Effect of Alloying on the Microstructure, Phase Stability, Hardness, and Partitioning Behavior of a New Dual-Superlattice Nickel-Based Superalloy

There is a global drive to improve engine efficiency and reduce emissions in civil aviation. This motivates the development of new Ni-based superalloys capable of withstanding the higher operating temperatures and increased rotational speeds demanded in the next generation of gas-turbine engine designs.^[1] Currently, Inconel® 718 (Ni-20.5Cr-18.85Fe-0.33Co-1.84Mo-1.18Al-3.3Nb-1.23Ti-0.34C-0.02B at. pct)^[2] remains the most widely used alloy for intermediate temperature applications due to its favorable combination of processability and mechanical strength up to 650 °C. Indeed, it constitutes approximately one-third of the weight of some engines.^[3] Inconel® 718 is strengthened by the precipitation of two ordered intermetallic phases, ~ 3 pct γ′ (based on Ni₃Al with the L1₂ structure) and ~ 20 pct γ″ (based on Ni₃Nb with the D0₂₂ structure) embedded coherently within the A1 γ Ni matrix.^[4] The γ″ phase confers additional strengthening over the γ′ phase alone (on a normalized volume fraction basis) due to the greater coherency strain resulting from the lattice misfit between the tetragonal c-axis of the γ″ and the γ matrix. However, the γ″ phase is also metastable, and rapidly transforms to the thermodynamically stable orthorhombic D0_a δ phase at temperatures above 650 °C, limiting the upper operational temperature ^[5,6].

Alloys with superior temperature capability and strength to Inconel® 718 have been developed, although such alloys typically rely on higher refractory metal contents and precipitate volume fractions. Consequently, these alloys possess narrow processing windows, exhibit rapid work hardening, are not readily weldable and are susceptible to hot shortness.^[7‐11] While such alloys may be processed using powder metallurgical routes, this significantly increases alloy costs compared to cast and wrought products. As a result, there remains significant interest in developing affordable cast and wrought alloys with superior properties to Inconel 718®, while simultaneously retaining the desirable processability characteristics.

Efforts to enhance the temperature capabilities of 718-type alloys have investigated compositional modifications as well as carefully defined heat treatment schedules to modify the precipitate morphology. Such efforts have been successful, but have only led to relatively modest improvements in thermal stability.^[4,12] More recently, studies of alloys based on Ni-Cr-Al-Nb have identified the possibility of alloy microstructures containing appreciable volume fractions of both γ′ and γ″.^[5,13] These alloys were reported to have yield strengths > 100 MPa higher than many commercial Ni-based superalloys such as 718Plus, Haynes 263, Haynes 282, IN718, Incoloy 909, ME3, Rene 88, Rene 95, RR1000, U720, Rene 220, and Waspaloy. Subsequent work has demonstrated that these alloys retain microstructural stability up to 750 °C^[14] and that further improvements to their properties may be achieved through judicious alloying.^[7] However, if these alloys are to be effectively optimized for high temperature service, a detailed understanding must be established of how alloying additions affect the relative stability, morphology and properties of the constituent phases.

The original studies of dual-superlattice superalloys reported that an alloy with composition of Ni-15Cr-4Al-6Nb (at. percent) contained ~ 25 γ′ and ~ 20 pct γ″. While this alloy demonstrated attractive properties, three alloy development opportunities may be identified based on established superalloy metallurgy. First, the phase fractions are higher than many other superalloys derived from Inconel® 718. This may adversely affect the processability of the alloy. It is therefore appropriate to consider alloys with lower Al and Nb contents to determine the extent to which the beneficial properties of a dual-superlattice microstructure may be retained with lower precipitate fractions. Second, the quaternary composition affords comparatively limited solid solution strengthening of the γ matrix. As such, further benefits may be derived from considering the addition of elements such as Mo and W to provide effective strengthening.^[15] Third, the elemental cost of the dual-superlattice alloy is higher than Inconel® 718. It is therefore appropriate to consider the extent to which Fe substitutions may be made as this will significantly reduce alloy cost and ease recycling.

