Annealing of cold-worked austenitic SS may involve complex microstructural processes such as martensite reversion, precipitation, recovery, recrystallization, and grain growth. Bearing this in mind, the obtained results can be explained as follows.
4.1 Isochronal Annealing
Figure
4(a) indicates a softening of the material caused by
α′-martensite reversion for temperatures above 475 °C in samples A and B. For sample C, with a lower amount of
α′-martensite, a decrease in the microhardness above 700 °C follows the values for samples A and B. It is caused by recrystallization.[
21] Optical micrographs in the Figure
4 inserts show that after annealing at 750 °C, sample C was completely recrystallized, while after annealing at 550 °C, its microstructure did not differ significantly from the initial microstructure. Cios
et al. showed that the completely recrystallized material does not contain DIM.[
22]
An increase in the
S parameter at 475 °C coincides with the small maximum of microhardness for samples A and B with a higher amount of
α′-martensite. The increase in microhardness may be caused by precipitation of carbides. Carbide precipitation in austenitic SS at approximately 500 °C is known to be responsible for steel’s sensitization to intergranular corrosion. However, at temperatures lower than 450 °C, the rate of Cr diffusion is slow and the Cr content of M
23C
6 carbides decreases with the decreasing sensitization temperature. The carbides that precipitate during low-temperature aging have a composition that approaches the matrix composition.[
23]
When considering an increase in the microhardness at 475 °C for samples A and B, one can speculate that it may have been caused by an increase in the volume fraction of
α′-martensite. The anomalous evolution of deformation-induced
α′-martensite,
i.e., an increase in its volume fraction was reported for isothermal annealing, however, performed at slightly lower temperatures,
i.e., from 300 °C to 400 °C.[
24,
25] One of the explanations for this phenomenon also takes into account the precipitation of carbides. This leads to composition inhomogeneities that locally increase
MS in the regions depleted of alloying elements, causing the formation of more martensite on cooling.[
24]
Precipitation of carbides was proposed as an explanation of an increase and subsequent decrease in the
S parameter or mean PL values with increasing annealing temperature.[
17] The effect was reported for Ti-doped austenitic SS.[
26,
27] This behavior was attributed to trapping of positrons in misfit dislocations at precipitate-matrix interfaces, more so as TiC precipitates inside grains. The increasing number of precipitates causes an increase in the
S parameter or mean PL value. The growth and coarsening of precipitates reduce the number density of precipitates and decrease the
S parameter as well as the mean positron lifetime. Similar behavior was also observed for the annealing of electron-irradiated, deformed, or sensitized Type 304 austenitic SS by Yabuuchi
et al. and was attributed to the formation of metal carbide precipitates M
23C
6.[
19] In the case of not-deformed samples, the increase in the
S parameter was explained by the Kirkendall effect rather than by positron trapping in the misfit dislocations at the M
23C
6 precipitate interface. This explanation is based on the fact that if grains are large, nearly all positrons annihilate inside the grains. The probability of positrons being trapped in the misfit dislocations at the precipitate interface at grain boundaries should be low. Additionally, migration of Cr atoms to grain boundaries results in a flow of vacancies into the grains. The quenching of samples after annealing should retain the vacancies as positron trapping sites. However, this effect cannot be used for the explanation of the results presented in this article because the samples were slowly cooled after each annealing step, which should have caused the removal of vacancies. Vacancy defects inside precipitates were also suggested as possible candidates for positron trapping sites.[
11]
It is known that cold working accelerates the precipitation of both carbides and intermetallic phases in SS.[
27,
29] When deformation-induced
α′-martensite is present, the interphase boundaries arising as a result of its reversion are the sites of M
23C
6 nucleation and growth.[
28] In our case, it should be noted that the higher the
α′-martensite volume fraction, the higher the values of the
S parameter in the temperature range from 475 °C to 600 °C, in spite of the lower deformation degree (Figure
4 (b)). This behavior indicates the influence of the initial
α′-martensite volume fraction on the formation of positron trapping defects. The reversion of
α′-martensite provides sites for metal carbide nucleation, whereas volume excess resulting from the bcc/fcc change may contribute to an excess of vacancy-type defects at the interphase boundaries.
The influence of the initial
α′-martensite fraction is visible for the
Wr parameter in the region of its maximum (Figure
4(c)). Moreover, the
S-W
r plot indicates changes in the chemical surrounding of the sites of positron annihilation, which are more pronounced for samples A and B with a higher
α′-martensite fraction. The divergence of the experimental points from the straight line in the
S-
Wr plot already starts at 375 °C (Figure
5(a)). The changes continue (the experimental points form a loop), and this behavior is reflected in an increase and subsequent decrease in the concentration of defects with a chemical surrounding different from that introduced by plastic deformation. Moreover, this chemical surrounding also evolves with altering carbide composition,
i.e., Fe and Cr content of M
23C
6 carbides precipitating during annealing at increasingly higher temperatures. The decrease in the
Wr-parameter value for temperatures from 425 °C to 625 °C may be caused by the influence of annihilation with core electrons of carbon since the coincidence DB spectrum for graphite has much lower values for core electron annihilation energies than those for steel, as observed by Rajaraman
et al.[
27]
Our previous studies of martensite reversion in 1.4307 SS deformed in uniaxial tension revealed the presence of clusters of six to nine vacancies for the sample deformed at LN temperature to 5 pct elongation with the
α′-martensite volume fraction of 0.25. For the sample deformed to 32 pct elongation with the
α′-martensite volume fraction of 0.19, the vacancy clusters, if present, were smaller and had a lower concentration.[
10] The PL spectra measured for the rolled samples after annealing at 500 °C, in the region of the higher
S-parameter values,
i.e., a step in Figure
4(b), exhibit only one component, which can be treated as the mean positron lifetime,
i.e., 139, 136, and 130 ps for samples A, B, and C, respectively. If the differences in the deformation mode can be neglected, the results for the tension-deformed samples and the lack of evidence of vacancy clusters for the rolled samples seem to indicate that the conditions favorable for vacancy cluster formation are the following: a relatively high initial
α′-martensite volume fraction and a low degree of deformation.
4.2 Isothermal Annealing
The linear relationship of the experimental points in Figure
8(a) for all three samples annealed at 650 °C suggests no noticeable change in the type of positron trapping sites. However, there are differences between the samples in the
S- and
Wr-parameter dependences due to an
α′-martensite volume fraction. For samples with higher
α′-martensite volume fractions, the higher value of the
S parameter for an annealing time from 1.8 to 25.8 ks indicates higher concentrations of defects (Figure
7(b)). These differences disappear after annealing for 51 ks. A similar influence of the
α′-martensite fraction can be seen for samples annealed at 550 °C, and it does not change when annealing time is increased. A closer inspection of the
S-
Wr plots for samples isochronally annealed at 550 °C reveals that the experimental points for samples B and C diverge from the straight line fitted for sample A. Then they converge for the annealing time of 43.1 ks, for which there is a local maximum of the
S-parameter value. This effect can be explained by the nucleation and then coarsening of carbide precipitates, the presence of which was confirmed by XRD patterns (Figure
6). For longer annealing times, the points diverge again from the straight line. This behavior, which starts at 87.6 ks, indicates a subsequent alteration of positron trapping sites, which can be connected to the early stages of intermetallic phase precipitation.