*3.4. Q&P Experiments, Steel C*

The XRD measurements and tensile test results for Steel C are shown in Figures 10 and 11. The behavior of Steel C is very similar to Steel A, as the retained austenite fraction drops with decreasing *QT*. The higher retained austenite fraction of ≈11 vol % at *QT* = 100 ◦C corresponds to a significant increase in both uniform and total elongations, as well as a drop in yield strength.

**Figure 10.** (**a**) The retained austenite fraction of Steel C with respect to initial quench temperature *QT*. (**b**) The carbon content of the retained austenite with respect to *QT*. Specimens annealed at 850 ◦C, quenched to *QT* and partitioned at 450 ◦C for 100 s.

**Figure 11.** (**a**) *Rp*<sup>05</sup> and *Rm* and (**b**) uniform and total elongations for Steel C with respect to initial quench temperature *QT*. Specimens annealed at 850 ◦C, quenched to *QT* and partitioned at 450 ◦C for 100 s.

#### *3.5. EBSD Measurements*

Figure 12 shows representative results of the EBSD measurements as band contrast maps overlaid with austenite grains (shown in random coloring). Prior austenite and packet boundaries were determined using a previously developed iterative reconstruction algorithm [8,27] and are shown in red (packet boundaries) and black (prior austenite grain (PAG) boundaries and ferrite boundaries).

The observed microstructures in Figure 12a,c show that the martensitic transformation has been heterogeneous on a grain-by-grain basis for Steels A and C: untransformed, partially transformed and almost completely transformed austenite grains can be found in the microstructure. It should be noted that some austenite has probably transformed to martensite during EBSD specimen preparation, as the austenite fraction in EBSD measurements was much lower than in XRD.

Steel B does not exhibit a typical martensitic lath structure, although crystallographic analysis with the iterative method shows the presence of block- and packet-type subunits within prior austenite grains. This is an indication that some degree of bainite transformation has taken place either during the initial quenching or the partitioning stage of the heat treatment. In addition, several large, irregular-shaped intercritical ferrite grains are present in the microstructure.

**Figure 12.** Electron backscatter diffraction (EBSD) band contrast image overlaid with indexed retained austenite grains (random coloring). White boundaries indicate Kurdjumov–Sachs type relationship between γ and α. Black boundaries mark prior austenite grains and red boundaries indicate packet boundaries. (**a**) Steel A quenched to 150 ◦C, (**b**) Steel B quenched to 275 ◦C and (**c**) Steel C quenched to 100 ◦C. All specimens partitioned at 450 ◦C for 100 s.

#### **4. Discussion**

Based on the image analysis of the dilatometry specimens, there is an approximately 27 vol % intercritical austenite fraction in the microstructure of the Steel A after annealing. Assuming that all of the untransformed austenite that remains directly after the interrupted quenching is stabilized with carbon and is also retained at room temperature; 55 vol % is transformed at 175 ◦C, 63 vol % at 150 ◦C and 70 vol % at 125 ◦C. Fitting the Koistinen-Marburger equation to these values does not give a meaningful result, because the apparent martensitic transformation is too gradual with respect to temperature to obtain a good fit. Besides, martensite finish temperature *Mf* is an indistinct term. It is therefore probable that the martensitic transformation is not actually homogeneous in the microstructure and the degree of transformation varies from grain to grain. This conclusion is supported by the EBSD maps in Figure 11a, which shows a heterogenous martensitic transformation. The behavior of Steel C appears to follow a similar trend, based on the image analysis, XRD and EBSD results.

The behavior of Steel B differs from that of Steels A and C. Instead of a steady reduction in austenite fraction, there is an initially high amount at *QT* = 300 ◦C followed by an appreciable drop at *QT* = 275 ◦C to a nearly stable austenite fraction irrespective of further *QT* temperature reduction.

As shown by Figure 4b, there is much more austenite in the microstructure of Steel B after intercritical annealing compared to Steels A and C. This has two consequences—the average carbon content of the austenite is significantly lower (assuming total partitioning of carbon) and austenite grain size is higher. Both factors lower the critical driving force necessary for martensite nucleation, contributing to the rapid formation of martensite when lowering *QT* past 300 ◦C. It is possible that autocatalytic nucleation ("burst martensite" [28]) accelerates the rate of transformation. The rapid martensite formation is shown in Figure 13, which displays the retained austenite content with respect to *QT* overlaid with the dilatation curve in the temperature regime of the martensitic transformation.

**Figure 13.** The dilatation curve of Steel B in the regime of martensite transformation, with the retained austenite content with respect to the initial quench temperature *QT* on the secondary *y*-axis.

Interestingly, the uniform elongation *Ag* does not correlate with the high retained austenite content at 300 ◦C. Instead, there is a significant drop in yield strength, accompanied with a significant rise in *Rm*, as shown by Figure 9. This behavior is likely to be caused by a combination of both high retained austenite fraction and the effect of the different austenite morphology characteristic to this *QT*. Figure 8b shows that the average carbon content of the austenite phase after quenching to 300 ◦C is lower, which should also affect mechanical stability. The presence of unstable austenite grains results in a very high degree of strain hardening at the initial stages of deformation and consequently results in a high *Rm* combined with a low initial yield point. The unstable austenite grains are unable to contribute to ductility during later stages of deformation, having been completely transformed at an earlier stage and resulting in a lower total elongation *A*, as shown by Figure 9b.

