UC Berkeley

It is increasingly and urgently obvious that the world needs to move toward green, renewable energy sources. Key in realizing this infrastructure are energy-conversion devices such as hydrogen fuel cells and water electrolyzers. As their name suggests, these devices allow energy to be converted from one form (a fuel) to another (electricity), and vice versa, enabling a distributed and modular energy network. The efficiency and cost of these devices have vastly improved over the last few decades, but those gains (in particular for fuel cells) have stagnated in recent years due to performance limitations and high costs associated with their catalyst layers (CLs). This dissertation focuses on fuel-cell CLs, although the fundamental principles and insights gained are applicable to electrolyzers and other energy-conversion devices that rely on similar CL paradigms and architectures.

CLs are heterogeneous porous electrodes comprised of agglomerates of catalyst particles (typically platinum supported on carbon), ion-conducting polymer (ionomer), and void space. The microstructure of the CL, including the size of the agglomerates, the ionomer coverage, porosity/tortuosity, etc. controls the complex gas, liquid, ion, and electron transport networks and impacts the overall kinetic, ohmic, and mass-transport performance of these devices. Characterization efforts have spanned micro-, meso-, and macro-scale techniques to probe structure-property-performance relationships of the CL.

CLs are made from precursor CL inks, which are colloidal dispersions of the ionomer and catalyst particles, dispersed in solvent. CL studies are complicated by the multiple material types used (varying ionomer chemistry, catalyst loading, carbon support type, solvents) and disparate ink deposition and drying methods that render no two CLs alike; this makes it difficult to compare across these different studies. Additionally, within the community, CL fabrication has traditionally been treated as a black art, and the details of the fabrication process (ink composition, casting method, etc.) are typically inconsistently reported, because emphasis has been placed on understanding CL properties and performance and not the forces controlling that formation process. However, it is increasingly clear that simply being able to characterize CLs is not enough. To enable predictive control and rational design of CLs, it is vital to understand how and why (in addition to what) specific microstructures form.

This dissertation sets out to uncover systematically the underlying fundamental interactions between the ink components in solution, which ultimately govern CL microstructure (agglomerate structures and sizes, ionomer coverages, etc.) and performance once cast. We begin by introducing all relevant parameters in the CL ink fabrication process and conducting a literature-based parameter screening to test correlation between ink variables and performance metrics. The analysis reveals that while no single parameter is controlling, solvent identity and ionomer-to-catalyst-particle ratio correlate well with performance metrics. This suggests ionomer/solvent, ionomer/particle, and ionomer/solvent/particle interactions merit further investigation.

To probe the specifics of these interactions, we build our understanding piecewise. The first part of this dissertation is dedicated to the investigation of ionomer/solvent interactions. We use a range of water/propanol ratios to probe the influence of solvent on these interactions. Water/propanol ratios are chosen because they are (1) the most commonly used solvents, (2) represent a range of water contents (relevant for studying hydrophobic/hydrophilic interactions), and (3) encompass a wide range of dielectric permitivities. We first explore the solution structure of ionomer dispersed in various water/propanol ratios, and how these different conformations affect the in situ evolution of thin-film morphology via x-ray scattering. Higher-water-content dispersions exhibit enhanced ordering of primary ionomer aggregates, and fewer secondary aggregates, which affect the final structure of the films; this results in thin films and membranes with improved transport.

Having established a strong link between ionomer/solvent dispersion interactions and the cast state, this dissertation focuses on additional details of the ionomer/solvent interaction and how it may influence catalyst-particle aggregation behavior. We uncover that solvent has a strong effect on the acidity of these dispersions, and that this acidity change is likely due to solvent-induced conformational differences. Additionally, this ionomer/solvent interaction is sensitive to ionomer concentration, wherein acidity does not scale linearly with ionomer concentration, again suggesting it is in part attributed to conformational differences (i.e. aggregation) across the different compositions. The increased acidity with increasing water content propagates to alter particle interactions: acidity strongly influences the electrostatic interactions between the ionomer and catalyst particles in an ink. These pH measurements are then used to probe a third dimension to this solvent/ionomer parameter space by considering the temporal stability of these dispersions. The ionomer slowly equilibrates to new solvent environments over the course of many days: lower ionomer concentrations and water/propanol composition extremes affect this change more drastically. Once again, these conformations persist to membranes upon casting and annealing.

To study ionomer/particle interactions further, isothermal titration calorimetry measurements are used to extract quantitative binding affinities of ionomer to both platinum and carbon nanoparticle surfaces as a function of ionomer charge density. There exists a nonmonotonic relationship between the number of charged groups in the ionomer and its binding affinity. The binding affinity is modeled to extract the entropic and enthalpic contributions to the binding free energy, critically revealing that binding to both platinum and carbon surfaces is governed by a similar entropy-dominated mechanism. This finding is counter to the prevailing hypothesis in literature that ionomer/particle ink association is due to inherent, strong specific-ion (enthalpic) interactions between the ionomer and platinum, and it reveals that ink interaction mechanisms operate in different modes than those occurring operando in the fuel cell. Therefore, extrapolation from well-studied fuel cell systems to ink systems should be avoided because of differences in charge states of the platinum surface; this leads to a new reinterpretation of existing data.

We compliment the calorimetry results with a wide parameter screening using quartz crystal microbalance: we probe ionomer adsorption from solution onto crystal surfaces with a range of different surface functionalities (varying hydrophobicity, metal type, and surface chemistry). Importantly, the findings from the above ionomer/solvent interaction studies inform this ionomer/particle interaction investigation by considering adsorption to these different surfaces both as a function of ionomer charge density as well as water/propanol ratio. In inks (under no applied potential) data reveal that higher-water-content dispersions promote adsorption to all surface types, and adsorption to hydrophobic surfaces (like carbon) is higher than adsorption to platinum. In sum, these ionomer/particle interaction investigations suggest that the ionomer/catalyst particle interactions in inks are dominated by ionomer/carbon interactions, rather than ionomer/platinum interactions.

Finally, the results of the preceding sections are used to inform a case study examining the influence of platinum versus carbon nanoparticle surface area on ink interactions, CL microstructure, and fuel-cell performance. Catalysts with varying platinum loadings are used to explore the zeta potential of inks with varying ionomer-to-particle ratios. A direct correlation between ink zeta potential and CL local-transport resistance is established. Furthermore, we demonstrate that particles with higher carbon surface areas can support more ionomer, likely leading to more uniform coverage and lower CL transport resistance.

The fundamental ink interactions uncovered here provide insights into CL performance and lay the groundwork for design rules to select for desired CL microstructures by manipulating ink parameters not only for fuel cells, but similar energy-conversion technologies from CO2 electrolysis to chemical synthesis to water treatment.