Thickness and composition of native oxides and near-surface regions of Ni superalloys

The surface chemistry and thickness of the native oxide, hydroxide, and modified sub-surface layer of three Ni superalloys (alloy 59, 625, and 718) were determined by synchrotron X-ray Photoelectron Spectroscopy (XPS) and X-ray Reflectivity (XRR). Taking advantage of the synchrotron radiation techniques, a procedure for normalizing the photoelectron intensity was employed, which allowed for accurate quantitative analysis revealing a total oxide thickness for all samples of 12–13 Å, a hydroxide layer of 2–3 Å


Introduction
Metallic materials are a cornerstone and a major building block in our modern society.They are used to build our cities, art, vehicles, and industries.One factor limiting the lifetime of metallic materials is the spontaneous material degradation due to corrosion driven by the thermodynamic laws.Many metals spontaneously oxidize and corrode in an ambient environment.Some metals and engineering alloys, however, can resist this spontaneous corrosion for relatively long periods.This is due to the inherent property of the metals and alloys to form a thin protective layer of native oxide that prevents the underlying metal from fast oxidation and hence acts as corrosion protection [1,2].A theory of oxidation of metals was developed by Cabrera and Mott in the 1950 s, predicting a self-limiting thickness of the stable oxide formed on metals [3,4].This model was further developed later, and it was shown that diminishing adsorption energy of oxygen as a function of oxide thickness is the physical phenomenon behind the self-limiting thickness of thin oxides on metal https://doi.org/10.1016/j.jallcom.2021.1626570925-8388/© 2021 The Authors.Published by Elsevier B.V. CC_BY_4.0
For some very demanding applications in harsh environments, especially at high temperature in combination with high mechanical strain, for example, in gas turbines, sour oil and gas wells, and nuclear industry, stainless steel sometimes does not meet the material requirements.There, high-alloyed Ni alloys or so-called Ni superalloys can be used, the later exhibiting excellent mechanical and corrosion properties even at elevated temperatures and often in very corrosive media [16][17][18][19][20][21][22][23].However, the nature of the native oxide film formed on industrial Ni superalloys is not well studied, and the thickness and composition of the native oxides are not well known.
New paragraph: Recent advancements in electrochemical measurements, namely inductively coupled plasma mass spectrometry (ICP-MS), have allowed for detailed electrochemical studies of Ni superalloys to better understand the corrosion mechanism [24][25][26][27][28].In addition, the effect of surface structure, temperature, pH, and heat treatment on the corrosion resistance of Ni superalloys have recently been studied [29][30][31][32][33][34].The studies are crucial for the practical understanding of these important engineering materials.However, the studies do not directly address the nature of the passive film on Ni superalloys and how it is affected by the material's composition, which is crucial for understanding the corrosion resistance.To date, only a limited number of publications address the composition and thickness of Ni alloys with direct experimental evidence [26,[35][36][37][38]70].Furthermore, the native oxides on many commercial Ni superalloys are virtually unexplored.
In this work, we study three industrial-grade Ni alloys: alloy 59, alloy 625, and alloy 718 using state-of-the-art large-scale synchrotron radiation facilities to perform X-ray Photoelectron Spectroscopy (XPS) and X-ray Reflectivity (XRR) measurements to study the composition and thickness of the native oxide.The high flux monochromatic synchrotron x-ray beam with tunable energy allows us to study the native oxide with excellent surface sensitivity, the same probing depth for all alloy components, and higher resolution than traditional laboratory source XPS and XRR instruments.The state-of-the-art instrumentation allowed for accurate determination of the composition and thickness of the native oxide and near-surface region of these Ni alloys, which are essential to improve the understanding of the corrosion properties of these materials.

