You are on page 1of 24

Polycrystalline Strengthening

NIELS HANSEN
The strength of polycrystalline specimens can be related to interaction phenomena taking place during
elastic and plastic deformation. Such phenomena are reviewed in term of macroscopic and microscopic
strain accommodation processes required to maintain strain continuity across the grain boundaries. The
strength-grain size relationships can be described in a number of empirical equations relating the yield
stress and the flow stress in tension to various structural parameters. A number of such equations are
reviewed and their predictive capability is discussed. Structural information of importance for the
understanding of polycrystalline strengthening is obtained mainly from surface relief patterns and from
bulk structures observed by transmission electron microscopy of thin foils. The results obtained by
these methods are discussed and correlations are proposed. A number of features characterizing the
deformed structure are summarized and the behavior of a number of metals and alloys is reviewed with
emphasis on the structural changes in the interior of the grains and in the vicinity of the grain
boundaries. The models for strain accommodation during deformation are discussed on the basis of
the microstructures found, and this structural information is correlated with a number of strengthstructural equations. Finally, the flow stress of fcc and bcc polycrystalline specimens is related to the
occurrence of microstructures formed by macroscopic and microscopic strain accommodation processes during plastic straining, and it is concluded that macroscopic processes may be strength
determining at larger strains whereas microscopic effects may be of importance at small strains.

I.

INTRODUCTION

POLYCRYSTALS consist of grains which, due to a mutual


interaction, cannot deform freely during a deformation process. Higher stresses are therefore required to deform polycrystals than are needed for isolated single crystals, which
can deform by glide on a single slip system. This polycrystalline strengthening relates to macroscopic deformation
processes which normally are considered to be independent
of the grain size. Beyond this an additional factor affecting
.the strength of polycrystals is that of the size of the grains.
This grain-size dependent contribution to the strength is
demonstrated by the well-known experimental observation
that the low temperature strength of polycrystalline specimens normally increases when the grain size is decreased.
A major part of the theoretical and experimental work in
the last 50 years has concentrated on the macroscopic deformation of polycrystalline specimens with the main aim of
understanding the stress-strain relationship and the texture
development during cold deformation. However, in the last
20 years increasing attention has been paid to the effect of
grain size on the yield stress and the flow stress of polycrystalline specimens. An important part of this work has
been to interpret empirical strength-grain size relationships
on the basis of the microstructural observation of the surface
and the bulk behavior. Similarly, the changes in the dislocation structures in or at the grain boundaries during deformation have been studied in detail.
This paper deals primarily with the microstructural aspects of polycrystalline strengthening at low and medium
strains andat temperatures where dislocation climb is not
important. The strength-structure relationships under such
NIELS HANSEN is Head of the Metallurgy Department, Rise National
Laboratory, 4000 Roskilde, Denmark.
This paper is based on a presentation made at the symposium "50th
Anniversary of the Introduction of Dislocations" held at the fall meeting
of the TMS-AIME in Detroit, Michigan in October 1984 under the TMSAIME Mechanical Metallurgy and Physical Metallurgy Committees.
METALLURGICALTRANSACTIONS A

conditions have been described in a number of models based


on various assumptions regarding the deformed microstructures. Such models are reviewed and their predictive capability discussed below, as well as the degree to
which their basic assumptions accord with microstructural
observations.
In this presentation the first section is a review of polycrystalline deformation concentrating upon the aspects of
macroscopic and microscopic strain accommodation during deformation. These subjects are recapitulated in the
discussion and related to the observations of deformation
microstructures in fcc and bcc polycrystalline specimens.
II.

POLYCRYSTALLINE STRENGTHENING

The subject of deformation and strengthening of polycrystalline specimens has been reviewed extensively; see,
e.g., References 1 through 13. The strengthening processes
may be related in general to interaction phenomena taking
place during deformation. Such phenomena will be discussed in terms of the macroscopic and microscopic accommodation processes required to maintain strain continuity
across the grain boundaries.
A. Macroscopic Strain Accommodation
An early observation was that the deformation behavior of
single crystals and polycrystals is different. The first theoretical attempt to unify the behavior was made by S a c h s 14
who suggested that the individual grains in a polycrystal
deform like free single crystals (i.e., by single glide). However, voids are known not to open up at grain boundaries
during plastic deformation; thus, strain continuity must be
maintained. This basic requirement was fulfilled in the theory by Taylor ~5who assumed that all the grains in the polycrystal were undergoing the same homogeneous strain as the
bulk material. Such a homogeneous strain requires slip on
at least five independent slip systems according to von
VOLUME 16A, DECEMBER 1985--2167

Mises. m Taylor calculated an average orientation factor (M)


relating the stress in tension (o') to the stress in shear (~-) as
~r = M~'. For a random fcc polycrystal Taylor predicts a
value of 3.06 for M (in contrast the prediction o f M = 2.24
by Sachs14). The Taylor theory has been developed and
extended notably by Bishop and Hill J7 and Budianski and
Wu.X8 In its modified version this theory has been applied
extensively and in particular to work hardening and texture
calculations. 19-28
A recent modification to the Taylor theory has been proposed by Kocks and Canova ~ suggesting that strain continuity can be maintained with less than five slip systems for
grains of special shapes (e.g., flat or elongated after large
deformations) or in various regions (e.g., near grain boundaries and surfaces). Here, it is proposed that the strain continuity is maintained across individual boundaries and not
simultaneously for the whole grain. Figure 1 shows the expected behavior of a surface grain. The flow stress of the
polycrystal will decrease as the number of required slip
systems is reduced. This effect is, however, not considered
to be of great significance. 2~
The relationship between the behavior of single crystals
and of polycrystalline specimens, as described in the Taylor
model, has been discussed extensively, especially by
K o c k s . 3'm'2~ He has suggested that the basis of the comparison should be with single crystals which deform by multislip, as this deformation pattern may well be representative
of the deformation that occurs in polycrystals. Such a comparison between the stress-strain curve of a (liD-single
crystal and a polycrystalline specimen with a large grain size
is illustrated in Figure 2(a) for aluminum and in Figure 2(b)
for copper. The Taylor factor of 3.06 is used both for converting shear stress to tensile stress and shear strain to tensile
strain (multiplication and division by the Taylor factor,
respectively).
These two figures (2(a) and 2(b)) show that the shapes of
the single crystal curves are comparable to the polycrystal
curves with the latter on a lower level. This discrepancy may
be related to differences in the deformation substructure;
however, such differences may not be recognized easily.
The structure of strained O l D single crystals consists of
dislocation tangles and cells which, both in size and shape,
bear a close resemblance to features observed in polycrystalline specimens 29-33 which, however, have not been
analyzed in similar depth. A more detailed understanding of
the relative position of the curves in figures such as 2(a) and
2(b) may require a better knowledge of stage III hardening.3

<111> R~F(M=3.06)

60

Polycrystol
(d=O.2 mrn

z
l.lJ

z.o

I-'s
LIJ

:3

N 20
t
0

0.0

0.05

0.10
0.15
e-TRUE STRAIN

0.20

(a)
r

300

RT

<111>
{M=3.06)

~
/

I,I

200

~_~

u~

LU

t
O

/(d=0.2

Polycrystal
mm)

/ / < 1 1 1
/ /

>

(M ==2.2Z.1306).

100

I_

0.0

0.05

0.10
0.15
E -TRUE STRAIN

0.20

(b)
Fig. 2 - - Stress-strain curves at room temperature of polycrystalline specimens (grain size 0.2 mm) and theoretical polycrystal curves derived on
basis of the Taylor model from 0 1 D single crystals. 29'3~A stress-strain
curve calculated according to the modified Sachs model is included (see
text) a: 99.99 pct aluminum, b: 99.99 pct copper.

Alternatively, the observed difference may be related to


the presence of high permanent internal stresses which are
necessary in order to obtain the assumed multiple slip.
An alternative to the Taylor model has been proposed by
Leffers ~ where pile-up stresses are generated at positions
where primary slip planes meet the grain boundary and these
then may be relaxed by secondary slip in the neighboring
grain (Figure 3). In this model (termed the modified Sachs

(a)

(b)

Fig. 1 - - Estimated number of slip systems in a surface grain (a) and expected slip
line pattern (b). From Kocks and Canova. 2'
2168--VOLUME 16A, DECEMBER 1985

METALLURGICAL TRANSACTIONS A

Taylor

modified Sachs

<111>

= 0.05rnm
= 0.2 m m

./...-../

IOO 9 , S U / -

I-

//~/"

7"

'

0
0.0

Fig. 3--Deformation patterns according to (a) the Taylor model (from


Mecking 9) and (b) the modified Sachs model (from Leffers').

d =

//;;-;..-'-

11)

(b)

..1

/
j . ~ - d
/,,~//~,~i"~t,,~d

,,,=
l.U
=
ct-

,~-~f.Poly.crystals

(M = 3o6)2

tn
200 t/)
~,L

(a)

RT -

13_
~

"L"t

300

i
0.05

0.10
0.15
c - T R U E STRAIN

t
0.20

Fig. 4--Stress-strain curves for 99.99 pct copper as in Fig. 2(b), but

including polycrystalline specimens of different grain sizes.

model) the stress and the strain conversion is based, respectively, on the Sachs factor (2.24) and on the Taylor
factor (3.06). A better agreement between the converted
single crystal curve and the experimental curve for the polycrystalline specimen is achieved as shown in Figure 2(b).
In the Taylor model and its modifications the effect of
grain size on the flow stress is not considered. An allowance
for this effect was introduced by KochendiSrfer,34 who suggested that the interior of the grains in polycrystalline specimens may deform like isolated single crystals and that the
misfit where the grains meet might be accommodated elastically or plastically. It follows from this theory that the flow
stress will depend on the grain size, and such a relationship
has been suggested in a number of succeeding papers. 3'35'36
The effect of grain size on the stress-strain behavior of
polycrystalline copper at room temperature is illustrated in
Figure 4.
A model for polycrystalline strengthening formulated in
terms of dislocations was proposed by Ashby. 37'38 In this
model, the deformation of each grain is separated into a
uniform deformation (by slip on one system) and a local,
nonuniform deformation in the grain boundary region.
During the uniform deformation a dislocation density (p')
of statistically stored dislocations accumulates and causes
work hardening within the grains, assumed to be the same
as in an equivalent single crystal. In a grain undergoing a
uniform strain in an aggregate of grains overlaps or voids
occur (see Figure 5). These were corrected for by introducing geometrically necessary dislocations 39 of density pg.
The accumulation of p~ causes the grain size dependent part
of the stress-strain curve. The density of geometrically necessary dislocations was calculated as
E

pg -

4bd

1 4y
Zg b

[21

where y is the shear strain. For polycrystals L g was assumed


to be proportional to the grain size.
METALLURGICALTRANSACTIONSA

t======.-

r = r0 + a l G ~ / ~

[3]

where r0 is a friction stress, a] is a constant, and G is the


shear modulus. Eq. [3] gives a d -]a relationship for the
flow stress-grain size dependence in accord with the HallPetch relation to be discussed in Section III-B. Equation [3]
does not include any contribution from long-range stresses
but it was noted by Ashby that the long-range stress field
may control the flow stress during the first 1 to 2 pct strain
and then remain constant. In a discussion of the dislocation
concept Mecking 1~ proposed that dislocations stored in
the grain boundary region consisted, predominantly, of
geometrically redundant dislocations with no long-range
stresses and some geometrically necessary dislocations giving rise to long-range stresses. The dislocations near the
grain boundaries contributed to the forest hardening, and
their density was estimated to be of the same order as suggested in Eq. [ 1].
In the models discussed above, the main problem considered has been that of macroscopic compatibility during
plastic deformation. However, the deformation of the individual grains is also inhomogeneous, which gives rise to the
problem of microscopic strain accommodation, which is
now discussed.
B. Microscopic Strain Accommodation