Previous studies have shown that Mo and W additions partition strongly to the γ matrix, where they provide potent solid solution strengthening on a per atomic percent basis.^[16,17] W preferentially sits at Al sites in γ′, but if Ta is present in the alloy system, W will instead partition to the γ matrix.^[18] However, these elements are also known to promote the formation of topologically close-packed phases (TCPs) such as σ and μ phase,^[19‐21] which are considered deleterious and lead to reduced creep lives. The concentrations of these elements must therefore be carefully controlled to ensure microstructural stability.

In sufficient concentrations, Nb promotes the formation of Ni₃Nb γ″ precipitates, as well as contributing to the γ′ precipitates where it substitutes for Al. As it does not partition strongly to the γ′, Nb additions usually have a limited effect on the γ–γ′ lattice misfit.^[22] Nb also increases the anti-phase boundary (APB) energy of the γ′, as well as acting as a solid solution strengthener in both the γ matrix and γ′ precipitates.^[22,23] The addition of Nb has also been reported to reduce precipitate coarsening rates, which has been attributed to its sluggish diffusion kinetics in Ni.^[24]

Fe is an important constituent of Inconel® 718 and related alloys. It significantly reduces raw material costs, while simultaneously improving processability and recyclability. However, if the content is too high, deleterious Laves phases (Ni,Fe,Cr)₂(Nb,Mo,Ti) can be formed, which compromise the mechanical properties. Fe has also been reported to aid in the formation of γ″ precipitates by lowering the electron:atom ratio to values around 2.50 to 2.62, which, as predicted using Engel–Brewer correlations, are preferred for stable D0₂₂ precipitation.^[25]

The present study involves a combination of microstructural and chemical characterization experiments and thermodynamic modeling, used to understand the properties of the resulting alloys. Six systematic compositional variations of the dual-superlattice alloy were produced to study the effect of Mo, Nb, W, and Fe on the phase stability, solvus temperature, and elemental phase partitioning. The results obtained give insight into the extent to which each of these additions may be accommodated without compromising the alloy microstructure and may therefore help inform future development of these alloys.

A series of six polycrystalline Ni-based superalloys were produced through vacuum arc melting from raw elements of > 99.9 wt pct purity, the alloy designations and nominal compositions of which are shown in Table I. The first alloy in this series, the Base alloy, was designed to contain the same Cr, Al, and Nb contents of the original dual-superlattice superalloy investigated in Reference [13]. In addition, small concentrations of C, B, and Zr were included to provide grain boundary strengthening, in line with other polycrystalline Ni-based superalloys. The other five alloys were designed with consecutive compositional changes. From the composition of the base alloy, 1.8 at. pct Mo was substituted for Ni to produce the variant + 1.8Mo, having a composition similar to that in Inconel® 718. Next, the Nb composition was reduced to 5.4 at. pct, a reduction of 10 pct of the overall Nb content, to produce the alloy – 0.6Nb. For the third variation + 9Fe, 9 at. pct Fe was added to the composition of 0.6Nb, similar to the composition in ATI 718Plus®. In the – 0.4Al alloy, the Al content was reduced to 3.6 pct, a reduction of 10 pct of the overall Al content. Finally, + 0.9W included 0.9 at. pct W, a composition corresponding to ~ 3 wt pct. In this alloy, Fe was removed in order to avoid complications arising between these two additions. Compositions were measured by SEM–EDX, operated at a voltage of 10 kV. For each alloy, two large area scans over approximately 11mm² were taken for 500 s each, the averaged data can be found in Table II. Within the uncertainties that arise from EDX bulk compositional analysis, there is generally a good agreement with the nominal compositions.

All six alloys were prepared for heat treatment by encapsulation in quartz ampoules under vacuum. Solution heat treatments to improve microstructural homogeneity and remove solidification-induced microsegregation were performed at 1200 °C for 24 hours. Subsequent heat treatments were performed at 800 °C for 10 and 100 hours to induce precipitation and coarsening of the dual-superlattice γ′ and γ″ microstructures.

