《金属材料强韧化与组织调控》教学资源（参考文献）MICROSTRUCTURES AND DISLOCATION CONFIGURATIONS IN NANOSTRUCTURED Cu PROCESSED BY REPETITIVE

Act MATERIALIA Pergamon Acta mater.49(2001)1497-1505 www.elsevier.com/locate/actamat MICROSTRUCTURES AND DISLOCATION CONFIGURATIONS IN NANOSTRUCTURED Cu PROCESSED BY REPETITIVE CORRUGATION AND STRAIGHTENING J.Y.HUANGi,Y.T.ZHU,H.JIANG and T.C.LOWE Materials Science and Technology Division,MS G 755,Los Alamos National Laboratory,Los Alamos, NM 87545.USA Received 6 November 2000:accepted 5 February 2001 Abstract-The microstructures and dislocation configurations in nanostructured Cu processed by a new tech- nique,repetitive corrugation and straightening (RCS),were studied using transmission electron microcopy (TEM)and high resolution TEM.Most dislocations belong to 60 type and tend to pile up along the (111] slip planes.Microstructural features including low-angle grain boundaries(GBs),high-angle GBs,and equilib- rium and non-equilibrium GBs and subgrain boundaries were observed.Dislocation structures at an intermedi- ate deformation strain were studied to investigate the microstructural evolutions,which revealed some unique microstructural features such as isolated dislocation cell (IDC),dislocation tangle zones (DTZs),and uncon- densed dislocation walls (UDWs).2001 Acta Materialia Inc.Published by Elsevier Science Ltd.All rights reserved. Keywords:Repetitive corrugation straightening;Microstructure;Transmission electron microscopy (TEM): Copper 1.INTRODUCTION has been paid to alternative procedures of introducing ultrafine grains in materials by severe plastic defor- Many methods have been used to synthesize materials mation (SPD)[9-121. with ultrafine grain sizes (10-1000 nm),including inert gas condensation [1],high-energy ball milling One of the SPD variants,equal-channel angular pressing (ECAP),has been used to refine bulk, [2],sliding wear [3],etc.These techniques are attract- ive for producing powders with grain sizes below coarse-grained metals and alloys to grain sizes rang- 100 nm,but cannot be used to make bulk samples. ing from <0.I to I um [9-12].However,ECAP is To consolidate the nanometer-sized powders into bulk difficult to scale up to process volumes of materials materials,high pressure and moderate temperature are much larger than the 20x20x100 mm3 samples that usually needed.Grains might grow during consoli- are typically produced today.Furthermore,current dation,making the bulk materials partially or com- implementations of ECAP are discontinuous,requir- pletely lose the nanocharacteristics.It is usually ing labor intensive handling of the work-piece impossible to completely eliminate porosity,even in between process steps.These difficulties in fabricat- materials consolidated under very high pressure and ing bulk,nanostructured materials have been substan- temperature.In addition,nanopowders are very sus- tial road-blocks to the structural applications of nano- ceptible to oxidation and absorb large quantities of structured materials.Other SPD techniques that have impurities such as O2.H2 and N2,making it difficult been reported in the literature include multipass-coin- to obtain clean bulk materials.The porosity as well forge (MCF)[13]and multi-axis deformation [14]. as impurities significantly affect the mechanical Both of them have certain advantages over the ECAP properties of the bulk materials,often making them process.However,they also employ batch processing, brittle [4-8].These problems prevent us from study- which is not efficient for large-scale production. ing the intrinsic properties of bulk nanomaterials.As Recently,we have developed a new technique, a consequence of these difficulties,much attention repetitive corrugation and straightening (RCS),that can not only create bulk nanostructured materials free of contamination and porosity,but can also be easily To whom all correspondence should be addressed.Tel.: adapted to large-scale industrial production [15].In +1-505-665-0835:fax:+1-505-667-2264. the RCS process,a work-piece is repetitively bent and E-mail address:jyhuang@lanl.gov (J.Huang) straightened without significantly changing the cross- 1359-6454/01/$20.00 2001 Acta Materialia Inc.Published by Elsevier Science Ltd.All rights reserved. P:S1359-6454(01)00069-6

Acta mater. 49 (2001) 1497–1505 www.elsevier.com/locate/actamat MICROSTRUCTURES AND DISLOCATION CONFIGURATIONS IN NANOSTRUCTURED Cu PROCESSED BY REPETITIVE CORRUGATION AND STRAIGHTENING J. Y. HUANG†, Y. T. ZHU, H. JIANG and T. C. LOWE Materials Science and Technology Division, MS G 755, Los Alamos National Laboratory, Los Alamos, NM 87545, USA ( Received 6 November 2000; accepted 5 February 2001 ) Abstract—The microstructures and dislocation configurations in nanostructured Cu processed by a new tech￾nique, repetitive corrugation and straightening (RCS), were studied using transmission electron microcopy (TEM) and high resolution TEM. Most dislocations belong to 60° type and tend to pile up along the {111} slip planes. Microstructural features including low-angle grain boundaries (GBs), high-angle GBs, and equilib￾rium and non-equilibrium GBs and subgrain boundaries were observed. Dislocation structures at an intermedi￾ate deformation strain were studied to investigate the microstructural evolutions, which revealed some unique microstructural features such as isolated dislocation cell (IDC), dislocation tangle zones (DTZs), and uncon￾densed dislocation walls (UDWs).  2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved. Keywords: Repetitive corrugation straightening; Microstructure; Transmission electron microscopy (TEM); Copper 1. INTRODUCTION Many methods have been used to synthesize materials with ultrafine grain sizes (10–1000 nm), including inert gas condensation [1], high-energy ball milling [2], sliding wear [3], etc. These techniques are attract￾ive for producing powders with grain sizes below 100 nm, but cannot be used to make bulk samples. To consolidate the nanometer-sized powders into bulk materials, high pressure and moderate temperature are usually needed. Grains might grow during consoli￾dation, making the bulk materials partially or com￾pletely lose the nanocharacteristics. It is usually impossible to completely eliminate porosity, even in materials consolidated under very high pressure and temperature. In addition, nanopowders are very sus￾ceptible to oxidation and absorb large quantities of impurities such as O2, H2 and N2, making it difficult to obtain clean bulk materials. The porosity as well as impurities significantly affect the mechanical properties of the bulk materials, often making them brittle [4–8]. These problems prevent us from study￾ing the intrinsic properties of bulk nanomaterials. As a consequence of these difficulties, much attention † To whom all correspondence should be addressed. Tel.: +1-505-665-0835; fax: +1-505-667-2264. E-mail address: jyhuang@lanl.gov (J. Huang) 1359-6454/01/$20.00  2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved. PII: S13 59-6454(01)00069-6 has been paid to alternative procedures of introducing ultrafine grains in materials by severe plastic defor￾mation (SPD) [9–12]. One of the SPD variants, equal-channel angular pressing (ECAP), has been used to refine bulk, coarse-grained metals and alloys to grain sizes rang￾ing from
0.1 to 1 µm [9–12]. However, ECAP is difficult to scale up to process volumes of materials much larger than the 20×20×100 mm3 samples that are typically produced today. Furthermore, current implementations of ECAP are discontinuous, requir￾ing labor intensive handling of the work-piece between process steps. These difficulties in fabricat￾ing bulk, nanostructured materials have been substan￾tial road-blocks to the structural applications of nano￾structured materials. Other SPD techniques that have been reported in the literature include multipass-coin￾forge (MCF) [13] and multi-axis deformation [14]. Both of them have certain advantages over the ECAP process. However, they also employ batch processing, which is not efficient for large-scale production. Recently, we have developed a new technique, repetitive corrugation and straightening (RCS), that can not only create bulk nanostructured materials free of contamination and porosity, but can also be easily adapted to large-scale industrial production [15]. In the RCS process, a work-piece is repetitively bent and straightened without significantly changing the cross-

