Delayed Parabolic Oxidation via Transient Thermal 1 Exposures on a Polycrystalline Nickel-Based Superalloy

10 Multiple oxides were observed during the transient oxidation stage of the polycrystalline Ni-based 11 superalloy, RR1000, before a protective Cr 2 O 3 scale formed. Thermogravimetric analysis, synchrotron 12 grazing incidence X-ray diffraction, and electron microscopy were performed on samples subjected to 13 isothermal exposures at 800 °C for up to 100 hours. Transient effects governed the early stages up to 40 14 hours. NiO, spinels (NiCr 2 O 4 , (Ni,Co)(Cr,Co) 2 O 4 ), Cr 2 O 3 /(Cr 0.88 Ti 0.12 ) 2 O 3 , NiTiO 3 , and CrTaO 4 formed 15 during the initial stage with pseudo-linear kinetics. At the onset of parabolic kinetics, extensive Cr 2 O 3 and 16 TiO 2 growth dominated scale formation with the former emerging as the major passivating oxide.

1 Introduction Historical superalloy development [1] has resulted in many commercially successful polycrystalline Nibased superalloys, e.g.Waspaloy, IN 718, Udimet 720, Alloy 625, and RR1000.The high temperature oxidation of polycrystalline superalloys typically results in the formation of an external compact scale of Cr2O3, with discontinuous finger-like intrusions of Al2O3 in the subscale [2][3][4].In addition to Cr and Al, the complex compositions of these alloys can lead to the formation of multiple oxides.For example, one study investigated the oxidation of several commercial polycrystalline Ni-based superalloys from 750-1000 °C and showed that Al only formed internal oxides, whilst Cr and Ti formed external oxides that resulted in complex multi-layered oxide scales [2].Similar results [5] were observed in alloys of varying Al, Co, Cr, and Ti concentrations where internal Al2O3 formed, with significant layers of Co, Cr, and Ti-based oxides on the external surface.The study also showed that the capacity of the alloys to form a continuous Cr2O3 layer could be improved by decreasing Co or Ti and increasing Al or Ni concentrations.For the Ni-based superalloy RR1000, significant enrichment of Cr and Ti in the external scale formed at 600-900 °C has been observed consistently in multiple studies [6][7][8][9].It has also been shown that Cr and Al work synergistically to form a multi-layered oxide scale, with the former promoting the formation of the latter [10].These studies suggest that Cr and Al concentrations must be carefully selected to achieve optimal environmental resistance.Further studies have also shown that the growth kinetics of Cr2O3, specifically, can be influenced by other alloying additions via doping, for instance [5,8,[11][12][13][14], and also by the formation of stable phases that tie up the dopant within the alloy and thus reduce the impact [15,16].
Transient oxidation in Ni-based alloys occurs where different oxides form before the eventual ascendancy of more stable oxides (e.g.Cr2O3) once steady-state isothermal oxidation is achieved after prolonged exposure to elevated temperatures [17].This primarily arises due to the competing reactions between the various alloying elements with adsorbed oxygen anions during initial exposure to the oxidising environment.In general, identifying the oxides formed during transient oxidation can be difficult given their small volumes and varying timescales of formation as well as the possible spallation of non-adherent oxide products.Several studies into the transient oxidation stages on Ni-based alloys have shown the early formation of NiO on the alloy surface [4,[18][19][20], due to the high Ni concentration in the alloy and the faster oxidation kinetics relative to the other potential oxides [21].Another group of oxides that form during transient oxidation are the spinels, which have the formula of AB2X4 where A is typically Ni, B is Co, Cr, Ta or Al, and X is O [17,22].Several mechanisms leading to the formation of spinels have been proposed [23][24][25][26].Moreover, the precise role of spinels in the transient and long-term steady-state oxidation regimes is controversial.While generally considered to be non-protective oxides that form discontinuous scales, it has been reported that the presence of spinel oxides could also have beneficial effects on the oxidation resistance of superalloys [27][28][29].It can be rationalised that the formation of spinels during the transient oxidation phase could act as barriers to oxygen ingress, giving sufficient time for a more stable oxide (e.g.Cr2O3) to form as a result of the reduced partial pressure of oxygen at the oxide-metal interface.However, the absence of a reliable method to monitor transient oxidation processes in real-time makes it challenging to confirm this theory.
It is acknowledged that thin oxide scales formed on the alloy surface, compared to the volume of the underlying bulk alloy substrate, can be problematic to characterise with conventional X-ray diffractometry (XRD) due to inherently low volume fractions.Under such circumstances, the early onset and evolution of oxide products can be difficult to detect.However, a recent study of thin oxide scales with synchrotron Xray diffraction [25] has shown promise to successfully characterise the oxide products during transient oxidation since the high intensities of the incident beam allow for the detection of even low-volume fraction phases.Moreover, high-sensitivity thermo-gravimetric analysis (TGA) equipment has also been shown [30] to be capable of detecting subtle changes in kinetics.This level of accuracy has enabled the development of algorithms to account for the formation of oxides and their interplay [30].
It is clear from the literature that the oxidation behaviour of Cr2O3-forming polycrystalline Ni-based superalloys is well-understood.In particular, the commercial superalloy RR1000 has been extensively studied and therefore provides an excellent opportunity for establishing an understanding of oxidation dynamics in polycrystalline Ni-based superalloys.The present study outlines the oxidation behaviour of RR1000 up to 100 hours at 800 °C with a focus on the transient oxidation stage.To study this behaviour, oxidation tests were performed on samples subjected to systematically increasing exposure times following a controlled heating stage.The results are compared with data from previous studies.