The effect of the different alloying additions on the γ′ solvus temperature was determined through differential scanning calorimetry (DSC). DSC thermograms were collected from samples after 100 hours of heat treatment at 800 °C, measuring 5 mm in diameter, and 1 mm thick, using a Netzsch 404 high temperature calorimeter. The heat flux was measured during heating at 10 °C min⁻¹ between 600 and 1400 °C. The tests were performed under flowing Ar to minimize sample oxidation. The resulting data were analyzed according to the NIST guidelines^[27] to determine the γ′ solvus temperatures.

X-ray diffraction (XRD) was performed using a Bruker D8 Advance Gen.9 θ–2θ diffractometer fitted with a LynxEye EX position sensitive detector (PSD). Diffraction data were acquired using Cu-Kα radiation, generated using a 40 kV accelerating voltage, a 40 mA current, and a 2θ collection range of 20°–100°, with an increment of 0.02° and a dwell time of 7 seconds. Diffractograms were analyzed using CrystalDiffract software by comparing them to reference spectra available through the Inorganic Chemical Structures Database (© FIZ Karlsruhe).

Hardness measurements were carried out on a Wilson® VH1202 micro hardness tester using a load of 2 kg and dwell time of 15 seconds. For each sample, 21 measurements were taken from various points spread over the full sample area and the mean values and standard deviations were calculated.

Samples were prepared for Atom Probe Tomography (APT) through cutting and electropolishing of matchsticks, performed using first a solution of 15 vol. pct perchloric acid in acetic acid and a 16–20 V DC voltage, then sharpened using a 2 vol. pct perchloric acid in butoxyethanol at 10 V DC voltage.^[28] Samples were analyzed in a Cameca LEAP 5000 XR atom probe, at 50 K, with a pulse frequency of 200 kHz, a detection rate of 1.5 pct, and a pulse fraction of 0.2 for specimens run in voltage mode. Reconstructions were performed using the Cameca Integrated Visualization and Analysis Software (IVAS) 3.6.12 and AP Suite 6, using mainly standard voltage curve reconstruction. Constant shank angle reconstructions were used and compared to the voltage curves. These were only employed in the event the voltage curve reconstruction visibly distorted precipitates due to artifacts (e.g., to mitigate the effect of potential small sample fractures during analysis). Iso-concentration surfaces were created at 7 at. pct Al and 13 to 15 at. pct Nb to identify the different phases present in the alloys. Where a peak overlap was observed (e.g., Cr–Al–Fe at 27 Da, Ni–Nb at 31 Da), it was labeled as the phase with only one isotope, in this case Al and Nb. The reason for this is while Cr, Fe, and Ni have other labeled primary isotope peaks, Al and Nb do not and therefore would not have been included in the reconstruction. Peak decomposition algorithms were then used to obtain phase composition values from isolated single-phase spectra, thereby reducing errors arising from peak overlaps.

Predictions of the phase fractions, phase compositions, and lattice parameters were completed using the Thermo-Calc 2019b software package, with the TCNi8 v8.2 Ni-alloy databases.^[29] All compositional predictions were obtained at an isothermal temperature of 800 °C. The simulations were run considering only the γ, γ′, and γ″ phases, consistent with the microstructural observations. Without phase selection, the equilibrium δ phase is predicted over the metastable γ″, in addition to several TCP phases. By excluding δ, the software is forced to calculate with the metastable γ″. The TCP phases were excluded for similar reasons—while they may be thermodynamically stable, their formation is typically kinetically inhibited over the short heat treatment durations studied here. Because of this, their exclusion is expected to increase the accuracy of the predictions, as TCP formation significantly depletes the matrix of alloying additions and affects the overall phase compositions and volume fractions.

