Microstructural Stability and Properties of New Nickel-Base Superalloys with Varying Aluminium: Niobium Ratio

New nickel-base superalloys with higher temperature capability are required for future, more ef ﬁ cient gas turbine engines. In designing such alloys, careful consideration is required of the elemental concentrations to ensure a suitable balance of properties is obtained. Herein, the phase equilibria and microstructural stability of new nickel-base superalloys with varying Al:Nb ratio are assessed via long-term thermal exposures at 700 °C. The alloys are analyzed using scanning and transmission electron microscopy, X-ray diffraction, differential scanning calorimetry


Introduction
Efforts to improve efficiency and meet ambitious legislative targets are motivating significant research to develop new alloys for high-temperature service. [1][2][3][4][5][6][7] Of particular interest are the polycrystalline Ni-base superalloys that are used extensively in aerospace propulsion. One of the most widely used polycrystalline superalloys is Inconel 718 (IN718), which is primarily strengthened by a dispersion of ordered γ 00 (Ni 3 Nb, D0 22 , tetragonal, I4=mmm) precipitates, with a small (%3%) volume fraction of γ 0 (Ni 3 Al, L1 2 , cubic, Pm3m). [8,9] While γ 00 has a potent strengthening effect on a per volume fraction basis due to the large coherency strains, it is metastable, and ultimately transforms to the thermodynamically stable δ phase (Ni 3 Nb, D0 a , orthorhombic, Pmmn) after long exposures at above 650°C. [10][11][12] This is also associated with a concomitant loss of mechanical properties. [13] At above 750°C, δ precipitation is observed after exposures of less than 10 h. [14] At typical service temperatures at or below 650°C, coarsening and dissolution of the strengthening precipitate phases is the primary factor resulting in a loss of tensile and creep properties. [15] Early attempts at enhancing the thermal stability of IN718-type alloys focused on improving the stability of the γ 00 precipitates. By modifying the Al:Nb ratio, different precipitate morphologies were obtained, for example, in the work by Cozar and Pineau, [16] where a "compact morphology" was produced. In these alloys, the γ 00 precipitates nucleate on the surface of the γ 0 , minimizing elastic distortions, acting as a diffusion barrier to limit precipitate coarsening, thereby enhancing their stability.
More recently, new alloys such as 718Plus [6,[17][18][19] and VDM780 [5,[20][21][22][23][24] have been developed to supersede IN718. These alloys rely primarily on strengthening by γ 0 precipitates and offer an increase in temperature capability of approximately 50°C, while avoiding the loss in processability associated with the more highly alloyed polycrystalline Ni-base superalloys. [25,26] A key feature of these alloys is the careful control of the Al:Nb ratio to promote γ 0 formation. In these alloys, Nb also provides potent solid solution strengthening, although this mechanism is limited by a relatively low solubility in the γ matrix. [27,28] In higher concentrations, Nb partitions to the γ 0 phase, where it substitutes for Al and increases the antiphase boundary (APB) energy. [29] However, Nb has also been linked to the formation of grain boundary η (where it substitutes for Ti), as well as the topologically close packed (TCP) σ phase. [19,30] These studies highlight the importance of understanding the combined roles of Al and Nb on the performance of superalloys of this type.
In this work, the microstructure and long-term thermal stability of a series of polycrystalline Ni-base superalloys with varying Al:Nb ratio have been studied. The alloys are based on the quaternary Ni-Cr-Al-Nb alloys first investigated by Mignanelli et al., [31] and incorporate multiple alloying additions including Mo and W for solid solution strengthening, Fe for processability, Co to provide a decreased stacking fault energy, and C, B, Zr for grain boundary strengthening. Critically, previous research has highlighted the sensitivity of the morphology and distribution of the γ 0 and γ 00 phases to the Al:Nb ratio as well as single elemental additions. [32] The effects of Al:Nb ratio in the presence of multiple alloying additions must therefore be understood if commercially useful alloys are to be realized.

