Theory of the Stability of Strongly Charged

Lyophobic Sols and of the Adhesion of Strongly

Charged Particles in Solutions of Electrolytes

By B. D e r j a g u i n and L. Landau

1. I n t r o d u c t i o n
The fundamental task of any quantitative theory of the stability of
lyophobic sols is the derivation of a quantitative criterion of their
s t a b i l i t y - - a criterion giving the relation between the quantities charac-
terizing the properties and state of a sol in the transition zone between
the unstable and stable states.
When we speak here of stable states we mean not the states which
are stable thermodynamically, but those which are stable in the sense
that the rate of coagulation of tile system is practically equal to zero.
Accordingly, calculations of the criterion of stability should be based
on the determination of the kinetics of the irreversible transition from
the unstable state of sol to the more stable state of coagulate in the
absence o f equilibrium between these two states, in this point our
treatment of coagulation differs in substance from that given by
A. M a r c h l, as well as from L a n g m u i r's treatment of coacervation a
according to which coacervation is considered as the transition between
two states capable of coexisting in equilibrium like two phases.
In order that a quantitative criterion of coagulation could exist,
it is obviously necessary that there should be a sufficiently abrupt
transition from kinetically stable states of the sol to unstable states.
And indeed, tile calculations which constitute the subject of this, as well

i A. M a r c h, Ann. Physik, 84, 605 (1927); Koll. Z., 45, 97 (1938).

I. L a n g m u i r , J. Chem. l'hy,s., 6, 873 (1938).

30
 Selected Works - 2 31

as the previous 3, investigations, indicate that this transition is actually

extremely abrupt. Moreover, the larger the radius of the sol particle,
the more abrupt is the transition (for a more detailed discussion see
Sect. 7). In connection with this last condition, the limit of completely
unstable states is particularly sharp for emulsions the radius of whose
particles increases during the process of coagulation, and for which
the conditions of complete coagulation therefore coincide with the coagu-
lation conditions of extremely coarsely disperse emulsions.
It is beyond all question that coagulation of a sol is due to the
forces of interaction arising between micelles while they are approaching
one another. Therefore, in calculating the rate of coagulation and the
degree of stability for the purpose of arriving at a criterion of stability,
one must know a quantitative expression for the force of interaction
of two particles, in the simplest case spherical ones, as a function of
the distance between them. After this, as may be shown, it is not
a particularly difficult matter to calculate the criterion of stability. Hence
the main, and at the same time the most difficult, part of a theory of
the stability of sols (and not only lyophobic ones) is the calculation
of the forces or energies of interaction of their particles as a function
of the distance between them. Such a problem was posed for the first
time by H. K a l l m a n n and bl. W i l l s t ~ t t t e r in a very short note 4
These authors consider the interaction of particles as the algebraic sum
of two terms: attraction, depending upon the v a n d e r W a a I s forces
acting between the molecules of file micelles, and repulsion, depending
upon the electrostatic forces between the ions of the electric double layers
surrounding each particle.
From this point of view coagulation should evidently be attributed
to the action of the v a n d e r W a a l s forces capable of bringing
about aggregation of the particles and inevitably causing it in the
absence of forces of repulsion. Thus, the stability of lyophobic sols
is due to the forces of electrostatic r e p u l s i o n - a point of view which,
by the way, had been advanced considerably earlier 5 than H. K a l l -
m a n n and M. W i l l s t ~ t t t e r ' s note appeared.
To find a criterion of coagulation corresponding to the moment
when the forces of attraction overcome the forces of repulsion, it is

s B. D e r j a g u in, Trans. Farad. Soc., 3~, 730 (1940); 36, 203 (1940).
4 H. K a l l m a n n u. M. W l l l s t , ' l t t e r , Naturwiss., 20, 952 (1932).
,~ F. D o n n a n, sec R. E ! lis, Z. physlk. Chem., 80, 597 (1912).
32 B. V. Derjaguin

evidently necessary to obtain an expression for both kinds of forces

as a function of the distance between the particles. For the energy
of the v a n d e r W a a l s interaction between two molecules at suffi-
ciently large distances R as compared with their radii, the following
expression holds:
U--- CR -e', (1)

where C is a constant depending upon the nature of the mole-

cules. As shown by H. K a l l m a n n and M. W l l l s t t l t t e r , using this
expression it is possible to find the energy of interaction (in vacuo)
between two micelles by simple integration with respect to their volumes.
The calculation of the energies and forces of attraction of miceUes
is thus quite elementary. Exact formulae for the case of two spherical
micelles not only of equal but also of different radii were given later
by H. t t a m a k e r ~
The calculation o f the electrostatic repulsive forces is a conside-
rably more complicated and interesting problem. At file same time,
a quantitative study of the forces of repulsion is the key to the problem
of the stability of lyophobic sols and to a number of other problems of
the physics of colloids and surfaces.
The main difficulty in calculating these forces lies in the determi-
nation of the distribution of ions. This must b e done before one may
sum up the forces of electrostatic interaction between the ions. In other
words, it is necessary to determine the concentration of the various
ions in the neighbourhood of the micelles, in the first place in the zone
located between them, where the ionic atmospheres of neighbouring
micelles overlap and deform one another. For this purpose it is neces-
sary to solve tile equation of the D e b y e - H f i c k e l theory of strong
electrolytes. Any calculations undertaken in an attempt to find the forces
of electrostatic interaction without clearing up the question as to the
changes in the distribution of the ions, which occur as a result of the
approach of two miceUes, may lead even in a qualitative respect,
to erroneous results, and all formulae thus derived are, in the best
case, only empirical in character. Similar formulae were used by
H. F r e u n d l i c h v, H. H a m a k e r ~, and R. H o u w i n k and

6 H. H a m a k e r, Physika, I0, 1058 (1937).

H. F r c u n d I i c 11, "Thixotropy", |Saris, 1935.
8 H. H a m a k c r, Rcc. Tray. Chim. Pays-Bas, 55, 1015 (1936); 56, 3 (1937);
56, 727 (1937).
 Selected Works - 2 33

W. B u r g e r s", to arrive at certain qualitative conclusions. Indepen-

dently of these investigations, and maiflly in connection with personal
experimental investigations of the "disjoining pressure" of thin liquid
films lo, one of us undertook a theoretical study of the repulsive forces
acting between lyophobic micelles, and of the stability of lyophobic sols.
In a paper which appeared in 1937 ~ calculations of the forces of
electrostatic repulsion of ions were carried out for the first time
by solving the D e b y e - H f i c k e l equation for the cases both of two
spheres and of two plates separatcd by a plane parallel layer of electro-
lyte. The solution arrived at holds ri#dly for the case of plates and
spheres with radii which are large in comparison with the D e b y e
thickness of the ionic atmospheres. Inasmuch as the calculations were
carried out by a simplified (linear) D e b y e- H ii c k e I equation, the
formulae obtained apply only to sols with sufficiently low charges.
In that paper the attraction forces between miceUes were taken into
account quite rottghly: by analogy with S m o I u c h o w s k i's theory
of rapid coagulatiotl they were supposed to possess an infinitely short
radius of action as compared with the thickness of the ionic atmospheres.
A stability criterion was eventually obtained which led to a critical
value of the ~-potential approaching that observed experimentally but
too strongly dependent upon the radius of particles (inversely propor-
tional to the square root of the radius).

2, T h e stability c r i t e r i o n of w e a k l y c h a r g e d sois
In a later work by B. D e r j a g u i n ~ 2 the theory of stability was
refined by introducing .'m expression for the v a n d e r W a al s forces
on the basis of equation (1). The stability criterion obtained in this
manner for weakly charged sols or emulsions has the following form:
xA
"/)':~]]._r--- const., (2)

u R. H o u w i n k and W. Btt r g e r s, "Elasticity, Plasticity :rod Strttctnre

of M a t t e r', Cambridge University Press. 1937, pp. 3 ~ -343; R. I-[ o tt w i n k,
"Second Report on Viscosity and Plasticity" (Comutittcc of the Acad. Sci.
Amsterdam) (Nordmann l~ublishing Company, N. Y.. 193~), pp. 233--237.
Jo B. D e r j a g u i n and M. K u s s a k o v , Acta Physicochimica U[{SS,
10, 25 (1939); Bull. Acad. Sci. URSS, St~r. Chim., No. 5, 1119 (1937).
11 B. D e rj a g u i n , Acta Physicochimica URSS, 10, 333 (1939): Trans.
F,'u'ad. Soc., 36, 203 (1940); Bull. Acad. Sci. URSS, S~r. Chitn., No. 5, 1152~
(1937}.
v, B. D c r j a g u i tt, Tratls. l'arad. Sot., 36, ?30 (1.940).
84 B. V. Derjaguin

where x is the reciprocal of the thickness of the ionic atmosphere,

A is the van der W a a l s constant (see Sect. 5), D is the dielectric
constant, and ~o is the potential of the surface of the particles. This formula
gives a theoretical basis to the empirical rule found by H. E i l e r s
and J. K o r f f is, according to which the coagulation of identical emul-
sions by different electrolytes occurs at a constant value of the expression
X
~ - - const., (2')

where ~ is the electrokinetic potential. Criterion (2) is applicable only

"to weakly charged and sufficiently large particles, such as the droplets
of ordinary emulsions, since it was derived on the basis of a simplified
D e b y e- H {l c k e I equation, applicable when the electrical potential in
the neighbourhood of the surface of the particles satisfies the inequality:
50
+ < }~ mV, (3)

where z' is the valency of the ion with the h i g h e s t c h a r g e .

However, at the threshold of coagulation for i,J > 1, condition (3)
is evidently fulfilled extremely rarely, particularly for the particles of
sols; and even when z'--=-1 it is probably far from being always ful-
filled. That is why it becomes necessary to alter the mathematical basis
of the theory and introduce the complete, non-simplified D e b y e-
ll il c k e 1 equation (see eq. 5).

3. Summary of contents
In Section 4 equation (5) is applied to a precise cal6ulation of the
force of interaction of two charged parallel planes of infinite dimenSions
separated by a solution of an electrolyte. It was found posSible to intro-
duce into the further calculations an extremely important Simplification
by assuming that the potential ~o to which the planes have been charged
is sufficiently large, vtz.:
100
+o> z, my, (4)
where z x is the electdcal valency of the dominating ion. It is beyond
all doubt that condition ( 4 ) - - t o a certain extent the antithesis of con-
dttion (3) - - should hold, even when z I --- 1, for a large number of cases,
which makes the assumed approximation quite natural.

zs H. E l l e r s and I. Korff, Trans. Farad. Sot., 36, 229 (1940).

 Selected Works - 2 35

In Section 5 the formulae obtained are used to find the condl,'ions

of the adhesion of plates in an electrolytic solution under the influence
of v a n d e r W a a I s f o r c e s - - in other words, the conditions of stabi-
lity of the electrolyte film separating them. However, the calculations
could be completed, i. e., the stability criterion could be obtained in an
explicit form, only for the case of a symmetrical electrolyte or an electro-
lyte in which the ratio of the valency of the auxiliary ion to the valency
of the dominating ion is equal to two. The results obtained may
be regarded as evidence of the insignificant r61e played by the valency
of the auxiliary ion in the effect of the electrolyte upon the stability of
the film.
The criterion obtained is close to, but by no means identical with,
W o. O s t w a I d's rule ~ the applicability of which to the phenomena
of the adhesion of quartz surfaces has been proved by A. B u z~ig h ~5
Section 6 contains calculations of the interaction and adhesion of
strongly charged spherical particles in the solution of a symmetrical
electrolyte. A criterion of adhesion is obtained which coincides to within
a numerical factor with an analogous criterion for the case of_ parallel
plates.
Calculations of the condition of stability of strongly charged sole
which lead to a criterion of the same type, are given in Section 7;
finally, Section 8 contains a summary and discussion of the data
obtained, as .well as a comparison of the results with experimental data.

4. T h e i n t e r a c t i o n of plane s u r f a c e s
The complete D e b y e - H ti c k e I equation for electrical potential
as a?plied to a binary electrolyte with ions of two types has the form:

~,e - - e-"~'~), (5)

where z I and z~ are the valencies of the dominating and auxiliary ions,
respectively;
8~ 8r. 8~
a = --~ e n a z I "f "-- - ~ ca.,. r.z "f ~ . ~ ec,

C --- n I z I " [ - - " n._, z z "{ (5')

and
e
b--
kT ~
J4 Wo. O s t w a [ d, KolI. Z., 73, 301 (1935); 88, I (1939).
1,5 A. B u z ~ig h, Ko[l. Z., 76, 2 (1935).
36 B. V. Derjaguin

where D is the dielectric constant of the solution, e is the charge of

an electron, T is the concentration in mols per cm.z, c is the concen-
tration in gram equivalents per cm.3, n I and n 2 are the number of anions
and that of cations, respectively, in a mo-
Z
lecule of the electrolyte, k is the B o l t z -
m a n n constant and T is the absolute tem- i
i

perature.
Considering the interaction of two in-
finite parallel plates situated at a distance
.Z"
h apart (Fig. 1) in an infinite volume of
the electrolyte solution, let us place the ----b
ori~u midway between the two plates and
direct the X-axis perpendicular to the surfaces
of the plates., Assuming ~ to be independent
of 3' and z, instead of equation (5) we F i g . I.

may write:
dx"- --- e " tr~ - - e -- ,~ b.~). (5")
Multiplying both sides of equation (5") by 9- ~x
dqJ dx--- 2d~ and inte-
grating, we obtain:
~']2 a F.e~," e - ' , H' C~
e x l --- b L ", 4 ,,.., , (6)

where C is the integration constant. We shall assume as before n , ~ that

the following boundary condition holds at the surfaces of the plates:

+ - - - +o - - const. (7)

For the intermediate electrolyte layer between the plates, C has the
following value:
e& H~, e - - , 2 vh
C~- C, - - - - -4 - - , (8)
Zl Z2

where d?l is tile value of ~ when x , - - 0 , since symmetry considerations

require that the electrical force E : dqJ should be equal to zero
dx
when x - - 0 . For the external volume of the electrolyte we have:

I._ 1
c-- G-- (8')
 Selected Works - 2 37

~dJ
since at x ~ ~ co, E "-- - - ~ --- 0 and d / ~ 0. Following lhe method
employed previously ~", we may represent the interaction (repulsion) o~
the plates P(h) calculated per unit area as follows:

D D a

(9)
_- - c k T E ( e " V l "zl- - l ) + ( e ' - " H"
z2 - - 1) 3

where E 2 and E 1 are the values of the electrical force at the external
surface of the plates (where it is larger) and at the internal surface,
respectively. In equation (9) the mechanical force acting on each plate
is equal to the difference in the tension of the lines of force on both
sides of the plate, whereas the osmotic pressures acting there are balan-
ced, since they are functions of the potential d/ only, which has the
same value d/----+o at both surfaces of each plate.
As may also be seen from (9), one may arrive at the same value
of P ( h ) by determining the excess osmotic pressure of the ions in the
plane x - - 0, as has been done, for instance, by I. L a n g m u i r 17. The
two methods of calculation are therefore absolutely equivalent. This
circumstance, which has been noted, in particular, by A. F r u m k i n is,
has for some reason escaped the attention of S. L e v i n e 1 9 , who
critidzed the works of B. D e r j a g u i n and [. L a n g~n u i r, refuting
them on d i f f e r e n t grounds. As a matter of fact his main reason for
rejecting them-lay in the disagreement between the formulae obtained
by B. D e r j a g u i n and I. L a n g m u i r , and those obtained b y S . L e -
v i n e himself. The reason for this disagreement has been pointed out
by one of us in a previous paper 2° and consists in the erroneous
expression for t h e free energy of charged surfaces placed in an electro-
lyte; in particular it has been shown that the expression for the free
energy found by S. L e v i n e by a statistical method incorrectly applied
contradicts thermodynamics. Expressions equivalent to that employed
by S. L e v i n e and, consequently, also erroneous, are used by

16 B. D e rJ a g u i n, Acta Physlcochimica URSS, I0, 333 (1939).

17 I. L a n g m u i r, J. Chem. Phys., 6, 873 (1938); Science, 88, 430 (1938).
Is A. F r u m k i n and A. G o r o d e t z k a j a, Acta Physlcochimica URSS,
9, 327 (1938).
19 S. L e v l n e and G. D u b e , Trans. Farad. Soc., 35, 1125 (1939).
~o B. D e rJ a g u i n, Trans. Farad. Soc., 36, 203 (1940).
38 B. V. Derjaguin

A. C o r k i l l and L. R o s e n h e a d ~ in calculations of the interaction

of parallel plales in an electrolyte; it ts thts error that accounts for t h e
discord between our results and tlietrs, in particular for the attraction
of plates with like charges which results from their calculations 22.
In order to derive from e q . ( 9 ) a n expression for P ( h ) in an explicit
[,,rm in terms of h, it is necessary to express C~ in terms of h in
this equation.
The relation between h and C~ is obtained by integrating equa-
tion (6) in the following form:
+o

.}- h --- ~' e - " v~

~, v zl -' "2 C1

-f ¢o

I / ~ , b*
!/--- +
e - - "~ b'[~ e . t b'h e - - . . b'~,
(10),

~, J" Z1 Z2 Z1 Z2

it being taken into account that when h - - O , ~q'--+o. We shall now

introduce an extremely important simplification: supposing that
+o>O,
we assume that the following relation holds:
b+o 1 (ll)
which practically coincides with condition (4).
This makes it possible to replace the upper limit ~0 of the integral
in (10) by oo without appreciably affecting the accuracy, since the
integrand for ~ ~ o will be extremely small unless in the case o f an
exceedingly small value Of h, the difference ~0--~,1 becomes too small.
Using the notations:
c k T ' - - Tn, z , k T -'- yn t z z k T --- p, (12)

(12')
V C ~ ' - - X ' - - V D kT'-- --xa-*- z2 x'
21 A. C o r k i l l and L. R os e n h e a d, Proc. Roy. Soc., London, (A) 172,
410 (1939).
22 The existence of repulsion, and not attraction, between similarly charged
air--thin water film interfaces is in agreement with the observations made
by B. D e r j a g u i n and A. T i t i j e v s k a j a on the stability o[ such films.
 Selected Works - 2 39

/,q, = l,+, (12")

1

where d is tile De b y e thickness of the ionic atmosphere, we may

rewrite equations (9) and (10) in the following form:

~ e Bt ~ - - 1 e-z2 ~ _ 17
P-- P -4 .j, (13)
and
CD

.........
~,zl lit
:-::- =:
e ~ , , ~ 7~t
" (14)
t / ~ ~'l TI --|- e--:' TI
-

"q! .71 2., "l Z2

Equations (13) and (14) express the relation between P and h in

parametric form: to express this relation explicitly, it is necessary
to eliminate Y,I from (13) and (14). it is easy to see that the relation
under consideration may be expressed with tile aid of elliptical integrals
for the cases where [.~m z.2
z~ __1 ' 2, T1 From examination of (13)
and (14)it is seen that with an increase in "~l front 0 to o0, /2 increases
steadily from 0 to co, whereas h, on the contrary, steadily decreases
from co ~o 0. Consequently, with a decrease in h, /2 increases conti-
nuously.
Let us now examine the approximate relations which are valid when

i. e., when, owing to the large value of "r,~, which is tile case when h
is sufficiently small, only the terms dependin~ upon the dominating ion
may be retained in equations (13) and (14).
Then, instead of (13) and (14), introducing into the latter equation
the integration variable ~ - - z i ('n re'n1), we shall have, appro.,dmately:

21
(16)
.and

0
40 B. V. Derjaguin

F.om (16) and (16') we find:

4~ ~ 1 2p (z 1 -*- z~) =.'-' 1
D (-k7~/" I . (17)
P - - - P zl~×''-' h'-' --= zl °- (×h)~ ~ 2 kz 1 e/ h-'

For the special case where z j - - z , , this formula has been found
by 1. L a n f f m u i r 17 From comparison of (13) and (14) with (16) and
(16') it follows that the approximation made increases the values of
P and decreases tile values of h corresponding to a given value of P
and, consequently, afortiori, increases the values of P corresponding
to the given values of It, i. e., formula (17) always gives exaggerated
values o f Pe:~. This is true not only with respect to devialions of P from
formula (17) at large values of h, but also with respect to deviations
from the same formula which should occur at very small values of h
at any finite value of tile potential of tile plates +0, however high it
may be.
This remark bears relation to the following conclusion of a physical
nature resulting from a similar examination: when account is taken
e - s ' ~ ~" - - 1
of the influence of the auxiliary ion [for example, the term
22
in formula (13)] the pressure P as calculated for a given value of h
may only be reduced.
Let us now examine the case of large values of ×fh and small
values of "~x and accordingly assume that
"~1.

Expanding the exponential functions involved in (13) in a series, we obtain

the following relation for this particular case:
p __ P (zl ÷2 z~) ~12" (18)

We shall further assume that the electrical potential at a point in the

intermediate layer ~emote from the surfaces of both plates may be found
by simply adding the potentials which would be produced at this point
by either of the plates in the absence of the other, i. e., assuming
these effects to be additive~L Then we may write

I. L a n g m u i r's opposite statement is, consequently, incorrect, and

equally erroneous in the same respect is hls graph of In o plotted against In h.
~4 This assumption m a y also be proved in a purely analytical way.
 Selected Works - 2 41

where ~ ( h ) is the value of the potential produced e t a distance(-~)

from one charged plate in the absence of the other.
Since for small values of ~ ( ~ - ) we may use the equation
\2/

d~, __=_x2~, (19)

d.r2
or
d2v] (19')
dx.p : x2.~

instead of (5"), for large values o! ~ ~,--" b+ has the [ollowing form:

- - ~o e-'~, (20)

where ~o is a constant which depends only upon z~ and z 2 and,

consequently,
1
-- lh (21)
"~t = 2"t),o e 2
and
P-- 2p (z 1 + z~) Zo~ e -I~ --- Po e-=h, (21')
where Po is another constant.
For a symmetric electrolyte with zl----z:---z and nl---n2---n
4
"~o'----z (22)
and, consequently,

P "--- Po e-Xh - - 64p e_Z~, ..._ 64TnkTe_=h"

z
(22')
For an asymmetric electrolyte, -% is calculated in a rather cumber-
some manner from the relation

In (z, ~ , o ) - r ( f 3 ) =
1 GO

~--- V/ 2. _1_ ,dt+

1 t /.~ -¢- e-t~ 1
/et e- ~ 1 . -V- 1
0
Z2
where ~ - - ~ - , which is obtained by comparison of tile value of xh
obtained from relation (20) with ihat which follows from the exact
42 B. V. Derjaguin

formula (14) after Introduction of the variable of integration t ---z~o

go
and the parameter [3.-------~ takin~ into account eq. (12r).
zl
Let us now examine in greater detail the exact solution for the case
of a symmetric electrolyte, assumin~ that

aod (23)
Il I -~ II,_ --'- II. I

After introduction of a t~ew variable of integration

-"L
k cos h vp -7"
where
1
L1
(23')
cos hyp 7 "

relation (14) in this case is transformed in the [ollowin~ manner:

co

ht
cO

zd~ du --- (24)

---
f t/; 2
..... - ~/e
hyp ~ ? - - cos hyp'-' --'21
.[ ~/[t -- uz) (1 -- kz .z)
o

---- k K (k) - ~ sin ~,.K(sin ~.),

where K ( k ) i s the usual symbol for a complete elliptical integral of the

first kind with a modulus k - - - s i n e..
With the aid of (12), (23) and (2a'), instead of (la) we may write

P'--4TnkT(-~=_,--1 ) - - 4 T n k r cos hyp~ 0~. (25)

Eqs. ( 2 4 ) a n d (25) deternfine tile P-h relation in a parametric form

which is somewhat more convenient that the analogous representation
Ol the same relation given by 1. L a n g m u i r ~7
Making use of the asymptotic expression for K(k) when k---~-1
one finds no difficulty in deriving the.relations (22) and (22').
 Selected Works - 2 43
In Figs. 2 end 3 the quantity
P
~nk T = co
is plotted against v.h using both the ordinary and the logarithmic scales.
The graph in Fig. 3 corrects the error present in [. L a n g m u ir's paper.

k/

I0
I ¢#
ez¢ct o¢lue'. ~::."
.I . . . .

--- - w'6~le"*l¢¢ ;'~'1 1£':

¢0

ZO
Ix
I [[J
,'[
t

G I IO x b

Fig. 2. Fig. 3.

The calculations were made with the aid of tables giving the nume-
rical values of the function K(k).
For comparison, curves expressing the approximate relations be-
tween ,.~ and xh according to fornnflae (17) and (22') are likewise
given in Fig. 2.

5. C o n d i t i o n s of a d h e s i o n u n d e r the action of v a n d e r W a a l s
f o r c e s for s u r f a c e s s e p a r a t e d by a l a y e r of e l e c t r o l y t e

Let us now take into account the v a n d e r W a a ls forces Which

decrease inversely as the seventh power of the distance between the
molecules and produce, per unit area of the film separating two parallel
plates of considerable thickness, an attraction

A
Q--~, (26)

where A depends upon the material of the plates and upon the inter-
mediate medium between them. From comparison of formulae (17),
44 B. V. Derjaguin

(21') and (26) it follows that tile forces of attraction preponderate

at very large and very small distances. DenoUng the resultant r e p u l -
s i o n by

re (h) --- P - - Q (26')

we see that the curve R, h

may be of the three basic
types shown in Fig. 4. g/
It is obvious that only
under conditions giving rise
to a curve of type III with a
maximum lying above the
axis of abscissae may the film
possess any stability and pre-
vent adhesion of the surfaces Fig. 4.
which it separates.
A limit of stability will be observed when the peak of this
maximum lies on a level with the axis of abscissae, i. e., when the
following relations exist simultaneously:

b)z---- 0, (27)

and
R -- 0 (27')

or, in accordance with (26),

OP A (28)
"Oh w 2.h4

p_ A (28')
6~hZ

whence we conclude that

d In P _ _ _ _ 3. (29)
dlnh

After the values of P and h from (13) and (14) expressed in terms
of the parameter z~-n~ have been substituted into equation (29) the
latter determines the value of this parameter at a given value of the
 Selected Works - 2 45

ratio [~'-- z'~" whereupon from the same relations (13) and (14), we
may also determine the quantities

('-~)o=wc = u (f~) (zo)

.and
(~h)c= ~(~). (30')
Substituting these quantifies into (28'), we obtain a criterion of the
stability of a film or the criterion of the adhesion of plates in the form:

x3 A
~.I kT= C(~), (31)

where C(~) is a function of the quantity [3 alone. An exact calculation

of the quantity C for different values of [3 involves laborious nume-
rical computations. Therefore, to start with, we will only examine i n
e x t e n s o the simplest case to calculate, v i z . , that of ~---1, correspond-
ing to symmetric electrolyte.
In this case, making use of equations (24) and (25), we may
write instead of (29):
l _ ,)
d In (t'-~-
dk
- - =--3. (32)
dlnkK
dk

Employing the weU-known relation from the theory of elliptical

functions ~5

dkK E
dk ~ l ~ k 2~

where E is a complete elliptical integral of the second kind with the

modulus h ---- sin ~:
1

E--" f ¢ 1 - - k'-'u*- l/t/,

• ql -- uz
o

z.~ See, e.g., E. W h i t t a k e r and G. W a t s o n , "A Course of Modern

\nalysls" 1V Ed. (Cambridge, University Press, 1927), Ch. 22, § 736.
46 B.V. Derjaguin

after simplification we obtain from (32)

K 3 (32')
~ - - - -3-"
Hence, employiug the tables of the values K and E, we find the
root of this equation:
k - - 0.773... (32")

Using this value of the modulus k we find from (24) and (23')
the "critical" values of the quantities 0'.h) and (z-e,1):

(nh)¢-- 3 . 0 1 4 . . . , (33)

(z~t) ~ - - 1 . 5 . . . , (339

Pc -'- 2.68 (33")

~'tzkT ....
Criterion (31) assumes the following form:
x3 A B ~ 1390, (31')
kT--
or

In ~
1 A~ e~ zI'; i
where I (31")
B ~ , - - 0 . 8 1 9 . . . . . 10-°" I
C ~ - - (8~)3

,:ld m, the reciprocal ',e coagulation concentration of the dominat-

ing ion, may be taken a.~ measure of the coagulating concentration
of the electrolyte.
The values ( 3 3 ) a n d (3:t')show that the "critical poinff lies at
such thicknesses for which one may, as a first approximation, attempt
to use formulae (21') and (21), and, in fact, if this attempt is made
and the value (2U) is substituted into (29), we obtain

(xh)J --- 3

which is close to the value given in (33). If we now substitute this

value for >~h into (22'), and also into (21) and (2Y), we obtain:

(Z~l)c' - - 1 . 7 8 . . . , (34')
 Selected Works - 2 47

D
' -- 3.20 (34")
•~')e ~ Tnlgl "'''

B' "~m1630. (34 m)

It follows that the error involved is not great.

The value of C found in (311') holds only for z 2 - - z ~.
However, since in all the formulae expressing P ( h ) the terms
depending upon the auxiliary ion are of secondary Importance, the
influence of z 2 upon the numerical value of C Is not great.
The above calculations may be extended ~e to the case of
z,
{~--- ----2 and it is found that the constant C then decreases only
Z1
by approximately 4°/0. It is thus confirmed that the influence of the
valency of the auxiliary ion is slight.

6. Conditions of adhesion of convex surfaces separated

by a l a y e r of electrolyte
In the theory of adhesion previously developed by one of us 27,
it was shown that one may pass over from consideration of the inter-
action of plane surfaces to the interaction of convex surfaces of the
same nature and in the same medium by employing the following for-
mula:
QD

N---- 0 .[ R (h) dh, (35)

H
where R(h) is the repulsion per unit area of surfaces separated

Fig. 5.

by a plane parallel layer of thickness h, N is the force of repulsion

of two spheres at the shortest distance H between their surfaces

e~ We omit these calculations as Mey comprise part of the calculations

undertaken by B. D e r j a g u l n with a view to extending the theory to
asymmetrical electrolytes.
'-'7 B. D e r j a g u i n, Koll. Z., 69, 155 (1934).
48 B.V. Derjaguin

(Fig. 5), and Q Is a quantity depending upon the shape (radii of cur-
vature) and the mutual orientation of the surfaces in the region of
their minimum separation.
For two spheres of radius r we have:

C -'- ::r. (35')

Formula (35) gives a good approximation for tile cases in which

r is sufficiently large.
In case the interaction of the surfaces is of a purely electrostatic
nature, the values of r satisfying the condition

z.r ,~: 1 (36)

will be "sufficiently large".

For the case of forces which obey formula (1), instead of (36]
we must examine the condition:

r
ft >~ 1. (36')

When the forces of both kinds are acting simultaneously, it is

necessary to examine both conditions, (36) and (36'), simultaneously.
Since, however, the values of N at values of H which are large
1
compared with d - - will only very slightly influence the further
X

calculations in comparison with the influence of the values of N at

smaller values of H,-for all practical purposes it is sufficient to take
into consideration condition (36) alone, assuming that (36 t) will, as it
were, be its corollary.
In any case, assuming that the conditions of applicability of equa-
tion (35) are fulfiUed, we obtain from (35), (28'), (26) and (26'),

N A 1 (37)
G - - - ] (H) 12~ 1t 2
where
Oo

1 (H) - - .[ P (h) dh (37')

H
 Selected Works - 2 49

and, Instead of (28) and (29), as a threshold condition of adhesion

Of the particles we get:
d in / (H) ~ ~ 2, (38)
dlnH
A
I (H) H 'z'-" 12~ • (383
For the particular case of a symmetrical electrolyte, we find from
(24) and (25):
oo

ll

i[
1
8"fnkT[ I ) 8ynkT" 2K
z k'~ ~ 1 kK-I---~ .[ die

or, in accordance with a property of complete el'iptical integrals that

may be readily proved,

(39)

H--- 2kK. (39')

From (38) and (39), after simple transformations we get

4E - - 3 (1 - - k-') K - - 4k = 0,
the root of this equation being
k ~ 0.543... (40)
From (40) and (39') we obtain the "critical" thickness,
:,.H.-- 1.858... (40')
and the criterion of adhesion

x'~A - - ' B " - - 1160, (41)

"~n k T"
or

II1 - - - -
1 .._ Cr A" eqlzta
~n I Ds (kT)n
50 B.V. Derjaguin

where
C'--- (8~)3
(B'p-- 1.174 . . . 1 0_~"

We can see that not only does the form of the criterion of adhe-
sion remain unaltered, but even the numerical value of its constant
changes but slightly on transition from plane surfaces to convex ones.
This is very fortunate and important from the point of view on an
experimental check of the theory or its applicability to systems contain-
ing particles irregular in shape but sufficiently large.
It should be noted that although the critical thickness, He, has
now been shifted, as compared with the case of flat plates, in the
direction of smaller values [compare with (33)], calculation of the cri-
teflon of stability on the basis of the simplified equation (22'), instead
of by the exact relations (24) and (25), leads to a practically identical
result. It is just with the aid of (37t), (38) and (38') that we find,
proceeding from (22'),
= 2, (40")

x3A "" 1300, (41 r)

"fn kT
so that in this case also the error has no particular significance.

7. Criterion of stability of strongly charged sols

The kinetic process of the coagulation of sols when adhesion
takes place as a result of the preliminary approach of particles in the
process of B r o w n i a n movement, differs from the adhesion of two
particles under static conditions. As has already been shown ~s, the
redprocal, L, of the rate of coagulation ~, considered as a measure
of the stability of a sol, is proportional to the expression
ets)
1 f -~- 2rdH (42)
L ~ - --~ .'..- (2r +/'/)'~ ~
o

where U(/-/) is the ener~o-y of interaction of particles for a distance H

between their surfaces at the point of least separation. From (42) it is
seen that the interaction only begins to play a positive r61e in the
stability of a sol when, at certain values of H, U(H) becomes posi-

N. F u c h s, Z. Phystk, 89, 736 (1934).

 Selected Works - 2 51
tire, L e., the maximum of the function U(H) lies above the axis
of abscissae--in other words, when an energy barrier exists.
Now the height of the barrier, as well as the force of repulsion N,
is, in accordance with (35) and (35"), proportional to the radius r
of the particles. Hence it follows that the larger the value of r, the
more abrupt, other conditions being equal, is the transition between
the unstable states of a sol and the stable ones which are realized
in the presence of a barrier, i. e., the more abrupt is the boundary
between the former and the latter states. In the limit, for extremely
"large" particles, any diffuseness of this boundary should disappear.
One should distinguish between the criterion of stability correspond-
ing to the beginning of coagulation (aggregation of primary particles),
and that corresponding to coagulation proceeding to the end.
If tile primary particles have radii considerably exceeding the thick-
ness of the ionic atmospheres, the diffuseness of the threshold
of coagulation will not be great. At the same time, with the formation
of large aggregates in the process,of coagulation, it will not decrease,
since the radii of curvature of the approaching sections of the surfaces
of any two aggregates will be equal to the radii of the primary par-
ticles. The situation changes if the primary particles have a radius that
is comparatively small in" comparison with the thickness of /he ionic
atmosphere °-9 The threshold of coagulation (of its initial stage) will be
strongly diffuse in this case; however, as these particles combine into
aggregates, the diffuseness will decrease, for the asperities of the sur-
face o f the aggregates formed will have dimensions that are small in
comparison with the thickness of the diffuse ionic atmospheres; in con-
sequence of this the interaction of the aggregates that collide and the
energy barrier due to this interaction will be such as if the surface
of the aggregates was smooth and had a radius of curvature corres-
p.3nding to the envelope of their microrelief.
The difference between the criteria at the beginning and completion
of coagulation which has been pointed out above will thus be parti-
cularly great here. Moreover, the diffuseness of the latter criterion
will be even less than in the case of a sol with larger primary
particles.

o.j A detailed investigation of the last caoe forms the sttbject of a paper
by L. L a n d a u (in preparation for press).
52 B.V. Derjaguin

Therefore, as a criterion of stability it is p o s s i b l e to assume

a condition under which the energy barrier vanishes, expressed by the
equations:
U--0,
dU
-a--E - - - O.

Tile criterion of stability for symmetrical electrolytes thus obtained

does not differ in form from (31 r) and (41). As regards the numerical
value 0f the criterion, tedious computations are required for its deter-
mination which have not as yet been carded out. However, an exact
calculation of the corresponding constants is hardly of any significance
owin~ to a certain diffuseness of the transition between the stable and
unstable states of the sol. For a rough calculation, however, one may
take advantage of the following method. One of us 1°- has previously
found a criterion of stability for weakly charged sols in the following
form:
A×
D~boo" - - C, (43)

where C has a value of the order of 2 and '~o is the potential of the
particle surface.
We obtain an approximate criterion of the stability of strongly
charged sols by substituting the "apparent" potential of the particle
surface s° for ~o in (43) with the aid of the formula:
4
7 -

[see formulae (22) and (22")].

After simple transformations we obtain:
×3A -...
.,c)~-~~.r.__ 80 0 (44)

30 The term apparent potential is here used to denote the potential which
must be assigned to a plane surface in order that, while decreasing exponen-
tially with the distance from this sildace in accordance not with the exact,
but with the approximate D e b y e - H tl c k e I equation (19), it should asymp-
totically tend to coincide with the potential values [see formula (20)] which
,~ctually exist in an electrolyte at a large distance from the surface. The appar-
' potential of a charged surface is less than the actual one (which may
t extremely large), t h e difference being due to that part of the screening
cloud of dominating ions which i~ not taken into account in employing the
simplified D e b y e - H fl c k e I equation instead of the complete o n e .
 Selected Works - 2 53

and
I __ C" Ale6 z16 (44')
7"I Da (kr); ")
where
C" ~ 2.5 X 10 -2. (44")
As a matter of fact, owing to the approximate character of the calcula-
tion, it gives only a rough estimate of the value of the constant C'.
As to the influence of the valency of the auxiliary ion, it remains
small. In accordance with an approximate estimate of this influence,
which is not claimed to be rigorous, the constant C and the coagulat-
ing power of the electrolyte seem to increase in all cases with an
increase in the valency of the auxiliary ion; this increase, if it exists,
must, however, be slower than that implied by the proportionality to
the quantity (1 --¢-~) (see below) which follows from Wo. O s t w a l d ' s
rule.

8. Discussion of results and their comparison with

experimental data
We have considered three problems connected with the stabilizing
action of ionic atmospheres with respect to three phenomena: adhesion
of plates, adhesion of particles with convex surfaces and coagulati6n
or aggregation of sols and suspensions.
Exact criteria of the stability are easily calculated for the case
of a symmetrical electrobie and in all three cases have the same form:
z3A
-7,kT "-- B
with close values of the constant B in all cases.
Maldng use of the known formula for x, we may write the sta,
bility cliterion in the following form:
1 __(8,-:)a A 2 e G z 6 ~ A2#zO
m --- "vr B~ 03 (kip = c ' ~ (-~--~ ; (45)

tile measure of the coagulating power of the electrolyte m - - - 1'Tit thus

turns out to be directly proportional to the sixth power of the ionic
charge. Owing to the dependence of A upon the nature of the particles,
-{ will not be the same for different sols, even when the values of z
are the same.
54 B. V. Derjaguin

For the case of a non-symmetrical electrolyte the calculations grow

more complicated. The coagulating power of an electrolyte proves to
depend mainly upon the dominating ion. The influence of the auxi-
liary ion manifests itself relatively slightly. The H a r d y - S c h u 1z e
rule thus turns out to be well-grounded. The criterion of stability in"
the presence of a non-symmetrical electrolyte has the following form,
for all three cases:
1 - - - C f ( ~ ) A2e6z~,
m = "fnz D~ (kT)~ ' (46)

where f ( 1 ) - - 1 ; an exact calculation of the function f([~)is an extre-

mely difficult task requiring laborious numerical and graphical computa-
tions, but theoretically presents no difficulties.
Let us compare condition (46) with the empirical criterion of sta-
bility of weakly solvated sols proposed by Wo. O s t w a l d14:

--In fl : const., (46')

where fl denotes the activity coefficient of the dominating ion. Substi-

tuting for lnfl its limiting expression taken from the D e b y e - H t i c k e 1
theory of strong electrolytes, we obtain instead of (46'):

1 e6 zl'; (46")
m'--- ~'nx "---const. (1--+- ~) D3(kT) Y.

It is known that the Wo. O s t w a l d rule has been confirmed

by a vast amount of experimental data obtained both for the coagu-
lation of sols 14 and for the adhesion of quartz particles as. By way of
illustration we give the accompanying Table 1 borrowed from W o.
O s t w a l d ' s 31 paper.
We can see that throughout the extremely wide range o! concen-
trations investigated there is a fair agreement with experiment within
the limits of experimental error and inevitable "diffuseness" of the
coagulation threshold.
Comparison of eqs. (46) and (46") shows that the theoretical cri-
teflon of coagulation which we have derived differs from that found.
empirically by W o. O s t w a 1 d by the factor ---l--r=-+--)2 and the factor

m Wo. Os twald, Ko]l. Z., 88, 1, Tab[e VII (1939).

 Selected Works - 2. 55

Table 1

T y p e of electrolyte I
m = -- "theoretical" 1
ml "I ~ t¢2 T HI m -- ~ experimenhtl*
a c c o r d i n g to (46") T Ht

I I I It I 1 **
122 1 1.5 1.65
.1331 2.0 2.5
1441 2.5 2.8
21 12 48 51
2121 64 47
31 13 486 573
3223 608 938
4 1 1 4 2560 1720
61 16 27216 14050

* T h e v a l u e s g i v e n In t h i s c o l u m n are t h e m e a n of a n u m b e r of e x p e r i m e n t a l
~letermlnations.
* * m Is t a k e n e q u a l t o u n i t y for t h e case a , = , 2 - - 1.

f(~) introduced instead of ( I - ' - 6 ) . The factor I~-/,) ~ is independent

of the nature of the electrolyte, whilst the ratio f(13) at different
(1 -~- 8)
values of [~ should retain an order of magnitude close to 1.
Therefore, the theory developed by the present authors has a twofold
significance.
Firstly, leading to formula (46), it explains the agreement existing
A
between the numerous experiments in which --[y- was constant (co-
agulation of the same sol in different electrolytes) and W o. O s t -
w a l d ' s rule, whereas formerly such agreement frequently met with
incredulous surprise. The main reason for such an attitude is, evidently,
the tendency to include the ~-potential in the expression for the crite-
rion of coagulation. Such a tendency is based on certain experimental
data (the experiments of P o w l s a2 and others) as well as on theo-
retical calculations in the case of weakly charged particles (for example
emulsions of oil in wate012.

s~ R. E l l i s , Z. physik. Chem., 78, 321 (1912); 80, 597 (1912); 84, 145
,(1915). F. P o w l s , Z. physik. Chem., 89, 91 (1915); 89, 179 (1915).
56 B. V. Derjaguin

The calculations given in this work show with all possible clarity
that, for strongly charged sols, the criterion of stability does not con-
tain the potential of the particles and should in the main be determined
by the concentration and nature of the electrolyte. For the case when
the ~-potenfial is found to be close to the "apparent" potential ~0 in
magnitude, the criterion of stability, as follows from (43), may assume
the following form:
Ax
DC -~- C, .

containing the T-potential. However, in this case (in contrast to the

case of weakly charged sots) it is quite unnecessary to introduce the
~-potential, since criterion (46), which contains the concentration alone,
is incomparably more useful, making it possible to precalculate the
be~nning of coagulation without making the frequently intricate and
rather inaccurate measurements of the ~-potential. Moreover, the obser-
ved agreement between ~ and ~0 may rather be an exception to the rule.
The above calculations thus furnish a sound basis for the concep-
(ion of coagulating concentrations and the H a r d y - S c h u I z e rule.
At the same time W o. O s t w a I d's opinion that the conceptions of com-
pression of the ionic atmosphere and of the preferential r61e of the
dispersion medium in coagulation are incompatible 33, proves to be
incorrect.
The calculations carded out above clearly show that coagulation
sets in when, as a result of compression of the diffuse ionic layer,
the range of the repulsive forces of ionic origin is reduced to such
an extent in comparison with the range of the v a n d e r W a a l s
forces of attraction, that the energy or, in the cases of static adhesion,
, the force-barrier disappears and, as an inevitable consequence, rapid
coagulation of the system sets in.
•Secondly, our theory corrects W o. O s t w a I d's rule, giving
strong theoretical grounds for rule (46) modified in comparison with
the former.
Such a modification compels us to look upon the relation between
the criterion of coagulation and the activity coeffident as acddental,
-for, in order that such a relation between the theoretical coagulating

s3 W o. O s t w a I d, Koll. Z., 73, 301 (1935); Z. physik. Chem., 42,

981 (1938).
 Selected Works - 2 57

power and the activity coefficient should exist, it is necessary, as

shown by comparison of formulae (46), (46") and (46'), that f([~)
should be exactly equal to (1 .-~[~), which in reality is not the case.
Further, the existence of the factor ( k T - ) p e r m i t s us to point out
the possible experiments which might be able to settle the question
conclusively in favour of the proposed criterion of coagulation, and
to reject W o. O s t w a 1 d's criterion. Namely, by studying the tempera-
ture effect on the coagulating concentration it might be possible
to reveal a discord with O s t w a I d's rule, provided A will not change
directly as T, which is rather'improbable. However, the methods availa-
ble for measuring the coagulating concentrations are so inexact that
such a solution of the problem would be very unreliable.
At any rate, the advantages of a theoretically substantiated r u l e
over an empirical one are so obvious that this fact alone is already
decisive.
It should be noted that not all the data on the coagulating power
of electrolytes may be described by rule (46).
The dependence of the coagulating concentration upon the concen-
tration of the sol ( B u r t o n ' s rule 34) and upon the nature of the
dominating ion (at equal valencies) does not agree, of course, with
formula (46). As to B u r t o n's rule, it is possible that it lies beyond
the scope of the above theory. This may be due to the difficulties
of a correct determination of t h e coagulating concentration, particularly
at small concentrations, since the use of a photoelectric cell for this
purpose can by no means dispel the doubts concerning the a c t u a l
comparability of the results of measurements at different concentrations
of the sol. It is possible that the influence of the nature of the ion ~
in the case of large organic ions, may be explained by taking account
of the ionic radius. However, this is a problem for the future. Finally,
the rSle of solvation m meaning the effects depending upon the repul-
sive forces of a non-electrostatic origin between surfaces separated
by layfirs of a dispersion m e d i u m - - i s still an open question. Another
problem to be solved is a more exact calculation of Ihe coagulation
criterion for asymmetric electrolytes.

s4 See e.g.H. Freundlich, "Kaplllarchemie", Bd. II, S. 162, Leipzig, 1932.

s5 Wo. O s t w a l d , Koll. Z., 85, 34 (1938).
58 B. V. Derjaguin
9. Conclusions
1. The calculations carried out permit us to determine the interac-
tion between strongly charged plane and convex surfaces B i n parti-
cular, spheres immersed in a solution of an electrolyte.
2. The interaction of similarly charged surfaces always results in
repulsion, in contradiction to the calculations of S. L e v i n e, A. C o r-
k i l l and L. l ~ o s e n h e a d , the error of which had already been
proved previously. Superposition of v a n d e r W a a ! s attraction
upon the repulsion of an electrical nature leads to curves of force or
energy as a function of the distance which have one or two minima
separated in the latter case by a maximum. The existence and height
of the latter determine the stability of the film of the electrolytic
solution separating the surfaces, i n the case of a force diagram, and
the stability of the sol or the suspension, in the case of an energy
diagram.
B. Taking into account the v a n d e r W a a l s forces, one may
•derive the following criterion for the adhesion of these surfaces:
1 Ae6 z 1e
m-- 3'rt 1 -~.C](~) D3(kT)5 ' (1)

where A is the v a n d e r W a a l s constant, e - - t h e charge of an

electron, T - - t h e concentration of the electrolyte in tools per cm. 3, n z - -
the numb~ of dominating ions in one molecule, D - - t h e dielectflc
constant, k - - the B o 1t z m a n n constant, T - - the absolute tempera-
ture, ~ - - t h e ratio of the valency of the auxiliary ion z 2 to that of the
dominating ion z l , C - - a constant, and f ( ~ ) - - a function o f the
"asymmetry of the electrolyte", ] ( 1 ) - - 1 , retaining the same order of
magnitude for other values of {~.
4. A criterion for the stability of strongly charged lyophobic sols
or suspensions determining the transition from rapid to extremely slow
coagulation is obtained in the same form as (1), in which the constants
on the right side have approximately the same numerical values.
5. The criterion of stability obtained differs from the well-known
empirical rule proposed by W o. O s t w a 1 d by the factor ---(~--~-)2.
On the one hand, this substantiates Wo. O s t w a l d ' s rule, explaining
:its good agreement with extensive experimental data, and, on the other
hand, points to the necessity of its modification.
 Selected Works - 2 59

6. A strict basis is given for tile H a r d y -S c h u I z e rule concern-

ing the influence of the valency of the dominating and auxiliary ions
on the coagulating concentration.

Academy of Sciences of the USSR, l?eceived

lnstitu~.e o.f Colloid-and Electrochemistry, .lanuarv 2.'~, 19:11.
Laboratory of Thin Films,
and
Instilute of Physical Problems,
Moscow.