Oxidation Behaviour of New Nickel-Base Superalloys with Varying Aluminium: Niobium Ratio

New Ni-base superalloys with higher temperature capability are required for future, more efficient gas turbine engines. In designing such alloys, careful consideration is required of the elemental concentrations to ensure that a suitable balance of mechanical properties and environmental resistance is obtained. In this study, the oxidation resistance of a series of new Ni-base superalloys with varying Al:Nb ratio has been assessed through long-term exposures in air at elevated temperature. The oxide scale was characterised using scanning electron microscopy, energy-dispersive X-ray spectroscopy, X-ray diffraction and quantitative measurements of oxide scale dimensions. The alloys were found to form continuous chromia scales at 700 °C and outperformed several current commercial superalloys. However, following exposure at 800 °C, significant microstructural degradation was observed due to precipitation of the δ phase.


Introduction
Ni-base superalloys continue to meet the needs of many key structural components in the hot sections of gas turbine aeroengines. In these applications, the alloys must operate under demanding conditions, often close to the limit of their capabilities. The need to deliver future, more efficient engine designs is therefore motivating parallel development of new superalloys that offer improvements in temperature capability, strength, or service lives. One subset of superalloys receiving particular attention is the polycrystalline superalloys used for turbine discs, in part due to the increasing use of rapid climb rates or "Continuous Climb Operations" that can offer significant fuel savings to airline operators [1,2]. As a result of these rapid climb rates, peak engine temperatures are being maintained for longer periods. This requires the polycrystalline turbine discs to endure increased durations in oxidative environments at elevated temperature.
High Temperature Corrosion of Materials (2023) 99:241-266 Historically, the most widely used polycrystalline Ni-base superalloy has been IN718 [6,7]. This alloy derives its exceptional strength from dispersions of γ″ precipitates (≈ 18 vol.%). However, it is limited to service below 650 °C as the metastable γ″ transforms to the thermodynamically stable δ phase following prolonged exposure at higher temperatures [8,9]. Efforts to develop new superalloys that may supersede IN718 at comparable cost have focussed on alloys with lower Nb and higher Al contents, which favour the formation of γ′ precipitates, e.g. Allvac®718Plus™ [10], VDM®780P [11] and AD730® [12]. For further improvements to be obtained, the interrelations between the composition, processing, and microstructure on the final properties need to be understood. These findings may then be exploited in empirical alloy design or incorporated into computational frameworks for the identification of optimal alloys to meet specific service requirements.
Amongst the many properties required, it is important that superalloys resist surface degradation from the oxidising gases to which they are exposed in service. Typically, surface protection of polycrystalline Ni-base superalloys is achieved through Cr additions, which promote the formation of a passivating chromia (Cr 2 O 3 ) scale [13]. Selective oxidation of Cr to form a passivating scale will only occur above a critical Cr concentration N Cr , the value of which can be calculated by following the classical diffusional analyses as performed by Wagner [14][15][16]. As the value of N Cr is a function of kinetic parameters, it has a strong temperature dependence, leading to two important criteria that must be met if the alloy is to be considered protected against oxidative attack. Firstly, a minimum concentration of Cr, N (1) Cr can be identified that suppresses the ingress of oxygen and prevents internal oxidation, see Equation 1. Additionally, a further value N (2) Cr can be defined as that required to sustain scale growth once a continuous scale has been established, Equation 2. These criteria have been successfully employed in studies of the Fe-Cr system [17,18] as well as the Ni-Cr system [19,20]. In such studies, it has been shown that Cr concentrations in excess of 10 wt% are necessary to form a protective chromia scale at temperatures below 1000 °C. However, due to the temperature dependence of the critical Cr concentrations, it is noted that higher levels are usually required for oxidation protection at higher temperatures, or for repassivation in the event of scale damage or spallation.

Equation 1 (Minimum Solute Concentration For Transition to External Oxidation)
Where f is the critical fraction of internal oxide beyond which an external scale begins to form, commonly taken as 0.3 from the work by Rapp [21]. is the stoichiometric ratio of metal to oxygen in the oxide species (1.5 in the case of chromia and alumina), N (S) O is the oxygen solubility in the alloy and D O the diffusivity of oxygen in the alloy. D Cr−Alloy is the interdiffusion coefficient for Cr in the alloy, V m the alloy molar volume and V CrO the oxide molar volume.

