2022 Volume 63 Issue 11 Pages 1533-1540
Under the conditions of deformation temperature of 900∼1200°C and strain rate of 0.01∼10 s−1, the single pass isothermal compression test of CF170 maraging stainless steel was carried out on the Gleeble-3800 thermal compression simulator, and its flow stress curve was measured. Based on the Arrhenius model and Avrami model, the flow stress model considering strain compensation and dynamic recrystallization volume fraction model were established according to the flow stress curve. The results showed that the dynamic recrystallization of CF170 steel occurs in advance compared with other steels, and the relationship between the critical strain of dynamic recrystallization and the peak strain is εc = 0.44 εp. CF170 steel is sensitive to the deformation temperature and strain rate. The temperature required for complete dynamic recrystallization is relatively low at a low strain rate. In addition, although the flow stress curve does not show the characteristics of dynamic recrystallization under the high temperature and high strain rate, partial dynamic recrystallization still occurs. In addition, the hierarchical relationship between the martensite and austenite was analyzed by Electron Back-Scattered Diffraction. The results showed that a prior austenite grain is divided into multiple martensite packets, subsequently a packet is divided into several martensite blocks.
Due to its corrosion-resistant and long-lasting, stainless steel is widely used in manufacturing and construction. The annual consumption of stainless steel has increased at a compound growth rate of 5% over the last 20 years, surpassing the growth rate of other materials.1) Martensitic stainless steel is often used to manufacture turbine blades, steam engines, and pressure vessels because of its excellent mechanical properties and corrosion resistance. In particular, ultra-low carbon maraging stainless steel has good plasticity and toughness and improves weldability and machinability.2) CF170 new Co-free maraging stainless steel is developed by Central Iron & Steel Research Institute based on maraging steel. Since there are no national and industrial standards for CF170 steel at present, its enterprise standard number is Q/GYB1121-2019, where CF refers to Cobalt free and 170 refers to 1700 MPa Grade. In recent years, researchers have done a lot of research on the microstructure and mechanical properties of maraging stainless steel. The results show that alloy elements effectively promote the dispersion precipitation of the strengthening phase,3) and heat treatment significantly affects the mechanical properties and microstructure.4) However, there are few reports about CF170 maraging stainless steel, and there are many bottlenecks in the process of large section forging, such as inhomogeneous microstructure after forging. Therefore, it is necessary to study the thermal deformation behavior and dynamic recrystallization behavior of CF170 steel. As an important method to study the thermal deformation behavior of materials, the constitutive model mainly includes empirical or semi-empirical model5) and model based on physical mechanisms.6) The Arrhenius model belonging to the former is widely used in many metal materials such as carbon steel,7) mold steel,8) and high temperature alloy steel.9)
In addition, dynamic recrystallization (DRX), as an important softening mechanism during hot deformation, directly affects the final microstructure. Therefore, controlling dynamic recrystallization in deformation is often regarded as an effective means to obtain refined microstructure. Although the occurrence of peak stress in the stress-strain curve is regarded as the main characteristic of DRX, DRX has already occurred before the peak strain. At present, there are mainly two ways to determine the critical strain for DRX: I) mathematical analysis for the plot of work hardening rate versus true stress10,11) and II) metallography investigation.12) The critical strain for DRX mainly depends on the chemical composition of the material, the grain size prior to deformation, and the deformation schedule.11) CF170 steel, a kind of steel with high alloy content, has excellent high-temperature deformation resistance, and its hot deformation behavior differs from that of traditional carbon steel and low alloy steel. Therefore, further study of the dynamic recrystallization behavior is of great significance for large-scale forging homogenization control of this kind of steel.
The as-cast CF170 maraging stainless steel was taken as the research object. Through the hot compression experiment in this work, the constitutive equation with strain compensation was constructed by using the Arrhenius model. At the same time, the DRX behavior was studied, and the DRX volume fraction model was constructed by using the Avrami model, which provides a certain theoretical basis for the homogenization control of large section forging.
