《金属材料强韧化与组织调控》教学资源（参考文献）Applying the texture analysis for optimizing thermomechanical treatment of high manganese twinning-induced plasticity steel

Available online at www.sciencedirect.com CrossMark ScienceDirect ACta MATERIALIA ELSEVIER Acta Materialia 80(2014)327-340 www.elsevier.com/locate/actamat Applying the texture analysis for optimizing thermomechanical treatment of high manganese twinning-induced plasticity steel Christian Haase*,Luis A.Barrales-Mora,Franz Roters,Dmitri A.Molodov", Gunter Gottstein A Institute of Physical Metallurgy and Metal Physics,RWTH Aachen University,Aachen 52074,Germany Max-Planck-Institut fur Eisenforschung GmbH.Max-Planck-Str.1.Dusseldorf 40237.Germany Received 4 July 2014;accepted 30 July 2014 Available online 3 September 2014 Abstract Texture analysis was applied to determine the optimal processing parameters for the thermomechanical treatment of an Fe-23Mn- 1.5Al-0.3C twinning-induced plasticity steel.A simple processing route consisting of cold rolling and recovery annealing was used to explore the possibility of tailoring the mechanical properties of this steel.The thermal stability of mechanically induced twin boundaries provided high retained yield strength after recovery annealing.In addition,recovery processes facilitated a significantly improved duc- tility compared to the cold-rolled material.It was shown that the analysis of texture evolution during deformation and annealing can be used as an effective tool to optimize cold rolling degree and annealing conditions.A dislocation-based constitutive model was used in order to validate that the CuT texture component can be used as an indirect indicator for the evolution of the deformation twin density. Furthermore,simulation results identified recovery as the dominating softening mechanism under the applied annealing conditions. 2014 Acta Materialia Inc.Published by Elsevier Ltd.All rights reserved. Keywords:TWIP steel;Texture;Microstructure;Recovery;Deformation twinning 1.Introduction ~20 and ~50 mJ m-2 at room temperature [5-71.As a con- sequence of the low SFE,deformation mechanisms such as Since the works of Grassel et al.[1,2]and Frommeyer deformation twinning and shear banding become activated et al.[3],high manganese twinning-induced plasticity during deformation in addition to dislocation glide,result- (TWIP)steels have gained strong interest of the automobile ing in high strain hardening rates and thus in high ductility industry for wide usage as structural components,and and high strength with a typical ultimate tensile strength therefore have moved into the focus of worldwide steel and elongation to fracture product of more than research.These face-centered cubic,fully austenitic steels 50,000MPa%[8-101. contain a high amount of manganese (15-30 wt.%)and A major shortcoming of TWIP steels in the fully recrys- typically,additions of C (0.05-1 wt.%),Al (0-3 wt.%) tallized,coarse-grained state is their relatively low yield and/or Si (0-3 wt.%)[4].This alloying concept results in strength,which is typically in the range between 200 and a low stacking fault energy(SFE)in the range between 400 MPa [2,11].Although a low onset of plastic deforma- tion facilitates energy-effective part shaping,it is detrimen- tal when the material is applied in crash-relevant structural Corresponding author.Tel.:+49 241 8026877:fax:+49 241 8022301. components(such as the A-or B-pillar in automobiles). E-mail address:haase@imm.rwth-aachen.de (C.Haase). http://dx.doi.org/10.1016/j.actamat.2014.07.068 1359-6454/2014 Acta Materialia Inc.Published by Elsevier Ltd.All rights reserved

Applying the texture analysis for optimizing thermomechanical treatment of high manganese twinning-induced plasticity steel Christian Haase a,⇑ , Luis A. Barrales-Mora a , Franz Roters b , Dmitri A. Molodov a , Gu¨nter Gottstein a a Institute of Physical Metallurgy and Metal Physics, RWTH Aachen University, Aachen 52074, Germany bMax-Planck-Institut fu¨r Eisenforschung GmbH, Max-Planck-Str. 1, Du¨sseldorf 40237, Germany Received 4 July 2014; accepted 30 July 2014 Available online 3 September 2014 Abstract Texture analysis was applied to determine the optimal processing parameters for the thermomechanical treatment of an Fe–23Mn– 1.5Al–0.3C twinning-induced plasticity steel. A simple processing route consisting of cold rolling and recovery annealing was used to explore the possibility of tailoring the mechanical properties of this steel. The thermal stability of mechanically induced twin boundaries provided high retained yield strength after recovery annealing. In addition, recovery processes facilitated a significantly improved duc￾tility compared to the cold-rolled material. It was shown that the analysis of texture evolution during deformation and annealing can be used as an effective tool to optimize cold rolling degree and annealing conditions. A dislocation-based constitutive model was used in order to validate that the CuT texture component can be used as an indirect indicator for the evolution of the deformation twin density. Furthermore, simulation results identified recovery as the dominating softening mechanism under the applied annealing conditions. 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: TWIP steel; Texture; Microstructure; Recovery; Deformation twinning 1. Introduction Since the works of Gra¨ssel et al. [1,2] and Frommeyer et al. [3], high manganese twinning-induced plasticity (TWIP) steels have gained strong interest of the automobile industry for wide usage as structural components, and therefore have moved into the focus of worldwide steel research. These face-centered cubic, fully austenitic steels contain a high amount of manganese (15–30 wt.%) and typically, additions of C (0.05–1 wt.%), Al (0–3 wt.%) and/or Si (0–3 wt.%) [4]. This alloying concept results in a low stacking fault energy (SFE) in the range between 20 and 50 mJ m2 at room temperature [5–7]. As a con￾sequence of the low SFE, deformation mechanisms such as deformation twinning and shear banding become activated during deformation in addition to dislocation glide, result￾ing in high strain hardening rates and thus in high ductility and high strength with a typical ultimate tensile strength and elongation to fracture product of more than 50,000 MPa% [8–10]. A major shortcoming of TWIP steels in the fully recrys￾tallized, coarse-grained state is their relatively low yield strength, which is typically in the range between 200 and 400 MPa [2,11]. Although a low onset of plastic deforma￾tion facilitates energy-effective part shaping, it is detrimen￾tal when the material is applied in crash-relevant structural components (such as the A- or B-pillar in automobiles). http://dx.doi.org/10.1016/j.actamat.2014.07.068 1359-6454/ 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. ⇑ Corresponding author. Tel.: +49 241 8026877; fax: +49 241 8022301. E-mail address: haase@imm.rwth-aachen.de (C. Haase). www.elsevier.com/locate/actamat Available online at www.sciencedirect.com ScienceDirect Acta Materialia 80 (2014) 327–340

