Effect of process parameters on the microstructure and mechanical properties of bars from Al-Cu-Mg alloy processed by multipass radial-shear rolling

The study of microstructure and mechanical properties formation of A2024 alloy obtained by the multipass radial-shear rolling (RSR) method is discussed in this article. FEM simulation was carried out that made it possible to evaluate the influence degree of rolling temperature–velocity parameters on the strain state of material. It has been found the increase in rotary velocity of rolls significantly influences on the deformation heating of bar after RSR (predominantly in its surface layer). The combination of rolling temperature–velocity conditions at selection of deformation regime has complex effect on structure and properties formation. The analysis of sizes and distribution of phase particles has shown that the rolling at lower temperatures allowed to increase the mechanical strength due to the more intensive refinement of undissolved Fe-containing phase. The gradual decrease in the rolling temperature in each pass makes possible to achieve the high strength (UTS ~ 430 MPa and YS ~ 255 MPa) while maintaining the ductility level ~ 15%, that are comparable to ones obtained at some severe plastic deformation (SPD) methods.


Introduction
The Al-Cu-Mg system alloys are basic materials for aerospace and automobile industries [1]. The alloy A2024 is a constructional material widely used for hulls and structural elements in mechanical engineering, aeronautics, and space where the minimum weight of structure plays a fundamental role [2,3]. The semi-finished products in shape of bars are produced by method of hot extrusion mainly. The alloy is hardened by the quenching and subsequent natural or artificial aging [4].
During deformation in industrial conditions, the materials are exposed to the complex combination of strain, stress, and temperature changes. The understanding of metals and alloys deformation behavior during hot working is of a great importance, first of all, from the point of view of product properties formation. To study machinability and to establish optimal hot deformation parameters for certain metals and alloys, a number of research groups have made efforts to carry out thermomechanical experiments (compression, tensile and torsion tests) at wide temperatures and strain rates [5][6][7][8]. At the same time, the processes of metal forming in industrial conditions are different from the ones in laboratory. In particular, during the radial-shear rolling (RSR) the zonal temperature change and strain rate in wide range (from 0.1 to 100 s -1 ) occur in depending on given rotary velocity of work rolls [9]. This fact indicates the need for research to find optimal deformation regime under conditions close to industrial ones.
Based on thermomechanical tests of samples for torsion, compression and tension, the different models of microstructure evolution during recrystallization have been constructed. The most famous models are McQueen [10], Hallberg [11,12], Gourdet and Montheillet [13,14], Sellars [15]. The most comprehensive review of dynamic recrystallization process in metallic materials was made by K. Huang and R.E. Logé [16]. One of the conclusions in this work is that there is a need to study the dynamic recrystallization mechanism under deformation conditions close to industrial ones since the microstructure evolution under these conditions is a more complex process than under constant ones. In addition, it depends on deformation history and the initial state of the metal because industrial processes most often have several stages of deformation and heat treatment. In [17], the author made attempt to describe and to control the microstructure at dynamic recrystallization of steels during hot rolling. Based on the experimental data and revealed dependencies, the processing maps were made which can be used to select the industrial rolling regimes and to build the rolling schedule. However, the results obtained also require confirmation by experimental studies in real conditions.
The radial-shear rolling method for obtaining the long bars with gradient functional structure in different materials is one of preferable due to its implementation simplicity and technological flexibility [18]. The equipment of RSR mills makes possible to obtain a wide range of product sizes and can provide different deformation degrees per pass [19]. The principle of this method and main distinctive parameters are described in detail earlier in publications of many authors [20][21][22].
The main purpose of this work is to determine the rational deformation regimes of aluminum alloy of system Al-Cu-Mg using RSR method from point of view of influence on microstructure and mechanical properties. To achieve this purpose, it is proposed to analyze the temperature conditions and deformation parameters at different rolling regimes using finite element method (FEM), as well as a study in laboratory conditions using industrial regimes.

