Effect of gradient temperature rolling process on precipitation behavior in Q690D steel

In this paper, the second phase precipitation model is incorporated into the finite element simulation system of hot rolling process through the secondary development subprogram VUSDFLD of ABAQUS, and the simulation analysis and calculation of nucleation rate, volume fraction and size of precipitates under different temperature fields and strain fields are realized. The size, morphology and distribution of the microalloyed carbonitride precipitates in Q690D steel under gradient temperature rolling (GTR) and uniform temperature rolling (UTR) were studied by laboratory rolling experiments. The results show that the nucleation sites of the microalloyed carbonitride precipitates in the matrix increase under GTR, which is conducive to the nucleation of new precipitates and promotes the re dissolution of large precipitates. Finally, after tempering, the precipitate particles are distributed randomly on the ferrite matrix, and the size is refined by 20 ∼ 35 nm, resulting in obvious precipitation strengthening and microstructure refining effects.


Introduction
Ultra-heavy steel plates are widely used in heavy machinery manufacturing, infrastructure construction, marine engineering construction and offshore oil exploitation platform building. This kind of steel plate requires not merely high strength but also low temperature toughness, good welding property, lamellar tearing resistance, and homogeneous microstructure [1][2][3]. However, it is a challenging problem to increase the reduction inside by conventional rolling process. Recently, Yu et al [4][5][6][7] presented a novel rolling method: gradient temperature rolling (GTR) process, corresponding to this is the traditional rolling process: uniform temperature rolling (UTR). The effect of GTR process on promoting grain size refinement and improving the mechanical property of ultra-heavy plate was verified through experiment researches.
In the process of thermoplastic processing, the precipitation behavior of microalloyed carbonitrides is affected by deformation degree, deformation temperature, deformation rate and cooling rate. Wang Fangli [8] studied the precipitation behavior of the second phase of V and Nb at high temperature by means of stress relaxation test, thermal expansion test and TEM (Transmission Electron Microscopy) test, and measured the 'C' type PTT curve (Precipitation Temperature Time Curve) of a certain composition steel grade. It is found that the increase of compression deformation at a certain temperature will promote the precipitation process, and the size of precipitates will also gradually decrease, and better fine grain effect will also be obtained. Wang Xiaonan et al [9] found through thermal simulation experiment and transmission electron microscope observation that the increase of cooling rate will inhibit the precipitation process, and the temperature of precipitation will also decrease with the increase of cooling rate. The precipitation behavior of the second phase accompanied by the thermal deformation process is usually called deformation-induced precipitation. With the increase of deformation in the non-recrystallization zone, the increase of crystal defects such as deformation bands, dislocations and substructures in austenite provides more favorable nucleation sites for the precipitation of microalloyed carbonitrides. The number of precipitates increases significantly and the size of precipitates is smaller, and the distribution of precipitates in the matrix is more dispersed [10].
Precipitation kinetics is a mathematical model for the volume fraction and size of precipitated phases changing with temperature and time. Pereda et al [11] obtained a model for predicting austenite evolution and precipitation behavior of Nb microalloyed steel by fitting and functionalizing the data through thermal simulation and stress relaxation experiments. The model comprehensively considers the interaction of macroscopic deformation, solute hindrance to recrystallization, and strain-induced precipitation. Medina et al [12] summarized the strain-induced precipitation kinetics model of V and Nb microalloyed steels with wider applicability. Through model prediction and experimental results, the increase of strain will shorten the nucleation and incubation period of precipitates. Zurob et al [13] established a mathematical model for the coupling of stress field, precipitation and recrystallization behavior of microalloyed steel, and comprehensively considered the effects of strain, temperature, time and other factors on austenite recrystallization and precipitation process.
Generally, in steel materials containing Nb, V and Ti, the carbonitride precipitates of micro-alloying elements is an important kind of structure. The precipitation of the second phase can produce a certain strengthening effect, and can also play a role in refining grains. The strengthening effect of the second phase and its influence on the plastic toughness of steel are closely related to its volume fraction, particle size, shape and distribution uniformity. Based on the deformation state of steel under hot working and heat treatment conditions, combined with the thermodynamic and dynamic models of precipitates, the finite element simulation and experimental methods are comprehensively used to study the influencing factors of the size, morphology and distribution of precipitates, and a reasonable hot working system is formulated to control the precipitation behavior of the second phase.
