Sei sulla pagina 1di 13

Single Particle Motion in a Sheared Colloidal Suspension

Toni Bechtel
Department of Chemical Engineering
Carnegie Mellon University
Pittsburgh PA
Ph.D. Qualier
August 15, 2014
Abstract
The motion of a single Brownian probe particle subject to a constant applied force
in a linear shear ow is studied. The average velocity of the probe particle is determined
from the spatio-temporal conguration (i.e. the microstructure) of other particles in
the limit of low P`eclet number (Pe 1), which is the ratio of advective to diusive
transport. Alift force is generated due to the coupling of the shear ow and applied
force. The coupling of the shear ow and external force generates regions of varying
pressure around the probe particle - regions of higher or lower pressure correspond to
a higher or lower density of bath particles, respectively. This lift force acts to push
the particle toward a region of lower pressure. From this we see that the probe particle
will not travel solely in the direction of the applied force, but it will migrate across
streamlines toward regions of lower pressure.
1 Introduction
Colloidal dispersions are a suspension of small particles (roughly 1 nm to 10 m) in an immis-
cible uid [1]. These particles are subject to Brownian motion where solvent molecules are
continuously colliding with the particles causing a random motion of the particle. Colloidal
dispersions will form a spatial-temporal conguration - known as a microstructure, which
aects how the uid will ow and deform under a given stress, i.e. the microstructure alters
the rheological behavior (how a uid ows and deforms under a given stress) of the uid.
Typically, commercially available rheometers (concentric cylinders, cone and plate, etc.) are
used to determine bulk viscoelastic properties such as shear viscosity, elastic moduli, and
normal stresses. The term viscoelastic refers to a material that exhibits both viscous (uid-
like) and elastic (solid-like) properties [2]. In order to accurately perform these experiments,
sample sizes of roughly 1 mL are required. However, for many rare syntheses or biological
systems, where sample sizes are roughly 1 L, this is not practical and traditional rheometers
cannot be utilized [3].
Recently, a class of techniques known as microrheology has been used to measure local
viscoelastic properties. These techniques can utilize small sample sizes and also have the
ability to study local inhomogeneities in a system, which is not generally possible in a
rheometer [3, 4]. There are two main forms of microrheology - active and passive tracking.
Passive tracking monitors the movement of colloidal probe particles due to Brownian motion.
The diusivity (and also the mean-squared displacement) of the probe is directly proportional
to the thermal forces encountered (from the collision with solvent molecules) [2, 5, 6]. Active
tracking applies an external force on a probe particle. This external force can be achieved by
use of optical tweezers, magnetic eld gradients, or magnetic tweezers [3, 6, 7] and a probe
particle can be pulled through the uid with either a constant force or constant velocity.
The applied force and average velocity are related through a friction coecient, , where
F = U with the brackets denote an ensemble average. For suciently small forces, the
linear-response regime is probed and is independent of the strength of the applied force.
1
As the external force is increased, however, becomes dependent on the strength of the
force, and the nonlinear-regime is probed.
What we are interested in is investigating how the combination of a linear shear ow
(much like what you would have in a rheometer) and an applied external force on a probe
particle aects the non-equilibrium microstructure of a colloidal suspension and ultimately
the average motion of the probe particle. Experimental work [8, 9] , along with numerical and
simulation techniques [10, 11, 12] have investigated the settling of spheres in a viscoelastic
uid that is exposed to a cross-shear ow (shear ow in the x-y plane). These methods
show that by combining a shear ow with an applied force on a probe particle (gravity in
their cases), the settling speed is decreased. We are investigating this phenomena from a
fundamental perspective in order to determine how changes in the uids microstructure
aect the motion of a probe particle.
2 Problem Set-Up
(a) Schematic of probe and bath particles in a shear ow (b) Spherical coordinate system
Figure 1: (a) The probe particle (depicted in red) is being pulled with a constant external force, while the
other bath particles (depicted in blue) remain force-free. The external force and shear ow are given by F
ext
and u

, respectively. (b) Denition of a spherical coordinate system.


