The Spreading of a Non-Isothermal

Liquid Droplet
Steven W. Benintendi
Marc K. Smith
The George W. Woodruff School of Mechanical Engineering
Georgia Institute of Technology
Atlanta, Georgia 30332-0405
Abstract
The effect of the slip coefficient and the mobility capillary number on the spreading of a thin
axisymmetric liquid droplet with uniform heating/cooling of the solid surface is examined. The results
show that increasing the slip coefficient reduces the spreading/shrinking behavior of the droplet and that
the final equilibrium states are slip dependent. These results are explained by the development of a return
flow inside the droplet. We show how a speed-dependent slip coefficient can be used to remove the
dependence of the final state on the slip coefficient. It is also shown that increasing the mobility capillary
number decreases the spreading/shrinking rate of the droplet. For thermocapillary-driven droplets, there
is a capillary-number-dependent time delay for the onset of motion. The entire effect of the mobility
capillary number on the spreading process is explained in terms of the deformability of the free surface.
PACS: 47.55.Dz, 68.15.+e, 68.10.Cr, 68.10.Gw
1
Introduction
The spreading of a liquid over a smooth solid surface is a complicated free-boundary problem
characterized by the presence of a moving contact line. A contact line is formed when three immiscible
material phases are in mutual contact, such as when a water droplet spreads on a glass substrate in an air
environment. This fundamental spreading process is of vital importance in a host of technological
processes such as coating, materials processing, and film-cooling applications. The motion of the liquid
during the spreading process is controlled by the fluid dynamics occurring in the immediate neighborhood
of the contact line. The importance of this contact-line region and the difficulties in modeling it have
been reviewed by Dussan
1
and de Gennes
2
. The main modeling difficulties stem from the fact that the
exact physics governing the behavior of the fluid in this microscopic region is unknown. Nevertheless,
several ad hoc assumptions can be made about the contact-line region and its effect on the overall motion
of the droplet in order to make the problem tractable in terms of continuum theory.
In classical fluid mechanics, the accepted boundary condition between a fluid and a solid surface is
the no-slip condition. However, in moving contact-line problems, the application of no-slip gives rise to a
non-integrable shear-stress singularity at the contact line
3 4 ,
. Several alternative methods have been
suggested to overcome this singularity, such as using a non-Newtonian description of the fluid near the
moving contact line or relaxing the no-slip condition by allowing the fluid to slip along the solid surface.
Imposing a slip law in the vicinity of the contact line is the most common method used to remove the
singularity and has been used extensively in spreading film and droplet geometries
5 10
.
Under static conditions, the solid-liquid-gas interactions in the vicinity of the contact line are
accounted for by prescribing the value of the contact angle, thereby providing a boundary condition for the
interface shape. As a natural extension to the static case, it is appealing to use a dynamic contact angle to
describe the interactions in the contact-line region under dynamic conditions. To this end, one approach
is to assume that the microscopic contact angle is always equal to its static value even under dynamic
conditions. In this approach, a local analysis near the contact line is used and matched to an outer
solution using asymptotic techniques. This approach has been taken by Hocking
6 7 ,
, Lowndes
11
, Cox
12
,
2
Dussan
13
, and others. Another approach is to assume a constitutive relation between the apparent
contact angle and the contact-line speed. Here the constitutive relation connects the mathematical
description of the outer region of the flow field to the contact-line speed. The precise micro-physics in the
contact-line region is contained in this relation and so a detailed analysis of this region is not required in
order to compute the flow for the bulk droplet. This approach is supported by the experiments of
Schwartz and Tejeda
14
, Hoffman
15
, Tanner
16
, and Chen
17
that demonstrate some dependence of the
dynamic apparent contact angle on the speed of the contact line. Greenspan
5
, Ehrhard and Davis
8
,
Haley and Miksis
9
, and Smith
10
have all used this approach to successfully compute the bulk droplet
behavior.
The spreading of an axisymmetric liquid droplet on a solid surface is a classic example of a geometry
that incorporates a moving contact line. Greenspan
5
was the first to examine the spreading of an
isothermal axisymmetric droplet due to a non-equilibrium initial shape. He adopted an analytical model
using a Maxwell slip law and a linear relationship between the apparent contact angle and the contact-line
speed. Using lubrication theory, a single evolution equation was derived in terms of the droplet shape and
was coupled to the dynamic contact-line boundary condition to describe the bulk motion of the droplet.
Under the limit of a small mobility capillary number, an asymptotic solution was found that described the
quasi-steady behavior of a spreading/retracting liquid droplet. Haley and Miksis
9
considered this same
spreading problem but with a goal of investigating various contact-line models. They did not invoke the
small mobility capillary number limit, but instead solved the full transient problem numerically. Their
results showed the dependence of the droplet spreading rates on the mobility capillary number, the
mobility exponent, and various formulations of the slip coefficient.
Ehrhard and Davis
8
considered non-isothermal effects by investigating an axisymmetric liquid
droplet spreading on a uniformly heated/cooled solid surface. As with Greenspan
5
these researchers
considered the quasi-steady limit of a small mobility capillary number and rightly set the slip coefficient to
zero. They showed that this type of thermal forcing induces a thermocapillary-driven flow that alters the
final equilibrium radius of the droplet. Since thermocapillarity causes fluid particles on an interface to
3
move from a hot region to a cold region, a heated solid surface establishes an interfacial flow from the
contact line toward the droplet center that decreases the final equilibrium radius of the droplet with
respect to the isothermal case. On the other hand, when the solid surface is cooled, an interfacial flow is
established from the droplet center toward the contact line that increases the final equilibrium radius of
the droplet with respect to the isothermal case.
In the current work, we shall extend the work of Ehrhard and Davis
8
and Haley and Miksis
9
by
examining the effect of the mobility capillary number and the slip coefficient on the transient spreading
behavior of a thermocapillary-driven axisymmetric droplet. Whereas Ehrhard and Davis
8
performed a
quasi-steady analysis emphasizing the final spreading radius as a function of the thermal forcing
parameter, we perform a fully transient analysis to determine the effect of the physical parameters on the
spreading rates under non-isothermal conditions. This essentially extends the work of Haley and
Miksis
9
to non-isothermal spreading. These thermal effects produce some interesting and counter-
intuitive results that we shall explain.
Problem Formulation
We follow the formulation of Ehrhard and Davis
8
by considering the motion of an axisymmetric
liquid droplet on a uniformly heated/cooled solid surface as shown in Fig. 1. A cylindrical coordinate
system with the radial direction embedded in the solid surface and the z-axis normal to the solid surface is
used. The droplet is composed of an incompressible Newtonian liquid with the density , thermal
conductivity k, dynamic viscosity , specific heat c
p
, and the unit surface thermal conductance h
g
all
considered constant. A passive gas at a temperature T