Experimental Section
Ingots of Alloys 1-3, with nominal compositions given in Table 1, were produced by vacuum arc melting from their constituent elements with purity ≥99.9%. These compositions fall within the range of United States Patent US10287654. [33] The ingots were homogenized for 96 h at 1080°C, and subsequently hot rolled at the same temperature to a reduction of 55%. Ageing of the alloys was achieved through a dual-step heat treatment comprising 750°C for 8 h, followed by 650°C for 8 h, with furnace cooling between stages and final air cooling. The homogenized, rolled, and aged samples will henceforth be referred to as the "standard" condition. Samples of each alloy were cut using a precision saw and encapsulated in argon backfilled quartz ampoules to minimize oxidation during the heat treatment. For microstructural stability studies, the samples were subsequently exposed at 700°C for 1000 h in a box furnace. To aid the identification of the δ solvus in Alloy 1, additional heat treatments were performed on samples in the standard condition for 100 h at 900, 950, 975, and 1000°C. Furnace temperatures were calibrated to AE1°C using an N-type thermocouple.
After thermal exposure, the samples were mounted in conductive phenolic resin, ground using successively finer SiC papers to a 3 μm finish, followed by diamond polishing to a 0.25 μm surface finish. Final chemical polishing was conducted using colloidal silica (oxide polishing solution) to a 0.04 μm finish. Electrolytic etching was performed at 3 V using a solution of 10% orthophosphoric acid, for approximately 2-3 s.
Scanning electron microscopy (SEM) was performed using a Zeiss GeminiSEM 300, operated between 3-15 kV, with images acquired using secondary electrons (SE) and a 30 μm aperture.
Samples were prepared for transmission electron microscopy (TEM) by electro-discharge machining (EDM) 3 mm diameter discs from 0.2 mm thick slices. These discs were then electropolished using a Struers TenuPol twin-jet electropolishing unit operated at 15 V with a 6% perchloric acid (HClO 4 ) in methanol electrolyte at À5°C. Diffraction patterns were obtained on film using a JEOL-JEM 200 CX TEM operated at 200 kV. Scanning TEM (STEM) was performed on an FEI Tecnai Osiris FEG-TEM also operated at 200 kV, using bright-field (BF) and high-angle annular dark-field (HAADF) detectors. Scanning TEM energy-dispersive X-ray spectroscopy (STEM-EDX) data were acquired on the same instrument, using the FEI Super-X system and Bruker silicon drift EDX detectors.
Samples for differential scanning calorimetry (DSC) measuring 5 mm in diameter by 1 mm thick were produced by EDM. DSC thermograms were obtained using a Netzsch 404 F1 Pegasus DSC between 50 and 1450°C, with a 10°C min À1 heating and cooling rate, under flowing argon at 50 mL min À1 .
X-ray diffraction (XRD) data were acquired with CuK α radiation and a 0.012 mm Ni filter, using a Bruker D8 ADVANCE diffractometer and a LynxEye XE position-sensitive detector. Scans were run using a variable slit width, a constant sample illumination of 6 mm, a time step of 1 s, and angular range between 30°a nd 110°2θ with a 2θ increment of 0.015°. During data acquisition, the samples were rotated to minimize any textural effects on peak intensities. The X-ray data were analyzed via full pattern refinement using the Pawley method in GSAS-II. [34] Vickers hardness data were obtained with a Qness Q30 Aþ automatic hardness tester, using a 20 kg load, and are given as the average of 10 indentations, with error bars representing the standard deviation.
Thermodynamic equilibrium predictions were performed using the Thermo-Calc software with the TTNi8 Ni-superalloy database (version 8.2). [35] The alloy compositions were run without any of the grain boundary-strengthening elements (C, B, or Zr) as these were found to produce instabilities in the calculations of phase fractions. The thermodynamically stable δ phase was deactivated for all calculations involving the metastable γ 00 phase.