The CF170 maraging stainless steel used in this investigation is directly sampled from 4 t ingot by vacuum self-consuming arc melting. Its chemical composition (mass%) is as follows: ≤0.01 C, 10.5∼12.5 Cr, 10.7∼11.3 Ni, 0.75∼1.25 Mo, 1.2∼1.4 Ti and balance Fe. The hot compression samples were machined into a cylinder with a diameter of 8 mm and a height of 15 mm, and then an isothermal compression experiment was carried out on Gleeble-3800 hot compression simulator in the temperature range of 900∼1200°C, and the strain rate range of 0.01∼10 s−1. The samples were first heated to 1250°C for 600 s at the speed of 10°C/s to eliminate segregation and obtain uniform austenite. Then cooled to the preset deformation temperature for 30 s at the speed of 10°C/s to obtain uniform temperature distribution before thermal compression. In order to retain the microstructure at high temperature, all specimens were compressed to a true strain ε = 0.6, then water quenched immediately. The experimental procedure is shown in Fig. 1(a). It is worth noting that due to the friction between the ends of the sample and the anvils, the compressed sample has bulged shape. At this time, the end of the flow stress curve is warped upward. Therefore, tantalum foil and high-temperature lubricate were used between the anvils and the sample to reduce the friction in this work. The compressed sample has a slight bulge, as shown in Fig. 1(b). In order to further reduce the experimental error, the experimental data with a true strain of less than 0.55 are selected when analyzing the flow stress curve. The quenched samples were sectioned along the axial direction, polished, and then electrolyzed at a voltage of 1.0∼1.2 V in nitric acid to observe the metallographic structure. EBSD (Electron Back-Scattered Diffraction) was used to determine the hierarchical relationship between martensite and austenite. Sample preparation for EBSD scans consisted of mechanical polishing and electropolishing at 14 V for 18 s in a perchloric acid alcohol solution (12.5% HClO4 + 87.5% C2H6O, volume%).
Thermal compression test: (a) schematic, (b) compressed sample.
Figure 2 shows the flow stress curves of CF170 under different deformation conditions. It can be seen that the peak value of stress decreases with the increase of deformation temperature at the same strain rate. The main reason is that the cross slip and dislocation climb become accessible with the increase in temperature, resulting in decreased stress. In addition, the flow stress curves also show two characteristics: i) dynamic recovery (DRV) and ii) DRX. As shown in Fig. 2(a), the stress curves at a deformation temperature range of 900°C to 1150°C with a strain rate of 0.1 s−1 show typical DRV character. That is, the stress first increases with the increase of strain, and then reaches a plateau. When the temperature is 1200°C at a strain rate of 0.1 s−1, the stress curve has obvious DRX character with a single peak. That is, after the stress reaches its peak, it gradually decreases with the increase of strain until it finally reaches a plateau. This is because when the deformation energy storage is the same, the early recovery releases most of the stored energy, and there is not enough driving force to make the recrystallization nucleation grow. Therefore, DRX can only occur at a higher temperature. As shown in Fig. 2(b), the stress curves at a strain rate range of 1 s−1 to 10 s−1 with a deformation temperature of 1200°C show DRV character, because there is not enough time for the nucleation and growth of DRX grains and dislocation annihilation at high strain rates.1,11)
Flow curves under different deformation conditions: (a) $\dot{\varepsilon } = 0.1$ s−1, (b) T = 1200°C.
The relationship between flow stress, strain, strain rate, and deformation temperature of material deformation can be expressed by a constitutive equation, which is a dynamic response to the deformation process of materials.13) It can be expressed by eq. (1):
| \begin{equation} Z = \dot{\varepsilon}\exp \left(\frac{\text{Q}}{\text{RT}}\right) = f(\sigma) \end{equation} | (1) |
In addition to eq. (1), the Z parameter can also be expressed by eq. (2)∼eq. (4):7)
| \begin{equation} \text{Z} = \text{f}(\sigma) = \text{A$_{1}$}\sigma^{\text{n}'} \end{equation} | (2) |
| \begin{equation} \text{Z} = \text{f}(\sigma) = \text{A$_{2}$}\exp(\beta\sigma) \end{equation} | (3) |
| \begin{equation} \text{Z} = \text{f}(\sigma) = \text{A}[\sinh(\alpha\sigma)]^{\text{n}} \end{equation} | (4) |
Substituting eq. (2)∼eq. (4) into eq. (1) and taking the natural logarithms for both sides:
| \begin{equation} \ln\dot{\varepsilon} = \text{n}' \ln\sigma + \ln\text{A$_{1}$} - \text{Q}/(\text{RT}) \end{equation} | (5) |
| \begin{equation} \ln\dot{\varepsilon} = \beta\sigma + \ln\text{A$_{2}$} - \text{Q}/(\text{RT}) \end{equation} | (6) |
| \begin{equation} \ln\dot{\varepsilon} = \text{n}\ln[\sinh(\alpha\sigma)] + \ln\text{A} - \text{Q}/(\text{RT}) \end{equation} | (7) |
Linear relationships between (a) ln $\dot{\varepsilon }$ and ln σ, (b) ln $\dot{\varepsilon }$ and σ, (c) ln $\dot{\varepsilon }$ and ln[sinh(ασ)], (d) ln[sinh(ασ)] and 1/T.