surrounds the droplet from above while a solid

surface held at a constant temperature T
0
bounds from below. The velocity
r
v u w ( , ) , pressure p, and
temperature T in the droplet are governed by the continuity equation, the Navier-Stokes equations, and the
energy equation,
4
r
r
v 0,
[ ]

r r
r
r r r
v v v p v g
t
+ + + ( )
2
,
[ ]
c T v T k T
p t
+
r
r
2
, (1)
where
r
g g ( , ) 0 is the gravitational vector.
The free surface of the liquid is defined by the function z h r t ( , ) . The boundary conditions imposed
on this surface are the kinematic condition, the normal- and tangential-stress conditions, and an energy
balance equation:
h uh w
t r
+ 0,
r r
n Sn
~
,
r
r
r
r
t Sn t
~
,
( )
( )

k T n h T T
g
r
r
, (2)
where
~
S is the liquid stress tensor, is the surface tension, and is the mean curvature of the free
surface. The normal and tangential unit vectors and the mean curvature of the free surface are defined as
r
n
h
h
r
r

+
( , )
( )
/
1
1
2 1 2
,
r
t
h
h
r
r

+
( , )
( )
/
1
1
2 1 2
,
( )

+ +
+
h
h
r
h
h
rr
r
r
r
1
1
2
2 3 2
( )
/
. (3)
To model thermocapillary effects, the surface tension is assumed to be a linear function of temperature
using the approximate equation of state
( )
0 0
T T , (4)
where
0
is the surface tension at the temperature T
0
, and > 0 is the negative of the rate of change of
surface tension with temperature.
The solid surface boundary conditions are a Navier slip law, a no-penetration condition, and a
specified surface temperature which can be greater than or less than the ambient temperature:
5
u u w
z r
+ ( ) , w 0 , T T
0
, (5)
where is the slip coefficient. Whereas Haley and Miksis
9
considered slip coefficients that were
constant, inversely proportional to the droplet height, and inversely proportional to the square of the
droplet height, this analysis will only consider constant slip coefficients.
The contact line is located at the point r c t ( ) with the contact condition and the value of the
apparent contact angle given by
h c t ( , ) 0 , h c t t
r
( , ) tan ( ) . (6)
We shall use a power-law form for the dynamic contact-line boundary condition that relates the apparent
contact angle to the contact-line speed U
cl
,
( )
( )
U
K
K
cl
A
m
A
R
m
R

>
<

'