Equation 2 (Minimum Solute Concentration for Scale Maintenance After a Continuous Scale Has Formed)
Where z a is the valence of the metallic element forming the oxide scale (i.e. 3 for chromia and alumina), 16 is the atomic mass of oxygen, and k p is the parabolic oxidation rate constant in g 2 cm −4 s −1 .
Surface oxidation of polycrystalline Ni-base superalloys has been shown to reduce fatigue and creep lifetimes as a result of crack initiation at the brittle oxide scales, or at oxidised microstructural features such as carbides or borides [22,23]. Similarly, oxidation is associated with an increased propensity for intergranular cracking [24][25][26][27][28][29][30][31]. Under dwell fatigue conditions, longer dwell times have been shown to significantly reduce the number of cycles to failure, which has been attributed to environmental attack at the crack tip during the dwell period. Higher dwell fatigue crack growth rates have also been reported when performing tests in air compared to vacuum. This has been ascribed to dynamic embrittlement as a result of oxidation damage [32,33].
Chromia scale formation is strongly affected by the concentrations of alloying additions. For example, Al additions have been shown to provide a synergistic effect, facilitating chromia formation at lower Cr concentrations than would otherwise be required [34][35][36]. However, additions below approximately 5 at.% typically do not confer any significant improvements in oxidation performance, instead resulting in the formation of discontinuous internal Al 2 O 3 precipitates [37,38] (Type II behaviour as categorised by Giggins and Petit [39]). The effect of minor Al additions on the oxidation resistance of Ni-Cr alloys is governed by a variety of factors, both thermodynamic and kinetic. These include: the relative diffusion rates [40,41], the chemical activities of the elements [42], and the solubility of oxygen in the sub-surface region [43,44]. Only if sufficient Al is added to the alloy such that it forms a continuous alumina scale will significant performance advantages be obtained, as shown in work on the wrought polycrystalline alloys Haynes 214 and 224 [45,46]. Whilst such alloys offer superior hightemperature oxidation resistance compared to most wrought superalloys, they are limited to low-stress applications. Polycrystalline superalloys that are utilised in high performance structural applications are typically highly alloyed in order to maximise a variety of strengthening mechanisms, and as such high aluminium contents are not usually viable due to the associated loss of workability that would result. As a consequence, alumina forming polycrystalline Ni-base superalloys tend to be limited to specialist applications, in contrast to the single-crystal superalloys used for turbine blade applications (where operating temperatures routinely exceed 1100˚C). These alloys rely on the formation of alumina scales for oxidation resistance. In alumina forming alloys, Cr additions have a significant beneficial effect on oxidation performance, lowering the partial pressure of oxygen, pO 2 , at the oxide-metal interface and supressing the oxidation of Ni [47][48][49]. They can also promote the nucleation of alumina precipitates, decreasing the time taken for a continuous protective scale to form [50,51]. In contrast, Ti additions have been implicated in accelerated oxidation kinetics [52]. These deleterious effects have been attributed to Ti 4+ ions doping the chromia scale, which increase the concentration of ionic defects and thereby the mass transport rates through the chromia scale [53,54]. The effects of the solid solution strengthening elements Mo and W have been shown to be largely detrimental to oxidation performance [55,56], although, some contrasting effects have been reported depending on the exact alloy composition. For example, Smialek et al. [57] found that moderate Mo additions in alloys containing sufficient Cr levels had minimal effects on oxidation kinetics.
Whilst Nb is an important alloying addition for strengthening, its influence on oxidation performance is complex, with a range of effects reported across different alloy classes with varying Nb contents. In some austenitic stainless steels, Nb additions have been reported to have no notable effect on the oxidation performance [58], whereas a so-called 'high-entropy superalloy' with additions of 0.9 at.% Nb was reported to have improved oxidation resistance at 900 °C, due to the formation of a Nb-rich layer between the chromia and alumina scales [59]. Notably, studies of the Ni-Cr-Fe system have concluded that Nb additions between 1 and 2 wt% are beneficial for oxidation performance, forming compact chromia scales, whereas the Nb-free alloys produced multi-layered spinels [60]. Significant improvements in both oxidation and hot corrosion resistance in Ni-base superalloys with 2 wt% Nb additions have also been reported [61,62]. This behaviour was attributed to the formation of a Nb-rich layer that acts as a diffusion barrier to the outward migration of Ni. Although, concentrations of Nb greater than 2.5 wt% were shown to be deleterious, due to the layer becoming discontinuous and allowing oxygen ingress [63]. Similar results were reported by Christofidou et al. [64] in a study of compositional modifications to the powder metallurgy Ni-base superalloy RR1000.
In a study of a wide range of Ni-and Co-base superalloys by Barret [55], Nb was found to be detrimental to oxidation resistance at temperatures above 1000 °C. These results are consistent with the study by Smialek and Bonacuse [56] who described synergistic effects with Ti that result in severe degradation of oxidation performance above 1100 °C, and also with the study by Alkmin et al. [65] who observed inferior oxidation performance in Mar-M246 with elevated Nb content. Interestingly, studies of IN617 [66], IN625 [67] and IN718 [68] all oxidised at the same temperature show very similar scale thicknesses and morphologies, despite IN617 containing no Nb, IN625 containing 3.5 wt% Nb and IN718 containing 5 wt% Nb. These studies of the effect of Nb additions collectively indicate that, whilst superalloy oxidation performance can be affected by Nb additions, the effects are dependent on both alloy composition and exposure temperature. Provided that a given alloy can form a protective Cr 2 O 3 scale, Nb does not appear to have a significant effect on oxidation, as expected given that Nb oxides are less stable than Cr 2 O 3 . When protective scaling is not possible, for example due to very high temperatures [69], Cr evaporation [70,71] or oxide spallation [72], then Nb may have a more prominent role in oxidation due to the formation of non-protective multi-layered oxide scales.

