Microstructure evolution and constitutive analysis of nuclear grade AISI-316H austenitic stainless steel during thermal deformation

Compression experiments were performed on AISI-316H austenitic stainless steel using Gleeble-3800 at temperatures ranging from 900 °C and 1200 °C and strain rates ranging from 0.01 and 10 s−1, up to the actual strain of 0.69. The tests aimed to examine the material’s microstructure evolution and flow stress behavior. Based on OM and EBSD studies, it was found that thermal deformation mostly induces discontinuous dynamic recrystallization (DDRX). The proportion of recrystallization nucleation increases steadily with increasing deformation temperature, while the impact of strain rate on recrystallization is complex. At the same deformation temperature, the recrystallization volume fraction initially declines and rises as the strain rate rises. In low strain rate regime, the longer (deformation) time available for grain boundary migration, the higher recrystallization volume fraction. In high strain rate regime, the higher stored energy (and thus the increased boundary velocity) raises the probability of nucleation events, stimulating twin formation. As a result, the twin promotes a dynamic recrystallization (DRX) process. An abundance of Σ3 twins was notably observed in uniformly refined recrystallized grains at a true strain of 0.69, at a temperature of 1200 °C, and a strain rate of 10 s−1. As a result, it was discovered that DRX occurs at higher strain rates and deformation temperatures. In addition, the flow stress curves were modified to account for adiabatic heating at strain rates exceeding 1 s−1. The findings demonstrated that adiabatic heating increased when strain level and strain rate increased and deformation temperature decreased. The strain compensation Arrhenius model is developed following the given stress–strain curve while considering strain. The model exhibits high accuracy, with a correlation value of 0.986. According to a kinetic study, the average activation energy for hot deformation of the tested steel was 444.994 kJ/mol. These findings provide fundamental insights into the microstructure control technology and the outstanding mechanical properties of the austenitic stainless steel AISI-316H.


