Investigation of spallation mechanisms in thermal barrier coating material systems

The failure of thermal barrier coatings (TBCs) for gas turbine applications is investigated in this thesis. This is an important avenue of research because failure of TBCs results in excessive thermal loading of turbine blades which can lead to catastrophic component failure. In addition, understanding the failure allows for maintenance of components to be more targeted and economical. Finally, a more in depth understanding of TBC failure can inform design improvements enabling gas turbine engines to reach higher temperatures and therefore operate more efficiently with reduced emissions. In terms of real-world applications, the findings from this thesis show that direct measurements of stresses, strains, and thickness variations of thermal barrier coatings can be used by turbine manufacturers and engine operators for the prediction of coating failure, and hence failure of blades themselves. This direct monitoring of turbine blades differs significantly from existing methods of blade life prediction where engine operating conditions are monitored, and blades are periodically replaced when blades have exceeded the safe operational life based on these operating conditions, which is statistically determined from historical turbine failure data. In this thesis, a novel finite element analysis (FEA) model is developed to identify the underlying mechanics of the observed TBC failure behaviour and analytical work is presented that develops and progresses the mechanical understanding of failure of TBC systems. One of the novel aspects of the FEA model is the implementation of a new technique for improving numerical robustness and repeatability of crack nucleation modelling between materials in the absence of contact algorithms when one continuous finite element mesh is used. This technique involves the insertion of high compressive stiffness (crack closing) and zero tensile stiffness (crack opening) interface elements between the separated layers immediately after crack nucleation occurs. These interface elements eliminate the need for contact algorithms which struggle to deal with the high in-plane loads coupled with the large difference in body size between the bulk material and thin coating. Furthermore, a novel method is used for the introduction of strain energy into the system where a small region is subjected to a different cooling rate compared to the bulk material which results in a pocket of increased strain energy. The three most important findings presented in this work are: a previously undocumented mechanism for crack nucleation between thermal barrier coating layers; a newly found driving force for sub-critical crack growth; and the analytical description of non-uniform creep relaxation which can drive failure by means of the introduction of additional strain energy. The mechanism for crack nucleation is a tensile stress (acting away from the metal substrate towards the thin oxide layer) normal to the flat material interface which has only ever been previously observed at the peaks of undulations on rough interfaces. The driving force for crack growth before the onset of thin film buckling is an upwards bending stress distribution within the blister region which generates an energy release rate for crack growth at the crack tip. A thorough literature review is conducted in this thesis which considers experimental and analytical works on the microscopic failure of TBC systems. The experimental work conducted by Tolpygo and Clarke is considered in great detail along with the analytical work conducted by Wang and Harvey. Tolpygo and Clarke identify failure behaviour in their work which cannot be explained with conventional mechanical approaches; Wang and Harvey have therefore developed the pockets of energy concentration (PECs) hypothesis to explain the behaviour. The hypothesis states that PECs form in and around the interface between a thin film and an underlying substrate when the system is subjected to dynamic and non-uniform stress relaxation or creep during cooling. These PECs take the form of regions of tensile and shear stress in and around the interface and are responsible for the nucleation and sub-critical growth of cracks. The PEC hypothesis is supported by an analytical model which predicts the experimentally observed behaviour very well. Much of the work in this thesis is focussed on developing and investigating the PEC hypothesis. In this work, the novel FEA model is verified by modelling a conventional postbuckling problem and comparing the results to predictions made using Hutchinson and Suo’s buckling theory. In addition, stress functions are developed to investigate the nature of PECs in more detail, and to show that a region of increased strain energy in a TBC system can be characterised analytically. Furthermore, the stress function results recreate the FEA results by a purely analytical method, thus validating the FEA. An extremely detailed FEA study of PECs with assumed energy concentrations is conducted. In this study, it is found that a region of increased strain energy, or PEC, in a TBC system is sufficient to drive crack nucleation by means of a tensile stress at the interface between layers. Further to this, the PEC can provide a driving force for crack growth when the crack is below the critical buckling size, and hence an alternative mechanism to buckling is identified for crack growth of thin film circular blisters in TBC systems. Wang and Harvey’s hypothesis has been validated analytically in great detail in published works, and the FEA model presented in this thesis is the first example of a numerical test and validation of the hypothesis. In addition to this validation, the FEA model is used to develop the understanding of TBC failure more generally. An analytical framework is presented to show that a PEC formed by non-uniform creep relaxation in a TBC system can describe experimentally observed failure behaviour. This nonuniform creep is the first potential cause of PECs to be investigated in detail. In addition, the creep behaviour of Kanthal (Fe-Cr-Al) alloys with alumina coatings is investigated and creep laws are developed to characterise the behaviour