the unenhancedspectra. These sorts of changes are common in SERS, where adsorption to the interface can change1324.6. SERS of Pt-PFSAFigure 4.22: Raman spectra of regular electrolessly deposited Pt (left) and after surface enhancementfrom Au coating (right).the symmetry and dominant vibrational modes of a molecule. The SER spectra show precisely the op-posite result: no new peaks are observed, the peaks at 773, 805, 972, and 1060 cm-1 correspond tothose from bulk Nafion, and even the relative intensity of each band matches the bulk measurement.These results indicate that the PFSA side chains are not adsorbed onto the metallic surface. Surfaceenhancement can be generated from layers of ionomer above the surface of the metal, but with a loweramplification than direct adsorption to the metal surface. Because the Au was deposited from a bathcontaining cyanide, it is very likely that the gold surface is protected by a strongly chemisorbed layerof cyanide ions. Cyanide ligands bound to the surface would not be competitively displaced by weaklycoordinating ionomer. Unfortunately, Raman spectra were not measured out to the characteristic CNstretch at 2133 cm-1 which would definitively confirm this surface modification.328The good agreement of the SERS and native Raman spectra indicate that the surface chemistry im-mediately surrounding the particles is unmodified. In particular, the prominent C-S and sulfonate stretchare consistent with the sidechains of the polymer remaining intact. Furthermore, this strategy for la-belling Pt catalysts with gold or silver could be very useful for investigating the agglomerate structure ofconventionally printed or sprayed CCMs. A key factor in optimizing the CCM morphology is the ionomercoverage of the Pt surface, for which SERS is simple and useful. Unfortunately, the SERS signals fromthe Pt-Au layers was extremely irreproducible. Multiple attempts to remeasure the CN stretch and con-firm the SHINERS effect on the interface failed. Many different films exhibited little to no enhancement,widely varying enhancement across the heterogeneously deposited surface, and large fluorescent back-grounds. Further work in tailoring the Au deposition, as well as the Raman measurement conditions are1334.7. Summaryrequired for a robust analytical screen of the local catalyst environment.4.7 SummaryThe chemistry of Pt-PFSA interfaces was investigated using X-ray and Raman spectromicroscopy.STXM revealed large changes in the PFSA near Pt particles after degradation testing including massloss and oxidation. However, a detailed understanding of X-ray induced beam-damage in PFSA is nec-essary to evaluate these changes. The spectroscopic understanding gained from the dose response ofthe pure PFSA and Pt loaded PFSA was used to successfully image degradation of the polymer duringa simulated fuel cell stress test. While the electrochemical treatment used here was quite harsh, theobserved changes were large enough that future work with conventional accelerated stress testing pro-tocols is promising. SERS was used to investigate the PFSA immediately surrounding catalyst particlesby electrolessly depositing an ultrathin Au film onto the Pt seeds.134Chapter 5In situ High Energy X-ray Diffractionof Pt Catalyst Layers5.1 Introduction to Pt oxidationDespite advances in materials for oxygen reduction electrocatalysis, Pt nanoparticles remain the stan-dard catalyst used in PEMFCs. However, further improvements are needed to meet competitive activityand durability criteria and this remains a key technological challenge in the ongoing commercializationof PEMFCs.329 For many PEMFC applications, the power demand is transient. Startup/shutdown ortransient loads lead to cycling of the nanoparticle electrocatalyst surface between oxidized and reducedstates. Oxidation of the Pt surface is directly linked to the separate processes of Pt dissolution andoxygen reduction, lowering the longevity and efficiency of the Pt catalysts.The degradation processes of Pt nanoparticles are themselves transient, and challenging to studyusing standard techniques under steady-state conditions.330 This has been most directly studied byMayrhofer using in situ electrochemical ICP-MS in a flow-cell configuration.228,331 Dissolution rates be-low 0.1 Pt atoms per nanoparticle per CV cycle can be detected on Pt foil and Pt/C. These ICP-MSexperiments have provided valuable time course data but the detection limit of Pt (~3 ppt) and the cellflushing time (up to 3 min) limit the speed at which kinetics can be accurately probed. An understand-ing of the dynamic surface chemistry on electrochemically active Pt nanoparticles from a time-resolvedperspective is therefore valuable but difficult to achieve.332A brief schematic of the processes occurring during Pt oxidation under inert, non-adsorbing elec-trolyte is shown in Figure 5.1. In the double layer region (green), the surface is relatively free of speciesexcept perhaps water molecules and nonspecifically bound ions. Cycling past 0.75 V, but before about1.1 V, oxygenated species are oxidatively electroadsorbed to the Pt surface. Above about 1.1 V the oxideundergoes place exchange (PE), where the outer oxygen atoms slip beneath the surface atoms of the Ptsurface. Recent advances in analytical and computational methods have extended the standard model1355.1. Introduction to Pt oxidationFigure 5.1: Pt oxide formation on 111 surface. CV (left) of Pt/C (left), with coloured bars indicating thesurface chemistry of the Pt at different potentials (right).of PE oxidation processes49,333 developed using single crystal surfaces in ultra-high vacuum towardselectrochemically active, polarized single crystal surfaces in liquid electrolytes.334,335 A basic diagramof the oxide formation process is shown in Figure 5.1. The exact details of this process remain poorlydetermined, and the cartoon schematics of the oxide should not be interpreted as precise atomisticstructures. However, even these well-defined surfaces have been shown to quickly reconstruct underelectrochemical potentials relevant to fuel cell operation.336 The characteristics of oxidation and dissolu-tion of nanoparticles dispersed on carbon also diverge from studies on single crystal or polycrystalline Ptelectrodes. The irreversible surface reconstruction phenomena characteristic of Pt single crystals suchas roughening and etching, are likely to occur differently on very small nanoparticles with no extendedfacets. Similarly, Wulff reconstruction, grain boundary mediated processes, and many strain-inducedeffects are exclusive to nanostructured interfaces. Therefore, direct structural probes of Pt nanopar-ticles under operationally relevant conditions are necessary to better understand their oxidation anddissolution phenomena.High energy synchrotron X-rays are an ideal probe, facilitating many in situ analytical approaches.Several reports have followed Pt oxidation using ambient-pressure photoelectron spectroscopy (XPS)337,338,X-ray absorption fine structure (XAFS)339,340, and X-ray diffraction (XRD)341,342. Dispersive X-ray spec-troscopic techniques (XPS, XAFS) generally suffer from energy throughput and duty cycle issues. Inquick-scan XAFS/XANES, dozens of precise energies need to be collected successively at high speedby a galvanometric monochromator with narrow slits. In contrast, with XRD signals are integrated con-tinuously, and with a much wider monochromator bandwidth (100 V FWHM vs 1 V). Both of these disad-1365.1. Introduction to Pt oxidationvantages lower the signal to noise in spectroscopic measurements by several hundred fold for a givenincident beam. Although chemically quite informative, XPS is even slower than XAFS. A single highquality XPS scan can require hours of collection time due to the strong scattering of photoelectrons bythe sample, even using a modern beamline. The development of next-generation light sources343 mayalleviate some of these issues, but for now, the disadvantages of energy-dispersive techniques can out-weigh the extra value gained in chemical speciation. Scanning wavelengths increases the total requireddose, causing radiation damage to extremely beam sensitive Pt oxide surfaces.189,344. Additionally, theX-ray wavelength range used in spectroscopy is very restricted (<~15 kV), whereas scattering measure-ments such as XRD or SAXS are compatible with arbitrarily high photon energies.High energy photons translate into two key benefits for in situ diffraction measurements: drasticallyincreased sample penetration and a larger range of observable reciprocal space. The former consider-ation allows X-ray transmission through conventional electrochemical cells up to several cm thick.345,346The latter advantage creates diffractograms containing a greater number of reflections, necessary forstudying the complex structure of nanomaterials using advanced data analysis techniques.347–349.Imai et al. previously established the surface oxidation of Pt nanoparticles leads to a reversibleloss of XRD intensity and a small lattice expansion.350 At potentials less than about 1.5V, the oxidationprocess is self-limited to a 1 or 2 monolayer oxide shell, with the remaining XRD signal derived fromthe strained metallic Pt core. This was later confirmed by Sasaki et al., who also showed the detailsof place-exchange are modified on alloyed catalysts.351 Time-resolved experiments have indicated thatthe oxidative place-exchange on Pt is unusually slow, and can require up to several minutes to reachcompletion. The rate constants for this process are unclear, and reported values vary, even for identicalcatalysts and experimental setups.352,353.In this chapter, it is shown that high energy XRD is an ideal probe for quantitatively studying theelectrochemistry of Pt based commercial fuel cell catalysts in real time. The oxidation process canbe followed, involving the adsorption of oxygenated species and PE, with unparalleled crystallographicdetail and with high time resolution, elucidating the details of oxidation/reduction processes on nanoparti-cles. Accelerated stress test protocol outcomes are shown to be directly linked to the oxidation/reductionchemistry, explaining the range of degradation results observed under different conditions. Furthermore,a transient state of the surface during reduction of Pt oxide is observed which may be responsible forthe dissolution and restructuring of nanoparticles. The use of high energy X-rays allows for a convenientelectrochemical cell design, overcoming limitations on sample transparency and beam damage placedby techniques using electron beams or lower energy photons. This non-destructive and simple method-1375.2. Interpreting XRD patternsology is not limited to platinum catalysts and can be used in situ to follow structural processes on othercrystalline or disordered materials.5.2 Interpreting XRD patternsMetallic Pt has an extremely simple face centred cubic unit cell containing only one atom position. Be-cause of this, the crystallographic analysis of catalyst layers is correspondingly simple. The high sym-metry of the face centred cubic Pt crystal structure can be described with only one variable (a=b=c,α=β=γ=90º) Therefore, each peak in the indexed diffractogram contains a complete description of theunit cell. Peak area, peak width, and peak positions are used as measurements of crystalline ordering,particle size and lattice parameter respectively. The reflection intensity is directly related to the peri-odic electron density of the lattice. The peak area is therefore directly proportional to the number ofatoms in the metallic Pt phase. If a surface Pt atom is oxidized, and moves into a different position,such as an α-PtO2 phase, the intensity of the Pt metallic lattice is proportionately reduced. There aremany physical phenomena which contribute to the observed diffraction linewidth, but the dominant formof peak broadening in the samples is from particle size. In small crystals, there are insufficient layersof atoms to produce perfect destructive interference of scattered photons at slightly off-peak orienta-tions.354 This imperfect cancellation results in peak broadening. The natural lineshape for diffractionpeaks is a Lorentzian function. A small Gaussian broadening contribution is often observed, and peaksare best understood as a convolution of those two components.355 Finally, the lattice parameter cap-tures the Pt-Pt bond distance. Changes in the lattice parameter report on the strain of the metallic phase.Crystallographic strain in nanoparticle systems can be quite severe, equivalent to many GPa of mechan-ical pressure. Changes of up to several percent from alloying356, surface tension357, or adsorption ofspecies to the surface358 are routine.To properly extract microstructural data from XRD powder patterns, Rietveld refinement should beused, which fits the pattern against a theoretical model using a large number of parameters.359 Thisprocess can be quite challenging to perform accurately for large heterogeneous datasets. The refine-ment process is quite sensitive to local minima traps, and the large number of fitted parameters (usually10-20) make the refinement computationally expensive. Because the crystal structure of Pt is highlysymmetrical, and described by a single parameter, all of the peaks in the diffractogram contain the sameangle-independent information for an isotropic crystal. The high level of crystallographic redundancy inthe model makes the refinement less subject to path dependence. Assumptions about the background1385.2. Interpreting XRD patternsshape are also relaxed, which is especially important for nanoparticle samples yielding broad Lorentzianpeaks, where a large amount of intensity is contained in the long tail regions between reflections.5.2.1 The synchrotron advantageDiffraction experiments were conducted ID31 at ESRF, a beamline for high energy XRD of buried in-terfaces. This beamline has several unique features necessary for analysis of fuel cell catalyst films.The energy of the beam is the most important component. There are relatively few beamlines that cantune into high photon energies with high spectral and spatial resolution, all without moving the beam.High energy X-rays have higher penetrating power. There are two other important effects to consider.The first is a direct consequence of Bragg’s law. Reflections are located at lower angle using shorterwavelengths of incident light, which has the effect of compressing the observed reciprocal space. Thediffracted ray therefore leaves the sample closer to the unscattered beam, which has practical conse-quences for detection. The attenuation of the diffracted rays depends on the path taken through thesample. If reflections all occur at low angles, their pathlengths will be similar, and the observed back-ground will be more constant. Sample holders can be designed to permit a single path for incoming andoutgoing rays, instead of requiring a dome configuration to detect diffracted rays leaving the sample inall directions. An area detector of a given size can observe a much larger portion of reciprocal space athigh energy. The Ewald sphere produced by the sample will not only be much larger with higher orderreflections, but also flatter, requiring less spherical correction of the detector. High q measurements areessential for good quality fitting of nanoparticle microstructure and disorder.360,361The second effect relates to beam damage. The scattering cross-sections of X-rays with electronsdecreases with wavelength. Several types of cross-sections arise from different scattering phenomena,of which two are relevant here. The diffraction signal represents the desired elastic Thompson scatter-ing. Inelastic scattering includes absorption and fluorescence processes. Inelastic scattering not onlycontributes to the background noise in XRD, but delivers substantial destructive energy into the sample.Since the dissociation energy of C-C bonds is approximately 3-4 eV, the total absorption of an 80 keVX-ray is sufficient to initiate many radical reactions.362 While both elastic and inelastic cross-sectionsdecrease at higher energies, the drop off for inelastic scattering is significantly faster.363 This means thatthe amount of useful diffraction information obtained per unit of damage incurred improves at higher X-ray energies.364 Historically, this benefit has been completely eliminated by a concurrent decrease indetector quantum efficiency. Recent advances in direct-detection single photon counting sensors withhigh stopping power have removed this limitation.3651395.2. Interpreting XRD patternsThe energy resolution of the beam is derived from a cryogenically cooled Laue-Lauemonochromator.Using a pair of Si wafers as diffraction gratings in transmission mode allows 10-100x higher flux thanstandard multilayer mirrors.366 In addition, the energy band pass of the beam can be freely chosen bybending the crystals, which tunes the optical acceptance angle. This strategy is muchmore efficient thanusing traditional energy dispersive slits, and provides control over several orders of magnitude FWHM.A critical feature of this optical configuration is that changing the energy does not change the exit angleof the monochromated X-rays. The illuminated sample volume is highly stable and remains fixed at alltimes.The spatial resolution of the beam is controlled by two transfocators. These are tuneable X-rayfocusing elements based on compound refractive lenses. The refractive index for X-rays is close to onefor all materials, but optical focusing can still be performed simply by stacking hundreds of concavealuminium lenses in series.367 High energy beamlines require very long distances for precise focusing,ID31 is 127 m from source to detector. The use of these transfocators preserves the brilliance of theX-rays over this distance, improving the flux at the sample by up to 105.368 At maximum focus, thebeam is approximately 2 μm vertically, 20 μm horizontally. The long working distance allows for the tightmicrofocus without serious divergence at the detector. Most beamlines use elements such as mirrorsand monochromators coated with Pt, Ir, Rh, or other high-density late transition metals in their opticaltrain to improve reflectivity and thermal performance. This has the unfortunate consequence of filteringout X-rays near the edges of those elements, and can make it very difficult to study samples containingthese materials at high energy. By using Si, Al, and Be optics, ID31 produces a broadband beam freeof spectral artifacts.5.2.2 Background subtraction strategiesThe low quantity of Pt in fuel cell membranes presents a serious problem for in-situ experiments on com-plex samples, where background scattering competes with diffraction from catalyst particles. Severalapproaches have been previously used to overcome this issue. The simplest method is to massivelyincrease the quantity of Pt on an MEA. Loadings of 5 mg/cm2 or higher are common in literature.351This cannot be reliably used to look at degradation inside fuel cells, which is loading-dependent.The bestapproaches to background correction are those that minimize its initial contribution. Careful design ofspectroelectrochemical cells and X-ray spectroscopy were both successfully used to isolate signal fromcatalysts inside complex environments. When used in combination, in-operando XRD experiments canbe performed on real-world samples, with data quality approaching ex situ measurements. A five term1405.2. Interpreting XRD patternsFigure 5.2: Raw powder diffractogram showing the subtracted polynomial background (red). Data wasanalyzed over the region bounded by the vertical red lines. The weighted residual for the fitted curve isshown below.polynomial background was used (red line) to remove the background scattering. Polynomial background correctionThe simplest approach towards XRD background subtraction is a simple polynomial fit. The relativecontribution of scattering from the special in-situ spectroelectrochemical cell is dramatically suppressedto the extent that explicit measurement and subtraction of the background scattering is unnecessary,even for fuel cell catalyst layers with low Pt loading. Differential SubtractionRemoving the background scattering from XRD through differential measurements, or multiphase de-convolution is popular. Spots on the sample with and without Pt are subtracted, or each of the referencebackground components is measured independently, and deconvoluted.369 This style of experiment of-ten complicates data analysis, especially with heterogeneous samples, or when the relative fraction ofeach phase is not constant over time. Each electrochemical experiment was run at least twice, collectedat different locations several microns apart on the electrode. If two locations can be found that differ only1415.2. Interpreting XRD patternsFigure 5.3: Raw (red) and differentially subtracted (blue) diffractograms showing contribution of variousnon-Pt cell components to overall signal (A). Signal-to-noise and background is massively improved bycorrecting for unwanted scattering. Pairwise differential diffractograms from four spots (0 to 3) on theGDE sample (B). Peaks at scattering vectors less than 2.8 correspond to the cell wall and graphite.in Pt content, the XRD signals from these locations can be directly subtracted (Fig. 5.3A). This yieldsa diffractogram containing (mostly) differences in Pt loading, while the background signal from the cellwalls, carbon electrode and electrolyte cancel out .Depending on the precise locations selected, the differentially subtracted diffractograms were verydifferent, indicating