Experimental procedure
Laboratory studies were carried out using radialshear rolling mills ''MISIS-100 T'' and ''14-40.'' These mills have the feed angle b and the toe angle d equal to 20°and 10°, respectively. Work rolls are arranged symmetrically relative to rolling axis with a step 120°. Each work roll has the entrance angle on section of grip and reduction and the exit angle on calibration section. The special design of rolls and their geometrical arrangement ensure the movement of the workpiece along the helicoidally trajectory and, at the same time, its reduction along the diameter (Fig. 1).
The cylindrical bar from industry alloy A2024 was used as an initial workpiece for rolling. Its chemical composition is presented in Table 1.
The diameter and length of initial workpiece were 60 mm and 130 mm, respectively. The initial workpieces were heated in batch-type furnace to rolling temperature T 0 for 2 h before deformation. The rolls were not additionally heated before deformation.
Temperature-velocity parameters of rolling are presented in Table 2.
In regimes 1, 2, and 4, the rolling was carried out to a final diameter of 14 mm in 5 passes (Ø60-42-31-24-17-14 mm) with varying the rolling temperature and rotation velocity of work rolls. The elongation ratio per pass l i is from 1.5 to 2.0 that makes it possible to avoid the significant deformation heating and to obtain the bar with an accurate geometric shape. The rolling with reduction along diameter of 2 mm per pass for regime 3 was carried out to determine the influence of large total number of deformation cycles on structure formation. After each pass, the bar was returned to the furnace to equalize the temperature across the section.
After rolling, the part of samples was subjected to heat treatment (HT) according to regime shown in the diagram (Fig. 2).

Microstructure, phase composition, and properties analysis
The microstructure was examined by means of scanning electron microscopy (SEM, TESCAN VEGA 3), transmission electron microscopy (TEM, JEM-2100), and electron microprobe analysis (EMPA, OXFORD AZtec). The metallographic samples were ground with SiC abrasive paper, polished with 1 lm diamond suspension and etched with a 1% hydrogen fluoride (HF) water solution. The thin foils for TEM were prepared by ion thinning with a PIPS (Precision Ion Polishing System, Gatan) machine and studied at an accelerating voltage of 160 kV.
The EBSD analysis of the alloys was carried out using a TESCAN VEGA LMH microscope with a LaB6 cathode and the NordlysMax2 detector (Oxford Instruments Advanced AZtecEnergy). MAD coefficient was less than 0, 3.
The phase composition of the alloy was assessed using Thermo-Calc software (TTAL6 database [23]).
The analysis of size and shape of secondary phases particles was carried out using the ImageJ (National Institute of Mental Health) program for images with a field size of 420 9 420 lm at a magnification of 9 500. To evaluate the shape of phases particles, the shape parameter C (circularity) was calculated, and it is determined by equation: where A is the area, mm 2 ; P is the perimeter, mm. Circularity with a value of 1 indicates a perfect circle. As the value approaches 0, it indicates an increasingly elongated shape.
The cross-sectional microhardness of as-processed bar samples was measured by the Vickers method (HV) on a DUROLINE MH-6 installation (load 300 g, dwell time of 30 s) with the step of 0.5 mm from surface to center direction.
Room-temperature tension tests were conducted for as-processed bar samples with final diameter of

FEM simulation
The simulation of RSR process was performed using QFORM 3D (QuantorForm Ltd., Russia) program. For this, the solid 3D model consisting of rolls and workpieces was imported. The sizes and shapes of work rolls and workpieces correspond to the real ones at running of experiment. Temperature-velocity parameters of simulation are presented in Table 3. The regimes 1, 2, and 5 correspond to the experimental ones of rolling. The regimes 3 and 4 were selected for a more comprehensive analysis of effect of heating temperature on deformation parameters.
After completion of calculation of each pass, the bar temperature was set uniform along section again that corresponds to conditions of real experiment. Other workpiece parameters were saved and imported to the next operation.
The calculation was carried out considering the heat exchange between the workpiece and tool and the environment. The main parameters were set for model: rolls temperature-25°C; ambient temperature-25°C; coefficient of heat transfer between the material and the tool -30,000 W/m 2 ÁK; coefficient of heat transfer between the material and the air -30 W/m 2 ÁK; friction factor-0.93; rolls material-40Cr steel. The chemical composition, physical, mechanical, and rheological properties of A2024 commercial aluminum alloy can be found in QFORM database. The rheological properties of material are given as a table function for temperatures 250, 330, 400, 470, and 550°C, strain rates 0.01, 0.1, 1.0, 10, 50, 200 s -1 , and deformation degree from 0.01 to 0.99.
