Evaporating Menisci of Wetting Fluids

STEVEN MOOSMAN AND G. M. HOMSY

Department o f Chemical Engineering, Stanford University, Stanford, California 94305

Received F e b r u a r y 18, 1979; accepted June 4, 1979

We model the transport p r o c e s s e s occurring in a horizontal evaporating m e n i s c u s which is af-

fected by both capillarity and by multilayer adsorption. Perturbation theory is used to describe the
profile change relative to the static isothermal profile. We solve for this deviation profile, as well as
for the evaporative flux from the interface, for a range of values appropriate to the n o n d i m e n -
sional parameters in the model. The results clearly d e m o n s t r a t e that large heat fluxes can occur in
the transition region between the capillary m e n i s c u s and the adsorbed layer. Moreover, the devia-
tion profile predicts a thinning in the m e n i s c u s region with a shift in the apparent zero toward the gap,
a m u c h thinner nonevaporating adsorbed film, and an increase in apparent contact angle.

INTRODUCTION at the same conclusion in studying heat

transfer characteristics of menisci formed at
Evaporation from a meniscus which wets the exit of a capillary tube. Renk and Way-
a heated solid surface is of critical im- net (3-5) used optical interferometry
portance in many heat transfer processes to determine the profile of an evaporating
which rely on the latent heat of vaporiza- meniscus formed at the exit of the gap be-
tion of a liquid to provide high localized heat tween two horizontal fiat plates. By fitting
fluxes, e.g., finned or grooved evaporators their experimental data to a fourth-order
and heat pipes. While it is well known that polynomial and differentiating, they con-
systems of this type can exhibit high heat cluded (3) that significant increases in curva-
transfer rates, particularly in the transition ture accompanied increases in the heat flux
layer between the liquid meniscus and the and that the pressure gradient for flow was
wetting film (1), the quantitative descrip- associated with a gradient in the meniscus
tion of the coupling between fluid flow, heat curvature.
transport, and evaporation in this region, Deryaguin and co-workers (6), using the
and the determination of the nondimen- dependence of disjoining pressure on film
sional physical parameters which most af- thickness, have argued that a gradient in
fect the heat flux remain problems of con- the disjoining pressure brought about by a
siderable interest. changing film thickness could yield an
Wayner and co-workers (1-5) have pub- evaporation rate from capillaries several
lished a series of papers on experimental times that based solely on diffusion. Later
observations of menisci during evaporation. work by Churaev and Ershova (7) employed
Wayner and Coccio (1) noted that the height this model to account for a higher evapora-
of an evaporating meniscus formed on a tion rate in hydrophilic capillaries than that
vertical flat plate was less than that in the observed in hydrophobic capillaries (in
isothermal case and concluded that fluid which no adsorbed film and hence no film
flow to the base of the meniscus was asso- transport occurs).
ciated with an increase in the average radius Mathematical analyses of these phe-
of curvature. Preiss and Wayner (2) arrived nomena are still in the stage of develop-

212
0021-9797/80/010212-12502.00/0
Copyright © 1980by AcademicPress, Inc.
All rights of reproduction in any form reserved. Journalof Colloidand InterfaceScience, Vol. 73, No. 1, January 1980
 EVAPORATING MENISCI 213