[1]

where e is the tensile strain, b is the Burgers vector, and d


is the grain size. pe may also be expressed in a more general
way by introducing the geometric slip distance L g thus
Pg

The total density of dislocations, (p'), in a grain may then


be taken as the sum of pg and pS neglecting interactions
between the arrays. In the case where pg dominates, the
relationship between the dislocation density and the flow
stress may be expressed as

The deformation of grains is inhomogeneous, as it is


concentrated on slip planes with a finite separation. The
grain boundaries are effective barriers to slip transfer and
thus slip steps, as observed on free surfaces, do not develop
(see Figure 6). Stress will, therefore, build up where the slip
bands end at a grain boundary, and accommodating elastic
strain 34or multislip will be required locally. This problem of
microscopic incompatibility has been dealt with in a number
of studies on bicrystal and multicrystal deformation, e.g.,
References 40 through 46.
VOLUME 16A, DECEMBER 1985--2169

void
overlap~
-

(a)

(c)

(b)

(d)

Fig. 5 - - Deformation of a polycrystalline specimen (a) as separate grains cause formation of overlaps and
voids (b). These are corrected for by introducing geometrically necessary dislocations (c and d). From
Ashby. 37

Fig. 6--111ustration of microscopic incompatibility at a grain boundary


caused by slip in adjacent grains. From Kochend6rfer. 34
2170--VOLUME 16A, DECEMBER 1985

Hauser and Chalmers 42 suggested that the main effect of


the localized stresses, produced by dislocation pile-ups at
grain boundaries, was to initiate multislip in the vicinity of
the grain boundary over a distance of the order of the spacing between the slip bands, i.e., a few micrometers. They
further suggested that the existence of such a multislip layer
(zone of misfit) of constant thickness could explain the grain
size effect on the strength of polycrystals. This microscopic
layer has not been investigated in detail in the bicrystal
experiments mentioned; rather, these structural studies have
concentrated on surface slip patterns. Observations of such
patterns indicated that secondary slip and multislip take
place in the vicinity of the grain boundary and that such
accommodating slip may extend for relatively large distances into the grains.
However, these effects of the grain boundary could be
related to the existence of either macroscopic or microscopic
incompatibility, or both. Furthermore, it was observed that
elastic incompatibility in the grain boundary plane may give
rise to local stresses initiating secondary slip. 43'44By considering a stress due to elastic interaction and adding this to the
applied stress, it could be explained that yielding took place
at grain boundaries in an Fe-Si bicrystal of an orientation in
which the applied stress was insufficient on its own to cause
yielding. 43,~
The magnitude of the microscopic incompatibility effects
in the grain boundary region may be related to the mechanisms by which dislocations and grain boundaries interact.
A number of such mechanisms have been discussed by
Hirth. 6 Direct transfer of slip across the boundary can take
place if slip planes in adjacent grains have a common line of
METALLURGICAL TRANSACTIONS A

intersection in the grain boundary plane. However, for high


angle boundaries such a condition is difficult to fulfill since
glide planes normally rotate across the boundary. The grain
boundary will therefore act as a barrier causing dislocations
to pile up. Dislocations ahead of the pile-up may be forced
into the grain boundary and can interact with the grain
boundary forming grain boundary dislocations (extrinsic
dislocations). 47'48An example of such a process is shown in
Figure 7. Grain boundary dislocations may also occur where
dislocations cut through the grain boundary in order to conserve Burgers vector during the cutting process. 49-53 The
individual reactions of matrix dislocations and grain boundaries may result in different dislocation arrangements in or
at the grain boundaries. An example given in Figure 8 illustrates the variation in behavior of different boundaries.
A matrix dislocation trapped in a grain boundary will
resist the trapping of the next dislocation approaching the
grain boundary on the same slip plane. However, grain
boundary dislocations may move in the boundary by glide,
climb, or both, and annihilation processes may take place.
The repulsive force may, thereby, be diminished, i.e., the
grain boundary has acted as a sink for dislocations. The
annihilation of grain boundary dislocations may, however,
be considered negligible at low temperatures. 47
Besides being dislocation sinks, grain boundaries may
also act as dislocation sources. In the yield stress-grain size
model proposed by Li5s (see Section III-A) it was assumed
that matrix dislocations were produced at grain boundary
ledges (i.e., grain boundary dislocations). A number of
other mechanisms for the operation of dislocation sources in
the grain boundaries have been proposed, and experimental
observations have indicated the significance of such sources
during the onset of plastic deformation. 56-6~

The assumption by Chalmers and c o - w o r k e r s 41'42 that


a microscopic incompatibility zone with a width of a
few micrometers may form at the grain boundary, both in
bicrystals and polycrystals, has led Kocks 3 to propose a
composite model for polycrystalline strengthening. He
considered each grain as consisting of a grain boundary rim
(with a volume fraction Vg and a flow stress o-g) and a grain
interior (with a volume fraction V~ and a flow stress tri). By
equating the forces over the whole grain with the sum of the
forces on each component (F = Fg + Fi) the following
equation was derived:
t r = Vgo" 8 + V,.o'i

III.

[4]

STRESS-GRAIN SIZE RELATIONS

The effect of grain size on yield stress and flow stress of


polycrystalline materials is an important aspect of polycrystalline strengthening. Following the different models
discussed above, there appears to be a general trend of
dividing the stress of a polycrystal into two parts: one
which is independent of grain size and one which is grain
size dependent. The idea of considering the stress of polycrystals as consisting of two components was originally
due to Hall62 and Petch, 63 who formulated the well-known
equation for the lower yield point (Ty) of iron
'/'y =

1"0 "~

ky 9d -1/2

[5]

where ~'0 and kr are constants. This Hall-Petch relation and


its extension from the yield stress to the flow stress region
is discussed below together with other strengthening models
relating the flow stress and the grain size.
In the context of polycrystalline strengthening the effect
of texture is important. The effect on the yield stress of
texture may be approximately allowed for on the basis of an
average Taylor factor determined from calculations of the
orientation distribution function (ODF). 64'65 The effect on
the flow stress of texture is not well understood, but an
explanation has recently been attempted of the stress strain
curve of copper on the basis of the reorientation of the grains
during deformation and their strain hardening behavior. 66
A . Yield S t r e s s - - H a l I - P e t c h Relations

The Hall-Petch relationship which arose from experimental observations was derived theoretically by calculating
the stress concentration where a slip band meets the grain
boundary on the basis of the number (n) of dislocations in
a pile-up. 67 This number can be expressed by the equation
n = 7r" L ' r k l / G b

[6]

where L is the length of slip plane containing the dislocations, ~- is the applied stress, and kl is a numerical factor
near to unity. The stress at the grain boundary ahead of the
pile-up ~'i equals nz, thus
T i =

Fig. 7 - - G r a i n boundary dislocation in lightly deformed austenitic steel


(20 pct Cr, 25 pct Ni). Dissociation of a matrix dislocation A into two
extrinsic dislocations (a and b) is observed. Further extrinsic dislocations
are observed in this boundary, and their origin may be the same slip plane
as for A. From Howell et a l . 47
METALLURGICAL TRANSACTIONS A

"ri'" L 9 T 2 k l / G b

[7]

By assuming that the applied stress ~- is opposed by an


internal stress r0 and by taking L = d / 2 , Hall 62 and Petch 63
derived the equation:
z = "Co + ~ / 2 7 i G b

. d_l/2

[8]

~kl
VOLUME 16A, DECEMBER 1985--2171

experimental verification is, however, difficult partly due to


the large variation in behavior between the different grain
boundaries. Empirically, the yield stress of metals and alloys follows the Hall-Petch equation. A few examples are
given in Figure 9 which shows a wide variation in slope.
Factors such as solute pinning of dislocations may increase
the measured slope, whereas a process such as cross slip at
the head of the pile-up may lead to a relatively small value
of the slope. 12
The different aspects of the yielding behavior of polycrystalline specimens have been dealt with in a number of
reviews;Z.a,,2,68 thus the remainder of this paper concentrates
on the flow stress behavior of polycrystalline specimens.
B. F l o w S t r e s s - - H a l l - P e t c h Relation

Fig. 8--Transmission electron micrograph of a 99.999 pct copper polycrystal (grain size 0.01 mm) strained 0.1 in tension. One boundary, BC, has
formed a relatively regular array of grain boundary dislocations (arrowed)
which has led to a depletion in the matrix dislocation density. The near
coherent twin boundaries (AB, CD) show a significantly different behavior.
From Ralph and Hansen. 54

The relationship between flow stress and grain size was


dealt with by Armstrong et al. 73 They extended Eq. [5] to
include the flow stress region and suggested that the relationship between the flow stress at a constant strain ~-(e),
and the grain size could be expressed by the equation:
[12]

~'(e) = To(e) + k d -'/2

where z0(e) and k are constants at a particular strain. For the


tensile flow stress Eq. [12] is written:
[13]

o'(e) = M(~'o(e) + k d -la)

which is equivalent to Eq. [5] where

which corresponds to

Petch interpreted Eq. [5] as the sum of a friction stress r0


within the grains and a stress equal to kyd -1/2 to overcome
the resistance to slip propagation associated with the grain
boundaries. Slip propagation may take place when the stress
ahead of the pile-up is sufficient to operate a dislocation
source in the next grain. By analogy with the stress concentration ahead of a crack, CottrelP 9 estimated the stress at a
distance, r, ahead of a pile-up and assumed yield to occur
when this stress was equal to the stress rc required to operate
a source. The following equation was derived:
r

ro

+ 2re " r 1/2 9d -1/2

or(e) = O'o(e) + k ( e ) d -1~2

[14]

g o ( e ) = M~co(e)

[15]

with

k(e)

[161

Mk

In the formulation by Armstrong et al. ,73 o.0(e) corresponds


to the flow stress within the grain; thus g0(e) is equivalent
to the flow stress of a single crystal oriented for multislip. 3
The term k(e) d-'/2 relates to the resistance at the end of the
slip bands reaching the grain boundary.

[10]
300

which corresponds to Eq. [5] if ky equals 2% 9 r '/z.