From a microstructure point of view, the expected response to the quenching and partitioning would be the partial transformation of each austenite grain into martensite, followed by the enrichment of the balance untransformed austenite with carbon. The final microstructure, shown in Figure 14a, would then be a mixture of intercritical ferrite and martensitic islands interspersed with carbon-enriched retained austenite in martensite.

**Figure 14.** Austenite grains in a ferritic matrix partially transformed into martensite for (**a**) an ideal quenching and partitioning (Q&P) scenario, (**b**) the observed behavior of Steel A and Steel C and (**c**) the observed behavior of Steel B.

This type of microstructure was not observed in any of the studied grades. Instead, a complex microstructure had emerged in all cases consisting of intercritical ferrite, untransformed austenite grains and prior austenite grains in which the martensitic transformation had progressed to some degree.

This behavior can be attributed to two factors—the local chemical composition and the size of each austenite grain. During intercritical annealing, austenite will form at low-energy sites that are favorable towards nucleation [10]. In practice, this means ferrite grain boundaries and grain corners where the dissolution of cementite or other carbides has formed a carbon-rich volume suitable for nucleation. The stage of the annealing cycle at which each austenite grain nucleates will be decided by these local conditions. The growth rate of a nucleated austenite grain will, in turn, initially depend on the carbon content of the nucleus and at later stages the diffusion barrier formed by ferrite-stabilizing elements. In the case of the studied experimental steels, the primary element limiting austenite growth is aluminum, which is a strong ferrite stabilizer. As the austenite growth front advances, more and more aluminum will diffuse across the advancing front, until the aluminum content is high enough in the interfacial ferrite neighborhood and the growth slows down significantly.

The behavior during intercritical annealing can thus be characterized by a slow and uneven growth of austenite. Just prior to cooling, a microstructure has formed where exists a range of austenite grains with different sizes and chemical compositions. Such intercritical austenite microstructures are outlined schematically in Figure 14b,c.

Steel A and Steel C exhibited microstructures and mechanical properties quite similar and will be discussed together. Both steels have a similar chemical composition, with the notable differences being an elevated manganese content and a lower aluminum content in Steel A. This difference, along with the lower aluminum content, resulted in a slightly elevated austenite volume fraction after intercritical annealing in Steel A. In any case, both steels exhibit the type of intercritical austenite structure shown in Figure 14b, in which there are both very small and slightly larger austenite grains in the microstructure after annealing. When the steel in this condition is quenched to the quench temperature *QT*, primarily the larger, less stable austenite grains undergo a martensitic transformation, while the smaller, more stable grains remain unchanged. The less stable retained austenite is then stabilized with carbon during the partitioning stage. The result is a microstructure where the large, less stable austenite grains have become more refined and chemically stable due to martensitic transformation and subsequent rejection of supersaturated carbon during partitioning. Although only the larger austenite grains exhibit the expected quenched and partitioned response, the final result is a more homogenous, refined microstructure in terms of austenite stability, leading to the observed improvement in uniform elongation. Similar results have also been previously observed for steels with higher aluminum contents [3].

The quenching and partitioning response of Steel B is mostly explained by a greater fraction of intercritical austenite, along with a large distribution of austenite grain sizes and a larger average grain size overall. For Steel B, it is likely that bainite or isothermal martensite formation has occurred during partitioning, indicated by the large, irregularly shaped laths in Figure 11b. Another factor supporting bainite formation during partitioning is the high volume fraction of intercritical austenite (approximately 65 vol %). At 65 vol % austenite, assuming full partitioning of carbon, the austenite carbon concentration would be approximately 0.17 wt %. Compared to the Steels studied here with lower intercritical austenite fractions and consequentially higher carbon concentrations, Steel B is more amenable towards bainite (or, keeping in mind the high aluminum content, carbide-free bainite) formation during partitioning.

When Steel B is quenched to a sufficiently high temperature (in this case, 300 ◦C), the situation is similar to that observed for Steel A and Steel C: the smaller grains are left almost fully austenitic, while the larger grains transform to a greater degree (either to martensite or to carbide-free bainite). However, in this case, the smaller grains are left more unstable in the final microstructure; these unstable grains are transformed to martensite during the early stages of deformation, becoming unable to promote ductility at later stages. The message of this result is that to increase ductility, the end goal of the heat treatment should not be a perfect quenching and partitioning response, but the presence of highly stable retained austenite that will transform at a controlled stage of deformation. For Steels A and C, the quenching and partitioning treatment can be successfully used to refine and stabilize blocky-type austenite and increase the ductility of the steel, while at the same time introducing martensite into the microstructure. For Steel B, this is also possible to some degree, even though the tendency to form bainite or isothermal martensite during partitioning affects the final austenite fraction.

#### **5. Conclusions**

The quenching and partitioning response of three aluminum-alloyed experimental steels following intercritical annealing were investigated in this study. The following conclusions can be drawn from these investigations:


The successful quenching and partitioning response of an intercritically annealed steel seems to largely depend on the state of the austenite prior to quenching. A sufficient carbon content and a small austenite grain size give the steels robust behavior with regard to heat treatment, when looking at Steels A and C. The martensitic transformation of these alloys can be controlled by varying *QT*.

**Author Contributions:** Conceptualization, T.N., O.O., P.J. and P.P.; methodology, T.N.; software, A.S.; validation, T.N.; formal analysis, T.N., M.S. and P.P.; investigation, T.N.; resources, O.O. and P.J.; data curation, T.N.; writing—original draft preparation, T.N.; writing—review and editing, T.N., P.P. and M.S.; visualization, T.N.; supervision, P.P.; project administration, T.N.; funding acquisition, P.P.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
