The materials and the structure of solid oxide fuel cells are designed to avoid thermo-mechanical damages under various operation conditions. However, inherent risk of chemo-mechanical failures are still not fully understood. This paper aims to review the recent works related to this topic, and to address some issues which have not been widely recognized.
The coupling of chemistry and mechanics are classified into four types, i.e. (1) chemically driven strain, (2) chemically modified mechanical properties, (3) mechanically driven chemical reactions, and (4) mechanically modified chemical (physical) properties. Since chemical energies are much larger than mechanical energy accommodated in SOFC, the former two types (type(1) and (2)) of chemo-mechanical coupling have been recognized as more important than the others, and have been studied intensively. An example of type-1 phenomena is chemical expansion of mixed conducting oxides with e.g. (La,Sr)(Co,Fe)O3 cathode, LaCrO3 based interconnect, and CeO2 based or (La,Sr)(Ga,Mg,Co)O3 electrolytes. Since the transient behavior as well as steady state distribution of oxygen potential inside the constituent solids is essential to know the effect of the chemical strain, Terada et al. developed a computer code “SIMUDEL” of an FEM-based calculation of oxygen potential. This code considers “chemical capacitance” due to nonstoichiometry of the materials to treat the transient responses, and the results of the calculation can be transported into some of major commercial programs for structure analysis. Volume change of a nickel cermet anode is also an important feature of type-1 coupling which must be considered in determining fabrication and operation processes. The electrode shrinks on reduction and expands on re-oxidation as expected from the lattice size of the metal and the oxide. However, under certain conditions, a porous cermet was found to “shrink” upon oxidation. It took place only during light re-oxidation around 400C. Under this condition the formation of NiO was not obvious from XRD, whereas weight gain was observed by thermo-gravimetry. Careful observation of the microstructure of a porous Ni revealed that, upon shrinkage, the particle-to-particle separation changed partly due to the neck growth between the particles and to the change of the connection angle of the particles. Further study is underway to elucidate the detailed mechanism of the oxidation-induced shrinkage.
The change of mechanical properties such as elastic moduli and fracture strength are also dependent on defect concentration and its motion in the lattice (type-2 coupling). Young’s modulus of nonstoichiometric oxides show dependences not only on temperature but also on pO2 through the change of defect concentration. Also, domain boundary shift of ferroelastic phase of LSCF was found to be correlated with the defect concentration. As is discussed for the anomaly of Young’s modulus of YSZ around 400˚C, the motion of oxide ion vacancies may also have correlation with the ferroelastic strain observed with Sc and Ce doped ZrO2 electrolyte above 300˚C. Another interesting type-2 coupling is with the lightly oxidized Ni cermet electrode. It was found that the creep rate of Ni-YSZ cermet at 400˚C was dramatically increased when oxygen-containing gas was introduced. This may be by a correlated mechanism with the above mentioned oxidation induced shrinkage.
Several reports, including those from our group, have been published on the effect of mechanical stress on defect formation (type-3 coupling) of nonstoichiometric oxides determined by experiments or by calculation. As is expected from thermodynamic consideration, the experimentally determined effect was not large, e.g. 1G Pa stress was equivalent to 1/5 order of magnitude shift of chemical potential of oxygen for nonstoichiometry of LSCF. Similarly, only minor effect on a practical system was reported for type-4 coupling. However, those phenomena can have significant effect on long-term stability if cation mobility and their driving force are modified at a strained interfaces or grain boundaries.
Generalized Ellingham diagrams for utilization in solid oxide fuel cells