Lack of experimental evidence of pile-ups in pure metals
led L P to suggest an alternative mechanism for yielding. In
this model, the grain boundary may act as a source of dislocations, and the capacity to emit dislocations may depend
on the character of the grain boundary, but not on the grain
size. 68 For a total length of dislocation (released to the interior) per unit area of grain boundary, m, the dislocation
density, p, at yielding can be calculated for a spherical grain
as p = 3 m / d . Assuming the usual relationship between r
and p the following equation is obtained: 55
z = ro + a , b G ~

d -1/2

RI"

ge

200
tn
tn
i,i
of

~-

70130

tn
UJ

8ross
100

[11]

Cu

Thus the slope in a Hall-Petch plot (Eq. [5]) is equal to


ot , b G k /'~m .

The different models for the yielding of polycrystalline


samples may be related to the various mechanisms for microscopic strain accommodation discussed in Section II-B.
A number of dislocation-grain boundary interactions have
been identified which may be part of a yielding process;
2172--VOLUME 16A, DECEMBER 1985

~
0

--'

'

'

'

AI
'

'

2
4
6
8
10
12
d-II2_(GRAIN SIZE)-u2 (ram-v2)

14

Fig. 9 - - Y i e l d stress-grain size relationship at room temperature for aluminum, ~ copper/~ 70:30 brass, 7~ and iron. n
METALLURGICAL TRANSACTIONS A

Equation [14] gives a good empirical description of the


behavior of many metals and alloys, e.g., aluminum and its
alloys, 69'~4 copper and its alloys, 7~
iron and ferrous
materials, 72,77-8~chromium, 8n niobium, 82zinc, 83beryllium, 84
and magnesium. 8~Examples are given in Figures 10 and 11.
However, a problem in the general acceptance of this empirical equation has been the frequent observation (especially
in copper) of stress-strain curves which cut through each
other (curve crossing). This phenomenon relates to specimens with a small grain size which are found to have a
larger flow stress at small strains than coarse-grained specimens, whereas, at larger strains, the strength ratio is reversed. Thus, a plot of the flow stress according to Eq. [14]
may result in negative values of k(e) at large strains and
grain sizes, s6
It has been suggested that curve crossing may be an effect
associated with texture, sT's8,89This problem has been considered for copper recently, 13 and it appears that the stressstrain behavior of coarse-grained specimens shows a very
large scatter in stress-strain behavior. This scatter may be
removed if a texture contribution to the flow stress is assumed. Figure 12 illustrates this effect by comparing the
flow stress of coarse-grained copper polycrystals having
different textures. 9~ The specimens were prepared by heat
treatment in the temperature range 400 to 750 ~ for 10 to
20 hours to cause grain growth. During this process the
volume ratio of (111 ) to (100) texture component increases,
as shown in Figure 12. The Taylor factors for the two components are 3.66 and 2.45, respectively; thus this texture
development should be followed by an increase in flow
stress, which is observed as shown in Figure 12. In conclusion, it is therefore proposed that Eq. [14] may be a valid
empirical equation for the flow stress-grain size relationship in polycrystalline specimens if texture effects are
accounted for.
The physical significance of the parameter k(e) is still
open to discussion. Some general trends have, however,
been established. An illustration is given in Figure 13 where
k(e) values at room temperature are plotted as a function of
strain in tension for aluminum, copper, and 70:30 brass. It
is seen that k(e) decreases with increasing stacking fault
energy, SFE, in accordance with results for a number of
copper alloys showing quite a wide variation in SFE. ~ This
behavior is related to the effect of SFE on the deformation
mode, ranging from easy cross slip in materials with a high
SFE to a planar slip mode when the SFE is low. Pure bcc

300

true
strain

RT
9

e~ _ .0~,04p ~ e e ~ ~

0.20 . . , ~ i . , ~o

200

_.,__.-

,
~-

O.lO

"

6 -il o ~ $ - - . ' - . "


~

-q's'

"

II

9~ - e ' o - q d b ~ o e - e -

oc 1 0 0 - 0 . 0 5
I--

:001
10.002
00

_~S
-.a

&

10

d-1/2_(GRAIN SIZE) -lt2 (ram -1/2)

Fig. l l - - F l o w stress-grain size relationship at room temperature for


99.999 pct copper. From Hansen and Ralph. 75

metals should show a comparable grain size dependency to


that exhibited by fcc metals and alloys due to an abundance
of slip systems. An example of this is given by pure iron
where k(e) values of the same order of magnitude as in
70:30 brass have been measured in the strain range 0.1 to
0 . 2 . 72,78 A more detailed discussion of the k( e )-parameter
may be found in a recent review by Armstrong.12
1

o
13_

300
True strain

to 250
co
I.tJ
rtI-

0.20

to 200
o
_J

0.10 -o

u_ 150
i

A
O
O

12

10

8
v

70
~ 60

true
strain

~ 50

020

It')

~o
i--

~n 30

_o

RT

0.10 -

o_ 9

~ o ~ - - - - - - - - I

0,05 -a.L--II~

--

o- 9

,m~o-----"---

e''~e'-~-''~--

LI..I

2O

,~ ~o

0.002_ef_@~.-~e--e~- e ~ "
2
3
4
d -lt2- (GRAIN SIZE }-1/2 (rnm-ll2)

Fig. 1 0 - - F l o w stress-grain size relationship at room temperature for


99.99 pct aluminum. From ltansen. 69

METALLURGICAL TRANSACTIONS A

4O0
6OO
8OO
ANNEALING TEMPERATURE (~

oo,

0
0

Fig. 1 2 - V o l u m e ratio of texture components (111) and (tOO) in coarse


grained 99.999 pct copper polycrystals calculated by different methods.
The grain size after heat treatment was 0.2 to 0.5 ram, and to obtain reliable
texture measurements neutron diffraction of large specimens was applied.
The upper curves show that the change in flow stress follows the variation
in texture."
VOLUME 16A, DECEMBER 1985--2173

15

o'(e) = t r o + aGX/ee
70/30
10

[20]

If it is assumed that the grain boundary strengthening dominates (compare Eq. [3]), i.e., Cz/d >> C~/L ~, Eq. [20] can
be written

E
z
I

tr(e) = O'o + aGX/-ee X/-'~zb

Cu

I
0.05

I
i
0.10
0.15
- TRUE STRAIN

I
0.21

C. Flow S t r e s s - - F o r e s t Hardening Models

In the early experiments on silver91 and copper92 it was


established that the flow stress of polycrystalline specimens
could be expressed by the equation o- oc p~/2. The grain size
of the specimens examined was quite small, of the order of
10/zm; thus an expected grain size contribution to the flow
stress must be contained in the dislocation term. The first
attempt to relate the dislocation density, at a given strain, to
the grain size was due to Meakin and Petch, 93 who assumed
that the average slip distance L' might be comparable to the
grain diameter. Based on this assumption and the equation
T = PbLs it follows that the dislocation density for a constant strain would increase with decreasing grain size. By
inserting the dislocation density in the hardening equation
the following equation was derived:
[17]

This differs from the usual Hall-Petch equation (Eq. [14])


since here or0 is independent of strain. Conrad 94'95 has also
suggested that the Hall-Petch relation reflected the influence
of grain size on the dislocation density, and later Conrad
et al. s2 observed that the flow stress of niobium could be
expressed by an equation like [17] based on the finding
that the dislocation density is proportional to e / d . In addition, the dislocation density has been observed to increase as
the grain size decreases in vanadium, 96 iron, 9v'98'99and titanium, 100and it has been concluded in all these investigations
that the effect of the grain size on the flow stress can entirely
be attributed to the grain size dependency of the dislocation
density at constant strain.
The dislocation model proposed by Ashby 37 allows a
separation of strength contributions into those independent
of and dependent on the grain size. The introduction of
statistically stored dislocations p' and geometrically necessary dislocations p8 leads to a flow stress relation based on
the following expressions for p~ and pg:
o' =

C~e
~
bL s

C2e
pg = ~
bd

[18]

Eqs. [21] and [22] show that the flow stress at a constant
strain may be a function of d -l/z or d -1 depending on the
magnitude of the grain size contribution.
Equation [20] has been applied to 99.99 pct aluminum 69
and 99.999 pct copper 7~by estimating the constants in the
equation. An example for aluminum is shown in Figure 14
showing an acceptable correspondence between the calculated curves and the experimental results. A similar correspondence was obtained for copper supporting the validity
of Eq. [20] in the intermediate strain range. The values for
C~/L' and C2 were estimated at 50 mm -J and 2 for aluminum and at 400 mm -1 and 3 for copper, respectively. The
difference in C~/L s values between aluminum and copper
indicates that the apparent slip length is much larger in
aluminum which may relate to the much easier cross slip in
this material. The values for Cz for aluminum and copper
coincide closely and are of the same order of magnitude as
the theoretical estimate. 37
D. Flow S t r e s s - - Composite Models

A composite model (see Eq. [4]) can be expressed by the


equation
cr = cri + (trg - o'i)(4 dx

[19]

xZ)/d

[23]

where x is the width of the grain boundary region (see Figure 15). For d ~> x this equation can be written 3

o r = o'i 1 + - - ~

- 1

)}

[24]

It follows from Eq. [24] that the flow stress may vary with
d -1. However, a d -1/~ relationship may also be obtained if
x and crg are allowed to vary with d. The dislocation density

n 60

f
'
'
'
~ 50 true s
t
r
~
t~
w L0 . . . . . . . ~ Z ' 7 ~
f- 3(3

' RT

w 20
rr 10

where Cl and C2 are constants. The flow stress thus depends


on the total dislocation density through the expression 13
2174-- VOLUME 16A, DECEMBER 1985

[22]

\Cjb] d

Fig. 13 --k(e)-values as defined in Eq. [14] plotted as a function of strain


in tension at room temperature for 99.99 pet aluminum, 69 99.999 pct copper, ~~ and 70:30 brass. 7~

tr(e) = Go + o~Gbl/2el/2 d -1/2

[21]

d -1/2

If the grain boundary contribution is small (e.g., in coarsegrained materials) Eq. [20] can be written

AI
0
0.0

Brass

0I
0

~
l
l
l
I
1
2
3
4
5
d-u2_(GRAIN SIZE)-u2 (ram-u2)

Fig. 1 4 - - F l o w stress-grain size relationship at room temperature for


99.99 pct aluminum, The continuous curves are plotted accordingly to
Eq. [20]; the hatched areas are scatter bands from Fig. 10, From Hansen. 69