DSC thermograms from all the alloys revealing the γ′ solvus temperature are shown in Figure 1. Full DSC thermograms taken from 200 to 1400 °C are found in the appendix, Figure A1. Both precipitate phases γ′ and γ″ strengthen the alloy, therefore knowledge of their solvus is important to understand the limits in operational temperature of the alloys. In this work, we have focused on understanding the effects on γ′ phase stability, which is particularly useful for high temperature applications. The stability of the γ″ phase will form part of a future work. Some additional phase changes are observed through kinks in the thermograms both below the γ′ solvus temperature and for the − 0.6Nb alloy also above the γ′ solvus temperature. These events are unlikely to be melting events as no intrinsic melting was observed for any alloys after heat treatment of up to 1200 °C. Further investigation as to the exact nature of the events is beyond the scope of this work. The measured γ′ solvus temperature is summarized and compared to thermodynamic predictions in Figure 2 (values also given in appendix Table A-I). The addition of 1.8 at. pct Mo did not significantly alter the γ′ solvus temperature as it led to a reduction of only 1 °C from 1010 to 1009 °C, which lies within the experimental error of the DSC. The reduction of Nb content to 5.4 at. pct reduced it by further 4 °C to 1005 °C and the addition of 9 at. pct Fe resulted in a decrease in γ′ solvus temperature of 17 °C from 1005 to 988 °C. A further 18 °C reduction to 970 °C was observed for the − 0.4Al alloy, which has an Al content of 3.6 at. pct, 10 pct lower than the 4 at. pct of the previous alloys. Al and Nb hence increase the γ solvus temperature, whereas Mo and Fe reduce it. Lastly, the + 0.9W alloy exhibited a γ′ solvus temperature of 983 °C. It is not as straightforward to determine the effect of W on the solvus temperature as it was for the other alloying additions. This is because the + 0.9W alloy differs in composition of at least two elements compared with any of the other investigated alloys. The + 0.9W alloy is best compared to either the − 0.4Al alloy with a lower γ′ solvus temperature of 970 °C or to the − 0.6Nb alloy with a higher γ′ solvus temperature of 1005 °C.

First, a comparison between the − 0.4Al alloy and the + 0.9W alloy is undertaken. The difference between these two alloys is a removal of the 9 at. pct Fe together with the addition of 0.9 at. pct W. As previously stated, the addition of Fe resulted in a reduction of the solvus temperature. Hence, it is not surprising that the γ′ solvus temperature increases as Fe is removed in the + 0.9W alloy. However, the total difference in solvus temperature from the − 0.4Al to the + 0.9W alloy amounts to only 13 °C, which is substantially less than what was observed for the addition of 9 at. pct Fe to the base alloy (17 °C). Hence, from the comparison of the − 0.4Al alloy and the + 0.9W alloy, it can be suggested that W reduces the γ′ solvus temperature.

The difference between the − 0.6Nb alloy and the + 0.9W alloy is a decrease of the Al content from 4 to 3.6 at. pct alongside the addition of 0.9 at. pct W. A comparison of these alloys reveals a strong reduction in γ′ solvus temperature of 22 °C. Comparing the base alloy and the – 0.4Al alloy, it was found that the reduction of Al led to a reduction of the solvus temperature by 18 °C. Again, the conclusion can be drawn that W led to a further reduction of the γ′ solvus temperature. For both comparisons, the excess reduction attributed to the addition of 0.9 at. pct W amounts to 4 °C.

The results collectively indicate that the γ′ solvus temperature is reduced by Mo, Fe, and W additions, among which W has the relatively strongest effect. However, more pronounced effects are observed for the alloying elements Nb and Al, which both increase the γ′ solvus temperature, with Al being by far the most potent element of all investigated elements. The potencies of these elements are summarized in Table III. The thermodynamic predictions from Thermo-Calc capture the overall trend of γ′ solvus temperature, particularly the correlation with Al content. However, the predicted effects of Mo and W do not appear to correlate with the experimental observations. Across the entire series of alloys, it is noted that there is an offset of approximately 70 °C between the Thermo-Calc predictions and the experimental values.

