Potential Dependent Mn Oxidation and Its Role in Passivation of Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ Multi-Principal Element Alloy Using Multi-Element Resolved Atomic Emission Spectroelectrochemistry

The Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA was produced using a computational design approach and characterized in our previous work. ⁹ A Ni₇₈Cr₂₂ binary alloy was also investigated for comparison. All alloys were arc-melted, then solution heat treated at T = 1100 °C for 96 h followed by water quenching. The alloy exhibited a compositionally homogeneous single-phase FCC characterized by energy dispersive spectroscopy (EDS) mapping and X-ray diffractometry (XRD). The sample surface was initially degreased with isopropanol in an ultrasonic bath for 15 min, rinsed with deionized (DI) water (Millipore^TM, 18.2 MΩ cm), and then dried with flowing N ₂ . The sample surface was ground with SiC paper up to P4000 under DI water then dried by flowing N ₂ again. The 0.1 M NaCl electrolyte was prepared from analytical grade materials in DI water. The final electrolyte pH was adjusted to 4.0 by adding 1.0 M HCl solution. The electrolyte pH was chosen to represent a slightly acidic corrosive environment based on the thermodynamic simulation. ²⁴ At this pH, thermodynamically stable soluble species of the alloying elements as well as non-soluble species will be present in a wide potential range. This pH is aggressive to challenge and interrogate passivity under conditions that were not benign, for comparison for stainless steel. The electrolyte was deaerated by bubbling Ar gas for 30 min prior to and during the experiments. All the experiments presented in this work showed reproducible results from at least three repeated experiments.

The elemental dissolution rate of each alloy component of the MPEA during the electrochemical test was monitored in situ, by the AESEC technique. The principles and detailed calculations used in this technique are available elsewhere. ^43,48 The sample was placed vertically in a specially designed electrochemical flow cell ⁴⁹ where the released cations were transferred within the electrolyte to an Ultima 2C Horiba Jobin-Yvon inductively coupled plasma atomic emission spectrometer (ICP-AES). A reference electrode (saturated calomel electrode, SCE) and a counter electrode (Pt foil) were positioned in a reservoir separated from the working electrode by a porous membrane enabling ionic current to pass between the two compartments but preventing bulk mixing. ^43,48

A Gamry Reference 600^TM was used for the electrochemical tests. The ICP-AES data acquisition system was specially adapted to measure the analog current and potential signals from the potentiostat in the same data file as the dissolved elemental emission intensities to facilitate the direct comparison between spectroscopic and electrochemical data. ⁴³

The detailed data analysis of the AESEC technique is presented elsewhere. ⁴³ The elemental concentration, C _M , was calculated from the atomic emission intensity, I _{M, λ} , at a characteristic wavelength of the element, λ, measured by the ICP spectrometer as:

$$\begin{eqnarray}&&{{\rm{C}}}_{{\rm{M}}}=\left({{\rm{I}}}_{{\rm{M}},\lambda }-{{\rm{I}}}_{{\rm{M}},\lambda }^\circ \right)/{\kappa }_{\lambda }\end{eqnarray} \tag{ 1 }$$

where I _{M, λ} ° is the background signal and κ _λ is the sensitivity factor for a given element, obtained from a standard ICP calibration method. The elemental dissolution rate per unit area, v _M , can be calculated from C _M with the flow rate (f) of the electrolyte controlled by a peristaltic pump and the exposed surface area A as:

$$\begin{eqnarray}&&{v}_{{\rm{M}}}=f{{\rm{C}}}_{{\rm{M}}}/{\rm{A}}\end{eqnarray} \tag{ 2 }$$

It is often convenient to present the elemental dissolution rate as an equivalent elemental current density (j _M ) using Faraday's law:

$$\begin{eqnarray}&&{{\rm{j}}}_{{\rm{M}}}={{\rm{z}}}_{{\rm{M}}}\,{\rm{F}}{v}_{{\rm{M}}}/{{\rm{M}}}_{{\rm{M}}}\end{eqnarray} \tag{ 3 }$$

where M _M is the atomic weight of M, F is the Faraday constant, and z _M is the valence of the dissolving ions for oxidation such as M → M^z+ + z _M e⁻. The oxidation states of the dissolved M in 0.1 M NaCl, pH = 4 environment are assumed from thermodynamic prediction as Ni(II), Fe(II), Cr(III), Mn(II), and Co(II). ²⁴ It is often convenient to present the normalized elemental current densities (j _M ') based on the bulk composition relative to j _Ni as:

$$\begin{eqnarray}&&{{\rm{j}}}_{{\rm{M}}}^{\prime} ={{\rm{j}}}_{{\rm{M}}}\left({{\rm{z}}}_{{\rm{Ni}}}{{\rm{X}}}_{{\rm{Ni}}}/{{\rm{z}}}_{{\rm{M}}}{{\rm{X}}}_{{\rm{M}}}\right)\end{eqnarray} \tag{ 4 }$$

where X _M is the mass fraction of the element in the bulk alloy. Congruent dissolution is indicated when j _M ' = j _Ni ; otherwise non-congruent dissolution is suggested. When j _M ' < j _Ni , M is being retained as a corrosion product or does not completely oxidize while when j _M ' > j _Ni preferential dissolution of M is indicated.

The total quantity of dissolved element M, Q _M (t), may be calculated as:

$$\begin{eqnarray}&&{{\rm{Q}}}_{{\rm{M}}}\left({\rm{t}}\right)=\displaystyle {\int }_{0}^{t}{v}_{{\rm{M}}}\left({\rm{t}}\right){\rm{dt}}\end{eqnarray} \tag{ 5 }$$

The quantity of the excess element M remained at the surface (Θ _M ) relative to Ni was calculated by a mass-balance as:

$$\begin{eqnarray}&&{{\rm{\Theta }}}_{{\rm{M}}}\left({\rm{t}}\right)=\left[\left({{\rm{X}}}_{{\rm{M}}}/{{\rm{X}}}_{{\rm{Ni}}}\right){{\rm{Q}}}_{{\rm{Ni}}}\left({\rm{t}}\right)\right]-{{\rm{Q}}}_{{\rm{M}}}\left({\rm{t}}\right)\end{eqnarray} \tag{ 6 }$$

It is worth mentioning that the interference between AESEC intensity signals of Fe (259.94 nm) and Mn (257.61 nm) were assessed by an individual experiment using the ICP-AES method in the case of a standard solutions of known concentration. The interference of Fe signal to Mn was found to be negligible (< 0.6 %), therefore, Fe and Mn signals were not corrected in this work.

The samples were characterized ex situ by XPS using PHI-VersaProbe III^TM X-ray photoelectron spectrometer using an Al Kα X-rays (1468.7 eV) under analysis pressure less than 10^–8 bar. A survey of XPS was conducted with a pass energy of 224 eV to obtain the maximum number of counts. A pass energy of 26 eV was used for the high-resolution spectra. XPS spectra was analyzed with KOLXPD^TM software. Both XPS 2p and 3p core level spectra were obtained to characterize the sample surface after the electrochemical tests. The 3p core level peak analysis has an advantage in this work because the 2p peaks for Ni-Fe-Cr-Mn-Co MPEA have interference with the Auger transitions from some of the other alloying elements. ³⁹

The electrochemical impedance spectroscopy (EIS) was conducted after potentiostatic experiments (not coupling with AESEC) in the frequency range from 10⁵ to 10^–2 Hz. Potentiostatic EIS was carried out at a given potential after a potentiostatic hold for 4000 s at that potential, following to a cathodic reduction step at −1.3 V vs SCE for 600 s. The data were recorded with 8 points per decade using a 10 mV_rms sine wave perturbation. The electrolyte resistance (R _e ) was corrected from the real part of impedance (R _r ) before the Bode plot construction for graphical analysis of constant-phase element (CPE) parameters ⁵⁰ as well as model circuit fitting.

The linear sweep voltammetry (LSV) was performed with following procedure; 1) a constant potential at E _ap = −1.3 V vs SCE was applied for 600 s to minimize the effect of air-formed oxide; 2) the potential was swept from −1.3 V to 0.8 V vs SCE with 0.5 mV s⁻¹ scan rate; and 3) spontaneous dissolution rates during open circuit exposure as well as open circuit potential were recorded.