2. Method 2.1. Numerical simulation of second phase precipitation behavior of steel Thermo-calc thermodynamic analysis software can be used to calculate the volume fraction of each phase and the elemental composition of the precipitated phase during the cooling process of the steel from the austenite zone. However, only taking the composition and temperature of steel as the influencing factors of precipitation behavior can not meet the actual conditions of steel plate rolling production process, and the influence of deformation conditions on the precipitation of the second phase in steel must be considered. Due to deformation, a large number of dislocations are generated in the matrix, which promotes the precipitation behavior, that is, the so-called deformation-induced precipitation. If the thermodynamic and kinetic mathematical models of precipitation behavior can be coupled into the numerical simulation software, the simulation analysis and calculation of nucleation rate, volume fraction and size of precipitates under different temperature and strain fields can be realized, which can play a certain reference significance for practical engineering production.
According to the research results obtained by Yong Qi long et al [14] , the thermodynamic and kinetic conditions of precipitation and phase transformation of the second phase in steel materials need to be comprehensively considered for the control of the volume fraction and size of the second phase. By studying and analyzing the PTT curve and NrT curve (Nucleation Rate Temperature Curve) of the second phase precipitation phase transformation obtained by theoretical calculation and the Ostwald ripening law of the second phase in the microalloyed steel with the determined composition, the precipitation amount and size characteristics of the second phase in the steel are determined from the perspective of theoretical calculation. The precipitation process of the second phase is essentially a nucleation-growth phase transformation, which is similar to recrystallization and conforms to the basic form of Avrami type equation. The calculation formula of volume fraction Xp [15] is (1-1): In the formula, t prs is the starting time of strain-induced precipitation. According to the calculation model proposed by Dutta and Sellars, it is considered that the time when 5% precipitation is obtained is the starting time of precipitation. The value of n p is related to the content of microalloyed elements in steel, and generally ranges from 1.54 to 2.05. For the second phase is precipitate containing Nb element, the calculation formula of t prs is shown in (1-2) and (1-3): In the formula, A and B are constants related to the content of Nb, C and N elements;ε represents Equivalent plastic strain;e ⋅ Equivalent plastic strain rate; Z represents the Zener-Hollomon parameter; T represents deformation temperature, K; R represents universalgas constant, 8.314J·mol −1 ·K −1 ; k s represents the supersaturation ratio at temperature T.
Based on Dutta Sellars model, Medina et al proposed a new improved calculation equation such as formula (1-4) according to the in-depth mining of experimental data [12]. This equation not only considers the effect of strain and strain rate on precipitation behavior, but also considers the effect of austenite grain size on precipitation nucleation. And this equation is no longer limited to the calculation of the volume fraction of precipitates containing Nb. The calculation formula of each parameter in the formula is shown in (1-4): In the formula, A and B are also constants related to the element content of the precipitate in the steel,ε represents Equivalent plastic strain, R is the universalgas constant, T is the rolling temperature, D is the austenite grain size, and Q is the diffusion activation energy.
According to the above formula and literature reference data, the calculation formula of each parameter is shown in table 1.