328 C.Haase et al.Acta Materialia 80 (2014)327-340 Nevertheless,pre-straining with medium or high degrees of main deformation texture components are retained due to plastic deformation can still be used to enhance the yield preservation of the deformed microstructure and slightly strength but this process reduces the remaining ductility strengthened due to a decrease of the dislocation density below the critical value that is required for the final shape [51-531.After complete recrystallization,a retained rolling forming.Apart from precipitation hardening due to micro- texture with a high degree of randomization was frequently alloying [12-14]or grain refinement by recrystallization observed in TWIP steels due to oriented nucleation and [15-191,a combination of pre-straining in the form of cold annealing twinning,respectively [51.54-61]. rolling and recovery annealing was proven to be a promis- In our previous study [25],the occurrence of recovery ing method to obtain significantly increased yield strength processes during annealing of a 30%cold-rolled Fe- along with appreciable elongation [20-26].The grain-scale 22.5Mn-1.2Al-0.3C TWIP steel was investigated.In the microstructure evolution during this processing procedure current study,the benefit of recovery annealing of the same is shown schematically in Fig.1.During cold rolling the material on its mechanical properties was analyzed.The microstructure is very effectively refined by the formation influence of initial cold rolling reductions and heat treat- of deformation twins,which is often referred to as a ment on microstructure evolution,yield strength-ductility dynamic Hall-Petch effect [27].A subsequent recovery combination and work-hardening capacity was addressed. annealing of the deformed sheet at heat treatment temper- Particular focus was put on the possibility of utilizing an atures and times below the onset of primary recrystalliza- analysis of the texture evolution during cold rolling and tion reduces the dislocation density and reestablishes annealing as a tool to predict the mechanical behavior of ductility.The deformation twins,which remain thermally the material investigated. stable during recovery annealing [20,21,28,29],act as strong barriers for dislocation movement in the same way 2.Applied methods as grain boundaries and facilitate a high retained yield strength[27,30-37刀. 2.1.Experimental In contrast to high and medium SFE materials,which form a Copper(Cu)-type rolling texture during cold roll- 2.1.1.Material chemistry and processing ing,low SFE materials,such as austenitic steel or a-brass, The chemical composition of the TWIP steel investi- are characterized by a Brass-type rolling texture after cold gated is given in Table 1.The corresponding SFE was cal- rolling [38-44].During texture evolution in TWIP steels, culated to be ~25 mJ musing a subregular solution pronounced (123](634)S and (110)(112)brass compo- thermodynamic model [62]. nents are readily formed due to dislocation glide.Upon The material was melted in a vacuum induction furnace further deformation,the (110)(100)Goss component, in argon atmosphere,cast into 100 kg ingots and subse- the (552)(115)copper twin (CuT)component and the quently homogenization-annealed at 1150C for 5 h in a y-fiber ((111)//ND)are strengthened [45-48].It was muffle furnace in order to reduce segregation.Afterwards. already proposed by Wassermann [49]that in silver and the initially 140 mm thick ingots were forged at 1150 C to brass the CuT component was formed as a consequence a height of 55 mm,followed by an additional homogeniza- of deformation twinning in (112)(111)Cu-oriented grains tion heat treatment at 1150C for 5 h.The forged slabs due to the preferable Schmid factor for twinning in these were then hot-rolled at 1150 C to a thickness of 2.4 mm. grains.This twinning effect causes the transition from the The material was then reheated between each of the 25 Cu-type to the Brass-type rolling texture with decreasing passes.A laboratory rolling mill was used to cold-roll the SFE.The development of the CuT component was also alloy at room temperature to thickness reductions in the observed in high manganese TWIP steels upon deforma- range between 10%and 80%.Finally,the samples were tion [45,46,50].In contrast to the other deformation texture subjected to isothermal recovery annealing in an air fur- components Cu,S,Brass and Goss,the CuT texture com- nace at annealing temperatures of 550 C or 630 C,which ponent is almost solely formed by deformation twinning. precluded recrystallization kinetics.In order to shorten the and therefore can be used as an indirect indication for an annealing time,recrystallization annealing was conducted increase of the volume fraction of deformation twins.Dur- at 700C for either 15 min (30%cold-rolled)or 10 min ing the recovery stage of a subsequent heat treatment,the (40%and 50%cold-rolled). prior to cold rolling after cold rolling after recovery annealing deformation twins Fig.1.Schematic diagram of the grain-scale microstructure evolution during the processing procedure applied

Nevertheless, pre-straining with medium or high degrees of plastic deformation can still be used to enhance the yield strength but this process reduces the remaining ductility below the critical value that is required for the final shape forming. Apart from precipitation hardening due to micro￾alloying [12–14] or grain refinement by recrystallization [15–19], a combination of pre-straining in the form of cold rolling and recovery annealing was proven to be a promis￾ing method to obtain significantly increased yield strength along with appreciable elongation [20–26]. The grain-scale microstructure evolution during this processing procedure is shown schematically in Fig. 1. During cold rolling the microstructure is very effectively refined by the formation of deformation twins, which is often referred to as a dynamic Hall–Petch effect [27]. A subsequent recovery annealing of the deformed sheet at heat treatment temper￾atures and times below the onset of primary recrystalliza￾tion reduces the dislocation density and reestablishes ductility. The deformation twins, which remain thermally stable during recovery annealing [20,21,28,29], act as strong barriers for dislocation movement in the same way as grain boundaries and facilitate a high retained yield strength [27,30–37]. In contrast to high and medium SFE materials, which form a Copper (Cu)-type rolling texture during cold roll￾ing, low SFE materials, such as austenitic steel or a-brass, are characterized by a Brass-type rolling texture after cold rolling [38–44]. During texture evolution in TWIP steels, pronounced {1 2 3}h634i S and {1 1 0}h112i brass compo￾nents are readily formed due to dislocation glide. Upon further deformation, the {1 1 0}h100i Goss component, the {5 5 2}h115i copper twin (CuT) component and the c-fiber (h111i//ND) are strengthened [45–48]. It was already proposed by Wassermann [49] that in silver and brass the CuT component was formed as a consequence of deformation twinning in {1 1 2}h111i Cu-oriented grains due to the preferable Schmid factor for twinning in these grains. This twinning effect causes the transition from the Cu-type to the Brass-type rolling texture with decreasing SFE. The development of the CuT component was also observed in high manganese TWIP steels upon deforma￾tion [45,46,50]. In contrast to the other deformation texture components Cu, S, Brass and Goss, the CuT texture com￾ponent is almost solely formed by deformation twinning, and therefore can be used as an indirect indication for an increase of the volume fraction of deformation twins. Dur￾ing the recovery stage of a subsequent heat treatment, the main deformation texture components are retained due to preservation of the deformed microstructure and slightly strengthened due to a decrease of the dislocation density [51–53]. After complete recrystallization, a retained rolling texture with a high degree of randomization was frequently observed in TWIP steels due to oriented nucleation and annealing twinning, respectively [51,54–61]. In our previous study [25], the occurrence of recovery processes during annealing of a 30% cold-rolled Fe– 22.5Mn–1.2Al–0.3C TWIP steel was investigated. In the current study, the benefit of recovery annealing of the same material on its mechanical properties was analyzed. The influence of initial cold rolling reductions and heat treat￾ment on microstructure evolution, yield strength–ductility combination and work-hardening capacity was addressed. Particular focus was put on the possibility of utilizing an analysis of the texture evolution during cold rolling and annealing as a tool to predict the mechanical behavior of the material investigated. 2. Applied methods 2.1. Experimental 2.1.1. Material chemistry and processing The chemical composition of the TWIP steel investi￾gated is given in Table 1. The corresponding SFE was cal￾culated to be 25 mJ m2 using a subregular solution thermodynamic model [62]. The material was melted in a vacuum induction furnace in argon atmosphere, cast into 100 kg ingots and subse￾quently homogenization-annealed at 1150 C for 5 h in a muffle furnace in order to reduce segregation. Afterwards, the initially 140 mm thick ingots were forged at 1150 C to a height of 55 mm, followed by an additional homogeniza￾tion heat treatment at 1150 C for 5 h. The forged slabs were then hot-rolled at 1150 C to a thickness of 2.4 mm. The material was then reheated between each of the 25 passes. A laboratory rolling mill was used to cold-roll the alloy at room temperature to thickness reductions in the range between 10% and 80%. Finally, the samples were subjected to isothermal recovery annealing in an air fur￾nace at annealing temperatures of 550 C or 630 C, which precluded recrystallization kinetics. In order to shorten the annealing time, recrystallization annealing was conducted at 700 C for either 15 min (30% cold-rolled) or 10 min (40% and 50% cold-rolled). Fig. 1. Schematic diagram of the grain-scale microstructure evolution during the processing procedure applied. 328 C. Haase et al. / Acta Materialia 80 (2014) 327–340