According to the calculation results, the parameters of the equivalent deformation e over the cross section of bar, the temperature distribution in deformation zone and after rolling were analyzed.
To determine the equivalent strain, the numerical integration of strain rate intensity for each mesh node of workpiece among the movement trajectory of element particle of deformed metal is used in QFORM program: where Dt n is the time step of calculation; n is number of time step of calculation.

Temperature analysis
At different velocity and temperatures of rolling, the temperature changes of bar immediately in deformation zone is of interest to study. As shown in [9], these changes can lead to different combination of structure over bar cross section during dynamic recrystallization. In this case, the temperature was controlled at three points: P1-bar center; P2-half of bar radius; P3-bar surface. The graphs of Figure 2 Heat treatment diagram. temperature changes at points P1-P3 obtained by tracing of points along their movement trajectory are presented in Fig. 3. It makes possible to see the dynamic of temperature changes in zones typical for RSR.
As can be seen, the greatest temperature fluctuations (the change reaches 20-50°C per deformation cycle) occur in the surface layer of bar where strains and stresses are localized. Since the deformation zone is open, the sharp change in temperature in contact zone occurs with each single reduction by roll. In the moment of point P3 passing between work rolls (noncontact area of deformation), the surface temperature can sharply change again.
At maximum heating temperature T 0 = 450°C, the surface temperature in the deformation zone decreases, while the temperature at points P1 and P2 smoothly increases in the reduction zone. At decrease in heating temperature of workpiece to 350-400°C, the tendency to decrease in surface temperature of bar reduces, and the temperature at point varies in temperature range of P1 and P2. (Fig. 3b and c). In the calibration zone of bar in all three cases, the temperature fluctuations of surface reduce since the reductions are insignificant, and the temperature of inner layers almost does not change.
The rolling velocity also significantly impacts on the changes in temperature field of workpiece in the deformation zone. At the rotary velocity of 90 rpm, the surface temperature in reduction zone intensively increases, and the graph is stepped (Fig. 3d). Since the time of one cycle decreases, the temperature in the contact zone with colder roll has no time to change and remains constant. It should be noticed in this case that the temperature gradient over section of bar DT sharply increases with the growth of rolling velocity. Difference between temperatures at points P1 and P3 is more 100°C, and it can additionally effect on structure formation processes.
Since the rolling in an industry is mostly done in several passes, it is effectually to analyze the temperature change after each pass. The graphs of temperature change of bar after each pass DT relatively heating temperature before rolling T 0 for five simulated regimes are presented in Fig. 4. The change in temperature DT was determined as the difference between T 0 and average temperature at points P1-P3 on the exit of deformation zone.
For regimes 2, 3, and 4 with constant temperature before each pass and the same rolling velocity, it should be noticed that DT decreases with reducing of bar diameter. For higher heating temperatures of 400 and 450°C, the temperature increase is 50-70°C in the first pass, and there is no increase in the last one. For regime 1 with gradual decrease in temperature, the deformation heating with each subsequent pass, on the contrary, increases. For the first pass, it is minimal and equals 25°C, and in the last pass it reaches approximately 100°C.
At the heating temperature 300°C and the increase in rotary velocity to 90 rpm, the temperature increase after RSR is the largest and it is 90-125°C, for first four passes, and it decreases to 60°C in last pass.
Based on data obtained, the following general conclusions can be made: • at the same heating temperature before rolling T 0 , the deformation heating of bars decreases with reduction in bar diameter; • reducing the heating temperature by 50°C leads to a decrease in deformation heating by approximately the same amount; • increase in the rotary velocity of rolls has a significant effect on the deformation heating of bar after RSR (mainly in surface layer of bar).
As can be seen, the combination of rolling temperature-velocity conditions at selection of deformation regime has complex effect that cannot be easily predicted or calculated analytically. Since the temperature is one of the main parameters effected on structure and properties formation, the simulation can be useful tool for preliminary analysis and forecasting of product properties formation.