The system we are investigating is a suspension of N colloidal particles of radius a in an
incompressible, Newtonian uid with density and viscosity , where is solely a function of
temperature. All particles are assumed to be Brownian hard-spheres with no hydrodynamic
interactions, meaning that the uid motion of one particle does not aect the uid motion of
another particle. The particles are subject to hard-sphere repulsion and thus are unable to
2
overlap [13]. The suspension is assumed to be in the dilute limit where the volume fraction,
dened as =
4
3
a
3
n with n equal to the number density of particles, is assumed to be much
less than one. A linear shear ow (u

) is applied to the system and a constant external force


(F
ext
) is applied to a single particle, referred to as the probe particle. All other particles,
referred to as bath particles, remain force-free. The system is schematically depicted in
Figure 1a. We assume that the Reynolds number (Re = Ua/) is suciently small such
that the uid ow is governed by the Stokes equations.
The system is driven out of equilibrium by the imposed shear ow and external force
while Brownian motion drives to restore the system to equilibrium. The extent to which the
system is driven out of equilibrium is dictated by the P`eclet number (Pe), which is a ratio of
the rate of advective transport to diusive transport. Since there are two modes of advective
transport, two dierent P`eclet numbers are relevant. The rst is the P`eclet number based on
the shear ow: Pe

= a
2
/2D
0
, where is the shear rate in reciprocal seconds and D
0
is the
diusivity of an isolated colloid determined from the Stokes-Einstein-Sutherland equation
(D
0
= kT/6a). The second P`eclet number is based on the force: Pe
F
= F
ext
/(2kT/a),
where F
ext
is the magnitude of the external force, k is Boltzmanns constant, and T is the
absolute temperature.
The microstructure of a dilute suspension is described by the pair distribution function
g(r). The pair distribution function describes the relative likelihood of nding a bath particle
at a center-to-center distance, r, from the probe particle [14]. For a uniform microstructure
(i.e. at equilibrium), g(r) = 1. In the limit of no hydrodynamic interactions, the pair
distribution function, g(r), satises a steady two-body Smoluchowski equation:
[Pe

s) Pe
F

F
ext
] g =
2
g (1)
where s = r/a,

is the velocity gradient of the uid scaled by the shear rate ( ), and

F
ext
is the applied external force scaled by the magnitude of the force. It should be noted
that from this point forward, all variables are dimensionless, unless otherwise stated.
Equation (1) is a second-order dierential equation and thus two boundary conditions
3
must be specied in order to fully solve for the pair distribution function. The rst condition
arises from the assumption that the particles are unable to overlap which leads to a no-ux
condition at the surface of the probe (s = 2):
s [(Pe

s) Pe
F

F
ext
)g g] = 0 (2)
where s is the radial unit vector dened as s = s/s and s = |s|. To obtain the second
boundary condition, it is assumed that the microstructure is uniform far away from the
probe (g(s) 1 as s ). Once a solution to the pair distribution function is obtained
the relative probe velocity can be determined:
U =
1
6a

F
ext
nkT

g(s) s dS

(3)
where U is the average probe velocity and the integral is evaluated over the surface of
the probe (dS = a
2
sin dd). The rst term in equation (3) is the average probe velocity
in isolation found from Stokes law. The second term arises because the presence of bath
particles causes a resistance to the movement of the probe particle [15]. From the pair
distribution function and the average probe velocity we are able to determine how the shear
ow and external force are driving the microstructure out of equilibrium and changing the
trajectory of the probe.
3 Asymptotic Analysis
At small P`eclet numbers (Pe

, Pe
F
1), the ratio of advective transport to diusive trans-
port is small, meaning that the microstructure is only slightly perturbed from its equilibrium
conguration. This fact allows us to assume that the pair distribution function can be de-
termined from a perturbation expansion in both Pe

and Pe
F
.
From previous work [16, 17] and due to the linearity of the Stokes equations, we propose
an expansion to the pair distribution function the following form:
g = 1 + Pe

g
1a
+ Pe
F
g
1b
+ Pe
2

g
2a
+ Pe
2
F
g
2b
+ . . . (4)
where g
1a
and g
2a
are the rst and second order solutions to the pair distribution due to
4
the shear ow, respectively, and g
1b
and g
2b
are the rst and second order solutions to the
pair distribution due to the external force. The solution for g
1a
was determined by Brady
& Vicic (1995) to be: g
1a
= (32/3s
3
)(s

E s), where

E is the symmetric rate of strain
tensor scaled by . The solution for g
1b
was determined by Khair & Brady (2006) to be:
g
1b
= (4/s
2
)(s

F
ext
). The solutions for g
2a
and g
2b
were also determined by Brady & Vicic
(1995) and Khair & Brady(2006), respectively. However, both g
2a
and g
2b
are irrelevant to
our analysis because they do not contribute to a change in the probes velocity.
We expect, from previous work [16, 17], that we are unable to neglect advection every-
where in the system. Therefore, we must dene and inner and outer region where in the
inner region the system is dominated by diusion and in the outer region advection and diu-
sion are balanced. To do so we dene an outer coordinate = Qs, where Q(Pe
F
, Pe

