A three-dimensional pore-scale model of the cathode electrode in polymer-electrolyte membrane fuel cell by lattice Boltzmann method

Highlights

• A 3-D lattice Boltzmann model of cathode GDL in PEM fuel cell is presented.

• Non-homogeneity and anisotropy of GDL, and electrochemical reaction are included.

• General orientation of GDL carbon fibers, shapes species and current distributions.

• Normally oriented carbon fibers cause larger current variations on catalyst layer.

• Microstructure must be regarded as an important characteristic of any GDL material.

Abstract

High power density, low operation temperature, high efficiency and low emissions have granted proton exchange membrane (PEM) fuel cells the most promising future among all types of fuel cells. The porous electrodes of PEM fuel cells have a complicated, non-homogeneous, anisotropic microstructure. Therefore, pore-scale modeling techniques such as lattice Boltzmann method, which can capture non-homogeneous and anisotropic microstructures, have recently gained a great attention. In the present study, a three-dimensional lattice Boltzmann model of a PEM fuel cell cathode electrode is proposed in which electrochemical reaction on the catalyst layer and microstructure of GDL are taken into account. The model enables us to simulate single-phase, multi-species reactive flow in a heterogeneous, anisotropic gas diffusion layer through an active approach. To show the capability of the proposed model, reactive flow in three reconstructed GDLs with different anisotropic characteristics is simulated to investigate the effects of GDL microstructure on species and current density distributions. The results demonstrate that when carbon fibers are more likely oriented normal to the catalyst layer, species density distribution is thicker and more disturbed. Current density also experiences a larger variation on the catalyst layer in such a case.

Introduction

A sustainable, high quality life is the basic driver for providing a clean, safe, reliable and secure energy supply in the world. Fuel cells and hydrogen systems are important technologies which may contribute positively to the future world energy demand. Among different types of fuel cells, proton exchange membrane (PEM) fuel cells currently hold the most promising prospect. They can be employed in a wide range of applications, ranging from very small portable devices such as mobile phones and laptops, through transport applications like cars, delivery vehicles, buses and ships, to combined heat and power generators in stationary applications in the domestic and industrial sectors [1].

Nevertheless, some barriers such as high cost and operational difficulties hinder PEM fuel cells' wide commercialization, which make fundamental research for their development inevitable. Modeling of reactant gas transport accompanied by electrochemical reactions in the electrodes is of great importance, especially in the cathode where the oxygen reduction reaction is sluggish and inefficient [2]. Numerous numerical models of the cathode electrode with different features can be found in the literature [3]. In most of these models, the gas diffusion layer (GDL) of cathode electrode is considered as a homogeneous and isotropic porous medium, while in reality that is not the case [4]. In this regard, pore-scale modeling techniques which can realistically capture the microstructure of the cathode GDL are extremely attractive.

Some researchers have used more conventional methods, such as the finite volume, to model electrodes of a PEMFC. Harvey et al. [5], for example, used such a method to investigate different approaches to modeling the catalyst layer of PEM fuel cells. They compared three catalyst treatments of thin-film model, discrete-catalyst volume model and agglomerate model, and concluded that only the agglomerate model could capture mass transport limitations that occur at high current densities. Sahraoui et al. [6] presented a two-dimensional, finite-volume model of a PEM fuel cell taking into consideration the finite thicknesses of the catalyst layer and membrane. Sun et al. [7] also introduced a two-dimensional model for PEM fuel cell cathode that treated the catalyst layer as agglomerates of polymer electrolyte coated catalyst particles. They showed that the cathode over-potential inside the catalyst layer was non-uniform influenced by the channel-land geometry.

In the past decade, lattice Boltzmann method has emerged as a powerful pore-scale modeling technique applicable in complicated, heterogeneous, anisotropic porous media such as GDLs of PEM fuel cells. Lattice Boltzmann method is superior to other conventional numerical modeling techniques through many aspects such as capability of dealing with complex boundaries of a complicated microstructure, easy parallelizable algorithm development, and facility in modeling multi-phase fluid flow in a porous medium [8].

However, a significant challenge for lattice Boltzmann modeling of the cathode electrode is a proper way to consider electrochemical reaction in the catalyst layer [9]. To the best of the authors' knowledge the electrochemical reaction is taken into account in only a few two-dimensional lattice Boltzmann investigations, such as those presented by Chen et al. [10], [11], [12]; moreover, no such three-dimensional modeling, taking into account the electrochemical reaction, was found in the literature. In the present study, a single-phase, three-dimensional model of the cathode electrode of a PEM fuel cell is presented in which the catalyst layer is considered as a thin interface where the electrochemical reaction occurs. The proposed model is implemented to investigate the effects of GDL microstructure on the distributions of species and current density. An advantage of the proposed model over previous lattice Boltzmann models [10], [11], [12] is the use of an active approach for species transport. In the preceding investigations [10], [11], [12], a passive approach was applied in which the velocity field is only solved for nitrogen that is taken as the solvent. This may lead to lack of accuracy especially when the density of other species is comparable to that of nitrogen which may happen at high over-potentials; in such cases no species can be considered as a solvent [13]. On the other hand, in an active approach the internally coupled velocity fields of all species are solved individually which can lead to more accurate results.