Calculation of critical nucleation size of second phase precipitation
For the second phase existing in the microalloyed steel, the nucleation process of precipitates in the hot working process basically belongs to the heterogeneous nucleation, that is, the new phase core is formed unevenly at the interface, dislocation and other specific positions in the parent phase. The existence of dislocation originates from plastic deformation, so it can be considered that the nucleation of the second phase in the deformation induced precipitation process mainly occurs on the dislocation line. On the basis of Cahn's [16] theory, researchers [14] put forward the optimized theory of precipitation nucleation on dislocation lines. The described calculation process of nucleation rate I d , critical nucleation size d c ⁎ (diameter) and subcritical nuclear embryo size d mc (diameter)is shown in (1-5)-(1-7): Nb: 270000 V: 264000 Where, DG v is the free energy of phase change per unit volume, and s is the specific interface energy of the twophase interface; A is a parameter related to shear modulus G, Poisson's ratio ν, and Burgers Vector b. A = Gb 2 /[4π(1-ν)] (edge dislocation) or A = Gb 2 /[4π] (screw dislocation). When β > −1, the critical nucleation energy DG d ⁎ formula is shown in (1)(2)(3)(4)(5)(6)(7)(8): The nucleation rate I d on the dislocation line is shown in (1)(2)(3)(4)(5)(6)(7)(8)(9): Where K is the influence parameter related to material composition, thermodynamic conditions and other factors; ρ is dislocation density, b is Burgers vector, k is Boltzmann constant; Q d is the diffusion activation energy of solute atoms along the grain boundary, which is generally 2/3 of the intragranular diffusion activation energy.
Because the current value of K is difficult to accurately calculate, it is difficult to accurately obtain the true value of nucleation rate. Dislocation density ρ is directly related to the work hardening and recrystallization softening of steel under hot deformation. During hardening ρ increases, but recovery and dynamic recrystallization make ρ decrease. The dislocation density changes with the thermal deformation under the competition between work hardening and recrystallization softening. The volume fraction of dynamic recrystallization is less than 2.9 % [17] due to the limitation of reduction rate per pass. The original austenite grain refinement of ultra-thick steel plate mainly comes from static recrystallization and sub-dynamic recrystallization. Therefore, a dislocation density calculation model without considering dynamic recrystallization softening and only considering dislocation climbing and subgrain merging caused by dislocation reduction is adopted. The calculation of dislocation density is shown in Formulas (1-10) [18].

Size model of second phase precipitates
The nucleation growth will occur in a period of time after the critical nucleus appears, and the process is controlled by the diffusion behavior of solute atoms in the matrix. According to the research results of literature [19], the change model of crystal nucleus size of spherical precipitates (radius) with time is shown in (1-11): where D is the diffusion coefficient, c 0 is the average concentration of solute atoms in the parent phase, c M is the concentration of solute atoms in the side of the phase interface out of the parent phase, c N is the concentration of solute atoms in the side of the precipitated phase. When the precipitation phase transformation process is completed, if it stays at a certain temperature for a period of time, the new phase particles with large size will continue to grow and coarsen, while the growth driving energy will be transformed into interface energy, and the second phase particles with small size will dissolve. This process is often referred to as the Ostwald ripening process of the second phase, which has a very large effect on the final size of the precipitate. The ripening process of the second phase precipitated on the dislocation line or grain boundary generally conforms to the law of t 1/3 or t 1/4 with time [14]. Assuming that the precipitate particles are spherical, the relationship between the average size (radius) and time is shown in (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12) and (1-13): Where r 0 is the initial particle size; σ is the interface energy; V P and V B are the molar volumes of precipitated phase and matrix, respectively; c 0 is the average concentration of solute atoms in the parent phase; c p is the concentration of solute atoms on one side of the precipitated phase; q is the segregation factor of solute on the dislocation line; k is the transport rate constant of solute per unit area; R is the gas constant; T stands for temperature.