Assuming that the $\dot{\varepsilon }$ is constant, taking the partial derivative of eq. (7) for 1/T, eq. (8) can be obtained:
| \begin{equation} \text{Q} = \mathrm{nR}[\partial \ln[\sinh(\alpha\sigma)]/\partial(1/\text{T})] \end{equation} | (8) |
Based on the results above, the constitutive equation of CF170 maraging stainless steel can be expressed as:
| \begin{align} \text{Z} & = \dot{\varepsilon} \exp\left(\frac{\text{Q}}{\text{RT}}\right) = \text{A}[\sinh(\alpha\sigma)]^{\text{n}}\\ & = \dot{\varepsilon} \exp\left(\frac{411.784 \times 10^{3}}{\text{RT}}\right)\\ & = 1.070\times 10^{15}\times [\sinh(7.529 \times 10^{-3}\,\sigma)]^{5.340} \end{align} | (9) |
| \begin{align*} \alpha& = 0.025 - 0.238\,\varepsilon + 1.300\,\varepsilon^{2} - 3.501\,\varepsilon^{3} \\ &\quad + 4.640\,\varepsilon^{4} - 2.414\,\varepsilon^{5} \end{align*} |
| \begin{align} \text{n}& = 4.390 + 36.606\,\varepsilon - 246.582\,\varepsilon^{2} + 717.343\,\varepsilon^{3} \\ & \quad- 993.909\,\varepsilon^{4} + 535.645\,\varepsilon^{5} \end{align} | (10) |
| \begin{align*} \text{Q} &= 459.426 - 405.194\,\varepsilon + 2187.147\,\varepsilon^{2} \\ & \quad - 6653.161\,\varepsilon^{3} + 9771.618\,\varepsilon^{4} - 5466.815\,\varepsilon^{5} \end{align*} |
| \begin{align*} \ln\text{A} &= 24.915 + 162.685\,\varepsilon - 918.880\,\varepsilon^{2} + 2454.649\,\varepsilon^{3}\\ & \quad - 3203.446\,\varepsilon^{4} + 1642.725\,\varepsilon^{5} \end{align*} |
Linear relationships between (a) ln[sinh(ασ)] and ln Z, (b) ε and α, (c) ε and n, (d) ε and Q, (e) ε and ln A, (f) verification of constitutive model.
Figure 4(f) exhibits the comparison between the experimental value and the predicted value of stress under different deformation conditions. It can be shown that the constitutive model taken into account strain compensation is in good agreement with the experimental value.
3.3 Dynamic recrystallization kineticsWith the intensification of deformation, dislocations will continue to increase and accumulate. When the deformation reaches the critical value, dynamic recrystallization nucleus will form and grow near large angle grain boundaries such as grain boundaries, twin boundaries, and deformation zones. In the plot of work hardening rate θ versus the true stress σ as shown in Fig. 5, the material mainly goes through three stages in the process of high-temperature deformation:17) I) In the dynamic recovery hardening stage, the work hardening rate decreases rapidly with the increase of strain. At this stage, on the one hand, the dislocation movement is hindered by the dislocation cell wall and then accumulates, which makes the dislocation density and stress increase continuously. That is, work hardening occurs. On the other hand, dislocations counteract each other so that the work hardening rate decreases. That is, dynamic recovery occurs. II) In the large strain hardening stage, the decreasing trend of the work hardening rate slows down. At this stage, with the increase of dislocation density, the dislocation cell wall continuously absorbs dislocations to form subgrain boundaries, and the cellular structure becomes a substructure, which slows down the decline of the work hardening rate. III) The inflection point appears in the work hardening rate curve in the dynamic recrystallization softening stage. With the increase of strain and the accumulation of dislocation density, subsequently dynamic recrystallization occurs. The inflection point of the work hardening rate curve, i.e. the junction point of stages II and III, is the critical strain point for dynamic recrystallization. Figure 6 shows that the ratio of dynamic recrystallization critical strain εc to peak strain εp is ∼0.44, showing a good linear correlation of ∼0.97. The Ref. 18) indicates that this value of a (a = εc/εp) is between 0.6 and 0.85, and the value is different for different steels. In this work, the value of a is less than 0.6, which may result from the lattice distortion caused by adding a large number of alloying elements. Lattice distortion will decrease the stacking fault energy and make it difficult to cross slip and climb the dislocation, reducing the recovered energy ratio to the total energy, ultimately leading to the early occurrence of recrystallization.
