The effect of predeformation on creep strength of 9% Cr steel

materials Article The Effect of Predeformation on Creep Strength of 9% Cr Steel Petr Král1,* , JiˇríDvoˇrák1, Wolfgang Blum 2, Václav Skleniˇcka 1, Zenji Horita 3,4,5, Yoichi Takizawa 6, Yongpeng Tang 7, Lenka Kunˇcická1, Radim Kocich 8, Marie Kvapilová1 and Marie Svobodová9 1Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Zizkova 22, 616 62 Brno, Czech Republic; [email protected] (J.D.); [email protected] (V.S.); [email protected] (L.K.); [email protected] (M.K.) 2Department of Materials Science, Institute I, University of Erlangen-Nuremberg, D-91058 Erlangen, Germany; [email protected] 3Department of Materials Science, Kyushu Institute of Technology, Kitakyushu 804-8550, Japan; [email protected] 4Magnesium Research Center, Kumamoto University, Kumamoto 860-8555, Japan 5Synchrotron Light Application Center, Saga University, Saga 840-8502, Japan 6Technology Department, Nagano Forging Co., Ltd., Nagano 381-0003, Japan; [email protected] 7World Premier International Research Initiative, International Institute for Carbon-Neutral Energy Research (WPI-I2CNER), Kyushu University, Fukuoka 819-0395, Japan; [email protected] 8Faculty of Materials Science and Technology, Technical University of Ostrava, 17. Listopadu 15, 708 00 Ostrava, Czech Republic; [email protected] 9UJP PRAHA a.s., 156 10 Praha-Zbraslav, Czech Republic; [email protected] *Correspondence: [email protected] Received: 2 November 2020; Accepted: 23 November 2020; Published: 25 November 2020   Abstract: Martensitic creep-resistant P92 steel was deformed by different methods of severe plastic deformation such as rotation swaging, high-pressure sliding, and high-pressure torsion at room temperature. These methods imposed significantly different equivalent plastic strains of about 1–30. It was found that rotation swaging led to formation of heterogeneous microstructures with elongated grains where low-angle grain boundaries predominated. Other methods led to formation of ultrafine-grained (UFG) microstructures with high frequency of high-angle grain boundaries. Constant load tensile creep tests at 873 K and initial stresses in the range of 50 to 300 MPa revealed that the specimens processed by rotation swaging exhibited one order of magnitude lower minimum creep rate compared to standard P92 steel. By contrast, UFG P92 steel is significantly softer than standard P92 steel, but differences in their strengths decrease with increasing stress. Microstructural results suggest that creep behavior of P92 steel processed by severe plastic deformation is influenced by the frequency of high-angle grain boundaries and grain coarsening during creep. Keywords: creep-resistant 9% Cr steels; severe plastic deformation; microstructure 1. Introduction Slight predeformation (up to 0.2) of materials can cause both improvement and deterioration of creep properties [ 1 ]. Investigation of creep behavior of predeformed heat resistant austenitic steels revealed that the effect of predeformation on creep behavior depends on deformation mode. It was demonstrated that predeformation in tension leads to improvement of creep resistance [ 2 ]. However, the opposite effect was found after application of compression predeformation [ 1 ]. Deterioration of Materials 2020,13, 5330; doi:10.3390/ma13235330 www.mdpi.com/journal/materials Materials 2020,13, 5330 2 of 17 creep properties was also observed in 2.25Cr–1Mo ferritic steel processed by tension predeformation at 873 K. Kikuchi and Ilschner [ 3 ] investigated the effect of 1% predeformation at different temperatures on creep and revealed that the higher the predeformation temperature, the smaller the difference in creep resistance with respect to the undeformed state. Similar results were found in P92 steel processed by hot bending at 1193–1233 K and subsequent creep at 873 K. It was shown [ 4 ] that hot bending caused insignificant differences in creep strength falling within a scatter band of the unbent pipe. However, insignificant differences in creep strength of 9% of Cr steels were observed after cold bending [5]. F. Abe [ 6 ] investigated creep behavior of P91 steel at 873 K and 78 MPa after cold rolling with plastic strains of 0.2 and 0.4. It was found that time to fracture decreases with increasing cold rolling deformation. Current methods of severe plastic deformation (SPD) allow significantly larger strain to be imposed in comparison with standard forming methods such as rolling or bending [ 7 – 9 ]. In the case where extremely large deformation is imposed to the material, transformation of a coarse-grained structure to an ultrafine-grained one (grain size 0.1–1 µ m) occurs [ 7 ]. Nevertheless, transformation of the microstructure using SPD methods proceeds subsequently. At the beginning of the transformation, the microstructure contains, in particular, cell walls containing tangles with very high dislocation density. The further deformation leads to subsequent increasing of misorientation between cells and formation of an ultrafine-grained microstructure. Thus, the microstructures formed by SPD contain various proportions of low-angle (LA) and high-angle (HA) grain boundaries (GB), depending on the level of imposed plastic strain. For this reason, SPD techniques provide a unique opportunity to study the influence of significantly different microstructures on creep behavior of metals. Recent results showed that differences in the creep strength of SPD-processed state and its undeformed counterpart depend on the value of plastic strain imposed into the material [ 10 – 14 ], temperature, and applied stress used during creep testing [15–17]. The aim of this work is to study the influence of predeformation on creep strength and the microstructure of P92 processed by severe plastic deformation (SPD), with equivalent strains ranging from 1 to 20. 