Introduction
AISI-316 austenitic stainless steel (ASS) and its modified versions have garnered widespread adoption in sodium-cooled fast neutron reactors, attributable to the excellent high temperature strength, good plasticity and toughness, high pressure water corrosion resistance, good processing and weldability [1,2].Compared to standard 316 austenitic stainless steel, 316H contains a higher carbon content, which provides improved oxidation resistance and increased strength at elevated temperatures [3,4].Adding molybdenum (2%-3%) improves the material's ability to resist pitting corrosion and creep at high temperatures [5].ASS has higher alloying elements than carbon steel, increasing deformation resistance during hot working and minor singlepass deformation.In addition, due to the difference of deformation parameters and temperature at each position of the plate during hot forging or hot rolling, the phenomenon of coarse and mixed grains or inhomogeneous performance often occurs in the ASS thick plate (the total compression ratio is less than 3), which is difficult to be improved by solution treatment [6].The existence of abnormal coarse grains will adversely affect the mechanical properties of materials [7,8], such as the deterioration of elongation [7].If the coarse grains cannot be refined, the whole steel plate needs to be scrapped, thereby incurring significant economic setbacks.Therefore, comprehensive analysis of the microstructural evolution and flow softening behavior during hot deformation under various conditions is critical to minimize the flaws arising from deformation [9].
In recent years, because the flow curves can reflect the hot workability of the material, numerous researchers have examined the deformation behavior by using the conventional Arrhenius relation to make precise flow stress predictions [10][11][12][13].The softening of the material occurs during thermoplastic deformation via dynamic recovery (DRV) and DRX [14].The low stacking fault energy (SFE) of conventional ASS leads to the critical accumulation of moving dislocations, resulting in the easiness of DRX [15,16].DRX mechanisms include DDRX and continuous DRX (CDRX).The generation of DDRX is closely related to the energy stored in the material [17], and its main form is dislocations generated by plastic deformation [18].CDRX commonly involves the formation of low-angle grain boundary (LAGBs) arrays and the gradual enhancement of boundary misorientations, steered by the accumulation and rearrangement of dislocations during hot compression process [19].Contrary to preceding discoveries [20], where 304-type ASS showed DDRX at reduced strain rates and CDRX at elevated strain rates, this investigation unveils distinct phenomena.
Recent research revealed that the strain rate substantially affects the process of recrystallization in ASS.In the low strain rate zone, grain growth influences the DRX kinetics because of the extended deformation duration [17,21].In the high strain rate zone, rapid deformation of the material results in accumulating many dislocations, leading to increased internal energy storage and promoting recrystallization nucleation.Therefore, it was determined that the recrystallization rate was governed by high-speed nucleation [16,20,22].However, according to some experts, this phenomenon is caused by the adiabatic temperature rise (ATR), which accelerates the DRX process while bogging the hardening work rate [23].At an intermediate strain rate, DRX kinetics are slow because of the absence of primary nucleation or DRX grain growth [17].
In addition, twins can also enhance recrystallization in high and low-strain rate areas.In the low strain rate zone, twins contribute to separating the bulging region of the initial austenite grain boundary.However, the formation of the Σ3 twin boundary is chiefly attributed to the 'growing accident' during the nucleation and growth processes.In the high strain rate zone, twinning changed the orientation of boundary fronts without needing boundary migration.As a result, the DRX area was extended through repeated treatment.Σ3 twins and their variations encourage the conversion of particular coincidence site lattice (CSL) grain boundaries to random high-angle grain boundaries (HAGBs), thereby promoting DRX [20].Additionally, the material's diverse initial states impact the recrystallization behavior.For instance, the DRX process is slower in as-cast materials than in as-forged materials [24].
The weak thermal conductivity caused by deformation heating can decrease the accuracy of the hot compression test flow curve [25].Substantial plastic work is converted to heat when the sample experiences significant plastic deformation during thermal deformation [26].Adiabatic heating occurs when the material is rapidly deformed, the deformation heat fails to diffuse into the surrounding environment in time, which will lead to an increase in the temperature of the material, which in turn leads to softening.Due to their weak thermal conductivity, adiabatic heating is essential for the researched alloys.Devadas et al [27] provided a more widely used technique for modifying the stress-strain curve.Many researchers have demonstrated that this approach is highly effective [12,23,28], with lower temperatures resulting in higher strain rates and wider gaps between the instantaneous deformation temperature and the target temperature.
In recent times, a multitude of academic inquiries have delved into the hot deformation mechanisms of various ASS, including 304L [10], 316LN [17], and 317L [29], under diverse conditions.However, despite these efforts, a detailed exploration focusing on the microstructural evolution, DRX mechanisms, and flow stress of forged AISI-316H steel during hot compression has not been pursued.In past research, scholars have formulated Static and Dynamic Kinetics models for cold-rolled 316 ASS, serving as boundary conditions in finite element analysis [30].However, it is crucial to note that the cold-rolled material utilized substantially diverges from the forged material examined in this study, especially regarding grain size and internal dislocation density.Such disparate initial microstructures can wield a significant influence on recrystallization behavior during deformation, and the effect of strain rate on recrystallization is not discussed.Jafari et al [31] studied the correlation between Zener-Hollomon parameter and necklace DRX of 316 austenitic stainless steel at 950 °C-1100 °C and 0.01-1s −1 .Nevertheless, their research encapsulated somewhat constricted ranges of deformation temperature and strain rate, and did not profoundly investigate the flow behavior.Currently, there is a clear gap in the understanding of the relationship between microstructural evolution and flow stress behavior under different strain rates and deformation temperatures in the research on high-temperature deformation of AISI-316H ASS.Additionally, most existing studies primarily focus on the macroscopic mechanical behavior of the material, while providing relatively limited explanations for phenomena at the microscale, such as twinning formation and the specific mechanisms of DRX.
Therefore, the objective of this research are to systematically investigate the impact of hot deformation parameters on the microstructural evolution and flows softening of ASS by considering adiabatic heating.A constitutive model that depends on the applied strain was established to predict the flow behavior of the alloy by modifying the flow stress.EBSD is utilized to assess the impact of deformation parameters on microstructures during hot deformation.In addition, this paper explores the impact of strain rate on DRX behaviour, the role of twin on DRX and DRX mechanisms of AISI-316H steel during hot deformation.

Materials and experimental procedures 2.1. Material
Table 1 shows the chemical contents of the tested ASS in this study, which was hot-forged as received.The steel microstructure comprised averagely-sized austenite grains, approximately 64 μm (figure 1).To create cylindrical test specimens for hot compression testing, we machined samples with dimensions of 8 mm in diameter and 12 mm in height.These specimens were precisely cut with their axes aligned parallel to the forging direction.Lathe machining was employed to ensure a smooth surface finish, and we maintained dimensional tolerance within ± 0.5 mm.This machining process was carried out on the central region of the original plate received for testing.

