Materials Science and Engineering A238(1997)219–274
Current issues in recrystallization:a review R.D.Doherty a,D.A.Hughes b,F.J.Humphreys c,J.J.Jonas d,D.Juul Jenn e, M.E.Kassner f,*,W.E.King g,T.R.McNelley h,H.J.McQueen i,A.D.Rollett j
a Department of Materials Engineering,Drexel Uni6ersity,Philadelphia,PA19104,USA
b Center for Materials and Applied Mechanics,Sandia National Laboratory,Li6ermore,CA94550,USA
六年级下册语文课本c Manchester Materials Science Center,UMIST an
d Uni6ersity of Manchester,Manchester M17H5,UK
d Department of Metallurgical Engineering,McGill Uni6ersity,Montreal,Quebec H3A2A7,Canada
e Materials Department,RisøNational Laboratory,Roskilde DK-4000,Denmark
f Department of Mechanical Engineering,Oregon State Uni6ersity,Co6allis,OR97331,USA
g Chemistry and Materials Science Directorate,Lawrence Li6ermore National Laboratory,Li6ermore,CA94550,USA
h Department of Mechanical Engineering,Na6al Postgraduate School,Monterey,CA93943,USA
i Department of Mechanical Engineering,Concordia Uni6ersity,Montreal63G1M8,Canada
j Department of Materials Science and Engineering,Carnegie-Mellon Uni6ersity,Pittsburgh,PA15213,USA
Received11April1997
Abstract
The current understanding of the fundamentals of recrystallization is summarized.This includes understanding the as-deformed state.Several aspects of recrystallization are described:nucleation and growth,the development of misorientation during deformation,continuous,dynamic,and geometric dynamic recrystallization,particle effects,and texture.This article is authored by the leading experts in the areas.The subjects are discusd individually and recommendations for further study are listed in thefinal ction.©1997Elvier Science S.A.
Keywords:As-deformed state;Nucleation;Recrystallization
民族团结月1.Introduction
The objectives of this article are two-fold.First,the current understanding of the fundamentals of recrystal-lization is summarized.This includes understanding the cold and hot-deformed state.Next,with the state of the art established,recommendations for future rearch are made.Several aspects of recrystallization are de-scribed.The authors of this paper are the contributors to each aspect described in a parate ction.The are listed below with the authors of each ction identified. Overall editing was performed by external reviewers as well as the contributors.
1.Introduction(R.D.Doherty and M.E.Kassner)
2.Theories of nucleation and growth during recrys-
简笔画雪花tallization(R.D.Doherty)
3.Formation of deformation induced high angle
boundaries and their effect on recrystallization
(D.A.Hughes and D.Juul Jenn)
4.Issues in texture development and simulation of
recrystallization(A.D.Rollett)
5.Second pha particles and recrystallization(F.J.
Humphreys)
6.Conventional dynamic recrystallization(DRX)
(J.J.Jonas)
7.Continuous reactions(T.R.McNelley)
8.Geometric Dynamic recrystallization(M.E.Kass-
ner)
9.The hot worked state(H.J.McQueen)
10.The role of grain boundaries in recrystallization
(W.E.King)
11.Recommendations for further study(all authors) It is,of cour,uful to carefully define the term ‘recrystallization’.The authors have agreed that recrys-tallization is the formation of a new grain structure in a deformed material by the formation and migration of high angle grain boundaries driven by the stored energy
*Corresponding author.Fax:+15417372600;e-mail:st.edu
0921-5093/97/$17.00©1997Elvier Science S.A.All rights rerved. PII S0921-5093(97)00424-3
R.D.Doherty et al./Materials Science and Engineering A238(1997)219–274 220
of deformation.High angle grain boundaries are tho with greater than a10–15°misorientation.Recovery can be defined as all annealing process occurring in deformed materials that occur without the migration of a high angle grain boundary.Grain coarning can,in turn,be defined as process involving the migration of grain boundaries when the driving force for migration is solely the reduction of the grain boundary area itlf. The definitions are consistent with some earlier defin-itions[1].螺丝面
2.Theories of nucleation and growth during recrystallization
2.1.Introduction
The theories and recent experimental insights into the process of nucleation and growth are reviewed with emphasis on what is not yet fully understood.In the light of the views,a range of needed new theoretical and experimental studies is propod to improve the understanding and modelling of recrystallization mecha-nisms.
