Effect of Post‐Swaging Annealing on Microstructure, Texture, and Mechanical Properties of Metastable Austenitic Stainless Steel

1. Introduction

The Metastable austenitic stainless steels (MASSs) exhibit attractive corrosion resistance, ductility, and impact toughness at both room and subzero temperatures [1,2,3,4]. Therefore, MASSs are widely used as structural materials in petroleum, chemical, and nuclear industries. However, their low strength limits potential applications for load-bearing elements. Conventional techniques of strengthening result in a substantial decrease in impact toughness due to the trade-off between strength and impact toughness [5,6]. The phenomenon is associated with the hindrance of energy dissipation at the crack tip through plastic strain during fracture in strengthened materials [7,8]. Indeed, cold working processes exhaust strain-hardening capacity by multiplication and accumulation of lattice defects such as dislocations, low- and high-angle grain boundaries, and mechanical twins. These defects decrease the energy consumption of failure under impact loading conditions [9].

In order to cope with the strength-to-impact toughness trade-off in artificial inorganic materials, several nature-inspired strategies of microstructure design have been suggested, which involve multiple toughening mechanisms [10,11,12,13]. The mechanisms operate at various structural scales during impact loading. Extrinsic toughening is associated with suppression of crack opening via bridging and fracture of fibers, whereas intrinsic toughening acts in an elastic/plastic zone ahead of the crack tip via crack deflection and longitudinal crack formation increasing energy dissipation during crack propagation. In particular, the gradient-structured materials fulfill the mechanisms [12,14,15,16,17]. According to Lin et al. [12], in gradient-grained pure Ni, a substantial increase in impact toughness occurs when a crack transits from a coarse-grain (CG) region to a nano-grain (NG) one. This phenomenon is attributed to ductile fracture via microvoid coalescence in the CG region, while microstructure coarsening in the NG region is observed during Charpy impact testing, also leading to ductile fracture. Huang et al. [14] reported that an AISI 321 MASS with a large-scale gradient structure possesses improved impact toughness due to increased crack initiation and propagation energy. During crack initiation, CG regions exhibit increased energy consumption due to the development of strain-induced martensitic transformation (SIMT), resulting in a high work-hardening capacity [18,19]. On the other hand, fine-grained regions suppress SIMT and hinder the crack growth because strain-induced martensite (SIM) is supposed to be a brittle phase [20,21].

Additionally, a texture substantially affects mechanical behavior [22,23,24]. So, Charpy V-notched impact specimens of an AISI 316L MASS with a uniform 〈100〉-fiber texture of austenite along its length show decreased fracture energy due to exaction of SIMT. The 〈110〉-textured material exhibits high energy adsorption via the development of SIMT [22]. Xu et al. [23] found that extraordinary high impact toughness can be achieved in a low carbon micro-alloyed steel at cryogenic temperatures, when a gradient structure and texture cause shielding and blunting of a crack. Thereby, tailoring of SIMT behavior, morphology of SIMT products (α’- and ε-martensite) and texture patterns can be proposed using gradient structuring techniques to reach the superior strength-impact toughness combinations in MASSs.

However, time-consuming processes and small-scale gradient conditions of workpieces impede potential applications of gradient-structured materials [14,25]. Meanwhile, several cost-effective industrial techniques can be used to manufacture structural materials with a large-scale gradient structure [26,27]. For instance, rotary swaging (RS) at ambient temperature develops a gradient structure [28,29,30]. During the RS processing, a gradient distribution of α’-martensite and a multicomponent gradient texture along the bar radius can be formed in MASSs, resulting in outstanding mechanical properties.

The present paper differs from previous studies by focusing on the effect of post-swaging annealing conditions on the mechanical behavior, microstructural response and fracture mechanisms of MASSs under impact loading conditions. Apparently, annealing temperature significantly affects a microstructure and tensile properties, but how the gradient structure contributes to Charpy impact behavior of MASSs has not been examined. Furthermore, the comprehensive characterization of as-swaged MASSs using orientation distribution functions (ODFs) has not been carried out to date yet. Thus, this paper aims to study the effect of post-swaging annealing on the microstructure/texture evolution and Charpy impact behavior of the AISI 321 MASS.

