Effects of Primarily Solidified Dendrite and Thermal Treatments on the M23C6 Precipitation Behavior of High-Chromium White Iron

The precipitation behavior of M23C6 carbide during thermal treatment of high-Cr white iron with various fractions of primarily solidified dendrite was studied and reviewed. M23C6 precipitation in the primarily solidified dendrite occurred preferentially during conventional heat treatment, whereas it occurred scarcely in the eutectic austenite. The reaction between M7C3 and austenite caused the dissolution of M7C3 into austenite, followed by precipitation of M23C6 along the periphery of eutectic M7C3. Relatively low-temperature thermal treatment (modified heat treatment) led to precipitation of M23C6 particles in the eutectic austenite, which is presumed to be caused by solubility difference depending on temperature.


Introduction
The superior wear resistance of high-Cr white iron enabled the application of this material in mining activities such as ore grinding, coal grinding, or in their transport. The superior wear resistance of these alloys is originated in their matrix microstructure and the existence of various carbides, both in as-cast and heat-treated conditions. As-cast high-Cr white irons may have martensitic or austenitic matrix depending on the post-casting cooling rate and section size. The heavy sections that cool slowly have a martensitic matrix with a small amount of relatively soft-retained austenite due to incomplete transformation to martensite. The maximum hardness of the matrix is achieved by the subsequent heat treatment that transforms the matrix to martensite. However, the pearlite matrix has less hardness and low toughness; thus, pearlite is not desirable. Pearlite formation occurs when alloying is insufficient [1].
The matrix structure and existing carbides have strong effects on the mechanical properties of high-Cr white cast irons. With a process similar to the solidification of hypoeutectic gray irons [2], hypoeutectic high-Cr white irons solidify primarily with the formation of austenite dendrites, followed by a eutectic reaction between austenite and carbide. The carbides in high-Cr white iron are very hard and wear resistant but very brittle [1]. Wear resistance could be improved by increasing the amount of carbide.
The types of existing carbides in high-Cr irons are M 7 C 3 , M 3 C, and M 23 C 6 . Generally, the eutectic reaction generates M 7 C 3 in high-Cr white iron. M 7 C 3 carbides contribute to improving the wear resistance of the alloy; however, those primarily precipitated from the melting process ahead of eutectic reaction are known to be quite deleterious to impact toughness and should be avoided [1]. Thus, the hypereutectic composition of the alloys is not desirable for engineering applications. In hypoeutectic alloys, increasing carbide fraction decreases toughness. However, the higher fraction of austenite or dendrite may improve toughness but reduce the hardness of the alloy. It is known that the fractions of austenite and M 7 C 3 carbide in hypoeutectic high-Cr white iron are sensitive to C and Cr contents [1]. Decreasing the amount of C and Cr increases the austenitic dendrite fraction and decreases the fraction of M 7 C 3 along the interdendritic regions.
The release of C and Cr from the saturated austenite and the precipitation of M 23 C 6 carbide, which is known as "destabilization", could occur at high temperatures. The degree of destabilization at high temperatures is expected to be related to the fraction of dendrite. Variation in dendrite fraction through modification of chemical compositions also means various fractions of eutectic structure. As reported earlier, the dendritically solidified austenite has a slightly different chemical composition from the eutectically solidified one [3,4]. Thus, the destabilization behavior in dendritic austenite might be different from that in eutectic austenite. M 23 C 6 precipitation temperature in the alloys that have various fractions of existing phases might be different from each other. Therefore, the purpose of this study was to understand the precipitation behavior of M 23 C 6 or destabilization of austenite in the hypoeutectic high-Cr white irons with various dendrite fractions. In particular, the modified heat treatment condition (direct aging (DA)), in which the maximum fraction of M 23 C 6 precipitation occurred for each alloy, was selected on the basis of ThermoCalc prediction. Precipitation behavior of M 23 C 6 in austenite (with both dendritic and eutectic structures), and the reaction between austenite and M 7 C 3 during the modified heat treatment were studied.

