Correlative EBSD and SKPFM characterisation of microstructure development to assist determination of corrosion propensity in grade 2205 duplex stainless steel

Abstract

Correlative electron backscatter diffraction (EBSD) and scanning Kelvin probe force microscopy (SKPFM) analysis has been carried out to characterise microstructure development and associated corrosion behaviour of as-received and 750 °C heat-treated grade 2205 duplex stainless steel. High-resolution EBSD analysis revealed the presence of σ- and χ-phase, secondary austenite, Cr₂N, and CrN after ageing treatment. SKPFM Volta potential measurements confirmed the formation of discrete reactive sites, indicating local corrosion propensity in the microstructure. Cr₂N, σ-phase, and inter-granular χ-phase had the largest net cathodic activity, followed by CrN and intra-granular χ-phase showing medium electrochemical activity, with ferrite and austenite (including secondary austenite) showing net anodic activity. Corrosion screening confirmed selective corrosion of ferrite in the as-received and 750 °C-aged conditions with the corrosion propensity of secondary phases staying in-line with SKPFM observations. Stress corrosion micro-cracks were also observed and are discussed in light of microstructure corrosion propensity.

Introduction

Duplex stainless steels (DSS) offer a synergistic combination of excellent mechanical properties with enhanced corrosion resistance, and these materials are now increasingly used, such as the manufacture of containers for the storage of intermediate-level radioactive waste (ILW) [1]. Heat treatment, welding, or prolonged exposure to elevated temperatures may lead to undesired phase reactions in these high-alloyed stainless steels [2–5]. For example, the ferrite can decompose into a series of meta-stable and thermodynamically stable phases, whilst the austenite has often been stated to be unaffected [2, 4, 6]. However, phase reactions can also occur in the austenite, increasing the volume fraction of secondary phase products in the microstructure [7–13]. Components made with large wall thicknesses, therefore, usually contain intermetallic phases due to variations in cooling rates after high temperature treatments, with the core of components typically containing large volume fractions [14, 15].

Phase reactions are more favoured in the ferrite due to the enrichment of Cr and Mo, combined with far higher diffusion rates compared to the austenite [3, 4, 16]. For example, phase reactions in the temperature range between 250 and 550 °C have been known as ‘475 °C embrittlement’ [2–5, 17, 18] where in addition to the degradation of mechanical properties a significant reduction in corrosion performance has also been observed [17–21]. Phase reactions occurring in the temperature range between 600 and 1000 °C have become known as ‘σ-phase embrittlement’ where numerous secondary phases, such as Frank–Kasper phases (σ and χ) and, to some extent, R-phase can be formed [5, 10, 16, 22–35]. These are often accompanied by the precipitation of nitrides (Cr₂N and CrN) and carbides [2, 4, 5, 16, 34, 36, 37].

Determination of the identity and the volume fraction of secondary phases have typically been carried out using image analysis of optical and/or electron microscopic micrographs, often supported by analytical semi-quantitative assessment of the chemical composition by energy-dispersive X-ray (EDX) analysis. However, uncertainty and lack of precision of these measurement methods did generally not allow to build-up a comprehensive mechanistic understanding of the behaviour of intermetallic phases during corrosion processes. Investigations to quantitatively describe microstructure development of aged duplex stainless steels at elevated temperatures, especially with respect to the formation of secondary phases, has often been associated with σ- and χ-phase formation based on eutectoid decomposition reaction of the ferrite. The effects of other precipitates such as Cr-nitrides were often not taken into account due to their small volume fractions and geometrical sizes, which makes them difficult to detect [3, 4, 16, 26, 34, 38].

The electron backscatter diffraction (EBSD) technique has developed into a tool for precise mapping of the microstructure for texture, crystallographic phases, and identification of misorientation gradients with information about local plastic strain. Mapping of multiple phases over large areas with high spatial resolution of up to 10’s of nm has become possible for quantitative identification of crystallographic information to allow in- and ex situ observation of microstructure development [39]. EBSD combined with scanning Kelvin probe force microscopy (SKPFM) can provide crystallographic information about the microstructure with local Volta potential information at high spatial resolution (10’s of nm’s). The latter is a quantitative measure to describe electrochemical reactivity of a metal surface [40–44]. The knowledge about local Volta potential differences allows characterisation of corrosion processes, for example, to understand local micro-galvanic coupling which has importance in bi-phase alloys such as duplex stainless steels [13, 45–50]. Selective dissolution of the ferrite phase in duplex stainless steels, for example, can be explained by the larger Volta potential difference with respect to the Pt reference (lower absolute Volta potential), compared to a smaller measured difference of the austenite [18, 45, 46]. With the introduction of cold work, grade 2205 duplex stainless steel tends to show preferential local corrosion sites associated with localised deformation in the austenite, with these regions related to local Volta potential extremes [45, 51].

The purpose of the work reported in this paper was to link the precipitation of secondary phases after 750 °C heat treatment using multi-scale EBSD analysis, to their Volta potential differences using SKPFM. Information from correlative EBSD and SKPFM analysis provides an insight into mechanistic understanding of the atmospheric corrosion and stress corrosion cracking (AISCC) behaviour of grade 2205 duplex stainless steel microstructure.