METALLURGICAL TRANSACTIONS A

Fig. 1 5 - - Cross-section of grain with a hard rim of thickness, x, and a soft


inner region. From Hirth. 6

in a grain boundary rim has been estimated by Mecking l~ to


be of the order of e / x b ; thus crt o~ N/-e-e/x. In considering
the width of the rim two cases were discussed,9'10 either a
width proportional to the grain size, x cx d, or a width independent of the grain size, x = const. For the two cases, the
width, x, was estimated at 0.1 d and at 1 ~m, respectively,
and reasonable correspondence with experimental values 7~
was obtained for the flow stress-grain size relationship of
polycrystalline copper.
The composite model as expressed in Eqs. [23] and
[24] has been further developed by Thompson and coworkers 7,s6't~ on the basis of the concept of statistically
stored and geometrically necessary dislocations, the latter
being assumed to be concentrated in the grain boundary
region. The dislocation density in this region pt is assumed
to be inversely proportional to the grain size but independent of the strain, in contrast to the suggestions of Eq. [19].
The dislocation density in the interior of the grains was
estimated to be inversely proportional to L', the average
slip length of the statistically stored dislocations. Finally,
the area of the grain boundary region was assumed to be
proportional to LS/d, and the following flow stress equation
was proposed: ~~

latter study the slip band spacings were measured in the


grain boundary region (S,) and in the grain interior (S~). The
strain range examined was 0.001 to 0.022, and the average
grain size varied from 0.049 to 0.2 mm. Generally slip
appeared as a single set of parallel bands (referred to as
primary slip) with a smaller spacing in the vicinity of the
grain boundary than in the grain interior. The grain boundary region was therefore not observed as a multislip zone
although it was noted that secondary slip also occurred in
this region. By relating the average stress to the average slip
band spacing, values for tr~ and ort (Eq. [23]) were calculated and, by measuring the boundary width, a composite
flow stress could be obtained. This stress coincided closely
with the experimental values. The volume fraction of the
grain boundary region increased with the strain to cover the
whole grain at a strain of about 0.02 (see Figure 16).
The slip band spacing S~ and Sg decreased as the strain
increased as illustrated in Figure 16 for two different grain
sizes. It is interesting to note, however, that S~ and St converge at a higher level for the specimens with the coarser
grain size. This trend indicates that the fine-grained specimens may be stronger than the coarse-grained specimens at
strains where a boundary rim can no longer be identified.
Such an effect is supported by the experimental data for the
applied stress as a function of grain size.
IV.

DEFORMATION STRUCTURES

According to the deformation models discussed in


Section II, compatibility during plastic deformation can be
maintained in various ways. The strain accommodation
may, therefore, give rise to a wide variation in the deformation patterns which have been studied by macroscopic
and microscopic techniques. Results from such studies are
summarized in the following five subsections. Section A
concerns macroscopic surface observations while Sections
B and C describe microscopical observations of surface
relief structures and bulk structures and discusses how they
may be correlated. In the last two sections a number of
specific observations are reviewed, emphasizing the characteristics of the deformation substructure in the interior of the
grains (Section D) and associated with the grain boundary
region (Section E).
120

T"

-~ ~oo

o-=o'o + ( 1 - L--~d)K----L+-L~k,d -~'2


L,

[25]
z

80

u',

60

where L ~ is a function of strain and grain size and Kl is a


constant. At small strains L ' approaches d and the equation
becomes equivalent to Eq. [5]. At increasing strains, LS/d
decreases, and thus the effect of the grain size on the flow
stress decreases.
Thompson and co-workers examined Eq. [25] for aluminum, copper, and brass.~~ The slip length was allowed to
vary to obtain a good fit between the experimental flow
stress values and the model predictions. The resultant slip
lengths were obtained as a function of strain and grain size,
and reasonable correspondence with assumed values was
observed. The validity of the composite model was tested
for fl-brass bicrystals ~~ and for /3-titanium Ti-9.8 pct
wt pct Mn polycrystals by Margolin and Stanescu. 103In the
METALLURGICAL TRANSACTIONS A

L
i
S~

O
Z

.j

0.049

0.2 mm

S g ~

20

Sg
0

0.0

i
0.01

i
0.02

t
0,03

E- TRUE STRAIN

Fig. 1 6 - - S l i p band spacings in 13-titanium Ti-9.8 wt pct Mn polycrystalline specimens (grain sizes 0.049 and 0.2 ram, respectively) tested
in compression at room temperature. S~ and Sg are~ respectively, the slip
band spacing in the interior of the grain and in the grain boundary region.
From Margolin and Stanescu.~~
VOLUME 16A, DECEMBER 1985--2175

A. Macroscopic Observations
A number of surface observations based on hardness
and elongation measurement have shown that the deformation of polycrystalline specimens is heterogeneous. 104-108
For example, it was observed in the study by Boas and
Hargreaves ~~ of tensile strained coarse-grained aluminum
(grain size 10 to 30 mm) that the local elongation varies
significantly within the grains and between the grains. Corresponding results have been reported recently ~~ based on
photo-grid measurements of tensile strained copper and iron
samples having grain sizes of the order of 0.1 to 0.2 mm.
This technique has also been applied by Urie and Wain ~~ in
a study of the deformation behavior in the grain boundary
region of coarse-grained aluminum (grain size 5 to 15 mm).
They observed a marked restriction in elongation near many
grain boundaries and that the affected region may extend for
up to Several millimeters away from the boundary.
Such a restriction in elongation causes grooving at the
boundary which is clearly illustrated in Figure 17 which
shows a scanning electron micrograph of strained aluminum
having a grain size of 0.3 ram. The formation of grooves
has been discussed, n~ and it was proposed that slip from a
source operating near the grain boundary could occur more
freely away from the boundary than toward the boundary,
thus creating a height difference between the grain boundary
zone and the areas farther away. However, Figure 17 shows
that the nonplanar nature of the surface is not only a characteristic of the grain boundary region, and height differences
of the order of 10/zm have been measured over a distance
of 100 ~m in typical grains. 109

~<

(bl
Ccl
DISTANCE
Fig. 18--Schematic illustration of the elongation behavior in the grain
boundary region in coarse-grained 99.99 pct aluminum strained 0.1 in
tension. From Urie and Wain.,0~
(ca}

Different types of interactions near the grain boundaries


were observed by Urie and Wain ~~ (see Figure 18). These
observations were correlated with the slip line pattern; e.g.,
it was observed for the grain boundary illustrated in Figure 18(b) that single slip occurred across the grain, whereas
a narrow zone of double slip was observed on each side of
the grain boundary. By contrast, the slip line pattern related
to Figure 18(c) extended right to the boundary, and one set
of slip lines extended into the neighboring grain.
In discussions of macroscopic surface observations it has
been argued that such results are not representative of the
bulk behavior since surface grains are under less constraint
than the grains in the interior of the samples. However, it
will emerge from the discussion which follows that the
macroscopic surface observations, although somewhat
qualitative, may complement the more precise information
obtained by applying various microscopical techniques.

Fig. 17--Scanning electron micrograph of a 99.99 pct aluminum polycrystal (grain size 0.3 mm) strained 0.16
by cold rolling. This figure demonstrates an extensive surface rumpling and a formation of grooves at the grain
boundaries. Well-defined slip bands can be seen in all grains although in some cases (as in grain A) they are
observed only near the grain boundary. Note that 3 slip line systems are present in grain B. From Barlow et a l . lo9

2176--VOLUME 16A, DECEMBER 1985

METALLURGICAL TRANSACTIONS A

B. Surface Relief Structures and Bulk Structures


Deformation structures are generally described on the
basis of observations of surface relief structures and bulk
observations by transmission electron microscopy of thin
foils. A number of structural features have been identified
and are summarized below.
The investigations of surface relief structures have developed from light microscopy m to higher resolution studies by
transmission electron microscopy, m-115 In general, a number of slip patterns have been identified in polycrystalline
specimens strained below 0.2 to 0.3.114.115Such patterns are
an elementary structure consisting of fine lines with a spacing of the order of 20 to 40 nm and a coarser structure of slip
lines which in medium and high SFE material may take the
form of slip bands of variable spacing. Other features observed are deformation twins (in low SFE materials) and
bands separating different regions of a grain in which different slip systems are operating. Such bands, termed transition
bands, may be narrow and carry a large misorientation
across them. ~16,1IV,118
The investigation of bulk structures of deformed metals
by transmission electron microscopy of thin foils dates back
more than 30 y e a r s . 119 A very large number of structural
features have been identified since then; a2~ however, only
the most characteristic ones will be summarized in this
paper. In high SFE fcc materials a cell (subgrain) structure
develops at increasing strain. This structure consists of relatively dislocation-free regions separated by dislocation
walls, normally with a small misorientation across (less than
0.5 to 1 deg). This cell structure is also characteristic of
deformed bcc metals.
In fcc materials with a low SFE, characteristic features of
the deformation structure are a relatively homogeneous distribution of dislocations and the occurrence of deformation
twins, m Another structural feature observed both in fcc
and bcc materials is that of microbands which may extend
for considerable distances across the grains, j22-~25 Such
bands, consisting of small cells, which are narrow (of the
order of 0.2 micrometers in width which are oriented parallel to slip planes), and which are only slightly misorientated
(of the order of a few degrees) in relation to the matrix.
Figure 19 shows such microbands in cold-rolled aluminum.
Transition bands are also observed in the bulk structure,
often as narrow bands carrying a large misorientation in
agreement with the surface observations.

C. Correlation between Surface Relief Structure and


Bulk Structure
The deformation structures observed on the surface and in
the bulk may, to a certain extent, be considered as complementary. However, some direct correlations have been
sought 1~ by investigating foils from the sample surface, in
order to observe, simultaneously, the surface relief structure
and the dislocation structure. (Such foils are prepared by
masking the side corresponding to the sample surface and
electropolishing to perforation from behind the surface.) It
has been observed in such foils that slip band traces were
visible in all the grains whereas microbands were not such
a common feature. However, when the microbands are
present there may be a good correspondence between a
number of these bands and the surface slip traces (see
Figure 20). However, in other areas, there was very little
METALLURGICAL TRANSACTIONS A

Fig. 19--Transmission electron micrograph of two grains in a 99.99 pct


aluminum polycrystal (grain size 0.4 mm) strained 0.16 by cold rolling.
The grains contain microbands in the form of narrow parallel bands containing small elongated subgrains. In two cases a slip band in grain 1 and
in grain 2 meets at the same place at the grain boundary (marked by curved
arrows). A few micrometers from the grain boundary the bands may stop.
(Two examples are marked by white arrows.) From Bay and Hansen. 12~

correspondence between the dislocation structure and the


surface structure (Figure 21).
. . . . . These types of observations form the basis of a hypothesis 1~ that the bulk structures formed during deformation of
high SFE fcc metals may contain a high concentration of
dislocation walls and bands which correspond to the slip
bands observed on the surface. However, the surface relief
structure forms a permanent record of the deformation,
whereas the dislocation structure may change continuously,
both during deformation and during unloading. The cell
structure observed as a characteristic structural element
in high SFE fcc metal may then be formed as a result of
processes involving relaxation of dislocations in microbands
and in dislocation walls, e.g., by cross slip. Structural evidence for such a dislocation relaxation has been obtained in
99.99 pct aluminum. 109
VOLUME 16A, DECEMBER 1985--2177

