《金属材料强韧化与组织调控》教学资源（参考文献）Extreme grain refinement by severe plastic deformation：A wealth of challenging science

Available online at www.sciencedirect.com SciVerse ScienceDirect Cta MATERIALIA ELSEVIER Acta Materialia 61 (2013)782-817 www.elsevier.com/locate/actamat Extreme grain refinement by severe plastic deformation:A wealth of challenging science Y.Estrin4,A.Vinogradovb.I Centre for Advanced Hybrid Materials,Department of Materials Engineering.Monash University.Clayton,VIC3800.Australia Laboratory for the Physics of Strength of Materials and Intelligent Diagnostic Systems.Togliatti State University.Togliatti 445667.Russia Abstract This article presents our take on the area of bulk ultrafine-grained materials produced by severe plastic deformation(SPD).Over the last decades,research activities in this area have grown enormously and have produced interesting results,which we summarise in this concise review.This paper is intended as an introduction to the field for the "uninitiated",while at the same time highlighting some polemic issues that may be of interest to those specialising in bulk nanomaterials produced by SPD.A brief overview of the available SPD technologies is given,along with a summary of unusual mechanical,physical and other properties achievable by SPD processing. The challenges this research is facing-some of them generic and some specific to the nanoSPD area-are identified and discussed. 2012 Acta Materialia Inc.Published by Elsevier Ltd.All rights reserved. Keywords:Severe plastic deformation;Ultrafine-grained materials;Modelling;Properties 1.Historical overview developed the scientific grounds and techniques for materi- als processing through a combination of high hydrostatic Grain size can be regarded as a key microstructural fac- pressure and shear deformation [5,6],which today are at tor affecting nearly all aspects of the physical and mechan- the core of SPD methods.Bridgman effectively introduced ical behaviour of polycrystalline metals as well as their the defining characteristics of SPD processing in the early chemical and biochemical response to the surrounding 1950s.In a strict sense generally accepted in the materials media.Hence,control over grain size has long been recog- engineering community,an SPD process is currently nized as a way to design materials with desired properties. defined as"any method of metal forming under an exten- Most of the mentioned properties benefit greatly from sive hydrostatic pressure that may be used to impose a very grain size reduction.As the race for better materials perfor- high strain on a bulk solid without the introduction of any mance is never ending,attempts to develop viable tech- significant change in the overall dimensions of the sample niques for microstructure refinement continue.A possible and having the ability to produce exceptional grain refine- avenue for microstructure refinement of metals is the use ment"[7].In this Diamond Jubilee issue of Acta Materialia of severe plastic deformation (SPD)a principle that is it is appropriate to mention that many of the modern ideas as old as metalworking itself.Recent essays [1-4]tell a fas- of thermomechanical processing involved in virtually all cinating story of the art of ancient swordmaking through SPD schemes were already addressed in the first volume SPD.The modern-day history of SPD technology has its of Acta Metallurgica in 1953.Carreker and Hibbard [8] beginnings in the seminal work by P.W.Bridgman who pointed out that the yield strength of high-purity copper benefits substantially from grain refinement and this effect Corresponding author.Tel:+61 420822164. is more pronounced at low temperatures.They also noticed E-mail address:yuri.estrin@monash.edu (Y.Estrin). that the effect of the initial grain size vanishes at strains lar- On leave from the Department of Intelligent Materials Engineering. ger than 0.1 and for that reason the grain size has little or Osaka City University,Osaka 558-8585,Japan. no influence on the strength under monotonic loading.A 1359-6454/S36.00 2012 Acta Materialia Inc.Published by Elsevier Ltd.All rights reserved. http://dx.doi.org/10.1016/j.actamat.2012.10.038

Extreme grain refinement by severe plastic deformation: A wealth of challenging science Y. Estrin a,⇑ , A. Vinogradov b,1 a Centre for Advanced Hybrid Materials, Department of Materials Engineering, Monash University, Clayton, VIC 3800, Australia bLaboratory for the Physics of Strength of Materials and Intelligent Diagnostic Systems, Togliatti State University, Togliatti 445667, Russia Abstract This article presents our take on the area of bulk ultrafine-grained materials produced by severe plastic deformation (SPD). Over the last decades, research activities in this area have grown enormously and have produced interesting results, which we summarise in this concise review. This paper is intended as an introduction to the field for the “uninitiated”, while at the same time highlighting some polemic issues that may be of interest to those specialising in bulk nanomaterials produced by SPD. A brief overview of the available SPD technologies is given, along with a summary of unusual mechanical, physical and other properties achievable by SPD processing. The challenges this research is facing—some of them generic and some specific to the nanoSPD area—are identified and discussed. 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Severe plastic deformation; Ultrafine-grained materials; Modelling; Properties 1. Historical overview Grain size can be regarded as a key microstructural fac￾tor affecting nearly all aspects of the physical and mechan￾ical behaviour of polycrystalline metals as well as their chemical and biochemical response to the surrounding media. Hence, control over grain size has long been recog￾nized as a way to design materials with desired properties. Most of the mentioned properties benefit greatly from grain size reduction. As the race for better materials perfor￾mance is never ending, attempts to develop viable tech￾niques for microstructure refinement continue. A possible avenue for microstructure refinement of metals is the use of severe plastic deformation (SPD)—a principle that is as old as metalworking itself. Recent essays [1–4] tell a fas￾cinating story of the art of ancient swordmaking through SPD. The modern-day history of SPD technology has its beginnings in the seminal work by P.W. Bridgman who developed the scientific grounds and techniques for materi￾als processing through a combination of high hydrostatic pressure and shear deformation [5,6], which today are at the core of SPD methods. Bridgman effectively introduced the defining characteristics of SPD processing in the early 1950s. In a strict sense generally accepted in the materials engineering community, an SPD process is currently defined as “any method of metal forming under an exten￾sive hydrostatic pressure that may be used to impose a very high strain on a bulk solid without the introduction of any significant change in the overall dimensions of the sample and having the ability to produce exceptional grain refine￾ment” [7]. In this Diamond Jubilee issue of Acta Materialia it is appropriate to mention that many of the modern ideas of thermomechanical processing involved in virtually all SPD schemes were already addressed in the first volume of Acta Metallurgica in 1953. Carreker and Hibbard [8] pointed out that the yield strength of high-purity copper benefits substantially from grain refinement and this effect is more pronounced at low temperatures. They also noticed that the effect of the initial grain size vanishes at strains lar￾ger than 0.1 and for that reason the grain size has little or no influence on the strength under monotonic loading. A 1359-6454/$36.00 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actamat.2012.10.038 ⇑ Corresponding author. Tel.: +61 420822164. E-mail address: yuri.estrin@monash.edu (Y. Estrin). 1 On leave from the Department of Intelligent Materials Engineering, Osaka City University, Osaka 558–8585, Japan. www.elsevier.com/locate/actamat Available online at www.sciencedirect.com Acta Materialia 61 (2013) 782–817

Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 783 similar effect is well known in fatigue where the grain size the reviews [21,22]and special issues of Advanced Engineer- of wavy-slip materials has no bearing on the fatigue limit. ing Materials [23],Materials Science and Engineering A [24] These observations can be associated with the vital role and Materials Transactions [25,26]. of the dislocation substructure,which forms during defor- What makes SPD processing techniques so popular is mation(be it monotonic or cyclic),and it is the size of the the possibility of using them to enhance the strength char- substructure which determines the strength characteristics acteristics of conventional metallic materials in a quite of metallic materials.Gow and Cahn [9]emphasized the spectacular way:by a factor of up to eight for pure metals significance of crystallographic texture for the deformation such as copper and by some 30-50%for alloys [7,27]. and recrystallization behaviour of metals and the effect of Despite the impressive property improvement achievable evolving texture on the resultant properties.Bell and Cahn with SPD techniques,their uptake by industry has been [10]outlined many fine features of mechanical twinning, rather sluggish.However,things are now starting to which play an important part in plastic deformation when change,and there is a common feeling in the nanoSPD accommodation by dislocation slip is hindered.Beck [11] community that major breakthroughs in terms of indus- highlighted the possibility of relieving the work-hardening try-scale applications of SPD-based technologies are immi- effects by post-processing recovery.As will be seen in the nent.We have been working in this area for more than a following sections,these ideas have had a great impact decade and have followed its developments closely.In this on the development of the SPD processing and are pivotal article we present our views on what has been achieved, to the modern concepts underlying these techniques.Now- what is possibly achievable,and what future trends are to adays,the subject of SPD processing is represented very be expected from SPD processing technologies.This article prominently on the pages of Acta,as illustrated by the does not represent a full review of the SPD area (one could analysis in Ref.[4].Its revival is due to the work of Segal almost say the discipline of SPD,considering the firm place et al.[12]in the Soviet Union in the mid-1970s.These this group of material processing techniques has taken in authors developed the method of equal-channel angular literature).Rather,it is our personal take on the SPD area pressing (ECAP),which later evolved into what is now and an attempt to foretell its future development.Empha- the most popular SPD technique.It should be mentioned, sis is placed on the scientifically challenging aspects of however,that in the time between the publication of Bridg- SPD,and not so much on technological issues,although man's studies and the reintroduction of this subject in the some insights into the promises and limitations of SPD metal science literature,exploration of the possibilities of technologies will also be given. changing the properties of materials through combined high pressure and shear deformation went on both in the 2.SPD methods Soviet Union and in the West.This less known work has been reviewed in Ref.[13].In particular,credit should be Among the procedures devised for grain refinement. given to the work by the group of N.S.Enikolopian con- SPD techniques are of particular interest and are the ducted mainly on polymers. focus of the present review.These techniques enjoy great A real appreciation for the new possibilities for improv- popularity owing to their ability to produce considerable ing the properties of metallic materials provided by SPD grain refinement in fully dense.bulk-scale work-pieces. techniques came with the work of the group of Valiev thus giving promise for structural applications.The [14,15],which demonstrated the relation between the achievable grain sizes lie within the submicrometer(100- enhanced strength and the extreme grain refinement 1000 nm)and nanometer (<100 nm)ranges.SPD-pro- imparted by SPD processing to a range of metals and cessed materials with such grain sizes are generally alloys.The seminal work of this group,emphasizing the referred to as nanoSPD materials [7],although only the great potential of SPD processing with regard to property latter ones can be regarded as being nanostructured improvement through grain structure modification,has according to the conventional definition.Several compre- heralded what has been described as the "microstructural hensive reviews have focused on various nanoSPD pro- age"of SPD research [4].Over the last decade,the nano- cessing techniques [22,28-33].We refer the reader to the SPD community (www.nanospd.org)has grown to an original works for specific details and only briefly outline impressive group of researchers,and thousands of publica- the general SPD methodology underlining the common tions on ultrafine-grained (UFG)and nanostructured features and the most important differences between the materials produced by SPD have been published.It is nanoSPD processes.By no means do we claim that our probably not surprising that in the year of the Diamond list of currently available manufacturing schemes is Jubilee of this journal,the Acta Materialia Gold Medal exhaustive. goes to Professor Terry Langdon-one of the world leaders After the landmark work by Bridgman mentioned above in the area of nanoSPD materials.A representative collec- [6.341.Langford and Cohen [35]and Rack and Cohen [36] tion of the relevant articles on the subject can be found in demonstrated in the 1960s that the microstructure of Fe- the proceedings of symposia on UFG materials [16,18]and 0.003%C subjected to high strains by wire drawing was nanoSPD conferences [19,20]the most recent ones in a refined to subgrain sizes in the 200-500 nm range.These series of five such forums.Further useful sources include microstructures could not be regarded as UFG proper in

similar effect is well known in fatigue where the grain size of wavy-slip materials has no bearing on the fatigue limit. These observations can be associated with the vital role of the dislocation substructure, which forms during defor￾mation (be it monotonic or cyclic), and it is the size of the substructure which determines the strength characteristics of metallic materials. Gow and Cahn [9] emphasized the significance of crystallographic texture for the deformation and recrystallization behaviour of metals and the effect of evolving texture on the resultant properties. Bell and Cahn [10] outlined many fine features of mechanical twinning, which play an important part in plastic deformation when accommodation by dislocation slip is hindered. Beck [11] highlighted the possibility of relieving the work-hardening effects by post-processing recovery. As will be seen in the following sections, these ideas have had a great impact on the development of the SPD processing and are pivotal to the modern concepts underlying these techniques. Now￾adays, the subject of SPD processing is represented very prominently on the pages of Acta, as illustrated by the analysis in Ref. [4]. Its revival is due to the work of Segal et al. [12] in the Soviet Union in the mid-1970s. These authors developed the method of equal-channel angular pressing (ECAP), which later evolved into what is now the most popular SPD technique. It should be mentioned, however, that in the time between the publication of Bridg￾man’s studies and the reintroduction of this subject in the metal science literature, exploration of the possibilities of changing the properties of materials through combined high pressure and shear deformation went on both in the Soviet Union and in the West. This less known work has been reviewed in Ref. [13]. In particular, credit should be given to the work by the group of N.S. Enikolopian con￾ducted mainly on polymers. A real appreciation for the new possibilities for improv￾ing the properties of metallic materials provided by SPD techniques came with the work of the group of Valiev [14,15], which demonstrated the relation between the enhanced strength and the extreme grain refinement imparted by SPD processing to a range of metals and alloys. The seminal work of this group, emphasizing the great potential of SPD processing with regard to property improvement through grain structure modification, has heralded what has been described as the “microstructural age” of SPD research [4]. Over the last decade, the nano￾SPD community (www.nanospd.org) has grown to an impressive group of researchers, and thousands of publica￾tions on ultrafine-grained (UFG) and nanostructured materials produced by SPD have been published. It is probably not surprising that in the year of the Diamond Jubilee of this journal, the Acta Materialia Gold Medal goes to Professor Terry Langdon—one of the world leaders in the area of nanoSPD materials. A representative collec￾tion of the relevant articles on the subject can be found in the proceedings of symposia on UFG materials [16,18] and nanoSPD conferences [19,20]—the most recent ones in a series of five such forums. Further useful sources include the reviews [21,22] and special issues of Advanced Engineer￾ing Materials [23], Materials Science and Engineering A [24] and Materials Transactions [25,26]. What makes SPD processing techniques so popular is the possibility of using them to enhance the strength char￾acteristics of conventional metallic materials in a quite spectacular way: by a factor of up to eight for pure metals such as copper and by some 30–50% for alloys [7,27]. Despite the impressive property improvement achievable with SPD techniques, their uptake by industry has been rather sluggish. However, things are now starting to change, and there is a common feeling in the nanoSPD community that major breakthroughs in terms of indus￾try-scale applications of SPD-based technologies are immi￾nent. We have been working in this area for more than a decade and have followed its developments closely. In this article we present our views on what has been achieved, what is possibly achievable, and what future trends are to be expected from SPD processing technologies. This article does not represent a full review of the SPD area (one could almost say the discipline of SPD, considering the firm place this group of material processing techniques has taken in literature). Rather, it is our personal take on the SPD area and an attempt to foretell its future development. Empha￾sis is placed on the scientifically challenging aspects of SPD, and not so much on technological issues, although some insights into the promises and limitations of SPD technologies will also be given. 2. SPD methods Among the procedures devised for grain refinement, SPD techniques are of particular interest and are the focus of the present review. These techniques enjoy great popularity owing to their ability to produce considerable grain refinement in fully dense, bulk-scale work-pieces, thus giving promise for structural applications. The achievable grain sizes lie within the submicrometer (100– 1000 nm) and nanometer (<100 nm) ranges. SPD-pro￾cessed materials with such grain sizes are generally referred to as nanoSPD materials [7], although only the latter ones can be regarded as being nanostructured according to the conventional definition. Several compre￾hensive reviews have focused on various nanoSPD pro￾cessing techniques [22,28–33]. We refer the reader to the original works for specific details and only briefly outline the general SPD methodology underlining the common features and the most important differences between the nanoSPD processes. By no means do we claim that our list of currently available manufacturing schemes is exhaustive. After the landmark work by Bridgman mentioned above [6,34], Langford and Cohen [35] and Rack and Cohen [36] demonstrated in the 1960s that the microstructure of Fe– 0.003% C subjected to high strains by wire drawing was refined to subgrain sizes in the 200–500 nm range. These microstructures could not be regarded as UFG proper in Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817 783