3
The formation of a Nb-rich layer beneath the overscale is widely reported in the literature and has been studied in detail in the alloy IN625 by Chyrkin et al. [73]. The authors rationalised this behaviour using thermodynamic modelling to assess the activities of the diffusing species as a function of composition from the bulk alloy to the oxide-metal interface. Cr depletion in the substrate, associated with the formation of an external Cr 2 O 3 scale results in a zone of low Nb activity, driving uphill diffusion of Nb from the bulk towards the oxide-metal interface. This causes the local enrichment of Nb below the Cr 2 O 3 scale, and concomitant precipitation of the δ phase. The same effect has been confirmed in conventional IN718 as well as additively manufactured IN718 and IN625 [67,74,75].
Due to the wide range of reported effects, it is vitally important that they are characterised and understood in new Nb-containing superalloys to ensure the appropriate balance of properties is achieved at the intended operating temperature. In this study, the oxidation characteristics of a series of new polycrystalline Ni-base superalloys with systematically varying concentrations of Nb and Al have been investigated. These alloys are based on a system first reported by Mignanelli et al. [76], with further additions of Mo and W for solid solution strengthening, Fe for processability, Co for a decreased stacking fault energy, plus C, B and Zr for grain boundary strengthening. Previous work on these alloys focussed on the effects of individual alloying elements on mechanical properties [77] and oxidation performance was not considered. In the present work, this behaviour has been investigated through a combination of furnace exposures and subsequent microstructural analysis. These alloys are intended for service applications around 700 °C and have therefore been studied at this temperature. Whilst 800 °C lies beyond the intended service conditions, the oxidation behaviour of the alloys has also been assessed at this temperature to give insight into microstructural changes that occur at this temperature.

Experimental Methods
The alloys studied in this work fall within the range of United States Patent US10287654. Ingots of Alloys 1-4, which had systematically decreasing Al and increasing Nb contents, were produced by vacuum arc melting from their constituent elements with purity ≥ 99.9%. The nominal compositions of the alloys in atomic per cent are given in Table 1. Alloys 1-3 were homogenised at 1080 °C for 96 h to minimise solidification induced micro-segregation, and subsequently hot rolled at 1080 °C to a reduction of 55%. Alloy 4 was homogenised at 1170 °C for 48 h and subsequently hot rolled at 1170 °C to a reduction of 55%. All alloys were then aged 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. Samples of each alloy measuring 20 mm × 10 mm × 3 mm were cut using a precision saw, ground using successively finer SiC papers to a 5 µm finish, and polished with a 0.25 µm diamond suspension. All corners and edges were chamfered to reduce stress concentration effects, and polished to the same surface finish. The samples were subjected to 1000-h exposures in a box furnace at either 700 or 800 °C in laboratory air. Furnace temperatures were calibrated to ± 1 °C with the use of an N-type thermocouple.
After thermal exposure, the samples were mounted in conductive phenolic resin and cross-sectioned using a precision cut-off saw in preparation for electron microscopy. The standard metallographic preparation route outlined previously was used, followed by chemical polishing using colloidal silica 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) of the oxidised cross sections was performed using a Zeiss Gemini300SEM, with images acquired using secondary electrons (SE). A low accelerating voltage of 3 kV, combined with a working distance of 1 mm and a 30 μm aperture was used in conjunction with an InLens SE detector, in order to maximise the contrast and highlight the microstructural features. Compositional information and elemental concentration maps were acquired by energy-dispersive X-ray spectroscopy (EDX) using an Oxford Instruments X-Flash N 50 EDX spectrometer fitted to the same instrument. For these measurements, an accelerating voltage of 20 kV, an aperture of 120 μm and a working distance of 8.5 mm were used to maximise the EDX signal. It is noted that the SE images acquired with this configuration offered inferior resolution than the imaging mode described above. EDX data were processed using the Oxford Instruments AZtec software package.
X-ray diffraction data were acquired from the oxidised alloy surfaces with CuKα radiation and a 0.012 mm thick Ni filter, using a Bruker D8 ADVANCE diffractometer fitted with a LynxEye EX 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 a 2θ increment of 0.015° over the angular range 30-110° 2θ. During data acquisition, the samples were rotated to minimise any textural effects on peak intensities. To identify the phases present, the X-ray data were compared with reference patterns for the expected equilibrium species using the CrystalDiffract software package and reference crystallographic information files from the Inorganic Crystal Structures Database (ICSD © FIZ Karlsruhe GmbH).
Measurements of total oxide damage depth and γ′ depleted zone thickness were made from the SEM images using the ImageJ software package. The values are reported as an average of 50 measurements taken from five sampling lines placed over ten separate images acquired at random locations across the scale cross section.
Discontinous thermogravimetric analysis (TGA) was performed on samples measuring approximately 25 mm × 13 mm × 0.5 mm, cut from each alloy. Each sample was prepared using successively finer SiC grinding papers to a 4000 grit finish (approximately 5 μm), cleaned with acetone and thoroughly dried. Sample masses were measured using a Mettler Toledo XP105DR balance, accurate to ± 0.01 mg. Samples of each alloy were placed into individual alumina crucibles and exposed at 700 or 800 °C in a laboratory box furnace, calibrated to ± 1 °C with an N-type thermocouple. The samples were removed after 1, 2, 5, 10, 20, 50 and 100 h for mass gain measurements, being reinserted into the furnace each time.
Thermodynamic equilibrium and kinetic predictions were performed using the Thermo-Calc software with the TCNi8 Ni-superalloy database (version 8.2) and the MOBNi3 mobilities database (version 3.2). Contour plots were generated considering only the γ, γ′ and δ phases, with equilibrium calculations performed isothermally at 700 and 800 °C. Interdiffusion coefficients for Wagnerian diffusion analyses were calculated with reference to Ni, for the required diffusing element in its own concentration gradient, as well as for all of the other alloying additions individually. The main interdiffusion term was added to the cross terms to give a total interdiffusion coefficient for the given element in the alloy.
Key parameters required for the Wagnerian analyses, the oxygen solubility N (S) O , and the diffusivity of oxygen in the alloy D O , were calculated using the empirical formulations described by Park and Altstetter [78]. These were 9.27 × 10 -5 (atomic fraction) and 7.72 × 10 -11 cm 2 s −1 at 700 °C, and 1.75 × 10 -4 (atomic fraction) and 5.10 × 10 -10 cm 2 s −1 at 800 °C. Values for the parabolic rate constant k p were estimated from Malacarne et al. [79] where summary data for several commercial polycrystalline Ni-base superalloys were presented. These were taken as 8.4 × 10 -16 and 4.0 × 10 -14 g 2 cm −4 s −1 at 700 and 800 °C, respectively.