Specimen Preparation
As reported earlier by the authors [5], the primarily solidified dendrite fractions in high-Cr white iron were adjusted in the composition range of 2.1~2.9 wt.% C and 24.0~27.0 wt.% Cr, which is in the composition range of ASTM A532 Class 3 Type A. Chemical analysis of the alloys was carried out by optical emission spectrometer (OBLF QSN-750) and carbon-sulfur determinator (ELTRA CS800). The chemical compositions of the specimens are given in Table 1. Phase prediction was carried out to understand the phase evolution during cooling, with the chemical composition provided in Table 2 through commercial software ThermoCalc (Thermo-Calc 2019b) on the basis of database DB TCFE9 for Steels/Fe-Alloys (V.9.1). The specimens were prepared by induction melting of the master alloy, followed by secondary melting and casting to a rod. The cast specimens were usually subjected to conventional heat treatment and modified heat treatment. The conventional heat treatment condition included soaking at 1065 • C for 4 h, followed by air cooling and tempering at 500 • C for 4 h, and 250 • C for 4 h, respectively [5]. To understand the precipitation behavior of M 23 C 6 , a modified heat treatment condition for each alloy was selected on the basis of ThermoCalc prediction. The modified heat treatment (DA) temperature for each alloy was chosen as the temperature at which the maximum M 23 C 6 fraction appeared on ThermoCalc calculation. The soaking time at each DA temperature was decided by the time (4 h) at which the highest hardness value was shown. Rockwell hardness values (WOLPERT / D-6700) of the specimens were measured with 10 points in the as-cast, conventionally heat-treated, and modified heat-treated conditions. The DA treatment included soaking at each temperature for each alloy, followed by air cooling in the absence of the tempering process.
The details of conventional heat treatment and modified heat treatment (direct aging (DA)) are as follows: 3. DA specimens were exposed under 1 × 10 −5 torr vacuum for 60 min to observe the movement of the M 7 C 3 carbide interface. Certain points of DA specimens were engraved to observe at the same position before and after the vacuum exposure. The vacuum exposure was carried out in a vacuum furnace (Jungmin 2010), followed by Ar gas fan quenching.

Microstructural Observation and Phase Identification
Specimens for optical microscopy (OM: Nikon ECLIPSE MA200) and scanning electron microscopy (SEM: JEOL JSM IT500LV) were prepared metallurgically and etched by swabbing with Vilella's reagent consisting of 45 mL glycerol, 15 mL nitric acid, and 30 mL hydrochloric acid. The volume fraction of the primarily solidified dendrites in the as-cast specimens was measured during optical microscopy by an Image analyzer (iMT iSolution DT). Specimens for transmission electron microscopy (TEM: JEOL JEM-2100F) were prepared by mechanical polishing down to 60-micrometer thickness, followed by twin jet polishing (Struers TenuPol-5). The solution for twin jet thinning was 10% perchloric acid in methanol. The existing phases in the heat-treated specimens were identified by TEM selected area diffraction pattern (SADP), and energy-dispersive X-ray spectrometer (EDS: Oxford X-MAX AZtec).

Prediction and As-Cast Microstructure
As reported earlier [6,7], the M 7 C 3 /austenite eutectic phase and primarily solidified dendrite coexist in the as-cast condition of high-Cr cast irons. The eutectic carbide in the white iron with more than 20% Cr content is M 7 C 3 [1]. Similar to gray irons, the primarily solidified dendrite forms during the solidification of high-Cr white irons. Oh et al. [5] reported various fractions of primarily solidified dendrite in high-Cr white irons, as shown in Table 1. It is known that the primarily solidified dendrite formation during solidification of cast irons is closely related to C equivalent value (Ceq). The equilibrium microstructural evolution during post-casting was able to be predicted through ThermoCalc calculation, as shown in Figure 1. Solidification began with austenite (or delta ferrite) in the alloys with low-Ceq-valued hypoeutectic high-Cr white iron, followed by the eutectic reaction of austenite and M 7 C 3 , as shown in Figure 1. The solidification of the austenite (or delta ferrite) forms dendrites, which are referred to as primarily solidified dendrites in this study. As displayed in Figure 1, phase transformation and precipitation of new phases occurred during the post-casting stage. It was possible to predict the precipitation behavior and phase transformation during cooling through commercial software ThermoCalc.