The plot of work hardening rate θ versus true stress σ for 1150°C-0.01 s−1.
The linear relationships between critical strain εc and peak strain εp.
The dynamic recrystallization kinetics is usually expressed by Avrami model. To simplify the model, eq. (11) is used to represent the dynamic recrystallization kinetic under a constant strain rate:8,19,20)
| \begin{equation} \text{X}_{\text{DRX}} = 1 - \exp\left[-\text{k}_{\text{d}}\left(\frac{\varepsilon - \varepsilon_{\text{c}}}{\varepsilon_{\text{p}}}\right)^{\text{n}_{\text{d}}}\right] \end{equation} | (11) |
| \begin{equation} \ln[-{\ln}(1-\text{X}_{\text{DRX}})] = \ln\text{k}_{\text{d}} + \text{n}_{\text{d}} \ln\left(\frac{\varepsilon - \varepsilon_{\text{c}}}{\varepsilon_{\text{p}}}\right) \end{equation} | (12) |
It can be seen from eq. (12) that XDRX needs to be known to determine the value of the constant. Reference 21) used peak stress σp and steady state stress σss to denote XDRX. However, as mentioned above, DRX has occurred before the peak strain. Therefore, in order to accurately calculate XDRX, the flow stress curve with DRV feature was restructured in this work, and then XDRX can be expressed by eq. (13):22)
| \begin{equation} \text{X}_{\text{DRX}} = \frac{\sigma_{\text{drvs}}^{2} - \sigma_{\text{drxs}}^{2}}{\sigma_{\text{sat}}^{2} - \sigma_{\text{ss}}^{2}} \end{equation} | (13) |
| \begin{equation} \theta = \frac{\text{d}\sigma}{\text{d}\varepsilon} = \text{m}_{0}\sigma_{\text{drvs}} + \text{C} \end{equation} | (14) |
| \begin{equation} \text{m}_{0} = \theta_{\text{c}}/(\sigma_{\text{c}} - \sigma_{\text{sat}}) \end{equation} | (15) |
| \begin{equation} \text{C} = -\theta_{\text{c}} \sigma_{\text{sat}}/(\sigma_{\text{c}} - \sigma_{\text{sat}}) \end{equation} | (16) |
| \begin{equation} \sigma_{\text{drvs}} = \cfrac{\sigma_{\text{c}} - \sigma_{\text{sat}}}{\exp\biggl[\cfrac{(\varepsilon_{\text{drvs}} - \varepsilon_{\text{c}})\theta_{\text{c}}}{\sigma_{\text{sat}} - \sigma_{\text{c}}}\biggr]} + \sigma_{\text{sat}} \end{equation} | (17) |
The DRX and DRV curves for 1150°C-0.01 s−1.
Figure 8 shows that the linear relationships between ln[−ln(1 − XDRX)] and ln[(ε − εc)/εp] under different deformation conditions were obtained according to eq. (12). After linear fitting, kd and nd can be obtained, and the average values are 0.342 and 3.028, respectively. Substituting the results into eq. (11) to obtain the XDRX under different deformation conditions, as shown in Fig. 9. It can be seen that the XDRX increases with the increase of temperature. When the strain rate is 0.01 s−1 at a strain of 0.6 and a deformation temperature range of 1150°C∼1200°C, completely dynamic recrystallization occurs, while the temperature is reduced to 1000°C, the maximum of XDRX is 0.89, which indicates the occurrence of the partial dynamic recrystallization. This is because the DRX driving force is reduced as the temperature decreases, resulting in incomplete DRX.
The linear fitting of ln[−ln(1 − XDRX)] versus ln[(ε − εc)/εp].
The plot of XDRX versus ε in different deformation conditions.
