Effects of conventional heat treatments and service conditions on selective laser-melted Ni-based superalloys for gas turbine applications

There has been increasing demand in recent years for industrial gas turbine components manufactured by Selective Laser Melting (SLM) and for these components to experience more and more aggressive environments. Therefore, in this research, two alloys, IN939 and Hastelloy X were subjected to a variety of industry standard heat treatments, conditions designed to simulate those experienced during service and mechanical testing to analyse the effect of these on the microstructure of the SLM material, in comparison to conventionally-produced variants of the same material.

Firstly, samples of selective laser melted (SLM) and cast IN939 were subjected to an industry standard heat treatment and the effects on the microstructure were compared between the two different materials. The SLM samples had a very different as-produced microstructure to the cast material, which is typical for SLM material. There were a number of differences in the effects of the heat treatment between the two materials. The SLM material underwent almost complete recrystallisation whereas the cast grain structure remained similar to the as-cast material throughout the heat treatment. The MC carbides which precipitated in the SLM material were also much smaller, (~0.1µm² compared to ~10µm² ) and more numerous (~0.1µm^-2 compared to ~0.0015 µm^-2 )than in the cast material, and also behaved differently throughout the heat treatment process. γ’ morphology was not significantly different between the SLM and cast materials examined.

Samples of SLM and Cast IN939 were then isothermally aged for long periods of time at temperatures representing the low and high end of the range of service temperatures expected for this alloy: 700°C and 950°C. The effects on the microstructure were then compared. There were similar effects from the aging on both cast and SLM materials. At 700°C after long aging periods, there was some slightly different precipitation behaviour of σ phase. Rather than precipitating on the MC carbides as the σ phase did in the cast material, the σ phase in the SLM material appeared to precipitate in isolation. At 950°C there was some abnormal growth of γ’ precipitates near grain boundaries and some formation of η phase within the SLM material, neither of which were observed in the cast samples.

Samples of SLM and cast Hastelloy X were then also isothermally aged for long periods of time at temperatures representing the low and high end of the range of service temperatures, in this case, 625°C and 850°C and the microstructural effects were compared. At 625°C the SLM samples showed linear precipitation patterns that were uniform across the grains in the material, whereas the cast material showed a gradient of secondary precipitation density with increasing distance from primary carbides. After a long period of time the morphology of the carbides in the SLM material changed and became lower aspect-ratio, while in the cast material it followed the same pattern of increasing aspectratio throughout the aging period studied. The samples aged at 850°C showed a similar initial distribution of carbides to the cast and SLM samples aged at 625°C, but it was less pronounced due to the higher temperature. µ phase was formed in both SLM and cast samples at 850°C, but in the SLM material the phase may have formed from the matrix itself whereas in the cast material it was mainly formed by degradation of M₆C carbides. Aging at 625°C caused an increase in tensile strength but a reduction in ductility of the SLM material, whereas aging at 850°C caused reductions in both strength and ductility.

An ex-service, unstressed Hastelloy X component was then analysed and showed some effects similar to the isothermally aged material. However, there were a number of differences; the hottest areas of the component had undergone recrystallisation, and some areas also showed different precipitate morphology due to cyclic heating and different operation regimes experienced by the component.

Samples of SLM Hastelloy X were then creep tested, some in the as-produced state and some after a 900°C, 1-hour heat treatment. The samples showed some unusual creep behaviour with a large reduction in creep rate after a certain period of time, which was not observed in data from creep testing of cast Hastelloy X. This was theorised to be due to the precipitate behaviour in the material; after a certain period of time there was a large amount of precipitation and growth of M₆C carbides in the material, causing the spacing between adjacent precipitates to reduce enough that the carbides imparted a significant strengthening effect to the material, causing a reduction in creep rate. After a further period of time, coarsening of the carbides occurred, increasing the spacing and reducing the strengthening effect, and causing the creep rate to increase again. There were some differences between the heat-treated and non-heat-treated samples. The heat treatment removed a large amount of the segregation in the material and accelerated precipitation in the material, resulting in large carbides and less of a linear pattern in the precipitate distribution. The strengthening effect of the carbides was also reduced, as shown by less of a reduction in the creep rate.