Prediction and As-Cast Microstructure
As reported earlier [6,7], the M7C3/austenite eutectic phase and primarily solidifi dendrite coexist in the as-cast condition of high-Cr cast irons. The eutectic carbide in t white iron with more than 20% Cr content is M7C3 [1]. Similar to gray irons, the primari solidified dendrite forms during the solidification of high-Cr white irons. Oh et al. [5] r ported various fractions of primarily solidified dendrite in high-Cr white irons, as show in Table 1. It is known that the primarily solidified dendrite formation during solidific tion of cast irons is closely related to C equivalent value (Ceq). The equilibrium micr structural evolution during post-casting was able to be predicted through ThermoCa calculation, as shown in Figure 1. Solidification began with austenite (or delta ferrite) the alloys with low-Ceq-valued hypoeutectic high-Cr white iron, followed by the eutec reaction of austenite and M7C3, as shown in Figure 1. The solidification of the austenite ( delta ferrite) forms dendrites, which are referred to as primarily solidified dendrites this study. As displayed in Figure 1, phase transformation and precipitation of new phas occurred during the post-casting stage. It was possible to predict the precipitation beha ior and phase transformation during cooling through commercial software ThermoCal In the as-cast microstructure of the alloys [5], the primarily solidified dendrite fra tion increased with decreasing Ceq, as predicted in ThermoCalc calculation. The lowe Ceq alloy, 2124 had a high fraction of dendrite, and little dendrite was found in the highe Ceq alloy, 2927.

Conventional Heat Treatment
Conventional heat treatment of the high-Cr cast iron followed by air cooling and tempering. During soaking, "d the solute elements such as C and Cr to be derived from the precipitate the secondary carbides, preferentially occurs [5] from the saturated austenite during destabilization is expres report [4,5].: where γ* is austenite with lower alloy content than that of th The conventionally heat-treated alloys have dendrite fe cast alloys [5], but dendritic boundaries were not distinct c Compared with those of the as-cast alloys, the dendrites we in the conventionally heat-treated condition. Little or no p drites in the as-cast condition [5], as displayed in Figures 2 an heat-treated alloys have a considerable number of precipita as displayed in Figure 4. The particles are secondary M23C study, TEM micrographs and selected diffraction pattern an to be M23C6, as shown in Figure 5. The diffraction patterns i the particles had an orientational relationship with the matr In the as-cast microstructure of the alloys [5], the primarily solidified dendrite fraction increased with decreasing Ceq, as predicted in ThermoCalc calculation. The lowest Ceq alloy, 2124 had a high fraction of dendrite, and little dendrite was found in the highest Ceq alloy, 2927.

Conventional Heat Treatment
Conventional heat treatment of the high-Cr cast iron included soaking at 1065 • C, followed by air cooling and tempering. During soaking, "destabilization", which allows the solute elements such as C and Cr to be derived from the saturated austenitic matrix to precipitate the secondary carbides, preferentially occurs [5]. M 23 C 6 carbide precipitation from the saturated austenite during destabilization is expressed as follows in the previous report [4,5].: where γ* is austenite with lower alloy content than that of the original matrix γ [4,8].
