Effect of Microstructure on the Morphology of Atmospheric Corrosion Pits in Type 304L Stainless Steel

Atmospheric pitting corrosion of austenitic stainless steels has been investigated by a number of researchers.^1-11  It is of significant interest to the nuclear waste industry in the United Kingdom as intermediate-level nuclear waste (ILW) is currently stored in containers made from Types 304L and 316L austenitic stainless steels (UNS S30403 and S31603, respectively⁽¹⁾). Future plans are for ILW containers to be held in an underground geological disposal facility (GDF).¹²  However, this facility has yet to be built and ILW containers are currently kept in warehouses above ground in which chloride-containing atmospheric aerosols¹³  may deposit at low levels on the container surfaces. The formation of small pits has been observed even when low chloride deposition densities are present.^11,14-15  It is important to establish any conditions under which atmospheric pitting corrosion may develop into atmospherically-induced stress corrosion cracking (AISCC), which might cause containers to fail when they are being moved to the GDF once it is completed.

It is well known that microstructure can influence pit initiation, with particular importance given to sulfide inclusions.^16-19  Propagation is also affected by factors such as grain orientation and grain boundary distribution. Recent evidence has shown the effect of centerline segregation of stainless steel alloys on atmospheric corrosion.²⁰  However, despite being labelled “fully austenitic,” many austenitic stainless steels have secondary phases²¹  such as delta ferrite (δ-ferrite) and the role of this phase in the atmospheric pitting corrosion has not been considered.

In austenitic stainless steels, the formation of delta-ferrite is explained in terms of the solidification process.²²  The thermodynamic preference for dendrites of solidifying stainless steel to maintain a eutectic composition causes Cr to be ejected into the solute liquid, giving interdendritic composition of relative Cr enrichment and Ni depletion. As Cr is a ferrite stabilizer and Ni is an austenite stabilizer, the interdendritic regions maintain a ferrite (body-centered cubic [bcc]) crystal structure when at room temperature. This ferrite is retained as a minor phase in the matrix of austenitic stainless steels once cooled to room temperature.²³  The presence of ∼2% to 5% ferrite is regarded as beneficial in fully austenitic stainless steels because ferrite phase improves hot workability.^24-26  However, it has been reported that this retained delta ferrite has detrimental effects on pitting corrosion resistance for fully immersed stainless steels.^27-28  These effects were explained in terms of segregation of impurities such as sulfur and phosphorus along the austenite/ferrite interface,^27,29  the formation of Cr-depleted zones,^29-31  the lower pitting resistance equivalent number (PRE_N) of austenite compared to ferrite phase,^31-33  and a low concentration of Cr and Mo in austenite phase.²⁴  In recent preliminary work, Davenport, et al.,³⁴  reported that the morphology of corrosion pits under atmospheric conditions was affected by the presence of residual ferrite in the microstructure, but the presence of residual ferrite was not confirmed.

Synchrotron x-ray absorption methods have been increasingly used to investigate corrosion propagation in steels in situ. Radiography has been used to measure influence of solution resistance on corrosion interface morphology^35-36  and the influence of oxidizing agents.^37-39  Microtomography has also been used to show propagation of pitting under atmospheric conditions.³⁴ 

In this paper, the role of delta ferrite in influencing pit morphology during atmospheric corrosion of Type 304L stainless steel is investigated.

A 3 mm thick plate of Type 304L austenitic stainless steel (Table 1) was supplied by Aperam. The plate had been cold-rolled and solution heat-treated (1,040°C to 1,100°C) after casting followed by forced air cooling. For synchrotron experiments, 2 mm pins were machined out of the ST plane (end grain) of the plate parallel to the rolling direction.