Alloy Characterisation
Compositional analysis of Alloys 1-4 was performed using Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES) by IncoTest UK, Special Metals Wiggin Ltd. Experimentally determined compositions are shown in Table 2, with elemental additions sufficiently close to the nominal values to be deemed suitable for the purposes of the study. The average grain sizes determined by the linear intercept method on data acquired via EBSD were 20 ± 12, 29 ± 18, 23 ± 14 and 60 ± 50 μm for Alloys 1-4 , respectively, (for further details please see supplementary information).

Oxide Microstructure After Exposure At 700 °C
Secondary electron (SE) micrographs showing the extent of oxidation damage across the alloy series are presented in Fig. 1 (back-scattered micrographs are provided for reference in the supplementary information). A compact, continuous oxide overscale is visible on the surface of all the alloys, alongside a sub-surface region containing acicular precipitates. This region is also denuded of the fine-scale cuboidal precipitates that can be seen in the bulk material far from the surface. The oxide overscale appears facetted for Alloys 1 and 4, whilst some scale rumpling was observed in Alloy 2. Dark contrast intrusions, penetrating into the alloy sub-surface can be seen prominently in Alloys 2 and 3. The thickness of the overscale and denuded zone appears to decrease across the alloy series (i.e. with increasing Al content, see Fig. 6 for quantitative measurements).

Compositional Analysis of Oxide Scales After Exposure at 700 °C
SE micrographs of the oxidised sample surface cross sections, accompanied by EDX elemental concentration maps corresponding to the same region are presented in Fig. 2 (the micrographs are less detailed than those in Fig. 1, due to the different imaging conditions used). The elemental concentration maps indicate the presence of a continuous chromium enriched layer, corresponding with enrichment in the oxygen map. Also visible from the elemental concentration maps is a sub-surface region depleted in Cr, with the depth of this region decreasing from approximately. 1.5 μm in Alloy 1, to 0.5 μm in Alloy 4.
There is clear evidence of Al enrichment in the overscale, coincident with the Cr enrichment. However, additional Al enrichment is seen beneath the overscale associated with the intrusions observed in the SE images. The regions of Al enrichment are also associated with elevated oxygen concentrations. This is most apparent in the overscale but is also present within the subscale intrusions. The extent of penetration of the intrusions is observed to decrease with increasing Al content (Alloy 1 to Alloy 4, left  Fig. 2). The scale also appears to become more uniform with increased Al content.
The Nb elemental concentration map shows significant enrichment within the elongated acicular precipitates that have formed within the Al and Cr depleted sub-surface region. The elemental concentration maps of Ni and Fe exhibited uniform contrast, with no evidence of any local enrichment, and are included in the supplementary information for reference.
EDX linescans with increasing distance normal to the oxidised surface for Alloys 1-4 are presented in Fig. 3. Cr depletion in the sub-surface zone is clearly observed for Alloys 1-3, although less so for Alloy 4. No notable enrichment in Nb or Al was detected in the sub-surface zone, although it is noted that due to the interaction volume of the electron beam limiting the resolution to approximately 1 μm, no quantitative analysis could be performed.  Fig. 4. As with the samples exposed at 700 °C, an external overscale can be observed with dark features penetrating the alloy sub-surface, as well as a zone denuded of fine precipitates. However, extensive formation of light contrast acicular precipitates can also be observed across the bulk of Alloys 1-3. These precipitates are observed to form with a common orientation, suggesting a strong crystallographic relationship with the γ matrix. In all four alloys the oxidationrelated damage extends to greater depths than observed for the samples exposed at 700 °C. The dark contrast intrusions beneath the overscale are mostly confined to Alloy 1, with limited evidence of their occurrence in the alloys with higher Al content. The light contrast acicular precipitates in Alloy 4 appear constrained to a region close to the alloy surface (in the region denuded of fine precipitates). This is in contrast with the other three alloys, in which the acicular precipitates extend  into the bulk material. Also noted in Alloy 4 are the darker regions in the bulk material. These are believed to be regions where residual oxide from the external scale has been mechanically spread over the cross-sectional surface during metallographic preparation.

