Entropy-Induced Dispersion of Bulk Liquids ~

H. REISS
Department of Chemistry, University of California, Los Angeles, California 90024
Received March 17, 1975; accepted July 8, 1975

The possibility that two-phase fluid systems will emulsify spontaneously, due to entropy
effects, is investigated theoretically. It is discovered that for interfacial tensions below about
2 dynes/cm, such spontaneous dispersions can occur, although with large tensions
the dispersed particles are not macroscopic. However, for interracial tensions below 5 X 10 -4
dynes/cm, macroscopic particles are formed.
Spontaneous dispersion due to entropy effects is to be distinguished from the formation of
micellar solutions that form because of energetic considerations.

1. INTRODUCTION micellar solutions in which the micelles are

Everyone knows that two immiscible liquid quite large, if not macroscopic, clusters of one
phases can coexist at equilibrium. Because of of the original phases. The motivating in-
gravity, the denser phase is always found at fluence for the formation of micellar solutions
the bottom of the fluid system. Even in the is largely energetic. For example, a typical
absence of gravity, the system, at equilibrium, micellar solution may consist of fatty acid
would consist of essentially two fragments, and water (1). By dispersing as micelles, the
each corresponding to a phase whose location fatty acid molecules are able to increase the
would be neither fixed nor predictable. area of contact between their polar heads and
If the binary system has an upper consolute water while maintaining their fatty hydro-
(critical) temperature, a broader range of phobic tails in close association. However, it is
fragmentation occurs at that point and the conceivable that there is another driving force,
clear distinction between the two phases besides dispersion. This is the entropy gained
disappears. The fluid then consists of clusters by the act itself. Such entropic effects also are
ranging continuously in size from macroscopic present in the micellar case, but are usually
lumps of both phases down to individual unnecessary in view of the prior effects of
molecules. However, even here, in the presence energy.
of gravity, stratification occurs. Ordinarily, dispersion is thermodynamically
All of this raises an interesting question. unfavorable because the extended interface
Are there situations, possibly in ternary between the two phases occasions an untenable
systems, in which the fluid, in the absence of increase in interface free energy. While the
gravity, would prefer to disperse, at equi- "entropy of mixing" that results from dis-
librium, into macroscopic fragments much persion produces a reduction in free energy,
larger than molecules rather than remain in it is rarely enough to overcome the increase
one cohesive lump? The answer is affirmative; due to the interfacial effects. As a result,
the situation at the critical temperature is a spontaneous dispersion does not occur. Micellar
case in point. Furthermore, there are so-called solution may be crudely regarded as a case in
1Research supported under NSF Grant No. MPS73- which the interfacial tension is negative so that
04743-A01. extension of the interface leads to a decrease in

61
Copyright ~ 1975 by Academic Press, Inc. Journal of Colloid and Interface Science, Vol. 53, No. I, October 1975
All rights of reproduction in any form reserved.
62 H. REISS

free energy, and no further decrease (e.g., free be positive and spontaneous dispersion will
energy of mixing) is required to effect dispersion. not occur.
I t is natural to ask whether cases exist in However, as indicated in the last section,
which the interracial tension is positive, but there is another contribution, &G", which
small enough that the increase in interface free comes from mixing, so the change in free energy
energy can be overcome by the mixing free should be
energy so that dispersion occurs. Dispersion a a = ~G' + a a " . E2~
at the critical point is a limiting example of
this phenomenon; it is well known that the If &G" is sufficiently negative, the net change
interracial tension tends to vanish as the &G will be negative, leading to a dispersed
critical temperature is approached. equilibrium state. We therefore attempt to
Recently, there have been reports (2) of the calculate zXG".
development of surfactants capable of reducing In performing this calculation, we note, as
interfacial tensions at an oil-water interface to it has been recognized before (3), that a
the range 10 -4 to 10.5 dynes/cm. One wonders spherical cluster of fluid not fixed to the
whether or not such a low interracial tension is laboratory frame of reference (e.g., not like a
enough to cause spontaneous dispersion. Such drop suspended from a capillary tube, itself
an end result could have a profound impact at rest in the laboratory frame) must be defined
(ignoring economics) on technological processes thermodynamically by a Gibbs dividing
like tertiary recovery. surface centered on its center of mass. Thus,
In the remainder of this paper, we perform the spherical clusters in our model will have
a statistical thermodynamic analysis of the this property. In contrast, a "drop" to which
possibility of such "entropy-induced" disper- the surface tension ~ refers, is fixed to the
sions in systems where the interfacial energy laboratory frame, and the dividing surface
is small, but still positive. used to define it is similarly fixed. Thus, in a
drop the position of the center of mass may
2. A SIMPLE MODEL fluctuate with respect to the dividing surface.
The question raised at the end of the previous The thermodynamic properties of the cluster
section can be studied conveniently by model- may therefore (as we shall show) be estimated
ing the dispersed fragments as spherical drops from those of the drop, but they are not
of one phase possessing the interfacial tension precisely the same.
corresponding to the fiat interface, even though To make matters even simpler, we assume
the drops may be very small, e.g., consisting that when dispersion occurs, clusters of only
of several thousand molecules. We denote the one size (one value of radius) are produced.
two phases by the subscripts 1 and 2, reserving Since the entropy of mixing would only be
the 2 for the less abundant phase, which we increased by admitting clusters of many sizes,
also regard as the dispersed phase. If the the tendency towards dispersion could only be
interracial tension is denoted by ~, and we increased. Thus, an estimate based on clusters
assume that the chemical potentials of both of one size gives a lower limit for the influence
components are not changed by dispersion, towards dispersion. If dispersion is indicated,
the change in free energy associated with the phenomenon is even more likely when
such a process (ignoring the free energy of many sizes are allowed.
mixing) will be Our general procedure will be equivalent to
the following. The "internal" partition func-
AG' = ~ N,4~rri2~, [1-] tions (motion relative to the centers of mass)
i of the clusters will be estimated, essentially,
where Ni are the number of drops of phase 2 by referring them to the free energies of a
of radius r~. If ~ is positive, G will necessarily laboratory drop of the same size. The spherical
Journal of Colloid and Interface Science, Vol. 53, No. 1, October 1975
 DISPERSION OF LIQUIDS 63

clusters will not be considered statistically present in the total volume

inflependent but will be allowed to interact
through their centers of mass. This interaction V = Vl -t- V2 [6]
will take the form of familiar interaction
between "hard spheres" of radii equal to that of the system. V2 is the volume of phase 2.
of the clusters. The configurational partition These molecules are assumed to be partitioned,
function of the disperse system then becomes always, between ND clusters, each containing
identical to that of the equivalent hard sphere n molecules. The clusters are spherical, and
system for which very good estimates are are defined by the requirement that all of the
available. This theory is developed in the molecules in a given cluster lie within a
next section. distance r of the center of mass of the group
comprising the cluster (4, 5). The magnitude
of r is determined by
3. PARTITION FUNCTION OF THE DISPERSED
SYSTEM
r = (3nv2/4~r)~, [7]
We begin by dividing the system into two
subsystems corresponding roughly to the two where v2 is the volume per molecule in bulk
phases, 1 and 2, and to which we assign the liquid phase 2. Thus, r is chosen as the radius
respective partition functions, (21 and Q2. of a drop of phase 2 consisting of n molecules.
Subsystem 2 is assumed to include the interface, Then the cluster resembles a drop but differs
i.e., the surface free energy is assigned to phase from it by the fact that its boundary surface
2. Furthermore, we assume that the partition (dividing surface) is fixed relative to the center
function for the total system is given by of mass, whereas in a drop the center of mass
can fluctuate with respect to the dividing
Q = Q1Q2, [3] surface.
As indicated in the last section, the molecules
an approximation not unreasonable as long
within a cluster may interact with each other
as effects due to the interaction of the two
by means of normal intermolecular forces, but
systems can be largely subsumed into surface
the interaction between clusters will be
effects invariant to the configuration of the governed by the potential,
clusters of the dispersed phase.
If ~1 is the chemical potential in phase 1 and
u~s = ¢(l~r..ll~), [S]
there are N1 molecules in that phase, then
the Helmholtz free energy of the phase is
where UHS is the potential for ND hard
A1 = N1~I - pV1, 1-4] spheres of radii r and R1 through RND are the
coordinates of the centers of mass of the
where p is the pressure and V1 is the volume clusters. The logic behind this is the following.
of the phase. Then, The clusters represent quasi-macroscopic (or
macroscopic) fragments of an emulsion (dis-
(21 = exp(--A { k T ) persion). As such, it is sufficient to account for
= e x p 1 - - - ( N l ~ l - - pV1)/kT]. 1-5-] their configurationai properties by disallowing
overlaps between particles. The hard sphere
The evaluation of Q2 is considerably more potential specified in Eq. [-8] takes care of
involved. Suppose there are N2 molecules in this. Interaction within a cluster is accounted
phase 2. Because of the assumed independence for by the ordinary intermolecular potential,
of the two phases or subsystems implicit in the while that between the clusters and phase 1 is
product in Eq. (3), we may proceed with the accounted for by the surface free energy.
system of iY2 molecules as if it alone were With these conditions, it is possible to
Journal of Colloid and Interface Science, Vol. 53, No. 1, October 1975
64 H. REISS

write Q~ as and Q2 can then be represented by (3, 4)

Q,, . . . . exp[-fl ~ p.2

__~ [.. \ j=l 2m \N2 !hsN~/\ (n ])NOND !)/

x dr~...drxJm...dpze~. [93
X
L L .. exp I - - f i e ( R e . . RND)]

X lI exp --fl
In this equation, rl and pl, etc., are the co- i=i 2m
ordinates of position and momentum of the
N2 molecules (which for simplicity we assume +
to be structureless and devoid of internal
degrees-of-freedom), U(rl.-.rN,) is the po- X dr~(i)dpl (~).. .dp,/~) 1 [125
tential for molecular interaction within clusters,
¢ is the hard sphere potential defined in
Eq. [8], m is the mass of a molecule, in which we allow contributions to the integral
h is Planck's constant, and/3 is 1/kT, where k only when each set of r l ( i ) " • ' r , S ) have values
is the Boltzmann constant and T is tem- such that
perature.
[ry) -- R~[ < r, [13]
The potential U can be represented by the
sum
where r is given by Eq. [7]. Eq. [13] implies
i=ND that any r/i) lies closer than the distance r to
U= ~ Ui(r1(~)--'r,Jl)), [10] the center of mass of the ith cluster. In Eq.
[12], the condition, Eq. [13], could be applied
where U~ is the potential for molecular inter- by simply introducing potentials which pass
action within the ith cluster, and (rl (o. . •rn (o) to infinity when the inequality in Eq. [13] is
are the coordinates of the molecules within violated. However, the condition is so simple
that cluster. This follows from the assumption that it is enough to state it.
that the interaction between clusters is deter- Next we transform to center of mass co-
mined by ¢(R1". "RN~), so that there are no ordinates (4, 5). Thus, we introduce r/~)' by
cross terms in U between clusters. For a fixed the relation
set of locations, rt (o.. .r~ (ND) (and conse- r/i) = r/i)' + R~, j = 1,2...n- 1, 04]
quently a fixed set of R1." "RND), there is a
large number of permutations of the N2 where
molecules which correspond to the same value
1 J=~
of the integrand in Eq. [9]. However, permu- ll~ = - E C i). [153
tations that permute "whole clusters" must n j=l
not be counted since these are accounted for
by the integration in Eq. [9-1. The molecules In terms of these coordinates, we may write
are of course considered distinguishable, the
actual indistinguishability being corrected for, Ui(rl(O...r. (1)) = fi(rl (i)'.. .rn_l(1)'), [16]
as usual, by the 1/2V2[ preceding the integral
in Eq. [9-1. The proper number of permutations since the molecular interaction can only depend
may easily be shown (4) to be given by on the internal coordinates of the cluster.
Substituting Eqs. [-14-1-1-16] into [12"], inte-
grating over the momenta, and performing
Journal of CoUoid and Interface Science, Vol. 53, No. 1, October 1975
 DISPERSION OF LIQUIDS 65

some algebraic condensation, yields

Q2
<°f £
. . . . .
NvtY
dRy.., d R ~ e~p [ - - ~ ( R 1 . . . R~)-]

X II . . . . exp E-flfi(rl (i)'.-.rn_l(i)')-]dri <i)'...dr~-i (°' [-17~

i< / n[ J o

. . . . dR1.., dRND exp [ - - ~ 4 , ( R v ' ' Rx~)-]

@ ~f
ND! fv
T,,-lna
[
X .
t n!
. _.
J /.
.
~.
exp
. [--¢~f(ri'. "r~-i)-]dri
' ' . . • d r~_l '
7 . [-18]

In this equation, nv2 is the volume of the a distance r of their common center of mass
cluster. The integration is limited to this are allowed.
volume by the condition, Eq. [-13]. The Comparison of Eq. [-18-] (with ~ = 0) with
quantity, % resulting from the integration over Eqs. [-20-] and [-21-] reveals that
momenta, is
"y = (2~rmkT/h2)~, [-19-1
i.e., the reciprocal of the de Broglie wavelength
for a molecule of mass m. In passing to the
last form in Eq. [-18-], use has been made of the
fact that all the integrals in the continued
4
product are identical and the i subscript and M dri'.., dry-1' I I [-223
superscript could be dropped. The n 3 in Eq.
[-18-] comes from the Jacobian of the trans- Now, within the framework of our model, we
formation from laboratory to center-of-mass wish to evaluate the quantity in braces in
coordinates (4, 5). Eq. E22-] (for use in Eq. [-183) by reference
As a point of reference, it might be men- to the free energy of a "drop" of n molecules.
tioned that if q~ were set equal to zero so that The partition function of a drop, fixed in the
the individual clusters did not interact with laboratory frame of reference, may be expressed
one another (an ideal gas of clusters) and the by a formula very similar to Eq. [-22-]. To
transformation back to laboratory coordinates avoid confusion, we denote this partition
were made, the result would be function by X~. Then we have (4),

Q~ = q~N~/X~ !, E20-] X =~f dRI7'~-l___

na
where nv~ L n!

qn
=':/ L
n! - exp[- --Su(rl. • •r,) -]
X drl...drn, [-211
X L ( m " • • f exp [--Sf(rl'" • •r~-l)-]

X dr[. • •dr~_l } i [23-]

and represents the classical phase integral of
a single cluster. Here the integral is over the The differences between Eqs. [ 2 2 - ] and [-23]
entire volume V, but it is understood that in are the following.
performing the integration, only those con- Because the bounding or dividing surface
figurations in which all n molecules are within of the drop is no longer centered on the center
Journal of Colloid and Interface Science, Vol. 53, No. 1, October 1975
 66 H. REISS

of mass, the volume over which the center-of- Comparison of Eqs. 1,26] and [27-] then yields
mass coordinates r~'.., rn_, ~ go in the integral
in Eq. 1,23-] will be a function of R (5). We
indicate this by symbolizing the limits of
integration by v(R) rather than nv2 as in Eq. z(R)dR
1,22-]. In addition, in Eq. [23-] the integration
J!
t,.~ nv2
over R is limited to the volume, nv2, of the where P (0) is the probability density measuring
fxed drop, rather than extending over V as it the chance that the center of mass will be
does in Eq. 1,22]. found at the origin. Substitution of Eq. [29-]
We can also write Eq. [23-] in the form into Eq. 1,28-] gives
standard for the partition function of a fluid
of n molecules confined to a volume nv2, "in-in3 f J]~
namely, n! "'" v2exp [---/sf(rl'...r~-l')-]
X . = ('i./n!)Zn, 1,24]
X drl'-..dr,_l' -- z(0) = - - P(0), 1,30-]
where z. is the configurationalpartition function n[ 7
for n molecules confined to nv2. Comparison of
Eq. [24] with Eq. 1,23] shows that and substitution of Eq. [30] into [-181 yields

z'~=f~,2dRI n 3f~(m exp[--/3f(rl'...r.-l')]

FXnP(O)y
f(..
X drt'..-dr~_l' 1 . 1,25-] × exp 1,-/3~(R1... RNv)~dR1... dRN~. [-31]

Now the Helmholtz free energy of a drop

Clearly, the quantity in brackets in Eq. [-25] consisting of n molecules of phase 2 may be
is the configurational partition function for a expressed as
drop under the constraint that its center of
mass is held at the point R. Denoting this
an = n#2 + O~n~ pnv2, 1,32] - -

configurational integral by z(R), Eq. 1,24-] in which u2 is the chemical potential in phase 2
becomes at temperature T and pressure p (note that
the pressure inside a drop may differ from the
Xn = z(R)dR. 1,26-1 pressure p outside), and a is given by
v2
a = 47r~(3v2/47r)~, 1,33"1
Since the integral over the center-of-mass
where ~ is the surface tension. In the canonical
coordinates in Eq. 1,22-] refers to the case
ensemble, a~ is related to the partition function
where the center of mass is constrained to the
k~ by
center of the spherical volume, nv2, i.e., to the
case R = 0, we can write a~ = -- (1//3) in Xn, [34]
so that combining Eqs. [32] and [34] gives
"In/V "In
q" n! z(0)dR n] Vz(O). 1,27] X~ = exp[--fl{n#2 + an~ -- pnv2}-]. 1,35]
The quantity P (0) has been estimated with
Furthermore, the quantity in braces in considerable accuracy, in another context
Eq. 1,22] is clearly (3, 4) to be
"in--1
- -
nt
z(0) =
~n
--
~V
1-283 P(0) = -- 1 I~r( 6n )'1'
, 1,36-]

Journalof Colloidand InterfaceScience,Vol. 53, No. 1, October 1975

 DISPERSION OF LIQUIDS 67

where (7). Defining

f = p/po, [37] y = 4rrraC/3 [-43]

in which p0 would be the density of the fluid and

if the quasi-hard cores of the molecules were = 4~rr3~PHs/3kT, [44]
close packed and p is the actual density. For this expression is
most liquids, f approximates 0.5, and typically,
y + y2 + ya

P(o) . .
, .r,2,7
.
n; (12)' [38]
= -

(1 - y)~
[4S]
nT)2 L. 71" .A T2 \ 7 1 " ] "
Substitution of this relation into Eq. [40]
With both M~ and P (0) determined, it remains and subsequent integration yields
to evaluate the integral in Eq. [31]. We
accomplish this in the next section.
ZHS= {
V(1--y) exPL2(l_y)Mj E<
4. T H E HARD SPHERE CONFIGURATION
INTEGRAL AND FREE ENERGY OF TIlE Thus, we have expressed ZHS as a function of
TOTAL SYSTEM N•, r, and V.
Substitution of Eqs. [35], [38], and [46]
The integral in Eq. [31] is simply the into [31] then yields,
configurational partition function for N•
hard spheres of radii r confined to the volume
V. We denote it by ZHs(ND, V). Now the
hard sphere equation of state may be expressed
in the form
(O lnZns~ , X exp [ - - ( ~ T ) { n u 2 q-au2 q-an~--pnv2,]
pH~ = kT 7V / T , ~ [393
ray(y - 2)-]} N~,
X V(1 -- y) exp l . . . . / 1-47]
from which we easily derive, by integration, L 2(1 -- y)2_l
the formula
where we have substituted 1/kT for 5. Finally,
ND substituting Eqs. [5] and [47] into [3] gives,
Nit S
Co for the total partition function of the system,

r
X exp --fiND
u. d CO
p~s(C) , [40] Q = exp --
kT ND!

in which X exp -- (1 -- y)
C = ND/V [41] ( V2

and
Co = lim C. [42] X exPL2(1 -- y)~dJ ' [48]
V -->~o

Co is of course zero, but we retain it in Eq. where we have used the relations
[40] because it is not immediately clear that
it introduces no singularity. : Y D n = IV'2 [-49]
and
Now an excellent analytical expression for
N~n~2 = V~ [50]
PHS (C) is available from scaled particle theory
(6), and confirmed by Percus-Yevick theory together with Eq. [6].
Journal of Colloid and Interface Science, Vol. 53, No. 1, October 1975
68 H. REISS

Although the system is in equilibrium, ~1 is AG', and AG", the free energy of mixing is
and ~ are different because they refer to
different species. Actually each phase is a
binary solution and ~ and #~ are separately AG"=--kTND l n t v2 (l--y)
equal in each phase. However, both phases
are assumed dilute so that the bulk free
J E55]
F3y(y - 2) -I 1 1-]
energy of phase 1 is essentially N ~ and that X exp [ - - --~-a-~2J/ -- in ND +
-2y(1 - y)
o

of phase 2, N2~2. We are ignoring the small

contributions from the attenuated solute in For spontaneous dispersion to occur, it is
each phase. necessary that
The Helmholtz free energy of the system
is then AO < O. [-561

A = - - k T In Q We investigate this possibility in the next

section.
= N1~1 + N2~2 - - p V

5. QUANTITATIVE ESTIMATES
-- k T i n - -- exp [--an~/kT-]
ND[ v2 \ T r ~ The quantitative consequences, for disper-
sion, may be quickly discerned through an
/
x (1 - y ) e x p - -
V3y(y- 2)7j}~,
- Z E511 application of Eq. [-53]. In this analysis, we
I_2(1 -- y) will regard the interracial tension ¢ which
enters the equation through a in the manner
and the Gibbs free energy is prescribed by Eq. [-33] as the controllable
variable. At constant temperature, such a
G = A + p V = N l # i + N2#2
variation of ¢ will require the addition of a
third component, a surfactant, to the system.
-krln 7o! L ,7) We can assume that the surfactant is strongly
localized to the surface, and that it is able
to reduce the surface tension (when enough
X (1 -- y) exp [_)~ - - y - ~ j . [52-] of it is present) to such small values as 10-4 to
10.5 dynes/cm. The general approach will
involve expressing AG in Eq. [53] as a function
The last term on the right side of Eq. E521 is of n, and determining what value of n will
AG of Eq. F21, N1~1 + N2~2 being the free cause AG = 0. Any smaller value of n will
energy before dispersion. We may express this then require AG < 0, and spontaneous dis-
term as persion will be indicated.
AG = an+ND On the other hand, it must be borne in mind
that the maintenance of ~ at a constant low
level as dispersion occurs will require the
G . (l-y) addition of more surfactant since more surface
will be generated. Therefore, to achieve the
equilibrium state of dispersion characteristic
× exp [3y(y--_ 2)1/ -ln V + 1] E531 of a given constant ¢, a very large amount of
surfactant may be required. To express AG or,
from which it is evident that equally convenient (since N1) is positive),
A G / N D as a function of n, we make use of the
a n i N D = 4~rr~aN z) [-541 relation ND = N2/n, Eq. [-491. In addition,
Journal of Collold and Interface Science, VoI. 53, No. 1, October 1975
 DISPERSION OF LIQUIDS 69

we note that from Eqs. [41] and [-43], TABLE I

~r n r
y = 47rraC/3 = 41rraND/3V (dynes/cm) (molecules) (ern ) (sites/par ticle)
= nv2Nv/V = nv2N2/Vn
1.68 10 6.20 X 10-s 2.25 X 10
= v2.¥2/V. [57] 1.00 102 1.34 X 10-7 1.04 X 102
3.53 X 10-1 103 2.88 X 10-7 4.84 X 102
S u b s t i t u t i n g Eqs. [--33] and [48] into, [53] 1.06 X 10-1 104 6.20 X 10-7 2.25 X 103
and setting AG -- 0, gives 2.91 X 10-2 105 1.33 X 10-6 1.04 X 10*
7.64 X 10-3 106 2.87 X 10-6 4.84 X 104
kT(4~) I Ve (12n')I 1.94 X I0-3 107 6.19 X 10-6 2.24 X 105
,r=-- in - - -- (l--y) 4.82 X 10-4 10s 1.33 X 10-5 1.04 X 106
47r t v2.¥2 \ ~r / 1.18 X 10 -4 109 2.87 X 10 -5 4 . 8 4 X 106
2.83 X 10-~ 10~° 6.19 X 10-5 2.25 X 107
6.73 X 10-~ 10n 1.33 X 10-4 1.04 X 108

T h e relationship between n and ~ in this a typical value, then

equation is just that required in order to have
V = (N1 + N2) X 10-22 cm a
AG = 0, so for a n y given value of ~, a value
of n smaller t h a n that required b y Eq. [-58-] = 2N2 X 10-22 cm a. [-60]
causes the system to be dispersed. We also set T = 300°K. With these choices,
To make matters simple, we choose each
phase to have the same volume per molecule y = 0.5, [-61-]
in the liquid and assume also that N1 = N2. and Eq. [58] becomes
If we choose
v2 = 10-2~ cm 3, [-59] 3.974
z = - { ( 1 . 5 0 0 I n n ) -- 1.490}. [62-]
n{
1011
Table I was prepared using this equation.
1010 I n the first column, values of interfacial
tension are provided. The second column lists
109
values of n for the particles that will arise
because of spontaneous dispersion at the
108
values of ~ appearing in the first column.
107 Column 3 gives the radius in centimeters of
the particles listed in the second column,
rl 106
assuming v2 = 10-22 cm a. Finally, the fourth
column lists the n u m b e r 2; of "surface sites"
10~
on each particle, assuming each site to occupy
304 the area @. Fig. 1 is a plot of n versus ~,
based on the data in Table I.
103
We note t h a t spontaneous dispersion is
indeed predicted for interfacial tensions below
10~
about 2 dynes/cm, and t h a t above this value
! no dispersion occurs. However, at 1.68
10
10 6 ~0 5 104 10 3 10"2 10 1 100
O-{DYNESI'CM) d y n e s / c m , the particles will hardly be macro-
scopic, containing only 10 molecules. On the
FIG. 1. Plot of the equilibrium size (number of
molecules in particle) of the dispersed particle as a other hand, at 1.17 X 10-4 d y n e s / c m , the
function of interfacial tension. clusters will contain 109 molecules and have a
Journal of Colloid and Interface Science, Vol. 53, No. 1, October 1975
 70 H. REISS

radius approximating a third of a micron. model becomes quite inadequate, and it m a y

These will be quite macroscopic. be necessary to view the calculated results as
The last column gives some idea of the qualitative at best.
amount of surfactant that will be required.
Assuming the surfactant molecule localized to REFERENCES
the surface and occupying one surface site
as defined above, we see that, to achieve 1. HAI~TI~EY,G. S., "Progress in the Chemistry of
Fats and Lipids," Vol. 3. Pergamon Press,
dispersion, particles containing 109 molecules London, 1955.
will require 4.84 ;< l0 G surfactant molecules 2. Private communication, NSF Workshop on Tertiary
per particle, or a mole fraction of about 0.005. Recovery, Austin, Texas, June 1974.
In closing, we should note that the clusters 3. REISS, H., in "Nucleation" (A. C. Zettiemoyer,
will be in dynamic equilibrium. If they are Ed.), Vol. 2, Sect. 8, to appear.
4. REISS, H., in "Nucleation" (A. C. Zettlemoyer,
small enough, they may disappear at one
Ed.), vol. 2, Sects. 9, 10, to appear.
point only to reappear at another. However, 5. REISS, H., KATZ, J. L., AND COHEN, E. R., Y. Chem.
quasi-macroscopic clusters are likely to main- Phys. 48, 5553 (1968).
rain continuous identities. Furthermore, when 6. REISS, H., "Advances in Chemical Physics" (I. Pri-
the surface tension becomes quite small, the gogine, Ed.), Vol. 9, p. 53. Interscience, New
York, 1965. Also, REISS, H., FI~ISCH,H. L., AND
thickness of the interfacial region will increase
LEBOWlTZ,J. L., J. Chem. Phys. 31, 369 (1959).
to the point where it becomes comparable to 7. WE~I~EI~, M. S., Bull. Amer. Phys. Soc. 8, 329
the dimensions of the cluster. In this limit, our (1963).

Journal of Colloid and Interface Science, VoL 53, No. 1, October 1975