The plate was cut using a Buehler Isomet 4000^† SiC blade into 20 mm to 25 mm squares and mounted in bakelite. Pieces of plate were orientated to expose LT plane (top surface), LS plane (side grain), and ST plane (end grain) surfaces. Samples were ground and polished to a mirror finish using oxide polishing suspension (OPS, Struers) and etched with Kalling’s 2 reagent⁴⁰  for 5 s. The microstructure of the base alloy was investigated by scanning electron microscope (SEM) (PHILIPS XL30^† and JEOL 7000^†) equipped with energy dispersive spectroscopy (EDX) detector. Vibrating sample magnetometer (VSM) and x-ray diffraction analysis (XRD, Bruker D8 Advance^† diffractometer using monochromatic CuKα radiation, λ = 1.54056 Å) were used to determine the presence of ferrite content on the base alloy. Standard electron backscatter diffraction (EBSD) sample preparation method was followed for EBSD measurements.⁴¹ 

The plate was cut in different orientations to expose surfaces of interest, with 20 mm to 25 mm square pieces used to examine the LT plane and 5 mm × 20 mm for LS and ST planes. Samples were cold mounted using Vari-Set^† (Metprep) and ground to 800 grit and washed with deionized water (15 MΩ/cm) and dried using cold air.

A 0.4 M MgCl₂ solution was made up from MgCl₂·6H₂O (Fisher Scientific). A micropipette (Eppendorf) was used to deposit 1.1±0.03 μL droplets onto the metal surfaces. The diameter of the droplet on the plate surface was ∼2 mm, giving a chloride deposition density of ∼1 mg/cm². Twenty droplets were deposited on the LT plane surface, and 10 droplets were deposited on the LS and ST plane surfaces.

Samples were held at 30°C for 1 week and 6 weeks at relative humidities (RH) of 33%, 43%, and 56%, giving equilibrium MgCl₂ concentrations of 5 M (saturated), 4.4 M, and 3.7 M, respectively.¹⁵  RH was controlled inside a desiccator using saturated salts MgCl₂ (33% RH), K₂CO₃ (43% RH), and NaBr (56% RH)⁴²  with RH and temperature recorded.

X-ray microtomography experiments were performed at Beamline I12^† at Diamond Light Source using 70 keV x-rays, as described.³⁴  The pixel size was 1.8 μm. The exposure time was typically 1 s per projection and 1,800 projections were collected over a 180° rotation for each data set. The total scan time was 40 min to 45 min. The projections were reconstructed using a filtered back projection algorithm and rendered into 3D images using Avizo^† software. 1.1 μL droplets of 0.4 M MgCl₂ were deposited on the end of pins as in lab-based corrosion tests above. The humidity around the droplet was controlled using saturated salt solutions as described above.

Some plate was solution heat treated in an attempt to remove residual ferrite. The plate was put into a sealed silica glass container that was evacuated and backfilled with Ar gas. It was then heat treated to 1,050±5°C for 1 h followed by water quenching. These were then mounted and ground to 800 grit, then 0.4 M MgCl₂ solution was used to deposit ∼1.1 μL droplets; these were corroded for 1 week at 30°C and 43% RH.

Figure 1(a) shows an SEM image of the ST plane following polishing and etching in Kalling’s 2 etchant. It can be seen that equiaxed grains dominate the bulk of the surface, with banded structures observed along the rolling direction. Comparison with EBSD data (Figure 1[b]) shows that the equiaxed grains are austenitic (blue). It is not possible to distinguish between ferrite and martensite using EBSD in this figure. The elongated features parallel to the rolling direction are red, indicating that this is likely ferrite (Point 1), but other regions that do not follow the rolling direction are also red. Some of these are likely to be twins (Point 2), and others (Point 3) in Figure 1(b) are likely to be deformation-induced martensite regions, which exist as a consequence of grinding and polishing the plate surface. The total mass of ferrite/martensite phases make up ∼6% of the steel.

The chemical composition of ferrite and austenite phases were obtained by EDX analysis. Five point scans were measured over the center of the ferrite band at different locations, and a similar number of scans were measured over an austenite region. The average chromium and nickel content of ferrite phase was 24±1% and 4±0.5%, respectively, while that for austenite phase was 19±0.5% and 8±0.4%, respectively (Table 2).

XRD results (Figure 2) show strong austenite peaks and one smaller peak that correlates to the ferrite (110) peak, with relative peak intensities indicating ∼4% ferrite. VSM (Figure 3) shows this material behaves in a soft magnetic manner in the presence of a magnetic field, indicating the presence of at least one nonaustenitic phase.