heterogeneity of the carbon background and water at the micron scale (Fig. 5.3B) .Because of the challenging background introduced by differential subtraction, a polynomial backgroundfit was preferred. Unfortunately, small amounts of mechanical sample drift render a simple two-positionsubtraction approach unreliable for experiments over several hours.A common in situ spectroelectrochemical approach is to monitor the differential instrument responseas a function of voltage (i.e. SNIFTIRS, DEMS, etc). This approach is not appropriate for diffraction,because an absolute measurement of the diffraction signal is required. Resonant diffractionOne technique for extracting background scattering is to measure multiwavelength anomalous disper-sion (MAD) effects, also known as resonant diffraction. This element selective contrast enhancementwas originally developed by Szillard in the 1920’s, now used routinely for solving protein crystal struc-tures. X-rays precisely matching the energy of a core-shell transition for an atom are scattered reso-nantly.370 By scanning the energy of the beam over the K-edge, the contrast of all scattering from Ptatoms can be specifically modulated. The amplitude of atomic scattering (ƒ ) can be described asƒtot(q,E) = ƒo(q)+ ƒ ′(E)+ ƒ ”(E) (5.1)1425.2. Interpreting XRD patternsFigure 5.4: Calculated total atomic scattering factors for Pt4 (a). Energy scan over the Pt K-edge showingintensity oscillation of the 220 reflection (b). The peak closest to the Pt K-edge is coloured black as aguide for the eye.where ƒo is the energy independent Thomson scattering cross-section in electron units, ƒ ′ is theanomalous dispersion, and ƒ ” is an additional photoabsorption component.371 Q and E are the scat-tering momentum transfer and the photon energy, respectively. The second two terms influence themeasured diffraction intensity. ƒ ′ is out of phase with ƒo, so the amplitude of the scattered radiationalways decreases (Fig. 5.4 A, black line). Real samples show photoabsorption effects in transmissiongeometry, adding to the background (red line) and the overall measured intensity is the convolution of allprocesses. This simplified calculation neglects multiple scattering from local coordination environment(EXAFS). The K edge of Pt is conveniently located at 78.3 keV.4 The linewidth of these transitions isquite narrow, so the absolute energy shift of the beam over the scan can be small, and scattering signalsfrom non-Pt background atoms are therefore unaffected. The diffraction signal as a function of X-rayenergy was recorded for a Pt foil (Fig. 5.4 B).Three effects are clearly visible in the diffractograms. First, the peak positions shift in accordancewith Bragg’s law as the wavelength changes. Above the edge (blue colours), the f” term contributes tothe background as fluorescence adding a constant offset. Third, the amplitude of the reflections reachesaminimum near the Pt edge (black trace). The observed diffraction intensity for energies above the edgeis also reduced because a large fraction of X-rays are absorbed.The technical problem with experimentally measuring these effects is that while edge linewidth isnarrow, the beamline optical performance is proportional to wavelength.372 Typical diffraction monochro-mators operate with a bandwidth of 0.1%, which corresponds to 78 eV resolution over an edge with anatural linewidth of 25 eV. The special monochromator used here allows for a bandwidth of better than1435.2. Interpreting XRD patternsFigure 5.5: Normalized powder diffractograms for Pt catalyst measured on (red) and off (black) theK-edge, showing the change in relative intensities for the 111, 220 and 200 reflections resulting fromanomalous contrast variation. Note that the red curve has been offset in the horizontal direction forclarity.25 eV, but at greatly reduced flux and stability. The amplitude of the intensity modulation is roughlyproportional to the fraction of electrons in the shell being excited over the total number of scatteringelectrons. For the Pt 1s orbital, the intensity fluctuation is only a few percent (note the y-axis in Fig.5.4 A), which means the diffracted intensity must also be recorded to a high level of precision. XRDenergy scans require every diffractometer component to scan together. The detector sensitivity must becarefully calibrated for each energy. The focal length of the transfocator lenses are altered, and wereactively tuned to keep a constant spot size on the sample. All of these factors become important formicro and nano beam experiments where it is critical that all signal derives from precisely the samesample volume. All measured intensities have been normalized to incident flux using beam monitordiodes immediately upstream of the sample.To perform the actual background correction, it is sufficient to merely subtract the on-edge and off-edge diffractograms in q-space.373 Unfortunately the anomalous signal does not scale with q as thestructure factor does, so the modulation biases the peak intensities and needs to be corrected. Forwhole pattern refinement, the subtracted data can bemodelled against a phase containing dummy atomswith a calculatedΔƒtot . Precise calculation of near-edge structure factors for core shells of heavyatoms remain challenging.374 The experimental Δƒtot(E) is a convolution with the monochromatorbandwidth, which is also difficult to measure. However, analysis not requiring accurate peak intensities(i.e.the Scherrer equation) can still be used directly without complication.The high degree of experimental complexity steeply reduces the possible acquisition speed when1445.2. Interpreting XRD patternsFigure 5.6: X-ray transmission liquid cell design, exploded view. CAD model drawn by Miika Rasola.performing resonant diffraction. The data analysis is also quite onerous. The background correction inthe liquid cell design is manageable enough that the additional challenge of anomalous contrast variationdoes not justify the extra beam time required in this case. Systems with more complex crystal structuresand cell designs where high data collection speed is unnecessary (e.g. metal oxides in batteries) stand tobenefit enormously from the elemental selectivity of in situ electrochemical resonant XRD. This strategyalso offers excellent background correction for other X-ray scattering techniques such as SAXS andPDF, which demand much higher quality background correction than standard XRD experiments.5.2.3 Cell design and measurement geometryAn electrochemical cell for in situ X-ray diffraction was designed and fabricated (Fig. 5.6). This cellcontains a number of features which significantly improve upon designs currently available in the litera-ture.375The design is based on the half-cell used in the previous chapters for electrolessly deposited cata-lysts, but modified so X-rays can be focused onto the sample. A threaded insert holds a disc-shapedworking electrode pressed against the cell body. The insert contains entry and exit slits for the X-raybeam. Viton o-rings compress polymer film windows sealed against endcap plates. The top and bottomwindows are made of a chemically inert, X-ray resistant polymer, either Mylar or PEEK, of approximately1455.2. Interpreting XRD patterns15 μm thickness. These optically transparent windows greatly reduce the difficulty of troubleshootingproblems, and allow any trapped bubbles to be easily detected during cell assembly. The endcap platesare fabricated out of aluminum, such that they can be bolted tightly onto the cell body without bending.The cell body of the first iteration (20 mm diameter working electrode) was fabricated out of Kel-F. Thispolymer is relatively cheap, undamaged by X-rays, chemically resistant, and easily machined. A sec-ond, smaller iteration with a 10 mm sample diameter was fabricated out of PEEK, to allow for easiercleaning in hot Caro’s acid or HNO3/H2SO4. While PTFE is a popular material for electrochemical cells,it cannot be used in X-ray measurements, since it is very easily damaged by radiation. Two ring gas-kets are visible in the diagram. These fluorosilicone sealing disks allow for excellent mechanical griponto the working electrode, and assist in clamping the sample into place such that no movement occurs.This element of the design is important for experiments with microbeams, and when good beam-samplealignment is required over several hours. Without o-rings and gaskets, mechanically brittle electrodessuch as carbon fibre paper are easily snapped when clamped.The cell includes multiple threaded ports in the upper and lower cell body. These ports allow fora miniature Ag|AgCl reference electrode, and inlets and outlets to pump liquid electrolyte through thechamber from a gas sparged reservoir. Ag|AgCl reference electrodes are not usually preferred for Ptelectrochemistry because of the possibility for chloride contamination. In practice, the leak rates ofmodern “leakless” electrodes are so low that high quality electrochemistry can still be achieved. Inaddition, they are compatible with hot acid cleaning. The counterelectrode was a flame-cleaned Pt wirewhich was lightly sanded to increase roughness.The electrolyte (Ar-sparged, UHP grade HClO4) was pumped throughout the cell at 1-2 mL/s, atroom temperature. The flow rates were optimized to maintain circulation while minimizing mechanicalvibration of the sample from the hydraulic stroke of the pump. Continuously exchanging the liquid re-moves residual products of X-ray radicals, flushing them away from the working electrode where theymay locally distort the electrochemical potential or interfacial chemistry.335,341 These species includeH2, H2O2, and various reactive oxygen species for which Pt is quite catalytically sensitive. The ClO4- ionis also known to be especially sensitive to X-ray exposure376, but there are limited alternatives for non-adsorbing acidic electrolytes on Pt. Beam-chopping experiments where the X-ray shutter was openedand closed while holding the electrode at a potential suitable for consuming these species showed onlyminute fluctuations in current (<10 nA). Open circuit voltages were also observed to be stable underX-ray illumination. Both of these control experiments indicate a lack of beam effects on the Pt surfacechemistry.3771465.2. Interpreting XRD patternsThe cell was designed such that the bottom compartment could be purgedwith oxygen, creating a halfcell. In practice, this mode of operation suffered from electrolyte leaks, and mechanical distortion of theMEA which complicated the sample alignment. Periodic, local flooding of the catalyst layer also changedthe attenuation of the beam in an unpredictable fashion. Half-cell operation offers no advantages whenstudying inert gas environments. Therefore, the cell was used in flooded mode, with both sides of theelectrode filled with electrolyte. A small hole was drilled through the sample, to allow electrolyte to flowfrom one side of the cell, through the electrode hole, and out the other side of the cell. This preventedstagnant electrolyte volumes inside the cell. Trapped oxygen gas inside the hydrophobic electrode waselectrochemically reduced prior to measurements.Many spectroelectrochemical cell designs require compromises in the quality of the electrochemicalmeasurements in order to achieve reasonable spectroscopic performance. This liquid cell design isrelatively free of such effects. Uncompensated resistance, and RC time constant are critical metrics forspectroelectrochemical cells. Cell designs with large resistivity or capacitance are challenging to use forscanning or kinetics experiments, because of voltage gradients and capacitive charging on the electrode.The experiments described here use relatively concentrated electrolytes (<0.25-1M) tominimize solutionresistance. Increased electrolyte concentrations increase the influence of contaminants, and change thedissolution behaviour of Pt, preventing direct comparison with literature experiments usually conductedat 0.1 M HClO4.331 The other major limitation of the cell design is the inability to perform acceleratedstress testing of the catalyst support, scanning repeatedly up to 1.5 V. Pushing the working electrodepotential past 1.3 V leads to dissolution of the Au working electrode contact, which can redeposit ontothe catalyst film surface.The cell is capable of operating in either one of two geometries: in-plane and through-plane. For thein-plane geometry, the effective pathlength of the Pt catalyst sample is 2 cm. This increases the signalfrom the catalyst by a factor of 2000-fold over previous ’through-plane’ transmission measurements, andallows either the anode or cathode of a membrane electrode assembly to be selectively illuminated us-ing a suitably focused beam. Literature experiments performed through-plane usually neglect the anodecontribution to the diffractogram on account of the lower Pt loading (10% of the total Pt signal).349 Often,Pd catalyst is used on the anode instead of Pt in X-ray spectroscopy measurements. This reduces thecell performance and creates uncertainty in the anode potential at voltages where Pd oxides form.353Since Pt and Pd have approximately the same fcc lattice (392 vs 389 pm lattice spacing), the broad re-flections from small particles still overlap in diffractograms. With the in-plane geometry, the X-rays passthrough the cell body twice, at the front and back of the sample. This creates ring doublets from polymer1475.2. Interpreting XRD patternsFigure 5.7: Cell design configured for through-plane and in-plane geometries. The black disc is theGDE sample. The white connectors visible on the right side are ports for electrolyte circulation.1485.2. Interpreting XRD patternsdiffraction, illustrating the instrumental broadening artifact discussed below. Because the distance be-tween these two rings are fixed, it is theoretically possible to apply an FFT filter on the diffractogram tostrip the background signals of the cell with this apparent spatial frequency. However, directly editing thediffractogram intensities by applying a real space mask onto reciprocal space can generate confusingartifacts, and this strategy was avoided.The attenuation of the beam by the assembled 10 mm cell plus sample in this configuration wasmeasured at approximately 0.45 absorbance units at 78 keV. This optical density is a good compro-mise between the efficiency of photons passing through the sample without interacting, and diffractedlight being internally absorbed. All experiments presented below were performed using the in-planeconfiguration of the X-ray cell.For through-plane geometry, X-rays pass through the windows of thin X-ray transparent film and liquidelectrolyte, before hitting the MEA sample and passing out the back of the cell. Through-plane modehas a number of advantages. The X-rays can be focused onto a particular spot on the sample, and iscompatible with imaging. The X-rays never pass through the relatively thicker polymer of the cell body,so the observed background of the polymer is low. The open back of the cell allows the backgroundto be constant regardless of orientation. Finally, the sample volume is very thin, so the diffraction doesnot suffer from sample-induced instrumental broadening. This last point is essential for collecting high-resolution reciprocal space measurements. Typical diffraction analysis assumes “far-field” optics.378More specifically, that the illuminated sample thickness is negligible compared to the distance betweenthe sample and the detector. The main disadvantage of the through-plane mode is low sensitivity.5.2.4 Instrumental broadeningErrors in the peak shape and position complicate crystallographic analysis. These contributions havebeen exaggerated for effect in the following cartoon diagrams, but explain some of the limitations inthe accuracy of the diffraction data. Instrumental broadening refers to the peak profile contributionof the diffractometer and sample geometry. For a hypothetically large perfect crystal illuminated by amonochromatic point source, diffraction peaks are infinitely narrow. In practice, reflections have mea-surable line width, both from the sample microstructure, and from instrumental geometry. For ID31,the primary contributions to instrumental broadening are monochromator bandwidth, slit width, air scat-ter, incident beam collimation, detector calibration, and sample thickness.379 These contributions mustnormally be subtracted before the lineshape analysis of the sample can be interpreted.380NIST SRM 674b CeO2 was used to calibrate all the XRD data. Unfortunately, some of the calibration1495.2. Interpreting XRD patternsFigure 5.8: Scanning the beam position vertically through the catalyst layer for alignment of a catalystlayer sample. The Pt 220 reflection is shown here, with a black line indicating optimal alignmentvalues historically provided by NIST for this standard reference material have been wildly incorrect. Thelatest published values (circa November 2017) are now allegedly reliable, but not certified. Multipointcalibration against a series of NIST XRD standards (SRMs 674b ZnO, TiO2, Cr2O3) yielded similarresults as the most recent reported CeO2 values.The lineshape produced by sample thickness can be quite severe for the in situ, in plane geometry.When sample thickness is a substantial fraction of the detector distance, near-field optical effects areproduced. The broadening imparted by a 1 cm sample is up to 1 degree, while the natural FWHM ofpeaks in these samples is less than 0.25 degrees. Detectors should be moved back as far as practicableto reduce these artifacts. A half cell design using a smaller diameter sample (10 mm) greatly reducedthe broadening while maintaining excellent diffraction signal. Technically, it is possible to deconvolutenear-field phenomena during data analysis with ray-tracing methods, but this is beyond the current scopeof high-throughput Rietveld analysis software.381–383 The instrumental profile is also very sensitive toeucentric alignment of the sample and goniometer. Strong changes are observed in the peak profilewhen scanning the beam’s vertical position through the MEA (Fig. 5.8). The black line corresponds tothe optimal alignment, while peak splitting and asymmetry is visible for even slightly misaligned samples.Misalignment as little as 0.005º can shift a flat sample out of the X-ray beam. For catalyst layerswhich are not extremely rigid and flat, different regions of the sample are illuminated depending on thevertical position of the beam. If high accuracy particle size measurements are not required, and thecrystallites do not contain substantial microstrain, the instrumental broadening can be ignored. Thepowder patterns were repeatedly scrutinized for traces of microstrain broadening, but this was found1505.2. Interpreting XRD patternsFigure 5.9: Real-space geometry and inset diffractogram of highly crystalline sample, free of broadeningartifacts (a) and with instrumental broadening from sample thickness (b).to be virutally absent from any of the data shown here. With XRD (unlike electron/neutron diffraction),the bandwidth is unaffected by sample attenuation, because photons have no rest mass. To illustratethe influence of instrumental or sample induced broadening, consider the lineshape of a perfect singlecrystal in the Bragg condition (Fig. 5.9a).Ignoring the precise line profile of the peak, a sharp diffracted ray is observed on the detector, cor-responding to a particular lattice parameter determined only by the crystal’s unit cell. With non-trivialsample thickness, the natural lineshape of the diffraction peak is convoluted with the projected sam-ple width into reciprocal space. This artifact can also produce absolute offsets in the observed latticeparameter depending on alignment. Relative changes in the lattice parameter are still reliable. Absorp-tion correction was also considered for the in-plane geometry. Depending on the origin position of thediffracting photon, significant bias in peak shape can be created. Rays diffracting from the initial portionof the sample experience a longer path through the cell, and therefore are more attenuated. Higherangle reflections suffer preferentially reduced intensity.Another important source of peak position error in these experiments is detector switching fromthe use of combined SAXS/WAXS measurements. The WAXS area detector is mounted onto an XYZrail which can move entirely out of the path of the beam. This allows SAXS and WAXS data to becollected in a near-simultaneous fashion. Although the collected SAXS data will not be presented here,this movement still creates calibration error in the WAXS data. The positioner has an optically encodedclosed loop with lowmicron position repeatability. Very small random errors in the position of the detectorare created when it is moved and replaced, on the order of one detector pixel. For high-precision XRD,this produces a noticeable variation in peak position. Offsets from this error were manually correctedduring data processing.1515.2. Interpreting XRD patternsFigure 5.10: Self-absorption inside the sample can alter peak shape of diffracting raysFigure 5.11: The large area XRD detector can be either inserted or moved out of the beam path to collectWAXS (left) or SAXS (right) data.1525.2. Interpreting XRD patterns5.2.5 CalibrationThe first step of data processing is the detector calibration. The raw data acquired from the detector(Fig. 5.12A) must be converted from a 2D image into continuously distributed polar coordinates, thenreduced into discrete XY values. The raw images from the detectors were calibrated and