Thermal barrier coatings (TBC's) are ceramic coatings applied to internally cooled gas turbine components to reduce the metal temperature by up to 150 C in order to realize important improvements in both part life and system performance. Such coatings almost universally are made of yttria stabilized zirconia. Such coatings are used on nearly all current generation aviation gas turbines and their use is rapidly increasing in industrial turbines.
Unfortunately TBC's eventually fail by spallation. There are a wide variety of failure mechanisms and each different TBC system and load history is it's own specific case. In the present paper we will briefly describe the failure mode of a specific system ( PtAl bondcoat EB-PVD coating, Single Crystal superalloy substrate). In this particular system there is a 10X scatter in the failure life for nominally identical samples. In the present paper we will show that of the several failure mechanisms considered previously, one of them provides a quantitative explanation for the 10X variation in failure life. In addition it will be shown that the proposed failure mechanism is quantitatively consistent with fracture mechanics. In order to apply fracture mechanics to this problem, parametric solutions for two different interface fracture problems were developed. The solutions were for elongated edge in a thin layer on an elastic quarter space and for an interface cracks in a pre-compressed film on a cylinder and sphere. In order to use fracture mechanics on this particular problem the available strain energy was determined experimentally using a combination of oxide thickness measurement and oxide stress measurement based on photo luminescent piezospectroscopy (PLPS). Consideration of PLPS measured stresses lead to a conclusion that accurate knowledge of the strain energy available to drive spallation at present is best obtained by measurement and that energy calculated from standard properties can be in error by as much as a factor of 4. Experimental determination of strain energy value is recommended in future studies of spallation mechanics. Progress in the ability to predict oxide stress evolution in such systems is needed.