Numerical simulation of second phase precipitation behavior under different rolling processes
The hot rolling process of steel plate is usually accompanied by complex evolution of temperature field, stressstrain field and microstructure, and the above three processes influence each other. In order to realize the multifield coupling simulation in this process, it is necessary to complete the setting of relevant parameters and solving conditions in ABAQUS software, so that the multi-field coupling calculation has complete mathematical conditions. The chemical composition of the steel (chemical composition is shown in table 2) is input into the material performance simulation software JMatPro to calculate the physical properties and thermodynamic properties of the material at different temperatures. The high temperature single-pass compression test ( the sample size is Ф10 × 15 mm cylindrical sample ) was completed on the Gleeble-3500 thermal simulation test machine. Firstly, it was heated to 1250°C at a heating rate of 5°C s −1 and held for 120 s, and then cooled to the deformation temperature at a cooling rate of 5°C s −1 . After holding for 20 s before compression, isothermal compression was carried out, and then quenched to room temperature in compressed air. The average cooling rate of quenching was 10.0 ∼19.0°C s −1 . After compression, the reduction of the sample in the length direction is 60%, the deformation temperature is set to 1200, 1100, 1000, 900, 800°C, and the strain rate is 0.1, 0.5, 1, 2, 5, 10, 15 s −1 . The flow stress model affected by recrystallition behavior under high temperature deformation was obtained by fitting the data of Gleeble simulation.The JMatPro calculation result function and the high temperature constitutive relation model are imported into the material library of ABAQUS. The second phase precipitation model is incorporated into the finite element simulation system of hot rolling process through the secondary development subroutine VUSDFLD of ABAQUS. The precipitation behavior model is written with subroutine code in FORTRAN language. The volume fraction of precipitation phase and the size of nucleation are defined as field variables FIELD that change with time, and saved as state variables (SDVs) for the retrieval and analysis of the calculated results in ABAQUS postprocessing. The VUSDFLD subroutine interface of ABAQUS is used to connect to the simulation program to realize the secondary development of simulation calculation of precipitated phase in ABAQUS.
In terms of the rolling process of extra thick plate, the deformation caused by the rolling from the surface to the center in the thickness direction of the plate and whether the microstructure and properties meet the quality  The rectangle represents a section in the steel plate, and the normal of the section is along the width direction of the steel plate. The S side is the steel plate surface, the C side is the steel plate center, and the plane represented by the line L is the symmetry plane. The circle represents the working roll on the reversing mill, with a diameter of 400 mm, which is consistent with the physical simulation experimental data conducted in the laboratory. The heat exchange with the rollers was taken into account by setting contact heat dissipation condition between rollers and steel plate. The heat exchange with environment was taken into account by setting convection heat transfer and radiation heat transfer conditions. The values of environment temperature, heat transfer coefficient, radiation emissivity, absolute zero temperature and Stefan-Boltzmann constant were keyed in the Abaqus software. The steel plate is set as a deformable body, and the two-dimensional temperaturedisplacement coupling element CPE4RT with size of 2.5×2.5 mm is adopted. In the present calculation, the definition of high temperature plastic behavior of material is created by setting the parameters of Johnson-Cook (JC) constitutive model [20,21]. Roller is modeled according to analytical rigid body. The simulated actual working condition is that the plate thickness before rolling is 250 mm, rolling to 120 mm, and rolling in the recrystallization zone for 5 passes on the reversible rolling mill. The schedule of passes for the rolling process is shown in table 3.
In the pretreatment stage, the temperature of steel plate leaving the heating furnace is defined as 1200°C, and various heat transfer conditions and parameters are set. The displacement constraint of the node in the thickness direction is defined on the symmetry plane of the steel plate. In addition, according to the actual process conditions, the necessary parameters such as the time of each rolling pass and gap, the steel plate moving speed, the roll rotating speed, the roll reduction and so on under different processes are defined. Figure 2 presents a contrast graph of the temperature distribution of plates in the thickness direction under UTR and GTR processes before rolling.

Mechanical property test
The steel plates in the rolling state and the heat treatment state were sampled for room temperature tensile test and Charpy pendulum impact test to detect the strength and low temperature toughness of the steel plates under various process conditions. The standard round bar tensile specimens (Diameter 10mm, Gauge length 50 mm) were cut parallel to the rolling direction at the surface, center and 1/4 position of the steel plate, and two were cut at each position and averaged. According to the national standard GB/T 228.3-2021 'Metallic materials -Tensile testing-Part 1: Method of test at room temperature', the tensile test of metal materials at room temperature was carried out on the universal testing machine CMT5105. The cross head speed of the tensile testing machine is 2 mm min −1 .
