The Effects of Chemistry Variations in New Nickel-Based Superalloys for Industrial Gas Turbine Applications

A set of six trial alloys, denoted in the following as SX1 to SX6, was pre-alloyed either from pure elements (Al, Cr, Co, Fe, Ni, Si, Ti) or from binary Ni-X alloys for refractory metals (Ni-Hf, Ni-Mo, Ni-Ta, Ni-W). Batches of 120 g were weighed, arc-melted under Ar using an Edmund Bühler AM 500 arc melter, and poured into water-cooled copper molds. Nominal chemical compositions in at. pct are presented in Table I. The base composition of the investigated alloy series, denoted as SX1, was derived from STAL15SX to the design point of 14.2 at. pct Cr and 12 at. pct Al. Four compositions were designed to assess the influence of adding 0.3 at. pct Hf (SX2), 5.2 at. pct Fe (SX3), and 1 at. pct Si (SX6) and of increasing the Co content from 5 to 10 at. pct (SX5). Finally, a composition with lower Al and Ta levels and containing 2.9 at. pct Ti (SX4) was also included as a ‘combination’ between the base alloy and IN-792.

Differential scanning calorimetry (DSC) testing was carried out using a Netzsch DSC 404 F1 Pegasus. Measurements were carried out at heating/cooling rates of 10 °C/min under a high purity Ar atmosphere at a purge flow rate of 40 mL/min. PtRh crucibles with PtRh lids and thin-walled Al₂O₃ liners were used in combination with a high-accuracy sample carrier system with type S thermocouples. Disk specimens of 3 mm diameter and 1 mm thickness were manufactured by wire-guided electro-discharge machining (EDM). Both circular surfaces were polished down to 4000 grit silicon carbide grinding paper to achieve good thermal contact with the sensor. Each specimen was ultrasonically cleaned in ethanol and weighed before the test.

For each test, two heating/cooling cycles were performed between 700 °C and 1450 °C to assess the influence of prior melting/solidification and of reduced segregation on the second cycle. To equalize the temperature in the system, constant temperature segments of 15 minutes duration were added between ramp steps. Following the ASTM method, three measurements were carried out as part of each test: (1) a baseline measurement with an empty sample pan, (2) a calibration measurement with a synthetic sapphire disk of known mass and heat capacity, and (3) a sample measurement with the actual superalloy disk. The reference pan remained empty during all three runs. Finally, changes in alloy heat capacity, \(C_{\text{p}}\), with temperature, T, were calculated using the Netzsch Proteus analysis program.

Preliminary DSC studies on as-received material gave an indication of the \(\gamma ^\prime \) solvus and solidus temperatures for each alloy. Using this information, solutioning heat treatments were carried out to reduce microsegregation between dendrite core and interdendritic regions, while avoiding incipient melting. Samples were placed in a pre-heated furnace at 1170 °C. The temperature was then slowly ramped up at a constant rate of 0.1 °C/min up to 20 °C below the expected solidus, following the method proposed by Hegde et al.[15] After reaching the peak temperature, specimens were cooled down at 10 °C/min to 850 °C and were then removed from the furnace and air-cooled. Subsequent primary and secondary aging heat treatments were chosen at 1120 °C for 4 hours and 845 °C for 24 hours, both completed by air-cooling.

The chemistries of fully heat-treated specimens were checked at an independent laboratory using inductively coupled plasma optical emission spectrometry (ICP-OES) for metallic elements and combustion analysis for C and S impurities. To further assess the remaining degree of microsegregation, chemistries were also checked on polished specimens by electron probe microanalysis (EPMA) using a JEOL JXA 8800 Superprobe equipped with four wavelength-dispersive X-ray spectrometers (WDS) for parallel detection of elements. Line scans of 200 points were carried out at a spacing of 10 µm, a probe diameter of 10 µm, an accelerating voltage of 25 keV, a beam current of 32 nA, and a dwell time for each point of 100 ms. The diffracting crystals used for WDS were LIF for Ni, Co, W, Ta, Hf, and Fe, PETJ for Cr, Mo, and Ti, and TAP for Al and Si. X-ray counts were converted to concentration values using standard correction procedures and signals from pure element standards. Average compositions obtained from ICP-OES and EPMA are shown in Table I below the nominal values.