Investigated materials and sample preparation
The materials investigated were commercial-grade Ni alloys supplied by Sandvik Materials Technology, Sweden.The composition of the main alloying elements of the three alloys is shown in Table 1.A table of minor and trace alloying elements can be found in the Supplementary Information (ST1).
The samples used for the XPS investigation had a cylindrical shape with a diameter of 7 mm and a thickness of 3 mm.The surface of interest was polished using a ¼ µm silica suspension, followed by a short polishing step using oxide polishing suspension (Struers, Denmark).After polishing, the samples were stored in air for several weeks before performing the experiments.The samples were cleaned in acetone and ethanol using an ultra-sonic bath and then blow-dried using clean air before the experiments.The samples dedicated for the XRR experiments had a cylindrical shape of 7 mm in diameter and a thickness of 4 mm.The samples were polished to an R a value of less than 30 nm by Surface Preparation Laboratory (SPL, Netherlands).After polishing, the samples were cleaned in acetone, ethanol, and ultra-pure water using an ultra-sonic bath and then blow-dried using clean air.After the polishing, the samples were stored in air prior to performing the experiments.Storage of the samples in air for several weeks was performed to ensure that the steady-state thickness of the native oxide was achieved since it has been shown that formation kinetics for native oxide film formation can be slow and more than 24 h might be needed before reaching steady state [39].

Synchrotron-based x-ray photoelectron spectroscopy study
The XPS experiments were conducted at the FlexPES beamline at MAX IV Laboratory, Lund, Sweden.The tunable energy at modern synchrotron facilities allows the photon energy to be varied such that the kinetic energy of the photoelectrons is the same for each core level.Here the photon energy was chosen such that the kinetic energy was 200 eV for each core level.This results in a constant probing depth of the XPS signal from each alloying element.The probing depth can be calculated from the inelastic mean free path (IMFP) of the electrons, λ, in the material where 3λ corresponds to the depth above which ~95% of the signal originates [10], corresponding to ~20 Å for Cr 2 O 3 at 200 eV kinetic energy for a value of λ = 7 Å calculated using QUASES-IMFP-TPP2M software.A survey spectrum was measured at 1200 eV for each sample, followed by high-resolution spectra collected at the following core levels: Ni 2p, Fe 2p, Cr 2p, O 1s, Mo 3d, and Nb 3d.A summary of the core levels and the respective photon energy used are listed in Table 2.The photoelectrons were detected by a SES2002 hemispherical analyzer (Scienta Omicron, Sweden) at normal emission.Table 2 also shows the IMFP, calculated using QUASES-IMFP-TPP2M software which is based on the formula from Ref. [40], and the photoelectron crosssection, σ, taken from Ref. [41].The binding energy (BE) for each spectrum was calibrated against the measured Fermi edge.
XPS was measured in ultra-high vacuum (UHV) in the pristine state, that is, after polishing and several weeks of air exposure.The samples were then sputtered with 1 keV argon plasma for 30 min to remove the oxide, after which XPS was measured again.The XPS results from the sputtered samples were used to calibrate the photoelectron intensities of the different core levels at different photon energies.The photon flux of the beamline varies significantly (approximately a factor of 10) between the Ni core level measured at 1045 eV and the Mo core level measured at 420 eV photon energy.
Using the survey spectra at 1200 eV photon energy and individual spectra after sputtering, the intensities could be calibrated for the difference in photon flux and photoelectron cross-section resulting in a good agreement between the measured and tabulated composition as shown in the Supplementary Information (ST2).The whole calibration procedure is described in detail in the Supplementary Information.
A benefit of varying the photon energy for each core level is that the kinetic energy of the photoelectrons is constant for each alloy element.This ensures that the attenuation of the photoelectrons is the same through the surface layer of carbon that is always present on samples exposed to air.It also ensures that the probing depth is constant, allowing for a more accurate quantitative analysis.

Synchrotron based x-ray reflectivity study
The XRR experiments were conducted at the Swedish Materials Science beamline P21.2 at PETRA III, DESY, Hamburg, Germany.The high photon flux, small beam size, and highly monochromatic x-ray beam at modern synchrotron sources enable high-resolution XRR studies of thin films.The x-ray energy was 37.5 keV (λ = 0.331 Å), and the x-ray beam size was 6 × 15 µm 2 (VxH).The sample was mounted on the surface diffractometer, and the sample surface normal was aligned in the vertical scattering plane and perpendicular to the incoming beam.The reflectivity was measured in a θ/2θ geometry in a q-range of up to q= 0.8 Å -1 .