In correlating surface relief structures and bulk structures


on the basis of observations of the behavior of surface
grains, it is important to know if such grains are representative of the grains in the interior of the specimen. This problem has been investigated, ~~
and it appears that both
the deformation pattern and the density and arrangement of
dislocations is not significantly affected by the position of
the thin foil in relation to the surface plane of the specimen.
D. Deformation Structures within the Grains

Fig. 2 0 - - H V E M picture of a foil prepared from the sample surface of a


99.99 pet aluminum polycrystal. The surface relief contrast is marked by
dotted lines, and it is seen that in a number of cases there is a close
correspondence between the surface slip traces and the dislocation bands.
From Barlow e t a l . ~~

The structural descriptions to be given both in this and in


the following section concentrate on results which are considered relevant to the phenomenon of polycrystalline
strengthening, and most of the examples discussed concern
the behavior of typical fcc and bcc materials (aluminum,
nickel, copper, 70:30 brass, and iron). In both sections the
surface observations and bulk observations are dealt with
separately.
In the analysis of surface relief structures emphasis has
been given to the number of slip line systems in the individual grains. 1~176
This number is typically from one to
two, possibly three, increasing with the strain. Each slip line
system may correspond to two independent slip systems on
the same slip plane, 128and, generally, it has been suggested
that two slip line systems may imply slip on three slip
systems. 2~ In addition to the slip on the main systems, socalled accommodating (or alien) slip may take place. This
type of slip may start at, or near, the grain boundaries and
extend far into the grains, as illustrated in Figures 22
through 24, showing surface relief structures of deformed
copper, 70: 30 brass, and iron.
The effect of grain size on the slip line pattern is, in
general, not very large. In large grain size specimens there
is a tendency for grains to separate into regions deforming

Fig. 21 - - S a m e specimen as in Fig. 20. HVEM picture showing a set of parallel surface steps (arrowed) and an underlying
structure of ordinary subgrains. There is very little correlation between the surface steps and the subgrain walls. From
Barlow et al. ~~
2178--VOLUME 16A, DECEMBER 1985

METALLURGICAL TRANSACTIONS A

Fig. 22--Surface replica of a 99.98 pct copper polycrystal (grain size 0.03 mm). Accommodating slip lines
extend from a grain boundary (arrowed). The specimen has been strained 0.07 by cold rolling. From Leffers. 23

Fig. 23--Surface replica of 70:30 brass polycrystal (grain size 0.02 mm). Accommodating slip lines extend
from a grain boundary (arrowed). The specimen has been strained 0.15 by cold roiling. From Leffers. 2s

METALLURGICALTRANSACTIONS A

VOLUME 16A, DECEMBER 1985--2179

Fig. 2 4 - - L i g h t micrograph of a 99.9 pct iron polycrystal (grain size 0.26 mm). Accommodating slip lines extend from a
grain boundary (arrowed). The specimen has been strained 0.15 in tension. From Jago and Hansen. 72

on different slip planes as observed in aluminum, ~~ copper, ~28'132and nickel. 129 The formation of transition bands
also indicates a subdivision of grains into regions of different slip. Data for slip line lengths as a function of strain and
grain size are not available; however, it has been observed
in aluminum that the slip band spacing for a given strain
decreases with decreasing grain size at small strains (0.04),
whereas at larger strains (0.16) the spacing is not significantly affected by the grain size. ~09
Deformation structures observed in thin foils are generally described by the density and arrangement of dislocations, local lattice misorientations, and the occurrence
of features such as deformation (and annealing) twins.
The most common parameters reported are the average
dislocation density and the cell size, which are interrelated. 75'~33 In a number of metals such as aluminum, 69
copper, 13 niobium, 9~ vanadium, 96 iron, 97'98'99and titanium 1~176
it has been observed, for a constant strain, that the average dislocation density increases when the grain size is
decreased. Figure 25 shows this effect for 99.99 pct
aluminum. 69

E. Deformation Structures in the Vicinity of Grain


Boundaries
The presence of a grain boundary may affect the deformation structure in a number of ways. This is illustrated
schematically in Figure 26 for 99.99 pct aluminum, showing correlations between typical surface relief structures and
dislocation structures observed in thin foils. 109Many of the
features shown in Figure 26 have been reported elsewhere
for different metals and alloys, e.g., vanishiog and bending
of slip lines near the boundary, 103.114,134anomalous slip,4~ slip
lines crossing the boundary,I~ wavy slip due to cross slip, H5
and different sizes and shapes of cells near a boundary 75'125
2180--VOLUME 16A, DECEMBER 1985

as well as a high dislocation density in this region. 135Some


of these features are illustrated in Figures 27 through 30.
Furthermore, the accommodating slip also characterizes the
grain boundary region (see Figures 22 through 24) as well
as the occurrence of distorted deformation twins. 23
The complex deformation pattern frequently observed in
the grain boundary region is illustrated in Figure 31 for three
adjacent grains in cold rolled aluminum. Grain A shows
small subgrains in a band of about two micrometers width
along the boundary between A and B (but not along the
boundary up to grain C), whereas grain B shows a relatively
large increase in subgrain size when the boundary between
l
Grain

i
size. mm

0.046

'E
el

b
X

fz
~3
z
o

3
2

0.130

o
o

,..1

0.49

1
I
Q.

0 -0.0

J
0.05

_1
J
0.10
0.15
E - T R U E STRAIN

Fig. 25--Dislocation density as a function of grain size and strain for


99.99 pct aluminum polycrystals. From Hansen. 69

METALLURGICAL TRANSACTIONS A

SURFACE
1. BOUNDARY HAS NO EFFECT
ON STRUCTURE

THIN FOIL

_==I---

a. With bands

b. Without bands
2. BANDS ONLYAT BOUNDARY
a. Bands penetrating grain

b. Bands parallel to boundary

3. BANDS IN GRAIN INTERIORBUT ~


NO BANDS AT BOUNDARY

lll

4. ANOMALOUSSLIP

S. BENDINGOF B A N D S

(may also become parallel to

boundary)

6. STEPPINGOF BOUNDARY

7 ~CROSS-SLIP
B
TO AGIVE WAVY
N

~D

Fig. 2 6 - - S c h e m a t i c illustrations of different surface relief structures and dislocation structures observed in the vicinity of grain boundaries. From Barlow e t a l . ~o9

B and C is approached. This zone of large subgrains in grain


B with a width of about eight micrometers may be caused by
enhanced recovery in the grain boundary zone which can
also be observed in copper and iron. Figure 31 also illustrates that subgrains on each side of a boundary may match
in size, as the subgrain size is relatively large in the interior
of grains A and C and relatively small in the interior of
grain B. The origin of some of the structures observed in the
vicinity of grain boundaries has been tentatively analyzed.
However, a more detailed mechanism for the formation of
the individual structural features has yet to be proposed.
V.

DISCUSSION

The models for polycrystalline deformation are normally


validated by their capability of predicting mechanical properties and texture. It is, however, of equal importance to
examine the structural evidence for the different models.
Most of the information about deformation structures has
been obtained by examinations of surface relief patterns and
of dislocation structures in the bulk. These two types of
observations may be considered complementary and are, as
such, of great significance. However, an improved understanding of deformation structures must rely quite heavily
on studies that explore the extent to which surface and bulk
METALLURGICAL TRANSACTIONS A

observations can be reconciled. Present knowledge about


deformation structures gives, however, important qualitative information both about strain accommodation processes taking place during plastic deformation and about
specific structural features, which are the result of the deformation. This is discussed below together with the effect of
various structural elements on the flow stress of polycrystalline specimens.

A. Polycrystalline Deformation
The models dealing with macroscopically compatible
plastic deformation fall into two categories assuming, respectively, a homogeneous (Taylor] 5 Kocks 3) and an inhomogeneous deformation pattern (Kochend6rfer34).
Macroscopically compatible deformation may take place
locally by slip on several systems, typically from three to
five (see Figure 1). The number of slip line systems observed is normally from one to three, but the number of slip
systems may be somewhat higher. Furthermore, accommodating slip extending inward from the grain boundaries for
relatively large distance (see Figure 23) has frequently been
observed, and such slip may add to the operating slip systems in maintaining the macroscopic compatibility. (However, accommodating slip has, in a number of cases, been
VOLUME 16A, DECEMBER 1985--2181

Fig. 27--Surface replica of a 99.99 pct aluminum polycrystal (grain size


0.4 mm) strained 0.16 by cold rolling (the line (arrowed) marks the intersection of the rolling plane and the replica). Slip bands in the two neighboring grains meet at the grain boundary (marked by a dotted line). In a
number of cases the bands penetrate only a small distance into the neighboring grain; some of these bands are marked (c). From Hansen and Bay. 136

observed to be present only in a relatively narrow grain


boundary region; see, e.g., Figure 24.) In such cases this
type of slip is more readily classified as a type of microscopic strain accommodation.
The relaxation of the number of slip systems required to
ensure compatible deformation locally is based on an assumption that grains may fragment into regions deforming
by slip on different systems (see Figure 1). Such a fragmentation has been observed in large grains and is illustrated
in Figure 32 showing channeling contrast micrographs of
deformed aluminum. Grains may also be subdivided by
transition bands separating regions of nearly uniform
deformation. It follows from these observations among
others that macroscopically compatible deformation may
take place in fcc and bcc materials in a heterogeneous
manner. This finding is in disagreement with the homogeneous deformation assumed in the Taylor theory. However,

2182--VOLUME 16A, DECEMBER 1985

the discrepancies between observed and postulated structural development have, to some extent, been resolved
by the recent modification of the Taylor model by Kocks
and Canova. zl
A number of structural observations indicate that local
inhomogeneities are present to such a degree that they also
may be of some significance for a macroscopically compatible deformation. Such inhomogeneities are most frequently observed extending from the grain boundaries.
Typical features are, for example, a disturbance of the slip
band symmetry (Figure 27), presence of multislip regions
(Figure 28), and formation of regions containing subgrains
of reduced size (Figure 29). Such inhomogeneities are
observed in many grains, although their magnitude and
frequency do not appear to be large enough to support
the general hypothesis that inhomogeneous deformation
of the grain boundary region can account for a large part
of the macroscopic strain accommodation as assumed in the
Kochend6rfer model. It appears, however, that compatible
deformation in many grains requires a combination of macroscopically and microscopically strain accommodation.
Some structural features seem to suggest that the macroscopically compatible deformation can be somewhat dependent on the grain size. For instance, it seems that a
phenomenon such as grain fragmentation may be more characteristic for large grains, whereas the deformation structure
of small grains appears more homogeneous across the individual grains. Strain accommodation may, therefore, locally
require the operation of fewer slip systems when the grain
size is large, and consequently the flow stress will be reduced. Supporting evidence for such behavior is the observation that plastic flow in a polycrystalline specimen may
start in the large grains I29'131 and that such grains appear to
accommodate more plastic strain than the smaller grains in
the beginning of deformation. 138
An assumption that small grains may be somewhat harder
than large ones may explain, in macroscopic terms, why the
flow stress increases when the grain size (or the ratio between large and small grains) decreases. A difference in the
macroscopic deformation behavior of specimens having different grain sizes may result in an effect of grain size on the
texture development during deformation, and preliminary
results indicate that such a phenomenon can be observed
(see Figure 33).
It appears from the microstructural features observed in
the grain boundary region that a microscopic incompatibility
zone extending for a few micrometers (see Section II-B)
may be present at a number of grain boundaries (see
Figures 28 and 29). However, the microstructures at the
grain boundaries in deformed specimens vary significantly
from boundary to boundary and even along the same boundary. The width of the boundary zone (see Section III-D) is
also undefined, and it may show large variations between
the extremes of dislocation arrays at (and within) grain
boundaries to multislip regions extending quite far into the
grains. It may therefore be concluded on the basis on the
microstructural observations that the grain boundary region
cannot in general be Classified as a well-defined structural
element. However, the structural differences observed in
many grains between the matrix and the grain boundary
region make it relevant to discuss the individual contribution
of the two regions to the flow stress.