784 Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 Table 1 Schematic illustration of some basic and modern SPD techniques. Process Schematic illustration Equivalent strain Ref. Basic processes (a)Equal-channel angular pressing ef=N头cot(p) [39 (ECAP) N,the number of ECAP passes m⊙ → (b)High-pressure torsion (HPT) ei=N头g 34 r.the distance from the axis,t,the thickness of the sample,N,the number of revolutions (c)Accumulative roll bonding (ARB) e=N孟ln(》) [54 to.the initial thickness of the sample..the thickness of the sample after rolling,N.the number of passes (d)Multi-axial forging ew=N孟ln(g [57 Strain is non-uniform.N,the number of processing steps (e)Twist extrusion (TE) ≈0.4+0.1iar:时≈N1amm [61] y is the twist line slope:N is the number of passes. Deformation is non-uniform Dericative processes (f)Repetitive side extrusion ECAP equivalent [65] (continued on next page)

Table 1 Schematic illustration of some basic and modern SPD techniques. Process Schematic illustration Equivalent strain Ref. Basic processes (a) Equal-channel angular pressing (ECAP) eeff ¼ N 2ffiffi 3 p cotð/Þ N, the number of ECAP passes [39] (b)High-pressure torsion (HPT) eeff ¼ N 2ffiffi 3 p pr t r, the distance from the axis, t, the thickness of the sample, N, the number of revolutions [34] (c) Accumulative roll bonding (ARB) eeff ¼ N 2ffiffi 3 p lnð t0 tÞ t0, the initial thickness of the sample, t, the thickness of the sample after rolling, N, the number of passes [54] (d) Multi-axial forging eeff ¼ N 2ffiffi 3 p lnða bÞ Strain is non-uniform. N, the number of processing steps [57] (e) Twist extrusion (TE) emin eff 0:4 þ 0:1tanc; emax eff N 2ffiffi 3 p tanc c is the twist line slope; N is the number of passes. Deformation is non-uniform [61] Derivative processes (f) Repetitive side extrusion ECAP equivalent [65] (continued on next page) 784 Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817

Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 785 Table 1 (continued) Process Schematic illustration Equivalent strain Ref. (g)Rotary-die ECAP ECAP equivalent [66 (h)Cyclic extrusion-compression terr N4In() [75] (CEC) N,number of cycles (i)Cyclic close-die forging (CCDF) teff =Nn(#) [76) N,number of cycles (k)Repetitive corrugation and teff =Nn(ts) [72] straightening(RCS) N.number of cycles Integrated processes (1)Integrated extrusion+ECAP [110,11 (m)Parallel channel ECAP (PC. ECAP equivalent [67] ECAP) Continuous processes (n)ECAP-Conform [37][120]

Table 1 (continued) Process Schematic illustration Equivalent strain Ref. (g) Rotary-die ECAP ECAP equivalent [66] (h) Cyclic extrusion–compression (CEC) eeff ¼ N4 lnðD d Þ N, number of cycles [75] (i) Cyclic close-die forging (CCDF) eeff ¼ N 2ffiffi 3 p lnðH W Þ N, number of cycles [76] (k) Repetitive corrugation and straightening (RCS) eeff ¼ N 4ffiffi 3 p lnð rþt rþ0:5t Þ N, number of cycles [72] Integrated processes (l) Integrated extrusion + ECAP [110,111] (m) Parallel channel ECAP (PC￾ECAP) ECAP equivalent [67] Continuous processes (n) ECAP- Conform [37] [120] Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817 785

786 Y.Estrin.A.Vinogradov/Acta Materialia 61 (2013)782-817 Table 1 (continued) Process Schematic illustration Equivalent strain Ref. (o)Con-shearing [121] (p)Continuous confined strip [122] shearing (C2S2) (q)Continuous repetitive corrugating [73) and straightening (RCS) (r)Incremental ECAP(I-ECAP) [126 (s)Continuous high-pressure torsion [127] (t)Continuous manufacturing [129] of bolts the sense of the commonly accepted definitions [7],because between nanoSPD materials and more conventional mate- most of the sub-boundaries were low angle.Indeed,it is the rials with subgrain structures produced by cold rolling or prevalence of high-angle grain boundaries that is com- other common metal forming techniques.This distinction monly believed to be a signature of UFG materials manu- notwithstanding,these works opened the gates for micro- factured by SPD.This constitutes a clear demarcation line structure refinement by deformation to gigantic strains

the sense of the commonly accepted definitions [7], because most of the sub-boundaries were low angle. Indeed, it is the prevalence of high-angle grain boundaries that is com￾monly believed to be a signature of UFG materials manu￾factured by SPD. This constitutes a clear demarcation line between nanoSPD materials and more conventional mate￾rials with subgrain structures produced by cold rolling or other common metal forming techniques. This distinction notwithstanding, these works opened the gates for micro￾structure refinement by deformation to gigantic strains. Table 1 (continued) Process Schematic illustration Equivalent strain Ref. (o) Con-shearing [121] (p) Continuous confined strip shearing (C2S2) [122] (q) Continuous repetitive corrugating and straightening (RCS) [73] (r) Incremental ECAP (I-ECAP) [126] (s) Continuous high-pressure torsion [127] (t) Continuous manufacturing of bolts [129] 786 Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817

Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 787 Imparting large plastic strains to a work-piece is a chal- first requirement stimulated development of dies with lenging and technically formidable task.It requires a con- reduced friction by implementing surface hardening of siderable investment in tool design,which on one hand the channel walls,mobile walls [37,43],etc.,as well as the should be durable enough to sustain repetitive high loads introduction of new effective lubricants [441.The third during material forming,and on the other hand be suitable requirement led to the understanding of the significance for materials processing without causing damage to the of back-pressure for processing of billets with uniform work-piece.A distinctive feature of SPD processing,which microstructure and improved mechanical properties meets these requirements,is that the high strain is imposed [43,45,46].Segal showed that the performance of a die without any significant change in the overall dimensions of may be compromised by perceived simplicity of the die the work-piece.This is achieved due to special tool geom- design [38].In particular,he warned against the use of dies etries that prevent free flow of the material and thereby with a corner arc which leads to the occurrence of a widely produce a significant hydrostatic pressure.The presence spread fan-shaped plastic zone.This is equivalent to artifi- of this hydrostatic pressure is a clue for achieving the high cially increased friction that spreads shear and gives rise to strains required for exceptional grain refinement.Many significant heterogeneity of strain.Unfortunately,this crystalline materials,including those which are brittle important warning was disregarded in many later studies, under normal conditions(e.g.tungsten oxide,B2O3 glasses which utilized a"simplified"die design with a rounded and amorphous materials),gain substantial ductility under outer corner and had to pay a high price in form of sub- high hydrostatic pressure and can be deformed to large stantial heterogeneity of the deformed structure.By con- strains without failure.Nowadays many variants of SPD trast,by following Segal's philosophy,samples with techniques,which expressly or tacitly employ this generic uniform microstructure throughout the billet could be fab- feature of high hydrostatic pressure,are readily available ricated [47,48].While standard laboratory-scale ECAP rigs for fabrication of a great variety of UFG materials. can handle billets with cross-sectional dimensions in the range of 10-20 mm,Segal's ideas enabled development of 2.1.Principal processing schemes industry-scale ECAP facilities for processing of billets as large as 50 x 50 mm-in cross-section and 500 mm in length Equal-channel angular pressing (ECAP),less appropri- 「431 ately referred to as equal-channel angular extrusion High pressure torsion (HPT)refers to processing that (ECAE)in some publications,is at present the most highly evolved from Bridgman's anvils [6],cf.Table 1b,and developed SPD processing technique(Table la).A simple involves a combination of high(GPa range)pressure with shear strain is introduced when the billet passes through torsional straining.Today this technique is appreciated the plane where the two channels meet.The cross-sectional by many researchers as the one that allows the most effi- dimensions of the billet remain unchanged,thereby permit- cient grain refinement.A handicap of the method is that ting repetitive pressing,leading to accumulation of very only small coin-shaped samples,typically 10-15 mm in large strains.For example,the equivalent (von Mises) diameter and I mm in thickness,can be processed.The strain,Eeg,introduced per pass in ECAP with a 90 angle readers are referred to a comprehensive review on the sub- between the channels amounts to 1.15 [37,38].Different ject [30]for details.Because of size restrictions,the samples ECAP variants involving rotations of the billet about the manufactured by HPT are used primarily for research pur- pressing axis between the passes are possible,and they gen- poses.An important issue for many SPD processing erally lead to different results in terms of the microstructure schemes,including HPT,is the non-uniformity of deforma- and texture produced.The definitions of these ECAP tion.For instance,during HPT straining,Table Ib,the routes to which we refer below can be found in Refs. shear strain at the rotation axis should be zero,increasing [14.151. linearly in the radial direction if the geometry of the work- Against the backdrop of a flood of publications on piece does not change.This means that the material near ECAP processing,it is easy to forget where it all started. the rotation axis of the sample should stay undeformed. It is therefore timely to recall that the key advantages This is not supported by numerous microstructural obser- and fundamentals of ECAP.including the mechanics of vations and microhardness measurements showing a rea- extrusion,the derivation of the optimal process conditions sonably uniform distribution of grain dimensions and involving a balance between friction,tool geometry,strain microhardness,provided the compressive pressure and path and its efficiency for grain refinement,were formu- the number of revolutions of the anvil are sufficiently large lated by V.Segal in a series of early publications [12,38- (as in Fig.1)[49-51].Vorhauer and Pippan [52]explained 42].He defined ECAP as "a deformation technique to this discrepancy by the fact that it is virtually impossible to impart intensive,uniform and oriented simple shear for realize an ideal HPT deformation due to the misalignment materials processing".He also demonstrated that ECAP of the axes of the anvils.Alternatively,the development of is effective if(i)friction between the billet and the die walls a reasonably uniform strain (Fig.2)and homogeneous is kept at a minimum;(ii)the angle between the channels is microstructure was explained in terms of gradient plasticity close to 90:and (iii)the sharp outer corner is fully filled theory coupled with the microstructurally based constitu- ensuring that the shear zone is as narrow as possible.The tive modelling [53].This model will be addressed in the

