3.1. Etching Method Development
To develop a quantitative assessment technique, it is necessary to find an etchant that allows obtaining an image with distinguishable phases for automatic analysis. The structure of the Steel 1 specimen with 21% Cr (
Table 1) quenched from a temperature of 1200 °C was investigated (
Figure 1). Three Vickers indentations were made for investigated area identification after each subsequent etching. A polishing with a minimal metal removal was carried out to always observe the same plane after each etching iteration.
Some etchants (reagents 1–3,
Table 2) recommended by other authors are not applicable to the investigated DSSs. Thus, the chemical etching with inhibited ferric chloride or with sodium hydroxide did not reveal the structure. The modified Beraha’s reagent showed a very indistinguishable structure.
Electrolytic etching is one of the most common techniques for revealing the structure of DSSs [
18].
Figure 1a shows a dark ferrite matrix and light austenite areas [
23] revealed by electrolytic etching using NaOH etchant. Using 60% nitric acid as an electrolyte provides the same result. This etching method consistently provides high-quality images of the DSSs’ structure, but the contrast between phases is insufficient for an automatic classification by the image analyzer. This etching technique is only suitable for measuring via the manual point count method [
24].
Other etchants according to
Table 2 (Glyceregia in
Figure 1b, Carpenter in
Figure 1c, Murakami’s etchant in
Figure 1d) showed similar results. Similar to electrolytic etching these etchants do not provide sufficient contrast on images for automatic quantitative description of the DSSs’ microstructure. The Murakami etchant solution requires heating to a temperature of 80–100 °C, which is unsafe and requires additional equipment.
Etching with Beraha’s reagent (solution 4,
Table 2) makes it possible to distinguish ferrite (
Figure 1e), which is darkened opposed to unetched austenite. The luminosity difference is sufficient to carry out an image binarization and to automatically determine the volume fraction of the phases according to the ASTM E1245 (
Figure 1f). The evaluation was carried out on panoramic images with an area of 1.4 mm
2. To assess the error, a series of preliminary analyzes were carried out, in which it was found that the relative error of measurement by the metallographic method does not exceed 5%.
The approval of the research methodology was carried out on Steels 1–3. Depending on the chemical composition and on the heat treatment of the DSSs, it is possible to obtain different austenite to ferrite ratios and to provoke the σ-phase formation. Taking into account the calculations carried out earlier in [
14], a series of heat treatments of experimental steels was carried out (
Table 3). The content of δ-ferrite in specimens of the studied steels (
Table 1), quenched after holding at different temperatures, was determined using a technique based on etching with Beraha’s reagent and subsequent automated image analysis. The measurement results are in good agreement with phase contents predicted by thermodynamic calculations (
Table 3 and
Figure 2). The XRD analysis results have some inaccuracies. For example, for Steel 1 quenched from 1200 °C, the δ-ferrite content determined using XRD is 77 wt.%, while the automated image analysis and thermodynamic calculations show 57 vol.% and wt.%, respectively. In this case, thermodynamic modeling was used to check the measurement results. In our experiments with prolonged (more than 60 min for 35 g specimens) isothermal holdings, the processes were almost complete and the thermodynamic equilibrium was practically achieved.
The σ-phase formation leads to the strengthening of DSSs and to a simultaneous decrease in corrosion properties.
Figure 3 shows the microstructure of a Steel 4 specimen after provoke annealing at 850 °C and revealed using the Beraha’s reagent (reagent 4,
Table 2), which showed the best result discussed earlier in identifying ferrite and austenite.
This etching darkens ferrite areas that have not transformed into a σ-phase, light gray areas of the σ-phase, and blue austenite (
Figure 3a). The obtained image can be binarized by grayscale threshold and used for the automated phase volume fraction analysis (
Figure 3b). Volume fraction of σ-phase (red in
Figure 3b) was found.
For a more detailed interpretation of the phase nature, the local chemical compositions were determined (
Figure 4 and
Table 4).
The content of ferrite-stabilizing elements in ferrite of Steel 1 quenched from 1200 °C (
Figure 4a,
Table 4, lines 1, 2) was as follows: 1.9–2.5% Mo, 23.0–23.3% Cr, and 5.4–5.7% Ni. Cr and Mo content in austenite (
Figure 4b,
Table 4, lines 3, 4) is lower (20.0–20.2% and 1.0–1.1%, respectively). Ni content in austenite is higher than in ferrite and reaches 7.2–7.5%. The content of the remaining elements (Mn and Si) is the same considering the error in their determination. The lowest content of Cr and Mo (25% and 4.2–4.6%, respectively) was determined in the austenite of Steel 4, which was subjected to provoked annealing for precipitation of a σ-phase at 850 °C for 15 min (
Figure 4b,
Table 4, points 5, 6). Ni content in this steel was found to be the highest. Cr and Mo content in the ferrite (
Figure 4b,
Table 4, points 7, 8) was higher (28.2–28.5% and 5.1%, respectively). Cr content in the σ-phase (
Figure 4b,
Table 4, points 9, 10) was slightly lower than in the ferrite, and Mo content was the highest, reaching 12.3–12.4%.
