RECOVERY AND RECRYSTALLIZATION

Microstructural change during recrystallization of cold rolled Armco iron

90% of metals are “wrought products”. Cast and thermo-
mechanically worked to shape: Plate, Sheet, Wire, Forgings…
Cold Worked alloys are strain hardened - increased dislocation
density.
Annealing - to soften a cold worked metal by heating to high
temperatures and to change its microstructure: Dislocation density
and Grain size, shape and texture are modified.
Recovery, recrystallization and grain growth
are the transformations that take place during anneling.
Recovery

Recrystallization
nucleation and growth of
strain free grains

Grain growth
 What drives these reactions?

Recovery and recrystallization:

Energy of the defects introduced during cold
working (Stored Energy of the deformed matrix)

Grain growth:
Boundary curvature
Most of the energy expanded during cold working is dissipiated
as heat, but a small fraction of it remains in the metal as stored
energy, ES.
 Recovery Recrystallization
CW Annealed CW Annealed

E E
Es Es

CW Annealed CW Annealed

E E
Es
Es
 STORED ENERGY

Elastic Strain 5-10% of the total stored energy.

Lattice Defects: dislocations, vacancies, interstitial atoms, stacking
faults, and twin boundaries.
The fraction of the stored energy produced by each of these defects
depends on the energy per defect and the density of defects produced
by deformation.
Major portion of the stored energy, 80-90%, is due to dislocations.
The vacancies and the others only account for a small fraction of the
total stored energy. But the effect of vacancies on physical properties
is still very important.
cold working : dislocation density, , increase: 107 → 1011 cm/cm3.
Gb2 r
Stored energy per unit volume → ED =  ln   = Gb2
4 (1 − )  r0 
Annealing behavior of a deformed metal is dependent not only on
the magnitude but more importantly on the spatial distribution of
the stored energy.
A uniform distribution
of dislocations is
unstable since the stored
energy can be reduced
by clustering into arrays.
As dislocations become
arranged into dense
tangles various stress
components can cancel
which leads to an
effective lowering of the r
strain energy.
The ability of the dislocations to arrange into tangles depends on
the mobility of dislocations (ability to climb and cross-slip)

High mobility dislocations Low mobility dislocations,

able to tangle, forming cells uniform distribution of slip
 The mobility of dislocations is closeley related to the stacking
fault energy of the metal. Stacking fault energy determines the
dissociation of dislocations into partials.

A A
C
B B B B

(111) Plane in FCC

A total dislocation carries green atom from B

position to the neighbouring B position. On the
other hand, partial 1 carries green atom from B
position to C, then partial 2 carries the atom to
the neighbouring B position.
The movement of a single partial dislocation, however, will
move atoms sitting on B sites to C sites, so that the stacking of
the atoms in between the two partial dislocations is ACACAC.
This region of different stacking is known as a stacking fault.
The energy increase associated with introducing a stacking fault
of unit area is known as the stacking fault energy.
Dislocations dissociate to lower their energy, which is Gb2
Energy balance of dissociation in FCC for partial seperation
of d is,

2*G(a6/6)2 + SFE*d < G(a2/2)2

d < Ga2 (1/2-1/3)/SFE

Partial dislocations are, therefore, promoted by

a low value of stacking fault energy.
d
The presence of the stacking fault makes it more difficult for
dislocations to move out of the slip-plane by cross slip,
because the two dissociated dislocations have to be brought
together first.
Stacking Fault
Cross Slip Plane
Slip Cross Slip Cross Slip Plane

b
b
b

Slip Plane Primary Slip Plane

Primary Slip Plane

Partial dislocations in high stacking fault metals (small separation

between partials), thus have a great tendency to combine to
become a total dislocation which can cross slip readily.
NOT A GENERAL RULE
FCC metals; Cu, g-Iron (except Al, Ni) have low stacking fault
energy. Thus dislocations can hardly cross slip and tangle.

BCC Metals; -Iron and as an exception Al and Ni have high

stacking fault energy.

Alloying generally reduces stacking fault energy.

Metal SFE (erg/cm2) Cross Slip

Brass <10 Difficult
Stainless Steel (304) 20
Silver ~25
Gold ~50
Copper ~80
Nickel ~150
Aluminum ~200 Easy
VARIABLES AFFECTING THE AMOUNT OF STORED
ENERGY

Purity. The addition of impurity atoms to a metal increases the

amount of stored energy at a given strain. Apparently, the impurity
atoms hinder dislocation motion and thereby produce an enhanced
dislocation multiplication.

Deformation. More complex deformation processes produce

higher stored energy. Simple tension may activate slip on only two
slip planes in an fcc metal, whereas more complex deformation
such as extrusion will generally activate slip on all four possible
sets of slip planes. In the latter case dislocation intersection will be
much more frequent, giving rise to higher dislocation densities.
Temperature. Deformation at lower temperatures increases the
amount of stored energy. This is because there is less thermal
energy to assist in the release of energy and to reduce the
interaction between defects during deformation.