Imparting large plastic strains to a work-piece is a chal￾lenging and technically formidable task. It requires a con￾siderable investment in tool design, which on one hand should be durable enough to sustain repetitive high loads during material forming, and on the other hand be suitable for materials processing without causing damage to the work-piece. A distinctive feature of SPD processing, which meets these requirements, is that the high strain is imposed without any significant change in the overall dimensions of the work-piece. This is achieved due to special tool geom￾etries that prevent free flow of the material and thereby produce a significant hydrostatic pressure. The presence of this hydrostatic pressure is a clue for achieving the high strains required for exceptional grain refinement. Many crystalline materials, including those which are brittle under normal conditions (e.g. tungsten oxide, B2O3 glasses and amorphous materials), gain substantial ductility under high hydrostatic pressure and can be deformed to large strains without failure. Nowadays many variants of SPD techniques, which expressly or tacitly employ this generic feature of high hydrostatic pressure, are readily available for fabrication of a great variety of UFG materials. 2.1. Principal processing schemes Equal-channel angular pressing (ECAP), less appropri￾ately referred to as equal-channel angular extrusion (ECAE) in some publications, is at present the most highly developed SPD processing technique (Table 1a). A simple shear strain is introduced when the billet passes through the plane where the two channels meet. The cross-sectional dimensions of the billet remain unchanged, thereby permit￾ting repetitive pressing, leading to accumulation of very large strains. For example, the equivalent (von Mises) strain, eeq, introduced per pass in ECAP with a 90 angle between the channels amounts to 1.15 [37,38]. Different ECAP variants involving rotations of the billet about the pressing axis between the passes are possible, and they gen￾erally lead to different results in terms of the microstructure and texture produced. The definitions of these ECAP routes to which we refer below can be found in Refs. [14,15]. Against the backdrop of a flood of publications on ECAP processing, it is easy to forget where it all started. It is therefore timely to recall that the key advantages and fundamentals of ECAP, including the mechanics of extrusion, the derivation of the optimal process conditions involving a balance between friction, tool geometry, strain path and its efficiency for grain refinement, were formu￾lated by V. Segal in a series of early publications [12,38– 42]. He defined ECAP as “a deformation technique to impart intensive, uniform and oriented simple shear for materials processing”. He also demonstrated that ECAP is effective if (i) friction between the billet and the die walls is kept at a minimum; (ii) the angle between the channels is close to 90; and (iii) the sharp outer corner is fully filled ensuring that the shear zone is as narrow as possible. The first requirement stimulated development of dies with reduced friction by implementing surface hardening of the channel walls, mobile walls [37,43], etc., as well as the introduction of new effective lubricants [44]. The third requirement led to the understanding of the significance of back-pressure for processing of billets with uniform microstructure and improved mechanical properties [43,45,46]. Segal showed that the performance of a die may be compromised by perceived simplicity of the die design [38]. In particular, he warned against the use of dies with a corner arc which leads to the occurrence of a widely spread fan-shaped plastic zone. This is equivalent to artifi- cially increased friction that spreads shear and gives rise to significant heterogeneity of strain. Unfortunately, this important warning was disregarded in many later studies, which utilized a “simplified” die design with a rounded outer corner and had to pay a high price in form of sub￾stantial heterogeneity of the deformed structure. By con￾trast, by following Segal’s philosophy, samples with uniform microstructure throughout the billet could be fab￾ricated [47,48]. While standard laboratory-scale ECAP rigs can handle billets with cross-sectional dimensions in the range of 10–20 mm, Segal’s ideas enabled development of industry-scale ECAP facilities for processing of billets as large as 50 50 mm2 in cross-section and 500 mm in length [43]. High pressure torsion (HPT) refers to processing that evolved from Bridgman’s anvils [6], cf. Table 1b, and involves a combination of high (GPa range) pressure with torsional straining. Today this technique is appreciated by many researchers as the one that allows the most effi- cient grain refinement. A handicap of the method is that only small coin-shaped samples, typically 10–15 mm in diameter and 1 mm in thickness, can be processed. The readers are referred to a comprehensive review on the sub￾ject [30] for details. Because of size restrictions, the samples manufactured by HPT are used primarily for research pur￾poses. An important issue for many SPD processing schemes, including HPT, is the non-uniformity of deforma￾tion. For instance, during HPT straining, Table 1b, the shear strain at the rotation axis should be zero, increasing linearly in the radial direction if the geometry of the work￾piece does not change. This means that the material near the rotation axis of the sample should stay undeformed. This is not supported by numerous microstructural obser￾vations and microhardness measurements showing a rea￾sonably uniform distribution of grain dimensions and microhardness, provided the compressive pressure and the number of revolutions of the anvil are sufficiently large (as in Fig. 1) [49–51]. Vorhauer and Pippan [52] explained this discrepancy by the fact that it is virtually impossible to realize an ideal HPT deformation due to the misalignment of the axes of the anvils. Alternatively, the development of a reasonably uniform strain (Fig. 2) and homogeneous microstructure was explained in terms of gradient plasticity theory coupled with the microstructurally based constitu￾tive modelling [53]. This model will be addressed in the Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817 787