Surface chemistry, composition, and thicknesses determined with XPS
Survey spectra measured with 1200 eV photons for each of the three samples are shown in Fig. 1.In the survey spectra, the core levels measured with high resolution are marked with dotted boxes.In addition to the XPS and Auger peaks of the expected alloying elements, a small quantity of copper was detected on the surface.This is disregarded from the further analysis since it was only in trace amounts.Oxygen and carbon were also present on the surface.Oxygen is due to oxide and potentially also in adsorbed water.Carbon is present in the alloy, as shown in the compositional Table of minor alloying elements in the Supplementary Information (ST1).However, carbon is mainly attributed to surface contamination due to exposure to an ambient atmosphere where hydrocarbon contamination is unavoidable.
The high-resolution XPS spectra measured for each alloying element in the alloys are shown in Fig. 2. All spectra are background-subtracted using a Shirley background.The experimental data are plotted as black dots, the cumulative fits are shown as solid red lines, and the individual components of the fit are shown as colored areas.Peak fitting was performed using IGOR PRO software using Voigt line profiles.Fig. 2a) shows the Ni 2p 3/2 XPS spectra for each of the three alloys.A clear metallic Ni peak can be seen with BE 852.8 eV, fitted with an asymmetric peak shape.Ni 2p is known to show a significant satellite peak observed at 859.4 eV, in agreement with the literature [36,37].Except for metallic Ni, a peak corresponding to Ni 2+ hydroxide can be seen at 855.8 eV with an associated satellite at 861.5 eV [37].No Ni oxide could be detected in the Ni 2p spectral range, NiO has a BE of 854 eV [42,43] where no peak is detected.This is in agreement with previous XPS studies of native oxides and passive films on Ni alloys showing the absence of NiO [35,36].Fig. 2b) shows the Fe 2p 3/2 XPS spectra for each Ni alloy.No significant Fe signal was detected with a kinetic energy of 200 eV for alloy 59 or alloy 625, which contains 0.8 at% Fe and 4.38 at% Fe, respectively.Only for alloy 718, which contains 19.5 at% Fe, a broad photoelectron peak was detected.The Fe 2p 3/2 spectrum for alloy 718 was fitted with an asymmetric metallic Fe peak at 706.7 eV and two oxide components, Fe 2+ oxide and Fe 3+ oxide at 709.6 eV and 710.8 eV, respectively [12,44,45].Fig. 2c) shows the Cr 2p 3/2 XPS spectra for each Ni alloy.The Cr spectra were fitted with an asymmetric metal component corresponding to the shoulder at low BE at 573.6 eV.The rest of the signal was deconvoluted into a Cr 3+ oxide peak and a Cr 3+ hydroxide peak at 576.0 eV and 577.3 eV, respectively [35,44].Fig. 2d) shows the Mo 3d XPS spectra for each of the three alloys.Mo has a small spin-orbit splitting of 3.15 eV resulting in an overlap of the 3d 5/2 and 3d 3/2 regions.Therefore, the whole doublet was fitted since the components could not be fitted individually as for the previous core levels of Ni 2p 3/2 , Fe 2p 3/2 and Cr 2p 3/2 .The distinct components in the Mo 3d spectra are metallic Mo at 228 eV (3d 5/2 ) and 231.15 eV (3d 3/2 ) [46], and Mo 6+ oxide at 332.2 eV (3d 5/2 ) and 235.35 eV (3d 3/2 ) [34,44].These two chemical states are not enough to fit the spectra, suggesting the presence of more oxidic chemical states.Therefore, doublets of Mo 4+ at 228.9 eV (3d 5/2 ) and 232.05 eV (3d 3/2 ), and Mo 5+ oxide at 231.1 eV (3d 5/2 ) and 234.25 eV (3d 3/2 ) are also included to fit the experimental data [47,48].Beyond that, including a doublet at higher BE 233.9 ± 0.4 eV and 237.1 ± 0.4 eV was necessary to obtain a reliable fit of the experimental data.Fitting the Mo 3d spectra without the component at high BE is shown in the Supplementary Information (SF3) to illustrate that the component at higher BE is necessary.This component can be attributed to Mo 6+ in a chemical environment different from the other peak of Mo 6+ oxide.The different chemical environment could be Mo 6+ binding to hydroxyl groups resulting in Mo 6+ hydroxide/oxyhydroxide, as in the case of Ni and Cr.Fig. 2e) shows the Nb 3d XPS spectra for alloy 625 and 718.Nb has a spin-orbit splitting of 2.7 eV resulting in an overlap of the 3d 5/2 and 3d 3/2 regions.Therefore, doublets were used to fit the Nb 3d spectra.No metallic Nb is observed, which should be located at BE 202.2 eV[49,50].The peak observed at low BE corresponds to Nb 2+ oxide at 203.1 eV (3d 5/2 ) and 205.8 eV (3d 3/2 ), and the clear doublet at BE of 206.6 eV (3d 5/2 ) and 209.3 eV(3d3 /2 ) correspond to Nb 5+ oxide [51][52][53][54].However, it is not enough to fit the spectra with only Nb 2+ and Nb 5+ oxide components.A component at higher BE is also needed to fit the shoulder detected at higher BE and what looks to be the asymmetry of the oxide peaks.We attribute this component at 207.7 eV (3d 5/2 ) and 210.4 eV (3d 3/2 ) to Nb 5+ hydroxide or oxyhydroxide.It has been shown that the BE and electronic structure for a metal core level with a specific oxidation state can change for complex mixed oxides [37,55].Further studies are needed to fully deconvolve the Mo and Nb 3d region of Ni superalloys to assign the chemical state of the small shoulders at high BE with certainty.