Hot deformation tests
Compression tests were conducted using a Gleeble-3800 thermo-mechanical simulator manufactured by DSI in the USA, with a single-pass method.This equipment is powered by a hydraulic pump that drives hydraulic oil into the cylinder, thereby moving the piston and controlling the ISO-T anvil system for adjusting loading configuration and speed.To measure strain during the deformation process, an L Gauge was utilized.In order to attain the required workpiece temperature, resistance heating was employed.Temperature control was achieved using a PID controller, which relied on feedback from a chrome-alum-type thermocouple attached to the midhigh section of the specimen, ensuring real-time temperature monitoring during testing [30].Simultaneously, the compression tests were carried out in an argon environment to prevent oxidation of the sample surface.
The tests were conducted at temperatures (900 °C-1200 °C in 50 °C increments), strain rates (0.01-10 s −1 ) and target true strain 0.69.The sample was rapidly heated to 1200 °C at 10 °C/s, followed by homogenization at that temperature and kept for 180 s.The sample was then cooled at 5 °C s −1 and kept for 30 s before loading.To  analyze DRX behavior and unstable microstructure (figure 2), samples were quenched to room temperature within 1-2 seconds using water after reaching the desired strain levels.Tantalum foils, each with a thickness of 0.05 mm, were applied to both sides of the specimen as a lubricant to reduce friction-induced bulging phenomena during compression.

Microstructural characterization
Microstructures of hot deformed samples were obtained by cutting them parallel to the compression direction.
The samples analyzed by a Zeiss optical microscope (magnified at 100 times) underwent a polishing process and an etching phase in a 10 g FeCl 3 + 20 ml HCl + 20 ml H 2 O solution for 15 min.The misorientation angle was determined by mapping the microstructures of deformed steels using a field-emission scanning electron microscope (ZEISS SUPRA 55 Compact) equipped with Oxford EBSD technology.The EBSD scan date was analyzed by The HKL Channel 5 software.An accelerating voltage of 20 kV at a working distance of 15 mm was used.The samples for EBSD analysis were obtained by mechanically grinding and polishing them and then electropolished with a solution of 10% perchloric acid in ethanol at 20V for the 60s at 25 °C-27 °C.EBSD maps were obtained with a step size of 0.8-1.5μmdepending on the grain size.Additionally, we categorized grain boundaries based on the misorientation angle (θ), such as LAGBs for 2 θ <10, HAGBs for θ 15 and mediumangle grain boundaries (MAGBs) for 10 θ <15.To identify CSL boundaries, we applied Brandon's criterion [20].In addition, in order to ensure the accuracy of the data statistics, each deformed sample is counted at least two graphs from different regions of the same sample, and the data provided in this study is its average.

Flow behavior circumstances
The flow stress is affected by the temperature of deformation and the rate of strain.Increasing the strain rate causes a rise in flow stress, while increasing the temperature causes a drop in flow stress.Furthermore, the flow stress curves indicate a significant rise in the first stage of thermal deformation due to work hardening.DRX occurs when the deformation temperature exceeds 1050 °C, creating a visible peak in the flow curve at a strain rate of 0.01-5 s −1 .When the deformation temperature is below 1050 °C, a 'flat-top' type curve with a possible broad or absent peak is observed, indicating the possible existence of DRV [32].However, when considering the microstructure observations in section 3.4 and the hardening work curves illustrated in figure 3, it becomes evident that the softening observed during hot deformation can be attributed to the combined influence of DRV and DRX.Consequently, the presence of a 'flat-top' type curve can also signify the coexistence of DRV and DRX, a phenomenon that has been observed in other materials as well [32,33].As strain increases, the initial deformation stage is characterized by increased dislocation density and interaction, increasing deformation resistance.As deformation increases, dislocation recovery gradually increases until work softening approximately equals work hardening.Following maximum stress, the predominance of dynamic restoration, specifically DRV and DRX, can be attributed to reduced dislocation density and increased DRX.Numerous research studies have demonstrated that as deformation temperature or strain rate decreases, DRX behavior becomes more challenging under high Zener-Hollomon parameter (Z) conditions.However, this study shows that under 10s −1, the stress-strain curve reveals great peak stress.We found a high proportion of DRX based on observations of the metallographic structure.According to some scholars, this can be attributed to twins, which will be discussed in section 4.3.

