Studying the Passivity and Breakdown of Duplex Stainless Steels at Micrometer and Nanometer Scales – The Influence of Microstructure

Duplex stainless steels (DSSs) consist of ferrite and austenite phases with approximately equal volume fraction. They exhibit combined superior mechanical strength and corrosion resistance, therefore are increasingly used in various applications. However, under certain conditions, passivity breakdown may occur, leading to corrosion initiation, which is often related to weak points in the heterogeneous microstructure. To understand the influence of microstructure on the passivity and breakdown of DSSs requires local probing techniques that can be used in situ, so that corrosion initiation process can be correlated to the microstructure. Recent studies employing advanced scanning probe microscopy and synchrotron-based techniques, in combination with electrochemical measurements, have contributed to a deep understanding of the passive film, passivity breakdown, and corrosion initiation of DSSs, as well as the influence of microstructural and environmental factors. This mini review presents a short summary of recent literature focusing on the studies utilizing local probing techniques and synchrotron-based analyses.


INTRODUCTION
Duplex stainless steels (DSSs) contain ferrite and austenite phases with approximately equal volume fraction. They are increasingly used in many industrial applications due to their superior mechanical properties and corrosion resistance (Charles, 1991(Charles, , 2007Nilsson, 1992;Sedriks, 1996). There are lean (low alloyed) (Charles, 2015), standard (e.g., 2205), and super and hyper (high alloyed) DSS grades (Chai and Kangas, 2016). Their corrosion resistance depends strongly on the alloying elements, especially Cr, Mo, and N, which determine the pitting resistance (Nilsson, 1992;Sedriks, 1996;Charles, 2007;Chai and Kangas, 2016). Super DSS grades (e.g., 2507) and hyper DSS grades (e.g., 3207) have a pitting resistance equivalent number over 40 and around 50, respectively. Corrosion behavior of DSSs is complicated due to the presence of phase boundaries that are preferential sites for precipitation of secondary phases/particles, and the partitioning of alloying elements, i.e., more Cr and Mo in ferrite, and more Ni and N in austenite (Combrade and Audouard, 1991;Olsson, 1995;Garfias-Mesias et al., 1996;Weber and Uggowitzer, 1998;Chai and Kangas, 2016). The type, size, and distribution of precipitate particles in the microstructure play a crucial role in the corrosion initiation (Alkire and Verhoff, 1998), and local interactions between the phases have an influence on the corrosion resistance (Combrade and Audouard, 1991;Olsson, 1995). Detailed knowledge of the microstructure, local electrochemical activities and corrosion processes occurring at micrometer and nanometer scales are needed to gain a fundamental understanding of the passivity and breakdown, and the corrosion resistance of DSSs, which is required for development of high performance stainless steels.
Traditionally, microstructure of alloys is characterized by using optical microscopy, scanning and transmission electron microscopy (SEM and TEM), energy dispersive spectroscopy (EDS), and X-ray diffraction (XRD). Corrosion behavior of the alloys in corrosive environments is evaluated by electrochemical measurements, e.g., potentiostatic/potentiodynamic polarization, and electrochemical impedance spectroscopy (EIS). Detailed information of the microstructure of DSSs can be obtained, however, traditional electrochemical measurements do not provide information of the local corrosion processes. Therefore, local probing techniques allowing in-situ measurements are needed to correlate the real time electrochemical information with the microstructure of the DSSs.