2. Materials and Methods The experimental material used in the present work was advanced tungsten-modified martensitic 9%Cr P92 steel. The chemical composition of P92 steel (wt.%) was as follows: 0.11C, 8.58Cr, 0.33Mo, 1.67W, 0.37Si, 0.48Mn, 0.23V, 0.06Nb, 0.013P, 0.037N, 0.005S, 0.0015B, and 0.017Al. Heat treatment of the as-received coarse-grained (CG) state consisted of normalization at 1323 K for 60 min, followed by tempering at 1013 K for 140 min [4]. Discs of 30 mm diameter and 1.1 mm thickness and sheets with dimensions of 10 × 100 × 1.1 mm 3 were cut from the as-received and relatively coarse-grained P92 steel (CG). The discs were processed by 1 rotation high-pressure torsion (HPT) at room temperature under the pressure of 6 GPa and rotation speed of 0.1 mm per second. The sheets were processed by high-pressure sliding (HPS) at room temperature with a sliding distance of 5 mm (HPS5) and 15 mm (HPS15) under the pressure of 4 GPa. The value of von Mises equivalent strain during HPT [ 7 ] grows with the distance, r, from the center of the disc according to εSPD = 2 πrN/√3t where Nis the number of turns and tis the thickness of disc. The value of von Mises equivalent strain during the HPS process [ 18 ] was estimated as εSPD =x √3t , where xis the sliding distance of the plunger with respect to the anvil and tis the thickness of the sample. The equivalent strain imposed by rotation swaging (RS) [ 19 , 20 ] was estimated by εSPD =ln(D 0 /D n ) 2 , where D 0 is the initial diameter ~30 mm and D n is the final diameter ~15 mm after application of RS at room temperature. The values of equivalent strain εSPD imposed by selected methods of SPD are shown in Table 1. Materials 2020,13, 5330 3 of 17 Table 1. Predeformation (εSPD) imposed by methods of severe plastic deformation (SPD). Material State of P92 Steel RS HPS5 HPS15 HPT εSPD 1.4 2.6 7.9 25 ±5 Constant load tensile creep tests were conducted at 873 K (~0.48 of melting temperature) in protective argon atmosphere using flat specimens with a gauge length of 10 mm and a cross section of 3 × 1 mm 2 . The tested HPT specimens were manufactured from the disc region with equivalent strain of about 25 ± 5 [ 12 , 21 ]. The gauge lengths of the HPS tensile specimens were taken from the central part of the sheet. Microstructure investigations were performed in a scanning electron microscope (SEM, Tescan Lyra 3, (Brno, Czech Republic), equipped with NordlysNano EBSD (High Wycombe, UK) detector operating at accelerating voltage of 20 kV with specimen tilted at 70 ◦ ) and a transmission electron microscope (TEM, JEOL 2100F, Tokyo, Japan). SEM was used to determine the misorientations θ between neighboring grains. θ =15 ◦ was taken to distinguish HAGBs from LAGBs consistent with previous works [ 10 – 12 ]. The mean spacing of LAGBs along test lines is called w, and the mean spacing of HAGBs is called d. The fraction of HAGBs is estimated as fHAGB =w/d. 3. Results 3.1. Microstructure before Creep In the following section we report microstructural details for the P92 variants investigated in this work regarding grains and their orientations (Figure 1), grain misorientations (Figure 2), and boundary spacings (Figure 3). They vary depending on the strain εSPD applied during SPD (Figure 3). As usual, the as-received state CG consists of prior austenite grains containing laths boundaries and other LA subgrain boundaries (not visible in Figure 1) in their interiors resulting from martensitic transformation. The microstructure of as-received state results from normalization at 1323 K for 60 min in air and subsequent tempering at 1013 K for 140 min. Materials 2020, 13, x FOR PEER REVIEW 3 of 19 Table 1. Predeformation (εSPD) imposed by methods of severe plastic deformation (SPD). Material State of P92 Steel RS HPS5 HPS15 HPT εSPD 1.4 2.6 7.9 25 ± 5 Constant load tensile creep tests were conducted at 873 K (~0.48 of melting temperature) in protective argon atmosphere using flat specimens with a gauge length of 10 mm and a cross section of 3 × 1 mm2. The tested HPT specimens were manufactured from the disc region with equivalent strain of about 25 ± 5 [12,21]. The gauge lengths of the HPS tensile specimens were taken from the central part of the sheet. Microstructure investigations were performed in a scanning electron microscope (SEM, Tescan Lyra 3, (Brno, Czech Republic), equipped with NordlysNano EBSD (High Wycombe, UK) detector operating at accelerating voltage of 20 kV with specimen tilted at 70°) and a transmission electron microscope (TEM, JEOL 2100F, Tokyo, Japan). SEM was used to determine the misorientations θ between neighboring grains. θ = 15° was taken to distinguish HAGBs from LAGBs consistent with previous works [10–12]. The mean spacing of LAGBs along test lines is called w, and the mean spacing of HAGBs is called d. The fraction of HAGBs is estimated as fHAGB = w/d. 3. Results 3.1. Microstructure before Creep In the following section we report microstructural details for the P92 variants investigated in this work regarding grains and their orientations (Figure 1), grain misorientations (Figure 2), and boundary spacings (Figure 3). They vary depending on the strain εSPD applied during SPD (Figure 3). As usual, the as-received state CG consists of prior austenite grains containing laths