Materials
The composition of coarse-grained RR1000, an established polycrystalline Ni-based superalloy used in turbine disc applications with a nominal grain size of approximately 30 µm [31], is outlined in Table 1.Samples were obtained from a test ring of a high-pressure turbine front cover plate forging, provided by Rolls-Royce plc.The test ring was commercially solution heat treated at 1170 °C for 1 hour, followed by a precipitation ageing heat treatment at 760 °C for 16 hours.The actual composition of RR1000, as given in Table 1, was measured with Energy Dispersive X-ray (EDX) analysis and averaged over three representative regions with approximate field-of-view dimensions of 23 x 18 µm 2 and an acquisition time of 10 minutes per region.In general, the results showed reasonable agreement between the actual and nominal compositions.

Oxidation Tests
TGA measurements were performed using a Setaram Setsys Evolution 18 apparatus to study the specific mass change associated with oxidation for RR1000 at 800 °C for 0 (the experiment was stopped immediately after the heating stage), 10, 20, 40, 60, 80, and 100 hours isothermal exposure in air.A testing temperature of 800 °C was selected to replicate the upper limit of operating temperatures that are typically expected for turbine discs.Samples for TGA analysis were prepared with approximate dimensions of 20 x 10 x 1 mm 3 .The final sample dimensions were measured using a digital vernier scale to allow specific mass changes to be calculated from associated mass change data.All faces and sides of the samples were polished to a 1 µm diamond finish.Before the TGA experiments, samples were cleaned in acetone followed by ethanol in an ultrasonic bath and dried for at least 24 hours in a drying cabinet.During the heating process, the sample was first heated from 20-200 °C at 10 °C/min and held for 10 minutes to stabilise.The sample was then heated from 200-775 °C at 20 °C/min.To avoid overshooting the temperature, the sample was heated from 775-800 °C at 5 °C/min.During the cooling process, the sample was cooled from 800 °C to 20 °C at 35 °C/min.
For the 100-hour oxidation test, the oxidation rate constant and mass gain exponent were acquired by fitting to a generalised form of the classical power law equation using the MATLAB Curve Fitting Tool: where ∆ is the specific mass change (mg/cm 2 ),   is the oxidation rate constant (mg n /cm 2n h), and  is the time (h).Parabolic oxidation behaviour is represented by an  ≈ 2, indicating that the growth is governed by diffusion of the oxidising species through the oxide.The 95% confidence bounds on the fitted coefficients were also computed with the MATLAB Curve Fitting Tool using Student's cumulative tdistribution function.