Compositional Analysis
The alloy compositions were experimentally determined by inductively coupled plasma optical emission spectroscopy, and are shown in Table 2. These measurements were performed at IncoTest UK, Special Metals Wiggin Ltd. The measured compositions were deemed suitable for the purposes of this study due to the similarity of the measured elemental concentrations to the nominal values.

Microscopy
Using low-magnification SEM imaging, the modal grain sizes of Alloys 1-3 were determined to be 43, 50, and 32 μm, respectively (see Supporting Information). Higher-magnification SEM imaging was used to reveal the precipitate distributions in the alloys in the standard condition and after exposure at 700°C for 1000 h, Figure 1. Ultrafine (<10 nm) dispersions of precipitates can be observed in the samples in the standard condition, with an increase in precipitate size observed after the long-term exposure.
In these samples, the pseudo-cuboidal morphology of the precipitates was much easier to resolve due to their increased size. No bulk precipitation (e.g., TCP or other deleterious phases such as δ) was detected in any of the alloys. However, in the sample of Alloy 1 following long-term exposure, a small volume fraction of elongated precipitates was observed, appearing on the surface of the pseudo-cuboidal precipitates. No such precipitates were observed in any of the other alloys. These elongated precipitates were investigated further using TEM, with a diffraction pattern acquired down [001] γ as shown in Figure 2, alongside STEM BF, HAADF images, and STEM-EDX elemental concentration maps taken from the same area.
The STEM images show evidence of the elongated precipitates forming on the faces of the rounded, pseudo-cuboidal precipitates. Reflections in the diffraction pattern consistent with a disordered γ matrix, alongside both γ 0 and γ 00 superlattice precipitates, were detected and are indicated with markers in Figure 2. The pseudocuboidal precipitates were enriched in Ni, Al, and Nb (and W to a lesser extent) and depleted in Co, Fe, Cr, and Mo, consistent with a Ni 3 Al γ 0 phase. The elongated precipitates were depleted in Al, with a notable enrichment in Nb greater than that observed for the pseudo-cuboidal precipitates, consistent with the γ 00 phase. [36] 3.3. XRD XRD patterns acquired from Alloys 1-3 in the standard condition are shown in Figure 3, where the fundamental reflections corresponding to the disordered matrix can be seen. Peak splitting due to the presence of CuK α1 and CuK α2 radiation was visible, and is more prominent at higher 2θ angles. Fundamental reflections for the γ 0 precipitates were visible as the asymmetry in peak shape toward the lower values of 2θ (left hand side) of the matrix peaks. The broad peak shape and reduced intensity are consistent with crystallite size broadening as a result of their dimensions and lower volume fraction, respectively. The superlattice reflections are known to be very weak using X-ray sources [37] and were not observed in these data, in line with in other XRD studies of Ni-base superalloys. [38,39] An example of the peak decomposition obtained from the Pawley refinement is shown in Figure 4, with extracted lattice parameters given in Table 3. The precipitates have a larger lattice parameter than the matrix, consistent with the peaks being at lower values of the diffraction angle, 2θ, as have also been observed for other alloys of similar composition. [32] The lattice parameters of the strengthening precipitates were not found to vary between the three alloys (at least not reliably resolvable using laboratory-based methods); whilst a very small increase in the Γ lattice parameter was detected. The results were consistent with predictions of molar volume obtained using    www.advancedsciencenews.com www.aem-journal.com Thermo-Calc that account for elemental partitioning between the two phases. These predictions captured the correct trend, albeit with a larger separation between the lattice parameters of the two phases. This has been attributed in other works to the prediction of unconstrained molar volumes from Thermo-Calc, which will naturally show a larger separation than the constrained values measured experimentally using XRD. [40]