Experimental

A solution-annealed (as-received) grade 2205 duplex stainless steel plate of 10 mm thickness was used in this work with a composition shown in Table 1. EDX spectroscopy measurements of the ferrite and the austenite phases were conducted to inform about alloy partitioning, with the results also provided in Table 1. Small miniature tensile samples were manufactured with an overall length of 50 mm, a thickness of 1 mm, with 25 mm gauge length, and 3 mm gauge width, as well as coupon samples with dimensions of 10 mm × 10 mm × 10 mm (L × W × T). All samples were aged for 5 h at 750 °C, followed by air-cooling to room temperature. The surface of these samples was mechanically ground to 4000-grit using SiC paper, followed by a 3, 1, ¼, and 0.1 µm diamond paste polishing finish. A final fine-polishing treatment using a modified OP-S suspension (OP-S with a few drops of concentrated HNO₃ and H₂SO₄) was performed for one hour to achieve a smooth, strain-free surface finish for EBSD and SKPFM analysis.

Table 1 Chemical compositions (wt%) of grade 2205 duplex stainless steel used in this study

Microstructure analysis

The microstructures of as-received and heat-treated specimens were analysed by EBSD using an FEI Magellan high-resolution scanning electron microscope (SEM). Data acquisition was performed with a Nordlys EBSD detector from Oxford Instruments with AZtec 2.2 software. An accelerating voltage of 10–20 kV was used with low current (spot size) to achieve effective spatial resolutions of 10’s of nm [39, 52]. High-resolution EBSD mapping was carried out with step-sizes between 56 and 75 nm over an area of 100 µm × 87 µm. For the identification of crystallographic phases, the parameters listed in Table 2 were used. The phase databases were included in AZtec software. Databases from HKL and Inorganic Crystal Structure Database (ICSD) were used for phase acquisition.

Table 2 Database with crystallographic geometry parameters for EBSD phase identification

Indexing rates were typically close to 99 % and at least six Kikuchi bands were selected for successful phase identification, with the minimum number of detected bands of the corresponding Kikuchi diffraction pattern for each phase given in Fig. 1. Extra care was taken for indexing secondary phases, such as CrN, Cr₂N, and χ-phase. Data acquisition was first performed using all listed phases. When secondary precipitates were indexed, the dataset was re-assessed by de-selecting individual phases and re-mapping the same area without the corresponding phase. Phase identification proved to be reliable when the previously indexed region of a secondary phase was non-indexed, which confirms and validates the indexing procedure for CrN, Cr₂N, χ, and σ-phase. Indexing of chromium carbides (Cr₂₃C₆) for example, revealed to be problematic, since the ferrite or austenite were mis-indexed when the Cr₂₃C₆ phase was selected for acquisition. EBSD assessment of Cr₂₃C₆ was, therefore, excluded in our assessment. However, carbide formation in the 750 °C temperature range is extremely retarded in modern duplex stainless steel, with nitrides usually formed in far larger quantities leading to a suppression of carbide formation [16, 34].

Fig. 1

Kikuchi diffraction bands of all indexed phases in our study. Each image pair shows unsolved (left) and solved (right) Kikuchi pattern with the minimum number of bands detected for each phase

All EBSD maps were processed using HKL Channel 5 software, and phase fractions and grain sizes were extracted. High-angle grain boundaries (HAGB’s) were determined with misorientation in excess of 15°, and low-angle grain boundaries (LAGB’s) were defined with 1°–15°. Phase maps were generated with all phases detected, including phase and grain boundaries.

Scanning Kelvin probe force microscopy

SKPFM and Magnetic Force Microscopy (MFM) measurements were carried out with a Dimension 3100 atomic force microscope (AFM) from Veeco interfaced with a Nanoscope 3a controller. Pt-coated OSCM-PT AFM probes from Olympus with 15 nm nominal radius were used to map the surface topography and the corresponding Volta potential difference of the metal with respect to the tip. The scan size of AFM map was between 10 and 80 µm, depending on the microstructural features to be characterised. The scan rate was adjusted to the tip velocity (10–30 µm/s), which corresponds to a scan rate of 0.2–0.5 Hz. The images contained 512 × 512 pixels yielding effective spatial resolution between 20 and 156 nm. All maps were processed using Nanoscope V1.5 software (Bruker). Topography/height and potential maps were flattened with 0^th flattening order to achieve best contrast within the microstructure; therefore, all data are semi-quantitative only.

In this paper, and for the probes used, higher potentials indicate a larger potential difference and correspond to anodic sites due to a larger absolute work function difference between the bias-controlled AFM tip and the microstructure feature of the grounded sample. According to this definition, cathodic sites have, therefore, lower Volta potentials than their anodic counter-parts [53]. More comprehensive explanation about the meaning of the potential and the Kelvin probe technique can be found elsewhere [43–45, 50, 54, 55].

Atmospheric-induced stress corrosion cracking testing (AISCC)

AISCC susceptibility was investigated on as-received and 750 °C heat-treated mini-tensile specimens, which were ground to 4000-grit using SiC sand papers. The heat-treated specimen, however, was afterwards electro-polished in an electrolyte of a mixture of 20 % perchloric acid and 80 % methanol at 20 V and at a temperature of −40 °C. This was done to achieve smooth surface finish for phase identification after the corrosion test.