Figure 4 shows SEM images of the three plate orientations (LT, LS, and ST plane) of Type 304L stainless steel plate after 5 s etching with Kalling’s 2. Bands along the rolling direction were preferentially dissolved by etching of polished samples. These correlate to ferrite bands observed in SEM and EBSD in Figure 1. Wider bands were revealed on LT plane, Figure 4(a), as this plane was not reduced during the rolling process. Coarse equiaxed grains were observed in all plate orientations.

Corrosion pits formed under droplets in all conditions tested, with results summarized in Table 3. In total, pitting was observed under 261 out of 274 droplets.

Figure 5 shows the optical image of a droplet on the ST plane after 1 week at 30°C and 33% RH. A single pit was observed near to the edge of the droplet.

Figure 6 shows SEM images of pit morphologies on the three plate orientations. The pit grown on the LT plane (top surface of the plate) has ridges in its wall and shows a “layered attack” morphology. Both LS and ST planes instead show a striped morphology of elongated strands of undissolved metal parallel to the rolling direction. The difference in ferrite band thickness between LS and ST planes may relate to the position of the pit on the plate, with a higher density of ferrite bands occurring toward the center of the plate. This may be a consequence of center-line segregation.

Figure 7 shows a droplet that has been corroding for 6 weeks on the LT plane. The pit growth is strongly affected by microstructure, with most growth at the surface advancing parallel to the rolling direction (Figure 7[b]). Bisecting the pit (Figure 7[c]) and polishing the ST plane shows growth propagates in both LS and ST planes, following the residual ferrite bands (Figure 7[d]). A magnified region (Figure 7[e]) of propagation into the LT plane after mirror polishing shows localized attack inside the plate. This is seen to propagate along elongated micrograins that have formed (Figure 7[f]). EDX results show that these micrograins are high in Cr and low in Ni, indicating the presence of ferrite (Figure 7[g]).

Propagation along the LS plane was also investigated (Figure 8). After 6 weeks at 30°C and 33% RH, a single pitting site was observed at the edge of the droplet (Figure 8[a]). After being washed (Figure 8[b]), the surface of the metal was polished to a mirror finish (Figure 8[c]) to allow the relationship between the pit morphology and the microstructure to be resolved (Figure 8[d]). Propagation along the LS surface was seen to follow micrograins (Figure 8[e]) that were high in Cr and low in Ni (Figure 8[f]), indicating presence of ferrite.

An SEM image of a pit grown for 1 week at 56% RH at 30°C on the LT surface is shown in Figure 9. The layered attack is clear from the surface (Figure 9[a]), with corrosion not only propagating down into the plate but also in parallel to rolling direction. After focused ion beam (FIB) milling adjacent to the pit mouth (Figure 9[b]), it is clear that dissolution parallel to rolling direction occurs in sheet-like morphology. The walls of the pit also have this preferential attack.

Tomography of a Type 304L pin corroding at 59% RH after 32 h is shown in Figure 10(a). Rendering the pit using software showed propagation along the rolling direction (Figures 10[b] and [c]). SEM image of the pit surface after 79 h (Figure 10[d]) shows grinding direction perpendicular microstructure.

In total, six tomography pin samples were examined, two samples at each exposure RH (33% RH, 43% RH, and 56% RH). It should be noted that there were always multiple pits per droplet “sample” (3 to 5 regions of attack per droplet compared with only one in lab-based tests). This was attributed to the effect of beam damage, which has been reported previously.^43-44 

Figure 11 compares the growth of pits for 1 week at 43% RH at 30°C on steel LT plane without an additional solution heat treatment (Figure 11[a]) and after heating at 1,050°C for 1 h followed by water quenching (Figure 11[b]). The layered morphology usually seen with pitting on the LT plane is not apparent after solution annealing. This loss of layered morphology is also observed on the LS plane after heat treatment (Figure 12), suggesting a reduction in the amount of residual ferrite. This has been confirmed by VSM (Figure 3).

There is strong evidence that appreciable amounts of ferrite are present in Type 304L, which is ordinarily considered a “fully austenitic” alloy, shown by EBSD which suggests ∼6 wt% and XRD suggesting ∼4 wt%. VSM confirms the presence of soft magnetic material in what should be a nonmagnetic alloy (Figure 3). This residual ferrite has also been observed elsewhere.^23,45-48  This ferrite is oriented in bands along the rolling direction with a strongly layered morphology with regular spacing on each surface of around 10 μm (Figure 4). Heat treatment is shown to reduce the magnetic response of the alloy, indicating a reduction in the ferrite volume. All of the pits showed etching on their surfaces that revealed the underlying microstructure, with no evidence of polishing observed.