A homogenizing heat treatment was chosen for all alloys at 1200 °C in order to dissolve all precipitates, including γ′ and γ″, that may have formed during fabrication. This temperature was selected based on the DSC data in Figures 1 and A1 which show the full DSC thermograms up to 1400 °C, thus including the solidus temperature of the alloys. The homogenization temperature was selected to be above all solvus events and below the solidus for each alloy. Then, samples were heat treated at 800 °C to induce the controlled levels of precipitation. Figure 3 shows the X-ray diffractograms produced after 100 hours at 800 °C. The base alloy was previously shown to precipitate primarily a combination of γ′, γ″, and δ phases, depending on the heat treatment.^[14] After up to 1000 hours at 750 °C, only γ, γ′, and γ″ were observed by synchrotron X-ray diffraction, while at 800 °C δ precipitation began to occur when the heat treatment exceeded 100 hours in duration.^[14]

In the present case, X-ray diffractograms, shown in Figure 3, confirmed that all alloys contained γ, γ′, and γ″ phases. Additional phases are confirmed by XRD in most alloys: the δ phase and another phase that is likely niobium carbide (NbC). The reduction of Nb may have reduced δ and NbC precipitation, whereas the reduction in Al and the addition of W promoted the precipitation of these phases. Analysis of the peak location allows to extract lattice parameters of the phases. As the diffraction peaks of the γ and γ′ phase overlap substantially, for most alloys only one lattice parameter for both the γ and γ′ phase was extracted as is presented in Table IV. For the Base and the − 0.4Al alloy, a single value for the lattice parameter was insufficient to match both peaks. It may be that the lower extracted value represents the γ lattice parameter and that the larger extracted lattice parameter belongs to the γ′ phase. Thermo-Calc predicted lattice parameters for the γ and γ′ phases were calculated using molar volumes from the following equation:where \({V}_{m}\) is the predicted molar volume and \({N}_{A}\) is Avogadro’s number.

The misfit was calculated according to the following expression:

The predicted data are also presented in Table IV.

Although there is an offset between the experimentally extracted and predicted lattice parameters, where the predicted lattice parameters are consistently larger, the trends in lattice parameter changes roughly coincide. Additions of Mo are found to increase the (γ) lattice parameter. The reduction of Nb reduces the lattice parameter, and the addition of Fe increases the lattice parameter where Thermo-Calc predicts this change to occur mostly in the γ′ phase. Thermo-Calc predicts an increase in γ and decrease in γ′ lattice parameter for the removal of Fe and addition of W in the 0.9W alloy, which coincides with the data extracted from XRD. The lattice misfit as predicted by Thermo-Calc effectively decreases for both the addition of Mo and W.

The measured hardness of the alloys is given in Figure 4. The most prominent effect is found for the reduction of Nb by 0.6 at. pct, which led to a reduction in the hardness of over 40 HV. In contrast, the addition of 9 at. pct Fe as well as the reduction of Al by 0.4 at. pct did not significantly affect the alloy hardness in comparison to their respective previous alloy. Lastly, the addition of the solid solution strengtheners Mo and W seem to have had a positive effect on the alloy hardness, although the relatively large standard deviations reduce the clarity of this observation.

SEM micrographs of the alloys following heat treatment at 800 °C for 10 hours and for 100 hours are presented in Figures 5(a) and (b), respectively. The micrographs were taken from within the grains, away from grain boundaries. The δ phase is not seen in the presented micrographs, as it was found only in small quantities, localized at grain boundaries. NbC particles were also found only in small quantities. The formation of the δ phase with respect to alloying alterations is beyond the scope of the present study and will be part of a future work.

As shown in Figure 5, the 10 hours of heat treatment yielded extremely fine nano-structured microstructures. Heat treating for 100 hours coarsened the precipitates and allowed the effect of individual alloying additions on the precipitate morphology and fraction to become clearer. The micrograph of the base alloy shows the precipitation of two distinct precipitate phases: cuboidal γ′ precipitates with rounded corners and thinner, lenticular γ″ precipitates. The two types of precipitates appear in a dual-superlattice and are not connected to each other. Some previous studies on coprecipitation of γ′ and γ″ had identified a different, compact morphology where the γ″ phase precipitated on the faces of the cuboidal γ′ precipitates and thus helping to reduce γ′ coarsening.^[4,30,31] The differences in precipitate morphology between those previous studies revealing a ‘compact coprecipitation' morphology and the current work revealing a precipitate dual-superlattice is likely due to differences in composition and processing, i.e., heat treatments. Besides the Base alloy, the precipitate dual-superlattice can also be observed in all the additional alloys, though slight changes in their morphology and distribution are found for each compositional variation.