The as-received and heat-treated mini-tensile specimens were strained to 3 and 1 %, respectively, in self-designed direct tension rigs [56]. A strain gauge was placed on the backside of each specimen and the sample extension monitored, in situ, using a LabVIEW programme. Water droplets containing MgCl₂ in different concentrations and volumes were applied onto the surface. The droplets were dispensed with an Eppendorf micropipette. Nominal deposition densities of magnesium chloride and chloride ions were calculated from area measurements as listed in Table 3. The deposited droplets changed their shape during exposure due to secondary spreading (see Fig. 2 for images after exposure); therefore, the determined ‘initial’ deposition densities most likely changed over time.

Table 3 Experimental conditions of the AISCC test

Fig. 2

Stereo-microscopy images of exposed electrolyte droplets containing magnesium chloride a–e of specimen aged at 750 °C (droplet 1–5) and f of as-received sample after exposure to 30 % relative humidity at 50 °C for 259 days

The direct tension rigs were placed in a climatically controlled KBF Binder cabinet for 259 days at 50 °C and 30 % relative humidity (RH); however, fluctuations in RH up to 60 % were observed during the last 2 weeks of exposure associated with equipment problems, therefore, the test was terminated. During the time of exposure, the direct tension rigs were periodically removed from the humidity chamber to assess whether corrosion had occurred. After terminating the test, the sample was rinsed in distilled water to dissolve salt and corrosion products, followed by an additional cleaning cycle in a 10 wt% citric acid solution at 80 °C for 2 h to remove remaining corrosion products.

Results and discussion

EBSD microstructure characterisation

The microstructure of the as-received and heat-treated condition is shown as EBSD phase maps in Fig. 3(a, b). The as-received microstructure consisted of 44 ± 2 % ferrite (δ) and 56 ± 2 % austenite (γ). The ferrite formed the matrix with the austenite present in the form of island-like discrete grains or clusters of small grains. The shape of ferrite and austenite grains was elongated due to the hot rolling process during manufacture. The austenite contained large fractions of twin boundaries, shown by the straight lines (yellow) in Fig. 3a, with interphase boundaries indicating more concave and convex shapes. The misorientation variation within ferrite and austenite grains was low. The average grain size of ferrite and austenite was 7 ± 1 and 6.5 ± 1 µm, respectively.

Fig. 3

EBSD phase maps showing a as-received microstructure and b microstructure after heat treatment at 750 °C for 5 h. Black lines are phase and high-angle grain boundaries. Note that Cr₂N and CrN are not visible at this scale (Color figure online)

The specimen aged at 750 °C in Fig. 3b contained a large fraction of secondary phases, including σ-, χ-, chromium nitrides (Cr₂N, CrN), and secondary austenite (γ₂) which have formed during the ageing treatment [16, 30]. The normalised phase fraction results are shown in Table 4. High-resolution EBSD maps allowed clear observations of the newly formed microstructure shown in Fig. 4a–g. The austenite grain boundaries at the interphase had convoluted morphologies, with newly formed secondary austenite protruding from primary austenite islands as can be seen in Fig. 4a.

Table 4 Normalised EBSD phase fractions

Fig. 4

High-resolution EBSD phase maps of the specimen aged at 750 °C for 5 h showing a σ, χ, and γ₂ formation, b intra-granular χ phase formation in ferrite with inter-granular CrN formation in ferrite and at σ/γ₂ interphase boundaries, c, d Cr₂N formation in austenite and at existing δ/γ phase boundaries, with d CrN formation at δ/γ phase boundaries, e inter- and intra-granular Cr₂N formation in ferrite, f intra-granular CrN formation in δ and χ formation at boundaries and interphases, and g CrN, χ, and Cr₂N formation at interphase boundaries

The ferrite is thermodynamically unstable and has decomposed [3, 4, 16, 34, 38], with three mechanisms proposed for secondary austenite formation where the eutectoid equilibrium δ → σ + γ₂ is the predominant reaction in the 700–900 °C temperature range. The overall fraction of secondary austenite formed was approximately 12 %, which was determined by the difference between pre- and post-ageing austenite contents. The eutectoid reaction mainly takes place at δ/γ-interphase boundaries due to increased diffusion rates, leading to segregation of Cr and Mo resulting in σ-phase formation [4, 12, 16, 26, 30, 32, 34]. The morphology of σ-phase often shows an allotriomorphic appearance due to its preferential growth along phase boundaries and towards ferrite grain interiors [57]. The surrounding regions are then depleted in Cr and Mo, resulting in the formation to secondary austenite [4, 16, 57]. The eutectoid secondary austenite can have a similar appearance to that of σ-phase, as can be seen in Figs. 3b and 4a, b. Secondary austenite in direct contact with primary austenite can also form by growth towards ferrite regions and assume coagulant-like shapes with similar grain orientation as the primary austenite, also shown in Fig. 4e.

