Microstructure Evolution and Tempering Transformation Kinetics in a Secondary Hardened M50 Steel Subjected to Cold Ring Rolling

Abstract

The microstructure evolution and tempering transformation kinetics of the M50 steel subjected to cold ring rolling (CRR) have been investigated. The results indicate that the brass R{110}<110> texture is weakened with the enhancement of the <111>//ND texture during CRR. Due to the increased low angle boundaries by CRR, the A_c1 temperature decreases while the carbon content and volume fraction of RA increase. During tempering, the activation energy of carbon atoms segregation and transition carbide precipitation decrease, while the activation energy of retained austenite (RA) decomposition increases after CRR. The kinetic analysis shows that the CRR is beneficial to the carbon atoms segregation during the beginning of tempering. Then, the CRR leads to the delay of the onset of transition carbide precipitation, but decreases the whole reaction time, which has been verified by the transmission electron microscopy (TEM) and hardness results. The lagging of transition carbide precipitation in the early stage is caused by the increased segregation trapping of carbon atoms, while the higher nucleation rate is responsible for the enhanced precipitation of transition carbide during the later stage. For the cementite formation, there are no significant changes in the predictive kinetics after the applied CRR. However, the kinetic transformation of RA decomposition is inhibited by the CRR, which is attributed to the higher carbon content and smaller grain size of RA. Additionally, the alloy carbides precipitation is also enhanced by the CRR process during secondary hardening.

1. Introduction

As a typical secondary hardened material, M50 steel has been widely used in the aerospace industry as main shaft bearing in gas-turbine engines due to its excellent elevated temperature performance.^1,2) Compared with the extensively used AISI 52100 steel containing low alloy content, the numerous alloying elements in M50 steel play a major role in favoring the formation of fine precipitated carbides to achieve secondary hardening during subsequent tempering.³⁾ Since the tempering behaviors of M50 steel directly determine the final property, how to optimize the properties by precisely controlling the tempering transformation of M50 steel has always been an important issue.

Generally, M50 steel is treated by martensitic quenching prior to the tempering to obtain a microstructure comprised of dominant martensite, a substantial amount of retained austenite (RA) and a few undissolved carbides.⁴⁾ These microstructures give a high hardness over 60 HRC for M50 steel in spite the large austenite content.⁵⁾ During the subsequent tempering, Bridge et al.⁶⁾ showed that the start transformation of RA took place at a temperature higher than the low-alloyed steel, and the secondary hardening could be characterized by a slight increase in hardness. Moreover, Lqbal et al.⁷⁾ indicated that the peak secondary hardness was observed at the tempering temperature of 530°C in both triple and quintuple tempering. The work of Fischmeister et al.⁸⁾ further explained that the tempering to peak hardness could lead to the formation of about 3 vol.% of precipitation. Recently, Hopkin et al.⁹⁾ studied the secondary carbides in M50 steel by combination of APT and TEM. They found that the cementite formed first during tempering, and then the carbon segregation assisted nucleation of Mo-richly M₂C and V-richly M₄C₃ carbides, which are more stable, and hence grow at the expense of the cementite. By summarizing the existing researches, it is found that the influence of the pre-treatment process on the tempering behaviors has not been considered, though there are plenty of detailed researches focused on the tempering response of M50 steel.

The cold ring rolling (CRR) technology, as an advanced forming process of bearing rings, has been proven to exert a crucial influence on the subsequent heat treatment and final mechanical property.^10,11,12) Ryttberg et al.¹³⁾ found the cold deformation occurs in the ferrite matrix, and the spheroidized carbides were nearly undeformed for the cold rolled bearing ring. When the imposed cold deformation is quite severe, the mismatch of strain between the carbides and ferrite could result in the opening-up of voids at the carbides/ferrite interfaces, which are detrimental to the final performance. However, the works of Li et al.¹⁴⁾ suggested that the imposition of 30% cold deformation combined with quenching and tempering process would increase the impact toughness due to the refined prior austenite grain size. In addition, Chakraborty et al.¹⁵⁾ and Lu et al.¹⁶⁾ showed that moderate cold deformation was effective in improving the impact toughness of bearing steel by the phase refinement. Moreover, Beswick¹⁷⁾ and Nanesa¹⁸⁾ indicated that the prior cold rolling would facilitate the ferrite to austenite transformation and then decrease the martensite start temperature. By reviewing the above studies, it can be found that the influence of prior cold rolling process on the quenched microstructure and mechanical properties has been extensively investigated in steels. Besides, some other researchers also focused on the effect of prior cold rolling on the direct aging behaviors without quenching. They proposed that the high-density dislocations induced by cold deformation could accelerate the aging response by promoting the precipitation formation.^19,20,21) However, the tempering process followed by quenching, as the important stage to realize secondary hardening, has not been systematically investigated in the context of cold ring rolling. Moreover, since the CRR process will directly affect the quenched microstructure, it is reasonable to believe that the altered quenched microstructures induced by CRR will also have inevitable consequences for the subsequent tempering transformation.

In view of these, the aim of the present work is to investigate microstructure evolution and tempering behaviors of a secondary hardened M50 steel subjected to CRR. As well, the kinetics model that explains the tempering transformation for the M50 steel without and with CRR is proposed. Lastly, the influencing mechanism of CRR on the whole tempering process of M50 steel is discussed in detail.

2. Experiments and Method

2.1. Specimen Preparation

The M50 steel used in this study with the chemical composition is given in Table 1. The material was received as spheroidize-annealed bar with a diameter of 60 mm and showing an initial microstructure of primary carbides embedded in ferritic matrix. The initial ferrite grain size of the tested material is about 15–25 μm and the Vickers hardness is approximately 178 HV.

Table 1. Chemical composition of M50 steel (wt.%).

