NUCLEATION AND INTERFACE KINETICS
HOMOGENEOUS NUCLEATION
When a solid forms within its own melt without aid of foreign materials, it is said to
nucleate homogeneously. Nucleation in this way requires a large driving force because
of the relatively large contribution of surface energy to the total free energy of very
small particles. From Eqs. (8-8) and (8-10), the radius r* of a spherical particle which
is just stable at an undercooling AT is given by
2oTyV,
AH AT
(9-1)
‘The particle of radius r* is termed the critical nucleus. Particles larger than this are
stable and grow; particles smaller than r* form spontaneously in the liquid metal both
above and below the equilibrium melting point. They do so because in this way they
increase the entropy of the system. Following Turnbull,” this is seen simply by viewing
the liquid as an ideal solution of various size clusters, each containing i atoms or
molecules (hereafter termed atoms). Let n equal the number of simple single atoms
(= 1) per unit volume and 1; be the number of clusters of i atoms formed from these
[> euwsteRs-p-muete! —e
FREE ENERGY, AG,
°
RaAIUs —=
FIGURE 9-1
Free-energy barrier to nucleation,
simple atoms. The free-energy change per unit volume of liquid on forming the
clusters AG, is then
AG, =n, AG, — T AS (9-2)
where AG; is the free energy of formation of one cluster of i atoms and AS is the
entropy of mixing nj clusters, which for an ideal solution is simply
= a8 ——K (min A +nin ) (9-3)
Now, at (metastable) equilibrium, 8G,/an; = 0. At this point, let n=, and assume
n> n,. Then, Eq. (9-3) becomes
nh = ne“ @6tAT) (9-4)and assuming Eq. (9-4) is valid for m; = nf,
nf = ne~aouery on
‘The thermodynamic barrier to formation of a critical nucleus AG} decreases rapidly
with increasing undercooling AT. Thus ¥, the number of such critical nuclei, is a
strong function of undercooling. ‘The greater the number of these critical nuclei, the
greater the probability that in a given time one will grow and initiate solidification.
We can now derive a rate law for homogeneous nucleation by making two
simple assumptions. The first is that the concentration ny of critical nuclei remains
that predicted by equilibrium considerations even after nucleation begins. The
second is that addition of a single atom to a critical nucleus makes that nucleus
supercritical and able to grow rapidly. Then, the rate of formation of nuclei per unit
volume J is
Tanto. (9-8)
where w* is the number of atoms surrounding a critical nucleus and vs is the fre-
quency with which atoms jump across the liquid-solid interface. For a spherical
nucleus, «* is approximately given by
Anr*?
y (9-9)
The jump of atoms or molecules in the bulk liquid is?
(9-10)
where 4 is jump distance and D, is liquid diffusion coefficient. The frequency with
which each atom strikes the liquid-solid interface is usually taken to be one-sixth its
jump frequency in the bulk liquid because it reaches the interface by jumping in only
‘one of six possible directions. Taking 2 ~ a, we arrive at the expression for v,5
By
ves ean
Substituting Eqs. (9-6) to (9-9) with the foregoing, we arrive finally at the classical
expression for homogeneous nucleation in bulk liquids. This equation is essentially
equivatent to that derived by Turnbull and Fisher? based on earlier work of nucleation
of droplets from the vapor. It is written
Di 620° Ty? V2
T= B, P+ exp| — 1620 Ta" ¥e— 12)
Bee sarrarir] C1
where the preexponential term B, depends on critical nucleus size and surface energy,
and Dry is the liquid diffusion coefficient at the equilibrium melting point Ty. For
liquid metals, D,/Dzy ~ 1 and the preexponential B, may be taken as a constant of
about 10** (Holloman and Turnbull*),@| é
5 5
gar gear
3 3
3 3
‘a rs
FIGURE 9-2
Characteristic shapes for nucleation rate curves. AT. is the critical undercooling
for a sensible nucleation rate. (a) Nonviscous liquids such as metals; (6) viscous
nonmetallics including oxides and organic polymers.
The nucleation rate is so sensitive to the term within the exponential that
changes in B, of several orders of magnitude do not appreciably affect the calculated
undercoolings for sensible nucleation, For materials whose liquid diffusivity drops
markedly with temperature, e.g., glasses and polymers, the preexponential term
D,/Dzy, becomes overriding at lower temperatures. As temperature is lowered from
the equilibrium transformation temperature, nucleation rate first increases from zero
to a maximum at some given undercooling. Then as the diffusivity D, becomes very
small, nucleation rate decreases toward zero again as temperature is further decreased.
Figure 9-2 shows schematically the nucleation behavior of a metal as compared with
that of a viscous material such as a polymer or oxide glass.
Many refinements and modifications of the classical nucleation theory given
here have been proposed, but that described above adequately explains most experi-
mental results thus far obtained. The most fruitful experimental approach employed
to confirm the theory is that of solidification studies on finely divided droplets. Two
methods have been employed: microscopical observation of small drops and thermal
analysis on aggregates of drops separated by thin films. In qualitative agreement
with the prediction of Eq. (9-12), these studies show that a given droplet supercool
without nucleating until it reaches a given critical temperature AT. after which
nucleation occurs rapidly. The droplet can be held at undercoolings less than AT,
for very long times without nucleating.
Quantitatively, results of Turnbull and coworkers!’ and of most studies
since theirs have shown that for metals the maximum undercooling achieved is about
0.18 of the absolute melting point (Table 9-1). Taking this undercooling as represent-
ing the undercooling required for homogeneous nucleation, Turnbull calculated
[from Eq. (9-12)] the liquid-solid surface energies o. These surface energies are of the‘Thus, a liquid at temperature either above or below its melting point will have a
distribution of solidlike clusters given by Eq. (9-4) at equilibrium. The free energy of
formation of each of these clusters AG, is
AG
AG, =cA+V
(9-5)
where A is the surface area and V is the volume of the cluster; AG is the thermo-
dynamic driving force for solidification as given by Eq. (8-3), and V, is the molar
volume. The form of this curve is as drawn in Fig. 9-1. AG, increases to a maximum
AG} at r =r", the critical radius; thereafter it decreases rapidly. For a spherical
cluster, substituting Eqs. (8-3) and (9-1) in Eq. (9-5) yields
AGT
ae O° Ty? V2
9-6)
AH? AT? sine