New grains nucleate

and grow into regions

of high dislocation High dislocation density
density.
New grain

1
Fraction (Re)crystallized

0.999

Experimental
data usually fits a 
1 exp    t
n  0.5
sigmoid curve

0 0
0 50 100
0 t 100
Time
Rate of recrystallization in the Rate of recrystallization in the Rate of recrystallization in the
early stages is limited middle stage is a maximum late stage is limited by
because of the presence of impingement and the lack of
by nucleation and many nuclei, high grain contact between grain
grain boundary area boundary area and limited boundaries and high
Rate of (Re)crystallization (fraction/s)

impingement dislocation density areas

0.03
0.022

0.02
d  1exp     t n  0.01
dt
0
 14
 1.03810 0.01
0 50 100
0 t 100
Time
of thermally activated, nucleation and growth processes can be approximated
can be determined.
Mehl-Avrami This allows
(JMA) equation. The one to examine
simplest form of how the temperature
the JMA of th
equation for the vo
ment controlsX(t),
ecrystallized, the is
rate of the process without actually having to do all the
Volume fraction transformed fits the curve
riments. It is also possible to use the values of Q and n to evaluate the m
rystallization
energy for the process.
process Toisdetermine 
implicitly n
Q contained
and n from equations [1] and [2], t
X( t) 1  exp T  G  t  in  and may be writte
riments at three or four temperatures needs to be massaged. Note that th
is a function of temperature, T, and free energy change, G for the (re)crysta
rimental data is the lhs of equation [1] which we attempt to fit to the rhs o
t iscan
the rearrange
time and nequation
is the exponent forthe  equation
Qprocess. An example of the form1off
We  a exp 
[1] by substituting  [2], subtracting
ntmultiply
(x-axis)each
recrystallization fraction
side by -1, take the natural  R T log of each
(y-axis) predicted by the JMA
side, equation
multiply byis-1g
and Ќ istake
then assumedthe tonatural
described
loga of
thermally
each activated
side process
again. Now (diffusion,
the data nucleation)
will have t
activation
.000007 energy, which
n  3 t  0 100 can be a function of the free energy of
en.onThea 3-D
JMA
Let
plotspecify
of thethe
X(t) equation
massaged
is usefuldata
experimental ifdata
the(z-axis) vs the natural
of exponent,
log as
volume fractionn,transformed
and the of time (x-
activation
a
nverse
termined.of temperature
function
Thisof time (1/T).
and
allows oneThis
Temperature has the form
to examine howofthe temperature of the h
rols the rate of the process without actually having to do all the
 
ln lnto   1    lnn
t to evaluate
Q
is[3]also possible use
 datathe
 t  values
  1of
  Q nand ln( a)  the mech
ion process. To  determine
  QTand  nfrom
 equations [1] and R T
[2], the
three or four temperatures needs to be massaged. Note that the
, n is the slope of the massaged data on the rhs of equation [3] versus ln(
ata is the lhs of equation [1] which we attempt to fit to the rhs of eq
e data versus 1/T is -Q/R. The intercept projected back to ln t =0 and 1/T=
arrange equation [1] by substituting equation [2], subtracting 1 from
xample of such a plot with n=3, a=0.01, Q/R=20 K.
each side by -1, take the natural log of each side, multiply by -1 on
n a 3-D plot of the massaged data (z-axis) vs the natural log of time (x-axi
rse of temperature (1/T). This has the form of

3]     1  
ln ln data t    1  n  ln t   ln( a) 
Q
   T    R T

is the slope of the massaged data on the rhs of equation [3] versus ln(t) a
ata versus 1/T is -Q/R. The intercept projected back to ln t =0 and 1/T=0 is
mple of such a plot with n=3, a=0.01, Q/R=20 K.

1/T
ln(t)
How can we get information about the physical process from n and Q?

So how can we get information about the process from the value of n and Q? As an example,
consider spherical JohnsonMehlAvrami Equation
growth of grains with boundary velocity, G, nucleating at a constant rate p
unit volume, N. If we pretend that growing grains do not impinge (run into) one another and tha
the total number of nuclei formed per unit time does not depend on the the fraction
Define the virtual volume fraction transformed-No impingement.
recrystallized, then we can write an equation for this imaginary (virtual) volume fraction
recrystallized, Xv ( t) as For spherical grains this is

  t 
X v ( t)   N   G   t   d 
4 3 3
[4a]
 3
 0 
 
t
s
 3 s 
 N 4   G t   3 d   1  t 4 N  G
3 s
G is the interface velocity (growth rate) of the grain
0 boundary (m/s)
Nwhere
is the marks
nucleation rate
the time per unit
at which volumeevent
a nucleation (1/(m^3 s))To be more explicit about the
occurs.
t difficult
is the clock of a nucleation time, , as opposed to a clock time, t, consider what the radi
concepttime
τ of
isathegrain would be after
nucleation a 20 second (t) heat treatment if it nucleated 10 seconds () after the
time
6 m
heat treatment began. If the value of G is 10  this radius would be given by
s

G  t   10 6 m  ( 20 s  10 s ) 10 5 m

 
 s 
The volume of this grain can then be determined from the radius. What we have to do to get t
 How can we get information about the physical process from n and Q?

R G  t  
R
R=G(540 s)

R=G(480 s)

1:00 am 1:01 am 1:02 am 1:10 am

time
Nucleation time for the upper particle is 1:01 am and lower particle is 1:02 am
Clock time for both particles is 1:10 am

Volume of individual grain is

4

  G t    3
3
 ythe
activated, nucleation and growth v
processes can d be approximated using X(of0the
)the
real volume fraction
the real
recrystallization,
volume d trecrystallized,
fraction then weX (
can
recrystallization,
t ) , with
write dan the
Xequation
( t ) ,
rate
for this
with the
virtual
imaginary
rate of
volu
(virtual)
the virtuv
mi (JMA) equation.
How can Thewesimplest form
get information of the JMA
about
recrystallized, ( tequation
vd
Xthe t) physical
as for the volume
process from n and Q?
dt
ed, X(t), is leading to the general solution
d
transformed, transformed,
Xv( t) , has thed
dt 1  exp  T dG
v ( t ) , hasX
Xspecific
n
form
the 1  exp
( t)specific   t  Xv ( t)
form 
X v ( t)   N   G   t   d 
4
 

t 3 3
X( t )
Noting the result   t
in equation 
[4a]
[4b] and substituting 
 0 3 equation  [4a] for
on of temperature, T, and free energy change, G for the (re)crystallization 
spherical growth and taking the derivative leads to the differential e
ts
me and n is the exponent for the process. An example 
 Nof 4 the form 3of time
recrystallization fraction
d
d (y-axis) predicted X( t)by the JMA 
  G
equation
  t   
s is givend  
1 4
 ts  N  G
X( t3)with X
X( t) dt 
3
X( t )  1 Xv(t )
[4b] 1  X( t )  or
0   d X d 
  bemore
Johnson, Mehl,dtAvrami d d  1 
n  3 [4b]t  0 100 dX
where
1difficult 
))( t)  ofor
marks
XX((ttconcept the time
X (at
t ) 
which
exp a
nucleation
X ( t ) event
d
( XX
occurs.
11X ( t
)(To
t ) )  1
 X
d
Xv( t)
e get information about the process from
doftd
a
the tgrain
v
value would
of n
d t
be
and
v  time, ,1vas
a nucleation
after
Q? a
As 
20an second
example, (t)

heat
opposed
X
X ((0t)) to a clock
treatment if it  X (
nucleated 
0)dt10
time, t, c

dboundary
rical growth of grains with t to the
leading general
velocity,
heat G, solution
nucleating
treatment began.at aIf the
constant
X ( 0
value of
)rate
G isper
 6 m
10  this radius would be gi
v

N. If we pretend that growing grains do not impinge (run X
leading
ber of nucleito the per
formed general solution
unit time
into)
does not depend on the the fraction
 exp and
( t) one1 another Xv (that
t)
s

then we can write an Noting
equationthe result
for this X (dtin
)X
imaginary
(1t 
equation exp
(virtual)
) ( 1 
[4b]
 
volume
G tX
X
and

v( 
(
 
tsubstituting

fraction
t ) )
 leads  
)10 64 m  (320 s equation
t  N  
10 
G s ) 10[4a]

5
 m for the s
to the differential equatio
equation  forthethe
Xv ( t) as spherical growth and taking the derivative s
dt and ofsubstituting 3
Noting the result in equation [4b]
The volume this grain can then [4a]from
be determined specific
radius. What w
total volume fraction transformed is to add up the volumes of all the grain
spherical The   t solution
growth is
and taking various
the derivative leads to theto the differential does theimagina
equation t
 
times, leading integral in eq. [4]. So how
X v ( t)   N   G   t   d us? dThere
4 3 3
is 
d based on Poisson
a theory     )statistics
1 ( 1 which
v 1  expv d  N G   ( t) 
 d4
3X( t)proposes
  
that
4t

 0
[5a]
3
the real
X ( t ) X X ( t ) exp
d t( t) recrystallization,
X ( t

)
the virtu3
 doft 
N
 
d t volume fraction
 dt 3 X ( t ) , with
  40
thet rate


ts
which

is

d
X( t )
equivalent 
d
to Xthe
v ( t )  exp
general
d  X ( t
result ) 
( 1
v the specific form  X ( t ) )  

d 
dt 
N   G 
 N 4   G t   3ddt  1  t 4 Nd
 
transformed, Xv ( t) , has 3
 t G
 3 s [5b] 3 s d d t X( t ) 1 
 exp
4 3  X v ( t ) 

0
X( t) ( 1  X( t) )    t  N  G 0
of thermally activated, nucleation and growth processes can be approximated
Mehl-Avrami (JMA) equation. The simplest
about the form of the JMAfrom equation for the vo
ecrystallized,
How can we
X(t), is
get information physical
 
process t n and Q?

    
d d d  4
X( t)  Xv( t)  exp  Xv( t) ( 1  X( t) )  N   G ts  
dt dt  dt  3
 exp  T  G  t  in  and 
n
energy for the process X( t) is 1implicitly contained 0 may be writte
is a function of temperature, T, and free energy change, G for the (re)crysta
t is the time
d and n is the exponent  4 forthe  
Qprocess. An example of the form of
   N  Gpredicted

3
t) )    t(y-axis)
X( t) ( 1  X(fraction a exp
nt (x-axis) drecrystallization
t  3  R T 
 by the JMA equation is g

onactivation
is energy,
n  3 t  which
0 100 can be a function of the free energy of
 
.000007
X( t) 1  exp  N G   ( t) 
]
n. The JMA equation is useful ifthe 1 exponent,
3 4 n, and the activation
termined. This allows one to examine 3 how the temperature of the h
rols the rate
quivalent to theofgeneral
the process
result without actually having to do all the
b]is also possible to XQ(use  exp
tis) the1the X
values
activation 
v(of
t) Q
energy 
for and
N andnGto evaluate the mech

gion process.
equations [1] To
anddetermine
[2]n with Q and[5]
equation
indicates
n from
allows
the geometry
equations
of the us to relate [1]the
particle--dimension
and
JMA[2], the
= n-1 equatio
s three or four temperatures
with physically needs For
relavant parameters. to be massaged.
example, Notegrowth
if we have that the
of sp
ata is the lhs
a constant of equation
nucleation [1] process
rate, the which we attemptn,towould
exponent, fit to be
theequal
rhs of eq
to 4.
arrange
thermallyequation
activated,[1] by substituting
a simple form is equation [2], subtracting 1 from
each sideby G  take the natural log of
Q-1,  each
QN  side, multiply by -1 on
G b exp  and N c exp 
 The solution is
   
