Skip to main content
eScholarship
Open Access Publications from the University of California

Ice-Crystallization Kinetics during Fuel-Cell Cold-Start

  • Author(s): Dursch, Thomas James
  • Advisor(s): Radke, Clayton J
  • et al.
Abstract

Proton-exchange-membrane fuel cells (PEMFCs) show promise in automotive applications because of their high efficiency, high power density, and potentially low emissions. To be successful in automobiles, PEMFCs must permit rapid startup with minimal energy from subfreezing temperatures, known as cold-start. In a PEMFC, reduction of oxygen to water occurs in the cathode catalyst layer (CL). Under subfreezing conditions, water generated during startup solidifies and hinders access of gaseous oxygen to the catalytic sites in the cathode CL, severely inhibiting cell performance and potentially causing cell failure. Achieving cold-start is difficult in practice, due to potential flooding, sluggish reaction kinetics, durability loss, and rapid ice crystallization. Currently, however, few studies focus on the fundamentals of ice crystallization during cold-start. Elucidation of the mechanisms and kinetics of ice formation within PEMFC porous media is, therefore, critical to successful cell startup and high performance at low temperatures.

First, an experimental method is presented for obtaining isothermal ice-crystallization kinetics in water-saturated gas-diffusion layers (GDLs). Ice formation is initially studied in the GDL because this layer retains a significant amount of product water during cold-start. Isothermal ice-crystallization and ice-nucleation rates are obtained in commercial Toray GDLs as functions of subcooling using differential scanning calorimetry (DSC). A nonlinear ice-crystallization rate expression is developed using Johnson-Mehl-Avrami-Kolmogorov (JMAK) theory, in which the heat-transfer-limited growth rate is determined from the moving-boundary Stefan problem. Predicted ice-crystallization rates are in excellent agreement with experiment. A validated rate expression is thus available for predicting ice-crystallization kinetics in GDLs.

Ice-crystallization kinetics is also considered under experimental settings similar to real PEMFC operating conditions where ice invariably forms non-isothermally. Non-isothermal ice-crystallization rates and ice-crystallization temperatures are obtained in water-saturated GDLs as a function of cooling rate. Our previously developed ice-crystallization rate expression is extended to non-isothermal crystallization to predict ice-crystallization kinetics at various cooling rates. For non-isothermal ice formation, we find that cooling rate has a negligible effect on the crystallization rate when crystallization times are much faster than the time to decrease the sample temperature by the subcooling. Therefore, a pseudo-isothermal method is proposed for non-isothermal crystallization kinetics using isothermal crystallization kinetics evaluated at the non-isothermal crystallization temperature.

Catalyst layers also retain a significant amount of product water during cold-start. Accordingly, ice nucleation and growth in PEMFC CLs are investigated using isothermal DSC and compared to isothermal galvanostatic membrane-electrode assembly (MEA) cold-starts. Measured ice-crystallization and ice-nucleation rates follow expected trends from classical nucleation theory. Following our previous approach, a quantitative nonlinear ice-crystallization rate expression is developed from the JMAK framework. To validate ice-crystallization kinetics within PEMFCs, we further measure and predict MEA cell-failure time during isothermal galvanostatic cold-start. Using a simplified PEMFC isothermal cold-start continuum model, MEA cell-failure times predicted using the newly obtained rate expression are compared to that predicted using a traditional thermodynamics-based approach. From this comparison, conditions are identified under which including ice-crystallization kinetics is critical and to elucidate the impact of freezing kinetics on low-temperature PEMFC operation.

During cold-start, the time for recovering cell performance strongly depends on the rate of melting residual ice by reactive heat generation. Non-isothermal ice melting in water-saturated GDLs is investigated using DSC with various heating rates. In all cases, ice-melting times decrease nonlinearly with increasing heating rate, whereas melting temperatures remain near the equilibrium melting temperature of bulk ice, demonstrating that melting is thermodynamic-based with a rate limited by heat transfer. Ice-melting endotherms are predicted from overall DSC energy balances coupled with a moving-boundary Stefan problem, where an ice-melting front within a GDL propagates with volume-averaged properties through an effective medium. Agreement between theory and experiment is excellent. Furthermore, an analytical expression is obtained for ice-melting time. Significantly, the new expression elucidates parameters controlling ice melting and allows for better design of both GDL materials and heating strategies to enhance the success of PEMFC cold-start.

Main Content
Current View