Quantitative Evaluation of Chemical Degradation of Polymer Electrolyte Membrane for Fuel Cells and the Method of Lifetime Prediction

Abstract

A fuel cell vehicle (FCV) is a vehicle that utilizes fuel cells as its power source. Development of fuel cells for FCV is always aimed at a more compactsize by means of higher power density and longer driving range through enhanced energy efficiency. Fuel cells also require endurance reliability comparable to internal combustion engines, and cost reductions will be required to encourage widespread consumer use. The power generating part of a fuel cell is called the membrane electrode assembly (MEA), which consists of a polymer electrolyte membrane (PEM) and electrodes with a catalyst. The performance of the fuel cell is mainly determined by the activity of the catalyst on each electrode and by the proton conductivity of the PEM. Because the performance, energy efficiency, durability, and cost of fuel cells are all related and involve trade-offs, optimization of the PEM design cannot be readily achieved. A particular issue that has faced optimization of the PEM design is the amount of time it takes to estimate PEM durability. The only method available has been to test it over a long period of time, and this is why FCV has had such an extended development period. In order to optimize the PEM design in less time, a new method capable of estimating durability without durability testing must be developed. To this aim, we first made it possible to quantitatively evaluate the rate of chemical degradation of PEM to estimate durability and predict its lifetime. Next, we developed a mathematical formula for the rate of chemical degradation of the PEM to predict its service life. Two possible reaction mechanisms have been proposed as the chemical degradation of the perfluorosulfonic acid (PFSA) membrane. One is the scission of the main chains of the polymer and the other is an unzipping reaction in which the end groups of the main chains progressively degrade. In order to quantitatively estimate the rate of chemical degradation of the PFSA membrane, we numerically modelized these two reaction mechanisms. When a scission of the main chain occurs, the location of the split becomes two new end groups, which suggests that the number of points of origin for an unzipping reaction increases exponentially. The fluoride release behavior was calculated based on this model. As a result, we confirmed that the experimental fluoride release data can be explained by these two reaction mechanisms of chemical degradation. In order to describe the fluoride release quantity F as a function of duration time t, exponential function Eq. (1) was formulated to closely match the experimental results. F(t) = a [exp(b･t) - 1] (1) One of the factors that influence the chemical degradation rate of the PEM is contamination by metal ions. Adding a radical quencher is a well-known method to assure durability of the PEM by mitigating chemical degradation. In order to estimate the acceleration of PEM degradation caused by metal ion contamination or the mitigation effects offered by a radical quencher, durability testing was carried out using different levels of metal ion contamination or radical quencher additives. The results of these tests show that fluoride release behavior can be represented by Eq. (1) by fixing coefficient b and varying only coefficient a. Coefficient ais therefore defined as the accelerating factor of chemical degradation. The accelerating factor can be obtained by making Eq. (1) fit the experimental fluoride release data, and it can be used as an indicator of the severity of chemical degradation. Introducing the concept of accelerating factor to represent the fluoride release rate made it possible to deal quantitatively with the effect of factors that accelerate or mitigate chemical degradation. Moreover, the size of acceleration or mitigation effect caused by multiple factors can be indicated with a single variable. We confirmed changes in accelerating factor during FCV operation with bench testing, which demonstrated the driving modes of an actual vehicle using a fuel cell stack with the same specifications as stacks installed in actual vehicles. The fluoride release rate was calculated from the accelerating factor envisioning actual FCV operation. Fluoride release behavior can be converted into membrane thickness loss. In this way, decrease in membrane thickness during FCV operation and the lifetime of the PEM can be predicted using calculations.