The longitudinal V-notch impact specimens were cut at the surface, center and 1/4 position of the steel plate. The size of the specimen was 10 mm × 10 mm × 55 mm. Three specimens were cut at each position and the average value was calculated. According to the national standard GB/T229-2020 'Metallic materials-Charpy pendulum impact test method', the test was carried out on ZBC2452-B pendulum impact testing machine. First, the sample was placed in alcohol, and the alcohol temperature was reduced to−25°C (The supercooling temperature compensation value was 5°C) for 10 min, and then the sample was taken out to measure the impact energy.

Metallographic observation
The metallographic samples of 10 × 10 × 10 mm were taken from different positions of the rolled steel plate, and then grinded and polished. A small amount of sodium dodecylbenzene sulfonate was added to the saturated picric acid aqueous solution and a few drops of hydrochloric acid were dripped as corrosion agent. The etchant was heated in a water bath furnace at 70°C ∼ 75°C, and then the sample was etched in the etchant for 40 s ∼ 60 s, washed with water and alcohol and blown dry. The microstructure and grain size were observed under LEXT OLS4100 laser confocal microscope (OM), and the grain size of the original austenite of Q690D steel was measured by the linear intercept point method in the national standard GB/T6394-2017 'Metal-methods for Estimating the Average Grain Size'.In order to ensure the accuracy and reliability of the data, five fields of view were selected for each metallographic sample for statistics, and then the average value was taken.

TEM observation
The metallographic samples were cut from the experimental rolled steel plate and the heat-treated steel plate respectively. After polishing and etching, a layer of carbon film is deposited on the eroded surface with a vacuum evaporation apparatus, and the carbon film is cut into 3 mm square pieces with a sharp knife, and then the sample is placed in 10 ∼12% nitric acid alcohol solution to make the carbon film detach from the attached surface. Take the floating carbon film and put it into anhydrous ethanol and distilled water in turn, and finally use the sample copper mesh as the film carrier to complete the preparation of TEM observation samples. The precipitates were observed by TEM and energy spectrum analysis. Use graphic processing tools to measure and count the quantity, size and distribution of precipitates.TEM counted 20 pictures at each position, and the average value of the statistical data was the size of precipitates.

Numerical simulation results
After completing the simulation analysis, the simulation data of equivalent plastic strain, instantaneous grain size at the end of recrystallization and grain size after growth in different regions of the thickness direction of the steel plate were statistically analyzed, as shown in figure 3. The peak value of the equivalent plastic strain of the steel plate under the UTR process is concentrated in the surface area of the steel plate, while the equivalent plastic strain of the 1/4 and core area of the GTR process is larger than that of the UTR process, indicating that the GTR effectively increases the deformation of the 1/4 and core area of the steel plate. The increase of deformation  promotes the increase of austenite recrystallization percentage, and the grain size of recrystallization after phase transformation is 10 um smaller than that of UTR process. After simulation completed by Abaqus software, the data results were exported by the post-treatment function. The volume fraction of precipitates after five passes of rolling is shown in table 5. The deformation of the GTR steel plate at the center and 1/4 is larger than that of the UTR steel plate, which promotes the occurrence of the second phase deformation induced precipitation, and the precipitation volume fraction is also larger.
Nucleation size of precipitates is shown in figure 4. It can be seen from the figure that the size of precipitates in the UTR steel plate is relatively small only within the thickness of less than 5 mm on the surface, while the critical size of the two types of precipitates at 1/4 of the steel plate is still relatively small under the GTR process,  and the smallest size of the crystal nucleus at the surface is 0.44 nm and 0.96 nm finer than that of the UTR. This is mainly due to the lower deformation temperature at this location during the hot rolling process, and the deformation induced precipitation of the second phase occurs, which makes the size of the crystal nucleus smaller.   Figure 5 shows metallographic images of austenite in different thickness areas during rolling tests using both UTR and GTR rolling processes. Table 6 shows the comparison between experimental measurement results and simulation results of austenite grains. From table 6, it can be seen that the 1/4 thickness and center austenite grain size of the GTR process are 15.5 μm and 10.4 μm lower than those of the UTR process. The GTR process has an excellent effect on refining the grain size in the center area of the steel plate. The prediction error of austenite grain size simulation is within the range of 10%. The microstructures of as-rolled steel plates by GTR and UTR are shown in figure 6. The metallographic composition of the steel plates under the two processes is basically similar, mainly bainite and polygonal ferrite. Image-Pro Plus6.0 software was used to count the microstructure content. The proportion of ferrite in the center of the GTR process was about 7 % −10 %, which was greater than that of the UTR process. The 1/4 thickness of the two rolling processes is all bainite. There are coarse granular bainite and lath bainite at 1/4 thickness of the UTR process. The 1/4 position of the thickness of the GTR process is fine lath bainite and some granular bainite. The ferrite grain size in the center of the plate rolled by GTR is between 12 ∼ 30 μm, while that of the plate rolled by UTR process is 20 ∼ 45 μm.

