Gregory

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Approximate Expressions for Retarded van der Waals Interaction

JOHN GREGORY
Deportment o f Civil Engineering, University College London, Gower Street,
London WC1E 6BT, United Kingdom
Received September 29, 1980; accepted February 16, 1981

Several approximations for retarded van der Waals interaction between plates, between spheres,
and for the sphere-plate system are considered. For the interaction of fiat plates, a single ex-
pression is shown to be an adequate substitute for two empirical formulae of Overbeek. This ex-
pression is then used to derive a simple result for retarded interaction between unequal spheres
at close approach which agrees fairly well with "exact" computations up to separations of about
10% of the sphere radius. An equivalent expression applies to the sphere-plate system but in this
case an even simpler result is shown to give reasonable agreement with exact computations up to
distances of about 20% of the sphere radius. These approximations take the form of simple unre-
tarded Hamaker formulae, together with a correction factor which depends on the separation
distance and the characteristic wavelength of the dispersion interaction. They should find application
in problems of colloid stability and particle deposition.

INTRODUCTION tions greater than a few nanometers.


Although the full Lifshitz treatment im-
In studies of colloid stability or particle plicitly incorporates the retardation effect,
deposition on surfaces it is often necessary whereas the original Hamaker approach
to calculate the van der Waals attraction be- does not, it is not difficult, in principle, to
tween approaching particles as a function of modify the latter to account for retardation.
separation distance. There are, essentially, The usual approach is to carry out Hamaker-
two approaches to such computations. In type integrations for the appropriate geom-
the first, due largely to Hamaker (1), the etry (e.g., interacting plates or spheres),
interaction between macroscopic bodies is using expressions for retarded intermolec-
derived by a pairwise summation of all the ular attraction. However, there is a dif-
relevant intermolecular interactions. The ficulty with this approach, which has not yet
second, more rigorous, approach is based been satisfactorily overcome.
on Lifshitz theory (2, 3) and depends en- This difficulty arises from the lack of a
tirely on macroscopic electrodynamic prop- convenient expression representing re-
erties of the interacting media. Despite a tarded interaction between molecules over
number of shortcomings, the older, Hamaker, the entire range of separation distances.
approach continues to be quite widely used, Overbeek (4) found that the Casimir-Polder
mainly because of its greater convenience (5) results for retarded interaction could
and the fact that, for many practical sys- be represented as an empirical correction
tems, the results are not far wrong. factor to the London dispersion energy, U,
Owing to the electromagnetic nature of between two molecules separated by a
the interaction, there is a finite time of distance R:
propagation between bodies and this leads
B
to r e t a r d a t i o n and a reduced interaction. U - --f(p), [1]
Re
This effect can become very important,
especially in particle deposition, at separa- where B is the London coefficient for the
138
0021-9797/81/090138-08502.00/0
Copyright© 1981by AcademicPress,Inc.
All rightsof reproductionin any formreserved. Journal of Colloidand InterfaceScience, Vol.83, No. 1, September1981
VAN DER WAALS INTERACTION 139

interacting molecules and the correction where A is the Hamaker constant for the
f a c t o r f ( p ) is a function of reduced distance, interacting media.
p = 2"n'R/h, where h is the "characteristic For very small distances Eq. [5] reduces
wavelength" of the interaction, often as- to the unretarded Hamaker result for plates:
sumed to be about 100 nm.
Unfortunately, Overbeek was unable to V = - A / 1 2 7 r x 2. [7]
find a single expression for f (p), but gave
The short-range expression, Eq. [5]
the following two empirical formulae:
contains more terms than Eq. [6] since
0<p<3: both Eqs. [2] and [3] have to be used in the
f ( p ) = 1.10 - 0.14p, [21 integration. For separations greater than
3h/2~-, all interacting molecules must be
3<p<~: separated by at least this distance, so that
f(p) = 2.45//) - 2.04/p 2. [3] only Eq. [3] needs to be used.
Equations [5] and [6] have recently been
Schenkel and Kitchener (6) found an ex- shown (8) to give excellent agreement with
pression valid for p > 0.5, matching the direct measurements of retarded van der
Overbeek results with an accuracy of better Waals attraction.
than 15%: Corresponding expressions for plates of
0.5 < p < 2: finite thickness have been given by Hunter
f(p) = 2.45//) - 2.17//) 2 + 0.59/p 3. [4] (9) and Vincent (10).
It has been found that Eqs. [5] and [6]
Equations [2]-[4] have been used to can be reasonably well represented by a
derive a number of expressions for retarded single expression covering the entire range
interaction in the plate-plate, s p h e r e - of separation distances. The result can best
sphere, and sphere-plate cases. be expressed as a correction factor to the
unretarded Hamaker fiat-plate expression,
INTERACTION BETWEEN FLAT PLATES

For semi-infinite flat plates separated


by a distance x, the integration is quite 1 + bx/h
straightforward and Overbeek (7) obtained
A
the following results for the interaction
energy, V, per unit area: - 1 2 r r x 2 f ' ( x ). [8]

x < 3X/2~-: The constant b can be chosen to give


exact agreement with the Overbeek ex-
V- A[0.0842~_/x
-7 0.093(Tr)x
pressions at any distance, but, for most
purposes, a value fitted at x = 3X/2~- should
be acceptable. At this distance, the required
0.58(~) "~ '~ ~
value is b = 5.32. The correction factor
f ' ( x ) obtained in this way is plotted in Fig.
+ 0.0026 (UI x2 , [5] 1, together with the corresponding factor
calculated from Eqs. [51 and [6] (each used
x > 3X/27r: over the appropriate region and the results
V - A [0.0408( X 1 plotted as a smooth curve). The agreement
is quite good over a wide range of separa-
~- L ~ \ ~ ) tions, the maximum error of about +20%
0.0086 ( h / 2 occurring at large distances. At close ap-
[6]
proach, Eq. [8] with b = 5.32 gives results
Journal of Colloid and Interface Science, Vol. 83, No. 1, September 1981
140 JOHN GREGORY

which are, at worst, about 10% lower than


lo
the Overbeek value. A value of b can be
chosen to give still better agreement at close
approach of the plates but the discrepancy
'[ o.o
at large separations then becomes greater. V(x)
04

INTERACTIONS INVOLVING SPHERES

For spheres, the Hamaker integration


procedure becomes considerably more o
×(nm)
ioo I~o

difficult, owing to the lack of a single


FIG. 1. The empirical retardation factor for plates
expression for retarded intermolecular at-
f'(x) from Eq. [8] with b = 5.32 and X = 100 nm
traction. One method is to use the Schenkel (dashed line) compared with the corresponding factor
and Kitchener expression, Eq. [4], even calculated from the Overbeek expressions, Eqs.
though this is not valid at close approach. [5] and [6] (full line).
Schenkel and Kitcbener (6) derived a result
for equal spheres and Ho and Higuchi The term in square brackets in Eq. [9]
(11) obtained equivalent expressions for is thus a correction factor to be applied to
unequal spheres and the sphere-plate case. the unretarded Hamaker expression.
The results quoted are only appropriate for By putting al = a2 in Eq. [9] the Schenkel
separation distances much less than the and Kitchener result for equal spheres is
sphere radii (d ~ a). Coupled with the recovered and, by letting a2 --~ ~, the corre-
d > M4~- condition (since p > 0.5) and sponding expression for sphere-plate inter-
assuming k = 100 nm, this restriction action is obtained.
means that the results are not appropriate Schenkel and Kitchener (6) also found
for particles smaller than about 2 ~m an empirical expression valid for intermedi-
diameter and for separation distances less ate distances by interpolation between the
than about 8 nm, which in many practical retarded and unretarded expressions, The
cases is the region of most interest. The Ho same correction factor was quoted by Ho
and Higuchi result for the energy of inter- and Higuchi (11) for the case of unequal
action between unequal spheres is: spheres and the resulting expression, valid
d > k/4~-, d ~ a~: for distances up to about k/~- is:
0 <d< k/~'id~al:
Aa,a2 F2.45i k \
V~ = - 6(al +a2)d[-i-~!---£) Aala2 ( 1 )
Vs- 6(al + a 2 ) d 1 + ll.12d/k [11]
It is worth pointing out that the correc-
60~.2 ~- 280~.------T , tion factor in Eq. [11] for retarded interac-
where ai and a2 are the sphere radii (ai < a2) tion between spheres is of just the same form
and d is the separation distance. as that found for flat-plate interaction in
When d ~ h retardation is insignificant Eq. [8].
and the simple, unretarded, Hamaker ex- Recently, Czarnecki (12) has carried out
pression may be used: complete Hamaker-type integration of Eq.
[4] for the sphere-plate case, so that the
d ~ a~: d ~ a restriction is lifted. However, his
Vs
Aa~a~ [10]
expression must still become very inac-
6(al + a~)d curate at separations less than about 8 nm.

Journal of Colloid and Interface Science, Vol. 83, No. 1, September 1981
VAN DER WAALS INTERACTION 141

The result is: expression, suitable for, say, programmable


calculators, giving an adequate representa-
d > M4~r:
tion of the retarded interaction between
[2.45h(d - a d+3a ) spheres over the important region of separa-
Vsp = A L6---~ \ (d + Ua)2 tion. Such an expression can be derived
by the Deryagin method (14), which enables
2.17X2 {d - 2a d+4a I the interaction energy between spheres at
7207r 2 ~ -~a- close approach to be derived from the
corresponding flat-plate expression. The
0.59X 3 /d - 3a ' +5a ll interaction energy between two spheres,
+ 50407r a
/ ~a- (d + 2-~4]J " (12) radii al and a2, separated by a distance
d is given by:
Expressions for retarded interaction be-
tween spheres, based on the two Overbeek Vs- 2-zra--ja2 V(x)dx, [13]
expressions, Eqs. [2] and [3], have been (al + a2)
derived by Vincent (10) and Clayfield et al. where V(x) is the interaction energy per unit
(13). Vincent carried out separate integra- area between parallel flat plates separated
tions of Eqs. [2] and [3], for unequal by a distance x.
spheres, to obtain" short-range" and "long- Using Eq. [8] for the retarded plate inter-
range" formulae, and assumed that the action energy, the integration can be carried
"cross-over point," where the two expres- out immediately to give:
sions gave the same result, represented the
limit of validity for each. This point was d ~ al:
determined graphically for several different Aaaa2
sphere sizes. This procedure is open to gs -
doubt since, even for the short-range case, 6(al + a2)d
the total interaction includes contributions
from regions of the spheres interacting × 1- In 1 + -- . [14]
h bd
over considerably greater distances than
the distance of closest approach of the The term in square brackets is a correc-
sphere surfaces. Thus, Vincent's short- tion factor to the unretarded Hamaker ex-
range expression is unlikely to be reliable pression, Eq. [10] and tends to unity as
except at very close approach of the d/X ~ 0. The constant b could be chosen
spheres. The long-range expression is only to give the best fit with " e x a c t " computa-
valid at separation distances which are too tions, such as those of Clayfield et al. (13),
large to be of much interest for colloidal but, for the sake of comparison, the value
particles. found appropriate for plate interaction (b
Clayfield et al. (13) carried out lengthy = 5.32) will be retained.
integrations for equal spheres and for the Equation [14] only applies to separations
sphere-plate case incorporating both Eqs. much smaller than the particle radius. This
[2] and [3], giving results which should is a purely geometric restriction, imposed
be valid at all separation distances. Within by the use of the Deryagin integration
the limitations of the Hamaker approach, procedure, and has nothing to do with re-
the Clayfield expressions can be regarded tardation. The same restriction applies to
as " e x a c t , " although they are extremely Eqs. [9] and [11] and is not a serious
cumbersome. limitation in practice, as will be shown
It would be useful to have a much simpler below.

Journal of Colloid and Interface Science, Vol. 83, No. 1, September 1981
142 JOHN GREGORY

RESULTS OF COMPUTATIONS
A. Interaction of Equal Spheres \
I n Figs. 2 and 3, the r e t a r d e d interac-
tion e n e r g y b e t w e e n equal s p h e r e s given b y
the n e w a p p r o x i m a t i o n , Eq. [14] (with al \\

= az), is c o m p a r e d with values f r o m o t h e r


expressions:
(1) T h e full, u n r e t a r d e d , H a m a k e r ex-
t~7
pression:

A [ 2a_2 + 2a2
Vs -
6 [ d ( 4 a + d) (2a + d) z

d(4a + d) ] aOI
1o 1oo
+ In (SSa +-dS J " tlSl d(nm)

Fia. 3. Same as for Fig. 2, but for spheres of 100


(2) T h e a p p r o x i m a t e , u n r e t a r d e d , Ha- nm radius and omitting the results based on Eq. [9].
m a k e r e x p r e s s i o n f o r v e r y close a p p r o a c h o f
the spheres, o b t a i n e d f r o m Eq. [10] with a I = a 2 ----a:
d~a:
V~ = -Aa/12d [16]

(3) T h e " e x a c t " r e t a r d e d results o f Clay-


field et al. (13).
(4) T h e " s h o r t - r a n g e " r e t a r d e d result o f
V i n c e n t (10) (his Eq. [10]).
(5) T h e e x p r e s s i o n o f S c h e n k e l and
K i t c h e n e r (Eq. [9] with at = a2), valid f o r
d > h/47r and d ~ a. (N.B. these results
are not i n c l u d e d in Fig. 3 since the ex-
p r e s s i o n is not valid f o r particles with
a = 100 nm.)
(6) T h e interpolated result o f S c h e n k e l
o.1
i ~o 1oo
and K i t c h e n e r , Eq. [11], valid for 0 < d
d (nrn) < h/Tr and d ~ a.
FIG. 2. The effect of retardation on the interaction T h e c o m p u t a t i o n s h a v e b e e n c a r d e d out
of equal spheres, radius 1 /.~m. The reduced interac- for particles o f t w o different sizes: a = 1
tion Vs/A is plotted against separation distance d. Cal- ~ m and a -- 100 n m , with the c h a r a c t e r i s t i c
culations based on: (1) The full, unretarded Hamaker w a v e l e n g t h , h, a s s u m e d to be 100 nm. T h e
expression, Eq. [15]. (2) The approximate, unretarded
Hamaker expression, Eq. [16]. (3) The "exact" re- results are plotted as r e d u c e d interaction
tarded results of Clayfield et al. (13). (4) The "short- e n e r g y , Vs/A, w h e r e A is the H a m a k e r
range" retarded result of Vincent (10). (5) The Schen- constant.
kel and Kitchener expression, Eq. [9], with al = a~. Several points e m e r g e f r o m t h e s e results,
(6) The interpolated result of Schenkel and Kitchener, s o m e o f t h e m already well k n o w n and re-
Eq. [11]. (7) The new approximation, Eq. [14]. The
"exact" expressions for unretarded and retarded inter- stated here:
action, curves (1) and (3), are plotted as full lines and (a) T h e c o m m o n l y u s e d s h o r t - r a n g e unre-
the approximate expressions as broken lines. t a r d e d H a m a k e r e x p r e s s i o n , Eq. [16], is a

Journal of Colloid and Interface Science, Vol. 83, N o . 1, S e p t e m b e r 1981


VAN DER WAALS INTERACTION 143

rather poor approximation and begins to a), i.e., Eqs. [9], [11] and [14], it might
depart seriously from the full expression, be thought that departure from the exact re-
Eq. [15], at separations greater than a few sults would occur at about that distance
percent of the particle radius. When d where the simple Hamaker expression,
= a/lO, Eq. [16] gives results which are Eq. [16] begins to depart from the full,
about 70% too high. unretarded expression, Eq. [15]. Figures 2
(b) Retardation begins to be appreciable and 3 show, however, that the retarded
at very small separation distances, espe- expressions are considerably better in this
cially for larger particles. For a = 1/~m and respect and have a usefully extended range
d = 1 nm (only 1% of the characteristic of validity.
wavelength) retardation already reduces the
interaction energy by 11% and at 10 nm B. S p h e r e - P l a t e I n t e r a c t i o n
separation by about 50%.
(c) The short-range expression of Vin- The case of a sphere and an infinite flat
cent is quite good for the smaller particles, surface is of considerable interest in prob-
where it matches the Clayfield results up lems of deposition and filtration. Usually,
to about 10 nm separation reasonably the particles are so much smaller than the
well, but changes sign at d -~ 40 nm. For collector that the sphere-plate model is a
the larger particles Vincent's expression justifiable assumption.
is of much more limited range. In Figs. 4 and 5, the reduced sphere-
(d) The approximate expression of Schen- plate interaction energy, Vsp/A, is plotted as
kel and Kitchener, Eq. [9], clearly should a function of separation distance for two
not be used at small separations, where different sphere sizes, a = 1 /zm and 100
it gives results which are much too high. nm, as used previously. The computations
For the 1-/zm particles, a practical lower are based on the following expressions:
limit would be d = 5 nm. For smaller (1) The complete unretarded Hamaker
particles, this expression should not be used expression:
at all.
(e) The interpolated result of Schenkel Vsp -
A[a
+ - -
a
and Kitchener, Eq. [11], agrees well with 6 d d+2a
the Clayfield results for separations up to
about 5% of the particle radius. The de- + In d + 2a [17]
partures at greater separations appear to be
due primarily to the geometric restriction (2) The approximate, unretarded, Ha-
d ~ a, rather than to exceeding the stated maker expression, valid only for very close
upper limit of d -- h/Tr (about 30 nm). approach:
(f) The new approximation Eq. [14] gives d ~ a :
the best overall agreement with the Clay-
Vsp = - A a / 6 d . [18]
field results, although for 1-/zm spheres the
agreement is only marginally better than (3) The "exact" retarded results of Clay-
Eq. [11]. Again, the approximation begins field et al. (13).
to become inaccurate at separations greater (4) The expression of Czarnecki (12),
than about 5% of the particle radius, al- valid for distances greater than about 8
though for practical purposes Eq. [14] nm, Eq. [12].
should be acceptable up to d ~- a/lO. (5) A modified version of the Schenkel
(g) For those retarded expressions which and Kitchener interpolation formula Eq.
are restricted to small separations (d [11], in which the numerical constant in the

Journal of Colloid and Interface Science, Vol. 83, No. 1, September 1981
144 JOHN GREGORY

correction factor has been chosen to give an


exact fit with the Clayfield computations
at a separation distance of 10% of the
sphere radius, d = a/lO. For a = 1/xm and
a = 100 nm the required constants are 13.88
and 14.02, respectively. Since these are so
similar, a value of 14 has been used in both
cases, giving:

Vs, ( 1
Aa_6_d\1 + 14d/)t) " [19]

(6) The previously derived approxima-


tion, Eq. [14] with a2 = o% and retaining
\
the value of b = 5.32.
a(nm}
The following comments concerning
these sphere-plate results can be made: FIG. 4. T h e effect of retardation on the interaction
(a) The simple unretarded Hamaker ex- of a sphere, radius 1/zm, with a semi-infinite fiat plate.
Calculations based on: (1) T h e full, unretarded Ha-
pression, Eq. [18], is a slightly better
m a k e r expression, Eq. [17]. (2) The a p p r o x i m a t e unre-
approximation than the corresponding ex- tarded H a m a k e r expression, Eq. [18]. (3) T h e " e x a c t "
pression for equal spheres, Eq. [16], but still retarded results of Clayfield et al. (13). (4) T h e ex-
gives results about 35% higher than the full pression of Czarnecki, Eq. [12]. (5) T h e modified
expression at d = a/lO and becomes much interpolation formula, Eq. [19], and ;t = 100 nm. (6)
The n e w approximation, Eq. [14] with a2 = ~,
worse at larger separations.
b = 5.32, and )t = 100 nm.
(b) Retardation has about the same sig-
nificance as in the sphere-sphere case.
For a = 1 /xm and d = 10 nm the retarded that the results should be accurate where
sphere-plate interaction energy is less than the interaction is large enough to be
half of the unretarded value. significant. Since Hamaker constants for
(c) The expression of Czarnecki, Eq. [12], colloidal dispersions rarely exceed 5kT
gives excellent agreement with the Clay- (2 x 10-20 J), the interaction energy be-
field results at all separations except be- comes insignificant relative to the thermal
low about 5 nm, where it gives values energy of the particles when V/A is less
which are much too high. than about 0.1. Inspection of Figs. 2-5
(d) The modified interpolation formula, shows that the retarded interaction energy
Eq. [19], is an adequate approximation falls to below this value before the geometric
for all separation distances up to about restriction, d ~ a, begins to cause serious
d = a/5. (Because of the fitting procedure, inaccuracies in the short-range approxima-
agreement with the Clayfield results is exact tions. Consequently, because of their
at d = a/10). greater simplicity, the short-range expres-
(e) Equation [14] is a good approximation sions are likely to be acceptable ap-
up to about d = a/10, but becomes inac- proximations.
curate at larger separations. For most practical problems, retarded
interaction can be represented by the ap-
propriate unretarded Hamaker expression
CONCLUSIONS
modified by a simple correction factor,
In seeking a convenient approximation which depends on the separation distance
for practical purposes it is only necessary and the characteristic wavelength.

Journal of Colloid and Interface Science, Vol. 83, No. 1, September 198t
VAN DER WAALS INTERACTION 145

sphere radius. At greater distances, the


interaction will usually be insignificant, but
if computations have to be carried out in
this region the Czarnecki expression, Eq.
[12], can be used. This must not be used for
interaction at close approach.
In calculating van der Waals interaction
for fairly large particles (1 /zm or greater)
neglect of the retardation effect leads to a
serious overestimate of the interaction,
even at separation distances of only a few
nanometers. The use of very simple ap-
proximations for retarded interaction, such
as Eq. [19], requires hardly any more ef-
lO d ( n m / lOO 2o0
fort and gives greatly improved results.
FIG. 5. Same as for Fig. 4, but with the sphere
radius 100 nm. REFERENCES

1. Hamaker, H. C., Physica 4, 1058 (1937).


For the interaction of plates, Eq. [8] with
2. Lifshitz, E. M., Soy. Phys. JETP 2, 73 (1956).
b = 5.32 is an adequate representation of 3. Parsegian, V. A., In "Physical Chemistry: En-
the two Overbeek expressions, Eqs. [5] riching Topics from Colloid and Surface Science
and [6]. Theory" (H. van Olphen and K. J. Mysels,
Retarded interaction between spheres can Eds.), pp. 27-72. Theorex, La Jolla, Calif.,
be adequately approximated by the simple, 1975.
4. Overbeek, J. Th. G,, In "Colloid Science"
short-range, Hamaker expression, Eq. (H. R. Kruyt, Ed.), Vol. 1, p. 266. Elsevier,
[10], together with a correction factor. Amsterdam, 1952.
This factor can take the form given in 5. Casimir, H. B. G., and Polder, D., Phys. Rev.
Eq. [11] or in Eq. [14]. While the latter 73, 360 (1948).
6. Schenkel, J. H., and Kitchener, J. A., Trans.
gives somewhat better agreement with exact
Faraday Soc. 56, 161 (1960).
results (and might be still further improved 7. Overbeek, J. Th. G., quoted by A. van Silhout,
by adjustment of the constant b), the simple Proc. Konigs. Akad. Wetensh. B. 69, 501
interpolated formula, Eq. [11], may be good (1966).
enough for most purposes. Both of these 8. Gregory, J., J. Chem. Soc. Faraday 1, 73,
1983 (1977).
approximations become better as the par-
9. Hunter, R. J., Austral. J. Chem. 16, 774 (1963).
ticle size increases. 10. Vincent, B., J, Colloid Interface Sci. 42, 270
For the important sphere-plate case simi- (1973).
lar remarks apply, but here the short- 11. Ho, N. F. H., and Higuchi, W. I., J. Pharm.
range Hamaker expression, Eq. [18], with a Sci. 57, 436 (1968).
12. Czarnecki, J., J. Colloid Interface Sci. 72, 361
suitable correction factor, gives even better
(1979).
agreement with exact results. In this case, 13. Clayfield, E. J., Lumb, E. C., and Mackey, P. H.,
the factor given in Eq. [19], can be used J. Colloid Interface Sci. 37, 382 (1971).
up to separations of about 20% of the 14. Deryagin, B. V., KolloidZ. 69, 155 (1934).

Journal of Colloid and Interface Science, Vol. 83, No. 1, September 1981

You might also like