2. Material and Methods

2.1. Materials Processing

The chemical composition of the commercial-grade AISI 321 MASS used in this study is presented in Table 1. Ingots of the selected material were subjected to RS in a temperature range of 900-1200 °C to obtain a bar shape with 38.5 mm in diameter, which was annealed at 1050°С for 2 hours. The microstructure of the as-received bar consisted of equiaxial grains of austenite with an average diameter of ~10 μm and 2% of fine-grained δ-ferrite. The uniform microstructure and weak texture were observed throughout the transversal section of the as-received bar. Then, the as-received bar underwent RS with external water cooling at ambient temperature according to earlier suggested regimes [31,32]: a feed rate of 180 mm/min; a strike frequency of 1000 bpm; and a workpiece rotation speed of 25 rpm. The following sequence of diameter reductions during RS was carried out: from 38.5 mm to 34 mm; from 34 mm to 30 mm; from 30 mm to 24 mm; from 24 mm to 15 mm; from 15 mm to 11.5 mm, which corresponded to a swaging reduction of 22% (RS22), 39% (RS39), 61% (RS61), 85% (RS85), and 91% (RS91), respectively. After RS91, annealing was conducted at 500 °C (A500), 550 °C (A550), 600 °C (A600), 650 °C (A650), 700 °C (A700), and 800 °C (A800) for 2 hours followed by air cooling.

2.2. Experimental Procedures

For microstructure and texture characterization, templets with 0.3 mm in thickness were cut using an electric discharge machine (EDM) in the transversal section of the as-processed bar. Then, the templets were ground using sandpaper and polished using a colloidal silica to 0.1 mm in thickness. Finally, discs with 3 mm in diameter were punched from the core, half the radius, and subsurface layers of the bar templets and polished in an electrolyte (5% perchloric acid, 35% butanol, and 60% methanol) using TenuPol-5. The microstructure was characterized using a JEOL JEM-2100 transmission electron microscope at an accelerating voltage of 200 kV. The EBSD-analysis was conducted using a FEI Nova NanoSEM 450 scanning electron microscope equipped with a Hikari EBSD camera at an accelerating voltage of 20 kV. All points with the confidence index (CI) below 0.1 were excluded in order to improve the quality of the texture characterization. To estimate the volume fraction of textured gains, the tolerance angle was set to 15°. The volume fraction of the ferromagnetic phases (δ-ferrite/α’-martensite) along the bar diameter was evaluated at nine points with a step of 0.25×r/R (where r—radial distance; R—bar radius) using a FERITSCOPE FMP30 tester, which was initially calibrated using reference samples. At least three measurements were carried out for each point and then averaged.

Charpy impact tests were carried out on V-notched specimens with dimensions of 5×10×55 mm³ (Figure 1a) using an Instron 450 J pendulum impact tester. The specimens were cut along the bar axis using an EDM, as shown in Figure 1b. The V-notched specimens were tested at room temperature. Macro- and microscopic fractography studies were carried out using a FEI Quanta 600 scanning electron microscope at an accelerating voltage of 30 kV. Post-mortem EBSD analysis was conducted under fracture surfaces in the longitudinal section of tested V-notched specimens (Supplementary Materials, Figure 1). To estimate the general yield stress (σ_gy) and maximum stress (σ_m) perpendicular to the minimum transversal section of a V-notched specimen, the following formula was applied [33]:

$$\sigma ={\displaystyle \frac{\beta ·L·P}{2{·C}_{ƒ}·{(W-a)}^2·B}},$$

(1)

where L—distance between supports (40 mm); P– the load, corresponding to the finish of an elastic stage (P_el for σ_gy) or maximum in impact load-deflection plots (P_max for σ_m); W –width of a specimen (10 mm); B—thickness of a specimen (5 mm); a—V-notch depth (2 mm); C_f—constant coefficient (1.363 for an ASTM hammer); and β—constant coefficient (2 [34]).

Using the compliance changing rate (CCR) method [35], regions of crack initiation and crack propagation in impact load-displacement plots were distinguished (Figure 2). The crack initiation point was located in a CCR plot that was evaluated using the subsequent formula [35,36]:

$${\displaystyle \frac{\Delta C}{C}}={\displaystyle \frac{(C-C_{\mathit{el}})}{C_{\mathit{el}}}},$$

(2)

where ${\displaystyle \frac{\Delta C}{C}}$—compliance changing rate; C—current compliance; C_el—elastic compliance.

3. Results

3.1. Texture and Microstructure Characterization

To carry out multi-scale texture and microstructure characterization of the as-processed material, eddy-current testing, EBSD analysis, and TEM observations are employed. The texture of the material is studied in the transversal section of the bar after RS91 and post-swaging annealing with various conditions. Therefore, the core, half the radius, and subsurface layers are considered to reveal phase composition and texture patterns. The morphology of microstructural elements is investigated both in transversal and longitudinal sections.