Grain Size. The amount of stored energy increases as the grain size
decreases. Consider a large grain that is deformed a given amount.
Now imagine this same volume to be divided into many small
grains and then deformed the same amount. In this second case the
deformation would result in many more grain-boundary-dislocation
interactions. Since the grain boundaries are effective in blocking
dislocations, the smaller grain size promotes dislocation interaction
and multiplication.
Second phase particles. Depends on the size, distribution,
volume fraction and nature of the second phase. For few
undeformable particles of large size, there would be increased
dislocation density only around particles, thus non uniform
distribution of deformation.
 For the uniform distribution of fine particles, because of
dislocation particle interaction, stored energy should be higher.
However the distribution of deformation should be uniform as
well, since particles are not only acting as barriers to
dislocations but also as sources.

Uniform deformation
structure in an
aluminum alloy with
fine Al4CuMg5Si4
dispersoids.
Comparison of driving forces:

G(fusion)  thousands of cal/mole

G(polymorphic)  hundreds of cal/mole
Stored energy  tens of cal/mole
Progress of recovery and recrystallization can be followed through,
•The changes in certain property values,
•Microstructural observations under SEM and TEM,
•Thermal analysis (Calorimetry).

In thermal analysis, two specimens (one cold worked and one

annealed) are heated in separate identical furnaces.

DSC—Differential Scanning Calorimetry

Measures the power requirement difference for keeping the reference
sample and cold worked sample at the same temperature during
heating at a constant rate or at a constant temperature
DTA –-Differential Thermal Analysis
Measures the temperature difference of the reference and cold worked
samples subjected to the same power or power rate.
 Release of Stored Energy
Isothermal Annealing: Constant temperature anneal. Stage I
(recovery), II (recrystallization), III (grain growth)

Energy release rate

I II III

Time
Isochronal Annealing: Two specimens (one cold worked and one
annealed) are heated in separate identical furnaces and power
necessary to maintain the same rate of rise of temperature in each
specimen is measured. Power needed for deformed specimen will
be less because it will release its stored energy of cold work. So
power difference P is proportional to the released stored energy.

Stage I (recovery),
II (recrystallization),
the III (grain growth)
recrystallized

Energy released
grains first
appear at the
onset of the
large power I II III
peak
Temperature
The fraction of the stored energy released during recovery is small for type A
and large for type C. Type A curves are usually obtained for pure metals and
type B or C for impure metals.

The fraction released stored energy due to recovery has been found to vary
from 0.03 for high-purity metals to as high as 0.7 for some alloys. It is apparent
that impurity atoms inhibit the nucleation of recrystallized grains and thereby
allow a larger fraction of the stored energy to be released by recovery processes
prior to the onset of recrystallization.
PROPERTY CHANGES
Hardness. There is usually only a small change in hardness
during recovery, about one-fifth of the total change. Since
hardness (and strength) are decreased with decreasing dislocation
densities, Figure indicates only a small drop in dislocation density
during recovery but a large drop during recrystallization. We
expect the latter result because the new recrystallized grains are
essentially strain free.
Resistivity. The electrical resistivity of a metal is a measure of the
resistance offered by the metal lattice to the flow of electrons
produced by an electric field. Defects in the lattice may act as
scattering sites for the moving electrons and thereby increase
resistivity. Point defects such as vacancies and interstitial atoms
are more effective at scattering than dislocations, so that changes
in resistivity reflect changes in vacancy and interstitial
concentrations. The drop in resistivity during recovery means that
a significant decrease in point-defect concentration must occur
during recovery.
Density. The density of a cold-worked metal is decreased due to
the generation of vacancies. Edge dislocations also make a small
contribution to the lower density because of the dilation of the
lattice about this type of dislocation. The recovery mechanisms
must involve a decrease in vacancy concentration and probably
some decrease in edge dislocation density.

Cell Size. The cell size grows only slightly in the first portion of
the recovery stage but shows a definite increase just prior to
recrystallization.
RECOVERY MECHANISMS

The mechanisms operating in

the low temperature range
involve vacancy motion;
those operating in the
intermediate temperature
range involve dislocation
motion without climb;
those in the high temperature
range involve dislocation
motion with climb.
This order of occurrence is a
reflection of the relative
thermal activation required for
the different mechanisms
Temperature Mechanism

1. Migration of point defects to sinks (GB)

Low
2. Combination of point defects

1. Rearrangenment of dislocations within

Intermediate tangles
2. Annihilation of dislocations
3. Subgrain growth
1. Dislocation climb
High 2. Subgrain coalescence
3. Polygonization
Dislocation tangle condensation into subgrain boundaries
Subgrain Growth: After deformation the dislocation tangles
isolate cellular regions of relatively low dislocation density. Upon
annealing the dislocation tangles condense into sharp two dimensional
boundaries and the dislocation density within the cells decreases. Near
the end of the recovery stage these subgrains begin to increase in size.
Subgrain Coalescence : Electron microscope studies have
shown that in some cases the boundaries between subgrains simply
disappear during the recovery stage. By processes that are not clear,
the orientation mismatch between two neighboring grains
disappears.