boundaries and other LA subgrain boundaries (not visible in Figure 1) in their interiors resulting from martensitic transformation. The microstructure of as-received state results from normalization at 1323 K for 60 min in air and subsequent tempering at 1013 K for 140 min. Figure 1. Cont. Materials 2020,13, 5330 4 of 17 Materials 2020, 13, x FOR PEER REVIEW 4 of 19 Figure 1. Grain microstructure (orientation contrast with high-angle grain boundaries (HAGB) only, normal direction (ND) and texture (poles of {110} planes) at room temperature before creep for (a) asreceived coarse-grained (CG), (b) rotation swaging (RS), and (c) high-pressure sliding, HPS5 and (d) HPS15; high-pressure torsion (HPT) is similar to HPS15. Rolling directions (RD) are parallel with the pipe axis in CG state and shear direction in SPD-processed states. RS has a heterogeneous grain structure with bands of fine nearly equiaxed submicron grains and large grains exceeding 30 μm that are significantly elongated parallel to the direction of swaging. The grain structure of HPS5 is also relatively coarse but more homogeneous. HPS15 has much more refined grains similar to HPT. Figure 2. Misorientation: cumulative frequency (F) of boundaries with misorientations θ in CG (asreceived) and predeformed P92. Intersection with the dashed line θ = 15° gives the low-angle grain boundaries (LAGB) fraction 𝑓 =𝐹(15°). Figure 1. Grain microstructure (orientation contrast with high-angle grain boundaries (HAGB) only, normal direction (ND) and texture (poles of {110} planes) at room temperature before creep for ( a ) as-received coarse-grained (CG), ( b ) rotation swaging (RS), and ( c ) high-pressure sliding, HPS5 and ( d ) HPS15; high-pressure torsion (HPT) is similar to HPS15. Rolling directions (RD) are parallel with the pipe axis in CG state and shear direction in SPD-processed states. Materials 2020, 13, x FOR PEER REVIEW 4 of 19 Figure 1. Grain microstructure (orientation contrast with high-angle grain boundaries (HAGB) only, normal direction (ND) and texture (poles of {110} planes) at room temperature before creep for (a) asreceived coarse-grained (CG), (b) rotation swaging (RS), and (c) high-pressure sliding, HPS5 and (d) HPS15; high-pressure torsion (HPT) is similar to HPS15. Rolling directions (RD) are parallel with the pipe axis in CG state and shear direction in SPD-processed states. RS has a heterogeneous grain structure with bands of fine nearly equiaxed submicron grains and large grains exceeding 30 μm that are significantly elongated parallel to the direction of swaging. The grain structure of HPS5 is also relatively coarse but more homogeneous. HPS15 has much more refined grains similar to HPT. Figure 2. Misorientation: cumulative frequency (F) of boundaries with misorientations θ in CG (asreceived) and predeformed P92. Intersection with the dashed line θ = 15° gives the low-angle grain boundaries (LAGB) fraction 𝑓 =𝐹(15°). Figure 2. Misorientation: cumulative frequency Fof boundaries with misorientations θ in CG (as-received) and predeformed P92. Intersection with the dashed line θ =15 ◦ gives the low-angle grain boundaries (LAGB) fraction fHAGB =F(15◦). Materials 2020,13, 5330 5 of 17 Materials 2020, 13, x FOR PEER REVIEW 5 of 19 Figure 1 shows a slight tendency in CG to orientate {101} crystallographic planes nearly parallel to the pipe axis, which is identical to the stress axis during creep testing. The microstructure of the RS state contains a strong fiber texture {hkl}<101> parallel to the swaging axis (Figure 1b). The HPS-processed specimens (Figure 1c,d) exhibited a fiber texture <110> perpendicular to the shear direction of HPS with {110}<111> and {110}<100> strongly pronounced variants. A similar texture was also observed in the HPT-processed specimens [21]. Figure 3. Boundary spacings d and w versus equivalent SPD strain (εSPD) (a) after SPD and (b) after annealing; CG data are displayed at εSPD = 0.6 for comparison reasons; the vertical extension of shaded area shrinks with increasing 𝑓. Figure 2 displays the cumulative frequency (F) of b oundaries with misorientations up to θ. For θ = 15°, F equals the LAGB fraction 𝑓 ≡1−𝑓 . The steep inclination of the GC curve shows predominance of LAGBs with θ < 15° and HAGBs with high θ > 40° and a concentration near Σ 3 (60°/111). After SPD, such a gap in the distribution is missing. In RS and HPS5, fHAGB is strongly raised by SPD-induced LAGB generation. With increase of εSPD, strain-induced LABGs get converted into HAGBs, and fHAGB increases to 0.85 for HPT. Figure 3a visualizes the existing data for the mean boundary spacings in the investigated material variants after SPD. The wand d-lines are not primarily precise fits of the limited results. The Figure 3. Boundary spacings dand wversus equivalent SPD strain ( εSPD ) ( a ) after SPD ( b ) after annealing; CG data are displayed at εSPD =0.6 for comparison reasons; the vertical extension of shaded area shrinks with increasing fHAGB. RS has a heterogeneous grain structure with bands of fine nearly equiaxed submicron grains and large grains exceeding 30 µ m that are significantly elongated parallel to the direction of swaging. The grain structure of HPS5 is also relatively coarse but more homogeneous. HPS15 has much more refined grains similar to HPT. Figure 1shows a slight tendency in CG to orientate {101} crystallographic planes nearly parallel to the pipe axis, which is identical to the stress axis during creep testing. The microstructure of the RS state contains a strong fiber texture {hkl}<101>parallel to the swaging axis (Figure 1b). The