XPS study of the thermal stability of passivated NiCrFeCoMo multi‐principal element alloy surfaces

X‐ray photoelectron spectroscopy analysis was applied to investigate the thermal stability under ultra‐high vacuum environment of the surface oxide film formed by electrochemical passivation of a newly designed Cr15Fe10Co5Ni60Mo10 (at.%) multi‐principal element alloy and providing the alloy superior localized corrosion resistance compared to conventional stainless steels and alloys. A spectral decomposition methodology involving the subtraction of Auger peaks overlapping the Fe 2p and Co 2p core level regions was applied for quantification of the oxide film composition and thickness. The results show that, at 100°C, the passive oxide film is mainly dehydrated and dehydroxylated. Obvious loss of Ni hydroxide and conversion of Mo (VI) to Mo (IV) species are observed at 200°C, with further reduction of Mo species to Mo (III) observed at 300°C. In this temperature range, the total cation quantity in the oxide film remains stable despite the compositional alteration. At 400°C, Cr (III) oxide forms at the expense of Fe and Mo oxides, resulting in an oxide film essentially consisting of chromium oxide. At 500°C, Cr (III) oxide is eliminated, making the passive film unstable at this temperature. Possible Cr oxide removal mechanisms are discussed.


| INTRODUCTION
Unlike conventional alloys composed of one major element and one or more minor elements modulating the alloy properties, multiprincipal element alloys (MPEA), first manufactured in 2004, 1,2 are composed of at least two major elements and contain more than three elements in total. 3 The innovative alloying strategy of MPEAs, based on a predictive computational thermodynamic approach, opens up a wide and unexplored phase and composition space, which represents enormous potential for optimizing desired properties. Enhanced corrosion resistance is one of the properties being searched for. [4][5][6][7][8] Cr-containing alloys are prime examples of stainless alloys that self-protect against corrosion by virtue of surface formation of a passive oxide film. [9][10][11] This protective oxide film has been extensively studied using diverse surface analytical techniques such as XPS, AES, ToF-SIMS, and STM. [10][11][12][13][14][15][16][17] For conventional Cr-containing stainless steels and Ni-based alloys, it is generally accepted that the passive oxide films formed under atmospheric conditions are only a few nanometers in thickness and have a bilayer structure, with the inner oxide layer mainly composed of Cr oxide, ensuring the protective character. 11,12,16,18,19 It has also been suggested that, instead of stoichiometric oxides, the species in the oxide film may exist under nonstoichiometric form due to nonequilibrium solute capture. [20][21][22] The addition of Mo to Cr-containing alloys is well known to stabilize passivity and enhance the localized corrosion resistance. Mechanisms such as impeding the penetration of aggressive ions, 23,24 stabilizing the oxide components, [25][26][27] and removing adsorbed sulfur 26,28 have been proposed and are still debated. In the case of MPEAs, the literature concerning passive oxide film composition is less abundant but a stratified in-depth distribution of the alloying elements in the oxide films formed on the CoCrFeMnNi Cantor alloy and CrFeCoNiMo x alloys, 14,29 and a non-equilibrium, non-stoichiometric solid solution state oxide film formed on a NiCrFeRuMoW alloy 13 have been reported.
Great efforts have been undertaken to study the stability of passive oxide films in various environments in order to extend the range of applications of self-protected alloys. Regarding thermal stability, many works addressed the behavior of the surface oxide film formed on conventional alloys upon heating under reducing, oxidizing, or humid conditions for practical applications. [30][31][32][33][34] Some works also focused on the heating-induced alteration of the surface oxide film under inert gas atmosphere 35 or vacuum (technically low pressures) [36][37][38] in order to provide basic insight into the thermal stability of the oxide components, which is crucial for further understanding the behavior of oxide films upon heating in a specific environment. 39,40 As for surface oxide films formed on MPEAs, such investigations are still lacking. Most of the relevant works focused on the oxidation resistance at high temperature in oxidizing atmospheres, [41][42][43][44] while investigation of the thermal stability of the passive oxide films is still in its infancy. 40 In this work, we committed to investigate the thermal stability of the highly protective passive oxide films providing enhanced localized corrosion resistance to some newly-designed CrFeCoNiMo MPEAs.
Such work is a prerequisite needed for interrogating the oxide growth and ion transport mechanisms in such protective oxide films via hightemperature reoxidation experiments. These new MPEAs were designed using the CALPHAD approach and manufactured to verify whether a much more corrosion-resistant CrFeCoNiMo MPEA could be obtained by elevating the Mo content 26,45 while ensuring the structural and compositional homogeneity of the alloy. 46,47 The thermodynamic modeling, the electrochemical investigations demonstrating enhanced passivity, and the characterization of the passive oxide films formed at room temperature are reported elsewhere. 48,49 Here, we report the XPS investigation of the modifications brought by thermal heating under ultra-high vacuum (UHV) environment of the passive oxide film formed electrochemically on the  The thermal stability of the passive oxide film was studied by heating the passivated sample under UHV (residual pressure of 10 À8 -10 À7 mbar during heating) in the preparation chamber of the XPS platform. To do so, the sample was heated up from room temperature (RT) to the target temperature and maintained at this temperature for a time period of 15 min. The temperature ramp rate was $6 C/s. The tested temperatures were 100 C, 200 C, 300 C, 400 C, and 500 C.
After each heating experiment, the sample was cooled down to RT and then transferred to the XPS analysis chamber for measurement.
The highest tested temperature was 500 C at which the surface oxides were eliminated.