Figure 1 reports the elementally resolved polarization data enabled by coupling AESEC with conventional linear sweep voltammetry (LSV), denoted as AESEC-LSV. The results are presented for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA in deaerated 0.1 M NaCl at pH = 4.0. The normalized elemental dissolution rates (j _M ') are presented as equivalent dissolution current densities to facilitate comparison (Eqs. 3 and 4) with the net electrical current density, j _e , obtained from the potentiostat. Figure 2 shows a close-up in the vicinity of the zero-current potential (E _j=0 ) in the AESEC-LSV.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Prior to the AESEC-LSV, a constant potential of E _ap = −1.3 V vs SCE was applied for 600 s to minimize the pre-existing air-formed oxide. All five elemental dissolution rates showed two distinctive elemental dissolution rate peaks in the cathodic reduction step, indicating that the dissolution was associated with the electrochemical process. This was distinct from the perturbations of the j _e signal which may be related to the hydrogen evolution reaction as shown for other Ni-based alloys and an MPEA in a similar potential range. ^22,42

The AESEC-LSV results in the potential range from −0.7 V to 0.3 V vs SCE are magnified in Fig. 2 including active and passive potential domains. The Ni dissolution equivalent current density, j _Ni , is shown as a dashed curve to indicate what the congruent dissolution rate of each element would be assuming complete Ni dissolution as described in the experimental section. The onset of anodically polarized elemental dissolution occurred slightly below E _j=0 , near −0.46 V vs SCE. The sum of j _M (j _Σ = j _Ni + j _Fe + j _Cr + j _Mn + j _Co ) is also presented in the top panel of Fig. 2. j _e is the equivalent faradaic electrical current density obtained by potentiostat, simultaneously with the AESEC elemental dissolution rates (j _M ). The formation of oxides is suggested by j _Σ < j _e while j _Σ > j _e may be due to either a significant cathodic current or a non-electrochemical dissolution mechanism. For E > E _j=0 , two anodic peaks were observed near E = −0.25 V (a ₁ ) and E = −0.15 V vs SCE (a ₂ ). The total elemental dissolution was nearly faradaic, indicated by j _Σ ≈ j _e near a ₂ . At potentials above a ₂ , the formation of a passive film is indicated by j _Σ < j _e in the passive potential domain. Selective Ni, Fe and Co dissolution is suggested in the passive potential domain where j _M ' = j _Ni (M = Ni, Fe, and Co) while non-congruent Mn and Cr dissolution is indicated by j _M ' < j _Ni (M = Mn and Cr). The latter may be due to the accumulation of oxidized Mn, and Cr species on the surface, probably in a form of oxide or hydroxide. In the potential domain above the passive potential range (Figs. 1 and 2), Mn and Cr continued to exhibit non-congruent dissolution as shown in the passive domain.

After AESEC-LSV, the spontaneous elemental dissolution rates and the corresponding open circuit potential were recorded (Fig. 1). Monitoring elemental dissolution rate in this period can give insight into the elements joining the passive film formed during the polarization experiment and subsequent relaxation. ^22,42,51 Only Mn showed less than congruent dissolution (j _Mn ' < j _Ni ) in this period suggesting a continual build-up of Mn on the surface during the E _oc relaxation.

Two potentials in the passive potential domain were chosen from Fig. 2 (i.e., 0.0 V and 0.1 V vs SCE). AESEC chronoamperometry (AESEC-CA, Fig. 3) was conducted while the net anodic current was monitored from the potentiostat. Previously, a significant Cr enrichment in the early stage of passivation was observed at a more negative potential (E _ap = −0.25 V vs SCE) in the same electrolyte for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA. ²⁸

Zoom In Zoom Out Reset image size

Figure 3a gives the AESEC-CA at E _ap = 0.0 V vs SCE with j _M ', j _e and j _Σ measured after a cathodic potential hold for 600 s. The complete elemental oxidation was not accounted for by dissolution shown as j _e > j _Σ , which may indicate passive film formation. j _e as well as j _M approach stable values, indicative of a steady state between passive film formation and dissolution. Mn and Cr showed non-congruent dissolution with j _M ' < j _Ni (dashed line) for the initial 100 s indicating their retention on the surface. Non-congruent Cr dissolution was also observed (j _Cr ' < j _Ni ) throughout the AESEC-CA experiment indicating a continuous Cr-species surface accumulation.

For E _ap = 0.1 V vs SCE (Fig. 3b), passive film formation is again suggested by j _Σ < j _e . In this case, however, passive film dissolution or local breakdown and pitting may be faster than oxide formation as both j _e and j _M ' increased with time. Cr and Mn again showed non-congruent dissolution as both j _Cr ' and j _Mn ' were lower than j _Ni throughout the measurement. This is in agreement with the previous AESEC-LSV curve (Figs. 1 and 2) where non-congruent Cr and Mn dissolution were observed at 0.1 V vs SCE, indicating Cr- and Mn-based passive film formation.