The conventionally heat-treated alloys have dendrite features similar to those in ascast alloys [5], but dendritic boundaries were not distinct compared with as-cast types. Compared with those of the as-cast alloys, the dendrites were filled with many particles in the conventionally heat-treated condition. Little or no particles exist within the dendrites in the as-cast condition [5], as displayed in Figures 2 and 3; however, conventionally heattreated alloys have a considerable number of precipitated particles in the dendrites, as displayed in Figure 4. The particles are secondary M 23 C 6 carbide [4]. In the present study, TEM micrographs and selected diffraction pattern analysis identified the particles to be M 23 C 6 , as shown in Figure 5. The diffraction patterns in Figure 5 show that most of the particles had an orientational relationship with the matrix. drites in the as-cast condition [5], as displayed in Figures 2 and 3; however, conventiona heat-treated alloys have a considerable number of precipitated particles in the dendrit as displayed in Figure 4. The particles are secondary M23C6 carbide [4]. In the pres study, TEM micrographs and selected diffraction pattern analysis identified the partic to be M23C6, as shown in Figure 5. The diffraction patterns in Figure 5 show that most the particles had an orientational relationship with the matrix.    The micrographs of dendritic and eutectic austenites in conventionally heat-treated conditions are different from each other. As mentioned earlier, dendritic austenite was filled with M23C6 particles, while little or no M23C6 particles existed in the eutectic austenite, as displayed in Figures 6 and 7. This means the nature of austenite with a dendrite structure is different from that with a eutectic structure. As Dupin showed schematically, the alloying elements are believed to contribute to forming M7C3 carbide during a eutectic The micrographs of dendritic and eutectic austenites in conventionally heat-treated conditions are different from each other. As mentioned earlier, dendritic austenite was filled with M 23 C 6 particles, while little or no M 23 C 6 particles existed in the eutectic austenite, as displayed in Figures 6 and 7. This means the nature of austenite with a dendrite structure is different from that with a eutectic structure. As Dupin showed schematically, the alloying elements are believed to contribute to forming M 7 C 3 carbide during a eutectic reaction; thus, C and Cr in the austenite enveloping the M 7 C 3 carbides are depleted [3,4]. As the austenite in the dendrite formed before the eutectic M 7 C 3 /austenite reaction, the solute concentration of the austenite in dendrite might be rich compared with that of the eutectic austenite. This means that the primarily solidified dendrites have a high potential to form the precipitates due to saturation of solute elements such as Cr and C. Therefore, easier and more precipitation of M 23 C 6 occurred during destabilization in the dendritic austenite than in the eutectic austenite. Except for 2927 alloy (eutectic alloy), precipitation of M23C6 scarcely occurred in t interdendritic eutectic area. Many studies also explained that the secondary carbides not nucleate and grow on eutectic carbides but form preferentially within the dendri matrix [9][10][11]. The present study found a similar result in hypoeutectic alloys, as display in Figures 6 and 7. When comparing Figures 6 and 7, the precipitation is strongly relat to the existence of primarily solidified dendrites in the hypoeutectic alloys, as reported earlier studies [5,[9][10][11].
However, the dendrite-free and fully eutectic 2927 alloy had a little amount of M2  Except for 2927 alloy (eutectic alloy), precipitation of M 23 C 6 scarcely occurred in the interdendritic eutectic area. Many studies also explained that the secondary carbides do not nucleate and grow on eutectic carbides but form preferentially within the dendritic matrix [9][10][11]. The present study found a similar result in hypoeutectic alloys, as displayed in Figures 6 and 7. When comparing Figures 6 and 7, the precipitation is strongly related to the existence of primarily solidified dendrites in the hypoeutectic alloys, as reported in earlier studies [5,[9][10][11].
However, the dendrite-free and fully eutectic 2927 alloy had a little amount of M 23 C 6 carbides in the austenite, which formed during the eutectic reaction, as displayed in Figure 7e. This is presumed to be related to the solubility of the alloying elements depending on temperature, which is addressed in direct aging treatment (DA).

Basic Background of M 23 C 6 Precipitation
According to the ThermoCalc prediction, the fraction of M 23 C 6 carbide increases at the cost of M 7 C 3 and austenite reduction at high temperatures. The fraction of M 23 C 6 carbide in each alloy was inversely proportional to that of M 7 C 3 carbide, as shown in Figure 1. Previous studies, mentioned above, revealed that M 23 C 6 carbides precipitated during destabilization; thus, precipitation preferentially occurs in the dendrites [5,[9][10][11]. This also occurred in the present study.
Then, a question arises from the ThermoCalc results: How does the fraction of M 23 C 6 increase with the reduction of M 7 C 3 and austenite? This is shown in Figure 1 and Table 2. The precipitation-initiating temperature of M 23 C 6 carbide during post-cast cooling is listed in Table 2 for each alloy, in comparison with its volume fraction and the temperature of maximum M 23 C 6 fraction. The phase fraction of M 23 C 6 carbide in the alloy 2927 (high Ceq and high fraction of M 7 C 3 ) was relatively low, compared with that of the other alloys. The amount of M 23 C 6 carbide was less in high-Ceq alloys than in low-Ceq alloys (with a high fraction of dendrite). The amount of M 7 C 3 reduction in the high dendrite fraction alloys (low-Ceq alloys 2124, 21,217, and 2427) was more than in high-Ceq alloys (2827 and 2927).