Fig. 3 shows the O 1s XPS spectra for each Ni alloy.A peak at 529.9 eV can be seen corresponding to O 2-in the oxide lattice, and the shoulder at higher BE at 531.1 eV corresponds to the metal hydroxide bond [11,35,56].The detected hydroxide peak in the O 1s spectra confirms the presence of hydroxide on the surface, which is also detected in the Ni 2p, Cr 2p, Mo 3d, and Nb 3d spectra.In the O 1s spectra, there is an additional shoulder at even higher BE at 533.6 eV, attributed to water [11,35,54,56].The water peak is not considered in the further analyses, and it is seen as natural of a sample exposed to an ambient atmosphere with no further sample preparation in the UHV chamber.
In a more detailed analysis, a three-layer model proposed in literature was adopted where the hydroxide forms a uniform outmost layer on top of a homogeneous oxide layer, and underneath the oxide, there is a sub-surface alloy layer having a different composition than the bulk [35,36,57,58].A schematic illustration of such a model is shown in Fig. 2f).This model is used to calculate both the composition and thickness of each layer.The area determined from fitting the XPS data of each alloying element was normalized to take into account the photoelectron cross-section and the difference in photon flux for each core level, as described in the Supplementary Information.After normalizing the photoelectron intensity, a good agreement between the experimental and tabulated values of the bulk composition was achieved, as shown in the Supplementary Information (ST2), making the quantitative analysis of the composition of the native oxide and surface region trustworthy.To calculate the composition of the sub-surface alloy layer underneath the oxide layer, the areas of the metallic components were used.To calculate the oxide composition, only the area of each oxide component was considered, and to calculate the composition of the hydroxide layer, only the area of the hydroxide components was used.Table 3 shows the calculated composition of the sub-surface layer, oxide, and hydroxide, respectively.By comparing the subsurface composition to the bulk composition, it can be seen that for all three alloys, Ni is substantially enriched, whereas Cr and Fe are depleted in the sub-surface alloy layer, and the concentration of Mo is slightly lower than in the bulk.An increasing trend in Ni enrichment with decreasing Mo and Cr content can also be seen, where alloy 59 has the lowest Ni enrichment and alloy 718 the highest.Minor amounts (unquantifiable) of Fe are found in the sub-surface layer of alloy 59 and 625 even though they have a bulk Fe content of 0.8 at% and 4.38 at%, respectively.For both alloy 59 and 625, Fe was detected in significantly higher amounts after sputtering, as seen in the Supplementary Information (SF2), suggesting that Fe segregates away from the surface region probed with XPS (depth≈20 Å) during oxidation of the bare metal surface.Nb was not detected in the metallic state for any of the alloys and is completely depleted under the oxide layer within the probing volume.When comparing the oxide composition to the bulk alloy composition, it can be seen that Cr is heavily enriched as Cr 3+ oxide in the native oxide film, while Ni does not contribute to the oxide film.Mo is also enriched in the oxide film, with a higher content in the oxide than in bulk.Nb is the element that is enriched to the highest degree and contributes the most to the oxide film relative to its bulk composition.Fe is only detected in quantifiable amounts for alloy 718 and contributes the least to the oxide except for Ni.The hydroxide layer consists mainly of Cr and Ni hydroxide, and also small amounts of Mo and Nb hydroxide were detected for alloy 625 and 718 containing Nb.
The oxide and hydroxide layer thickness was calculated from the fitted XPS data by considering the ratios between the deconvoluted peak areas of oxide, hydroxide, and metal components.The equation used is shown below [15,36,58].More details and the value of all constants can be found in the Supplementary Information.In the formula above, d is the thickness, λ M N is the electron IMFP of species M in matrix N, θ is the take-off angle of the photoelectrons (90°), N M N is the atomic density of species M in matrix N, and I M N is the normalized area of the XPS signal for species M in matrix N. The thickness of each oxide and hydroxide component, as well as the weighted average, is shown in Table 4.The variation of the calculated thickness for each chemical state in the oxide is reasonably low, making the results, analysis, and calculations trustworthy.The average oxide thickness of the three alloys is around 12 Å, with a maximum deviation of 1.8 Å.The thickness of the hydroxide layer is around 2-3 Å, determined from both the O 1s spectra and metal core level spectra from the different alloying elements.Nb and Mo  hydroxide from the thickness calculation since it was detected in small amounts, and no data for the density, molar mass, or stoichiometry is available in the literature.The thickness of the hydroxide layer was also calculated from the fitting of the O 1s spectra using two simple models: i.e. homogeneous Ni 2+ hydroxide on Cr 3+ oxide and homogeneous Cr 3+ hydroxide on Cr 3+ oxide.An average of the values from the two models gives a hydroxide thickness of around 3.5 Å, as shown in Table 4.The difference in the thickness determined from the core levels of the alloying elements and the O 1s spectra can be explained by the fact that the thickness based on the spectra of the alloying elements is that of the metal cations in the hydroxide, while the thickness determined from the O 1s spectra is that of OH -anions in the hydroxide.

