0% found this document useful (0 votes)
28 views19 pages

10 1108 - Ec 03 2018 0153

Uploaded by

rs21merer01
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
28 views19 pages

10 1108 - Ec 03 2018 0153

Uploaded by

rs21merer01
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/0264-4401.htm

EC
36,3 Nanoparticles migration near
liquid-liquid interfaces using
diffuse interface model
1036 Ali Daher
Angevin Laboratory of Mechanics, Processes and Innovation,
Received 30 March 2018 Arts et Métiers ParisTech d’Angers, Angers, France and Physics Department,
Revised 22 September 2018
22 November 2018 Faculty of Sciences, Lebanese University, Beirut, Lebanon
Accepted 11 December 2018
Amine Ammar
Angevin Laboratory of Mechanics, Processes and Innovation,
Arts et Métiers ParisTech d’Angers, Angers, France, and
Abbas Hijazi
Physics Department, Faculty of Sciences, Lebanese University, Beirut, Lebanon

Abstract
Purpose – The purpose of this paper is to develop a numerical model for the simulation of the dynamics of
nanoparticles (NPs) at liquid–liquid interfaces. Two cases have been studied, NPs smaller than the interfacial
thickness, and NPs greater than the interfacial thickness.
Design/methodology/approach – The model is based on the molecular dynamics (MD) simulation in
addition to phase field (PF) method, through which the discrete model of particles motion is superimposed on
the continuum model of fluids which is a new ide a in numerical modeling. The liquid–liquid interface is
modeled using the diffuse interface model.
Findings – For NPs smaller than the interfacial thickness, the results obtained show that the concentration
gradient of one fluid in the other gives rise to a hydrodynamic drag force that drives the NPs to agglomerate
at the interface. Whereas, for spherical NPs greater than the interfacial thickness, the results show that such
NPs oscillate at the interface which agrees with some experimental studies.
Practical implications – The results are important in the field of numerical modeling, especially that the
model is general and can be used to study different systems. This will be of great interest in the field of
studying the behavior of NPs inside fluids and near interfaces, which enters in many industrial applications.
Originality/value – The idea of superimposing the molecular dynamic method on the PF method is a new
idea in numerical modeling.
Keywords Numerical simulation, Nanoparticles, Diffuse interface, Liquid-liquid interface,
Molecular dynamics, Phase field method
Paper type Research paper

1. Introduction
Particles and macromolecules in fluids, near surfaces and at interfaces are related to many
physical and biological applications. Many researchers have addressed different aspects of
the dynamics of particles suspended in a flowing solution (Mackaplow and Shaqfeh, 1998;
Engineering Computations
Vol. 36 No. 3, 2019
pp. 1036-1054
© Emerald Publishing Limited The authors A. DAHER and A. HIJAZI would like to thank the Lebanese University for the Grant
0264-4401
DOI 10.1108/EC-03-2018-0153 (2017).
Butler and Shaqfeh, 2002; Yamamoto and Shaqfeh, 1996; Altenbach et al., 2007; Moses et al., Nanoparticles
2001). In addition, the behavior of particles and macromolecules at ideally flat surface migration
boundaries (Hijazi et al., 2003), and uneven solid surface boundaries (Hijazi and Khater,
2008) have been investigated. Besides that, the rheology of colloidal dispersions was studied
by Grmela et al. (2014), and colloidal suspensions of structureless particles were investigated
by Maitrejean et al. (2012).
As their appearance, nanoparticles have attracted significant interest as promising
candidates for various applications in the field of materials nanotechnology. They are used 1037
in the production of cosmetics, surface coating and drugs (Shukla et al., 2007; Wittstock
et al., 2010; Gewin, 2006) in addition to many other applications.
NPs at liquid–liquid interface arise in many industrial processes such as in the
stabilization of emulsions and foams, in addition to nanoparticle-armored fluid droplets
(Cauvin et al., 2005), phase-arrested gels (Clegg, 2008) and crude oil extraction.
The adsorption of nanoparticles at liquid interfaces is a new field of study during the
past decade, but unfortunately their behavior at such interfaces is not understood well yet.
In this work, we present a numerical method to study the dynamic behavior of nanoparticles
near a liquid–liquid interface, based on the diffuse interface model.
The topology of interfaces between partially miscible or nominally immiscible fluids is a
fundamental phenomenon in fluid dynamics. Classical models of fluid-fluid systems, which
consider the interfacial region between two fluids as a sharp, dividing surface, fail to
describe the transition that occurs when the interfaces merge and reconnect.
The diffuse-interface modeling is one of the phase field models (PFMs) which replace the
sharp fluid interface by thin but nonzero thickness transition regions, where the interfacial
forces are smoothly distributed. Cahn and Hilliard (1958) developed an approach that is well
suited to model such systems through the coupling of thermodynamic and hydrodynamic
forces.
PFMs have been widely used for investigating complex multiphase systems behavior
such as phase separation under shear near-critical interfacial phenomena and microstructure
evolution during solidification. The capabilities of the diffuse-interface model were discussed
in a previous work by Wise et al. (2011) where they have shown the efficiency of this model in
simulating the progression of tumors with complex morphologies by simulating the tumor
growth in two and three dimensions. The evolution of solid particles segregated at fluid–fluid
interfaces was studied through introducing a novel mesoscale simulation approach as can be
seen by the work of Millett and Wang (2011).
To solve the coupled Cahn–Hilliard/Navier–Stokes system, Badalassi et al. (2003)
presented an accurate and efficient numerical method known as Model H, which constitutes
a PFM for density-matched binary fluids with variable mobility and viscosity. And the case
of fluids with different densities was studied in a detailed manner by Lowengrub and
Truskinovsky (1998). Furthermore, Ding et al. (2007) investigated the applicability of an
incompressible diffuse interface model for two-phase incompressible fluid flows with large
viscosity and density contrasts. Recently, Choi and Anderson (2012) presented a numerical
method for the dynamic of a particle in two-phase flow based on the Cahn–Hilliard theory.
In another research, Choi and Djilali (2016) studied the agglomeration of circular colloidal
particles in two-dimensional shear flow, and they showed that external shear and particle
fraction have a direct impact on the structure formation of colloidal particles in a
suspension.
In this paper, we present a numerical method for the dynamics of nanoparticles near a
liquid–liquid interface by coupling the particles and fluid motion, so that we superimpose
the discrete model of particles motion, on the continuum model of fluids which is a new idea
EC in computational physics. First we consider the case of nanoparticles smaller than the
36,3 interfacial thickness, and then we studied the case of nanoparticles greater than the
interfacial thickness. For our knowledge, there are no previous studies that differentiate
between these two cases. The content of this paper is as follows. In Section 2, we give a brief
mathematical description for the diffuse interface model, based on the Cahn–Hilliard theory,
and then we present the intermolecular interactions between multiple particles, in addition
1038 to the fluid-particles interactions. In Section 3, the numerical algorithm is discussed, and
numerical results for the dynamics of nanoparticles at liquid–liquid interface are presented
in Section 4 where we study the effect of different parameters on the behavior of the system.
Finally, a conclusion follows in Section 5.