Additionally, another question arises: Is M 23 C 6 formed only due to the transformation [9][10][11] of M 7 C 3 to M 23 C 6 occurring at high temperatures? If so, alloys with a high fraction of M 7 C 3 have more possibility of transition than alloys with a low fraction of M 7 C 3 . Hence, the reduction in M 7 C 3 in the alloys with a high fraction of M 7 C 3 would be more than that in the alloys with a low fraction of M 7 C 3 during post-casting cooling or under equilibrium conditions. However, the alloys with the low fraction of M 7 C 3 (high fraction of dendrite) showed a relatively high fraction (including destabilized carbide) of M 23 C 6 , as listed in Table 2. Similar to the findings Pearce reported earlier [11], the result showed that the formation of M 23 C 6 may not result only from the transformation from M 7 C 3 to M 23 C 6 but also from the precipitation from the adjacent austenite.
The fractions of M 7 C 3 and M 23 C 6 did not always have a linear relationship with Ceq; for instance, the M 7 C 3 fraction of 2127 was the lowest among the alloys, whereas that of M 23 C 6 was highest. This might be related to the initiation of solidification with delta ferrite in the alloy. Delta ferrite in the alloy might contain a high concentration of Cr and C; thus, 18.1% of M 7 C 3 fraction at the fully solidified temperature (1285 • C) increased to 19.5% at the cost of delta ferrite reduction (from 9.9% at 1284 • C to 0 at 1205 • C). Below this temperature, M 23 C 6 precipitation increased due to the destabilization that allowed Cr and C to arise from the austenite matrix, as well as the reduction in a small portion of M 7 C 3 . The fraction of M 7 C 3 decreased to 2.4% at 900 • C, at which maximum precipitation of M 23 C 6 occurred. In particular, the fraction of M 23 C 6 in 2127 alloy was the highest, and that of M 7 C 3 was the lowest among the experimental alloys. Solidification of 2127 alloy begins with delta ferrite before austenite solidifies. The relatively small fraction of M 7 C 3 in 2127 alloy might be related to the delta ferrite. Additionally, it is possible to suppose that delta ferrite might have a high concentration of Cr and C. The consumption of Cr and C at the early stage of solidification may cause deficiency of the elements in the remaining liquid for sufficient eutectic reaction. Thus, the alloy (2127) may have less amount of eutectic reaction compared with the other alloys. This would be indirect evidence indicating that delta ferrite has a relatively high content of Cr. Unfortunately, it was impossible to measure the Cr concentration in delta ferrite due to the rapid transformation of delta ferrite to austenite and finally to martensite. However, based on the ThermoCalc calculation and microstructural observations, the above assumption of early consumption of Cr and C during delta ferrite formation might have caused a deficiency in solute elements in the remaining liquid; thus, a relatively small amount of M 7 C 3 and eutectic reaction occurred during the final freezing. In fact, the fraction of M 7 C 3 in 2127 was lower than that of the other alloys. The difference in M 23 C 6 precipitation among the alloys can be found in Figures 8 and 9 (DES + Temp of both AC and WQ). As listed in Table 2, the fraction of M 23 C 6 was the highest in 2127 alloy, while it was medium in 2124 and 2427, and low in 2827 and 2927 because of having little or no dendrites when destabilization occurred.

Basic Background of M23C6 Precipitation
According to the ThermoCalc prediction, the fraction of M23C6 carbide increases at the cost of M7C3 and austenite reduction at high temperatures. The fraction of M23C6 carbide in each alloy was inversely proportional to that of M7C3 carbide, as shown in Figure  1. Previous studies, mentioned above, revealed that M23C6 carbides precipitated during destabilization; thus, precipitation preferentially occurs in the dendrites [5,[9][10][11]. This also occurred in the present study.