Synchrotron X-ray Diffraction Experiments
Synchrotron grazing incidence X-ray diffraction (SGIXD) was used to characterise the thin oxides formed during the thermal exposures.This technique provides very high angular resolution, high intensity and low instrumental peak broadening compared to conventional laboratory-based diffractometry, which greatly facilitates the identification of the phases present even at low-volume fractions.In this study, the procedure reported in [25] was employed and a schematic illustration of the setup geometry is provided in Figure 1.
The incident beam was inclined to the sample surface by an angle of 1 ° so that the resulting diffraction signal probes only the top few micrometres of the sample, making it advantageous for oxide analysis on the surface of the RR1000 samples.The SGIXD method was performed at the I11 high-resolution powder diffraction beamline at the Diamond Light Source, UK.
The X-ray measurements were acquired with a monochromatic 15 keV incident beam energy (with an equivalent wavelength of 0.8264 Å), which was calibrated against a NIST 640c Si standard.Each sample was mounted onto the sample stage of the diffractometer using a carbon tab.Diffraction patterns were acquired over a 3-150° 2θ scan range using the high-resolution multi-analysing crystals (MAC) detector (the primary detector on the I11 beamline).The scan time for each sample was 30 minutes, collected with an angular increment of 0.001°, determining the angular resolution.To improve the signal-to-noise ratio, the data was re-binned into 0.02° intervals.

Figure 1 -Schematic illustration of the SGIXD configuration used on the I11 beamline. An incident beam was impinged on the surface of the sample with the diffracted X-rays being fed through Si 111 crystals and received by multi-analysing crystals (MAC) detectors. Directions and rotational axes with respect to
the incident beam and sample are also shown.Image adapted from [25] with permission.

Oxide Analyses
Samples for oxide examination were Ni-plated for oxide preservation and sectioned with a precision saw utilising a SiC blade.A low-impact dry-cut method was used to minimise potential damage to the oxide layers.The sectioned samples were cold mounted in resin and subsequently ground down by at least 1 mm using SiC grinding papers to ensure the prepared oxide cross-sections were free from any cutting damage.Samples were polished to a 0.06 µm colloidal silica suspension finish.Samples were attached to an SEM stub using Cu-tape for enhanced conductivity during Scanning Electron Microscope (SEM) examination.
Microstructural examination of the polished cross-sections was carried out in a Zeiss GeminiSEM 300 SEM operated at 20 kV and a working distance of approximately 8.5 mm.The instrument was also equipped with an Oxford Instruments EDX detector to facilitate compositional analyses of microstructural features.Oxide cross-sections were examined in Backscattered Electron (BSE) mode and EDX elemental concentration maps of the alloy cross-sections were acquired alongside BSE SEM micrographs.
To investigate the evolution of the oxides and phases with increasing time, a general Rietveld analysis was performed on the SGIXD data using TOPAS-Academic V6 software [33].Rietveld refinement of the data allowed for the determination of individual phase fractions from the detected oxides.To this end, relevant CIF files were acquired from the Inorganic Crystal Structure Database (ICSD) through the Physical Sciences Data-Science Service (PSDS) [34].In addition, the appropriate instrumental parameters for the I11 beamline were acquired from the Diamond Light Source.The RR1000 sample that was oxidised at 800 °C for 100 hours was expected to possess the greatest quantity of surface oxides and was therefore deemed to be the most amenable for Rietveld analysis and, as such, was refined first.Once a satisfactory fit was achieved, its weighted profile R-factor (Rwp) was used as an approximate guide in the refinement of subsequent datasets.By subjecting each dataset to a similar level of refinement, it was possible to identify trends in the data when investigating the transient oxides formed.

Thermo-Gravimetric Analysis
The specific mass changes recorded using the TGA instrument for RR1000 during isothermal oxidation at 800 °C for 0-100 hours are presented in Figure 2. A zoomed-in view of the first 30 hours is shown in Figure 3.During the initial 30-40 hours of oxidation, the specific mass change curve for RR1000 had an approximately linear profile.Beyond ~40 hours, the curves followed a parabolic profile.The specific mass change curves for RR1000 were reproducible and virtually identical in the early stages.2 with 95% confidence bounds in parentheses.For comparison with literature values, the data were also fitted to an  = 2.Where parabolic kinetics was assumed,  is expressed as "2.000".Assuming parabolic behaviour, the   value for RR1000 in this study was similar to that reported in the literature, as shown in Table 2.