Compositional Analysis of Oxide Scales After Exposure at 800˚C
SE micrographs of the oxidised alloy cross sections following exposure at 800 °C and accompanying elemental concentration maps are shown in Fig. 5. As with the samples exposed at 700˚C, the overscale is enriched in Cr and O. There is also evidence of globular precipitates with elevated Cr content that extend deep into the bulk material that were not apparent in the SE images. These globular precipitates are also rich in Nb and Mo (see supplementary information). The composition and morphology of these precipitates is consistent with them being carbides, which are known to form in Ni-base superalloys. However, full characterisation The elongated acicular precipitates show depletion in Cr and Al, and enrichment in Nb, indicating a compositional difference between these acicular precipitates and the globular precipitates. This suggests that the two different precipitate morphologies correspond to different crystallographic phases. At the base of the Cr-enriched surface scale, Al enrichment is detected, however, this appears to become less prevalent with increasing Al concentration. Al-rich intrusions can also be seen, and these are associated with a depletion in Cr, similar to that observed for the samples oxidised at 700 °C.
At the interface between the Cr-rich surface scale and the Al-rich layer a pronounced Nb enrichment can be seen. This is associated with a depletion in Cr and an absence of O. These observations suggest that the enrichment does not relate to the presence of a Nb-containing oxide, but to a metallic or intermetallic phase.
Finally, Alloy 4 shows a bright particle near the bottom left-hand corner of the elemental concentration maps that is rich in Nb and Mo but depleted in Cr, Fe and Ni. This is the only observed occurrence and is putatively NbC, which is known to be stabilised in Nb-bearing Ni-base superalloys [80].
EDX linescans with increasing distance normal to the oxidised surface for Alloys 1-4 at 800 °C are shown in Fig. 6. As with the results obtained following exposure at 700 °C, Cr depletion can be seen, corresponding to a reduction in the local concentration to approximately 10 at.%. Additionally, Al and Nb depletion is observed in the sub-surface zone, with a small zone of enrichment found immediately below the oxide scale.

Oxidation Damage Depth Measurements
Measurements of the extent of surface damage for all the alloys after oxidation at 700 and 800 °C are presented in Fig. 7. In this figure, the total damage depth and the width of the precipitate denuded zone are shown. At 800 °C, measurements of the total damage depth were not possible due to significant spallation of oxide from the surface. Following exposure at 700 °C, there is a decreasing trend in both the total damage depth and the extent of the precipitate denuded region with increasing Al content. Both measures of damage decrease approximately linearly with Al content over this composition range. Following exposure at 800 °C, a similar trend is observed as Al content increases. However, the alloy with the lowest Al content (Alloy 1) demonstrates significantly inferior performance compared to the other alloys in the series, with the γ′ depleted zone that develops at 800 °C being approximately four times greater than that observed at 700 °C.

X-ray Diffraction
XRD patterns acquired from the surfaces of Alloys 1-4 after oxidation at 700 and 800 °C are shown in Fig. 8, with the reflections corresponding to the principal phases indicated by markers. The fundamental reflections from the γ and γ′ phases in the alloy substrate are seen in the XRD patterns from all the samples. However, no superlattice reflections from the γ′ are discernible, in-line with previous studies of superalloys and the intrinsically low intensity of these peaks in laboratory XRD [53,81]. Peaks from the oxide phases Cr 2 O 3 and Al 2 O 3 are seen in all samples, with stronger signals from the Cr 2 O 3 , consistent with this phase being the dominant constituent of the oxide scale. Additional peaks can also be seen that may be attributed to the δ phase. These are likely to arise from the acicular precipitates seen in the precipitate denuded regions in the SE micrographs in Fig. 1 and extending into the matrix in Fig. 4. In the data from the samples exposed at 800 °C, the intensities of the Cr 2 O 3 , Al 2 O 3 and δ phase Fig. 7 Average measurements of total damage depth and γ′ depleted zone after exposure at 700 and 800 °C for Alloys 1-4 peaks are much higher than those from the samples exposed at 700 °C, with peaks that were previously not resolvable from the background now present. This is consistent with a greater volume fraction of oxide species having formed at 800 °C, and therefore being present within the volume illuminated by the X-ray beam.