integratedusing either the PyFAI package, or with GSAS-II features. The detector is a flat plate, but the diffractedrays are projected in Debye-Scherrer rings from the sample. By fitting ellipses onto the first few intensepeaks, the calibration corrects for the conic intersection of the rays onto the image plate, along withthe sample-detector distance and any tilt in X and Y planes. Hot or dead pixels are corrected at thisstage. A ’dark’ frame (acquired with the shutter closed) is subtracted for the scintillation detector, whilethe single-photon counting detector has pixel-specific sensitivities applied. The details of the peak fittingand pixel-splitting process are quite sophisticated, but necessary to reach sub-pixel accuracy, typically<5% of the data spacing.For in-situ experiments, portions of the diffraction cone are often obscured by the cell hardware. Thatis, the detected intensity is not homogeneous in the azimuthal plane. Radially integrating the wholeimage would then alter the counting statistics and background. For the half cell, the complete imagewas used and no masking was required. Radial segmentation (’caking’) of the powder diffraction showshomogeneous signal intensity, even at high angles (Fig. 5.12B). For the phosphor detector, patternsco-added across radial angles generate the standard deviation of the signal. For the photon countingdetector, Poisson statistics are captured at the pixel level. The grid visible on the raw diffraction framereflects the dead spaces between detector sub-units. XRD MEA sampleThe working electrode was a 20mm diameter disc gas diffusion electrode. The electrode was rinsed withwater and electrolyte prior to use. The carbon fibre paper GDL substrate can be clearly distinguishedfrom the sprayed MPL and catalyst layer. For SEM imaging, a region of cleanly fractured catalyst layerwas located. There is no clear demarcation between where the sprayed catalyst layer begins and wherethe MPL stopped. Even using higher voltage and BSE imaging, the thickness of the carbon film con-taining Pt was not possible to clearly detect. At high magnification, the individual catalyst particles couldbe imaged directly on the carbon support.A simple calculation assuming parallel beam optics and excitation of a whole strip across the sample1535.2. Interpreting XRD patternsFigure 5.12: Raw diffraction detector image of the Pt electrode (A) and azimuthal texture map of the Pt220 reflection showing homogeneous intensity distribution (B).Figure 5.13: Cross-section SEM of the GDE showing MPL and GDL regions.1545.2. Interpreting XRD patternsFigure 5.14: Cross-section SEM images of the Pt/C catalyst layer sprayed on the surface of the MPL.Figure 5.15: High resolution SEM image of the GDE catalyst film showing Pt nanoparticles dispersedon carbon.1555.3. In situ diffraction of Pt oxidationFigure 5.16: TEM images of Pt catalyst from before (A) and after (B) cycling. Images collected byRaphael Chattot.allows estimation of diffraction sensitivity. Knowing the lateral spot size of the beam relative to the planeof the sample surface, the length of the electrode and the electrode loading, the amount of Pt in the pathof the incident beam can be calculated:(0.02cm∗2cm)∗ (0.3mgPt/cm2) = 12μgPtThis assumes the electrode is perfectly flat, and thin compared with the thickness of the X-ray beam.The thickness of the beam is approximately 2 μm, determined from knife-edge alignment tests. It is hardto determine exactly how thin the catalyst layer is, because it is sprayed onto a microporous carbon layerwhich looks very similar under the microscope. It can reasonably assumed that the minimum thicknessis about the same as sprayed CCM films, where the Pt/C films are as compact as possible (about 10umthick). It is estimated that the signal corresponds to a maximum of about 2 ug of Pt total in the diffractingvolume.5.3 In situ diffraction of Pt oxidationPrevious in situ XRD studies have relied on classical but simplified crystallographic models to determinecatalyst particle size (i.e.integrating Gaussian fits to single peaks342). A greater number of Bragg reflec-tions are observed at higher photon energy. Here, the first 28 peaks in the Pt diffractogram were mea-1565.3. In situ diffraction of Pt oxidationFigure 5.17: Powder diffractogram of Pt catalyst layer with Rietveld fit (A). Blue points, green line, andred line correspond to experimental data, fitted Pt pattern, and residual respectively. Black ticks indi-cate the calculated positions of Pt reflections. Detail of the Pt 220 reflection corresponding to oxidizedand reduced catalyst surface is in (B). Oxidation of the Pt surface decreases the diffraction intensity,broadens the peak, and shifts the peak position towards lower angle.sured. Simultaneously fitting the whole Pt pattern with the same crystallographic model over-determinesthe structural parameters, and allows precise tracking of small systematic changes in peak shape andsample microstructure as shown in Figure 5.17A. The higher angle reflections (only visible with highenergy X-rays) are virtually free of background signal from carbon or polymer cell components, furtherreducing systematic errors and improving the quality of the structural measurement. Rietveld refinementof the Pt in the catalyst layer reveals changes in peak profile as a function of electrochemical potential(Fig. 5.17B).Several structural parameters extracted from the Rietveld analysis can be used monitor the oxidationand reduction of the nanoparticle catalyst surface (Fig. 5.19). Surface oxidation of the Pt nanoparticlesappears as a 17% decrease in peak area. The apparent size of the Pt crystallite, determined from thepeak width, decreases from 2.53 to 2.30 nm. The lattice parameter, determined from the peak position,expands from an equilibrium value of 0.3895 to 0.3903 nm. The lattice parameter corresponds to thesize of the Pt unit cell, and is therefore a direct measurement of average Pt-Pt bond length and crys-tallographic strain. Bulk Pt metal has a lattice parameter of 0.3924 nm, indicating that highly curvednanoparticle surfaces exhibit a significant amount of surface tension related strain. Even at these highpotentials, no clear diffraction signal from any Pt surface oxide is detected, indicating a highly disor-dered, amorphous structure consistent with other surface scattering studies.188,189,351 Subtracting thediffractogram of the oxidized catalyst from the reduced form shows peaks matching the Pt lattice, and1575.3. In situ diffraction of Pt oxidationFigure 5.18: Differential signal showing powder diffractogram of oxidized Pt catalyst subtracted fromreduced material.a weak, broad feature overlapping with the Pt 111 reflection which could not be indexed to a known Ptoxide phase (Fig. 5.18).Ticks indicate peak positions for the Pt lattice (red) as well as the Pt3O4 (blue), and PtO2 (green)species that have been detected on very small Pt particles by other groups384. Any Pt oxidation productsshould appear here as negative going peaks. Rietveld refinementRietveld refinement of all patterns was performed by first initializing the fitting algorithm parameters withan identical diffractogram. Pt was modelled as the only phase present, ignoring the polymer, graphite,and cell components. The refinement was restricted to a portion of reciprocal space between about 0.2-1585.3. In situ diffraction of Pt oxidation1.4 nm-1. This range includes the lowest angle Pt peak, up to where no more signal is detected abovebaseline noise. Organic materials such as Kel-F have much larger unit cells than Pt. Combined withtheir low crystallinity and electron density, a large cell causes their diffracted intensity to occur almostentirely at low scattering angles, below the selected cutoff. The more electron-dense Pt lattice has lessdiffuse atomic scattering, which produces more intense reflections at higher angles. All of the visiblefeatures in the refined range can be indexed to Pt reflections.The largest challenge in refinement is avoiding local minima in fitting. It is customary to iterativelymodel the pattern in a number of steps, increasing the number of modelled variables over time. In thisexperiment, a difficult aspect is fitting the background, which is strongly structured compared with typicalpatterns obtained from bulk material in ex-situ experiments.The broad peaks from small particles makeit difficult to define a clear baseline.The quality of the fits is quite good given the complexity of the cell environment and background(Fig. 5.1).385 While there are a number of common statistical metrics used in assessing the quality ofcrystallographic models, none are well-poised to judge the quality of systematic errors in backgroundmodelling. Inspecting the raw diffractogram before polynomial background subtraction shows the vastmajority of the error derives from unmodelled scattering off water, graphite and the polymer cell wall(Fig. 5.2). These residuals reach almost 500 times the standard deviation of the measurement, whichcomplicates statistical analysis. The random error arising from imprecision and counting statistics in theunderlying data is much lower.Here, the RF factor is reported, which effectively measures how well the predicted electron density ofthemodel corresponds to the observed scattering amplitude. FO and FC are the observed and calculatedstructure factors respectively, being summed over individual Bragg reflections hkl.RF =Σhk|FO,hk|−|FC,hk|Σhk|FO,hk| (5.2)Values below 5% are generally considered publication quality. Also of importance to the model is theUiso parameter, sometimes called the temperature or B-factor, which relates to isotropic thermal motionof the Pt atoms. In nanoparticles, these values are quite high, indicating a low average coordinationnumber associated with surface atoms. This value contributes extra broadening to the peaks whichotherwise confuses the particle size determination.1595.3. In situ diffraction of Pt oxidationParameter Refined valueLattice Spacing (nm) 0.3896Crystallite Size (nm) 2.5Microstrain noneUiso (10-2 nm2) 0.012RF 3.6%Table 5.1: Structural parameters refined from XRD scans collected at 0.6V vs RHE.5.3.1 Equilibrium conditionsTo measure the Pt nanoparticle surface oxidation behavior under approximately static conditions, theelectrode potential was stepped in 100 mV increments from 0.28 V up to 1.38 V and returning to 0.28V vs RHE. The electrode was allowed to equilibrate for at least two minutes at each potential before adiffractogram was measured. Previous measurements using surface scattering indicate that the oxida-tion process takes an extremely long time to reach completion (>30 minutes), but two minutes is longenough that the Pt surface can be considered near equilibrium. The Pt lattice parameter begins to shiftbetween 0.8 and 0.9 V, coincident with the oxidation peak seen in the CV (Fig. 5.19A). However, the peakarea and particle size only begin to shift at more positive potentials, between 1 V and 1.1 V (Fig. 5.19Band C). This can also be clearly seen in the measurements taken at a slow sweep rate of 5 mV/s (Fig.5.22). These changes are attributed to the observation of two different processes: the electroadsorptionof oxygenated species (the early change of lattice parameter) and the onset of Pt place-exchange (theshift in particle size and peak area), in agreement with previous observations on Pt(111) surfaces341and nanoparticle systems351. Differentiation of these processes will be discussed later in more detail.The error bars in Figure 5.19 are derived from the Rietveld analysis covariance matrix, reflecting theaccuracy of the refinement on an absolute scale, and not the standard uncertainty from precision.386Covariance is a generous metric for analysis of systematic error, which estimates the accuracy of struc-tural prediction from cross-correlation of variables in the crystallographic model. Adding extra variablesto the model actually increases the covariance error bars, even if the RF fitting quality marginally im-proves. This is a useful metric to monitor, as it allows one to ensure the crystallographic model does notoverfit the data.These changes are only partially reversible upon reduction of the Pt oxide. The average particlesize increases slightly, and the peak area does not completely recover to its initial value. These shiftsare attributed to Ostwald ripening and dissolution of the Pt, respectively. Electrochemically cycling Ptnanoparticles to >1 V vs RHE triggers irreversible corrosion and microstructural evolution. TEM of1605.3. In situ diffraction of Pt oxidationFigure 5.19: Structural parameters extracted from diffractograms collected at each potential step. Latticeparameter (A), particle size (B), peak area (C) of the modeled Pt phase. Arrows indicate the datacorresponding to forward and reverse steps. Overlaid is the CV collected at 5 mV/s for reference.1615.3. In situ diffraction of Pt oxidationFigure 5.20: Comparison of the diffractograms measured in-situ after the CV experiment shown in Fig.3, with several as-received HiSPEC catalysts of varying particle size. The intensity of each diffractogramhas been normalized.the catalyst after the experiments shows an increase in particle size and agglomeration (Fig. 5.16).To prevent these changes from influencing in situ measurements while cleaning the Pt surface, theelectrode was cycled at 50 mV/s between 0 and 1.25 V until both the diffraction signal and hydrogenunderpotential deposition region of the CV was stable, about 25 cycles.The diffraction of several different Pt/C materials from the HiSPEC series were measured ex-situas reference standards. Approximately 25 mg of each catalyst was packed into a Kapton capillaryunder ambient conditions, and measured under the same beamline configuration as the GDE. The size-dependent peak broadening is clearly visible in the profile of the overlapping 331 and 420 peaks atapproximately 10.3 degrees (Fig. 5.20). The in-situ scan was collected after several rounds of potentialcycling then allowing the particles to fully reduce at 0.6 V for several minutes. The peaks for cycledparticles are significantly narrower than any of the uncycled materials, implying larger crystallites. Thecatalyst measured in-situ was the samematerial as the 4.5 nmPt, but from a different batch. The nominalparticle sizes represent the maximum value allowed by the manufacturer specification, measured byXRD.387 The 3.5, 4.5, 2.3 and 2.8 nm catalysts are HiSPEC 3000, 4000, 4100, and 9100 respectively.Finely divided Pt (and especially Pt-alloy) is known to sometimes exhibit surface oxidation after synthesisor under ambient storage, which is reduced upon cell conditioning.388 This effect might also partiallycontribute towards small increases in the apparent crystallite size of the cycled GDE.Catalyst nanoparticles exhibit disordered microstructure (defects, twinning, etc) and large atomicthermal displacement parameters which contribute additional line broadening to the observed XRD re-1625.3. In situ diffraction of Pt oxidationFigure 5.21: Particle size distributions of Pt catalyst from TEM images before (black) and after (red)electrochemical treatment, with normal distribution fits.flections. To better decouple these phenomena, the crystallite size determined by TEM, was used toguide the Rietveld refinement. Only individually isolated particles were counted in the particle size distri-bution used in the calibration of crystallite size, excluding heavily aggregated or agglomerated material.True crystallite size is difficult to measure precisely by TEM, because the fraction of agglomerated par-ticles is high. All particles were assumed to be perfect single crystals for the purposes of calculatingparticle size. The “before cycling” catalyst was sampled directly from the GDE, to account for any ripen-ing or aggregation that can occur during sonication steps in the preparation of catalyst inks.The raw particle size distribution from TEM needs to be weighted by volume for an accurate compar-ison to the XRD values. The weighted mean particle sizes for initial and aged catalysts were determinedat 2.9 and 3.6 nm, assuming spherical particles. 100 particles were counted for each sample.The commercial catalyst has a somewhat polydisperse particle size, which complicates a preciseXRD microstructural analysis. No currently available Rietveld software package easily models crys-tallite size distributions, and peak broadening induced by polydispersity is very difficult to differentiatefrom crystallographic packing defects.389 While the exact line profile contribution from polydispersity andinstrumental-geometry effects are not well-determined here, these factors are expected to remain rel-atively constant within one electrochemical cycle. Considering the XRD measurement underestimates1635.3. In situ diffraction of Pt oxidationcrystallite size while the TEM overestimates it, the agreement between the two values is good (2.5 vs2.9 nm, respectively).Peak area and particle size are both useful independent parameters for determining the degree ofPt oxidation. It was previously shown that the Pt loading varied across the sample at the micron scalewhich influences the measured diffraction intensity (Fig. 5.3B). By instead using the refined particlesize, derived from Rietveld analysis of the peak shape, the influence of any electrode inhomogeneity ormechanical drift during long experiments can be almost completely suppressed. The ability to monitormultiple independent parameters for determining the progress of oxidation within a single dataset is apowerful advantage in reducing experimental error during complex spectroelectrochemical experiments.The changes in the particle size and peak area are consistent with each other, and the existingphysical model of place exchange. The low ECSA vs Pt surface area from CO adsorption isothermindicates that about half the catalyst particles are electrically disconnected in this experiment. The signalfrom these disconnected particles should remain constant. The fraction of surface atoms on spherical3 nm particles is approximately 40%. If it is assumed half are disconnected, then the surface oxidationof a monolayer should result in an observed reduction of the peak area by 20%. Because oxidation andplace exchange probably does not occur on the Pt atoms bonded directly to the carbon support, it canbe anticipated that the observed changes in peak area will be reduced by some factor. The observedreduction in peak area of 17% is therefore in line with estimation from first principles. Oxidation of aPt monolayer reduces the crystallite size of a sphere by the thickness of two Pt atoms (2 x 0.278 nm).Again accounting for the disconnected fraction, and applying the carbon footprint calculated from thepeak area:1− 17%20%= 15%coeredthe calculated reduction in particle size is 0.235 nm, which is in excellent agreement with the ob-served change in particle size of 0.23 nm. The peak area and particle size converge on a single quan-tity of oxidized Pt that is both physically plausible and predicted by theory, building confidence in theexperimental result. Rietveld models including two Pt phases, one disconnected and constant and oneconnected and variable were attempted several times, but were not able to discriminate the two sub-populations of nanoparticles.1645.3. In situ diffraction of Pt oxidationFigure 5.22: Peak area collected during cyclic voltammetry at various scan rates (A). The ’step’ scancorresponds to the potential step experiment shown in Fig. 5.19. Lattice parameter (red points), peakarea (black line) and current (blue line) from 5 mV/s CV (B). The lattice parameter has been smoothedwith a three point moving average (red line).5.3.2 Cyclic voltammetryThe dynamics of the oxidation phenomena can be studied by measuring diffractograms collectedcontinuously during CVs at varying scan rates (Fig. 5.22). Monitoring the diffraction during CVs demon-strates how the slow kinetics of place-exchange limit oxide formation of Pt nanoparticles under typicallyencountered electrochemical conditions. Slower sweep rates result in longer periods of time spent athigh potentials, significantly increasing the extent of Pt oxidation (Fig. 5.22A). This trend is visible as asystematically lower peak area between 1.1 and 1.6 V for the cathodic scan using slower sweep rates,indicated by a black arrow. This phenomena has been previously observed in electrochemical “sweep-hold-sweep” experiments390, where the charge passed during the reductive scan increases after theelectrode potential is held at a oxidizing potential for longer periods of time. This can also be seen indi-rectly through Pt dissolution using scanning flow cell ICP-MS.331. Kinetic limitations are also visible inthe apparent width of the oxidation/reduction loop. Faster scans oxidize at systematically higher poten-tials, and are reduced at lower potentials than steady state measurements, indicating the oxide surfaceis not in equilibrium with the applied potential during the CV. The data were too sparse to extract a quan-titative scan rate dependence. These shifts help to explain why small variations in accelerated stresstesting protocols can lead to different degradation outcomes, and why rigorously identical degradationconditions must be used in order to meaningfully compare the stability of different catalysts.134,391 USDepartment of Energy guidelines recommend catalyst degradation testing at 50 mV/s, 80ºC, and in a50 cm2 PEM cell98. Very few studies follow the latter two conditions, and merely copy the electrochem-1655.3. In situ diffraction of Pt oxidationical cycling potential limits and sweep