Thickness and density of native oxide and sub-surface layer determined with XRR
XRR was measured for each alloy in a q-range from 0 to 1 Å -1 , as shown in Fig. 4a).The experimental data shown as blue circles were modeled using the program GenX [59], the fit is shown as a solid red line.A drop in intensity is observed at low scattering angles when the critical angle of total external reflection is surpassed.The fact that the critical angle is observable indicates that the surface is flat and well polished.The oscillations in the XRR data were fitted with a model consisting of a sub-surface alloy layer, an oxide layer, and a top layer to account for roughness, as shown in the inset in Fig. 4b).The composition of the sub-surface layer and oxide layer in the model was based on the composition determined from XPS.The top layer in the XRR model was also given the same composition as the oxide layer.The hydroxide layer was ignored in the model to fit the XRR data since it was very thin (3 Å).Fig. 4b) shows the real space model of scattering length density obtained from modeling the XRR data as a function of height, where the metal/oxide interface is set to height zero.The scattering length density is directly related to the electron density of the different layers, where metal has a higher electron density than oxide.The material's electron density depends both on the mass density and the atomic number of the elements within the material.The smooth transition between each layer is due to the roughness of each interface.A summary of the fit results is shown in Table 5.A table of additional fitting parameters is shown in the Supplementary Information (ST4).
From the XRR results, the thickness of the sub-surface layer can be determined, which was not possible with the very surface-sensitive XPS data.The results show a trend where both the density and thickness of the subsurface layer decrease from alloy 59 with a thickness of 36.6 Å and a density of 9.6 g/cm 3 to alloy 718 with a thickness of 22.5 Å and a density of 8.8 g/cm 3 .According to the XPS results, alloy 59 had the highest content of heavy elements, whereas alloy 718 had the lowest content in the sub-surface layer.Alloy 718 contained 91% Ni in the sub-surface layer, and the density of the subsurface layer for alloy 718 of 8.8 g/cm 3 matches the density of Ni 8.9 g/cm 3 almost perfectly.The scattering length density plot in Fig. 4b) shows that the sub-surface alloy layer has a gradual composition gradient towards the bulk composition indicated by the low slope in scattering length density at the interface between the bulk and sub-surface alloy layer.Based on the XRR results, all three alloys have an oxide thickness of 12-13 Å.This is in good agreement with the thickness of ~12 Å determined from XPS.A trend of decreasing density of the oxide layer can be seen from alloy 59 to alloy 718.The    3  2.0 3.5 2.6 density of oxides oxidation such as Mo 6+ and Nb 5+ is lower than, for example, Cr 3+ oxide due to the presence of more oxygen in the oxide lattice of the higher oxidation state of Mo and Nb.This gives rise to a lower overall oxide density for alloy 625 and especially 718 containing a significant amount of Nb 5+ in the oxide layer.
Regarding the nature of the top layer, it should not be a layer of hydroxide since XPS determined the hydroxide layer to be in the order of 2-3 Å, not 15-25 Å.Also, the samples were only mechanically polished and have not been UHV prepared with sputtering and annealing cycles.So, the surface is not expected to be atomically flat.In previous XRR studies where thin oxide films were grown on single-crystal substrates, such layers of low electron density were observed on top and attributed to islands and roughness [60].Therefore, the observed scattering length density in the top-layer in the present study is attributed mainly to roughness effects and potentially also to water and carbon contamination, which also was observed with XPS as seen in the survey spectra in Fig. 1.