The process of recrystallization of plastically de-formed metals and alloys is of central importance in the processing of metallic alloys for two main reasons.The first is to soften and restore the ductility of material hardened by low temperature deformation(that occur-ring below about50%of the absolute melting tempera-ture,0.5T m).The cond is to control the grain structure of thefinal product.In metals,such as iron,titanium, and cobalt that undergo a pha change on cooling,the grain structure is readily modified by control of the pha transformation.For all other metallic alloys, especially tho bad on copper,nickel,and aluminum, recrystallization after deformation is the only method for producing a completely new grain structure with a modified grain size,shape,and,in particular,mean orientation or texture.The subject has been recently given a long overdue review in the monograph by Humphreys and Hatherly[2]that nicely complements the much earlier multi-authored volume edited by Haessner[3].This ction aims to summarize the current status of the still
rather limited scientific understanding of the two central process of recrystallization—nucle-ation and growth of new grains—with the objective of focusing on what,in the authors’opinion,ems to be the necessary new studies for improved scientific under-standing of the process.Although there is a great deal of empirical knowledge of the microstructures that can be produced during current industrial processing,the ability to produce more nearly ideal microstructures for different applications is very limited and it is in order to gain improved control of recrystallization processing that incread scientific understanding is needed.
During deformation energy is stored in the material mainly in the form of dislocations.This energy is relead in three main process,tho of recovery, recrystallization,and grain coarning.The usual defin-ition of recrystallization[1]is the formation and migra-tion of high angle grain boundaries driven by the stored energy of definition.On this definition recovery includes all process releasing stored energy that do not require the movement of a high angle grain boundary.Typi-cally,recovery process involve the rearrangement of dislocations to lower their energy,for example by the formation of low-angle subgrain boundaries.Grain coarning is the growth of the mean grain size driven by the reduction in grain boundary area[4,5].Coarn-ing can take place by either‘normal’grain growth, who main mechanism is the disappearance of the smallest grains in th
e distribution,or‘abnormal’grain growth.The latter process involves the growth of a few grains which become much larger than the average. 2.2.Discussion
2.2.1.Nucleation and growth in recrystallization
波尔效应
In all structural transformations there are two alter-native types of transformation as originally recognized by Gibbs,e for example Doherty[6].In thefirst of the,Gibbs I,typically called‘nucleation and growth’, the transformation is extensive in the magnitude of the structural change but is,initially,spatially localized with a sharp interface between the old and new struc-tures.The cond type of transformation,Gibbs II, often described as‘continuous’or‘homogeneous’(the best known example being spinodal decomposition),the transformation is initially small in the magnitude of the structural change,but it occurs throughout the parent structure.In the range of process en on annealing plastically deformed materials,both dislocation recov-ery,that takes place before and during recrystallization and also normal grain growth are clearly Gibbs II transformations which occur uniformly throughout the sample while recrystallization and abnormal grain growth are Gibbs I transformations—at least on the obrvational length scales of about1–5m m for recrys-tallization or about0.1–1mm for abnormal grain growth[2].At the length scales,typically studied by optical microscopy,the new recrystallized g
rain or the abnormally large grains,are en to be growing into the prior structure with a sharp interface,a grain boundary,as the‘recrystallization front’between the deformed and new grains(e Fig.1).The usual name of‘nucleation and growth’for a Gibbs I transformation is bad on the two apparently distinct steps in the process:(i)the initial formation of the new grain;and (ii)its growth.
R.D.Doherty et al./Materials Science and Engineering A238(1997)219–274221
In the scientific study of pha transformations,one kinetic model of nucleation has been dominant.This is the thermalfluctuation model initially developed in physical chemistry[8,9]but applied very successfully to solidification and then to solid state pha transitions by Turnbull[10,11]as described very fully in the review by Christian[12]and recently updated[6].If there is a volume free energy driving pressure of D G v(in units of Jm−3or Pa),an interfacial energy of k(in units of Jm−2)between the old and new structures and,for heterogeneous nucleation of a defect interface,a con-tact angle q,there is an energy barrier,D G*,to the formation of a critically sized new region(usually called ‘embryos’)that are just stable and capable of growth.