Austenite forming onto primary austenite has also been described as new austenite in the literature [58], but in our case all newly formed austenite induced by ageing is denoted as ‘secondary austenite’ (γ₂). Both types of secondary austenite have been found to contain reduced Cr contents, particularly if the formation of Cr-rich precipitates is involved, such as σ-phase [30, 59]. The volumetric fraction of secondary austenite is typically twice that of formed σ-phase, clearly indicating faster reaction kinetics [30]. The corrosion performance of microstructures containing secondary austenite has been reported to cause reduced pitting corrosion resistance [37, 60]. Therefore, the fraction of σ-phase alone is an insufficient parameter for microstructure assessment regarding the corrosion behaviour of duplex stainless steels.

The newly formed secondary austenite had similar grain orientation to the existing austenitic phase, with neither new high-angle or low-angle grain boundaries seen in most newly formed austenite grains. The secondary austenite seemed to have developed through grain growth from primary austenite grains, indicated by the white dashed line in Fig. 4a, showing the approximate position of the initial δ/γ phase boundary. Secondary austenite formation was also observed through new austenite formation within ferrite regions associated with eutectoid σ-phase formation in the ferrite producing elongated lamellae-like shapes (allotriomorphic) also shown in Fig. 4a, b.

The decomposition of ferrite was accompanied by the formation of σ- and χ-phase. The σ-phase contained no high- and low-angle grain boundaries and showed allotriomorphic morphology. The σ-phase was located between pre-existing primary austenite islands, clearly evidencing that this phase formed by consuming ferrite. Ferrite–austenite phase boundaries and triple junctions of grain boundaries have been reported to act as nucleation sites for σ-phase [16, 22–24, 26–28, 30, 32]. The formation of σ-phase occurred at discrete regions in the microstructure, resulting in clusters of σ-phase and secondary austenite as can be seen in Fig. 3b. Furthermore, σ-phase was also found in ferrite grains, highlighted in Fig. 4c. The measured fraction of σ-phase was almost 6 % (Table 4), which is in good agreement with the work of Michalska et al. who obtained 6–7 % σ-phase in grade 2205 after ageing at 750 °C for 5 h via conventional image analysis [26]. Elmer et al. also investigated σ-phase formation in grade 2205, and ageing at 750 °C for 10 h produced 22.7 % of σ-phase determined via in situ synchrotron X-ray diffraction technique [24]. They reported 90 % completion of phase reactions after 10 h ageing, and predicted an overall fraction of 24 % of σ-phase by Thermocalc calculations [24]. The size of σ-phase was estimated in Fig. 4 using horizontal and vertical ferret diameters of 1–7 µm. However, the three-dimensional appearance may result in different ‘structural’ sizes due to the allotriomorphic shape of σ.

A total fraction of 2.1 % χ-phase was measured after ageing at 750 °C for 5 h in the microstructure, shown in Fig. 4. The χ-phase precipitated primarily at δ/γ phase boundaries, with expansion along the circumference of primary austenite grains but growth towards the ferrite. In addition, χ also consumed some austenite as can be seen in Fig. 4a–g. However, χ-phase precipitation is not restricted to ferrite or δ/γ phase boundary regions. In austenitic stainless steels, χ, σ, and other secondary phases can also form by consumption of austenite grains [10]. Therefore, similar reactions are also expected to occur in duplex stainless steels. The formation of χ-phase in grade 2205 duplex stainless steel has been reported to occur at temperature approximately 75 Kelvin below that of σ-phase formation, with fastest transformation kinetics at 750 °C [16, 30]. Therefore, χ-phase formation has been reported to be more favoured at the initial stages of ageing [16, 22, 26, 27, 30]. Padilha et al., for example, observed qualitatively larger fractions of χ than σ-phase at early ageing stages, with the χ-phase consumed by σ-phase formation after prolonged ageing treatment at 750 °C [30]. Several works have reported χ-phase as meta-stable [22, 30], although computational isothermal sections of Fe–Cr–Ni–Mo–N systems containing 22 % Cr and 5 % Ni predicted χ as a stable phase, co-existing with σ-phase and Cr₂N [16].

The size of χ-phase was estimated in Fig. 4 using horizontal and vertical ferret diameters of 0.1–2 µm. Transformation of χ occurred at multiple sites, resulting in a more homogenous formation in the microstructure compared to the clustered σ-phase appearance. The χ-phase precipitates also seemed to pin interphase boundaries, causing bulging of the newly formed secondary austenite, as shown in Fig. 4e. Moreover, it was noticed that χ-phase even nucleated within austenite grains, decorating part of a twin boundary, as shown in Fig. 4f.

Chromium nitrides, both CrN and Cr₂N, were observed after ageing at 750 °C. Both compounds can be seen in Fig. 4, with the orthorhombic CrN and the hexagonal Cr₂N decorating δ/γ, δ/γ₂, and also σ/γ₂ interphase regions. The Cr₂N had more discrete dimensions in the form of larger areas than CrN (Fig. 4). The morphology of Cr₂N was elongated and ellipsoid shaped. Their sizes varied between 130 nm to 4 µm with high aspect ratios. The CrN had smaller dimensions, typically between 80 and 620 nm, but CrN was often found in clusters as highlighted in Fig. 4f. The total fraction of Cr₂N was 0.33 % and the fraction of CrN was 0.41 %. There has been work published on Cr₂N and most is in agreement that Cr₂N is stable and co-exists with χ-phase and σ-phase precipitates [6, 10, 12–16, 27, 28, 34, 36, 48, 61, 62], but far less is known about CrN [11, 36].