2. Mathematical formulatiom
2.1 Diffuse interface model
PFM is a particular class of diffuse-interface models through which the state of the system at
any given time is described by an order parameter U which is a function of the position vector.
In this work, we consider the flow of two incompressible immiscible fluids (A and B) of
different densities and viscosities.
In Figure 1, the blue color represents the first liquid (A) of density r 1 and viscosity h 1,
and the red color represents the second fluid (B) of density r 2 and viscosity h 2, and the
transition layer between the two represents the interface. Note that our simulation box is
bounded between 0 and 1 along the two axes as shown by the limiting lines in the figure, but
we present periodic repetition of this box in Figure 1 for clarity, and this will be done in all
our 2D figures in the paper.
Let c be the mass fraction of one of the fluids in a specific domain element. If the mass
fraction of component A is denoted by c, then the local densities of the species A and B in a
domain element are given by:
0
r A ¼ cr A (1)
and
0
r B ¼ cr B (2)

Figure 1.
Two incompressible
immiscible fluids
where r A and r B are the bulk densities. Then the local averaged density at an arbitrary Nanoparticles
domain element will be denoted by: migration
r ¼ c r A þ ð 1  cÞ r B (3)

We begin the analysis of the conservation of mass of species A in an arbitrary element fixed
in space.
The corresponding continuity equation is written as: 1039
@ r 0A
þ r  nA ¼ 0 (4)
@t

with nA is the mass flow rate. 0


In the bulk domain, nA ¼ r A u where u is the velocity of the fluid flow. Whereas, in the
interfacial region the total mass flux should be modified due to diffusion.
The total mass flow rate of species A can be expressed by:
0
nA ¼ r A u  r A jA (5)

The second term on the right side of equation (5) contributes for the diffusive mass flow rate
of component A, with jA represents a volume diffuse flow rate.
In the diffuse interface model, the convective Cahn–Hilliard equation is given by:

@c
þ u  rc  r:jA ¼ 0 (6)
@t

The Cahn – Hilliard theory (Cahn and Hilliard, 1959) assumes that the driving force for
diffusion is the gradient of the chemical potential, and thus the above equation is generally
written as:

@c
þ u:rc ¼ r  ð M r m Þ (7)
@t

With M is the mobility, and m is the chemical potential obtained from the variational
derivative of the free energy with respect to the mass fraction as shown below:

1 1 1
f ðc; rcÞ ¼  ac2 þ b c4 þ « jrcj2 (8)
2 4 2

where « is the gradient energy parameter, and a and b are positive constants:

df
m¼ ¼ ac þ b c3  « r2 c (9)
dc
The fluid dynamics are described by the Navier–Stokes (NS) equations with a phase field-
dependent surface force (Choi and Anderson, 2012):
 