rate. This sweep rate creates conditions where the Pt oxidationand reduction are under kinetic control, and therefore sensitive to temperature. Conventional benchtopcatalyst testing using rotating disc electrodes (RDE) at room temperature203 likely induces less Pt oxi-dation versus actual PEM cell testing, explaining why the RDE tests greatly overestimate the durabilityof catalysts when applying the same potential cycling profile.392 Allowing for a short pause at the upperpotential limit during accelerated stress testing with an RDE would allow for the Pt oxidation to reachequilibrium, possibly increasing the accuracy of simulated degradation without compromising experi-mentally convenient conditions or vastly increasing the time required.331The lattice parameter shows a narrow hysteresis loop for Pt oxide formation (Fig. 5.22B, red line).Starting at 0.35 V the lattice parameter holds relatively constant until it starts to increase at approxi-mately 0.85 V (note that the lattice parameter axis in Fig. 5.22 is reversed) until reaching saturation atapproximately 1.2 V (Fig. 5.22B). As mentioned above, this potential corresponds to the onset of ad-sorption of oxygenated species. The potential at which the Pt(111) surface reaches complete coveragewith oxygenated species (O or OH) is about 1.1 V.393,394 On nanoparticles, this range is expanded,because defects and low coordinate edge sites have higher oxygen affinity.338,395 These changes inlattice parameter are assigned primarily to the adsorption of oxygenated species, which are known toinduce strain in the Pt surface.396 It is expected that the formation of the PE oxide layer also makes asmall contribution to the total strain at the nanoparticle surface. Therefore, these changes in the latticeparameter during CVs should be considered a combined measure of both oxygenated species adsorp-tion and PE at high potentials. These two processes can be further decoupled by exploiting differencesin their rate constants.The degree to which this strain propagates from the surface towards the Pt core remains unclear.Atomic resolution STEM has shown that lattice strain from surface tension or support interactions canbe either homogeneously distributed throughout the particle’s volume, or restricted to surface atoms397.Surface diffraction on Pt single crystals indicates that the surface oxide creates measurable strain withinat least the 2nd layer of subsurface Pt atoms335,395, which would account for a large majority of thenanoparticle’s volume.398 These changes in lattice parameter represent small shifts in the peak max-imum and their accuracy is limited by sub-pixel interpolation, requiring careful attention to the data re-duction parameters, as well as the subsequent fitting.399At lower potentials (<0.35 V) a continuously reversible shift in the lattice parameter is observed,which can be understood as strain induced by the adsorption of underpotentially deposited hydrogen(H-UPD) (Fig. 5.22B). This sensitivity towards minute structural changes highlights the usefulness of1665.3. In situ diffraction of Pt oxidationin-situ XRD for probing catalyst surfaces, especially when the actively bound species, e.g.H atoms oroxygenated species, cannot be directly detected. H-UPD serves as a convenient control, because it isfast, reversible, specific for Pt, and confined to a monolayer on the surface. If the strain from H-UPDwas not evenly distributed throughout the nanoparticle, a range of lattice parameters would be estab-lished, broadening the peaks. No such peak broadening or correlation between lattice parameter andthe apparent particle size was detected, which supports a hydrostatic “liquid-drop” model where straingradients across nanoparticles are minimized.350 On Pt(111), H-UPD expands the spacing between thefirst and second surface layers of Pt by approximately 2%.395 Here, the lattice expansion is only about0.15%. If the nanoparticles behaved as single crystal surfaces, with a strained monolayer skin over arelaxed core, the expected strain would be much higher, even accounting for the disconnected fractionas before.While CVs are relevant to degradation experiments, they are not the best tool for probing electro-chemical reaction rates, because time and potential are convoluted in the response.5.3.3 Pt oxidation/reduction kineticsThree time-resolved potential step experiments were performed to better understand the oxidation/reductionkinetics (Fig 5.23). First, the Pt surface was conditioned in the double layer region (0.48 V). Then, theelectrode was stepped to higher potential, allowed to evolve for two minutes, and stepped back to 0.58V (Fig. 5.23A). The upper potential step was varied between 0.98, 1.18 and 1.28 V. The potential ofthe first step is sufficient for partial adsorption of oxygenated species, but well before place-exchangeoccurs. The second and third steps are sufficient to initiate place-exchange of the adsorbed oxygenatedspecies, and the growth of surface-limited Pt oxide.It can be seen that the lattice parameter rapidly equilibrates in both the oxidation and reductiondirections (Fig 5.23B). Conversely, the peak area and particle size respond sluggishly upon steppingto the upper potential, but return quickly to the reduced state (Fig 5.23C and D). For the potential stepto 0.98 V, the peak area is only slightly effected, and no change in particle size can be detected. Thisobservation is consistent with the previous discussion: the lattice parameter responds to the faster, initialstep of adsorption of oxygenated species, while the peak area and particle size follow the much slowerplace-exchange process. Both mechanistic steps of the surface oxidation on Pt nanoparticle catalystscan therefore be independently tracked using different structural parameters determined using in-situhigh energy XRD. All the structural parameters are measured in a precisely synchronized fashion in real1675.3. In situ diffraction of Pt oxidationFigure 5.23: Structural evolution of Pt catalyst during potential steps. Potential step profile over time(A). Lattice parameter (B), Particle size (C), Peak area (D), as obtained from Rietveld refinement. Thepotential step was made at the time point of zero from a resting potential of 0.48 V. (Black 0.98 V, red1.18 V, blue 1.28 V). The error bars represent the refinement covariance.1685.3. In situ diffraction of Pt oxidationFigure 5.24: Curve fitting on particle size and scale factor for Pt oxidation steps to 1.18 (A) and 1.28 V(B), as shown in Figure 5.23.Potential Step Particle Size Peak Area0.98 V no change observed1.18 V 0.040±0.01 0.053±0.0081.28 V 0.047±0.007 0.043±0.006Table 5.2: First order rate constants (s-1) determined for Pt place-exchange.time.The particle size and peak area parameters from Figure 5.23 were fit to determine first order rateconstants for Pt place exchange. The data were modelled with a simple exponential function, beginningfive seconds after the potential step, so as to avoid problems with the electrochemical time constant.y= y0+Ae−kt (5.3)The rate constants observed for potential steps to 1.18 and 1.28 V are similar (Table 5.2). Thesevalues are in reasonable agreement with the literature rate constants for Pt oxidation measured at roomtemperature with time-resolved XAFS and XRD (extracted from350). Studies using in situ XAFS andXRD in PEM fuel cells at 80℃ show consistently faster oxide formation rates, approximately 0.14 s-1.342Differences in the extent of Pt oxidation could be responsible for the much lower degradation rates ofcatalysts exposed to cycling at room temperature, compared with the same cycling in a functional PEM1695.3. In situ diffraction of Pt oxidationFigure 5.25: RC time constant determination from current decay after potential stepfuel cell at 80 ℃. Accounting for the quantity of surface oxide formed at high potentials may allow forthe design of catalyst stress testing protocols which better approximate real-world aging conditions.It is necessary to keep in mind, however, that a quantitative understanding of the adlayer formationand place exchange kinetics is limited by capacitive charging. Porous electrodes such as fuel cell catalystlayers have large polarized surface areas, which require a relatively long time to reach the desired voltageafter a large potential step. Oxidation or reduction at the electrode cannot occur until the voltage atthe Pt surface has been meaningfully established. Rate constants for electrochemical reactions whichprogress at similar speeds or faster than this capacitive response cannot be easily determined usingsuch a configuration, even with arbitrarily fast spectrometer sampling. While well understood within theelectrochemistry community400,401, capacitive delay has been only rarely350 recognized in spectroscopystudies.With this time delay limitation in mind, diffractometer transients from potential steps cannot be treatedindependently until after the potential on the electrode has been suitably established. The RC timeconstant of the electrochemical cell is approximately 1.5 seconds (Fig. 5.25). The time constant wasdetermined by measuring the capacitive current decay after steps between two potentials in the doublelayer region (0.58 to 0.48V).Therefore, reactions requiring substantially longer than 4.5 seconds to reach completion can be fitadequately using the raw diffractometer transient, discarding the data collected during the initial fewseconds. In this cell design, the relatively slow place-exchange on the Pt surface meets this criteria,but the fast adsorption and desorption of oxygenated species does not. It should be noted that thepercentage of electrochemically connected nanoparticles affects the absolute accuracy of peak area1705.3. In situ diffraction of Pt oxidationFigure 5.26: Curve fitting of lattice parameter and peak area for reduction of Pt oxide after conditioningat 1.28 V, as shown in Figure 5.23.and particle size measurements but not the rate constants derived from analyzing changes in thesevalues.The reduction of Pt oxide occurs much faster than the oxidative place-exchange. Previous studiesusing in-situ spectroscopy at elevated temperatures have modeled Pt oxidation as a two-step process,but proposed a swift, concerted step for the oxide reduction.342 In this room temperature experiment,additional detail in the Pt reduction kinetics can be seen. The rate constant obtained from exponentialfitting of the lattice parameter during the oxide reduction from Figure 5.23 is 1.8 times faster than for thepeak area (Fig 5.26).Note that unlike the oxidation analysis, the fitting of the Pt reduction did not discard the first few sec-onds of data, and these points are convoluted with capacitive response. However, for a truly concertedreduction mechanism, this convolution would apply equally to both parameters, and not influence theirrelative rates.These data are consistent with a model where fast reduction of the amorphous oxide layer yieldsa momentarily disordered Pt surface, prior to the reconstruction of the metallic lattice. The peak areaonly detects Pt atoms located in their crystal lattice positions, a disordered metallic surface would not1715.3. In situ diffraction of Pt oxidationFigure 5.27: Transient response of lattice parameter and peak area for the reduction of Pt oxide duringa potential step from 1.3 to 0.5 V. Solid lines show first order exponential fits.contribute to the diffracted intensity. This model is supported by several previous reports using in situXAFS spectroscopy.342,352,353 Upon reduction, Pt-O bonds in the surface oxide disappeared signifi-cantly faster than Pt-Pt bonds were reformed. The difference in this balance implies the existence oftransiently undercoordinated Pt species during the reduction. These XAFS measurements were per-formed inside PEM cells at 80ºC, lending further weight to the hypothesis that the disordered surfacesdetected in this potential step experiment are also relevant in real operating conditions. Characterizingthe transient states generated by oxide reduction and the mobility of surface atoms during this period iskey to understanding the dissolution of Pt.228,330 Under-coordinated, weakly bound Pt species are themost likely to become involved in oxidation and surface detachment processes.402,403The data sampling rate used for Figures 5.23 and 5.26 is too slow, and convoluted with capacitiveresponse for an unambiguous quantitative analysis of the Pt reduction. A similar limitation applies tothe measurement of the rate constant for the adsorption of oxygenated species during the oxidationstep. The Pt oxide reduction kinetics were revisited with a setup optimized for better time resolution(Fig. 5.27). The exposure time of each diffractogram was shortened to 100 ms with no loss in signalto noise by increasing the incident flux of the beam. A five times smaller working electrode was usedto reduce capacitance, and the electrolyte concentration was increased four fold to 1 M to minimize thesolution resistance.The results from this high-speed potential step measurement are very similar to the transient shownin Figure 5.26. The time constant for the reduction of the Pt oxide while monitoring the lattice param-eter is approximately 1 s, but the peak area is significantly slower, with a time constant of about 2 s.The sensitivity improvement facilitated by the grazing incidence cell design allows for smaller samples,1725.3. In situ diffraction of Pt oxidationenabling the study of faster electrochemical processes than previously feasible. Future experimentsutilizing variable temperatures and with fast data acquisition will build a more complete understandingof Pt nanoparticle oxidation and degradation phenomena.The positive potential limits used in this study for CVs and potential steps (up to 1.6 V) are moreaggressive than experienced in the typical, start-stop catalyst cycling protocol which lie between 0.6 and1 V. For deep oxidation (i.e.1.6 V) most of the dissolution occurs upon the reduction of the oxide.331 Forpotentials below place-exchange (approximately 1.1 V), most of the dissolution occurs on the oxidationstep. A better understanding of Pt oxide reduction at low surface coverage is necessary to determineto what extent similar processes occur under those more gentle conditions, and during the life cycle ofoperational fuel cell systems.5.3.4 Pair distribution function analysisThe nature of the disordered and amorphous Pt oxide and reduced Pt species is challenging to analyzedirectly with traditional XRD, because no long range order exists. In addition to the crystal packingwhich gives rise to Bragg reflections, wide-angle X-ray scattering curves also contain local structureinformation in the form of diffuse elastic scattering.360,361 This scattering is approximately 106 timesweaker than for Bragg reflections, but can still be detected in these high quality data. By performing aFourier transform on the scattering curves, one can generate pair-distribution functions (PDFs). ThesePDFs can be represented as radially correlated electron densities in real space, usually referred to asG(r). This is similar to the local coordination environment data usually probed with EXAFS (Fig. 5.28A).The scattering is dominated from the positions of the Pt atoms, since the scattering cross section oflighter atoms are much smaller (i.e.581 vs 0.18 barn for O).The PDF of nanoparticles exhibit a characteristic “ring-down” shape (Fig. 5.28B).349 At a low radius,the Pt atoms are, on average, surrounded by other Pt atoms. However, when the PDF probe radiusexceeds that of the particle size, the electron density becomes uncorrelated. In the data from the catalystlayer, the electron density drops to background noise levels at approximately 5 nm, which can be usedto roughly estimate particle size. To probe the structure of amorphous monolayer films such as the Ptoxide, the shorter range structure needs to be considered (Fig. 5.29).Several changes in the local coordination sphere of the Pt catalyst are visible. The easiest wayto understand the experimental PDF data is by comparison to simulated PDFs generated from singlecrystal structures. While not perfectly accurate for nanoparticles and surfaces, the structural changes incoordination environment for Pt oxidation are quite large, and this technique is sufficient. In bulk single1735.3. In situ diffraction of Pt oxidationFigure 5.28: Pair distribution function schematic showing radial scattering function of coordination envi-ronment around Pt atom (A). G(r) for oxidized Pt nanoparticles. (B)Figure 5.29: PDF G(r)s for the oxidized and reduced Pt catalyst (red and black, respectively) as wellsimulated patterns from single crystal structures of PtO2 and Pt metal (green and blue). The simulatedPDFs have been vertically offset for visual clarity.1745.3. In situ diffraction of Pt oxidationcrystal Pt, each atom is coordinated by 12 other close packed atoms, with bond distances of about 277pm, corresponding to the first blue peak. The second and third coordination shells are visible at 390and 480 pm. In crystalline PtO2, all Pt-Pt bonds are disrupted by oxygen insertion, with a Pt-O distancebetween 200 and 210 pm, visible as a broad weak feature. DFT studies suggest that the Pt-O bondlength in a crystal is similar to that found adsorbed on Pt single crystal surfaces.404 The next nearestin-plane and out-of-plane Pt-O-Pt distances are visible as peaks between 300 and 400 pm. In the PDFof the oxidized catalyst, the features corresponding to Pt metal have reduced intensity, and the featureswhich indicate PtO2 are increased. A small shoulder at 140 pm indicates the C-C bond of graphite,present in very large quantities in the catalyst layer. A more thorough, time resolved study of in situ PDFmeasurements on Pt/C is currently underway, and expected to provide a better understanding of theoxide formation, reduction, and dissolution.5.3.5 SummaryThe oxidation and reduction are key processes affecting the activity and stability of Pt based nanocat-alysts, therefore directly influencing performance of fuel cells. High energy X-ray diffraction provides aconvenient probe for studying these processes, and more generally changes in the surface chemistry,with high temporal resolution. Grazing incidence high energy XRD also greatly simplifies the complexityof both cell design and data analysis, due to the high signal to background ratio.Structural parameters obtained by Rietveld refinement during changes in potential are used to dy-namically characterize the surface oxidation of Pt nanoparticles and clearly differentiate two processes:Lattice strain is recognized as an independent parameter sensitive to the adsorption of electroactivespecies on the surface, while refined peak area and nanoparticle size is attributed to the place ex-change step in the formation of Pt oxide. Therefore, both steps of the oxidation mechanism (adsorptionand place exchange) could be simultaneously quantified and decoupled in one XRD acquisition. Mea-suring changes in lattice strain allows indirect detection of both UPD hydrogen and oxygenated specieson the Pt surface.Time-resolved measurements demonstrate that the slow kinetics of place-exchange can lead to largevariation in oxide surface coverage during dynamic electrochemical conditions, depending on cyclingspeed and potential. This slow oxidation has significant implications for the design and interpretationof accelerated stress testing protocols, as the amount of oxide grown directly affects the quantity ofplatinum dissolved over time. High speed measurements also allow direct observation of the transientsurface structure formed during the reduction step. This disordered structure consists of large fraction of1755.3. In situ diffraction of Pt oxidationunder-coordinated surface atoms and facilitates dissolution and restructuring of the nanoparticle, leadingto degradation of the catalyst. Developing a fundamental atomistic and time resolved understanding ofPt catalyst degradation phenomena will be essential in managing the life cycle of fuel cell devices anddesigning higher durability catalysts.It is anticipated that enabling high quality in-situ diffraction will be useful for a variety of surface sci-ence and catalysis applications, especially in the hands of non-experts. The capability of high energyX-ray diffraction to probe atomic-level ordering, alloys, and anisotropically shaped nanoparticles at lowloading makes it an ideal technique for understanding the structure and function of advanced electro-catalyst materials.176Chapter 6Conclusion6.1 SummaryIn this thesis, the chemistry of the Pt electrolyte interface was explored in three ways. Electroless depo-sition was utilized to manufacture robust ultrathin films directly from solution onto polymer electrolytesin a single step. Controlling the nucleation and growth of Pt nanoparticles to a much greater extent thanprevious literature allowed for the optimization of the particle size and interconnectivity of the high sur-face area metallic film. These catalysts display unique properties due to the strong interaction betweenthe electrolyte and metallic particles, including very high electrochemical stability. The optical and elec-tronic properties of the embedded nanoparticles were mediated by hydration dependent swelling of thepolymer, which was probed at a variety of length scales. The PFSA was used to direct the growth of thenanoparticles into single crystal nanowires, although the solution chemistry behind this process remainsunresolved. The deposition chemistry was expanded to include Pt alloys such as PtRu, and these filmsbehave equivalently to conventional bimetallic