The addition of 1.8 at. pct Mo had a refining effect on the microstructure and the 0.6 at. pct reduction of Nb seems to have led to a reduction in the precipitate phase fraction, both of which can be observed after 10 hours and even more so after 100 hours of heat treatment. Alloying with 9 at. pct Fe led to substantial coarsening of the microstructure as compared to the − 0.6Nb alloy. The formerly cuboidal precipitates are now more irregularly shaped. The clear morphology difference between the γ′ and the γ″ precipitates is reduced as these phases seem to have precipitated adjacent to each other instead of being separated from one another by the matrix phase. With the reduction of the Al content in the − 0.4Al alloy, the precipitate phase fraction further decreased and the precipitate phases were still clustered together rather than separated by the matrix. Finally, the removal of Fe and the addition of W in the + 0.9W alloy restored the more refined microstructure similar to the + 1.8Mo alloy, with more distinct precipitate phases than in the Fe-containing alloys. The exact phase fractions of γ′ and γ″ of the alloys cannot be calculated from the SEM micrographs due to the similarity in contrast between the precipitates.

Atom probe tomography was utilized to measure accurate chemical compositions and assess the effect of individual alloying additions on partitioning behavior in the γ′ and γ″ phases. The samples heat treated for 10 hours were chosen for this study because the finer microstructure would allow detection of multiple phases within each atom probe needle. In comparison to the SEM analysis, APT allows the unambiguous identification of the precipitate phases by their compositional differences.

An atomistic scale reconstruction of the APT sample of the base alloy is shown in Figure 6(a) in which the γ, γ′, and γ″ phases may be separately distinguished. Due to the complex shapes and number of phases present, proximity histograms could not be utilized to analyze their interfaces. Instead, regions of interest (50 × 20 × 20 nm) were selected and located perpendicular to the interface to be analyzed. This method, which provides 1-D composition profiles, is very effective for measuring chemical compositions at the interfaces, although if the selected area is not perfectly parallel to the interface, its width can be artificially inflated. Figure 6(b) shows a 1-D composition profile going through all the phases present. The solid solubility of Nb in the γ′ phase is substantially higher than that of Al in γ″, which is consistent with the previous semi-quantitative EDX observations.^[13]

The alloy with 1.8 at pct Mo was also analyzed by atom probe tomography and a reconstruction is shown in Figure 7(a), where the cuboidal precipitate morphology of the γ′ precipitates can be observed.

One-dimensional composition profiles across the γ–γ′–γ″ phase interfaces are shown in Figures 7(b) through (d). The added Mo is predominantly accommodated within the γ phase as well as in the γ″ phase but is absent in the v phase. Figure 7(d) reveals a depletion in Ni at the γ′–γ″ interface. While the interface between the γ and the γ′ phase is clear, the interface between the γ and γ″ phases as well as between the γ′ and γ″ phases (Figures 7(c) and (d)) are diffuse, which can be observed through the “serrations” in the iso-concentration surfaces shown in Figure 7(a). The serrations are likely to be an artifact of the data binning algorithm.

The 3D reconstruction of the − 0.6Nb APT specimen is shown in Figure 8(a). The lenticular γ″ shape is clearly seen in this specimen. It is also noteworthy that the γ″ precipitate has grown around a γ′ precipitate, indicating that the γ′ precipitates formed first. 1-D composition profiles through all three phases are presented in Figures 8(b) through (d). No apparent differences to the previous alloy are observed from the reduction of Nb, with the only exception that the γ′–γ″ interface is slightly enriched in Ni, whereas the same interface was depleted in Ni for the previous alloy. The alloying alterations may have caused a change in interfacial energy of the precipitate phase interfaces, leading to decoration or expulsion of Ni at the interface.

The APT reconstruction of the + 9Fe alloy is shown in Figure 9(a). Based on the SEM analysis shown in Figure 5, the presence of Fe coarsened the microstructure and led to more spherical than cuboidal γ′ phase precipitates which have grown into one another as well as into the γ″ phase. The spherical shape of the γ′ precipitates and their merging are also observable in the APT reconstruction, as can be seen for the irregular precipitate in the back of the reconstruction, likely having merged from two or three γ′ phase precipitates.