1498 HUANG et al:NANOSTRUCTURED Cu section geometry of the work-piece,during which grain size is desired to effectively demonstrate the large plastic strains are imparted into the materials, grain-refinement capability of the RCS process.A which leads to the refinement of the microstructure. basic RCS cycle consists of two steps:corrugation Hansen and coworkers [16-27]have systematically and straightening.The corrugation is carried out in a studied the evolution of microstructures and defined die set as shown in Fig.2(a),which is the discontinu- microstructural features in rolled face-centered cubic ous version of the RCS process.It was used to simu- (fcc)metals with medium to high stacking fault ener- late the continuous version of the RCS process as gies,such as Al and Cu.However,the deformation shown in Fig.2(b).It is obvious that the continuous mode in the RCS process is different from that of RCS process can be easily adapted to a rolling mill rolling and is expected to result in different micro- for industrial production of nanostructured metals and structural evolution and consequently different micro- alloys.The straightening is accomplished by pressing structures. the corrugated work-piece between two flat platens. A controversial microstructural feature in nanos- It is well known that lower deformation temperature tructured materials processed by SPD techniques is impedes dynamic recovery and consequently non-equilibrium grain boundary (non-equilibrium improves the grain-refinement efficiency [35].There- GB)[28,29].Valiev et al.[28]defined it as GB that fore,the copper bar was immersed in liquid nitrogen contains extrinsic dislocations that are not needed to for 3 min before each RCS cycle.It took about 1 min accommodate the misorientation across the GB.The to take the sample out of liquid nitrogen,place it in extrinsic dislocations are usually lattice dislocations the die and start pressing it.The sample temperature trapped at the GB.They cause lattice distortion near was unknown but is expected to rise with time during the GB and increase the GB energy [30].Although the RCS process.This corrugation-straightening cycle the non-equilibrium GB has been mentioned by many was repeated 14 times with 90 rotations along the researchers [12,30-34],it has not been directly longitudinal axis of the sample between consecutive proved experimentally and has been controversial. cycles.Lubricant was used to reduce friction between The objectives of this work are:(1)to study the the work-piece and the die,although some amount of microstructural features and dislocation configur- stretching was still present.The microstructure of the ations in nanostructured Cu processed by RCS:(2)to investigate the microstructural evolutions and grain- refinement mechanisms through TEM observations of (a) Press dislocation structures at an intermediate deformation strain;and (3)to clarify the existence of non-equilib- rium GBs produced by SPD. 2.EXPERIMENTAL PROCEDURES A high purity (99.99 at.%)copper bar with 6x6x50 mm3 in dimension was used in this study.It was annealed at 900C for I h to increase its average Work piece grain size to about 765 um (see Fig.1).The large (b) 500m Fig.2.(a)Die set up for discontinuous RCS process and (b) Fig.1.As-annealed copper has an average grain size of set up for continuous RCS process that can be easily adapted 765μm. to large-scale industrial production

1498 HUANG et al.: NANOSTRUCTURED Cu section geometry of the work-piece, during which large plastic strains are imparted into the materials, which leads to the refinement of the microstructure. Hansen and coworkers [16–27] have systematically studied the evolution of microstructures and defined microstructural features in rolled face-centered cubic (fcc) metals with medium to high stacking fault ener￾gies, such as Al and Cu. However, the deformation mode in the RCS process is different from that of rolling and is expected to result in different micro￾structural evolution and consequently different micro￾structures. A controversial microstructural feature in nanos￾tructured materials processed by SPD techniques is non-equilibrium grain boundary (non-equilibrium GB) [28, 29]. Valiev et al. [28] defined it as GB that contains extrinsic dislocations that are not needed to accommodate the misorientation across the GB. The extrinsic dislocations are usually lattice dislocations trapped at the GB. They cause lattice distortion near the GB and increase the GB energy [30]. Although the non-equilibrium GB has been mentioned by many researchers [12, 30–34], it has not been directly proved experimentally and has been controversial. The objectives of this work are: (1) to study the microstructural features and dislocation configur￾ations in nanostructured Cu processed by RCS; (2) to investigate the microstructural evolutions and grain￾refinement mechanisms through TEM observations of dislocation structures at an intermediate deformation strain; and (3) to clarify the existence of non-equilib￾rium GBs produced by SPD. 2. EXPERIMENTAL PROCEDURES A high purity (99.99 at.%) copper bar with 6×6×50 mm3 in dimension was used in this study. It was annealed at 900°C for 1 h to increase its average grain size to about 765 µm (see Fig. 1). The large Fig. 1. As-annealed copper has an average grain size of 765 µm. grain size is desired to effectively demonstrate the grain-refinement capability of the RCS process. A basic RCS cycle consists of two steps: corrugation and straightening. The corrugation is carried out in a die set as shown in Fig. 2(a), which is the discontinu￾ous version of the RCS process. It was used to simu￾late the continuous version of the RCS process as shown in Fig. 2(b). It is obvious that the continuous RCS process can be easily adapted to a rolling mill for industrial production of nanostructured metals and alloys. The straightening is accomplished by pressing the corrugated work-piece between two flat platens. It is well known that lower deformation temperature impedes dynamic recovery and consequently improves the grain-refinement efficiency [35]. There￾fore, the copper bar was immersed in liquid nitrogen for 3 min before each RCS cycle. It took about 1 min to take the sample out of liquid nitrogen, place it in the die and start pressing it. The sample temperature was unknown but is expected to rise with time during the RCS process. This corrugation-straightening cycle was repeated 14 times with 90° rotations along the longitudinal axis of the sample between consecutive cycles. Lubricant was used to reduce friction between the work-piece and the die, although some amount of stretching was still present. The microstructure of the Fig. 2. (a) Die set up for discontinuous RCS process and (b) set up for continuous RCS process that can be easily adapted to large-scale industrial production

HUANG et al:NANOSTRUCTURED Cu 1499 deformed sample was characterized by transmission electron microscopy (TEM)as well as high-resolution TEM (HRTEM).The HRTEM was carried out in a JEOL 3000 FEG electron microscope operated at 300 kV.The point-to-point resolution is about 1.8 A. TEM and HRTEM samples were prepared by jet elec- tro-polishing at room temperature.The electrolyte consists of 33%orthophosphoric acid and 67%water. To enhance the image contrast,most of the HREM images were reconstructed from Fast Fourier Trans- formation,during which the diffuse scattering from the background or inelastic scattering was filtered. DTZ 3.EXPERIMENTAL RESULTS 3.1.Microstructures and dislocations in nanostruc- tured Cu processed for 14 RCS passes Figure 3(a)is a TEM micrograph showing that individual grains were produced with sizes ranging from less than 100 nm to a few hundred nanometers, 100nm separated by high-angle grain boundaries(high-angle GBs).Most grains are heavily strained and contain high density of dislocations.The corresponding elec- tron diffraction pattern (EDP)in Fig.3(b)exhibits diffraction rings,indicating a polycrystalline struc- ture.The diffraction rings show significant 011 tex- ture.Figure 4(a)shows a TEM micrograph of a grain with a diameter of about 500 nm.A number of fine structures were observed in the interior of the grain. As pointed out by two arrowheads,an array of dislo- 3nm cations piled up along the (111)plane.Consequently, two subgrains (denoted by I and 2)with a misorien- tation of about 1 were produced(measured from the Fig.4.(a)A TEM micrograph showing fine deformation struc- tures in a grain.The numbers 1-3 denote three subgrains:the HRTEM image,not shown here).The dislocations are two arrowheads point out an array of dislocations:the four stars mostly 60 type (as shown later in Fig.4(c))and are mark a low-angle GB:the white circle marks a dislocation tan- glissible along the (111}planes. gle zone (DTZ):the white square marks a transition from DTZ A low-angle GB was also found in this grain,as to dislocation cells.(b)An HRTEM image of the low-angle GB pointed out by the four stars in (a).(c)A Fourier filtered marked by four stars in Fig.4(a).The low-angle GB HRTEM image of a 60 dislocation.A Burgers circuit was was formed by the accumulation of a number of gliss- drawn to enclose the dislocation core marked by a "T".The ile dislocations.It is not edge-on but overlapped,as electron beam and the dislocation line is parallel to [11O].and revealed by the periodic Moire Fringes.Figure 4(b) the Burgers vector b =1/2[011]or 1/2[101]. is an HRTEM image of a local region of this low- angle GB.The misorientation of the two grains is about 5.The spacing of the Moire pattern can be d=2.08 A.D is calculated to be 23.84 A.which calculated using the formula:D=d/o,where d is the agrees well with the experimental value of 24 A,as measured from Fig.4(b). lattice spacing and a is the rotation angle.For Cu. Dislocation cell structure was also observed in subgrain 3 in Fig.4(a).These cells may form individ- ual subgrains upon further plastic straining.Dislo- cation tangling was frequently observed in the interior of grains,as marked by a white circle in Fig.4(a). where the grain is heavily strained.We shall refer such a region as dislocation-tangle zone (DTZ).Fig- ure 4(c)shows a Fourier filtered HRTEM image of a 60 dislocation which was frequently observed in RCS-deformed Cu.Assuming the electron beam and the dislocation line is parallel to [110].the Burgers vector of the dislocation is determined to be 1/2[101] Fig.3.TEM micrographs showing:(a)nanostructured Cu pro- or 1/210111.which has an angle of 60(or 120)with duced by the RCS process:and (b)the corresponding EDP. respect to the dislocation line.For this reason,the