METALLURGICAL TRANSACTIONS A

Fig. 28--Transmission electron micrograph of a 99.99 pct aluminum polycrystal (grain size 0.05 ram) strained
0.16 by cold rolling. Slip bands in grain A are associated with short bands in grain B extending 4 / z m into that
grain. It appears that the short bands are on slip planes which are generally not favored in grain B. From
Barlow et al.~~

Fig. 2 9 - - Dark-field electron micrograph from the grain boundary of a 99.999 pct copper polycrystal strained 0.2
in tension. An area of small subgrains is observed in the grain boundary region. From Hansen and Ralph. 75

METALLURGICAL TRANSACTIONS A

VOLUME 16A, DECEMBER 1985--2183

Fig. 30--Transmission electron micrograph of a 99.9 pct iron polycrystal (grain size 0.26 mm) strained O. 15 in tension.
Continuous slip bands are observed at a grain boundary, which is stepped at the points of incidence of the bands. From Jago
and Hansen. 72

Fig. 31 --Transmission electron micrograph of a 99.4 pct aluminum polycrystal (grain size 0.37 mm) strained
0.36 by cold rolling. Three grains, A, B, and C, are shown at a grain edge marked by curved arrows. Identical
subgrains in the montages of grains B and C are connected by dotted lines. In grain A small subgrains are seen
in a band of 2/xm width along the boundary up to grain B. The width of the band is marked by white arrows.
This figure illustrates differences in the microstructure between the grain boundary region and the grain interior
(see text). From Bay and Hansen.~Z5
2184--VOLUME 16A, DECEMBER 1985

METALLURGICAL TRANSACTIONS A

B. Strengthening of the Matrix Region

(a)

In models considering macroscopically compatible deformation and in a number of strength-structural relationships,


the flow stress of the matrix (or grain interior) is considered
to be independent of the grain size. However, it follows
from the previous discussion that large grains and small
grains may not behave in a completely identical manner
during a deformation process and, consequently, the structure and strength of the grain matrix may be partially dependent on the grain size.
Surface observations of the effect of grain size and strain
on the slip length are not available, but an effect of grain size
on the deformation pattern is illustrated by the observation
that the slip band spacing for a given strain may decrease
with decreasing grain size. From bulk observations it appears that the dislocation density may vary between grains
and within grains, and it has consistently been observed that
the average dislocation density for a given strain increases
with decreasing grain size. These findings are supported by
the observation that, for a given strain, more energy is
stored in fine-grained than in coarse-grained specimens. ~40
Dislocations in the grain boundary region may be included
in the measurements of an average density, but, in general,
the values quoted from measurements in thin foils seem to
be quite representative of the matrix region of the grain. It
follows from these observations that the flow stress of the
matrix may increase with decreasing grain size, and such an
effect has been observed directly in the microhardness testing of polycrystalline mild steel.~4~ The effect of grain size
on the matrix stress increased with the plastic strain, and at
a strain of 0.2 this component could account for about half
of the grain size dependent part of the flow stress.

C. Strengthening of the Grain Boundary Region


(b)

(c)

Fig. 32--Channeling-contrast micrographs illustrating the deformation


behavior of a grain in the center section of a commercially pure aluminum
polycrystal (grain size 1 to 2 ram). The specimen (cross-sectional area
30 x 30 m m 2) has been strained 0.05 in tension. The orientation of the
different regions of the grain has been measured by selected area channeling, and misorientation up to 4 deg has been observed across a grain. 137
METALLURGICAL TRANSACTIONS A

The grain boundary region may, in general, be more


strained than the matrix, as has been observed in a number
of materials by selected area diffraction. 23,~4z.143At a number
of grain boundaries a relatively well-defined deformation
zone has been observed (see Figures 28, 29, and 31), and
the width of this zone is typically from one to ten micrometers. This grain boundary zone may have a higher dislocation density and a smaller subgrain size than the matrix.
Moreover, in materials where cross slip is relatively easy it
has been observed that subgrains at some grain boundaries
are larger than those in the matrix, probably due to enhanced
dynamic recovery (see Figure 31).
Although the latter type of grain boundary regions may
have a weakening effect, most of the structural information
indicates that a grain boundary zone, if present, will be
somewhat stronger than the matrix. Supporting evidence has
been found by hardness measurements of a multislip region
in 70:30 brass TM with a grain size of 0.16 mm, where the
hardness of the grain boundary region was found to be about
20 pct higher than the matrix at a strain of 0.02 and about
5 pct higher at a strain of 0.10. The width of the boundary
region was 0.046 mm and 0.025 mm, respectively, and a
calculation based on Eq. [23] shows that the hardness of the
polycrystalline specimen for a strain of 0.1 is close to the
hardness of the matrix. This finding can be correlated with
the results given in Figure 13, which show that the grain size
contribution to the flow stress of 70: 30 brass is significant
for strains as high as 0.3. Thus it appears likely that the
VOLUME 16A, DECEMBER 1985~2185

),-

C2.J"

c=:::30t,..,

c::::

~'C)~

~,~

+
*
x
n

ooo

1110} <112>
1123} (634>
(1121<111>
{100}<001>

',,

+ 1110} <11 2>

* {123} <63z,>
x {112}(111)
[3 (1001 <001 >
1

Levels 1.2, 3

~P2 sections 0, S........... 90 ~

Levels 1, 2,3

(a)

fl

~P2 sections 0, 5 ........... 90 ~


(N

--~ 41

---"- ~1

r?

0
Oil101 <001)
+ {110}<112)
* {1231 (63L>
x {112}(111)
rr~ u 11001<001>

<> 1110l <001 >


+ {110} (112>
* 11231 (63z.>
x {112}<111)
n 11001 <001>
1

Levels 1, 2

~P2 sections 0.,5 ........... 900


(c)

Levels 1, 2, 3

~P2 sections 0,5 ........... 900


(d)

Fig. 3 3 - - T h e effect of grain size on the texture development during cold rolling of commercially pure aluminum. (a, b, c): Grain size 0.05 mm;
strain: 0 (a), 0.16 (b), 0.36 (c). (d, e,f): Grain size 0.35 mm; strain: 0 (d), 0.16 (e), 0.36 (f). It appears from the orientation distribution functions that the
development of a rolling texture occurs faster in the specimens having the smaller grain size. 9~
2186--VOLUME 16A, DECEMBER 1985

METALLURGICAL TRANSACTIONS A

-~
-

~01

vt..~

* ? ~"+

v~.;

Ls
0

r,

2t

,~+ ~ O~ /"
f~'_.

Levels 1.2

i.

r"

~P2 s e c t i o n s

r
011101 (001>
+ {110}<112)
* 1123}(634>
x {112}(111)

. ,b~'~ 0
+
*
x

[] {IOO}<ooi>

'~O~r O {100}(001)

O, 5 ........... 9 0 ~

(e)

L e v e l s 1, 2

II10) (001)
{110}<112)
{123l<634>
1112}<111)

~P2 S e c t i o n s (3,5 ........... 9 0 0

(f)

Fig. 3 3 - - T h e effect of grain size on the texture development during cold rolling of commercially pure aluminum. (a, b, c): Grain size 0.05 mm; strain: 0 (a),
0.16 (b), 0.36 (c). (d, e,f): Grain size 0.35 mm; strain: 0 (d), 0.16 (e), 0.36 (f). It appears from the orientation distribution functions that the development
of a rolling texture occurs faster in the specimens having the smaller grain size. 9~

presence of grain boundaries primarily affects the matrix


strength when the strain is not too small.
An indirect indication that a grain boundary related
strengthening component may exist has been observed by
AI-Haidery et al. 74 in aluminum. Straight lines relating the
flow stress to pl/2 for different grain sizes were displaced
from one another, and the stress displacement increased with
decreasing grain size. Similar behavior has been identified
in iron. 75'98
D. Flow Stress of Polycrystals

The flow stress of polycrystalline materials has been satisfactorily described empirically by the Hall-Petch relationship (Eq. [14]) or by taking or(e) ~ pV2 where p is the
average dislocation density. Both equations are valid for
many metals and alloys and cover a large number of grain
sizes and strains. These strength-structural relations are very
Useful for correlating parameters which are relatively easy to
determine experimentally. However, both equations are
based on very general assumptions which should be related
to the structural features characterizing the deformed state.
The microstructural parameters and their relation to the
flow stress are now discussed in terms of the strengthstructural relations reviewed in Section III. These relations
are based either on a direct barrier effect of grain boundaries
to dislocation pile-ups or on an indirect strengthening by
forest hardening. Supporting structural evidence for the
former mechanism is the extension of slip bands across the
grain and the observation that slip bands formed at small
strains may remain active at higher strains. 145 In general,
METALLURGICAL TRANSACTIONS A

the pile-up model may be relevant for an interpretation


of the flow stress-grain size relationship, at small plastic strains. However, a more general application of this
model in the flow stress region will require a more detailed
analysis of the operation of slip bands in a work-hardened
microstructure.
The forest hardening models are based on the concept that
the dislocation density at a given strain increases when the
grain size decreases. This suggestion is compatible with
many experimental observations showing that the extra dislocations may contribute both to the average dislocation
density and to the local density near or at the grain boundaries. Such a distribution has been discussed in the model by
Thompson et al. ~o~(Eq. [25]), where it is suggested that the
flow stress of a polycrystalline specimen is composed of a
matrix contribution and a grain boundary contribution where
the magnitude of both depends on the grain size.
A verification of this model would require the determination of a number of structural parameters related especially to the grain boundary region, supposing this to be a
structural entity. However, as discussed above, the grain
boundary regions show such wide variation that a quantitative description of their geometry and strength does not
seem feasible. The order of magnitude of a strength contribution from such boundary regions may, however, be
estimated on the basis of microstructural analysis and/or
hardness measurements. 103.135.144 Such observations indicate that, for increasing strains, the extra strength contribution from boundary regions may be relatively small in
comparison with the total grain size contribution to the
flow stress.
VOLUME 16A, DECEMBER 1985--2187

In conclusion, the effect of grain size on the flow stress


of fcc and bcc polycrystalline specimens may be caused by
different elements as the strain is increased. At small strains
dislocation structures at or near the grain boundaries may be
of significance, whereas at larger strains the effect of grain
size on the overall deformation microstructure may dominate. In terms of strain accommodation processes, this
means that microscopic effects may be of importance at
small strains, whereas macroscopic effects may be strength
determining at larger strains.
VI.