The formation of Cr-nitrides is usually associated with increased ferrite contents in duplex stainless steels [11, 36, 61, 62], and large fractions of nitrides have, therefore, been observed in welds and heat-affected zones [11, 34]. The formation of Cr₂N has been reported to either occur during cooling, when ferrite is supersaturated with nitrogen leading to the formation of elongated and discretely-shaped intra-granular precipitates with $\left\langle {0001} \right\rangle_{{{\text{Cr}}_{ 2} {\text{N}}}} \left\| {\left\langle {011_{\delta } } \right\rangle } \right.$ orientation relationship, or during isothermal ageing in the 700–900 °C temperature range where inter-granular Cr₂N precipitates are formed on δδ- and δγ-boundaries [16]. In our study, Cr₂N was observed not only at interphase boundaries, but also within austenite grains indicating that nitride precipitation may be possible in austenite as well, which can be clearly seen in Fig. 4c, d.

CrN formation occurred within the ferrite, mostly on δ/δ and δ/γ grain boundaries as shown in Fig. 4b–g. No CrN was observed in the austenite, but only on γ₂/σ phase boundary regions as shown in Fig. 4b. An orientation relationship between CrN and ferrite $\left\langle {110} \right\rangle_{\text{CrN}} \left\| {\left\langle {111_{\delta } } \right\rangle } \right.$ and $\left\langle {001} \right\rangle_{\text{CrN}} \left\| {\left\langle {1\bar{1}0_{\delta } } \right\rangle } \right.$ has been reported [11]. In a grade 2505 duplex stainless steel with 0.14 % N both types of Cr-nitrides were observed, when the microstructure was cooled with rapid cooling in the range of 40–150 K/s [11]. The reported CrN precipitates showed a film- or platelet-like appearance, with Cr₂N having a rod-like shape, with the latter forming in larger sizes than CrN precipitates [56].

Both Cr-nitrides are typically enriched in Cr, N, Fe, and Mo, but Cr₂N contains more Cr than CrN, whilst the opposite holds true for the N content [36, 62]. The extent of elemental depletion zones around these precipitates usually develops as a function of element enrichment within the precipitate and the volumetric size. Since Cr₂N is more enriched in Cr and has a larger size than CrN, it is, therefore, expected that Cr₂N yields larger Cr-depleted regions, resulting in an increased electrochemical activity associated with reduced corrosion resistance [13, 36, 47, 48, 61]. This assumption will be discussed in more detail in light of the SKPFM results.

SKPFM: Volta potential measurements

Maps showing the Volta potential differences measured over regions containing ferrite and austenite in the as-received microstructure with corresponding topography and magnetic frequency maps are shown in Fig. 5a–c. Regions showing low Volta potential differences were determined to be non-ferromagnetic and hence austenitic as can be seen in the magnetic frequency map in Fig. 5c, indicating net cathodic character. Vice versa, regions of high potential with net anodic character are ferritic, showing a magnetic frequency response. The Volta potential reflects the electronic activity of a metal, and the larger the potential difference with respect to Pt the lower the actual electronic activity leading to facilitated charge transfer during electrochemical reactions [40, 41]. However, this describes thermo-dynamic equilibrium conditions, and kinetic information cannot be extracted.

Fig. 5

SKPFM analysis of as-received microstructure with a topography of a scanned area of 80 × 80 µm² containing ferrite and austenite with a line profile measurement along the white arrow given below, b the corresponding Volta potential differences map with a line profile, and c correlated magnetic frequency measured over the same area (note a small off-set of the scanned area) (Color figure online)

The ferrite showed 50–70 mV in average larger Volta potential values than the austenite indicating higher electrochemical activity. The Volta potential difference also indicated possible micro-galvanic activity between austenite and ferrite, at which the latter was expected to form the net anode whilst austenite the net cathode. The potential variation across ferrite and austenite interphases was smooth with only minor potential gradients. A smooth potential gradient is indicative for non-heterogeneous activity, such as selective attack. Earlier work on grade 2205 duplex stainless steel clearly demonstrated the susceptibility of ferrite when exposed to chloride-containing environment, with the introduction of plastic deformation leading to local Volta potential extremes (hot spots) [63].

A Volta potential differences map with topography to screen the 750 °C-aged microstructure is given in Fig. 6a, b. The corresponding SEM image with EBSD phase map of the same region is given in Fig. 6c, d. Volta potential differences over ferrite and austenite in certain regions were nearly similar, which means that the driving force for galvanic coupling between those regions is reduced. However, local Volta potential extremes were developed indicating enhanced electronic activity and galvanic coupling between net anodic and net cathodic regions. The Volta potential variations within ferrite and austenite regions increased, indicating higher micro-galvanic activity within each phase. At this magnification, potential differences over secondary phases could not be discerned, and the mapped area in Fig. 6 was, therefore, re-scanned with higher resolution as shown in Figs. 7 and 8.