Then, a question arises from the ThermoCalc results: How does the fraction of M23C6 increase with the reduction of M7C3 and austenite? This is shown in Figure 1 and Table 2. The precipitation-initiating temperature of M23C6 carbide during post-cast cooling is listed in Table 2 for each alloy, in comparison with its volume fraction and the temperature of maximum M23C6 fraction. The phase fraction of M23C6 carbide in the alloy 2927 (high Ceq and high fraction of M7C3) was relatively low, compared with that of the other alloys. The amount of M23C6 carbide was less in high-Ceq alloys than in low-Ceq alloys (with a high

Effect of Direct Aging (DA) on the M 23 C 6 Precipitation
On the basis of ThermoCalc calculation, direct aging of each alloy in the absence of destabilizing treatment or solutionizing was conducted at the temperature of the maximum fraction of M 23 C 6 , as shown in Table 2. Unlike conventionally heat-treated alloys, DAtreated alloys had M 23 C 6 precipitates in both dendritic and eutectic austenites of M 7 C 3, as displayed in Figure 10. This is significantly different from the conventionally heat-treated alloys, which are displayed in Figures 8 and 9, where M 23 C 6 precipitation occurred in the dendrite region of austenite but not in eutectic austenite. As mentioned above, M 23 C 6 precipitation during the conventional heat treatment is mainly caused by the destabilization of the saturated austenite, as expressed in Equation (2). This also leads to a difference in composition (or degree of saturation) between the dendrite austenite and that in the eutectic region. The primarily solidified austenite may have a saturation of solute elements such as Cr and C; however, the austenite in the eutectic region may have fewer solute contents due to Cr-rich M 7 C 3 carbide formation during the eutectic reaction. It is known that the transition or replacement of carbides from metastable M7C3 to stable M23C6 occurs during destabilization heat treatment [11,12]. It results in M23C6 shells surrounding the eutectic M7C3 carbide core [11]. They also reported that the transition from M7C3 to M23C6 increased with raising destabilization temperature [11]. The direct transformation from M7C3 to M23C6 was not readily observed in the present study. Although the transformation of carbides was not directly observed, the formation of M23C6 is expected to be caused by destabilization of austenite, and the reaction between matrix and M7C3 is expressed with Equation (2) as follows: where γ* is austenite with lower alloy content than that of the original matrix γ [4,8]. The interface of M7C3 had a relatively smooth interface and a narrow precipitationfree zone (PFZ) with fine M23C6 particles at the outer region of PFZ in the conventionally heat-treated alloys (in Figures 6-9). This means that the reaction of Equation (2) initiated at the interface of the carbide was directly contacted with a matrix-like peritectoid reaction It is known that the transition or replacement of carbides from metastable M 7 C 3 to stable M 23 C 6 occurs during destabilization heat treatment [11,12]. It results in M 23 C 6 shells surrounding the eutectic M 7 C 3 carbide core [11]. They also reported that the transition from M 7 C 3 to M 23 C 6 increased with raising destabilization temperature [11]. The direct transformation from M 7 C 3 to M 23 C 6 was not readily observed in the present study. Although the transformation of carbides was not directly observed, the formation of M 23 C 6 is expected to be caused by destabilization of austenite, and the reaction between matrix and M 7 C 3 is expressed with Equation (2) as follows: where γ* is austenite with lower alloy content than that of the original matrix γ [4,8].
The interface of M 7 C 3 had a relatively smooth interface and a narrow precipitation-free zone (PFZ) with fine M 23 C 6 particles at the outer region of PFZ in the conventionally heattreated alloys (in Figures 6-9). This means that the reaction of Equation (2) initiated at the interface of the carbide was directly contacted with a matrix-like peritectoid reaction [11]. Thus, the M 23 C 6 carbides appeared along the PFZ in the vicinity of eutectic M 7 C 3 reaction in Equation (2) due to dissolution of M 7 C 3 into the enveloped austenite PFZ at destabilizing temperature, followed by precipitation from the saturated austenite (PFZ) (Figure 11). The formation of a small gap of PFZ, which meant M 7 C 3 at the interface dissolved into the matrix (Equation (2) reaction), led to the precipitation of M 23 C 6 from the saturated PFZ.