MICROSTRUCTURE AND PRECIPITATES
High levels of alloying elements are added to achieve excellent corrosion resistance and mechanical properties of DSSs. It is wellknown that Cr is the key element for passive film formation, Ni enhances the passivity and balances the ferrite-austenite phase structure, Mo and N are added to strengthen the passivity and improve the resistance to localized corrosion and they have beneficial synergistic effects (Marcus, 2012). Microstructure of DSSs, e.g., grain size and austenite spacing, may vary due to heat treatment and processing. Because of high contents of the alloying elements, precipitation of secondary phases including intermetallics such as sigma and chi phases, as well as carbides and nitrides, can occur when DSSs are subjected to improper heat treatment and processing (Chan and Tjong, 2014;Paulraj and Garg, 2015). Generally, these precipitates have an adverse effect on the mechanical properties and the corrosion resistance due to depletion of alloying elements in the area adjacent to the precipitates (Nilsson, 1992;Chan and Tjong, 2014;Paulraj and Garg, 2015;Chai and Kangas, 2016). For example, precipitation of sigma phase rich in Cr and Mo leads to depletion of these alloying elements in the surrounding matrix, and precipitation of Cr nitrides causes depletion of Cr and N in the area adjacent to the particles (Chan and Tjong, 2014). However, under what conditions (e.g., location and size of sigma phase and Cr nitrides) these precipitates become detrimental are still unanswered questions.

Characterization of Phases in DSS and Their Relative Nobility
Volta potential mapping by SKPFM was employed firstly to evaluate relative nobility of ferrite and austenite phases in DSSs, showing several tens of mV difference in Volta potential between the two phases and also a potential drop at the phase boundary (Femenia et al., 2003). The austenite exhibited a nobler Volta potential than the ferrite, in agreement with a higher dissolution rate of the ferrite next to the austenite observed by in-situ STM measurements in chloride solutions. Later, magnetic force microscopy (MFM) was combined with SKPFM for characterization of the phases in DSSs including ferrite, austenite, sigma phase, and Cr nitrides (Sathirachinda et al., 2008(Sathirachinda et al., , 2009(Sathirachinda et al., , 2010(Sathirachinda et al., , 2011. MFM maps revealed different magnetic properties of the ferrite and austenite, allowing easy phase identification (Ramirez-Salgado et al., 2013). Combined MFM and SKPFM results provided information of the location of the precipitate particles and their corrosion tendency (Sathirachinda et al., 2008;Ramirez-Salgado et al., 2013). The high lateral resolution (better than 100 nm) of MFM and SKPFM enabled characterization of corrosion propensity of micro-and nanometer sized precipitate particles and the surrounding matrix. A potential drop at the phase boundary was observed in slowly cooled 2205 DSS with sigma phase precipitate, and the depletion effects at the phase boundary were confirmed by Volta potential mapping and TEM/EDS analyses (Sathirachinda et al., 2009). Moreover, SKPFM study of 2507 DSS indicated that isothermally precipitated Cr nitrides (80-230 nm in size) at ferrite/austenite phase boundaries could be detrimental, whereas small quenchedin Cr nitrides (50-100 nm in size) formed inside the ferrite seemed to have small or no adverse effect on the corrosion behavior (Sathirachinda et al., 2010(Sathirachinda et al., , 2011.