788 Y.Estrin.A.Vinogradov/Acta Materialia 61 (2013)782-817 700 Al-Mg alloy AA5083 and interstitial-free steel,and was also used to process Al-and Mg-based laminated struc- 600 tures and composites [56].In addition,ARB can be applied for the production of metal-matrix composites by sheath- Annealed ing mixed powders and subjecting them to a roll-bonding 500 N=0 process [57].Similarly to conventional rolling,the UFG N=1 structure formed in the course of repetitive rolling is not -N=2 400 -N=4 equiaxed but rather exhibits pancake-like grains,irrespec- N=8 tive of the type of metal or alloy processed. Multi-directional forging (MDF),Table ld,was proposed 300 as a technique for structure refinement in the first half of the 1990s [58-60].Multiple free-forging operations include 200 repeated setting in three orthogonal directions.Since MDF is commonly performed in the temperature interval of 0.1- -2 .10 0.5T where T is the melting temperature,grain refinement Distance,mm during MDF is usually associated with dynamic recrystalli- Fig.1.Vickers microhardness,Hv,of HPT samples after different zation.The homogeneity of the strain produced by MDF numbers of turns,N,as a function of the distance from the specimen is lower than for ECAP or HPT.However,the method can centre.After Ref.[53](reprinted with permission). be used for microstructure refinement in rather brittle mate- rials owing to elevated temperatures and the low specific following section.Axial inhomogeneities observed in an loads on tooling involved.Furthermore,MDF was demon- HPT-processed Zr3Al intermetallic [54]were associated strated to be a potent technique for manufacturing large-size with softening effects related to nanostructuring. billets with microcrystalline(UFG)structures and was suc- Accumulative roll-bonding (ARB),Table Ic,was intro- cessfully applied to a wide range of materials [61]. duced by Saito et al.[55]as a process which was supposed Twist extrusion (TE),Table le,is another variant of a to overcome major limitations of the ECAP and HPT. simple shear deformation process that was introduced by namely the low productivity of the former and the small Beygelzimer et al.some ten years ago [62-64].Under TE work-piece size of the latter.The greatest technological processing,a prismatic billet is extruded through a"twist advantage of ARB is that it makes use of a conventional die".While the advantage of the process is its high upscal- rolling facility.A metal sheet is rolled to 50%thickness ing capacity,it suffers from essentially the same generic reduction.Then,the rolled sheet is cut in two and both problem as the HPT:deformation is non-uniform,being halves are stacked together,thus restoring the original smallest near the extrusion axis.Investigation by Orlov thickness of the sheet.The contact faces are degreased et al.[65]showed that this technique is slightly less effective and wire-brushed before stacking of the sheets,which are in producing UFG structure than ECAP or HPT. then rolled together to half the thickness.This sequence of rolling,cutting,brushing and stacking operations is 2.2.'Derivative'SPD processes repeated so that ultimately a large strain is accumulated in the sheet.ARB was successfully applied to a wide range Inspired by the success of the above "classical"SPD of materials,including commercial-purity (CP)Al,the techniques,more "exotic"methods were developed a)15 (b)80 1/3 turn -e-1/2turn 70 5 turns ◆-4/5tum 号1turn 60 HS 10 4 turns 50 40 3 turns 30 2 turns 20 10 2 0 2 DISTANCE FROM CENTRE(mm) DISTANCE FROM CENTRE(mm) Fig.2.Accumulated equivalent strain as a function of the distance from the torsion axis for the first-order gradient model [53].The degree of homogeneity is seen to increase with the overall strain,i.e.with the number of revolutions

following section. Axial inhomogeneities observed in an HPT-processed Zr3Al intermetallic [54] were associated with softening effects related to nanostructuring. Accumulative roll-bonding (ARB), Table 1c, was intro￾duced by Saito et al. [55] as a process which was supposed to overcome major limitations of the ECAP and HPT, namely the low productivity of the former and the small work-piece size of the latter. The greatest technological advantage of ARB is that it makes use of a conventional rolling facility. A metal sheet is rolled to 50% thickness reduction. Then, the rolled sheet is cut in two and both halves are stacked together, thus restoring the original thickness of the sheet. The contact faces are degreased and wire-brushed before stacking of the sheets, which are then rolled together to half the thickness. This sequence of rolling, cutting, brushing and stacking operations is repeated so that ultimately a large strain is accumulated in the sheet. ARB was successfully applied to a wide range of materials, including commercial-purity (CP) Al, the Al–Mg alloy AA5083 and interstitial-free steel, and was also used to process Al- and Mg-based laminated struc￾tures and composites [56]. In addition, ARB can be applied for the production of metal–matrix composites by sheath￾ing mixed powders and subjecting them to a roll-bonding process [57]. Similarly to conventional rolling, the UFG structure formed in the course of repetitive rolling is not equiaxed but rather exhibits pancake-like grains, irrespec￾tive of the type of metal or alloy processed. Multi-directional forging (MDF),Table 1d, was proposed as a technique for structure refinement in the first half of the 1990s [58–60]. Multiple free-forging operations include repeated setting in three orthogonal directions. Since MDF is commonly performed in the temperature interval of 0.1– 0.5Tm, whereTm is the melting temperature, grain refinement during MDF is usually associated with dynamic recrystalli￾zation. The homogeneity of the strain produced by MDF is lower than for ECAP or HPT. However, the method can be used for microstructure refinement in rather brittle mate￾rials owing to elevated temperatures and the low specific loads on tooling involved. Furthermore, MDF was demon￾strated to be a potent technique for manufacturing large-size billets with microcrystalline (UFG) structures and was suc￾cessfully applied to a wide range of materials [61]. Twist extrusion (TE), Table 1e, is another variant of a simple shear deformation process that was introduced by Beygelzimer et al. some ten years ago [62–64]. Under TE processing, a prismatic billet is extruded through a “twist die”. While the advantage of the process is its high upscal￾ing capacity, it suffers from essentially the same generic problem as the HPT: deformation is non-uniform, being smallest near the extrusion axis. Investigation by Orlov et al. [65] showed that this technique is slightly less effective in producing UFG structure than ECAP or HPT. 2.2. ‘Derivative’ SPD processes Inspired by the success of the above “classical” SPD techniques, more “exotic” methods were developed Fig. 1. Vickers microhardness, Hv, of HPT samples after different numbers of turns, N, as a function of the distance from the specimen centre. After Ref. [53] (reprinted with permission). Fig. 2. Accumulated equivalent strain as a function of the distance from the torsion axis for the first-order gradient model [53]. The degree of homogeneity is seen to increase with the overall strain, i.e. with the number of revolutions. 788 Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817

Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 789 recently with the aim of processing samples other than sim- diate annealing and/or post-ECAP processing by conven- ple rod or disk stock and/or enabling a higher throughput. tional rolling,drawing or extrusion.The advantageous Some of them are illustrated in Table 1.A list of these tech- effect of post-processing was confirmed by many research- niques (which is admittedly not exhaustive)includes: ers who combined different post-ECAP techniques to fur- ther enhance strength [103-105],modify texture [106]or repetitive side extrusion [66; improve ductility by subsequent annealing [107-109]. ·rotary die ECAP[67 Finally,new integrated processing schemes.which adopt .parallel channel ECAP [68]: features of different processes and combine them in a sin- hydrostatic extrusion [69-71]and hydrostatic extrusion gle-step integrated processing workflow [110-112],have combined with torsion [72]: recently been developed,cf.Table 11.The use of the inte- repetitive corrugating and straightening(RCS)for pro- grated semicontinuous processing techniques may be a cessing of sheets or plates [73-751 promising way of overcoming obstacles to uptake of SPD constrained groove pressing [76]; techniques by industry. cyclic extrusion-compression(CEC)[77]; Among the recent developments of SPD methods one cyclic closed-die forging (CCDF)[78]: can recognize a trend to target thin products,particularly cone-cone method(CCM)[79]; thin-walled tubes,and produce grain refinement by fric- cryogenic rolling [80,81]; tion-induced shear.One of the work-piece dimensions used asymmetric rolling (ASR)[82]; in such processes,namely the thickness,is much smaller continuous frictional angular extrusion(CFAE)[83,84]; than the other two dimensions.The cone-cone method friction stir processing(FSP)[85,86]; [79,113]and high-pressure tube twisting [96]are in that cat- super short interval multi-pass rolling(SSMR)[87,88]; egory,as is a modified tube-twisting technique suggested in severe torsion straining(STS)[89,90]; Ref.[114].Depending on the wall thickness,grain refine- torsion extrusion [91]; ment can be achieved throughout the tube wall thickness ECAP with rotation tooling in which the conventional or only within near-surface regions of the wall.This fixed die is replaced by rotating tools [92]; method was also applied for producing bimetallic Al-Cu reversed shear spinning [92]; tubes with ultrafine grain size(as small as about 140 nm transverse rolling [92]; near the interface of the two metals)[115]. non-equal channel angular pressing (NECAP)for plate- In a similar vein,Umemoto [116-118]made a point that shaped billets [93]; conventional metal processing techniques,such as shot tube channel pressing [94]; peening,drilling and wear [119],can be used as an effective ·KOBO forming[95l: way to create UFG structure and concomitant strengthen- high-pressure tube twisting (HPTT)for thin-walled ing in near-surface regions of metals and alloys. tubes [96]; cyclic expansion-extrusion CEE-a modified CEC pro- 2.3.Continuous SPD techniques and post-processing cess [97]: simple shear extrusion [98,99]; Many of the SPD methods presented above involve a vortex extrusion [100]; large number of discrete steps and are not labour and cost helical rolling [101]; efficient.Furthermore,they suffer from the inability to deli- .high-pressure sliding [102]. ver sufficiently large work-pieces as required for industry- scale applications.A number of approaches to SPD pro- From this list alone one can see that there are really no cessing seek to alleviate these disadvantages.In what fol- bounds to the imagination and resourcefulness of SPD pro- lows,we touch upon them briefly. cess designers.and more and more new SPD techniques Continuous forming (CONFORM),Table Im,is a well- have been emerging recently.From a purist's viewpoint, known process,the principles of which were first formu- not all of them would qualify to be termed "nanoSPD" lated by Etherington [120]with the aim of improving the processing according to the definition in Ref.[6],but most efficacy of materials recycling.They were later adapted of these techniques are cognate with the principal pro- by Segal and co-workers to continuous ECAP of bulk cesses-ECAP.HPT.TE or ARB-they derive from and materials [37].These principles were implemented by Raab bear some semblance with.Most of these processes use et al.in a rig for production of Al and Ti rods [121].In this shear deformation in conjunction with hydrostatic pressure process,the rod is placed in a groove within a rotating to produce large strains.The potential benefits of these shaft and is driven forward by frictional forces and then "derivative"techniques include simplified tool design, extruded through an outlet cannel of the die similarly to lower loads,reduced material loss,the possibility to pro- ECAP.A modification of this process was proposed by Sai- cess larger work-pieces,automated handling and/or poten- to et al.[122]as continuous shearing,Table 1o,for process- tial continuous operation. ing of sheets or strips. It is broadly recognized that strength and ductility may Continuous confined strip shearing (C2S2),sometimes greatly benefit from a combination of ECAP with interme- referred to as ECA-rolling process,Table 1p,is a

recently with the aim of processing samples other than sim￾ple rod or disk stock and/or enabling a higher throughput. Some of them are illustrated in Table 1. A list of these tech￾niques (which is admittedly not exhaustive) includes:  repetitive side extrusion [66];  rotary die ECAP [67];  parallel channel ECAP [68];  hydrostatic extrusion [69–71] and hydrostatic extrusion combined with torsion [72];  repetitive corrugating and straightening (RCS) for pro￾cessing of sheets or plates [73–75];  constrained groove pressing [76];  cyclic extrusion–compression (CEC) [77];  cyclic closed-die forging (CCDF) [78];  cone–cone method (CCM) [79];  cryogenic rolling [80,81];  asymmetric rolling (ASR) [82];  continuous frictional angular extrusion (CFAE) [83,84];  friction stir processing (FSP) [85,86];  super short interval multi-pass rolling (SSMR) [87,88];  severe torsion straining (STS) [89,90];  torsion extrusion [91];  ECAP with rotation tooling in which the conventional fixed die is replaced by rotating tools [92];  reversed shear spinning [92];  transverse rolling [92];  non-equal channel angular pressing (NECAP) for plate￾shaped billets [93];  tube channel pressing [94];  KOBO forming [95];  high-pressure tube twisting (HPTT) for thin-walled tubes [96];  cyclic expansion–extrusion CEE—a modified CEC pro￾cess [97];  simple shear extrusion [98,99];  vortex extrusion [100];  helical rolling [101];  high-pressure sliding [102]. From this list alone one can see that there are really no bounds to the imagination and resourcefulness of SPD pro￾cess designers, and more and more new SPD techniques have been emerging recently. From a purist’s viewpoint, not all of them would qualify to be termed “nanoSPD” processing according to the definition in Ref. [6], but most of these techniques are cognate with the principal pro￾cesses—ECAP, HPT, TE or ARB—they derive from and bear some semblance with. Most of these processes use shear deformation in conjunction with hydrostatic pressure to produce large strains. The potential benefits of these “derivative” techniques include simplified tool design, lower loads, reduced material loss, the possibility to pro￾cess larger work-pieces, automated handling and/or poten￾tial continuous operation. It is broadly recognized that strength and ductility may greatly benefit from a combination of ECAP with interme￾diate annealing and/or post-ECAP processing by conven￾tional rolling, drawing or extrusion. The advantageous effect of post-processing was confirmed by many research￾ers who combined different post-ECAP techniques to fur￾ther enhance strength [103–105], modify texture [106] or improve ductility by subsequent annealing [107–109]. Finally, new integrated processing schemes, which adopt features of different processes and combine them in a sin￾gle-step integrated processing workflow [110–112], have recently been developed, cf. Table 1l. The use of the inte￾grated semicontinuous processing techniques may be a promising way of overcoming obstacles to uptake of SPD techniques by industry. Among the recent developments of SPD methods one can recognize a trend to target thin products, particularly thin-walled tubes, and produce grain refinement by fric￾tion-induced shear. One of the work-piece dimensions used in such processes, namely the thickness, is much smaller than the other two dimensions. The cone–cone method [79,113] and high-pressure tube twisting [96] are in that cat￾egory, as is a modified tube-twisting technique suggested in Ref. [114]. Depending on the wall thickness, grain refine￾ment can be achieved throughout the tube wall thickness or only within near-surface regions of the wall. This method was also applied for producing bimetallic Al–Cu tubes with ultrafine grain size (as small as about 140 nm near the interface of the two metals) [115]. In a similar vein, Umemoto [116–118] made a point that conventional metal processing techniques, such as shot peening, drilling and wear [119], can be used as an effective way to create UFG structure and concomitant strengthen￾ing in near-surface regions of metals and alloys. 2.3. Continuous SPD techniques and post-processing Many of the SPD methods presented above involve a large number of discrete steps and are not labour and cost efficient. Furthermore, they suffer from the inability to deli￾ver sufficiently large work-pieces as required for industry￾scale applications. A number of approaches to SPD pro￾cessing seek to alleviate these disadvantages. In what fol￾lows, we touch upon them briefly. Continuous forming (CONFORM), Table 1m, is a well￾known process, the principles of which were first formu￾lated by Etherington [120] with the aim of improving the efficacy of materials recycling. They were later adapted by Segal and co-workers to continuous ECAP of bulk materials [37]. These principles were implemented by Raab et al. in a rig for production of Al and Ti rods [121]. In this process, the rod is placed in a groove within a rotating shaft and is driven forward by frictional forces and then extruded through an outlet cannel of the die similarly to ECAP. A modification of this process was proposed by Sai￾to et al. [122] as continuous shearing, Table 1o, for process￾ing of sheets or strips. Continuous confined strip shearing (C2S2), sometimes referred to as ECA-rolling process, Table 1p, is a Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817 789

790 Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 modification of the CONFORM method for processing of sheets or strips [123,124]. 250um 0☐255 Continuous ECAP for sheet manufacturing was dis- cussed by Lapovok et al.[125,126].For the Al alloys tested it was established that just a single ECAP pass was suffi- cient to obtain a significant reduction of normal and in- plane anisotropy.A variant of the process is continuous equal-channel angular drawing [125,126]. Repetitive corrugating and straightening (RCS)has an obvious advantage of providing a simple modification of rolling to enable continuous SPD processing,as illustrated graphically in Table 1q [74,75]. Incremental ECAP (I-ECAP).Rosochowski and co- authors extended general knowledge of incremental metal forming operations,such as rolling,swaging or rotary forging,and adapted it to ECAP by modifying it for pro- cessing of long billets.This process was dubbed incremen- tal ECAP (I-ECAP)[127].The basic version of I-ECAP is shown in Table Ir.The deformation mode is simple shear. and it is uniform within the marked zone similarly to the Fig.3.Steel preform with 100 um porosity infiltrated with Al (seen in red) shear deformation in "classical"ECAP.Separation of the as revealed by energy dispersive spectroscopy.A90 ECAP-like"knee"is feeding and deformation stages reduces or eliminates fric- seen in the bottom left part of the picture.After Ref.[132](reprinted with permission). tion during feeding;this substantially reduces the feeding force and enables processing of very long or continuous billets. Concluding this section we would like to note that a great Continuous manufacturing of bolts by ECAP.The group variety of SPD techniques are now available.Their com- of Prof.Y.-T.Im at KAIST in Korea developed a method mon features are the high hydrostatic pressure and the tool [128.129]which overcomes the discrete character of ECAP geometry permitting multipass operation to achieve ultra- by integrating what they call"spring-loaded ECAp"in a high strains.Differences are mainly related to the deforma- continuous bolt manufacturing process (Table It). tion mode,the work-piece shape.the efficacy with respect to AA6016 bolts produced using this technology were shown the strain imposed per pass and the load involved.All these to be superior to those manufactured in a conventional way factors affect,to a varying extent,the resultant microstruc- in terms of their tensile strength and fatigue strength. ture,the properties of the product and the upscaling capac- Continuous high-pressure torsion.An advanced version ity of the technique used.These aspects of SPD processing of the HPT technique was proposed by Edalati and Horita will be addressed in the next sections. [130],who demonstrated its viability as a method to pro- A great advantage of the SPD techniques is that they are duce sheets 0.6-mm thick and 3 mm wide,which possess based on a "top-down"approach involving grain refine- UFG structure,in a continuous fashion,cf.Table Is. ment through "breaking down"the microstructure of the While for most structural applications upscaling of the bulk to the submicron scale.SPD processing is thus free SPD technologies is required,there may be niche applica- from problems of excessive residual porosity and contami- tions where downscaling would be desirable.The feasibility nation,which are common in nanostructured materials of such downscaling was demonstrated for the ECAP pro- manufactured in a"bottom-up"fashion,e.g.by consolida- cess [131].Miniaturized dies with channel diameters in the tion of nanopowders.Furthermore,no health hazards millimeter range were used to deform Al specimens and potentially associated with handling of nanopowders are achieve grain refinement in a single pass. involved in SPD processing. An SPD-like process of an entirely different type was As will be seen below,perhaps the most important dis- proposed by Estrin et al.[132].In this "solid-state infiltra- advantage of SPD is that the efficiency of grain refinement tion"method,solid aluminium was forced to fill a porous drops with strain [134].A way to overcome the problem by steel preform under high pressure in much the same way suppressing dynamic recovery,e.g.by using SPD process- vias are filled with metal in fabrication of metallic intercon- ing at cryogenic temperatures,was suggested in Ref.[81]. nects by the force-fill process in microelectronics [1331.The However,the microstructures obtained in this way retain random paths taken by the plastically flowing Al are pretty a large volume fraction of low-angle boundaries,giving rise tortuous and involve numerous kinks.Some of them may to considerable thermal instability and coarsening of the be similar to those seen in ECAP channels and induce ultrafine microstructure produced. ECAP-like localized shear zones and ensuing grain refine- Given the rapid progress in the field,we are confident ment.Penetration of Al into the porous steel preform is that new processes with higher throughput,upscaling illustrated in Fig.3. capacity and greater cost-effectiveness will emerge,meeting

modification of the CONFORM method for processing of sheets or strips [123,124]. Continuous ECAP for sheet manufacturing was dis￾cussed by Lapovok et al. [125,126]. For the Al alloys tested it was established that just a single ECAP pass was suffi- cient to obtain a significant reduction of normal and in￾plane anisotropy. A variant of the process is continuous equal-channel angular drawing [125,126]. Repetitive corrugating and straightening (RCS) has an obvious advantage of providing a simple modification of rolling to enable continuous SPD processing, as illustrated graphically in Table 1q [74,75]. Incremental ECAP (I-ECAP). Rosochowski and co￾authors extended general knowledge of incremental metal forming operations, such as rolling, swaging or rotary forging, and adapted it to ECAP by modifying it for pro￾cessing of long billets. This process was dubbed incremen￾tal ECAP (I-ECAP) [127]. The basic version of I-ECAP is shown in Table 1r. The deformation mode is simple shear, and it is uniform within the marked zone similarly to the shear deformation in “classical” ECAP. Separation of the feeding and deformation stages reduces or eliminates fric￾tion during feeding; this substantially reduces the feeding force and enables processing of very long or continuous billets. Continuous manufacturing of bolts by ECAP. The group of Prof. Y.-T. Im at KAIST in Korea developed a method [128,129] which overcomes the discrete character of ECAP by integrating what they call “spring-loaded ECAP” in a continuous bolt manufacturing process (Table 1t). AA6016 bolts produced using this technology were shown to be superior to those manufactured in a conventional way in terms of their tensile strength and fatigue strength. Continuous high-pressure torsion. An advanced version of the HPT technique was proposed by Edalati and Horita [130], who demonstrated its viability as a method to pro￾duce sheets 0.6-mm thick and 3 mm wide, which possess UFG structure, in a continuous fashion, cf. Table 1s. While for most structural applications upscaling of the SPD technologies is required, there may be niche applica￾tions where downscaling would be desirable. The feasibility of such downscaling was demonstrated for the ECAP pro￾cess [131]. Miniaturized dies with channel diameters in the millimeter range were used to deform Al specimens and achieve grain refinement in a single pass. An SPD-like process of an entirely different type was proposed by Estrin et al. [132]. In this “solid-state infiltra￾tion” method, solid aluminium was forced to fill a porous steel preform under high pressure in much the same way vias are filled with metal in fabrication of metallic intercon￾nects by the force-fill process in microelectronics [133]. The random paths taken by the plastically flowing Al are pretty tortuous and involve numerous kinks. Some of them may be similar to those seen in ECAP channels and induce ECAP-like localized shear zones and ensuing grain refine￾ment. Penetration of Al into the porous steel preform is illustrated in Fig. 3. Concluding this section we would like to note that a great variety of SPD techniques are now available. Their com￾mon features are the high hydrostatic pressure and the tool geometry permitting multipass operation to achieve ultra￾high strains. Differences are mainly related to the deforma￾tion mode, the work-piece shape, the efficacy with respect to the strain imposed per pass and the load involved. All these factors affect, to a varying extent, the resultant microstruc￾ture, the properties of the product and the upscaling capac￾ity of the technique used. These aspects of SPD processing will be addressed in the next sections. A great advantage of the SPD techniques is that they are based on a “top-down” approach involving grain refine￾ment through “breaking down” the microstructure of the bulk to the submicron scale. SPD processing is thus free from problems of excessive residual porosity and contami￾nation, which are common in nanostructured materials manufactured in a “bottom-up” fashion, e.g. by consolida￾tion of nanopowders. Furthermore, no health hazards potentially associated with handling of nanopowders are involved in SPD processing. As will be seen below, perhaps the most important dis￾advantage of SPD is that the efficiency of grain refinement drops with strain [134]. A way to overcome the problem by suppressing dynamic recovery, e.g. by using SPD process￾ing at cryogenic temperatures, was suggested in Ref. [81]. However, the microstructures obtained in this way retain a large volume fraction of low-angle boundaries, giving rise to considerable thermal instability and coarsening of the ultrafine microstructure produced. Given the rapid progress in the field, we are confident that new processes with higher throughput, upscaling capacity and greater cost-effectiveness will emerge, meeting Fig. 3. Steel preform with 100 lm porosity infiltrated with Al (seen in red) as revealed by energy dispersive spectroscopy. A 90 ECAP-like “knee” is seen in the bottom left part of the picture. After Ref. [132] (reprinted with permission). 790 Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817

Y.Estrin,A.Vinogradov/Acta Materialia 61 (2013)782-817 791 the demand for advanced high-performance structural dp ko materials in modern industries. de bL-kap, (1) with the intrinsic length scale L corresponding to a charac- 3.Mechanisms of grain refinement by SPD teristic length scale of the cell structure,e.g.the cell size, that determines the dislocation mean free path.Here ko is The main aim of SPD processing,its ultimate raison a constant or a slowly varying quantity and k2 is a mecha- d'etre,is extreme grain refinement and the ensuing nism-dependent phenomenological parameter sensitive to strengthening of the processed material.There is no longer strain rate and temperature.The Kocks-Mecking model any doubt that this is achievable with most malleable and has been extremely successful in providing a description even with many hard-to-deform materials,and innumera- of stages II and III of strain hardening.However,an ade- ble experimental results documented in review articles quate description of stages IV and V of strain hardening, and conference proceedings(e.g.[16-20))are a convincing which are predominant at large strains [146],requires a testimony to that.Despite this body of experimental evi- more detailed representation of the dislocation population. dence,the mechanisms of grain refinement,which are piv- It involves treating dislocations in the dislocation cell walls otal in designing the routes to property improvement,are and cell interiors as separate entities,thus introducing two far from being understood.In particular,there is no gener- distinct evolving dislocation densities.This was done by ally accepted scenario of grain fragmentation by subdivi- Prinz and Argon [147]and later by Nix et al.[148]who sion of grains,and the underlying processes have adopted the approach proposed by Mughrabi [149,150], remained a riddle for researchers to the present day. in which the cell walls and cell interiors are treated as two distinct phases of a 'composite'.These models were 3.1.Disclination models of grain fragmentation able to account for stage IV and stage V hardening.More complex models with three internal variables [151]also Early attempts at unravelling this riddle go back to the provided an adequate description of mechanical response work of Honeycombe [135],Rybin [136-137],Vladimirov at large strains and were successfully applied for metal and Kusov [1381,Mughrabi et al.[139].Indenbom and forming simulations [152]. Orlov [140]and others.This research has prepared the Estrin et al.[153,154]and Zehetbauer et al.[146,155,156] ground for models introducing grain fragmentation by dis- proposed constitutive models based on Mughrabi's com- location wall formation,notably in form of rows of dislo- posite principle and detailing the evolution of the disloca- cation dipoles.These concepts have led to a description tion densities in the cell walls and cell interiors,including of the grain fragmentation process in terms of disclinations interactions between the two "phases"of the composite. [141,142].The process of grain subdivision is represented While Zehetbauer's model postulated constancy of the vol- by the nucleation and propagation of incomplete disclina- ume fraction of cell walls,Estrin et al.emphasized that this tions,producing misorientation between the adjacent grain volume fraction must decrease during stage IV of harden- fragments.Models based on coupled disclination-disloca- ing in order to account for the nearly constant hardening tion dynamics were shown to provide a reasonably good coefficient commonly found experimentally in stage IV. description of the microstructures formed at large strains The two-dislocation density models [153,154]have [142].A recent review of disclination-based models become a useful platform for modelling SPD processes addressing,in particular,bulk nanostructured materials (cf.[155,157,158)),and we shall present them in a condensed can be found in Ref.[143] form here.The models apply to dislocation cell-forming materials and address sufficiently large strains when a cell 3.2.Dislocation density based models structure is already formed.The "primordial"stage of the deformation in which this happens is not considered. The most commonly accepted type of models of grain Understanding the self-organization processes leading to refinement due to large strain,particularly under SPD con- the emergence of a dislocation cell structure is one of the ditions,are based on the notion that a dislocation cell greatest unsolved problems of the dislocation theory,and structure,which forms already in the early stages of plastic numerous modelling efforts,including discrete dislocation deformation,gradually transforms to the final fine grain dynamics simulations [159],are being devoted to this prob- structure.This is believed to occur through continual lem.An assumption made in the models [153,154]is that the decrease in the average grain size accompanied by accumu- average dislocation cell size,d,scales inversely with the lation of misorientation between neighbouring dislocation square root of the total dislocation density,p: cells.This type of model goes back to Kocks and Mecking [144,145]who described the deformation behaviour of met- d=K/vp, (2) als and alloys in terms of a single internal variable:the total where K is a proportionality constant.Eq.(2)goes back to dislocation density p.Within this approach,the dislocation an early work of Holt [160]in which patterning in a kinetics equation governing the evolution of the total dislo- dislocation population was first considered.While a sub- cation density is represented in its simplest,yet rather gen- stantial body of evidence supports the validity of Eq.(2) eral,way as: for steady-state deformation,extending it to the dynamic

the demand for advanced high-performance structural materials in modern industries. 3. Mechanisms of grain refinement by SPD The main aim of SPD processing, its ultimate raison d’eˆtre, is extreme grain refinement and the ensuing strengthening of the processed material. There is no longer any doubt that this is achievable with most malleable and even with many hard-to-deform materials, and innumera￾ble experimental results documented in review articles and conference proceedings (e.g. [16–20]) are a convincing testimony to that. Despite this body of experimental evi￾dence, the mechanisms of grain refinement, which are piv￾otal in designing the routes to property improvement, are far from being understood. In particular, there is no gener￾ally accepted scenario of grain fragmentation by subdivi￾sion of grains, and the underlying processes have remained a riddle for researchers to the present day. 3.1. Disclination models of grain fragmentation Early attempts at unravelling this riddle go back to the work of Honeycombe [135], Rybin [136–137], Vladimirov and Kusov [138], Mughrabi et al. [139], Indenbom and Orlov [140] and others. This research has prepared the ground for models introducing grain fragmentation by dis￾location wall formation, notably in form of rows of dislo￾cation dipoles. These concepts have led to a description of the grain fragmentation process in terms of disclinations [141,142]. The process of grain subdivision is represented by the nucleation and propagation of incomplete disclina￾tions, producing misorientation between the adjacent grain fragments. Models based on coupled disclination–disloca￾tion dynamics were shown to provide a reasonably good description of the microstructures formed at large strains [142]. A recent review of disclination-based models addressing, in particular, bulk nanostructured materials can be found in Ref. [143]. 3.2. Dislocation density based models The most commonly accepted type of models of grain refinement due to large strain, particularly under SPD con￾ditions, are based on the notion that a dislocation cell structure, which forms already in the early stages of plastic deformation, gradually transforms to the final fine grain structure. This is believed to occur through continual decrease in the average grain size accompanied by accumu￾lation of misorientation between neighbouring dislocation cells. This type of model goes back to Kocks and Mecking [144,145] who described the deformation behaviour of met￾als and alloys in terms of a single internal variable: the total dislocation density q. Within this approach, the dislocation kinetics equation governing the evolution of the total dislo￾cation density is represented in its simplest, yet rather gen￾eral, way as: dq de ¼ k0 bL  k2q; ð1Þ with the intrinsic length scale L corresponding to a charac￾teristic length scale of the cell structure, e.g. the cell size, that determines the dislocation mean free path. Here k0 is a constant or a slowly varying quantity and k2 is a mecha￾nism-dependent phenomenological parameter sensitive to strain rate and temperature. The Kocks–Mecking model has been extremely successful in providing a description of stages II and III of strain hardening. However, an ade￾quate description of stages IV and V of strain hardening, which are predominant at large strains [146], requires a more detailed representation of the dislocation population. It involves treating dislocations in the dislocation cell walls and cell interiors as separate entities, thus introducing two distinct evolving dislocation densities. This was done by Prinz and Argon [147] and later by Nix et al. [148] who adopted the approach proposed by Mughrabi [149,150], in which the cell walls and cell interiors are treated as two distinct phases of a ‘composite’. These models were able to account for stage IV and stage V hardening. More complex models with three internal variables [151] also provided an adequate description of mechanical response at large strains and were successfully applied for metal forming simulations [152]. Estrin et al. [153,154] and Zehetbauer et al. [146,155,156] proposed constitutive models based on Mughrabi’s com￾posite principle and detailing the evolution of the disloca￾tion densities in the cell walls and cell interiors, including interactions between the two “phases” of the composite. While Zehetbauer’s model postulated constancy of the vol￾ume fraction of cell walls, Estrin et al. emphasized that this volume fraction must decrease during stage IV of harden￾ing in order to account for the nearly constant hardening coefficient commonly found experimentally in stage IV. The two-dislocation density models [153,154] have become a useful platform for modelling SPD processes (cf. [155,157,158]), and we shall present them in a condensed form here. The models apply to dislocation cell-forming materials and address sufficiently large strains when a cell structure is already formed. The “primordial” stage of the deformation in which this happens is not considered. Understanding the self-organization processes leading to the emergence of a dislocation cell structure is one of the greatest unsolved problems of the dislocation theory, and numerous modelling efforts, including discrete dislocation dynamics simulations [159], are being devoted to this prob￾lem. An assumption made in the models [153,154] is that the average dislocation cell size, d, scales inversely with the square root of the total dislocation density, q: d ¼ K= ffiffiffi q p ; ð2Þ where K is a proportionality constant. Eq. (2) goes back to an early work of Holt [160] in which patterning in a dislocation population was first considered. While a sub￾stantial body of evidence supports the validity of Eq. (2) for steady-state deformation, extending it to the dynamic Y. Estrin, A. Vinogradov / Acta Materialia 61 (2013) 782–817 791