Adiabatic heating impact on flow curves
Due to the slow reaction rate, the temperature control device of the experimental equipment cannot instantaneously adjust its temperature during hot deformation procedures with more excellent strain rates.Due to the large amount of plastic deformation energy converted into heat, the thermal history of the sample can transition from almost isothermal to almost adiabatic, leading to an accelerated recrystallization process [34].To mitigate the impact of temperature increase induced by deformation on the flow curve of the specimen, the current temperature of the specimen can be calculated using the following method [35].
Where ρ represents the density of test sample (7.99 g/cm 3 ), C p denotes the specific heat of test sample changed with temperature, which was calculated by laser-pulse method (as shown in table 2) [36], σ represents the flow stress, ε denotes the strain, and η corresponds to the thermal efficiency being calculated for austenitic stainless steel as follows [37]: 0.001s 0.316 log 0.95 0.001s 1s 0.95 1s Where  e is the strain rate.Select representative strains within the range of 0.05-0.69and calculate ΔT values at intervals of 0.05 by using equations (1) and (2) to investigate temperature elevation throughout the deformation process.In lower strain rates range (< 1 s −1 ), ATR was negligible as a result of prolonged deformation; thereby, curve correction procedures were unnecessary [38,39].Figure 4 illustrates the correlation between ART and deformation temperature and strain rate in the range of 900 °C-1200 °C and with strain rates of 1, 5, and 10 s −1 .A almost a linear relationship existed between ATR and strain under the current experimental parameters.It was observed that ΔT increased with the decrease of deformation temperature, increase of strain  rate and strain, and the maximum ΔT exceeded 31.59°C at 900 °C with 10 s −1 .Figure 5 shows the correlation between ATR with deformation temperature and strain rate under a strain level of 0.69. Figure 5 illustrates that the discrepancy between the measured and actual values is due to a sudden increase in deformation heat.Adjustment of flow stress curves is necessary for strain rates exceeding 1s −1 .Multiple strategies have been proposed [40][41][42].Devadas et al [27] presented the following as a more prevalent strategy: Where σ 2 represents the flow stress under corrected for deformation heating, σ 1 denotes the measured flow stress, Δσ represents the stress difference caused by deformation heat was calculated by the following [27]: Where T 0 is the set value of material deformation temperature, T i = T 0 +ΔT is the instantaneous temperature of hot deformation.However, the parameters in equation (4), such as n, Q, α, are based on measuring the actual flow curve, and the parameters affected by the deformation heat are inaccurate.Therefore, we utilized a novel approach proposed by Zhang et al [34] to correct the stress-strain curve in this study.The uncorrected flow stress values will be computed using the isothermal flow stress at the given strain values, strain rate, and target temperature T 0 under adiabatic conditions.  T T , , , , 5    s e e ¢ ( )and is the instantaneous softening rate due to adiabatic heating at T 0 .The correlation between temperature and actual stress was established to acquire the isothermal flow stress, as shown in figure 6.
It follows from equations (5) and (6) that.Therefore, the isothermal flow stress is determined by equation (6.2).In figure 6(b), in the case of 10s −1 , it is apparent that both the σ 2 -T i and σ 1 -T 0 data can be fitted by a linear relation with a high correlation R 2 0.992.In this case, equation (6.2) becomes.Therefore, the isothermal flow curve at the target temperature can be obtained through equation (7) or directly read from the σ 2 -T i curve.In figure 6(b), it is apparent that both the σ 2 -T i and σ 1 -T 0 data can be fitted by a linear relation with a high correlation R 2 0.992 at 10 s −1 .While in the case of 1.0 s −1 (figure 6(a)), a linear relationship is insufficient to adequately represent the σ 2 -T i and σ 1 -T 0 data.Thus, a second-order polynomial is introduced with a correlation coefficient of R 2 0.996, from which the isothermal flow stresses can be calculated.
Figure 7 illustrates the actual measured flow curves with the corrected flow curves for adiabatic heating.Notably, the differences between these curves become more pronounced under higher strain levels and lower deformation temperatures.This results from accumulated deformation heat with increased strain levels, causing higher thermal softening during deformation, as depicted in figure 4. Additionally, at higher strain rates, the  aforementioned phenomenon is more pronounced.Similar behavior has been noted in other materials [41,43].Greater strain rates lead to less time for heat dissipation during deformation and more temperature rise per unit of time.Thus, the difference between the curves amplifies with increasing strain rates.The flow stress softening caused by deformation heating is about 23 MPa, at 900 °C and 10s −1 .It was initially a dynamic softening curve, and after correction, it showed a hardening work curve, such as the 900 °C deformations rheological curve in figure 7(c).

