Variation of the Passive Film on Compositionally Concentrated Dual-Phase Al0.3Cr0.5Fe2Mn0.25Mo0.15Ni1.5Ti0.3 and Implications for Corrosion

The compositions of three synthesized alloys are listed in Table I, including the previously studied^[12,26] Al_0.3Cr_0.5Fe₂Mn_0.25Mo_0.15Ni_1.5Ti_0.3 dual-phase CCA, henceforth referred to as the parent alloy, and two single-phase CCAs selected to match the compositions of the FCC and L2₁ constituents of the parent alloy with methods further discussed below. The compositions are used to calculate the listed thermodynamic indicators used originally for identification of high-entropy alloys.^[7,10] Equations for the thermodynamic indicators and ranges commonly used to predict the presence of disordered solid solution single-phase microstructures are shown in the electronic supplementary material.^[8‐10] The decreasing atomic radius mismatch (δ), a less negative enthalpy of mixing (ΔH_mix), and an increasing Ω parameter of the single-phase FCC matrix CCA all suggest an increased likelihood of a stable disordered single-phase microstructure, consistent with the matrix phase of the parent alloy. These parameters did not predict a disordered solid solution for the second phase L2₁ CCA, which meets expectation given the ordered nature of the alloy.

CCAs were synthesized via arc-melting from pure elements (Alfa Aesar, Cr > 99.2 pct purity, all other metals > 99.9 pct purity), flipped five times to ensure homogeneity, and suction cast into 1 cm diameter buttons in a water-cooled copper mold. Samples were annealed at 1070 °C, the temperature previously utilized to target the dual-phase microstructure, for five hours before quenching in water. Each sample was ground with SiC paper to a 1200 grit finish prior to electrochemical testing and polished with diamond paste to a 1 μm finish prior to microstructural and surface analysis.

The phases present in each alloy were identified via x-ray diffraction (XRD) on a Malvern PANalytical (Malvern, GBR) Empyrean Diffractometer^TM with Cu Kα x-rays (1468.7 eV) and at a scan rate of 0.15 deg/s. Microstructures were imaged via scanning electron microscopy (SEM) in backscattered electron (BSE) mode with an FEI (Hillsboro, OR, USA) Quanta 650^TM. The chemical compositions of the parent alloy were first identified via energy dispersive spectroscopy (EDS) in point scan and mapping mode, and analyzed with Oxford Instruments (Abingdon, GBR) Aztec^TM software. The spot sizes for SEM and EDS were selected to ensure the probe diameter was below 10 nm, well below the size of microstructural features. EDS mapping of a representative two-phase region is shown in Figure 1. The compositions identified from point scans taken over each phase were used to isolate single-phase CCA compositions from the dual-phase CCA. The composition of the point scan of the FCC matrix of the parent alloy yielded a single-phase microstructure when synthesized. However, synthesis of the composition defined by the point scan obtained over the L2₁ phase led to the formation of a dual-phase microstructure with approximately equal area fractions of FCC and L2₁ phases suggested by micrograph threshold analysis. The presence of a multi-phase microstructure in spite of a predicted single-phase microstructure was attributed to EDS spill over from the L2₁ to the FCC phase due to the x-ray generation being possible at depths of up to 1 μm,^[57] which can exceed the diameters of many of the second phase particles in the parent alloy. Thus, it is likely that some characteristic emission from the FCC matrix was detected in the L2₁ point scan, pushing the initial composition of this phase beyond regions of single-phase stability. Therefore, compositions obtained from point scans over the L2₁ phase of the equal area fraction alloy with larger regions of L2₁ was selected to represent the single-phase L2₁ CCA shown in Table I. Average compositions obtained from point scans of the single-phase alloys, parent alloy, and equal area fraction intermediate alloy along with a micrograph of the intermediate alloy are reported in Table S.1 and Figure S.1 of the electronic supplementary material, respectively.

Bild vergrößern

A series of electrochemical methods were selected to indicate the characteristics of the passivity and identify the relative resistance to localized corrosion behavior. All measurement were global and lack spatial resolution to characterize individual phases within the parent alloy. Electrochemical characterization of the synthesized single-phase alloys was thus used to evaluate the performance of the individual phases. A Gamry Instruments (Warminster, PA, USA) Reference 600 + ^TM potentiostat connected to a conventional three-electrode cell with the CCA sample cold-mounted in epoxy as the working electrode, a platinum mesh counter electrode, a saturated calomel reference electrode (SCE, + 0.241 V vs standard hydrogen electrode), and 0.01 M NaCl¹ (pH ~ 5.75) electrolyte solution was used for all electrochemical experimentation unless otherwise noted. All electrochemical experimentation was repeated threefold to ensure reproducibility. Samples were first potentiodynamically polarized using two procedures. In both cases, the scan rate was 0.5 mV/s. To evaluate the formation of the passive film, the air-formed oxide was first exposed to a − 1.3 V_SCE treatment for 600 seconds before conducting cyclic polarization between − 1.3 and 0.8 V_SCE. Additionally, to evaluate passivity and self-healing in the absence of Cl⁻ ions capable of initiating localized corrosion, potentiodynamic polarization from a cathodically pre-treated surface was evaluated in 0.1 M H₂SO₄ with a procedure further detailed in the electronic supplementary material. Second, to characterize effects of the solution-exposed air-formed oxide, the sample was exposed at open circuit potential (OCP) for 1800 seconds before conducting electrochemical impedance spectroscopy over the frequency range of 100 kHz to 1 mHz measuring 5 points/decade with a 20 mV_RMS AC voltage. All EIS spectra were fit to an equivalent circuit model discussed further below. Following EIS, the air-formed film was polarized from 0.1 V below the final OCP to 0.8 V_SCE in the upward direction followed by downward polarization from 0.8 to − 1.3 V_SCE.

To evaluate the formation and aging of a solution-formed passive film, a potentiostatic exposure procedure was used. A BioLogic SP-200^TM potentiostat was instead utilized to decrease the timestep between current density and impedance measurements with the electrochemical setup otherwise identical to the description above. The samples were first exposed to − 1.3 V_SCE for 600 seconds to reduce the effect of the air-formed passive film. A step increase in potential to − 0.25 V_SCE, determined to be within the alloys’ passive ranges during potentiodynamic polarization, was then applied for 40 ks. The impedance was monitored in situ every 200 ms using a single frequency within the capacitive range of 5 Hz and a 20 mV_RMS AC voltage. Following film growth, the film was characterized with full frequency range EIS at − 0.25 V_SCE and all other parameters maintained from the air-formed oxide evaluation (100 kHz to 1 mHz, 5 points/decade, 20 mV_RMS). The solution-formed film was subsequently characterized with x-ray Photoelectron Spectroscopy (XPS), scanning Auger Electron Spectroscopy (AES), and scanning transmission electron microscopy (STEM) as discussed further below.

The degree of microgalvanic coupling within the parent alloy was evaluated with a zero-resistance ammeter (ZRA). The FCC single-phase CCA was used as the working electrode with the L2₁ single-phase CCA of equal area as the counter electrode. Open circuit potential and current of the coupled system with air-formed oxides present were monitored for 40 ks directly after mechanical grinding of each sample at both equal area ratios (0.785 cm²) as well as a 10:1 FCC to L2₁ area ratio (0.785 to 0.0785 cm²) selected to more accurately represent the phase volume fractions. Current densities for both area ratios were normalized relative to the area of the FCC working electrode (0.785 cm²). The coupled potential was compared to the OCPs of the parent and single-phase alloys during a 40 ks exposure air-formed oxides to open circuit corrosion in the conventional three-electrode cell.

The chemical compositions and molecular constituents of the solution-formed oxide films were characterized with XPS. Samples were transported under N₂(g) directly after the 40 ks exposure at − 0.25 V_SCE, subsequent EIS characterization, and OCP exposure described above to a PHI (Chanhassen, MN, USA) VersaProbe III XPS system. Samples were exposed to lab-air for a maximum of 20 minutes at room temperature. High-resolution spectra over the Al 2p, Cr 2p_3/2, Fe 2p_1/2, Mn 2p_1/2,² Mo 3d, Ni 2p_3/2, and Ti 2p_3/2 core series were obtained with Al K_α x-rays (1,468.7 eV) at a 26 eV pass energy, 100 μm spot size, and a 45 deg take off angle. The large spot size indicates the XPS lacks the lateral resolution to probe individual phases. Thus, the synthesized phases were treated as representative of individual phases on the local scale. Spectra were calibrated with C 1s set to 284.8 eV as a reference. Spectra were deconvoluted with KolXPD^TM (Žďár nad Sázavou, CZE) software with a combination of Shirly background substitutions, Doniach–Sunjic peaks for metallic features, and Voigt functions for oxidized features with characteristic peak positions, intensities, widths, and multiplet splitting from reference spectra obtained elsewhere.^[58‐60] The Cr 3s series, which overlaps the Al 2p series, was fit to a single Voigt peak with the intensity defined by the total Cr 2p_3/2 signal adjusted with relative sensitivity factors. Surface cation fractions (X) were obtained via total intensity (I) of the peaks fit to oxidized species for each metal (M). The total intensity was normalized via relative sensitivity factors (R) as shown below in Eq. [1]. Elements for which surface cation fractions exceed the bulk microstructural composition for either a single-phase CCA or the parent alloy were considered enriched in the passive film.^[61]

To evaluate the lateral variation of the passive film chemistry, the film growth procedure described above was repeated on a polished sample before transfer to a PHI (Chanhassen, MN, USA) 710^TM Field Emission Scanning Auger Nanoprobe operating with a 10 keV, 1 nA electron beam, a beam diameter of approximately 3 to 5 nm, and a 0 deg take off angle.³ The instrument was equipped with an SEM which facilitated the location of phases for spot probe analysis. AES spectra were obtained for five point obtained over each phase and at least 1 μm from the interface and averaged to obtain representative surface cation fractions. Additionally, qualitative analysis of surface cation fractions was measured in mapping and line-scan modes to demonstrate the changes in passive film chemistry across the FCC-L2₁ interface with Eq. [1] modified to include system specific sensitivity factors.

Finally, a cross-section of the passive film was characterized via STEM-EDS. The film growth procedure was repeated on a polished sample before transfer under N_2(g) to a ThermoFischer Scientific (Waltham, MA, USA) Helios UC G4^TM Dual Beam Focused Ion Beam (FIB) and SEM system for cross-sectioning. The surface was deposited with a 0.4 μm protective Pt layer with a 5 keV electron beam followed by a 1.5 μm layer with the ion beam. The film cross-section was then prepared via FIB lift-out technique with a Ga⁺ ion beam at an accelerating voltage of 30 keV followed by subsequent passes at 5 keV. The sample was transferred to a ThermoFisher Scientific (Waltham, MA, USA) Themis^TM 60-300 kV transmission electron microscopy system and imaged in dark-field mode operating at a 200 kV and a 350 pA beam current with probe correction, allowing for resolution as low as 0.1 nm. The bulk microstructure and passive film were mapped with STEM-EDS with a Super-X detection system over a representative area near the phase interface with three line integrals obtained to evaluate the passive film depth profile over each phase and lateral homogeneity of the passive film across the phase interface. Compositions were obtained from intensities normalized with Velox software.

XRD patterns shown in Figure 2 index the matrix and second phase single-phase synthesized CCAs as FCC and Heusler (L2₁)^[62] respectively, while peaks are present for both phases in the parent alloy XRD pattern. The presence of many peaks in the single-phase L2₁ CCA that were not observed in the parent alloy is attributed to the low volume fraction of the L2₁ phase. The microstructures are confirmed by BSE micrographs, as shown in Figure 3. EDS point scans listed in Table S.1 of the electronic supplementary material confirm similar compositions between the synthesized single-phase CCAs and constituent phases within the parent alloy. The FCC CCA is shown to be single-phase, while the L2₁ single-phase CCA has small (generally less than 1 μm) regions enriched in Fe and Cr,⁴ whose formation may be attributed to spill-over inaccuracy from EDS measurements of the L2₁ phase of previous alloys, which could have incorporated signal from the FCC phase. However, given that the L2₁ volume fraction is much greater than that of the previously established 4.33 pct L2₁ area fraction for the parent alloy,^[26] the composition is assumed to be representative of the L2₁ behavior. The dual-phase microstructure consists of an FCC matrix with L2₁ features generally between 2 and 10 μm in size, well-distributed throughout the matrix. L2₁ regions generally have rounded interfaces with the matrix; however, some, such as the case of Figure 3(d), possess sharp corners and linear boundaries running in parallel with potential grain boundaries (GBs) indicated by contrast between FCC regions in the BSE micrograph.

Bild vergrößern

Bild vergrößern

E–log(i) plots comparing the corrosion behavior of the parent alloy on a global scale in 0.01 M NaCl with the single-phase CCAs are shown in Figure 4. Key values are tabulated in Table II. The use of a dilute chloride solution leads to the formation of a broad passive range for all three alloys that may be characterized prior to breakdown, while the cathodic pre-treatment minimizes the effect of the air-formed oxide. Passivity of both phases is attributable to each phase having passivating elements enriched relative to the overall alloy composition (e.g., Cr in the FCC phase, Ti and Al in the L2₁). The zero-current potential (E_i=0), defined as the potential at which the current density switches from anodic to cathodic, of all three alloys is similar. Similar Tafel slopes in the anodic and cathodic regions of each CCA suggest E_i=0 may be used to assess trends in the CCA corrosion potentials. Likewise, there is little variation in the passive current density (i_pass). For both parameters, the parent alloy resembles the values of its constituent phases. However, the lack of statistically significant variation between the alloys limits evaluation of E_i=0 and i_pass as a function of phase volume fraction. Pitting was the dominant breakdown mechanism for all three alloys with limited crevice corrosion near the sample-epoxy interface also observed. Pits disproportionately occurred near the FCC-L2₁ interface, as in the case of the micrograph shown in Figure 5. The pitting potential (E_pit) for the L2₁ phase was lower than that of the matrix phase, indicating a more modest breakdown potential in L2₁ relative to the FCC matrix. However, the interphase boundary is the weak site. Intermittent increases in current density prior to breakdown observed for all three scans suggest metastable pitting and repassivation, with the behavior most noticeable in the matrix single-phase CCA. Notably, E_pit for both the parent and FCC alloys is above the range of stability for Cr₂O₃ formed on pure Cr. Enduring passivity at these potentials is attributable to stability due to the presence of Al and Ti passive species which retain a partially protective oxide.^[12] Additionally, cyclic polarization scans shown in Figure S.2 of the electronic supplementary material also reveal a lower repassivation potential (E_rep) for the L2₁ phase. The decrease in both E_pit and E_rep for the L2₁ phase may be attributable in part to significantly lower Mo concentrations in, which has been shown elsewhere to have strong effects both of these parameters in this alloy series.^[33] Both E_pit and E_rep for the parent alloy were in between the respective values for the single-phases. The parent alloy values more closely resembled those of the matrix phase, likely resulting from its higher matrix volume fraction.

Bild vergrößern

Bild vergrößern

The corrosion behavior was also evaluated in 0.1 M H₂SO₄ to characterize passivity in the absence of localized corrosion-inducing Cl⁻ ions (electronic supplementary material Figure S.2, Table S.2). i_pass for the parent alloy is lower than both constituent phases; however, similar values for both i_pass and the critical current density suggest similar rates of passive film formation and passive film strength between the three CCAs, as in the case of polarization in NaCl.

Experiments were also conducted on the alloys with the air-formed oxide without the reduction step. The air-formed solution-exposed passive film was first characterized with EIS after 1800 seconds exposure at the film OCP in 0.01 M NaCl solution. EIS spectra of selected representative runs for each alloy are shown in Figure 6. Table III shows parameters for the fit of each spectra with the equivalent circuit model shown in Figure 6(c). The equivalent circuit model includes resistances attributable to the solution (R_S) and passive film charge transfer (R_CT), a constant phase element (CPE) used to identify the non-ideal (i.e., frequency dependent) capacitive behavior of the film. The CPE consists of admittance (Y_CPE) and CPE coefficient terms (α). A physical basis for the CPE parameters, behavior, and applications is provided elsewhere.^[63‐65] Additionally, a bounded Warburg impedance component (consisting of both an admittance term, Y, and a length-based bound, B) is used to represent finite mass-transport-limited processes, mainly diffusion and/or migration through the passive film. The goodness of fit values as well as discussion of the mathematical relationships between circuit parameters, impedance, and capacitance are provided in the electronic supplementary material. All fits suggest near-ideal capacitive behavior with the α values above 0.8 and have high R_CT values, suggesting protectiveness of the passive films for both the single-phase and parent alloys in dilute chloride solutions. While the parent alloy R_CT is slightly lower than both phase subcomponents, the low-frequency impedance modulus data shown in Table S.2 of the electronic supplementary material, a commonly used surrogate for the polarization resistance, shows that all phases possess good passivity and that differences between the CCAs are within the range of statistical scatter. Therefore, neither constituent phase is indicated to have an inferior passive film, nor can the parent alloy with interphase boundaries be shown to be superior to or inferior to either constituent phase.

Bild vergrößern

The CCAs with air-formed oxides were also characterized via polarization beginning slightly below the OCP of the films, bypassing any cathodic pre-treatment to avoid reducing the film. E–log(i) plots are shown in Figure 7 with key parameters identified in Table IV. The parent and matrix phase alloys had similar E_i=0 values, while E_i=0 for the L2₁ CCA was slightly more negative. Pitting was the dominant breakdown mechanism, with metastable pits that repassivate prior to E_pit also indicated for the parent and FCC alloys. However, formation and repassivation of metastable pits were less prominent for L2₁ single-phase CCA. The L2₁ CCA also showed a less positive E_pit, implying inferior resistance to film breakdown. E_pit for the parent alloy was in between that of the two single-phase CCAs.

Bild vergrößern

The passive film growth kinetics during exposure to a − 0.25 V_SCE potential, within the passive range of all three evaluated alloys, on a cathodically pre-treated (600 seconds, − 1.3 V_SCE) surface for each alloy are shown in Figure 8. Current density magnitudes for each alloy decrease with time after step potentiostatic hold at passive potentials, while the magnitude of the imaginary component of impedance (Z”) at an intermediate frequency, which has been previously shown to be directly proportional to film thickness,^[65,66] increases with time. For all three alloys, the current density switches from positive to negative between ~ 2000 and ~ 10,000 seconds, indicating a change from anodic- to cathodic-dominated kinetics. Prior to the transition, all three alloys show similar film growth kinetics, consistent with similar current density measurements. The anodic current densities of the single-phase L2₁ CCA are the lowest, whereas they are the highest for the parent alloy. Sharp spikes the current density measurements for the single-phase L2₁ CCA may indicate metastable film breakdown and repassivation. Z” increases with time for all alloys, indicating steady film growth and increasing protectiveness. Changes in growth rate, such as those most visible in the step-like behavior of the single-phase L2₁ CCA, may indicate different metals’ passive species and/or passive film layers form at different times.

Bild vergrößern

The film grown within the passive range was also characterized via EIS as shown in Figure 9. Selected representative spectra were fit to the equivalent circuit model shown in Figure 6(c) with the parameters listed in Table V. Similar to the case of the air-formed oxides, high α values suggest nearly ideal capacitive behavior for all three alloys. The L2₁ CCA shows different time constant behavior, with the maximum phase angle magnitude obtained at a higher frequency than the other two CCAs. This may contribute to the higher Z” values during film growth and a higher impedance modulus at 5 Hz, which is shown by the phase angle plot to be firmly within the capacitive range, despite the comparatively lower R_CT of the single-phase L2₁ CCA. Low-frequency impedance data tabulated in Table S.2 of the electronic supplementary material confirms passivity of all three alloys but suggest that deviations between the constituent phases may be considered statistically insignificant and that there is no strong evidence for the passivity of either phase being comparatively weaker.

Bild vergrößern

Possible galvanic interaction was evaluated utilizing ZRA. The galvanic couple potentials and net current measurements at the conclusion of a 40 ks coupling of the constituent phase alloys with air-formed oxides are shown in Table VI. The current densities of the coupled system following 40 ks exposure reached average values below 50 nA cm⁻² for both area ratios with maximum values generally below 750 nA cm⁻² during the exposure. Such magnitudes are well below those of i_pass for the parent alloy obtained during polarization (Table II). Following extended immersion, the current is generally negative for both area ratios, indicating the FCC matrix (working electrode) may be slightly cathodic or more passive than that of the L2₁ passive film (counter electrode). A cathodic FCC phase would suggest the galvanic couple potential following oxide exposure aging would be slightly below the OCP of the FCC phase. However, changes in the sign of the current were frequently observed, as in the case of Figure S.4 of the electronic supplementary material, indicating similarity between phases and the change in the polarizability of the passive film with exposure time. Significant variability was present with regard to the timescale and frequency of cathodic-anodic transitions. These results indicate negligible galvanic coupling of consequence. Furthermore, the galvanic couple potential is below E_pit observed during potentiodynamic polarization of the air-formed oxides for both alloys, suggesting there is insufficient driving force for galvanic coupling-induced pitting for either phase. The coupled potential may also be compared to the OCP measurements following 40 ks solution exposure for the parent alloy shown in Table VII. Both area ratios displayed galvanic couple potentials within the statistical scatter of the final OCP measurement, possibly indicating the OCP of the parent alloy is representative of the coupling of its constituent phases.

The passive film is suggested by XPS to contain passivated species for many constituent elements, consistent with the stability of multiple passive species including Al(III), Cr(III), and Ti(IV) oxides at the − 0.25 V_SCE potential in the ~ 5.75 pH environment for which the passive film was grown.^[67] Figure 10 shows the deconvolutions of high-resolution spectra collected over Al 2p, Cr 2p_3/2, Fe 2p_1/2, Ni 2p_3/2, and Ti 2p_3/2 core series for both single-phase CCAs compared to previously obtained^[12] measurements of the parent alloy. Calculated surface cation fractions are reported in Table VIII. The XPS spot size (100 μm) was significantly larger than the L2₁ regions in the parent alloy (3 to 5 μm), ensure that XPS spectra spectra for the alloy cover a representative region of the microstructure and are not disproportionately characteristic of either phase. All films have high concentrations of Ti that were suggested to take the form of Ti(IV) oxide. The single-phase FCC passive film is dominated by Cr(III) signal attributable to oxide, hydroxide, and possible Fe–Cr spinel presence while also being enriched in Al(III) and Ti(IV) relative to bulk composition. The film formed over the L2₁ phase is dominated by Ti(IV) while also being enriched in Al(III), Cr(III), and Mo(VI) relative to the bulk composition. Al(III) and Ti(IV) surface cation fractions and surface cation fractions are higher than those of the film formed over the matrix phase, while Cr(III) surface cation fractions are lower. The decreased presence of Cr in the L2₁ film follows the decreased Cr composition in the bulk L2₁ phase. Contrastingly, the high Ti(IV), Al(III), and Ni(II) signal in the L2₁ film, attributable to Ti(IV) oxide, Al(III) oxide, and Ni(II) oxide, hydroxide, and possibly Ni–Cr spinel, respectively, is likely aided by the higher concentrations of Ti, Al, and Ni in the bulk L2₁ phase. Fe, Mn, and Mo are present at low concentrations over both phases at similar magnitudes, and cannot be established as more enriched over either phase. Although Cr was fit to show a intensity of spinel in both samples, the companion metal was suggested by the Cr 2p_3/2 spectra to change from Fe in films of the parent and FCC alloys to Ni in the L2₁ film⁵. Notably, XPS deconvolution does not provide any structural information; thus, spinel deconvolutions may merely be indicative of disordered solid solution oxide nearest-neighbor interactions as opposed to long-range ordered complex oxides.^[14] Further structural characterization that would be necessary to confirm long-range ordering would require further high intensity, hard x-ray techniques.^[68‐70]

Bild vergrößern

Figure 11(b) shows AES maps characterizing the local chemical composition of the passive film formed. Distinct regions of increased Al, Ti, and Ni signal are present over the L2₁ phase regions indicated by the secondary electron SEM micrograph. Fe and Cr are more prominent in the passive film formed over the matrix where higher bulk concentrations were present, while no conclusions were obtained regarding Mn and Mo, the two elements with the lowest compositions in the bulk alloy. Similar enrichment is present in the AES line scan (Figure 11(c)) with Al, Ti, and Ni enriched over the L2₁ phase, Fe, Cr, and Mn enriched over the FCC phase, and little Mo signal observed.

Bild vergrößern

Local passive film compositions evaluated over each phase are shown in Table IX. Although both the XPS and AES point scan data indicate the film formed over the L2₁ phase has more Al, Ni, and Ti than that which is formed over the FCC phase, quantitative cation fractions severely differ. This may be attributable to the XPS-obtained cation fractions only representing signals attributed to oxidized features (e.g., oxides, hydroxides, and spinels). AES point scans signal were obtained from within the photoelectron escape depth of approximately 10 nm. However, AES did not have the necessary resolution for deconvolution and thus surface cation fractions include both metallic and oxidized features. For example, the surface cation fraction for Ni is higher over both phases when measured by AES than by XPS, but this may be inflated by the high intensity of the Ni⁰ metal peak indicated by the XPS deconvolutions. Alternatively, for elements where the majority of escaped photoelectrons observed by XPS may be fit to oxidized features such as Cr and Ti, the majority of AES signal is also likely attributable to passivated features.

The passive film chemistry was also evaluated with the HAADF-STEM image and STEM-EDS mapping shown in Figure 12. High-angle annular dark-field (HAADF) imaging reveals an oxide layer with a thickness of 2 to 3 nm. Both the HAADF image and the O EDS map indicate the film thickness is fairly consistent across the FCC-L2₁ interface. EDS mapping shows the differing phase chemistries in both bulk microstructures and their oxides. Fe and Ni show high signals in the FCC and L2₁ phases, respectively, and in the passive films grown above each phase, where a Cr-dominated passive film over the FCC phase is suggested. This is in contrast with the Ti- and Al-dominated film formed over the L2₁ phase. Differences in compositional homogeneity of the passive film are confirmed by the line profile in the horizontal direction (i.e., in-plane with the oxide surface). The concentration of Al and Ti decreased when crossing the interface from the L2₁ to the FCC phase while the concentrations of Fe and Cr were higher in the film grown over the FCC phase. The changes in passive film composition occurred over a 5 to 10 nm lateral distance before leveling out. O concentrations were roughly similar across both phases.

Bild vergrößern

Line profiles in the out-of-plane direction (i.e., perpendicular to the oxide surface), from each bulk microstructure phase though the oxide to the platinum deposition, illustrate the enriched elements in the passive film relative to the bulk microstructure. In the film grown over the L2₁ phase, Al and Ti concentrations were higher in the oxide region of the profile than in the metal region, signifying enrichment. Increasing Cr and Fe signal was also observed with the highest signal observed nearest the oxide/platinum interface, suggesting outer-layer enrichment, whereas inner-layer enrichment was suggested for Al and Ti. Ni was also shown to be present in the passive film, although at concentrations steadily decreasing with distance from the metal/oxide interface. The Ni concentration in the FCC region of the metal sharply increased near the interface but in the absence of strong O signal, demonstrating enrichment in the metal near the oxide interface. In both line profiles, limited Pt signal in the outer regions of the passive film was observed but is not expected to affect comparisons between constituent element cation fractions.

Independent passive films are formed over each phase. The composition of the passive film tracks with the bulk microstructure composition over which the film was grown, as summarized in Table X, which follows trends previously established models for passive film chemistry predicted based on bulk composition.^[61] For example, the passive film formed over the FCC phase is suggested to have higher surface cation fractions of Cr, Fe, and Mo, while the film formed over the L2₁ phase has higher surface cation fractions of Ni, Al, and Ti (Figures 11(c), 12(e), Tables VIII, IX). Passive films form quickly in solution. Z”, and therefore film thickness, rises rapidly for the first 100 seconds before reaching near-steady state values after 10 ks (Figure 8(b)). The distinct phases within the passive film along with rapid film formation suggest surface diffusion is minor compared to cation flux perpendicular to the passive film by migration in the electric field perpendicular to the interface.

Phase composition may also be compared to traditionally established metrics for desirable passive film to limit aqueous corrosion in chloride containing environments. The matrix phase bulk Cr concentration of 10 pct is below the ~ 12 at. pct limit traditionally suggested to be a minimum required for the formation of a stable Cr-dominated passive film in Fe–Cr^[71] and Ni–Cr^[72] alloys. This is noteworthy particularly given reports suggesting possible stabilization of Cr passive species at subcritical bulk Cr concentrations in Al- and/or Ti-containing CCAs.^[12,73] Alternatively, the bulk Cr concentration is much lower in the L2₁ phase (Table I). Thus, the matrix bulk composition contributes to significantly higher Cr surface cation fractions in the film formed over the matrix phase (Figures 11(c), 12(e), Tables VIII, IX), following previous reports. Likewise, the critical Al concentration thresholds must be obtained for the formation of a stable Al-dominated passive film^[74,75] are exceeded for the L2₁ phase but not for the matrix. While high bulk Al concentrations promote increased Al surface cation fractions over the L2₁ phase, Al is still present over both phases, showing stable Al presence outside of such thresholds possible in CCAs.^[12] The L2₁ phase has over three times the bulk Ti concentration as the FCC matrix, leading to higher Ti surface cation fractions, although the Ti concentrations remains below those of Fe–Ti alloys showing strong passivity).^[76‐80]

Passive film chemistry may be influenced by the thermodynamic stability of oxides for each constituent species. Al, Cr, and Ti were present in the passive films of all three alloys at higher concentrations than those of each alloy’s respective bulk composition (Table VIII). The stability of the passivated species is likely aided by favorable formation energies of the Al, Cr, and Ti oxide species listed in Table XI relative to other stoichiometric oxides. Similar trends are observed in the surface cation fractions obtained via AES (Table IX), although such values may be influenced by signal coming from unoxidized species in the bulk microstructure. Extended exposure or aging during the 40 ks film growth procedure may further promote increasing surface cation fractions of more thermodynamically favorable passive species. Gradual enrichment of the prevalent oxides may occur through preferential chemical dissolution of less stable oxides to form metal chlorides.^[81,82] Furthermore, favorable enthalpies of mixing or other beneficial interactions may contribute to Al, Cr, and Ti stability within the film.^[12] Beneficial interactions with Cr and/or Al (akin to a third element effect) may influence the high surface cation fractions of Ti within the passive film of the matrix and parent alloy, despite comparatively lower bulk concentrations (Tables I, VIII, IX). Al, Cr, and Ti presence in the passive film is further supported by the predicted stability of their passive species from E-pH diagrams of their respective pure metals, while Fe, Mn, Mo, and Ni are predicted to exists in a dissolved state at − 0.25 V_SCE in pH 5.75 conditions.^[67] For all elements, the applied potential was well above the standard reduction potentials, indicating a sufficient driving force for passivation and/or dissolution was present. While E-pH diagrams of pure metals inform the effect of environment on passivity, stability may also be affected by interactions with other metals in the alloy, either through thermodynamic or kinetic considerations.^[16] For example, despite pure Fe and Ni being suggested to dissolve, the stability of Fe and Ni in the passive film may be improved through the formation of spinel species. NiCr₂O₄ that is suggested to form over the L2₁ phase (Figure 10) has a comparable free energy of formation to pure Cr₂O₃ on both a per-atom and per-anion level (Table XI). FeCr₂O₄, which has a less negative free energy of formation than NiCr₂O₄ is instead suggested to form over the Fe-enriched FCC phase. Therefore, while free energy of formation may inform thermodynamically possible chemical species in the passive film, the local composition in bulk microstructure over which the film is formed also plays a prominent effect.

The transition between chemically distinct compositions in the passive film scale provides evidence of a heterophase interface (HI) within a passive film formed over a CCA (Figures 11, 12). Compositions for each phase are suggested to be uniform over each phase beyond a transition region of less than 10 nm. This indicates little to no localized depletion of passivating elements near either the FCC-L2₁ interface of the bulk microstructure (Figure 1) or passive film phase interface (Figures 11(c), 12(e)), limiting the risk of localized corrosion at sites hypothetically depleted in passivating elements following solutionizing heat treatments. The film thickness and limited resolution in the HAADF-STEM image (Figure 12(b)) prohibit characterization of the orientation and nanostructure of the oxides and subsequently the structural aspects of the FCC-L2₁ interface. Structural considerations may be of note given the miscibility of Fe(III) and Al(III) in corundum-like structures that likely form over the Cr(III)-enriched FCC film,^[14] whereas solubility, and therefore possible synergistic behavior, may be comparatively less likely in Ti(IV) oxides, which are generally suggested to be of rutile or amorphous structure.^[83‐85] It is unclear to what degree the crystallinity of Ti(IV) oxides may affect compatibility with the corundum-like structures suggested to form over the FCC phases. Despite such limitations, the presence of a defined HI suggests the applicability of the comparatively well-established study of polycrystalline oxides^[86] to explain electrochemical phenomena. For instance, oxygen vacancies, well established charge carriers under the point defect model,^[87] have been shown to congregate near GBs in cerium oxides and affect local electrical resistivity^[88] in addition to their well-established effect on global film-formation rates.^[89] Local structural changes and increased oxygen vacancy concentration may further alter the local valence of passivating cations such as Ti,^[90,91] despite valence changes not being observed in the global XPS measurements (Figure 10). Such effects may be enhanced by the multi-principal cation nature of films formed over CCAs. While the effects of HIs on localized corrosion, specifically pit initiation, are unclear, the nature of the interface must be considered.

Pitting is considerably more likely at phase interfaces and appears to propagate equally in all directions (Figure 5). It is difficult to determine the connection between interfacial pitting and attributes of the passive film given no significant depletion of passivating elements was observed at the interfaces (Figures 11(c), 12(e)). Despite being slightly anodic to the FCC phase, low current density magnitudes observed during galvanic coupling (Table VI) and the equal growth of pits in each direction of the interface (Figure 5) suggest that the L2₁ phase is not subject to preferential dissolution from microgalvanic coupling. However, the consistently lower E_pit and E_rep values for the L2₁ CCA may imply inferior resistance to localized corrosion in the absence of galvanic coupling (Figures 4, 7, Tables II, IV). The lack of significant microgalvanic corrosion may have been aided by the similar open circuit potentials following both 1800 seconds (Figure 7, Table IV) and 40 ks solution exposure (Table VII). Similar values indicate a reduced driving force for microgalvanic corrosion. Additionally, passivation has been shown to decrease the current of a galvanic couple,^[54] indicating the ability of both phases to independently passivate may contribute to decreased influence of microgalvanic effects.

Individual phases could be prone to preferential dissolution in dual-phase alloys if a phase is depleted in passivating elements, however, such behavior is not observed (Figure 5). Passivity of both phases is further justified by similar EIS results for both the air-formed and solution-formed passive films, with neither phase having a lower low-frequency impedance at a statistically significant level (Figures 6, 9, Tables III, V, electronic supplementary material Table S.2). In contrast, the Al-Ni rich BCC phase is strongly depleted in Cr relative to the Al_xCoCrFeNi CCA, leading to pitting and eventual preferential dissolution in chloride environments.^[23] The partitioning of Cr and Ti to different phases herein in the evaluated CCA ensures passivity over both phases, and limits preferential dissolution. Additionally, unlike previously reports for duplex stainless steel,^[46,50] the film is suggested to be of similar thickness on both sides of the HI (Figure 12(b)), which could additionally contribute to similar levels of passivity obtained over both phases.

Simultaneous passivation observed over both phases demonstrates that selecting combinations of multiple phases capable of independent passivation, such as the evaluated FCC-L2₁ microstructure, may be utilized as a strategy for the design of multi-phase, corrosion resistant CCAs. Therefore, high-throughput methods frequently utilized to narrow a broad CCA compositional space may filter compositions to target the preset phases with both desirable structures and compositions that possess the independent ability to passivate as a method to help promote corrosion resistance.^[92] Phase volume fraction in a multi-phase CCAs can be used as a tunable feature within the design process. For example, predictive measurements of mechanical properties from changing area fractions of constituent phases have also become well established within the CCA field.^[7,21,93] However, the effect of changing phase fraction on passive film chemistry and corrosion behavior must also be considered. The surface cation fractions (Figure 10, Table VIII), and by extension the corrosion behavior (Figures 4, 7), obtained for the parent alloy is generally contributed to by both constituent phases, but often not in a quantitative manner characteristic of the single-phase CCAs weighted by area fractions, as in the case of duplex stainless steel.^[55] The findings indicate that second phase area fractions and composition may not be freely adjusted to tailor mechanical properties without regard to the effects on corrosion, even if the global composition does not change. Determination of the quantitative relationships between phase area fractions, compositions, and overall corrosion behavior provides opportunity for future study.

The morphology of HIs in the passive film is also likely to significantly alter corrosion behavior. Grain size, and thus the morphology of GBs in the microstructure, has been shown to strongly affect the corrosion behavior of CoCrFeMnNi,^[94] potentially through localized thinning of the passive film.^[95] HIs, with differing passive film chemistry on either side of the boundary, initiate more significant cation segregation than single-phase GBs; however, such partitioning may be limited by the high solubility of constituent elements aided by the high entropy nature of the passive film.^[96] Ensuring the passive film possesses adequate concentrations of passivating elements in the film on both sides of the interface is demonstrated as a viable strategy for ensuring passivity in the design of corrosion resistant CCAs.^[13] While local characterization of the passive film chemistry remains inefficient, the presence of HIs at similar locations to microstructural phase interfaces (Figures 11(b), 12(b)) ensures local passive film chemistry is governed by the local composition in the bulk microstructure. Thus, the local composition within the passive film, and therefore the passivity, is heavily influenced by each phase in the bulk microstructure having concentrations of passivating elements above critical threshold values, which may inform global alloy compositions and processing techniques. Additionally, microstructural analysis of and/or predictive tools for volume fraction, location, and second phase regions may be useful as an initial predictive metric for dual-phase CCA passivity over each phase, allowing for targeting of compositions with global corrosion resistance.