C.Haase et al.Acta Materialia 80 (2014)327-340 329 Table 1 orientation distribution functions (ODFs)were calculated Chemical composition of the investigated alloy. in MTEX.The volume fractions of the corresponding tex- Element Fe Mn Al Si N & ture components were calculated using a spread of 15 (wt.%) Bal. 0.32522.461.210.0410.0150.01 from their ideal orientation. The Vickers hardness (ASTM E384-10e2)of the cold- rolled and annealed samples was examined using a Shima- 2.1.2.Specimens and characterization techniques dzu HMV microhardness tester with a load of I kg(HV1). Specimens with the dimensions 10 mm x 12 mm(trans- Ten indents per sample were performed. verse direction (TD)and rolling direction (RD),respec- The mechanical properties of the material in deformed, tively)were cut from the cold-rolled and annealed sheets. recovered and recrystallized condition were evaluated by The samples were then mechanically ground with 800, uniaxial tensile tests at room temperature and a constant 1200,2400 and 4000 SiC grit paper and mechanically pol- strain rate of 10-3s-along the previous rolling direction ished using a 3 um and I um diamond suspension.For on a screw-driven Zwick 1484 mechanical testing device. scanning electron microscopy (SEM)and electron back- Flat bar tension specimens were used with a gauge length scatter diffraction(EBSD)the RD-ND (ND:normal direc- of 44 mm,gauge width of 12 mm,fillet radius of 20 mm tion)section was electropolished at room temperature for and variable thickness depending on the rolling degree. 20s at 22 V.For texture analysis and hardness testing, the middle layer of the RD-TD section was electropolished 2.2.Simulation setup for 2 min at 22 V.The used electrolyte consisted of 700 ml ethanol (C2HsOH),100 ml butyl glycol (CH1402)and The model used for simulations is a finite element model 78 ml perchloric acid(60%)(HCIO).The same electrolyte (FEM)implementation of the analytical model described in was used for preparing the samples for transmission elec- Ref.[66].To achieve this,a number of modifications simi- tron microscopy(TEM).In order to reveal the microstruc- lar to those described in Ref.[67]were made.First,the dis- ture by SEM,the specimens were additionally etched at location cell structure was neglected.Second,all evolution room temperature using a 2%Nital solution (95 ml equations were rewritten in a per slip/twin system formula- C2HsOH and 5 ml HNO3). tion.Since the twin volume fraction is in the focus of this SEM and EBSD analyses were performed in a LEO work,the evolution equations used to calculate the twin 1530 field emission gun SEM operated at 20 kV accelerat- volume fraction are recalled in the following.The equa- ing voltage and a working distance of 10 mm.EBSD map- tions for the dislocation part of the model are rather stan- pings were generated with a step size of 0.28 um.The HKL dard and can be found in Ref.[68]. Channel 5 software was utilized for data post-processing The twinning process is split into two parts,namely twin and removal of wild spikes and non-indexed points,taking nucleation and twin growth.For the nucleation,the model at least five neighbor points into account.Furthermore, of Mahajan and Chin [69]is adopted.It relies on the EBSD mappings were subdivided into subsets including reaction of two dislocations to form a twin nucleus: only recrystallized (RX),recovered (RC)or deformed 号(01i)+号(10i)=3×g(1I2).The twin nucleation rate (DEF)grains using an algorithm of the MATLAB-based (NB)per twin system B is calculated by multiplying the total MTEX package [63,64].The internal grain/subgrain mis- number density of potential twin nuclei per unit time (p), orientation was determined using the grain reference orien- by the probability that a sufficient stress concentration for tation (GROD-AO)technique,which takes the average grain/subgrain orientation as a reference.An internal grain the formation of the nucleus exists(p),and by the prob- misorientation threshold of RX<1.5<RC<6<DEF ability that one of those nuclei grows into a twin (p): was used [65].Grains containing fewer than 10 data points NB pPncsPtw (1) were disregarded. In order to prepare the TEM samples,the initial speci- In the FEM implementation p and p are calculated analogous to Ref.[66].However,as slip systems are differ- mens were ground to a thin layer of ~100 um,from which disks 3 mm in diameter were stamped out.The disks were entiated,pr can be calculated individually for each twin system B based on the slip activity of the slip systems then electropolished using a double jet Struers Tenupol-5 involved: with a voltage of 29 V at 15C.The RD-TD sections of the final specimens were analyzed in a JEOL JEM 2000 p22+2p1 (2) FX II analytical TEM operated at 200 kV. P= Lo The crystallographic texture was characterized by means where,2 are the shear rates on slip system / of X-ray pole figure measurements.Three incomplete (0-85)pole figures,(111),(200}and {220),were p,p are the dislocation densities on slip system acquired at the mid-layer of the sheet thickness on a Bruker and Lo is the length of the sessile partial dislocations D8 Advance diffractometer,equipped with a HI-STAR area detector,operating at 30 kV and 25 mA,using filtered IIn Ref.[68]a slightly different formulation is used for the twin volume iron radiation and polycapillary focusing optics.The evolution

2.1.2. Specimens and characterization techniques Specimens with the dimensions 10 mm  12 mm (trans￾verse direction (TD) and rolling direction (RD), respec￾tively) were cut from the cold-rolled and annealed sheets. The samples were then mechanically ground with 800, 1200, 2400 and 4000 SiC grit paper and mechanically pol￾ished using a 3 lm and 1 lm diamond suspension. For scanning electron microscopy (SEM) and electron back￾scatter diffraction (EBSD) the RD–ND (ND: normal direc￾tion) section was electropolished at room temperature for 20 s at 22 V. For texture analysis and hardness testing, the middle layer of the RD–TD section was electropolished for 2 min at 22 V. The used electrolyte consisted of 700 ml ethanol (C2H5OH), 100 ml butyl glycol (C6H14O2) and 78 ml perchloric acid (60%) (HClO4). The same electrolyte was used for preparing the samples for transmission elec￾tron microscopy (TEM). In order to reveal the microstruc￾ture by SEM, the specimens were additionally etched at room temperature using a 2% Nital solution (95 ml C2H5OH and 5 ml HNO3). SEM and EBSD analyses were performed in a LEO 1530 field emission gun SEM operated at 20 kV accelerat￾ing voltage and a working distance of 10 mm. EBSD map￾pings were generated with a step size of 0.28 lm. The HKL Channel 5 software was utilized for data post-processing and removal of wild spikes and non-indexed points, taking at least five neighbor points into account. Furthermore, EBSD mappings were subdivided into subsets including only recrystallized (RX), recovered (RC) or deformed (DEF) grains using an algorithm of the MATLAB-based MTEX package [63,64]. The internal grain/subgrain mis￾orientation was determined using the grain reference orien￾tation (GROD-AO) technique, which takes the average grain/subgrain orientation as a reference. An internal grain misorientation threshold of RX < 1.5 < RC < 6 < DEF was used [65]. Grains containing fewer than 10 data points were disregarded. In order to prepare the TEM samples, the initial speci￾mens were ground to a thin layer of 100 lm, from which disks 3 mm in diameter were stamped out. The disks were then electropolished using a double jet Struers Tenupol-5 with a voltage of 29 V at 15 C. The RD–TD sections of the final specimens were analyzed in a JEOL JEM 2000 FX II analytical TEM operated at 200 kV. The crystallographic texture was characterized by means of X-ray pole figure measurements. Three incomplete (0–85) pole figures, {1 1 1}, {2 0 0} and {2 2 0}, were acquired at the mid-layer of the sheet thickness on a Bruker D8 Advance diffractometer, equipped with a HI-STAR area detector, operating at 30 kV and 25 mA, using filtered iron radiation and polycapillary focusing optics. The orientation distribution functions (ODFs) were calculated in MTEX. The volume fractions of the corresponding tex￾ture components were calculated using a spread of 15 from their ideal orientation. The Vickers hardness (ASTM E384-10e2) of the cold￾rolled and annealed samples was examined using a Shima￾dzu HMV microhardness tester with a load of 1 kg (HV1). Ten indents per sample were performed. The mechanical properties of the material in deformed, recovered and recrystallized condition were evaluated by uniaxial tensile tests at room temperature and a constant strain rate of 103 s 1 along the previous rolling direction on a screw-driven Zwick 1484 mechanical testing device. Flat bar tension specimens were used with a gauge length of 44 mm, gauge width of 12 mm, fillet radius of 20 mm and variable thickness depending on the rolling degree. 2.2. Simulation setup The model used for simulations is a finite element model (FEM) implementation of the analytical model described in Ref. [66]. To achieve this, a number of modifications simi￾lar to those described in Ref. [67] were made. First, the dis￾location cell structure was neglected. Second, all evolution equations were rewritten in a per slip/twin system formula￾tion. Since the twin volume fraction is in the focus of this work, the evolution equations used to calculate the twin volume fraction are recalled in the following. The equa￾tions for the dislocation part of the model are rather stan￾dard and can be found in Ref. [68]. 1 The twinning process is split into two parts, namely twin nucleation and twin growth. For the nucleation, the model of Mahajan and Chin [69] is adopted. It relies on the reaction of two dislocations to form a twin nucleus: a 2 h0 11i þ a 2 h1 01i ¼ 3  a 6 h1 12i. The twin nucleation rate (N_ b) per twin system b is calculated by multiplying the total number density of potential twin nuclei per unit time (pb st), by the probability that a sufficient stress concentration for the formation of the nucleus exists (ptw), and by the prob￾ability that one of those nuclei grows into a twin (pncs): N_ b ¼ pb stpncsptw ð1Þ In the FEM implementation ptw and pncs are calculated analogous to Ref. [66]. However, as slip systems are differ￾entiated, pst can be calculated individually for each twin system b based on the slip activity of the slip systems involved: pb st ¼ c_ a1qa2 þ c_ a2qa1 L0 ð2Þ where c_ a1, c_ a2 are the shear rates on slip system a1=a2, qa1, qa2 are the dislocation densities on slip system a1=a2 and L0 is the length of the sessile partial dislocations Table 1 Chemical composition of the investigated alloy. Element Fe C Mn Al Si N P (wt.%) Bal. 0.325 22.46 1.21 0.041 0.015 0.01 1 In Ref. [68] a slightly different formulation is used for the twin volume evolution. C. Haase et al. / Acta Materialia 80 (2014) 327–340 329

330 C.Haase et al.Acta Materialia 80 (2014)327-340 forming the twin nucleus.In order to make the nucleus required to introduce a high fraction of deformation twins, grow,a critical stresswhere sr is the SFE and thus to attain an effective reduction of the mean free and bs is the Burgers vector of the Shockley partial,has glide distance of dislocations.Subsequent recovery anneal- to be overcome [66].Since energy is always gained during ing offers the possibility of regaining ductility and must be grain growth,it is assumed that twins grow instantaneously carried out for a period of time that is,on the one hand, until they encounter an obstacle such as a grain boundary long enough to initiate recovery processes and,on the other or a twin on a non-coplanar twin system.A new twin is hand,short enough to impede recrystallization [25].Both considered to be disk-shaped,where the radial dimension the necessary degree of deformation by cold rolling and is based on the average twin spacing.The twin volume is the right heat treatment regime can be determined by anal- then given by: ysis of texture evolution.In the following we will show the Vw-jep efficiency of texture analysis for choosing the optimal pro- (3) cessing parameters for the aforementioned approach. where t is the average twin spacing and e is the average twin width. 3.1.2.Degree of reduction by cold rolling Finally the twin volume fraction evolution is calculated The texture evolution of the investigated material after by the product of the nucleation rate with the volume that cold rolling in the range between 10%and 80%thickness a new twin occupies,and the untwinned volume: reduction is illustrated by 2=45 ODF sections in Fig.2.A schematic illustration of the main texture compo- fnm=(1-fm)）NVnw (4) nents observed in the 2=45 ODF section and the corre- Further details such as the incorporation of temperature sponding definitions of these components are given in are given in Refs.[66,68]. Fig.2 (top left corner)and Table 2,respectively.Fig.3 depicts the calculated volume fractions of selected texture 3.Results components.At low rolling degrees of 10-20%,the compo- nents Cu,S,Goss and Brass developed and formed a weak 3.1.Determination of processing parameters by texture Cu-type texture.Increased rolling reduction (30-50%) analysis facilitated a shift of the maximum intensity of the ODF from the Brass component into a position between Brass 3.1.1.Required deformation and heat treatment for and Goss ({110)(115)G/B)along the a-fiber (Fig.2). beneficial mechanical properties Furthermore,a spread from the Goss towards the CuT In order to achieve a TWIP steel with high yield strength component along the r-fiber and a weakening of the Cu along with high ductility,a suitable level of deformation is component were observed.As a consequence,fractions of 1 Φ Cu y-fiber 10 E a-fiber CuT 10% 20% 30% 40% 50% 60% 70% 80% Fig.2.Texture evolution of the investigated Fe-23Mn-1.5Al-0.3C steel during cold rolling,ODF sections at 2=45

forming the twin nucleus. In order to make the nucleus grow, a critical stress sc ¼ cSF 3bs þ 3Gbs L0 , where cSF is the SFE and bS is the Burgers vector of the Shockley partial, has to be overcome [66]. Since energy is always gained during grain growth, it is assumed that twins grow instantaneously until they encounter an obstacle such as a grain boundary or a twin on a non-coplanar twin system. A new twin is considered to be disk-shaped, where the radial dimension is based on the average twin spacing. The twin volume is then given by: V tw ¼ p 4 et2 ð3Þ where t is the average twin spacing and e is the average twin width. Finally the twin volume fraction evolution is calculated by the product of the nucleation rate with the volume that a new twin occupies, and the untwinned volume: _ f tw ¼ ð1 f twÞN V_ tw ð4Þ Further details such as the incorporation of temperature are given in Refs. [66,68]. 3. Results 3.1. Determination of processing parameters by texture analysis 3.1.1. Required deformation and heat treatment for beneficial mechanical properties In order to achieve a TWIP steel with high yield strength along with high ductility, a suitable level of deformation is required to introduce a high fraction of deformation twins, and thus to attain an effective reduction of the mean free glide distance of dislocations. Subsequent recovery anneal￾ing offers the possibility of regaining ductility and must be carried out for a period of time that is, on the one hand, long enough to initiate recovery processes and, on the other hand, short enough to impede recrystallization [25]. Both the necessary degree of deformation by cold rolling and the right heat treatment regime can be determined by anal￾ysis of texture evolution. In the following we will show the efficiency of texture analysis for choosing the optimal pro￾cessing parameters for the aforementioned approach. 3.1.2. Degree of reduction by cold rolling The texture evolution of the investigated material after cold rolling in the range between 10% and 80% thickness reduction is illustrated by u2 = 45 ODF sections in Fig. 2. A schematic illustration of the main texture compo￾nents observed in the u2 = 45 ODF section and the corre￾sponding definitions of these components are given in Fig. 2 (top left corner) and Table 2, respectively. Fig. 3 depicts the calculated volume fractions of selected texture components. At low rolling degrees of 10–20%, the compo￾nents Cu, S, Goss and Brass developed and formed a weak Cu-type texture. Increased rolling reduction (30–50%) facilitated a shift of the maximum intensity of the ODF from the Brass component into a position between Brass and Goss ({1 1 0}h115i G/B) along the a-fiber (Fig. 2). Furthermore, a spread from the Goss towards the CuT component along the s-fiber and a weakening of the Cu component were observed. As a consequence, fractions of Fig. 2. Texture evolution of the investigated Fe–23Mn–1.5Al–0.3C steel during cold rolling, ODF sections at u2 = 45. 330 C. Haase et al. / Acta Materialia 80 (2014) 327–340