Metals 2021, 11, x FOR PEER REVIEW 15 of 24 [11]. Thus, the M23C6 carbides appeared along the PFZ in the vicinity of eutectic M7C3 reaction in Equation (2) due to dissolution of M7C3 into the enveloped austenite PFZ at destabilizing temperature, followed by precipitation from the saturated austenite (PFZ) (Figure 11). The formation of a small gap of PFZ, which meant M7C3 at the interface dissolved into the matrix (Equation (2)  It is known that the eutectic M7C3/austenite interface does not act as a preferential heterogeneous site for M23C6 precipitation due to the following two reasons [9][10][11]: (1) formation of M23C6 at the shell of M7C3 along eutectic structure is perhaps a peritectoidtype reaction consuming both M7C3 and austenite and (2) the formation of M23C6 may not result only from the transformation of M7C3 to M23C6 but also from secondary precipitation from the adjacent matrix. As shown in Figures 6-9, the formation or precipitation of M23C6 in the dendritic regions and along the periphery of M7C3 occurred in all conventionally heat-treated alloys. However, precipitation of M23C6 in the eutectic austenite scarcely occurred in the conventionally heat-treated alloys in the present study.
The DA-treated alloys (2124, 2127, and 2427) with a relatively high fraction of the primarily solidified dendrites had both the reaction product M23C6 (between M7C3 and austenite) and destabilized product (from saturated dendritic and eutectic austenites). The M23C6 precipitation in the dendritic austenite occurred due to the destabilization of austenite, which led the solute elements to emerge to release saturation. However, some portion of M23C6 in the eutectic austenite is supposed to occur with the reaction between the matrix and M7C3 (Equation (2)) and destabilization (less than that in the dendritic austenite). As shown in Figure 1, the fraction M7C3 suddenly decreased at a high temperature below the freezing temperature and that of M23C6 took over the portion of M7C3 reduction. Though the inverse proportion between the fractions of stable M23C6 and metastable M7C3 might explain the partial transformation or replacement, as mentioned in the previous reports [11,12], the M23C6 precipitation began to occur in the vicinity or at the interface of M7C3, rather than in its core, at the early stage of DA (less than 60 min at each DA treatment), as shown in Figures 10-17. This phenomenon indicates that some portion of the precipitation was caused by Equation (2) reaction between M7C3 and austenite. In particular, precipitates around M7C3 in 240 min or 300 min DA exposure of each alloy (in Figures  12-17) were very fine, compared with coarse particles in the dendritic austenite that grew after destabilization. In fact, the total amount of M23C6 was composed of the precipitation amount in both dendritic and eutectic austenites resulting from destabilization, and the reaction product of Equation (2). M23C6 precipitation preferentially occurred in the dendritic austenite in the conventional heat treatment; however, it also occurred in the eutectic austenite during DA treatment, which might be due to the solubility limit in austenite It is known that the eutectic M 7 C 3 /austenite interface does not act as a preferential heterogeneous site for M 23 C 6 precipitation due to the following two reasons [9][10][11]: (1) formation of M 23 C 6 at the shell of M 7 C 3 along eutectic structure is perhaps a peritectoidtype reaction consuming both M 7 C 3 and austenite and (2) the formation of M 23 C 6 may not result only from the transformation of M 7 C 3 to M 23 C 6 but also from secondary precipitation from the adjacent matrix. As shown in Figures 6-9, the formation or precipitation of M 23 C 6 in the dendritic regions and along the periphery of M 7 C 3 occurred in all conventionally heat-treated alloys. However, precipitation of M 23 C 6 in the eutectic austenite scarcely occurred in the conventionally heat-treated alloys in the present study.