HUANG et al.: NANOSTRUCTURED Cu 1499 deformed sample was characterized by transmission electron microscopy (TEM) as well as high-resolution TEM (HRTEM). The HRTEM was carried out in a JEOL 3000 FEG electron microscope operated at 300 kV. The point-to-point resolution is about 1.8 A˚ . TEM and HRTEM samples were prepared by jet elec￾tro-polishing at room temperature. The electrolyte consists of 33% orthophosphoric acid and 67% water. To enhance the image contrast, most of the HREM images were reconstructed from Fast Fourier Trans￾formation, during which the diffuse scattering from the background or inelastic scattering was filtered. 3. EXPERIMENTAL RESULTS 3.1. Microstructures and dislocations in nanostruc￾tured Cu processed for 14 RCS passes Figure 3(a) is a TEM micrograph showing that individual grains were produced with sizes ranging from less than 100 nm to a few hundred nanometers, separated by high-angle grain boundaries (high-angle GBs). Most grains are heavily strained and contain high density of dislocations. The corresponding elec￾tron diffraction pattern (EDP) in Fig. 3(b) exhibits diffraction rings, indicating a polycrystalline struc￾ture. The diffraction rings show significant 011 tex￾ture. Figure 4(a) shows a TEM micrograph of a grain with a diameter of about 500 nm. A number of fine structures were observed in the interior of the grain. As pointed out by two arrowheads, an array of dislo￾cations piled up along the (111) plane. Consequently, two subgrains (denoted by 1 and 2) with a misorien￾tation of about 1° were produced (measured from the HRTEM image, not shown here). The dislocations are mostly 60° type (as shown later in Fig. 4(c)) and are glissible along the {111} planes. A low-angle GB was also found in this grain, as marked by four stars in Fig. 4(a). The low-angle GB was formed by the accumulation of a number of gliss￾ile dislocations. It is not edge-on but overlapped, as revealed by the periodic Moire´ Fringes. Figure 4(b) is an HRTEM image of a local region of this low￾angle GB. The misorientation of the two grains is about 5°. The spacing of the Moire´ pattern can be calculated using the formula: D = d/a, where d is the lattice spacing and a is the rotation angle. For Cu, Fig. 3. TEM micrographs showing: (a) nanostructured Cu pro￾duced by the RCS process; and (b) the corresponding EDP. Fig. 4. (a) A TEM micrograph showing fine deformation struc￾tures in a grain. The numbers 1–3 denote three subgrains; the two arrowheads point out an array of dislocations; the four stars mark a low-angle GB; the white circle marks a dislocation tan￾gle zone (DTZ); the white square marks a transition from DTZ to dislocation cells. (b) An HRTEM image of the low-angle GB pointed out by the four stars in (a). (c) A Fourier filtered HRTEM image of a 60° dislocation. A Burgers circuit was drawn to enclose the dislocation core marked by a “T”. The electron beam and the dislocation line is parallel to [11¯0], and the Burgers vector b = 1/2[011] or 1/2[101]. d(111) = 2.08 A˚ . D is calculated to be 23.84 A˚ , which agrees well with the experimental value of 24 A˚ , as measured from Fig. 4(b). Dislocation cell structure was also observed in subgrain 3 in Fig. 4(a). These cells may form individ￾ual subgrains upon further plastic straining. Dislo￾cation tangling was frequently observed in the interior of grains, as marked by a white circle in Fig. 4(a), where the grain is heavily strained. We shall refer such a region as dislocation-tangle zone (DTZ). Fig￾ure 4(c) shows a Fourier filtered HRTEM image of a 60° dislocation which was frequently observed in RCS-deformed Cu. Assuming the electron beam and the dislocation line is parallel to [110], the Burgers vector of the dislocation is determined to be 1/2[101] or 1/2[011], which has an angle of 60° (or 120°) with respect to the dislocation line. For this reason, the

1500 HUANG et al:NANOSTRUCTURED Cu dislocation is referred as a 60 dislocation.In many subgrain boundary as non-equilibrium subgrain cases,such a dislocation is dissociated or extended to boundary. more than 10 atomic planes. Figure 6(a)shows a Fourier filtered HRTEM image Figure 5(a)shows another example of subgrain of a low-angle GB,which is delineated by periodic generation from larger grains.The parallelogram- dislocations and Moire Fringes.The GB plane is shaped subgrain with a size of about 250 nm is delin- curved and changes from the (55 12)plane to the eated clearly by the dense-dislocation walls (DDWs) (002)plane.The corresponding EDP (Fig.6(b)) [18,19],which are almost parallel to the two sets of shows that the two grains are misoriented for about (111)planes.This subgrain is clearly inside a larger 9.HRTEM images from the upper-left and lower- grain and is isolated from other subgrains.We shall right part of the low-angle GB(see the framed areas) call it isolated subgrains since its boundary does not are shown in Fig.6(c)and (e),respectively.Figure meet with other subgrain boundaries.The dislocation 6(d)is a structural model corresponding to the low- density is higher inside the subgrain than outside it.angle GB in Fig.6(c).From this model,it is seen that Some of the dislocations are forming cell structures two types of dislocations with Burgers vectors inside the subgrain.Figure 5(b)is a Fourier filtered b=1/2[1011 and b2 1/2[1011.hereafter referred as HRTEM image of the dislocation wall at the point type 1 and type 2 dislocations,respectively,are marked by an arrowhead in Fig.5(a).The dislocation needed to accommodate the geometrical misorien- density is estimated as 3x1017 m-2 at the DDW tation.In other words,these dislocations are geo- subgrain boundary.The estimation is based on the metrically necessary.Valiev et al.[28]referred these Fourier filtered HREM image shown in Fig.5(b).The geometrically necessary dislocations as intrinsic dis- number of dislocations was counted,and divided by locations.According to Fig.6(d),the spacing of type the area in this figure.Such an estimation assumes I dislocations is about 18 A,which is consistent with that the dislocation line goes straight from the top of that measured from Fig.6(c).However,there are the grain through the bottom of it,which may not be three more type 2 dislocations in Fig.6(c)than in(d), the case for most dislocations,since most of them are which indicates that three extrinsic (or non-geometri- curved,or they may terminate in the grain interior. cally necessary)dislocations exist at the GB shown Therefore such estimation is only qualitative but not in Fig.6(c).Therefore,this segment of low-angle GB quantitative.Interstitial loops (marked by black is in a high energy configuration and should be called circles)and vacancy loops (marked by white circles) non-equilibrium grain boundary also exist.The dislocations are again mostly 60 type In Fig.6(e),dislocations are periodically spaced. ones.In addition,the lattice planes near the cell walls The Burgers vector is determined as 1/2[101].The are heavily distorted.The width of the subgrain dislocation spacing in a low-angle GB can be calcu- boundary is about 10 nm.The misorientation across lated using the formula:D b/e,where b is the Burg- the subgrain boundary is measured as about 5.There ers vector of the GB dislocation and 6 the rotation are significantly more dislocations than required to angle of the two grains.The calculated dislocation accommodate the misorientation.These dislocations spacing is 20 A,which is in reasonable agreement are not arranged in the lowest-energy dislocation with experimentally measured values of 22 A in Fig. structure (LEDS)[17,19,36],which makes the 6(e).No extrinsic dislocation is found in Fig.6(e). subgrain boundary unstable.We shall refer such This segment GB is equilibrium GB. Besides the 60 dislocations,other dislocations such as screw dislocations and Frank dislocations were also frequently observed.Figure 7(a)shows a number of dislocations in a grain.The Fourier filtered HRTEM image shown in Fig 7(b)reveals that these dislocations are Frank dislocations.The Burgers vec- tor was determined as 1/2[110].which is in (110) plane.Therefore,they are immobile or sessile dislo- cations. 20m 3.2.Deformation microstructures at intermediate plastic strain Fig.5.(a)A TEM micrograph of a subgrain;inset is an To investigate the microstructural evolutions and HRTEM image from the subgrain showing the crystalline planes.Note that the subgrain is delineated by DDWs which grain-refinement mechanisms,deformation structures are almost parallel to two sets of (111)planes.(b)A Fourier at an intermediate deformation strain (6 RCS passes) filtered HRTEM image from the DDW as pointed out by an were studied.Shown in Fig.8 is a TEM micrograph arrowhead in (a).The electron beam and the dislocation line that depicts many microstructural features.The is parallel to [1fo],and the Burgers vector b 1/2[011]or description of deformation microstructures is contro- 1/2[101].The dislocations are all 60 type.The longer arrow points out the grain boundary orientation.The black circles versial and is confusing in the literature [37].In a mark interstitial loops and the white circles mark vacancy series of recent papers [16-27].Hansen and cowork- loops. ers systematically studied the evolution of microstruc-

