Effects of Solid Solution Strengthening Elements Mo, Re, Ru, and W on Transition Temperatures in Nickel-Based Superalloys with High γ′-Volume Fraction: Comparison of Experiment and CALPHAD Calculations

This study covers four groups of alloys. The first two groups were fabricated as polycrystalline alloys by arc melting with graded amounts of 1 and 2.5 at. pct Mo, Re, Ru, and W and without Ta (Group 1—EXP alloys without Ta). A group of Ta-containing materials (Group 2—EXP alloys with Ta) was also fabricated, as listed in Table I. The “as-solidified” microstructure was annealed in two steps: a homogenization heat treatment (10 K/min up to 1300 °C/1 hour, 1310 °C/1 hour, and 1320 °C/4 hours, cooling in ambient air) and an aging heat treatment (4 K/min up to 860 °C, 1 K/min up to 900 °C/30 hours, cooling in ambient air). Complete homogeneity of the elements was achieved in all group 1 and 2 alloys.

The third group referred to as “Astra alloys” (Table II) is based on the work of Heckl et al.[17] In our earlier study, the Re and Ru contents were varied, starting from the base composition of CMSX-4. These directionally solidified alloys were heat treated as described in Reference 17. The fourth group (referred to as “ERBO alloys”) consisted of experimental nickel-based superalloys for blade applications, which were investigated at our institute in the framework of the Collaborative Research Center SFB TR 103. All elements could be completely homogenized, except for some residual segregation of Re in the dendritic core.

From each of these 41 heat-treated alloys, cubes having dimensions of 3 mm were cut and used for differential scanning calorimetry (DSC) measurements on a Netzsch STA 409C/CD (Netzsch GmbH, Selb, Germany) apparatus. The heat-treated state was used instead of the as-cast state because otherwise the γ′-solvus temperature could not be determined clearly owing to peak broadening and microsegregation occurring during the solidification process. The solidus temperature is also strongly influenced by these effects (see Figure 1(a)). Additionally, the homogenized state is much closer to the equilibrium conditions (CALPHAD calculation) than the as-cast state. The DSC device was calibrated against the solidus temperature of pure elements (Ni, Co, and Cu) with an absolute error of less than 1 K. The transformation temperatures were determined by plotting the DSC-signal (W/mg) against temperature (°C) and tracking the deviation to the base line of the heating curve. The γ′-solvus was not clearly visible in the cooling curves and therefore, the heating curves are used. Also, the transition temperatures were assumed to be closer to the equilibrium conditions when heating curves are used. In our experience, a heating rate of 5 K/min enables spectra of sufficient quality for peak picking by hand with a reasonable experimental time, and the accuracy of the determined data is suitable for comparison with equilibrium conditions (CALPHAD Calculation). Nevertheless, experiments with different heating rates and an extrapolation of the 0 K/min heating rate would be best for determining the equilibrium phase transition temperatures.

The compositions of the matrix and γ′-precipitate phases of EXP10-18 were measured by electron probe micro-analyzer (EPMA). For this purpose, the size of γ′-precipitates was deliberately increased to be several micrometers by a special heat treatment originally developed by Heckl et al.[17]

To create extra-large γ′-precipitates, the fully heat-treated specimens were further heated to 1280 °C at a rate of 10 K/min, followed by annealing at a rate of 1 K/min up to 1290 °C, and holding at that temperature for 1 hour to achieve complete dissolution of the γ′-phase. A slow cool down to 1100 °C was then performed at approximately 20 K/h to create a small number of nucleating γ′-precipitates, which later grew into large regions. The specimens were held for 12 hours at 1100 °C to allow growth of the precipitates. Afterwards, the temperature was slowly lowered for 8 hours to 700 °C to reduce the amount of tertiary γ′-precipitates formed. A coarsened microstructure with a precipitate size greater than 5 µm was formed and then investigated by EPMA (parameter set: electron beam diameter smaller 1 µm, spacing of 0.25 µm, and 100 × 100 points). The compositions of the matrix and precipitation phase were analyzed at five different spots for each phase. The specimens were clearly subjected to a wide range of temperatures during the coarsening anneal. For the purposes of comparison with calculated data, we assumed that experimental data represented the equilibrium phase compositions at a temperature of approximately 1000 °C.

For every alloy composition, a property diagram (phase fraction vs temperature) was calculated by Thermo-Calc version. 3.1 (Thermo-Calc, Solna, Sweden) with the databases TTNI8 (TermoTech, Surrey, United Kingdom) and TCNI6 (Thermo-Calc, Stockholm, Sweden); MatCalc version 5.61 (MatCalc Engineering, Vienna, Austria) was also used with its database ME-NI (MatCalc Engineering, Vienna, Austria). Note that the MatCalc database was improved by consideration of the experimental data presented in this paper. From each property diagram, the transformation temperatures γ′-solvus, solidus, and liquidus were calculated. In addition, for the alloys from EXP10 to EXP18 the compositions of the matrix and γ′-phase at 1000 °C were extracted. All these values were compared with their corresponding experimental results.