@u
r þ u  ru ¼  r rg þ r  h ðru þ ruT Þ þ r m rc (10)
@c
EC We consider the continuity equation for incompressible fluid:
36,3 ru¼0 (11)
The generalized chemical potential defined in equation (9), allows for the description of the
interface between the two materials by a continuously varying concentration profile. In
equation (7), the diffusion flux is assumed to be proportional to the gradient of the chemical
1040 potential. The physical mechanism that produce the Cahn–Hilliard equation can be seen as a
negative diffusion, and thus this fourth-order partial differential equation that depends on
the diffusion phenomena allows for the separation of the two fluids and the formation of the
interface.
In the above equations, the dynamic viscosity of the fluid is denoted by h , u is the
velocity field, and g is the Gibbs free energy given by:
g = f þ p/ r , with p is the pressure, and r is the mass density. The superscript T stands
for the transpose operator.
The viscosity h is assumed to have the following relationship with the concentration c
because the two fluids have different viscosities h 1 and h 2 respectively:
   
cþ1 1c
h ¼ h1 þ h2 (12)
2 2

As we focus on the coupling of the particles with the interface, we neglect inertia in the
momentum equation, so that equation (10) will be written as:
r rg  r  h ðru þ ruT Þ ¼ r m rc (13)

2.2 Motion of nanoparticles


The nanoparticles are randomly placed within the simulation box, and each one is
considered as a rigid spherical body whose velocity (vi) and position (xi) are updated by
using the dimensionless Newton’s equation of dynamics, which allows for computing the
particle acceleration (ai):
d2 xi
F i ð t Þ ¼ mi ¼ mi ai ðtÞ (14)
dt2
where mi is the mass of the particle, and Fi is the applied force on the particle. The applied
forces can be classified into forces between multiple particles, and external forces due to the
fluid, (in our case we consider Brownian and drag forces).
2.2.1 Particle-particle interactions. In our simulation model, the force between particles
is obtained using the truncated Lennard–Jones (LJtrunc) potential given by:
(
VLJ ðrÞ  VLJ ðrc Þ for r < rc
VLJtrunc ðrÞ ¼ (15)
0 for r > rc

where:
"   6 #
s 12 s
VLJ ðrÞ ¼ 4e  (16)
r r
with e is the depth of the potential well, s is the finite distance at which the inter-particle Nanoparticles
potential is zero, r is center to center separation between two particles, and rc is the cut-off migration
distance which is taken to be 3s . This potential is used in numerical simulation to model the
particle-particle interactions (Choi and Djilali, 2016; Hmady et al., 2013).
Based on this potential, the interaction force acting on the i – th particle induced by the
j – th particle is given by:

@VLJtrunc ðrij Þ 1041


f ji ¼ rVLJtrunc ðrÞ nji ¼  nji (17)
@rij

where rij = ri – rj is the distance between two particles, i and j correspond to different
nanoparticles and nj–i represents the unit vector that point from xj to xi.
Considering N particles, the force acting on each particle is composed of the individual
interactions with the rest of the particles:

X
N
Fi ¼ f ij (18)
j6¼i

In our model, the parameter e governs the strength of the interaction and s defines a length
scale, and they are chosen as reduced units.
2.2.2 Brownian force. A colloidal particle immersed in a fluid phase shows an irregular
and random walk, due to the collision between the molecules of the surrounding fluid and
the particle itself. Such an effect is known as Brownian motion. A random force Fr(t) alters
the particle’s trajectory after each molecular collision. This force is given by:
pffiffiffiffiffiffiffiffiffiffiffiffi
F r ðtÞ ¼ 2 D Dt x (19)

where x a normal random number is whose average is zero and variance is one, and D is the
diffusion coefficient.
2.2.3 Drag force. The drag force is applied in the opposite direction of the particle’s
motion in fluid. It is proportional to the relative velocity of the fluid with respect to the
particle.
Consider a sphere of radius a moving through an incompressible fluid with a relative
speed U.
'The drag D is a function of the parameters on which the problem depends, meaning
that:

D ¼ f ðU; a; r 0 ; h Þ

with r 0, h are the density of the fluid, and the dynamic viscosity, respectively.
The drag D has the dimension of force (Kgs2 m), so dimensional analysis implies that:
 
Ua
D ¼ r 0 U 2 a2 f
h

In the case of two immiscible fluids, the variation of concentration from each fluid toward
the interface imposes a drag force on the particles given by the following equation:
EC F d ¼ w ðvf  vp Þ (20)
36,3
where w is the drag coefficient, vp is the velocity of the particle and vf is the velocity of the
fluid given by the Navier–Stokes Cahn–Hilliard equation (Lowengrub and Truskinovsky,
1998):

1042 vf ¼ div ðrc  rcÞ (21)

where  corresponds to the tensor product.