Fig. 6

SKPFM analysis of aged microstructure with a topography map, b corresponding Volta potential map, c corresponding SEM micrograph in secondary electron imaging mode, and d corresponding EBSD phase map showing γ/γ₂ (blue), δ (red), χ-phase (green), σ-phase (aqua), Cr₂N (yellow), CrN (white), and grain/phase boundaries (black). Highlighted region in c, d indicate the scanned area (Color figure online)

Fig. 7

Medium-resolution SKPFM analysis of aged microstructure: a Volta potential map, b, c measured line profiles (1, 2) in a, d corresponding SEM image, and e corresponding EBSD phase map. The large highlighted area in e shows the SKPFM measured region in a (Color figure online)

Fig. 8

High-resolution SKPFM analysis of an area containing interphase χ (IP-χ) and CrN: a Volta potential map, b–d measured line profiles (1, 2, 3 in a), e 3D height map overlaid with Volta potential data, and f corresponding EBSD map (Color figure online)

Figure 7 gives the Volta potential with corresponding EBSD map of an area containing intermetallic phases, showing numerous discrete regions with low and high potential variations and gradients. The line profile 1 in Fig. 7a, b, e shows that Cr₂N precipitates have the lowest Volta potential followed by austenite regions, with the latter possibly secondary austenite. This indicates low electronic activity of these phases. However, large potential gradients were found surrounding Cr₂N precipitates which gives an indication of local micro-galvanic cells with enhanced electronic coupling activity. Such large gradients adjacent to second phase precipitates are indicative of element depletion zones; hence, the adjacent regions often form local anodes. Cr₂N has been reported to contain higher Cr concentrations than CrN and χ, most likely resulting in concentration gradients during thermal treatments [11, 36]. Volta potential gradients measured adjacent to CrN and χ precipitates were far lower than those measured adjacent to Cr₂N. Sathirachinda et al. investigated the effect of thermally treated microstructure of grades 2205 and 2507 duplex stainless steel with SKPFM [13], measuring potential difference of 10–15 mV between γ and Cr₂N with respect to a Pt–Ir tip. The study concluded that Cr₂N has the highest practical nobility in the microstructure, which would mean net cathodic behaviour, in-line with our observations.

The χ-phase showed variations of its electronic behaviour over different areas, with some regions indicating net cathodic activity, whereas others showed net anodic character as can be seen in Fig. 7a, e. The χ-phase at δ/γ boundaries was denoted as ‘interphase/inter-granular-χ’ (IP-χ) and the χ-phase within ferrite labelled as ‘in-ferrite/intragranular-χ’ (IF-χ). The IP-χ had more net cathodic character, whilst IF-χ seemed to be net anodic, suggesting that they have different chemical compositions, with the IP-χ phase probably more enriched in Cr and Mo. However, the potential gradients adjacent to IP-χ were far steeper than those observed surrounding IF-χ, indicating the presence of depletion effects.

High-resolution Volta potential mapping was, therefore, performed on a region containing γ/γ₂, χ-phase, and CrN (Fig. 8). Small finger-like globular features were apparent which were indexed by EBSD as CrN precipitates and shown in Fig. 8e, f. The CrN particles were next to χ-phase precipitates, embedded in the ferrite. Line profile measurements across the CrN precipitates showed a potential difference of 5-10 mV higher than the surrounding ferrite phase, with the depleted areas around these precipitates indicating 5–6 mV lower potentials than the ferrite. This clearly indicated that depleted regions would behave net anodic, whilst the CrN would behave net cathodic with respect to the ferrite phase. The IP-χ in Fig. 8 had 30–40 mV larger Volta potentials than the IF-χ shown in Fig. 7 despite their comparable dimensions. This potential difference supported different chemical compositions of this phase. Seemingly, the cathodic character was not only limited to the χ-phase.

The potential differences measured over all crystallographic phases are shown in Fig. 9. The Cr₂N phase showed the largest net cathodic behaviour, together with σ and IP-χ. The intra-granular IF-χ phase showed net anodic properties whilst the IP-χ phase was more cathodic. Therefore, both χ-phase compositions are expected to behave differently. The nobility of CrN seemed to lie in-between the matrix phases and most other secondary phases. Volta potential values of ferrite and austenite had large scatters, with ferrite showing slightly more net anodic character than austenite. The scatter is partly due to large micro-galvanic activities observed within both matrix phases.

Fig. 9

Comparison of relative Volta potential differences of all observed phases after ageing in descending order from left (ferrite) to right (Cr₂N)

Stress corrosion screening tests

As-received microstructure

The as-received mini-tensile specimen dosed with 1450 µg/cm² MgCl₂ was exposed to 50 °C and 30 % RH, and after removing and cleaning the sample substantial corrosion attack was observed, which can be seen in Fig. 10a. Primarily the ferrite was corroded under the droplet deposit with some minor corrosion observed on the austenite. The same corrosion pattern has previously been reported under similar exposure conditions [45, 63, 64], with ferrite clearly the electrochemically more active phase leading to selective dissolution [17, 18, 44, 45, 56, 64–72]. The ferrite phase is, usually, the electrochemically more active phase in 2205 duplex stainless steel due to its lower corrosion potential in contrast to the austenite in mild chloride-bearing environments leading to micro-galvanic coupling between ferrite and austenite [66, 70, 73]. Selective corrosion of the ferrite, therefore, is often manifested as the main corrosion mechanism whilst the austenite is galvanically protected [17, 63, 72, 74–77].

