Study of Phase Transformations and Interface Evolution in Carbon Steel under Temperatures and Loads Using Molecular Dynamics Simulation

Abstract

Carbon steel materials are widely used in mechanical transmission. Under different temperature and pressure service conditions, the microscopic changes of stress and strain that are difficult to detect and analyze by experimental means will lead to failure deformation, thus affecting their operational stability and life. In this study, the molecular dynamics method is used to simulate the heating–cooling phase transition process of common carbon steel materials. Austenite transformation temperatures of 980 K (0.2 wt.%) and 1095 K (0.5 wt.%) are acquired which is determined by the volume hysteresis before and after transformation, which is consistent with the results of JMatPro phase diagram analysis. The internal stress state of the material varies between compressive stress and tensile stress due to the change of phase structure, and the dislocation characteristics during the phase transition period are observed to change significantly. Then, an $\mathsf{\alpha }/\mathsf{\gamma }$ two-phase interface model is constructed to study the migration of the phase interface and the change of the phase structure by applying a continuously changing external load. At the same time, the transition pressure of $\mathsf{\alpha }\to \mathsf{ϵ}$ is obtained with a value of 37 GPa under three different initial loads showing the independence of the initial load and the historical path. Based on the molecular dynamics simulation and the phase diagram calculation of the carbon steel, the analysis method for the microstructure transformation and the stress–strain behavior of the phase interface under the external load can provide a reference for the design of microstructure and mechanical properties of alloy steel in the future.

1. Introduction

The heat treatment state of a material is the main factor that affects the external force resistance behavior of the parts, especially the atomic scale structure of the two-phase interface which has an important influence on the migration behavior of the phase interface. Using molecular dynamics (MD) simulation to study the phase transition process in the heat treatment state of a material has become a hot topic in the study of the microscopic deformation mechanism of materials. In recent years, more progress has been made in the study of phase transition of iron BCC$\to$FCC/HCP based on MD simulation. For example, Kadau et al. [1,2] conducted shock-induced simulations of single crystal iron and polycrystalline iron using MD. The HCP phase (1000) plane was observed to form by the relative movement of the BCC close-packed plane (110) along the [110] crystal direction; thus the possibility of the existence of a certain number of FCC structures was clarified. Song et al. [3] simulated the migration of Fe; they found the boundary very mobile, and that the transformation was not due to the martensitic mechanism. The transition process is affected by both the heating/cooling rate and carbon concentration. Wang et al. [4] simulated the phase transition of austenite and martensite of the Fe-C system within 0–1 at.% carbon content and found that the phase transition temperature determined by the hysteresis of system volume showed negative dependence of carbon content. The elevated rate would delay the phase transition.

The actual carbon steel materials are a variety of structures coexisting with phase interfaces between different structures [5]. Studying the evolution of the phase interface and analyzing the microstructure is also of great significance for guiding engineering practice [6]. Ou et al. [7] studied the effect of different orientation relationships of FCC and BCC phases on the propagation of the BCC/FCC interface in pure iron at 300 K using MD. The phase transition in the low potential region was found with martensitic properties, while in the high potential region it was involved with accidental atomic diffusion jumps. Wang et al. [8] used molecular dynamics simulation to study the energy and dynamics of the phase transition process by establishing a Nishiyama–Wassermann (N-W) oriented BCC/FCC Fe bicrystal interface model. The homogeneous and heterogeneous transformations were observed. The phase boundary propagation velocity reached 24 m/s at the low temperature (100 K), an order of magnitude lower than that at the high temperature (1300 K). The dependence of martensitic (FCC$\to$BCC) and austenitic (BCC$\to$FCC) transformations in Fe twins was then investigated [9]. Homogeneous nucleation in bulk materials was found at the expense of heterogeneous growth and the interface geometry also had an impact on the phase transition behavior [10].

However, most studies only focus on the microstructure of pure iron materials, and few are found to research carbon steel materials. Although Wang et al. [9] studied the heat treatment phase transition process of the Fe-C system, their research adopted the percentage of carbon atoms which were relatively small values (within the low carbon steel range) when converted into engineering practice. Therefore, the study of carbon steel materials with carbon content more in line with engineering practice is urgently needed. In addition, the two-phase interface is mostly studied based on pure iron, while in practice, the existence of carbon atoms may have a great influence [11]. At present, there are few studies to explore the migration of phase interface and the evolution of phase structure under continuous changing pressure.