nanoparticles which require high temperature process-ing. Ultimately, the ionomer blocked the availability of gaseous reagents to the Pt surface, reducing thecatalytic activity versus dispersed, porous systems. Several niche applications for Pt-PFSA compositefilms were explored, grounded by the capability to control the nucleation and growth of nanoparticlesin a patterned fashion. While probably not suited for high performance PEMFCs, electroless Pt-PFSAcomposites are most promising as both high-value materials in the burgeoning field of flexible electronicmaterials, and as convenient models of the Pt-PFSA interface.The chemistry of PFSA inside catalyst layers was characterized using STXM. The absence of interfer-ing carbon support particles in the electroless film facilitates spectroscopic measurements by reducinglight scattering, porosity, and carbonaceous background, and allows for a relatively clean and robustprobe of the ionomer structure at the surface, which would otherwise be obscured. High resolutionimaging of the delicate PFSA nanostructure is strictly limited by beam damage artifacts. Soft X-raydamage of PFSA was evaluated using correlative AFM, FTIR, and fluorescence microscopy. By com-1776.1. Summarybining the results from identical location measurements, a clear picture of the radiation chemistry anddose-response for these sensitive materials was constructed. Data collection strategies which minimizeand could possibly account for artifacts were developed. These guidelines will be critical in evaluat-ing the next-generation of X-ray microscopes, and their capabilities in 3D and 4D chemical mapping ofelectrode materials.405 Electroless deposition was used to assemble Pt-PFSA composites of controlledparticle size and dispersion. Pt was shown to influence the radical decomposition of the PFSA in acomplex fashion, which influences the interpretation of data from commercial PEMFC samples. A dam-aged SEI layer at the electrolessly deposited Pt-PFSA interface was directly imaged for the first time.The ability to characterize structural changes to the ionomer at the catalyst interface will be critical indiagnosing its degradation phenomena, and for the development of more stable ionomer films. Dam-age caused by X-ray imaging created very similar chemical and morphological effects to the damageinduced by electrochemical aging, further emphasizing the necessity to account for artifacts generatedduring measurements.The Pt electrolyte interface also includes water, and at high potentials, Pt oxide. The formation, re-duction, and dissolution of this oxide is deeply responsible for the limited stability of Pt nanoparticlesin catalyst films. The dynamics of the Pt oxidation and subsequent reduction were probed using in situelectrochemical X-ray diffraction. Advances in cell design, measurement geometry, and data analysisallowed for characterization of the Pt surface with unprecedented structural detail using high energyX-rays. The two steps of the oxidation process, the adsorption of oxygenated species and place ex-change could be separately quantified using different structural parameters, onset potentials, and rateconstants. The sluggish place-exchange reaction was shown to be rate-limiting under typical cyclicvoltammetry conditions. While Pt oxidation has been extensively characterized, little is known regardingthe reduction of the oxide back to the metallic surface. High-speed XRD revealed evidence of a disor-dered intermediate species on the nanoparticles during the reduction of the Pt oxide. A more completefundamental mechanistic understanding of the Pt oxidation and oxide reduction reactions would be ex-tremely valuable in the development of improved electrocatalysts. The strategy described here allowsfor a facile, unified perspective of the surface chemistry of buried interfaces using in operando X-rayscattering on single crystals, half cells, and full PEMFCs. Technological advances in state-of-the-artsynchrotron characterization have enabled facile anomalous and total scattering techniques, which areextremely promising for probing not only amorphous species, but for interpreting scattering data fromcomplex samples such as fuel cell devices.1786.2. Outlook6.2 OutlookThe ongoing development of PEM fuel cell technology requires attention to the activity and durabilityof the electrode catalyst layers. The complex nanostructure of these films is critical for achieving goodperformance, but is poorly understood, especially in the hot, humid conditions of an operating cell. Thesimple Pt/C electrocatalysts used in (most) devices today remain effectively unchanged, despite morethan a decade of apparent breakthroughs in the academic ORR literature.While “conventional” cell de-signs have achieved commercial success, many opportunities remain for improvements in catalysts,catalyst supports, electrolytes, and film fabrication techniques. Many classes of chemistry problems aresusceptible to empirical problem solving. The trending popularity of design-of-experiment and machine-learning approaches are quite valuable in this regard. However, the large number of variables involved,operational complexity, high cost, and low throughput have largely precluded parametric screening fromefficiently advancing the state of PEM fuel cell engineering. Furthermore, any improvements to cellpower and device lifetime derived from simple screening are likely to limit the performance envelopeof the system, and may therefore be impractical for automotive applications which require high flexi-bility. Efficient solutions for the materials science problems in fuel cells will require a better physicalunderstanding of their interfaces. Unfortunately, traditional model systems, such as Pt single crystals,face serious limitations in both simulating and explaining the complex phenomena occurring inside de-vices. Therefore, a large opportunity exists for middle-of-the-road, somewhat well-defined surfaces,which capture some, but not all aspects of the nanostructure intrinsic to PEMFCs.For modelling applications, the electrolessly deposited Pt allows quite a fine level of control overthe placement of nanoparticles and structure of the interface. The use of electrolessly deposited Pt asa cathode catalyst layer in PEMFCs is fundamentally liminted by poor electrocatalytic activity. Unlessionomers with substantially different Pt surface adsorption characteristics are developed, such electro-less films are not anticipated to be useful in power producing devices. The role of “Pt in the mem-brane” towards MEA longevity remains unresolved, but seems remarkably beneficial under certain cir-cumstances. The chemistry occuring on these surfaces inside real fuel cell devices should be studiedfurther, to understand what role they play in controlling performance and durability. From a fabricationperspective, these interactions can be more easily controlled or investigated by dispersing nanoparticlesin the ionomer during membrane casting, so electroless deposition is not an especially useful avenue forprobing such chemistry. From an academic perspective, the wide preexisting scope of literature on elec-trolessly deposited platinum group metals contrains the novelty of these materials for PEMFCs, unless1796.2. Outlooktheir performance can be significantly improved by nearly an order of magnitude. The use of these filmsas model systems for the catalyst-electrolyte interface, and as advanced functional MEMS materials isnot similarly constrained.A more detailed picture of Pt-PFSA interactions is necessary for understanding how advanced elec-trocatalysts perform inside sprayed films. The degradation of PFSA at the electrified interface is likely toinfluence fuel cell performance. Future work in ink formulation stands to benefit from the STXM modelsystem developed in this thesis. Using X-ray sensitivity as a proxy for ionomer stability in fuel cells is apromising idea for future investigation. While X-ray radicals will not precisely recapitulate the phenom-ena inside functioning devices, the speed and quantitative control over dose provided by a STXM aresubstantial improvements over exposure to baths of Fenton reagents.Operando measurements of full electrochemical devices is an obvious next step for the high energyXRD technique demonstrated in this thesis. These measurement strategies are directly extendable toboth more complex sample environments and electroactive materials. While the surface chemistry ofPt is complex, Pt metal has the simplest possible crystal structure, and the full crystallographic power ofdiffraction was not used to its full potential. Future improvements in time and spatial resolution, as wellas data analysis are also likely to advance the state of the art. These advances can be fully applied to notonly PEM fuel cells, but also to batteries, electrolyzers, solar cells, and a number of other electrochemicaldevices.The development of novel, high performance materials (e.g. graphene) is in many cases limited bythe availability of analytical tools to characterize them, and of techniques for fabricating them at scale.While this work focused on the Pt-PFSA interface, these lessons in nanomaterials synthesis and tooldevelopment could be easily applied to any number of interfaces. Certainly, there exists no shortage ofpoorly understood electrochemical systems, and the desire for more efficient devices appears sure toincrease in the future.180References[1] Commons, W. Proton Exchange Fuel Cell Diagram. 2015; https://commons.wikimedia.org/wiki/File:ProtonExchangeFuelCellDiagram.svg.[2] Atomic force microscope block diagram. 2015; https://commons.wikimedia.org/wiki/File:Atomicforcemicroscopeblockdiagram.svg.[3] ESRF, I. SR-FTIR Endstation. http://www.esrf.eu/home/UsersAndScience/Experiments/XNP/ID21/SrFtir.html.[4] Chantler, C. T. Theoretical Form Factor, Attenuation, and Scattering Tabulation for Z=1 to 92 fromE=1 to 10 eV to E=0.4 to 1.0 MeV. Journal of Physical and Chemical Reference Data 1995, 24,71–643.[5] Ade, H.; Hitchcock, A. P. NEXAFS microscopy and resonant scattering: Composition and orien-tation probed in real and reciprocal space. Polymer 2008, 49, 643–675.[6] Polte, J. Fundamental growth principles of colloidal metal nanoparticles, a new perspective. Crys-tEngComm 2015, 17, 6809–6830.[7] Negro, E.; Vittadello, M.; Vezzù, K.; Paddison, S. J.; Di Noto, V. The influence of the cationic formand degree of hydration on the structure of Nafion. Solid State Ionics 2013, 252, 84–92.[8] Grove, W. On voltaic series and the combination of gases by platinum. The London, Edinburgh,and Dublin Philosophical Magazine and Journal of Science 1839, 86–87.[9] Shao, M.; Chang, Q.; Dodelet, J.-P.; Chenitz, R. Recent Advances in Electrocatalysts for OxygenReduction Reaction. Chemical Reviews 2016, 116, 3594–3657.[10] Durst, J.; Siebel, A.; Simon, C.; Hasché, F.; Herranz, J.; Gasteiger, H. A. New insights into theelectrochemical hydrogen oxidation and evolution reaction mechanism. Energy and Environmen-tal Science 2014, 7, 2255–2260.181References[11] Sui, S.; Wang, X.; Zhou, X.; Su, Y.; Riffat, S.; jun Liu, C. A comprehensive review of Pt electrocat-alysts for the oxygen reduction reaction: Nanostructure, activity, mechanism and carbon supportin PEM fuel cells. Journal of Materials Chemistry A 2017, 5, 1808–1825.[12] Shinozaki, K.; Zack, J. W.; Richards, R. M.; Pivovar, B. S.; Kocha, S. S. Oxygen Reduction Re-action Measurements on Platinum Electrocatalysts Utilizing Rotating Disk Electrode Technique.Journal of the Electrochemical Society 2015, 162, F1144–F1158.[13] Shao, Y.; Yin, G.; Gao, Y. Understanding and approaches for the durability issues of Pt-basedcatalysts for PEM fuel cell. Journal of Power Sources 2007, 171, 558–566.[14] Debe, M. K. Nanostructured Thin Film Electrocatalysts for PEM Fuel Cells - A Tutorial on theFundamental Characteristics and Practical Properties of NSTF Catalysts. ECS Transactions 2012,45, 47–68.[15] Devaraj, V.; Lopez, L. F.; Beaman, J. J.; Prudhomme, S. Model-based control of a continuouscoating line for proton exchange membrane fuel cell electrode assembly. International Journal ofChemical Engineering 2015, 2015.[16] Rikkinen, E.; Santasalo-Aarnio, A.; Airaksinen, S.; Borghei, M.; Viitanen, V.; Sainio, J.; Kaup-pinen, E. I.; Kallio, T.; Krause, A. O. I. Atomic layer deposition preparation of Pd nanoparticleson a porous carbon support for alcohol oxidation. Journal of Physical Chemistry C 2011, 115,23067–23073.[17] Yan, W. M.; Wu, D. K.; Wang, X. D.; Ong, A. L.; Lee, D. J.; Su, A. Optimal microporous layer forproton exchange membrane fuel cell. Journal of Power Sources 2010, 195, 5731–5734.[18] Weber, A. Z.; Newman, J. Effects of Microporous Layers in Polymer Electrolyte Fuel Cells. Journalof the Electrochemical Society 2005, 152, A677.[19] Rouquerolt, J.; Avnir, D.; Fairbridge, C.W.; Everett, D. H.; Haynes, J. H.; Pernicone, N.; Ramsay, J.D. F.; Sing, K. S.W.; Unger, K. K. Recommendations for the characterization of porous solids.Pure& Applied Chemistry 1994, 66, 1739–1758.[20] Ronne, J. A.; Wozniczka, B. M.; Chow, C. Y.; Voss, H. Electrochemical fuel cell stack with improvedreactant manifolding and sealing. US20030203246A1, 1997.[21] Creager, S. Handbook of Electrochemistry; Elsevier, 2007; pp 57–72.182References[22] Hurd, J. A.; Vaidhyanathan, R.; Thangadurai, V.; Ratcliffe, C. I.; Moudrakovski, I. L.; Shimizu, G.K. H. Anhydrous proton conduction at 150C in a crystalline metal-organic framework. NatureChemistry 2009, 1, 705–710.[23] Yousfi-Steiner, N.; Moçotéguy, P.; Candusso, D.; Hissel, D.; Hernandez, A.; Aslanides, A. A reviewon PEM voltage degradation associated with water management: Impacts, influent factors andcharacterization. Journal of Power Sources 2008, 183, 260–274.[24] Mauritz, K. A.; Moore, R. B. State of understanding of Nafion. Chemical Reviews 2004, 104,4535–4585.[25] Kusoglu, A.; Weber, A. Z. New Insights into Perfluorinated Sulfonic-Acid Ionomers. Chemical Re-views 2017, 117, 987–1104.[26] Schmidt-Rohr, K.; Chen, Q. Parallel cylindrical water nanochannels in Nafion fuel-cell membranes.Nature Materials 2008, 7, 75–83.[27] Hiesgen, R.; Morawietz, T.; Handl, M.; Corasaniti, M.; Friedrich, K. A. Insight into the Structureand Nanoscale Conductivity of Fluorinated Ionomer Membranes. Journal of the ElectrochemicalSociety 2014, 161, F1214–F1223.[28] Song, C.; Zhang, J. PEM Fuel Cell Electrocatalysts and Catalyst Layers: Fundamentals and Ap-plications; Springer London: London, 2008; pp 89–134.[29] McNaught, A. D.; Wilkinson, A. The Gold Book, 2nd ed.; Blackwell Scientific Publications: Oxford,1997; p 1670.[30] Briega-Martos, V.; Herrero, E.; Feliu, J. M. Effect of pH and Water Structure on the Oxygen Re-duction Reaction on platinum electrodes. Electrochimica Acta 2017, 241, 497–509.[31] Ramaswamy, N.; Mukerjee, S. Fundamental mechanistic understanding of electrocatalysis of oxy-gen reduction on Pt and non-Pt surfaces: Acid versus alkalinemedia.Advances in Physical Chem-istry 2012, 2012, 1–17.[32] Sepa, D. B.; Vojnovic, M. V.; Damjanovic, A. Kinetics and mechanism of O2 reduction at Pt INalkaline solutions. Electrochimica Acta 1980, 25, 1491–1496.[33] Markovic, N.; Gasteiger, H.; Ross, P. N. Kinetics of Oxygen Reduction on Pt(hkl) Electrodes:Implications for the Crystallite Size Effect with Supported Pt Electrocatalysts. Journal of the Elec-trochemical Society 1997, 144, 1591.183References[34] Dong, Q.; Santhanagopalan, S.; White, R. E. Simulation of the Oxygen Reduction Reaction at anRDE in 0.5 M H2SO4 Including an Adsorption Mechanism. Journal of the Electrochemical Society2007, 154, A888.[35] Climent, V.; Feliu, J. M. Thirty years of platinum single crystal electrochemistry. Journal of SolidState Electrochemistry 2011, 15, 1297–1315.[36] Furuya, Y.; Mashio, T.; Ohma, A.; Tian, M.; Kaveh, F.; Beauchemin, D.; Jerkiewicz, G. Influenceof Electrolyte Composition and pH on Platinum Electrochemical and/or Chemical Dissolution inAqueous Acidic Media. ACS Catalysis 2015, 5, 2605–2614.[37] Grgur, B. N.; Marković, N. M.; Ross, P. N. Temperature-dependent oxygen electrochemistry onplatinum low-index single crystal surfaces in acid solutions. Canadian Journal of Chemistry 1997,75, 1465–1471.[38] Sánchez-Sánchez, C. M.; Solla-Gullón, J.; Vidal-Iglesias, F. J.; Aldaz, A.; Montiel, V.; Herrero, E.Imaging structure sensitive catalysis on different shape-controlled platinum nanoparticles. Journalof the American Chemical Society 2010, 132, 5622–5624.[39] Kuzume, A.; Herrero, E.; Feliu, J. M. Oxygen reduction on stepped platinum surfaces in acidicmedia. Journal of Electroanalytical Chemistry 2007, 599, 333–343.[40] Rabis, A.; Rodriguez, P.; Schmidt, T. J. Electrocatalysis for polymer electrolyte fuel cells: Recentachievements and future challenges. ACS Catalysis 2012, 2, 864–890.[41] Dubau, L.; Nelayah, J.; Moldovan, S.; Ersen, O.; Bordet, P.; Drnec, J.; Asset, T.; Chattot, R.;Maillard, F. Defects do Catalysis: CO Monolayer Oxidation and Oxygen Reduction Reaction onHollow PtNi/C Nanoparticles. ACS Catalysis 2016, 6, 4673–4684.[42] Hodnik, N.; Baldizzone, C.; Cherevko, S.; Zeradjanin, A.; Mayrhofer, K. J. J. The Effect of the Volt-age Scan Rate on the Determination of the Oxygen Reduction Activity of Pt/C Fuel Cell Catalyst.Electrocatalysis 2015, 6, 237–241.[43] Okada, T.; Ayato, Y.; Satou, H.; Yuasa, M.; Sekine, I. The effect of impurity cations on the oxygenreduction kinetics at platinum electrodes covered with perfluorinated ionomer. Journal of PhysicalChemistry B 2001, 105, 6980–6986.[44] Bard, A. J.; Faulkner, L. R. Electrochemical methods : fundamentals and applications, 2nd ed.;Wiley, 2001; p 833.184References[45] Rumble, J. R. In CRC handbook of chemistry and physics, 91st ed.; Rumble, J. R., Ed.; CRCPress, 2017; p 2560.[46] Hoare, J. P. In Berichte der Bunsengesellschaft für physikalische Chemie; Vielstich, W., Ed.; In-terscience Publishers, 1969; Vol. 73; pp 628–628.[47] Bard, A. J.; Parsons, R.; Jordan, J. In Standard Potentials in Aqueous Solution, 1st ed.; Bard, A. J.,Parsons, R., Jordan, J., Eds.; New York and Basel : Marcel Dekker: New York, NY, 1985; p 46.[48] Birss, V. I.; Chang, M.; Segal, J. Platinum oxide film formation-reduction: an in-situ mass mea-surement study. Journal of Electroanalytical Chemistry 1993, 355, 181–191.[49] Conway, B. E. Electrochemical At Noble Metals Oxide Film Formation As a Surface-ChemicalProcess. Progress in Surface Science 1995, 49, 331–452.[50] Ingle, N. J. C.; Sode, A.; Martens, I.; Gyenge, E.; Wilkinson, D. P.; Bizzotto, D. Synthesis andcharacterization of diverse Pt nanostructures in Nafion. Langmuir 2014, 30, 1871–9.[51] Sode, A.; Ingle, N. J. C.; McCormick, M.; Bizzotto, D.; Gyenge, E.; Ye, S.; Knights, S.; Wilkin-son, D. P. Controlling the deposition of Pt nanoparticles within the surface region of Nafion. Journalof Membrane Science 2011, 376, 162–169.[52] Bradley, J. S. In Clusters and Colloids: From Theory to Applications, 1st ed.; Schmid, G., Ed.;Wiley-VCH Verlag GmbH: Weinheim, Germany, 1994; Chapter 6, pp 459–544.[53] Roebben, G. et al. Interlaboratory comparison of size and surface charge measurements onnanoparticles prior to biological impact assessment. Journal of Nanoparticle Research 2011, 13,2675–2687.[54] Rahman, M.; Rebrov, E. Microreactors for Gold Nanoparticles Synthesis: From Faraday to Flow.Processes 2014, 2, 466–493.[55] Wilson, J. J.; Lippard, S. J. Synthetic methods for the preparation of platinum anticancer com-plexes. Chemical Reviews 2014, 114, 4470–95.[56] Magnus, G. Ueber einige verbindungen des platinchlorrs. Annalen der Physik 1828, 90, 239–242.[57] Caseri, W. Derivatives of Magnus’ Green Salt. Platinum Metals Review 2004, 48, 91–100.185References[58] Liao, H.-G.; Zheng, H. Liquid Cell Transmission Electron Microscopy Study of Platinum IronNanocrystal Growth and Shape Evolution. Journal of the American Chemical Society 2013, 135,5038–5043.[59] Fang, Z.; Zhang, Y.; Du, F.; Zhong, X. Growth of Anisotropic Platinum Nanostructures Catalyzedby Gold Seed Nanoparticles. Nano Research 2008, 1, 249–257.[60] Leong, G. J.; Schulze, M. C.; Strand, M. B.; Maloney, D.; Frisco, S. L.; Dinh, H. N.; Pivovar, B.;Richards, R. M. Shape-directed platinum nanoparticle synthesis: nanoscale design of novel cat-alysts. Applied Organometallic Chemistry 2014, 28, 1–17.[61] Wuhrer, R.; Moran, K. Low voltage imaging and X-ray microanalysis in the SEM: Challenges andopportunities. IOP Conference Series: Materials Science and Engineering 2016, 109.[62] Seiler, H. Secondary electron emission in the scanning electron microscope. Journal of AppliedPhysics 1983, 54.[63] Schindelin, J. et al. Fiji: An open source platform for biological image analysis. Nature Methods2012, 9, 676–682.[64] Maniguet, L.; Robaut, F.; Meuris, A.; Roussel-Dherbey, F.; Chariot, F. X-ray microanalysis: Thestate of the art of SDD detectors and WDS systems on scanning electron microscopes (SEM).IOP Conference Series: Materials Science and Engineering 2012, 32.[65] Statham, P. J. Limitations to accuracy in extracting characteristic line intensities from x-ray spectra.Journal of Research of the National Institute of Standards and Technology 2002, 107, 531.[66] Brown, L. M. Expunging diffraction contrast: EELS and HAADF in STEM. Journal of Physics:Conference Series 2014, 522.[67] Egerton, R. F.; Malac, M. EELS in the TEM. Journal of Electron Spectroscopy and Related Phe-nomena 2005, 143, 43–50.[68] Smits, F. M. Measurements of Sheet Resistivity with the Four-Point Probe. The Bell System Tech-nical Journal 1958, 37, 711.