HPS-processed specimens (Figure 1c,d) exhibited a fiber texture <110>perpendicular to the shear direction of HPS with {110}<111>and {110}<100>strongly pronounced variants. A similar texture was also observed in the HPT-processed specimens [21]. Figure 2displays the cumulative frequency (F) of b oundaries with misorientations up to θ . For θ =15 ◦ ,Fequals the LAGB fraction fLAGB ≡ 1 −fHAGB . The steep inclination of the GC curve shows predominance of LAGBs with θ <15 ◦ and HAGBs with high θ >40 ◦ and a concentration near Σ 3 (60 ◦ /111). After SPD, such a gap in the distribution is missing. In RS and HPS5, f HAGB is strongly raised by SPD-induced LAGB generation. With increase of εSPD , strain-induced LABGs get converted into HAGBs, and fHAGB increases to 0.85 for HPT. Figure 3a visualizes the existing data for the mean boundary spacings in the investigated material variants after SPD. The wand d-lines are not primarily precise fits of the limited results. The w-line with w∝ 1 /ε0.5 SPD is motivated by parabolic work hardening, σ∝ε0.5 SPD , during SPD at room temperature toward the final stationary (saturation) stage and formation of quasi-stationary subgrain structures Materials 2020,13, 5330 6 of 17 with wqs ∝ 1 /σ . The d -line with d∝ 1 /εSPD decreases more strongly with εSPD than w because misorientation of θ -distribution shifts toward higher θ with εSPD by transformation of LAGBs into HAGBs. Vertical extension of the shaded area marks f HAGB ; as the lines for dand wapproach each other, fLAGB shrinks and fHAGB approaches 1. Imposing εSPD =1.4 by RS reduces the grain size dto 1.6 µ m, and 35% of the boundaries are HAGBs (Figure 2) with relatively equal θ -distribution between 15–65 ◦ . Processing with HPS with a sliding distance of 5 mm to εSPD =2.6 generates a grain size of about 1.25 µ m and a similar portion of HAGBs. Imposing εSPD =7.9 by HPS15 led to a significant decrease of mean grain size down to 0.35 µ m and an increase of f HAGB up to 0.7. The lowest grain size, d, and the highest, f HAGB , are found for HPT subjected to the highest SPD strain, εSPD =25 ±5. However, grain parameters after SPD at room temperature are not the ones determining creep, as the grains coarsen during heating to the test temperature and subsequent soaking until the beginning of creep. While this annealing treatment before creep has little effect for micron-sized grains, it causes significant coarsening of submicron grains. This is seen in Figure 3b (see arrow “recovery”). Figure 4shows the effect of annealing for 5 h at 873 K; this treatment simulates the annealing effect of heating to the test temperature. Despite different degrees of εSPD in RS, HPS5, HPS15, and HPT, the subgrain structures look qualitatively similar. Recovery of dislocation line length and boundary area involves motion of both free dislocations and boundaries to sites of recovery by recombination and annihilation. While this occurs, many relatively straight free dislocations (see arrows) are seen to extend between neighboring boundaries. The subgrain sizes w(mean intercepts) estimated from micrographs like those of Figure 5are plotted in Figure 3b. The wand d-coarsening is small for the relatively coarse grains of RS and HPS5, but it is prominent for the ultrafine grains of HPT where dgoes up to 0.35 µm with a fraction of HAGB area (fHAGB) ~90%. Materials 2020, 13, x FOR PEER REVIEW 6 of 19 w-line with 𝑤∝1/𝜀  . is motivated by parabolic work hardening, 𝜎∝𝜀  . , during SPD at room temperature toward the final stationary (saturation) stage and formation of quasi-stationary subgrain structures with 𝑤 ∝1/𝜎. The 𝑑-line with 𝑑∝1/𝜀  decreases more strongly with 𝜀 than w because misorientation of 𝜃-distribution shifts toward higher 𝜃 with 𝜀 by transformation of LAGBs into HAGBs. Vertical extension of the shaded area marks f HAGB; as the lines for d and w approach each other, fLAGB shrinks and fHAGB approaches 1. Imposing εSPD = 1.4 by RS reduces the grain size, d, to 1.6 μm, and 35% of the boundaries are HAGBs (Figure 2) with relatively equal θ-distribution between 15–65°. Processing with HPS with a sliding distance of 5 mm to εSPD = 2.6 generates a grain size of about 1.25 μm and a similar portion of HAGBs. Imposing εSPD = 7.9 by HPS15 led to a significant decrease of mean grain size down to 0.35 μm and an increase of fHAGB up to 0.7. The lowest grain size, d, and the highest, fHAGB, are found for HPT subjected to the highest SPD strain, εSPD = 25 ± 5. However, grain parameters after SPD at room temperature are not the ones determining creep, as the grains coarsen during heating to the test temperature and subsequent soaking until the beginning of creep. While this annealing treatment before creep has little effect for micron-sized grains, it causes significant coarsening of submicron grains. This is seen in Figure 3b (see arrow “recovery”). Figure 4 shows the effect of annealing for 5 h at 873 K; this treatment simulates the annealing effect of heating to the test temperature. Despite different degrees of εSPD in RS, HPS5, HPS15, and HPT, the subgrain structures look qualitatively similar. Recovery of dislocation line length and boundary area involves motion of both free dislocations and boundaries to sites of recovery by recombination and annihilation. While this occurs, many relatively straight free dislocations (see arrows) are seen to extend between neighboring boundaries. The subgrain sizes w (mean intercepts) estimated from micrographs like those of Figure 5 are plotted in Figure 3b. The wand d-coarsening is small for the relatively coarse grains of RS and HPS5, but it is prominent for the ultrafine grains of HPT where d goes up to 0.35 μm with a fraction of HAGB area (fHAGB) ~90%. (a) RS εSPD = 1.4 (b) HPS5 εSPD = 2.6 Materials 2020, 13, x FOR PEER REVIEW 7 of 19 (c) HPS15 εSPD = 7.9 (d) εSPD ~ 25 ± 5 Figure 4. Microstructure (TEM) after annealing at 873 K for 5 h in (a) RS, (b) HPS5, (c) HPS15, and (d) HPT; note the higher magnification in (c,d). Figure 5. Microstructure of P92 steel processed by HPT and annealed at 923 K for 500 h. Even more (sub)grain coarsening occurs in annealing for 500 h at 923 K (Figure 5). Here the grains coarsen to d ~0.85 μm with fHAGB ~87%. The high value of fHAGB is understandable as no LABGs are formed during static annealing where no net plastic deformation occurs. 