3.1.1. Eddy-Current Testing

In the as-received bar, the volume fraction of ferromagnetic phases is 2-3%. It is uniformly distributed over the transversal section and corresponds to δ-ferrite. The RS91 material performs an uneven distribution of ferromagnetic (δ-ferrite + α’-martensite) phases along the bar radius. The volume fraction of ferromagnetic phases in the bar core reaches ≈40%, whilst, in the direction of subsurface layers, it increases to ≈80%. Beneath the start temperature (A_S) of α’→γ reversion estimated at 520 °C [15], the volume fraction of ferromagnetic phases changes negligibly (Figure 3). An increase in the annealing temperature from 550°С to 800°С provokes a significant decrease in the volume fraction of ferromagnetic phases to the value of the as-received bar. Apparently, the volume fraction of δ-ferrite remains stable after all applied treatments (RS and subsequent short-term (up to 2 hours) annealing). Hence, changes in the volume fraction of ferromagnetic phases can be attributed to α’→γ reversion. Obviously, the uneven distribution of ferromagnetic phases is detected after RS91 and following annealing at 550-650°С.

3.1.2. EBSD analysis

The CI value of the SW91 material hardly reaches 0.13 that decreases the quality of EBSD data due to high density of lattice defects. Hence, quantitative EBSD characterization of the cold-worked bar can be obstructed. On the other hand, post-swaging annealing at 500 °C (A500) develops static recovery, while static recrystallization and α’→γ reversion are suppressed [15]. Consequently, the CI value increases to 0.28 or more, whereas grain size, texture parameters, and austenite/α’-martensite ratio are stable. Nevertheless, after A500, texture parameters, structure morphology, and phase ratio are similar to those of the SW91 material.

Expected texture components of austenite and the BCC/BCT phase on ODF sections developed during applied thermomechanical processing, and a list of the Miller indices ({hkl}〈uvw〉) of the texture components along with their Euler angles are presented in Figure 4. According to the EBSD data, the A500 material possesses 30-40% of the BCC/BCT phase in the bar core (Figure 5a₁). In the direction of subsurface layers, it increases up to ≈70% (Figure 5a₂) that corresponds well to the results of eddy-current characterization (Figure 3a). Furthermore, microstructure refinement occurs during RS processing (Figure 5b₁ and b₂). KAM (Kernel Average Misorientation) maps demonstrate increased local misorientations throughout the bar radius (Figure 5c₁ and c₂). Texture analysis reveals a pronounced multi-component texture over the transversal section of the bar. The strong triple-fiber texture γ (〈111〉||bar axis (BA)) and θ (〈100〉||BA) of austenite and α (〈110〉||BA) of the BCC/BCT phase dominates in the bar core, while distinct reflections of textures $B/\overline{B}$ ({112}〈110〉), Cube ({001}〈001〉), and Rotated-Cube ({110}〈001〉) of austenite as well as D₁/D₂ ({112}〈111〉) and F ({110}〈001〉) of the BCC/BCT phase are also observed herein (Figure 5d₁ and e₁). Due to the low content of austenite, no austenite textures are identified in subsurface layers (Figure 5d₂). Meanwhile, the textures D₁/D₂, F, and Cube are sharpened in ODF sections of the BCC/BCT phase (Figure 5e₂).

The α’→γ reversion during post-swaging annealing at 550-650 °C is associated with a decrease in the volume fraction of the BCC/BCT phase over the transverse section of the bar. For instance, after annealing at 650°С (A650), the volume fraction of the BCC/BCT phase in the bar core decreases to ≈7% (Figure 6a₁), whereas the content of the BCC/BCT phase in subsurface layers is ≈10% (Figure 6a₂). According to the KAM characterization, grains with reduced local misoriantations are formed in the core of the A650 bar (Figure 6c₁). Yet, the KAM value is still increased (Figure 6c₁ and c₂). The microstructure morphology and grain size are similar to that of the RS91/A500 material over the transversal section after annealing at 550 °C (Supplementary Materials, Figure S2), but, after annealing at 650 °C, grain size increases only in the bar core (Figure 6b₁ and b₂). The strong triple-fiber textures γ and θ of austenite and α of the BCC/BCT phase, as well as reflections of shear textures $B/\overline{B}$ and Reverted-Cube of austenite and D₁/D₂ and F of the BCC/BCT phase are also found in the bar core (Figure 6d₁ and e₁). Interestingly, the θ-fiber texture of austenite is detected only in the bar core, while the Reverted-Cube texture is derived over the transversal section of the bar (Figure 6d₁ and d₂). In the direction of subsurface layers, the intensity of other fiber textures decreases, whilst the shear textures appear to be more pronounced (Figure 6d₁, d₂, e₁, and e₂).

In the A700 material, the uniform distribution of the BCC/BCT phase at a value of 4-5% is observed throughout the bar transverse section (Figure 7a₁ and a₂). In the case of the A800 material, the volume fraction of the BCC/BCT phase decreases to a value of the initial state (2-4%) (Supplementary Materials, Figure S3), mainly corresponding to the content of δ-ferrite [9]. Microstructure coarsening occurs with increasing annealing temperature compared to the A650 sample (Figure 7b₁ and b₂; Supplementary Materials, Figure S3). The KAM maps depict a decrease in local misorientations due to static recovery and recrystallization (Figure 6c₁ and c₂; Supplementary Materials, Figure S3). Meanwhile, textures $B/\overline{B}$, Cube, and Rotated-Cube of austenite in the bar core are smoothened (Figure 7d₁; Supplementary materials, Figure S3), whereas the pronounced patterns of austenitic textures such as $B/\overline{B}$, γ-fiber, and Rotated-Cube are found in subsurface layers (Figure 7d₂). Textures of the BCC/BCT phase are similar to those of the A650 material (Figure 7e₁ and e₂).

As a result of the quantitative analysis of texture patterns, maps of the textured grain distribution are built (Figure 8 and Figure 9). In the material subjected to RS91 and annealed beneath A_S, the large-scale multi-gradient structure is observed. Specifically, the volume fraction of α-fiber textured grains decreases in the direction from the core to subsurface layers (Figure 8a). The volume fraction of BCC/BCT grains with D₁/D₂ and F orientations decreases in the same direction (Figure 8b and c). An increase in the annealing temperature above 650°С smoothens the gradient texture of the BCC/BCT phase due to a substantial decrease in its volume fraction (Figure 3). On the other hand, annealing at temperatures up to 700°С causes the development of a large-scale multi-gradient texture of austenite (Figure 9), but a further increase in annealing temperature results in its weakening. Obviously, the volume fraction of austenitic grains with γ-fiber orientation decreases in the radial direction (Figure 9a), when θ-textured grains of austenite are only detected only in the bar core. Additionally, the volume fractions of austenitic grains with $B/\overline{B}$ and Rotated-Cube orientations increase in the radial direction (Figure 9c and d).

Figure 10 presents the effect of post-swaging annealing temperature on the grain size and volume fraction of the BCC/BCT phase (Vα), fraction of recrystallized areas, and grain size of austenite in the core and subsurface layers of the bar. Although the volume fraction of the BCC/BCT phase of the A500 material is higher in subsurface layers compared to that in the core, it gradually decreases to the value corresponding to the volume fraction of δ-ferrite as the annealing temperature increases to 800 °C (Figure 10a). Meanwhile, the grain size of the BCC/BCT phase is stable in the temperature range of 500-650 °C. After dissolution of fine grains of α’-martensite at 700 °C, the increased size of BCC/BCT grains is attributed to δ-ferrite, which is stable against dissolution. In subsurface layers, the BCC/BCT phase is still fine and uniformly distributed. Recrystallization occurs in the temperature range of 650-800 °C (Figure 10b), which is accompanied by formation of new grains with decreased KAM value. Apparently, in the bar core, a slight increase in grain size is observed at 650°С compared to subsurface layers. However, the grain size of austenite grows dramatically with increasing annealing temperature from 700 to 800°С.

3.1.3. TEM Characterization

To understand microstructural changes at a higher magnification level, the TEM characterization is conducted in transversal and longitudinal sections and the results are presented in Figure 11 and Figure 12. In the transversal section, the bar core of the RS91 material possesses both lamellar and cellular regions with austenite-martensite and fully austenitic structures, respectively (Figure 11a). Increased dislocation density is observed herein. Concurrently, an ultrafine-grained martensite microstructure is derived in subsurface layers of the bar (Figure 11d). In the longitudinal section, the areas are elongated along the bar axis and are fragmented by dislocation boundaries (Figure 12a and d).

Although α’→γ reversion, dislocation redistribution, and Me₂₃C₆-carbide precipitation occur, similar microstructures to those seen just after RS91 are found over the transversal section of the A550 bar (Supplementary Materials, Figure S4). In the A650 material, globular austenitic grains with low dislocation densities nucleate, corresponding to the development of primary recrystallization (Figure 11b and e). Meanwhile, the lamellar morphology of the longitudinal microstructure is observed (Figure 12b and e). The A700 material possesses a partially recrystallized microstructure, but the dislocation cellular regions of austenite are not recrystallized (Figure 11c and f). Moreover, annealing at 700 °C provokes the precipitation of large particles of σ-phase elongated along the bar axis (Figure 12c and f).

3.2. Charpy V-Notch Impact Testing

Charpy V-notch impact tests of the material subjected to RS91 and subsequent annealing are carried out at room temperature. During the testing, all specimens are broken into two parts and then undergo fractographic studies. Further, the microstructural response of materials with outstanding dynamic properties is investigated.