Thermogravimetric Analysis (TGA)
Discontinuous TGA measurements are presented for Alloys 1-3 at 700 and 800 °C in Fig. 9. Alloy 4 exhibited significant internal cracking as a result of the hot rolling process, resulting in the TGA sample having a much larger effective surface area than measured. This resulted in very large mass gains compared to the other alloys. These data are included in the supplementary information for reference.
The normalised mass gains were fitted to a function of the form given in Eq. 2, with the exponent n tabulated in Table 3.

Equation 2 (TGA Fitting)
where Δm n is the normalised mass gain, A is a constant that accounts for the initial transient oxidation period, B is a rate constant, t is time, and n is the exponent describing the oxidation kinetics (parabolic with n = 2, cubic with n = 3, etc.) All of the alloys studied exhibited sub-parabolic oxidation kinetics. At both 700 and 800 °C, Alloy 1 exhibited the largest normalised mass gains and fastest oxidation rates. Alloys 2 and 3 both demonstrated lower normalised mass gains and slower oxidation rates, which were very similar at 700 °C, but showed greater separation at 800 °C.

Microstructure
The microstructures of the bulk material across the alloy series consisted of a γ matrix containing γ′ precipitates with a similar size, volume fraction and distribution. These appeared unchanged in the bulk of the alloys after exposure at 700 °C, and may be seen towards the bottom of the images in Fig. 1. These two phase γ/γ′ Δm n = A + (Bt) 1∕n Fig. 9 Discontinuous mass gains for Alloys 1-3 after oxidation at a 700 °C and b 800 °C microstructures are very similar to those found in commercial polycrystalline alloys such as Allvac®718Plus™, VDM780P and AD730® [10][11][12]. It should also be noted that some fine elongated precipitates (on the same scale as the γ′) are visible in the bulk of Alloy 1, likely to be γ″, which is known to form in alloys with elevated Nb content such as IN718 [3,82] as well as the dual-superlattice alloys that this alloy series is based on [76,83,84].
As confirmed by the localised Nb enrichment in the elemental concentration maps, there is extensive precipitation of the δ phase after long-term oxidation exposure, which is confined to the γ′ denuded zone at 700 °C, and extends into the bulk at 800 °C (in Alloys 1-3). Precipitation in this sub-surface zone is oxidation induced, with the Nb enrichment a consequence of the Cr depletion due to oxide formation. The precipitation of δ phase in the bulk of Alloys 1-3 after exposure at 800 °C does suggest that the alloy compositions are thermodynamically unstable with respect to δ formation at this temperature. As this phase only forms in significant volume fractions (in the bulk) after 1000 h at 800 °C, its formation is likely kinetically inhibited at 700 °C. Comparing these results to IN718, where δ precipitation is observed after thermal exposures of approximately 1 h at 800 °C, and 60 h at 700 °C [85] suggests that these alloys have a greater microstructural stability, with the precipitation of the δ phase retarded relative to IN718.
The EDX linescan data obtained from the samples exposed at 700 and 800 °C confirm the presence of a Cr depletion zone below the oxide scale, as expected given the diffusional supply of Cr from the bulk to the oxide-metal interface. At 800 °C, the Al and Nb depletion zone in the sub-surface region is indicative of the oxidation damage depth, and extends the furthest into the bulk in Alloy 1. In Alloys 2 and 3 the depleted zone does not extend as far, with Alloy 4 having the smallest distance from the oxide-metal interface to the depleted zone. Also present in the 800 °C linescans was notable Al and Nb enrichment immediately below the oxide scale, consistent with the microscopy results obtained, as well as the Nb enrichment phenomenon described by Chyrkin et al. [73] To help visualise the composition space occupied by these alloys, and rationalise some of the microstructural features described above, Thermo-Calc was used to generate contour plots showing the predicted volume fraction of δ precipitates as a function of Al and Nb concentration at 700 and 800 °C, shown in Fig. 10. It is important to recognise that at 700 °C, where no bulk precipitation of the δ phase was observed, the thermodynamic calculations predict that the δ phase will be present. The equilibrium volume fraction of δ increases with Nb content (and decreases with Al content) as would be expected considering that the δ phase is based on the nominal composition Ni 3 Nb. At 700 °C, the predicted equilibrium volume fractions of δ phase are greater than at 800 °C, providing further confirmation that within the bulk, δ precipitation is kinetically inhibited at 700 °C over the 1000 h exposures performed in this work.
To rationalise the oxidation induced δ precipitation observed in the γ′ denuded zone, the local composition, rather than the bulk nominal composition, needs to be considered. The zone formed below the metal-oxide interface is enriched in Nb as a consequence of uphill diffusion, and depleted in Al (which has been oxidised to alumina). From the contour plots, this would be expected to result in an increased δ volume fraction. The observed subscale δ precipitation is most likely driven by the Nb enrichment, as it has been observed in Ni alloys with very low Al content such as IN625 [67]. Whilst the Al depletion may contribute to the increased δ volume fraction by locally affecting the diffusivity, it is unlikely to be the primary cause of the oxidation-induced precipitation observed.