Therefore, the proposed analysis method (etching with Beraha’s reagent with subsequent automated image analysis) makes it possible to clearly distinguish austenite, δ-ferrite, and σ-phase in DSSs in images for automatic quantitative assessment according to ASTM E1245.
Three specimens of Steel 4 quenched from 1050 °C and subjected to provoked annealing at 850 °C were investigated to verify the method of the σ-phase quantifying. The measurement results shown in
Table 5 are consistent with the data on the kinetics of the σ-phase [
25]. The σ-phase was completely precipitated in the first 15 min and then its amount remained unchanged.
In addition to the σ-phase the precipitation of other detrimental secondary phases is possible in highly alloyed DSSs with a high PREN. The possibility of recognizing secondary phases using automated image analysis after etching with Beraha’s reagent is shown using high-alloy Steel 4 as example.
Secondary phases on the images obtained after etching with the Beraha’s reagent were identified accurately using automated image analysis (
Figure 5a) and are consistent with the images obtained by SEM (
Figure 5b,c). The undesirable secondary phases precipitations in Steel 4 are presented as various morphology particles of Laves phase (long stripes and granules) [
26]. The results of secondary phases volume fraction evaluating depending on the quenching temperature are presented in
Table 6.
Violation of the melting conditions leads to the formation of non-metallic inclusions which should be considered when evaluating the quality of DSSs. In addition to assessing the volume fraction of secondary phases, the contamination of the Steels by non-metallic inclusions (NMIs) was assessed on images of unetched specimens (
Table 7).
Steel 1, produced in the laboratory induction furnace, has 0.012 vol.% of NMIs. Steel 4 and Steel 5, produced in the industrial induction furnace, contain 0.146 vol.% and 0.214 vol.% of NMIs, respectively. The usage of this method and the accumulation of assessment results in different DSSs can become the basis for the development of metal products quality requirements.
3.2. Corrosion Properties
Automated image analysis after etching with the Beraha’s reagent made it possible to reliably estimate the amount of δ-ferrite in the steels depending on the quenching temperature, to investigate the impact of the δ-ferrite amount in DSSs on the resistance to pitting and crevice corrosion.
The crevice corrosion rate values of Steels 1, 2, and 3 with a lower PREN are higher than the corrosion rates of Steels 4 and 5. The behavior of the curves obtained for steels with different PRENs is different (
Figure 6). For Steels 1, 2, and 3, the crevice corrosion rate values gradually increase from 11 to 14 g/m
2·h. Apparently, in steels with a low PREN, crevice corrosion processes are less sensitive to structural changes (the ratio of phase components) than in steels with a high PREN. The crevice corrosion rate increases with distance from the point of equilibrium of δ-ferrite and austenite on phase diagram.
The crevice corrosion rate values of Steel 4 are maximum at the δ-ferrite content of 65–70 vol.% (3–4 g/m
2·h), then with an increase in the δ-ferrite content to 80 vol.% the corrosion rate drops rapidly to near zero values. This steel has the highest PREN and is also Nb-rich with about 0.1 wt.% of Nb. Because of this, Steel 4 is susceptible to the formation of secondary phases—Laves phases and carbonitrides [
27]. However, with an increase in quenching temperature, their volume fraction decreases. Therefore, with distance the point of equilibrium of δ-ferrite and austenite on phase diagram, the resistance to crevice corrosion increases.
The crevice corrosion rate values of Steel 5 have the opposite tendency compared to Steel 4. With an increase in the amount of δ-ferrite from 45 to 70 vol.%, the crevice corrosion rate rapidly increases from near zero values to 3 g/m2·h.
Despite the similarities in mechanisms of the pitting and crevice corrosion damage initiation the results of evaluating the corrosion resistance of the Steels show completely opposite results. Steels 1, 2, and 3 showed a smooth growth of the crevice corrosion rate with an increase in the δ-ferrite content from 38 to 100 vol.%. Epit values changes extremely with the highest value at 70 vol.% of δ-ferrite. For Steels 4 and 5 Epit values are at maximum for 50 vol.% of δ-ferrite and gradually decrease farther from the point of equal ratio of ferrite and austenite on phase diagram.
The pitting potential (Epit) for all steels is determined by the alloying level (PREN). Thus, the pitting potentials of steels with high PREN (Steels 4 and 5) are around 1000 mV, but for steels with low PREN (Steels 1, 2, and 3) it is lower: 100–500 mV. Analysis of steels with close PREN values shows that the Epit values also depends on the structural state.
Thus, it is necessary to search for the optimal phase ratio of δ-ferrite and austenite for each grade of DSSs. An equal ratio is not always the optimum.