1500 HUANG et al.: NANOSTRUCTURED Cu dislocation is referred as a 60° dislocation. In many cases, such a dislocation is dissociated or extended to more than 10 atomic planes. Figure 5(a) shows another example of subgrain generation from larger grains. The parallelogram￾shaped subgrain with a size of about 250 nm is delin￾eated clearly by the dense-dislocation walls (DDWs) [18,19], which are almost parallel to the two sets of {111} planes. This subgrain is clearly inside a larger grain and is isolated from other subgrains. We shall call it isolated subgrains since its boundary does not meet with other subgrain boundaries. The dislocation density is higher inside the subgrain than outside it. Some of the dislocations are forming cell structures inside the subgrain. Figure 5(b) is a Fourier filtered HRTEM image of the dislocation wall at the point marked by an arrowhead in Fig. 5(a). The dislocation density is estimated as 3×1017 m2 at the DDW subgrain boundary. The estimation is based on the Fourier filtered HREM image shown in Fig. 5(b). The number of dislocations was counted, and divided by the area in this figure. Such an estimation assumes that the dislocation line goes straight from the top of the grain through the bottom of it, which may not be the case for most dislocations, since most of them are curved, or they may terminate in the grain interior. Therefore such estimation is only qualitative but not quantitative. Interstitial loops (marked by black circles) and vacancy loops (marked by white circles) also exist. The dislocations are again mostly 60° type ones. In addition, the lattice planes near the cell walls are heavily distorted. The width of the subgrain boundary is about 10 nm. The misorientation across the subgrain boundary is measured as about 5°. There are significantly more dislocations than required to accommodate the misorientation. These dislocations are not arranged in the lowest-energy dislocation structure (LEDS) [17, 19, 36], which makes the subgrain boundary unstable. We shall refer such Fig. 5. (a) A TEM micrograph of a subgrain; inset is an HRTEM image from the subgrain showing the crystalline planes. Note that the subgrain is delineated by DDWs which are almost parallel to two sets of {111} planes. (b) A Fourier filtered HRTEM image from the DDW as pointed out by an arrowhead in (a). The electron beam and the dislocation line is parallel to [11¯0], and the Burgers vector b = 1/2[011] or 1/2[101]. The dislocations are all 60° type. The longer arrow points out the grain boundary orientation. The black circles mark interstitial loops and the white circles mark vacancy loops. subgrain boundary as non-equilibrium subgrain boundary. Figure 6(a) shows a Fourier filtered HRTEM image of a low-angle GB, which is delineated by periodic dislocations and Moire´ Fringes. The GB plane is curved and changes from the (5 5 12) plane to the (002) plane. The corresponding EDP (Fig. 6(b)) shows that the two grains are misoriented for about 9°. HRTEM images from the upper-left and lower￾right part of the low-angle GB (see the framed areas) are shown in Fig. 6(c) and (e), respectively. Figure 6(d) is a structural model corresponding to the low￾angle GB in Fig. 6(c). From this model, it is seen that two types of dislocations with Burgers vectors b1 = 1/2[101] and b2 = 1/2[101], hereafter referred as type 1 and type 2 dislocations, respectively, are needed to accommodate the geometrical misorien￾tation. In other words, these dislocations are geo￾metrically necessary. Valiev et al. [28] referred these geometrically necessary dislocations as intrinsic dis￾locations. According to Fig. 6(d), the spacing of type 1 dislocations is about 18 A˚ , which is consistent with that measured from Fig. 6(c). However, there are three more type 2 dislocations in Fig. 6(c) than in (d), which indicates that three extrinsic (or non-geometri￾cally necessary) dislocations exist at the GB shown in Fig. 6(c). Therefore, this segment of low-angle GB is in a high energy configuration and should be called non-equilibrium grain boundary. In Fig. 6(e), dislocations are periodically spaced. The Burgers vector is determined as 1/2[101]. The dislocation spacing in a low-angle GB can be calcu￾lated using the formula: D = b/q, where b is the Burg￾ers vector of the GB dislocation and q the rotation angle of the two grains. The calculated dislocation spacing is 20 A˚ , which is in reasonable agreement with experimentally measured values of 22 A˚ in Fig. 6(e). No extrinsic dislocation is found in Fig. 6(e). This segment GB is equilibrium GB. Besides the 60° dislocations, other dislocations such as screw dislocations and Frank dislocations were also frequently observed. Figure 7(a) shows a number of dislocations in a grain. The Fourier filtered HRTEM image shown in Fig 7(b) reveals that these dislocations are Frank dislocations. The Burgers vec￾tor was determined as 1/2[1¯10], which is in (11¯0) plane. Therefore, they are immobile or sessile dislo￾cations. 3.2. Deformation microstructures at intermediate plastic strain To investigate the microstructural evolutions and grain-refinement mechanisms, deformation structures at an intermediate deformation strain (6 RCS passes) were studied. Shown in Fig. 8 is a TEM micrograph that depicts many microstructural features. The description of deformation microstructures is contro￾versial and is confusing in the literature [37]. In a series of recent papers [16–27], Hansen and cowork￾ers systematically studied the evolution of microstruc-

HUANG et al:NANOSTRUCTURED Cu 1501 55 002 nm 2nm Fig.6.(a)A TEM micrograph of a low-angle GB.The GB plane is curved from(55 12)to (002)plane.(b) An EDP corresponding to (a).The grain misorientation measured from the EDP is 9.(c)and (d)A Fourier filtered HRTEM image and a structural model of the upper-left of the low-angle GB shown in (a)(marked by a black frame).Burgers vectors are b =1/2[101]and b2=1/2[101].(e)A Fourier filtered HRTEM image from the lower-right part of the low-angle GB shown in (a).The Burgers vectors b of the dislocations is 1/2[101].The unit cells of the images are also outlined. As shown.the deformation structure consists of 110 dislocation cells and cell-blocks (CBs)[20].Within each CB,the deformation is accommodated by one set of slip systems that are different from those in neighboring CBs.The boundary between the neighb- oring CBs is called geometrically necessary boundary (GNB)since the boundary is necessary to accommo- date the glide-induced lattice rotation in the adjoining a CBs [20].Figure 8 shows that GNBs may be parallel to the (111}slip planes or other crystallographic Fig.7.(a)A TEM micrograph showing the high density of planes.The dislocation cell boundaries are called dislocations in a grain.(b)A Fourier filtered HRTEM image showing the dislocation type in (a).The Burgers vector determ- incidental boundaries since they are generated by the ined from the Burgers circuit is b 1/2[110]. statistical mutual trappings of glide dislocations. Hansen and coworkers [16,17,19]observed two types of features on the GNBs in the rolling-induced tures and defined microstructural features in rolled fcc microstructure.One is DDWs.The other is metals with medium to high stacking fault energies, microbands(MBs)that are composed of small pan- such as Al and Cu.In Fig.8,we try to follow their cake-shaped cells(SPCs).The definition of MBs has nomenclatures in defining the microstructural fea- been controversial and confusing since the nomencla- tures,but find it necessary to make our own ture has been used to describe the long,thin plate-like nomenclatures in several occasions. regions observed in deformed copper [37-39].The

HUANG et al.: NANOSTRUCTURED Cu 1501 Fig. 6. (a) A TEM micrograph of a low-angle GB. The GB plane is curved from (5 5 12) to (002) plane. (b) An EDP corresponding to (a). The grain misorientation measured from the EDP is 9°. (c) and (d) A Fourier filtered HRTEM image and a structural model of the upper-left of the low-angle GB shown in (a) (marked by a black frame). Burgers vectors are b1 = 1/2[101] and b2 = 1/2[101¯]. (e) A Fourier filtered HRTEM image from the lower-right part of the low-angle GB shown in (a). The Burgers vectors b of the dislocations is 1/2[101]. The unit cells of the images are also outlined. Fig. 7. (a) A TEM micrograph showing the high density of dislocations in a grain. (b) A Fourier filtered HRTEM image showing the dislocation type in (a). The Burgers vector determ￾ined from the Burgers circuit is b = 1/2[1¯10]. tures and defined microstructural features in rolled fcc metals with medium to high stacking fault energies, such as Al and Cu. In Fig. 8, we try to follow their nomenclatures in defining the microstructural fea￾tures, but find it necessary to make our own nomenclatures in several occasions. As shown, the deformation structure consists of dislocation cells and cell-blocks (CBs) [20]. Within each CB, the deformation is accommodated by one set of slip systems that are different from those in neighboring CBs. The boundary between the neighb￾oring CBs is called geometrically necessary boundary (GNB) since the boundary is necessary to accommo￾date the glide-induced lattice rotation in the adjoining CBs [20]. Figure 8 shows that GNBs may be parallel to the {111} slip planes or other crystallographic planes. The dislocation cell boundaries are called incidental boundaries since they are generated by the statistical mutual trappings of glide dislocations. Hansen and coworkers [16, 17, 19] observed two types of features on the GNBs in the rolling-induced microstructure. One is DDWs. The other is microbands (MBs) that are composed of small pan￾cake-shaped cells (SPCs). The definition of MBs has been controversial and confusing since the nomencla￾ture has been used to describe the long, thin plate-like regions observed in deformed copper [37–39]. The