Deformation parameters analysis
The equivalent strain parameter can be used for the complex characteristic of technological process conditions influence on material properties. During the severe plastic deformation processes, the large values of equivalent strain (4-6 or more) are achieved with relatively small changes in the overall dimensions of workpiece. It makes possible to obtain the materials with ultrafine-grain or nanostructure and significantly increase the strength properties. [24,25].
During RSR process, the field of equivalent strain is formed with pronounced gradient of distribution over section of rolled product. The equivalent strain e reaches the maximum values in peripheral zone of workpiece where the greatest shear strains occur locally. The values of e are minimal in the central zone and, mainly, determined by reducing of crosssectional area of bar.
The temperature and rotary velocity of rolls significantly effect on the gradient of equivalent strain field. With decrease in the initial temperature and increase in the rotary velocity of rolls, the growth of maximum values of e at insignificant change in level of minimum values in the center is observed. In these dependencies, the influence of rheological component on the strain state of workpiece is appeared. As known, the deformation heating is due to dissipation of plastic deformation power which is proportional to flow stress and strain rate. Disregarding some conventions, it can be argued that the following self-activating scheme is implemented. The initial nonuniform distribution of e creates nonuniform deformation heating. In the most heated near surface layers, the resistance to deformation of metal (flow stress) decreases, which contributes to an even greater localization of the accumulated strain on these layers with an increase in its maximum values. This, in turn, further enhances the deformation-temperature gradient. It is obvious that when implementing such scheme, the force of metal on roll decreases due to a decrease in the level of contact stresses during local heating, which is confirmed by the simulation data (Fig. 5). For all studied regimes, the growth of maximum values of e and the increase in the gradient with the decrease in the rolling force are observed.
For a comparative analysis, the development of severe plastic deformation area (SPD area) expressed in relative fraction of bar cross-sectional area with the value e C 8 was determined for each regime (Fig. 6).
The greatest values of SPD area were recorded for regime 1 with gradual decrease in rolling temperature (83.86%) and regime 5 (82.30%). The decrease in deformation temperature from 450 to 400°C leads to increase to SPD area by 4.5%, while the decrease in temperature to 350°C increases this value almost by 14%. In the central zone of bar occupied approximately 1/3 of the radius, where the effect of strain rate intensity is minimal [26,27], the equivalent strain parameter is mainly determined by elongation in a given section (by reducing its area).
From the data obtained, it can be assumed that regimes 1 and 5 can have the greatest influence on properties and microstructure. The SPD area defined in this case by parameter of equivalent deformation (e C 8) is localized in bar surface layer and depends not only on the magnitude of shear strain but also on the temperature-velocity regimes of processing.

Microstructure characterization and phase analysis
The microstructures of the as-processed (RSR) and heat-treated (RSR ? HT) bars obtained at different regimes (see Table 2) are presented in Fig. 7. Table 4 represents the quantitative data on the number, shape, and size of particles observed in the microstructures. For ease of reading, the latter data are also presented as histograms in Fig. 8. As can be seen that heat treatment leads to substantial decrease in the number density of particles observed in the asprocessed structures. It may be deduced that the decrease in the number density after heat treatment is associated with dissolution of the fine Cu and Mg precipitates (represented by the Al 2 Cu and Al 2 CuMg phases) formed during the cooling of bars from the RSR temperature. As can be seen from Table 4 and Fig. 8, the fraction of the precipitates depends on the RSR temperature. Indeed, the higher fraction of precipitates can be observed for the samples obtained at low RSR temperature (regimes 1 and 4). The latter is due to the natural decrease in the Cu and Mg solubility with the temperature decrease. The precipitation structure formed during processing was also observed by using TEM (Fig. 9). The analysis was performed for the bar obtained at regime 1. The distinguish feature of this regime is continuous decrease in the deformation temperature down to 250°C leading to the formation of some number of precipitates during RSR, while for the higher temperature regimes, most of Cu and Mg should be dissolved in the aluminum matrix. The analysis reviled the quit uniform distribution of precipitates ( Fig. 9a) with relatively wide range of size (possibly depends on the actual temperature of formation). Some of the precipitates can be observed on the subgrains boundaries pinning them and thus additionally inhibit recrystallization (Fig. 9a). Moreover, around of the coarser particles one can also observe the formation of the intensely deformed zones with high number density of dislocations (Fig. 8b) indicating the localization of deformation. As it is known [28], the localization of deformation in a such manner is beneficial for the more intensive strain hardening.