C	Cr	Mo	V	Mn	Si	Ni	Cu	Fe
0.82	4.00	4.20	1.05	0.32	0.20	0.10	0.06	Bal.

The cold ring rolling experiments were conducted using a self-designed radial ring rolling machine (D56G90). The general forming principle of the cold ring rolling as applied in this machine is shown in Fig. 1. In this principle, the driven roll makes active rotation and linear feed motion, and then the idle roll rotates freely due to the friction force between the ring and the tool. With the cooperative motion of the driven roll and idle roll, a ring blank is continuously bitten into the radial rolling cavity. The guide roll is applied to stabilize the ring during forming and is required to guarantee the final roundness of the ring. Besides, oil is continuously flowing to the ring to decrease the sliding friction during ring rolling and limit the temperature rise. In order to investigate the effect of cold ring rolling process on the microstructure evolution, the rings were rolled to different final dimensions as indicated in Table 2. The final thickness reductions of cold ring rolling were respectively determined to be 14%, 22%, 32% and 40%, which is referred as CRR1, CRR2, CRR3 and CRR4 in this study (Fig. 2).

Fig. 1.

The forming principle of cold ring rolling. (Online version in color.)

Table 2. The final dimension of the cold rolled rings with different thickness reductions.

Specimens	Out diameter (mm)	Inner diameter (mm)	Thickness (mm)	Thickness reduction (%)
Without CRR	54.50	34.50	10.00	0
CRR1	57.76	40.62	8.57	14
CRR2	61.40	45.80	7.80	22
CRR3	66.22	52.58	6.82	32
CRR4	72.28	60.28	6.00	40

Fig. 2.

Cold rolled rings with different thickness reductions.

After cold ring rolling tests, the rolled rings were austenitized at 1090°C for 30 min in a vacuum furnace and followed by oil quenching at 60°C. To investigate the tempering behaviors of the M50 steel subjected to cold ring rolling, the as-quenched specimens were finally subjected to the vacuum tempering processes with different tempering temperatures (100, 200, 300, 400 and 500°C) for 2 h.

2.2. Microstructure Investigation

The microstructure and texture of the specimens without CRR and with CRR4 were examined using a JEOL JSM 6500F electron microscope (SEM) equipped with an electron backscatter diffraction (EBSD) detector. The data was recorded with a scan step size of 0.5 μm and analyzed using the HKL CHANNEL 5 software. The specimens used for EBSD analysis were prepared by mechanical grinding and polishing with diamond suspension down to 1 μm, followed by vibration polishing using colloidal silica suspension.

The dilatometry tests were performed in a Bähr DIL 805A dilatometer from ambient temperature to 1090°C at a heating rate of 10°C/min. The cylindrical dilatometry specimens with a length of 10 mm and a diameter of 4 mm were cut from the center of the cold rolled rings. The A_c1 temperature of the specimens subjected to different thickness reductions were determined based on the tangent method.²²⁾ Besides, to evaluate the volume fraction and lattice parameter of RA after quenching, the X-ray diffraction (XRD) data were recorded with a scanning speed of 1°/min on a Rigaku D/MAX-RB diffraction analyzer at 12 kW. The XRD diffraction profiles were fitted by Lorentz function to obtain the intensities and angles of diffraction peaks. The volume fraction of RA was evaluated by the following equations:²³⁾   

$$V_{\alpha }+V_{\gamma }=1$$(1)

  

$$V_{\gamma }=\left(1.4I_{\gamma }\right)/\left(I_{\alpha }+1.4I_{\gamma }\right)$$(2)

where V_α and V_γ represent volume fractions of martensite and austenite, respectively; I_γ represents the average intensity of diffraction peaks of 200_γ, 220_γ and 311_γ; and I_α represents the average intensity of diffraction peaks of 200_α and 211_α. The average lattice parameter of austenite was obtained based on the angles of diffraction peaks of 200_γ, 220_γ and 311_γ.

2.3. Kinetic Analysis

To determine the tempering kinetic parameters, the differential scanning calorimetry (DSC) tests were performed using a PerkinElmer Pyris 1 calorimeter. The specimens for DSC tests with a thickness of 0.5 mm and diameter of 4 mm were cut from the as-quenched rings, and then heated from ambient temperature to 500°C under a vacuum condition at different heating rates of 5, 10, 15, 20°C/min, respectively. Meanwhile, a rerun heating process was performed at the same heating rate to exclude the interfering peaks.²⁴⁾

There are several methods to study the tempering transformation kinetics and determine the activation energies associated with the different tempering stages. In this work, two models, Kissinger model^25,26) and isoconversional model,^27,28) were applied. The Kissinger mode is based on the fact that the peak temperature depends on the heating rate. The effective activation energies during different tempering stages can be calculated by using Kissinger equation:²⁵⁾   

$$\text{In}\left(T_P^2/\varnothing \right)=Q/R{T}_P+const$$(3)

where T_P is the peak temperature, Ø is the heating rate, and R is the gas constant (R=8.314 kJ/mol). By plotting $\text{In}\left(T_P^2/\varnothing \right)$ as a function of 1/T_P, the effective activation energy (Q) can be obtained.

