3.1. Microstructure Evolution
Figure 1 shows the microstructure of the original rolled pure molybdenum sheets with four different deformations. According to
Figure 1a, the grain of the pure molybdenum sheet with a deformation of 70% extended along the rolling direction, which was larger than that of the sample in
Figure 1b, with a deformation of 80%. With the increasing degree of deformation, the morphology of the grains deviated from the approximately spherical sintered state, and the grains of the deformed sheet gradually extended along the rolling direction. When 95% of severe plastic deformation was reached, the grains were deformed to a greater extent, and the morphology showed a fibrous “pancake” shape, as shown in
Figure 1d. When analyzing the grains’ aspect ratio in
Figure 1, it was concluded that the deformation of the pure molybdenum sheet increased from 70% to 95%, and the corresponding grain aspect ratio increased from 2.4 to 10.3.
After heat treatment at 900 °C for 1 h (
Figure 2a), since the temperature was lower than the theoretical recrystallization temperature, no fine recrystallized grains were observed in the metallographic microstructure diagram, and the microstructure morphology of the rolled state was maintained. The annealing temperature was raised to 1000 °C (
Figure 2b), reaching a higher temperature than the starting temperature of theoretical recrystallization of pure molybdenum. The narrow structure of the deformed molybdenum sheet appeared mixed with extremely fine recrystallized grains; as the annealing temperature continued to increase, the recrystallized grains continued to grow. Gradually, the grains in the structure became uniformly equiaxed. As shown in
Figure 2d, the grain size in most of the structure of the pure molybdenum sheet with deformation of 90% was 19.2 ± 8.0 μm when recrystallization was completed.
In
Figure 3a, a pure molybdenum sheet with 95% deformation was heat-treated at 900 °C. New recrystallized grains, which were small and roughly circular, preferentially appeared between two long, narrow, and fibrous grains or were concentrated at the tip of the grains and around broken grains (as indicated by the arrows in
Figure 3a). The temperature was increased at intervals of 50 °C to continue the isochronous heat treatment. The number of recrystallized nuclei increased. The gradually recrystallized grains can be observed in
Figure 3b. During the process, the “pancake”-like rolling structure gradually faded and was replaced by a uniform structure with small equiaxed grains. The grain size of a pure molybdenum sheet with a deformation of 95% was about 12.68 μm when recrystallization was almost completed (
Figure 3f).
Compared with
Figure 2 and
Figure 3, it was found that the higher the heat treatment temperature, the larger the proportion of recrystallized grains in the structure. The average mobility
at the grain boundaries is proportional to
according to Arrhenius formula [
14] (1):
Here, m
0 is related to the diffusion constant,
Qm is the grain boundary migration activation energy, R is the perfect gas constant, 8.314 J/(mol
K), and T is the heat treatment temperature. When the amount of deformation of the pure molybdenum sheet is constant, the atomic diffusion energy
Qm is a fixed value. The higher the temperature, the more the energy provided for atomic motion. The faster the mobility of the grain boundaries, the faster the macroscopic growth of the recrystallized grains, and the larger their proportion in the structure.In addition, the pure molybdenum sheet with a larger deformation had smaller grains after recrystallization for a period of time, which was certainly related to the grain size of the original rolled sheet. In the 900 °C annealing below the recrystallization temperature, recrystallized grains first appeared in the pure molybdenum sheet with a large amount of rolling (
Figure 2d). Comparing
Figure 2a,d, the degree of recrystallization was higher in pure molybdenum sheets with a large deformation.
3.2. Texture Evolution
The formation mechanism of rolling texture is mainly related to the stress analysis of Dillamore and Roberts [
15]. It is also useful to study the orientation of rolling texture by means of local orientation, that is, the free rotation of a certain angle around a fixed axis. A {110}//RD-TD texture, {110}<110> formed in pure molybdenum sheets with different deformation amounts, together with weak <100>{211}, <001>{211}, and {211}<110> texture [
16].
For the study of pure molybdenum sheet rolling texture evolution behavior during the process of annealing in this paper, the EBSD method was used to characterize the large plastic deformation after annealing. Using the Bunge method [
17], the Euler coordinate system was established, involving the Euler angle (φ
1, Φ, φ
2) texture orientation. The ODF diagrams were determined by the Euler angle (φ
1, Φ, φ
2). In the ODF diagram, the Miller index {ND}<RD> = {hkl}<uvw> can be known from the Euler angles (φ
1, Φ, φ
2) of the texture. The analysis of the typical texture type of a cubic-structure metal in this paper used a section of φ
2 = 45°, and the standard texture ODF diagram of φ
2 = 45° is shown in
Figure 4. The texture of a sample of 95%-deformed pure molybdenum sheet after recrystallization annealing at 800–1300 °C for 1 h was studied. The orientation density of the recrystallized texture components with the annealing system is shown in
Figure 5.
