| Protective coatings effectively make the underlying component more tolerant of harsh conditions without compromising structural integrity. Thermal barrier coatings are deposited on superalloy turbine blades to protect them from engine temperatures in excess of 1000°C. Failure of these coatings is known to involve undulations that develop in the oxide layer, between the ceramic top-coat and the metallic bond coat. The equi-biaxial stress in the readily creeping bond coat, that results from thermal mismatch with the superalloy substrate and a reversible phase transformation, dramatically reduces its ability to resist transverse deformation at elevated temperatures. The nonlinear interaction between the equi-biaxial stresses in the bond coat, and, to a lesser extent, the ceramic top-coat, with the tractions applied on both creeping layers by the compressed oxide film allows an increment of growth to occur each cycle before the equi-biaxial stress decays by creep. The thermal barrier layer, although creeping, constrains growth significantly, resulting in large tensile stresses above the undulations and normal to the interface with the oxide layer.; A model providing an analytical approximation of undulation growth in a four-layer thermal barrier system, incorporating nearly all details, is presented for a realistic cyclic thermal history. A set of model problems is used to elucidate the mechanics of undulation growth, and in some cases, replicate trends observed in experiments. Common observations for PtNiAl bond-coats without the thermal barrier layer present are explained, including that oxide undulation growth does not occur on unconstrained bulk PtNiAl, that far more undulation growth occurs when the system is thermally cycled than when it is held at the peak temperature for the same total time, and that larger amplitudes are observed for systems that are cycled with a .1 hour isothermal period than for a 1 hour period for the same total time at high temperature. Complete simulations are performed, and selected results are compared with a finite element program written expressly for this purpose. Large undulation amplitudes are possible, and the tensile stresses predicted by this model are sufficient to initiate cracking in the top-coat parallel to the interface with the oxide layer. |