In our model, the nanoparticles are of very small weight, and they have no effect on the
kinematics of the fluids, so we do not include the effect of the particle’s velocity on the fluid.
2.2.4 Scaling the equations. For fluid simulation, it is important to use a suitable set of
reduced units (dimensionless units) to simplify the equations and minimize the effects of round-
off errors. So we scale the governing equations by defining Uc, and Lc as the characteristic
velocity and characteristic length, respectively, Tc ¼ LUc c as the characteristic time, and « c as
the characteristic energy. The characteristic length of the phase field scale is related to that of
the molecular dynamics as Lc = 100 s (fixed ratio chosen for our studies).
We define:
s as the characteristic length for molecular dynamics.
m as the characteristic mass.
pffiffiffiffiffiffiffiffi
«c m
s as the characteristic drag coefficient.
m j2
We introduce r = r/s , «  = « /« c, c* ¼ ccB , u* ¼ Uucq
, t*ffiffiffi¼ tU
Lc , m ¼ « cB .
c *
pffiffi«ffi
j ¼ a Is the interfacial thickness and cB ¼
a
b is the bulk concentration which
represents the mean field equilibrium value.
Dropping the asterisk notations, we get:

dc 1
¼ u  rc þ r2 m (22)
dt Pe

m ¼ c3  c  C 2 r 2 c (23)

ru ¼0 (24)

rg  r  h ðru þ ruT Þ ¼ A m rc (25)

The surface tension parameter, A is related to the Cahn number and the capillary number,
and it is introduced in equation (25). In our model, A is a constant parameter it is not varying
with time:
 12  6 !
48 « s s
FLJ ¼  0:5 r (26)
r2 r r

Fd ¼ w * ð vf  vp Þ (27)
where the dimensionless drag coefficient is defined as: Nanoparticles
w s migration
w * ¼ pffiffiffiffiffiffiffiffiffiffi
«c m

It is a chosen dimensionless parameter.


where:
h ¼ hhc is the normalized viscosity. 1043
The dimensionless groups used above are:
2
Pe ¼ j ML«c Uc is the Peclet number representing the ratio between the diffusive time scale,
and the convective time scale.;
2
A ¼ Ljc 2r «hcUB gives the relative magnitude of viscous and interfacial tension forces at the
c

interface; and
C ¼ Ljc is the dimensionless interfacial thickness.
The effect of the Brownian force is represented by introducing the diffusion coefficient for a
spherical particle:

KB T

6p h a

where KB is the Boltzmann constant, T is the temperature (in kelvin), h is the dynamic
viscosity of the fluid and a is the particle’s radius.
In our work, we are considering the case of nanoparticles of size a smaller than the
interfacial thickness j , so that aj < 1.

3. Simulation methodology
3.1 Molecular dynamics method (MD)
In macroscopic systems, the fraction of the particles near the wall is negligible, whereas, in
the MD simulations, this fraction is more significant, and the surface effect is of great
importance. To reduce surface effect, and conserve the number of particles in the simulation
box, periodic boundary conditions must be used, so that when a particle enters or leaves the
simulation region, an image particle must leave or enter this region. Moreover, provided
the potential range is not too long, we use the minimum image convention by truncating the
potential at a certain (cut-off) distance, so that VLJ(r) = 0 for rij > rcut.

3.2 Numerical implementation


In this section, we will give some details about the numerical implementation.
The concentration c is represented as indicated in equation (22).
To find the weak form, we proceed as it is usual in finite element method (FEM) (Olek
et al., 2005). 
We multiply by a weighting function c and integrate over the whole fluid domain X to
get:
ð ð ð
dc 1 * 2
c* dX þ c* u  ðrcÞdX  c r m dX ¼ 0 (28)
dt Pe

where c* ¼ c * TN
EC dc
¼ N T c_ (29)
36,3 dt

rc ¼ B c (30)

Solving equation (28), we get:


1044 ð ð ð
*T 1
c *T T
_
N N dX c þ c *T
u  B dX c  c N r2 m dX ¼ 0 (31)
Pe

This integration allows obtaining a linear system that has to be solved at each time step:

M c_ þ G c þ F ðc Þ ¼ 0 (32)

Similarly, to solve the velocity equation, equation (25) we can write:


ð ð
u* r2 u dX þ u* A m rc dX ¼ 0 (33)

where:

u* ¼ u * T N (34)
and
r2 u ¼ K u (35)
Solving equation (33), we get:
ð ð
u * T N K dX u þ u * T N A m B dX c ¼ 0 (36)

This integration gives the following linear system to be solved at each time step:

T uþH c ¼0 (37)