Other particles also observed in the structures (especially well resolved after heat treatment) are relatively coarse and belong to the insoluble Fe-containing phase. According to the thermodynamic calculation made using Thermo-Calc program, this Fecontaining phase should belong to the a(AlFeSiMn) compound [29]. For the central section of the obtained bars, it can be also observed the presence of the fine spherical and elongated voids accompanied the insoluble a(AlFeSiMn) phase particles. The origin of these voids is mainly associated with the fracturing of the insoluble particles at RSR due to the acting in some extend the tensile stresses in the central part of the bar (near its longitudinal axis). However, as it is shown below, the integral influence of these voids on the mechanical properties is insignificant.
The quantitative analysis (Table 4 and Fig. 8) well illustrates that after heat treatment the fraction of particles decreases to * 2% area for all samples. These particles correspond to the undissolved Fecontaining phase, the fraction of which depends only on the concentration of Fe in the alloy. The linear size of these particles varies in the range of 1.5-4.0 lm and strongly influenced by the RSR regime. Indeed, for the regimes 1 and 4 characterized by the lowest processing temperature range from 250-350°C, the size of particles is smallest. For the bars processed at 450°C, the average linear size of the Fe-containing particles increases about two times. The obtained data on the influence of processing temperature on the particle liner size are in good agreement with the numerical simulation data on the influence of processing temperature on the equivalent strain (Fig. 5).   The obtained data suggest development of much higher SPD area for the bars processed at low temperature regimes 1 and 4. The latter should lead to a more intensive refinement of the structure. It should be noted that the size of particles near the surface is noticeably less for most regimes except 4. The observed difference in particle size over the cross section of workpiece is due to the natural gradient distribution of strains and stresses at RSR. According to the previous studies [30,31], higher strains and stresses are observed near the surface of workpiece. For the regime 4, the about equivalent size of the  particles over the cross section of workpiece can be observed. The latter result is due to the increased fraction of SPD area coupled with the highest equivalent strain which allows to process the workpiece more intensively through entire cross section. In addition, from Fig. 8d it can be also observed that the morphology of all particles slightly depends on the processing regimes and most particles strongly tend to acquire a spherical morphology. It should be noted that the size and shape of undissolved Fecontaining particles strongly influence on the alloy mechanical properties (especially for ductility), and fine structure coupled with the spherical morphology of the particles is most desirable one since it minimizes the detrimental influence of Fe-containing phase on the mechanical properties. The grain structure of the 14 mm bar obtained at regime 1 was analyzed using EBSD in the near surface area and central area after both processing and processing and followed by solid solution treatment (annealing at * 495°C for 1 h before quenching). Color reveals the crystallographic orientation in the unit triangle of the standard stereographic projection. EBSD grain orientation maps of the longitudinal structure of the 14 mm bars are shown in Fig. 10. In the EBSD micrographs, black lines denote high angle boundaries (HAGBs) with misorientations greater than 15°, and gray lines are used to identify low angle boundaries (LAGBs) with misorientations between 2°a nd 15°. As can be seen from Fig. 10a, the structure in the near surface area is mainly deformed and characterized by the formation of fibrous grains elongated in the rolling direction. It should be noted that according to the numerical simulation the equivalent strain in this area reaches more than 20 (Fig. 5) which should be enough to trigger the dynamic recrystallization. However, the continuous decrease in the deformation temperature down to 250°C coupled with pinning of dislocations and sub-grains boundaries by the precipitates (Fig. 9) obviously hinders recrystallization and formation of extensive net of LAGB can be instead observed. According to Fig. 4, the deformation heating during the last passes should be more than 100°C. However, due to the relatively small radius of the final bar and the contact with the cold rolls, most of this heat is dissipated and the resulting temperature is insufficient to activate recrystallization. Taking into account the size of subgrains formed, the average grain size in this area is about 4 lm. Due to the increased strain energy stored during deformation, the annealing at 495°C for 1 h leads to intensive development of recrystallization (Fig. 10, b) result in the substantial increase in the grain size and reduction in LAGB.