1502 HUANG et al:NANOSTRUCTURED Cu 4.DISCUSSION cscw 4.1.Microstructures and dislocations in nanostruc- tured Cu Microstructural investigations by TEM have DDW revealed that the RCS process can produce bulk nano- structured materials.The average grain size is CSCW reduced from about 765 um to a range of less than 100 nm to a few hundred nanometers,which is com- parable with that attained by ECA process [9-12]. Non-equilibrium subgrain boundaries (Fig.5(b)) and non-equilibrium,low-angle GBs (Fig.6(c))are observed.Valiev and co-workers [10,28,29,42,43] proposed the existence of non-equilibrium UDW grain/subgrain boundaries in nanostructured materials processed by SPD.The non-equilibrium grain/subgrain boundaries are characterized by high density of extrinsic dislocations and lattice distortion near the boundaries [10,30].In other words,there are more dislocations at and near non-equilibrium subgrain/grain boundaries than required to geometri- cally accommodate the misorientations across the Fig.8.A TEM micrograph showing the formation of dislo- boundaries.These dislocations are not in a LEDS cation cells,IDC,CBs,CSCWs,DDWs,UDWs and DTZs Note that the CSCWs,DDWs and UDWs are almost parallel configuration [17,19,36]and render the to the two sets of (111)planes. subgrain/grain boundaries to have higher energies. Such boundaries are unstable and may reconfigure to form equilibrium boundaries.Although referred to by many researchers [12,30-34],the non-equilibrium grain boundaries have been controversial because of definition by Hansen and coworkers has been critic- the lack of direct observation.Figures 5(b)and 6(c) ized by Hatherly [38].To make it more confusing, show all the features of non-equilibrium Hansen and coworkers [17]defined MBs referred by subgrain/grain boundaries,and therefore prove their other researchers as the "second generation MBs"and existence the SPCs-formed GNB wall as the "first generation Interestingly,while the segment of low-angle GB MBs."although the first-and second-generation MBs parallel to (55 12)plane in Fig.6(a)is in non-equilib- neither physically resemble each other nor have the rium configuration(Fig.6(c)),another segment paral- same formation mechanism. lel to (002)plane is in equilibrium configuration with- Figure 8 shows three types of the GNBs.The first out any extrinsic dislocations.Note that(002)is a low is DDWs.The second is composed of strings of equi- index plane.It is not clear whether GBs oriented axed,small dislocation cells,which we shall define along a low index plane tend to be in an equilibrium as clustered-small-cell walls(CSCWs).The CSCWs state.Attention should be paid to this issue in future are similar to the MBs defined by Hansen et al.[16, studies.These observations suggest that the GBs pro- 17.191.We do not use the nomenclature MB to avoid duced by SPD can be in either non-equilibrium state the aforementioned confusion.We also believe that or equilibrium state. CSCW is more descriptive.The third is uncondensed Figure 5 shows an isolated subgrain delineated by dislocation walls (UDWs).The UDWs have been DDWs that form its boundaries.Such an isolated observed in fatigued polycrystalline copper and subgrain was most likely formed from an IDC shown defined by Liu et al.[40]. in Fig.8.The subgrain boundaries shown in Fig.5(b) In the rolling-induced microstructures,the dislo- is composed of DDWs,a concept proposed by cation cells in a CB are usually interconnected to Hansen and coworkers [16,17,19].Figure 5(b)is the form a cell network [17,19,261.However,such cell first time that dislocation structures in a DDW are networks are not well developed in Fig.8.An isolated observed by HRTEM.As shown,the DDW is not dislocation cell (IDC)is clearly shown.Such a micro- in an LEDS configuration.Dislocations with opposite structural feature (IDC)has not been reported in signs are close to each other and dislocation loops rolled microstructures.Another feature not reported also exist.The lattice planes in the DDW are severely in rolling-induced microstructures is DTZ,in which distorted.With further straining and/or annealing.the there is very high dislocation density.The DTZ is DDS could become an equilibrium boundary.On the similar to some dislocation structures in fatigued other hand,these excessive dislocations could also polycrystalline copper [40,41]. glide away from the subgrain boundaries to accom-

1502 HUANG et al.: NANOSTRUCTURED Cu Fig. 8. A TEM micrograph showing the formation of dislo￾cation cells, IDC, CBs, CSCWs, DDWs, UDWs and DTZs. Note that the CSCWs, DDWs and UDWs are almost parallel to the two sets of {111} planes. definition by Hansen and coworkers has been critic￾ized by Hatherly [38]. To make it more confusing, Hansen and coworkers [17] defined MBs referred by other researchers as the “second generation MBs” and the SPCs-formed GNB wall as the “first generation MBs,” although the first- and second-generation MBs neither physically resemble each other nor have the same formation mechanism. Figure 8 shows three types of the GNBs. The first is DDWs. The second is composed of strings of equi￾axed, small dislocation cells, which we shall define as clustered-small-cell walls (CSCWs). The CSCWs are similar to the MBs defined by Hansen et al. [16, 17, 19]. We do not use the nomenclature MB to avoid the aforementioned confusion. We also believe that CSCW is more descriptive. The third is uncondensed dislocation walls (UDWs). The UDWs have been observed in fatigued polycrystalline copper and defined by Liu et al. [40]. In the rolling-induced microstructures, the dislo￾cation cells in a CB are usually interconnected to form a cell network [17, 19, 26]. However, such cell networks are not well developed in Fig. 8. An isolated dislocation cell (IDC) is clearly shown. Such a micro￾structural feature (IDC) has not been reported in rolled microstructures. Another feature not reported in rolling-induced microstructures is DTZ, in which there is very high dislocation density. The DTZ is similar to some dislocation structures in fatigued polycrystalline copper [40, 41]. 4. DISCUSSION 4.1. Microstructures and dislocations in nanostruc￾tured Cu Microstructural investigations by TEM have revealed that the RCS process can produce bulk nano￾structured materials. The average grain size is reduced from about 765 µm to a range of less than 100 nm to a few hundred nanometers, which is com￾parable with that attained by ECA process [9–12]. Non-equilibrium subgrain boundaries (Fig. 5(b)) and non-equilibrium, low-angle GBs (Fig. 6(c)) are observed. Valiev and co-workers [10, 28, 29, 42, 43] proposed the existence of non-equilibrium grain/subgrain boundaries in nanostructured materials processed by SPD. The non-equilibrium grain/subgrain boundaries are characterized by high density of extrinsic dislocations and lattice distortion near the boundaries [10, 30]. In other words, there are more dislocations at and near non-equilibrium subgrain/grain boundaries than required to geometri￾cally accommodate the misorientations across the boundaries. These dislocations are not in a LEDS configuration [17, 19, 36] and render the subgrain/grain boundaries to have higher energies. Such boundaries are unstable and may reconfigure to form equilibrium boundaries. Although referred to by many researchers [12, 30–34], the non-equilibrium grain boundaries have been controversial because of the lack of direct observation. Figures 5(b) and 6(c) show all the features of non-equilibrium subgrain/grain boundaries, and therefore prove their existence. Interestingly, while the segment of low-angle GB parallel to (5 5 12) plane in Fig. 6(a) is in non-equilib￾rium configuration (Fig. 6(c)), another segment paral￾lel to (002) plane is in equilibrium configuration with￾out any extrinsic dislocations. Note that (002) is a low index plane. It is not clear whether GBs oriented along a low index plane tend to be in an equilibrium state. Attention should be paid to this issue in future studies. These observations suggest that the GBs pro￾duced by SPD can be in either non-equilibrium state or equilibrium state. Figure 5 shows an isolated subgrain delineated by DDWs that form its boundaries. Such an isolated subgrain was most likely formed from an IDC shown in Fig. 8. The subgrain boundaries shown in Fig. 5(b) is composed of DDWs, a concept proposed by Hansen and coworkers [16, 17, 19]. Figure 5(b) is the first time that dislocation structures in a DDW are observed by HRTEM. As shown, the DDW is not in an LEDS configuration. Dislocations with opposite signs are close to each other and dislocation loops also exist. The lattice planes in the DDW are severely distorted. With further straining and/or annealing, the DDS could become an equilibrium boundary. On the other hand, these excessive dislocations could also glide away from the subgrain boundaries to accom-