Discussion
XPS and XRR give complementary information.XPS is sensitive to the exact chemical nature of the native oxide and the sub-surface alloy layer.XPS can also be used to calculate the thickness of the native oxide and hydroxide layer.XRR is sensitive to the thickness and electron density of the different layers in the surface region, which is helpful to distinguish the oxide layer and sub-surface alloy layer.Here XRR is used to complement and confirm the model and quantitative analysis of the XPS data.XRR also probes deeper than the XPS in the present study, making it possible to quantify the thickness of the sub-surface alloy layer and detect the gradual concentration gradient between the modified sub-surface alloy and the bulk.XPS was measured in UHV, while XRR was measured in atmospheric pressure under ambient conditions.The fact that the XPS and XRR analysis resulted in a similar value for the oxide layer thickness confirms that the native oxide is not altered upon exposure to UHV, and there is no significant effect of x-ray beam-induced damage.The highly element-specific and surface-sensitive information obtained from XPS allowed the hydroxide terminated surface to be quantified with a thickness of only 2-3 Å.This is not accessible in XRR due to surface roughness and a small difference in electron density between the oxide and hydroxide for such a thin layer.However, XPS and XRR gave an impressively similar result for the oxide layer thickness.
The thickness of the oxide layer was practically the same for all three alloys, varying only ~1 Å.From XPS, the oxide thickness was determined to be around 12 Å and from XRR 12-13 Å, differing only 1 Å.That difference is less than the radius of a chromium atom, so in practice, the XPS and XRR give the same thickness value.To put the oxide and hydroxide thickness in perspective, it can be compared to structural atomic models of pure crystalline oxide phases of Cr, Mo, and Nb oxide, and Ni and Cr hydroxide, as shown in Fig. 5.The crystallographic information is taken from the Crystallography Open Database [61], and the models are generated using VESTA [62].As shown in Fig. 5, ~2-3 Å of hydroxide should correspond to around one atomic plane of hydroxide, and 12-13 Å of oxide should correspond to a thickness of around four planes of metal cations in the oxide.As seen in the atomic model of the Cr 3+ oxy-hydroxide, it is a hydrogen-terminated oxide surface.Intuitively, the Cr 3+ hydroxide layer cannot be thicker than one monolayer, and the hydroxide must be the topmost layer since the hydrogen atom bonded to the oxide surface cannot form any more chemical bonds to continue the solid structure.However, the atomic model of Ni 2+ hydroxide displays a monolayer of Ni oxide terminated with hydrogen on each side.Such a layer cannot bond with covalent or ionic bonds to the oxide surface, as for Cr 3+ oxy-hydroxide.One explanation of how the Ni 2+ hydroxide layer could bond with the surface is through hydrogen bonding between the hydrogen of Ni hydroxide and an oxygen terminated oxide surface.These atomic models are not claimed to show the exact atomic structure of the oxides or hydroxides present on these samples, which are amorphous for X-ray diffraction.Surfacesensitive synchrotron x-ray diffraction, shown in the Supplementary Information (SF4), revealed additional peaks due to oxide phases indicating that the oxides are amorphous in the light of diffraction.The presented atomic models are meant to put the measured thicknesses in perspective and give some atomistic insight into what a hydroxide or oxyhydroxide could look like and why the hydroxide should be the topmost layer and be restricted to a thickness of one monolayer.To strengthen the argument that hydroxide must be the topmost layer, the high-resolution XPS spectra of Ni 2p measured at 1045 eV were compared to the survey measured at 1200 eV.This reveals that Ni hydroxide is present on top since the Ni hydroxide peak is larger for the spectra measured at 1045 eV, which is more surface sensitive, as shown in the Supplementary Information (SF5).
The composition of the native oxide formed on the three Ni superalloys can, to some extent, be predicted and explained by thermodynamics [63].The oxide composition determined from XPS was compared to the bulk composition to see how much each alloying element contributes to the oxide composition.This is, in turn, compared to the enthalpy of formation for each oxide [64], as shown in Fig. 6.Here the enthalpy of formation for an oxide can be seen as the affinity or driving force towards oxidation.That is, how energetically favorable it is to form the metal-oxygen bond.In Fig. 6, the enthalpy of formation is taken as the weighted average for the different oxidation states based on the XPS results.Fig. 6 shows that the oxide contribution increases with the increasing enthalpy of formation.The XPS Ni 2p spectra revealed the absence of NiO, which has the smallest (negative) enthalpy of formation, − 240 kJ/mol [65,66], compared to the oxides of the other alloying elements.Nb2O5 has the largest (negative) enthalpy of formation, and the XPS results reveal that all Nb within the XPS probing volume is oxidized, and Nb contributes the most to the oxide relative to the bulk composition.Oxidation kinetics also seem to play a crucial role in determining the composition of the native oxide film.All alloying elements should spontaneously form oxides at room temperature and ambient pressure according to thermodynamics since their enthalpy of formation is negative.Therefore, the Ni in the alloy should oxidize.However, the self-limited steady-state thickness of the oxide is just a few monolayers thick.If the kinetics for the formation of, for example, Nb oxide is much faster than Ni oxide formation, Nb will react with oxygen and may reach the self-limited thickness in a time frame where no significant amount of Ni oxide could form.It can be seen as a first-come, first-serve scenario where a limited amount of oxide can form due to the self-limiting thickness according to the Cabrera-Mott model.The oxide that will form on the surface will first and foremost contain alloying elements with the highest affinity for oxygen and the fastest kinetics for oxide formation.
Surface-sensitive synchrotron radiation techniques such as XPS and XRR are excellent tools to study thin oxides, surfaces, and interfaces and have been used in many fields of science and technology [67][68][69].Synchrotron radiation can also be a suitable tool to answer some challenging questions that are still open.Extended X-ray Absorption Near-Edge Structure (EXAFS) or Pair Distribution Function (PDF) analysis could be applied to study the distances between atoms in the oxide to better understand the atomic structure of the native oxide.XPS can also be performed in more detail, where depth profiling can be performed to further verify the proposed layered model.However, synchrotron radiation is not extensively used in the field of corrosion science.This paper demonstrates the benefits of performing well-controlled studies of the structure and chemistry of the surface region of applied materials, which has a significant impact on corrosion properties.Synchrotrons are superior to lab-based x-ray terms of photon energy resolution, divergence, coherency, and brilliance.This becomes important when studying very thin films such as the native oxide and passive film on metals and alloys.Another key benefit of using synchrotron radiation is the tunability of the x-ray energy.This tunability was utilized in the present study to vary the x-ray energy during the XPS experiment in order to keep the kinetic energy of the photoelectrons constant for each alloying element.This ensures a constant probing depth and a constant attenuation of the photoelectrons from each alloying element through the carbon contamination layer on the surface.The hemispherical analyzers used in XPS may also have different efficiency at different photoelectron kinetic energies, which can be ignored if the photon energy is varied, and constant kinetic energy of the photoelectrons is achieved for each alloying element.This allows for very well-controlled studies with high accuracy when it comes to quantitative analysis of the composition and thickness of the different layers.The agreement between the two techniques, XPS and XRR, confirms the high quality of the data that allowed the quantitative analysis, ensuring the reliability of the obtained results.This study and the proposed normalization procedure of the XPS intensity can serve as an example for further surface-sensitive studies of applied materials where detailed knowledge of the surface chemistry and accurate quantitative analysis is of importance.