ment. Potash and Wayner (8) have modeled Wayner and co-workers (10, 11) have also
the transport processes occurring in an sought to model fluid flow in the transition
evaporating meniscus formed on a heated region based on a gradient in the disjoining
vertical flat plate immersed in carbon tetra- pressure. Numerical results for the profile
chloride. Fluid flow was described by a near the adsorbed layer indicated, how-
simple wedge flow model, with the pressure ever, that curvature effects become im-
gradient near the adsorbed layer based on a portant at small distances from the film.
disjoining pressure gradient and that in the Based on these results, these authors echo
meniscus based on a curvature gradient. the suggestion (3, 8, 9) that a complete fluid
While the presence of the adsorbed super- flow model should account for the presence
heated layer in this model allowed for a of a transition region in which both dis-
smooth transition between evaporating and joining pressure and curvature effects are
nonevaporating portions of the meniscus, significant.
the computed profiles exhibited sharp dis- We note that the concept of a transition
continuities in slope due to the abrupt region between a curvature controlled
transition from one model to the other. meniscus and an adsorbed thin film has
They suggested that, while the proposed recently been exploited (12, 13) in deter-
model may adequately represent many of mining the shape of isothermal nonevapor-
the features observed in experiment (1, 2) ating menisci. Derjaguin et al. (12) analy-
and can account for the required fluid flow tically determined the profile in the transi-
in the meniscus (curvature controlled) re- tion zone between a horizontal capillary
gion, a model which includes a transition re- meniscus and a wetting film. By numerical
gion in which both disjoining pressure and evaluation of the solution, they noted that
curvature effects are present might better the extent of the transition zone increased
represent the actual profile. with increasing thickness of the adsorbed
Renk and Wayner (3, 4, 9) have used film. By employing the method of matched
experimentally determined profiles and a asymptotic expansions (14), Renk et al. (13)
model for fluid flow based solely on curva- were able to determine the width of the
ture effects to compute the fluid flow and transition layer based on the physical
heat flux distribution near the transition parameters of the system. The results were
layer of a horizontal evaporating meniscus. consistent with those of Derjaguin et al. (12)
Difficulties implicit in differentiating the ex- and the asymptotic treatment aided in
perimental profiles three times (3) were later elucidating the competing physical effects of
circumvented (4, 9) by assuming a form for curvature and disjoining pressure.
the volumetric flow rate and using the data The purpose of this paper is threefold.
to evaluate the adjustable parameters in the First, we model the transport processes
model. They concluded that, while the which occur in a specific geometry, namely,
general characteristics of the model seemed a horizontal evaporating meniscus formed
reasonable, the predicted location of the at the exit of the gap between two horizontal
intersection of the capillary meniscus and flat plates. This is the same geometry stud-
the adsorbed film differed appreciably from ied experimentally by Renk and Wayner
the measured value in several instances. (3-5). Second, we wish to solve for the
This was attributed to the neglect of thin shape of an evaporating meniscus for which
film effects (disjoining pressure) in the all relevant physical effects have been
transition region. They also note that the na- accounted. Specifically, we study the nature
ture of their model (fitting a microscopic of the transition between an evaporating
model to macroscopic data) admits poten- meniscus relatively unaffected by thin film
tially large errors in this region. forces and multilayer adsorption of a wet-
Journal of ColloM and Interface Science, Vol. 73, No. 1, January 1980
214 MOOSMAN AND HOMSY

ting fluid on a heated solid. We do so in a LIQUID FILM ]

rigorous manner, without an artificial
separation of the meniscus into regions , /
,, //
where one or another physical effect may be T=T w >l ,,/
neglected. It is convenient to use perturba- lh
~o
tion theory (14) to describe the change in
shape of the evaporating meniscus relative T=%
to the static isothermal profile, where the
latter is the same as that determined by Renk
LIQUID POOL
et al. (13). We show that the evaporative
flux displays a sharp maximum in the transi-
tion layer and vanishes in the thin film re-
FIG. 1. Definition sketch.
gion. Similarly, we show that the change in
the meniscus profile is a maximum in the
transition region and is strongly affected by fluid flow are balanced by a reduction in
the disjoining pressure in the adsorbed fluid pressure due to curvature and by dis-
layer. By a limited parameter study, we then joining pressure, with 6zrA being the Ha-
analyze the relative effects of the various maker constant. This expression differs
dimensionless groups on the meniscus pro- from that considered by Renk et al. (13) only
file and evaporative flux and discuss the by the inclusion of the fluid stress term
range of validity of the results. h .o-.h. We note that by including the hydro-
static component of the pressure pgh in the
BASIC EQUATIONS
suction term, the stress o- is defined as the
Model Definition and Scaling stress field associated with the flow due to
evaporation.
The geometry of interest is depicted in We note that the form of the disjoining
Fig. 1. An evaporating two-dimensional pressure in [1] neglects a number of poten-
meniscus is formed at the exit of the gap be- tially important physical effects, including
tween two horizontal fiat plates. The menis- the possible formation of diffuse electric
cus is assumed to be in equilibrium with double layers in ionic fluids and the effect of
an adsorbed layer in which the pressure is electromagnetic retardation in thicker films
governed by L o n d o n - v a n der Waals disper- (15, 16). Moreover, the inverse cube rela-
sion forces. Hydrostatic suction is imposed tionship of disjoining pressure to thickness
by a liquid pool a distance L below the up- is derived only for planar films (15). With
per plate surface. The upper wall tempera- regard to the latter objection, we note that
ture Tw is assumed constant and is elevated recent isothermal analyses (12, 13) indicate
with respect to the uniform gas temperature that thin film effects become important only
Tg. Under these conditions, the steady state
in the transition layer and the adsorbed film.
profile h(x) is determined from a normal In both these regions, the meniscus slope
stress balance at the interface. Neglecting is negligible and the assumption that the
the density of the gas relative to the density disjoining pressure can be based on a model
of the liquid, this may be written: appropriate to planar films is justified. In
pg(L - h) + h.~r.h addition, electromagnetic retardation ef-
fects appropriate for thicker films become
d2h / [ ( dh A t2]3'2
= T-07/[ 1 + (22] + [1] insignificant for fluid films which are less
V"
than approximately 100 & in thickness (17,
In this expression, we see that the com- 18) provided the films exceed a monolayer
bined effects of hydrostatic suction and or so of coverage. It would thus appear that
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
 EVAPORATING MENISCI 215