TEM and OM observation of precipitates
The microstructure of the rolled state has a direct effect on the microstructure after quenching. The grain size and M/A component size obtained by GTR are smaller than those obtained by UTR. After quenching, the width of martensite lath bundles in the center and 1/4 of the GTR steel plate is finer and the orientation is more diverse. After tempering, the center position of the GTR steel plate is mainly uniform and fine tempered sorbite and a very small amount of polygonal ferrite, and the boundary of polygonal ferrite is difficult to distinguish. The distribution of tiny carbide particles in the structure is very uniform, and no coarse carbide or concentrated distribution is found. The microstructure of the UTR steel plate is tempered sorbite and a small part of upper bainite. The tempered sorbite maintains the orientation of martensite generated during quenching. The size of carbides in the structure is generally large, and it shows island or banded segregation distribution, and the uniformity is obviously less than that of the GTR steel plate.
The microstructure at 1/4 position of the GTR steel plate is composed of tempered sorbite and polygonal ferrite. The number of ferrite is significantly increased compared to the center structure, with a small amount of intragranular ferrite appearing. The grain boundaries are clearer and there are more granular and banded hard phases precipitated at the interface, which will have a certain adverse impact on toughness. At this thickness, the microstructure of the UTR steel plate is similar to that of the center, while the degree of segregation of carbides is higher, and the microstructure zoning is obvious.
On the surface of the steel plate, the microstructure of the GTR steel plate is finer than that at 1/4, with more intragranular ferrite generated by small precipitates as nucleation points, and the distribution of elongated carbides at grain boundaries is reduced. The granular carbides in the microstructure of UTR steel plates exhibit strip or cluster like distribution along a certain direction.
The morphology of precipitates at the center, 1/4 and surface of the rolled steel plate is shown in figure 7. Table 6 shows the statistical results of the average size of precipitates and austenite grains. The amount of precipitates in the center of gradient temperature rolled steel plate is more than that of uniform temperature steel plate, and the average size is smaller; The amount of precipitates at 1/4 is less than that at the center, and the average size difference is not significant; However, the proportion of fine precipitates on the surface is higher, mainly because the surface temperature before rolling is about 900°C, and the second phase with deformationinduced precipitation has a smaller critical nucleation radius. At the same time, low temperature also inhibits the growth and aging process of the precipitated phase nucleus. There are many large precipitates in the center of uniform temperature rolled steel plate, while there are many precipitate clusters at 1/4 of the center. According to the determination results of austenite grain in table 6, UTR process has excellent effect on refining the steel plate center and 1/4 thickness grain. Figure 8 shows the precipitates at different thickness positions of the steel plate after quenching treatment. The number and size of precipitates on different rolling processes and different thicknesses are reduced, and the size difference is not as obvious as that of rolled state, indicating that some precipitates are redissolved during quenching heating. The shape of the precipitates in the quenched gradient temperature rolled steel plate is mainly round or oval, while the precipitates in the uniform temperature rolled steel plate are square, rivet and other irregular shapes. Figure 9 is the precipitate picture of steel plate after quenching and tempering at 550°C. The size of precipitates at the center and 1/4 is close, while the size of precipitates at the surface is the smallest, mainly distributed near grain boundaries and dislocations. Most of the precipitates in the differential temperature rolled steel plate are round, while the short rod or large size precipitates with sharp corners appear in the uniform temperature rolled steel plate. Figure 10 is the morphology of precipitates in the steel plate tempered at 630°C. Due to the increase of tempering temperature, a large number of supersaturated carbon atoms in the quenched structure are dissolved, and the amount of precipitation increases. The size of carbides at some positions is larger than that at 550°Ctempering, and the distribution is more uniform. Table 7 shows the statistical results of mechanical properties and average size of precipitates of steel plate after quenching and tempering treatment.
The stress-strain curves for each condition is shown in figure 11. These results show in table 7 and figure 11 that the GTR rolling improves the strength and impact energies after quenching and tempering. After quenching and tempering at 550°C, the yield strength increases on average by 10MPa and the impact energy at −20°C by 30J. Tempering at 630°C, results in even greater improvements in properties on GTR rolling, the strength increasing on average by 30MPa and the impact energy by 50J. This can be related to a refinement in the average particle size for tempering at 550°C from 74 nm after UTR rolling to 61 nm after GTR rolling with again a much bigger change at 630°C from 79 nm with UTR rolling to 64 nm after GTR rolling.
Through quenching treatment, some large precipitates in the rolled steel plate can be remelted, and a large number of dislocations and sub-grain boundaries are produced in the quenched structure, and then carbide precipitates near these positions during tempering. The second precipitate in ferrite has finer size and more spherical morphology than that precipitated in austenite. Nano-sized tiny precipitate particles are evenly distributed in the matrix, which can produce considerable precipitation strengthening effect, and also avoid the toughness damage caused by coarse precipitates with irregular shape.
The particle size distribution of the precipitates in the UTR and GTR samples with different process states is shown in figure 12. This indicates that the precipitate size is mostly relatively small, and the diameter distribution is between 0-200 nm. The frequency of GTR process is higher than that of UTR process in the range of particle size less than 75 nm, especially in the core area, which indicates that with the increase of deformation, the size of precipitates is more concentrated in a smaller scale range.

Discussion
The research points out that [22], the increase of deformation can improve the nucleation rate of precipitates and refine their size. Under the UTR process, the deformation at the center and 1/4 of the steel plate is smaller than that on the surface of the steel plate. The insufficient deformation at these positions leads to less precipitation. And due to the less nucleation positions, the precipitation particles are easy to gather and grow at the grain boundary, with larger size. The extent of the coarse carbide remelting during quenching is very limited, which ultimately leads to the relatively large size and irregular shape after tempering. The precipitation strengthening effect of this kind of center second phase is very small, and it has great harm to the toughness of steel. However, GTR can produce a large amount of plastic deformation near the center of the steel plate, refine the size of precipitates and improve their morphology and distribution, so that the center of the steel plate has good strength and toughness. Because the lattice distortion near the dislocation line is very serious, the distortion energy is high, which is beneficial to the nucleation of the precipitate phase. Moreover, the dislocation provides a high-speed channel for the diffusion of microalloy element atoms and carbon and nitrogen atoms, effectively promoting the nucleation of the precipitate. The nucleation size of the second phase is smaller, and the growth degree is limited. Plastic deformation is the main source of dislocations in the metal, so a greater degree of plastic deformation can obtain a higher volume fraction of precipitated phase, increase the nucleation position of precipitates in the ferrite matrix, make the distribution of precipitates more uniform, and it is not easy to gather and grow up in individual positions to form a large size inclusion crack source, to ensure good precipitation strengthening effect and avoid causing sharp deterioration of toughness.