In order to further study the effect of temperature and strain rate on the DRX during hot deformation, Fig. 10 shows the microstructure under different deformation conditions. The typical incomplete dynamic recrystallization microstructure is shown in Fig. 10(a). When the temperature is 1000°C with a strain rate of 0.01 s−1, scattered deformed grains are elongated perpendicular to the compression direction. From the enlarged local figure, it can be seen that necklaced refined grains distribute at the grain boundaries of the deformed microstructure. It is worth noting that there are some small recrystallization microstructures between the two large deformed microstructures, indicating that DRX occurs at the large angle grain boundaries at first, and then continues to advance towards the inside of the deformed microstructure until the deformed microstructure is entirely exhausted. In Fig. 10(b)∼Fig. 10(d), the deformed microstructure is almost completely replaced by fine grains, indicating that the material has undergone complete dynamic recrystallization, which is consistent with the calculated value of XDRX in Fig. 9. However, the microstructure in Fig. 10(b) is finer, indicating the recrystallized grains will continue to grow when the temperature exceeds 1100°C at a strain rate of 0.01 s−1. As shown in Fig. 10(d), the grain boundaries become tortuous under the high temperature and low strain rate, indicating that the recrystallized grains are continuously merging and growing. As shown in Fig. 10(e) and Fig. 10(f), when the strain rate increases to 0.1 s−1, a small number of coarse grains without recrystallization exist in the microstructure at a deformation temperature of 1150°C, and the grain boundaries present a serrated shape. However, there are almost no coarse deformed grains at a temperature of 1200°C, indicating that with the increase in temperature, the activation energy of atoms increases, and the driving force of recrystallization increases, which are conducive to the occurrence of recrystallization. At the same time, due to the increase in strain rate, the recrystallized grains do not have enough time to grow completely, making the microstructure more refined than that in Fig. 10(d). It is worth noting that when the temperature is 1200°C, the flow stress curves with strain rates of 1 s−1 and 10 s−1 (Fig. 2(b)) do not show single peak stress, but partial dynamic recrystallization still occurs, as shown in Fig. 10(g) and Fig. 10(h). It can be shown that it is not wholly applicable for high-strength stainless steel to directly judge whether DRX has occurred by flow stress curves. Similar phenomena also exist in 17-4PH steel24) and TA15 titanium alloy.17) By judging whether there is an inflection point in the plot of the work hardening rate versus true stress, it can better determine whether DRX has occurred in the process of thermal deformation.
Microstructure in different deformation conditions: (a) 1000°C-0.01 s−1, (b) 1100°C-0.01 s−1, (c) 1150°C-0.01 s−1, (d) 1200°C-0.01 s−1, (e) 1150°C-0.1 s−1, (f) 1200°C-0.1 s−1, (g) 1200°C-1 s−1, (h) 1200°C-10 s−1.
In order to study the hierarchical relationship between the martensite and austenite, EBSD orientation mapping was shown in Fig. 11(a). By reconstructing the parent grain, the prior austenite grains are shown in Fig. 11(c), which agrees with Fig. 10(a). Within the prior austenite grain that is outlined by green dash lines, there are packets that are outlined by the yellow lines, which are divided into some of the same color parallel blocks, as shown in Fig. 11(b). It is worth noting that block boundaries have a misorientation angle around 50∼70°, which are calculated based on the ideal KS orientation relationship.25) Figure 11(d) shows a multi-level structure of martensite. A prior austenite grain is divided into multiple martensite packets, then a packet is divided into several martensite blocks. In addition, with the decrease of austenite grain, the size of martensite packets within austenite grain decrease.
Results of the EBSD scanning in deformation condition of 1000°C-0.01 s−1: (a) the EBSD misorientation mapping, (b) enlarged local boundary mapping, (c) reconstructing parent grain, (d) illustration for its hierarchical relationship.
Figure 12 shows the kernel average misorientation (KAM) mapping in different deformation conditions. With the increase of deformation temperature, the misorientation within prior austenite grains are relatively decreased, which infers that the density of geometrically necessary dislocations is reduced. On the one hand, there are a large number of dislocations within grains, resulting in dislocations difficult to climb, which makes DRX occur in advance, as mentioned above. On the other hand, the decrease in dislocation density within a grain leads to a decrease in the number of martensite packets and blocks.
The KAM mapping in different deformation conditions: (a) 1000°C-0.01 s−1, (b) 1200°C-0.01 s−1.