The DA-treated alloys (2124, 2127, and 2427) with a relatively high fraction of the primarily solidified dendrites had both the reaction product M 23 C 6 (between M 7 C 3 and austenite) and destabilized product (from saturated dendritic and eutectic austenites). The M 23 C 6 precipitation in the dendritic austenite occurred due to the destabilization of austenite, which led the solute elements to emerge to release saturation. However, some portion of M 23 C 6 in the eutectic austenite is supposed to occur with the reaction between the matrix and M 7 C 3 (Equation (2)) and destabilization (less than that in the dendritic austenite). As shown in Figure 1, the fraction M 7 C 3 suddenly decreased at a high temperature below the freezing temperature and that of M 23 C 6 took over the portion of M 7 C 3 reduction. Though the inverse proportion between the fractions of stable M 23 C 6 and metastable M 7 C 3 might explain the partial transformation or replacement, as mentioned in the previous reports [11,12], the M 23 C 6 precipitation began to occur in the vicinity or at the interface of M 7 C 3 , rather than in its core, at the early stage of DA (less than 60 min at each DA treatment), as shown in Figures 10-17. This phenomenon indicates that some portion of the precipitation was caused by Equation (2) reaction between M 7 C 3 and austenite. In particular, precipitates around M 7 C 3 in 240 min or 300 min DA exposure of each alloy (in Figures 12-17) were very fine, compared with coarse particles in the dendritic austenite that grew after destabilization. In fact, the total amount of M 23 C 6 was composed of the precipitation amount in both dendritic and eutectic austenites resulting from destabilization, and the reaction product of Equation (2). M 23 C 6 precipitation preferentially occurred in the dendritic austenite in the conventional heat treatment; however, it also occurred in the eutectic austenite during DA treatment, which might be due to the solubility limit in austenite depending on temperature. In gray iron [2], graphitization occurs due to the solubility limit of C in austenite depending on temperature and C content. Thus, the experimental alloys may have their own solubility limit depending on the temperature and nature of austenite. As mentioned earlier, the eutectic austenite may have relatively low contents of Cr and C, in contrast to those in the dendritic austenite, due to the formation of M 7 C 3 eutectic carbide during the eutectic reaction (which is expected to consume Cr and C). Thus, destabilization of the eutectic austenite was less active during the conventional heat treatment at 1065 • C; however, destabilization of eutectic austenite might be active even at relatively low temperatures in DA treatments.
Metals 2021, 11, x FOR PEER REVIEW 16 of 24 limit of C in austenite depending on temperature and C content. Thus, the experimental alloys may have their own solubility limit depending on the temperature and nature of austenite. As mentioned earlier, the eutectic austenite may have relatively low contents of Cr and C, in contrast to those in the dendritic austenite, due to the formation of M7C3 eutectic carbide during the eutectic reaction (which is expected to consume Cr and C). Thus, destabilization of the eutectic austenite was less active during the conventional heat treatment at 1065 °C ; however, destabilization of eutectic austenite might be active even at relatively low temperatures in DA treatments.   The PFZ along the periphery of M 7 C 3 in the conventionally heat-treated alloys lacks the alloying elements C and Cr, as reported earlier [3,4]. The dissolution of M 7 C 3 would occur preferentially at the interface during DA treatment, at the temperature at which maximum M 23 C 6 precipitation occurred and M 7 C 3 was unstable or metastable, which might then supply alloying elements to the PFZ. As time elapsed at high temperature (during DA treatment), the concentration of the alloying elements in the PFZ was saturated, followed by precipitation of M 23 C 6 . Therefore, M 23 C 6 precipitation along the periphery of M 7 C 3 was very active during DA treatment due to the stability of M 7 C 3 at the DA treatment temperature. Figure 18a,b display the interface morphology of M 7 C 3 in the DA-treated specimens. The reaction between M 7 C 3 and austenite occurred regardless of the alloy composition. As already shown in Figure 11, the interface had a very small PFZ gap between M 7 C 3 and M 23 C 6 fine precipitates. The small PFZ gap presumably formed by the dissolution of M 7 C 3 at the very interface, followed by M 23 C 6 precipitation from the saturated PFZ gap. Then, the M 23 C 6 precipitation depleted the solute elements in the PFZ, and thus, the reaction between M 7 C 3 and PFZ (austenite) progressively continued. In order to observe the progressive reaction and PFZ movement with exposure time, vacuum heat exposure on the DA-treated specimens was conducted at each DA treatment temperature for 60 more minutes. The DA-treated (for 240 min) 2124 and 2127 alloys were subjected to 60 more minutes under 1 × 10 −5 mmHg vacuum, followed by Ar gas fan quenching. The microstructural observations were carried out at the same area of each specimen before and after the vacuum heat exposure. In order to observe the progressive reaction and PFZ movement with exposure time, vacuum heat exposure on the DA-treated specimens was conducted at each DA treatment