Fig. 10

Corrosion morphology of as-received microstructure after exposure to 1450 µg/cm² MgCl₂ showing a the entire corroded area, b localised intra-granular corrosion attack on austenite, c, d stress corrosion micro-cracks on austenite, e multiple corrosion sites on austenite with a crack inside a crack (within the highlighted area), f selectively dissolved ferrite regions with multiple cracks on austenite. The stress acted along the horizontal direction

High-resolution SEM assessment confirmed the presence of localised attack in the form of superficial sub-micrometre sized intra-granular corrosion pits on the austenite, with typical images shown in Fig. 10b, e. The attack on the austenite seemed to be related to slip bands and strain. The evolution of strain localisation in correlation with the development of local Volta potential extremes in austenite leading to enhanced propensity to localised corrosion with the introduction of cold deformation was earlier demonstrated [45]. Pitting corrosion on the austenite tended to suppress the selective dissolution of the ferrite. Seemingly, heterogeneous nucleation of numerous discrete corrosion pits on the austenite indicated those sites as more susceptible, possibly associated with strain localisation.

Some stress corrosion micro-cracks were also found in the austenite, oriented perpendicular to the applied stress direction (stress acted horizontal in Fig. 10c, e, f). The longest crack observed was less than 15 µm in length. Nucleation sites of cracks within existing cracks were also found, suggesting that some cracks may have stifled and stopped growing, before conditions for re-nucleating inside existing cracks were satisfied again. This is different from the classic stress corrosion cracking theory where cracks propagate with discrete steps. Figure 10e shows features related to crack nucleation inside an existing crack.

Retardation effects may have played an important part in crack development, for example, when the crack encounters γ/γ grain boundaries or δ/γ interphase boundaries. The stress corrosion micro-cracks observed were all transgranular in nature, and located in the austenite phase, as concluded from EBSD analysis seen in Fig. 11. Slip planes are easily noticed from SEM images, which play an active role in crack nucleation and growth. The crack in Fig. 11 initiated on a grain with close 101 orientation and grew towards the grain interior before changing its direction slightly after encountering a boundary. A difference in crack opening is also observed with the crack wide open on one side of the grain boundary, and more tightly closed on the other.

Fig. 11

Corrosion morphology of as-received microstructure: a selective corrosion of ferrite (partially dissolved only) and b selective stress corrosion cracking of austenite with c corresponding EBSD phase map overlapped with band contrast map (austenite is blue and ferrite is red coloured), and d corresponding inverse pole figure map in x direction. Step size = 135 nm (Color figure online)

Heat-treated microstructure

Figure 12a shows the specimen aged at 750 °C and exposed to 14.5 μg/cm² chloride for 259 days. Minor corrosion attack was only observed located at interphases and on the austenite, with some attack around secondary phases. Increasing the chloride concentration resulted in more severe attack as can be seen with a deposition density of 1450 μg/cm² chloride (drop 3) in Fig. 12b. The attacked area and corroded volume of the 750 °C-aged sample was significantly larger than that observed on the as-received specimen exposed with equivalent deposition density of chloride in Fig. 10a. However, no evidence for stress corrosion micro-cracking was found in these specimens. Stress corrosion cracking can only occur when the corrosion rate and crack velocity in that corrosion system have similar order of magnitude, but when corrosion reactions advance too fast, then crack nucleation and propagation are hindered.

Fig. 12

Corrosion morphology of the aged specimen with 1 % elastic–plastic strain after 259 days exposure at 50 °C and 30 % RH: a minor corrosion attack under drop 1 (14.5 µg/cm²) and b corrosion area under drop 3 (1450 µg/cm²)

The corrosion morphology of the aged sample after exposure to 145 µg/cm² chloride (drop 2) indicated selective corrosion of the ferrite. However, high-resolution SEM analysis showed that selective corrosion of ferrite with some attack on the austenite occurred, shown in Fig. 13a, b. The σ-phase often remained in corroded regions, clearly indicating its net cathodic character, whereas ferrite and austenite were attacked, indicating their more net anodic character. In Fig. 13b, however, secondary austenite remained unaffected after the corrosion test, indicated by the characteristic shape. A large number of smaller discrete precipitates can also be seen and these observations are in good agreement with the earlier Volta potential assessments, where such precipitates were measured to be net cathodic (Fig. 9). Thus, these precipitates are assumed to be Cr₂N, χ, and possible CrN on basis of their Volta potential response and morphological appearance.

Fig. 13

Corrosion morphology of the aged sample under drop 2 (145 µg/cm²) with a a corroded area, b higher-resolution image of the highlighted box in a, c selective dissolution of the ferrite with remainders of secondary phases, and cracked σ, and d fracture in σ phase

A large σ-phase precipitated with a transgranular crack was also found, shown in Fig. 13c, d. A number of smaller cracks were also found, all located in the σ-phase indicating its brittleness and susceptibility to cracking. The presence of these cracks indicates an increased microstructure crack nucleation propensity, which was clearly enhanced by the presence of the brittle σ-phase in the microstructure.

