Test pieces were machined from vacuum plasma sprayed (VPS) deposits of the bond coat materials Ni-22Cr-10Al-1Y and Co-32Ni-21Cr-8Al-0.5Y. Changes in gauge length at high temperatures under various applied stresses were measured using a scanning laser extensometer. In this way, isothermal creep data and, using no applied load, expansivity data were obtained. The CoNiCrAlY creeps faster than the NiCrAlY at low stresses, but the reverse is true at high stresses. The CoNiCrAlY has an appreciably higher expansivity than the NiCrAlY. These data were used in a numerical process model to evaluate the effect of bond coat creep on the stress state of a TBC system. Comparisons between measured and predicted curvature histories during deposition were used to evaluate the quenching stress for the two materials. This is considerably higher for the CoNiCrAlY. Although creep generally results in reduced stress levels at service temperatures, it can generate residual stresses which are raised after cooling down to ambient temperature, particularly for the CoNiCrAlY. Evaluation of the strain energy release rates associated with various stress distributions, and comparison with measured interfacial fracture energy values, confirmed that debonding will tend to occur at the top coat / bond coat interface, rather than between the bond coat and the substrate. However, bond coat spallation is more likely with CoNiCrAlY than with NiCrAlY.

Thermal barrier coating systems have been heat treated in order to study the oxidation kinetics of the bond coat. All the surfaces of Ni superalloy substrates were sprayed with ~100 μm of a NiCrAlY bond coat, with or without ~250 μm of a ZrO 2 top coat. Thermogravimetric analysis (TGA) was used to monitor continuously the mass change as a result of oxidation of the bond coat during heating at 1000°C for 100 hours in flowing air. In addition, some specimens were heated to 1000°C in static air, cooled to room temperature, weighed and re-heated cyclically. The total exposure time was 1000 hours. Rates of weight gain were found to be higher for the cycled specimens, despite the absence of air flow. This is attributed to damage to the oxide film, which was predominantly α-Al 2 O 3 , as a consequence of differential thermal contraction stresses. The changing residual stress state during heat treatment was predicted using a previously-developed numerical model. A thin (1 mm) substrate with ~100 μm bond coat and ~250 μm ZrO 2 top coat was used in these simulations, which incorporated creep of the bond coat and the lateral strain associated with oxidation. It is concluded from these computations that, while high stresses develop in the oxide layer, the associated driving forces for interfacial debonding remain relatively low, as do specimen curvature changes. It seems likely that coating spallation after extensive oxide layer formation arises because the interface is strongly embrittled as the layer thickens.

An analytical model has been developed to predict the residual stress distributions in thermal spray coatings on substrates of finite thickness. This is based on the concept of a misfit strain, caused by either the quenching of splats or by differential thermal contraction during cooling. During spraying, the coatings are asssumed to deposit on the substrate in a progressive (layer-by-layer) manner. Although the misfit strain ("the quenching strain") is the same for each successive incremental layer of deposit, this is imposed each time on a "substrate" of changing thickness. The final stress distribution will in general differ from that which would result if the coating were imposed on the substrate (with the same misfit strain) in a single operation. The model is straightforward to apply: for example, it can be implemented using a standard spreadsheet program. The required input data are the quenching strain (or stress), the spraying temperature, material properties and specimen dimensions. Comparisons have been made between the predictions from this model and from a numerical model for two plasma sprayed systems. Good agreement is observed. The effects of varying certain parameters, such as coating thickness, substrate thickness, coating stiffness, etc, are readily explored, so that the model provides a useful tool for controlling residual stress levels. Application of the model to determine the quenching stress, in conjunction with the use of a curvature monitoring technique, is briefly outlined. In addition, an analysis is made of the errors introduced by using Stoney's equation to deduce stress levels from curvature measurements.

The adhesion of various interfaces in thermal barrier coating systems strongly affects their stability and thermal cycling life. A new spontaneous debonding technique and the four point bend delamination test have been applied to measure the interfacial fracture toughnesses of various interfaces in several thermal barrier coating systems. The spontaneous debonding technique is based on spraying a relatively stiff layer on top of the ZrO 2 coating. This raises the strain energy release rate for debonding, the magnitude of which is monitored via modelling of the stress distribution. The critical strain energy release rate for debonding (interfacial fracture energy) was then determined from the stress states before and after debonding. Thermal barrier coatings (TBCs), consisting of a Ni-22wt.%Cr-10wt.%Al-lwt.%Y bond coat and a ZrO 2 -8%Y 2 O 3 top coat, were deposited on a nickelbased superalloy. Two methods, air (APS) and vacuum (VPS) plasma spraying, were used to produce the bond and the top coats. The corresponding as-sprayed residual stress distributions and the interfacial fracture energies were evaluated. It was found that a VPS bond coat and an APS top coat produced the most mechanically stable structure. A layer of vacuum plasma sprayed AI 2 O 3 was then introduced between the top and the bond coat, designed to act as an oxygen diffusion barrier. The effect on residual stress distributions, and associated crack driving forces for debonding, at different interfaces were determined. The effect of the alumina layer on the oxidation behaviour was also studied. It is shown that the oxidation barrier could significantly enhance the coating life-time.