CONCLUDING REMARKS

Macroscopic and microscopic strain accommodation processes may affect the flow stress behavior of polycrystalline
specimens. The importance of these processes may depend
on the plastic strain, and it is suggested that the relative
contribution of macroscopic processes increases as the strain
is increased. The strain accommodation processes give rise
to numerous characteristic, structural elements related to
the matrix, the grain boundary region, and the boundaries
themselves. The macroscopic strength properties may be
composed of contributions from these regions, although no
detailed models exist, neither for the strength contribution of
the individual structures nor for the way in which they
should be added. The structure and the strength contribution
from the individual regions of the grains may be greatly
affected by the presence of impurities, solute elements, and
particles. Such effects as well as the influence of texture
should be further investigated to improve the fundamental
understanding of strengthening processes and to advance the
development of engineering materials.
ACKNOWLEDGMENTS
It is a pleasure to thank many colleagues in the Metallurgy
Department for their assistance. A special thanks to Eva
SCrensen for typing this paper. For helpful discussion I am
grateful to T. Leffers, H. Lilholt, O. Br
Pedersen, and
B. Ralph.
REFERENCES
1. A. KochendiSrfer: Physikalische Grundlagen der Formiinderungsfestigkeit der Metalle, Verlag Stahleisen, Diisseldorf, 1963, 55 pp.
2. E. Macherauch: Z. Metallk., 1964, vol. 55, pp. 60-82.
3. U. E Kocks: Metall. Trans., 1970, vol. 1, pp. 1121-43.
4. R.W. Armstrong: Metall. Trans., 1970, vol. 1, pp. 1169-76.
5. T.L. Johnston and C.E. Feltner: Metall. Trans., 1970, vol. 1,
pp. 1161-67.
6. J.P. Hirth: Metall. Trans., 1972, vol. 3, pp. 3047-67.
7. A.W. Thompson: Work Hardening in Tension and Fatigue, A. W.
Thompson, ed., AIME, New York, NY, 1977, pp. 89-126.
8. D. McLean: Strength of Metals andAlloys, E.N.S.M.I.M., Nancy,
France, 1976, pp. 958-75.
9. H. Mecking: Strength of Metals and Alloys, P. Haasen et al., eds.,
Pergamon Press, Oxford, 1980, pp. 1573-94.
10. H. Mecking: Deformation of Polycrystals: Mechanisms and
Microstructures, N. Hansen et al., eds., Rise National Laboratory,
Roskilde, Denmark, 1981, pp. 73-86.
11. T. Leffers: Deformation of Polycrystals: Mechanisms and
Microstructures, N. Hansen et al., eds., Riso National Laboratory,
Roskilde, Denmark, 1981, pp. 55-71.
12. R.W. Armstrong: Yield, Flow and Fracture of Polycrystals, T.N.
Baker, ed., Applied Science Publishers, London, 1983, pp. 1-31.
13. N. Hansen: YieM, Flow and Fracture ofPolycrystals, T. N. Baker,
ed., Applied Science Publishers, London, 1983, pp. 311-50.
14. G. Sachs: Z. Verein. Deut. lng., 1928, vol. 72, pp. 734-36.

2188--VOLUME 16A, DECEMBER 1985

15. G. 1. Taylor: J. Inst. Metals, 1938, vol. 62, p. 307-24.


16. R.v. Mises: Z. Angew. Math. Mech., 1928, vol. 8, pp. 161-85.
17. J. E W. Bishop and R. Hill: Phil. Mag., 1951, vol. 42, pp. 414-27
and pp. 1298-1307.
18. B. Budianski and T.T. Wu: Fourth U.S. National Congress of
Applied Mechanics, ASME, New York, NY, 1962, pp. 1175-85.
19. U.F. Kocks: Acta Metall., 1958, vol. 6, pp. 85-94.
20. U. E Kocks: Acta Metall., 1960, vol. 8, pp. 345-52.
21. U.F. Kocks and G.R. Canova: Deformation of Polycrystals:
Mechanisms and Microstructures, N. Hansen et al., eds., Rise
National Laboratory, Roskilde, Denmark, 1981, pp. 35-44.
22. T. Leffers: Textures in Research and Practice, J. Grewen and
G. Wassermann, eds., Springer Verlag, Berlin, 1969, pp. 120-29.
23. T. Leffers: A. Kinematical Model for the Plastic Deformation of
Face Centred Cubic Polycrystals, 1975, Risci Report No. 302,
114 pp.
24. T. Leffers: Textures of Materials, G. Gottstein and K. Liicke, eds.,
Springer Verlag, Berlin, 1978, vol. 1, pp. 277-87.
25. H. Honeff and H. Mecking: Textures of Materials, G. Gottstein and
K. Lticke, eds., Springer Verlag, Berlin, 1978, vol. 1, pp. 265-75.
26. U.F. Kocks and H. Chandra: Acta Metall., 1982, vol. 30,
pp. 695-709.
27. G.R. Canova, U. E Kocks, and J.J. Jonas: Acta Metall., 1984,
vol. 32, pp. 211-26.
28. I". Leffers and O. B. Pedersen: Strength of Metals andAlloys, R. C.
Gifkins, ed., Pergamon Press, Oxford, 1982, pp. 75-82.
29. T. Tabata, H. Fujita, M. Hiraoka, and S. Miyake: Phil. Mag. A,
1982, vol. 46, pp. 801-16.
30. Y. Kawasaki and T. Takeuchi: Scripta Metall., 1980, vol. 14,
pp. I83-88.
31. P. Ambrosi, E. Grttler, and Ch. Schwink: Scripta Metall., 1974,
vol. 8, pp. 1093-98.
32. W.L. Price and J. Washburn: J. Austral. Inst. Met., 1963, vol. 8,
pp. 1-7.
33. P. Ambrosi, W. Homeier, and Ch. Schwink: Scripta Metall., 1980,
vol. 14, pp. 325-29.
34. A. Kochendrrfer: Plastische Eigenschaften yon Kristallen und
metallischen Werkstoffen, Springer Verlag, Berlin, 1941,
pp. 203-53.
35. E R. N. Nabarro: Some Recent Developments in Rheology, V. G. W.
Harrison, ed., United Trade Press, London, pp. 38-52.
36. W. Boas: Helvetica Phys. Acta, 1950, vol. 23, pp. 159-66.
37. M.F. Ashby: Phil. Mag., 1970, vol. 21, pp. 399-424.
38. M. E Ashby: Strengthening Methods in Crystals, A. Kelly and R. B.
Nicholson, eds., Wiley, New York, NY, 1971, pp. 137-92.
39. A.H. Cottrell: The Mechanical Properties of Matter, Wiley, New
York, NY, 1964, pp. 277-83.
40. B. Chalmers: Proc. R. Soc. Lond., 1937, vol. A162, pp. 120-27.
41. 1. D. Livingston and B. Chalmers: Acta Metall., 1957, vol. 5,
pp. 322-27.
42. J.J. Hauser and B. Chalmers: Acta Metall., 1961, vol. 9,
pp. 802-18.
43. R.E. Hook and J. P. Hirth: Acta Metall., 1967, vol. 15, pp. 535-51.
44. R.E. Hook and J.P. Hirth: Acta Metall., 1967, vol. 15,
pp. 1099-1110.
45. C. Elbaum: Trans. TMS-AIME, 1969, vol. 218, pp. 444-48.
46. R.L. Fleischer and W.A. Backofen: Trans. TMS-AIME, 1969,
vol. 218, pp. 243-51.
47. P.R. Howell, A. R. Jones, A. Horsewell, and B. Ralph: Phil. Mag.,
1976, vol. 33, pp. 21-31.
48. D.J. Dingiey and R.C. Pond: Acta Metall., 1979, vol. 27,
pp. 667-82.
49. E.S.P. Das and M.J. Marcinkowski: J. Mat. Sci. Eng., 1971,
vol. 8, pp. 189-97.
50. E. S.P. Das: Grain Boundaries in Engineering Materials, J.L.
Walter et al., eds., AIME, New York, NY, 1974, pp. 319-26.
51. J.P. Hirth and R.W. Baluffi: Acta Metall., 1973, vol. 21,
pp. 929-42.
52. C.T. Forwood and L. M. Clarebrough: Phil. Mag. A, 1981, vol. 44,
pp. 31-41.
53. C.T. Forwood and L.M. Clarebrough: Strength of Metals and
Alloys, R.C. Gifkins, ed., Pergamon Press, Oxford, 1982,
pp. 27-32.
54. B. Ralph and N. Hansen: Deformation of Polycrystals: Mechanisms
and Microstructures, N. Hansen, et al., eds., Ris~ National Laboratory, Roskilde, Denmark, 1981, pp. 473-78.

METALLURGICALTRANSACTIONS A

55. J.C.M. Li: Trans. TMS-AIME, 1963, vol. 227, pp. 239-47.
56. G. BSro, H. Gleiter, and E. Hombogen: Mater. Sci. Eng., 1968/69,
vol. 3, pp. 92-104.
57. H. Gleiter, E. Hornbogen, and G. BSxo: Acta Metall., 1968, vol. 16,
pp. 1053-67.
58. C.W. Price and J.P. Hirth: Mater. Sci. Eng., 1972, vol. 9,
pp. 15-18.
59. L.E. Murr: Metall. Trans. A, 1975, vol. 6A, pp. 505-13.
60. T. Malis and K. Tangri: Acta Metall., 1979, vol. 27, pp. 25-32.
61. L.E. Murr: Mater. Sci. Eng., 1981, vol. 51, pp. 71-79.
62. E.O. Hall: Proc. Phys. Soc., 1951, vol. B64, pp. 747-53.
63. N.J. Petch: J. Iron Steel Inst., 1953, vol. t74, pp. 25-28.
64. G.Y. Chin: Work Hardening in Tension and Fatigue, A . W .
Thompson, ed., AIME, New York, NY, 1977, pp. 45-66.
65. R. Penelle: Textures of Materials, G. Gottstein and K. Liicke, eds.,
Springer Verlag, Berlin, 1978, pp. 129-53.
66. C. Tome, G.R. Canova, U. E Kocks, N. Christodoulou, and J. J.
Jonas: Acta Metall., 1984, vol. 32, pp. 1637-53.
67. J.D. Eshelby, E C. Frank, and E R. N. Nabarro: Phil. Mag., 1951,
vol. 42, pp. 351-64.
68. J . C . M . Li and Y.T. Chou: Metall. Trans., 1970, vol. 1,
pp. 1145-59.
69. N. Hansen: Acta Metall., 1977, vol. 25, pp. 863-69.
70. N. Hansen: Strength of Metals and Alloys, P. Haasen et al., eds.,
Pergamon Press, Oxford, 1979, pp. 849-54.
71. J.D. Meakin and N.J. Petch: Phil. Mag., 1974, vol. 29,
pp. 1149-56.
72. R. Jago and N. Hansen: unpublished research.
73. R. Armstrong, I. Codd, R.M. Douthwaite, and N.J. Petch: Phil.
Mag., 1962, vol. 7, pp. 45-58.
74. J.T. A1-Haidary, N. J. Petch, and E. R. De Los Rios: Phil. Mag. A,
1983, vol. 47, pp. 869-90 and pp. 891-902.
75. N. Hansen and B. Ralph: Acta Metall., 1982, vol. 30, pp. 411-17.
76. W.L. Phillips and R.W. Armstrong: Metall. Trans., 1972, vol. 3,
pp. 2571-77.
77. N.J. Petch: Phil. Mag., 1956, vol. 1, pp. 331-37.
78. H.H. Tjerkstra: Acta Metall., 1961, vol. 9, pp. 259-63.
79. Y. Bergstr6m and H. Hallen: Met. Sci., 1983, vol. 17, pp. 341-47.
80. W. Roberts and Y. Bergstr6m: Z. Metallkde., 1971, vol. 62,
pp. 752-57.
81. M.J. Marcincowski and H. A. Lipsitt: Acta Metall., 1962, vol. 10,
pp. 95-110.
82. H. Conrad, S. Feuerstein, and L. Rice: Mater. Sci. Eng., 1967,
vol. 2, pp. 157-68.
83. G.W. Greenwood and A.G. Quarrell: J. Inst. Met., 1953-54,
vol. 82, pp. 551-60.
84. N.J. Perch and E. Wright: Proc. R. Soc. Lond., 1980, vol. A370,
pp. 29-39.
85. R.W. Armstrong: Canad. Met. Quart., 1974, vol. 13, pp. 187-202.
86. A.W. Thompson and M.I. Baskes: Phil. Mag., 1973, vol. 28,
pp. 301-08.
87. S. Riegger, O. Vfhringer, and E. Macherauch: Metall, 1979,
vol. 33, pp. 1139-47.
88. E. Macherauch: Z. Metallk., 1968, vol. 59, pp. 669-88.
89. N. Ono and S. Karashima: Scripta Metall., 1982, vol. 16,
pp. 381-84.
90. N. Hansen and D. Juul Jensen: unpublished research.
91. J.E. Bailey and P. B. Hirsch: Phil. Mag., 1960, vol. 5, pp. 485-97.
92. J.E. Bailey: Phil. Mag., 1963, vol. 8, pp. 223-36.
93. J. Meakin and N. J. Petch: Role of Substructure in Mechanical Behavior of Metals, Report ASD-TDR-63-324, U.S. Air Force W-P.
AFB, OH, 1963, pp. 243-51.
94. H. Conrad and B. Christ: Recovery and Recrystallization of Metals,
L. Himmel, ed., Interscience, New York, NY, 1963, pp. 124-30.
95. H. Conrad: Electron Microscopy and Strength of Crystals, G.
Thomas and J. Washburn, eds., Interscience, New York, NY, 1963,
pp. 299-300.
96. J.W. Edington and R.E. Smallman: Acta Metall., 1964, vol. 12,
pp. 1313-28.
97. D.J. Dingley and D. McLean: Acta Metall., 1967, vol. 15,
pp. 885-901.
98. J.-P. Bailon, A. Loyer, and J.-M. Dorlot: Mater. Sci. Eng., 1971,
vol. 8, pp. 288-98.
99. A.S. Keh and S. Weissman: Electron Microscopy and Strength of
Crystals, G. Thomas and J. Washburn, eds., Interscience, New
York, NY, 1963, pp. 231-300.

METALLURGICALTRANSACTIONS A

100. R.L. Jones and H. Conrad: Trans. TMS-AIME, 1969, vol. 245,
pp. 779-89.
101. A.W. Thompson, M.I. Baskes, and W. F. Flanagan: Acta Metall.,
1973, vol. 21, pp. 1017-28.
102. Y.D. Chuang and H. Margolin: Metall. Trans., 1973, vol. 4,
pp. 1905-17.
103. H. Margolin and M.S. Stanescu: Acta Metall., 1975, vol. 23,
pp. 1411-18.
104. W. Boas and M.E. Hargreaves: Proc. R. Soc. Lond., 1948,
vol. A 193, pp. 89-97.
105. V.M. Urie and H.L. Wain: J. Inst. Metals, 1952-53, vol. 81,
pp. 153-59.
106. W. Boas and G. J. Ogilvie: Acta Metall., 1954, vol. 2, pp. 655-59.
107. G.J. Ogilvie: J. Inst. Metals, 1952-53, vol. 81, pp. 491-95.
108. T. Bretbeau and D. Caldemaison: Deformation of Polycrystals:
Mechanisms and Microstructure, N. Hansen et al., eds., Risr National Laboratory, Roskilde, Denmark, 1981, pp. 157-61.
109. C. Barlow, B. Bay, and N. Hansen: Phil. Mag., A, 1985, vol. 51,
pp. 253-75.
110. H. Margolin and A.W. Thompson: Deformation of Polycrystals:
Mechanisms and Microstructures, N. Hansen et al., eds., Ris~
National Laboratory, Roskilde, Denmark, 1981, pp. 197-203.
111. J.A. Ewing and W. Rosenhain: Trans. R. Soc. Lond., 1900,
vol. A 193, pp. 353-75.
112. E.S. Heidenreich and W. Schockley: J. Appl. Phys., 1947, vol. 18,
pp. 1029-31.
113. R.D. Heidenreich and W. Schockley: Strength of Solids, Phys. Soc.
Lond., 1947, pp. 57-75.
114. H. Wilsdorf and D. Kuhlmann-Wilsdorf: Z. Angew. Phys., 1952,
vol. 4, pp. 361-70,409-18, 418-24.
115. D. Kuhlmann-Wilsdorf and H. Wilsdorf: Acta Metall., 1953, vol. 1,
pp. 394-413.
116. J.L. Walter and E. E Koch: Acta Metall., 1963, vol. 11, pp. 923-38.
117. H. Hu: Acta Metall., 1962, vol. 10, pp. 1112-16.
118. I.L. Dillamore, P. L. Morris, C. J. E. Smith, and W. B. Hutchinson:
Proc. R. Soc. Lond., 1972, vol. A 329, pp. 405-20.
119. R.D. Heidenreich: J. AppL Phys., 1949, vol. 20, pp. 993-1010.
120. J. Gil Sevillano, P. van Houtte, and E. Aemoudt: Progress in Materials Science, 1981, vol. 25, pp. 69-412.
121. B.J. Duggan, M. Hatherly, W. B. Hutchinson, and P. T. Wakefield:
Metal Sci., 1978, vol. 12, pp. 343-51.
122. F. Bourelier and J. LeH~ricy: Ecrouissage, Restauration, Recristallation, Presses Universitaires de France, Paris, 1963, pp. 33-39.
123. A.S. Malin and M. Hatherly: Metals Sci., 1979, vol. 13,
pp. 463-72.
124. M. Hatherly: Strength of Metals and Alloys, R.C. Gifkins, ed.,
Pergamon Press, Oxford, 1982, pp. 1181-95.
125. B. Bay and N. Hansen: Deformation of Polycrystals: Mechanisms
and Microstructures, N. Hansen et al., eds., Ris~i National Laboratory, Roskilde, Denmark, 1981, pp. 137-44.
126. M.E. Hargreaves: J. Austral. Inst. Metals, 1956, vol. 1, pp. 125-33.
127. P.R. Swann: Acta Metall., 1966, vol. 14, pp. 900-03.
128. U. Essmann, M. Rapp, and M. Wilkens: Acta Metall., 1968, vol. 16,
pp. 1275-87.
129. V.G. Zankl: Z. Naturforschg., 1963, vol. 18a, pp. 795-809.
130. Ch. Schwink and W. Vorbrugg: Z. Naturforschg., 1967, vol. 22a,
pp. 626-42.
131. Ch. Schwink: phys. star. sol., 1965, vol. 8, pp. 457-74.
132. C. Rey and A.M. Vroux: Non-Microscopical Techniques, N.H.
Andersen et al., eds., Ris6 National Laboratory, Roskilde, Denmark,
1984, pp. 451-55.
t33. M . R . Staker and D.L. Holt: Acta Metalt., 1972, vol. 20,
pp. 569-79.
134. H. Margolin, K. Hashimoto, and H. Yagushi: Scripta Metall., 1981,
vol. 15, pp. 181-84.
135. L.E. Murr and S.S. Hecker: Scripta Metall., 1979, vol. 13,
pp. 167-71.
136. N. Hansen and B. Bay: Strength of Metals andAlloys, R. C. Gifkins,
ed., Pergamon Press, Oxford, 1982, pp. 401-06.
137. J.B. Bilde-SOrensen and N. Hansen: Strength of Metals andAlloys,
H.J. McQueen et al., eds., Pergamon Press, Oxford, 1985,
pp. 141-46.
138. E N. Rhines, R. A. Ellis, Jr., and A.B. Gokhale: Scripta Metall.,
1981, vol. 15, pp. 783-85.
139. N. Hansen, B. Bay, D. Juul Jensen, and T. Leffers: Strength of
Metals and Alloys, H.J. McQueen et al., eds., Pergamon Press,
Oxford, 1985, pp. 317-22.
VOLUME 16A, DECEMBER 1985--2189

140. L.M. Clarebrough, M. E. Hargreaves, and M. H. Loretto: Recovery


and Recrystallization of Metals, L. Himmel, ed., Interscience, New
York, NY, 1963, pp. 63-130.
141. R.M. Douthwaite and N.J. Petch: Acta Metall., 1970, vol. 18,
pp. 211-16.
142. B. Ralph, R.C. Ecob, A. J. Porter, C.Y. Barlow, and N. R. Ecob:

Deformation of Potycrystals: Mechanisms and Microstructures, N.


Hansen et al., eds., RisO National Laboratory, Roskilde, Denmark,

2190--VOLUME 16A, DECEMBER 1985

1981, pp. 111-24.


143. B. Bay and N. Hansen: RecrystaUization and Grain Growth of Multiphase and Particle Containing Materials, N. Hansen et al., eds.,
Risct National Laboratory, Roskilde, Denmark, 1980, pp. 51-56.
144. A.W. Thompson: Strength of Metals and Alloys, E.N.S.M.I.M,,
Nancy, France, 1976, pp. 405-10.
145. A.R. Rosenfield and G.T. Hahn: MetalL Trans., 1970, vol. 1,
pp. 1080-81.

METALLURGICALTRANSACTIONS A

You might also like