**1. Introduction**

Intercritical annealing of low-alloy steel occurs in the temperature regime between the *Ac1* and *Ac3* temperatures, where thermodynamic equilibrium corresponds to some mixture of the austenite and ferrite phases. The amount of austenite at the conclusion of intercritical annealing depends on several factors, such as the starting microstructure, heating rate [1] and, more importantly, the alloying contents of the steel. Upon quenching at higher than critical cooling rates, the intercritical austenite phase is transformed to martensite, resulting in low-alloy dual-phase steel consisting of martensitic islands in a matrix of ferritic grains.

If the quenching is interrupted somewhere between the martensite start temperature *Ms* and the temperature where the last of the austenite transforms (*Mf*), the martensitic transformation will be incomplete and the martensitic islands will include some untransformed metastable austenite. If the steel is then transitioned for a certain duration to a suitable intermediate holding temperature (also commonly referred as 'partitioning temperature'), this metastable austenite can be either partially or fully stabilized down to room temperature by enriching it with carbon partitioned from the carbon-supersaturated martensite. The resulting microstructure then would consist of martensitic islands interspersed with carbon-enriched retained austenite, suspended in the ferritic matrix. In effect, the intercritical austenite phase has undergone a heat treatment known as quenching and partitioning (Q&P), originally proposed in 2003 by Speer et al. [2].

The end goal of the Q&P treatment of intercritical austenite is to improve the ductility of a dual-phase microstructure via the strain-induced transformation of the retained austenite to martensite during deformation, essentially increasing the strain hardening capability of the steel. It was found by some of the present authors in a previous study [3] that austenite stabilized in this manner can increase ductility even for a low-carbon, high-aluminum steel. However, it was recently observed by Tan et al. [4] that the ability of the retained austenite to contribute to strain hardening depends on its localization, morphology and stability essentially with respect to the carbon-content. They found that if the deformation occurs primarily in the ferrite phase, the transformation of retained austenite to martensite remains limited, thus limiting the total elongation. A fine film-like morphology (resulting in high mechanical stability) and localization of the austenite away from ferrite-prior austenite interfaces will also promote this type of undesirable mechanical stability.

Successfully promoting quenching and partitioning in a steel imposes some requirements on alloying. First, the steel must have the ability to suppress or delay the formation of carbides or other precipitates during partitioning. This is typically achieved by adding a combination of ferrite-stabilizing elements such as aluminum [5], silicon [6] or chromium [7]. Second, the steel must contain a sufficient amount of carbon to enable stabilizing an optimum retained austenite fraction to achieve the desired increase in ductility. Third, the steel must be sufficiently hardenable in order to reach the initial quench temperature without allowing undesired ferritic or bainitic phase transformations during the initial cooling step.

Intercritical annealing opens up some interesting possibilities for controlling the condition of the austenite at the conclusion of annealing. The grain size and the relative fraction of the intercritical austenite can be controlled to a large extent by suitably varying the annealing parameters [1,8,9]. The growth of the austenite phase in low- and medium-carbon steels has been observed in various research works to be controlled essentially by the diffusion of carbon across the interphase boundary. Therefore, carbon can be expected to partition nearly completely to austenite [1,10], while the partitioning of heavier elements (such as silicon, aluminum or manganese) is usually limited and will depend on the annealing temperature and duration [11].

The conditioning of the austenite will drastically affect the quenching response, both in terms of martensite start temperature *Ms* as well as the hardenability. Assuming total partitioning of carbon from ferrite to austenite, a smaller initial fraction of austenite will greatly increase its carbon content. This will both decrease *Ms* and improve hardenability [12], at least up to the eutectoid composition. To illustrate the point further, three preliminary calculations were made with the thermodynamic and kinetic calculation software JMATPRO® [13] for two of the experimental steels studied in this work: Steel A and Steel B (see Table 1 for chemical composition). JMATPRO was used to calculate continuous cooling transformation (CCT) curves based on the model by Kirkaldy et al. [14] modified by Lee and Bhadeshia [15]. Figure 1 shows the CCT curves for the steels with modified carbon contents, from nominal composition to carbon contents corresponding to 50 and 25 vol % fraction of intercritical austenite at the conclusion of annealing (assuming full partitioning of carbon). In the figure, the 50 vol % fraction of austenite is indicated by an assumed carbon content of 0.32C for Steel A and 0.22C for Steel B, and 25 vol % austenite content is indicated by 0.64C for Steel A and 0.44C for Steel B. Dashed lines in the figure indicate continuous cooling curves corresponding to a linear cooling rate of 10 ◦C/s, which is a realistically achievable cooling rate in modern continuous annealing lines. Referring to Figure 1, the onset of bainite transformation is delayed due to an increase in carbon content at lower austenite fractions. At an austenite fraction of 25 vol %, bainite transformation is avoided completely at a cooling rate of 10 ◦C/s, based on the JMATPRO® calculations.