3.2.1. Charpy V-Notch Impact Behavior

Figure 13 shows impact load-displacement plots obtained from instrumental Charpy impact tests. Using the CCR method, regions of crack initiation and its propagation are distinguished in the plots. During impact loading (Figure 13a), the as-received material performs elastic and yield-point-like deformation with further substantial strain hardening in the region of crack initiation (Stage I). Then, the load gradually decreases as the crack propagates (Stage II).

After RS91 and subsequent annealing in a temperature range of 550-700 °C, a weak strain-hardening capacity and increased maximum load (P_MAX) are found in impact load-displacement plots (Figure 13b-e). Therefore, yield-point-like deformation is not attained in the crack initiation stage. Interestingly, pronounced deflections are observed during the crack propagation stage. After A800, the yield-point-like deformation and significant strain hardening occur during the crack initiation, whereas a negligible deflection is also detected in the impact load-displacement plot when the crack propagates (Figure 13f).

3.2.2. Dynamic Properties

Charpy V-notch impact toughness (KCV), the displacement of the maximum load point (d_MAX), and dynamic strength properties of the material under study after different regimes of thermomechanical treatment are shown in Figure 14. The KCV value is presented (Figure 14a) as a sum of relative energy for crack nucleation (KCV_in) and relative energy for crack propagation (KCV_pr). In comparison to KCV of the as-received material (KCV = 2.95 MJ/m²), the RS91 sample demonstrates decreased KCV to 0.83 MJ/m² due to the lack of both KCV_in and KCV_pr. An increase in the annealing temperature from 550 °C to 650 °C causes a slight decrease in KCV_in, while KCV_pr increases dramatically, resulting in overall toughening. After annealing at 700 °C (A700), KCV drops to а value of 0.73 MJ/m² because of low levels of KCV_in and KCV_pr. However, annealing at 800°С (A800) almost doubles KCV_in and KCV_pr compared to the A700 material and thereby increases the KCV value to 1.28 MJ/m².

The RS91 material performs a d_MAX value of 1.5 mm. Then, a negligible decrease in the d_MAX value occurs as annealing temperature increases from 550 °C to 700 °C. Yet, it increases to ≈3 mm after A800, as shown in Figure 14b. The RS91 material possesses the higher general yield stress (σ_gy=1370 MPa) and maximum stress (σ_m=1500 MPa) in comparison to the as-received steel (σ_gy=455 MPa; σ_m=725 MPa). Dynamic strength properties of the A550 material show slight improvement (σ_gy=1415MPa; σ_m=1540 MPa) than those of the RS91 material (Figure 14c). Apparently, the strength properties gradually decrease with increasing annealing temperature from 600 °C to 800 °C.

3.2.3. Fractography After Charpy V-Notch Impact Tests

The fracture surfaces of tested specimens are captured using SEM characterization, as shown in Figure 15. The materials in the as-received state, as well as after A650 and A700 are selected for further detailed fracture analysis. The as-received material shows large shear lips and the flat region between the strained regions on the fracture overview (Figure 15a₁). The fine-dimpled microstructure is observed over the fracture surface that corresponds to the ductile failure mechanism (Figures 15a₂ and a₃). Transition regions noted by dashed line ellipses are clearly visible in A650 and A700 specimens (Figures 15b₁ and c₁). The microstructure of the transition fracture regions is captured step-wise (Figures 15b₂ and c₂). The dimpled microstructure is also found over the fracture (Figures 15b₃ and c₃) that is associated with the formation of microvoids and their growth during the failure.

3.2.4. Post-Mortem EBSD Observations

The origin of increased KCV is discovered via observations of the microstructural response after Charpy V-notch impact tests. Post-mortem EBSD analysis of A650 and A700 materials is conducted under the fracture surface in the longitudinal section (Figure 16; Supplementary Materials, Figure S5). The step-wise fracture profile for the A650 material is presented in Figure 16a. At the V-notch tip, crack nucleation occurs transversely to the bar axis and grows until the bar core. Below the V-notch tip, a volume fraction of α’-martensite is detected at 82% (Figure 16b). In the vicinity of the bar core (at the half the bar radius), the crack turns along the bar axis where wide microzones of γ-, θ-, and α-fibers are observed (Figure 16c). Meanwhile, lines of microvoids aligned with the bar axis are found in the core (Figure 16d and e). These lines are located mainly within austenitic microzones with the γ-fiber orientation and along phase boundaries (Figure 16c). The next turn of the main crack in the direction of the external load forms a step-wise profile (Figure 16d) until the half the bar radius. Here, the main crack expands along the bar axis (Figure 16f). Afterwards, the crack turns transversely back to the bar axis and grows until its periphery. Despite the lamellar morphology of the microstructure at the bar periphery, the grains are fragmented and the wide fiber-textured microzones are not developed (Figure 16b and g).