while inappropriate to the curvature con- of viscous force to capillary force and will
trolled portion of the meniscus (where thin be an order one quantity.
film effects are insignificant anyway), the For a constant property, Newtonian fluid,
model of the disjoining pressure in [1] ade- the stress tensor takes the form:
quately represents the thin film effects in the
cri~ = - p l i j + 21JEij
regions where these effects are significant,
where
provided that the adsorbed film is thin
(~<100 ]k) and we restrict ourselves to non-
l ( Ou~ Ouj)
I~ = 6~; Eij = - ~ \ Oxj + Ox~ . [7]
polar fluids. Extension of the analysis to
thicker films with isotherms of the form B~ This can be made nondimensional with
h 4 is clearly possible. respect to the characteristic pressurep0. The
The normal stress balance can be non- resulting nondimensional expression for the
dimensionalized once the dependent and in- normal stress balance [1] then becomes:
dependent variables are scaled. We begin
62
by defining:
1- a~q(() - C , ( P + ( 1 + ~(~,)2))
x =x0~:; Y =h0Y; h ( x ) =h0~(~) [2]
where the characteristic length x0 is selected ×
in order that the characteristic curvature
balances the hydrostatic suction in the capil- = ~/'/(1 + 6e(~')e)3/e + ea/~)3 [8]
lary meniscus region, i.e.,
where ' = d/d~. We neglect terms propor-
vho/xg = o g L ~ Xo = ( y h o / p g L ) 1/2. [31 tional to a and 6 2 and consider the first order
Then, consistent with [2] and the continuity expression for [8]:
equation: 1 - C1P(~) = a~" + e3/-03. [9]
Ou Ov
- - + - - = 0, [4] This expression can be solved for the profile
Ox Oy
~(~:) provided the pressure variation P ( O
we scale the pressure and velocities: due to evaporative flow is known. The de-
termination of this pressure profile, how-
u = uoU; v = (ho/xo)uoV; p = poP. [5]
ever, requires that we solve a coupled fluid
The characteristic pressurep0 is selected on flow, heat transfer problem which we now
a viscous scale with p0 = tZUoxo/h~ and the consider.
scaling u0 is left undetermined at this point.
The following nondimensional param- Fluid Mechanics
eters are now introduced: Under the scalings in [2, 5], the nondimen-
pgho
sional N a v i e r - S t o k e s equations become:
a- - - - ho/L ; 6 = ho/xo;
~holx~ OP 02 U
-- - -- + O(RM) + 0 ( 8 2) [10]
0~ Oy ~
ea _ A/h3o ," C1 _ P o _ __tzu°x° [6]
pgL pgL pgLh~ OP
- 0(82) + 820(RM). [11]
OY
We note that here ~ is a Bond number, ex-
pressing the ratio of body force to capillary The pressure here is the pressure associated
force. 6 is the characteristic slope, and d is with fluid flow, the effect of hydrostatic
the ratio of dispersion force to hydrostatic variation in the pressure having already
force. In cases of practical interest a, 6 and been included in [1]. The nondimensional
e3 are typically small. C1 represents the ratio parameter RM, defined as:
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
216 MOOSMAN AND HOMSY

RM = ~2R where R = pUoXo/lX [12] lem of determining the temperature profile

in the liquid and, particularly, to expressing
is a modified Reynolds number which ap-
the interfacial temperature in terms of the
pears commonly in lubrication problems
normal velocity F and wall temperature Tw.
involving flow in narrow channels. RM in
such situations is typically small and terms
of this order can be neglected. The pressure E n e r g y Transfer
and fluid flow may then be ,determined by Nondimensionalizing the temperature
solving the continuity Eq. [4] and the mo- with respect to the gas temperature Tg:
mentum equation:
T = TgO [16]
OP 02U
[131
and assuming the velocity and coordinates
- - - - -

O~ OY 2
scale as in [2, 5], the steady-state thermal
under the appropriate no slip condition at energy equation for a constant property
Y = 0 and no tangential stress condition Newtonian fluid can be written:
at the free surface, Y = ,/((). We spe-
cifically neglect thermocapillary convection 020
- O(PrRM) + 0 ( 8 2)
driven by the coupling between the gradient Oy 2
in surface temperature and the variation of
surface tension with temperature. For a con- where now the Prandtl number appears mul-
stant wall temperature, the thermocapillary tiplying RM. Since the Prandtl number is
stress will drive a flow opposite to that typically of order one for most fluids and
computed here. Thus the results reported since RM and 82 are assumed to be much less
represent upper limits to the strength of the than unity, the temperature field may be
evaporative flux. The resulting expressions computed from a simple locally one-dimen-
for the velocity distribution are: sional heat conduction model which ne-
glects convection, and conduction in the
longitudinal direction.
The resulting linear temperature distribu-
tion is given by:
V=P'~'--2 + P" ~ 2 6 . [14]
0=0w+
{\ 0~(D - 0,v)Y
~(~ [17]
From these, the volumetric flow rate Q and
the normal velocity at the interface (F
= - Q ' ) m a y be calculated: where 0w is the nondimensional wall temper-
ature (Tw/Tg) and 0i is the nondimensional
Q = _ __P'~3 ," r - --(P'~/~)' [151 and as yet unknown interfacial temperature
3 3 (Ti/Tg) which we wish to express in terms
of the normal velocity F and wall temper-
We see that the velocity distribution, fluid
ature (0w). Continuity of heat flux at the
flow rate, and normal velocity at the inter-
interface requires that the heat conducted to
face can be expressed in terms of the inter-
the surface be removed by convection to the
facial position ~(~:) and the pressure distri-
gas and by evaporation from the surface.
bution P(~:). Determination of the latter,
Thus we have:
however, requires that we consider the
balance between evaporation from the inter- Ow - - 0i(~: )
face and the normal velocity at the surface. - C2F
But, since the former depends on the inter-
facial temperature, we first turn to the prob- + Nu(Oi(() - 1) + 0(82) [18]
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
 EVAPORATING MENISCI 217

where the Nusselt number Nu and the non- by disjoining pressure [9], while the temper-
dimensional group C2 are defined: ature is increased from Tg at the pool to Ti
at the meniscus interface. As a consequence,
hho oH~uo
Nu- ; C2- [19] the equilibrium vapor pressure computed
k kTg/ho from [22] is given by:
The Nusselt number represents the ratio of 1 -Y
interfacial convection of conduction in the In
liquid and may be assumed to be negligible Pro /
for small conduction pathlengths h0. C2,
however, represents the ratio of heat trans- + RTi " T-g . [23]
port by evaporation to heat transport by
conduction and must be retained. Neglect- The reduction in vapor pressure above the
ing terms in Nu and 82, we can express liquid pool can be computed from:
the interfacial temperature 0~(~) in terms of
the meniscus profile ,/(~) and the normal In ~0v0 - pt?Tg [24]
velocity at the free surface F.
Since it is to be expected that Wv and Pv will
0i(~) = 0w - (C2~)F. [20]
be close to the value at the pool surface (Pv0),
The final relation we require is the balance the logarithms in these two expressions are
between the normal flux at the interface and expanded in a Taylor series and only the
the evaporation rate from the free surface. first term is retained. The resulting expres-
Plesset and Prosperetti (19) have proposed a sion for the evaporative flux at the meniscus
simple kinetic theory model for evaporation interface (from [22, 23, 24]) then becomes
from a fluid-vapor interface. Based on their (for Ti = Tg):
model, the normal flux is given by:
_- (Ovl(Rrg/1, 2
[ Ov~{/~Ti \1/2 [ p~ 1
(tT'u) = [---~-)[-~----~-) [a,<.-~-~--- fivj • [21]

In this expression ~ and/3v are accommoda- x okT~+ ~Rrg ~ . [25]

tion coefficients for evaporation and con-
densation, respectively,/? is the gas constant If we now equate the evaporative flux [25]
per unit mass of fluid, Pv and P~v are the to the normal velocity at the interface [15],
vapor pressure and equilibrium vapor pres- and use [20] to express the interracial
sure at the interface, and Pv and P are the temperature in terms of the normal velocity
density of the vapor and the liquid density, and shape, we obtain the following form of
respectively. Near equilibrium, we note that the mass flux condition:
oe~ and fi~ will be nearly equal and for sim-
plicity we set them to unity in what follows. (°'/(Rrgl"2( " /
The equilibrium vapor pressure of the fluid 8.0r= ) tYT-Tf
at the interface can be computed from the
equations of equilibrium thermodynamics: x [(fi-C2Fr/)+ Pop]. [26]
oH
1 H
d In P~v - dP + dT. [22] If we select for the characteristic velocity
pl?T RT 2
at the interface:
We note that the liquid pressure at the
meniscus interface is reduced relative to the
8uo = (Ov7 ))
\-p )\-'~ ) \---t~T-~gg [27]
pressure at the pool (ogL) by curvature and
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
218 MOOSMAN AND HOMSY