3.2. Creep Behavior Figure 6 shows evolution of strain rate (𝜀󰇗) with strain in creep at σ0 = 150 MPa for all P92 variants. All curves display a pronounced relative minimum of creep rate (𝜀󰇗). HPT, HPS5, and HPS15 have similar 𝜀󰇗. CG and RS creep much more slowly. Ductilities of different SPD specimens differ greatly. RS has the lowest ductility. The results will be discussed below in relation to microstructure. Figure 7 shows the creep rates of CG, RS, and HPS versus stress. Tests at constant load have the advantage to show both the evolution with strain and the influence of stress in a single plot. This is because stress varies at constant load with strain, as Figure 4. Microstructure (TEM) after annealing at 873 K for 5 h in ( a ) RS, ( b ) HPS5, ( c ) HPS15, and (d) HPT; note the higher magnification in (c,d). Materials 2020,13, 5330 7 of 17 Materials 2020, 13, x FOR PEER REVIEW 7 of 19 (c) HPS15 εSPD = 7.9 (d) εSPD ~ 25 ± 5 Figure 4. Microstructure (TEM) after annealing at 873 K for 5 h in (a) RS, (b) HPS5, (c) HPS15, and (d) HPT; note the higher magnification in (c,d). Figure 5. Microstructure of P92 steel processed by HPT and annealed at 923 K for 500 h. Even more (sub)grain coarsening occurs in annealing for 500 h at 923 K (Figure 5). Here the grains coarsen to d ~0.85 μm with fHAGB ~87%. The high value of fHAGB is understandable as no LABGs are formed during static annealing where no net plastic deformation occurs. 3.2. Creep Behavior Figure 6 shows evolution of strain rate (𝜀󰇗) with strain in creep at σ0 = 150 MPa for all P92 variants. All curves display a pronounced relative minimum of creep rate (𝜀󰇗). HPT, HPS5, and HPS15 have similar 𝜀󰇗. CG and RS creep much more slowly. Ductilities of different SPD specimens differ greatly. RS has the lowest ductility. The results will be discussed below in relation to microstructure. Figure 7 shows the creep rates of CG, RS, and HPS versus stress. Tests at constant load have the advantage to show both the evolution with strain and the influence of stress in a single plot. This is because stress varies at constant load with strain, as Figure 5. Microstructure of P92 steel processed by HPT and annealed at 923 K for 500 h. Even more (sub)grain coarsening occurs in annealing for 500 h at 923 K (Figure 5). Here the grains coarsen to d~0.85 µ m with f HAGB ~87%. The high value of f HAGB is understandable as no LABGs are formed during static annealing where no net plastic deformation occurs. 3.2. Creep Behavior Figure 6shows evolution of strain rate . ε with strain in creep at σ0 =150 MPa for all P92 variants. All curves display a pronounced relative minimum of creep rate . εmin . HPT, HPS5, and HPS15 have similar . εmin . CG and RS creep much more slowly. Ductilities of different SPD specimens differ greatly. RS has the lowest ductility. The results will be discussed below in relation to microstructure. Materials 2020, 13, x FOR PEER REVIEW 8 of 19 σ = σ0 exp(ε) (1) provided that deformation is uniform. Empirically it is known that uniformity of deformation along the gauge length is provided in relatively ductile materials until the strain has reached about half the fracture strain [22]. According to Equation (1), the logarithmic stress axis corresponds to a linear strain axis. Figure 6. Evolution of strain rate with strain in tensile creep at constant load with σ0 = 150 MPa for different SPD strains εSPD. Thus, the individual curves in Figure 7 can be viewed as 𝜀󰇗–ε curves like those of Figure 6 but shifted along the abscissa. Deformation begins at 𝜎=𝜎 ; the scale bar marks an interval Δ𝜀 = 0.5 of uniform strain. Not only do the minima of the curves displayed in Figure 7 show dependence on stress, they also show the full evolution of rate from the beginning, via strengthening and softening up to the second half of the curves where external and internal necking cause an over-exponential increase of rate until final fracture. Figure 7 shows two main features. First, the global stress dependences of rates differ strongly for the variants HPS and HPT with the highest εSPD and the variants RS and CG with the lowest εSPD. Roughly speaking, the stress exponent 𝑛 = 𝑑𝑙𝑜𝑔𝜀󰇗/𝑑𝑙𝑜𝑔𝜎 of the rates is about 6 to 7 for HPS and HPT and twice that high for RS and CG. This large n-difference leads to the result that the minimum creep rates of all variants are similar at a high stress of about 400 MPa, while CG and RS creep much more slowly at the lower stress of 150 MPa. Figure 7 also displays softening during creep that is evident from the pronounced minimum due to the fact that the rate increases distinctly more strongly after the minimum than would be expected from stress sensitivity alone. In the literature, this steep increase is termed tertiary creep and is often linked to “damage” and fracture. However, if deformation is fairly uniform after the minimum, as in our ultrafine-grained (UFG) materials with fracture strains > 0.5, the rate increase must be attributed to softening by internal microstructural changes, e.g., coarsening processes (see Discussion) rather than fracture. The following subsection describes the observed changes. Figure 6. Evolution of strain rate with strain in tensile creep at constant load with σ0 =150 MPa for different SPD strains εSPD. Materials 2020,13, 5330 8 of 17 Figure 7shows the creep rates of CG, RS, and HPS versus stress. Tests at constant load have the advantage to show both the evolution with strain and the influence of stress in a single plot. This is because stress varies at constant load with strain, as σ=σ0exp(ε) (1) provided that deformation is uniform. Empirically it is known that uniformity of deformation along the gauge length is provided in relatively ductile materials until the strain has reached about half the fracture strain [22]. According to Equation (1), the logarithmic stress axis corresponds to a linear strain axis. Materials 2020, 13, x FOR PEER REVIEW 9 of 19 Figure 7. Creep rate–stress curves for HPT, CG, and RS. 3.3. Changes of Microstructure during Creep Testing The microstructure of fractured tensile specimens was investigated at different locations along their length. In the grip parts, the local stress, σloc, is negligible. In the gauge length, the stresses and strains vary with location. The local strain was roughly estimated as εloc = ln(Sloc/S0) from the local cross section, S, and the initial cross section, S0. The local stress follows from Equation (1) as σloc = σ0exp(εloc). Figures 8 and 9 show examples of the microstructure in SEM and TEM for the moderately predeformed variant RS after creep to fracture at different σ0 and different σloc. (a) grip part, local strain (εloc) ~ 0 Figure 7. Creep rate–stress curves for HPT, CG, and RS. Thus, the individual curves in Figure 7can be viewed as . ε – ε curves like those of Figure 6but shifted along the abscissa. Deformation begins at σ=σ0 ; the scale bar marks an interval ∆ε =0.5 of uniform strain. Not only do the minima of the curves displayed in Figure 7show dependence on stress, the curves also show the full evolution of rate from the beginning, via strengthening and softening up to the second half of the curves where external and internal necking cause an over-exponential increase of rate until final fracture. Figure 7shows two main features. First, the global stress dependences of rates differ strongly for the variants HPS and HPT with the highest εSPD and the variants RS and CG with the lowest εSPD . Roughly speaking, the stress exponent n=dlog . ε/dlogσ of the rates is about 6 to 7 for HPS and HPT and twice that high for RS and CG. This large n-difference leads to the result that the minimum creep rates of all variants are similar at a high stress of about 400 MPa, while CG and RS creep much more slowly at the lower stress of 150 MPa. Figure 7also displays softening during creep that is evident from the pronounced minimum due to the fact that the rate increases distinctly more strongly after the minimum than would be expected from stress sensitivity alone. In the literature, this steep increase is termed tertiary creep and is often linked to “damage” and fracture. However, if deformation is fairly uniform after the minimum, as in our ultrafine-grained (UFG) materials with fracture strains >0.5, the rate increase must be attributed to softening by internal microstructural changes, e.g., coarsening processes (see Discussion) rather than fracture. The following subsection describes the observed changes. Materials 2020,13, 5330 9 of 17 3.3. Changes of Microstructure during Creep Testing The microstructure of fractured tensile specimens was investigated at different locations along their length. In the grip parts, the local stress, σloc , is negligible. In the gauge length, the stresses and strains vary with location. The local strain was roughly estimated as εloc =ln(S loc /S 0 ) from the local cross section, S loc , and the initial cross section, S 0 . The local stress follows from Equation (1) as σloc =σ0exp(εloc). Figures 8and 9show examples of the microstructure in SEM and TEM for the moderately predeformed variant RS after creep to fracture at different σ0and different σloc. Materials 2020, 13, x FOR PEER REVIEW 9 of 19 Figure 7. Creep rate–stress curves for HPT, CG, and RS. 3.3. Changes of Microstructure during Creep Testing The microstructure of fractured tensile specimens was investigated at different locations along their length. In the grip parts, the local stress, σloc, is negligible. In the gauge length, the stresses and strains vary with location. The local strain was roughly estimated as εloc = ln(Sloc/S0) from the local cross section, S, and the initial cross section, S0. The local stress follows from Equation (1) as σloc = σ0exp(εloc). Figures 8 and 9 show examples of the microstructure in SEM and TEM for the moderately predeformed variant RS after creep to fracture at different σ0 and different σloc. (a) grip part, local strain (εloc) ~ 0 Materials 2020, 13, x FOR PEER REVIEW 10 of 19 (b) gauge near fracture surface, εloc ~ 0.09 Figure 8. Microstructure and texture of RS after creep at σ0 = 200 MPa in (a) grip part and (b) gauge length near the fracture. RD directions are parallel with the stress axis. Grain structure and texture, subgrain, and dislocation structure look relatively similar to before creep (Figures 1b and 4a). It is noted that the dislocation structures have all undergone recovery, either before