This is probably
accomplished by
movement of
interface
dislocations
involving climb
so that diffusion
is required
 Subgrain Coalescence

HVEM bright field images of the same area after in situ annealing
at 700 °C for 5 min (a) and 30 min (b) in a deformed (55%)
Zr–2Hf alloy. Subgrain A, B and C form a larger subgrain D
Polygonization: After the single crystal has been bent, an
excess of positive edge dislocations is generated. Annealing
causes these edge dislocations to line up over one another in
small-angle tilt boundaries. It should be clear that in order for the
dislocations to become lined up they must undergo both glide and
climb. Polygonization is not as well defined in polycrystalline
materials because of the complexity of the deformation. However,
we may say that, in general, (a) polygonization requires an excess
of edge dislocations, (b) it is only produced at higher recovery
temperatures because dislocation climb is involved, and (c) it
produces subgrains that are roughly 10 times larger than those
produced via dislocation tangle condensation.
 KINETICS OF RECOVERY
Zn single crystals were strained in pure shear by some amount γ1.
This increased the yield stress from to to tmax.
Recovery causes the yield stress to decay from tmax back toward to ,
and the rate is a function of time and temperature

t (t,T ) = ?
Let P be some physical property that changes during the
recovery stage (e.g., resistivity, yield stress etc.). We may write

P = P0 + Pd
P0 is the value of the physical property in the annealed state prior
to deformation, base value, and Pd is the increment in the
physical property due to the defects produced by deformation.
Assume that Pd is proportional to the volume concentration of
defects, Cd , produced by deformation,
P = P0 + const Cd
Cd  ( P − P0 )
Time rate of change of the physical property

d ( P − P0 ) dCd
= const (1)
dt dt

The decay rate of the defects is a function of the concentration of

defects and the mobility of these defects. This problem may be
treated as in chemical reaction rate theory and we obtain
dCd
= − K (Cd )n e−Q / kT
dt (2)
P-P0
where Q is the activation energy for the process of defect
annihilation occurring and n is an integer, being 1 for first-order
kinetics, 2 for second-order, and so on, and K is a constant.
Combining equations 1 and 2

d ( P − P0 )
= − K ( P − P0 )n e−Q / kT
dt

d ( P − P0 ) − Q / kT
= − Ae dt
( P − P0 ) n

Let n=1

d ( P − P0 )
 ( P − P0 ) 
− Q / kT
= − Ae dt

ln( P − P0 ) − cons. = − Ae−Q / kT t

To find integration constant substitute initial condition,
which is P=PCW at t = 0.

ln( PCW − P0 ) − cons. = 0

cons. = ln( PCW − P0 )
Substitute the constant,

ln( P − P0 ) − ln( PCW − P0 ) = − Ae−Q / kT t

( P − P0 ) − Q / kT
ln = − Ae t
( PCW − P0 )
 PCW − P
Define fraction recovered, R=
PCW − P0

P − P0
and residual recovery, 1− R =
PCW − P0

Then,
( P − P0 )
ln = − Ae − Q / kT t
( PCW − P0 )
ln(1 − R) = − Ae − Q / kT t
 1  − Q / kT
ln   = Ae t
 1− R 
 1  Q
ln ln   = ln A − + ln t
 1− R  kT
 Then is it possible to obtain information concerning the mechanisms of recovery
from analysis of the kinetics of recovery?
From a plot of lnt versus 1/T at a constant value of P, we determine a value of Q
from the slope.

 1  Q
ln ln   = ln A − + ln t
 1− R  kT

constant
Q
Rearrange, ln t = const +
kT

lnt Q/k

1/T
If n>1, then
d ( P − P0 )
 ( P − P0 )n 
− Q / kT
= − Ae dt

1 1− n − Q / kT
( P − P1− )
 n( P − P0 ) 0
11− n −cons . = − Ae
A t
− Q / kT
 − 1 = − e t
1 − n  ( PCW − P0 )1−n ( PCW − P0 )1−n
Applying the initial condition that, t = 0, P=PCW

 ( P − P0 )1−n  A (1 − n ) −Q / kT
 1− n
− 1 = − 1− n
e t
 ( PCW − P0 )  ( PCW − P0 )

Residual recovery Constant = B

(1 − R )
1−n
Then, = 1 − Be−Q / kT t
The above example, data on Zn gave a linear relation between lnt and 1/T ,
indicating that only one activation energy and, hence, only one defect
annihilation process was involved.
Q was found to be approximately the same as the activation energy for self-
diffusion, implying that vacancy migration is involved in this recovery process.
It therefore indicates a dislocation climb controlled recovery process.

In recovery of pure Fe, Q value is close to ΔE(vacancy migration) at short

recovery times but close to Q(selfdiffusion) at long recovery times, indicating
that the recovery mechanism is time dependent.
Normally, activation energy for thermally activated single atomic processes is
constant and independent of time or temperature.
Most probably, the reason of the change in activation energy is annihilation of
strain induced vacancies during the early stages of recovery, which reduces the
vacancy controlled diffusion rate by increasing the diffusion activation energy,
during dislocation climb.
Q(selfdiffusion) = ΔE(vacancy migration) + ΔE(vacancy formation)
CONCLUSIONS
− Q / kT
ln( P − P ) − ln( P − P ) = −
1.Recovery usually0 occursCWexponentially
0 Ae t
with
time. ( P − P0 )
ln = − Ae−Q / kT t
( PCW − P0 )
2. Proper analysis of kinetic data allows one to
determine Q in some cases.
 1  Q
ln ln   = ln A − + ln t
 1− R  kT
3. Generally, more than one recovery mechanism
operates so that Q is not a constant.
 NUCLEATION FOR RECRYSTALLIZATION
For a recrystallization nucleus three criteria have to be met:
Thermodynamic instability: a viable recrystallization nucleus has to
exceed a critical size, r*, so that the Gibbs free energy decrease with
the growth of nucleus,
* 2g 2g
r =
D.F. = Gb 2

Because of the low driving force for recrystallization, the predicted

radius of the critical sized cluster (embryo) is too large compared
to experiments amd improbable.
Therefore, one must assume that a supercritical nucleus is
already present in the deformed microstructure (pre-existent
nucleus), for instance in form of a dislocation cell or subgrain but
then additional recovery processes are necessary, to activate such
a cell as a nucleus.
Mechanical instability: The grain boundary of the nucleus has to
move at the expense of the cold worked structure, which requires
a local imbalance of the driving forces. This condition can be
met by an inhomogeneous dislocation distribution or by a local
imbalance of subgrain sizes that develops during the incubation
period by recovery processes.

Kinetic instability: The boundary of a nucleus must be mobile to

make the nucleus grow. Only high-angle boundaries have a
sufficiently high mobility.
The constraint to comply with all three instability criteria
concurrently favors nucleation in specific regions of the deformed
microstructure, especially in deformation inhomogeneities and at
prior grain boundaries. By subgrain growth, a grain boundary with
high misorientation and, thus, high mobility can be generated.

Subgrain
Growth
 Nucleation of grains results from the sudden growth of a high-
mobility boundary. These boundaries are either:
1. An original high-angle boundary.
2. (a) A high-angle subgrain boundary formed by a subgrain
enlargement mechanism.
(b) An existing high-angle subgrain boundary modified by
unknown atomic rearrangements.
Mechanisms 1 and 2(a) are more common in lightly deformed
metals and mechanism 2(b) is more common in highly deformed
metals.
The "nucleation" event appears to be a growth phenomenon so that
variables affecting growth will also affect "nucleation" in a similar
way. Even though the new grains do not spontaneously nucleate in
the sense of the classical nucleation picture, we may still treat the
recrystallization transformation as being composed of nucleation
and growth stages.
Subgrain Growth
The growing subgrain, as its boundary acquires more dislocations,
will become progressively more misoriented with respect to its
neighbours. Eventually they undergo a transition to a conventional
high angle boundary.

Subgrain Coalescence
Similar to previous
model based on the
hypothesis of
coalescence of
neighbouring
subgrains.
Strain Induced Boundary Migration (SIBM)
Nucleation at high mobility grain boundaries usually is preceeded
by grain boundary bulging. A critical bulge size has to be exceeded
for a bulge to be a nucleus..
Usually the driving forces on both sides of a grain
boundary are assumed to be equal, but actually an
imbalance can be present due to a dislocation
density or subgrain size distribution difference
across the grain boundary.
For the method to operate, the dislocation
distribution must be inhomogeneous on a
sufficiently large scale, i.e., a particulary large
subgrain neighbouring to a grain boundary may
cause that grain boundary to bulge across the
entire width of the subgrain.
Strain induced grain boundary migration can occur after a relatively
small or moderate amounts of cold work. Differring from
recrystallization in that no new crystals are formed. Rather the
boundaries between pairs of grains move so as to increase the size
of one grain while the other to decrease.
 No new crystals are formed, instead, a boundary between two
grains move at the expense of the more deformed one. Hence,
the "nucleation" event is essentially a growth phenomenon.
Assume that the mobile boundary is pinned at two points as it bulges out. Let the
volume change in going from position I to II be dV and the overall free-energy
change per volume be ΔG. The surface free energy of the boundary is taken as γ,
and the stored energy per volume in the cold-worked grain is Es. For the
boundary moving from position I to II a free-energy balance gives,

dA
G  dV = − E s  dV + g  dA G = − Es + g
dV
Released
stored energy
is used to
create a larger
surface area.
 dA 1 1 2g
Where, = + For spherical boundaries; G = − Es +
dV r1 r2 r
r
In order to have growth, the value of ΔG
must be negative, so that Es must be larger
than 2γ/r. Since the boundary is pinned at a a
distance of 2a , the radius of curvature will progress

go through a minimum, rmin= a. Therefore, G

in order to have growth of the bulging
boundary it is necessary that
2g
Es  rmin=a
a
The incubation time would
be the time required for the
bulge to grow out to the
critical position of rmin=a
Even without bulging of the boundary the area adjacent to a grain
boundary is favored for nucleation because usually the dislocation
density is higher and more inhomogeneous due to compatibility
constraints arising from joint deformation of adjacent grains.

The same argument holds

for large particles too.
Heterogeneous nucleation at
a particle surface and a high
and inhomogeneous
dislocation density
distribution promote rapid
nucleation in two phase
alloys with a coarse
dispersion.
GENERAL CONCLUSIONS
Nucleation of grains results from the sudden growth of a high-
mobility boundary. These boundaries are either:
1. An original high-angle boundary.
2. (a) A high-angle subgrain boundary formed by a subgrain
enlargement mechanism. (b) An existing high-angle subgrain
boundary modified by unknown atomic rearrangements.
Mechanisms 1 and 2(a) are more common in lightly deformed
metals and mechanism 2(b) is more common in highly deformed
metals.
The "nucleation" event appears to be a growth phenomenon so that
variables affecting growth will also affect "nucleation" in a similar
way. Even though the new grains do not spontaneously nucleate in
the sense of the classical nucleation picture, we may still treat the
recrystallization transformation as being composed of nucleation
and growth stages.
 KINETICS OF RECRYSTALLIZATION

Recrystallization occurs by the nucleation of new strain-free

grains that grow and consume the cold-worked matrix.
Recrystallization rate, d(volume recrystallized)/dt, is a function
of both the nucleation rate and the growth rate of these new
grains.
After a time t new grain start to nucleate in the cold-worked
matrix and begins to grow. Frequently, these will increase in size
at a constant rate until it impinges upon a neighboring grain.
 NUCLEATION AND GROWTH

t0 t1 t2 t3

100
Sigmoidal kinetics
% Product

50 y = 1 − exp( −ktn )
Incubation
period
t 50 log (time)
0
5
 impingment
R

Growth rate R = G (t − t )
G=dR/dt

t t
4 3 4 3
= R = G ( t − t )
3
Volume of a spherical nucleus; Vnuclei
3 3
Number of nuclei formed in volume V in time interval dt is

n = N  dt V
number nuclei formed/unit time
N=
untransformed volume
The total volume transformed until time t = n*dt
ndt
Vtransformed =  (vol. per nuclei )*(# of nuclei /dt)
0

t
Vtransformed =   G (t − t ) * N Vdt
4
3
3 3

0
It is more convenient to work with the fraction of the volume
transformed,
t
Vtransformed
X transformed = =  43 G 3 (t − t ) 3 * N dt
V 0


X transformed = G 3 N t 4
3
  Experimental
X transformed = G N t 4
3

3 sigmoidal curve

Xtransformed

time

The error is due to predicting

“n” in terms of the total volume
rather than untransformed
volume. Nuclei may not form in
the already transformed volume.
 t
= 3 − t V
4 3 3
Vreal transf ormed G ( t ) * N untransf ormeddt
0

dVr = 43  G 3 (t − t ) 3 * N Vu dt
V = Vr + Vu
dVr =  G (t − t ) * N (V − Vr )dt
4
3
3 3

  V − Vr 
dX r =  G (t − t ) * N  dt = 43  G (t − t ) * N (1 − X r )dt
4 3 3 3 3
3
 V 
Xr t
dX r
0 1 − X r 0 3
=  − t  dt
4 3 3
G ( t ) * N
assume t=0

− ln(1 − X r ) = G 3t 4 N
3
Johnson-Mehl
Eq’n X r = 1 − exp( − 3 G 3 N t 4 )
   3 4
Johnson-Mehl eq. X real = 1 − exp  − G Nt 
 3 
applicable to any heterogeneous phase transformation
subject to the four restrictions :
random nucleation, constant Ń, constant G, and small t.

“Sigmoidal" shape of these curves is characteristic

of all nucleation and growth transformations.
Volume fraction transformed is a much stronger function of G
than of Ń.

  3 4
X real = 1 − exp  − G Nt 
 3 
Avrami considered the case where the nucleation rate decayed
exponentially with time. In this case the above analysis with
X r = 1 − e− kt
n

variable Ń leads to,

 1 
ln ln   = ln k + n ln t
− kt n  1 − Xr 
Xr = 1− e n
 1 
ln ln   = ln k + n ln t
 1 − Xr 
ln k
ln t
where k and n are constants. For rapid decay the results predict n=3
and for slow decay n=4.
k is a thermal activation related term for the recrystallization process;

 Q 
k = k0 exp  − 
 RT 
Growth rate G is a usually constant but the nucleation rate
Ń is not in solid state phase transformations, particulary at
the early stages, probably because of the experimental
difficulties in the determination of small nuclei.
 EXPERIMENTAL DETERMINATION OF G AND Ń

The determination of the nucleation

rate, Ń, and growth rate, G, is a
very tedious process requiring
much quantitative metallography.
Identically strained samples are
annealed at some temperature
and individually quenched after
varying times. After
metallographic preperation,
number of new recrystallized
grains are counted, Ns, and
plotted as a function of time.
Slope of which gives the
nucleation rate, Ń
Growth rate is determined from the slope of the curve of radii of
the largest unimpinged grains in each sample plotted versus
time. It is assumed that the largest grains nucleated first, t=0.
The intercept of the curve on the time axis gives the incubation
time.
Note that the growth rate was
a constant in this study.
2x10-4
However it may not be so mm/sec 7x10-4
because of the fact that mm/sec
during the recrystallization,
recovery processes will
occur thereby changing the
stored energy Es , and hence
the growth rate.
FACTORS AFFECTING GROWTH & NUCLEATION RATE
Growth Rate. DB E s
G = B = 
 kT 
Increasing the stored energy will increase the growth rate.
Therefore, the growth rate is increased by increasing strain or by
decreasing grain size, both of which increase Es .
The effect of impurity upon the growth rate of the recrystallized
grains is very strong. Although they increase the stored energy,
because of their strong influence upon the mobility, apparently
due to a drag effect, they lower the growth rate

The temperature dependence of the growth rate is found to

follow an Arrhenius equation.
− Qg / RT
G = G0e
we expect the measured Qg values for grain growth to
correspond to the activation energy for grain-boundary self-
diffusion, since the only term in growth equation strongly
temperature dependent is DB .
Nucleation Rate. The nucleation rate of recrystallization is found
to increase with strain. There is a critical amount of strain
necessary for nucleation and therefore recrystallization to occur,
due to homogeneity of deformation at low strains.
The effect of prior grain size upon nucleation rate is quite
pronounced. Smaller grain size causes a much higher rate of
nucleation, and shorter incubation periods.
Smaller grains produce a more complex stress pattern that results
in higher local deformations, which enhances the nucleation rate.
The effect of purity upon nucleation rate does not seem to have
been well studied. Since impurities increase the amount of stored
energy for a given strain one expects the nucleation rate to be
increased by the presence of impurities. However in grain boundary
bulging mechanism of nucleation presence of impurity may
affectively retard nucleation.

The temperature dependence of the nucleation rate follows an

Arrhenius equation,

− Qn / RT
N = N 0e
Activation energy for Ń and G is roughly the same except at
low strains.
 OVERALL EFFECT ON RECRYSTALLIZATION

Strain
Grain size
 Impurity

Xr
 CONTROL OF
RECRYSTALLIZATION TEMPERATURE
AND GRAIN SIZE
From a practical point of view we are interested in
controlling the recrystallization temperature and the
recrystallized grain size. To estimate how these are
affected by the parameters which we may control,
Johnson-Mehl equation will be used

  3 4
X real = 1 − exp  − G Nt 
 3 
 RECRYSTALLIZATION TEMPERATURE

Recrystallization temperature is defined as the temperature at

which 95% recrystallization (practically almost complete
transformation) occurs in 1 hour (which is a reasonable time)

Using the Johnson-Mehl equation,

time required for 95%
ln t
recrystallization, t0.95, is

1/ 4 Low Driving force

 2.85 
t0.95 = 3
 NG 
High Driving force

Ln t0.95=K + Q/RT
Trecry  0.4Tm
 RECRYSTALLIZED GRAIN SIZE

R=G(t-t) G*t
t  R/G

  3 4
X real = 1 − exp  − G N ( R G ) 
 3 
14
G 
d = 2 R = const   
 N 
Small recrystallized grain size is promoted by a high nucleation
rate and a slow growth rate; that is, the grains nucleate at a high
density and grow very little.
The predominant factors that affect Ń and G are:
(1) amount of prior strain,
(2) temperature of anneal,
(3) prior grain size, and
(4) purity.

How these factors influence

(a) the value of Trecry and
(b) the value of d.
 Effect of Parameters on Trecry & Grain size

Prior Strain
Trecry is decreased by
high strains since high
strains increase both Ń
and G.

High strains decrease

recrystallized grain size,
d, because Ń/G increases
with strain.
Annealing Temperature

Higher annealing temperature produce faster recrystallization

since both Ń and G increase with temperature.

d is probably a weak function of annealing temperature

because Ń and G both follow an Arrhenius equation with Q
nearly the same, so one expects Ń/G to be nearly constant with
temperature.
Prior Grain Size

Fine-grained metals store more energy during deformation

because of the more complex deformation required.

Trecry is lowered by fine grains since both Ń and G increase

due to increased stored energy.

d is decreased by fine
grains.
The smaller grains cause
inhomogeneous
deformation, which
increases Ń faster than G.
Impurities
Addition of impurities increases the amount of stored energy at a
given strain, so that one might expect the growth and nucleation
rates to increase. However, impurities are extremely effective at
decreasing the interface mobility and this latter effect is dominant.
So, impurities decrease the growth and possibly the nucleation rate.

Trecry is increased by addition of impurities.

Smaller recrystallized grain size is promoted by impurities because

Ń/G increases with impurity content.
Deformation Temperature
Straining at higher temperatures allows more recovery to occur
so that one expects less stored energy to drive both Ń and G.