,
,
, (7)
where
A
is the static advancing contact angle,
R
is the static receding contact angle, K is a constant,
and m is the mobility exponent. This constitutive relation was used previously by Ehrhard and Davis
8
,
Haley and Miksis
9
, and Smith.
10
We recognize that the use of a single constitutive relation for the dynamic contact angle, valid for all
capillary numbers and spreading characteristics (e.g. partial- or fully-wetting liquids, forced or non-
equilibrium motion, isothermal or non-isothermal surface conditions), is not very realistic. However, in
order to gain some qualitative understanding of the bulk spreading behavior of thermally-forced droplets,
we have elected to use the simple relation given by equation (7). This choice is reasonable and justified in
several ways. First, with a mobility exponent of m = 1, equation (7) reduces to that used by
Greenspan
5
which is a small-angle approximation for the expression derived by Blake and Haynes
18
6
and later experimentally verified by Schwartz and Tejeda
14
for small contact-line velocities. Secondly,
Ehrhard and Davis
8
showed that a boundary condition in the form of equation (7) with a mobility
exponent of m = 3 gives long-time asymptotic isothermal spreading rates consistent with the experimental
results of Tanner
16
, Cazabat and Cohen Stuart
19
, and Chen
17
. This choice of the mobility exponent also
provides agreement with the theoretically-predicted isothermal asymptotic spreading rates of Tanner
16
,
Starov
20
, and de Gennes
2
for fully-wetting liquids. Lastly, Ehrhard
21
performed experiments on
partially-wetting liquids spreading under non-isothermal conditions and found that the theoretical model
of Ehrhard and Davis
8
gave results that compared reasonably well with his experiment. (We note that in
his experiment, the constant K and the mobility exponent m where evaluated in an indirect manner thus
preventing equation (7) from being directly verified experimentally.) The above results show that the use
of the constitutive relation (7) with a mobility exponent of m = 3, provides a reasonable basis with which
to investigate the bulk spreading characteristics of liquid droplets, at least for a limited range of
experimental conditions. Thus, we feel it is appropriate to use such a condition for the droplet-spreading
problem considered here.
Finally, we apply a zero-slope condition at the droplet center, due to the symmetry of the
axisymmetric shape, and maintain a constant droplet volume V
0
since evaporative effects are neglected.
These conditions are:
h t
r
( , ) 0 0 , V h r t r dr
c t
0
0
2

( , )
( )
. (8)
These equations and boundary conditions (1)-(8), along with an initial droplet and contact-line shape,
dictate the motion of an axisymmetric droplet due to uniform heating /cooling of the solid surface.
The system is examined in the limit of a thin droplet using the scales of Smith
10
:
7
r
r
L
*
, z
*

z
L
s

, u
u
K
s
m
*

, v
v
K
s
m
*

,
t
t
L K
s
m
*

, p
p
K L
s
m
*

2
, T
T T
T T
*

0
,

s
. (9)
The length scale is based on the volume, L V
s
( )
/
0
1 3
, the velocity scale is based on the contact-line
speed, a convective time scale and a viscous pressure scale are used, and
s A
is a measure of the
contact angle. The dimensionless groups that result from this scaling are
R
K L
s
m

1
, B
h L
k
g s

, C
K
s
m

0
3
, G
gL

2
0
,
( )
C
K
T T
s
m

0
1
,

L
s
. (10)
These are the Reynolds number, Biot number, mobility capillary number, Bond number, thermocapillary
number, and the dimensionless slip coefficient, respectively.
The thin-droplet limit is invoked by taking
s
0 to obtain the lubrication equations with the
appropriate free-surface and solid-surface boundary conditions. The scaled governing equations to
leading order are
u
u
r
w
r z
+ + 0 , p u
r zz
, p
G
C
z
, T
zz
0, (11)
with free-surface and solid-surface boundary conditions
h uh w
t r
+ 0, +

_
,
p
C
h
h
r
rr
r
1
, ( ) u
C
T h T
z r r z

+
1

, T BT
z
u u
z
, w 0, T 1. (12)
8
Note we have suppressed any notation relevant to the asymptotic expansions used to produce these
equations. The solution to this problem in terms of the droplet height h and the relevant dimensionless
parameters is:
p
G
C
h z
C
h
h
r
rr
r
+

_
,

( )
1
, (13)
T
B h z
Bh

+
+
1
1
( )
, (14)
( ) u p
z
h z
B
C
h z
Bh
r
r
+

1
]
1
1
+
+
+
2
2
2
1

( )
( )
, (15)
( )
w p
p
r
z hz
h z p h
B
C Bh
h
h
r
Bh
Bh
z
z
rr
r
r r rr
r r
+

_
,

_
,
+
+
+
+

1
]
1
1

'