SGIXD Analysis
The SGIXD diffractograms obtained from the surface oxide scales of RR1000 after exposure at 800 °C for 0-100 hours following the heating stage are shown in Figure 4. Strong diffraction peaks corresponding to the γ matrix and γ′ precipitation phases in RR1000 could be identified, which dominated the diffraction signal from the 0-hour sample.Peaks associated with NiO, Cr2O3, NiCr2O4 (tetragonal), TiO2, (Cr0.88Ti0.12)2O3,(Ni,Co)(Cr,Co)2O4, NiTiO3, and CrTaO4 oxides were also identified.It is noted that NiCr2O4 has a cubic structure at high temperatures but transforms to a tetragonal structure below 46°C [36]; each allotrope can be distinguished from other phases present.Due to the significant similarities in crystal structure and lattice parameters for the cubic spinels, these compounds are referred to herein the general form (Ni,Co)(Cr,Co)2O4.
The spectra for the oxide phases showed clear and systematic increases in intensity with increasing time of exposure, which is most prominent in the approximate 13, 15, 18, 19, 22, and 32.5/33° 2θ positions.In contrast, the spectra for the γ and γ′ phases showed systematic decreases in intensity with increasing time of exposure as evidenced by the approximate 38, 45, 47, 60, and 62° peak locations.Rietveld refinement was performed on the SGIXD data in Figure 4 and the corresponding fits are shown in Figure 5 on a square root intensity scale.To compare the fidelity of the refinements, the difference between the experimental data and the calculated fit (i.e.residual) is shown for each experimental run.For some peak locations, it was challenging to describe the intensity of the experimental data with the calculated fits which resulted in slight underfitting.However, the residuals were considered small and demonstrated good agreement between the experimental data and the calculated fits.The identified phases in Figure 4 and fractions in Figure 6 are tabulated in Table 3 with associated uncertainties calculated using TOPAS-Academic V6.In addition, the relevant CIF file code for each phase used and the weighted profile R-factor (Rwp) for each Rietveld refinement are shown.3.3 Cross-Sectional Oxide Analysis BSE SEM images of RR1000 cross-sections and associated EDX elemental concentration maps after exposures at 800 °C in air for 0-100 hours are presented in Figure 7.In the initial case (0 hours of isothermal exposure following the heating stage), the oxide scale on the surface of RR1000 was too thin to be positively identified using SEM EDX.It was noted that the Ni plating appeared to have separated from the oxide surface which could potentially have resulted in some oxide scale loss.Oxygen was detected, however, the exact composition of the oxide formed could not be measured using SEM EDX.As the relative amounts of Ni in the alloy and NiO are similar, it is postulated that the most likely oxide formed is that of Ni, as revealed by the SGIXD above.After 10 hours, early oxidation products were visible near the surface which EDX showed as predominantly Cr and Ti oxides.The thickness of the oxide formed varied across the surface of the sample with faster growth noted at emergent grain boundaries, identifiable by the presence of internal oxidation (finger-like intrusions) of Al.After 20 hours, a more continuous external scale of Cr2O3 was beginning to form across the surface of the sample with internal Al2O3 intrusions.The thickness of the surface layer of Cr2O3 after this period of exposure continued to show non-uniform growth, with the faster oxidation kinetics shown at emergent grain boundaries.The presence of TiO2 in the external scale was observed and an obvious Ti-depleted zone developed with the alloy beneath the external scale.Similar to the 0-hour sample, the Ni plating appeared to have separated from the oxide surface and may have also resulted in some oxide scale loss.After 40 hours, the external Cr2O3 scale had thickened on the surface of the alloy.After 60 hours, the Cr2O3 scale remained relatively constant while TiO2 increased in thickness.Moreover, the Ni plating appeared to have separated from the oxide scales and a significant amount of compacted debris from the grinding/polishing procedure was entrapped in the resulting gap.However, this does not appear to have adversely affected the Cr2O3 and TiO2 scales as these are still distinguishable from the debris.The thickness of the Cr2O3/TiO2 scale did not increase markedly from 80 to 100 hours, but the size and depths of the discontinuous Al2O3 intrusions increased and continued to extend into the alloy.While no specific areas of significant Co and Ni concentrations were observed, faint protrusions enriched in Co and Ni were detected at the alloy surface.Regions of Mo enrichment were also observed near the surface of the alloy and surrounding the Al2O3 intrusions.No significant Ta enrichment was observed in the initial condition but gradually increased in intensity from 10-40 hours, reaching an approximately constant thickness from 60-100 hours.