[69] Alshwawreh, N. K. Microstructure evolution in electrodeposited copper thin films for advancedmicroelectronic applications. Doctoral thesis, University of British Columbia, 2012.186References[70] Still, B. M.; Kumar, P. G. A.; Aldrich-Wright, J. R.; Price, W. S. 195Pt NMR - Theory and application.Chemical Society Reviews 2007, 36, 665–686.[71] Levason, W.; Pletcher, D. 195-Platinum Nuclear Magnetic Resonance Spectroscopy. PlatinumMetals Review 1993, 17–23.[72] Holz, M. New developments in NMR of simple electrolyte solutions. Progress in Nuclear MagneticResonance Spectroscopy 1986, 18, 327–403.[73] Villamagna, F.; Whitehead, M. A. Determination of the ionic strength and degree of crystallizationin nitrate salt based emulsions using14N NMR. Journal of Molecular Structure 1995, 356, 149–158.[74] Paddock, S. W. Confocal laser scanning microscopy. BioTechniques 1999, 27, 992–1004.[75] Olympus, LEXT OLS5000 manual. https://www.olympus-ims.com/en/metrology/ols5000.[76] Ge, G.; Han, D.; Lin, D.; Chu, W.; Sun, Y.; Jiang, L.; Ma, W.; Wang, C. MAC mode atomic forcemicroscopy studies of living samples, ranging from cells to fresh tissue. Ultramicroscopy 2007,107, 299–307.[77] Nečas, D.; Klapetek, P. Gwyddion: an open-source software for SPM data analysis. Central Eu-ropean Journal of Physics 2012, 10, 181–188.[78] Park, J. Y.; Maier, S.; Hendriksen, B.; Salmeron, M. Sensing current and forces with SPM. Mate-rials Today 2010, 13, 38–45.[79] Hecht, E. Optics, 5th ed.; Addison-Wesley, 2002; p 698.[80] Griffiths, P. R. "Trading rules" in infrared Fourier transform spectroscopy. Analytical Chemistry1972, 44, 1909–1913.[81] Webber, M.; Dimakis, N.; Kumari, D.; Fuccillo, M.; Smotkin, E. S. Mechanically Coupled InternalCoordinates of Ionomer Vibrational Modes. Macromolecules 2010, 43, 5500–5502.[82] Bass, M.; Berman, A.; Singh, A.; Konovalov, O.; Freger, V. Surface Structure of Nafion in Vaporand Liquid. Journal of Physical Chemistry B 2010, 114, 3784–3790.[83] Carlson, O. N. The N-Si (Nitrogen-Silicon) system. Bulletin of Alloy Phase Diagrams 1990, 11,569–573.187References[84] Scardera, G.; Puzzer, T.; Conibeer, G.; Green, M. A. Fourier transform infrared spectroscopy ofannealed silicon-rich silicon nitride thin films. Journal of Applied Physics 2008, 104.[85] Lanford, W. A.; Rand, M. J. The hydrogen content of plasma-deposited silicon nitride. Journal ofApplied Physics 1978, 49, 2473–2477.[86] Egerton, R. F.; Li, P.; Malac, M. Radiation damage in the TEM and SEM. Micron 2004, 35, 399–409.[87] Du, Q.; Huang, Y.; Ogbuu, O.; Zhang, W.; Li, J.; Singh, V.; Agarwal, A. M.; Hu, J. Gamma radiationeffects in amorphous silicon and silicon nitride photonic devices. Optics Letters 2017, 42, 587.[88] McKellar, A. R. W. High-resolution infrared spectroscopy with synchrotron sources. Journal ofMolecular Spectroscopy 2010, 262, 1–10.[89] Hirschmugl, C. J.; Gough, K. M. Fourier Transform Infrared Spectrochemical Imaging: Reviewof Design and Applications with a Focal Plane Array and Multiple Beam Synchrotron RadiationSource. Applied Spectroscopy 2012, 66, 475–491.[90] Rightor, E. G.; Hitchcock, A. P.; Ade, H.; Leapman, R. D.; Urquhart, S. G.; Smith, a. P.; Mitchell, G.;Fisher, D.; Shin, H. J.; Warwick, T. Spectromicroscopy of Poly(ethylene terephthalate): Compar-ison of Spectra and Radiation Damage Rates in X-ray Absorption and Electron Energy Loss.Journal of Physical Chemistry B 1997, 101, 1950–1960.[91] Kilcoyne, A. L. D.; Tyliszczak, T.; Steele, W. F.; Fakra, S.; Hitchcock, P.; Franck, K.; Anderson, E.;Harteneck, B.; Rightor, E. G.; Mitchell, G. E.; Hitchcock, A. P.; Yang, L.; Warwick, T.; Ade, H.Interferometer-controlled scanning transmission X-ray microscopes at the Advanced Light Source.Journal of Synchrotron Radiation 2003, 10, 125–136.[92] Leontowich, A. F. G.; Hitchcock, A. P. Experimental investigation of beam heating in a soft X-rayscanning transmission X-ray microscope. The Analyst 2011, 265, 65–95.[93] Li, J.; Hitchcock, A. P.; Stöver, H. D. H.; Shirley, I. A new approach to studying microcapsule wallgrowth mechanisms. Macromolecules 2009, 42, 2428–2432.[94] Blom, D. A.; Dunlap, J. R.; Nolan, T. A.; Allard, L. F. Preparation of Cross-Sectional Samples ofProton ExchangeMembrane Fuel Cells by Ultramicrotomy for TEM. Journal of the ElectrochemicalSociety 2003, 150, A414.188References[95] Smith, A. P.; Coffey, T.; Ade, H. In X-ray Microscopy and Spectromicroscopy; Thieme, J.,Schmahl, G., Rudolph, D., Umbach, E., Eds.; Springer Berlin Heidelberg: Berlin, Heidelberg,1998; pp 3–7.[96] Hauptman, H. The Direct Methods of X-ray Crystallography. Science 1986, 233, 178–183.[97] Petkov, V. Nanostructure by high-energy X-ray diffraction. Materials Today 2008, 11, 28–38.[98] U.S. DRIVE Partnership/Fuel Cell Technical Team, Fuel Cell Technical Team Roadmap; 2013.[99] Toby, B. H.; Von Dreele, R. B. GSAS-II: The genesis of a modern open-source all purpose crys-tallography software package. Journal of Applied Crystallography 2013, 46, 544–549.[100] Giovannelli, G.; Natali, S.; Bozzini, B. Silver coated lead coins: An appraisal of ancient technology.Journal of Applied Electrochemistry 2006, 36, 951–956.[101] Brenner, A.; Riddel, G. Nickel plating by chemical reduction. US2532283A, 1947.[102] Brenner, A.; Riddell, G. E. Nickel plating on steel by chemical reduction. Journal of Research ofthe National Bureau of Standards 1946, 37, 31–34.[103] Ohno, I. Electrochemistry of electroless plating. Materials Science and Engineering A 1991, 146,33–49.[104] Kloke, A.; Von Stetten, F.; Zengerle, R.; Kerzenmacher, S. Strategies for the fabrication of porousplatinum electrodes. Advanced Materials 2011, 23, 4976–5008.[105] Ohno, I. In Modern Electroplating, 5th ed.; Schlesinge, M., Paunovic, M., Eds.; John Wiley andSons, Inc, 2010; pp 477–482.[106] Rhoda, R. N.; Vines, R. F. Bath and process for platinum and platinum alloys. US3486928A, 1965.[107] Okinaka, Y.; Sard, R.; Wolowodiuk, C.; Craft, W. H.; Retajczyk, T. F. Some Practical Aspects ofElectroless Gold Plating. Journal of the Electrochemical Society 1974, 121, 56.[108] Gurrappa, I. Platinum Aluminide Coatings for Oxidation Resistance of Titanium Alloys. PlatinumMetals Review 2001, 45, 124–129.[109] Rao, C. R. K.; Trivedi, D. C. Chemical and electrochemical depositions of platinum group metalsand their applications. Coordination Chemistry Reviews 2005, 249, 613–631.189References[110] Strangman, T. E.; Raybould, D.; Kozlov, A. Platinum electrolyte for use in electrolytic plating.US6306277B1, 2000.[111] Enea, O.; Duprez, D.; Amadelli, R. Gas phase electrocatalysis onmetal/Nafionmembranes.Catal-ysis Today 1995, 25, 271–276.[112] Takenaka, H.; Torikai, E.; Kawami, Y.; Wakabayashi, N. Solid polymer electrolyte water electroly-sis. International Journal of Hydrogen Energy 1982, 7, 397–403.[113] Liu, R.; Her, W.-H.; Fedkiw, P. S. In Situ Electrode Formation on a Nafion Membrane by ChemicalPlatinization. Journal of the Electrochemical Society 1992, 139, 15.[114] Katayama-Aramata, A.; Nakajima, H.; Fujikawa, K.; Kita, H. Metal electrodes bonded on solidpolymer electrolyte membranes (SPE)-The behaviour of platinum bonded on SPE for hydrogenand oxygen electrode processes. Electrochimica Acta 1983, 28, 777–780.[115] Holze, R.; Ahn, J. Advances in the use of perfluorinated cation exchange membranes in integratedwater electrolysis and hydrogen/oxygen fuel cell systems. Journal of Membrane Science 1992, 73,87–97.[116] Liu, R.; Fedkiw, P. S. Partial oxidation of methanol on a metallized Nafion Polymer ElectrolyteMembrane. Journal of the Electrochemical Society 1992, 139, 3514.[117] Fedkiw, P. S.; Potente, J. M.; Her, W. H. Electroreduction of gaseous ethylene on a platinizedNafion membrane. Journal of the Electrochemical Society 1990, 137, 1451.[118] Fedkiw, P. S. Preparing in situ electrocatalytic films in solid polymer electrolyte membranes, com-posite microelectrode structures produced thereby and chloralkali process utilizing the same.US4959132A, 1988.[119] Fedkiw, P. S.; WeiHwa, H. An Impregnation-Reduction Method to Prepare Electrodes on NafionSPE. Journal of the Electrochemical Society 1989, 136, 899.[120] Sheppard, S. A.; Campbell, S. A.; Smith, J. R.; Lloyd, G. W.; Walsh, F. C.; Ralph, T. R. Electro-chemical and microscopic characterisation of platinum-coated perfluorosulfonic acid (Nafion 117)materials. The Analyst 1998, 123, 1923–1929.[121] Hawut, W.; Hunsom, M.; Pruksathorn, K. Platinum electroless deposition on Nafion membranefor PEM fuel cells. Korean Journal of Chemical Engineering 2006, 23, 555–559.190References[122] Kucernak, A. R.; Muir, B. Analysis of the electrical and mechanical time response of solid polymer– platinum composite membranes. Electrochimica Acta 2001, 46, 1313–1322.[123] Bessarabov, D.; Michaels, W. Morphological diversity of platinum clusters deposited on proton-exchange , perfluorinated membranes. Membrane Technology 2001, 139, 5–9.[124] Shen, M.; Roy, S.; Scott, K. Preparation and characterisation of Pt deposition on ion conductingmembrane for direct methanol fuel cell electrodes. Journal of Applied Electrochemistry 2005, 35,1103–1109.[125] Millet, P.; Durand, R.; Dartyge, E.; Tourillon, G.; Fontaine, A. Precipitation of Metallic Platinum intoNafion Ionomer Membranes. Journal of the Electrochemical Society 1993, 140, 1373.[126] Millet, P.; Pineri, M.; Durand, R. New solid polymer electrolyte composites for water electrolysis.Journal of Applied Electrochemistry 1989, 19, 162–166.[127] Verbrugge, M. W. Selective Electrodeposition of Catalyst within Membrane-Electrode Structures.Journal of the Electrochemical Society 1994, 141, 46–53.[128] Thompson, S. D.; Jordan, L. R.; Forsyth, M. Platinum electrodeposition for polymer electrolytemembrane fuel cells. Electrochimica Acta 2001, 46, 1657–1663.[129] Grot, S. Corrugated Membrane Fuel Cell Structures -DOE Hydrogen and Fuel Cells Program, FY2013 Annual Progress Report; 2011; pp 694–695.[130] Aziznia, A.; Oloman, C. W.; Gyenge, E. L. A Swiss roll liquid gas mixed reactant fuel cell. Journalof Power Sources 2012, 212, 154–160.[131] Shao, M. H.; Huang, T.; Liu, P.; Zhang, J.; Sasaki, K.; Vukmirovic, M. B.; Adzic, R. R. Palladiummonolayer and palladium alloy electrocatalysts for oxygen reduction. Langmuir 2006, 22, 10409–10415.[132] Maass, S.; Finsterwalder, F.; Frank, G.; Hartmann, R.; Merten, C. Carbon support oxidation inPEM fuel cell cathodes. Journal of Power Sources 2008, 176, 444–451.[133] Meyers, J. P.; Darling, R. M. Model of Carbon Corrosion in PEM Fuel Cells. Journal of the Elec-trochemical Society 2006, 153, A1432.[134] Pizzutilo, E.; Geiger, S.; Grote, J.-P.; Mingers, A.; Mayrhofer, K. J. J.; Arenz, M.; Cherevko, S.On the Need of Improved Accelerated Degradation Protocols (ADPs): Examination of Platinum191ReferencesDissolution and Carbon Corrosion in Half-Cell Tests. Journal of The Electrochemical Society 2016,163, F1510–F1514.[135] Watanabe, M.; Uchida, H.; Emori, M. Analyses of Self-Humidification and Suppression of GasCrossover in Pt-Dispersed Polymer Electrolyte Membranes for Fuel Cells. Journal of the Electro-chemical Society 1998, 145, 1137.[136] Rodgers, M. P.; Pearman, B. P.; Bonville, L. J.; Cullen, D. A.; Mohajeri, N.; Slattery, D. K. Evalu-ation of the Effect of Impregnated Platinum on PFSA Degradation for PEM Fuel Cells. Journal ofthe Electrochemical Society 2013, 160, F1123–F1128.[137] Helmly, S.; Eslamibidgoli, M. J.; Friedrich, K. A.; Eikerling, M. H. Local Impact of Pt Nanodepositson Ionomer Decomposition in Polymer Electrolyte Membranes. Electrocatalysis 2017, 8, 501–508.[138] Macauley, N.; Ghassemzadeh, L.; Lim, C.; Watson, M.; Kolodziej, J.; Lauritzen, M.; Holdcroft, S.;Kjeang, E. Pt Band Formation Enhances the Stability of Fuel Cell Membranes. ECS Electrochem-istry Letters 2013, 2, F33–F35.[139] Macauley, N.; Wong, K. H.; Watson, M.; Kjeang, E. Favorable effect of in-situ generated platinumin the membrane on fuel cell membrane durability. Journal of Power Sources 2015, 299, 139–148.[140] Macauley, N.; Alavijeh, A. S.; Watson, M.; Kolodziej, J.; Lauritzen, M.; Knights, S.; Wang, G.;Kjeang, E. Accelerated Membrane Durability Testing of Heavy Duty Fuel Cells. Journal of theElectrochemical Society 2014, 162, F98–F107.[141] Helmly, S.; Hiesgen, R.; Morawietz, T.; Yuan, X.-Z.; Wang, H.; Andreas Friedrich, K. MicroscopicInvestigation of Platinum Deposition in PEMFC Cross-Sections Using AFM and SEM. Journal ofthe Electrochemical Society 2013, 160, F687–F697.[142] Helmly, S.; Ohnmacht, B.; Hiesgen, R.; Gulzow, E.; Friedrich, K. A. Influence of Platinum Pre-cipitation on Properties and Degradation of Nafion® Membranes. ECS Transactions 2013, 58,969–990.[143] Le Penven, R.; Levason, W.; Pletcher, D. Studies of platinum electroplating baths Part I: Thechemistry of a platinum tetrammine bath. Journal of Applied Electrochemistry 1992, 22, 415–420.[144] Rao, C. R. K.; Pushpavanam, M. Electrodeposition of platinum from platinum ammine basedbaths. 2006, 84, 79–82.192References[145] Basirun, W. J.; Pletcher, D. Studies of platinum electroplating baths Part VI: Influence of someexperimental parameters on deposit quality. Journal of Applied Electrochemistry 1997, 28, 167–172.[146] Chattot, R.; Asset, T.; Drnec, J.; Bordet, P.; Nelayah, J.; Dubau, L.; Maillard, F. Atomic-ScaleSnapshots of the Formation and Growth of Hollow PtNi/C Nanocatalysts. Nano Letters 2017, 17,2447–2453.[147] Berners-Price, S. J.; Applewood, T. G. In Platinum-based drugs in cancer therapy; Farrell, N.,Kelland, L. R., Eds.; Humana Press, 2000; p 341.[148] Elding, L. I. Preparation and Properties of the Tetraaqua-Platinum(II) Ion in Perchloric Acid Solu-tion. Inorganica Chimica Acta 1976, 20, 65–69.[149] Gregory, A. J.; Levason, W.; Noftle, R. E.; LePenven, R.; Pletcher, D. Studies of platinum electro-plating baths .3. The electrochemistry of Pt(NH3)(4-x)(H2O)(x)(2+) and PtCl4-x(H2O)(x)((2-x)-).Journal of Electroanalytical Chemistry 1995, 399, 105–113.[150] Lippert, B. In Cisplatin: Chemistry and Biochemistry of a Leading Anticancer Drug; Lippert, B.,Ed.; Verlag Helvetica Chimica Acta: Zürich, 2006; pp 1–563.[151] Priqueler, J. R. L.; Butler, I. S.; Rochon, F. D. An Overview of 195 Pt Nuclear Magnetic ResonanceSpectroscopy. Applied Spectroscopy Reviews 2006, 41, 185–226.[152] Millet, P.; Andolfatto, F.; Durand, R. Preparation of solid polymer electrolyte composites: investi-gation of the precipitation process. Journal of Applied Electrochemistry 1995, 25, 233–239.[153] Fisher, E. A.; Leung, K. K.; Casanova-Moreno, J.; Masuda, T.; Young, J.; Bizzotto, D. Quantifyingthe Selective Modification of Au(111) Facets via Electrochemical and Electroless Treatments forManipulating Gold Nanorod Surface Composition. Langmuir 2017, 33, 12887–12896.[154] Djokic, S. S. In Modern Aspects of Electrochemistry; Conway, B. E., White, R. E., Eds.; KluwerAcademic Publishers: Boston, 2002; pp 51–133.[155] Ma, J.; Choudhury, N. A.; Sahai, Y. A comprehensive review of direct borohydride fuel cells. Re-newable and Sustainable Energy Reviews 2010, 14, 183–199.[156] Guella, G.; Patton, B.; Miotello, A. Kinetic features of the platinum catalyzed hydrolysis of sodiumborohydride from11B NMR measurements. Journal of Physical Chemistry C 2007, 111, 18744–18750.193References[157] Liu, B. H.; Li, Z. P. A review: Hydrogen generation from borohydride hydrolysis reaction. Journalof Power Sources 2009, 187, 527–534.[158] Wei, H.; Huang, K.; Zhang, L.; Ge, B.; Wang, D.; Lang, J.; Ma, J.; Wang, D.; Zhang, S.; Li, Q.;Zhang, R.; Hussain, N.; Lei, M.; Liu, L. M.; Wu, H. Ice Melting to Release Reactants in SolutionSyntheses. Angewandte Chemie 2018, 57, 3354–3359.[159] Thanh, N. T. K.; Maclean, N.; Mahiddine, S. Mechanisms of nucleation and growth of nanoparticlesin solution. Chemical Reviews 2014, 114, 7610–7630.[160] Yamamoto, J. The Sodium Borohydride Digest; 2003; pp 1–212.[161] Domenech, B.; Munoz, M.; Muraviev, D. N.; Macanas, J. Uncommon patterns in Nafion filmsloaded with silver nanoparticles. Chemical Communications 2014, 50, 4693–4695.[162] International, A.ASTMD3359-17, Standard Test Methods for Rating Adhesion by Tape Test; 2017.[163] Grillo, F.; Van Bui, H.; Moulijn, J. A.; Kreutzer, M. T.; Van Ommen, J. R. Understanding andControlling the Aggregative Growth of Platinum Nanoparticles in Atomic Layer Deposition: AnAvenue to Size Selection. Journal of Physical Chemistry Letters 2017, 8, 975–983.[164] Viswanatha, R.; Sarma, D. D. Nanomaterials Chemistry: Recent Developments and New Direc-tions; Wiley, 2007; pp 139–170.[165] Chen, X.; Schröder, J.; Hauschild, S.; Rosenfeldt, S.; Dulle, M.; Förster, S. SimultaneousSAXS/WAXS/UV-Vis Study of the Nucleation and Growth of Nanoparticles: A Test of ClassicalNucleation Theory. Langmuir 2015, 31, 11678–11691.[166] Sun, Y.; Ren, Y.; Haeffner, D. R.; Almer, J. D.; Wang, L.; Yang, W.; Truong, T. T. Nanophase evolu-tion at semiconductor/electrolyte interface in situ probed by time-resolved high-energy synchrotronX-ray diffraction. Nano Letters 2010, 10, 3747–3753.[167] Abécassis, B.; Testard, F.; Spalla, O.; Barboux, P. Probing in situ the Nucleation and Growth ofGold Nanoparticles by Small-Angle X-ray Scattering. Nano Letters 2007, 7, 1723–1727.[168] Wu, L.; Fournier, A. P.; Willis, J. J.; Cargnello, M.; Tassone, C. J. In Situ X-ray Scattering Guidesthe Synthesis of Uniform PtSn Nanocrystals. Nano Letters 2018, 4053–4057.194References[169] Peng, S.; Okasinski, J. S.; Almer, J. D.; Ren, Y.; Wang, L.; Yang, W.; Sun, Y. Real-time probingof the synthesis of colloidal silver nanocubes with time-resolved high-energy synchrotron X-raydiffraction. Journal of Physical Chemistry C 2012, 116, 11842–11847.[170] Harada, M.; Kamigaito, Y. Nucleation and aggregative growth process of platinum nanoparticlesstudied by in situ quick xafs spectroscopy. Langmuir 2012, 28, 2415–2428.[171] Langford, J. I.; Wilson, A. J. C. Scherrer after sixty years: A survey and some new results in thedetermination of crystallite size. Journal of Applied Crystallography 1978, 11, 102–113.[172] Avanesian, T.; Dai, S.; Kale, M. J.; Graham, G. W.; Pan, X.; Christopher, P. Quantitative andAtomic-Scale View of CO-Induced Pt Nanoparticle Surface Reconstruction at Saturation Cover-age via DFT Calculations Coupled with in Situ TEM and IR. Journal of the American ChemicalSociety 2017, 139, 4551–4558.[173] Abedini, A.; Bakar, A. A. A.; Larki, F.; Menon, P. S.; Islam, M. S.; Shaari, S. Recent Advances inShape-Controlled Synthesis of Noble Metal Nanoparticles by Radiolysis Route. Nanoscale Re-search Letters 2016, 11.[174] Liao, H. G.; Cui, L.; Whitelam, S.; Zheng, H. Real-time imaging of Pt3Fe nanorod growth in solu-tion. Science 2012, 336, 1011–1014.[175] Morris, D. R. P.; Liu, S. P.; Villegas Gonzalez, D.; Gostick, J. T. Effect of water sorption on the elec-tronic conductivity of porous polymer electrolyte membrane fuel cell catalyst layers. ACS AppliedMaterials and Interfaces 2014, 6, 18609–18618.[176] Soboleva, T.; Malek, K.; Xie, Z.; Navessin, T.; Holdcroft, S. PEMFC catalyst layers: The role ofmicropores and mesopores on water sorption and fuel cell activity. ACS Applied Materials andInterfaces 2011, 3, 1827–1837.[177] Guo, S. et al. Electrodeposition. ACS Applied Materials and Interfaces 2013, 5, 1–5.[178] Hiesgen, R.; Helmly, S.; Galm, I.; Morawietz, T.; Handl, M.; Andreas, F. K. Microscopic analysisof current and mechanical properties of nafion studied by atomic force microscopy. Membranes2012, 2, 783–803.[179] Drouin, D.; Couture, A. R.; Joly, D.; Tastet, X.; Aimez, V.; Gauvin, R. CASINO V2.42 - A fast andeasy-to-use modeling tool for scanning electron microscopy and microanalysis users. Scanning2007, 29, 92–101.195References[180] Goswami, S.; Klaus, S.; Benziger, J. Wetting and absorption of water drops on nafion films. Lang-muir 2008, 24, 8627–8633.[181] O’Dea, J. R.; Economou, N. J.; Buratto, S. K. Surface morphology of nafion at hydrated anddehydrated conditions. Macromolecules 2013, 46, 2267–2274.[182] James, P. J.; Elliott, J. A.; McMaster, J.; Newton, J. M.; Elliott, A. M. S.; Hanna, S.; Miles, M. J.Hydration of Nafion studied by AFM and X-ray scattering. Journal of Materials Science 2000, 35,5111–5119.[183] James, P. J.; Antognozzi, M.; Tamayo, J.; McMaster, T. J.; Newton, J. M.; Miles, M. J. Interpretationof contrast in tapping mode AFM and shear force microscopy. A study of Nafion. Langmuir 2001,17, 349–360.[184] McLean, R. S.; Doyle, M.; Sauer, B. B. High-Resolution Imaging of Ionic Domains and CrystalMorphology in Ionomers Using AFM Techniques. Macromolecules 2000, 33, 6541–6550.[185] Kwon, O.; Wu, S.; Zhu, D. Characterization of Proton Conduction in Nafion Membranes UsingCurrent Sensing Atomic Force Microscopy. ECS Transactions 2010, 33, 1035–1044.[186] El-Kharouf, A.; Mason, T. J.; Brett, D. J. L.; Pollet, B. G. Ex-situ characterisation of gas diffusionlayers for proton exchange membrane fuel cells. Journal of Power Sources 2012, 218, 393–404.[187] Fazeli, M.; Hinebaugh, J.; Bazylak, A. Investigating Inlet Condition Effects on PEMFC GDL LiquidWater Transport through Pore Network Modeling. Journal of the Electrochemical Society 2015,162, F661–F668.[188] van Spronsen, M. A.; Frenken, J. W. M.; Groot, I. M. N. Observing the oxidation of platinum.NatureCommunications 2017, 8, 429.[189] Parkinson, C. R.; Walker, M.; McConville, C. F. Reaction of atomic oxygen with a Pt(1 1 1) surface:Chemical and structural determination using XPS, CAICISS and LEED. Surface Science 2003,545, 19–33.[190] Cheng, T. T. H.; Colbow, V.; Wessel, S.; Chuy, C.; He, P. Impacts of Ru Dissolution and Crossoveron Polymer Electrolyte Membrane Fuel Cell Performance and Anode Functionality. ECS Transac-tions 2010, 28, 243–251.196References[191] Zhang, L.; Kim, J.; Singh, S.; Tsay, K.; Zhang, J. Novel Half-cell Design and Fabrication for In-situEvaluation of PEM Fuel Cell Electrocatalysts. International Journal of Green Energy 2012, 5075,120917081427006.