Trecrys will increase with deformation temperature, because both

G and Ń decreases.

Recrystallized grain size, d, will increase with increased

deformation temperature because Ń/G will be smaller due to
homogenization of strains.
 Summary

Strain , Es ;G ,t also N
Impurity , Es but due to drag ; G , effect on N ?
Grain size , Es ;G and complex strain pattern ; N
Temperature ; G, N both arrhenius dependence with similar Q

So how does Trecrys and d depend on these parameters?

Strain , G N ; t0.95 , Trecry and G/N ;d

Anneal T , no meaning for Trecry; G, N both Arrhenius with similar

Q then d is a weak function of T

Grain size ,G N ; t0.95 , Trecry

G/N ; d

Impurity ,G N~ ; t0.95 , Trecry

G/N ; d

Temperature of prior strain , more recovery , Es ;G N

Trecry and decreased local strains G/N , d
 Recrystallization in Two Phase Alloys

Dispersed Second Phase

The presence of dispersed second phase particles may
accelerate or even completely suppress recrystallization.
Generally, recrystallization kinetics are speeded up by coarse
particles, whereas a fine dispersion of particles strongly
hinders recrystallization.

This complex influence of particles on recrystallization is due

to the fact that particles influence both the deformation
structure, recovery during nucleation, and eventually grain
boundary migration.
 Hard coarse particles cannot be deformed and geometrically
necessary dislocations will generate an inhomogeneous dislocation
structure around those particles. If the particles are large, this can
facilitate particle stimulated nucleation. Hence recrystallization
may be accelerated.

Nucleation
site
Finely dispersed particles are detrimental for nucleation, because
usually they homogenize slip. Moreover they hinder dislocation motion
(recovery) and grain boundary migration, lowering the nucleation and
growth rates. Hence recrystallization is always retarded.

Small particles impeding the motion of

dislocations at a subgrain boundary
Effect of second phase particles on recrystallization, depending
on particle size and spacing.
 Precipitation-Recrystallization Interractrion
During the annealing of cold rolled supersaturated solid solutions,
either or both precipitation and recrystallization are found to occur.

In the TTT diagram,

start and finish lines
of recrystallization
are assumed to be
the ones in alloys
without
precipitates. It is
seen that three
regimes are present.
Regime I: Precipitation Before Recrystallization
There is significant precipitation before any recrystallization has
occurred, and recrystallization is prevented. However this is a very
slow process since the temperature should be kept at low values.
Regime II: Simultaneous Precipitation and Recrystallization
Recrystallization starts before precipitation, but subsequently will
inhibit the completion of recrystallization.
Regime III: Precipitation After Recrystallization
Recrystallization occurs
before precipitation, therefore
it is unaffected by second
phase particles.
In Regime II, where recrystallization is retarded by precipitation,
actual (affected) TTT diagram will be as shown below.
Note that the increased CW (dislocation density) shift the
recrystallization curves to shorter times.
 Duplex Alloys
Examples of duplex alloys include /b brasses, /g steels and /b
titanium alloys. A typical case would be /b brasses, where the
microstructure comprises equiaxed grains of -brass (FCC-low
stacking fault energy) and grains of ordered b-brass (BCC).

Upon annealing b-brass

recovers readily and
forms subgrains, while
-brass recrystallizes,
resulting in a duplex
microstructure.
For annealing temperatures different from the original
precipitating temperature, the volume fractions of  and b must
change, due to the phase diagram, so that an  nuclei may form
within the b grains which then act as recrystallization nuclei for
the deformed  grains.
 Texture
The distribution of orientations in a polycrystal is referred to as
crystallographic texture. When a metal is severely cold worked
the individual grains will all tend to rotate so as to produce a
preferred crystallographic orientation in the direction of
deformation, which is generally referred to as a deformation
texture.

When a metal containing a deformation texture is annealed to

produce recrystallization one might expect the new grains to be
randomly oriented. This is frequently not the case and one often
obtains what is termed an annealing texture or a recrystallization
texture. This result means that the new recrystallized grains must
form with a preferred crystallographic orientation relative to the
deformed grains.
The preferred orientation of the recrystallized grains must be due
to one or both of two causes.

Oriented Nucleation Theory: The nuclei form with a preferred

orientation.

Oriented Growth Theory: Only those grains with the preferred

direction grow to an appreciable volume; the mobility of other
grains is too low and they are suppressed by competitive growth.

Over the years there was some debate on the relative merits of
these mechanisms; but there are many evidences now that shows
that both can take place.
 The cube texture,
{001}<100>, is often
obtained as the
recrystallization
texture in fcc metals.

A particular example where this texture is advantageous is in Fe-Si

alloys. Because of the anisotropy of the magnetic properties of iron
the cube texture makes a much superior transformer sheet material.