Thermo-Gravimetric Analysis
The TGA results in Figure 2 and Figure 3 showed approximately linear specific mass change kinetics over the early oxidation stages of RR1000 at 800 °C (up to 30-40 hours) followed by decreasing oxidation rates.This change in kinetic law with increasing time at temperature contrasts with previous studies where parabolic kinetics were recorded from the outset of isothermal exposure [9,35].A value for  of 1.286 was determined for the collected data (Figure 2 and Figure 3) and suggests a kinetics relationship that is closer to linear than parabolic.The earlier studies found the alloy, at 800 °C, followed parabolic kinetics over the same time period, i.e. up to 100 and 200 hours, respectively [9,35].The difference between the results obtained in this study and those of the similar studies on the same alloy was unexpected.The linear kinetics recorded in this study up to 40 hours showed a prolonged period of transient oxidation before the protective and continuous scale of Cr2O3 had been established.The EDX results in Figure 7 revealed the development of the oxides on the alloy surface.At 0 hours (i.e., following the heating rate), oxygen was detected at the surface of the alloy, with SGIXD confirming this to be NiO.At 10 hours, a discontinuous Cr2O3 scale had formed along with TiO2, corresponding to the onset of linear kinetics as shown in the TGA results.It is believed that the formation of these two oxides, the observed differences in growth rate across the surface of the samples, and the additional identified oxides contributed to the linear kinetic region of the TGA results.It was also noticeable that the concentration of NiO decreased over this time period.At 40 hours, the EDX maps showed a continuous Cr-rich oxide had formed at the surface of the alloy and the majority of the additional oxides were found using SGIXD to be either stable in concentration or decreasing in intensity.At 40 hours, a classical parabolic profile began.
The difference in the value of  in this study to that reported in [9,35] could be explained by examining the specific protocol details of the testing performed.The heating rate used in this study is presented in Figure 8 along with the heating rates used in the systematic study by Reynolds et al. [25] which investigated the effect of heating rates (2, 5, 10, 20, 40, 100 °C/min) from ambient conditions to a test temperature of 650 °C on the oxidation of RR1000.The average heating rate in this study is approximately 13 °C/min and is therefore between the 10 °C/min and 20 °C/min heating rates observed in [25].Thus, in this study it took approximately 1 hour for the samples to reach the test temperature of 800 °C, enabling a significant quantity of oxide formation to occur over the thermal transient stage.In this regard, the results agree with [25] where the formation of NiO and (Ni,Co)(Cr,Co)2O4 spinels were observed.Therefore, the oxides formed during the heating stage may have contributed to the relatively linear kinetics observed over the time period of up to 40 hours.However, it is noted that the test temperature was 650 °C in [25] while this study was conducted at 800 °C.It is possible that a faster heating rate may limit the extent of the transient oxidation stage in this alloy.
Figure 8 -Heating rates of RR1000 in a thermo-gravimetric apparatus.The heating profile used in this study (black) and the average heating rate (dotted) are shown alongside heating rates investigated in [25].
In addition to the heating rates, the different surface finishes adopted in other studies compared to this investigation must be considered as they may influence the oxidation behaviour.The sources in Table 2 [7,8,35] had polished samples to either 0.25 µm or 6 µm diamond surface finishes which differ from the samples in this study (1 µm diamond finish).However, the final mass gain in Figure 2 shows close agreement with the final mass gain at 800 °C in [9], which suggests that the surface roughness range quoted for these studies does not significantly affect the overall mass gain.It has been proposed [37] that higher surface roughness can reduce oxidation damage and mass gains.It is theorised that the increased concentration of defects at the surface due to a higher roughness could create fast diffusion paths for Al or Cr, expediting protective oxide formation [37].This hypothesis could explain why the results in [7,8] showed comparatively better (i.e.parabolic) kinetics than the present study.However, the reason why the result in [35] also showed parabolic kinetics despite having a smoother surface finish than the present study suggests that the different heating rates used may be the cause of the difference in the observed kinetics laws.