the resulting expression for the flux balance Since the pressure gradient associated with
becomes (from [15, 26, 27]): a given flow rate Q is inversely propor-
tional to the cube of the film thickness
[15], and since the evaporation rate (F
1 + C2~ [281
= - Q ' ) is small over much of the meniscus
region, we expect the largest reduction in
Two new nondimensional groups have been
pressure to occur in the transition region.
introduced, C3 and/3, which are defined:
There the pressure gradient associated with
fluid flow acts over a short distance to
C3 = P--2-° ; /3- Tw-Tg [29]
drastically reduce the fluid pressure and
pH T~
choke off evaporation.
/3 represents the degree of wall heating, The normal stress distribution [9] ex-
while C3 represents the ratio of pressure ef- presses the relation between curvature and
fects on the evaporation rate to thermal ef- fluid thickness (disjoining pressure) on the
fects. The latter is extremely small; how- one hand, and the pressure reduction in the
ever, as we point out below, the term C3P liquid on the other. In the meniscus region
becomes significant in both the transition disjoining pressure effects are small and the
region and the adsorbed film, and cannot be pressure gradient accompanying fluid flow is
neglected. associated with a moderate curvature gradi-
The meniscus profile and normal flux ent. In the transition region large gradients
distribution are governed by the two coupled in both curvature and fluid thickness occur
equations, [9] and [28] (together with the ap- as a result of fluid flow. Finally, we note
propriate boundary conditions which we that in the adsorbed layer the increased
discuss in the next section). The effects of disjoining pressure associated with the re-
both heat transport and fluid flow on the duced fluid pressure results in a consider-
normal flux distribution are apparent from ably thinner adsorbed film than that which
the form of [28]. exists under isothermal conditions.
We note that thermal resistance to heat
conduction (C2~9) limits the normal flux in Boundary Conditions
the meniscus region, where the conduction
path ~7 is relatively large, but has progres- The process by which evaporation is
sively less effect on the flux in the transi- choked off may be examined by consider-
tion region and adsorbed film. This is con- ing the situation near the adsorbed layer.
sistent with the fact that the temperature At any point in this region where evapora-
gradient associated with a given heat con- tion occurs, a pressure gradient will ac-
duction rate must act over a much larger company the resulting flow. From [9] we
distance in the meniscus region. Since heat see that such a pressure gradient is asso-
conducted to the surface must be carried ciated with both a gradient in curvature and
away as the latent heat of the evaporating a variation in thickness. As a consequence,
fluid, and since the evaporation rate de- the pressure and fluid thickness are reduced
pends critically on the interfacial tempera- as the adsorbed layer is approached. F r o m
ture via [25], the thinner portions of the [28] we see that the resulting evaporation
fluid can support a much higher heat flux rate is also reduced, while the pressure
than the meniscus region. gradient associated with any given flow rate
The effects of fluid flow on the normal is increased (since the thickness decreases).
flux (CzP) b e c o m e important only in regions When the pressure has been reduced to
where the magnitude of the pressure is -/3/C3, no further evaporation occurs and
large (recall that C3 is extremely small). the fluid thickness, and therefore the dis-
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
 EVAPORATING MENISCI 219

joining pressure, becomes constant. Phys- the channel flow region (4P'(O+)) and the
ically, this occurs when the intermolecular length of the channel (10 in Fig. 1). The
forces of attraction between molecules of resulting boundary conditions at the origin
the solid and liquid molecules at the inter- then become:
face are sufficient to prevent further evapora-
P(0) = C,P'(0); 7(0) = 1 [33]
tion in spite of the increased kinetic energy
of the liquid associated with the increased where we note that C4 is a nondimensional
interfacial temperature. The appropriate parameter given by:
boundary conditions in the adsorbed film
C4 = (41o/Xo). [34]
are thus (from [9, 28]):
P ~ -fl/C3; "O---> e/(1 + /3C1/C3) 1/3 SOLUTION BY PERTURBATION ABOUT
ISOTHERMAL CONDITIONS
as
---> ~. [30] While the complete problem is thus
The boundary conditions at the origin are specified in terms of the normal stress
derived from a volumetric flow balance and balance [9] and the mass flux balance [28],
the condition that the fluid initially fills the together with the boundary conditions at the
gap. With the regard to the former, we note base of the meniscus [33] and in the down-
that in the channel flow region (~ < 0) the stream adsorbed film [30], the nonlinear
pressure gradient is constant and the volu- coupled nature of the equations precludes
metric flow rate is given by: a simple solution. We note, however, that
we are interested primarily in describing
e'(o_) both the change in the evaporating meniscus
Q - [31]
12 profile from the static isothermal profile and
the evaporative flux from the interface, for
The total evaporative flux from the inter- small wall superheats,/3. As a consequence
face, however, is given by the value of the we seek a solution by expanding the pres-
volumetric flow rate at the origin, Q(O), sure and shape in a regular perturbation
given by [15]. As a consequence we con- series in/3:
clude that a discontinuity in the pressure
gradient occurs near the origin with: P = Po + /3P1 + • • • ;