{100} pole figures of

(a) primary, (b) secondary
recrystallization in Fe-Si
 Secondary Recrystallization

If the microstructure remains essentially stable except for the

growth of a few grains, this is referred to as discontinuous grain
growth. It requires that normal grain growth is globally suppressed
in the specimen, but is triggered at a few locations.
Secondary recrystallization is found to occur in metals that have
achieved a relatively stable small grain size. Such a condition is
obtained in three common ways:
(1) In the cold worked metal with a deformation texture,
recrystallization texture is formed after primary recrystallization.
Since the recrystallized grains have a similar orientation there is an
absence of high-angle boundaries. Hence, boundary mobilities are
low and the grain structure is relatively stable.
(2) The metal contains inclusions that result in a limiting grain size.
(3) The metal is in thin sheet form. The boundaries intersecting the
surface tend to form grooves at the surface intersections and these
grooves retard the boundary motion.
A technically important example is discontinuous grain
growth in two-phase materials during annealing close to
the solvus temperature. During such an annealing
process, precipitates locally dissolve and initiate grain
growth, while the remaining granular microstructure
remains stabilized by the precipitates.

Some of the locally growing grains gain such an

overwhelming size advantage that they can overcome
Zener pinning and, therefore, become extremely large
grains by consuming the nearby small grains pinned by
Zener drag.
 Hot Deformation;
Dynamic Recovery & Recrystallization
If a rod of soft solder is bent back and forth several times it is
found that the rod remains soft. The same operation done on a
rod of annealed copper results in a rapid work hardening so
that the copper can no longer be bent by hand after only a few
bends.
Why the solder does not work harden? The reason is that the
recrystallization temperature of solder is sufficiently low to
allow recrystallization to occur during the bending process.

Hardening by cold working

Simultaneous
Softening by recovery & recrystallization
At temperatures above approximately 0.5 Tm, recovery and
recrystallization occurs during deformation. These softening
mechanisms operative during high temperature straining are called
dynamic recovery and dynamic recrystallization.
Deformation zone
Annealing zone

The simultaneous ??? occurrence

of recrystallization and
deformation during hot working
???????
 HOT WORKING vs CREEP

Low   = B( ) n

 = B ( Sinh   )
 n

High   = B exp( b )
B = A exp(-Q/RT)
CREEP → Low  → power law → tested under constant
load (or ) → independent (measured) parameter is
strain rate → recovery dominated
HOT WORKING → High  → exponential law → tested at
constant strain rate → independent (measured)
parameter is stress → recrystallization dominated
  = B ( )
 n
where

 =  − i
 is the applied true stress and i the true internal stress.
i is modified by dynamic recovery and recrystallization

Static conditions Dynamic conditions

Ferritic Alloys Recovery & Only Recovery
recrystallization
Austenitic Alloys Limited recovery Recovery at small 
& recrystallization & recrystallization
In Al and ferritic alloys dynamic recovery is usually the only
softening mechanism.
Hardening mechanism---reduction in subgrain size by deformation
which is constant at constant strain rate.
Softening mechanism---increase of subgrain size by recovery, rate
of which is inversly proportional with subgrain size.

2 , T2

1 , T1

1  2 ; T1  T2


Dynamic recrystallization
is observed as the
dominant mechanism in
austenitic alloys and
appears as a single peak or
oscillations in the flow
curve.

Critical values of stress

and strain at which
dynamic recrystallization
is initiated depend on
material and
deformation.
Flow curves of a carbon steel at
1100 C and various strain rates
 Many concentrated
alloys and dispersion
strengthened materials
do not recrystallize
dynamically.

Torsion flow curves of copper and copper

alloys at a strain rate of ε˙ = 2 · 10−2 s−1
The flow stress to start dynamic recrystallization
decreases with increasing deformation temperature
and decreasing strain rate.
The flow stress behavior can be explained and formulated in a
phenomenological theory by the assumption that a static
recrystallization process in different regions of the material is
superimposed to deformation. A flow curve with oscillating flow
stress is observed, if the cycles of recrystallization occur
separately. If the recrystallization cycles overlap only a single
flow stress peak is observed.
This phenomenon is closely related to the critical strain c,
required for recrystallization and that recrystallization should
occur repeatedly each time the critical strain is attained in
previously recrystallized grains.
At the same time, once recrystallization has started, a finite strain
x, should be required in order to attaine a structure that is
effectively completely recrystallized (e.g. 98% recryst.).
Difference between stress-strain curves should be due to whether
x is smaller or larger than c.
If x<c a given cycle of recrystallization should be completed
before the next one starts. On the other hand, if x>c, the first
wave of recrystallization will not have had enough time to be
completed before the next one starts.
The process of dynamic recrystallization is important for hot
forming. It keeps the flow stress level low, i.e., requires only low
deformation forces, and substantially improves ductility .

Moreover, the dynamically recrystallized grain size is directly

correlated with the flow stress in such a way that with increasing
flow stress the grain size decreases.
 Summary

Dynamic Recovery Dynamic Recrystallization

A .
T1 or  1
T1< T2
. > .
B .
T or 
1 2
2 2 A .
T 1 or  1
T1< T2 B T2
.
or 
. . 2
1> 2

strain strain

A: Slower annealing kinetics or faster deformation

B: Faster annealing kinetics or slower deformation