One Million Atoms Large-Scale Reactive Molecular Dynamics Simulations for Design of Cathode Catalyst Layer in Polymer Electrolyte Fuel Cell Toward Boosting Its Performance

Abstract

Listen

Translate article

Boosting polymer electrolyte fuel cell (PEFC) performance is required in transportation. PEFC performance depends on the electrode reaction activity related to proton conductivity and oxygen diffusivity in catalyst layer (CL), consisting of Pt nanoparticles (Pt NPs), ionomer, water, and carbon supports. To achieve high proton conductivity and oxygen diffusivity, the optimization of CL structures such as Pt composition, ionomer/water distribution, and carbon support structures is essential by computational approaches, such as first-principles molecular dynamics (MD) and classical MD methods. The first-principles MD can describe chemical reactions, however can only simulate a small part of CL structures, consisting of a few hundred atoms. In contrast, classical MD can perform over 1 million atoms simulation to calculate whole CL structures, while the simple inter-atomic potential which enables 1 million atoms calculation, can not describe chemical reactions. Then, developing our original MD simulator "Laich", implementing MPI and OpenMP with ReaxFF inter-atomic potential, enabled us to calculate million atoms system reproducing whole CL structures and to simulate chemical reactions in whole CL structures. In this work, to design the higher-performance CL, we performed reactive MD simulations using a 1 million atoms CL model. A CL structure model was constructed with ionomers, water, Pt NPs, and carbon support. The carbon support composed of six meso pores with a size of 6 nm was constructed. To improve the hydrophilicity, hydroxyl groups and hydrogen terminated 32% and 31% of its surface carbon, respectively. Pt NPs were put on the carbon support at the exterior and in the interior of the meso pores. The Pt-supported carbon was coated with ionomers and water. Protons and oxygen were introduced into the system. Hereafter, we refer to this structure as a catalyst particle (CP) model (Fig. 1). To assess the influence of the carbon support structures on the electrode reactions, the oxygen diffusivity was evaluated from the trajectories of oxygen at the exterior and in the interior of the meso pores. Ideally, oxygen should be supplied to the Pt NPs without hindering its diffusion by obstacles. At the exterior of the meso pore, oxygen could not approach the Pt NPs, because ionomers and water were obstacles to the oxygen diffusion (Figs. 2(a)). Conversely, in the interior of the meso pores, oxygen could approach the Pt NPs via the gas phase since ionomers and water did not inhibit oxygen diffusion (Figs. 2(b)). From this analysis, the oxygen diffusivity showed higher value in the interior of the meso pores than at the exterior of the meso pores. Furtherly, it is reported that oxygen diffusivity in ionomers varies with temperature [1]. However, the influence of temperatures on oxygen diffusivity at the exterior and in the interior of the meso pores has not been analyzed. In this study, we calculated the mean squared displacement (MSD) and diffusion coefficients at 300 K and 365 K. At 300 K, oxygen diffusion coefficients (the linear regressions of MSD) at the exterior (Dexterior ) and in the interior of the meso pores (Dinterior ) were 1.67×10-10 m2/s and 3.02×10-6 m2/s, respectively (Fig. 3(a)). At 365 K, Dexterior and Dinterior were 12.3×10-10 m2/s and 2.59×10-6 m2/s, respectively (Fig. 3(b)). Therefore, Dexterior and Dinterior respectively increased 7.37 and 0.86 times with increasing temperature. Next, to investigate whether the change of Dexterior was caused only by temperatures, we calculated the oxygen diffusion coefficient at 1 atm (D1atm ) in an environment consisting of oxygen atoms only. As a result, D1atm increased 1.05 times from 300 K to 365 K. By rising temperature, the increase of the oxygen diffusion coefficient showed 1.05 time at D1atm , while it showed 7.37 time at D exterior. Therefore, other factors besides temperature affect the oxygen diffusivity. To identify the factors contributing to D exterior , we focused on the size of the water cluster, one of the obstacles to oxygen diffusion. Water clusters are defined as groups of water with an inter-molecular distance of less than 3.5 Å. The cluster size is defined as the number of water molecules forming the water cluster. Fig. 4 depicts the sizes of the 10 largest water clusters. As the temperature rose, the size of the largest water cluster decreased, while the size of other water clusters increased. These analyses indicate that the morphology of water clusters changed from continuous to discontinuous clusters with increasing temperature. Therefore, as water cluster sizes decrease by increasing temperature, oxygen diffuses through the gas phase. These results indicate that the control of the water cluster size improves the oxygen diffusivity.[1] K. Kudo et. al., Electrochim. Acta, 209, 682-690 (2016). Figure 1