In the bar core of the A700 material, microvoids, a step-wise profile, and axial secondary cracks are observed (Supplementary Materials, Figure S5a-с). The fracture profile is similar to that of the A650 material. However, the volume fraction of α’-martensite reaches 25% and 11% in the core and subsurface layers of the bar, respectively. The somewhat lamellar character of the longitudinal bar microstructure can be seen in the bar core (Supplementary materials, Figure S5b-d) where microzones with the θ-fiber orientation are detected. Meanwhile, a fine fiber-like microstructure is found in subsurface layers of the bar (Supplementary Materials, Figure S5d).

4. Discussion

The metastable austenitic stainless steel AISI 321 undergoes the RS process and gains a gradient structure, which is lamellar, large-scale, and multicomponent. The gradient structure is transformed during post-swaging annealing via α’→γ reversion, static recovery, and static recrystallization. The phenomena affect the Charpy V-notch impact behavior of the material under study. Charpy V-notched specimens prepared with special care include all layers over the bar transversal section (Figure 1). Therefore, the effect of post-swaging annealing temperature on microstructure/texture evolution, impact strength properties, crack initiation energy, crack propagation energy, mechanical behavior, microstructural response to Charpy impact testing, and fracture mechanisms is discussed in detail below.

4.1. Microstructure and Texture Evolution of RS Material with Annealing Temperature

The as-received material has a uniform structure over the bar transversal section, consisting of coarse grains of austenite with a diameter of 10 μm and fine grains of δ-ferrite. After RS91, a multi-gradient structure forms in the bar, which is associated with increased plastic strain accumulation in subsurface layers during RS processing [37]. Following regions are distinguished within the bar core (Figure 11a): (i) cellular regions of austenite and (ii) lamellar austenitic regions fragmented by martensitic laths or mechanical twins. Apparently, the cellular and lamellar regions of austenite possess θ- and γ-fiber orientations, respectively, due to operating tensile RS stress [38]. However, patterns of shear textures $B/\overline{B}$ and Rotated-Cube are also detected herein. Meanwhile, UFGs of α’-martensite are found in subsurface layers of the as-swaged bar (Figure 11d) that exhibits mostly D₁/D₂ and F orientations due to high compression RS stress [29]. The later enhances SIMT and hence increases the volume fraction of α’-martensite in the radial direction toward the bar surface. It is worth mentioning that the gradient α-fiber texture of the BCC/BCT phase is also derived over the transversal section.

Post-swaging annealing in the temperature range of 550-650 °C provokes α’→γ reversion through the martensitic mechanism that increases the volume fraction of austenite. Yet, the morphology of the microstructure remains unchanged [15,39,40]. This occurs through the variant selection of reverted austenite [41,42]. Thereby, α’-martensitic grains with α-fiber, F and D₁/D₂ orientations are replaced by austenite grains with γ-fiber, Rotated-Cube and $B/\overline{B}$ orientations (Figure 8 and Figure 9). Expectedly, regions of retained austenite can serve as a substrate for the nucleation of reverted austenite [43] and define its orientation [40]. More pronounced α’→γ reversion and static recrystallization occur during annealing at 700 °C, resulting in a strong θ-fiber texture in the bar core, as cellular austenitic regions stay unrecrystallized and new austenitic grains with low dislocation densities replace lamellar regions (Figure 11с). Recrystallization is inhibited in subsurface layers (Figure 11f) due to the pinning effect of fine BCC/BCT grains [40]. Annealing at higher temperatures smoothens microstructure and texture gradients due to enhancement of α’→γ reversion and static recrystallization [30].

Thus, the applied post-swaging annealing conditions can be divided into low- (A550-A650), medium- (A700), and high-temperature (A800) regimes. To summarize the results of the microstructure and texture characterization (Section 3.1), a novel scheme of the microstructure evolution of the RS material with annealing temperature is proposed (Figure 17). Obviously, the pronounced gradient structure obtained after low-temperature annealing at 550-650 °C resemble bamboo one that is also lamellar, large-scale, and multicomponent [25,44,45,46].