The results show that [23,24], the basic structure of some complex precipitates containing two or more metal elements does not belong to the composite at the cell level (the different metal elements exist in the cell as replacement atoms and form a replacement solid solution). The formation of some composite precipitates is due to the nucleation and growth of the post-precipitates attached to the pre-formed precipitates. Generally, these  precipitates have large size, and the final morphology is irregular geometry with sharp edges. They can not play the role of precipitation strengthening, but will become an adverse component that seriously endangers the plasticity and toughness of materials. Figure 13 shows the morphology of (Nb/Ti) C composite precipitate. The calibration of its diffraction spot determines that its crystal structure is FCC. Use Image Pro Plus software to calculate the area of rolled carbonitride precipitates in figure 4, and the area ratio of GTR and UTR center precipitates is 0.44% and 0.34% respectively. The morphology of precipitates can be regarded as spherical. When the area fraction is converted into volume fraction, it can be regarded as the ratio of sphere to cylinder [25][26][27], and the volume ratio of precipitates in the center of GTR and UTR states is 0.29% and 0.23% respectively. The volume sum of NbC and VC precipitates in GTR and UTR states in table 5 is 0.058% and 0.031% respectively. The main reasons for the analysis are as follows: First, the precipitates obtained from the experiment include NbC, VC, TiC, MnS, TiN, etc, while the simulated state only calculates NbC and VC precipitates. The second is that the experimental state is the precipitate result obtained by cooling to 500°C after rolling and air cooling to normal temperature. There is precipitation growth phenomenon during cooling, so the volume fraction is larger than the simulation data. In the follow-up study, we will continue to pay attention to the comparative analysis of theoretical and experimental values of precipitation physics, and further revise the key parameters in the theoretical model. Figure 3 which show that GTR rolling results in a refinement of the critical nucleus precipitate size. However, because these precipitates, precipitate out on each other the size is much greater than the predicted and the theoretical calculations need to be modified.

Conclusion
Through numerical simulation analysis and precipitation TEM experiment, the precipitation dynamics behavior, size and distribution characteristics of precipitation under the influence of different hot working processes were studied. The effect of GTR process on the precipitation behavior of microalloyed carbides in the microstructure of extra-thick steel plate was analyzed. The main conclusions are as follows: Figure 11. The stress-strain curves for each condition (a) UTR-Rolling (b)UTR-550°C tempering (c) UTR-630°C tempering (1) The coupling simulation of austenite recrystallization behavior and rolling process was realized by the secondary development of ABAQUS software. The results show that GTR process can refine the grains in the center of the ultra-thick plate by increasing the strain, and the recrystallized grain size is about 10 μm smaller than that of UTR. The actual effect of the GTR process to refine the recrystallized grains in the center of the extra-thick plate was confirmed by the rolling experiment, and the calculation results of the developed microstructure simulation model were also proved to be highly reliable.
(2) According to the calculation results of the coupled finite element model of precipitation kinetics and thermodynamics in the hot rolling process, the GTR process can effectively increase the volume fraction of the precipitates in the steel plate, and the critical nucleation size of the precipitates at the surface and 1/4 is smaller than that under the normal uniform temperature process. The size of precipitates in the UTR steel plate is relatively small only within the thickness of less than 5 mm on the surface, while the critical grain size of the two types of precipitates at 1/4 of the steel plate is still relatively small under the differential temperature process, and the minimum grain size at the surface is 0.44 nm and 0.96 nm finer than that of the UTR. This is mainly due to the low deformation temperature of the position during the hot rolling process, which results in the deformation induced precipitation of the microalloyed carbonitride precipitated phase and the smaller size of the crystal nucleus.
(3) Increasing the amount of deformation can make more microalloyed carbonitride precipitates nucleate near the dislocation in the matrix, and can control the number of complex precipitates with coarse size and poor morphology, and reduce the effect of toughness deterioration caused by the precipitates with poor morphology.
(4) The amount of precipitates in the rolled steel plate is small, and some precipitates are very coarse and have large size fluctuations; The quantity of precipitates after quenching is less, and the size is also reduced; After tempering, the precipitation increases obviously and the size decreases further. The GTR process increases the deformation near the center of the steel plate, and the high-density dislocation generated increases the nucleation position of the microalloyed carbonitride precipitates in the matrix, which is conducive to the nucleation of the new precipitate and promotes the dissolution of large precipitated particles. Finally, after tempering, the precipitate particles are dispersed and randomly distributed on the ferrite matrix, and the size is refined by 20 ∼ 35 nm, resulting in obvious precipitation strengthening and microstructure refining effect.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).