C.Haase et al.Acta Materialia 80 (2014)327-340 331 Table 2 Definition of texture components illustrated in Fig.2. Component Symbol Miller indices Euler angles (,2) Fiber Brass(B) △ {1101(112) (55,90,45) x.B Goss (G) 口 {1101(100 90,90.45) ，t Cube (C) O {0011100 (45,0.45) ◇ ◆ {1111110 0/60,55.45) 《 0 {111}112) (30/90,55,45 Copper(Cu) 7 {112}111) (90,35,45) B,t CopperTwin (CuT) V {552}115》 90,74,45) x-Fiber (110)parallel to ND B-Fiber (110)tilted 60 from ND towards RD t-Fiber (110)parallel TD y-Fiber (111)parallel ND 10% rolling degrees of 60-80%,the y-fiber components,E and 30 20% 30% F,developed due to successive rotation of twin-matrix 40% lamellae into the rolling plane and severe shear banding. 50% Furthermore,the decrease of the Cu component and the 20 60% 70% stagnation of the CuT component reflected further defor- mation twinning in Cu-oriented grains and,on the other OA] 80% hand,elimination of the twin containing grains by shear 10 bands,which prohibited a further increase of the CuT tex- ture component. Shear bands,as microstructure heterogeneities of high localized shear deformation [70],are hardly capable of CuT Copper Goss Brass E+F Random accommodating further strain during deformation,and therefore are undesirable microstructure constituents in Fig.3.Volume fractions of the main texture components developed during cold rolling. structural components that require uniform elongation. In addition,shear bands act as preferential nucleation sites for primary recrystallization due to the high,localized the Goss and CuT components increased,whereas those of stored energy in these bands,and thus accelerate the onset the Cu as well as the S components decreased(cf.Fig.3). of recrystallization [71,72],which then results in a decrease As seen in Fig.2,with further rolling reduction,at of the yield strength.Furthermore,the elimination of twin- 60-80%deformation,a weak y-fiber consisting of the matrix lamellae,and thus of the necessary twin boundaries (111)(011)E and (111(112)F components developed. by shear bands [73],is also an undesired effect for the The increase of the volume fraction of the E+F compo- approach applied in this study.Due to the aforementioned nents was accompanied by a stagnation of CuT and a fur- effects,shear banding should be avoided during cold rolling ther decrease of the Cu component as well as an increase of when applying the processing route introduced in this the volume fractions of the Goss and randomly oriented work.Since the volume fraction of shear bands is directly grains(cf.Fig.3). related to the fraction of the E and F texture components, The relation between microstructure and texture evolu- samples with the highest fraction of the E and F texture tion in a similar cold-rolled TWIP steel is described in components,namely the 60%,70%and 80%cold-rolled detail in our previous paper [50].Most likely due to the samples,were excluded.In addition to the absence of shear similar chemical composition and the same SFE value of bands in the microstructure,a high density of deformation ~25 mJm-2,similar features of the deformed structure twins is required in order to attain a high yield strength and deformation mechanisms,as in the steel studied previ- after recovery annealing.Therefore,the deformed speci- ously [50],were found to relate to the specific texture com- mens with the highest twin density,which corresponds to ponents in the Fe-23Mn-1.5Al-0.3C steel investigated in the largest volume fraction of the CuT component,were the current work.Deformation at low rolling degrees of chosen,i.e.the 30%,40%and 50%cold-rolled samples 10-20%was dominated by dislocation glide,and thus (highlighted in Fig.3). resulted in a Cu-type texture.At medium rolling degrees, i.e.30-50%thickness reduction,the contribution of defor- 3.1.3.Annealing time determined by texture analysis mation twinning to the accommodation of strain increased The texture evolution in the investigated material after significantly.Consequently,the decrease of the volume 30%,40%and 50%reduction by rolling during annealing fraction of the Cu component was accompanied by an is shown by selected ODF sections at 2=45 in Fig.4. increase of the E3 twin related CuT component.At high Even though the texture intensities and indices(T)are very

the Goss and CuT components increased, whereas those of the Cu as well as the S components decreased (cf. Fig. 3). As seen in Fig. 2, with further rolling reduction, at 60–80% deformation, a weak c-fiber consisting of the {1 1 1}h011i E and {1 1 1}h112i F components developed. The increase of the volume fraction of the E + F compo￾nents was accompanied by a stagnation of CuT and a fur￾ther decrease of the Cu component as well as an increase of the volume fractions of the Goss and randomly oriented grains (cf. Fig. 3). The relation between microstructure and texture evolu￾tion in a similar cold-rolled TWIP steel is described in detail in our previous paper [50]. Most likely due to the similar chemical composition and the same SFE value of 25 mJm2 , similar features of the deformed structure and deformation mechanisms, as in the steel studied previ￾ously [50], were found to relate to the specific texture com￾ponents in the Fe–23Mn–1.5Al–0.3C steel investigated in the current work. Deformation at low rolling degrees of 10–20% was dominated by dislocation glide, and thus resulted in a Cu-type texture. At medium rolling degrees, i.e. 30–50% thickness reduction, the contribution of defor￾mation twinning to the accommodation of strain increased significantly. Consequently, the decrease of the volume fraction of the Cu component was accompanied by an increase of the R3 twin related CuT component. At high rolling degrees of 60–80%, the c-fiber components, E and F, developed due to successive rotation of twin-matrix lamellae into the rolling plane and severe shear banding. Furthermore, the decrease of the Cu component and the stagnation of the CuT component reflected further defor￾mation twinning in Cu-oriented grains and, on the other hand, elimination of the twin containing grains by shear bands, which prohibited a further increase of the CuT tex￾ture component. Shear bands, as microstructure heterogeneities of high localized shear deformation [70], are hardly capable of accommodating further strain during deformation, and therefore are undesirable microstructure constituents in structural components that require uniform elongation. In addition, shear bands act as preferential nucleation sites for primary recrystallization due to the high, localized stored energy in these bands, and thus accelerate the onset of recrystallization [71,72], which then results in a decrease of the yield strength. Furthermore, the elimination of twin￾matrix lamellae, and thus of the necessary twin boundaries by shear bands [73], is also an undesired effect for the approach applied in this study. Due to the aforementioned effects, shear banding should be avoided during cold rolling when applying the processing route introduced in this work. Since the volume fraction of shear bands is directly related to the fraction of the E and F texture components, samples with the highest fraction of the E and F texture components, namely the 60%, 70% and 80% cold-rolled samples, were excluded. In addition to the absence of shear bands in the microstructure, a high density of deformation twins is required in order to attain a high yield strength after recovery annealing. Therefore, the deformed speci￾mens with the highest twin density, which corresponds to the largest volume fraction of the CuT component, were chosen, i.e. the 30%, 40% and 50% cold-rolled samples (highlighted in Fig. 3). 3.1.3. Annealing time determined by texture analysis The texture evolution in the investigated material after 30%, 40% and 50% reduction by rolling during annealing is shown by selected ODF sections at u2 = 45 in Fig. 4. Even though the texture intensities and indices (T) are very Table 2 Definition of texture components illustrated in Fig. 2. Component Symbol Miller indices Euler angles (u1,U,u2) Fiber Brass (B) {1 1 0}h112i (55, 90, 45) a, b Goss (G) {1 1 0}h100i (90, 90, 45) a, s Cube (C) {0 0 1}h100i (45, 0, 45) / E {1 1 1}h110i (0/60, 55, 45) c F {1 1 1}h112i (30/90, 55, 45) c Copper (Cu) {1 1 2}h111i (90, 35, 45) b, s CopperTwin (CuT) {5 5 2}h115i (90, 74, 45) s a-Fiber h110i parallel to ND b-Fiber h110i tilted 60 from ND towards RD s-Fiber h110i parallel TD c-Fiber h111i parallel ND CuT Copper Goss S Brass E+F Random 0 10 20 30 f [vol.%] 10% 20% 30% 40% 50% 60% 70% 80% Fig. 3. Volume fractions of the main texture components developed during cold rolling. C. Haase et al. / Acta Materialia 80 (2014) 327–340 331