DSC and Solvi Investigation
DSC thermograms for Alloys 1-3 in the standard condition are shown in Figure 5a, with the locations of the γ 0 solvi marked by the dashed region. The measured γ 0 solvus temperature for each alloy increased across the series (with increasing Al:Nb ratio), with the same trend predicted by Thermo-Calc as shown in Figure 5b. There was an offset of approximately 20°C between the experimentally determined γ 0 solvi and those predicted computationally, which is consistent with other reports in the literature from studies of nickel-base superalloys. [41] In all three of the alloys studied, thermodynamic calculations predicted the δ phase to be stable. When the δ phase was excluded from the phases considered in the calculations, the metastable γ 00 phase was predicted to occur instead. The predicted volume fractions of the γ, γ', and δ phases obtained assuming thermodynamic equilibrium are summarized in Table 4, along with the γ, γ 0 , and γ 00 phases when the δ phase is suppressed. In both sets of predictions, increasing the Al:Nb ratio resulted in a decrease in the expected fraction of both the δ and γ 00 phases. However, it is noted that evidence of γ 00 precipitation was only observed experimentally in Alloy 1 after a long-term thermal exposure at 700°C for 1000 h. Figure 5b also presents the predicted solvus temperatures for the δ and γ 00 phases, both of which decreased with increasing   www.advancedsciencenews.com www.aem-journal.com Al:Nb ratio. The δ solvi decreased from 1033°C in Alloy 1 to 1005°C in Alloy 3, while the γ 00 solvi varied from 941°C in Alloy 1 to 907°C in Alloy 3. As no δ-phase precipitation was visible in the alloys in the standard condition, or after long-term thermal exposure at 700°C for 1000 h, further heat treatments were performed to assess the microstructural stability and propensity for δ-phase formation. For this study, Alloy 1 was selected as it possessed the largest difference between the measured γ 0 solvus and the predicted δ solvus, and therefore the greatest potential for unambiguous determination of the δ solvus free from any effects of the γ 0 precipitates. Samples of Alloy 1 were exposed for 100 h at 900, 950, 975, and 1000°C and prepared for metallographic examination using the protocol described in Experimental Section. SE micrographs of the alloys following these thermal exposures are presented in Figure 6. Significant intragranular precipitation of an acicular phase was observed in the sample exposed at 900°C. The volume fraction of the acicular phase decreased in the sample exposed at 950°C, while none was observed after heat treatment at 975 or 1000°C. This suggests that the acicular phase has a solvus temperature in Alloy 1 between 950 and 975°C. This is approximately 70°C lower than the value predicted by Thermo-Calc. DSC thermograms obtained from Alloy 1 following exposure at each of the four temperatures were compared to material in the standard condition; however, no unambiguous peaks in the region of the expected solvus could be identified. The reader is directed to the Supporting Information for the DSC data.
To confirm the identity of the acicular precipitates, XRD was performed on the sample of Alloy 1 exposed at 900°C for 100 h, as this sample contained the largest volume fraction of the precipitate phase, Figure 7. Peaks consistent with the γ and γ 0  phases can be observed, as detected in the standard condition. Additional peaks attributable to the δ phase were also present, and are indicated with markers in the diffraction pattern.

Hardness
A preliminary investigation of the mechanical properties of these alloys was performed using Vickers indentation as a proxy for mechanical strength. The hardness of the alloys was assessed in both the standard condition, as well as after long-term thermal exposure, with the results presented in Figure 8. A pronounced hardening response was observed following long-term exposure; however, the magnitude of the response decreased with increasing Al:Nb ratio. Alloy 1 demonstrated a hardening response of %50 HV, compared to %25 HV in Alloy 3.