The corrosion morphology under a chloride deposition density of 2472 µg/cm² (drop 4) is shown in Fig. 14a. Selectively corroded ferrite regions can clearly be seen indicating the net anodic character during the corrosion process. Several corroded regions located along δ/γ interphases are also shown in Fig. 14b, decorated with secondary phases (SP) that seemed almost unaffected by the corrosion process, supporting their net cathodic behaviour. Their discrete shape and morphology indicated that these secondary phases are possibly Cr₂N, χ, and/or CrN (Fig. 14c, d).

Fig. 14

Corrosion morphology of the aged sample under drop 4 (2472 µg/cm²) with a selective corrosion of the ferrite, b selective attack along interphases with remaining secondary phases, c, d corrosion along interphase boundary regions with remaining secondary phases (possibly Cr₂N)

The same corrosion morphology as with exposure to lower chloride deposition densities was observed under the droplet containing 2856 µg/cm² chloride (drop 5), showing selective attack on the ferrite with minor attack on the austenite, as can be seen in Fig. 15a, b. The most severe corrosion was observed under this droplet, due to largest deposition density, which in turn affects the corrosion potential. A large number of transgranular stress corrosion micro-cracks were observed in areas containing σ-phase and austenite.

Fig. 15

Corrosion morphology of the aged sample under drop 5 (2856 µg/cm²) with a selective corrosion of the ferrite and some attack on austenite, b transgranular cracks through an area containing σ-phase and austenite, c–f multiple micro-cracks nucleated within austenite

The austenite is susceptible to stress corrosion cracking due to its critical nickel content which was determined to be in the order of 7 % (see Table 1) [78]. The Copson curve shows lowest time to failure, i.e. highest stress corrosion cracking susceptibility for chloride-induced stress corrosion cracking of stainless steels with nickel contents between ~7 and ~15 % [78]. Stress corrosion micro-cracks were observed in regions containing secondary austenite and σ-phase as shown in Fig. 15b, a region eutectoidically transformed from ferrite. Secondary austenite is depleted in Cr, Mo, and N, and therefore, more prone to corrosion. The nickel content of secondary austenite is somewhat lower than that of primary austenite [30], and the cracking susceptibility may have been enhanced by the presence of sigma phase. Multiple stress corrosion cracking events were also seen in the primary austenite as shown in Fig. 15c, f, with maximum length of cracks in the aged microstructure of up to 30–35 µm, spanning across several grains. The latter confirms stress corrosion cracking propensity of aged sample when exposed to environment with extremely high chloride concentrations.

Practical relevance

Grade 2205 duplex stainless steel in the as-received condition is typically resistant to localised corrosion and stress corrosion cracking and the material only shows propensity towards corrosion and micro-cracking with exposure to harsh environmental conditions, such as the high chloride concentrations used in Fig. 10. High deposition densities of chloride in high concentrations (close to deliquescence equilibrium of MgCl₂) led to a maximum crack length of 15 µm, which showed minor stress corrosion cracking propensity of the as-received material. Grade 2205 duplex stainless steel has superior stress corrosion cracking resistance to the most common austenitic counterpart alloys 304L and 316L which showed more severe crack developments in similar (or under less aggressive) conditions [74–76, 79–84]. The 750 °C-aged microstructure, however, was clearly more susceptible to localised corrosion and stress corrosion cracking, which may have implications for welding practice of duplex stainless steels. Grade 2205 duplex stainless steel can be rendered susceptible to AISCC at a temperature of 50 °C, and the calculated threshold temperature (>50 °C) for AISCC in grade 2205 should be re-visited under MgCl₂ exposure conditions [75].

Conclusions

The effect of 750 °C ageing treatment on microstructure development using correlative EBSD and SKPFM techniques has been investigated. The crystallographic phases present were correlated to their local Volta potentials and corrosion behaviours.

• 1. CrN, Cr₂N, χ-, σ-, and secondary austenite (γ₂) phases precipitated in the microstructure with γ₂ and σ forming the largest secondary phases in size and fraction. The formation of σ, γ₂, and χ was allotriomorphic, with the σ-phase showing no internal grain boundaries.

• 2. The σ-phase and γ₂ precipitated within ferrite and at δ/γ interfaces, and χ-phase at δ/γ interfaces only, whilst CrN/Cr₂N was found at δ/γ and σ/γ₂ interfaces and δ/δ boundaries.

• 3. The as-received microstructure showed 50–70 mV Volta potential differences with δ acting as net anode and γ as net cathode. The Volta potential difference between δ and γ after ageing treatment decreased in certain regions, but Volta potential variations within each phase increased. Local Volta potential extremes were developed indicating enhanced corrosion susceptibility.

• 4. The σ-phase, Cr₂N, CrN, and inter-granular (IP) χ-phase indicated low electronic activities, i.e. expecting net cathodic behaviour, whilst the intra-granular (IF) χ-phase indicated only enhanced net anodic activity among secondary phases.

• 5. Selective corrosion of the ferrite was observed in the as-received and the 750 °C-aged condition, with the overall corrosion propensity of secondary phases being in-line with SKPFM observations.

• 6. Stress corrosion micro-cracks were found in the austenitic phase in the as-received and 750 °C heat-treated conditions after exposure to atmospheric MgCl₂ electrolyte.