Microstructure-based solid oxide fuel cell seal material design using statistical continuum mechanics

In this dissertation, the glass-ceramic solid oxide fuel cell seal material G18 is examined using several modeling techniques. The seal of planar solid oxide fuel cells is one of the main challenges in commercializing solid oxide fuel cells. The seal material separates the air and fuel sides of the electrode, and also bonds several components of the fuel cell. The seal keeps fuel from leaking and prevents mixing of the fuel and oxidants. These fuel cells operate at high temperatures, generally between 600-1000°C, and are used for large stationary application and auxiliary power for mobile applications. Glass-ceramic seal G18, developed by the Pacific Northwest National Laboratory, is the focus of this work. The goal of this thesis is to provide tools to enable the design of a suitable glass-ceramic seal by modeling.;The microstructure of G18 is examined by SEM analysis. SEM micrographs of G18 aged for 4 h and 1000 h. The 4 h-aged G18 showed ∼54% crystallization of the microstructure. In the micrograph, the barium silicate phase is shown as needles dispersed uniformly throughout the microstructure, with no preference to direction. Smaller amounts of hexacelsian needles also appear, which are smaller and shorter than the barium silicate needles. In the 1000 h-aged sample, the phases in the microstructure have developed blurred phase boundaries. The hexacelsian has disappeared, and only the barium silicate phase, glass matrix phase, and pores are clearly observed. Micro-voids have also developed, which is most likely due to shrinkage of the crystalline phase from cooling. Residual stress may also be a factor in weakening of the fuel cell seal material.;The elastic properties and creep deformation have been measured by nanoindentation at room and high temperatures. Results show that the elastic modulus and hardness of the samples aged for short times (4 or 5 h) decrease with increasing temperature, showing significant decrease above the glass transition temperature, 619°C. The 100 h-aged sample has an elastic modulus and hardness that is below the short aged samples at room temperature. This is believed to be from damage induced from cracking during cooling. Above 400°C, the elastic modulus and hardness of the 100 h-aged sample increases, and is above the short aged samples values. This phenomenon in the glass seal has been studied, and it is concluded that the increase in elastic modulus and hardness is due self-healing of micro-voids and cracks in the seal material. The crystalline phases are temperature independent, therefore increasing the 100 h-aged sample hardness and elastic modulus at high temperatures, due to increased crystallization in the 100 h-aged sample.;The effective elastic stiffness and coefficient of thermal expansion are predicted using homogenization relations. These models are based on volume fraction and individual phase properties, while making assumptions about microstructure morphology. The effective elastic stiffness is predicted using several models, which assume different geometries of the microstructure. The results for the Mori-Tanaka model and self-consistent composite inclusion model fall within the Voigt and Reuss upper and lower bounds. The effective elastic stiffness models underestimate the measured value of G18. In comparing the coefficient of thermal expansion models, the Levin model fell within the upper and lower bounds, but the predicted results can not accommodate the lowering in thermal expansion due to the introduction of monoclinic celsian in the 1000 h-aged sample.;A statistical continuum mechanics approach was used to predict the effective elastic stiffness, effective inelastic behavior, and also thermal expansion coefficient. This type of approach allows the inclusion of several phases once the models are established. Also, the models can predict microstructure properties using the actual morphology, represented by two-point correlation functions. The effective elastic stiffness scheme incorporates the use of two-point correlation functions to characterize the microstructure. The model currently assumes no damage or self-healing occurs. The effective elastic stiffness model considers the morphology of the microstructure, as well as the individual phase properties. Borosilicate glass values were used for the glass phase due to unavailability of glass phase properties. The effective elastic stiffness results showed results comparable to measured values, and fall between the upper and lower bounds. (Abstract shortened by UMI.)...