Influence of Microstructure on Localized Corrosion Initiation
Effect of nano-sized precipitate Cr nitrides in DSSs on their corrosion behavior was studied by electrochemical polarization and in-situ AFM measurements in 1M NaCl solution (Bettini et al., 2013(Bettini et al., , 2014. 2205 DSS exhibited a passive behavior despite of quenched-in Cr nitrides present in the ferrite, and preferential ferrite dissolution only occurred under transpassive conditions. At 50 • C, selective austenite dissolution occurred, but the ferrite remained to be stable. The results show that finely dispersed quenched-in Cr nitrides in the ferrite do not cause localized corrosion, but elevated temperature can trigger localized corrosion initiation (Bettini et al., 2013). 2507 DSS at room temperature remained stable in a wide potential range of applied potential despite of presence of the Cr nitrides. At 90 • C (above the critical pitting temperature), the isothermal Cr nitrides precipitated along the phase boundaries largely reduced the corrosion resistance of the austenite, while the small quenched-in Cr nitrides reduced the corrosion resistance slightly (Bettini et al., 2014). Moreover, the low Volta potential areas in weld fused zone were found to be preferential nucleation sites for pitting initiation of 2205 DSS (Sun et al., 2018). These in-situ local probing observations of DSSs with precipitated Cr nitride particles are in general agreement with the ex-situ SKPFM mapping of the relative nobility of the phases including the precipitates. However, the Volta potential measured at room temperature did not correlate to the corrosion behavior at elevated temperatures (Bettini et al., 2013(Bettini et al., , 2014. Influence of Heat Treatment, Mechanical Loading, and Hydrogen Charging AFM/SKPFM/MFM have also been used for studies of other factors that affect corrosion behavior of DSSs. A study of the effect of annealing temperature on corrosion behavior of 2507 DSS showed that annealing at 1100 • C resulted in a greater Volta potential difference between the two phases, with the austenite exhibited a higher Volta potential than the ferrite, in agreement with preferential dissolution of the ferrite in HCl solution. Whereas, annealing at 1050 • C led to a smaller Volta potential difference and a lower dissolution rate (Guo et al., 2011). SKPFM was used to investigate hydrogen-induced pitting corrosion of 2507 DSS. The hydrogen charging resulted in low potential areas at the ferrite/austenite boundaries, associated with pitting initiation. Different effects of hydrogen on ferrite and austenite were observed and attributed to different hydrogen behavior (solubility and diffusivity) in the two phases Guo et al., 2013a). In-situ SKPFM investigation of tensile deformation of 2205 DSS revealed that the slip bands generated by tensile deformation decrease the surface potential. The work function increased in the elastic deformation stage but decreased and stabilized in the plastic deformation stage (Wang et al., 2014). Combing electron back-scattering diffraction (EBSD) and SKPFM enabled correlation between Volta potential and strain localization in the microstructure of 2205 DSS, which facilitates the understanding of the influence of cold-work and heat treatment on the corrosion behavior Engelberg, 2015, 2016;Örnek et al., 2017). Furthermore, combining digital image correlation and SKPFM allowed in-situ study of effects of hydrogen and tensile strain on local Volta potential evolution and micro-deformation of 2507 DSS, providing information helpful for understanding the effects of microstructure and strain on hydrogen embrittlement of the DSS (Örnek et al., 2018b). Besides, a theoretical effort was made by first-principle calculations, showing that the difference in the properties between ferrite and austenite can be attributed to their electron work functions (Guo et al., 2016).

Probing Semiconducting Properties of Passive Film
Recently, current sensing AFM (CSAFM) was employed to investigate semiconducting properties of passive films on 2507 DSS. From the current maps, band-gap energies could be extracted from I-V curves obtained on the ferrite and austenite areas. The results showed that the conductivity of passive film on ferrite and austenite is different and decreases with increasing film formation potential that determines the film thickness and composition (Guo et al., 2013b(Guo et al., , 2014. Hydrogen charging was found to significantly increase the conductivity of the passive film, which was higher on austenite than that on ferrite due to more hydrogen in austenite. Moreover, hydrogen charging caused an inversion of the conductivity from p-type to n-type. The highest passive film conductivity was seen at phase/grain boundaries, where pitting corrosion initiated (Guo et al., 2017;Yakubov et al., 2018). In some cases, I-V curves could be obtained through scanning tunneling spectroscopy (STS), the data provide similar information as CSAFM (Rahimi et al., 2019). Moreover, combined SKPFM and electrochemical measurements were used for characterization of passive films of DSS. Natural oxidation in air and chemical oxidation in concentrated nitric acid improved the nobility and reduced the Volta potential difference between ferrite and austenite, whereas hydrogen charging increased the Volta potential difference, resulting in a decreased nobility, more for ferrite than austenite (Örnek et al., 2019b). The same ROI was imaged in photoemission mode and defined as the area for PEEM measurement. (E) The measured ROI was later re-accessed using EBSD to allocate the crystallographic information (phases, orientation, etc.). (F) The spectra obtained were then associated with the local microstructure as exemplified by the ferrite and austenite grains with (001) orientation before and after anodic polarization in 1M NaCl [reprint with permission (Långberg et al., 2019b)].