_
,

3 2
2
2 2
6 2
1
2
1 2

( )
. (16)
The pressure is controlled by a hydrostatic component, produced by gravity, and a component due to the
curvature of the interface, the temperature is conduction dominated, and the flow field is due to capillary-
pressure gradients and thermocapillary effects.
The evolution equation that governs the shape of the droplet is a height-averaged continuity equation
of the form h
r
rq
t r
+
1
0 ( ) , where q is the net radial flow in the droplet. For small droplets, we may
neglect the Bond number. Then by assuming that the Biot number is small, but still retaining it in the
thermocapillary term, we produce an evolution equation of the form
Ch
r
r h
h
r
h
h rMh
h
h
t rr
r
r
r
r
+ +

_
,
+

_
,

+ +

_
,

1
]
1
1

1
3 2
0
3
2
2

$
. (17)
Here
9
$
( )
M
CB
C
T T h L
k
g
s

0
0
, (18)
is an effective Marangoni number that is a ratio of the thermocapillary force to the mean surface tension
force. Note that this Marangoni number is proportional to the applied temperature difference between the
solid surface and the overlying passive gas. If
$
M > 0 , the solid surface is uniformly heated, and if
$
M < 0 ,
the solid surface is uniformly cooled. In this analysis we have assumed that
$
M is an order-one
parameter.
The boundary conditions and other constraints applied to the evolution equation (17) are zero film
thickness at the contact line, zero slope at the center, and constant volume:
( ) h c t , 0 , (19)
( ) h t
r
0 0 , , (20)
( ) h r t r dr
c
,
0
1
2

, (21)
The scaled constitutive relation governing the contact-line speed is
( )
( )
c
t
A
m
A
R
m
R

>
<

'