4.2. Evolution of Strength with Annealing Temperature

The general yield strength (σ_gy) and maximum stress (σ_m) values decrease with increasing annealing temperature (Figure 14c). However, for the A550 material, the strength increases in comparison to that of the RS91 sample that can be ascribed to the pinning effect of alloying element segregations [47] and Cr₂₃C₆-nanocarbides [28], although α’→γ reversion occurs. According to the earlier study [30], α’→γ reversion at 550-650°С in the AISI 321 steel develops via the martensitic mechanism. Therefore, reverted austenite inherits defects of SIM, resulting in the enhanced hardness and strength properties [48]. Furthermore, the development of the γ-fiber texture of austenite in the bar core and the half the radius also contributes to the increased strength properties due to the low Schmid factor for dislocation glide and mechanical twinning (Figure 17). The diffusional α’→γ reversion, carbide coagulation, static recovery, and static recrystallization (Section 3.1) provoke microstructure coarsening and thereby a gradual decrease in strength when the annealing temperature increases. Indeed, the σ_gy and σ_m values are higher than those under the quasi-static tensile condition [30]. Charpy V-notch impact tests realize a strain rate of more than 10² s^-1 that is a dynamic condition, restricting the stress redistribution capacity at the crack tip and leading to higher operating stress [49].

4.3. Evolution of the KCV Value with Annealing Temperature

The effect of post-swaging annealing temperature on the crack nucleation energy (KCV_in) and crack propagation energy (KCV_pr) is considered separately (Figure 14a). KCV_in and KCV_pr are calculated based on the area under the corresponding parts of impact load-displacement plots (Figure 13) [12]. In comparison to the as-received steel, the RS91 material shows the increased maximum load (P_MAX) and the lack of strain-hardening capacity. The latter results in a substantial decrease in KCV_in (Figure 14a). Despite the development of α’→γ reversion, strain hardening is suppressed after A550, A650, and A700 (Figure 13c-e) that is associated with postponed recrystallization and thereby the UFG microstructure in the annealed samples. Martensitic α’→γ reversion preserves the former UFG microstructure of the RS91 material. Meanwhile, static recovery (Figure 13c-f) reduces the P_MAX value that in turn causes a slight decrease in the d_MAX value (Figure 14b). Hence, annealing treatment at temperatures ranging from 550 to 700 °C provokes a gradual decrease in KCV_in. It is worth noting that partial static recrystallization leads to the formation of the strong θ-fiber texture in the A700 bar, which is referred to the lack of strain hardening because SIMT is inhibited within θ-fiber austenitic grains [22]. Due to complete static recrystallization, microstructure coarsening, and weaker texture patterns, the A800 material performs the double d_MAX value (Figure 14b) due to an increase in strain hardening (Figure 13f), although the maximum load decreases. Moreover, KCV_in of the A800 material is twice higher than that of the A700 material.

The volume fraction of α’-martensite is detected as increased under the V-notch region (Figure 16b) of the A650 specimen that is attributed to operating high stress under impact loading. The lack of strain hardening is derived in the crack initiation stage because its UFG microstructure hinders SIMT [50,51,52,53]. The SIM morphology always depends on austenite grain size [6]. So, globular/blocky α’-martensite is attained near the fracture surface in the AISI 321 steel within an UFG microstructure, while lamellar α’-martensite can be detected in the coarse-grained steel. Interestingly, only lamellar α’-martensite can deflect a crack, increasing crack propagation resistance during Charpy impact testing [6,54]. This phenomenon is not yet observed in the present study due to the fine microstructure of the as-treated material.

On the other hand, an increase in KCV_pr after post-swaging annealing in a temperature range of 550-650 °C is ascribed to the effect of the multi-gradient structure. Therefore, pronounced deflections are observed in the impact load-displacement plots (Figure 13b-d). Moreover, transition zones are detected on fracture overviews of gradient-structured materials (Figure 15b₁). The fracture profile of the A650 material depicts the step-wise patterns of the crack growth (Figure 16a). After nucleation, the crack propagates in subsurface layers along with the external load where the A650 material is UFGed. Turns of the main crack along the bar axis occur before and after the bar core where wide austenitic microzones with the γ-fiber orientation are detected (Figure 16c). These microzones are surrounded by thinner bands of austenite and the BCC/BCT phase with θ- and α-fiber orientations, respectively. After Charpy V-notch impact tests, the γ-fiber microzones possess increased porosity inside (Figure 15c-e). Expectedly, starting from the half the bar radius, the crack grows along the lines of microvoids found within the γ-fiber microzones under impact loading condition.

According to Ref. [55], the formation of microvoids within the γ-fiber microzones of austenite is simplified via the dislocation-dominated mechanism. Moreover, due to the low Schmid factor of the γ-fiber microzones (Figure 18), strain transfer from ductile ambient regions to the γ-fiber microzones is restrained, which causes the microvoid formation at intersections of twin/slip planes and grain boundaries under tension in the upper part of the bending specimen [56,57]. Thus, a step-wise pattern of fracture profile is observed in the region corresponding to the bar core (Figure 16d). Obviously, wide γ-textured microzones can be found from the half the bar radius to the core. In subsurface layers, the UFG lamellar microstructure is not involved such microvoids. Hence, the main crack propagates directly at the onset and end of the failure.

In the A700 material, γ-fiber microzones of austenite are also derived at the half the bar radius (Supplementary Materials, Figures S6a₁-d₁) that is accompanied by the formation of microvoids and secondary cracks. In the bar core, the pronounced θ-fiber microzones of austenite are found (Supplementary Materials, Figure S5a-c). Wang et al. [22] published that θ-fiber microzones are stable against SIMT during Charpy V-notch impact testing. Furthermore, the precipitation of σ-phase diminishes the Cr concentration in solid solution and thereby rise stacking fault energy (SFE) [58], resulting in a decrease in SMIT ability [59]. Therefore, less SIM is derived in the microstructure beneath the fracture, causing the lack of strain hardening despite the high Schmid factor of dislocation glide and mechanical twinning (Supplementary Materials, Figures S6a₂-d₂). Apparently, the σ-phase is brittle that deteriorates the KCV_pr value of the A700 material.

4.4. Strategy of Overcoming the Strength-Toughness Trade-Off

The strategy of overcoming the strength-impact toughness trade-off for MASSs is suggested in the present paper via the development of the multi-gradient structure. The gradient-structured AISI 321 MASS consists of wide γ-textured microzones surrounded by ductile bands in the bar core and the UFG lamellar structure in bar subsurface layers. This gradient structure is observed to form after RS91 and following annealing in the low-temperature range of 550-650 °C. During the first stage of Charpy V-notch impact testing, the crack nucleation at the V-notch tip occurs and microvoids form in γ-fiber microzones and along phase boundaries. Then, during the second stage, the main crack turns along the bar axis, propagating along the microvoids. Hence, the crack propagation energy increases via the rise of the crack growth path (Figure 14a), causing deflections in the impact load-displacement plots (Figure 13b-d). Finally, during the third stage, the crack grows along the direction of the external load within the UFG layers. The revealed toughening mechanism can be attributed to intrinsic mechanisms operating in the elastic/plastic zone ahead of the crack tip [7]. The refined microstructure and strong γ-fiber texture enhance strength properties due to Hall-Petch strengthening and the decreased Schmid factor, respectively. Thus, the program material with the multi-gradient structure performs the superior dynamic strength-impact toughness combination (general yield stress (σ_gy) = 1275-1540 MPa; maximum stress (σ_m) = 1180-1415 MPa; KCV = 0.83-1.38 MJ/m²), overcoming the strength-impact toughness trade-off (Figure 19).

5. Conclusions

In the present paper, the effect of post-swaging annealing on microstructure/texture evolution and Charpy impact toughness behavior of the metastable austenitic stainless steel (i.e., AISI 321) has been studied. The following conclusions are drawn:

• After RS91 and subsequent annealing at 550-650 °C, the bar gains a multi-gradient structure, which is lamellar, large-scale and multicomponent. In the bar core, cellular grains and austenitic regions fragmented by martensitic laths or mechanical twins are observed. In subsurface layers, an ultrafine-grained austenite-martensite microstructure is formed. The gradient distribution of phases and textured grains is derived. So, in the radial direction, the content of the BCC/BCT phase increases. Meanwhile, the volume fraction of austenitic grains with the γ-fiber texture and BCC/BCT grains with the α-fiber texture decreases, whereas the volume fraction of austenitic grains with $B/\overline{B}$ and Rotated-Cube orientations and BCC/BCT grains with D₁/D₂ and F orientations increases. Austenitic grains with the θ-fiber texture are detected only in the bar core.
• The gradient-structured material (after RS91, A550, A600, and A650 treatments) demonstrates increased Charpy V-notch impact toughness (KCV) because of enhanced crack propagation energy (KCV_pr). The phenomenon is attributed to deflections in the impact load-displacement plots that is associated with crack turning along the bar axis. The toughening mechanism involves the formation of microvoids within γ-textured microzones and along phase boundaries elongated in the bar axial direction. The microzones are located in the core of the as-treated bar. The main crack turns and grows along the microvoids under impact loading, resulting in step-wise patterns of fracture after Charpy V-notch impact tests.
• Apparently, the AISI 321 steel subjected to post-swaging annealing in low-temperature range of 550-650 °C performs excellent dynamic mechanical properties (general yield stress (σ_gy) = 1275-1540 MPa; maximum stress (σ_m) = 1180-1415 MPa; KCV = 0.83-1.38 MJ/m²), overcoming the trade-off between strength properties and Charpy V-notch impact toughness. Diffusional α’→γ reversion, static recovery, and static recrystallization cause a gradual softening of the RS91 material when the annealing temperature increases from 550 °C to 800 °C. Annealing at 700 °C provokes the development of θ-fiber microzones of austenite and the precipitation of brittle σ-phase, resulting in the lack of Charpy V impact toughness.