332 C.Haase et al.Acta Materialia 80 (2014)327-340 1+ cold rolled T=3.1 T=3.2 T=3.2 recovered 11 10 T=3.2 T=3.3 T=3.3 early RX X50% T=2.2 T=2.9 T=2.6 complete RX X=100% T=1.6 T=2.1 T=2.0 (a) (b) (c) Fig.4.ODF sections at 2=45 of the (a)30%cold-rolled,(b)40%cold-rolled and (c)50%cold-rolled material after cold rolling,recovery annealing. partial and complete recrystallization annealing. 40 low,a clear trend in texture development during annealing 50%CR is observed,which is consistent with the behavior of similar 2 min TWIP steels [51.54].During the recovery stage,the texture 15 min 30 intensity was slightly strengthened,which was accompa- 30 min 1h nied by an increase of the texture indices.During further 2h (recrystallization)annealing,the main texture components 20 8h were retained due to oriented nucleation,whereas addi- 24h tional annealing twinning facilitated evolution of a com- plete a-fiber [51]as well as further randomization [56.74] (cf.Fig.4).This behavior can also be seen from the volume fractions of the main texture components during annealing, which is exemplarily shown for the 50%cold-rolled mate- rial annealed at 550C(cf.Fig.5).As a result of disloca- CuT Copper Goss Brass Random tions'recovery,the volume fraction of randomly oriented Fig.5.Volume fractions of the main texture components of the 50%cold- grains decreased,whereas the fractions of the main rolled material during annealing at 550 C

low, a clear trend in texture development during annealing is observed, which is consistent with the behavior of similar TWIP steels [51,54]. During the recovery stage, the texture intensity was slightly strengthened, which was accompa￾nied by an increase of the texture indices. During further (recrystallization) annealing, the main texture components were retained due to oriented nucleation, whereas addi￾tional annealing twinning facilitated evolution of a com￾plete a-fiber [51] as well as further randomization [56,74] (cf. Fig. 4). This behavior can also be seen from the volume fractions of the main texture components during annealing, which is exemplarily shown for the 50% cold-rolled mate￾rial annealed at 550 C (cf. Fig. 5). As a result of disloca￾tions’ recovery, the volume fraction of randomly oriented grains decreased, whereas the fractions of the main Fig. 4. ODF sections at u2 = 45 of the (a) 30% cold-rolled, (b) 40% cold-rolled and (c) 50% cold-rolled material after cold rolling, recovery annealing, partial and complete recrystallization annealing. CuT Copper Goss S Brass Random 0 10 20 30 40 f [vol.%] 50% CR 2 min 15 min 30 min 1 h 2 h 8 h 24 h Fig. 5. Volume fractions of the main texture components of the 50% cold￾rolled material during annealing at 550 C. 332 C. Haase et al. / Acta Materialia 80 (2014) 327–340

C.Haase et al.Acta Materialia 80 (2014)327-340 333 deformation texture components.such as CuT,S and and/or deformation twins of primary,respectively primary Brass,increased.As indicated in Fig.4,this was accompa- and secondary twin systems(cf.Fig.7a).In specimens with nied by the slight strengthening of the texture index.It 40%and 50%reduction the grains were further elongated must be noted.however,that the measured change in vol- and the density of the aforementioned microstructural ume fraction of the deformation texture components does features increased.After a deformation of 40%first micro- not correspond to an increased volume fraction of grains shear bands developed in individual grains(cf.Fig.7c and with specific orientations,but is rather a consequence of e).The microstructures of the cold-rolled material after less distorted lattice planes due to lower dislocation den- recovery annealing with annealing times determined by tex- sity,which in turn resulted in less scattering of X-rays ture analysis are illustrated in Fig.7b,d and f.It was found and thus in a higher detected intensity.With progressive that the deformed grains with their elongated shape and the recrystallization,the volume fraction of randomly oriented microstructural features observed in the cold-rolled state grains increased and the deformation texture components were still present after recovery annealing.Since the grain decreased.Therefore,annealing for 30 min at 550C refinement effect due to the mechanically induced twin (highlighted in Fig.5),after which the fraction of the boundaries is essential for the efficiency of the processing CuT component was maximum and the fractions of the route applied in this study,their thermal stability during main deformation texture components still remained at recovery annealing was analyzed qualitatively using TEM the highest level,was chosen as optimal heat treatment (Fig.8).Regardless of the cold rolling degree and anneal- parameters for the recovery annealing of the 50%cold- ing temperature applied,the deformation twins introduced rolled samples.The same approach was applied to the during cold rolling were found to be thermally stable in the 30%and 40%cold-rolled material and resulted in recovery non-recrystallized grains(cf.Fig.Sa-d).Even though pri- annealing parameters of 630C/10 min and 550C/1 h, mary recrystallization was locally initiated at the annealing respectively. temperatures chosen,the deformation twins were main- In order to control the reliability of the annealing tained in the microstructure unless they were consumed parameters determined by texture analysis,hardness mea- by the growth of recrystallized grains into the deformed surements after annealing were conducted and the micro- matrix,as illustrated in Fig.8c. structure prior to and after recovery annealing was In addition to the thermal stability of deformation characterized.The hardness development of the 30%. twins,the occurrence of recovery processes during recovery 40%and 50%cold-rolled material with annealing time is annealing was also checked using EBSD measurements,as shown in Fig.6.As seen,recovery-annealing times deter- described in detail in Ref.[25].The EBSD mapping of the mined by texture analysis for all three specimens (encircled 40%cold-rolled and recovery-annealed material is shown data points)were just at the beginning of the steep decrease in Fig.9 as inverse pole figure(IPF)mapping with orienta- of hardness,which can be associated with the onset of tions parallel to the sheet normal direction (ND).Due to primary recrystallization. the varying intragranular misorientations of RX and non-RX grains,the EBSD data could be broken down into 3.2.Microstructure evolution during recovery annealing RX(cf.Fig.9c)and non-RX grains by applying an intra- granular misorientation threshold [25,65].Furthermore, After cold rolling to a thickness reduction of 30%,the the non-RX grains could be further subdivided into RC microstructure consisted of grains elongated along the roll- (Fig.9a)and DEF (Fig.9b)grains using the same ing direction and contained slip lines,deformation bands approach,where DEF grains can be considered as slightly recovered grains with a high residual dislocation density, whereas RC grains underwent stronger recovery than 450 DEF grains.In contrast to the typical equiaxed shape of the RX grains(cf.Fig.9c),both RC and DEF grains in 400 [LAH] Fig.9a and b,respectively,revealed an elongated grain 350 morphology,inherited from their previous plastic deforma- tion during cold rolling.The corresponding microtexture of 300 the RX,RC and DEF grains is shown in Fig.9d-f.The microtexture of the DEF and RC grains was dominated 250 by the main texture components Goss,Brass and Cu, 200 ■一30%-630°C whereas the RX grains revealed the same components with 。-40%-550°C lower intensity and further formation of widespread orien- 150 +50%-550°℃ tations leading to texture randomization.The grain bound- 0.1 0 1 00 100010000100000 ary misorientation profiles of the DEF,RC and RX grains of the deconvoluted EBSD data of the material after 30%. Time [s] 40%and 50%cold rolling and recovery annealing are Fig.6.Microhardness evolution of the 30%,40%and 50%cold-rolled shown in Fig.10a-c.The RX grains of the three different material after annealing at 550C and 630C for various annealing times. EBSD mappings were characterized by two peaks at 390