, (22)
where h c t t
r
( ( ), ) is the apparent contact angle. As discussed earlier, the micro-physics near the
contact line is contained in this constitutive relation (22) and we obtain the apparent contact angle from
the droplet profile as described by lubrication theory.
10
Solution Techniques
The droplet equations (17-22) were solved numerically using a combination of the backward Euler
method, the Newton-Kantorovich method, and the pseudo-spectral method. The time derivatives were
discretized using backward Euler. With this method, an equation such as h f h c
t
( , ) would be
discretized as
h h t f h c
i i i i + + +
+
1 1 1
( , ) , (23)
where the superscript refers to the discretized time coordinate and t is the time step. Although this is a
first-order method, it is very robust to instabilities and is easy to implement. We use a time step of 0.001
for most of the results shown, however, for the small-time results, a time step of 0.0001 is used.
Once equation (17) and (22) are discretized in time, the resulting two non-linear equations in h
i+1
and c
i+1
are linearized using the Newton-Kantorovich method. Here the non-linear equation N h c ( , ) 0
is expanded in a series using Frechet derivatives of the operator N as follows:
... c N h N ) c , h ( N ) c c , h h ( N
c h
+ + + + + . (24)
Subscripts on N refer to the appropriate Frechet derivative and refers to the change in the
corresponding variable. This reduces to the standard Newton iteration equation
) c , h ( N c N h N
c h
+ , (25)
for the change in the solution h and c . Equation (25) is solved with a standard pseudo-spectral
method using Chebyshev polynomials. We used 100 Chebyshev polynomials in all the results that follow.
Further details for all of the above techniques can be found in Boyd
22
and Canuto et al.
23
. Our reference
11
parameter set is C 01 . , 001 . , and
A R
1, with
$
M 1 for heating,
$
M 1 for cooling, and
$
M 0 for an isothermal droplet. We will vary both the mobility capillary number and the slip coefficient
in order to examine their effect on the spreading process of the droplet. Note that the effect of the heating
parameter
$
M on the final spreading radius of the droplet can be found in Ehrhard and Davis
8
and will
not be pursued in detail in this work.
Slip coefficient
We have used the Navier slip law of equation (5) to overcome the shear-stress singularity associated
with the moving contact line. Haley and Miksis
9
also used this model and showed that when an
isothermal droplet spreads due to a non-equilibrium initial condition, increasing the slip coefficient
increases the radius of the droplet during the entire spreading process. For easy reference, we duplicate
this result in Fig. 2 in terms of our reference parameter set. Note that the slip parameter values we have
used are unrealistically large in order to emphasize their effect. The increased spreading behavior seen
here is the result of a radially-outward volume flux that increases with increasing slip for a spreading
droplet. This is easily understood since the positive wall shear stress for a spreading droplet produces a
radially-outward slip velocity on the solid surface increasing the slip coefficient increases this slip
velocity and thus the radially-outward volume flux. The extra volume flux gives an extra outward push to
the contact line resulting in a larger droplet during the spreading process. Likewise, for a shrinking
droplet the effects are reversed and the droplet shrinks faster when the slip coefficient is increased. For
this non-equilibrium system, slip only affects the spreading or shrinking rates of the droplet it does not
affect the final equilibrium radius, for which the droplet is stationary. Integrating the time-independent
form of the evolution equation (17) with
$
M 0 and imposing a zero net radial volume flux shows that
the stationary equilibrium state of the droplet is independent of the slip coefficient , as expected.
To examine the effect of slip on the dynamics of thermocapillary-forced spreading, we used the
isothermal equilibrium shape as the initial condition. This insures that any motion of the droplet is due to
12
thermal forcing. Fig. 3a shows the effect of the slip coefficient on thermocapillary-induced spreading.
We see here that for cooled (heated) droplets, increased slip actually reduces the spreading (shrinking)
and decreases (increases) the final equilibrium radius of the droplet, in contrast to the effect of the slip
coefficient in an isothermal system. The spreading behavior is a surprising result in that our physical
intuition, built from the non-equilibrium spreading case, says that increased slip should enhance
spreading or shrinking. Figure 3b shows that at small times, increasing the slip coefficient does increase
the spreading radius of a cooled droplet, as we would expect, but that this effect is soon reversed when the
curves cross. Similar trends in the opposite direction can be observed for a heated droplet. To clearly see
the effect of the slip coefficient on the final equilibrium droplet radius, the steady form of the evolution
equation is solved for various values of the slip coefficient in Fig. 4. The physical explanation of this
behavior centers on the emergence of a return flow inside the droplet as it spreads. We will first explain
these results for the spreading process and then show how the same physics explains the observed slip-
dependent final equilibrium states.
The initial condition we used to generate Fig. 3 is the isothermal equilibrium droplet shape in which
there is no capillary-pressure gradient or contact-line mobility to cause fluid motion. As a result, when
surface-tension gradients are induced on the free surface by cooling (heating) the solid surface, the initial
fluid motion is an outward (inward) uni-directional thermocapillary shear flow and slip has the same
effect as in the non-equilibrium case, i.e., increasing the slip coefficient increases the spreading
(shrinking) of the droplet. As the droplet spreads (shrinks), a capillary-pressure gradient forms that tends
to push the liquid away from (toward) the contact line. As this capillary-pressure gradient increases, the
velocity profile changes from being uni-directional to having a return flow, and so there is a reversal of
the wall shear stress and the slip velocity. This evolution of the velocity profile is shown in Fig. 5 for a
cooled droplet. As the wall slip velocity changes from radially outward to radially inward, an increase in
the slip coefficient decreases the radially-outward volume flux. This decrease in flux slows the motion of
the contact line which reduces the spreading of the droplet. Likewise for a heated droplet, the flow
directions are reversed and increasing the slip coefficient reduces shrinking by reducing the radially-
inward volume flux.
13
The evolution equation (17) clearly shows that when the system is thermally forced, the final time-
independent equilibrium state for which there is a motionless contact line is still affected by the slip
coefficient. Figure 4 indicates that when the droplet is cooled (heated), increasing the slip coefficient
decreases (increases) its final equilibrium radius. For a thermally-forced droplet in its equilibrium
position, the thermocapillary-driven surface flow is balanced by a capillary-pressure-driven return flow.
Imagine that the return-flow profile in Fig. 5 is for a cooled droplet in its equilibrium shape with some
value of the slip coefficient and that there is no contact-angle hysteresis. If the slip coefficient were
increased, the inward volume flux due to the capillary-pressure gradient would increase over the outward
flux due to thermocapillarity and the droplet would have a net inward radial volume flux. This would
drain fluid away from the contact line causing the contact angle to decrease and consequently causing the
droplet radius to decrease in order to regain its equilibrium shape. A similar argument can be made for a
heated droplet. Thus, the development of a return flow in the droplet explains the observed effect of slip
on the spreading process and on the final equilibrium droplet shape.
As mentioned earlier, slip models are used to remove the shear-stress singularity that appears at a
moving contact line. Typically, the effect of these slip models is localized near the contact line. They have
little effect on the bulk motion of the fluid. As a corollary to this, when the contact line is motionless, slip
should not be a part of the flow physics. However, the results we have just explained show that the final
equilibrium radius of a thermocapillary-forced droplet depends on the slip coefficient even though the
contact line is motionless at this time. Despite the fact that this behavior is properly explained using the
physics built into the lubrication model of the spreading process, the behavior itself is unphysical. Since it
is due to the interaction of the dynamics of the droplet and the slip model that was used, we must conclude
that the slip model itself is defective.
One way to address this kind of a defect in the Navier slip model is to modify the slip coefficient so
that it depends on the speed of the contact line in such a way that the slip coefficient is zero when the
contact line is stationary. While the precise form of this dependence is uncertain at this time, the
existence of a speed-dependent slip parameter has been recognized and discussed by Marsh, Garoff, and
Dussan
24
.
14
To demonstrate this idea, consider a slip coefficient that depends linearly on the speed of the contact
line, i.e.,
c
t
. (26)
This model was chosen so as to remove the observed defect in the Navier slip law. It is not, at this time, a
fully justified slip law. A comprehensive derivation of a slip law with a similar property still needs to be
done. We computed the response of a cooled droplet using the slip law (26). The results are compared to
the constant-slip case by choosing such that the maximum value of the slip coefficient in Eq. (26)