The differential isoconversional model was also employed to obtain the kinetic parameters for different tempering stages. In this model, the reaction rate is assumed to be a function of temperature (k(T)) and fractional conversion function (f(α)), which can be expressed as:²⁹⁾   

$$d\alpha /dt=k\left(T\right)f\left(\alpha \right)$$(4)

where α and T represent the converted fraction and temperature, respectively. k(T) can be obtained by the Arrhenius equation as follow:   

$$k\left(T\right)=A\cdot \text{exp}\left(-E_{\alpha }/RT\right)$$(5)

where A and E_α are the pre-exponential factor and activation energy, respectively. By substituting Eq. (5) into (4) and taking the equation to a logarithm, the differential isoconversional method can be proposed as follow:   

$$\text{In}\left(d\alpha /dt\right)=ln\left[A\cdot f\left(\alpha \right)\right]-E_{\alpha }/RT$$(6)

For the non-isothermal process with a constant heating rate Ø, Eq. (6) can be expressed as:   

$$\text{In}\left[\varnothing \cdot \left(d\alpha /dt\right)\right]=ln\left[A\cdot f\left(\alpha \right)\right]-E_{\alpha }/RT$$(7)

By plotting In[Ø·(dα/dt)] as a function of 1/T, the slope indicates a value of –E_α/R and the intercept for ln[A·f(α)], and then the curve of the activation energy varying with the converted fraction can be determined.

Furthermore, using the kinetics parameters of E_α and A·f(α) obtained from Eq. (7), the kinetic prediction of the converted fraction at any specific temperature can be made as Eq. (8):   

$$t_{\alpha }={\int }_0^{t_{\alpha }}dt={\int }_0^{\alpha }\left\{1/\left[A_{\alpha }f\left(\alpha \right)\text{exp}\left(-E_{\alpha }/RT\right)\right]\right\}d\alpha$$(8)

Then, the relationship between conversional fraction α and isothermal times t can be described using the empirical approach of JMAK theory²⁵⁾ involving the kinetic parameters of k and n.   

$$\alpha =1-\text{exp}\left(-k{t}^n\right)$$(9)

2.4. Validation of Kinetic Analysis

In order to verify the tempering transformation behaviors, the Vickers hardness of the specimens subjected to different tempering temperatures was measured by using the HV-1000 A Vickers hardness tester under a load of 4.9 N and a dwell time of 10 s. Noted that the test points were distributed at the center of the ring along the thickness direction, and at least eight indents were applied at each condition to obtain the error bar. Additionally, the microstructure of the specimens tempered at 150°C for 10 min were characterized by a transmission electron microscope (TEM, JEOL 2100 F). The thin foil specimens for TEM experiments were prepared on a twin-jet polisher at 20 V using a solution of 10% perchloric acid and 90% acetic acid.

3. Results

The EBSD inverse pole figure (IPF) maps, misorientation angle distribution and {111} pole figures from the specimens without CRR and with CRR4 are presented in Fig. 3. From the initial microstructure of ring blanks (Fig. 3(a)), it is found that the equiaxed ferrite grains with uniform intragranular orientation can be found, and a large amount of carbides are embedded within the grains. After CRR4, the IPF map is obtained at the central region of RD-TD surface of the rolled rings, as shown in Fig. 3(b). Obviously, the deformed ferrite grains are elongated along the rolling direction (RD). Figures 3(b) and 3(e) further show a comparison of the misorientation angle distribution before and after CRR4. It can be seen that the proportion of low angle grain boundaries (2° < θ < 15°) in the specimens without CRR is 18.1%, while that of the rolled specimens reaches 66.3%. This indicates a significant increase of relative frequency of low angle grain boundaries after CRR4, which can be ascribed to the strain heterogeneity in the distorted ferrite grains.³⁰⁾ The {111} pole figures from the specimens without and with CRR4 are shown in Figs. 3(e) and 3(f). The texture analysis of the ring blanks shows a relatively strong brass R{110}<110> texture, which may be remained from the forging forming of initial bars. However, the amount of brass texture decreases with the increase of the <111>//ND texture after the applied CRR4. The maximum texture intensity also increases from 3.61 to 5.08 after CRR4, indicating a developed sharp rolling texture in the rolled rings.

Fig. 3.

EBSD images showing the IPF maps taken along the RD-TD plane, misorientation angle distribution and {111} pole figures from the specimens (a–c) without CRR and (d–f) with CRR4. (Low angle grain boundaries: 2° < θ < 15°; RD: rolling direction; TD: transverse direction; ND: normal direction). (Online version in color.)

To investigate the effect of CRR on the austenitization process of M50 steel, the A_c1 temperatures of the specimens with different thickness reductions are determined from the heated-up dilatation curves, as shown in Fig. 4. According to the magnification drawing in Fig. 4(a), the A_c1 temperature obviously decreases from 818.8°C to 806.1°C after the applied CRR4. Meanwhile, it is found that the specimens subjected to CRR4 exhibits a higher gradual expansion during the heating process before the A_c1 temperature, which should be ascribed to the fact that more relaxation of stresses occurs during the annealing process for the specimens subjected to CRR. The variations of A_c1 temperature with increasing thickness reduction are further summarized in Fig. 4(b), where it is illustrated that the A_c1 temperature gradually decreases with the increasing thickness reduction. During the transformation of ferrite to austenite, the presence of the low angle boundaries in the microstructure could be the favored nucleation sites to accelerate austenite formation upon the heating process.¹⁷⁾ It has been shown in Fig. 3 that the high density of low angle boundaries is generated by the CRR process. Therefore, the CRR process makes the transforming from ferrite to austenite easier, thereby leading to the decrease of A_c1 temperature.

Fig. 4.

(a) The dilatation curves for the specimens without CRR and with CRR4. (b) the determined A_c1 temperatures for the specimens with different thickness reductions. (Online version in color.)

Figure 5 shows the prior austenite grain boundaries of the as-quenched specimens without and with CRR4. The average grain size of prior austenite was apparently refined from 13.5 μm to 7.9 μm after subjecting to CRR4, which could be attributed to the increasing nucleation sites (dislocation and low angle grain boundary) supplied by the CRR during the ferrite to austenite transformation.

Fig. 5.

The prior austenite grain boundaries of the as-quenched specimen (a) without CRR and (b) with CRR4. (Online version in color.)