Temporal Evolution of Oxides
From the studies by Giggins and Pettit [4], the Ni, Cr, and Al content of this alloy (approximately Ni-16Cr-3Al wt% or Ni-17Cr-6Al at%) should form a continuous surface scale of Al2O3 instead of Cr2O3.This difference in behaviour can be attributed to the lower temperature used in this study, reducing the diffusion rate of Al [38].Indeed, the increased diffusion rate of Al at higher temperatures has been used to promote Al2O3 formation on a Cr2O3-forming superalloy through high-temperature pre-oxidation treatments [39].
In addition, the role of the other alloying elements also needs to be considered.Critically, Ti additions have been shown to promote Cr2O3 scale formation in Ni-based alloys and suppress Al2O3 [15] through accelerated Cr2O3 growth kinetics [8,[11][12][13][14].Conversely, in alloys where more stable phases containing Ti can form, lower Cr2O3 kinetics have been recorded [15].
In this investigation, the development of the surface oxides formed over the first 100 hours of exposure to 800 °C has been revealed using SGIXD (Figure 4) showing a gradual formation of various oxides with time.
While peaks associated with NiO, NiCr2O4 (tetragonal), (Cr0.88Ti0.12)2O3,(Ni,Co)(Cr,Co)2O4, NiTiO3, and CrTaO4 were identified, the peak intensities were relatively low (0-20 hours) until the later stages (80-100 hours) where more reliable identification was possible.The formation of NiO, NiCr2O4, and (Ni,Co)(Cr,Co)2O4 was expected and has been reported in similar studies [25,[40][41][42][43][44][45].This phenomenon is typically attributed to the alloy forming spinels and/or NiO upon initial exposure to air at high temperatures and has been shown by Evangelou et al. [44] to predominate at high oxygen partial pressures.Although the temperatures investigated in [44] were significantly lower than in this study, similar oxidation mechanisms were proposed for the formation of NiO and spinels.Moreover, the formation of NiCr2O4 may have contributed to decreased oxidation resistance as it has been previously associated with non-protective oxidation behaviour [4].
The detection and formation of (Cr0.88Ti0.12)2O3 is not frequently reported in similar studies [8,11] on RR1000 from 600-900 °C potentially due to the limited resolution of conventional XRD techniques.However, Reynolds et al. [25] reported the formation of this phase at 650 °C for RR1000, which was attributed to a Ti-doped Cr2O3 scale.In addition, Wang et al. [46] have also reported the formation of (Cr0.88Ti0.12)2O3 in the GH4720Li Ni-based superalloy at temperatures ranging from 750-900 °C.The effect of Ti-doping on the Cr2O3 scale in Ni-based alloys has also been previously reported by several researchers [8,[11][12][13][14] therefore its occurrence in this study is expected given the Ti concentration of RR1000.The use of SGIXD has thus been able to provide a more detailed investigation of the oxides formed even at short thermal exposures providing greater detail of this important stage in the oxidation process.Similarly, NiTiO3 (nickel titanate) is seldom observed, however, its formation is speculated to result from a chemical reaction between NiO and TiO2, as shown in the reaction quoted in Equation 2 [47][48][49].When detected, NiTiO3 has been associated with a fast-growing and non-protective scale [48].In addition, several other reactions are possible that would result in a decrease and subsequent disappearance of the initially formed NiO.These include: where  indicates a solid phase and  indicates a gaseous phase.Equations (2), (3), and (4) are solid-state reactions with no associated mass change, while reaction (5) would result in a change of mass.
The increasing Ta concentrations on the surface in Figure 7 are consistent with the detection of CrTaO4 in the SGIXD results.The identification of Ta-rich particles in the oxide layer is consistent with Das et al. [40] who observed Ta-rich oxide particles in the superalloy CM-247LC after exposure at 1000-1200 °C.The occurrence of CrTaO4 aligns well with previous studies [43,50,51] where complex oxides of the form (Cr, Ti, Al)(Ta, Ti)O4 tend to form at temperatures below 1100 °C.Pieraggi et al. [51] surmised that CrTaO4 can form a continuous solid solution with rutile TiO2 given that CrTaO4 also possesses a rutile structure.Indeed, the addition of Ta was shown to be beneficial for the oxidation performance of Co-Cr-Ta alloys by Irving et al. [52].It has also been reported by Jalowicka et al. [15] that Ta and Ti can react with oxygen to form TiTaO4, which reduces Ti-doping of Cr2O3 and promotes dense Al2O3 sub-scale formation.However, it is noted that the alloys in that work (Rene 80 and PWA 1483) had significantly higher Ti and Ta concentrations than RR1000, which prevents a direct comparison.No evidence of Ta2O5 was detected in this study, contrasting to studies undertaken at different temperatures [43,53].However, it can be inferred from the work of Ren et al. [50] that Ta2O5 may initially form as a precursor oxide that subsequently reacts with Cr2O3 to form CrTaO4.It is therefore possible that any Ta2O5 that may have formed in RR1000 was rapidly developed into CrTaO4.The absence of NiTa2O6 tri-rutiles in RR1000 differs from the work of Nychka et al. [42] which is likely due to PWA 1484 having a much higher Ta concentration.
In general, the SGIXD data (Figure 4) shows that the Cr2O3 peaks become much more prominent after 40 hours, aligning well with the TGA data where the specific mass change curve begins to adopt a parabolic profile.This agreement suggests that the onset of transient oxidation and the shift towards dominance of a specific oxide (e.g.Cr2O3) could be monitored with the use of TGA experiments alone.
To summarise, it is proposed that the oxidation behaviour of RR1000 at 800 °C could be broadly divided into two stages: early and late.The early stage, up to 40 hours, was characterised by approximately linear growth kinetics with the initial formation of NiO, during the heating stage, followed by a mixture of spinels (NiCr2O4/(Ni,Co)(Cr,Co)2O4) and Cr2O3/(Cr0.88Ti0.12)2O3and other minor oxide phases, without a clear oxide preference.The formation of NiO subsided due to competing oxide formation and participation in chemical reactions.As discussed in the literature, the spinels and (Cr0.88Ti0.12)2O3 in the early stage could have acted as barriers to oxygen ingress [27], which may have facilitated the formation of Cr2O3 and some TiO2.In the late stage, from 40 hours onward, Cr2O3 emerged as the dominant oxide as well as the continued formation of TiO2.In addition, the formation of spinels decreased while relatively small levels of NiTiO3 and CrTaO4 remained.It was shown in this study that oxide phase characterisation during the transient oxidation regime was enabled using the high intensity and angular resolution of SGIXD, which was subsequently able to rationalise the pseudo-linear oxidation kinetics.