HUANG et al:NANOSTRUCTURED Cu 1503 modate further deformation,especially when the ing,large subgrains may further divide into smaller strain path changes. subgrains.and the misorientation between subgrains Another salient feature of the deformation structure may increase to form low-angle GBs and high-angle is the DTZ,as shown in Figs 4(a)and 8.Such a DTZ GBs (>15).The above theory has worked pretty well may form cell structure through recovery.In fact, with rolling-deformation of metals with a medium to transformation of the DTZ to cell structure is in pro- high stacking fault energy such as Cu and Al [16,17, gress as shown in the area marked by a white square 19,26],although the formation of subgrain structure in Fig.4(a) from dislocation cells has not been experimentally observed. 4.2.Microstructural evolution During the rolling deformation,the work-piece Grain refinement is caused by dislocation gliding, is deformed in one direction (i.e.under constant accumulation,interaction,tangling and spatial strain path)with increasing strain.This is different rearrangement.Deformation in polycrystalline from the RCS process,in which the work-piece materials has been described by a number of mod- was rotated between consecutive RCS cycles, els,including Sach's zero constraint model [441. resulting in the change of strain path.To a certain Taylor's full constraint model [45]and relaxed con- extent,the rotation of work-piece makes the defor- straint model [46].For equiaxed grains,it is gener- mation mode of RCS process resemble that of ally agreed that the Taylor's model is most appro- fatigue.However,unlike fatigue,larger plastic priate [46,47].According to Taylor's model,slip is deformation is introduced to the work-piece during uniform within each grain and strain compatibility each RCS cycle. is achieved by simultaneous operation of at least The unique deformation mode in the RCS process five slip systems.As discussed below,the uniform is expected to affect the deformation microstructure deformation within each grain as hypothesized by and is indeed shown to do so in Fig.8.Similar to Taylor is often not followed in the deformation of rolling-induced microstructure,Fig.8 shows that the real materials.Consequently,the Taylor's model grains of Cu deformed by RCS are divided into CBs has been modified. and dislocation cells.However,new microstructural It has been observed in coarse-grained fcc features including UDWs,IDCs,and DTZs are also materials such as copper that each grain is divided observed.The UDWs and DTZs are,to some extent, into many volume elements during plastic defor- similar to dislocation structures observed in fatigued mation [16-27,48-52]and there are differences in polycrystalline Cu [40,41].In addition,unlike the the number and selection of active slip systems rolling-induced microstructure.the dislocation cells among neighboring volume elements [20,21].Each are not well networked. volume element deforms under a reduced number During the RCS process,even in the same CB or (less than 5)of slip systems,but a group of adjacent subgrain,slip systems will change when the strain volumes act collectively to fulfill the Taylor cri- path changes from one RCS cycle to the next.As a terion.Each volume element is usually subdivided consequence.the dislocations not only interact with into cells with dislocations forming cell boundaries. other dislocations in the current active slip systems, For this reason,the volume elements are referred but also interact with inactive dislocations generated as CBs.Dislocations from neighboring CBs meet in previous RCS cycles.This will promote the forma- at their boundaries and interact to form CB bound- tion of DTZs and IDCs.Liu et al.[40]proposed a aries.This type of boundary is named GNB since mechanism for the formation of dipolized dislocation they are needed to accommodate the misorientation tangle during fatigue.It is not clear if the same mech- in neighboring CBs.The dislocation cell bound- anism applies to the formation of DTZs in RCS-pro- aries are called incidental boundaries since they are cessed Cu. generated by statistical mutual tapping of glide dis- As marked by white triangles in Fig.8,dislocations locations [20].often supplemented by "forest"dis-may pile up on one side of DDWs to form UDWs. locations [36]. This indicates that DDWs formed first and dislo- The misorientations are very small across cell cations then piled up against the DDWs.The other boundaries but much larger across cell-block bound- type of UDWs was formed by the interaction of dislo- aries.With increasing strain,the misorientations cations from CBs on both sides (see the place marked across cell and CB boundaries increase,and the size by white squares).Both types of UDWs may sub- of the CBs become smaller due to further division. sequently transform to small dislocation cells.for- At a certain strain,the misorientation between ming two types of CSCWs.The former form CSCWs neighboring cells becomes so high that additional slip whose boundaries on one side are delineated by system may be triggered in the cells,which converts DDWs (see the place marked by a black triangle), incidental boundaries into GNBs and make the dislo-while the latter form CSCWs whose both boundaries cation cells act like CBs.Domains surrounded by are composed of rough,small cell boundaries (see the GNBs,such as CBs and CB-like dislocation cells are place marked by a black square). called subgrain structures,and the GNBs are also With increasing RCS strain (cycles),the IDCs may called subgrain boundaries [20].With further strain- become an isolated subgrain (e.g.Fig.5(a)).Also

HUANG et al.: NANOSTRUCTURED Cu 1503 modate further deformation, especially when the strain path changes. Another salient feature of the deformation structure is the DTZ, as shown in Figs 4(a) and 8. Such a DTZ may form cell structure through recovery. In fact, transformation of the DTZ to cell structure is in pro￾gress as shown in the area marked by a white square in Fig. 4(a). 4.2. Microstructural evolution Grain refinement is caused by dislocation gliding, accumulation, interaction, tangling and spatial rearrangement. Deformation in polycrystalline materials has been described by a number of mod￾els, including Sach’s zero constraint model [44], Taylor’s full constraint model [45] and relaxed con￾straint model [46]. For equiaxed grains, it is gener￾ally agreed that the Taylor’s model is most appro￾priate [46, 47]. According to Taylor’s model, slip is uniform within each grain and strain compatibility is achieved by simultaneous operation of at least five slip systems. As discussed below, the uniform deformation within each grain as hypothesized by Taylor is often not followed in the deformation of real materials. Consequently, the Taylor’s model has been modified. It has been observed in coarse-grained fcc materials such as copper that each grain is divided into many volume elements during plastic defor￾mation [16–27, 48–52] and there are differences in the number and selection of active slip systems among neighboring volume elements [20, 21]. Each volume element deforms under a reduced number (less than 5) of slip systems, but a group of adjacent volumes act collectively to fulfill the Taylor cri￾terion. Each volume element is usually subdivided into cells with dislocations forming cell boundaries. For this reason, the volume elements are referred as CBs. Dislocations from neighboring CBs meet at their boundaries and interact to form CB bound￾aries. This type of boundary is named GNB since they are needed to accommodate the misorientation in neighboring CBs. The dislocation cell bound￾aries are called incidental boundaries since they are generated by statistical mutual tapping of glide dis￾locations [20], often supplemented by “forest” dis￾locations [36]. The misorientations are very small across cell boundaries but much larger across cell-block bound￾aries. With increasing strain, the misorientations across cell and CB boundaries increase, and the size of the CBs become smaller due to further division. At a certain strain, the misorientation between neighboring cells becomes so high that additional slip system may be triggered in the cells, which converts incidental boundaries into GNBs and make the dislo￾cation cells act like CBs. Domains surrounded by GNBs, such as CBs and CB-like dislocation cells are called subgrain structures, and the GNBs are also called subgrain boundaries [20]. With further strain￾ing, large subgrains may further divide into smaller subgrains, and the misorientation between subgrains may increase to form low-angle GBs and high-angle GBs (>15°). The above theory has worked pretty well with rolling-deformation of metals with a medium to high stacking fault energy such as Cu and Al [16, 17, 19, 26], although the formation of subgrain structure from dislocation cells has not been experimentally observed. During the rolling deformation, the work-piece is deformed in one direction (i.e. under constant strain path) with increasing strain. This is different from the RCS process, in which the work-piece was rotated between consecutive RCS cycles, resulting in the change of strain path. To a certain extent, the rotation of work-piece makes the defor￾mation mode of RCS process resemble that of fatigue. However, unlike fatigue, larger plastic deformation is introduced to the work-piece during each RCS cycle. The unique deformation mode in the RCS process is expected to affect the deformation microstructure and is indeed shown to do so in Fig. 8. Similar to rolling-induced microstructure, Fig. 8 shows that the grains of Cu deformed by RCS are divided into CBs and dislocation cells. However, new microstructural features including UDWs, IDCs, and DTZs are also observed. The UDWs and DTZs are, to some extent, similar to dislocation structures observed in fatigued polycrystalline Cu [40, 41]. In addition, unlike the rolling-induced microstructure, the dislocation cells are not well networked. During the RCS process, even in the same CB or subgrain, slip systems will change when the strain path changes from one RCS cycle to the next. As a consequence, the dislocations not only interact with other dislocations in the current active slip systems, but also interact with inactive dislocations generated in previous RCS cycles. This will promote the forma￾tion of DTZs and IDCs. Liu et al. [40] proposed a mechanism for the formation of dipolized dislocation tangle during fatigue. It is not clear if the same mech￾anism applies to the formation of DTZs in RCS-pro￾cessed Cu. As marked by white triangles in Fig. 8, dislocations may pile up on one side of DDWs to form UDWs. This indicates that DDWs formed first and dislo￾cations then piled up against the DDWs. The other type of UDWs was formed by the interaction of dislo￾cations from CBs on both sides (see the place marked by white squares). Both types of UDWs may sub￾sequently transform to small dislocation cells, for￾ming two types of CSCWs. The former form CSCWs whose boundaries on one side are delineated by DDWs (see the place marked by a black triangle), while the latter form CSCWs whose both boundaries are composed of rough, small cell boundaries (see the place marked by a black square). With increasing RCS strain (cycles), the IDCs may become an isolated subgrain (e.g. Fig. 5(a)). Also