Based on the above analysis, the phase structure transformation and corresponding dislocation characteristics of two common carbon steel materials with the carbon content of 0.2 wt.% and 0.5 wt.% under different heat treatment states were studied using the MD method, and the change of residual stress during the heat treatment was analyzed. Subsequently, the phase diagram simulation was carried out by JMatPro [12] and the transition temperature was obtained. Then, a martensitic–austenitic two-phase interface model of carbon steel was established, and unilateral continuously increasing pressure was applied in the form of piston impact pressure to study the migration of the interface and the microstructure evolution of the material under pressure. The work presented in this paper has certain enlightening significance for studying the heat treatment of carbon steel materials and the deformation of carbon steel materials under external loads. In addition, our work provides a reference for obtaining the transition pressure by applying unilateral continuous load. Further studies can focus on adding alloying elements based on Fe-C materials to better simulate the actual carbon steel materials to guide the engineering practice.

2. Simulation Methods

2.1. Model Setup

2.1.1. Carbon Steel Modelling

There are two types of interstitials in the body-centered cubic and close-packed crystal structures of iron: tetragonal sites (T-sites) and octahedral sites (O-sites). Unlike close-packed iron, the O site of body-centered cubic iron is smaller than the T site. However, the O site is always the most energy-preferred site for carbon in BCC and FCC iron [13,14]. In addition, it should be noted that the crystal will not be tetragonally distorted by the randomly distributed C atoms [15].

The simulation box is set to be a BCC crystal structure cube with a volume of 86.1 × 86.1 × 86.1 ų. The carbon atoms are randomly inserted into the octahedral interstitial sites. Because of the random insertion, the simulation box still maintains the shape of a cube. Periodic boundary conditions are used in all three directions. Figure 1 is a general model diagram of Fe-C alloys with 0.2 wt.% and 0.5 wt.% carbon content.

2.1.2. α/γ Interface Modelling

The crystal orientation relationships between austenite and ferrite phases in steel are Nishiyama–Wasserman (N-W) and Kurdjumov–Sachs (K-S) [16], described as:

$$\mathrm{N}–\mathrm{W}: \left(110\right) \mathrm{BCC}// \left(111\right) \mathrm{FCC} \mathrm{and} \left[001\right] \mathrm{BCC}// \left[1\overline{1}0\right]\mathrm{FCC}$$

$$\mathrm{K}–\mathrm{S}: \left(110\right) \mathrm{BCC}// \left(111\right) \mathrm{FCC} \mathrm{and} [1\overline{1}1] \mathrm{BCC}// [1\overline{1}0]\mathrm{FCC}$$

The N-W orientation relationship has an additional 5.26° deflection relative to the interphase boundary normal. Two relationships are most closely aligned in the (111) FCC/(110) BCC interface contact zone for both FCC and BCC phases. Bos et al. [17] studied several planar interfaces in Bain orientation and K-S orientation in their MD work of iron from the atomic point of view and observed no interface motion on the MD time scale when periodic boundary conditions were used for simulation. In this work, the BCC/FCC interface model is established based on the research of Song et al. [3]. The actual direction relationship is given by:

$$\mathrm{BCC}: \mathrm{x} \left[110\right], \mathrm{y} \left[\overline{1}10\right], Z \left[001\right]$$

$$\mathrm{FCC}: \mathrm{x} \left[776\right], \mathrm{y} [33\overline{7}], \mathrm{z} [\overline{1}10]$$

Figure 2 shows the side view of the analog unit observed from the z direction, where the BCC phase is on the left side of the interface and the FCC is on the right side. Note that this is only the initial unrelaxed configuration, the interface orientation is as mentioned above, and x is parallel to the interface normal. The y and z directions are set as periodic boundary conditions, and the x direction is the free boundary.