) 1
must be determined from the governing equation. Equation (1) can be written in terms of
the outer coordinate:
Pe

Q
2
[

(
Pe
F
Pe

)Q

F
ext
]

G =
2

G (5)
where G is the pair distribution function in the outer region. We will focus exclusively
on a shear dominated outer region where (Pe
F
/Pe

)Q 1 and the outer coordinate is then


= Pe
1/2

s. By comparing the magnitude of the term Pe

g
1a
to Pe
F
g
1b
we nd that this
can only occur if Pe
3/2

Pe
F
Pe
1/2

for Pe

1. Therefore, the shear dominated outer


region is governed by:
( )

G =
2

G (6)
The solution in the outer region must match the solution in the inner region as 0 and
be equal to 0 as , denoting a uniform microstructure far away from the probe.
4 Results and Discussion
We now look specically at a system with a linear shear ow in the x-y plane given by:
u

= y e
x
and an external force solely in the x-direction given by: F
ext
= e
x
. Figure 2 shows
an intensity plot for the two terms g
1a
and g
1b
. By convention, regions of red/orange indicate
that g
i
is positive, blue/purple indicate that g
i
is negative, and green indicate g
i
0, where
5
(a) Intensity plot of g
1a
(b) Intensity plot of g
1b
Figure 2: (a) The shear ow and movement of bath particles around the probe generates an overall stress
around the surface of the probe. Due to symmetry, the overall stress is balanced and the probe will continue
to travel in the direction of the applied force (b) The probe is being pulled in the positive x-direction (toward
the right), forcing it into contact with bath particles. This causes an increased pressure in front of the probe
and decreased pressure behind the particle (due to a sink of bath particles). The probe will naturally try to
move toward the region of lower pressure behind it, slowing it down. Ultimately it will continue to move in
the direction of the force.
g
i
is any term in the perturbation expansion. In gure 2a we see the quadrupole nature of the
shear ow where there is a higher probability of bath particles being present in quadrants II
and IV and a lower probability in quadrants I and III. The increased probability of spheres in
quadrants II and IV creates a normal stress on the probe particle, but this stress is balanced
by an equal and opposite stress in quadrants I and III. Thus the probe will continue to travel
in the direction of the applied force.
Figure 2b shows the dipole nature of the external force where there is a higher probability
of bath particles being present in front of the sphere. The probe is pulled in the positive
x-direction (toward the right) and forced into closer contact with the bath particles, thus
inducing a normal stress on the probe. Naturally, the probe attempts to move to the region
of lower pressure, creating a resistance to the external force and eectively slowing it down
[15].
From equations (3) and (4) and the solutions to the terms g
1a
and g
1b
, we determined
that the average probe velocity, based on a rst-order approximation to the pair distribution
6
function, to be U = (6a)
1
(1 2)e
x
. From this we are only able to predict that the
probe will continue to travel in the direction of the force (i.e. in the x-direction). However,
other theoretical work [10, 11, 12] and experimental studies [8, 9] predict that the probe
should not travel solely in the direction of the force, but it should experience a lift force
that pushes it up in the positive y-direction. Therefore, a rst-order approximation to the
pair distribution function does not completely describe how the microstructure is changing
due to the shear ow and external force and higher-order terms in the expansion must be
determined.
The rst-order terms from equation (4) can be written in terms of the outer coordinate
system:
Pe

g
1a
+ Pe
F
g
1b
= Pe
5/2

(
32
3
1

3
)(s

E s) + Pe

Pe
F
(
4

2
)(s

F
ext
) (7)
indicating that the leading order solution to the pair distribution function in the outer region
is O(Pe
F
Pe

). This suggests than an O(Pe


F
Pe

) term is required in the inner region to


satisfy the matching condition. The relevant expansion in the inner region then becomes:
g = 1 + Pe

g
1a
+ Pe
F
g
1b
+ Pe
F
Pe

g
3
+ . . . (8)
and the term g
3
in the inner region must satisfy:

2
g
3
=

s g
1b


F
ext
g
1a
(9a)
s g
3
= 2(s

E s) g
1b
(s

F
ext
) g
1a
(9b)
From equations (9a) and (9b) we determined g
3
to be:
g
3
= (
112
15s
2
+
224
15s
4
)(

F

E s) + 2 (

F

s) + (1
16
3s
2
+
112
3s
4
)(s

E s)(s

F) (10)
where

is the anti-symmetric vorticity tensor scaled by and is dened from the veloc-
ity gradient (