Constitutive analysis
During hot deformation, the DRX process in metals is governed by thermal activation energy, which is correlated with the temperature and the deformation rate.When the deformation temperature is low and the deformation rate is high, the flow behavior is comparable, indicating that these two factors have a similar impact on the material during deformation.Generally, parameters of the heating process, such as deformation temperature and rate, are characterized using the Zener-Hollomon parameter Z [44].
Where Q represents the deformation activation energy, R represents the gas constant (8.314J mol −1 K −1 ),  e represents the strain rate, and T represents the thermodynamic temperature, σ which can be the peak, steadystate, or flow stress at a specific strain.Due to the wide range of peak stress in the flow curve, we utilize the stress value associated with a strain of 0.5 when calculating the constitutive equation.A 1 , A 2 , β, α, n, n 1 are material constants that are independent of temperature, and it's worth noting that α is equal to β/n 1 .
By applying a mathematical transformation to equation (8), it can be obtained that Multiple linear regression analysis was conducted on the stress data at a strain of 0.5, obtained at various strain rates and temperatures, utilizing the least squares method.The resulting coefficient α was determined to be 0.006647.Figures 8(a) and (b) illustrate the use of linear regression for the data points, allowing for the calculation of n 1 and β values from the reciprocal slope of lines in lnσ - ln e and σ - ln , e respectively.The mean value of n 1 and β is calculated to be 7.4925 and 0.0498.The value of α = β/n 1 = 0.0046647 can be obtained.Using linear regression for the data points, the values of n, Q, and A can be calculated from the reciprocal slope of lines in  ln sinh ln , as e -[ ( )] ln[sinh(ασ)−1000/T and lnZ-ln[sinh(ασ)], respectively.Through the above method, the values of n Q and A are 5.258, 431625.52,and 1.364 × 1016, respectively.Therefore, based on the above data, the constitutive equation for AISI-316H austenitic stainless steel can be expressed as the following equations: Nonetheless, the equation mentioned earlier should take into account the influence of strain on flow stress during high-temperature deformation.A constitutive equation has been established that model strains ranging from 0.1 to 0.69, focusing on magnitudes of 0.1.Values of n, Q, and lnA were computed at different stress levels to enhance the precision of material flow stress forecasting.The material constants and thermal activation energy can be represented as polynomial functions of the strain.The results demonstrate that five-order polynomial fitting is more accurate.Figure 9 displays the relevant results of each parameter.Significant differences were observed in α, n, Q, and lnA under different strains.Table 3 displays the coefficients of the polynomial function (13).
Figure 9 shows that α decreases gradually from 0.0077 MPa −1 to 0.0066 MPa −1 , with increasing strain, reaching its minimum at a strain of 0.4, and then increasing with further strain, reaching 0.00683 MPa −1 at 0.69.In addition to the influence of the alloy, α value is also associated with the deformation conditions.The variation range of the n value is 5.09342-6.63303.The calculated value of n in this study is similar to that reported in the literature for ASS [46].Decreasing deformation activation energy Q with strain means increased deformation storage energy during high-temperature deformation.The change range of Q value is 426.78518-475.66937KJ/ mol, with an average value is 444.99421KJ/ mol.The calculated values are also close to the deformation activation energy (350-510 KJ / mol ) of ASS reported in the literature [46].These values are notably higher than the activation energy for self-diffusion in γiron, which is approximately 280 kJ/mol.This disparity suggests that the dominant deformation mechanism is DRX rather than diffusion or DRV [47].Furthermore, these values are in line with reported values for the hot deformation of austenitic stainless steel, such as 375 kJ mol −1 in AISI-304L [48], 442 kJ mol −1 in AISI-316LN [49], and 433 kJ mol −1 in AISI-321 [50].This implies that the deformation resistance of 316H is comparatively higher when compared to 304L, 316LN, and 321 stainless steels.
To evaluate the predictive ability of the established constitutive model, the calculated Flow Stress based on the model in all experimental conditions is compared with the experimental values, as shown in figure 10.Overall, the established model demonstrates a high level of prediction accuracy.However, a notable discrepancy exists between the calculated and experimental values under certain deformation conditions.To assess the prediction accuracy of the established constitutive model quantitatively, the correlation coefficient (R) is introduced [51].
In the above equations, E represents the experimental data, P represents the predicted data, represents the average values of E and P , and n is the amount of data points.Figure 11 illustrates a strong linear correlation between the experimental and predicted values.The high correlation coefficient R of 0.986 provides strong validation for the accuracy of the constitutive equation in predicting the flow stress of the alloy.

