Recrystallization kinetics of larger yttria particle dispersion tungsten alloy warm rolled to 50% thickness reduction

Tungsten-based materials are considered one of the candidate materials for the key components (such as the first wall and diverter) of future thermonuclear fusion reactors because of their excellent properties. At high operation temperatures, restoration processes such as recovery, recrystallization and grain growth will unavoidably embrittle ductile tungsten achieved by rolling. Dispersed second-phase particles will affect the recrystallization behavior. The thermal stability of a yttria dispersion-strengthened tungsten plate warm-rolled to 50% thickness reduction is investigated. Isochronous annealing is used to predict its recrystallization temperature range and isothermal annealing is used to study its recrystallization kinetics. The tungsten alloy in the present study has a faster recrystallization kinetics and a smaller activation energy compared to that of W-2 vol% Y2O3 alloy with small Y2O3 particles, indicating that the larger Y2O3 particles have little effect on impeding grain boundaries during recrystallization.


Introduction
Tungsten-based materials are considered to be the most promising candidate for plasma-facing materials due to their high melting point, good high-temperature stability, good radiation resistance, low sputtering rate, and low tritium retention [1].In fusion reactors, tungsten-based materials that are strengthened by deformation will experience recovery, recrystallization, and grain growth problems when subjected to long-term high-heat loads [2], resulting in the degradation of mechanical properties, especially recrystallization embrittlement.Embrittlement can pose a hazard to fusion reactors [3], but the use of yttrium oxide dispersion strengthening and deformation strengthening can improve the strength and high-temperature stability of tungsten [4,5].Currently, there is insufficient research on the recrystallization behavior and microstructure of oxide particle-reinforced rolled tungsten plates under steady-state thermal loads, and further research is needed.
The second-phase particle can affect the recrystallization kinetics [6].Generally, larger secondphase particles are more effective in promoting recrystallization because they provide more nucleation sites and have higher interfacial energy with the matrix, which drives the grain boundary migration.This nucleation mechanism is also known as particle-stimulated nucleation.On the other hand, small second-phase particles can hinder recrystallization by pining grain boundary migration and reducing the number of available nucleation sites.Additionally, the presence of large second-phase particles can lead to heterogeneous nucleation, which can result in a non-uniform grain size distribution and texture [7].
This paper focuses on a yttria particles reinforced tungsten plate as the research object.The annealing experiment is used to simulate its long-term high-temperature service environment in the fusion reactor.Recrystallization kinetics was analyzed by fitting the JAMK equation to hardness evolution during isothermal annealing.The rolling microstructure and yttria particle is studied using metallography, SEM and EBSD, and the influence of the yttria particles on the subsequent recrystallization of tungsten-based materials is preliminarily studied.

Materials and experiments
The material studied in this paper is a W-0.5 wt.%Y2O3 plate with 50% rolled thickness reduction (referred to as WY50), prepared by powder metallurgy technology and a warm rolling process.Refer to the literature [8,9] for the specific preparation process.Different from the yttria-strengthened tungsten plate with the same rolling thickness reduction before, the yttria particles of tungsten alloy in this study have wider size distribution and larger average size.
Small samples of 6×5×7 mm 3 were cut along the rolling direction (RD), transversal direction (TD) and normal direction (ND) from the WY50 plate for subsequent annealing treatment.To prevent oxidation during annealing, the samples were encapsulated in quartz glass ampoules.The samples were placed in a KSL-1500X batch-type furnace, removed after the desired annealing time, and cooled to room temperature by forced air cooling.
After annealing, the micro-hardness of specimens on the RD/ND surface was measured.A diamond Vickers indenter is applied with a load of 10 kgf and removed after a dwell time of 10 s. 12 indents were performed for each condition.The smallest and the largest hardness value were discarded.and the average value and standard deviation of the remaining values were reported.
The details of the deformed microstructure were characterized by optical microscope, scanning electron microscope (SEM) and electron backscatter diffraction (EBSD).EBSD was performed by a Zeiss Gemini 500 SEM with an Oxford C-Swift EBSD detector using a voltage of 20 kV and a step size of 1 µm.The open-source MTEX toolbox [10] was used to analyze the EBSD data.Yttria particles are characterized by backscattered electron imaging in the Zeiss Gemini 500 SEM.The method of metallographic observation is referred to as a reference [11].The orientation differences between adjacent pixels are measured, and the existence of a boundary is inferred when the misorientation angle is at least 2°.Misorientation angles between neighboring pixels of 2° to 15° are categorized as low-angle boundaries (LABs), while angles of at least 15° are classified as high-angle boundaries (HABs).Based on the assumption of HABs between individual grains, grains were detected and reconstructed using a threshold angle of 15° with the ability to determine their size.The grain contains a minimum of six pixels inside.