D G*is given by Eq.(1):
D G*={hk3/D G2v}f(cos q)(1) h is a number that varies with the shape of the new
region,for example h is16y/3for a spherical nucleus, and the function f(cos q),that is typically between0.1and0.5,depending on the geometry of the defect[12].
In a matrix containing N v atoms per unit volume,the
density of critical embryos,n v*,is given by Eq.(2):
n v*=N v exp(−D G*/kT)(2) and the rate of formation of new grains,I v(m−3s−1)
is:
I v=i n v*(3) The kinetic parameter i involves various terms that
include the rate of atom addition to the embryo(pro-
形容知识渊博portional to the interface mobility)and the reduction in
the equilibrium value of n v*,due to the loss of embryos
as they evolve into growing new particles.
As reviewed recently[6],the predictions of the kinetic
theory are found to be in excellent qualitative agree-
ment with a vast range of experimental behavior and,in
a few cas,for example,for homogeneous nucleation
in solidification and homogeneous precipitation reac-
tions with a low energy fully coherent interface between
phas of very similar structure(GP zones in Cu–Co
and ordered k%precipitates,Ni3Al,in Ni–Al),quantita-
tive agreement as well.A major problem for the study
of nucleation in recrystallization is that it is easily
shown,for example[13],that given the typically low
values of the stored energy of deformation,D G v:0.1–1MPa[2]and the high value of the energy of a high
angle grain boundary,k:0.5Jm−2[2]that D G*is so
large,of order108kT,that new grains cannot form by
the mechanism of thermalfluctuation even at tempera-
tures(T\0.5T m)where atomic and grain boundary
mobility are significant and where grains do indeed
‘nucleate’and grow.That is,the obrved rate of
formation of new grains is found to be almost infinitely
larger,by some impossibly large factor such as1050
times,than the nucleation rate predicted by the thermal
fluctuation model,Eq.(3).
As a result of this disagreement,it is now universally
recognized[2,14]that,asfirst propod by Cahn[15]in
1949,the new grains do not‘nucleate’as totally new
grains by the atom by atom construction assumed in
the kinetic model.What happens is that the new
grains grow from small regions,recovered subgrains or
cells,that are already prent in the deformed mi-
crostructure.One of the many important conquences
of this idea is that the orientation of each new grain
aris from the same orientation prent in the de-
formed state[2].This results has been experimentally
confirmed many times,e for example[14].As dis-
cusd,for example,by Hatherly[16]and clearly
demonstrated by Haan[17],new orientations can
弯道超越develop in low stacking fault energy materials,that
form annealing twins,by growth twinning of a growing
new grain.In the cas,however,the original orienta-
tion in the deformed state can be tracked back from the
resultingfirst or,in some cas,in thin transmission
Fig.1.Optical micrograph of partially recrystallized coar grained aluminum compresd40%.The large grain has fragmented into two misoriented regions,A and B,misoriented by about40°.New grains, 13–17,and19have an A orientation and are growing into B;1–8, 11,and18have a B orientation and are growing into A.As in all examples of recrystallization,the inhomogeneous nature of the pro-cess is clear.Bellier and Doherty[7]courtesy of Acta Metallurgica.