Metallographic preparation was carried out on a Struers Tegramin-25 system down to 1 µm diamond suspension. An optimal finish was achieved by polishing for 5 minutes with a 1 to 1 mixture of Struers OP-S 0.04 µm colloidal silica suspension and water on a Struers MD-Chem polishing cloth. Specimens were cleaned in an ultrasonic bath with acetone and then with ethanol. Backscatter electron (BSE) images of fully heat-treated microstructures were obtained using a Zeiss Merlin field emission gun (FEG) SEM operating at an accelerating voltage of 10 keV and a probe current of 500 pA. The working distance was set at 8.5 mm to yield optimal energy-dispersive X-ray spectroscopy (EDX) results using an Oxford Instruments X-Max^N silicon drift detector with a size of 150 mm². EDX data were processed using the TruMap function of the Oxford Instruments AZtec 3.3 software package to eliminate artifacts, correct background and peak overlaps, and enhance compositional variations.

Matchsticks of \(0.5 \times 0.5\times 2\hbox { mm}^{3}\) were manufactured by EDM from fully heat-treated material. Sharp tips for atom probe tomography (APT) with a radius below 100 nm were produced in a two-stage electro-polishing process. The matchsticks were first coarse-polished in a solution of 10 vol pct perchloric acid in acetic acid at a voltage of 12 to 20 VDC and then fine-polished in a solution of 2 vol pct perchloric acid in 2-butoxyethanol at 8 to 12 VDC. APT data were collected using a CAMECA local electrode atom probe (LEAP) 5000 XR at a specimen base temperature of 50 K and an ambient gauge pressure of less than 10^-10 Pa. Samples were run in laser-assisted mode with an ultraviolet laser wavelength of 355 nm, a pulse energy of 50 pJ, a pulse frequency of 125 kHz, and a constant ion detection rate of 0.5 pct. 3D reconstructions were generated using the CAMECA IVAS 3.8.2 software package. Peak overlaps of multiple elemental species were deconvolved using the standard IVAS decomposition parameters.

Throughout the present study, experimental results are compared with predictions made using version 2019b of the Thermo-Calc software package and two associated databases for nickel-based superalloys—TCNi8, provided by Thermo-Calc Software AB, and TTNi8, provided by ThermoTech Ltd. The chemical compositions measured by ICP-OES were used as inputs. Phase diagrams were generated between 600 °C and 1400 °C. Equilibrium phase compositions were calculated at the final aging temperature of 845 °C. In both cases, a simplified three-phase system consisting of \(\gamma \), \(\gamma ^\prime \), and liquid at high temperatures was assumed and all other phases were suspended. SEM results support this assumption, as only insignificant levels of small MC carbides and no topologically close-packed (TCP) phases were observed after heat treatment.

Chemical compositions determined by ICP-OES and EPMA are in good overall agreement with nominal values. Trace levels of Fe were detected, but no other impurities are present. Furthermore, EPMA line scans show little variation across the samples, implying good homogenization and low microsegregation after the heat treatment. Nonetheless, some discrepancies can be noted; the levels of Cr in SX2 and SX6, Co in SX6, W in SX2, and Ta in SX6 are somewhat lower than intended, while the levels of Ta in SX2 and Fe in SX3 are higher.

The microstructures shown in Figure 2 are characteristic of IGT superalloys and are composed of cuboidal, secondary \(\gamma ^\prime \) precipitates with side lengths on the order of 1 µm and spherical, tertiary \(\gamma ^\prime \) particles with diameters on the order of 10 nm, embedded in the \(\gamma \) matrix. Secondary particles form after the homogenization step during cooling from around 1260 °C, while tertiary ones precipitate during the subsequent aging heat treatments at 1120 °C and 845 °C. Secondary precipitate morphology varies between the alloys and appears more cuboidal for SX2, SX3, SX5, and SX6 and more rounded for SX1 and SX4. The substitution of Ta for Ti in SX4 results in a marked reduction in contrast between \(\gamma \) and \(\gamma ^\prime \), as the heavy Ta makes precipitates appear brighter in BSE micrographs.

Particle size distributions were analyzed using the image processing software package MIPAR. Cross-sectional area fractions comprising both secondary and tertiary \(\gamma ^\prime \) were measured for each BSE micrograph. These can be considered equal to \(\gamma ^\prime \) volume fractions from a stereological standpoint as long as the precipitates do not have a preferred orientation.[16] Values obtained by averaging the results from at least five micrographs are shown in Table II, along with Thermo-Calc predictions for equilibrium conditions at 845 °C.