corresponds to the constant-slip value. The results shown in Fig. 6 indicate that the speed-dependent-slip
model works as intended. It produces results that are qualitatively similar to the constant-slip results, and
in addition, shows a final radius that is appropriate for a droplet with no moving contact line. We also see
that the results are much less sensitive to the actual value of the speed-dependent slip coefficient than is
the case for the constant-slip model. Since the actual value of the slip coefficient is difficult to determine,
this lack of sensitivity is appealing in that we do not have to determine the slip coefficient very accurately.
Furthermore, these results reinforce the physical argument put forth earlier concerning the development of
a return-flow profile in the droplet and its effect on the spreading rate. A small-time analysis of Fig. 6
shows a crossover pattern just as in Fig. 3b. Indeed, this is an expected result in that the return-flow
profile develops regardless of which Navier slip model is used and the reversal in the effect of slip is a
direct consequence of this flow pattern. However, this mechanism no longer has any effect on the final
equilibrium radius in the speed-dependent slip model because the slip coefficient is zero when there is no
contact-line motion.
In a final comparison, Fig. 6 also shows spreading results for a droplet with no slip, i.e., 0.
These computations are possible because the shear-stress singularity at the moving contact line is relieved
by truncation in the numerical solution. These results are of interest because they show the spreading
15
process is not significantly affected by the size of the slip coefficient when a more appropriate Navier slip
model is used. This is evident in Fig. 6 since the 0 and the 1 3 / curves are nearly identical.
Mobility capillary number
An examination of the evolution equation (17) shows that the mobility capillary number does not
affect the final equilibrium state of the droplet, but that it does affect the spreading rate. Haley and
Miksis
9
showed that for non-equilibrium spreading of an isothermal droplet, increasing the mobility
capillary number decreases the radius of the droplet during the entire spreading process and increases the
total spreading time. Again for easy reference, we duplicate this result in Fig. 7 in terms of our reference
parameter set. These same researchers also derived an asymptotic representation of the spreading rate for
C 0 as
( )
c V c
t A
8
3
3
, where V is the volume of the drop. The solution of this ODE is also
plotted in Fig. 7.
Figure 8a shows that the effect of increasing the mobility capillary number also decreases
thermocapillary-driven spreading or shrinking. However, one interesting feature of this flow not
previously seen is a capillary-number-dependent time delay for the onset of spreading or shrinking. This
delay, shown in detail for small times in Figure 8b, increases as the mobility capillary number increases.
The time delay, as well as the entire effect of the mobility capillary number on the spreading process,
is a result of the deformation of the free surface. The capillary number is a ratio of viscous forces to
surface tension forces and is a measure of the deformation of the free surface, as explained by Davis and
Homsy
25
. Small mobility capillary numbers indicate little free-surface deformation. When a droplet is
not in equilibrium, either due to initial conditions or thermal forcing, the resulting interior flows are
accommodated by deformation of the free surface, motion of the contact line, or a combination of the two.
When the mobility capillary number is small, the contact line moves easily because the free surface does
not deform very much in response to the interior motion of the droplet. When the mobility capillary
number is large, the free surface deforms easily and so the contact-line motion is not affected as much.
16
When a temperature field is imposed on a droplet in equilibrium, the resulting thermocapillary flow is
accommodated first by the deformation of the free surface away from the contact line. After a short time,
the free surface near the contact line begins to deform, the contact angle changes, and the contact line
begins to move. The more deformation the droplet can withstand, i.e., the larger the mobility capillary
number, the longer the time delay for contact-line motion, as shown in Fig. 8b. This behavior is not
observed in isothermal non-equilibrium droplets because the initial condition has a contact angle greater
than the advancing contact angle and so spreading starts immediately.
Conclusion
We have examined the effect of the slip coefficient and the mobility capillary number on the
spreading process of a thermocapillary-driven liquid droplet on a uniformly heated or cooled solid surface.
The results were compared with previous work by Haley and Miksis
9
and Ehrhard and Davis
8
and
explained in terms of the physics built into the lubrication model used in this work. Our most interesting
result, shown in Fig. 3, is how the slip coefficient influences the motion of the contact line. For
isothermal droplets, increasing slip increases the contact-line speed and thus the rate of spreading.
However, the final equilibrium radius of the droplet is unaffected by slip. For a thermocapillary-driven
droplet uniformly cooled from below, increasing slip causes the contact-line speed to first increase and
then decrease. In addition, increased slip decreases (increases) the final equilibrium radius of a uniformly
cooled (heated) droplet. The physical inconsistency of this last result was removed by modifying the slip
coefficient to depend on the speed of the contact line. This resulted in spreading behavior that was similar
to the constant-slip-coefficient case, but now the final equilibrium state is independent of slip. We
conclude that the Navier slip model with constant slip does not do a very good job in capturing the physics
for these thermally-forced droplet motions. Alternative slip models, such as the one suggested here, or
other contact-line formulations, such as the one suggested by Shikhmurzaev
26 27 ,
, should be further
explored to see if these models can adequately describe contact-line motion in thermally-forced systems.
17
References
1
E.B. Dussan, "On the spreading of liquids on solid surfaces: static and dynamic contact lines," Ann.
Rev. Fluid Mech., 11, 371 (1979).
2
P.G. de Gennes, "Wetting: statics and dynamics," Rev. Mod. Phys., 57, 827 (1985).
3
C. Huh, L.E. Scriven, "Hydrodynamic model of steady movement of a solid/liquid/fluid contact line," J.
Colloid Interface Sci., 35, 85 (1971).
4
E.B. Dussan, S.H. Davis, "On the motion of a fluid-fluid interface along a solid surface," J. Fluid
Mech., 65, 71 (1974).
5
H.P. Greenspan, "On the motion of a small viscous droplet that wets a surface," J. Fluid Mech., 84, 125
(1978).
6
L.M. Hocking, "Sliding and spreading of thin two-dimensional drops," Q.J. Mech. Appl. Maths., 34, 37
(1981).
7
L.M. Hocking, "The spreading of a thin drop by gravity and capillarity," , Q.J. Mech. Appl. Maths., 36,
55 (1983).
8
P. Ehrhard, S.H. Davis, "Non-isothermal spreading of liquid drops on horizontal plates," J. Fluid
Mech., 229, 365 (1991).
9
P.J. Haley, M.J. Miksis, "The effect of the contact line on droplet spreading," J. Fluid Mech., 223, 57
(1991).
10
M.K. Smith, "Thermocapillary migration of a two-dimensional droplet on a solid surface," J. Fluid
Mech., 294, 209 (1995).
11
J. Lowndes, "The numerical simulation of the steady movement of a fluid meniscus in a capillary tube,"
J. Fluid Mech., 101, 631 (1980).
12
R.G. Cox, "The dynamics of the spreading of liquids on solid surface. Part 1. Viscous flow," J. Fluid
Mech., 168, 169 (1986).
13
E.B. Dussan, "The moving contact line: the slip boundary condition," J. Fluid Mech., 77, 665 (1976).
18
14
S.W. Schwartz, S.B. Tejeda, "Studies of dynamic contact angles on solids," J. Colloid Interface Sci.,
38, 359 (1972).
15
R.L. Hoffman, "A study of the advancing interface. Interface shape in liquid-gas system," J. Colloid
Interface Sci., 50, 228 (1975).
16
L.H. Tanner, "The spreading of silicone drops on horizontal surfaces," J. Phys. D: Appl. Phys., 12,
1473 (1979).
17
J.-D. Chen, "Experiments on a spreading drop and its contact angle on a solid," J. Colloid Interface
Sci., 122, 60 (1988).
18
T.D. Blake, J.M. Haynes, "Kinematics of liquid/liquid displacement," J. Colloid Interface Sci., 30, 421
(1969).
19
A.M. Cazabat, M.A. Cohen Stuart, "Dynamics of wetting: effects of surface roughness," J. Phys.
Chem., 90, 5845 (1986).
20
V.M. Starov, "Spreading of droplets on nonvolatile liquids over a flat surface," Colloid J. USSR (Eng.
transl.), 45, 1009 (1983).
21
P. Ehrhard, "Experiments on isothermal and non-isothermal spreading," J. Fluid Mech., 257, 463
(1993).
22
J.P. Boyd, Chebyshev and Fourier Spectral Methods, Lecture Notes in Engineering, 49, ed. C.A.
Brebbia and S.A. Orszag (Springer-Verlag, Berlin, 1989).
23
C. Canuto, M.Y. Hussaini, A. Quarterone, T.A. Zang, Spectral Methods in Fluid Dynamics (Springer-
Verlag, New York, 1988).
24
J.A. Marsh, S. Garoff, E.B. Dussan, "Dynamic contact angles and hydrodynamics near a moving
contact line," Phys. Rev. Lett. 70, 2778 (1993).
25
S.H. Davis, G.M. Homsy, "Energy stability theory for free-surface problems: bouyancy-thermocapillary
layers," J. Fluid Mech., 98, 527 (1980).
26
Y.D. Shikhmurzaev, "The moving contact line on a smooth solid surface," Int. J. Multiphase Flow, 19,
589 (1993).
19
27
Y.D. Shikhmurzaev, "Moving contact lines in liquid/liquid/solid systems," J. Fluid Mech., 334, 211 (1997).
20
Figure Captions
Figure 1. An axisymmetric liquid droplet on a uniformly heated/cooled solid surface.
Figure 2. The effect of the slip coefficient on non-equilibrium spreading, first shown by Haley and
Miksis
9
. Increasing slip increases the radius of the droplet during the entire spreading process. The
equilibrium radius is c