R.D.Doherty et al./Materials Science and Engineering A238(1997)219–274 222
electron microscopy(TEM)foils[17],higher order twins.The critical difference between nucleation in recrystallization and in the other types of structural transformation such as solidification or the precipita-tion of cond solid pha,P,from a supersaturated matrix pha,M,is that,in the latter cas the required atomic arrangements characteristic of the new structures do not exist in the parent structure(liquid or M)and so must be built up atom by atom to the critical size.This will take place at the rate given by Eq.(3).In a deformed metal,for example heavily rolled alu-minum,copper,or brass,even though there is a very high defect density,the equilibrium fcc arrangement of atoms is still prent everywhere.The diffraction pat-tern remains that of the fcc metallic structure in all deformed fcc metals,though the diffraction peaks are broadened.Strain broadened X-ray diffraction lines is heavily cold rolled aluminum alloys with very high strain hardening promoted by various solutes has been reported recently[18].In abnormal grain growth,the equivalent situation occurs.The reaction involves a very small minority of the existing grains starting to grow at the expen of the vast majority of the other grains,which do not grow at any significant rate[2]. Here again the special grains do not have to form;they, like the embryos of the new grains in recrystallization, are prent in the starting structure.The question in both cas,recrystallization and abnormal grain growth,is how are t
he successful embryos or special grains lected?The prent review will address only the ca of the subgrain lection in successful regions in recrystallization,although,in the opinion of the author, the two topics have much in common,at least as regards the problem of‘nucleation’.
2.2.2.Grain boundary energy and mobility
Following the suggestion of Cahn[15]that nuclei grew from deformed subgrains,the question aro over 40years ago of why only a very small minority of subgrains made this transition.A simple calculation [13,14]indicates the magnitude of the problem.Moder-ately deformed,polycystalline aluminum develops a subgrain size of about1m m but after complete recrys-tallization the sample can evolve to a grain size of about100m m,e,for example,recrystallization of moderately deformed aluminum[9].An increa in diameter of about100indicates a volume increa, from the subgrain embryo to thefinal recrystallized grain,of about106.This estimate indicates that only about one subgrain in a million becomes a successful recrystallization nucleus in moderately deformed alu-minum.Cottrell[19]suggested that a critical reason for this small probability of success was the low mobility of most subgrain boundaries since most of the subgrains have only a small misorientation with their neighbors.Only subgrains with a high misorientation angle to the adjacent deformed material appear to have the neces-sary mobility to evolve into new recrystallized grains. This old ide
a is completely supported by extensive experimental evidence that‘nucleation’only takes place at regions in the microstructure with high local misori-entation.Evidence for this was reviewed in1978[14] and subquent studies strongly confirm this conclusion [2].Typical nucleation sites,all of which have high local misorientations,include:
1.pre-existing high angle grain boundaries;
2.misoriented‘transition’bands inside grains between
different parts of the grain that have undergone different lattice rotations due to different slip sys-tems being activated(Fig.1shows an example of a misoriented transition band between regions A and
B at which nucleation of new grains has occurred
[7]);
3.at highly misoriented deformation zones around
large particles;
怨天尤人是什么意思
4.within highly misoriented regions within shear
bands(the are bands of highly localized deforma-tion en in materials with high stored energies);and 5.at many places within very heavily deformed materi-
als(m\3–5),such as highly drawn wires. Humphreys and Hatherly[2]recently reviewed the surprisingly limited studies of the orientation depen-dence of grain boundary mobility in metals and con-cluded that there were indeed very large mobility differences of100–1000times,directly measured,be-tween low angle(2–5°)and high angle(\15°)grain boundaries.In high purity copper,the low angle boundaries showed activation energies clo to that of bulk diffusion(204kJ mol−1)while the high angle boundaries had the lower activation energies of boundary diffusion(125kJ mol−1)(e Fig.2).In low-angle,dislocation,boundaries,the rate determining step appears to be vacancy diffusion between disloca-tions,in near perfect crystal,while in high angle grain boundaries the rate determining step appears to be the atom transport by single atom jumps from the shrink-ing to the growing grains in the defect structure of a high angle grain boundary.Very recently Ferry and Humphreys[20]produced direct evidence for the in-crea of mobility in Al–0.05%Si subgrain boundaries of about14times as the misorientation incread from 2to5°and2500times from2°subgrain boundaries to high angle recrystallization boundaries on annealing at 300°C.Fig.1shows examples of this effect.New grains 3a
nd17are only growing into the deformed regions A and B,respectively,with which they are strongly mis-oriented and not into the regions with which they share a common orientation;17has a low angle misorienta-tions with A and3with B[7].