Conclusions
This paper presents a unique body of results where the native oxide film on Ni superalloys was studied using synchrotron based XPS and XRR.Careful calibration of the XPS intensities allowed for accurate quantification of the thickness and chemistry of the native oxide and near-surface region.The oxide films contained mainly Cr 3+ , Mo 4,5,6+ , Fe 2+ , and Nb 5+ mixed oxides with varying compositions for the three alloys.Cr and Nb oxides were highly enriched in the oxide film and depleted in the sub-surface alloy layer.Ni oxide was not detected for any of the three alloys, and metallic Ni was enriched in the sub-surface alloy layer underneath the oxide layer.For all samples, the sub-surface alloy layer was 20-35 Å thick, while the oxide layer was 12-13 Å thick for all samples corresponding to four planes of metal cations in the oxide lattice.The high sensitivity of the applied techniques allowed us to detect a hydroxylated monolayer (2-3 Å) containing Ni 2+ , Cr 3+ , Mo 6+ , and Nb 5+ hydroxide, which terminates the oxide surface.The oxide composition could be explained with thermodynamics, where the enthalpy of formation could predict which alloying element contributes the most to the oxide film.Synchrotron radiation with high brilliance and tunable energy is shown to be an excellent tool to study the surface chemistry and thickness of the near-surface region of relevant industrial alloys.
of Competing Interest declare that they have no known competing fiinterests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Survey XPS spectra for each alloy measured using 1200 eV photons.Core levels marked with dotted boxes were measured with high resolution and described in more detail below.