2.2. Potential Energy Functions

In MD simulations, the potential function describes the interactions between atoms, and different atoms require different types of potential functions to be set. For composites composed of multiple atom types, a common strategy is to use a hybrid potential function if the specialized potential function is not available. Meyer–Entel interaction potential [18] as an embedded-atom method (EAM) potential [19] is adopted as the interaction between Fe atoms. The interaction potential between atoms is the physical basis of classical MD simulation. The form of the potential function and adaptation parameters are generally given by empirical or semi-empirical methods. EAM many-body potential model can well describe the interaction of atoms in metals, and it can be expressed as:

$$E_i=F_{\mathsf{\alpha }}\left({\sum }_{j\ne i}{\mathsf{\rho }}_{\mathsf{\beta }}\left(r_{ij}\right)\right)+\frac{1}{2}{\sum }_{j\ne i}{\mathsf{\varphi }}_{\mathsf{\alpha }\mathsf{\beta }}\left(r_{ij}\right),$$

(1)

where $E_i$ is the total energy and F_α is the embedded energy, which is a function of the atomic electron density $\mathsf{\rho }$. ${\mathsf{\phi }}_{\mathsf{\alpha }\mathsf{\beta }}$ is a pair of potential interactions, and $\mathsf{\alpha }$ and $\mathsf{\beta }$ are the element types of atoms. The many-body property of the EAM potential is the result embedded in the energy term. Both summations in the formula are on all neighbors of atoms within the cut-off distance.

The modified embedded atom method (MEAM) was first proposed by Baskes [20]. In MEAM, the total energy of the system is written as:

$$E={\sum }_i\left[F_i\left({\overline{\mathsf{\rho }}}_i\right)+\frac{1}{2}{\sum }_{j\ne i}{\mathsf{\varphi }}_{ij}\left(R_{ij}\right)S_{ij}^{\mathsf{\varphi }}\right],$$

(2)

where ${\overline{\mathsf{\rho }}}_i$ is the background electron density at position i, and F_i, is the energy ${\overline{\mathsf{\rho }}}_i$ embedded in the background electron density of atom i. ${\mathsf{\varphi }}_{ij}$ and $R_{ij}$ are the interaction functions and distances between atoms, respectively. $S_{ij}^{\mathsf{\varphi }}$ is the many-body screening function. The outer sum is all atoms in the system. The electron density contribution ${\overline{\mathsf{\rho }}}_i$ and the summation in square brackets include all atoms close to i, within the cutoff distance. The electron density ${\overline{\mathsf{\rho }}}_i$ has both a spherically symmetric electron density term and an angular dependence term. The above function has an analytical expression of 15 tunable parameters of pure elements. MEAM has been widely used to develop interatomic potentials for various materials, including metals and alloys, and semiconductors. In this work, the Fe-C interaction adopts a modified embedded atom potential referring to the work of Liyanage et al. [21]. This potential function is well-suitable for various phase structures. The interaction between C atoms adopts the Tersoff potential [22], which is a three-body potential.

2.3. Simulation Process

In the first part, we simulated the heat treatment process of two carbon steel materials. The system was relaxed for 50 ps under the NPT ensemble at 50 K with the pressure in all three directions maintained at 0 Pa. After equilibration, the system was heated from 50 K to 1600 K with a heating rate of 1 K/ps, and then cooled from 1600 K to 50 K at the same rate. The NPT ensemble was used for the entire heating/cooling cycle. The temperature was controlled by the Nosé–Hoover thermostat and the pressure was kept at 0 Pa in all three directions.

In the second part, we simulated the interface migration and microstructure evolution of the martensite–austenite two-phase interface of carbon steel under continuously changing external loads. The time step was chosen as 1 fs. In the first stage, the constant temperature layer and the working layer are controlled at 300 K by the velocity scaling method under the NVE ensemble. After the first stage, 30 MPa, 40 MPa and 50 MPa were selected to study the evolution of ferrite/austenite under different initial loads by applying pressure on the right thermostatic layer, respectively. The fixed layer was set to prevent the atoms from escaping. The constant temperature layer could absorb the heat released by the working layer and therefore maintain the thermodynamic equilibrium. The left side was fixed during the whole process, and it seemed to push the piston to the left. In this stage, the working layer was kept under the NVT ensemble with the constant temperature layer remaining unchanged.

Both simulations were carried out using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) [23]. The atomic phase structure and dislocation structure were analyzed by the common neighbor analysis (CNA) and dislocation line analysis (DXA) provided by OVITO [24], respectively.

3. Results and Discussion

3.1. The Volume Hysteresis of Phase Transition

As pointed out in the literature [4], the occurrence of phase transition and the phase transition temperature can be determined by volume hysteresis. Figure 3 is the relationship between the volume of two Fe-C system materials with different carbon content and temperature in the heating/cooling cycle. The 0.2 wt.% system is heated, and the volume expands until 1095 K followed by a jump, which indicates that Fe-C changes from $\mathsf{\alpha }$ structure to $\mathsf{\gamma }$ structure, as shown in Figure 3a. The volume then bounces back when cooled down until 540 K, indicating that the structure is completely transformed from $\mathsf{\gamma }$ to $\mathsf{\alpha }$, with a volume difference of 2.21 × 10⁴ ų. Hysteresis also occurs in the 0.5 wt.% material shown in Figure 3b. The volume expands, jumps at 980 K and bounces back at 320 K indicating the change from structure and inverse process, with a volume difference of 2.94 × 10⁴ ų.

3.2. Atomic Structure and Dislocation Characteristics of Phase Transition

Figure 4 shows the microstructure of the system with a carbon content of 0.2 wt.% at different temperatures during the simulation. Figure 4a–f shows the heating process, corresponding to the system temperature of 300 K, 500 K, 700 K, 1000 K, 1095 K and 1200 K, respectively. Figure 4g–j is the cooling process, corresponding to the system temperature of 700 K, 600 K, 540 K, and 500 K. As shown in Figure 4, FCC accounts for 31.5% and BCC accounts for 7.1% for Fe-C alloy with 0.2 wt.% carbon during the process of heating up to 1095 K. In the previous temperature, it can be seen intuitively that BCC dominates. Therefore, it can be considered that 1095 K is the phase transition temperature. Similarly, the cooling treatment of the system, until 540 K, is BCC dominant, accounting for 66.0%. Similarly, we find that the temperature of 540 K is a critical value during which the BCC-dominated atomic configuration is almost unchanged, so the critical temperature of 540 K can be considered as the final transition temperature of $\mathsf{\gamma }\to \mathsf{\alpha }$ [25].

Figure 5 shows the simulated microstructure of 0.5 wt.% Fe-C alloy at different temperatures. As shown in Figure 5, for Fe-C alloy with 0.5 wt.% carbon content, during the heating process until 980 K, the phase structure distribution, FCC accounted for 27.9%, BCC accounted for 8.2%, and FCC is far more than BCC. It can be seen that BCC apparently dominates until 980 K. Therefore, 980 K can be considered as the phase transition temperature of $\mathsf{\alpha }\to \mathsf{\gamma }$. Similarly, during the cooling process, the BCC accounting for 71.7% dominates. In addition, 320 K is found at a critical point below which the BCC-dominated atomic configuration is almost unchanged, so 320 K can be considered as the final transition temperature. Moreover, the consistency can be found by comparing Figure 4 and Figure 5.

From the above simulation, in both the 0.2 wt.% and 0.5 wt.% materials, C content exceeds the maximum solubility of carbon in iron at room temperature. Therefore, with the increase in temperature, the carbide transition phase of iron will occur. From the simulation structure, the HCP structure exists mainly as the $\mathsf{ϵ}$ phase carbide in the BCC structure transformation process. In addition, FCC and HCP structure atoms are more stable than BCC structures due to the poor stability in high temperatures [26]. Therefore, during the continuous heating process, the microstructure of Fe undergoes a transition from BCC to FCC/HCP, and the microstructure of Fe changes oppositely during the continuous cooling process. At the same time, it can also be found that the temperature change rate is too fast, which can induce the stacking fault in the crystal growth process and promote the alternating layered distribution of FCC and HCP crystals in the heating and cooling process of Fe. This is due to the similar atomic energy, the same crystal density, and the same atomic space stacking mode of FCC and HCP crystals [27].

3.3. Phase Diagram Calculation of Carbon Steel

SW JMatPro (JMP) is a simulation software which can calculate various material properties of alloys, mainly focusing on multi-component alloys used in industrial practice [10]. In this work, considering the overall composition of the studied alloy, the phase diagram simulation of two carbon steel materials was carried out. Figure 6a,b are the phase diagrams of iron–carbon alloy materials with 0.2 wt.% and 0.5 wt.% C content calculated by JMatPro, respectively. From Figure 6a, we found that the initial structure ferrite occupies the majority. As the temperature increases, the austenite structure begins to appear and gradually increases, and the transition temperature is 1043 K. In Figure 6b, a similar phenomenon was found, and the corresponding transition temperature was 993 K. In the previous sections, we obtained the transition temperatures of two carbon steel materials at 1095 K and 980 K by MD simulation, which was basically consistent with the temperature obtained by phase diagram calculation using JMatPro in this section. The reliability of molecular dynamics simulation is illustrated.

3.4. Stress–Time Relationship under Temperature Effect

Figure 7 shows the variation of stress with time under 0.2 wt.% (Figure 7a) and 0.5 wt.% (Figure 7b) carbon content. For the 0.2 wt.% C content system, BCC structure dominates until reaching the phase transition temperature of 1095 K marked as region I in the figure. The system then maintains the FCC/HCP state marked as region II. At 540 K, it regresses to the BCC structure, corresponding to region III in the figure. Through observation, it can be found that the stress in the I region is generally negative, and the bias stress leads to a decrease in volume. In region II, due to the generation of FCC/HCP structure, there is a big difference with the initial BCC atomic structure, which in turn causes the overall stress imbalance. In the beginning, due to the presence of the iron carbide phase, the system is mainly dominated by obvious compressive stress. As the temperature changes, the structure gradually changes to BCC and the system stress begins to bias towards tensile stress with a decrease in the volume. For region III, the overall stress fluctuation is weak, and the volume is basically unchanged. For the system with 0.5 wt.% C content, BCC is at the topmost before reaching the phase transition temperature of 980 K, marked as region I in Figure 7a. The FCC/HCP structure then dominates, and the corresponding region is marked as II in the figure. At 320 K, it goes back to the BCC structure, corresponding to region III in the figure. Through observation, it can be found that the bias stress in region I leads to a decrease in volume. The stress state distribution of region II is similar to that of 0.2 wt.% C content, showing an overall deviatoric tensile stress and the volume increase. In region II, the stress fluctuation is weak, and the volume remains almost unchanged. This is consistent with the transformation of Fe-C alloy from BCC to HCP crystal, volume compression, and volume expansion from HCP to BCC.

The dislocations of the two materials at 1000 K are analyzed by OVITO. Figure 8a,b shows the distribution characteristics of dislocation lines with a carbon content of 0.2 wt.%. According to the dislocations of the HCP phase structure shown in Figure 8a, it was found that there are two kinds of dislocation lines in HCP: the lines colored red are 1/3<1$\overline{1}$00> type with a length of 587.542 Å, and the lines colored blue are <1$\overline{1}$00> type whose length is 6.02 Å. For FCC structure analysis illustrated in Figure 8b, there are three kinds of dislocation lines, which are red 1/3<100> with a length of 128.538 Å, yellow 1/6<112> 1429.24 Å, and blue 935.985 Å representing other dislocation types. Figure 8c shows the distribution characteristics of dislocation lines with a carbon content of 0.5 wt.%. For FCC structure analysis, the type of dislocation line is 1/3<100>, and the total length is 1968.94 Å.

3.5. Migration of α/γ Phase Interface under Varying Loading

After the first stage of relaxation, the atomic phase structure diagram Figure 9a and the dislocation structure diagram Figure 9b are obtained. For Figure 9a, blue, green, and red represent the BCC, FCC, and HCP structures identified by the common neighbor analysis, respectively, and the white atoms are identified as unknown structures. After the first stage, the temperature of the working layer and the constant temperature layer reached 300 K. Some atoms in the BCC region of the left half are not identified (white atoms in the image) due to the existence of C atoms. After energy minimization and temperature control, the BCC atoms in the perfect structure are misplaced, and the lattice is distorted, resulting in internal stress. At this time, the temperature is not high, and it is impossible to provide sufficient internal energy to resist this part of the stress through temperature, thus failing to be identified by the algorithm. Luu et al. [28] proposed that in actual Fe-C, dislocations can be temporarily pinned by impurities, and additional stress is required to move the reduced dislocation mobility of dislocations. The existence of unidentified white atoms in the FCC structure of the right half is also for the same reason. Figure 9b shows the different dislocation structures obtained by dislocation analysis. When the FCC structure is used as the input model to analyze the dislocation structure, we can see that there are many types of dislocations in the right FCC region, among which the 1/6<112> type is the most common, as well as a few 1/2<110> and 1/3<110> dislocations and other types of dislocations. These dislocations do not show obvious crosslinking and ringing.

Figure 10 is the atomic structure diagram of BCC starting to transform into FCC/HCP and completing the transformation under three different initial loadings. From left to right are the start time and the end time, from top to bottom corresponding to 30 MPa, 40 MPa, and 50 MPa three initial loading states, respectively. By observing the left half of Figure 10, it is found that although the starting time is different, the atoms of the BCC part of the two-phase interface begin to transform into the HCP structure when pressure reaches a certain degree. We consider this moment as the beginning of the transformation, and the corresponding force is the transformation stress. We define the end time as follows: the atomic structure has no obvious change, and the larger the loading is, the system will collapse. The right half of Figure 10 shows the atomic snapshot at the end of the transition. It can be found that the right BCC structure has been completely transformed into FCC/HCP structure. There is no obvious interface between the two phases.

Figure 11 is the dislocation structure diagram corresponding to Figure 10 when BCC begins to transform into FCC/HCP under three different initial loadings, and the transition is completed. From left to right are the start time and the end time, corresponding to 30 MPa, 40 MPa and 50 MPa three initial pressures from top to bottom, respectively. The transition of the atomic structure from BCC to FCC/HCP is also the direct cause of the significant change of the dislocation structure at the beginning of the transition relative to the end of the first stage. From the evolution of dislocation structures in the figure, the carbides formed by HCP are supersaturated solid solutions of carbon in iron with the emergence of FCC/HCP structures. Meanwhile, this carbide becomes the second phase particle entangled by dislocations. When the dislocation moving on the slip plane encounters the second phase particle, it will be blocked by the particle and bend. With the increase of the applied stress, the bending of the blocked part of the dislocation line is intensified, and the dislocation winding and plugging here are significant. As the HCP phase decreases, the dislocation lines around the particles meet on the left and right sides. The positive and negative dislocations cancel each other out, forming a dislocation loop surrounding the growing particles.

By diamond anvil and shock experiments, Gunkelmann et al. [29] found that iron undergoes a phase transition from BCC to FCC structure at about 13 GPa. By using the low-velocity impact of single crystal iron, Jensen et al. [30] found that the transition pressure was 14.26 GPa. Crowhurst et al. [31] reported the α to ε phase transition of iron is about 25 GPa at a strain rate of 1 × 10⁹ s⁻¹. At present, there are few reports on the transition pressure of carbon steel materials. The data of three different initial pressure transition moments are summarized in

Table 1. Pressure changes at the beginning and end of transition from BCC to HCP under different initial loadings.

Initial Pressure	30 MPa	40 MPa	50 MPa
Start Time (ps)	32	29	27
Final Time (ps)	42	38	35
Start Pressure (GPa)	36	38	37
Final Pressure (GPa)	228	249	258

. Even if the initial pressure is different, the final transition pressure is almost the same, with a value of about 37 GPa, indicating that the transition pressure is independent of the initial load and the historical path.

4. Conclusions

In this study, the heating/cooling processes of two common carbon steel materials with different C content were simulated by MD. Subsequently, the phase diagram was calculated by JMatPro, and transition temperatures of two carbon steel materials were obtained which agreed with the result of MD. Then, an $\mathsf{\alpha }/\mathsf{\gamma }$ two-phase interface model was established. The migration of the phase interface and the change of the phase structure were studied by applying a continuously changing external load, and the transition pressure of $\mathsf{\alpha }/\mathsf{\gamma }$ was obtained. The conclusions are as follows:

(1) The $\mathsf{\alpha }/\mathsf{\gamma }$ phase transition exists in the Fe-C alloy affected by carbon concentration, increasing with the carbon content and showing a negative correlation. The austenite transformation temperature of Fe-C alloy with 0.2 wt.% carbon content is 1095 K, and that of Fe-C alloy with 0.5 wt.% carbon content is 980 K. This also leads to different transformation volumes of two Fe-C alloys with different carbon contents. A high carbon content relates to a higher volume difference.

(2) With the increase of heating temperature, the thermal motion of the internal atoms is intensified. In the process of heating and cooling, due to the supersaturated solid solution of carbon, the HCP structure transformation exists in the phase structure. And the FCC and HCP crystals are alternately layered. The change of phase structure is accompanied by the change of volume, which in turn causes the tensile/compressive stress fluctuation in the internal stress state of the material and the significant change of dislocation characteristics in the phase transition.

(3) The $\mathsf{\alpha }/\mathsf{\gamma }$ two-phase interface model is established to study the phase interface migration and the desired structure change of the two phases under continuously increasing pressure. It is found that even if the initial loading is different, the stress which eventually induces the phase structure transition is almost the same, about 37 GPa. And the dislocation proliferation, offset and growth at the phase interface during the phase structure transition process are directly related to the HCP structure during the phase structure evolution process.