3.2.1 Discretization. Using the notation c(t) for the known values of the current time step
t, c(t þ t ) at time t þ t is estimated by solving the following equation:

cð t þ t Þ  cð t Þ
M þ G cðtÞ þ F ðcðtÞÞ ¼ 0 (38)
t
This gives:
cðtÞ  t G cðtÞ  t F ðcðtÞÞ
cð t þ t Þ ¼ (39)
M
The position and velocity of the NPs (xi and vi respectively) at each new time are calculated
using the Velocity Verlet algorithm time integration scheme:
1 Nanoparticles
xðt þ d Þ ¼ xðtÞ þ vðtÞ d þ aðtÞd 2 (40)
2 migration
1 
vðt þ d Þ ¼ vðtÞ þ aðtÞ þ aðt þ d Þ d 2 (41)
2
t and d indicate the time increments for the phase field and for the molecular dynamics
(MD) respectively, with d  t . 1045
3.2.2 Problem description. The simulations have been carried out using 100 nanoparticles
in the medium of two immiscible liquids. First, we consider the interactions of the particles
with each other under the effect of Lennard – Jones force only. Initially, the particles are
randomly distributed inside a simulation box as shown in Figures 2 and 3.
The Lennard–Jones potential parameters are chosen as: e = 1, the particle’s mass (m) is
chosen to be unity, and s = 0.02. We can examine different values of s in our work.

Figure 2.
Periodic boundary
conditions

Figure 3.
Simulation cell with
100 nanoparticles
randomly distributed
EC The diameter of the nanoparticle is chosen to be related to the dimensionless Lennard–Jones
36,3 parameter sigma, d = 2 sigma = 0.04 as dimensionless unit, so that the minimum distance
that two nanoparticles are allowed to approach each other is (2 sigma).
The dimensionless parameters for the size of the nanoparticle are chosen so that the
nanoparticle’s size is consistent with the size of a few hundreds of atoms.
As time goes on, the distributed particles gradually form clusters of agglomeration under
1046 the effect of Lennard–Jones force as shown in Figure 4.
Similar results were obtained by Choi and Djilali (2016) who showed that as time goes on,
homogeneously distributed particles driven by the direct LJ inter-particle forces gradually
form clusters of agglomeration.
After examining the agglomeration of the NPs under the effect of Lennard – Jones
potential, we will introduce them into the system of two immiscible liquids.
We first modeled the two liquids according to the diffuse interface model discussed in
Section 2.
Then we introduce the nanoparticles into the medium by randomly distributing them,
and study their motion under the effect of the inter-particle force obtained from LJ potential
to examine their agglomeration with time. After that we study the effect of Brownian force
exerted by the fluid particles on the motion of nanoparticles. And as a third step, we
introduce the drag force exerted by the fluid motion at the interface on the previous system,
so that the final model represent the dynamics of nanoparticles interacting between each
other via LJ potential, placed within the medium of two immiscible liquids. First, we will
study the simulation in one dimension, and then we will consider the 2D case. All the results
will be represented in the following section below.

4. Results and discussion


4.1 1D Case
4.1.1 Particle–particle and fluid – particle interactions. The randomly distributed 100
nanoparticles interact with each other via the force represented by Lennard–Jones potential,
and they are affected by the Brownian force, and the drag force. This simulation was carried

Figure 4.
Snapshot of
agglomeartion of
nanoparticles under
the effect of Lennard–
Jones force for
s = 0.02
out using the following parameters: e = 1, the particle’s mass (m) is chosen to be unity, s = Nanoparticles
0.02, w  = 0.001 and C = 0.05. migration
The dashed blue curve represents the variation of the concentration between the two
liquids, it appears to vary from (1) to (1) and the continuous red one represents the PDF of
the nanoparticles.
In Figure 5, the interface is the region showing the continuously varying concentration
profile; it is limited by the green lines in the figure. The red curve representing the
probability density function (PDF) shows that the nanoparticles accumulate at the interface,
1047
which is mainly due to the drag force, and this result is insured in the 2D case also.
4.1.2 Brownian force. If we neglect the effect of the drag force, and consider only
Lennard Jones forces, and Brownian force, the nanoparticles will no longer accumulate at the
interface.
In Figure 6, we are showing the result of the dynamics of nanoparticles in the case when
two liquids have different viscosities, and the nanoparticles are under the effect of Lennard –
Jones force and Brownian force, the nanoparticles tend to accumulate in the medium of
higher viscosity h .
This result can be explained by the fact that in the medium of less viscosity, the
nanoparticles have more mobility driving them toward the medium of higher viscosity.