Microstructure
While these alloys were based on the dual-superlattice system investigated by Mignanelli et al., [31] they are principally γ 0 forming, with the presence of ordered cubic precipitates confirmed using XRD and microstructural examination. The pseudocuboidal precipitate morphologies are consistent with alloys of similar composition, with lattice misfits <0.5%. [32] Notably, the dual-superlattice microstructure, consisting of appreciable volume fractions of γ 0 and γ 00 , was not replicated in this alloy series. This can be attributed to the higher Al:Nb ratio of these alloys (0.83 to 1.21) compared to the dual-superlattice alloy (0.67). However, a small volume fraction of γ 00 precipitates was detected in Alloy 1 after thermal exposure for 1000 h at 700°C. The occurrence of these precipitates on the faces of the γ 0 is similar to the "compact morphology" reported by Cozar & Pineau [16] and others, [42,43] but typically only on a single {100} γ 0 face, rather than multiple faces. Such morphologies have been explored by Phillips [44] and Shi, [45] and termed "single lobe," with the morphology formed being highly sensitive to composition, heattreatment time, temperature, and cooling rate. Nevertheless, provided there is a fine dispersion of γ 0 precipitates, coprecipitation of γ 00 was shown in most cases to limit the coarsening rate of the γ 0 by limiting the diffusion of γ 0 forming elements to the precipitates. Thermodynamic calculations predicted the occurrence of γ 00 in all three of the alloys studied. Despite this, precipitation of the γ 00 phase was only observed in Alloy 1, which had the lowest Al:Nb ratio, and highest predicted γ 00 volume fraction (0.09). While it is possible that longer duration exposure in the other alloys may have resulted in the precipitation of γ 00 , its occurrence cannot be assured as superalloys with similar Al:Nb ratio have  . Vickers hardness data for Alloys 1-3 in both the standard and long-term exposed conditions. www.advancedsciencenews.com www.aem-journal.com been shown to be solely γ/γ 0 forming. [32] Notably, the predicted γ 00 solvus in Alloy 1 of 941°C was approximately 40°C higher than the experimentally determined γ 0 solvus for the same alloy. This is significant, as the coprecipitation phenomena observed rely on the precipitation of γ 0 prior to γ 00 . Therefore, while thermodynamic calculations are useful in assessing relative phase stabilities as a function of composition, the absolute solvus temperatures should be treated with caution.
Higher-temperature exposures of Alloy 1 at 900°C for 100 h resulted in extensive precipitation of the δ phase. Critically, however, all of the alloys studied remained stable with respect to δ-phase formation after 1000 h exposures at the target service temperature of 700°C. These results compare favorably to IN718, which precipitates δ after approximately 50 h when exposed to the same temperature. [14] The predicted δ solvi decreased across the alloy series with increasing Al:Nb ratio, demonstrating similar behavior to those predicted for the metastable γ 00 . Considering the similar stoichiometry and nominal phase compositions, it is perhaps unsurprising that the effects of Al:Nb ratio on the solvi of both phases demonstrate similar trends. While the experimentally determined δ solvus was somewhat lower than the predicted value, there appears to be at least a 50°C window between the γ 0 and δ solvi in Alloy 1. Such a window could facilitate conventional deformation processes such as forging, where it may be possible to utilize a minor volume fraction of the δ phase to control grain growth in supersolvus forging operations. The good microstructural stability demonstrated by these alloys should also ensure that excessive δ precipitation, expected to be deleterious to the mechanical properties, is avoided during these operations. Further work is currently being undertaken to assess the response of these alloys to deformation processing.
The window between the γ 0 and δ solvi in Alloy 1 is comparable to that observed in the δ-forming, γ 0 strengthened alloy, ATI 718Plus (γ 0 = 960°C, δ = 998°C), which has superseded IN718 in several applications. [46] As the alloys with higher Al:Nb ratio are predicted to have lower δ solvus temperatures, smaller windows between the γ 0 and δ solvi would be expected. This makes these alloys somewhat less attractive as enhanced temperature control would be required for their thermomechanical processing. However, it is noted that the alloys with higher Al:Nb ratio may exhibit further improved microstructural stability and be less susceptible to δ-phase formation as a result of their reduced Nb concentrations.
The observed hardening response after long-term thermal exposure at 700°C indicates that the initial dual-aging heat treatment did not achieve the optimal mechanical properties, and that the alloys were underaged. Critically, the significant hardening of approximately 50 HV in Alloy 1, the alloy with the lowest Al:Nb ratio, may be attributed to the precipitation of γ 00 after the longterm thermal exposure. It is well reported in the literature that γ 00 has a potent strengthening effect on a normalized volume fraction basis. [47,48] However, it is noted that no evidence of γ 00 precipitation was detected in any of the alloys studied in the standard condition, nor was it seen after long-term thermal exposure in the alloys with higher Al:Nb ratio.
However, for all three alloys, the absolute values of hardness after long-term thermal exposure were comparable to IN718 after exposure at 700°C for 1000 h (%410 HV). [49,50] However, it is noted that the standard commercial age used for IN718 was not effective at optimizing the hardness in these alloys. It is likely that shorter duration heat treatments at higher temperatures may achieve comparable hardening to commercially aged IN718 (%460 HV), although further tests are required to determine the optimal aging conditions to achieve this. Furthermore, while superior mechanical properties may be obtained with sufficient aging to produce precipitates of the optimum size (at the weak-to strong-pair coupling transition), it is important to recognize that maximizing the strength in this way is unlikely to provide the appropriate balance of properties required for structural materials. For example, it has been reported in the literature that alloys with narrow monomodal precipitate dispersions at or close to the optimal precipitate size are likely to demonstrate very limited ductility and poor creep performance. [51][52][53] To better understand the origins of the variations in hardness between the alloys, the expected contributions to the critical resolved shear stress, Δτ CRSS , were examined. The difference in Δτ CRSS arising from grain size alone could not account for the trends in alloy hardness (see Supporting Information). An analysis was therefore performed to assess the strengthening contribution from the different dislocation-precipitate interaction modes, through either strong-or weak-coupled dislocation pairs. This analysis was performed following the approach summarized by Kozar et al. [54] utilizing Equations (1) and (2). (1))