Figure 6(a) shows the XRD patterns for the as-quenched specimens without CRR and with CRR4. In addition to the dominant diffraction peaks of martensite (α) and austenite (γ), the weak diffraction peaks of MC and M₂C carbides is observed in the XRD patterns as well. The results confirm that the undissolved carbides mainly consist of MC and M₂C after the austenitisation process of 1090°C, which thus coincides with the experimental results reported by Bridge.⁶⁾ Meanwhile, the peak intensities of carbides distinctly decrease after the applied CRR4, indicating that more primary carbides have dissolved into the matrix. Quantitative results of volume fraction and lattice parameter of RA are shown in Fig. 6(b). With the increase of thickness reduction, the volume fraction of RA rises from 26.2% to 31.2% corresponding with an increase of lattice parameter of RA. It is known that the lattice parameter of RA is positively related with its carbon content.³¹⁾ The increasing carbon content in RA should be mainly due to the improvement of carbide dissolution induced by the CRR. It has been reported that austenite is preferentially nucleated at the junction of carbides and ferrite grain boundaries during austenitization.³²⁾ For the spheroidised microstructure subjected to CRR, the surface to volume ratio between carbides and matrix will increase due to the fragmentation of carbides, thereby leading to the increase of overall interfacial energy of carbide-ferrite.³³⁾ During the transformation of ferrite to austenite, the unstable carbide is prone to dissolve at the interfacial of carbide-ferrite to accelerate the austenite transformation. Consequently, the carbide dissolution behaviors would be accelerated for the specimens subjected to CRR. In addition, the increased carbon content of austenite will make it thermally stable and inhibit the transformation into martensite during quenching, which should be responsible for the promoted volume fraction of RA. On the other hand, the refinement of prior austenite grain could result in a decrease of martensite transformation start (M_s) temperature,³⁴⁾ thereby decreasing the undercooling degree of the transformation from austenite to martensite. As a result, more austenite will be retained after quenching to ambient temperature. It is therefore inferred that the finer austenite grain may also be responsible for the increased RA in the specimens subjected to CRR.

Fig. 6.

(a) The XRD patterns for the as-quenched specimens without CRR and with CRR4. (b) The volume fraction and lattice parameter of RA for the specimens with different thickness reductions. (Online version in color.)

By summarizing the above results, it is easy to understand that the CRR process would exert a nonnegligible impact on the austenitisation process and quenched microstructure of secondary hardened M50 steel. Nevertheless, how the various quenched microstructure induced by CRR affect the tempering behaviors is still not clear. In this work, the tempering transformation kinetics of the specimens without and with CRR are compared, with the aim of clarifying the influencing mechanism of CRR on the tempering process. Figure 7(a) displays the DSC curves of the specimens without CRR at a heating rate of 10°C. After subtracting the rerun curves from the first run curves, the revised DSC curves can be obtained. As shown in Fig. 7(b), the revised DSC curves of the specimens without CRR at different heating rates are summarized, where it can be seen that transformation peak temperature slightly increases with the increase of heating rate.

Fig. 7.

(a) The DSC curves of the specimens without CRR at a heating rate of 10°C/min. (b) The summary of revised DSC curves of the specimens without CRR at different heating rates. (Online version in color.)

To identify the different stages of tempering, a Gaussian multi-peaks fitting method was employed for analyzing the revised DSC curves. As shown in Fig. 8, the revised DSC curves are fitted by five exothermic peaks. Consistent with our previous dilatometry results,³⁵⁾ the fitted peaks correspond to five different stages during tempering of M50 steel. (I) The first stage, referred to as the “pre-precipitation process” and occurring below 100°C, consists of the segregation of carbon atoms to lattice defects and the formation of clusters of carbon atoms in the matrix. (II) The second stage, associated with the precipitation of transition carbides, occurs at the temperature above 80°C. (III) The third stage, taking place at the temperature above 250°C, involves the conversion of the transition carbide into cementite. (IV) The fourth stage, occurring at a temperature higher than the low-alloyed steel, relates to the decomposition of retained austenite. (V) The fifth stage, assigned to the dissolution of cementite particles and the formation of more stable alloy carbides.

Fig. 8.

The revised DSC curve and fitted peaks for the specimens without CRR at a heating rate of 10°C/min. (Online version in color.)

Based on the fitted results of the revised DSC curves, the peak temperature (T_P) of different tempering stages can be determined. Then, the activation energy (Q) can be calculated using the Kissinger method by substituting the various peak temperatures at different heating rates. The calculated activation energy for different tempering stages as a function of the thickness reduction are illustrated in Fig. 9. It is observed that, with the increase of thickness reduction to 40%, the actvation energy of stage I shows a decrease from 99.3 kJ/mol to 75.6 kJ/mol. For the stage II, the actvation energy gradually decreases from 130.7 kJ/mol to 106.4 kJ/mol with the increase of thickness reduction as well. This demonstrates that the energy barriers of carbon atoms segregation and transition carbide precipitation both are decreased after the applied CRR process. However, the stage III, which refers to the cementite formation process, seems to be not sensitive to the increasing thickness reduction of CRR. For the stage IV, the activation energy of RA decomposition significantly increase from 147.4 kJ/mol to 188.4 kJ/mol after the applied CRR4 process. Previous researches³⁶⁾ have indicated that the activation energy of RA decomposition is closely related to the thermal stability of RA during tempering. Moreover, the increase of carbon content in RA is beneficial to the imprvoment of its stability.^37,38) Since it have proved that the the CRR process could increase the carbon content of RA after quenching (Fig. 6), the increased activation energy of RA decompostion could be well interpred by the carbon enrichment of RA induced by CRR. The activation energy of alloy carbides formation (stage V) is remarkably higher than that of other stages, which ranges from 357.5–380.3 kJ/mol. This may be acribed to the fact that the formation of highly stable alloy carbides requires more potential energy during tempering. Moreover, the activation energy of stage V shows a slight decrease with the increase of thickness reduction, thereby verifying that the CRR process contributes to the alloy carbides formation during tempering of M50 steel.