Conclusions
In this study, three techniques were used to characterise the transient oxidation behaviour of the commercial polycrystalline Ni-based superalloy, RR1000, in air at 800 °C from 0-100 hours.The main findings are as follows: • RR1000 exhibited approximately linear specific mass change kinetics up to approximately 40 hours of oxidation, beyond which parabolic kinetics was evident.The  and   values were 1.286 and 1.623 × 10 -3 mg n /cm 2n h, respectively.When fitted to parabolic kinetics, the values compared well to literature data.The changes in kinetics observed in this study were rationalised as an extended period of transient oxidation before a continuous and protective Cr2O3 scale was established.• A continuous external Cr2O3 scale formed on the surface of RR1000 with internal Al2O3 finger-like intrusions, which is characteristic of the oxidation behaviour for alloys of this chemical composition.TiO2 was also detected at the surface.• SGIXD results showed a clear and systematic evolution of oxide spectra with increasing oxidation time.Peaks associated with Cr2O3 and TiO2 increased in intensity with increasing time at temperature.To a lesser extent, the presence of NiO, NiCr2O4, (Ni,Co)(Cr,Co)2O4, (Cr0.88Ti0.12)2O3,NiTiO3, and CrTaO4 was also detected.• Rietveld refinement of the SGIXD data revealed two main stages in the oxidation process of RR1000: The early stage, demonstrated by approximately linear kinetics, was characterised by the formation of NiO during the heating stage followed by the formation of spinels (NiCr2O4/(Ni,Co)(Cr,Co)2O4), and Cr2O3/(Cr0.88Ti0.12)2O3and a number of minor oxide phases.The