Weak-Pair Coupling (Equation
where b is the burgers vector, d s is the precipitate size, φ is the precipitate volume fraction, Γ γ 0 is the γ 0 APB energy, T is the line tension, taken to be equal to Gb 2 2 for screw dislocations, and A is a geometric factor taken to be 0.72 for spherical precipitates. (2))

Strong-Pair Coupling (Equation
where G is the shear modulus of the material, and w describes the repulsion between a pair of dislocations, which can be approximated to 1, as discussed in the work by Hüther et al. [55] Calculations were performed with APB energies estimated using the Miodownik and Saunders method. [56] The fine scale of the precipitates meant that the precipitate size distribution and volume fraction were difficult to obtain experimentally (using etching techniques) without introducing significant stereological errors. [57,58] As a consequence of this, the precipitate volume fractions were estimated using Thermo-Calc based on the observed phases (γ, γ 0 , and γ 00 if present). The burgers vector b was assumed to be of the conventional a 2 < 110 > type, and the shear modulus G equal to that of a similar polycrystalline nickelbase superalloy, in this case 718Plus (64 GPa at 700°C). [59] All calculations were performed using a nominal precipitate size of 30 nm.
For all three alloys, weak coupling gave the lowest values of Δτ CRSS and, hence, is expected to be the controlling strengthening mechanism, with predicted Δτ CRSS of 115, 135, and 137 MPa for Alloys 1-3, respectively. The lower value of Δτ CRSS for Alloy 1 is a result of the lower predicted γ 0 volume fraction (due to the precipitation of γ 00 ) which reduces the strengthening contribution of the γ 0 precipitates. The similar values obtained for Alloys 2 and 3 arise from the similar values of predicted γ 0 volume fraction based on the thermodynamic calculations (0.269 and 0.271 for Alloys 2 and 3, respectively).
The optimum γ 0 precipitate diameter was calculated by solving the weak-and strong-pair coupling equations for the value d Ã s that gives the greatest Δτ CRSS for a given precipitate volume fraction. This was calculated as 34 nm for Alloy 1, and 39 nm for Alloys 2 and 3, which are all within the typically reported range for the similar polycrystalline Ni-base superalloy Udimet 720Li. [60] As Alloy 1 contained γ 00 precipitates, the strengthening contribution from this phase should also be considered. Such analyses have been performed by Oblak et al. [47,48] who assessed both the coherency strengthening and order hardening arising from a dispersion of tetragonally distorted particles. While they concluded the coherency strengthening was dominant, both contributions have been assessed in this work, with the relevant equations reproduced later. (3))

Coherency Strengthening from γ 00 (Equation
where ε is the tetragonal misfit, h is the precipitate half thickness, and R is the precipitate radius.

Order
Hardening from γ 00 (Equation (4)) where Γ γ 00 is the γ 00 APB energy, and β is a factor that represents the fraction of γ 00 particles in which order is restored after cutting, and is equal to 1 3 in the case that equivalent amounts of the three variants are present.
Using the equations, the contributions to Δτ CRSS for Alloy 1 are calculated to be 236 and 173 MPa from coherency strengthening and order hardening, respectively. These values are very large compared to the values predicted from the γ 0 and imply a much greater strength in Alloy 1 than observed experimentally. It is noted however that the calculation of coherency strengthening is highly sensitive to the tetragonal misfit, and the order hardening to the γ 00 APB energy. The values used were taken from refs. [47,48] where they were derived for IN718, which possesses a conventional dispersion of γ 00 . In contrast, Alloy 1 exhibits "single-lobe" coprecipitation. As such, the actual misfit is likely to be significantly lower, resulting in a smaller value of Δτ CRSS .
The γ 00 APB energy will also vary from that in IN718 due to the different bulk composition of Alloy 1.
These results highlight that further refinement of the strengthening models is required for their reliable application to these alloy systems, particularly in cases where coprecipitation of the strengthening species takes place. These morphologies result in lattice distortions that are likely to be very different from those encountered in conventional γ 00 strengthened alloys. However, from the data presented before, Alloy 1 arguably offers the most attractive balance of properties, having the largest difference in temperature between the γ 0 and δ solvi, as well as the most pronounced hardening response, as a result of the low Al:Nb ratio promoting the coprecipitation of γ 00 . These characteristics are expected to translate into good processability, while offering comparable alloy performance.

Conclusions
The microstructure and thermal stability of a series of new polycrystalline Ni-base superalloys with varying Al:Nb ratio were investigated, with comparisons made to commercially used alloys such as IN718 and 718Plus. The alloys were principally γ 0 forming, with γ 0 solvus temperatures determined using DSC, varying between 905 and 945°C. However, as the Al:Nb ratio was reduced, the propensity to form the metastable γ 00 phase increased, with a minor volume fraction detected in Alloy 1 (with the lowest Al:Nb ratio) after a long-term thermal exposure at 700°C.
Thermal exposures demonstrated that all of the alloys studied were resistant to δ-phase formation after exposure at 700°C for 1000 h, providing a significant increase in thermal stability compared to IN718 exposed at the same temperature. Metallographic assessment of the δ solvus in Alloy 1 indicated that a wide processing window, comparable to 718Plus, exists for this alloy. Such a window is expected to be beneficial in cast and wrought manufacturing routes, providing a cost advantage over powderprocessed alternatives.
A hardening response was confirmed across the entire alloy series after long-term thermal exposure, suggesting that the alloys were underaged after the initial heat treatment. In addition, different hardening responses were observed as a function of the Al:Nb ratio, with Alloy 1 (possessing the lowest Al:Nb ratio) demonstrating a pronounced hardening response, greater than that of the other alloys. This behavior was attributed to the potent strengthening contribution of the metastable γ 00 precipitates, which exceeds the hardening provided by γ 0 precipitates alone. All the alloys demonstrated comparable hardness to standard aged IN718, with the results showing that careful optimization of the heat-treatment schedule could lead to further improvements to the mechanical properties.
Modeling of the strengthening contributions of the strengthening precipitates highlighted that the conventional models for estimating the strengthening arising from γ 00 precipitates overestimated their contribution, and required refinement for alloys that exhibit coprecipitation phenomena.
The mechanical properties, combined with the good thermal stability demonstrated by these alloys, indicate that they are attractive candidates for development of low-cost, polycrystalline Ni-base superalloys. [61]