Figure 5.
PDF of nanoparticles
interacting via
Lennard – Jones
potential, Brownian
force and drag force
in 1D

Figure 6.
PDF of nanoparticles
interacting via
Lennard – Jones
potential, and
Brownian force
EC 4.2 2D Case
36,3 After we examine our model in 1D, we extrapolate the work to consider the 2D case.
We consider the simulation box to be a square of side S = 1, and the non-dimensional

parameters stated above are considered to be Pe = 100, C = 0.02, A = 0.02, w = 0.001 and
D = 0.001 and the results are shown in the figures below.
The side bar in Figure 7 represents the evolution of the concentration between the two
1048 fluids in order to form the interface. The concentration is taken to vary from (1) in the first
fluids to (1) in the second one.
4.2.1 Particle–particle interactions. The physical mechanism that produce the Cahn-
Hilliard equation can be seen as a negative diffusion, that allows the separation of the two
liquids and the formation of the interface, as discussed briefly is Section 1. regarding
equation (7) and equation (9).
As time goes on, the liquids start to separate and the interface is formed between them.
The blue medium represents the first liquid, and the red represents the second liquid, and
the concentration is varying according to the diffuse interface model. Note that our
simulation box is bounded between 0 and 1 along the two axes, and we present periodic
repetition of this box in Figure 8 for clarity.
The separation of the two fluids and the formation of the interface between them are
obtained using the diffuse interface model based on the Cahn – Hilliard theory. The phase
separation of the two liquids obtained in our model matches previous results obtained by
Badalassi et al. (2003).
The nanoparticles distributed in the mixture tend to agglomerate and form clusters,
under the effect of Lennard–Jones forces, but they are still normally distributed within the
two fluids as shown in Figure 8.
4.2.2 Drag force. Although we start from static case liquids, the fluid velocity field is
seen at the interface. This field is related to the concentration gradient at the interface.
If we consider equation (10), compared with the Navier–Stokes equations, only one extra
capillary term r m rc appears reflecting the interfacial tension. This modiflcation accounts
for hydrodynamic interactions, that is, the influence of the concentration c or the
morphology on the velocity field and, hence, describes the spatial variations of the velocity
field because of the presence of interfaces.

Figure 7.
Initial state of the
mixture of two fluids
with randomly
distributed
nanoparticles just at
the beginning of the
simulation
The velocity field is seen from both sides in opposite direction normal to the interface as Nanoparticles
shown in Figure 9. migration
This fluid velocity at the interface will drive the nanoparticles to the interfacial region.
Note that in our model, we are introducing the nanoparticles to the medium at the initial state
when the two fluids are mixed. And as time goes, the two liquids will start to separate and
the interface will be formed, and at each moment, during phase separation the nanoparticles
are affected by the fluids velocity at the interface resulting from the concentration gradient.
This is the main factor that drives the nanoparticles to the interface as shown in Figure 10. 1049
The nanoparticles approach the interface from both sides and agglomerate there forming
a layer between the two mediums as shown in Figure 10 where the black dots represent the
nanoparticles accumulating at the interface.
The NPs are driven to the interface under the effect of the drag force arising from the
concentration gradient as shown in equation (27).
These results match the results obtained experimentally (Herzig et al., 2007) for the
spinodal decomposition in the presence of solid particles.

Figure 8.
2D representation of
nanoparticles
suspended in two
immiscible liquids,
and interacting via
Lennard–Jones forces
at the end of the
simulation

Figure 9.
Fluid velocity at the
interface from both
sides
EC 4.2.3 Effect of non-dimensional parameters. In the following subsection, we will study the
36,3 effect of different non-dimensional parameters on the motion of the nanoparticles within the
two fluids. We will examine the effect of different values on the time needed by these
nanoparticles to reach the interface and approach the equilibrium position, taking into
account that in our simulation model, the drag force is driving the NPs toward the interface
during the formation of the interface.
1050 4.2.3.1 Surface tension. The non-dimensional measure of the surface tension is described
by the parameter A defined in Section 2.
Figure 11 shows the variation of the time taken by the nanoparticles to reach the
interface and approach their equilibrium position as function of different values of A.
As the parameter A increases, that is the surface tension increases, the nanoparticles
needs less time to reach the interface and belong to the interfacial region. This result is
expected because the surface tension plays an essential role in pushing the nanoparticles
toward the interface.
4.2.3.2 Interfacial thickness. The non-dimensional measure of the interfacial thickness is
the Cahn number C. Figure 12 shows the effect of different values of C on the time needed by
the nanoparticles to reach the interface, and we have shown that as the Cahn number
increases, i.e. the interfacial thickness increases the nanoparticles moves faster toward the
interface.