Global Characterization of Passive Films
Passive films on metals have been studied extensively by using different analytical techniques (Strehblow, 2016;Maurice and Marcus, 2018). Both ex-situ chemical analysis techniques, e.g., X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES), and in-situ microscopic techniques (AFM and STM), have been used for characterization of composition, structure and properties of the passive films. Synchrotron based X-ray diffraction (XRD) and absorption spectroscopy (XAS) have also been employed for in-situ analyses of the surface films (Virtanen et al., 2002;Monnier et al., 2014;Watanabe et al., 2015). XPS and AES were used to study passivation of 2205 DSS, focusing on lateral and depth distribution of Mo and N in and near the passive layer. The passive film seemed to be homogeneous. The results showed Ni and N enrichment at the oxide/metal interface and also indications for Mo and N interaction both in and near the passive layer (Olsson and Hörnström, 1994;Olsson, 1995;Olsson and Landolt, 2003). Combined XPS analysis and electrochemical measurements of 2205 DSS showed that the passive film composition depends on pH of the solution and the polarization potential (Luo et al., 2011(Luo et al., , 2012. For 2507 DSS, it was found that the passive film is composed of hydroxide and oxide of Fe and Cr, with Cr oxide close to the film/metal interface. Moreover, Cr enrichment and passive film growth occur upon increasing exposure temperature (Cui et al., 2017). By using angle-resolved XPS combined with electrochemical analyses, it was shown that the composition of the passive film formed on 2205 DSS in borate buffer solutions varies with applied potential and exhibits n-type and p-type semiconductor character in different potential regions (Yao et al., 2019). The composition of passive film on 2205 DSS was substantially affected by hydrogen charging, which not only reduced the content of Cr 2 O 3 , N and O 2− species inside the passive film, but also decreased its thickness (Luo et al., 2017).

Local Analysis of Passive Films
Due to intrinsic heterogeneous microstructure of DSSs, local analysis of the passive films is needed to assess local degradation of DSSs in corrosive environments. Different approaches have been used to gain local chemical information of passive films on individual phases. AES analysis during local ion sputtering provided elemental information of the surface layer with a lateral resolution of 5 µm (Vignal et al., 2010(Vignal et al., , 2013, but no information about oxidation states of the elements. Selective etching of one phase and analyzing the remaining phase by XPS showed that the passive film composition is quite different on the two phases (Wang et al., 2015). However, the etching may have caused significant changes of the passive film. Further, single phase materials with chemical composition similar to the two phases in the DSS were analyzed by XPS, assuming that the passive film on the single phase is the same as that on the corresponding phase in the DSS (Gardin et al., 2018). This assumption is questionable because the chemical composition of the respective phases in the DSS depends on the processing and heat treatment, and the grain boundaries and residual strains may be different from the single phase due to different deformation and thermal expansion of the two phases. More suitable method is needed for local analysis of the passive films on DSSs.

Analysis of Passive Films by X-Ray Photoemission Electron Microscopy
State-of-the-art synchrotron techniques enable microscopic chemical analysis of passive films on DSS without the need for sputtering, etching or use of single-phase materials. In a recent study, air-formed oxide and anodic passive film on 2507 DSS were analyzed by using hard X-ray photoemission electron microscopy (HAXPEEM) with lateral resolution of 1 µm. By varying the photon energy, information depth could be changed in order to probe different depths of the surface region. Pre-deposited Ptmarkers in combination with EBSD allowed local analysis of the oxide film on individual grains of the ferrite and austenite, before and after electrochemical polarization (Långberg et al., 2019b). The measurement protocol showing the experimental procedures is displayed in Figure 1. Preliminarily data analysis showed a certain difference in the composition (e.g., Cr 2 O 3 content) of the surface films between the two phases, and, anodic polarization up to 1000 mV/Ag/AgCl in 1M NaCl solution led to a growth of Crand Fe-oxides, diminishing of Cr-hydroxide, and an increased proportion of Fe 3+ species (Långberg et al., 2019b).