The solution is
 t exp  1  N G13   ( t)34 4
[5a] X ( t ) 
1   
)
[5a]
 4X ( t ) 4 3 3
1  exp  N 
3 
G   ( t ) 
d
X( t) X ( 1( t) X1(t)exp   t3 N  N  
 GG 
 
 t 
3   d  
which is which
equivalent to the general
dt is equivalent 3 result
result
to the general 3  
ion is
[5b] [5b] X( t)  1X0( t)exp 1Xexp  
v ( t) Xv( t)  
Comparing
a] parameters
equations
Comparing [1] and [1]
equations
withXphysically
with physically
parameters
[2] and
( t)relavant
with[2] 
equation
with 1equation
exp parameters.
1 relavant
parameters.  NFor  
[5] allows3 [5] allows

G example, For
us 4to 
( t)example,
relate
us to relate
 if we have
the JMA equatio
the JMA
if we growth
eq
of spo
have growth
grains and a constant
grains nucleation
and a constant rate, the
nucleation rate, 3process
the process exponent, n, would
exponent, be equal
n, would to 4. t
be equal
equivalent
and G are andto Gthe
are general
thermally activated,
thermally result
a simple
activated, form isform is
a simple
5b]  X
G b  expG b exp and and 
Q (Gt) Q1G exp  X ( t)  QNQ N 
N v c exp

N c exp  
R T  R T 
 [2]  R Tus  Rto T 
ng equations [1] and with equation [5] allows relate the JMA eq
then the activation energy, Q, in the JMA equation is
rsthen
with the activation energy, Q, in the JMA equation is
physically
Growth relavant
rate controlled parameters.
by diffusion For  1)example,
nNucleation  QG rate/volume if we controlled
have growth o
Q QNQ ( nQN  1) (Q
d a constantActivationnucleation energy rate, the processG byexponent,
nucleation activation n, would be equal t
energy
It can alsoIt can
be also be
seen seen
that that
this for
forsimple case, this case,
e thermally activated, a form is  (QnN1)(nQG1) QG 
  Q     Q
 QGaexp a exp   ( n  1) (bnn11 )cbexpcQ 
n 1
Q  exp N
  
Q RQ 
NN c exp  G     R T
N
G b exp   R T and  
T ( 3 ) Q R  T

It isalso
toconsider time dependent behavior
R Tpossible  Rbehavior
 Tofparameters
to consider time dependent of parameters within the J
It is also possible within the JMA
For example, QN, can be represented as a function of time if heterogeneous nuc
activation
For example, energy,QN, canQ, be in represented
the JMA equation as a function is of time if heterogeneous nucleatio
have a range Q
have a range of activation QN  energies,
of activation
energies,  1)the
( nwith Q with the lowest activation energy sites being
G
lowestparameter activationisenergy sites being used
Another common case of a time dependent where the temperature
o Another
be seen common
the that case
for thisofcase,
recrystallization a time
process.dependent
In theseparameter cases,  is is awhere
function theoftemperature chan
time,  ( t). The g
the recrystallization process. In these cases,  is a function of time,  ( t). The genera
 4 t 
)  3 3 
d
X( t) ( 1( t) X1(t)exp
X  t3
  3 N
4
N  
GG
  t    d
dt  3   n corresponds
to shape
tion is
 0 

5a] X( t) 1  exp

1  3
 N G   ( t)
4

 3 
equivalent to the general result
 
t

5b]  exp n=4
X( t) 1Sphere  Xv( t)  N 4   G3  t   3 d 
 3

0
ng equations [1] and [2] with equation [5] allows us to relate the JMA eq
ers with physically relavant parameters. For example, if we have growth
nd a constant nucleation rate, the process exponent,t n, would be equal

 N h  2  G2  t   d 
e thermally activated, a simple form is 2
Plate n=3
h 
 G
Q  0N 
Q
G b exp  and N c exp 
 R Tr   R T 
activation energy, Q, in the JMA equation is t

Q QN  ( n Rod
1) n=2  N 2  r  G  t   d 
2
QG

0
so be seen that for this case,
Q  ( n  1)  Q
 So how does this data on n and Q apply to a ‘real’ problem?

Hot drawing of wire

T variation with time (position)

Modify the JMA equation so that we add up the amount of

recrystallization occurring at each place (at each temperature)on the
wire.
  t 
X( t) 1  exp  
d  a exp Q     d
n
    
 d   RT   
 0 
 