The orientation of ferrite in the plate influenced the morphology of atmospheric pitting strongly. Pitting on both the LS and ST planes followed the ferrite bands and continued into the pit during corrosion, showing that this was not just a surface phenomenon (Figure 6). This suggests that the ferrite phase corrodes preferentially, promoting lateral pit growth that then propagates into the surrounding austenite phase. Pitting on the LT plane after 1 week showed the same layered structure, with attack into the metal appearing to form in sheets (Figure 6). FIB milling next to a pit on the LT plane (Figure 9) showed this attack undercut the surface and propagated along the ferrite bands, which was confirmed by tomography (Figure 10). SEM images on the ST (Figure 7) and LS planes (Figure 8) showed that the attack along the rolling direction into the planes is led by deep attack along ferrite bands. EDX in both cases show the grains that are being dissolved at the tip of the corrosion front are high in Cr and low in Ni, suggesting ferrite. Figures 11 and 12 show the layered morphology ordinarily present in these pits is no longer observed after heat treatment. Instead, Figure 11 shows a hexagonal pit formed that the pit grew inside a single grain, as seen by others.^49-51  This may be a consequence of grain growth during the solution treatment. This agrees with reports that delta ferrite content can be reduced by solution annealing.^22-23,52 

In duplex stainless steels under chloride deposits in atmospheric conditions, it is often observed that one phase, either ferrite or austenite, corrodes preferentially.^53-56  Prosek, et al.,^53-54  studied the atmospheric corrosion resistance of eight stainless steel grades, including Types 304L and 316L austenitic stainless steels, under salt solutions of pure MgCl₂, CaCl₂, and NaCl as a function of temperature (20°C to 50°C) and RH (15% to 70%) in stress corrosion cracking experiments. They reported that, for duplex stainless steels (DSS) exposed to a number of different conditions, the corroded phase was ferrite, which contained more Cr (identified by the use of EDX analysis). Örnek, et al.,⁵⁶  found that ferrite is more susceptible to corrosion than austenite in DSS 2205 and 2507 when under MgCl₂ droplets exposed to 80°C and 40% RH for 7 d. They also found that heat treatment (at 750°C for 3 h followed by water quenching) led to selective attack of ferrite and ferrite/austenite interfaces in DSS 2205. However, for in DSS 2507, mainly primary and secondary austenite was attacked.

The active-passive transition behavior of phases is sensitive to the environment in which the reaction happens. The active-passive transitions for austenite and ferrite occur at similar potentials; for example, the active peak for ferrite is slightly lower in potential for that of austenite for DSS 2205 in 2 M H₂SO₄ + 0.5 M HCl solutions.⁵⁷  In atmospheric corrosion conditions, such as those with highly concentrated chloride solutions, it is possible that small local differences in potential may change the relative reactivity of the two phases during corrosion, promoting the dissolution of ferritic phases in favor of austenitic phases.

• •

  Type 304L “austenitic” stainless steel plate was observed to contain between 4 wt% and 6 wt% ferrite, confirmed using XRD, VSM, and EDX, which is most likely residual delta-ferrite that formed early in the solidification process. This ferrite was orientated in bands along the rolling direction.

• •

  Atmospheric pitting corrosion of Type 304L under MgCl₂ droplets was found to be strongly affected by the presence of ferrite. Pitting corrosion propagated in the direction of these ferrite bands on all three orientations, often causing significant undercutting of the surface.

• •

  Ferrite was observed to be preferentially attacked in all cases, with austenite being attacked subsequently.

• •

  Solution annealing of the stainless steel largely removed the residual ferrite and reduced the elongated geometry of the corrosion pits as compared to as-received plate, and resulted in the formation of equiaxed etched faceted pits.

Haval Mohammed-Ali was funded by the Kurdistan Regional Government (KRG), Ministry of Higher Education and Scientific Research. Thanks also go to I12 – JEEP beamline at Diamond Light Source, U.K. for tomography results and Cem Örnek for EBSD results.