Further insight into the surface enrichment may be obtained by the mass-balance (Θ_M) of AESEC results calculated from Eq. 6, shown in Fig. 4. At E _ap = 0.0 V vs SCE, the excess amount of oxidized Cr (Θ _Cr ) increased with time to approximately 150 ng cm⁻² after 4000 s. At E _ap = 0.1 V vs SCE (Fig. 4b) the Θ _Cr was approximately 500 ng cm⁻², and Θ _Mn ≈ 290 ng cm⁻² at the end of the experiment. The fraction of the excess alloying element at the surface during the AESEC-CA experiment is estimated from Fig. 4, and shown in Fig. 5. In both cases, Cr enrichment occurred in early stage of passivation (after several hundred seconds) in accordance with previous observation. ²⁸ Mn accumulation was much greater at E _ap = 0.1 V vs SCE where the excess Mn at the surface increased as a function of time (Fig. 5b).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The electrochemical response of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA was compared with that of a Ni-Cr binary alloy with the same at.% of Cr to investigate the effect of minor alloying elements such as Mn or Co. The electrical current density as a function of potential (LSV) is shown in Fig. 6a, and the current density decay during a constant potential hold in the passive potential domain is given in Fig. 6b. The dissolution and passivation kinetics during the potential hold experiments were investigated by potentiostatic EIS as shown in Fig. 7. The j _e in the cathodic potential domain in the LSV curve (Fig. 6a) for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA was lower than that of Ni₇₈Cr₂₂ binary alloy. The Ni₇₈Cr₂₂ binary alloy showed an anodic peak at a slightly more positive potential than that of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 6b gives the j _e decay as a function of time (in log-scale for both j_e and time) when a constant potential of either 0.0 V or 0.1 V vs SCE was applied to the two alloys after 600 s of cathodic potential hold at E _ap = −1.3 V vs SCE. The j _e decay for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA showed multiple metastable oxide breakdown events indicated by anodic perturbations in j _e for t > 80 s at both applied potentials, whereas the Ni ₇₈ Cr ₂₂ binary alloy showed much fewer and lower perturbations in j _e .

EIS was conducted after a potentiostatic hold for 4000 s at E _ap = 0.0 V and 0.1 V vs SCE, after a cathodic potential hold for 600 s at −1.3 V vs SCE. These experiments provide insights into the dissolution kinetics as well as a possible oxide layer formation verified or corroborated independently by a simplified equivalent electrical circuit model, as shown in Figs. 7 and 8. The Bode plots and fitting results are presented in Fig. 7a. Layered oxide circuit models were used to simulate the effect of a duplex oxide, as reported for a near equiatomic Ni-Fe-Cr-Mn-Co MPEA following high temperature oxidation ^52,53,54 and aqueous passivation at room temperature. ^26,27,39 For a Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA at E _ap = 0.0 V vs SCE ^27,39 and Ni₇₈Cr₂₂ binary alloy, ^41,55,56 a two time-constant circuit model with constant-phase elements (CPE) was used; CPE ₁ /R ₁ represents the inner oxide and CPE ₂ /R ₂ represents the outer oxide, shown in Fig. 8a. For the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA at E _ap = 0.1 V vs SCE, a circuit model with a relatively compact porous inner covered by an outer layer ^57,58 was utilized (Fig. 8b). The CPE _{l, 1} represents the CPE contribution of the inner oxide, R _{l, 1} is the inner oxide resistance within the pore length, CPE _dl is the double layer component at the interface in parallel with a faradaic impedance (Z _F ), and CPE _{l, 2} /R _{l, 2} represents a thin outer oxide layer. Both circuits could be reasonably well fitted to the experimental data shown in Fig. 7. For the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA at E _ap = 0.1 V vs SCE, the modulus impedance (Z _mod ) in the low-frequency domain (f < 10^–1 Hz, Fig. 7b) was three orders of magnitude lower than that observed for E _ap = 0.0 V vs SCE. This was likely due to the formation of less protective Mn oxide as previously indicated by a non-congruent Mn dissolution, Figs. 3b and 4b, at this potential. Lower Z _mod in the low-frequency domain at E _ap = 0.1 V vs SCE was also monitored for the Ni₇₈Cr₂₂ binary alloy (Fig. 7b) as compared to that at E _ap = 0.0 V vs SCE. For the Ni₇₈Cr₂₂, the difference in Z _mod between two applied potentials was only one order of magnitude. The inner oxide resistance (R _{l, 1} ) of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA at E _ap = 0.1 V vs SCE obtained from the circuit model fitting using Fig. 8b was 1.6 x 10² Ω cm² whereas that the inner oxide resistance (R ₁ ) at E _ap = 0.0 V vs SCE was 1.2 × 10⁵ Ω cm² using Fig. 8a. This may indicate a less protective oxide at E _ap = 0.1 V vs SCE compared to that at E _ap = 0.0 V vs SCE for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA.

Zoom In Zoom Out Reset image size

The surface of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA after each potentiostatic hold experiment was characterized by ex situ XPS, shown in Figs. 9 and 10. Analyzing the 2p XPS data for a Ni-Fe-Cr-Mn-Co based MPEA is difficult due to the overlap of 2p and Auger transitions. ³⁹ To this end, the high-resolution XPS 3p core level spectra were obtained from 35 to 80 eV, shown in Fig. 10.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The Cr 2p_3/2 spectra were analyzed because there is no interference between 2p Auger transitions from the other elements and 2p photoelectron signal of Cr. At both applied potentials, the Cr 2p_3/2 spectra could be fitted by Cr(0) metallic (574.2 eV), Cr(III) hydroxide (577.3 eV), and Cr(III) oxide (576.2 eV) peaks. ⁵⁹ For Mn, the XPS 2p core level spectrum near 640 eV binding energy is sometimes explored to examine Mn(II) or Mn(III), and Mn-oxide species. ^60,61 However, this binding energy largely overlaps with the Ni Auger signal, ³⁹ which makes both quantitative and qualitative analysis of Mn 2p difficult. A shoulder in the peak of the Mn 2p photoelectron spectrum near 655 eV for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA polarized to 0.1 V vs SCE was observed, which does not have any interference with Ni Auger transitions. This may indicate the presence of Mn(II) species, possibly a Mn-oxide. ⁶² This peak was not observed for the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA at E _ap = 0.0 V vs SCE. The suggestion of minimal Mn-based oxide formation at this potential is consistent with the AESEC results shown in Figs. 3 and 4.

The XPS 3p core level spectrum of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA shows distinct peaks indicative of the presence of Mn-oxide species at E _ap = 0.1 V vs SCE (Fig. 10b). A Mn metallic peak at 47.0 eV and a Mn oxide peak at 48.6 eV, possibly attributed to Mn(II), ⁶³ were similarly characterized by the XPS 3p analysis of a near equiatomic Ni-Fe-Cr-Mn-Co MPEA. ³⁹ At E _ap = 0.0 V vs SCE (Fig. 10a), only a Mn metallic peak was detected.

In this work, it is posited that Mn has a detrimental effect on the passivity of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA in a 0.1 M NaCl, at pH = 4 solution. The effect is potential-dependent and degrades the corrosion resistance of Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA compared to a Ni₇₈Cr₂₂ binary alloy under the same conditions. EIS and DC electrochemistry of Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA indicated that a change in potential from 0.0 V to 0.1 V vs SCE was more detrimental to the passive film than the same potential applied to a Ni₇₈Cr₂₂ binary alloy. As a consequence, increased instability of the passive film (j _e decay as a function of time punctuated with breakdown events, Fig. 6b) and lowered Z_mod (and the inner oxide resistance, Fig. 7) are consistent with XPS 3p characterization (Fig. 10), which corroborates with the AESEC mass-balance to indicate excess Mn at the surface (Figs. 4 and 5).

The results herein demonstrate that Mn can congruently dissolve or remain on the surface depending on the applied potential in the passive potential domain of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA in a 0.1 M NaCl at pH = 4 solution. A layered oxide is indicated by EIS analysis using a circuit model fit, similar to previous observation in the case of an equiatomic Ni-Fe-Cr-Mn-Co MPEA in 0.05 M H ₂ SO ₄ . ³⁹ The cation release detected at E _ap = 0.0 V vs SCE suggested nearly congruent elemental dissolution rates for Ni, Fe, Co, and Mn but not for Cr, accounting for Cr enrichment in the oxide (Fig. 3a). The surface did not reach a stable state at E _ap = 0.1 V vs SCE as indicated by both j _e and j _M increasing as a function of time (Fig. 3b). At E _ap = 0.1 V vs SCE, nearly congruent dissolution of Ni, Fe and Co was observed. In contrast, Cr and Mn dissolution was found to be non-congruent. For an equiatomic Ni-Fe-Cr-Mn-Co MPEA, both the natively formed oxide and electrochemically formed passive film were characterized to have an outer Cr, Fe and Co-based oxide, and an inner Cr- and Mn-based oxide as analyzed by the XPS 3p core level spectra and time of flight-secondary ion mass spectroscopy (ToF-SIMS). ³⁹ A strong segregation of Cr and Mn in the presence of O on the surface and in oxide phases is predicted for the Cr-Mn-Fe-Co-Ni MPEA by an ab initio thermodynamic simulation. ²³ The result at E _ap = 0.1 V vs SCE may be explained by this thermodynamic prediction assuming the outer Cr-, Fe-, and Co-based oxide layer dissolved at least partially at this potential whereas the inner Cr- and Mn-based oxides remained (Fig. 3b) due to the strong low segregation free energy of Cr and Mn. ²³

It should be noted that the Cr oxide formation and congruent Mn dissolution at E _ap = 0.0 V vs SCE observed in Fig. 3 is in accordance with the thermodynamic predictions for the same MPEA composition and electrolyte used in this work. ²⁴ However, the apparent build-up of Mn at E _ap = 0.1 V vs SCE contradicts the thermodynamic calculations which predicted that Mn would be completely oxidized to soluble Mn(II) over a broad potential range at pH = 4. ²⁴ The thermodynamically stable formation of (Fe, Cr) ₂ O ₃ was predicted due to the low mixing Gibbs energy of Cr ₂ O ₃ and Fe ₂ O ₃ . In contrast, Mn ₂ O ₃ mixing on the corundum lattice was not favored. ²⁴ Limited Fe ₂ O ₃ formation, if any, and non-congruent Mn dissolution was observed by AESEC at E _ap = 0.1 V vs SCE (Fig. 3b). Therefore, it can be speculated that Mn-based oxides such as MnO, Mn ₂ O ₃ , Mn ₃ O ₄ , or a MnCr ₂ O ₄ spinel are more likely possibilities in this system at this potential. Except for the spinel, these oxides are not likely mixed with Cr ₂ O ₃ . These Mn oxides may have similar free energy of formation to each other and compared to other transition metal oxides, the thermodynamic probability of the formation of one versus the other is similar. ⁶⁴ Hence, the Cr ₂ O ₃ and Mn oxides, whether MnO, Mn ₃ O ₄ or Mn ₂ O ₃ , likely exist as phase separated layers as seen during high temperature oxidation. ⁶⁵ This situation is detrimental to corrosion resistance of an oxide film. Hence the Mn oxide formed on other MPEAs and steel surfaces are not protective in aqueous solutions. ^30,31,65

Ni was observed to dissolve into solution and only minor or trace amounts of Fe and Co were found in oxides on the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA (Figs. 4 and 5). The Fe 3p spectrum of the Ni₃₈Fe₂₀Cr₂₂Mn₁₀Co₁₀ MPEA after E _ap = 0.1 V vs SCE (Fig. 10b) showed metallic and oxide state peaks at 52.8 eV and 55.2 eV, respectively. Fe dissolved congruently during AESEC-CA experiments as shown in Fig. 4 and was estimated to be accumulated at the surface at concentrations of only 50 ng cm⁻² after 4000 s at both applied potentials (Fig. 5). A small amount of oxidized Co is also seen in XPS 3p spectra. It is proposed here that remaining oxidized Fe and Co might form spinels with Cr but these cannot be a source of protection as they could not account for all the Cr which is oxidized and also would not be present in sufficient amount to cover the surface. It should be also noted that an air-formed oxide possibly formed during sample transfer to the XPS instrument. The Fe 3p oxide component after AESEC-CA experiments might be due to the Fe-based oxide formation during sample transfer after polishing, which was not maintained in a vacuum. Hence, the dominant oxides are Cr and Mn-rich and likely in a layered unmixed state with dissolution of Fe, Ni, and Co.