P'(O_) ~9 = "O0 + /3"01 + • • • [35]

e'(o+) - - - [321
4 The perturbation functions P1 and rh will
elucidate the nature of the evaporation at
Such a discontinuity has been demon- menisci and allow us to draw some general
strated to occur near the junction of a free conclusions.
surface and two parallel plates in studies If we substitute these expressions into
by Richardson [22, 23]. Solving for the [9, 28, 30, 33] and collect terms of the same
Stokes flow between two flat plates which order in/3 we recover, at lowest order, the
undergo an abrupt transition from no-slip to equation for the static isothermal meniscus
no-stress boundary conditions, Richardson profile given by:
demonstrated that the flow associated with
1 = ~ + C/'0o3; Po = 0
such a transition undergoes a change in pro-
with
file and pressure gradient over approxi-
r/0(0) = 1; -00(oo)= E. [36]
mately a half channel width either side of
the transition. This is the problem considered by Renk et
The pressure at the origin can be com- al. (13), who obtained an implicit analytical
puted in terms of the pressure gradient in expression for ~q0 given by:
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
220 MOOSMAN AND HOMSY

E
70 = ~ (3 z2 - 1). [37] values of the five nondimensional groups
{~LC1,C~,C3,C4}.
The implicit variable (z) in this expression The five nondimensional parameters were
is related to the independent variable (() evaluated based on physical properties of
through the following relation: carbon tetrachloride on glass at 20°C (22)
and assuming for the gap width (h0), lower
plate length (10), and distance between the
(z-z°)+~ln[ ( z °- l+) l ) ] ( (z z+0l )- - ] )
upper plate and liquid pool (L); 10-3 cm,
¼ cm, and 1 cm, respectively. The H a m a k e r
(3e)v2 , z0
=12+o/1'
\--'~e ] " [38]
constant for carbon tetrachloride-glass sys-
tems was set at 5.6 × 10-21 J (17). Under
these conditions the five nondimensional
By employing the method of matched
asymptotic expansions, these authors (13) groups, rounded to the nearest multiple of
showed that as e ~ 0, this profile can five, become:
be considered as being made up of a capil- ea = (A/h~o)/pgL = 10-9
lary meniscus region, where thin film ef-
fects are negligible; and an adsorbed film of Ca = po/pgL ~- 5
thickness e, where curvature effects vanish. Cz = pH6uo/(kTg/ho) ~- 50 [41]
By matching the meniscus solution with the
thin film solution through an intermediate Ca = po/pH ~- 10-6
transition region, in which both curvature Ca = lo/4Xo ~- 100.
and disjoining pressure effects are sig-
nificant, they found that the capillary Although • and Ca are small, their neglect
meniscus merges with the thin adsorbed film is not uniformly valid. At the same time,
at ~:0 = 2 l/2, and does so with zero apparent the groups which have been assumed to be
contact angle. Moreover, they found that negligible are given by:
the width of the transition region scales as
(~ - 2,/2) - e,/2. { ~,62,R M,Nu }
At next order in /3 the shape 71(0 and {10 -3 , 5 × 10 2, 5/3, 10-3}. [42]
flux FI(~) are determined from the solution
of the two equations: The two differential equations [39] were
solved subject to the boundary conditions in
3e3 71 [40] and with the choice of parameters given
7~ - - - C1P,;
7~ 7o in [41]. The integration was carried out us-
ing a variable order ordinary differential
equation routine with automatic error con-
F, = - 1 + C27o trol which is described by Shampine and
With boundary conditions: Gordon (23). The downstream boundary
was set at ( = 2.0 after noting that shift-
7a(O) = O, Pa(O) = C4PI(O); ing this location produced no change in any
EC~ 1 of the computed quantities to at least six
' r h ( o o ) ----) - - - - , Pi(oo) ---> - - - [40] significant figures. We then alter each of the
3C3 C3 five parameters in [42] by a factor of two (as
The isothermal profile 70(~:) is of course might correspond to different choices of
known. We note that the flux equation is fluids or physical dimensions) to determine
now decoupled from the normal stress equa- the relative effect of each on the deviation
tion, and that both equations are linear. The profile and the normal flux.
solution will depend, of course, on the In Figs. 2 and 3 we display the results for
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
 EVAPORATING MENISCI 221

1.0
the increase in the equilibrium vapor pres-
k•o (:103
sure of the liquid with temperature. As a
~,~o consequence, the disjoining pressure must
05
be higher than in the isothermal film and the
film thickness is therefore less.
With an increase in e, we note that the
O0
O0 05 I0 15 20 normal flux is slightly spread out near the
transition region (Fig. 2) and that the de-
FIG. 2. The evaporative flux as a function of distance crease in the thickness T~I is greater. The
a n d • 3.C1 = 5, C2 = 50, C3 = 10 ~,C4 = 100. former result is consistent with the observa-
tion that larger values of • are associated
the normal flux and deviation profile for with an increased importance of disjoining
both the choice of parameters given in [41] pressure, the influence of which is sig-
and for those corresponding to a factor of nificant over a larger distance. The result
two increase in the parameter •. We include that the deviation in the thin film region
the isothermal profile T0 for reference and (-7/1) is greater for larger values of • reflects
note that the transition layer between the the fact that larger decreases in the thick-
capillary meniscus and the adsorbed film oc- ness are required to bring about a given
curs at approximately sc = 21/2. The normal increase in the disjoining pressure when the
flux (Fig. 2) is small over much of the menis- isothermal thickness • is larger (since dis-
cus region but increases rapidly as the tran- joining pressure depends on the inverse
sition region is reached, and the thermal cube of the thickness). The actual film thick-
resistance to conduction (C2~) becomes ness is of course larger for increased •
small. A sharp maximum in the evaporation values [30].
then occurs, b e y o n d which the effects of Increased values of the parameter C1,
increased disjoining pressure with thinning while not affecting the normal flux distribu-
act to rapidly choke off further evaporation. tion, act to substantially increase the thin-
In Fig. 3 we note that the deviation profile ning (Fig. 4). Since Ca is a measure of the
71 also displays a maximum near the transi- relative effect of flow on the pressure in the
tion region, and that the adsorbed film is fluid, an increase in C1 represents a larger
uniformly thinner than the isothermal layer. suction for a given flow rate. As a conse-
The latter fact is consistent with the ob- quence [9], the disjoining pressure in the
servation that the intermolecular forces be-
tween the solid and the liquid molecules at
the interface must offset both the hydro-
static suction imposed by the liquid pool and 7.0

6O
IO~x~ / ~ 50
50
7/o -!40
"qo ~ ~2×1d3 C1:100
05 "ql/lO
2()
~,. C~: 5.0
I© C1:25
10
O0 m I I ~ I 90
O0 05 I0 1.5 2.0 O0 115 2 I0
O0 05 I0

FIG. 3. The perturbation profile as a function of Fro. 4. The perturbation profile as a function of Ca.
distance and e3. P a r a m e t e r values are as in Fig. 2. e3 = 10 9, C2 = 50, C~ = 10 6, q = 100.

Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
222 MOOSMAN AND HOMSY

'°I C2:2
is in agreement with our intuition that the
channel flow should substantially decouple
from that in the meniscus.
0.5-
In summary, we point out that the com-
puted results display many of the features
we expect for the evaporating meniscus pro-
0.0 I
0.0 05 1.0 1.5 2.0 file and the normal flux distribution. In par-
ticular, the flux displays a sharp maximum
FIo. 5. T h e e v a p o r a t i v e flux as a f u n c t i o n of C2. near the transition region where thermal re-
E3 = 10 -9, C1 = 5, C3 = 10-6, C4 = 100. sistance effects (which limit the evapora-
tion in the meniscus region) are small and
thin film increases with C1 and the thickness yet viscous stresses associated with fluid
of the layer decreases. flow have not reduced the fluid pressure
A change in the thermal resistance of the (and thickness [9]) to the point where
system C2 has relatively little effect on the evaporation is choked off. The deviation
deviation profile. We note that the principal profile displays a uniform thinning in the ad-
effect o f decreasing this resistance is to ex- sorbed layer, which is associated with the
tend the region o f significant evaporation increased disjoining pressure acting to sup-
into somewhat thicker portions of the press evaporation, and a maximum near the
meniscus region (Fig. 5). Changing values of transition region. As a consequence the ac-
C2 have relatively little effect in the thinner tual profile is expected to thin in the menis-
regions where resistance to conduction cus region, and the transition region should
(C2~) is insignificant anyway. shift toward the inlet. In addition, the ap-
Doubling or halving the parameter C3 has parent contact angle should increase.
relatively little effect on t h e normal flux For practical applications, the assump-
distribution. This is due to the fact that the tion of a regular perturbation expansion in
effects of pressure on the evaporation rate /3 [35] is strictly applicable only for values of
are confined to a very small region near the /3 for which/31~711 ~ T0. From the results,
adsorbed layer. There large pressure gradi- as well as from the boundary conditions
ents associated with fluid flow act to rapidly [40], we see that this corresponds to wall
reduce the pressure to -fl/C3 and choke off superheats/3 which are much less than the
evaporation. On the other hand, the devia- r a t i o 3 C 3 / C 1 ~ 10-6. For more reasonable
tion profile (Fig. 6) is considerably affected values of/3 the qualitative features of both the
by changes in C3. For larger values of this profile and the normal flux distribution
parameter, the relative effect of pressure on
the evaporation rate is increased. As a re-
7O
suit, the disjoining pressure in the adsorbed
layer necessary to suppress evaporation is
reduced. Hence, the film thickness in- 5O
creases with increasing values of Cz.
_'r/l/10 C3:QSxlO-6
Finally we note that C4 has virtually no 30
effect on either the evaporation rate or the
20 Cs= 10-6
deviation profile. Since the largest fluid
EO Cs= 2 xlO-6
pressures occur in the thin fluid regions (as
a consequence of pressure gradients asso- O0
00 05 I0 115 2.10
ciated with flow). The comparatively small (
fluid pressure at the inlet (set by C4) is not FIG. 6. The perturbation profile as a function of
expected to display much of an effect. This C3. e3 = 10-9,C1 = 5, C2 = 50, C4 = 100.
Journal of Colloid and Interface Science, Vol. 73, No. 1, January 1980
 EVAPORATING MENISCI 223

should be similar to those outlined above. 8. Potash, M., Jr., and Wayner, P. C., Jr., Int. J.
However, to obtain these requires that the Heat Mass Tran. 15, 1851 (1972).
9. Renk, F. J., and Wayner, P. C., Jr., J. Heat
full nonlinear problem [9, 28, 30, 33] must Tran. 101, 59 (1979).
be considered. 10. Wayner, P. C., Jr., Kao, Y. K., and La Croix,
L. V., Int. J. Heat Mass Tran. 19, 487 (1975).
11. Wayner, P. C., Jr.,J. Heat Tran. 100, 155 (1978).
ACKNOWLEDGMENT
12. Derjaguin, B. V., Starov, V. M., and Churaev,
This work was supported in part by the Office of N. V., Colloid J. 38, 786 (1976).
Basic Sciences of the U. S. Department of Energy. 13. Renk, F., Wayner, P. C., Jr., and Homsy, G. M.,
J. Colloid Interface Sci. 67, 408 (1978).
14. Van Dyke, M., "Perturbation Methods in Fluid
REFERENCES Mechanics." Parabolic Press, Stanford, Calif.,
1975.
1. Wayner, P. C., Jr., and Coccio, C. L., AIChE J. 15. Sheludko, A., Adv. Colloid Interface Sci. 1, 392
17, 569 (1971). (1967).
16. Churaev, N. V., Colloid J. 36, 283 (1974).
2. Preiss, J., and Wayner, P. C., Jr., J. Heat Tran.
17. Churaev, N. V., Colloid J. 36, 287 (1974).
98, 178 (1976).
18. Blake, T. D., J. Faraday Soc. 71, 192 (1974).
3. Renk, F., and Wayner, P. C., Jr., Proc. 5th Int.
19. Plesset, M. S., and Prosperetti, A., J. Fluid Mech.
Heat Transfer Conf. 5, 252 (1974).
78, 433 (1976).
4. Renk, F. J., Ph.D. Thesis, Rensselaer Poly- 20. Richardson, S., Proc. Comb. Phil. Soc. 67, 477
technic Institute, Troy, N. Y. (1977). (1970).
5. Renk, F. J., and Wayner, P. C., Jr., J. Heat 21. Richardson, S., Rheologica Acta 9, 193 (1970).
Tran. 101, 55 (1979). 22. Weast, "Handbook of Chemistry and Physics."
6. Derjaguin, B. V., Nerpin, S. V., and Churaev, Chemical Rubber Co., 1967.
N. V., Bull. RILEM, 29, 93 (1965). 23. Shampine, L. F., and Gordon, M. K., "Computer
7. Churaev, N. V., and Ershova, I. G., Colloid J. Solution of Ordinary Differential Equations."
33, 764 (197I). W. H. Freeman, San Francisco, 1975.

Journal of Colloidand Interface Science, Vol. 73, No. 1, January 1980