Fig. 9.

The calculated activation energy (Q) for different stages determined by the Kissinger method. (Online version in color.)

By using the differential isoconversional model, the the fraction converted for different rates as a function of temperature can be obtained, as typically shown in Fig. 10. The results show that the transformation temperature range of each stage shifts toward higher temperature with the increase of heating rate. Meanwhile, based on the differential isoconversional model as plotted in Eq. (7), the isothermal transformation of different stages as a function of holding time are predicted in the form of JMAK theory, as indicated in Fig. 11.

Fig. 10.

The fraction converted during stage II as a function of temperature for different heating rates: (a) without CRR, (b) with CRR4. (Online version in color.)

Fig. 11.

The predictive fraction converted of different stages as a function of holding time using the differential isoconversional model for the specimens without CRR and with CRR4: (a) isothermal tempering at 50°C for stage I, (b) isothermal tempering at 150°C for stage II, (c) isothermal tempering at 250°C for stage III, (d) isothermal tempering at 350°C for stage IV, (e) isothermal tempering at 450°C for stage V. (Online version in color.)

Furthermore, the predictive kenetics of different stages at specific temperatures are obtained using the differential isoconversional model. Figure 11(a) presents the predictive kenetic of stage I at the isothermal tempering of 50°C for the specimens without and with CRR4. It is found that the “pre-precipitation process” is easily to occur and the Avrami exponent n decreases from 0.637 to 0.508 for the specimens subjected to CRR4. Compared with the specimens without CRR, the transition carbide precipitation of the specimens subjected to CRR4 lags behind in the initial stage, while the reaction is completed in a shorter time (Fig. 11(b)). The Avrami exponent n for the specimens without and with CRR4 are determined to be 0.631 and 0.443, respectively. For the stage III related to the process of cementite formation, there are no significant change in the predictive kinetics after the applied CRR4. Meanwhile, the Avrami exponent n are close for the specimens without and with CRR4, indicating that the kinetic parameters of cementite formation will not be affected by the CRR process. The comparsion of RA decomposition kenetics at 350°C for the specimens without and with CRR4 are also illustrated in Fig. 11(d), where it can be seen that longer isothermal time will be required to realize the same transformed fraction after the applied CRR4. Moreover, the Avrami exponent n increases from 0.503 to 0.641 for the specimens subjected to CRR4. As shown in Fig. 11(e), the kenetic analysis of stage V indicates that the alloy carbides will precipitate in a shorter time during a isothermal process of 450°C for the specimens subjected to CRR4.

In order to verify the kinetic results of transition carbide precipitation, the TEM micrographs of the specimens tempered at 150°C for 10 min are presented in Fig. 12. The lath martensite (M) and blocy RA both can be observed in the microstructure, while the specimens subjected to CRR exhibit finer martensite packets. In addition, for the specimens without CRR, some ultrafine precipitates are distributed within the martensite along the same direction. The precipitates, which are identified as transition carbides (ε-Fe₂C) by selected area electron diffraction (SAED) pattern inserted in Fig. 12(a), are short rod-like shape approximately with a length of 150 nm and a width of 20 nm. However, there are almost no carbides precipitated in the specimens subjected to CRR. The results thus verifies that the precipitation of transition carbides is restrainted by the CRR at the begining of stage II, which is well consistent with the kinetic prediction as shown in Fig. 11(b).

Fig. 12.

The TEM micrographs of the specimens after tempering at 150°C for 10 min (a) without CRR and (b) with CRR4. (Online version in color.)

The variations of Vickers hardness with increasing tempering temperature for the specimens with CRR and with CRR4 are measured, as provided in Fig. 13. As can be seen, the as-queched specimens subjected to CRR4 exhibts a higher hardness than that of the as-quenched specimens without CRR. This can be explained by the fact that the improved carbide dissolution induced by CRR leads to the carbon enrichment in qenched martensite, which will provide higher hardness to the matrix due to the increased solution strengthening effect.³⁹⁾ Moreover, it should be noted that the hardness of martensite is not only dependent on the solution strengthening but also decided by the boundaries strenthening.⁴⁰⁾ As a result of the refinement of martensite microstructure induced by CRR (discussed in Fig. 13), the increase of martensite boundaries should also be responsible for the higher hardness in the as-quenched specimens with CRR. After tempering of 100°C, the hardness shows a slight decrease regardless of the applied CRR or not. This may be attributed to the decrease in solution strenthening of martensite due to the formation of carbon cluster or the carbon segregation to dislocations and boundaries. With the increase of tempering temperature to 200°C, the hardeness of the specimens without and with CRR4 both significantly decrease. It has been shown from the DSC results (Fig. 8) that the precipitation of transition carbide mainly occurs at the tempering temperature of 200°C. Therefore, the significant weakening of solution strengthening resulted from the lower soluted carbon in matrix should be the main reason for the decrease of hardness, although the transition carbide precipitation will play a role of precipitation strengthening. Moreover, at the tempering temperature of 200°C, the hardness of the specimens subjected to CRR4 decreases by 109.4 HV, while the hardness of the specimens without CRR only falls by 76.0 HV. It can be inferred that more transition carbides will precipitate in the specimens subjected to CRR4 due to the more significant decrease in solution strengthening, which is also well agree with the kinetic predication in Fig. 11(b). When the tempering temperature increases to 300°C, the hardness of the specimens further decreases regardless with the apllied CRR. This indicates that the soluted carbon atoms of martensite is comsumed by the cementite formation, thereby further weaking the solution strengthening effect. With the increase of tempering temperature to 400°C, the hardness of the specimens exhibits a slight increase trend, which should be associated with the stage of RA decomposition. After tempering of 500°C, the increase of hardness can be observed in the specimens without and with CRR, which could be attributed to the secondary hardening effect caused by the precipitation of nano-scale alloy carbides. In addition, the specimens sujected CRR4 shows a more significant increase in hardness, thus verifying that the CRR is favorable for the alloy carbides precipitation as indicated in Fig. 11(e).