Similar studies for the central parts of the same bar revealed that after RSR the partially recrystallized structure is formed (Fig. 10c). According to Fig. 6, the maximum equivalent strain in the central part reaches 6-8, which is more than two times less than for the surface. However, the increase in the temperature due the deformation heating cannot be prevented in the same way as for the near surface area. The latter fact leads to the heating of this area up to the recrystallization initiation temperature. Taking into account the size of grains and sub-grains formed, the average grain size in this area is about 6 lm. It should be noted that the increase in the temperature also leads to a partial dissolution of the precipitates which, as it shown above, can act as pinning agents. By comparison the fraction of particles (''Area %'' Table 4) for the center and surface part for the as-processed rod, it can be seen * 10 and 12%, respectively, which is defined by the temperature difference in these areas. The solid solution treatment at 495°C for 1 h also leads to intensive development of recovery processes leading to formation of recrystallized structure (Fig. 10d).
Thus, from the presented analysis, it can be observed that the deformation with a gradual decrease in temperature to 250°C leads to formation of the intensely deformed and partly recrystallized grain structure with an extensive net of LAGB. On the contrary, solid solution heat treatment activates recovery processes including recrystallization result in the formation of much coarse grain structure.

Mechanical properties
The graphs of microhardness distribution in cross section of 14 mm bars obtained at different RSR regimes are presented in Fig. 11. The analysis showed the microhardness of as-processed bars is higher for lower processing temperatures (regimes 1 and 4). The latter is due to the more intensive strain hardening. An increase in the deformation temperature up to 450°C leads to a gradual decrease in hardness (Fig. 11a) due to the development of recovery processes. The distribution of microhardness in cross section of bars obtained at constant initial processing temperature (regimes 2, 3 and 4) is quit uniform. In case of bars obtained at 450°C, the increased processing temperature neutralizes the difference in equivalent strain between surface and center due to the intensive development of recovery processes toward the surface. For the bar obtained at low temperature 300°C and highest rotary velocity 90 rpm, numerical simulation suggests the highest difference in equivalent strain between surface and center (28 vs 8). However, the deformation heating, especially for near-surface regions, should also be maximum for this regime (Fig. 3, d). Indeed, the temperature in the near surface area can reach 470°C which is even higher than that for the bars obtained at regimes 2 and 3. Thus, the high temperature similarly compensates for the high difference between the equivalent strain over the section of bar. Slightly higher hardness in the center of bar obtained at regime 1 can be explained by the higher temperature (due to the deformation heating at last passes) result in the increased solubility of Mg and Cu in aluminum solid solution.
The bars in the as-processed state were subjected to uniaxial tensile tests (Fig. 11b). It is found out that the increase in the processing temperature provides low strength of the as-processed bar. For the regimes 2 and 3 with the highest processing temperature, UTS is less than 300 MPa and YS is less than 180 MPa. The relative low strength of these bars is due to the high processing temperature leading to the intensive development of recovery processes in the microstructure. On the contrary, a gradual decrease in the processing temperature to 250°C (regime 1) led to significant strengthening. As shown above, the RSR processing at regime 1 leads to formation of the intensely deformed (for a significant part of the bar) grain structure with an extensive net of LAGB. Moreover, it can be also observed the presence of relatively fine precipitates (Fig. 9) which form during processing and as discussed above, effect on the dislocation structure formed. Result in this structure the strength properties of the as-processed bar obtained at regime 1 UTS * 430 MPa and YS * 255 MPa are close to ones usually obtained for this alloy after a full cycle of heat treatment, including quenching and aging. In addition, the analysis revealed that the ductility for all samples is not less than 10%. However, the bars obtained at low temperature regime 1 with highest strength properties also have a highest ductility up to 16%. The latter result is due to the superposition of some factors. The first one is more intensive refinement of undissolved Fe-containing phase at low processing temperature. As follows from Fig. 4, the average size of these particles 1.21-2.53 lm (depends on the area) is lowest compared to high temperature regimes 2 and 3 (2.33-4.01 lm). And the second one factor is the observed formation of the intensely deformed zones