[192] Zalitis, C. M.; Kramer, D.; Kucernak, A. R. Electrocatalytic performance of fuel cell reactions atlow catalyst loading and high mass transport. Physical Chemistry Chemical Physics 2013, 15,4329–4340.[193] Gasteiger, H.; Kocha, S.; Sompalli, B.; Wagner, F. Activity benchmarks and requirements for Pt,Pt-alloy, and non-Pt oxygen reduction catalysts for PEMFCs. Applied Catalysis B: Environmental2005, 56, 9–35.[194] Zhang, C.; Hwang, S. Y.; Trout, A.; Peng, Z. Solid-state chemistry-enabled scalable productionof octahedral Pt-Ni alloy electrocatalyst for oxygen reduction reaction. Journal of the AmericanChemical Society 2014, 136, 7805–7808.[195] Zhou, Z.-Y.; Kang, X.; Song, Y.; Chen, S. Enhancement of the electrocatalytic activity of Ptnanoparticles in oxygen reduction by chlorophenyl functionalization. Chemical Communications2012, 48, 3391.[196] Debe, M. K. In Handbook of Fuel Cells Fundamentals, Technology and Applications; Vielstich, W.,Lamm, A., Gasteiger, H. A., Eds.; Wiley: New York, NY, 2003; Chapter 45.[197] O’Hayre, R.; Lee, S. J.; Cha, S. W.; Prinz, F. B. A sharp peak in the performance of sputteredplatinum fuel cells at ultra-low platinum loading. Journal of Power Sources 2002, 109, 483–493.[198] Ohma, A.; Fushinobu, K.; Okazaki, K. Influence of Nafion film on oxygen reduction reaction and hy-drogen peroxide formation on Pt electrode for proton exchangemembrane fuel cell.ElectrochimicaActa 2010, 55, 8829–8838.[199] McGovern, M. S.; Garnett, E. C.; Rice, C.; Masel, R. I.; Wieckowski, a. Effects of Nafion as abinding agent for unsupported nanoparticle catalysts. Journal of Power Sources 2003, 115, 35–39.[200] Shinozaki, K.; Pivovara, B. S.; Kocha, S. S. Enhanced Oxygen Reduction Activity on Pt/C forNafion-free, Thin, Uniform Films in Rotating Disk Electrode Studies. ECS Transactions 2013, 58,15–26.197References[201] Yuan, X. Z.; Zhang, S.; Sun, J. C.; Wang, H. A review of accelerated conditioning for a polymerelectrolyte membrane fuel cell. Journal of Power Sources 2011, 196, 9097–9106.[202] Kocha, S. S.; Zack, J. W.; Alia, S. M.; Neyerlin, K. C.; Pivovar, B. S. Influence of Ink Composition onthe Electrochemical Properties of Pt/C Electrocatalysts. ECS Transactions 2013, 50, 1475–1485.[203] Kocha, S. S.; Shinozaki, K.; Zack, J. W.; Myers, D. J.; Kariuki, N. N.; Nowicki, T.; Stamenkovic, V.;Kang, Y.; Li, D.; Papageorgopoulos, D. Best Practices and Testing Protocols for BenchmarkingORR Activities of Fuel Cell Electrocatalysts Using Rotating Disk Electrode. Electrocatalysis 2017,8, 366–374.[204] Shinozaki, K.; Morimoto, Y.; Pivovar, B. S.; Kocha, S. S. Suppression of oxygen reduction reactionactivity on Pt-based electrocatalysts from ionomer incorporation. Journal of Power Sources 2016,325, 745–751.[205] Iqbal, S.; Podgaynyy, N.; Ahmed, M.; Attard, G. A.; Baltruschat, H. Surface morphological studiesof Nafion®/Pt(100) interface. Electrochimica Acta 2014, 144, 141–146.[206] Zeng, J.; Jean, D.-i.; Ji, C.; Zou, S. In Situ Surface-Enhanced Raman Spectroscopic Studies ofNafion Adsorption on Au and Pt Electrodes. Langmuir 2012, 28, 957–964.[207] Kodama, K.; Motobayashi, K.; Shinohara, A.; Hasegawa, N.; Kudo, K.; Jinnouchi, R.; Osawa, M.;Morimoto, Y. Effect of the Side-Chain Structure of Perfluoro-Sulfonic Acid Ionomers on the OxygenReduction Reaction on the Surface of Pt. ACS Catalysis 2018, 8, 694–700.[208] Gómez-Marín, A. M.; Berná, A.; Feliu, J. M. Spectroelectrochemical Studies of the Pt(111)/NafionInterface Cast Electrode. Journal of Physical Chemistry C 2010, 114, 20130–20140.[209] Hanawa, H.; Kunimatsu, K.; Watanabe, M.; Uchida, H. In Situ ATR-FTIR Analysis of the Structureof Nafion Pt/C and Nafion Pt 3 Co/C Interfaces in Fuel Cell. Journal of Physical Chemistry C 2012,116, 21401–21406.[210] Kendrick, I.; Kumari, D.; Yakaboski, A.; Dimakis, N.; Smotkin, E. S. Elucidating the ionomer-electrified metal interface. Journal of the American Chemical Society 2010, 132, 17611–17616.[211] Yagi, I.; Inokuma, K.; Kimijima, K.; Notsu, H. Molecular Structure of Buried PerfluorosulfonatedIonomer/Pt Interface Probed by Vibrational Sum Frequency Generation Spectroscopy. Journal ofPhysical Chemistry C 2014, 118, 26182–26190.198References[212] Masuda, T.; Fukumitsu, H.; Kondo, T.; Naohara, H.; Tamura, K.; Sakata, O.; Uosaki, K. Struc-ture of Pt(111)/ionomer membrane interface and its bias-induced change in membrane electrodeassembly. Journal of Physical Chemistry C 2013, 117, 12168–12171.[213] Masuda, T.; Sonsudin, F.; Singh, P. R.; Naohara, H.; Uosaki, K. Potential-Dependent Adsorp-tion and Desorption of Perfluorosulfonated Ionomer on a Platinum Electrode Surface Probed byElectrochemical Quartz Crystal Microbalance and Atomic Force Microscopy. Journal of PhysicalChemistry C 2013, 117, 15704–15709.[214] Garrick, T. R.; Moylan, T. E.; Yarlagadda, V.; Kongkanand, A. Characterizing Electrolyte and Plat-inum Interface in PEM Fuel Cells Using CO Displacement. Journal of the Electrochemical Society2017, 164, F60–F64.[215] Tymoczko, J.; Calle-Vallejo, F.; Colic, V.; Koper, M. T. M.; Schuhmann, W.; Bandarenka, A. S.Oxygen Reduction at a Cu-Modified Pt(111) Model Electrocatalyst in Contact with Nafion Polymer.ACS Catalysis 2014, 4, 3772–3778.[216] Kodama, K.; Jinnouchi, R.; Suzuki, T.; Murata, H.; Hatanaka, T.; Morimoto, Y. Increase in adsorp-tivity of sulfonate anions on Pt (111) surface with drying of ionomer. Electrochemistry Communi-cations 2013, 36, 26–28.[217] Ahmed, M.; Morgan, D.; Attard, G. A.; Wright, E.; Thompsett, D.; Sharman, J. Unprecedentedstructural sensitivity toward average terrace width: Nafion adsorption at Pt{ hkl } electrodes. Jour-nal of Physical Chemistry C 2011, 115, 17020–17027.[218] Subbaraman, R.; Strmcnik, D.; Stamenkovic, V.; Markovic, N. M. Three phase interfaces at elec-trified metal-solid electrolyte systems 1. study of the pt(hkl)-nafion interface. Journal of PhysicalChemistry C 2010, 114, 8414–8422.[219] Markovic, N. M.; Schmidt, T. J.; Stamenkovic, V.; Ross, P. N. Oxygen Reduction Reaction on Ptand Pt Bimetallic Surfaces: A Selective Review. Fuel Cells 2001, 1, 105–116.[220] Subbaraman, R.; Strmcnik, D.; Paulikas, A. P.; Stamenkovic, V. R.; Markovic, N. M. Oxygen re-duction reaction at three-phase interfaces. ChemPhysChem 2010, 11, 2825–2833.[221] Sleightholme, A. E. S.; Kucernak, A. An anomalous peak observed in the electrochemistry of theplatinum/perfluorosulfonic acid membrane interface. Electrochimica Acta 2011, 56, 4396–4402.199References[222] Curnick, O. J.; Pollet, B. G.; Mendes, P. M. Nafion-stabilised Pt/C electrocatalysts with efficientcatalyst layer ionomer distribution for proton exchange membrane fuel cells. RSC Advances 2012,2, 8368.[223] Liu, Z.; Tian, Z. Q.; Jiang, S. P. Synthesis and characterization of Nafion-stabilized Pt nanoparti-cles for polymer electrolyte fuel cells. Electrochimica Acta 2006, 52, 1213–1220.[224] Sarma, L. S.; Lin, T. D.; Tsai, Y. W.; Chen, J. M.; Hwang, B. J. Carbon-supported Pt-Ru catalystsprepared by the Nafion stabilized alcohol-reduction method for application in direct methanol fuelcells. Journal of Power Sources 2005, 139, 44–54.[225] Gupta, G.; Sharma, S.; Mendes, P. M. Nafion-stabilised bimetallic Pt-Cr nanoparticles as electro-catalysts for proton exchange membrane fuel cells (PEMFCs). RSC Advances 2016, 6, 82635–82643.[226] Novitski, D.; Holdcroft, S. Determination of O2 Mass Transport at the Pt | PFSA Ionomer Interfaceunder Reduced Relative Humidity. ACS Applied Materials and Interfaces 2015, 7, 27314–27323.[227] Novitski, D.; Xie, Z.; Holdcroft, S. Time-Dependent Mass Transport for O2 Reduction at the PtPerfluorosulfonic Acid Ionomer Interface. ECS Electrochemistry Letters 2014, 4, F9–F12.[228] Cherevko, S.; Keeley, G. P.; Geiger, S.; Zeradjanin, A. R.; Hodnik, N.; Kulyk, N.; Mayrhofer, K. J.Dissolution of Platinum in the Operational Range of Fuel Cells.ChemElectroChem 2015, 2, 1471–1478.[229] Proietti, E.; Jaouen, F.; Lefèvre, M.; Larouche, N.; Tian, J.; Herranz, J.; Dodelet, J. P. Iron-basedcathode catalyst with enhanced power density in polymer electrolyte membrane fuel cells. NatureCommunications 2011, 2.[230] Neitzel, A.; Figueroba, A.; Lykhach, Y.; Skála, T.; Vorokhta, M.; Tsud, N.; Mehl, S.; Ševčíková, K.;Prince, K. C.; Neyman, K. M.; Matolín, V.; Libuda, J. Atomically Dispersed Pd, Ni, and Pt Speciesin Ceria-Based Catalysts: Principal Differences in Stability and Reactivity. Journal of PhysicalChemistry C 2016, 120, 9852–9862.[231] Cheng, N.; Stambula, S.; Wang, D.; Banis, M. N.; Liu, J.; Riese, A.; Xiao, B.; Li, R.; Sham, T. K.;Liu, L. M.; Botton, G. A.; Sun, X. Platinum single-atom and cluster catalysis of the hydrogenevolution reaction. Nature Communications 2016, 7, 1–9.200References[232] Kibsgaard, J.; Gorlin, Y.; Chen, Z.; Jaramillo, T. F. Meso-structured platinum thin films: Activeand stable electrocatalysts for the oxygen reduction reaction. Journal of the American ChemicalSociety 2012, 134, 7758–7765.[233] Cullen, D. A.; Lopez-Haro, M.; Bayle-Guillemaud, P.; Guetaz, L.; Debe, M. K.; Steinbach, A. J.Linking morphology with activity through the lifetime of pretreated PtNi nanostructured thin filmcatalysts. Journal of Materials Chemistry A 2015, 3, 11660–11667.[234] Papandrew, A. B.; Atkinson, R. W.; Goenaga, G. A.; Kocha, S. S.; Zack, J. W.; Pivovar, B. S.;Zawodzinski, T. A. Oxygen Reduction Activity of Vapor-Grown Platinum Nanotubes. Journal ofthe Electrochemical Society 2013, 160, F848–F852.[235] Chen, C. et al. Highly crystalline multimetallic nanoframes with three-dimensional electrocatalyticsurfaces. Science 2014, 343, 1339–1343.[236] Godínez-Salomón, F.; Mendoza-Cruz, R.; Arellano-Jimenez, M. J.; Jose-Yacaman, M.;Rhodes, C. P. Metallic Two-Dimensional Nanoframes: Unsupported Hierarchical Nickel-PlatinumAlloy Nanoarchitectures with Enhanced Electrochemical Oxygen Reduction Activity and Stability.ACS Applied Materials and Interfaces 2017, 9, 18660–18674.[237] Clavilier, J.; Feliu, J. M.; Aldaz, A. An irreversible structure sensitive adsorption step in bismuthunderpotential deposition at platinum electrodes. Journal of Electroanalytical Chemistry and In-terfacial Electrochemistry 1988, 243, 419–433.[238] Mustain, W. E.; Kim, H.; Narayanan, V.; Osborn, T.; Kohl, P. A. Electroless Deposition and Charac-terization of Pt[sub x]Ru[sub 1−x] Catalysts on Pt/C Nanoparticles for Methanol Oxidation. Journalof Fuel Cell Science and Technology 2010, 7, 041013.[239] Onodera, T.; Suzuki, S.; Takamori, Y.; Daimon, H. Improved methanol oxidation activity and stabil-ity of well-mixed PtRu catalysts synthesized by electroless plating method with addition of chelateligands. Applied Catalysis A: General 2010, 379, 69–76.[240] Maillard, F.; Lu, G. Q.; Wieckowski, A.; Stimming, U. Ru-decorated Pt surfaces as model fuelcell electrocatalysts for CO electrooxidation. Journal of Physical Chemistry B 2005, 109, 16230–16243.[241] Povar, I.; Spinu, O. Ruthenium redox equilibria: 3. Pourbaix diagrams for the systems Ru-H2Oand Ru-Cl–H2O. Journal of Electrochemical Science and Engineering 2016, 6, 145.201References[242] Lin, W. F.; Iwasita, T.; Vielstich, W. Catalysis of CO Electrooxidation at Pt, Ru, and PtRu Alloy. Anin Situ FTIR Study. Journal of Physical Chemistry B 1999, 103, 3250–3257.[243] Gasteiger, H. A.; Markovic, N.; Ross, P. N.; Cairns, E. J. Temperature-Dependent MethanolElectro-Oxidation on Well-Characterized Pt-Ru Alloys. Journal of the Electrochemical Society1994, 141, 1795.[244] Larson, C.; Smith, J. R. Recent trends in metal alloy electrolytic and electroless plating research:a review. Transactions of the IMF 2011, 89, 333–341.[245] Gasteiger, H.; Markovic, N.; Ross, P. N.; Cairns, E. J. CO Electrooxidation on Well-CharacterizedPt-Ru Alloys. Journal of Physical Chemistry 1994, 98, 617–625.[246] Mayrhofer, K. J. J.; Blizanac, B. B.; Arenz, M.; Stamenkovic, V. R.; Ross, P. N.; Markovic, N. M.The impact of geometric and surface electronic properties of Pt-catalysts on the particle size effectin electocatalysis. Journal of Physical Chemistry B 2005, 109, 14433–14440.[247] Arenz, M.; Mayrhofer, K. J. J.; Stamenkovic, V.; Blizanac, B. B.; Tomoyuki, T.; Ross, P. N.;Markovic, N. M. The Effect of the Particle Size on the Kinetics of CO Electrooxidation on HighSurface Area Pt Catalysts. Journal of the American Chemical Society 2005, 127, 6819–6829.[248] Mayrhofer, K. J. J.; Hanzlik, M.; Arenz, M. The influence of electrochemical annealing in COsaturated solution on the catalytic activity of Pt nanoparticles.Electrochimica Acta 2009, 54, 5018–5022.[249] Spendelow, J. S.; Lu, G. Q.; Kenis, P. J. A.; Wieckowski, A. Electrooxidation of adsorbed COon Pt(1 1 1) and Pt(1 1 1)/Ru in alkaline media and comparison with results from acidic media.Journal of Electroanalytical Chemistry 2004, 568, 215–224.[250] Brimaud, S.; Pronier, S.; Coutanceau, C.; Léger, J. M. New findings on CO electrooxidation atplatinum nanoparticle surfaces. Electrochemistry Communications 2008, 10, 1703–1707.[251] Ciapina, E. G.; Santos, S. F.; Gonzalez, E. R. Electrochemical CO stripping on nanosized Ptsurfaces in acid media: A review on the issue of peak multiplicity. Journal of ElectroanalyticalChemistry 2018, 815, 47–60.[252] Jerkiewicz, G. Electrochemical Hydrogen Adsorption and Absorption. Part 1: Under-potential De-position of Hydrogen. Electrocatalysis 2010, 1, 179–199.202References[253] Uchida, H.; Izumi, K.; Watanabe, M. Temperature dependence of CO-tolerant hydrogen oxidationreaction activity at Pt, Pt-Co, and Pt-Ru electrodes. Journal of Physical Chemistry B 2006, 110,21924–21930.[254] Garzon, F.; Lopes, T.; Rockward, T.; Sansiena, J.-M.; Kienitz, B.; Mukundan, R. The Impact ofImpurities on Long-Term PEMFC Performance. ECS Transactions 2009, 25, 1575–1583.[255] Garzon, F. Effects of Fuel and Air Impurities on PEM Fuel Cell Performance 2010 Annual MeritReview; 2011.[256] Imamura, D.; Ebata, D.; Hshimasa, Y.; Akai, M.; Watanabe, S. Impact of Hydrogen Fuel Impuritieson PEMFC Performance; 2007; pp 1–6.[257] Rajalakshmi, N.; Jayanth, T. T.; Dhathathreyan, K. S. Effect of Carbon Dioxide and Ammonia onPolymer Electrolyte Membrane Fuel Cell Stack Performance. Fuel Cells 2003, 3, 177–180.[258] Scanlon, D.; Fang, P.-p.; Chen, S.; Deng, H.; Pereira, C. M.; Yang, Z. Conductive Gold Nanopar-ticle Mirrors at Liquid / Liquid Interfaces. ACS Nano 2013, 9241–9248.[259] Strathmann, M.; Strathmann, M. Optically Transparent Porous Medium for Nondestructive Studiesof Microbial Biofilm Architecture and Transport Dynamics. Microbiology 2005, 71, 4801–4808.[260] DuPont, Nafion PFSA technical data sheets; 2009; pp 0–3.[261] Rakic, D.; Djurisic, B.; Elazar, J. M.; Majewski, M. L. Optical properties of metallic films for vertical-cavity optoelectronic devices. Applied optics 1998, 37, 5271–5283.[262] Aspnes, D. E. Optical properties of thin films. Thin Solid Films 1982, 89, 249–262.[263] Ung, T.; Liz-Marzan, L. M.; Mulvaney, P. Optical properties of thin films of Au@SiO2 particles.Journal of Physical Chemistry B 2001, 105, 3441–3452.[264] Huang, P.-J.; Su, G.-D. J. Making hexagonal electrodes on Nafion for deformable mirrors appli-cations. Proceedings of SPIE 2014, 9202, 92021A.[265] Farid, M.; Zhao, G.; Khuong, T. L.; Sun, Z. Z.; Ur Rehman, N.; Rizwan, M. Biomimetic Applicationsof Ionic Polymer Metal Composites (IPMC) Actuators - A Critical Review. Journal of Biomimetics,Biomaterials and Biomedical Engineering 2014, 20, 1–10.203References[266] Nguyen, T. T.; Goo, N. S.; Nguyen, V. K.; Yoo, Y.; Park, S. Design, fabrication, and experimentalcharacterization of a flap valve IPMC micropump with a flexibly supported diaphragm. Sensorsand Actuators, A: Physical 2008, 141, 640–648.[267] Tailoka, F.; Fray, D. J.; Kumar, R. V. Application of Nafion electrolytes for the detection of humidityin a corrosive atmosphere. Solid State Ionics 2003, 161, 267–277.[268] Korotcenkov, G.; Han, S. D.; Stetter, J. R. Review of Electrochemical Hydrogen Sensors.ChemicalReviews 2009, 109, 1402–1433.[269] Sakthivel, M.; Weppner, W. Electrode kinetics of amperometric hydrogen sensors for hydrogendetection at low parts per million level. Journal of Solid State Electrochemistry 2007, 11, 561–570.[270] Sakthivel, M.; Weppner, W. Development of a hydrogen sensor based on solid polymer electrolytemembranes. Sensors and Actuators, B: Chemical 2006, 113, 998–1004.[271] Kruusamäe, K.; Punning, A.; Aabloo, A.; Asaka, K. Self-Sensing Ionic Polymer Actuators: A Re-view. Actuators 2015, 4, 17–38.[272] Tiwari, R.; Garcia, E. The state of understanding of ionic polymer metal composite architecture:A review. Smart Materials and Structures 2011, 20.[273] Bhandari, B.; Lee, G. Y.; Ahn, S. H. A review on IPMC material as actuators and sensors: Fabri-cations, characteristics and applications. International Journal of Precision Engineering and Man-ufacturing 2012, 13, 141–163.[274] Palmre, V.; Kim, S. J.; Pugal, D.; Kim, K. Improving electromechanical output of IPMC by highsurface area Pd-Pt electrodes and tailored ionomer membrane thickness. International Journal ofSmart and Nano Materials 2014, 5, 99–113.[275] Shahinpoor, M.; Kim, K. J. Ionic polymer metal composites: IV. Industrial andmedical applications.Smart Materials and Structures 2005, 14, 197–214.[276] Kruusamäe, K.; Brunetto, P.; Punning, A.; Kodu, M.; Jaaniso, R.; Graziani, S.; Fortuna, L.;Aabloo, A. Electromechanical model for a self-sensing ionic polymer metal composite actuatingdevice with patterned surface electrodes. Smart Materials and Structures 2011, 20, 124001.204References[277] Riddle, R. O.; Jung, Y.; Kim, S.-M.; Song, S.; Stolpman, B.; Kim, K. J.; Leang, K. K. Sectored-electrode IPMC actuator for bending and twisting motion. Electroactive Polymer Actuators andDevices. 2010; p 764221.[278] Chen, Z.; Tan, X. Monolithic fabrication of ionic polymer-metal composite actuators capable ofcomplex deformation. Sensors and Actuators, A: Physical 2010, 157, 246–257.[279] Chen, L.; Zhao, W.; Jiao, Y.; He, X.; Wang, J.; Zhang, Y. Characterization of Ag/Pt core-shellnanoparticles by UV-vis absorption, resonance light-scattering techniques. Spectrochimica ActaA 2007, 68, 484–490.[280] Ahn, C.-Y.; Jang, S.; Cho, Y.-H.; Choi, J.; Kim, S.; Kim, S. M.; Sung, Y.-E.; Choi, M. Guidedcracking of electrodes by stretching prism-patterned membrane electrode assemblies for high-performance fuel cells. Scientific Reports 2018, 8, 1257.[281] Sasahara, J.; Cha, S. W.; Chang-Chien, A.; Kubota, T.; Kuriyama, N.; Lee, S.-J. J.; Liu, Y.;O’Hayre, R.; Prinz, F. B.; Saito, Y. Fuel cell with patterned electrolyte/electrode interface.US6835488B2, 2004.[282] Susac, D.; Berejnov, V.; Hitchcock, A. P.; Stumper, J. STXM Study of the Ionomer Distribution inthe PEM Fuel Cell Catalyst Layers. ECS Transactions. 2011; pp 629–635.[283] Susac, D.; Berejnov, V.; Stumper, J.; Hitchcock, A. P. STXM Characterization of PEM Fuel CellCatalyst Layers. ECS Transactions 2012, 50, 405–413.[284] Ade, H.; Smith, A.; Cameron, S.; Cieslinski, R.; Mitchell, G.; Hsiao, B.; Rightor, E. X-ray mi-croscopy in polymer science: prospects of a ânew’ imaging technique. Polymer 1995, 36, 1843–1848.[285] Ade, H.; Urquhart, S. NEXAFS spectroscopy and microscopy of natural and synthetic polymers.Adv. Ser. Phys. Chem. 2002, 12A, 285–355.[286] Rightor, E. G.; Hitchcock, A. P.; Ade, H.; Leapman, R. D.; Urquhart, S. G.; Smith, A. P.; Mitchell, G.;Fischer, D.; Shin, H. J.; Warwick, T. Spectromicroscopy of Poly(ethylene terephthalate): Compar-ison of Spectra and Radiation Damage Rates in X-ray Absorption and Electron Energy Loss. TheJournal of Physical Chemistry B 1997, 101, 1950–1960.[287] Hitchcock, A. P. In Handbook on Nanoscopy Vol.II; Tendeloo, G. V., Dyck, V. D., Pennycook, S. J.,Eds.; Wiley-VCH Verlag GmbH & Co. KGaA, 2012; Chapter 22, pp 745–791.205References[288] Ade, H. In VacuumUltraviolet Spectroscopy II; Samson, J., Ederer, D., Eds.; Elsevier, 1998; Chap-ter 11, pp 225–262.[289] Wang, J.; Botton, G. A.; West, M. M.; Hitchcock, A. P. Quantitative evaluation of radiation dam-age to polyethylene terephthalate by soft X-rays and high-energy electrons. Journal Of PhysicalChemistry B 2009, 113, 1869–76.[290] Wang, J.; Morin, C.; Li, L.; a.P. Hitchcock,; Scholl, a.; Doran, a. Radiation damage in soft X-raymicroscopy. Journal of Electron Spectroscopy and Related Phenomena 2009, 170, 25–36.[291] Schulze, M.; Lorenz, M.; Wagner, N.; Guelzow, E. XPS analysis of the degradation of Nafion.Fresenius’ Journal of Analytical Chemistry 1999, 365, 106–113.[292] Almeida, S. H.; Kawano, Y. Effects of X-ray radiation on Nafion membrane. Polymer Degradationand Stability 1998, 62, 291–297.[293] Wu, J.; Melo, L. G. A.; Zhu, X.; West, M. M.; Berejnov, V.; Susac, D.; Stumper, J.; Hitch-cock, A. P. 4D imaging of polymer electrolyte membrane fuel cell catalyst layers by soft X-rayspectro-tomography. Journal of Power Sources 2018, 381, 72–83.[294] Allen, F. I.; Comolli, L. R.; Kusoglu, A.; Modestino, M. A.; Minor, A. M.; Weber, A. Z. Morphology ofHydrated As-Cast Nafion Revealed through Cryo Electron Tomography. ACS Macro Letters 2015,4, 1–5.[295] Cullen, D. A.; Koestner, R.; Kukreja, R. S.; Liu, Z. Y.; Minko, S.; Trotsenko, O.; Tokarev, A.; Gue-taz, L.; Meyer, H. M.; Parish, C. M.; More, K. L. Imaging and Microanalysis of Thin Ionomer Layersby Scanning Transmission ElectronMicroscopy. Journal of the Electrochemical Society 2014, 161,F1111–F1117.[296] Lee, V.; Susac, D.; Kundu, S.; Berejnov, V.; Atanasoski, R. T.; Hitchcock, A. P.; Stumper, J. STXMCharacterization of Nanostructured Thin Film Anode Before and After Start-Up Shutdown andReversal Tests. ECS Transactions 2013, 58, 473–479.