Fig. 13.

The Vickers hardness as a function of tempering temperature for the specimens without CRR and with CRR4. (Online version in color.)

4. Discussion

In the specimens subjected to CRR4, the carbon atoms segregation is enhanced during the beginning of tempering. Figure 14 illustrates the EBSD grain/phase maps for the specimens after tempering of 500°C. It can be found that the specimen subjected to CRR4 exhibits finer phase packets, which confirms that CRR leads to the finer martensite microstructure after quenching. Figure 5 has shown that the prior austenite grain has been significantly refined after the applied CRR. The finer prior austenite grain may play a role as obstacles to the growth of the martensite, and thus refine the fresh martensite microstructure.^41,42) Besides, the increase in nucleation frequency of martensite in the finer austenite grain boundaries could also contribute to the finer martensite. Consequently, the CRR process is expected to result in a higher density of grain or phase boundaries in the microstructure. Askeland⁴³⁾ and Kang⁴⁴⁾ have indicated that carbon atoms are also likely to segregate to the grain boundaries or lath interfaces during stage I, due to the fact that these sites have higher lattice distortion energy than the body of grain or phase. Then, it is easy to understand why the carbon atoms segregation is enhanced for the as-quenched specimens subjected to CRR4.

Fig. 14.

The EBSD grain/phase maps for the specimens after tempering of 500°C (a) without CRR and (b) with CRR4. (Online version in color.)

In addition, the precipitation process of transition carbide lags behind in the beginning stage, as compared with the specimens without CRR (Figs. 11(b) and 12). It has been reported that the boundary trapping of carbon atoms during the stage may inhibit the subsequent transition carbide precipitation because the state of segregated carbon to the dislocations is more stable than that of an iron carbide precipitate.⁴⁵⁾ Thus, the increase of grain or phase boundaries may be also responsible for the lagging of transition carbide precipitation due to the increased carbon segregation during stage I.

Though the transition carbide precipitation lags behind in the early stage, the whole reaction is completed in a shorter time for the specimens subjected to CRR4 (Fig. 11(b)). As well, the hardness results reflect that the amount of precipitated transition carbides is increased by the CRR process at a tempering process of 2 h (Fig. 13). These results should be mainly attributed to the nucleation behavior of transition carbides. The Turnbull’ theories have indicated that the nucleation rate (I) of precipitation can be expressed by:⁴⁶⁾   

$$I=C_{\alpha }{N}_0\frac{KT}{h}\text{exp}\left(-\frac{G+Q}{RT}\right)$$(10)

where, C_α represents the concentration of the carbide forming element in matrix, N₀ is the number of nucleation sites per unit volume, K is the Boltzmann constant, h is the Planck constant, G is the critical nucleation energy, Q is the activation energy for solute diffusion and R is gas molar constant.

In the present work, since the higher density of phase interfaces can be obtained in the specimens subjected to CRR, it is anticipated that number of nucleation sites per unit volume (N₀) will be higher. On the other hand, the XRD analysis in Fig. 6 have indicated that the carbide dissolution is promoted by the CRR process. There will be a larger amount of carbon atoms dissolved in the martensitic matrix for the specimens subjected to CRR. This means that the higher content of the carbide forming element (C_α) will be contained within the quenched martensite to prepare for the subsequent tempering process. As plotted in the Eq. (10), it can be concluded that the increase of N₀ and C_α induced by CRR together contributes to the enhanced nucleation rate (I) of transition carbide precipitation.

Considering the above combining effects of increased carbon segregation and improved nucleation rate, the reason why the transition carbide precipitation is first reduced and then promoted by the CRR can be explained. At the beginning of stage II, the precipitation of transition carbide is delayed due to the increased segregation trapping of carbon atoms. However, once the trapping sites of carbon atoms are enriched, the reaction rate of transition carbide precipitation will increase rapidly because of the higher density of nucleation sites and higher carbon content in matrix. Finally, the amount of precipited transition carbides increases obviously and thus leads to a sharper decrease of hardness for the specimens subjected to CRR.

In the case of cementite formation, since the cementite is formed by the the transformation from metastable transition carbides,⁴⁷⁾ the increased nucleation sites resulted from CRR will exert little influence on this process. In addition, for the specimens subjected to CRR, due to the enhanced precipitation of transition carbide prior to cementite formation, the martensitic matrix will no longer contain higher content of soluted carbon than that of the specimens without CRR. This indicates that the there will be no significant difference in the content of cementite forming elements (Fe or C) after the transition carbide precipitation. As a result, the kinetics of cementie formation is not sensible to the CRR process in M50 steel (Fig. 11(c)).

It is well known that the RA enriched with more carbon content exhibits a higher termally stability during tempering. In the present work, Fig. 6 has clearly indicated the CRR leads to the carbon enrichment of RA after quenching, which should be the main reason for that the kinetic transformation of RA is inhibited by CRR (Fig. 11(d)). Moreover, Jimenez et al.⁴⁸⁾ suggested that the finer packets structure can be achieved during a lower austenitization temperature, which will leads to the increasing thermal stability the RA. In this work, since the packets is significantly refined by the CRR (Fig. 14), the resulting finer grain of RA is expected to increase the stacking fault energy (SFE) of the RA at a given temperature.⁴⁹⁾ As the SFE increases, it will act as a barrier to the further transformation of RA, thus making the RA more stable.⁵⁰⁾ This means that the more energy will be required for the transformation of finer RA, which is the reason why the specimens subjected to CRR exhibits a higher activation energy of RA decomposition. Therefore, the delayed tranformation kinetic of RA as a result of CRR can be attributed to its higher carbon content and smaller grain size of RA.