creep or after fracture; therefore, Figure 9 probably is not representative for the dislocation structure in situ under creep stress. Figure 9. Dislocation microstructure in the gauge lengths of specimens tested: (a) 300 MPa and (b) 150 MPa. Next, we consider the most severely predeformed materials HPS15 and HPT. Figure 10 shows grain structure and texture for HPS15 in dependence of εloc. Compared to the state before creep (Figure 4c), the ultrafine grains have coarsened statically in the grip part and even more so in gauge length. The texture did not change in the grip part and after slight creep strain in the gauge length (Figure 10a,b). The misorientation distribution evolves during creep (Figure 11). There is a significant increase of HAGB fraction (fHAGB) during creep as εloc increases to 0.12 and 0.38. It appears that LAGBs migrate faster during grain coarsening by recovery through recombination with other boundaries than HAGBs do and so disappear earlier. Near the fracture surface, the HAGB fraction, fHAGB, decreased again. This suggests massive deformation connected with formation of small subgrains in large grains at the high local stresses acting there. (a) (b) Figure 8. Microstructure and texture of RS after creep at σ0 =200 MPa in ( a ) grip part and ( b ) gauge length near the fracture. RD directions are parallel with the stress axis. Materials 2020, 13, x FOR PEER REVIEW 10 of 19 (b) gauge near fracture surface, εloc ~ 0.09 Figure 8. Microstructure and texture of RS after creep at σ0 = 200 MPa in (a) grip part and (b) gauge length near the fracture. RD directions are parallel with the stress axis. Grain structure and texture, subgrain, and dislocation structure look relatively similar to before creep (Figures 1b and 4a). It is noted that the dislocation structures have all undergone recovery, either before creep or after fracture; therefore, Figure 9 probably is not representative for the dislocation structure in situ under creep stress. Figure 9. Dislocation microstructure in the gauge lengths of specimens tested: (a) 300 MPa and (b) 150 MPa. Next, we consider the most severely predeformed materials HPS15 and HPT. Figure 10 shows grain structure and texture for HPS15 in dependence of εloc. Compared to the state before creep (Figure 4c), the ultrafine grains have coarsened statically in the grip part and even more so in gauge length. The texture did not change in the grip part and after slight creep strain in the gauge length (Figure 10a,b). The misorientation distribution evolves during creep (Figure 11). There is a significant increase of HAGB fraction (fHAGB) during creep as εloc increases to 0.12 and 0.38. It appears that LAGBs migrate faster during grain coarsening by recovery through recombination with other boundaries than HAGBs do and so disappear earlier. Near the fracture surface, the HAGB fraction, fHAGB, decreased again. This suggests massive deformation connected with formation of small subgrains in large grains at the high local stresses acting there. (a) (b) Figure 9. Dislocation microstructure in the gauge lengths of specimens tested: ( a ) 300 MPa and ( b ) 150 Mpa. Materials 2020,13, 5330 16 of 17 5. The heterogeneous microstructure of RS containing large elongated grains with subgrains exhibits the highest creep resistance but the lowest ductility. The observed creep softening related with grain refinement by SPD indicates that there are only limited means to improve creep resistance. However, the enormous increase in ductility may be a significant advantage in some cases. In any case, the present findings seem to be of relevance for the problem of creep at very low stresses and creep rates in long-term application of tempered martensite Cr steels. Due to inverse stress dependence, the qs subgrain size becomes very large under these conditions so that the grains become more and more subgrain-free and the fraction of LAGBs approaches and exceeds 0.5. In this situation, one must expect that the stress dependence of the creep rate changes compared to the conventional CG case with d>> wbecause dynamic recovery is accelerated by HAGB control. This expectation of softening, accompanied by declining stress exponent of creep rate, agrees with many observations made in the range of very slow creep, which so far have not found a satisfactory and generally accepted explanation. Author Contributions: Conceptualization, P.K., W.B., V.S., and Z.H.; methodology, P.K., W.B., V.S., J.D., R.K. and Z.H.; software, W.B.; validation, P.K., J.D., Z.H., Y.T. (Yoichi Takizawa), Y.T. (Yongpeng Tang) and M.S.; formal analysis, W.B. and P.K.; investigation, P.K., J.D., V.S., L.K., M.K., M.S. and Y.T. (Yoichi Takizawa); resources, P.K., J.D., R.K., M.S., Y.T. (Yoichi Takizawa) and Y.T. (Yongpeng Tang); data curation, P.K., J.D. and W.B.; writing—original draft preparation, P.K., W.B. and V.S.; writing—review and editing, P.K., W.B. and V.S.; visualization, P.K., W.B.; J.D., L.K., R.K. and M.K.; supervision, W.B. and V.S. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the Czech Science Foundation, grant number 19-18725S. Acknowledgments: The authors acknowledge financial support from the Czech Science Foundation (grant No. 19-18725S). The work was supported in part by a grant-in-aid from MEXT, Japan, for scientific research (A) (No. 19H00830). The HPT process was carried out at the International Research Center for Giant Straining for Advanced Materials (IRC-GSAM) at Kyushu University, Fukuoka, Japan. Conflicts of Interest: The authors declare no conflict of interest. References 1. Li, D.F.; Dowd, N.P.O.; Davies, C.M.; Nikbin, K.M. A review of the effect of prior inelastic deformation on high temperature mechanical response of engineering alloys. Int. J. Press. Vessel. Pip. 2010 ,87, 531–542. [CrossRef] 2. Usami, S.; Mori, T. Creep deformation of austenitic steels at medium and low temperatures. Cryogenics 2000 , 40, 117. [CrossRef] 3. Kikuchi, S.; Ilschner, B. Effects of a small prestrain at high temperatures on the creep behaviour of AISI 304 stainless steel. Scr. Metal. 