**Table 1.** Compositions of the experimental steel grades.


These calculations do not take into account prior austenite grain size, which has been observed to have a significant effect on the martensite transformation of intercritically annealed steel in a previous study [8]. Essentially, the prior austenite grain size potentially remains very small during intercritical annealing, further reducing the *Ms* temperature. Such an effect of fine austenite grain size on *Ms* has been exhaustively investigated, for instance, by Yang and Bhadeshia [16].

Austenite conditioning via intercritical annealing was shown to affect the quenching and partitioning response of a high-aluminum steel in a previous study [3]. In this case, the prior austenite grain size was of the order of 1.5 μm and the austenite had a high carbon content directly after intercritical annealing. This decreased *Ms* significantly beyond the value predicted by empirical equations and promoted the formation of martensite with a small packet size and coarse, irregularly-shaped lath morphology. This, in turn, resulted in blocky rather than film-like retained austenite formation, situated primarily at prior austenite and packet boundaries. The amount of retained austenite was found to correlate well with the initial quench temperature, as well as the uniform and ultimate elongations of the studied steels. The effect of intercritical annealing on *Ms* was also observed by Yi et al. [17], who showed that with a suitably high carbon and aluminum content in the steel, intercritical annealing can be used to lower the initial quench temperature *QT* to ambient temperature or lower. Q&P-aided dual-phase microstructures were also produced for a Fe-0.2C-2.0Mn-1.5Si (wt %) steel after intercritical annealing by Wang et al. [18].

**Figure 1.** Partial CCT (continuous cooling transformation) curves showing the bainite start lines, calculated for (**a**) Steel A and (**b**) Steel B (see Table 1 for compositions), assuming full partitioning of carbon from ferrite to austenite during intercritical annealing. The carbon concentration for each annealing condition is indicated in wt % next to the corresponding bainite start line.

The above discussion should emphasize the point that the condition of the austenite at the conclusion of intercritical annealing should be known (at least at the level of prior austenite grain size and its carbon content), if any reliable estimation for Q&P heat treatment parameters should be made beforehand. Otherwise, it is necessary to determine *Ms* experimentally by suitable methods, such as dilatometry.

Despite several promising early results [3,4,17,18], the alloy compositions that have been investigated thus far are challenging to produce in an industrial setting with the current level of knowledge and technology. Furthermore, either the high silicon content of the steels makes hot dip galvanizing difficult [19] or the high aluminum or other alloying contents complicate the casting procedure.

With this in mind, the quenching and partitioning response was investigated for three experimental alloys following intercritical annealing: a conventional aluminum-alloyed TRIP-type steel and two novel lean-alloyed complex phase-type compositions with a mixture of aluminum, silicon and chromium elements as suppressors of carbide formation. The steels have been cast, rolled and could be hot dip galvanized in an industrial setting, at the current level of technology. The annealing temperatures and timeframes investigated here were also selected from the viewpoint of industrial relevance.

The phase transformation behavior during quenching of the steels was investigated via dilatometry and suitable quenching and partitioning parameters were determined based on the results. The effect of quench temperature on the retained austenite content and the morphology of the various phases in the final microstructures were investigated via optical microscopy, X-ray diffraction (XRD) and electron backscatter diffraction (EBSD). Preliminary evaluation of Q&P treated samples was made in respect of tensile properties.

#### **2. Materials and Methods**

The compositions of the investigated steels are shown in Table 1. The steels were supplied in a cold rolled, fully hard condition as 1.3-mm thick sheets.

Dilatometry experiments were conducted with a Gleeble 3800 thermomechanical simulator (Dynamic Systems Inc., Poestenkill, NY, USA) to determine the *Ms* temperatures of the steels after intercritical annealing. 10 mm × 60 mm specimens were cut from the sheets for the experiments. The specimens were subjected to a computer-controlled resistance heating followed by controlled cooling, with the aid of compressed argon gas below 400 ◦C down to room temperature. The temperature of each specimen was monitored with a K-type thermocouple, and the dilatation was measured in the transverse direction of the specimen at the thermocouple location using an extensometer fitted with quartz rods. The heat treatment parameters are shown in Table 2.

**Table 2.** Dilatometry parameters for the steels. *HR* = heating rate, *HT* = time at annealing temperature, *AT* = annealing temperature and *CR* = cooling rate.


The transformation curves were extracted from the dilatometric data by line fitting over the linear thermal contraction portion of both the austenite and martensite phases. Martensite and austenite cooling contraction curves were extrapolated from the linear portions of the curve. The extent of the martensitic transformation was calculated by using the lever rule. A curve was fitted to this data corresponding to the Koistinen-Marburger [20] equation:

$$V\_m = 1 - e^{-K(Ms - T)} \tag{1}$$

The martensite start temperature *Ms* and empirical fitting constant *K* were determined by curve fitting of Equation (1), using Trust-Region-Reflective Least Squares Algorithm in the Matlab® Curve Fitting Toolbox in Matlab® R2018b (Mathworks Inc., Natick, MA, USA). The transformation data up to a martensite fraction of 0.2 were excluded from the fitting, in order to minimize the effect of initial gradual martensite start on the fit. Two additional fits were also made, excluding data up to 0.4 and 0.6 martensite fraction. Of these three fits, the best fit according to the adjusted R-square statistics, when compared to the whole transformation data, was chosen as the optimum fitting solution. The *Ms* temperature was obtained directly from the fitting parameters for Equation (1). The fitting procedure is shown schematically in Figure 2.

The specimens were then sectioned in the transverse direction at the thermocouple location and mounted into cold setting resin for metallography. The specimens were ground and polished with colloidal silica Buehler Mastermet2 used in the final polishing step. The specimens were then treated with the color etching procedure proposed by LePera [21]. The polished and etched specimens were examined and micrographed with the Alicona InfiniteFocus G5 profilometer (Optimax IIM Ltd., Market Harborough, UK).

Typical quenching and partitioning parameters were designed for the steels based on the *Ms* values measured in the dilatometry experiments and the fitted martensitic transformation behavior, Figure 2. The parameters are shown in Table 3. Figure 3 shows a schematic diagram of the Q&P treatment cycle including linear heating and cooling rates.

Heat treatment specimens of 10 mm × 60 mm size were cut from the steel sheets. The Q&P treatments were conducted in the Gleeble 3800 simulator with the pocket jaw grips set at a free span distance of 35 mm. Two specimens were heat treated for each Q&P cycle: one specimen was later used for tensile testing and another for microstructural evaluation and retained austenite measurements using XRD. The test setup was essentially similar to the one employed in dilatometry, with the only difference that the extensometer was not used, if deemed unnecessary.

To determine if significant heating or cooling gradients exist in the specimens during the heat treatment cycles and to assess uniform temperature zone, three thermocouples were fitted on one of the Steel C specimens: one at the center, one 3 mm from center to the side and one 6 mm from center to the other side. It was found that at a distance of 3–6 mm from the controlling thermocouple, the temperature gradient remains below 3–5 ◦C at all stages of the heat treatment.

**Figure 2.** (**a**) The cooling contraction curve around the martensitic transformation. (**b**) The martensite volume fraction with respect to temperature extrapolated from the data shown in (**a**).

The retained austenite contents of the Q&P specimens were measured at room temperature using X-ray diffraction. A thickness of ~0.2 mm was ground from the specimen surface using a P800 SiC emery paper. The specimens were then ground with progressively finer papers, ending with the roughness P2000 in accord with standard polishing practice. The specimens were then electrolytically polished for 12 s at 40 V with the A2 electrolyte in a Lectropol-5 polisher (Struers Inc., Cleveland, OH, USA). The XRD analyses were conducted with the Panalytical Empyrean X-Ray diffractometer (Malvern Panalytical Ltd., Malvern, UK) using Co Kα-radiation (40◦ < 2θ < 102◦, 40 kV, 45 mA). The site of the thermocouple location was used for centering the X-ray beam on the specimens. The peaks used in the analysis were (110), (200), (211) and (220) for martensite and (111), (200), (220) and (311) for austenite. The method for retained austenite calculation was the four-peak method described in SP-453 [22] (four peaks for both ferrite and austenite). The carbon content was estimated from the measured average austenite lattice parameter using Equation (2) [23]:

$$C\_{\gamma} = (a\_{\gamma} - 0.3555) / 0.0044 \tag{2}$$

in which *aγ* stands for the average austenite lattice parameter in Å.