The alloy carbides precipiation as the major tempering stage realizing secondary hardening will directly determin the final properties of M50 steel. In the present work, the predictive kinetic indicates that the alloy carbides precipitation is enhanced by the CRR process. Since the grain boundaries or lath interfaces is known as the nucleation sites of alloy carbides precipitation, the higher density of boundaries in the specimens subjected to CRR (Fig. 14) is considered to be the main reason for the enhanced precipitation of alloy carbides.

5. Conclusion

In this work, the microstructure evolution and whole tempering transformation of the scondary hardened M50 steel without and with CRR were investigated, and following conclusions can be drawn:

(1) The proportion of low grain boundary increases from 18.1% to 66.3% due to the applied CRR. The amount of brass R{110}<110> texture decreases with the increase of the <111>//ND texture for the specimens subjected CRR4. Meanwhile, the maximum texture intensity increases from 3.61 to 5.08, indicating a developed sharp rolling texture in the rolled rings.

(2) The CRR not only results in the decrease of A_c1 temperature during austenisation, but also increases the carbon content and volume fraction of RA after quenching. The weakened diffraction peaks of undissolved carbides and increased carbon content of RA together confirm that the carbide dissolution is enhanced by the CRR.

(3) During the tempering, the calculated results using Kissinger model show that the activation energy of carbon atoms segregation and transition carbide precipitation gradually decrease with the increasing thickness reduction, while no significant change is observed in the activation energy of cementite formation. Additionally, the activation energy of RA decomposition significantly increases from 147.4 kJ/mol to 188.4 kJ/mol and the activation energy of alloy carbides formation decreases from 380.3 kJ/mol to 357.5 kJ/mol after the applied CRR.

(4) The kinetic transformation predicted by isoconversional model indicates that the CRR is beneficial to the carbon atoms segregation during the beginning of tempering. Moreover, the CRR leads to a delay on the onset of transition carbide precipitation, but makes the whole reaction completed in a shorter time, which has been verified by the TEM and hardness results. The lagging of transition carbide precipitation in the beginning stage is mainly caused by the increased segregation trapping of carbon atoms, while the higher nucleation rate induced by CRR is responsible for the enhanced precipitation of transition carbide during the later stage.

(5) For the process of cementite formation, there are no significant change in the predictive kinetics after the applied CRR, demonstrating that the CRR has little influence on the cementite formation process. The kinetic transformation of RA decomposition is obviously inhibited by the CRR, which can be attributed to the higher carbon content and smaller grain size of RA. In addition, the CRR is favorable for the alloy carbides precipitation during secondary hardening.

Acknowledgements

The work was supported by the National Natural Science Foundation of China (No. 51875426), 111 Project (B17034), Innovative Research Team Development Program of Ministry of Education of China (No. IRT_17R83) and Important Science and Technology Innovation Program of HuBei Province (No. 2019AAA001).

Conflicts of Interest Statement

The authors declared that they have no conflicts of interest to this work.

