This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy

Physics Letters A 376 (2012) 35633566

Contents lists available at SciVerse ScienceDirect

Physics Letters A
www.elsevier.com/locate/pla

Rapid colloidal solidications under local nonequilibrium diffusion

conditions
S.L. Sobolev
Institute of Problems of Chemical Physics, Academy of Sciences of Russia, Chernogolovka, Moscow Region 142432, Russia

a r t i c l e

i n f o

Article history:
Received 12 August 2012
Received in revised form 18 October 2012
Accepted 19 October 2012
Available online 24 October 2012
Communicated by C.R. Doering
Keywords:
Rapid solidication
Colloidal suspensions
Diffusionless solidication
Local nonequilibrium diffusion
Segregation

a b s t r a c t
Partition coecient for rapid solidication of colloidal suspensions has been calculated under local
nonequilibrium diffusion conditions typically used when processing advanced materials. It has been
demonstrated that the local nonequilibrium diffusion effects stabilize the planar solid liquid interface
and lead to an abrupt transition to diffusionless solidication with complete particle trapping. The
effective diffusion coecient, which depends on interface velocity and particle size, has been introduced.
It explains the strong dependences of the partition coecient and the velocity leading to absolute
stability of a planar solidliquid interface on particle size.
2012 Elsevier B.V. All rights reserved.

In materials science, the rapid solidication of colloidal or

nanosuspensions is receiving increasing interest owing to the novel
microaligned structures that can be produced for a wide range
of applications such as organic electronics, microuidics, molecular ltration, nanowires, cryopreservation and tissue engineering [18]. Colloidal particles are often seen as big atoms that can
be directly observed in real space. The rapid solidication of colloidal systems is therefore of scientic interest as model systems to
study meltingfreezing processes under local nonequilibrium conditions. In order to understand and control the rapid solidication
of colloidal suspensions, a predictive and general theory is required
for the number of particles trapped by an advancing solidication
front as a function of front velocity and particle radius. In the recent work of Elliott and Peppin [1] used the Boltzmann velocity
distribution of a particle to derive an expression for the nonequilibrium segregation coecient K ( V ) at an ice-colloidal suspension
interface as a function of the freezing interface velocity V and particle radius R.
Substantial progress has been realized over the past decades
in the modeling of solute trapping in rapid solidication of binary alloys on the bases of thermodynamic theory [3], molecular
dynamics [9], and Monte Carlo simulations [10,11]. The analogy
between the rapid solidication of binary alloys and the colloidal
systems will allow us to use the thermodynamical approach devel-

Tel.: +79032478152; fax: +74965223507.

E-mail address: SSL55@yandex.ru.

0375-9601/$ see front matter 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.physleta.2012.10.031

oped for the rapid solidication of binary alloys for analyzing the
rapid solidication of the colloidal systems.
The work extends local nonequilibrium diffusion model (LNDM)
of solute trapping during rapid solidication of binary alloy [1315]
by accounting for several specicities of colloidal systems, including the variation of the particle diffusivity with the particle
size and interface velocity. LNDM is based on hyperbolic diffusion
equation, which has been successfully used for analyzing different
kinds of local nonequilibrium phenomena [16]. We focus on relatively high interface velocity of the order of 10100 m/s typical
used when processing materials through solidication of colloidal
suspensions [2]. The system in these conditions is far from equilibrium and the local nonequilibrium diffusion effects [1315] should
be taken into account. We can tackle the problem considering the
colloids as big atoms [1,2], applying Brownian diffusion mechanisms and modied Ficks law [1316], which accounts for deviation from local equilibrium of diffusion process in the bulk liquid
ahead of a fast moving interface [1315]. The diffusion coecient
of the particles D ( R ) can be estimated using the StokesEinstein
diffusivity of an isolated colloidal particle [2]

D(R) =

kB T
6 R

(1)

where R is the particle radius, is the dynamic viscosity, k B is the

Boltzmann constant.
The modied Ficks law [16] results in hyperbolic diffusion
equation for the particle concentration ahead of the interface
C ( X ) [1315]