Microstructures of hot deformed
Figure 12 presents optical microscopy images of samples subjected to compression at varying temperatures and strain rates.These images reveal that, for a given strain rate, the volume fraction of recrystallization grows with higher deformation temperatures, accompanied by an increase in grain size.However, the degree of recrystallization initially decreases with an increased strain rate but then increases.At all deformation temperatures, recrystallization kinetics slowed at an intermediate strain rate.At 900 °C, the initial grains were elongated, and the grain boundary became wavy, as indicated by the red circles.The observations reveal that strain hardening occurred, matching the predicted deformation conditions of 900 °C-0.1 s −1 .In some areas, flow is highly localized into deformation bands aligning with the shear direction.When the deformation temperature increased to 1050 °C, the structure comprised refined recrystallized grains and elongated original grains.Deformation at 1200 °C resulted in a completely recrystallized structure containing abundant annealing twins.

Analysis of DRX mechanisms
To gain a deeper understanding of the recrystallization mechanism and perform a quantitative analysis of the microstructure in the hot deformation samples, we utilized EBSD technology.Figure 14 presents the EBSD grain boundary maps and distributions of grain boundary misorientation for the samples subjected to different temperatures and strain rates.At 900 °C deformation temperature, the proportion of recrystallization in the microstructure is the lowest, and most of the sample being deformed, as shown in figure 14.As illustrated in figures 14(a)-(c), a limited number of DRX grains nucleated along the parent grain boundaries by bulging, forming a distinctive 'necklace' structure [52].This phenomenon has also been found in other ASS [53].This mechanism is mainly due to dislocation slipping and piling-up processes near the grain boundaries, which cause preferred nucleation and orientation evolution at these sites [54].Prior research has demonstrated that ASS of the 304 type experiences DDRX at low strain rates and CDRX at high strain rates [16].In contrast to the previously mentioned observations, DDRX was observed in both low and high strain rate regions during the deformation process.According to related investigations, the DDRX procedure is characterized by sawtooth and bulging at the original coarse grain boundary [55].Clear serrations and bulges, highlighted by yellow arrows, can be observed at the original grain boundary in figures 14(a)-(c).These features are closely linked to strain-induced grain boundary migration [10].Additionally, all of these serrated grain boundaries are accompanied by regions of high Kernel Average Misorientation (KAM), as depicted in figures 15(a)-(c).The data suggests that DDRX, marked by grain boundary bulging, was evident in this study.At a deformation temperature of 900 °C, as depicted in figures 15(a)-(c), newly developed strain-free DRX grains could be identified near the initial HAGBs.This observation strongly supports the occurrence of DDRX [56,57].It's well-understood that for CDRX to initiate through the gradual rotation of subgrains, the presence of MAGBs is essential [58].If the softening characteristics of the material in regions of high strain rate were majorly influenced by continuous dynamic recrystallization, a decrease in MAGBs would be expected with an escalating strain rate [20].However, the experimental results of this study are opposite to this phenomenon,  as shown in the figure 13.As a result, the DRX observed in this study can be classified as a DDRX process.Moreover, studies have highlighted that DDRX is more prevalent in stainless steel alloys with a reduced stacking fault energy during deformation [17,59].Therefore, this study is mainly about the deformation mechanism dominated by DDRX.