After the XRD measurements, the specimens were sectioned in the transverse direction at the location of the thermocouple and prepared for electron backscatter diffraction (EBSD). The specimens were polished in a manner similar to the practice adopted for dilatometry specimens prepared for light optical microscopy. However, the specimens were not etched after polishing. The polished specimens were removed from the mounts, washed in ethanol and placed in a low pressure desiccator overnight to remove moisture. Prior to electron microscopy, the specimens were cleaned in an evacuated chamber fed with ionized oxygen plasma to remove any leftover organic residues on the surface.

**Figure 3.** Schematic of the heat treatment cycle for the quenched and partitioned specimens.


**Table 3.** Annealing temperature (*AT*), initial quench temperature (*QT*), partitioning temperature (*PT*) and holding time at partitioning temperature (*tH*) for the heat treated specimens.

The specimens were then subjected to electron backscatter diffraction (EBSD) studies. The scanning electron microscope (SEM) used was a Zeiss ULTRAPLUS UHR FEG-SEM system (ZEISS International, Oberkochen, Germany) fitted with a field emission gun (FEG) and an HKL Premium-F Channel EBSD system with a Nordlys F400 detector (Oxford Instruments plc, Abingdon, UK), which was used for phase contrast and orientation mapping. The parameters for EBSD analysis were 20 kV acceleration voltage, 14 mm working distance and a tilt angle of 70◦ with a step size of 0.05 μm.

Tensile testing was carried out using an Instron 8800 servohydraulic materials testing machine (Instron, Norwood, MA, USA). Non-standard tensile specimens were prepared by precision milling a 6 mm long, 3.5 mm wide gage area with 1 mm roundings into the center of each specimen. A cooling lubricant jet was used to reduce thermal effects on the specimen during milling. Each specimen was tested in tension to fracture at an engineering strain rate of 0.001 s<sup>−</sup>1. Elongation was measured using a miniature axial extensometer. The total elongation *A* was recalculated to correspond to standard test geometry of a 120 mm × 20 mm gage section using the Oliver equation as implemented by ISO 2566/1 [24]:

$$A\_2 = A\_1 \times \left(\frac{k\_1}{k\_2}\right)^n \tag{3}$$

where *A*<sup>2</sup> is the calculated elongation value, *A*<sup>1</sup> is the known elongation value, *k*<sup>1</sup> and *k*<sup>2</sup> are the proportionality ratios of the two test pieces, and *n* is a material dependent constant. The standard adopts *n* = 0.4. The proportionality ratios *k*<sup>1</sup> and *k*<sup>2</sup> were calculated with the equation:

$$k\_1 = \frac{w\_s}{\sqrt{w\_a t\_a}}\tag{4a}$$

$$k\_2 = \frac{l\_s}{\sqrt{w\_s t\_a}}\tag{4b}$$

where *ws* and *ls* are the new width and length, respectively, and *wa*, *la* and *ta* are the measured width, length and thickness, respectively.

#### **3. Results**

#### *3.1. Dilatometry Experiments*

Table 4 shows the measured *Ms* temperatures for the steels estimated from dilatation curves based on dilatometer data. Figure 4 shows examples of the optical micrographs taken from the quenched dilatometry specimens. For Steel A and Steel C, the microstructures consist of martensite (including prior austenite) (lighter shade) and intercritical ferrite (darker shade). For Steel B, Le Pera etching produced an unexpected coloring result: the microstructure appears to consist of a primarily martensitic matrix (brown) with islands of intercritical ferrite (white).

Assuming that no ferrite transformation occurred during quenching, the intercritical austenite fraction can be estimated from the amount of the lighter phase (martensite) in the optical micrographs by image analysis [21]. The intercritical austenite contents could be estimated from the micrographs for Steel A and Steel C and they are shown in Table 4. In Steel B, the amount of intercritical austenite was judged to be approximately 65 vol %, although a reliable analysis is not possible due to the poor etching response. Assuming full partitioning of carbon, 65 vol % of austenite would correspond to an *Ms* temperature of 339 ◦C when calculated with the method proposed by Bhadeshia [25,26], which is in line with the measured *Ms* value shown in Table 4.

**Figure 4.** Typical examples of the optical micrographs of the dilatometry specimens for (**a**) Steel A, (**b**) Steel B and (**c**) Steel C.

**Table 4.** Experimentally determined *Ms* temperatures.