Operando Synchrotron X-Ray Analyses of Passive Film and Passivity Breakdown
Passivity breakdown of highly-alloyed stainless steels is commonly believed to occur at a fixed potential (breakdown potential) due to further oxidation of Cr (III) in the passive film to soluble Cr (VI) species. However, the anodic current due to oxygen evolution at high potentials may be misleading in the interpretation of the measured polarization curves. In a recent work, combined with electrochemical measurements, synchrotron-based X-ray reflectivity (XRR), X-ray fluorescence (XRF), and XRD were employed, in operando, to study the passivity and breakdown of 2507 DSS in 1M NaCl solution . Schematic illustration of the experimental setup is shown in Figure 2. Combined XRR, XRD, and XRF results revealed that the passivity breakdown is in fact a continuous degradation of the passive film over a certain potential range, accompanied by enhanced Fe dissolution before rapid Cr dissolution caused by increased potential. The breakdown process involves structural and compositional changes of both the passive film and the underlying alloy surface layer, as well as selective metal dissolution depending on the applied potential . The passivity breakdown of the 2507 DSS started to occur around 1000 mV/Ag/AgCl, and Fe dissolved more on the ferrite than the austenite. Upon further increasing potential, the passive film became thicker but less dense, while the underlying alloy surface layer became denser, indicating Ni and Mo enrichment, and rapid Cr dissolution occurred at ≥ 1300 mV/Ag/AgCl .
The grazing incidence XRD (GIXRD) provided structural information of the native oxide and anodic passive film on the 2507 DSS. The native oxide was composed of nano-crystalline mixed-oxides and hydroxides. Electrochemical polarization to higher potential in the passive region resulted in film thickening, preferential Fe dissolution, and partial loss of crystallinity (Örnek et al., 2018a). Comparing mechanically polished and electropolished samples, the GIXRD results revealed that the surface strain induced by surface preparation has an influence on the passive film formation and evolution during anodic polarization. Specifically, surface strain affected crystallinity of the passive film, crystalline CrOOH diminished upon immersion in the NaCl solution, whereas crystalline FeOOH increased during anodic polarization in the passive potential range (Örnek et al., 2019a). Moreover, associated with metal dissolution, strain relaxation occurred in both austenite and ferrite grains during immersion in the solution. Polarization to transpassive regime led to maximum strain relaxation, and selective dissolution was significantly reduced due to large strains in the austenite (Örnek et al., 2019a).

CONCLUSION
DSSs possess intrinsic heterogeneous microstructure, and microor nanometer-sized particles of intermetallic phases, carbides and nitrides, etc., may form in the microstructure due to improper heat treatment and processing. Ex-situ and in-situ local probing techniques have been employed, in combination with electrochemical measurements, for characterization of the phases present in the microstructure and the composition, structure, and properties of the passive films. The microstructure components may lead to micro-galvanic effects, and the precipitate particles may cause depletion of alloying elements in the boundary regions, all have an influence on the corrosion susceptibility. Passive films on DSSs are composed of mixed oxides and hydroxides of mainly Cr and Fe, but also minor Mo-compounds. The thickness, composition, and structure of the passive film depend on the alloying contents and the formation conditions. The alloy surface layer underneath the passive film also plays an important role in the passivity. There are heterogeneities in the passive films of DSSs, which are related to the microstructure, while heat treatment, mechanical deformation, hydrogen charging, as well as pH and temperature of the aqueous media, all can alter the passive film. Weak sites of the passive film may trigger localized corrosion.
Combination of different ex-situ and in-situ local probing techniques provides complementary information about corrosion propensity and localized corrosion initiation, and enables the correlation with the microstructure. In-situ and operando measurements utilizing advanced synchrotron-based techniques yield fundamental understanding of structural, chemical and electrochemical processes, at nanometer or atomic scale, involved in the passivity and breakdown of DSSs.

AUTHOR CONTRIBUTIONS
JP wrote the manuscript.

FUNDING
Financial support by the Swedish Science Council (Vetenskapsrådet) through project Grants Nos. 2015-04490 and 2015-06092 was greatly acknowledged.

ACKNOWLEDGMENTS
I wish to thank the Ph.D. students, academic and industrial collaborators who have contributed to our own works on DSSs.