Inspection of the interfacial composition profiles, shown in Figures 9(b) through (d), indicate that Fe segregated preferentially to the γ phase, although low levels were found in solid solution both in the γ′ and in the γ″ phases. In contrast to the + 1.8Mo alloy but similarly to the − 0.6Nb alloy, there is an excess Ni segregation at the γ′–γ″ interface.

Figure 10(a) shows an APT reconstruction of the − 0.4Al alloy. The lenticular nature of the γ″ phase is, again, clearly seen and the γ′ phase is found to sit on the side of the γ″ phase in accordance with the observations made from the SEM micrographs. One γ′ precipitate is found sitting partially within the γ″ phase. This is an observation that the SEM had not been able to provide due to the lack of contrast between the precipitate phases.

Figures 10(b) through (d) show 1-dimensional composition profiles across phase interfaces. There are no apparent differences in the phase compositions and interface segregation of this alloy compared to the + 9Fe alloy.

Figure 11(a) shows an APT reconstruction of the alloy with 0.9 at pct W. The + 0.9W alloy does not contain Fe and exhibited phase compositions comparable to the − 0.6Nb alloy apart from low levels of W in each phase. W partitioned preferentially to the γ″ phase (Figure 11(d)), although low levels (< 0.5 at. pct) were detected within the γ and γ′ phases as well (Figures 11(b) and (c)). In contrast to the previous alloys, the γ′–γ″ interface exhibits neither a depletion nor an enrichment of Ni.

For accurate elemental compositions of the γ, γ′, and γ″ phases, peak decomposition analysis was performed on the atom probe tomography data. This analysis yields more precise phase compositions than the 1-D elemental composition profiles, which are used to analyze the interfaces. The decomposition analysis was performed on spectra from isolated single-phase volumes, which were delineated with the iso-concentration surfaces shown in Figure 11, with the interfacial region excluded due to its data not being representative. This was done by exporting the individual phases as a separate file, then selecting a central region of interest (ROI) within them. The peak decomposed APT compositions are shown in Figures 12(a) through (c), respectively (values also given in appendix Tables A-II, A-III and A-IV). From this, conclusions can be drawn on the effects of the alloying additions on the microsegregation of the elements. The alloying additions changed the composition as follows:

Figure 12 also shows the phase compositions as predicted by Thermo-Calc at 800 °C. For the γ matrix, the predicted data matched the experimental data well with only slight deviations, such as the predicted Ni content being slightly above the measured data but following the trends and the Cr prediction being slightly below the measured content. While the prediction of Ni content in the γ′ phase matches the experimental data very well, larger deviations are found for the two precipitate forming elements Al and Nb. It is worth noting that this is likely related to the observed temperature offset in the Thermo-Calc predictions, as has been observed in other studies.^[32‐34] The comparison of the γ′ solvus temperature as measured by DSC and as predicted by Thermo-Calc (Figure 2) revealed an offset of approximately − 70 °C. Thermo-Calc predictions of the phase compositions calculated at an isothermal temperature of 700 °C, i.e., at a temperature which compensates somewhat for the observed offset, are in very good agreement to the measured compositions for the γ and γ′ phase, see appendix Figure A2. For both the compositions calculated at 800 °C shown in Figure 12 and those calculated at 700 °C, the Thermo-Calc predictions for the γ″ phase showed large discrepancies from the experimentally obtained data. The greatest deviation was found for Nb, followed by Cr, which is likely due to the difficulties in performing calculations involving metastable phases. The deviations may also be a reflection of the fidelity of the thermodynamic databases across this particular region of composition space.

In this study, we have assessed the effect of individual alloying additions of Mo, W, Fe and variations of Nb, and Al on the properties of a new dual-superlattice reinforced Ni-based superalloy. Thermal stability, hardness, elemental partitioning, microstructure, and phase fractions were investigated. The results presented lead to the following conclusions: