Oxidation and degradation of nickel-base alloys at high temperatures

Abstract: This master’s thesis work is a study of oxidation and degradation of nickel-base alloys at high temperatures. The materials studied are designed for use in critical gas turbine components such as turbine blades and vanes. Some of the alloys are used today, whereas others have not yet entered commercial application. In order to maximize the efficiency of gas turbines, there is an ambition to maximize the operating temperatures. There is therefore a demand for materials which can withstand the damage mechanisms active at high temperatures. Among these damage mechanisms are oxidation and microstructural degradation. To investigate the oxidation resistance of 7 different monocrystalline and polycrystalline alloys, samples have been exposed isothermally in still air at temperatures between 850 and 1000°C, for exposure times of up to 20000h. Two of the alloys were also exposed cyclically at 950°C. Oxidation during the heat treatment resulted in significant weight changes, which were measured after each cycle for cyclically exposed samples and after completed heat treatment for isothermally exposed samples. The weight change data was used to evaluate the relative oxidation resistance of the alloys. The ranking of the alloys with respect to oxidation resistance was generally in agreement with the oxidation resistance predicted by a simple consideration of the Cr and Al contents of the alloys. However, the single-crystal alloy PWA1483 displayed better oxidation resistance than predicted from its chemical composition. Metallographic analysis of the samples indicated that the oxide scales formed consisted of several different types of oxides. The oxide scales were mainly composed of Cr2O3 and Al2O3. Fragments of the oxide scales spalled off, primarily during cooling but also in some cases during the long-term heat treatments. Spalling of the oxide scale accelerated the oxidation process, since the ability of the oxide scale to impede diffusion decreased with its decrease in thickness. Oxidation caused depletion of Al and thereby local dissolution of the aluminum-rich γ′ particles, which are of vital importance to the mechanical properties of the material. A γ′ depleted zone thereby formed underneath the oxide scale. In this zone nitrides and needle-like particles, believed to be topologically close packed μ phase, precipitated during heat treatment. Recrystallization in the depletion zone was observed in some of the monocrystalline materials. MC carbides (M=metal) present in the virgin material decomposed during heat treatment and M23C6 carbides were formed. The γ′ particles coarsened during heat treatment, which resulted in decreased hardness. The hardness decreased with exposure temperature up to 950°C, as expected due to the increased coarsening rate. At 1000°C an unexpected increase in hardness was observed for all sample materials except one. A possible explanation for this hardness increase is redistribution of γ′, by dissolution of γ′ during heat treatment and reprecipitation during cooling as much finer particles. A fine dispersion of γ′ is expected to contribute more to the hardness than a corresponding volume of γ′ in the form of larger particles. For some of the sample series, clear correlations between hardness and γ′ particle size or exposition time were found. These relationships could potentially be used to estimate the exposure temperature of service-exposed material. A numerical model was implemented in Matlab to describe the process of oxide growth and spalling, cycle by cycle. The model was successfully adapted to experimental data from the cyclic oxidation measurements. The general applicability of the model to cyclic oxidation data at different temperatures and cycle frequencies was not investigated. At long times of cyclic exposure, the net weight loss of the samples could be well approximated as a linear function of the number of cycles. However, during the last few cycles the amount of oxide spalled in each cycle suddenly decreased. This change in spallation behavior was mainly observed for the samples cooled in air between every cycle and to a much smaller extent for the samples cooled in water. The proposed explanation is that spalling occurred preferentially at a weak subscale interface and that the spalling propensity decreased with decreasing area of this weak interface. The deviating results of the last few cycles were not included in the modeling of the cyclic oxidation process.