1504 HUANG et al.:NANOSTRUCTURED Cu larger CBs may further divide into smaller CBs. REFERENCES DTZs may transform into dislocations cells. Subgrains will develop from both CBs and dislo- 1.Gleiter,H.,Prog.Mater.Sci.,1989,33,223 cation cells.The misorientation across subgrain 2.Koch,C.C.and Cho,Y.S..Nanostruc.Mater.,1992,1. 207. boundaries increases with further RCS strain.and 3.Rigney,D.A..Annu.Rev.Mater.Sci..1988.18.141. eventually becomes large enough to transform the 4.Alexandrov,I.V.,Zhu.Y.T..Lowe,T.C.,Islamgaliev. subgrain boundaries into low-angle GBs or high- R.K.and Valiev,R.Z.,Metall.Mater.Trans.,1998. angle GBs.Note that the change of strain path gener- 29A.2253 ally enhances the effectiveness of grain refinement 5.Stolyarov,V.V..Zhu,Y.T..Lowe,T.C.,Islamgaliev, R.K.and Valiev.R.Z.,Mater.Sci.Engng.2000, [53].Other factors that affect the efficiency of grain A282.78. refinement include crystal structure,orientation and 6.Alexandrov,I.V.,Zhu,Y.T.,Lowe,T.C.and Valiev,R. deformation mode [53]. Z..Powder Metall.,1998,41,11. 7.Weertman,J.R..Mater.Sci.Engng.1993.A166,161. 8.Sanders,P.G.,Eastman,J.A.and Weertman.J.R..Acta 5.CONCLUSIONS nater..,1997,45,4019. 9.Segal,V.M.,Mater.Sci.Engng.1995,A197,157 The RCS process effectively reduced the grain size 10.Valiev,R.Z.,Islamgaliev,R.K.and Alexandarov,I.V., of a high-purity copper bar from 765 um to about Prog.Mater.Sci.,2000,45,103. 500 nm,demonstrating the RCS as a promising new 11.Iwahashi,Y..Horita,Z..Nemoto,M.and Langdon,T.G.. Acta mater.,1998.46.3317. technique for producing bulk nanostructured metal 12.Ferrase,S.Segal,V.M.,Hartwig.K.T.and Goforth,R. materials.The change of strain path during the RCS E..Metall.Mater.Trans..1997.28A.1047. process generally enhances the effectiveness of 13.Ghosh,A.K.and Huang.W.,Investigations and Appli- grain refinement. cations of Severe Plastic Deformation,in:T.C.Lowe and The development of the microstructure during the R.Z.Valiev (Eds.).NATO Science Series.Series 3.High RCS process was characterized by TEM and Technology,Vol.80.Kluwer Academic,Boston,2000, p.29. HRTEM.Dislocations cell structures,IDCs,cell- 14.Chen,W.,Ferguson,D.and Ferguson,H.,Ultrafine blocks (CBs),dense-dislocation walls (DDWs),clus- Grained Materials,in:R.S.Mishra,S.L.Semiatin,C. tered-small-cell walls (CSCWs),UDWs,DTZs, Suryanrayana.N.N.Thadhani and T.C.Lowe (Eds.). subgrains,low-angle GBs and high-angle GBs were TMS,Warrendale,PA,2000,p.235. 15.Zhu,Y.T.,Jiang,H.,Huang,J.and Lowe,T.C.,Metall. observed.The UDWs,DTZs and IDCs are new Mater.Trans.A (submitted for publication). microstructural features not observed in rolling- 16.Hansen,N.,Mater.Sci.Technol.,1990,6,1039. induced Cu or Al.The dislocation is mostly 60 type 17.Bay,B..Hansen,N..Hughes,D.A.and Kuhlmann- and it tends to pile-up along the (111}glide planes Wilsdore,D..Acta mater..1992,40,205. to form DDWs,CSCWs,CBs,etc. 18.Hughes,D.A.and Hansen,N.,Acta mater.,1997,45. 3871. Most dislocations are 60 type.Screw dislocations 19.Bay,B.,Hansen,N.and Kuhlmann-Wilsdorf,D.,Mater. and Frank dislocations are also frequently observed. Sci.Engng,1989,A113,385. DDWs contain high-density dislocations,interstitial 20.Kuhlmann-Wilsdorf,D.and Hansen,N.,Scr.Metall. loops and vacancy loops.The dislocation density is Mater.,1991,25,1557. as high as 3x1017m-2.Subgrain boundaries formed 21.Hansen,N.,Scr.Metall.Mater.,1992,27,1447 22.Liu,Q.and Hansen,N.,Scr.Metall.Mater.,1995,32, by DDWs are in non-equilibrium state. 1289. This work for the first time observed the existence 23.Ananthan,V.S.,Leffers,T.and Hansen,N.,Scr.Metall. non-equilibrium GBs.However,equilibrium GBs are Mater.,1991.25.137. also observed.Therefore,equilibrium and non-equi- 24.Hansen,N.and Huang,X.,Acta mater.,1998,46,1827 librium GBs coexist in RCS-processed Cu.Further 25.Hansen,N.and Hughes,D.A.,Phrys.Status Solidi B,1995 149.155. study is needed to find out what affects the equilib- 26.Hansen,N.and Juul Jensen,D..Philos.Trans.R.Soc. rium state of GBs. Lomd.,1999.A357.1447. The grain refinement and microstructural evol- 27.Liu,Q.,Juul Jensen,D.and Hansen,N.,Acta mater.,1998. ution during RCS is as follows:at low strains, 46.5819. grains is first divided into CBs,which contain dis- 28.Valiev,R.Z..Yu Gertsman,V.and Kaibyshev,O.A., Phys.Status Solidi A,1986,97,11. location cells.DTZs may also develop inside CBs. 29.Valiev,R.Z.,Kaibyshev,O.A.and Khnnanov.Sh.Kh.. With increasing RCS strains,CBs may further sub- Phys.Status Solidi A,1979,52,447. divide into smaller CBs and DTZs may transform 30.Jiang.H.,Zhu,Y.T.,Butt,D.P.,Alexandrov,I.V.and into dislocations cells.Subgrains will develop from Lowe,T.C.,Mater.Sci.Engng,2000,A290,128. 31.Furukawa,M..Iwahashi,Y..Horita,Z.,Nemoto,M., both CBs and dislocations cells.The latter become Tsenev.N.K..Valiev,R.Z.and Langdon,T.G.,Acta subgrains when the misorientation across their maer.,1997,45,4751. boundaries are so large that they develop their own 32.Islamgaliev,R.K.,Chmelik,F.and Kuzel,R.,Mater.Sci. unique slip systems.The misorientation across Engng,1997,A234-236.335. subgrain boundaries increases with further RCS 33.Horita,Z..Smith,D.J.,Nemoto,M.,Valiev,R.Z.and strain,and eventually becomes large enough to Langdon,T.G..J.Mater.Res..1998,13,446. 34.Horita,Z.Smith,D.J..Furukawa,M..Nemoto,M.,Val- transform the subgrain boundaries into low-angle iev,R.Z.and Langdon,T.G..J.Mater.Res..1996.11, GBs or high-angle GBs. 1880

1504 HUANG et al.: NANOSTRUCTURED Cu larger CBs may further divide into smaller CBs. DTZs may transform into dislocations cells. Subgrains will develop from both CBs and dislo￾cation cells. The misorientation across subgrain boundaries increases with further RCS strain, and eventually becomes large enough to transform the subgrain boundaries into low-angle GBs or high￾angle GBs. Note that the change of strain path gener￾ally enhances the effectiveness of grain refinement [53]. Other factors that affect the efficiency of grain refinement include crystal structure, orientation and deformation mode [53]. 5. CONCLUSIONS The RCS process effectively reduced the grain size of a high-purity copper bar from 765 µm to about 500 nm, demonstrating the RCS as a promising new technique for producing bulk nanostructured metal materials. The change of strain path during the RCS process generally enhances the effectiveness of grain refinement. The development of the microstructure during the RCS process was characterized by TEM and HRTEM. Dislocations cell structures, IDCs, cell￾blocks (CBs), dense-dislocation walls (DDWs), clus￾tered-small-cell walls (CSCWs), UDWs, DTZs, subgrains, low-angle GBs and high-angle GBs were observed. The UDWs, DTZs and IDCs are new microstructural features not observed in rolling￾induced Cu or Al. The dislocation is mostly 60° type and it tends to pile-up along the {111} glide planes to form DDWs, CSCWs, CBs, etc. Most dislocations are 60° type. Screw dislocations and Frank dislocations are also frequently observed. DDWs contain high-density dislocations, interstitial loops and vacancy loops. The dislocation density is as high as 3×1017 m2 . Subgrain boundaries formed by DDWs are in non-equilibrium state. This work for the first time observed the existence non-equilibrium GBs. However, equilibrium GBs are also observed. Therefore, equilibrium and non-equi￾librium GBs coexist in RCS-processed Cu. Further study is needed to find out what affects the equilib￾rium state of GBs. The grain refinement and microstructural evol￾ution during RCS is as follows: at low strains, grains is first divided into CBs, which contain dis￾location cells. DTZs may also develop inside CBs. With increasing RCS strains, CBs may further sub￾divide into smaller CBs and DTZs may transform into dislocations cells. Subgrains will develop from both CBs and dislocations cells. The latter become subgrains when the misorientation across their boundaries are so large that they develop their own unique slip systems. The misorientation across subgrain boundaries increases with further RCS strain, and eventually becomes large enough to transform the subgrain boundaries into low-angle GBs or high-angle GBs. REFERENCES 1. Gleiter, H., Prog. Mater. Sci., 1989, 33, 223. 2. Koch, C. C. and Cho, Y. S., Nanostruc. Mater., 1992, 1, 207. 3. Rigney, D. A., Annu. Rev. Mater. Sci., 1988, 18, 141. 4. Alexandrov, I. V., Zhu, Y. T., Lowe, T. C., Islamgaliev, R. K. and Valiev, R. Z., Metall. Mater. Trans., 1998, 29A, 2253. 5. Stolyarov, V. V., Zhu, Y. T., Lowe, T. C., Islamgaliev, R. K. and Valiev, R. Z., Mater. Sci. Engng, 2000, A282, 78. 6. Alexandrov, I. V., Zhu, Y. T., Lowe, T. C. and Valiev, R. Z., Powder Metall., 1998, 41, 11. 7. Weertman, J. R., Mater. Sci. Engng, 1993, A166, 161. 8. Sanders, P. G., Eastman, J. A. and Weertman, J. R., Acta mater., 1997, 45, 4019. 9. Segal, V. M., Mater. Sci. Engng, 1995, A197, 157. 10. Valiev, R. Z., Islamgaliev, R. K. and Alexandarov, I. V., Prog. Mater. Sci., 2000, 45, 103. 11. Iwahashi, Y., Horita, Z., Nemoto, M. and Langdon, T. G., Acta mater., 1998, 46, 3317. 12. Ferrase, S., Segal, V. M., Hartwig, K. T. and Goforth, R. E., Metall. Mater. Trans., 1997, 28A, 1047. 13. Ghosh, A. K. and Huang, W., Investigations and Appli￾cations of Severe Plastic Deformation, in: T. C. Lowe and R. Z. Valiev (Eds.), NATO Science Series, Series 3, High Technology, Vol. 80. Kluwer Academic, Boston, 2000, p. 29. 14. Chen, W., Ferguson, D. and Ferguson, H., Ultrafine Grained Materials, in: R. S. Mishra, S. L. Semiatin, C. Suryanrayana, N. N. Thadhani and T. C. Lowe (Eds.). TMS, Warrendale, PA, 2000, p. 235. 15. Zhu, Y. T., Jiang, H., Huang, J. and Lowe, T. C., Metall. Mater. Trans. A (submitted for publication). 16. Hansen, N., Mater. Sci. Technol., 1990, 6, 1039. 17. Bay, B., Hansen, N., Hughes, D. A. and Kuhlmann￾Wilsdore, D., Acta mater., 1992, 40, 205. 18. Hughes, D. A. and Hansen, N., Acta mater., 1997, 45, 3871. 19. Bay, B., Hansen, N. and Kuhlmann-Wilsdorf, D., Mater. Sci. Engng, 1989, A113, 385. 20. Kuhlmann-Wilsdorf, D. and Hansen, N., Scr. Metall. Mater., 1991, 25, 1557. 21. Hansen, N., Scr. Metall. Mater., 1992, 27, 1447. 22. Liu, Q. and Hansen, N., Scr. Metall. Mater., 1995, 32, 1289. 23. Ananthan, V. S., Leffers, T. and Hansen, N., Scr. Metall. Mater., 1991, 25, 137. 24. Hansen, N. and Huang, X., Acta mater., 1998, 46, 1827. 25. Hansen, N. and Hughes, D. A., Phys. Status Solidi B, 1995, 149, 155. 26. Hansen, N. and Juul Jensen, D., Philos. Trans. R. Soc. Lond., 1999, A 357, 1447. 27. Liu, Q., Juul Jensen, D. and Hansen, N., Acta mater., 1998, 46, 5819. 28. Valiev, R. Z., Yu Gertsman, V. and Kaibyshev, O. A., Phys. Status Solidi A, 1986, 97, 11. 29. Valiev, R. Z., Kaibyshev, O. A. and Khnnanov, Sh. Kh., Phys. Status Solidi A, 1979, 52, 447. 30. Jiang, H., Zhu, Y. T., Butt, D. P., Alexandrov, I. V. and Lowe, T. C., Mater. Sci. Engng, 2000, A290, 128. 31. Furukawa, M., Iwahashi, Y., Horita, Z., Nemoto, M., Tsenev, N. K., Valiev, R. Z. and Langdon, T. G., Acta mater., 1997, 45, 4751. 32. Islamgaliev, R. K., Chmelik, F. and Kuzel, R., Mater. Sci. Engng, 1997, A234-236, 335. 33. Horita, Z., Smith, D. J., Nemoto, M., Valiev, R. Z. and Langdon, T. G., J. Mater. Res., 1998, 13, 446. 34. Horita, Z., Smith, D. J., Furukawa, M., Nemoto, M., Val￾iev, R. Z. and Langdon, T. G., J. Mater. Res., 1996, 11, 1880.