Oxidation
The results from the microscopy and diffraction experiments suggest that these alloys form two principal oxides after oxidation at elevated temperature. These can be spatially resolved to show a continuous Cr 2 O 3 overscale and a discontinuous Al 2 O 3 subscale. These strata are consistent with conventional knowledge on oxide formation in Ni-Cr-Al alloys [34][35][36]86], and would therefore be classed as Type II alloys in the designation used by Giggins and Petit [39]. It is well reported that superalloys based on Ni-Cr-Al with significant (approx. 20 wt% Cr) typically form transient surface NiO and Ni(Cr,Al) 2 O 4 spinels in the early stages of oxidation. However, due to the long duration exposures, it not surprising that no significant traces of such transient oxide products could be detected.
Other commercial polycrystalline Ni-base superalloys such as Allvac®718Plus™ and the powder disc alloy RR1000 have been shown to produce dense, compact Cr 2 O 3 scales, with an internal oxidation region containing Al 2 O 3 fingers, in addition to a diffusion zone depleted in γ′ precipitates [87][88][89][90][91]. This Type II oxidation behaviour has also been reported in studies of Astroloy, Waspaloy and U720Li [92]. Comparing these findings to the results shown in Fig. 6, it would appear that this alloy series offers a modest improvement in terms of oxidation resistance over RR1000 across the range of temperatures and durations studied. This is likely due to the higher Cr and significantly lower Ti content of these alloys, with higher Ti contents generally associated with accelerated oxidation kinetics as a result of increased defect concentrations following doping of the oxide with tetravalent Ti cations [53,93].
Increasing the Al content from 3.35 at.% in Alloy 1 to 4.40 at.% in Alloy 4 resulted in a decreased total damage depth, despite Al not being the principal scale forming element. The role of Al with regards to the oxidation resistance of chromia forming alloys is complex, but has been attributed to modified chemical activities of the oxide forming elements in the alloy [43]. However, thermodynamic calculations performed using the Thermo-Calc software show only minor changes in Cr activity (0.33678 in Alloy 1 to 0.33604 in Alloy 4). These modest changes are too small to be able to fully rationalise the observed behaviour. Whilst removing Al or Nb on their own would be expected to decrease the Cr activity [41,42,73], the substitution of one for the other is unlikely to have a significant effect given that both Nb and Al form compounds with Ni.
The main difference between the alloys is the evolution of the subscale zone, where the density of alumina intrusions decreases with increasing Al content. In order to assess whether these observations are a result of increased Al content leading to a gradual transition towards a continuous alumina subscale, a Wagnerian analysis was performed. The molar volume of the oxide species V Ox AlO 1.5 was taken to be 12.5 cm 3 mol −1 , with molar volumes for the individual alloys calculated using Thermo-Calc.
In these calculations it is important to consider how the oxygen solubilities will be affected by the partial pressure of oxygen ( p O 2 ), as they pertain to alumina formation below an external chromia overscale. As reported by Young et al. [94] when investigating internal oxidation, lower p O 2 values were found to result in reduced oxygen permeability and slower oxidation kinetics than would otherwise be expected. Effectively, the oxygen solubility governing the formation of internal alumina will be controlled by the equilibrium partial pressure of oxygen associated with Cr 2 O 3 dissociation. These partial pressures were calculated from data in [95] as 7.9 × 10 -32 atm at 700 °C and 4.3 × 10 -28 atm at 800 °C, which are several orders of magnitude lower than the oxygen partial pressure for NiO dissociation at the same temperatures. Using these partial pressures, the oxygen solubilities were calculated from Sieverts law using data from Park and Altstetter [78]. The reduced oxygen solubilities N (S*) O were calculated to be 3.9 × 10 -12 (atomic fraction) at 700 °C, and 3.7 × 10 -11 (atomic fraction) at 800 °C.
The critical Al concentrations required to supress the ingress of oxygen and promote continuous Al 2 O 3 formation, N (1) Al , were calculated for each alloy at 700 and  Table 4, alongside the predicted molar volumes and interdiffusion coefficients used in the calculations. The values for N (1) Al are very small (< < 0.1 at.%), which is expected given the extremely low p O 2 beneath the external scale. This results in a very low oxygen ingress into the alloy, meaning that negligible concentrations of aluminium are predicted to be sufficient to suppress it. Despite this, such low aluminium concentrations are unlikely to be able to sustain the scale, and hence the second Wagnerian criterion becomes critical when assessing the behaviour in the sub-surface zone. Values for N (2) Al , the critical Al concentration required to sustain a continuous scale were calculated for each alloy at both temperatures of interest, and are also tabulated in Table 4.
Both parameters N (1) Al and N (2) Al decrease with increasing Al content, with the predicted values for N (2) Al at 700 °C only slightly larger than the nominal compositions of the alloys studied. Alloy 4 has the closest nominal composition to the predicted value of N (2) Al at 700 °C (nominal = 4.05 at.%, N (2) Al = 4.23 at.%), which may rationalise the trend towards a continuous alumina subscale shown in Fig. 2. At 800 °C, the values for N (2) Al are higher than at 700 °C, but nevertheless display the same trend across the alloy series. These results are consistent with the reduced number density of internally oxidised alumina penetrations observed moving across the alloy series in Fig. 5, and may also account for why the continuous subscale observed in Alloy 4 at 700 °C was not seen at 800 °C.
Whilst quantitative assessments of the subscale enrichment in Al and Nb were not possible due to the small lengthscale over which they occur (as a result of relatively low oxidation damage at 700 °C), these phenomena are also likely influencing the observed behaviour. The amount of alumina formation below the external overscale will directly control the extent of Al depletion in the sub-surface region. This has been shown to influence local Cr activity and hence the diffusional flux of Cr through the sub-surface zone to the growing oxide scale [41]. Greater Al depletion in this zone causes enhanced Cr diffusivity, providing increased diffusional supply of Cr, more rapid establishment of a protective external overscale, and slower oxidation kinetics. Predictions of the Cr interdiffusion coefficients increase across the alloy series (with increasing Al content) from 5.42 × 10 -15 cm 2 s −1 in Alloy 1 to 6.52 × 10 -15 cm 2 s −1 in Alloy 4 at 700 °C, and 1.56 × 10 -13 cm 2 s −1 to 1.90 × 10 -13 cm 2 s −1 at 800 °C, supporting the observations of oxidation rate decreasing across the alloy series.
Another factor known to affect relative oxidation rates is doping of the oxide scales with elements added as alloying additions to the bulk alloy. The effects of doping can be considered using the Wagner-Schottky-Hauffe semiconductor valence approach [35,96]. As chromia is known to act as a p-type semiconductor at oxygen partial pressures greater than 10 -5 atm [54,97,98], doping with cations of a higher valency than Cr 3+ results in an associated increase in the vacancy concentration. The result is a concomitant increase in ionic conductivity and the facilitation of faster oxidation kinetics. Based on the changing Al:Nb ratio across this alloy series, Alloy 1, with the highest Nb content would be expected to have the fastest oxidation rate, consistent with the microscopy and mass gain measurements obtained.
However, in order to assess the contribution from doping of the oxide scales, it is important to consider the solubility limits of the respective alloying additions within the scale being formed. These are likely to be significantly different from the nominal alloy composition, which is often used in these types of analyses. The composition and solubility limits of oxide scales are not trivial to predict computationally but have been investigated experimentally using atom probe tomography (APT) [93,99]. These studies have confirmed that there exists significant solubility for Ni, Al and Ti in chromia scales. Notably, Lapington et al. [100] investigated the composition of chromia and alumina scales in Nb-bearing Ni-base superalloys, finding a small solubility for Nb and Ti in the bulk of both scales, and higher solubility at grain boundaries. Such grain boundary enrichment has been implicated in faster oxidation kinetics, and may contribute to the effects observed [53].