Fig. 2 .
Fig. 2. XPS spectra of high-resolution core levels for each alloying element and each alloy.Black dots are experimental data, the solid red line is the fit, and the shaded areas are the deconvoluted contributions to the spectra.a) Ni 2p spectra showing Ni metal and Ni hydroxide.b) Fe 2p region showing Fe metal and Fe 2+ and Fe 3+ oxides for alloy 718 while no Fe was detected in the near-surface region for alloy 59 and 625.c) Cr 2p region showing Cr metal and Cr 3+ oxide and hydroxide.d) Mo 3d region showing Mo metal and Mo 4+ , Mo 5+ and Mo 6+ oxides and Mo 6+ hydroxide.e) Nb 3d region showing Nb 2+ and Nb 5+ oxides as well as a component attributed to Nb 5+ hydroxide.f) schematic model of the surface region.

Fig. 3 .
Fig. 3. O 1s XPS spectra for each alloy showing oxygen present on the sample surface in the form of O 2-, OH -and H 2 O.

Fig. 4 .
Fig. 4. a) XRR data shown as the scattered intensity as a function of the scattering vector q.Experimental data are shown as blue circles, and the fit is shown as a solid red line.b) Scattering length density as a function of height extracted from the fitting of XRR data, and the regions corresponding to bulk alloy, sub-surface alloy, oxide, and top-layer are indicated in the schematic model.

Table 1
Main alloying element composition of the investigated alloys in at% as provided by the supplier.

Table 3
Composition of the sub-surface alloy layer, oxide layer, and hydroxide layer determined from XPS.

Table 4
Thickness of the native oxide and hydroxide layer for each sample determined from XPS.

Table 5
Thickness and density values of the different layers for each sample determined from XRR.