Characterization of rolling microstructure and yttria particles
To study the variation of grain size along the thickness direction (ND) of the rolled plate, the microstructure of the RD/ND section of the rolled plate was obtained in the different regions along the thickness direction, as shown in figure 1.The grains are colored according to their area.Figure 1a-c correspond to the surface layer, 1/4 transition layer, and center layer, respectively.The identified grains in figure 1 are complete grains with grain boundaries, with 581, 543, and 610 grains, respectively.Rolling grains show fibrous morphology.The average equivalent circular diameter (ECD) and aspect ratio (AR) were further calculated and presented in table 1.The ECD and AR for the three regions are numerically similar.Considering the relatively small number of samples in the statistical analysis and the possibility of some edge grains not being fully counted, the results may be affected.Overall, the grain size from the surface layer to the center layer is homogeneous.Figure 2 shows a micrograph obtained through backscattered electron imaging, where black regions represent yttria particles and gray regions represent the tungsten matrix, to analyze the size and distribution of yttria particles.In figure 2a, 166 yttria particles were detected and the equivalent circular diameter of the particles was calculated to obtain a frequency distribution histogram (as shown in figure 2b) showing a unimodal distribution.The equivalent circular diameter of the yttria particles ranged from 0.8 to 5.1 μm, with an average size of 1.8 μm.The particle size distribution peaked at 1.0-1.5 μm, accounting for 30% of the total particles.

Prediction of recrystallization temperature
To explore the approximate range of recrystallization temperature, the temperature range at which the hardness value decreases rapidly can be obtained through isochronal annealing experiments, thereby obtaining the interval of recrystallization temperature.Recrystallization temperature is affected by factors such as the initial deformation structure, temperature and strain of the processing process.Recrystallization is not a phase transformation, and there is no constant temperature but a temperature range.Reasonable design of isothermal annealing experiments can be achieved through the obtained range of recrystallization temperature.
The evolution of hardness value with increasing temperature obtained by isochronal annealing is shown in figure 3. It can be seen that after annealing for 1 h at a temperature ranging from 1100 ℃ to 1300 ℃, the hardness change is small, and it is relatively close to the starting hardness value of 442.0 HV10 in the rolled state, and the standard deviation is small.At 1350 ℃, the hardness value drops to 415.3 HV10, and the standard deviation increases.Recrystallization may have started at some locations.Subsequent metallographic observations confirmed this phenomenon, as shown in figure 4. Some recrystallized zone with obvious changes in grain morphology was found in the lower left region of the metallographic image after annealing at 1350 ℃ for 1 h.Some large Y2O3 particles can be observed within the recrystallized zone.This indicate that large Y2O3 particles preferentially accelerate recrystallization through the PSN nucleation mechanism (i.e.increased nucleation sites) [6].During the recrystallization process, the size and orientation of the grains will change, which will cause the mechanical properties to change [7].The hardness value of the deformed grains is larger than that of the recrystallized grains.The coexistence of the two types of grains leads to a large scatter in hardness values.In the range of 1350 ℃-1400 ℃, the hardness drops sharply to around 360.4 HV10.At this condition, complete recrystallization is about to occur.Therefore, the recrystallization temperature range of WY50 is predicted to be between 1350 °C and 1400 °C.

Recrystallization kinetics
The hardness curve of WY50 after high-temperature annealing at different temperatures is shown in figure 5a.At all three temperatures, with the extension of annealing time, the hardness value of the samples showed a monotonically decreasing trend.The hardness value decreased from around 442.0 HV10 in the initial rolled state to around 360 HV10 in the fully recrystallized state, but the rate of hardness decrease at different annealing temperatures was different.As shown in figure 5a, the hardness value decreased at a relatively flat rate during annealing at 1250 ℃ for 0-4 h.The hardness value only dropped from 442.0 HV10 in the rolled state to 435.5 HV10.The recovery mainly occurred at this stage, and there was no obvious recrystallization behavior.When the annealing time continued to increase to 24 h, the hardness value dropped significantly from 435 HV10 to around 360 HV10.At this stage, Recrystallization dominates the softening of hardness.By the annealing time of 24 h, the sample had been fully recrystallized.During isothermal annealing at 1300 ℃, the degradation of hardness value was similar to that at 1250 ℃, but the time for the hardness value to decrease slowly was greatly reduced.Afterward, with the further increase of annealing time, the hardness value remained unchanged.When the isothermal annealing experiment was carried out at 1350 ℃, the early stage of the hardness curve did not show the stage of slow decrease as at 1250 ℃, and the hardness value directly decreased at a faster rate to around 360 HV10.The hardness value dropped to around 360 HV10 when the annealing time was 4 h.The time required for the hardness value to drop to around 360 HV10 was further shortened.When the final equilibrium hardness value was reached at all three temperatures, the material had been fully recrystallized.It can be concluded that the completion time of the recovery stage in the isothermal annealing experiment at the three temperatures was different.The completion time of the recovery stage at 1250 ℃ and 1300 ℃ was 4 h and 1 h, respectively, while there was no obvious completion time for the recovery stage at 1350 ℃.As the annealing temperature increased, the time required for the completion of the recovery stage became shorter.The time required to reach the fully recrystallized state in the isothermal annealing experiment at the three temperatures was different.
It can be seen that the standard deviation of the hardness values of the samples in the partial recrystallization stage is large.This was caused by some regions beginning to undergo recrystallization while other regions may remain deformed structures, resulting in a large difference in the microstructure.This may be related to the addition of second-phase particles.If the distribution of yttrium oxide particles is not uniform in the tungsten matrix, this can lead to pinned grain boundaries in regions containing more yttria particles and hinder recrystallization, while those regions containing fewer yttria particles will preferentially undergo recrystallization, resulting in lower hardness values.This requires subsequent metallographic microstructure analysis.When the recrystallization was complete, the standard deviation of the sample's hardness value decreased, indicating that the hardness was more uniform at this time.
The mixing rule can be used to obtain a relationship between the hardness value at the different conditions and the volume fraction of recrystallized material:

HV=X•HV rex+(1-X)•HVrec
(1) HVrec represents the hardness value when recovery is complete, and HVrex represents the hardness value of the recrystallized matrix.
From this equation, by substituting the hardness values of each condition, the volume fraction of recrystallized material represented by each point can be obtained.figure 5b shows the variation of the volume fraction of recrystallized material in the center layer of the sample with annealing time.
The analysis of recrystallization kinetics can be carried out using the Johnson-Mehl-Avrami-Kolmogorov (JMAK) model, which is used to describe the typical curve of recrystallized volume fraction from the beginning of nucleation to the completion of recrystallization.The JMAK model is expressed by the following equation [12]: ) where X represents the recrystallized volume fraction, an Avrami exponent n and a rate coefficient b characterize nucleation and growth of recrystallized nuclei in a combined manner.All three kinetic parameters are obtained by fitting of the experimental hardness data.
Figure 5b shows the fitting results of the recrystallized volume fraction, and the difference between the original measurement data and the fitting curve is small, with the original data distributed on or near the curve.The parameters of the three curves are shown in table 2. With the increase of annealing temperature, the b value gradually increases.The b value describes the nucleation rate, and the faster the recrystallization rate, the higher the temperature.The n values of the three curves are similar because the selected hardness values all come from the center region of the sample thickness, and the nucleation and growth modes are similar.The incubation time tinc gradually decreases with increasing temperature, with values of 4.05 h, 2.11 h, and 0.48 h, which are close to the results reflected by the measured hardness curves of 4 h, 2 h, and 0.5 h, respectively.Recrystallization requires grain boundary migration to form new stress-free grains, and the initially deformation grains be merged by new recrystallized grains.The migration rate of grain boundaries is related to temperature.This relationship can be expressed by the Arrhenius equation [13]: ) where Q represents the activation energy, R is the gas constant, and T is the annealing temperature.For the same material, the activation energy and structure-dependent pre-factor t* remain the same regardless of the annealing temperature and time.Typically, the value of Q is determined by using the time tX = 0.5 corresponding to the recrystallized volume fraction reaching 50% and the corresponding temperature T.
Figure 6 shows the relationship between ln(t) and 1/T.Based on the fitting curve, the activation energy for recrystallization is calculated to be Q = 383 (1±5%) kJ/mol, which falls within the range of activation energies for self-diffusion along grain boundaries in pure tungsten (377-460 kJ/mol) [14].The activation energy of tungsten alloy in this study is less than that of tungsten alloy rolled plate with small yttria particles (508 kJ/mol) but greater than that of pure tungsten rolled plate (320 kJ/mol) with the same rolling amount [9].A reasonable assumption is that the pinning force caused by small particles hinders the diffusion of grain boundaries, which leads to an increase in activation energy.However, the distribution of yttria particles in this paper is relatively wide, and the majority of yttria particle size is large, which has little effect on the inhibition of recrystallization, which is also confirmed by the recrystallization kinetics.

Conclusion
The recrystallized temperature of a larger yttria particle dispersed tungsten alloy was predicted by isochronal annealing and its recrystallization kinetics were analyzed by isothermal annealing.The obtained activation energy is higher than that of pure tungsten plate due to the dispersion strengthening effect.However, compared with tungsten alloys containing small yttria particles, the activation energy is significantly smaller.Larger yttria has less effect on the inhibition of recrystallization during annealing.

Figure 1 .
Figure 1.Grains colored according to their area on the RD/ND section of the rolling plate in different zones: (a) surface layer, (b) 1/4 transition layer and (c) center layer.

Figure 2 .
Figure 2. (a) Backscattering images on RD/ND section of WY50; (b) Equivalent circular diameter distribution of yttria particles.

Figure 3 .
Figure 3. Hardness evolution of WY50 during isochronal annealing at different temperatures for 1 h.

Figure 4 .
Figure 4. Metallography test on the RD-ND section of WY50 during annealing at 1350 ℃ for 1h.

Figure 5 .
Figure 5. Evolution of the Vickers hardness (a) and recrystallized volume fraction X (b) of WY50 with time during annealing at different temperatures; the solid lines in figure 5b present the obtained description of the recrystallization kinetics by equation (1) and (2).Table2.Parameter values of the fitting curve.

Figure 6 .
Figure 6.Arrhenius relationship between annealing temperature and the time to half-recrystallization.

Table 1 .
Statistic of average equivalent circular diameter and average aspect ratio of grains in different rolling regions.

Table 2 .
Parameter values of the fitting curve.