Conclusions
A series of new polycrystalline Ni-base superalloys have been evaluated in terms of their oxidation resistance in air at 700 and 800 °C.
The alloys all formed continuous external chromia scales, with internal alumina precipitation below the external scale. The depth of oxidation damage decreased with increasing nominal Al content in the alloys.
Bulk precipitation of the δ phase was detected for Alloys 1-3 at 800 °C, but not at 700 °C. Thermodynamic phase equilibria calculations confirmed the alloys to be unstable with respect to δ phase formation at these temperatures, suggesting that at 700 °C the transformation is kinetically inhibited over the duration of the 1000 h thermal exposure. In the alloy with the highest Al content (Alloy 4), no evidence of bulk δ precipitation was observed at either exposure temperature.
An increased propensity for δ phase formation occurred in the substrate immediately below the oxide scale. Compositional assessment of this zone indicates that there is a zone of Al and Nb enrichment immediately below the oxide scale. The uphill diffusion of Nb from the bulk to this zone is widely reported, and is a result of the Cr depletion (due to chromia formation) causing a zone of low Nb activity directly below the oxide scale. Wagnerian diffusional analyses considering the low p O 2 beneath the external scale were able to rationalise the behaviour of the sub-surface zone. The value of N (2) Al , the critical concentration of Al required to sustain a continuous subscale, decreased across the series, consistent with the reduced number density of alumina penetrations observed.
Normalised mass gain measurements showed that the oxidation rate decreased with increasing Al and decreasing Nb content, consistent with considerations of doping of the oxide scales, and the results from the microscopy performed.