Author's personal copy

3564

S.L. Sobolev / Physics Letters A 376 (2012) 35633566

C
2C
2C
+ 2 = D 2
(2)
t
t
X
here is the relaxation time to local equilibrium. Eq. (2) predicts
a nite speed of the diffusive wave V D = ( D / )1/2 i.e. maximum
speed with which the diffusion perturbations can propagate in
the bulk liquid [1316]. From microscopic point of view the diffusive speed V D can be treated as the maximum speed with which
an individual particle moves in X -direction in the bulk liquid,
i.e. the molecular velocity of the particles V D = (3k B T /m p )1/2 ,
where m p is the mass of a particle, k B is Boltzmanns constant,
and T is temperature. In reality there may be asperities on the
particles surface leading to smaller value for V D but as a rst order approximation it is reasonable to assume that the particles are
spherical. As it will be demonstrated later, this assumption leads
to theoretical prediction consistent with experimental result. Thus,
after some algebra the diffusive speed for colloidal particles takes
the form

V D ( R ) = 9k B T /4 R 3

1/2

= R 3/2

(3)

where is the particle density, = (9k B T /4 ) .

In the moving reference frame for steady-state regimes solution
to Eq. (2) takes the form [13,14]
1/ 2

(C i C 0 ) exp[ V X / D (1 V 2 / V D2 )] + C 0 ;
LNDM
C
(X) =
V < VD

C0; V > V D

(4)

where C 0 and C i are the particle concentration in the liquid

far from ( X ) and at the interface ( X = 0), respectively.
The comparison of solute concentration given by LNDM, Eq. (4),
with solute concentration predicted by classical parabolic diffusion
equation, allows us to introduce the effective (velocity dependent)
diffusion coecient in the bulk liquid D LNDM ( V ) [1315] as

D LNDM ( V ) =

D (1 ( V / V D )2 );
0;

V < VD
V > VD

(5)

The local nonequilibrium solute concentration prole C ( X ), Eq. (4),

implies that when the interface velocity V passes through the
critical point V = V D , an abrupt transition from diffusion controlled to diffusionless (kinetic controlled) solidication occurs.
When V < V D , there is solute diffusion ahead of the interface and
the solidication is mainly diffusion controlled. If V > V D , there is
no diffusion ahead of the interface with C ( X ) C 0 and the growth
is kinetic controlled [1315]. In this case the solidliquid interface
does not reject solute molecules or particles and they remain in
their initial positions and are trapped immediately upon arrival of
the interface. Combining (3) and (5), the effective diffusion coecient for the colloidal growth takes the form

D LNDM ( V , R ) =

k

B T D
(1 ( V / V D ( R ))2 );
6 R

0;

V < V D (R)
V > V D (R)

(5a)

The effective diffusion coecient D LNDM ( V , R ), Eq. (5a), takes into

account both the local nonequilibrium diffusion effects and particle
size effects, i.e. the dependency of R. The transition to diffusionless solidication during rapid solidication of colloidal particles is
analogues to that of binary metallic alloys. Specicities of colloidal
systems are the dependency on particle size of the transition velocity V D ( R ) and its low value of the order of 10100 m/s [2] in
comparison with V D 130 m/s for metallic alloys [15]. The low
critical velocity makes colloidal suspensions increasingly important
as model systems to study nonequilibrium processes in condensedmatter physics such as melting, freezing and glass transitions.
The degree of particles trapping is usually quantied by the partition coecient K , dened as a ratio of the particle concentration
in the solid to that in the liquid at the interface. The most resent

Fig. 1. Partition coecient K LNDM as a function of interface velocity V (m/s) for

different particle radii R = 20 m, 6 m, and 2 m. Solid lines K E = 104 , dash
lines K E = 106 .

theory of Jackson et al. [10,11] for nonequilibrium segregation during rapid rst-order phase transformations is based on the idea
that the interface can transfer, between components, the chemical potential differences which are driving the transformation. The
partition coecient can be extended for the local nonequilibrium
diffusion case based on insights gained from hyperbolic LNDM,
Eqs. (2), (4), and (5). Following Jackson et al. [10,11] we introduce
the trapping probability as the ratio of the distance that the interface advances during time t to the distance that the particle
can diffuse during this time. Under local nonequilibrium condition,
i.e. on a short time scale, a particle propagates with molecular
velocity V D and, hence, the trapping probability LNDM is dened
as LNDM = V /( V D V ) at V < V D and LNDM at V < V D .
In this case the modied partition coecient takes the form

K LNDM ( V , R ) =

1/(1+ A V /( V D ( R ) V ))

KE
1;

; V < V D (R)
V > V D (R)

(6)

where K E is the equilibrium partition coecient, A is a constant governing energy transfer between species [10,11]. K LNDM
as a function of V is plotted in Fig. 1 for different R and K E .
At low interface velocity V 0, i.e. near equilibrium, K LNDM
K E , which value is extremely small for colloidal size particles [1].
As the interface velocity increases above the characteristic velocity V = V D ( R )/(1 + A ), the partition coecient deviates strongly
from K E and results in complete particle trapping K LNDM = 1 at
V  V D ( R ). It should be stressed that the critical point at which
the complete particle trapping occurs V = V D ( R ) R 3/2 strongly
depends on the particle radius R. Increasing the particle radius
decreases the critical velocity and shifts the transition to diffusionless solidication with complete particle trapping to lower velocity
range while the dependency of K LNDM on K E is relative weak (see
Fig. 1). The possibility to reach the complete trapping during rapid
solidication at a nite interface velocity is conrmed by experimental results [2,58] (see also references in [15]) and molecular
dynamic simulations [9].
The particles rejected by the interface create a concentration
gradient in the liquid, Eq. (4), that causes the liquid ahead of the
interface to be undercooled. This is the basis of the constitutional
undercooling which leads to instability of a planer solidliquid
interface [12]. At high interface velocity V > V D ( R ) the growth
occurs in the diffusionless regime C ( X ) C 0 and the concentration gradient in the liquid ahead of the interface is zero. It implies
an absence of constitutional undercooling at V > V D ( R ), which,

Author's personal copy

S.L. Sobolev / Physics Letters A 376 (2012) 35633566

Fig. 2. Critical concentration C a as a function of interface velocity V (m/s) at

R = 2 m. Solid curve LNDM, dash curve quasi-equilibrium approach [12]. Insert
shows C a vs. V for different particle radii R = 20 m, 6 m, and 2 m.

in turn, leads to a stable planar solidliquid interface. Thus the local nonequilibrium diffusion effects play a stabilizing role in rapid
solidication of colloidal particles as well as in rapid solidication
of binary metallic alloys [14]. Mullins and Sekerka [12], assuming
equilibrium at the interface and in the bulk liquid, showed that
the stabilizing effect due to the interface energy (GibbsThomson
effect) leads to absolute stability of the solidication front if the
solidication velocity approaches the critical value V a

Ca =

K E2 T m V a

( K E 1)mD

(7)

where C a is the critical bulk concentration in the liquid, T m is the

melting point of pure solvent, is the capillarity constant, m is
the slope of the equilibrium liquidus line. From (1) and (7) one
can conclude that at a given particle concentration the velocity
leading to a transition between the stability regimes V a is proportional to D R 1 . This is valid for quasi-equilibrium regimes
when V a  V D . Far from local equilibrium the MullinsSekerka
theory should be modied using velocity- and radius-dependent
K LNDM ( V , R ), Eq. (6), D LNDM ( V , R ), Eq. (5a), and mLNDM ( K LNDM )
(m vs. K dependence can be found in [1,14]). The LNDM version of
Eq. (7) is shown in Fig. 2 as solid line. The area under the curves is
a stable planar interface region, above the curve unstable. Fig. 2
clearly demonstrates that the local nonequilibrium diffusion effects
as well as increasing the particle radius (see insert) play a stabilizing role (compare areas under solid and dash curves). For any
value of critical concentration C a at a given particle radius R holds
inequality V aLNDM < V D , where V aLNDM is the velocity of absolute
stability limit with allowance for local nonequilibrium diffusion effects (solid curve). When V aLNDM < V < V D , the planar interface is
stable and there is particle diffusion ahead of the interface and particle segregation at the interface with K LNDM < 1. At higher interface velocity V > V D , the diffusionless solidication with a stable
planar interface and complete particle trapping K LNDM = 1 occurs.
Fig. 3 shows the stability diagram for solidication of colloidal
particles. The solid curve represents critical velocity for absolute
stability limit based on LNDM version of Eq. (7) as a function of
particle radius. The diffusive velocity V D ( R ), leading to the transition from diffusion controlled to diffusionless solidication with
complete particle trapping, is shown as the dashed curve. The
hatched area is an estimation of the domain in which the transition to unstable regimes occurs during colloidal solidication of
alumina particles investigated in [2]. The area above solid curve