1986,20, 159–162. [CrossRef] 4. Skleniˇcka, V.; Kuchaˇrov á , K.; Kr á l, P.; Kvapilov á , M.; Svobodov á , M.; ˇ Cmakal, J. The effect of hot bending and thermal ageing on creep and microstructure evolution in thick-walled P92 steel pipe. Mater. Sci. Eng. A 2015,644, 297–309. [CrossRef] 5. Caminada, S.; Cumino, G.; Cipolla, L.; Di Gianfrancesco, A. Cold bending of advanced steels: ASTM grades T23, T92, T92. Int. J. Press. Vessel. Pip. 2009,86, 853–861. [CrossRef] 6. Abe, F. Effect of quenching, tempering, and cold rolling on creep deformation behavior of a tempered martensitic 9Cr-1W steel. Metall. Mater. Trans. A 2003,34, 913–925. [CrossRef] 7. Valiev, R.Z.; Islamgaliev, R.K.; Alexandrov, I.V. Bulk nanostructured materials from severe plastic materials. Prog. Mater. Sci. 2000,45, 103–189. [CrossRef] 8. Renk, O.; Pippan, R. Saturation of Grain Refinement during Severe Plastic Deformation of Single Phase Materials: Reconsiderations, Current Status and Open Questions. Mater. Trans. 2019 ,60, 1270–1282. [CrossRef] 9. Ganeev, A.; Nikitina, M.; Sitdikov, V.; Islamgaliev, R.; Hoffman, A.; Wen, H. Effects of the Tempering and High-Pressure Torsion Temperatures on Microstructure of Ferritic/Martensitic Steel Grade 91. Materials 2018 , 11, 627. [CrossRef] 10. Blum, W.; Dvorak, J.; Kral, P.; Eisenlohr, P.; Sklenicka, V. Effect of grain refinement by ECAP on creep of pure Cu. Mater. Sci. Eng. A 2014,590, 423–432. [CrossRef] Materials 2020,13, 5330 17 of 17 11. Kral, P.; Dvorak, J.; Sklenicka, V.; Langdon, T.G. The Characteristics of creep in metallic materials processed by severe plastic deformation. Mater. Trans. 2019,60, 1506–1517. [CrossRef] 12. Kral, P.; Dvorak, J.; Sklenicka, V.; Masuda, T.; Horita, Z.; Kucharova, K.; Kvapilova, M.; Svobodova, M. The effect of ultrafine-grained microstructure on creep behaviour of 9% Cr steel. Materials 2018 ,11, 787. [CrossRef] [PubMed] 13. Kostka, A.; Tak, K.-G.; Hellmig, R.J.; Estrin, Y.; Eggeler, G. On the contribution of carbides and micrograin boundaries to the creep strength of tempered martensite ferritic steel. Acta Mater. 2007 ,55, 539–550. [CrossRef] 14. Kral, P.; Dvorak, J.; Sklenicka, V.; Horita, Z.; Takizawa, Y.; Tang, Y.; Kvapilova, M.; Svobodova, M. Effect of ultrafine-grained microstructure on creep behaviour in 304L austenitic steel. Mater. Sci. Eng. A 2020 , 785, 139383. [CrossRef] 15. Blum, W.; Dvorak, J.; Kral, P.; Eisenlohr, P.; Sklenicka, V. Effects of grain refinement by ECAP on the deformation resistance of Al interpreted in terms of boundary-mediated processes. J. Mater. Sci. Technol. 2016,32, 1309–1320. [CrossRef] 16. Blum, W.; Eisenlohr, P. A simple dislocation model of the influence of high-angle boundaries on the deformation behavior of ultrafine-grained materials. In Proceedings of the 15th International Conference on the Strength of Materials (ICSMA-15), Dresden, Germany, 16–21 August 2009; IOP Publishing: Bristol, UK, 2010; Volume 240, p. 012136. [CrossRef] 17. Blum, W.; Dvorak, J.; Kral, P.; Eisenlohr, P.; Sklenicka, V. Quasi-stationary strength of ECAP-processed Cu-Zr at 0.5 Tm.Metals 2019,9, 1149. [CrossRef] 18. Fujioka, T.; Horita, Z. Development of high-pressure sliding process for microstructural refinement of rectangular metallic sheets. Mater. Trans. 2009,50, 930–933. [CrossRef] 19. Kunˇcick á , L.; Kocich, R.; Dvoˇr á k, K.; Mach á ˇckov á , A. Rotary swaged laminated Cu-Al composites: Effect of structure on residual stress and mechanical and electric properties. Mater. Sci. Eng. A 2019 ,742, 743–750. [CrossRef] 20. Mach á ˇckov á , A.; Kr á tk á , L.; Petrmichl, R.; Kunˇcick á , L.; Kocich, R. Affecting structure characteristics of rotary swaged tungsten heavy alloy via variable deformation temperature. Materials 2019,12, 4200. [CrossRef] 21. Kral, P.; Dvorak, J.; Sklenicka, V.; Masuda, T.; Horita, Z.; Kucharova, K.; Kvapilova, M.; Svobodova, M. Microstructure and creep behaviour of P92 steel after HPT. Mater. Sci. Eng. A 2018 ,723, 287–295. [CrossRef] 22. Straub, S.; Meier, M.; Ostermann, J.; Blum, W. Development of microstructure and strengthening in the ferritic Steel X20CrMoV121 at 823K during long-term creep tests and during annealing. VGB Kraftw. 1993 , 73, 646–653. 23. Blum, W.; Zeng, X.H. A simple dislocation model of deformation resistance of ultrafine-grained materials explaining Hall–Petch strengthening and enhanced strain rate sensitivity. Acta Mater. 2009 ,57, 1966–1974. [CrossRef] 24. Li, Y.; Zeng, X.; Blum, W. Transition from strengthening to softening by grain boundaries in ultrafine-grained Cu. Acta Mater. 2004,52, 5009–5018. [CrossRef] 25. Raj, S.V.; Pharr, G.M. A compilation and analysis of data for the stress dependence of the subgrain size. Mater. Sci. Eng. A 1986,81, 217–237. [CrossRef] 26. Frost, H.J.; Ashby, M.F. Deformation-Mechanism Maps; Pergamon Press: Oxford, UK, 1982; pp. 62–63. 27. Abe, F. Effect of fine precipitation and subsequent coarsening of Fe2W Laves phase on the creep deformation behaviour of tempered martensitic 9Cr-W steels. Metal. Mater. Trans. A 2005,26, 321–332. [CrossRef] Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).