( / ) .
/
4 1084
1 3
. A similar trend is seen for non-equilibrium shrinking. The
parameter set is C 01 . ,
A R
1, and
$
M 0 .
Figure 3. (a) The effect of the slip coefficient for thermocapillary-induced motion. Note the slip-
dependent final equilibrium states. (b) The small-time behavior shows a reversal in the effect of slip for
the cooled case. Similar behavior is observed in the heated case. The parameter set is C 01 . and

A R
1.
Figure 4. Final equilibrium radius as a function of the heating parameter
$
M and the slip coefficient .
Increased slip decreases (increases) the final equilibrium radius for a cooled (heated) droplet.
Figure 5. The temporal development of the radial velocity profile in a cooled droplet. The initial profile
is linear, it then passes through a point with zero shear stress at the solid, and then a return flow appears.
For heated droplets, the flow directions are reversed.
Figure 6. Spreading of a cooled droplet using a speed-dependent slip coefficient as compared to a
constant slip coefficient. The spreading behavior of the two models is similar, as seen by the two curves
with the same line type, but the final radius is independent of the slip coefficient for the speed-dependent
slip law. The 0 curve is also plotted and shows slip is not very significant when a speed-dependent
slip law is used. The parameter set is C 01 . ,
A R
1, and
$
M 1 .
21
Figure 7. The effect of the mobility capillary number on non-equilibrium spreading, first shown by Haley
and Miksis
9
. Increasing the mobility capillary number decreases the spreading rate of the droplet. The
same trend is seen in non-equilibrium shrinking. The parameter set for the numerical results is 001 . ,