HUANG et al.:NANOSTRUCTURED Cu 1505 35.V.V.Stolyarov.Y.T.Zhu.T.C.Lowe,R.Z.Valiev,53.Zhu,Y.T.and Lowe,T.C..Mater.Sci.Engng,2000, Mater.Sci.Engng A (in press). A291.46. 36.Kuhlmann-Wilsdorf,D.,Mater.Sci.Engng,1989,A113,1. 37.F.J.Humphreys,M.Hatherly,Elsevier,Oxford,UK. 1995,Pp.24-26. APPENDIX A 38.Hatherly,M.,Scr.Metall.Mater.,1992,27,1453. 39.Hansen,N.,Scr.Metall.Mater.,1992,27,1457. A.1.Glossary 40.Liu,C.D.,Bassim,M.N.and You,D.X.,Acta metall. maer.,1994.42.3695. 41.Winter,A.T.,Pederson,O.B.and Rasmussen,K.V.,Acta CB:cell block metall..,1981,29.735. CSCW:clustered-small-cell-wall 42.Valiev,R.Z,Korznikov,A.V.and Mulyukov,R.R.. DDW:dense-dislocation wall Mater.Sci.Engng.1993.A168.141. 43.Valiev,R.Z.,Gertsman,V.Y.and Kaibyshev,O.A.,Scr. DTZ:dislocation-tangle zone Mater.,1983,17.853. EDP:electron diffraction pattern 44.Sachs,G..Z.Ver.Dtsch.Ing.,1928,72,734. ECAP:equal-channel angular pressing 45.Taylor,G.L,J.Inst.Met.,1938,62,307. GB:grain boundary 46.U.F.Kocks,Texture and Anisotropy,in:U.F.Kocks,C. N.Tome,H.-R.Wenk (Eds.).Cambridge University Press, GNB:geometrically necessary boundary Cambridge,UK,1998,p.391. HRTEM:high-resolution transmission electron 47.Stout,M.G.and O'Rourke.J.A..Metall.Trans..1989 microcopy 20A.125. IDC:isolated dislocation cell 48.Rey.C..Rev.Phys.Appl.,1988,23,491. 49.Barret,C.S.and Lewenson,L.H.,Trans.Metall.Soc., LEDS:lowest-energy dislocation structure AME.1940.137.113. MB:microband 50.Barret.C.S..Trans.Am.Inst.Min.Metall.Engng.1945. RCS:repetitive corrugation and straightening 161.15. SPC:pancake-shaped cell 51.Wood,W.A.and Rachinger,W.A..J.Inst.Metall.,1950, 76.237. SPD:severe plastic deformation 52.Wood,W.A.and Scrutton,R.F..J.Inst.Metall.,1950, TEM:transmission electron microcopy 77,423. UDW:uncondensed dislocation wall

HUANG et al.: NANOSTRUCTURED Cu 1505 35. V. V. Stolyarov, Y. T. Zhu, T. C. Lowe, R. Z. Valiev, Mater. Sci. Engng A (in press). 36. Kuhlmann-Wilsdorf, D., Mater. Sci. Engng, 1989, A113, 1. 37. F. J. Humphreys, M. Hatherly, Elsevier, Oxford, UK, 1995, pp. 24–26. 38. Hatherly, M., Scr. Metall. Mater., 1992, 27, 1453. 39. Hansen, N., Scr. Metall. Mater., 1992, 27, 1457. 40. Liu, C. D., Bassim, M. N. and You, D. X., Acta metall. mater., 1994, 42, 3695. 41. Winter, A. T., Pederson, O. B. and Rasmussen, K. V., Acta metall., 1981, 29, 735. 42. Valiev, R. Z., Korznikov, A. V. and Mulyukov, R. R., Mater. Sci. Engng, 1993, A168, 141. 43. Valiev, R. Z., Gertsman, V. Y. and Kaibyshev, O. A., Scr. Mater., 1983, 17, 853. 44. Sachs, G., Z. Ver. Dtsch. Ing., 1928, 72, 734. 45. Taylor, G. I., J. Inst. Met., 1938, 62, 307. 46. U. F. Kocks, Texture and Anisotropy, in: U. F. Kocks, C. N. Tome´, H. -R. Wenk (Eds.). Cambridge University Press, Cambridge, UK, 1998, p. 391. 47. Stout, M. G. and O’Rourke, J. A., Metall. Trans., 1989, 20A, 125. 48. Rey, C., Rev. Phys. Appl., 1988, 23, 491. 49. Barret, C. S. and Lewenson, L. H., Trans. Metall. Soc., AIME, 1940, 137, 113. 50. Barret, C. S., Trans. Am. Inst. Min. Metall. Engng, 1945, 161, 15. 51. Wood, W. A. and Rachinger, W. A., J. Inst. Metall., 1950, 76, 237. 52. Wood, W. A. and Scrutton, R. F., J. Inst. Metall., 1950, 77, 423. 53. Zhu, Y. T. and Lowe, T. C., Mater. Sci. Engng, 2000, A291, 46. APPENDIX A A.1. Glossary CB: cell block CSCW: clustered-small-cell-wall DDW: dense-dislocation wall DTZ: dislocation-tangle zone EDP: electron diffraction pattern ECAP: equal-channel angular pressing GB: grain boundary GNB: geometrically necessary boundary HRTEM: high-resolution transmission electron microcopy IDC: isolated dislocation cell LEDS: lowest-energy dislocation structure MB: microband RCS: repetitive corrugation and straightening SPC: pancake-shaped cell SPD: severe plastic deformation TEM: transmission electron microcopy UDW: uncondensed dislocation wall