The impact of strain rate on DRX behaviour
The related KAM maps of the examined samples are displayed in figure 15.High KAM values indicate the accumulation of strain, which suggests high stored energy, while low KAM values imply the presence of recrystallized grains, which are dislocation-free [60].As can be inferred from figures 15(a)-(c), the specimen with the highest KAM value is shown in red, indicating a compression at 900 °C with a small amount of DRX, as shown in blue.This illustrates an accumulation network of dislocations, confirming the occurrence of dislocation slip activities [61,62].The KAM value of the specimen deformed at temperatures below 1100 °C-1200 °C is low, implying that the high deformation temperature causes a relaxation of stored deformation energy.As a result, the proportion of recrystallization nucleation increases steadily with increasing deformation temperature.
On the other hand, the impact of strain rate on recrystallization is complex.At the same deformation temperature, the recrystallization volume fraction initially declines and rises as the strain rate rises, as showened in figure 16.At the same time, the KAM value exhibited an initial increase and then a sharp decrease in figures15.This phenomenon can be attributed to the lowering of the critical strain requirement for the occurrence of DRX with the strain rate decreasing in low strain rate region [63].Moreover, the longer (deformation) time available for grain boundary migration, the higher recrystallization volume fraction.In high strain rate regime, higher stored energy (and thus the increased boundary velocity) raises the probability of nucleation events, stimulating twin formation.As a result, the twin promotes the DRX process, which have be discussed carefully in section 4.3.

The role of twin on DRX
The Σ3 and Σ9 CSL boundaries of the compression-tested samples at various hot deformation processes are displayed in figure 17.Initially, the number of twins decreases during high-temperature deformation but then increases with increasing strain rate until it reaches a minimum at 1s −1 .The volume proportion of LAGBs steadily decreased when twins emerged, whereas the volume fraction of HAGBs, special Σ3 twin boundaries, rapidly grew, as illustrated in figure 17(a).At temperatures of 1100 and 1200 °C and 10 s −1 , the twinning rate significantly increased, resulting in recrystallization fractions of 81.7% and 96.2%, respectively, and decreased the KAM value.Figure 15(i) and figure 17(a) demonstrate that the twinning mechanism primarily achieved a homogeneous, refined microstructure.In the process of dynamic recrystallization, twins significantly affect the nucleation and growth of recrystallized grains.When grain growth stagnation occurs during DRX, twinning will help restore growth [64].In cases where recrystallized grains interface with the deformed matrix after growth stagnation, these interfaces often exhibit low misorientation and/or low dislocation density.Twinning occurring at these interfaces brings about a significant change in misorientation, supplying the necessary additional boundary energy for migration and growth.This twinning not only facilitates the perpendicular expansion of the DRX volume concerning the interface but also promotes growth in the parallel direction [65].Furthermore, twinning within the interface can expedite the bulging process [58] and assist in the separation of bulged portions from the parent grain [66].This study confirms that recrystallization is influenced by the twinning mechanism at high temperatures and high strain rates, which accelerates the nucleation and growth of DRX.Similar phenomena have been found in other study [17].In general, the interaction between the existing Σ3 boundaries and 'growth accidents' during grain boundary migration will increase CSL boundaries fraction (special Σ3) during hot deformation of low SFE materials [67][68][69].The interaction mechanism between the preexisting Σ3 boundaries usually leads to an increased Σ9 boundary fraction, which is known to generate higherorder twin boundaries (i.e., Σ3 n , n > 1) boundaries [67].However, in this study, the fraction number of the Σ9 boundary did not increase significantly.Consequently the emergence of Σ3 twin boundaries in this study was mainly due to 'growth accidents' during grain boundary migration.Previous research has shown that during high-temperature deformation, the activation of twinning is mainly influenced by the strain rate.Moreover, the impact of twinning on DRX and microstructure evolution in ASS at a high strain rate has also been reported in earlier studies [17].In addition, the CSL boundaries are found to be effective in preventing passive intergranular  corrosion [70], improving the resistance of grain coarsening [71], reducing the propagation rate of fatigue crack [72], increasing ductility and uniform elongation [73].
In this study, figure 18 displays the proportion of recrystallized, substructured, deformed microstructure, and twinned regions within ASS under diverse compression conditions.The results suggest that the alloy contains numerous LAGBs at the lower deformation temperature.Moreover, as the deformation temperature increases, the fraction of HAGBs in the alloy also increases.These findings illustrate that a higher deformation temperature enhances the recrystallization process.