deformation texture components, such as CuT, S and Brass, increased. As indicated in Fig. 4, this was accompa￾nied by the slight strengthening of the texture index. It must be noted, however, that the measured change in vol￾ume fraction of the deformation texture components does not correspond to an increased volume fraction of grains with specific orientations, but is rather a consequence of less distorted lattice planes due to lower dislocation den￾sity, which in turn resulted in less scattering of X-rays and thus in a higher detected intensity. With progressive recrystallization, the volume fraction of randomly oriented grains increased and the deformation texture components decreased. Therefore, annealing for 30 min at 550 C (highlighted in Fig. 5), after which the fraction of the CuT component was maximum and the fractions of the main deformation texture components still remained at the highest level, was chosen as optimal heat treatment parameters for the recovery annealing of the 50% cold￾rolled samples. The same approach was applied to the 30% and 40% cold-rolled material and resulted in recovery annealing parameters of 630 C/10 min and 550 C/1 h, respectively. In order to control the reliability of the annealing parameters determined by texture analysis, hardness mea￾surements after annealing were conducted and the micro￾structure prior to and after recovery annealing was characterized. The hardness development of the 30%, 40% and 50% cold-rolled material with annealing time is shown in Fig. 6. As seen, recovery-annealing times deter￾mined by texture analysis for all three specimens (encircled data points) were just at the beginning of the steep decrease of hardness, which can be associated with the onset of primary recrystallization. 3.2. Microstructure evolution during recovery annealing After cold rolling to a thickness reduction of 30%, the microstructure consisted of grains elongated along the roll￾ing direction and contained slip lines, deformation bands and/or deformation twins of primary, respectively primary and secondary twin systems (cf. Fig. 7a). In specimens with 40% and 50% reduction the grains were further elongated and the density of the aforementioned microstructural features increased. After a deformation of 40% first micro￾shear bands developed in individual grains (cf. Fig. 7c and e). The microstructures of the cold-rolled material after recovery annealing with annealing times determined by tex￾ture analysis are illustrated in Fig. 7b, d and f. It was found that the deformed grains with their elongated shape and the microstructural features observed in the cold-rolled state were still present after recovery annealing. Since the grain refinement effect due to the mechanically induced twin boundaries is essential for the efficiency of the processing route applied in this study, their thermal stability during recovery annealing was analyzed qualitatively using TEM (Fig. 8). Regardless of the cold rolling degree and anneal￾ing temperature applied, the deformation twins introduced during cold rolling were found to be thermally stable in the non-recrystallized grains (cf. Fig. 8a–d). Even though pri￾mary recrystallization was locally initiated at the annealing temperatures chosen, the deformation twins were main￾tained in the microstructure unless they were consumed by the growth of recrystallized grains into the deformed matrix, as illustrated in Fig. 8c. In addition to the thermal stability of deformation twins, the occurrence of recovery processes during recovery annealing was also checked using EBSD measurements, as described in detail in Ref. [25]. The EBSD mapping of the 40% cold-rolled and recovery-annealed material is shown in Fig. 9 as inverse pole figure (IPF) mapping with orienta￾tions parallel to the sheet normal direction (ND). Due to the varying intragranular misorientations of RX and non-RX grains, the EBSD data could be broken down into RX (cf. Fig. 9c) and non-RX grains by applying an intra￾granular misorientation threshold [25,65]. Furthermore, the non-RX grains could be further subdivided into RC (Fig. 9a) and DEF (Fig. 9b) grains using the same approach, where DEF grains can be considered as slightly recovered grains with a high residual dislocation density, whereas RC grains underwent stronger recovery than DEF grains. In contrast to the typical equiaxed shape of the RX grains (cf. Fig. 9c), both RC and DEF grains in Fig. 9a and b, respectively, revealed an elongated grain morphology, inherited from their previous plastic deforma￾tion during cold rolling. The corresponding microtexture of the RX, RC and DEF grains is shown in Fig. 9d–f. The microtexture of the DEF and RC grains was dominated by the main texture components Goss, Brass and Cu, whereas the RX grains revealed the same components with lower intensity and further formation of widespread orien￾tations leading to texture randomization. The grain bound￾ary misorientation profiles of the DEF, RC and RX grains of the deconvoluted EBSD data of the material after 30%, 40% and 50% cold rolling and recovery annealing are shown in Fig. 10a–c. The RX grains of the three different EBSD mappings were characterized by two peaks at 39 0.1 1 10 100 1000 10000 100000 150 200 250 300 350 400 450 Vickers hardness [HV1] Time [s] 30% - 630°C 40% - 550°C 50% - 550°C Fig. 6. Microhardness evolution of the 30%, 40% and 50% cold-rolled material after annealing at 550 C and 630 C for various annealing times. C. Haase et al. / Acta Materialia 80 (2014) 327–340 333

334 C.Haase et al.Acta Materialia 80 (2014)327-340 RD (a) (b) ND twins 30 um 30 um 30 um 30 um (e) (⑤ SB 30μm 30μm Fig.7.SEM micrographs of the material after:(a)30%cold rolling,(b)30%cold rolling recovery annealing,(c)40%cold rolling,(d)40%cold rolling recovery annealing,(e)50%cold rolling and (f)50%cold rolling recovery annealing. and60°，which indicated38.9(10l)Σ9and60°(111）3 obtained discrepancy between fractions of 3 boundaries in CSL boundaries,respectively.In contrast to the low frac- DEF and RC grains can be understood from the fact that in tion of low angle grain boundaries (<15)of the RX highly twinned grains due to the additional accommodation grains,the fraction of these boundaries was significantly of plastic strain by deformation twinning the dislocation higher for the DEF and RC grains.Furthermore,the low density remained at a lower level compared to grains angle grain boundary fraction of the DEF and RC grains containing a lower fraction of deformation twins.As a was found to increase with increasing rolling degree,and consequence,the intragranular misorientation caused by thus indicated a higher residual dislocation density with dislocations was smaller in grains with high fraction of higher rolling reduction.The occurrence of recovery deformation twins(with respective E3 boundaries)and thus processes is further supported by the lower fraction of a relatively high number of these grains was detected as RC low angle grain boundaries in the interior of the RC grains grains. compared to the DEF grains. Finally,it is stressed that the capability of the EBSD 3.3.Mechanical properties technique to resolve deformation twins with nanoscale width and separation distance quantitatively is strongly The results of uniaxial tensile tests of the material after limited.which explains the lower fraction of X3 boundaries cold rolling,recovery annealing and recrystallization in the DEF and RC grains compared to the RX grains. annealing are illustrated in Fig.11.With increasing rolling Moreover,the deteriorating indexing rate with increasing reduction the yield strength increased continuously up to deformation level lowered the detected fraction of E3 1220 MPa after 50%deformation,whereas the elongation boundaries in the DEF and RC grains significantly(more decreased dramatically,as evident from the engineering in DEF than in RC grains),and thus was not representative stress-strain curve in Fig.Ila.Regardless of the previous for the true density of deformation twins.Furthermore,the rolling degree,recovery annealing resulted in a decreased

and 60, which indicated 38.9h101i R9 and 60h111i R3 CSL boundaries, respectively. In contrast to the low frac￾tion of low angle grain boundaries (H < 15) of the RX grains, the fraction of these boundaries was significantly higher for the DEF and RC grains. Furthermore, the low angle grain boundary fraction of the DEF and RC grains was found to increase with increasing rolling degree, and thus indicated a higher residual dislocation density with higher rolling reduction. The occurrence of recovery processes is further supported by the lower fraction of low angle grain boundaries in the interior of the RC grains compared to the DEF grains. Finally, it is stressed that the capability of the EBSD technique to resolve deformation twins with nanoscale width and separation distance quantitatively is strongly limited, which explains the lower fraction of R3 boundaries in the DEF and RC grains compared to the RX grains. Moreover, the deteriorating indexing rate with increasing deformation level lowered the detected fraction of R3 boundaries in the DEF and RC grains significantly (more in DEF than in RC grains), and thus was not representative for the true density of deformation twins. Furthermore, the obtained discrepancy between fractions of R3 boundaries in DEF and RC grains can be understood from the fact that in highly twinned grains due to the additional accommodation of plastic strain by deformation twinning the dislocation density remained at a lower level compared to grains containing a lower fraction of deformation twins. As a consequence, the intragranular misorientation caused by dislocations was smaller in grains with high fraction of deformation twins (with respective R3 boundaries) and thus a relatively high number of these grains was detected as RC grains. 3.3. Mechanical properties The results of uniaxial tensile tests of the material after cold rolling, recovery annealing and recrystallization annealing are illustrated in Fig. 11. With increasing rolling reduction the yield strength increased continuously up to 1220 MPa after 50% deformation, whereas the elongation decreased dramatically, as evident from the engineering stress–strain curve in Fig. 11a. Regardless of the previous rolling degree, recovery annealing resulted in a decreased RD ND (a) 30 µm 30 µm (f) 30 µm (e) 30 µm (d) 30 µm (c) 30 µm (b) twins SB Fig. 7. SEM micrographs of the material after: (a) 30% cold rolling, (b) 30% cold rolling + recovery annealing, (c) 40% cold rolling, (d) 40% cold rolling + recovery annealing, (e) 50% cold rolling and (f) 50% cold rolling + recovery annealing. 334 C. Haase et al. / Acta Materialia 80 (2014) 327–340