around of the particles. In case of uniform distribution of these particles, the localization of deformation around them should activate new arias providing deformation thus inhibit macro-localization of deformation. Due to the latter fact, it can be expected prolonged deformation leading to higher strain hardening. Inspecting the sample after tensile testing, we can see that most of cross section was involved in the deformation (Fig. 12a), and the fracture surface has a fine dimple structure characteristic for ductile fracture (Fig. 12b). On the bottom of the dimples, it can be observed the presence of the fractured particles of the Fe-containing particles. It follows that the average hardness of the bar obtained at regime 4 is about equal to that for the bar obtained at regime 1 and the average size of the undissolved Fe-containing phase particles is even less (* 1.5 lm). Despite the latter facts, the mechanical properties both strength and ductility of the bar obtained at regime 4 are much lower. One of possible reason of the obtained result can be associated with the peculiarity of deformation and fracturing of the bar. Although the fracture surface also has a characteristic fine dimple structure (Fig. 12d), the macro-analysis (Fig. 12, b) revealed a tendency to localization of deformation in the central region. In addition, it seems reasonable to assume that extremely high rotary velocity (90 rpm) together with the relatively low temperature (300°C) leads to the micro-fracturing especially in the middle and near surface areas resulting in a weakening of the cross section of the bar.
Thus, after processing the bar obtained at regime 1 has simultaneously highest strength and ductility which are comparable to ones obtained at different SPD methods [32,33]. However, unlike the latter, RSR method used to obtain the bars can be applied in mass industrial production.

Conclusion
The analysis of temperature-velocity regimes of deformation for the A2024 aluminum alloy obtained by RSR method was discussed in this article.
1. All other conditions being equal, the decrease in heating temperature before rolling promotes to reducing of surface temperature fluctuations in deformation zone and to a more intensive deformation heating of bar during deformation. The increase in rotary velocity of rolls leads to a sharp growth of temperature gradient over bar cross section. 2. The analysis of changes in temperature after RSR for different regimes shown that the deformation heating of bar decreased at equal heating temperature before rolling with a decrease in bar diameter. The decrease in temperature by 50°C leads to the reduction in deformation heating approximately by the same value. The increase in rotary velocity of rolls significantly influences on the deformation heating of bar after RSR (predominantly in surface layer of bar). 3. Analysis of the microstructure of bars obtained at different regimes revealed that the linear size of the excess phases (primarily Fe containing) varies in the range of 1.5-4.0 lm and strongly influenced by the RSR regime. The size of particles near the surface is noticeably less; it is due to the natural gradient distribution of strain and stresses across the cross section of bars obtained at RSR. Most particles strongly tend to acquire a spherical morphology. The size and shape of the undissolved Fe-containing particles strongly influence on the alloy mechanical properties (especially for ductility), and fine structure coupled with the spherical morphology of particles obtained at RSR is most desirable one since it minimizes the detrimental influence of Fe-containing phase on the mechanical properties. 4. The grain structure of the 14 mm bar obtained at regime with a gradual decrease in the deformation temperature from 450 to 250°C was analyzed using EBSD in the near surface area and central part after both processing and processing and followed by heat treatment. Analysis showed that the deformation with a gradual decrease in temperature to 250°C leads to formation of the intensely deformed and partly recrystallized grain structure with an extensive net of LAGB. On the contrary, solid solution heat treatment activates recovery processes including recrystallization result in the formation of much coarse grain structure. 5. The bars in the as-processed state were subjected to uniaxial tensile tests. The analysis revealed that an increase in the processing temperature to 450°C provides lower strength (UTS * 285 MPa and YS * 178 MPa) and ductility (* 11%). On Figure 12 Fracture surfaces after tensile testing for the bars obtained at regime 1 (a, c) and regime 4 (b, d).
the contrary, a gradual decrease in the processing temperature to 250°C led to significant strengthening. The properties of bar processed under these conditions are as follows: UTS * 430 MPa and YS * 255 MPa and El * 15% and close to ones usually obtained for this alloy after a full cycle of heat treatment, including quenching and aging. The mechanical properties obtained are comparable to ones obtained at different SPD methods.