EDX maps of elements added to the base composition of SX1 are presented in Figure 3. Hf and Ti are found to partition preferentially to \(\gamma ^\prime \), whereas Fe and Co are enriched in \(\gamma \). EDX results for Si are inconclusive, in part due to peak overlaps with the Ta largely present in \(\gamma ^\prime \), as shown for the SX1 reference.

Figure 7 illustrates an exemplary 3D APT dataset collected for the base alloy SX1. The \(\gamma \) matrix and a \(\gamma ^\prime \) precipitate are identified unambiguously as a result of the preferential partitioning of Cr, Co, and Mo to \(\gamma \) and of Al and Ta to \(\gamma ^\prime \). To highlight the \(\gamma \)/\(\gamma ^\prime \) interface, an iso-concentration surface was generated at 20 at. pct Cr. This cut-off value is used consistently for all six alloys.

Hf became a key addition to superalloys after it was found to significantly increase creep ductility and the time to rupture by influencing grain boundary microstructure in both conventionally cast and directionally solidified components.[23,52,53] Further advantages of adding Hf include improved castability, as the probability of hot tearing is reduced,[25,26] and increased resistance to sulfidation and to isothermal and cyclic oxidation, as Hf acts as a getter of S and improves the adherence of oxide scales and of thermal barrier coatings.[54‐56] A main downside of Hf is its strong reduction of the available HTW, although for SX2 this has been shown to remain large enough to enable full homogenization.

The results in Table V and Figure 8(a) reveal a further benefit of Hf in SX2, as the partitioning coefficients of W and Mo are reduced. A higher content of slow-diffusing W and Mo in the \(\gamma \) matrix implies a reduced rate of diffusion-controlled dislocation movement, which constitutes a key damage accumulation mechanism during service for IGT alloys.[11] The effect of Hf on partitioning behaviors can be understood by considering the preference of different elements for either Ni \(\alpha \)-sublattice sites or Al \(\beta \)-sublattice sites within the \(\gamma ^\prime \) phase. Results of previous APT studies on nickel-based superalloys are compiled in Table VIII, while Thermo-Calc predictions for equilibrium conditions at 845 °C are shown in Table IX. It emerges that Hf, Mo, and W all prefer Al sublattice sites, suggesting that W and Mo are displaced by Hf into the \(\gamma \) matrix in SX2.

From an alloy design perspective, it must be pointed out that the preferential partitioning of Hf towards \(\gamma ^\prime \) in SX2, which directly influences its efficiency at displacing W and Mo, has been shown to vary with alloy chemistry. Recent studies[38,57] proposed that \(K_{\text{Hf}}^{\gamma ^\prime /\gamma }\) depends on the ratio of Cr to Co additions. At a low Cr/Co ratio, it is anticipated that Co will stabilize the \(\gamma ^\prime \) phase for Hf, whereas at a high Cr/Co ratio, Cr will increase the solubility of Hf in the \(\gamma \) matrix and will reverse partitioning.[58,59] However, the data compiled from literature in Table X suggest that Hf partitioning is not as simple as previously suggested. Alloy SX2, with double the Cr/Co ratio compared with ME-9 and ME-15, exhibits a \(K_{\text{Hf}}^{\gamma ^\prime /\gamma }\) value two orders of magnitude higher. Furthermore, alloys CMSX-4 and RR1000, with similar Cr/Co ratios to ME-9 and ME-15, still exhibit a clear partitioning of Hf towards \(\gamma ^\prime \). It is therefore suggested that the dependence of \(K_{\text{Hf}}^{\gamma ^\prime /\gamma }\) on the Cr/Co ratio only holds true for low Hf additions on the order of 0.05 at. pct.

APT observations and first-principle calculations of substitutional formation energies have revealed a strong effect of Ta additions for displacing W into the \(\gamma \) matrix and reversing its partitioning.[60‐62] This has been related to the high affinity of Ta for the \(\gamma ^\prime \) phase and to the repulsive interatomic interactions of Ta-W dimers. More recently, a similar, albeit weaker effect was found for Ti,[63,64] in agreement with the site preferences in Tables VIII and IX. Despite its downsides of lowering oxidation resistance[65‐67] and the available HTW,[19,41] Ti has the advantage of significantly lower density and cost compared with Ta, and it has been given preference in recent design studies of superalloys with optimized solid solution strengthening.[63, 64]

Results from image analysis of SEM micrographs indicate a significant increase in the \(\gamma ^\prime \) volume fraction of SX4, in agreement with the raised solvus temperature and with Thermo-Calc predictions. Thus, the reduced partitioning coefficients of W and Mo in SX4 can be related to both a higher precipitate volume fraction and to the effects of Ti and Ta atoms occupying Al sublattice sites.[64] The value of \(K_{\text{Cr}}^{\gamma ^\prime /\gamma }\) also decreases, suggesting that Cr may preferentially occupy \(\beta \)-sites in the trial IGT alloys considered here. However, it should be noted that the sublattice preference of Cr can vary strongly with composition and heat treatment,[43] as summarized in Table VIII.

In agreement with the reduced solvus temperatures and with Thermo-Calc predictions, the \(\gamma ^\prime \) volume fractions of alloys SX3 and SX5 are lower than for SX1. Accordingly, it might be expected that the concentrations of refractory elements in the \(\gamma \) matrix will decrease with added Fe and Co. Furthermore, Table VIII shows only a weak preference of Fe for Al sublattice sites in a model Ni-Al-B-Fe alloy, while Co is found to either occupy Ni sites preferentially or both Al and Ni sites in an equal measure. Nonetheless, a noticeable reduction of \(K_{\text{Mo}}^{\gamma ^\prime /\gamma }\) and \(K_{\text{W}}^{\gamma ^\prime /\gamma }\) is observed for both SX3 and SX5, thus warranting a careful analysis.

A review of relevant literature reports reveals a complex and highly composition-dependent influence of Co on the partitioning coefficients of other elements. For Udimet 700 alloys with Co contents between 0 and 16 at. pct, Jarrett and Tien[31] found using STEM-EDX that \(K_{\text{Cr}}^{\gamma ^\prime /\gamma }\) decreases and \(K_{\text{Ni}}^{\gamma ^\prime /\gamma }\) increases with added Co, in agreement with the present work, while the partitioning behaviors of Co, Al, Ti, and Mo were not affected. Surprisingly, Harf[33] found no significant changes in partitioning for the same set of Udimet 700 compositions, when these were prepared by a powder metallurgical route. In an extensive study on the influence of Co, Ta, and W on the microstructure of single-crystal superalloys similar to NASAIR 100, Nathal and Ebert[24] measured phase chemistries after electrolytic phase extractions and only found significant changes for Ni, which partitioned more strongly to \(\gamma ^\prime \) as the Co content was increased from 0 to 10 at. pct. The same results were reported by Zhi’An et al.,[68] who examined a set of Udimet 500 alloys with 0 to 23.5 at. pct Co. A recent APT study on model Ni-Cr-Co-Al-Ti alloys containing between 0 and 57 at. pct Co[35] found that, between 0 and 19 at. pct Co, \(K_{\text{Cr}}^{\gamma ^\prime /\gamma }\) decreases and \(K_{\text{Ni}}^{\gamma ^\prime /\gamma }\) strongly rises, in accordance with results for SX1 and SX5, however, also contrarily that \(K_{\text{Al}}^{\gamma ^\prime /\gamma }\) increases. The best agreement of results for SX5 is found with a report by Nathal et al.[34] on conventionally cast Mar-M247 containing between 0 and 10 at. pct Co. These authors found that \(K_{\text{Ni}}^{\gamma ^\prime /\gamma }\) strongly increases and that \(K_{\text{Cr}}^{\gamma ^\prime /\gamma }\), \(K_{\text{Al}}^{\gamma ^\prime /\gamma }\), and \(K_{\text{W}}^{\gamma ^\prime /\gamma }\) gradually decrease as more Co is added. All in all, this brief review demonstrates that the impact of Co in superalloys has not yet been fully understood and that its effects are highly dependent on composition and even heat treatments.[32]

The scarcity of data on alloys containing relatively low levels of Fe makes it difficult to assess the result for SX3. The seminal review by Ochiai et al.[69] on site preferences in Ni₃Al showed a mixed behavior for Fe and a roughly equal preference for Ni and Al sublattice sites. An earlier study by Nicholls and Rawlings[70] on Ni₃Al(Fe) alloys found the site preference of Fe changes from Al \(\beta \)-sublattice sites at 2.5 at. pct Fe to mixed at 9.3 at. pct. This is in agreement with the work of Miller and Horton,[71] who found a weak \(\beta \) preference in a Ni-Al-B-Fe alloy with 6 at. pct Fe. Thus, it can be surmised that most of the 2 at. pct Fe measured for the \(\gamma ^\prime \) phase of SX3 will occupy Al sublattice sites. This can explain why the partitioning coefficients of Mo, W, Al, and Ta all decrease. From an alloy design perspective, potential improvements in oxidation resistance associated with Fe must be carefully considered against the significant reduction in \(\gamma ^\prime \) volume fraction, which will negatively impact high-temperature mechanical properties.

A similar situation to Fe emerges for Si additions in superalloys. Previous research has shown that while Si improves hot corrosion and oxidation resistance, it can also negatively affect mechanical properties and promote the formation of deleterious TCP phases.[5,27‐29,72] An optimal Si level of around 0.5 at. pct was reported for alloy STAL15SX,[5,72] which possesses a similar chemical composition to the trial IGT alloys considered here.

Analogue to Hf, the partitioning preference of Si towards \(\gamma \) or \(\gamma ^\prime \) is influenced by the overall chemical composition. Based on results from several APT studies compiled in Table X, a good correlation is found between \(K_{\text{Si}}^{\gamma ^\prime /\gamma }\) and the ratio of nominal Cr/Co compositions. Therefore, the aforementioned considerations regarding the stabilizing effects of Cr and Co on \(\gamma \) and \(\gamma ^\prime \), respectively, may explain the stronger partitioning of Si towards \(\gamma \) for an increasing Cr/Co ratio. It should be noted that Yeh et al.[29] previously related this shift towards \(\gamma \) to the presence of Ti; however, this hypothesis is not supported by the present work nor by APT results for STAL15SX.[73] A comparison of partitioning coefficients for alloy CM247LC without Si and with additions of 0.53 and 1.06 at. pct suggested that Si shifts all \(K^{\gamma ^\prime /\gamma }\) values towards unity.[29] This trend is also observed in SX6 for Ni, W, and Ta, but other results are inconclusive. Given the reduced nominal level of Al in this alloy, it is challenging to isolate the effects of Si.

An interesting result observed in Figure 9 for alloys SX2, SX4, and SX5 is an enrichment of W at \(\gamma \)/\(\gamma ^\prime \) interfaces. No such effect could be observed for SX1, SX3, or SX6, suggesting that it is related to the reversed partitioning of W towards the \(\gamma \) matrix. Extensive work on the partitioning behavior of W and the influence of Ta[60‐62,74] clearly showed that W segregation only occurs in alloys with a high enough Ta content to effectively displace W atoms from their preferred Al \(\beta \)-sublattice sites and into the \(\gamma \) phase. Conversely, no W enrichment was observed in model ternary, quaternary, and quinary alloys with lower levels of Ta or in the Ta-depleted dendrite core regions of multi-component superalloys.[60‐62,74] Considering the literature reports compiled in Table XI, it appears the condition \(K_{\text{X}}^{\gamma ^\prime /\gamma }<1\) is indeed a prerequisite for the enrichment of solute X at \(\gamma \)/\(\gamma ^\prime \) interfaces, but it does not guarantee this will indeed occur.

From an alloy design perspective, it is important to assess what gives rise to interfacial solute segregation and, more so, how this might impact high-temperature creep resistance. Early APT studies hypothesized that slower-diffusing solutes like Ta, W, or Re may preferentially segregate to \(\gamma \)/\(\gamma ^\prime \) interfaces, and this was indeed observed for the case of Re in several single-crystal superalloys.[75‐79] However, it was concluded that this enrichment does not simply arise for reasons of thermodynamic equilibrium, but rather that it is related to the growth of \(\gamma ^\prime \) precipitates during cooling[77,79]. Phase field modeling conclusively showed that the diffusion rate of Re is too slow to allow equilibrium particle growth and that, as the \(\gamma \)/\(\gamma ^\prime \) interfaces migrate during cooling, a Re ‘bow wave’ forms ahead of them.[77,79] The same type of profile has also been reported for W segregation[39,61,62] and is observed in the present work. Furthermore, W enrichment has only been observed at the interfaces of secondary, cuboidal precipitates, but not at the curved interfaces of tertiary, spheroidal precipitates. As the latter form during isothermal aging heat treatments, there is no comparable supersaturation as during cooling after the solutioning step that could give rise to interfacial segregation. All these considerations suggest that the excess W will no longer be present at interfaces during long-term creep exposure, as the alloy microchemistry will gradually approach an equilibrium state. Nonetheless, the higher molar fractions of W and Mo in the \(\gamma \) phase of alloys like SX2, SX4, and SX5 are expected to increase creep resistance. As slow-diffusing atomic species, W and Mo have the beneficial effect of imposing a drag effect on dislocation movement, thus reducing the creep strain rate.[51,80,81]