[297] Paul, D. K.; Giorgi, J. B.; Karan, K. Chemical and Ionic Conductivity Degradation of Ultra-ThinIonomer Film by X-ray Beam Exposure. Journal of the Electrochemical Society 2013, 160, F464–F469.206References[298] Berejnov, V.; Rubinstein, B.; Melo, L. G. A.; Hitchcock, A. P. First-principles X-ray absorption dosecalculation for time-dependent mass and optical density. Journal of Synchrotron Radiation 2018,25.[299] Zhang, X.; Jacobsen, C.; Lindaas, S.; Williams, S. Exposure strategies for polymethyl methacry-late from in situ x-ray absorption near edge structure spectroscopy. Journal of Vacuum Science &Technology B: Microelectronics and Nanometer Structures 1995, 13, 1477.[300] Leontowich, A. F. G.; Hitchcock, A. P.; Tyliszczak, T.; Weigand, M.; Wang, J.; Karunakaran, C.Accurate dosimetry in scanning transmission X-ray microscopes via the cross-linking thresholddose of poly(methyl methacrylate). Journal of Synchrotron Radiation 2012, 19, 976–987.[301] Leontowich, A. F. G. Utility of the G value and the critical dose to soft X-ray radiation damage ofpolyacrylonitrile. Radiation Physics and Chemistry 2013, 90, 87–91.[302] PermaPure, Nafion: Physical and Chemical Properties. http://www.permapure.com/products/nafion-tubing/nafion-physical-and-chemical-properties/.[303] Wetteland, C. J.; Gosnell, G.; Hollander, M. G.; Tesmer, J. R.; Mooday, R.; Wang, Y. Q. Ionbeam irradiation of polytetrafluroethylene-loaded polyetheretherketone. Nuclear Instruments andMethods in Physics Research B 2006, 250, 396–401.[304] Yakovlev, S.; Balsara, N. P.; Downing, K. H. Insights on the study of nafion nanoscale morphologyby transmission electron microscopy. Membranes 2013, 3, 424–439.[305] Yan, Z. B.; Hayes, R.; Melo, L. G. A.; Goward, G. R.; Hitchcock, A. P. X-ray Absorption and Solid-State NMR Spectroscopy of Fluorinated Proton Conducting Polymers. The Journal of PhysicalChemistry C 2018, 122, 3233–3244.[306] Urquhart, S. G.; Smith, A. P.; Ade, H. W.; Hitchcock, A. P.; Rightor, E. G.; Lidy, W. Near-Edge X-rayAbsorption Fine Structure Spectroscopy of MDI and TDI Polyurethane Polymers. The Journal ofPhysical Chemistry B 1999, 103, 4603–4610.[307] Webber, M.; Dimakis, N.; Kumari, D.; Fuccillo, M.; Smotkin, E. S. Mechanically coupled internalcoordinates of ionomer vibrational modes. Macromolecules 2010, 43, 5500–5502.[308] Cable, K. M.; Mauritz, K. A.; Moore, R. B. Effects of hydrophilic and hydrophobic counterions onthe Coulombic interactions in perfluorosulfonate ionomers. Journal of Polymer Science Part B:Polymer Physics 1995, 33, 1065–1072.207References[309] Almeida, S. H.; Kawano, Y. Effects of X-ray radiation on Nafion membrane. Polymer Degradationand Stability 1998, 62, 291–297.[310] Nagao, Y. Highly Oriented Sulfonic Acid Groups in a Nafion Thin Film on Si Substrate. Journal ofPhysical Chemistry C 2013, 117, 3294–3297.[311] Leontowich, A. F. G.; Hitchcock, A. P. Secondary electron deposition mechanism of carbon con-tamination. Journal of Vacuum Science & Technology B, Nanotechnology and Microelectronics:Materials, Processing, Measurement, and Phenomena 2012, 30, 030601.[312] Baruchel, J., Hodeau, J.-L., Lehmann, M., Regnard, J.-R., Schlenker, C., Eds. Neutron and syn-chrotron radiation for condensed matter studies: Theory, instruments and methods; Springer-Verlag, 1993.[313] Forsythe, J. S.; Hill, D. J. T. Radiation chemistry of fluoropolymers. Progress in Polymer Science(Oxford) 2000, 25, 101–136.[314] Escobedo, G.; Raiford, K.; Nagarajan, G. S.; Schwiebert, K. E. Strategies for Mitigation of PFSAPolymer Degradation in PEM Fuel Cells. ECS Transactions 2006, 1, 303–311.[315] Pugmire, D. L.; Wetteland, C. J.; Duncan, W. S.; Lakis, R. E.; Schwartz, D. S. Cross-linking ofpolytetrafluoroethylene during room-temperature irradiation. Polymer Degradation and Stability2009, 94, 1533–1541.[316] Feng, M.; Qu, R.; Wei, Z.; Wang, L.; Sun, P.; Wang, Z. Characterization of the thermolysis prod-ucts of Nafion membrane: A potential source of perfluorinated compounds in the environment.Scientific reports 2015, 5, 9859.[317] Yakovlev, S.; Balsara, N. P.; Downing, K. H. Insights on the study of nafion nanoscale morphologyby transmission electron microscopy. Membranes 2013, 3, 424–439.[318] Schulze, M.; Lorenz, M.; Wagner, N.; Gülzow, E. XPS analysis of the degradation of Nafion.Fresenius Journal of Analytical Chemistry 1999, 365, 106–113.[319] Ishimoto, T.; Koyama, M. A review of molecular-level mechanism of membrane degradation in thepolymer electrolyte fuel cell. Membranes 2012, 2, 395–414.[320] Ono, Y.; Nagao, Y. Interfacial Structure and Proton Conductivity of Nafion at the Pt-DepositedSurface. Langmuir 2016, 32, 352–358.208References[321] Tant, M. R.; Darst, K. P.; Lee, K. D.; Martin, C. W. In Multiphase Polymers: Blends and Ionomers;Utracki, L. A., Weiss, R., Eds.; 1989; pp 370–400.[322] Kusoglu, A.; Weber, A. Z. New Insights into Perfluorinated Sulfonic-Acid Ionomers. Chemical Re-views 2017, 117, 987–1104.[323] Ghassemzadeh, L.; Kreuer, K. D.; Maier, J.; Müller, K. Chemical degradation of Nafion mem-branes under mimic fuel cell conditions as investigated by solid-state NMR spectroscopy. Journalof Physical Chemistry C 2010, 114, 14635–14645.[324] Ghassemzadeh, L.; Holdcroft, S. Quantifying the Structural Changes of Perfluorosulfonated AcidIonomer upon Reaction with Hydroxyl Radicals. Journal of the American Chemical Society 2013,2–5.[325] Paul, D. K.; Karan, K.; Docoslis, A.; Giorgi, J. B.; Pearce, J. Characteristics of Self-AssembledUltrathin Nafion Films. Macromolecules 2013, 46, 3461–3475.[326] Mosier-Boss, P. Review of SERS Substrates for Chemical Sensing. Nanomaterials 2017, 7, 142–172.[327] Campion, A.; Ivanecky, J. E.; Child, C. M.; Foster, M. On the Mechanism of Chemical Enhance-ment in Surface-Enhanced Raman Scattering. Journal of the American Chemical Society 1995,117, 11807–11808.[328] Lauridsen, R. K.; Rindzevicius, T.; Molin, S.; Johansen, H. K.; Berg, R. W.; Alstrøm, T. S.; Alm-dal, K.; Larsen, F.; Schmidt, M. S.; Boisen, A. Towards quantitative SERS detection of hydrogencyanide at ppb level for human breath analysis. Sensing and Bio-Sensing Research 2015, 5, 84–89.[329] USDRIVE Fuel Cell Technical Team Roadmap 2017 .[330] Cherevko, S.; Kulyk, N.; Mayrhofer, K. J. Durability of platinum-based fuel cell electrocatalysts:Dissolution of bulk and nanoscale platinum. Nano Energy 2016, 29, 275–298.[331] Topalov, A. A.; Cherevko, S.; Zeradjanin, A. R.; Meier, J. C.; Katsounaros, I.; Mayrhofer, K. J. J.Towards a comprehensive understanding of platinum dissolution in acidic media.Chem. Sci. 2014,5, 631–638.209References[332] Baroody, H. A.; Jerkiewicz, G.; Eikerling, M. H. Modelling oxide formation and growth on platinum.Journal of Chemical Physics 2017, 146.[333] Climent, V.; Feliu, J. M. Thirty years of platinum single crystal electrochemistry. Journal of SolidState Electrochemistry 2011, 15, 1297–1315.[334] Huang, Y. F.; Kooyman, P. J.; Koper, M. T. Intermediate stages of electrochemical oxidation ofsingle-crystalline platinum revealed by in situ Raman spectroscopy.Nature Communications 2016,7, 1–7.[335] Drnec, J.; Ruge, M.; Reikowski, F.; Rahn, B.; Carlà, F.; Felici, R.; Stettner, J.; Magnussen, O. M.;Harrington, D. A. Initial stages of Pt(111) electrooxidation: dynamic and structural studies bysurface X-ray diffraction. Electrochimica Acta 2017, 224, 220–227.[336] Ruge, M.; Drnec, J.; Rahn, B.; Reikowski, F.; Harrington, D. A.; Carlà, F.; Felici, R.; Stettner, J.;Magnussen, O. M. Structural Reorganization of Pt(111) Electrodes by Electrochemical Oxidationand Reduction. Journal of the American Chemical Society 2017, 139, 4532–4539.[337] Takagi, Y.; Wang, H.; Uemura, Y.; Nakamura, T.; Yu, L.; Sekizawa, O.; Uruga, T.; Tada, M.; Sam-jeské, G.; Iwasawa, Y.; Yokoyama, T. In situ study of oxidation states of platinum nanoparticleson a polymer electrolyte fuel cell electrode by near ambient pressure hard X-ray photoelectronspectroscopy. Phys. Chem. Chem. Phys. 2017, 19, 6013–6021.[338] Saveleva, V. A.; Papaefthimiou, V.; Daletou, M. K.; Doh, W. H.; Ulhaq-Bouillet, C.; Diebold, M.;Zafeiratos, S.; Savinova, E. R. Operando Near Ambient Pressure XPS (NAP-XPS) Study of thePt Electrochemical Oxidation in H2O and H2O/O2 Ambients. The Journal of Physical ChemistryC 2016, 120, 15930–15940.[339] Allen, P.; Conradson, S.; Wilson, M.; Gottesfeld, S.; Raistrick, I.; Valerio, J.; Lovato, M. In SituStructural Characterization of a Platinum Electrocatalyst by Dispersive X-ray Absorption Spec-troscopy. Electrochimica Acta 1994, 39, 2415–2418.[340] Merte, L. R.; Behafarid, F.; Miller, D. J.; Friebel, D.; Cho, S.; Mbuga, F.; Sokaras, D.; Alonso-Mori, R.; Weng, T. C.; Nordlund, D.; Nilsson, A.; Roldan Cuenya, B. Electrochemical oxidationof size-selected pt nanoparticles studied using in situ high-energy-resolution X-ray absorptionspectroscopy. ACS Catalysis 2012, 2, 2371–2376.210References[341] Nagy, Z.; You, H. Applications of surface X-ray scattering to electrochemistry problems. Elec-trochimica Acta 2002, 47, 3037–3055.[342] Sekizawa, O.; Uruga, T.; Higashi, K.; Kaneko, T.; Yoshida, Y.; Sakata, T.; Iwasawa, Y. Simultane-ous Operando Time-Resolved XAFS-XRD Measurements of a Pt/C Cathode Catalyst in PolymerElectrolyte Fuel Cell under Transient Potential Operations. ACS Sustainable Chemistry and Engi-neering 2017, 5, 3631–3636.[343] McNeil, B. W. J.; Thompson, N. R. X-ray free-electron lasers. Nature Photonics 2010, 4, 814–821.[344] Dickinson, T.; Povey, A. F.; Sherwood, P. M. A. X-ray photoelectron spectroscopic studies of oxidefilms on platinum and gold electrodes. Journal of the Chemical Society, Faraday Transactions 1:Physical Chemistry in Condensed Phases 1975, 71, 298–311.[345] Thurston, T. R.; Jisrawi, N. M.; Mukerjee, S.; Yang, X. Q.; Mcbreen, J.; Daroux, M. L.; Xing, X. K.Synchrotron x-ray diffraction studies of the structural properties of electrode materials in operatingbattery cells. Applied Physics Letters 1996, 69, 194–196.[346] Borkiewicz, O. J.; Wiaderek, K. M.; Chupas, P. J.; Chapman, K. W. Best Practices for OperandoBattery Experiments: Influences of X-ray Experiment Design on Observed Electrochemical Re-activity. The Journal of Physical Chemistry Letters 2015, 6, 2081–2085.[347] Scardi, P.; Leoni, M. Advances in Line Profile Analysis for the Study of Nanocrystalline Systems.ECS Transactions 2006, 3, 125–132.[348] Weidenthaler, C. Pitfalls in the characterization of nanoporous and nanosized materials.Nanoscale 2011, 3, 792.[349] Redmond, E. L.; Setzler, B. P.; Juhas, P.; Billinge, S. J. L.; Fuller, T. F. In-Situ Monitoring of ParticleGrowth at PEMFC Cathode under Accelerated Cycling Conditions. Electrochemical and Solid-State Letters 2012, 15, B72.[350] Imai, H.; Izumi, K.; Matsumoto, M.; Kubo, Y.; Kato, K.; Imai, Y. In Situ and Real-Time Monitoringof Oxide Growth in a Few Monolayers at Surfaces of Platinum Nanoparticles in Aqueous Media.Journal of the American Chemical Society 2009, 131, 6293–6300.[351] Sasaki, K.; Marinkovic, N.; Isaacs, H. S.; Adzic, R. R. Synchrotron-Based in Situ Characterizationof Carbon-Supported Platinum and Platinum Monolayer Electrocatalysts. ACS Catalysis 2016, 6,69–76.211References[352] Ishiguro, N.; Saida, T.; Uruga, T.; Sekizawa, O.; Nagasawa, K.; Nitta, K.; Yamamoto, T.;Ohkoshi, S.; Yokoyama, T.; Tada, M. Structural kinetics of a Pt/C cathode catalyst with practi-cal catalyst loading in an MEA for PEFC operating conditions studied by in situ time-resolvedXAFS. Physical Chemistry Chemical Physics 2013, 15, 18827–18834.[353] Ishiguro, N.; Saida, T.; Uruga, T.; Nagamatsu, S.-i.; Sekizawa, O.; Nitta, K.; Yamamoto, T.;Ohkoshi, S.-i.; Iwasawa, Y.; Yokoyama, T.; Tada, M. Operando Time-Resolved X-ray AbsorptionFine Structure Study for Surface Events on a Pt3Co/C Cathode Catalyst in a Polymer ElectrolyteFuel Cell during Voltage-Operating Processes. ACS Catalysis 2012, 2, 1319–1330.[354] Patterson, A. L. The scherrer formula for X-ray particle size determination. Physical Review 1939,56, 978–982.[355] Ida, T.; Ando, M.; Toraya, H. Extended pseudo-Voigt function for approximating the Voigt profile.Journal of Applied Crystallography 2000, 33, 1311–1316.[356] Leppert, L.; Albuquerque, R. Q.; Kümmel, S. Gold-platinum alloys and Vegard’s law on thenanoscale. Physical Review B - Condensed Matter and Materials Physics 2012, 86, 2877–2884.[357] Fitzpatrick, M.; Fry, A.; Holdway, P.; Kandil, F.; Shackleton, J.; Suominen, L. Determination ofResidual Stresses by X-ray Diffraction - Issue 2; 2005; p 74.[358] Watari, M.; McKendry, R. A.; Vögtli, M.; Aeppli, G.; Soh, Y. A.; Shi, X.; Xiong, G.; Huang, X.;Harder, R.; Robinson, I. K. Differential stress induced by thiol adsorption on facetted nanocrystals.Nature Materials 2011, 10, 862–866.[359] McCusker, L. B.; Dreele, R. B. V.; Cox, D. E.; Louer, D.; Scardi, P. Rietveld refnement guidelines.Journal of Applied Crystallography 1999, 32, 36–50.[360] Hong, X.; Ehm, L.; Zhong, Z.; Ghose, S.; Duffy, T. S.; Weidner, D. J. High-energy X-ray focus-ing and applications to pair distribution function investigation of Pt and Au nanoparticles at highpressures. Scientific Reports 2016, 6, 1–8.[361] Billinge, S. J. L. The atomic pair distribution function: past and present. Zeitschrift fur Kristallogra-phie - Crystalline Materials 2004, 219, 117–121.[362] Blanksby, S. J. J.; Ellison, G. B. B. Bond dissociation energies of organic molecules. Accounts ofChemical Research 2003, 36, 255–263.212References[363] Doerr, A. Diffraction before destruction. Nature methods 2011, 8, 283.[364] Fourme, R.; Honkimäki, V.; Girard, E.; Medjoubi, K.; Dhaussy, A. C.; Kahn, R. Reduction of ra-diation damage and other benefits of short wavelengths for macromolecular crystallography datacollection. Journal of Applied Crystallography 2012, 45, 652–661.[365] Ruat, M.; Ponchut, C. Characterization of a pixelated CdTe X-Ray detector using the timepixphoton-counting readout chip. IEEE Transactions on Nuclear Science 2012, 59, 2392–2401.[366] Shastri, S. D.; Fezzaa, K.; Mashayekhi, A.; Lee, W. K.; Fernandez, P. B.; Lee, P. L. Cryogenicallycooled bent double-Laue monochromator for high-energy undulator X-rays (50-200 keV). Journalof Synchrotron Radiation 2002, 9, 317–322.[367] Snigirev, A.; Kohn, V.; Snigireva, I.; Lengeler, B. A compound refractive lens for focusing high-energy X-rays. Nature 1996, 384, 49–51.[368] Vaughan, G. B. M.; Wright, J. P.; Bytchkov, A.; Rossat, M.; Gleyzolle, H.; Snigireva, I.; Snigirev, A.X-ray transfocators: Focusing devices based on compound refractive lenses. Journal of Syn-chrotron Radiation 2011, 18, 125–133.[369] Koprinarov, I. N.; Hitchcock, A. P.; McCrory, C. T.; Childs, R. F. Quantitative mapping of structuredpolymeric systems using singular value decomposition analysis of soft X-ray images. Journal ofPhysical Chemistry B 2002, 106, 5358–5364.[370] Hendrickson, W. A. Determination of Macromolecular Structures from Anomalous Diffraction ofSynchrotron Radiation. Science 1991, 254, 51–58.[371] Hodeau, J.-L.; Favre-Nicolin, V.; Bos, S.; Renevier, H.; Lorenzo, E.; Berar, J.-F. Resonant Diffrac-tion. Chemical Reviews 2001, 101, 1843–1867.[372] Krause, M. O.; Oliver, J. H. Natural widths of atomic K and L levels, Ka X-ray lines and severalKLL Auger lines. Journal of Physical and Chemical Reference Data 1979, 8, 329–338.[373] Bazin, D.; Guczi, L.; Lynch, J. Anomalous wide angle X-ray scattering (AWAXS) and heteroge-neous catalysts. Applied Catalysis A: General 2002, 226, 87–113.[374] Cromer, D. T.; Liberman, D. A. Anomalous dispersion calculations near to and on the long wave-length side of an absorption edge. Acta Crystallographica Section A 1981, 37, 267–268.213References[375] Kaito, T.; Mitsumoto, H.; Sugawara, S.; Shinohara, K.; Uehara, H.; Ariga, H.; Takakusagi, S.;Asakura, K. A new spectroelectrochemical cell for in situ measurement of Pt and Au K-edge X-rayabsorption fine structure. Review of Scientific Instruments 2014, 85.[376] Heal, H. G. The Decomposition of Solid Potassium Perchlorate By 50 Kilovolt X-Rays. CanadianJournal of Chemistry 1952, 21, 91.[377] Nagy, Z.; You, H. Radiolytic effects on the in situ investigation of buried interfaces with synchrotronX-ray techniques. Journal of Electroanalytical Chemistry 1995, 381, 275–279.[378] Bernier, J. V.; Barton, N. R.; Lienert, U.; Miller, M. P. Far-field high-energy diffraction microscopy:A tool for intergranular orientation and strain analysis. Journal of Strain Analysis for EngineeringDesign 2011, 46, 527–547.[379] Drnec, J.; Zhou, T.; Pintea, S.; Onderwaater, W.; Vlieg, E.; Renaud, G.; Felici, R. Integrationtechniques for surface X-ray diffraction data obtained with a two-dimensional detector. Journal ofApplied Crystallography 2014, 47, 365–377.[380] Balzar, D.; Audebrand, N.; Daymond, M. R.; Fitch, A.; Hewat, A.; Langford, J. I.; Le Bail, A.;Louër, D.; Masson, O.; McCowan, C. N.; Popa, N. C.; Stephens, P. W.; Toby, B. H. Size-strainline-broadening analysis of the ceria round-robin sample. Journal of Applied Crystallography 2004,37, 911–924.[381] Doebelin, N.; Kleeberg, R. Profex: A graphical user interface for the Rietveld refinement programBGMN. Journal of Applied Crystallography 2015, 48, 1573–1580.[382] Hansford, G. M. PoDFluX: A new Monte Carlo ray-tracing model for powder diffraction and fluo-rescence. Review of Scientific Instruments 2009, 80.[383] Nervo, L.; King, A.; Wright, J. P.; Ludwig, W.; Reischig, P.; Quinta Da Fonseca, J.; Preuss, M.Comparison between a near-field and a far-field indexing approach for characterization of a poly-crystalline sample volume containing more than 1500 grains. Journal of Applied Crystallography2014, 47, 1402–1416.[384] Banerjee, R.; Liu, Q.; Tengco, J. M. M.; Regalbuto, J. R. Detection of Ambient Oxidation of Ul-trasmall Supported Platinum Nanoparticles with Benchtop Powder X-Ray Diffraction. CatalysisLetters 2017, 147, 1754–1764.214References[385] Toby, B. H. R factors in Rietveld analysis: How good is good enough? Powder Diffraction 2006,21, 67–70.[386] Larson, A.; Von Dreele, R. B. General Structure Analysis System (GSAS); 2004; pp 1–231.[387] Johnson Matthey Fuel Cells, HiSPEC Fuel Cell Catalysts.[388] Stoupin, S.; Chung, E. H.; Chattopadhyay, S.; Segre, C. U.; Smotkin, E. S. Pt and Ru X-rayabsorption spectroscopy of PtRu anode catalysts in operating direct methanol fuel cells. Journalof Physical Chemistry B 2006, 110, 9932–9938.[389] David, W. I. F.; Leoni, M.; Scardi, P. Domain Size Analysis in the Rietveld Method. MaterialsScience Forum 2010, 651, 187–200.[390] Conway, B. E.; Barnett, B.; Angerstein-Kozlowska, H.; Tilak, B. V. A surface-electrochemical basisfor the direct logarithmic growth law for initial stages of extension of anodic oxide films formed atnoble metals. The Journal of Chemical Physics 1990, 93, 8361–8373.[391] Riese, A.; Banham, D.; Ye, S.; Sun, X. Accelerated Stress Testing by Rotating Disk Electrode forCarbon Corrosion in Fuel Cell Catalyst Supports. Journal of the Electrochemical Society 2015,162, F783–F788.[392] Dubau, L.; Maillard, F. Unveiling the crucial role of temperature on the stability of oxygen reductionreaction electrocatalysts. Electrochemistry Communications 2016, 63, 65–69.[393] Gilroy, D.; Conway, B. E. Surface oxidation and reduction of platinum electrodes: Coverage, kineticand hysteresis studies. Canadian Journal of Chemistry 1968, 46, 875–890.[394] Holby, E. F.; Greeley, J.; Morgan, D. Thermodynamics and hysteresis of oxide formation andremoval on platinum (111) surfaces. Journal of Physical Chemistry C 2012, 116, 9942–9946.[395] Kondo, T.; Masuda, T.; Aoki, N.; Uosaki, K. Potential-Dependent Structures and Potential-InducedStructure Changes at Pt(111) Single-Crystal Electrode/Sulfuric and Perchloric Acid Interfaces inthe Potential Region between Hydrogen Underpotential Deposition and Surface Oxide Formationby In Situ Su. The Journal of Physical Chemistry C 2016, 120, 16118–16131.[396] Drnec, J.; Harrington, D. A.; Magnussen, O. M. Electrooxidation of Pt(111) in acid solution.CurrentOpinion in Electrochemistry 2017, 4, 69–75.215References[397] Daio, T.; Staykov, A.; Guo, L.; Liu, J.; Tanaka, M.; Matthew Lyth, S.; Sasaki, K. Lattice StrainMapping of Platinum Nanoparticles on Carbon and SnO2 Supports. Scientific Reports 2015, 5,13126.[398] Bucher, J. P.; Buttet, J.; Van Der Klink, J. J.; Graetzel, M. Electronic properties and local densitiesof states in clean and hydrogen covered Pt particles. Surface Science 1989, 214, 347–357.[399] Kieffer, J.; Karkoulis, D. PyFAI, a versatile library for azimuthal regrouping. Journal of Physics:Conference Series 2013, 425.[400] Osteryoung, J.; Wechter, C. ACS Symposium Series; 1989; Vol. 390; pp Chapter 25, 380–395.[401] Robinson, R. S.; McCurdy, C. W.; McCreery, R. L. Microsecond Spectroelectrochemistry by Ex-ternal Reflection from Cylindrical Microelectrodes. Analytical Chemistry 1982, 54, 2356–2361.[402] Grillo, F.; Van Bui, H.; Moulijn, J. A.; Kreutzer, M. T.; Van Ommen, J. R. Understanding andControlling the Aggregative Growth of Platinum Nanoparticles in Atomic Layer Deposition: AnAvenue to Size Selection. Journal of Physical Chemistry Letters 2017, 8, 975–983.[403] Comsa, G.; Heinig, K. H.; Michely, T.; Kalff, M.; Strobel, M. Step edge diffusion and step atomdetachment in surface evolution: Ion erosion of Pt(111). Physical Review Letters 2001, 86, 2589–2592.[404] Pedersen, T. M.; Xue Li, W.; Hammer, B. Structure and activity of oxidized Pt(110) and α-PtO2.Physical Chemistry Chemical Physics 2006, 8, 1566–1574.[405] Leontowich, A. F. G.; Berg, R.; Regier, C. N.; Taylor, D. M.; Wang, J.; Beauregard, D.; Geilhufe, J.;Swirsky, J.; Wu, J.; Karunakaran, C.; Hitchcock, A. P.; Urquhart, S. G. Cryo scanning transmissionX-ray microscope optimized for spectrotomography. ArXiv 1808.05702 2018,216