Combining the Euler angle and the Miller index in
Figure 4, it can be seen that the pure molybdenum sheet with a deformation amount of 95% in
Figure 5a mainly contained a strong Brass texture, a weak Copper texture, and a weak Goss texture after annealing at 800 °C for 1 h. During the rolling process, due to the sliding movement of dislocations, the orientation change caused by the dislocations during deformation was such that the sliding surface tended to be parallel to the rolling direction. The body-centered cubic structure metal would cause stable Goss texture formation due to shear forces. During the rolling process, the {110}<001> texture appeared first, parallel to the rolling direction {110}. Under the action of shear stress, two stable plastic deformation textures, Goss and Goss Twin, formed [
18]. In this paper, the rolling texture components of the pure molybdenum sheet without recrystallization and the deformed pure molybdenum sheet were basically consistent. Ideally, these alignment lines always correspond to high-intensity regions of the orientation distribution, but the maximum region for any given ODF map may deviate from the ideal position.
In
Figure 5b, the Goss texture, Brass texture, and Copper texture are present in the microstructure of the pure molybdenum sheet with a deformation of 95% at 900 °C. In
Figure 5c, the structure of the pure molybdenum sheet after annealing at 1000 °C includes Goss texture and {001}<001> oriented texture. The annealing temperature continued to increase to 1100 °C, and a strong Goss texture remained in the structure. There were only two component textures in the annealed structure at 1300 °C, which were Goss texture and cubic texture. It is known that the type of texture in a sheet changes as the heat treatment temperature is increased. The Goss texture always exists, and the orientation of other weak textures changes. In our sample, the Copper texture changed to {001}<001> as the heat treatment temperature increased, and then turned to cubic texture {001}<0
0>. The {001}<0
0> texture was the main texture component after recrystallization was completed. The texture composition in
Figure 5a–f shows a change in intensity, with the Goss texture decreasing from 21.99 to 2.67, and the cube texture appearing and increasing in intensity from 0 to 2.77, with no new texture types appearing during recrystallization. This indicated that during the recrystallization process, the grains with {001}<0
0> texture in the deformed structure might undergo “directional nucleation”. The mechanism of texture recrystallization is considered to be related to a dislocation movement in the microstructure. Increasing the heat treatment temperature can provide energy for grain boundary movement during recrystallization, which can affect the change in orientation of some weaving structures. The annealing time was extended by an hour, and the 95%-deformed pure molybdenum sheet was heat-treated for 2 h at 900–1300 °C, as shown in
Figure 6.
When extending the annealing time, as shown in
Figure 6a–d, the intensity of the Goss texture, the major texture component in the pure molybdenum sheet, decreased from 19.69 to 4.36. After annealing at 1000 °C for 2 h, the Copper texture {112}<111> had a tendency to transition to the plane (113). After annealing at 1100 °C for 2 h, the Copper texture was transformed into the cube texture, and the path passed from {001}<110> to {110}<100>. As shown in
Figure 6c, the reason for the weakening of the Goss texture strength was that the orientation of the Goss texture was converted to γ-fiber, which transitioned from {110}<100> to {111}<112>. In order to examined the texture transformation, the texture components in
Figure 6 were statistically analyzed (
Figure 7).
In
Figure 5c and
Figure 6b, the recrystallization texture at 1000 °C is slightly different. It is possible that the area measured by EBSD was under a relatively concentrated stress, in relation to the orientation distribution of stress caused by anisotropic density in the system. Previous researchers [
19] explained that when the direction of the minimum elastic modulus of recrystallized grains coincides with the direction of the largest absolute value of stress in a deformed matrix, the released strain is the largest. At this time, the driving force provided to the recrystallization is also the largest. The crystalline grains grow rapidly, forming a recrystallized texture that is associated with a specific orientation of the deformed matrix.
Figure 7 shows the texture distribution after heat treatment at 900–1200 °C for 2 h.
Table 3 shows the strength of each texture type. The pure molybdenum sheet with a large plastic deformation still had a deformation texture after heat treatment. It is highly possible that the recrystallized texture of the pure molybdenum sheet had the same orientation as its rolling texture. The strength of the Goss weave weakened and tended to shift in the direction of γ-fiber, and the Copper weave deflected in the direction of the cubic weave during heat treatment-induced recrystallization of molybdenum plates with a large plastic deformation.
This phenomenon is inextricably linked to the process of recrystallization nucleation and grain growth, which weakens the deformation fabric anisotropy. The evolution of the recrystallization texture is complex and affected by many factors; therefore, it is difficult to study the formation and evolution mechanisms of the recrystallization texture. Therefore, the formation of the recrystallization texture can be discussed from the perspective of recrystallization nucleation and orientation change during grain growth.
3.3. Change of the Grain Boundary
Figure 8 shows a characteristic distribution diagram of EBSD grain boundaries after a 95%-deformed molybdenum sheet was heat-treated at 800–1300 °C for 1 h. Dislocations proliferated and became entangled during rolling deformation, and their distribution was not uniform. The local low-angle grain boundaries (LAGB) were dense, as shown in
Figure 8a. The LAGB/subgrain boundaries are represented by green lines. The grain boundaries were still tightly distributed between the grains. When annealing was performed at a lower temperature (900 °C), a recovery process of the molybdenum sheet structure occurred, and the dislocation density inside the metal decreased. In conjunction with
Figure 1c, when recovery occurred and large-area recrystallization did not occur, the microstructure of the sample was mainly composed of broken grains interspersed with extended grains after rolling, and most of the pancake-like high-angle grain boundaries (HAGB) (red lines) were still maintained; the migration of dislocations within the structure led to the consolidation and slippage of the subgrains, which made the substructure polygonal, forming grain boundaries, and the number of green LAGB was larger. This observation is similar to those reported in other studies [
20].
As the annealing temperature increased (
Figure 8d), with the completion of the recovery recrystallization process, the proportion of HAGB in the microstructure increased significantly. With the increase of the temperature, along with the increasing degree of recrystallization, until complete recrystallization (1200 °C) occurred, the LAGB were gradually transformed into HAGB during the recrystallization process, and high-angle grain boundary almost occupied all grain boundaries. It is worth noting that even if the corresponding boundary had high energy and mobility, its contribution would not disappear. During grain growth, some of these boundaries disappear, but others form because they are needed to maintain orthorhombic sample symmetry [
21].
The characteristic distribution of the LAGB (misorientation angle (θ) between 2° and 15°) and HAGB (misorientation angle (θ) higher than 15°) boundaries was found to be influenced by the rolling and the annealing processes. To avoid artifacts in the local misorientation data, the misorientation angles lower than 2° were removed, which might have cause experimental fluctuations between adjacent pixels within a single grain.
As shown in
Figure 8a, in the grain boundary distribution diagram of the 95%-deformed pure molybdenum sheet treated at 800–1300 °C for 1 h, fine HAGB existed between adjacent elongated rolled fibrous tissues and surrounded the rolled broken grain. If these substructures aggregated, the LAGB did not have high mobility and did not easily grow into recrystallization nuclei due to the small difference in orientation between the substructure and the deformed substrate. In
Figure 8, it can be observed that recrystallization occurred preferentially in the vicinity of certain regions with the same characteristics, i.e., these regions were substructures and deformed grains with very different orientations and high mobility and could be rapidly transformed into recrystallized nuclei. The evolution of the dislocations can be seen in the TEM results.
Figure 9 shows the direct observation of the dislocation substructures via TEM. After heat treatment at 900 °C, there were flat grain boundaries in the microstructure of the pure molybdenum sheets (
Figure 9), with differences in the number of dislocations within adjacent grains. By combining this figure with
Figure 8b, it is possible to observe a large number of entangled dislocations within the grains shown in
Figure 9a,b. In
Figure 9c, the entangled dislocations inside the grains gradually disappeared after isochronous heat treatment at 1000 °C, leaving only the dislocation outcrops.
However, the recrystallization process of the actual pure molybdenum sheet is very complicated, many factors affect the recrystallization, and various reasons for recrystallization nucleation cannot be peeled out by a microscopic analysis. It can also be seen in
Figure 8 and
Figure 9 that the distribution of the recrystallized grains at this stage was not uniform. Therefore, the directional nucleation caused by the deformation and storage energy due to the texture orientation is also another way of recrystallization nucleation in the pure molybdenum sheet.
In order to better visualize the overall changes of the LAGB and HAGB, the angular grain boundaries in
Figure 8 were counted, as shown in
Figure 9. After annealing at 1100 °C, the HAGB proportion reached 82.23%. The gradual disappearance of the LAGB and the uniform distribution of the HAGB indicated that the recrystallization process in the sheet was gradually completed.
Figure 10 is a statistical picture of the distribution of recrystallized grains, deformed grains, and subgrains after isochronous heat treatment of the 95%-deformed pure molybdenum sheet at 800–1300 °C. In red are the rolled grain, in yellow the sub-grain, and in blue the recrystallized grain.
Figure 11a shows the grain distribution of different states after heat treatment at 800 °C for 1 h. At this time, the structure was still in the rolled state, and recrystallized small grains appeared. Aggregated agglomerates were distributed in the small rolled grains, and the grain boundary density was high. Combined with
Figure 8, the recrystallization nucleation mechanism in the 95%-deformed pure molybdenum sheet was subcrystalline nucleation. In the recovery stage of subcrystalline nucleation, the dislocation movement on the adjacent subcrystalline boundary gradually transferred to other surrounding subgrain boundaries, resulting in the disappearance and merge of adjacent subgrain boundaries. Due to the increase in size and dislocation density of the merged sub-crystals, the phase difference between adjacent sub-grain boundaries increased, gradually turning them into HAGB, which have greater mobility than the LAGB. Dislocations during migration formed distorted recrystallized nuclei.
When the heat treatment temperature was increased to 900 °C, the number of nucleated recrystallized grains shown in
Figure 11b and of small previously recrystallized grains gradually increased, and the recrystallized grains were distributed substantially parallel to the rolling direction in the entire structure. Both the recrystallized nuclei and the recrystallized nuclei grown into grains in this stage consumed the energy stored in the rolled grains. As the heat treatment temperature was further increased, the recrystallized grains in
Figure 11d–f replaced the rolled grains and became the main grains in the structure.
Statistical analysis of the recrystallization volume fraction at each grain ratio in
Figure 11 (
Figure 12) showed that after heat treatment at 1100 °C, the recrystallization volume fraction of the 95%-deformed pure molybdenum sheet was as high as 97.96%, indicating an almost complete recrystallization. Compared with the heat treatment at 1000 °C, the recrystallization volume fraction increased by 33.05%.
Recrystallization is a thermally activated process. From a deformed unsteady state, through the typical nucleation and grain growth process of recrystallization, the distortion energy is released to a steady state. Discussing the recrystallization behavior and the kinetics of recrystallized grain growth is helpful to analyze the recrystallization mechanism of the annealing process, which can reveal its thermodynamic features, important for studying the growth of grains and the migration of grain boundaries. Therefore, by analyzing the recrystallization volume fraction to understand the evolution of these parameters, it is possible to obtain the nucleation and growth mode of the recrystallization process.
Assuming that nucleation is uniform, nuclei are spherical, and nucleation rate
and grain growth rate
do not change with time, the recrystallization volume fraction X can be expressed by the JMAK equation [
22] after t time at a constant temperature:
After simplifying Equation (2)
Here, , and Formula (3) is often called the JMAK equation; n is the Avrami index. It is assumed that nucleated grains grow in three dimensions. According to Formula (3), we obtain that n = 4.
For Equation (3), considering the logarithm twice on both sides, we obtain:
The analysis of heat-treated metallographs of pure molybdenum sheets was performed using the grid method to determine their recrystallization volume fraction. The statistical results are shown in
Table 4. Combined with the data in the
Table 4, according to Formula (2), the ln[−ln(1 − X)] − lnt curve in the recrystallization process could be obtained, shown by the straight line in
Figure 13.
The fitted results in
Figure 13 correspond to a primary function, indicating that the recrystallization mechanism did not change during recrystallization. In the JMAK equation, it is assumed that nucleation occurs randomly and uniformly in the matrix, the nucleation rate is constant, and the ideal value of the Avrami index is 4. The fitting Avrami index of the pure molybdenum sheet with 95% deformation in the process of recrystallization at 900 °C was 3.6, i.e., less than the ideal value of 4. This was mainly due to the unevenness of the tissue leading to an uneven distribution of the stored strain energy in the sheet. Some large grains did not have a subgrain structure with small grain boundaries inside; therefore, the recrystallization nucleation did not show a uniform random distribution.