A R
1, and
$
M 0 .
Figure 8. (a) The effect of the mobility capillary number on thermocapillary-driven spreading or
shrinking. (b) The small-time behavior of the motion in Figure 8a. The parameter set is 001 . and

A R
1.
22
T
o
n
t
g
z
r
Liquid
Solid

z = h (r,
Gas T
8
t)
,
Figure 1, Benintendi, Physics of Fluids
23
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
0.01
1.0
c ( t )
t
Figure 2, Benintendi, Physics of Fluids
24
0.95
1
1.05
1.1
1.15
1.2
1.25
0 10 20 30 40 50
0.01
1.0
0.1
0.01
1.0
0.1
M = 1.0
M = -1.0
^
^
c ( t )
t
( a )
Cooled Droplet
Heated Droplet
1.0835
1.084
1.0845
1.085
1.0855
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
0.01
1.0
0.1
c ( t )
t
M = -1.0
^
( b )
Figure 3, Benintendi, Physics of Fluids
25
0.95
1
1.05
1.1
1.15
1.2
1.25
1.3
-1 -0.5 0 0.5 1
0.01
0.1
0.0
1.0
r
steady
M
^
Figure 4, Benintendi, Physics of Fluids
26
Figure 5, Benintendi, Physics of Fluids
27
1.08
1.1
1.12
1.14
1.16
1.18
1.2
1.22
1.24
0 10 20 30 40 50
0.027
1.0
0.01
c ( t )
t
0.01
0.027
1/3
Constant slip Speed-dependent slip
1/3
1.0
Line
0.0
Figure 6, Benintendi, Physics of Fluids
28
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
1.05
0 0.5 1 1.5 2 2.5 3 3.5
t
c ( t )
C = 1.0
C = 5.0
C = 0.1
C = 0
Numerical
Asymptotic
Figure 7, Benintendi, Physics of Fluids
29
0.95
1
1.05
1.1
1.15
1.2
1.25
0 10 20 30 40 50
c ( t )
t
C = 0.1
C = 0.1
C = 5.0
C = 1.0
C = 1.0
C = 5.0
M = 1.0
M = -1.0
^
^
( a )
Cooled Droplet
Heated Droplet
1
1.05
1.1
1.15
0 0.5 1 1.5 2 2.5 3 3.5
c ( t )
t
C = 0.1
C = 0.1
C = 5.0
C = 1.0
C = 1.0
C = 5.0
M = 1.0
M = -1.0
^
^
( b )
Cooled Droplet
Heated Droplet
Figure 8, Benintendi, Physics of Fluids