Conclusions and future research
This study explored the heated deformation behavior of AISI-316 ASS under different hot compression conditions of 0.01-10s −1 and 900 °C-1200 °C.This study analyzes the microstructural changes under varying deformation conditions, constructs a strain-dependent constitutive model, and conducts a comprehensive investigation into correction procedures for adiabatic heating.The results are as follows.
(1) Deformation Heating and Flow Softening: Deformation heating significantly affects the dynamic flow softening of ASS steel at high strain rates.This effect is more pronounced at lower temperatures and higher strain rates, with a clear correlation between thermal softening magnitude, temperature decrease, and strain rate increase.
(2) Stress-Strain Curve Adjustment: To counteract the influence of ATR during high-speed deformation, adjustments were made to the stress-strain curve.This step helps mitigate the impact of temperaturerelated variations on material behavior.
(3) Constitutive Model and Strain Compensation: A constitutive equation was developed for predicting true stress at a true strain of 0.5 using kinetic analysis.Additionally, a strain compensation model based on the Arrhenius model was proposed.The model demonstrated high accuracy, as evidenced by a correlation coefficient of 0.986 when comparing experimental and predicted results.
(4) DRX Mechanisms: The study revealed that DRX in as-forged AISI 316H ASS is influenced by both dislocation and twinning mechanisms.The extent of control exerted by these mechanisms depends on the applied temperature and strain rate.Dislocation mechanisms dominate at lower temperatures, while twinning mechanisms play a more significant role at higher temperatures and strain rates, accelerating DRX kinetics.
(5) Σ3 CSL Boundaries and DRX Grains: As the volume fraction of DRX grains increases, the proportion of Σ3 CSL boundaries also rises.This suggests that a substantial portion of Σ3 boundaries forms during the DRX process due to 'growth accidents.'The highest fractions of Σ3 boundaries were observed at a strain rate of 10 s-1, and the evolution of Σ9 boundaries exhibited similar behavior due to interactions between Σ3 n boundaries.
In summary, the study's findings provide valuable insights into the complex interactions between deformation conditions and microstructural changes in ASS steel.These insights have implications for controlling and understanding the mechanical properties of such materials under varying deformation scenario.
Based on these conclusions, potential gaps and future research directions are identified.In this study, the deformation heat generated during hot compression was corrected.However, the bulging of the sample caused by friction during hot compression was not corrected by friction, so as to further improve the accuracy of the data.The next step is to establish a thermal processing map, and use the data obtained in this study to verify the recrystallization during heating and deformation by means of simulation and industrial production.

Figure 1 .
Figure 1.Original microstructure of the AISI 316H ASS used in this investigation.

Figure 2 .
Figure 2. Process flow diagram of hot compression experiment.

Figure 3 .
Figure 3.The flow stress curves of the hot deformed ASS were obtained at different deformation conditions.

Figure 5 .
Figure 5. Contour plot of ATR at a strain of 0.69.

Figure 6 .
Figure 6.Adiabatic heating correction method diagram: (a) linear relationship; (b) The nonlinear relationship between temperature and actual stress.

Figure 7 .
Figure 7. Comparisons between the actual stress-actual strain curves, both corrected and uncorrected for deformation heating at different temperatures for strain rates of (a) 1 s −1 , (b) 5 s −1 , and (c) 10 s −1 .

Figure 11 .
Figure 11.Correlation between experimental and strain-compensated predicted flow stresses.

Figure 13 .
Figure 13.The fractions of MAGBs at various strain rates.

Figure 16 .
Figure 16.Volume fractions of DRX grains at different strain rates.

Figure 18 .
Figure 18.The proportion of recrystallized, substructured, deformed microstructure and twin within ASS under diverse compression conditions.

Table 2 .
Values of specific heat of AISI-316H ASS.
[45]ars andMcTegart's research findings[45]indicate a distinct mathematical correlation between flow stress and the Z-parameter for thermal deformation, particularly under high temperatures:

Table 3 .
Polynomial coefficients of α, n, Q, and lnA for investigated alloy.