| XPS analytical conditions and spectral decomposition
XPS measurements were carried out in a Thermo Electron ESCALAB 250Xi spectrometer operating at a pressure of 10 À9 mbar using a monochromatic Al Kα X-ray source (hν = 1,486.6 eV). High-resolution spectra of the Cr 2p, Fe 2p, Co 2p, Ni 2p, Mo 3d, S 2p, and C 1 s core levels were recorded at photoelectron take-off angle of 90 , with a pass energy of 20 eV and a step size of 0.1 eV.
The XPS analytical methodology based on the decomposition of the 3p core level spectral region (30-80 eV BE), previously proposed for the equimolar CrFeCoNiMn Cantor alloy 29 and Al x CoCrFeNi alloys, 7 could not be employed in the present work because of the weak intensities of Co, Fe, and Mo constituents, of lower concentration, 10, 5, and 10 at.%, respectively, than in the Cantor alloy. Instead, we applied the spectral decomposition methodology previously developed for this alloy and involving the subtraction of Auger peaks overlapping the 2p core level spectra. 49 The area of the Co L 2 M 23 M 45 and Ni L 3 M 23 M 45 Auger peaks overlapping the Fe 2p core level region was determined according to the corresponding 2p 3/2 peak area and the reference Auger/2p 3/2 area ratios of 0.615 and 0.350 previously obtained for pure nickel for pure cobalt, respectively. Data processing (curve fitting) was performed with CasaXPS applying a Shirley-type background subtraction. For Ni, only the Ni 2p 3/2 spectral region was decomposed, while for Cr, Fe, Co, S, and Mo, the entire 2p or 3d core level region was decomposed due to the overlaps between the 2p 3/2 -2p 1/2 or 3d 5/2 -3d 3/2 spin-orbit doublets.

Constraints on Binding Energy (BE), Full Width at Half Maximum
(FWHM), intensity ratio and splitting of spin-orbit doublets, and line shape were applied. For the metallic components, we used in-house reference line shapes, either fitted with LF (α, β, w, m) line shapes for representing the asymmetry or without detailed decomposition when more than two peaks were needed for fitting. 49 The LF (α, β, w, m) line T A B L E 1 Parameters and assignments of the component peaks obtained by XPS curve fitting for the passivated MPEA-15Cr10Mo surface before and after heating at different temperatures Cr (0)  shape represents an asymmetric Lorentzian curve convoluted by a Gaussian curve, with α and β defining the spread of the Lorentzian tail of each side of the peak, w the damping of the tail, and m the width of the Gaussian. An in-house reference line shape was also used for Cr oxide to take into account its discrete structure resulting from multiplet splitting. 50 The other non-metallic peaks were fitted with symmetric Gaussian/Lorentzian line shape products GL(x) with x ranging from 0 to 100 and defining the percentage of Lorentzian character.
Similar to previous work, a 3d 5/2 -3d 3/2 doublet between the widely reported Mo (IV) and Mo (VI) doublets was detected and assigned to Mo (IV) in the present work. 49,51 For the sake of visibility, the two symmetric doublets of Mo (IV) are fused into an asymmetric one in the reported figures. After heating, BE shifts were observed for several components, which is also reported in the literature. 52  3 | RESULTS At 500 C, the sharp drop in the quantity of all species indicates that the passive oxide film is unstable under UHV at this temperature.
The tiny equivalent thickness of residual oxide (0.1 nm) suggests that the remaining oxide species should not form a fully covering layer.
The instability of Cr oxide at moderate temperature under dry conditions has been reported in many works 38,39,66 and attributed to the reaction: In this work, considering the ultra-low oxygen partial pressure and the enrichment of metallic Cr after the elimination of Cr oxide, another CrO 3(g) formation reaction is also suggested: Thermodynamically, at p(O 2(g) ) < 10 À7 mbar, the theoretical CrO 3 (g) pressure needed to trigger reaction (1) is lower than 10 À19 mbar, and that to trigger reaction (2) is even much lower, around 10 À42 mbar, independently of the oxygen partial pressure. Although the thermodynamic thresholds to generate CrO 3(g) are hard to reach, F I G U R E 4 Alteration mechanisms of the passivated MPEA-15Cr10Mo surface upon vacuum heating the low bulk Cr concentration, the fact that the Cr oxide in the passive oxide film is not well crystallized, 39,67 and the redeposition of CrO 3(g) in the form of Cr 2 O 3 35,36 in the colder parts of the chamber may reduce the threshold of the reactions. The possibility of having Cr oxide reduction by residual reductive gas, for example, H 2(g) or CO (g) , cannot be excluded.
On a thermodynamic basis, Cr oxide elimination via decomposition can also be considered. Theoretically, the oxygen partial pressure to trigger the direct decomposition of Cr 2 O 3(s) to Cr (s) and O 2(g) at 500 C is 10 À40 atm, which is far lower than the residual pressure in our experiment. Moreover, no evident increase of O 2 partial pressure was detected during heating. Therefore, a decomposition mechanism driven by oxygen dissolution into bulk alloy towards the thermodynamic equilibrium of M-O system is considered. For an oxygen solubility of 0.1 at.% in the alloy, 68-70 the total oxygen decrease at 500 C would lead to the oxygen dissolution into the substrate alloy reaching about 860 nm in depth, which is too deep to be achieved with an oxygen diffusion coefficient lower than 10 À13 cm 2 Ás À1 . 69,71,72 Thus, an oxide decomposition mechanism driven by oxygen diffusion into alloy is not excluded but cannot be the only mechanism of the oxide elimination observed at 500 C.

| CONCLUSION
Quantitative XPS analysis was applied to investigate the thermal sta- This work shows also that stepwise in situ XPS analysis combined with thorough quantitative data exploitation is insightful to study the thermal stability of surface oxide films. The employed XPS spectral decomposition approach is not only proven to be appropriate for the studied MPEA but also enlightening for developing a generalized systematic decomposition methodology applicable to all MPEAs, which requires intensive reference XPS measurements and proper data processing.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.