References

• 1)   W. F.  Liu,  Y. F.  Cao,  Y. F.  Guo,  M. Y.  Sun,  B.  Xu and  D. Z.  Li: J. Mater. Sci. Technol., 38 (2020), 170.
• 2)   F.  Wang,  D. S.  Qian,  L.  Hua,  H. J.  Mao and  L. C.  Xie: Sci. Rep., 9 (2019), 11315.
• 3)   T. A.  Harris,  M. A.  Ragen and  R. F.  Spitzer: Tribol. Trans., 35 (1992), 194.
• 4)   H. K. D. H.  Bhadeshia: Prog. Mater. Sci., 57 (2012), 268.
• 5)   H. E.  Kline: Technical Report 22-0751, Airsearch Manufacturing Company of Arizona, Arizona, (1978), 1.
• 6)   J. E.  Bridge,  G. N.  Maniar and  T. V.  Philip: Metall. Trans., 2 (1971), 2209.
• 7)   A.  Iqbal and  M. S.  Baig: Eur. J. Sci. Res., 17 (2007), 150.
• 8)   H. F.  Fischmeister,  S.  Karagöz and  H. O.  Andrén: Acta Metall., 36 (1988), 817.
• 9)   S. E.  Hopkin,  M.  Danaie,  G.  Guetard,  P.  Rivera-Diaz-del-Castillo,  P. A. J.  Bagot and  M. P.  Moody: Philos. Mag., 98 (2018), 766.
• 10)   T.  Zhou,  Y.  Xiong,  Y.  Yue,  Y.  Lu,  Y. N.  Chen,  T. T.  He,  F. Z.  Ren,  H.  Singh,  J.  Kömi,  M.  Huttula and  W.  Cao: Mater. Sci. Eng. A, 766 (2019), 138352.
• 11)   F.  Wang,  D. S.  Qian,  L.  Hua and  X. H.  Lu: Tribol. Int., 132 (2019), 253.
• 12)   X. D.  Song,  F.  Wang,  D. S.  Qian and  L.  Hua: Mater. Sci. Eng. A, 780 (2020), 139171.
• 13)   K.  Ryttberg,  M. K.  Wedel,  V.  Recina,  P.  Dahlman and  L.  Nyborg: Mater. Sci. Eng. A, 527 (2010), 2431.
• 14)   Z. X.  Li,  C. S.  Li,  J. Y.  Ren,  B. Z.  Li,  J.  Zhang and  Y. Q.  Ma: Mater. Sci. Eng. A, 674 (2016), 262.
• 15)   J.  Chakraborty,  D.  Bhattacharjee and  I.  Manna: Scr. Mater., 61 (2009), 604.
• 16)   X. H.  Lu,  D. S.  Qian,  W.  Li and  X. J.  Jin: Mater. Lett., 234 (2019), 5.
• 17)   J.  Beswick: Mater. Trans. A, 15 (1984), 299.
• 18)   H. G.  Nanesa,  J.  Boulgakoff and  M.  Jahazi: J. Manuf. Process., 22 (2016), 115.
• 19)   J. L.  Tian,  W.  Wang,  H. B.  Li,  M. B.  Shahzad,  Y. Y.  Shan,  Z. H.  Jiang and  K.  Yang: Mater. Charact., 155 (2019), 109827.
• 20)   Y.  Lian,  J. F.  Huang,  J.  Zhang,  C.  Zhao,  W.  Gao,  Z. J.  Zhang and  M. Y.  Ma: Mater. Sci. Eng. A, 712 (2018), 663.
• 21)   H. H.  Lu,  H. K.  Guo,  W. G.  Zhang,  W.  Liang and  Y. M.  Liu: J. Mater. Process. Technol., 281 (2020), 116645.
• 22)   Y. C.  Liu,  F.  Sommer and  E. J.  Mittemeijer: Acta Mater., 51 (2003), 507.
• 23)   Y. K.  Lee,  H. C.  Shin,  Y. C.  Jang,  S. H.  Kim and  C. S.  Choi: Scr. Mater., 47 (2002), 805.
• 24)   A. S. M. A.  Haseeb,  M.  Arita and  Y.  Hayashi: J. Mater. Sci., 36 (2001), 4739.
• 25)   E. J.  Mittemeijer: J. Mater. Sci., 27 (1992), 3977.
• 26)   E.  Donoso,  G.  Díaz and  J.  Criado: J. Therm. Anal. Calorim., 91 (2008), 491.
• 27)   S.  Primig and  H.  Leitner: Mater. Sci. Eng. A, 527 (2010), 4399.
• 28)   A.  Burnham and  L.  Dinh: J. Therm. Anal. Calorim., 89 (2007), 479.
• 29)   S.  Vyazovkin,  A. K.  Burnham,  J. M.  Criado,  L. A.  Pérez-Maqueda,  C.  Popescu and  N.  Sbirrazzuoli: Thermochim. Acta, 520 (2011), 1.
• 30)   B. L.  Li,  A.  Godfrey and  Q.  Liu: Scr. Mater., 50 (2004), 879.
• 31)   A. S.  Podder and  H. K. D. H.  Bhadeshia: Mater. Sci. Eng. A, 527 (2010), 2121.
• 32)   G. R.  Speich and  A.  Szirmae: Trans. Metall. Soc. AIME, 245 (1969), 1063.
• 33)   J. H.  Kang and  P. E. J.  Rivera-Díaz-del-Castillo: Comput. Mater. Sci., 67 (2013), 364.
• 34)   A.  García-Junceda,  C.  Capdevila,  F. G.  Caballero and  C.  García de Andrés: Scr. Mater., 58 (2008), 134.
• 35)   F.  Wang,  D. S.  Qian,  L.  Hua,  H. J.  Mao,  L. C.  Xie,  X. D.  Song and  Z. H.  Dong: Mater. Sci. Eng. A, 771 (2020), 138623.
• 36)   X. H.  Lu,  W.  Li,  C. L.  Wang,  H. S.  Zhao and  X. J.  Jin: Acta Metall. Sin. (Engl. Lett.), 28 (2015), 787.
• 37)   S. J.  Lee and  Y. K.  Lee: Mater. Sci. Forum., 475–479 (2005), 3169.
• 38)   X. H.  Lu,  W.  Li,  H. S.  Zhao and  X. J.  Jin: ISIJ Int., 56 (2016), 851.
• 39)   T.  Tsuchiyama,  K.  Inoue,  K.  Hyodo,  D.  Akama,  N.  Nakada,  S.  Takaki and  T.  Koyano: ISIJ Int., 59 (2019), 161.
• 40)   S.  Morito,  H.  Yoshida,  T.  Maki and  X.  Huang: Mater. Sci. Eng. A, 438–440 (2006), 237.
• 41)   F.  Wang,  D. S.  Qian and  X. H.  Lu: Acta Metall. Sin. (Engl. Lett.), 32 (2019), 107.
• 42)   T.  Karthikeyan,  M. K.  Dash,  Ravikirana,  R. Mythili, S. P. Selvi, A. Moitra and S. Saroja: J. Nucl. Mater., 494 (2017), 260.
• 43)   D. R.  Askeland and  P. P.  Fulay: Essentials of Materials Science and Engineering, Cengage Learning, Boston, (2010), 1.
• 44)   Z. H.  Jiang,  P.  Wang,  D. Z.  Li and  Y. Y.  Li: Mater. Sci. Eng. A, 742 (2019), 540.
• 45)   D.  Kalish and  M.  Cohen: Mater. Sci. Eng., 6 (1970), 156.
• 46)   D.  Turnbull and  J. C.  Fisher: J. Chem. Phys., 17 (1949), 71.
• 47)   A. T. W.  Barrow,  J. H.  Kang and  P. E. J.  Rivera-Díaz-del-Castillo: Acta Mater., 60 (2012), 2805.
• 48)   E.  Jimenez-Melero,  R.  Blondé,  M. Y.  Sherif,  V.  Honkimäki and  N. H.  van Dijk: Acta Mater., 61 (2013), 1154.
• 49)   R. D. K.  Misra,  B. R.  Kumar,  M.  Somani and  P.  Karjalainen: Scr. Mater., 59 (2008), 79.
• 50)   J.  Talonen and  H.  Hänninen: Acta Mater., 55 (2007), 6108.