C.Haase et al.Acta Materialia 80 (2014)327-340 335 RD (a) b TD win R 1um 600nm 1 um 4 (d) (e) matrix twin 8 (000) 1um Fig.8.TEM bright field micrographs of the material after:(a)30%cold rolling+recovery annealing.(b,c)40%cold rolling+recovery annealing,(d) 50%cold rolling recovery annealing;(e)selected area diffraction (SAD)pattern of(d)with [0 1 1],zone axis. (a) (61 ND 50μm 50μm >RD (d) (e) 111} 50m 001 101 Fig.9.IPF mappings of the 40%cold-rolled recovery-annealed material broken down into:(a)DEF grains,(b)RC grains,(c)RX grains and (d-f)the corresponding ODF sections at 2=45 of (a-c)(levels:1.0.2.0,3.0,4.0,5.0,6.0.7.0.8.0)

1 µm (a) RD TD 600 nm (b) 1 µm RX (c) 1 µm (d) twin matrix (000) (e) Fig. 8. TEM bright field micrographs of the material after: (a) 30% cold rolling + recovery annealing, (b, c) 40% cold rolling + recovery annealing, (d) 50% cold rolling + recovery annealing; (e) selected area diffraction (SAD) pattern of (d) with [0 1 1]c zone axis. Fig. 9. IPF mappings of the 40% cold-rolled + recovery-annealed material broken down into: (a) DEF grains, (b) RC grains, (c) RX grains and (d–f) the corresponding ODF sections at u2 = 45 of (a–c) (levels: 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0). C. Haase et al. / Acta Materialia 80 (2014) 327–340 335

336 C.Haase et al.Acta Materialia 80 (2014)327-340 (a) 0.5 (a) 1400 deformed grains 0.4 -recovered grains 1200 -recrystallized grains 1000 0.3 800 0.2 600 0.1 400 200 (b) 0.4 0 10 2030405060 70 Engineering strain,e[%] 0.3 —30%CR—30%CR+RC —30%CR+RX 02 ----40%CR----40%CR+RC----40%CR+RX ---50%CR---50%CR+RC---50%CR+RX 0.1 (b) 4000 (c) 3000 0.4 0.3 2000 g 0.2 1000 0.1 0.0 0 10 20 30 40 50 60 0.0 0.1 0.2 0.3 0.4 0.5 Grain boundary misorientation angle[] True strain,2 Fig.10.Grain boundary misorientation profiles of the DEF,RC and RX grains after:(a)30%.(b)40%and (c)50%reduction by cold rolling and Fig.11.(a)Engineering stress-strain curves and(b)true stress-true strain subsequent recovery annealing (630C/10 min,550C/1 h and 550C/ curves (dotted lines)and work-hardening rate-true strain curves of the 30 min,respectively) investigated Fe-23Mn-1.5Al-0.3C steel after various degrees of cold rolling (CR)and subsequent heat treatment.(RC annealing-630C/ 10 min after 30%CR,550C/1 h(40%CR),550C/30 min (50%CR):RX yield strength along with significantly improved ductility annealing -700C/15 min after 30%CR,700C/10 min (40%CR). This effect was most pronounced for the 50%cold-rolled 700C10min(50%CR). and recovery-annealed material,where the total elongation increased by a factor of 12 compared to the cold-rolled by cold rolling and recovery annealing should consist of state.Moreover,in comparison with the recrystallized a high density of deformation twins,a low fraction of shear samples the yield strength after recovery annealing bands and a significantly decreased dislocation density remained at a high level.The 40%and 50%deformed compared to the cold-rolled state.As reported above,in and recovery-annealed materials revealed a yield strength the current work the necessary cold rolling and recovery- of 831 MPa and 929 MPa,respectively,and thus raised annealing parameters to achieve this microstructure were the yield strength by 250%compared to the recrystalliza- determined by means of texture analysis.Due to the tion-annealed states.These trends can also be observed increased volume fraction of the y-fiber components from the true stress-true strain curves in Fig.I1b.In addi- (E+F),which are related to the occurrence of shear tion to the improved ductility,a clearly improved work- bands,materials with rolling degrees in excess of 50%were hardening capacity of the material after recovery annealing found to be unsuitable for the applied approach.This was compared to the cold-rolled samples was also observed (cf. confirmed by SEM images of the microstructure after cold Fig.11b). rolling (Fig.7e),where only occasional grain-scale shear bands were observed after 50%deformation.On the other 4.Discussion hand,the increased volume fraction of the CuT component in specimens deformed up to 50%thickness reduction In order to achieve the desired combination of high yield indicated an increased density of deformation twins.This strength and high ductility,the microstructure introduced relationship was confirmed by the calculation of the twin

yield strength along with significantly improved ductility. This effect was most pronounced for the 50% cold-rolled and recovery-annealed material, where the total elongation increased by a factor of 12 compared to the cold-rolled state. Moreover, in comparison with the recrystallized samples the yield strength after recovery annealing remained at a high level. The 40% and 50% deformed and recovery-annealed materials revealed a yield strength of 831 MPa and 929 MPa, respectively, and thus raised the yield strength by 250% compared to the recrystalliza￾tion-annealed states. These trends can also be observed from the true stress–true strain curves in Fig. 11b. In addi￾tion to the improved ductility, a clearly improved work￾hardening capacity of the material after recovery annealing compared to the cold-rolled samples was also observed (cf. Fig. 11b). 4. Discussion In order to achieve the desired combination of high yield strength and high ductility, the microstructure introduced by cold rolling and recovery annealing should consist of a high density of deformation twins, a low fraction of shear bands and a significantly decreased dislocation density compared to the cold-rolled state. As reported above, in the current work the necessary cold rolling and recovery￾annealing parameters to achieve this microstructure were determined by means of texture analysis. Due to the increased volume fraction of the c-fiber components (E + F), which are related to the occurrence of shear bands, materials with rolling degrees in excess of 50% were found to be unsuitable for the applied approach. This was confirmed by SEM images of the microstructure after cold rolling (Fig. 7e), where only occasional grain-scale shear bands were observed after 50% deformation. On the other hand, the increased volume fraction of the CuT component in specimens deformed up to 50% thickness reduction indicated an increased density of deformation twins. This relationship was confirmed by the calculation of the twin 0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0 10 20 30 40 50 60 0.0 0.1 0.2 0.3 0.4 deformed grains recovered grains recrystallized grains (a) Relative frequency (b) Grain boundary misorientation angle [°] (c) Fig. 10. Grain boundary misorientation profiles of the DEF, RC and RX grains after: (a) 30%, (b) 40% and (c) 50% reduction by cold rolling and subsequent recovery annealing (630 C/10 min, 550 C/1 h and 550 C/ 30 min, respectively). 0 10 20 30 40 50 60 70 0 200 400 600 800 1000 1200 1400 Engineering stress, σ [MPa] Engineering strain, e [%] 0.0 0.1 0.2 0.3 0.4 0.5 0 1000 2000 3000 4000 True stress, σtrue [MPa] / Work hardening rate (d σ/dε) [MPa] True strain, ε 30% CR 30% CR + RC 30% CR + RX 40% CR 40% CR + RC 40% CR + RX 50% CR 50% CR + RC 50% CR + RX (a) (b) Fig. 11. (a) Engineering stress–strain curves and (b) true stress–true strain curves (dotted lines) and work-hardening rate–true strain curves of the investigated Fe–23Mn–1.5Al–0.3C steel after various degrees of cold rolling (CR) and subsequent heat treatment. (RC annealing – 630 C/ 10 min after 30% CR, 550 C/1 h (40% CR), 550 C/30 min (50% CR); RX annealing – 700 C/15 min after 30% CR, 700 C/10 min (40% CR), 700 C/10 min (50% CR)). 336 C. Haase et al. / Acta Materialia 80 (2014) 327–340