3565

Fig. 3. Stability diagram for the solidication of colloidal suspension: dash curve
critical velocity V D (m) leading to a transition to complete particle trapping,
Eq. (3); solid curve critical velocity for absolute stability limit based on LNDM.
The hatched area is an estimation of the domain in which the transition to unstable regimes occurs during colloidal solidication of alumina particles investigated
in [2].

is a stable planar interface region. Under the curve a planar interface is unstable and the cellulardendritic and banded structures can form. Between solid and dash curves solidication is
mainly diffusion controlled with particle segregation at the interface, above dash curve diffusionless solidication with complete
particle trapping K LNDM = 1 occurs. The predictions are made with
no tting parameters and are in good agreement with experimental results [2,58], capturing the decrease in the transition velocity
to unstable regimes V aLNDM and the transition velocity to complete
particle trapping V D with the particle radius R.
In summary, by considering local nonequilibrium diffusion effects we have predicted a sharp transition from diffusion controlled to diffusionless solidication with complete particle trapping when the interface velocity passes through the critical point
V = V D ( R ). The transition velocity strongly depends on particle radius as R 3/2 , Eq. (3). We have obtained an expression for
the partition coecient, Eq. (6), which describes particle segregation at the solidliquid interface as a function of interface velocity and particle size. The local nonequilibrium diffusion effects
as well as increasing the particle radius substantially stabilize the
planar interface against cellular breakdown. The velocity leading
to a transition between the stabilityinstability domains V aLNDM
also strongly depends on particle radius. At low interface velocity V  V D , i.e. at quasi-equilibrium, the Mullins and Sekerka
approach leads to V a D R 1 . Under local nonequilibrium diffusion condition, i.e. at high interface velocity V V D , LNDM
predicts that V aLNDM ( R ) < V D ( R ) and V aLNDM ( R ) R n with n
between 1 and 1.5. The predictions are consistent with experiments [2,58]. Future work determining the interface temperature
and dendrite tip radius as a function of interface velocity and particle size will allow us to predict in more details the structure and
morphology of materials processed by rapid solidication of colloidal suspension under local nonequilibrium conditions.
References
[1] J.A.W. Elliott, S.S.L. Peppin, Phys. Rev. Lett. 107 (2011) 168301.
[2] S. Deville, E. Maire, G. Bernard-Granger, A. Lasalle, A. Bogner, C. Gauthier,
J. Leloup, C. Guizard, Nature Mater. 8 (2009) 966.
[3] M. Asta, C. Beckermann, A. Karma, W. Kurz, R. Napolitano, M. Plapp, G. Purdy,
M. Rappaz, R. Trivedi, Acta Mater. 57 (2009) 941.

Author's personal copy

3566

S.L. Sobolev / Physics Letters A 376 (2012) 35633566

[4] S.S.L. Peppin, J.S. Wettlaufer, M.G. Worster, Phys. Rev. Lett. 100 (2008)
238301.
[5] G. Lipp, C. Korber, J. Cryst. Growth 130 (1993) 475.
[6] G. Lipp, C. Korber, G. Rau, J. Cryst. Growth 99 (1990) 206.
[7] C. Korber, G. Rau, M. Cosman, E. Cravalho, J. Cryst. Growth 72 (1985) 649.
[8] S.N. Omenyi, A.W. Neumann, J. Appl. Phys. 47 (1976) 3956.
[9] Y. Yang, H. Humadi, D. Buta, B.B. Laird, D. Sun, J.J. Hoyt, M. Asta, Phys. Rev.
Lett. 107 (2011) 025505.

[10]
[11]
[12]
[13]
[14]
[15]
[16]

K.A. Jackson, K.M. Beatty, K.A. Gudgel, J. Cryst. Growth 271 (2004) 481.
K.M. Beatty, K.A. Jackson, J. Cryst. Growth 271 (2004) 495.
W.W. Mullins, R.F. Sekerka, J. Appl. Phys. 35 (1964) 444.
S.L. Sobolev, Phys. Lett. A 199 (1995) 383.
S.L. Sobolev, Phys. Rev. E 55 (1997) 6845.
S.L. Sobolev, Acta Mater. 60 (2012) 2711.
D. Jou, J. Casas-Vazquez, G. Lebon, Extended Irreversible Thermodynamics, third
ed., Springer, Berlin, 2001.