Figure 2 -
Figure 2 -TGA specific mass change curves for isothermal oxidation of RR1000 samples for times up to 100 hours at 800 °C in air.A separate sample was exposed for each period outlined in the legend.The traces show a period of approximately linear kinetics up to 40 hours followed by a decreasing oxidation rate.

Figure 3 -
Figure 3 -TGA specific mass change curves for isothermal oxidation of RR1000 samples at 800 °C over the time period of 0-30 hours in air showing linear or near-linear growth kinetics.The fitted mass change coefficients from Equation (1) for the 100-hour exposure at 800 °C are given in Table2with 95% confidence bounds in parentheses.For comparison with literature values, the data were also fitted to an  = 2.Where parabolic kinetics was assumed,  is expressed as "2.000".Assuming parabolic behaviour, the   value for RR1000 in this study was similar to that reported in the literature, as shown in Table2.

Figure 4 -
Figure 4 -Synchrotron grazing incidence X-ray diffractograms (SGIXD) from the surface of RR1000 after oxidation at 800 °C for 0-100 hours in air.

Figure 5 -
Figure5-Calculated fits and residuals using Rietveld refinement of synchrotron grazing incidence X-ray diffractograms (SGIXD) from the surface of RR1000 after oxidation at 800 °C for 0-100 hours in air.Individual phase fractions (wt.%) obtained from the Rietveld refinements using TOPAS-Academic V6 software were used to examine the evolution of the identified phases as a function of time at 800 °C and are displayed in Figure 6.The SGIXD spectra were initially dominated by signals attributed to the γ, γ′, and NiO phases.The first 0-10 hours were characterised by a reduction of NiO followed by increasing formation of spinels (NiCr2O4 (tetragonal)/(Ni,Co)(Cr,Co)2O4) and Cr2O3/(Cr0.88Ti0.12)2O3while γ and γ′ fractions decreased.From 10-40 hours, Cr2O3/(Cr0.88Ti0.12)2O3grew to become the major oxide while significant fractions of NiTiO3 and NiCr2O4 (tetragonal) were also present.The amount of NiO decreased almost completely to zero during the first 30 hours.From 40 hours onward, Cr2O3 and TiO2 were the major oxides present alongside (Cr0.88Ti0.12)2O3and NiTiO3.The cubic spinels, (Ni,Co)(Cr,Co)2O4), also decreased in

Figure 6 -
Figure 6 -Evolution of phase fractions (wt.%) calculated with Rietveld refinement of synchrotron grazing incidence X-ray diffractograms (SGIXD) from the surface of RR1000 after oxidation at 800 °C for 0-100 hours in air.

Figure 7 -
Figure 7 -Cross-sectional BSE SEM images (top row) and corresponding EDX elemental concentration maps of RR1000 after oxidation at 800 °C for 0-100 hours in air.It is noted that the Ni plating had separated from the oxide scales in the 0-hour, 20-hour, and 60-hour samples.The 60-hour sample showed compacted debris from grinding and polishing procedures entrapped in the gap between the oxide scales and detached Ni plating.

Table 1 -
[32]nal and actual compositions of RR1000 (at.%).It was not possible to accurately measure the actual concentrations for B and C with Energy Dispersive X-ray (EDX) and therefore these valueswere omitted from the table.Adapted from[32].

Table 2 -
Values of  and   from Equation (1) determined for RR1000 after oxidation at 800 °C for 100 hours with 95% confidence bounds in parentheses.Literature values are provided for comparison and sources are given.

Table 3 -
Identified phases in RR1000 during oxidation at 800 °C for 0-100 hours following the heating stage.CIF collection codes for each phase are provided.The uncertainties and Rwp values for each Rietveld refinement calculated in TOPAS-Academic V6 are also shown.