=

E

) and

F is the applied external force on the probe particle - the
superscript ext has been dropped for brevity.
From equations (3), (8), and (10) the average probe velocity is found to be
U = (6a)
1
[(1 2)e
x
+
44
15
Pe

e
y
] (11)
7
Figure 3: (a) The O(Pe
F
Pe

) term, g
3
shows
the coupling eect of the shear ow and exter-
nal force. The region of high pressure below the
probe and low pressure above the probe results
from a stress imbalance around the surface, driv-
ing the probe toward the region of lower pressure
and stress (positive y-direction).
By including the O(Pe
F
Pe

) term in the perturbation expansion we now see that the probe


will travel in both the x and y directions for an applied force solely in the x-direction. Figure
3 shows an intensity plot of the term g
3
in the x-y plane for a force in the x-direction. A
region of lower pressure exists above the sphere and a region of higher pressure below the
sphere. These regions of varying pressure arise due to a stress imbalance in the system. At
the top of the probe, the magnitude of the stress due to the shear ow is decreased because
the shear ow and external force are in the same direction. However, at the bottom of the
probe the direction of the shear ow and the force are opposite, increasing the stress on
the probe. The stress imbalance then causes the probe to rotate in a clock-wise direction
and migrate in the positive y-direction (across streamlines) to alleviate the stress imbalance,
which is qualitatively supported by equation (11). Therefore, from the coupling of the shear
ow and force we predict cross-streamline migration to occur.
It should be noted that g
3
= O(1) +O(s
2
) +O(s
4
) and the solution does not approach
0 as s , meaning that the far-eld boundary condition is not satised. Therefore,
the O(Pe
F
Pe

) term in the inner solution must be balanced by a term from the outer
region in order to satisfy the boundary condition. In the outer solution at O(Pe
F
Pe

),
equation (6) must be solved and match the leading order term from the inner solution (i.e.
G (4Pe
F
Pe

/
2
)(s

F
ext
) as 0). The fundamental solution to equation (6) was
originally determined by Elrick [18] for a point-source. However, the probe particle acts as
8
a point-dipole in the outer region by inducing the movement of nearby bath particles. By
following the work of Leal (1973) [19], we determined the solution to G in the outer region
to be:
G =
Pe
F
Pe


0
x
yw
2
w
5/2
(1 +
w
2
12
)
3/2
exp

(x
yw
2
)
2
4s(1 +
w
2
12
)

y
2
+ z
2
4w

dw (12)
The integral in equation (12) can be evaluated in the limit as 0, which is in fact the
solution to the pair distribution function in the inner region (due to the required matching
condition) and is determined to be:
g =
4Pe
F
s
2
sin cos Pe
F
Pe

(1 sin
2
cos
2
)

Pe
F
Pe
3/2

s sin (0.479 cos 0.683 sin ) + O(s


2
) (13)
The O(Pe
F
Pe
3/2

) that arises in the inner solution to the pair distribution function in equa-
tion (13) suggests that the relevant perturbation expansion should then be:
g = 1 + Pe

g
1a
+ Pe
F
g
1b
+ Pe
F
Pe

g
3
+ Pe
F
Pe
3/2

g
4
+ . . . (14)
where g
4
must satisfy:

2
g
4
= 0 (15a)
s g
4
= 0 on s = 2 (15b)
g
4

s sin

(0.479 cos 0.683 sin ) as s (15c)


From equations (15a) - (15c) we determined g
4
to be:
g
4
= (s +
4
s
2
)(0.385 sin sin 0.27 sin cos ) (16)
and the updated average probe velocity is:
U = (6a)
1

(1 2 + 1.62Pe
3/2

)e
x
+ (
44
15
Pe

2.31Pe
3/2

)e
y

(17)
From equation (17) we see that the point-dipole nature of the probe acts to decrease
its movement in the y-direction but enhances its movement in the x-direction. We can
qualitatively observe this in gure 4 where there is a region of lower pressure in quadrant
IV and a region of higher pressure in quadrant II. Even though this term suggests a region
9
Figure 4: The probe particle acts as a point-dipole
source in the outer region of the system by in-
ducing the motion of the bath particles around
it. This creates a region of lower pressure in the
positive x and negative y directions, hindering the
migration observed in the O(Pe
F
Pe