Figure 10.
Accumulation of
nanoparticles at
liquid–liquid
interface due to the
drag force

Figure 11.
Effect of the surface
tension on the time
needed by the
nanoparticles to reach
the interface
4.2.4 Spherical nanoparticle greater than the interfacial thickness. Many experiments deal Nanoparticles
with nanoparticles which are greater than the interfacial thickness. The dynamics of migration
particles breaching interfaces has been investigated experimentally (Kaz et al., 2012), and it
has been shown that sulfate-functionalized latex microspheres breaching an oil-water
interface relaxed logarithmically with time toward equilibrium. There are many techniques
that have been used to study particle interface interactions in the past years. For overviews,
see Albijanic et al. (2010) and McGorty et al. (2010).
4.2.5 Modeling of large nanoparticles. To model a large particle (greater than the 1051
interfacial thickness), we consider a group of small nanoparticles interacting between each
other via the Lennard–Jones potential, defined in equation (16). The small nanoparticles will
aggregate and form the larger one.
Another approach to model the large particle is to consider its effect on the system,
through which more inertia is applied on the system.
Both approaches have been tested in our model, and the same results regarding the
oscillation of the nanoparticle at the interface is obtained.
When such a particle is placed near to the interface, it will be affected by the velocity field
within the interfacial region, and it will be attracted from the first side of the interface
through the interfacial region to the other side where it is affected by the same field but in
the opposite direction, and thus it starts to oscillates, as shown in Figure 13.
The damped oscillation is mainly due to the drag effect, the amplitude of oscillation will
decrease with time due to the relative velocity between the particle and the fluid, and thus
the nanoparticle will undergo damped oscillation. This damped oscillation is seen even if we
neglect the Brownian force. In addition, we have found that the Lennard–Jones forces and
drag forces dominate over the Brownian force.
These obtained results match well with experimental data obtained by Chen et al. (2012)
in addition to the results obtained by Pouzot et al. (2006) which insures the consistency of
our model.

5. Conclusion
In this work, we have presented a numerical simulation method for the dynamics of solid
nanoparticles near a liquid–liquid interface. Our work was based on the diffuse interface
model of Cahn–Hilliard. The phase separation of the two liquids obtained in our model
matches previous results obtained by Badalassi et al. (2003). We find that nanoparticles
smaller than the interfacial thickness accumulate at the interface due to the drag force
related to fluid concentration gradient. In addition, we have shown that as the interfacial

Figure 12.
Effect of the Cahn
number on the time
needed by the
nanoparticles to reach
the interface
EC
36,3

1052

Figure 13.
Oscillation of
nanoparticles greater
than the interface at
the interfacial region

thickness and the surface tension increase these nanoparticles move faster toward the
interface where they agglomerate and form a layer in the interfacial region. On the other
hand, for the case of nanoparticles greater than the interface, we have addressed the damped
oscillation of such nanoparticles. Our results agree well with the experimental results
obtained by Chen et al. (2012) in addition to the results of Pouzot et al. (2006). These results
insure the consistency of the model used.

References
Albijanic, B., Ozdemir, O., Nguyen, A.V. and Bradshaw, D. (2010), “A review of induction and
attachment times of wetting thin films between air bubbles and particles and its relevance in the
separation of particles by flotation”, Advances in Colloid and Interface Science, Vol. 159 No. 1,
pp. 1-21.
Altenbach, H., Naumenko, K., Pylypenko, S. and Renner, B. (2007), “Influence of rotary inertia on the
fiber dynamics in homogeneous creeping flows”, ZAMM, Vol. 87, pp. 81-93.
Badalassi, V.E., Ceniceros, H.D. and Banerjee, S. (2003), “Computation of multiphase systems with
phase field models”, Journal of Computational Physics, Vol. 190 No. 2, pp. 371-397.
Butler, J.E. and Shaqfeh, E.S.G. (2002), “Dynamic simulations of the inhomogeneous sedimentation of
rigid fibres”, Journal of Fluid Mechanics, Vol. 468, pp. 205-237.
Cahn, J.W. and Hilliard, J.E. (1958), “Free energy of a nonuniform system. I. Interfacial energy”, Journal
of Chemical Physics, Vol. 28 No. 2, pp. 258-267.
Cahn, J.W. and Hilliard, J.E. (1959), “Free energy of a nonuniform system. III nucleation in a two- Nanoparticles
component incompressible fluid”, Journal of Chemical Physics, Vol. 31 No. 3, pp. 688-699.
migration
Cauvin, S., Clover, P.J. and Bon, S.A.F. (2005), “Pickering stabilized miniemulsion polymerization:
preparation of clay armored latexes”, Macromolecules, Vol. 38 No. 19, pp. 7887-7889.
Chen, L., Heim, L.O., Golovko, D.S. and Bonaccurso, E. (2012), “Snap-in dynamics of single particles to
water drops”, Applied Physics Letters, Vol. 101 No. 3, p. 031601.
Choi, Y.J. and Anderson, P.D. (2012), “Cahn – Hilliard modeling of particles suspended in two-phase
flows”, International Journal for Numerical Methods in Fluids, Vol. 69 No. 5, pp. 995-1015.
1053
Choi, Y.J. and Djilali, N. (2016), “Direct numerical simulations of agglomeration of circular
colloidal particles in two dimensional shear flow”, Physics of Fluids, Vol. 28 No. 1,
p. 013304.
Clegg, P.S. (2008), “Fluid-bicontinuous gels stabilized by interfacial colloids: low and high molecular
weight fluids”, Journal of Physics: Condensed Matter, Vol. 20, p. 113101.
Ding, H., Spelt, P.D.M. and Shu, C. (2007), “Diffuse interface model for incompressible two-phase flows
with large density ratios”, Journal of Computational Physics, Vol. 226 No. 2, pp. 2078-2095.
Gewin, V. (2006), “Nanotech’s big issue nature”, 443, p. 137.
Grmela, M., Ammar, A., Chinesta, F. and Maitrejean, G. (2014), “A mesoscopic rheological model of
moderately concentrated colloids”, Journal of Non Newtonian Fluid Mechanics, Vol. 212,
pp. 1-12.
Herzig, E.M., White, K.A., Schofield, A.B., Poon, W.C.K. and Clegg, P.S. (2007), “Bicontinuous emulsions
stabilized solely by colloidal particles”, Nature Materials, Vol. 6 No. 12, pp. 966-971.
Hijazi, A. and Khater, A. (2008), “Boëder PDF brownian simulations for macromolecular rod-like
particles near uneven solid surfaces”, European Polymer Journal, Vol. 44 No. 11,
pp. 3409-3416.
Hijazi, A., Ben Yahia, L., Khater, A. and Zoaeter, M. (2003), “Experimental study of the collision between a
rod like macromolecule and a solid surface”, European Polymer Journal, Vol. 39 No. 3, pp. 521-525.
Hmady, S., Hijazi, A. and Atwi, A. (2013), “Lennard–Jones interactions between nano-rodlike particles at
an arbitrary orientation and an infinite flat solid surface”, Physica B, Vol. 423, pp. 26-30.
Kaz, D.M., McGorty, R., Mani, M., Brenner, M.P. and Manoharan, V.N. (2012), “Physical ageing of
the contact line on colloidal particles at liquid interfaces”, Nature Materials, Vol. 11 No. 2,
pp. 138-142.
Lowengrub, J. and Truskinovsky, L. (1998), “Quasi-incompressible Cahn-Hilliard fluids and topological
transitions”, Proceedings of the Royal Society A, Vol. 454 No. 1978, pp. 2617-2654.
McGorty, R., Fung, J., Kaz, D. and Manoharan, V.N. (2010), “Colloidal self-assembly at an interface”,
Materials Today, Vol. 13 No. 6, pp. 34-42.
Mackaplow, M.B. and Shaqfeh, E.S.G. (1998), “A numerical study of the sedimentation of fibre
suspensions”, Journal of Fluid Mechanics, Vol. 376, pp. 149-182.
Maitrejean, G., Ammar, A., Chinesta, F. and Grmela, M. (2012), “Deterministic solution of the kinetic
theory model of colloidal suspensions of structureless particles”, Rheologica Acta, Vol. 51 No. 6,
pp. 527-543.
Millett, P.C. and Wang, Y.U. (2011), “Diffuse-interface field approach to modeling arbitrarily-shaped
particles at fluid–fluid interfaces”, Journal of Colloid and Interface Science, Vol. 353 No. 1,
pp. 46-51.
Moses, K.B., Advani, S.G. and Reinhardt, A. (2001), “Investigation of fiber motion near solid boundaries
in simple shear flow”, Rheologica Acta, Vol. 40 No. 3, pp. 296-306.
Pouzot, M., Nicolai, T., Benyahia, L. and Durand, D. (2006), “Strain hardening and fracture of heat-
set fractal globular protein gels”, Journal of Colloid and Interface Science, Vol. 293 No. 2,
pp. 376-383.
EC Shukla, A., Degen, P. and Rehage, H. (2007), “Synthesis and characterization of monodisperse poly
(organosiloxane) nanocapsules with or without magnetic cores”, Journal of the American
36,3 Chemical Society, Vol. 129 No. 26, pp. 8056-8057.
Wise, S.M., Lowengrub, J.S. and Cristini, V. (2011), “An adaptive multigrid algorithm for simulating
solid tumor growth using mixture models”, Mathematical and Computer Modelling, Vol. 53
Nos 1/2, pp. 1-20.
Wittstock, A., Zielasek, V., Biener, J., Friend, C.M. and Baumer, M. (2010), “Nanoporous gold catalysts
1054 for selective gas-phase oxidative coupling of methanol at low temperature”, Science, Vol. 327
No. 5963, pp. 319-322.

Further reading
Yamamoto, S. and Matsuoka, T. (1996), “Dynamic simulation of microstructure and rheology of fiber
suspensions”, Polymer Engineering and Science, Vol. 36 No. 19, pp. 2396-2403.
Zienkiewicz, O.C. and Taylor, R.L. (2005), The Finite Element Method for Fluid Dynamics, 6th ed.,
Butterworth-Heinemann, Oxford.

Corresponding author
Ali Daher can be contacted at: ali.daher@ensam.eu

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy