Dendritic Growth: 17.1 Formation of Dendrite
Dendritic Growth: 17.1 Formation of Dendrite
\
|
\
|
=
z
T
K
z
T
K
L
x v
L
L
S
s
S
1
(17.2)
( )
( )
|
\
|
=
z
c
k c
D
x v
L
1
*
(17.3)
where L is the heat of fusion.
S
is the density of solid metal. K
S
and K
L
are heat
conductivity of solid and liquid metals, respectively. T
S
and T
L
are temperature at
solid and liquid side of interface, respectively. c
L
* is equilibrium composition in
liquid. k is the equilibrium partition ratio which is given by k= c
S
*/c
L
*. D is the
solute diffusivity.
Graduate Institute of Ferrous Technology, POSTECH
Rongshan Qin (R. S. Qin)
The average moving rate of the solid-liquid interface, which is the average growth
rate of the solid, is an average of v(x) in one wavelength.
( )
0
1
dx x v V (17.4)
In a coordinate frame that moves with velocity V and with the origin at z=0, the
steady-state equations of the solute composition and temperature are
( )( ) 0 / /
2
= + z c D V c (17.5)
( )( ) 0 / /
2
= + z T V T
L
(17.6)
( )( ) 0 / /
2
= + z T V T
S
(17.7)
where
L
and
S
are thermal diffusivity of liquid and solid metals. For the
geometry of sinusoidal ripple, we suppose that the temperature and
concentration at the interface takes the expressions of
( ) ( )
a T x t a T T + = + =
0 0
sin (17.8)
( ) ( )
b c x t b c c + = + =
0 0
sin (17.9)
where T
0
and c
0
are the values for a flat interface and the second terms are the
first-order corrections corresponding to the infinitesimal ripple. By reference of
(6.15), we know that the steady state solution of equations (17.5)-(17.7) for flat
interface is in a format of
( )z D V
Be A
/
+ . We, therefore, assume that the solutions
of equations (17.5)-(17.7) in sinusoidal ripple interface are
( ) ( )
( )
( ) ( ) ( )
0
* /
sin 1 / , c e x G b e V D G z x c
z w
c
z D V
c
+ + =
(17.10)
( ) ( )
( )
( ) ( ) ( )
0
* /
sin 1 / , T e x G A e V G z x T
z w
c
z V
S S S
S S
+ + =
(17.11)
( ) ( )
( )
( ) ( ) ( )
0
* /
sin 1 / , T e x G A e V G z x T
z w
L
z V
L L L
L L
+ + =
(17.12)
Graduate Institute of Ferrous Technology, POSTECH
Rongshan Qin (R. S. Qin)
where G
S
and G
L
are thermal gradients at unperturbed flat interface in the solid
and liquid, respectively. G
c
is the concentration gradient in the liquid at the
unperturbed flat interface. Insulting (17.10) into (17.5) we have
0 * *
2 2
= w w
D
V
w (17.13)
which means that (17.5) will be satisfied by the solution of (17.10) when
2
2
2 2
* +
|
\
|
+ =
D
V
D
V
(17.14)
Similarly, we have
2
2
2 2
*
L
L L
L
V V
+
|
|
\
|
+ = (17.15)
and
2
2
2 2
*
S
S S
S
V V
+
|
|
\
|
+ = (17.16)
Equation (17.10) gives gradient of the diffusion fields at the interface to the first
order in
( ) ( )
( ) ( )
( )
( ) [ ] { } ( )
c c
x w
c
x D V
c
G x D V G b w
e x G b w e G
z
c
+
=
|
\
|
sin * / 1 *
sin *
sin * sin /
(17.17)
Similarly, (17.11) and (17.12) gives
( ) [ ] { } ( )
L L L L L
L
G x V G a w
z
T
+
|
\
|
sin * / 1 * (17.18)
and ( ) [ ] { } ( )
S S S S S
S
G x V G a w
z
T
+
|
\
|
sin * / 1 * (17.19)
Graduate Institute of Ferrous Technology, POSTECH
Rongshan Qin (R. S. Qin)
The solution for temperature should also satisfy the Gibbs-Thomson effect on the
curvature contribution. In dilute binary system, the Gibbs-Thomson effect gives
the temperature at curved interface as the format of following.
( ) x T T mc
x
z
T T mc K T T mc T
m m m m m m
sin
2
2
2
+ =
|
|
\
|
+ + = + + = (17.20)
where m is the slope of the liquidus at z=. T
m
is the melting temperature of pure
materials where c=0. is the capillary length.
Bring equations (17.17)-(17.19) into (17.2) and (17.3) and using the boundaty
condition of (17.20), it has
( ) ( ) [ ]( )
( ) ( ) [ ] p D V mG
D V G G T G
b
L S c
L S c L S c m c
/ * 2
/ * 2
3
+
+ + +
=
(17.21)
and
2
=
m
T mb a (17.22)
where ( ) [ ]
S L S S S
G K K K + = 2 / , ( ) [ ]
L L S L L
G K K K + = 2 / and p=1-k. Finally we
have
( ) [ ] ( ) ( ) [ ] ( ) [ ] { }
( ) ( ) [ ] p D V mG
D V mG p D V p D V T V
L S c
c L S m
/ * 2
/ * 2 / * / * 2
2
+
+ +
=
&
(17.23)
As mentioned before, the interface stability is determined by
&
. 0 >
&
implies the
instability of the interface and dendrite will grow.
17.3 Instability analysis
For to discuss equation (17.23), let us first see the phase diagram and
temperature/solute concentration distribution profiles in solidification. The phase
diagram of dilute alloy is illustrated in figure 3a. The corresponding solute
concentration profile is demonstrated in figure 3b. It can be see that the slope of
the liquidus and the gradient of solute concentration always has the same sign.
Graduate Institute of Ferrous Technology, POSTECH
Rongshan Qin (R. S. Qin)
This means that the first term in the denominator of equation (17.23) is always
positive. Figure 3c shows the temperature profile in the solidifying alloys without
liquid supercooling. Figure 3d shows the temperature profile in the solidifying
alloys with supercooling. Both figure 3c and figure 3d shows that 0 >
L S
.
From equation (17.14),
2
2
2 2
* +
|
\
|
+ =
D
V
D
V
, one has 0 * >
D
V
.
Furthermore, one has 0 * > p
D
V
because of 0<p<1. All of a word, the
denominator of equation (17.23) is positive.
Figure 3
The numerator of equation (17.23) can be expressed as
( ) [ ] ( ) S p D V w V / * 2
We know that ( ) [ ] 0 / * 2 > p D V w V . The question of whether
&
>0 turns into the
question of whether S()>0.
( ) ( )
( ) [ ]
( ) [ ] p D V
D V
mG T S
c L S m
/ *
/ *
2
2
+ + =
(17.24)
Graduate Institute of Ferrous Technology, POSTECH
Rongshan Qin (R. S. Qin)
S(() contains three terms. The first term is related to interface energy. The first term is
always negative because the interfacial energy is always in positive. The larger interface
energy possesses more stable interface. The second term is about the gradient of
temperature around interface. The positive gradient of temperature around interface helps
to maintain interface stable. The negative temperature gradient will cause the instability
of the solid-liquid interface. The third term is about the effect of solute diffusion on the
interface stability. When V/D<<1, the third term approaches to
c
mG 2 . If we ignore the
contribution from the interface energy, S()<0 will be satisfied when
( )
c L S
mG > +
2
1
(17.25)
References
1. W.W. Mullins and R.F. Sekerka, J. Appl. Phys., 35 (1964) 444-451.
2. W.W. Mullins and R.F. Sekerka, J. Appl. Phys., 34 (1963) 323-329.