) term (Fig-
ure 3). However, this eect is small and we still
predict an overall lift in the probe particle in the
positive y-direction.
of lower pressure, it is dependent on Pe
3/2

which is much less than one, making this a small


eect. Therefore, we can still predict that the probe particle will have velocity components
in the x and y directions and that cross-streamline migration will be observed.
5 Conclusions
We determined the average velocity of a forced probe particle in a dilute colloidal suspension
that is subjected to a linear shear ow. The average probe velocity was determined from the
pair distribution function, g(s), which was found using singular perturbation theory. The
coupling eect of the shear ow and external force generates regions of varying pressure
around the probe particle, driving the probe to migrate to regions of lower pressure (i.e.
lower density of bath particles).
Moving forward, it would be advantageous to investigate the eects of hydrodynamic
interactions between the probe and bath particles. We expect to still predict cross-streamline
migration of the probe particle, however the degree to which this occurs will be aected by
the strength of the hydrodynamic interactions. It would also be interesting the investigate
this system in the limit of high P`eclet numbers (Pe

, Pe
F
), where the pair distribution
function varies greatly in the boundary layer that forms near the surface of the probe.
10
References
[1] R. J. Hunter, Foundations of Colloid Science. Oxford University Press, 2nd ed., 2001.
[2] H. Barnes, J. Hutton, and K. Walters, An Introduction to Rheology, vol. 1. Amsterdamn,
The Netherlands: Elsevier Science Publishers, 1993.
[3] W. P. L. Wilson, Small-world rheology: an introduction to probe-based active mi-
crorheology., Physical chemistry chemical physics : PCCP, vol. 13, pp. 1061710630,
2011.
[4] P. Cicuta and A. M. Donald, Microrheology: a review of the method and applications,
Soft Matter, vol. 3, no. 12, pp. 14491455, 2007.
[5] F. C. Mackintosh and C. F. Schmidt, Microrheology, Colloid & Interface Science,
vol. 4, pp. 300307, 1999.
[6] A. M. Puertas and T. Voigtmann, Microrheology of colloidal systems., Journal of
physics. Condensed matter : an Institute of Physics journal, vol. 26, p. 243101, June
2014.
[7] P. Habdas, D. Schaar, A. Levitt, and E. Weeks, Forced motion of a probe particle near
the colloidal glass transition, Europhysics Letters, vol. 67, no. 3, pp. 477483, 2004.
[8] B. van den Brule and G. Gheissary, Eects of uid elasticity on the static and dynamic
settling of a spherical particle, Journal of Non-Newtonian Fluid Mechanics, vol. 43,
pp. 123132, 1993.
[9] R. Chhabra, P. Uhlherr, and D. Boger, The inuence of uid elasticity on the drag co-
ecient for creeping ow around a sphere, Journal of Non-Newtonian Fluid Mechanics,
vol. 6, pp. 187199, 1980.
[10] R. Vishnampet and D. Saintillan, Concentration instability of sedimenting spheres in
a second-order uid, Physics of Fluids, vol. 24, no. 7, p. 073302, 2012.
[11] S. Padhy, E. Shaqfeh, G. Iaccarino, J. Morris, and N. Tonmukayakul, Simulations of
a sphere sedimenting in a viscoelastic uid with cross shear ow, Journal of Non-
Newtonian Fluid Mechanics, vol. 197, pp. 4860, July 2013.
[12] F. Q. Roger I. Tanner, Kostas D. Housiadas, Mechanism of drag increase on spheres
in viscoelastic cross-shear ows, Journal of Non-Newtonian Fluid Mechanics, vol. 203,
pp. 5153, 2014.
[13] J. Prausnitz, Molecular Thermodynamics of Fluid-Phase Equilibria. Prentice Hall,
3rd ed., 1999.
[14] E. Guazzelli and J. F. Morris, A Physical Introduction to Suspension Dynamics. Cam-
bridge University Press, 2012.
11
[15] T. M. Squires and J. F. Brady, A simple paradigm for active and nonlinear microrhe-
ology, 2005.
[16] J. F. Brady and M. Vicic, Normal stresses in colloidal dispersions, Journal of Rheology,
vol. 39, pp. 545566, May 1995.
[17] A. S. Khair and J. F. Brady, Single particle motion in colloidal dispersions: a simple
model for active and nonlinear microrheology, Journal of Fluid Mechanics, vol. 557,
pp. 73117, June 2006.
[18] D. Elrick, Source functions for diusion in uniform shear ow, Australian Journal of
Physics, vol. 15, no. 3, p. 283, 1962.
[19] L. G. Leal, On the Eective Conductivity of a Dilute Suspension of Spherical Drops
in the Limit of Low Particle Peclet Number, Chemical Engineering Communications,
vol. 1, pp. 2131, Jan. 1973.
12

Potrebbero piacerti anche