R.D.Doherty et al./Materials Science and Engineering A238(1997)219–274223
Fig.2.The much lower mobility(K%)and higher activation energy of low angle grain boundaries in high purity copper.From Humphreys and Hatherley[2]derived from the results of Viswanathan and Bauer.[20]for the retention of mobility differences of1000
even in the prence of0.05%Si in Al is reassuring as
it indicates the large difference in boundary mobilities,
previously found in very high purity copper,are quanti-
tatively as well as qualitatively similar to the effects
obrved in commercially pure materials.
There is,in addition,the much discusd question of
the relative mobility of a few special high angle grain
boundaries,some of which are clo to so called coinci-
dent site boundaries.An important example is 7(that
is with one atom in ven coincident in position in both
grains)the38°misoriented boundary with a common 111 rotation axis.The higher mobility of this boundary is central to one model of the‘oriented
growth’of fcc recrystallization texture[2].It is clear
that,at least for the tilt boundaries of 7,boundaries
parallel to the 111 axis do have higher mobility than
average high angle boundaries,at least in the prence
of solute,but this appears to be offt by the signifi-
cantly lower mobility of twist boundaries[22].A nice
demonstration of this provided by Ardakani and
Humphreys[23]who found the new grains with a near
40° 111 misorientation relationship with the matrix
in deformed single crystals of Al–0.05%Si grew ten
times faster in the tilt than in the twist direction.(The
tilt boundary is one with the common 111 rotation
axis lying parallel to the boundary plane;the twist
boundary has the axis normal to the plane).
2.2.
3.Transition from subgrain embryo to growing new
grain
It has been recognized for many years[24]that for a
subgrain to make this transition,posssion of a high
angle misorientation is a necessary but not sufficient
criterion.The subgrain,to become a successful new
grain,must have,in addition,an energy advantage,
usually a larger size in order to be able to grow rather
than shrink and vanish.The need for both a size
advantage and a high local misorientation appears to
be a reasonable explanation of the rarity of the process,
discusd above.In an earlier review[14],it was noted
that the deformation process itlf may,in some cas,
give a favored subgrain both advantages,high local
misorientation and a size advantage simultaneously.
The original model of Beck and Sperry[25]for‘strain
induced grain boundary motion’in which large sub-
grains on one side of a grain boundary can immediately
grow into the matching grain is a clear example of this
effect.Experimental evidence for this suggestion,from
orientation dependent stored energy,for example in
heavily cold rolled iron[26],is well established.How-
ever,in other cas,for example in moderately com-
presd aluminum[7,27],there is often no significant
orientation dependent subgrain size differences pro-
duced directly by deformation.In this latter ca,it was
found that the necessary size advantage of the sub-
Humphreys and Hatherly[2]have also reviewed other topics of importance to the consideration of grain boundary mobility.The topics include the important effect of solute drag and the possible role of grain boundary structure.For solute drag,the main effects em quite well established in that solute,especially that with low very limited solubility in the metal,strongly adsorbs at grain boundary and acts to inhibit boundary motion.The effect is most noticeable at very low solute levels and one of its most dramatic influences on recrys-tallization can be en in ultra high purity aluminum which,when deformed at low temperatures,can readily recrystallize at or below room temperature(0.3T m),e for example Haessner and Schmidt[21],while with a more typical purity even as little as0.01wt.%Fe in solution,the recrystallization temperature is very much higher at250°C(0.6T m)[7].The activation energy of recrystallization,involving the movement of high angle grain boundaries in the prence of solute,ris from that expected for grain boundary diffusion(as en for very high purity material)to that of bulk diffusion of either the solvent or the solute in metals
of even moder-ate impurity levels.The increa of activation energy in the prence of solute rais the question of by how much the higher mobility of high angle versus low angle boundaries discusd above might be affected by solute. The recent demonstration by Ferry and Humphreys
R.D.Doherty et al./Materials Science and Engineering A238(1997)219–274 224
grains adjacent to high misorientations had to be devel-oped by a cond,thermally activated process sub-quent to the deformation.
In the ca of moderately deformed aluminum it was found[27–29]that the cond process was that of subgrain coalescence.The coalescence of veral sub-grains on one side of a grain boundary was obrved to yield a subgrain large enough to grow rapidly into the adjacent grain[27,28].This coalescence process was highly localized;it only occurred at isolated points on the grain boundaries in aluminum,where misoriented ‘transition bands’,in the grain in which coalescence occurred,reached the grain boundary[27].A recovery model bad on the climb and cross slip of dislocation loops that could successfully account for the location and kinetics of the coalescence was subquently pro-pod and tested[29].In work currently in progress, Woldt(E.Woldt,private communication,1996).is directly reporting subgrain coalescence as occurring at the grain boundaries in heavily cold rolled high purity copper.Here the process gives a measured energy re-lea preceding recrystallization[30]and so in this ca it appears to be occurring more generally than in the more moderately deformed aluminum[27–29].The de-velopment of new grains by thermal recovery of the ‘deformation zone’around large particles in cold de-formed metals as‘particle stimulat
ed’nucleation[2]is a further example of a clear two-step nucleation process. Thefirst step occurs in deformation when misorienta-tion develops in the deformation zone but with small subgrains in the zone.The cond occurs on annealing when subgrain growth occurs within the deformation zone giving the misoriented region the necessary size advantage.
A more recent example of the difference between a microstructure in which the necessary conditions of size and misorientation were produced by deformation alone and where a post-deformation anneal was re-quired to produce this microstructure has been reported by Samajdar[31].He studied the deformed microstruc-ture in a commercial purity aluminum that had been plane strain extruded to two strains,84and96%reduc-tion,at320°C.It was found that a portion of each pre-existing near‘cube’grain,{100} 100 ,retained its near cube orientation during deformation and,on quenching from the extrusion press,the cube oriented material had significantly larger subgrain sizes(and smaller subgrain misorientations)than were en in the adjacent material,which was parated from the de-formed cube regions by a sharp,high angle grain boundary.In the ca of the highest reduction,the ‘deformed cube bands’were only one to two subgrains thick,so the large subgrains were in direct contact with the high angle grain boundary.As a result,recrystal-lization started immediately on annealing and,in fact,a small amount of recrystallization growth of the cube embryos
had begun in the material after quenching from the extrusion press.However,the material with the smaller extrusion reduction had deformed cube bands that were about eight to ten subgrains thick.The largest subgrains,which were en to have the smallest deviations from the exact cube orientation,were found in the center of the cube band with smaller,and more misoriented from exact cube,subgrains between them-lves and the high angle grain boundary.On anneal-ing,the latter material at a temperature in which recrystallization went to completion in250s,there was an incubation period of50s before any detectable recrystallization occurred.During this period there was a slow growth of the large subgrains in the center of the deformed cube band to the edge of the band before the much more rapid‘recrystallization’growth occurred of the cube regions into the misoriented material of the adjacent deformed band.
2.2.4.Formation(nucleation)and growth in recrystallization texture de6elopment
A subject that has been a major dispute for over50 years has been the origin of the strong recrystallization texture often found after heavy deformation.In many cas,after only moderate deformation,nearly random textures are produced.On annealing after very heavy reductions,a strong recrystallization texture is usually found,which may involve the partial retention of the deformation texture but quite often a very different but very strong new texture forms.A classic example is the for
mation of a very strong cube texture in some(but not all)heavily rolled fcc metals[31].The cube orienta-tion is afinite but very small part of the deformation texture.Two major alternative models exist for the formation of a strong new texture—usually described as‘oriented nucleation’or‘oriented growth’[1,31]. Oriented nucleation is the hypothesis that grains, with an orientation that dominates the fully recrystal-lized texture,nucleate more frequently than do grains of all other orientations.To describe this quantitatively, for example for the most discusd ca of the forma-tion of‘cube’texture after the recrystallization of heav-ily rolled fcc metals such as Cu or Al,the fraction of grains,by number,within a lected misorientation,say 10or15°from exact cube,h c,must be normalized by the fraction expected in a random grain structure,h r [32].The condition for a strong‘oriented nucleation’effect is that:
h=h c/h r 1(4) That is,the frequency of the formation or birth of the new cube grains is much higher than the expected random frequency,so many of the grains will have the special orientation.The oriented growth factor,i,is determined by the relative sizes d c/d r of the cube to the average grains[32].That is,there is a strong oriented growth effect if:
