Electrical Double Layer and Electrocapillary

This paper is in a collection of
"Historic Papers in Electrochemistry"
which is part of
Electrochemical Science and Technology Information
Resource (ESTIR)
(http://electrochem.cwru.edu/estir/)
T H E ELECTRICAL DOUBLE LAYER AND T H E THEORY OF
ELECTROCAPILLARITY’
DAVID C. GRAHAME
Department of Chemistry, Amherst College, Amherst, Massachusetts
Origi,nal manuscript received July 24, 1945’
Duplicate manuscript received July 14, 1947
The theory of the electrical double layer is discussed with particular reference to
those aspects of the subject usually included under the heading “electrocapil-
larity”. Modern values for the thermodynamic properties of the double layer
are presented for typical substances, and references are given to that part of the
literature which is likely to be of interest in connection with the problems here
discussed.
CONTENTS
I. Introduction.. ... ............................. 441

11. The ideal polarized electrode.. . . . . . . . . . . . . . . . . . .
111. The electrocapillary curve and a definition of the
IV. The thermodynamic theory of electrocapillarity . . . . . . . . .
V.Verification and application of the thermodynamic theory
VI.The kinetic theory of the diffuse double layer. . . . . . . . . . .
VII.The theory of the compact double layer.. ....................
VIII.Potential differences between dissimilar phases, .............................. 487
IX. Time lag and polarizat ized electrode.. ......... 494
X. Conclusion . . . . . . . . . . . . ........................................... 495
XI. References.. . . . . . . . . . . ........ ....... 500
I. INTRODUCTION
The review3 which follows deals with the theory of electrocapillarity and its
relation to the problems of elucidating the structure of the electrical double layer.
The theory of electrokinetic phenomena is not’discussed.
1 New experimental work reported in this paper has been supported by a grant from the
Penrose Fund of the American Philosophical Society.
* Publication of this paper was delayed by the loss of the original manuscript in transit.
a The symbols used in this paper are listed below. All capacities and charges are per
square centimeter of interface.
C = differential capacity of the electrical double layer
Cd = differential capacity of the diffuse double layer
Co = differential capacity of the region between a metallic surface and the outer
Helmholtz plane
Ci = differential capacity of the region between a metallic surface and the inner
Helmholtz plane
D = dielectric constant
Do = diabattivity of free space (= 45r X permittivity of free space)
E = electrical potential difference a t an ideal polarized electrode as read on the po-
tential-fixing potentiometer
E = electrical potential difference a t an ideal polarized electrode relative to the
potential of the electrocapillary maximum
& = electrical potential difference a t an ideal polarized electrode as read on the
potential-fixing potentiometer when the reference electrode is a normal calo-
44 1
442 DAVID C. GRAHAME
The terms “electrocapillarity” and “electrical double layer” are not com-
monly well understood. A satisfactory definition of the first must await a dis-
cussion of the ideal polarized electrode, another unfamiliar term, but we may
state meanwhile that electrocapillarity is the study of t h surface tension of
metals in contact with inert salt solutions, together with the study of all of the
effects with which this is connected. The term “electrocapillarity” arises from
the fact that most of the early studies in the field were done with the Lippmann
capillary electrometer.
The electrical double layer is the array of charged particles and oriented dipoles
which is thought to exist a t every interface. Very often the term is used in a
narrower sense to mean the array of charges and dipoles between two immiscible
media when one is a salt solution. It is in this narrower sense that the term
is here used.
The concept of the electrical double layer is due to Quincke (80), although that
term was not used by him. The double layer was thought to consist of two layers
of charge, one positive and one negative, situated a t the interface. Today the
me1 electrode and when liquid-junction potentials have been eliminated as
well as possible
E-(E+) = electrical potential difference E when the reference electrode is simply reversible
to the anion (cation) of a salt solution
F = Faraday’s constant
K = integral capacity of the electrical double layer
Kd = integral capacity of the diffuse double layer
K O = integral capacity of the region to which Corefers
K i = integral capacity of the region to which Ci refers
k = Boltzmann’s constant
M = molar concentration of a solution
m = molal concentration of a solution
noiP number of ions of type X, in unit volume of solution measured at a point remote
from the double layer
ni = number of ions of type X, in unit volume of solution measured at a point within
the double layer
q = electronic charge of the electrical double layer
r radius of an adsorbed anion or distance from the interface to the inner Helm-
E
holtz plane
T = absolute temperature
tui = work required to move an ion of type Xi from the interior of a designated posi-
tion within the double layer
X, = an ion or neutral molecule of type designated by the subscript i
z = distances measured from the physical interface
z = absolute value of the valence of an ion
z-(z+) = valence of an anion (cation) including sign
z i = maximum number of ions of type X, which can be adsorbed at 1 sq. cm. of an
interface
= maximum number of ions of type Xi for which thcre is space in 1 cc. of solution
OL = the metallic phase of an ideal polarized electrode

@ = the non-metallic phase of a n ideal polarized electrode
ri = superficial density of a component of type X, as involved in the Gibbs adsorp-
tion equation
THEORY OF ELECTROCAPILLARITY 443
point of view is somewhat more complicated, although the name persists even
in those cases where it is plainly inappropriate.
Briefly the electrical double layer may consist of a layer of electrons (if the
non-electrolytic phase is a metal or electronic conductor), a layer of adsorbed
ions, and a diguse double layer consisting of an ionic atmosphere in which ions
of one sign are in excess of their normal concentrations whereas those of the other
sign are in defect. This atmosphere of abnormal concentrations of ions falls off
rapidly as one recedes from the surface, the half-thickness of the charge density
being seldom over 100 h g s t r d m s and usually much less. Finally, there may
exist at the interface a thin (often monomolecular) layer of neutral molecules
which, whether they are oriented or not, exert an influence on the thermodynamic
properties of the interface. Diagrams of the electrical double layer are given
later (figures 17-19), after the theory upon which they are based has been
presented.
Metal-solution interfaces lend themselves to the exact study of the double layer
better than other types because of the possibility of varying the potential differ-
r+(r-)= superficial density of cations (anions)

r, = superficial density of entropy a t a n interface
-y = mean activity coefficient of the ions of a strong electrolyte
= charge on a single univalent cation
9 = total charge of the solution phase of the electrical double layer (= - q)
T+ = charge of the diffuse double layer
d d
q-(~+= ) contribution of any excess or deficiency of anions (cations) to the charge of the
diffuse double layer
vi = charge of the compact (adsorbed) part of the double layer
q i ( q : ) = charge of the compact part of the double layer when the adsorbed ions are
anions (cations)
I = a measure of the reciprocal thickness of the diffuse double layer defined by equa-
tion 57
p = chemical potential of a neutral molecular species
p&+) = chemical potential of an anion (cation)
pi = chemical potential of a component Xi
v+(Y-) = number of cations (anions) furnished by the dissociation of one molecule of 8
salt
Y = Y+ + Y-
p = charge density (charge per unit volume)
u = interfacial tension
6: = specific adsorption potential of an anion
$ = electrical potential at any point within the double layer relative to the interior
of the solution phase
$' = rational potential difference between dissimilar phases
$* = electrical potential of outer Helmholtz plane relative to the interior of the solu-
tion phase
$i = electrical potential of inn3r Helmholtz plane relative to the interior of the
solution phase
p(@) = cavity potential (i-e., electrostatic potential) of the metallic (non-metallic)
phase
$I = cavity potential of a copper wire attached to the metallic phase
$I1 = cavity potential of a copper wire attached to the reference electrode
nAx = subscript or superscript referring to the electrocapillary maximum
ence between the phases without varying the composition of the solution. This
is done through the use of a reference electrode and a potentiometer which fixes
the potential difference in question. In favorable cases there is a range of poten-
tials for which a current does not flow across the interface in a system of this kind,
the interface being electrically similar to a condenser of large specific capacity,
The capacity of this condenser gives a fairly direct measure of the electronic
charge on the metallic surface, and this, in turn, leads to other information about
the double layer. No such convenient and informative procedure is possible
with other types of interfaces, a fact which explains the preoccupation with such
systems in the present review.
Of the metals which might be used for the purpose outlined, mercury is pre-
gminent. Being a liquid, its surface is readily cleaned, it is free from mechanical
strains, and its interfacial tension is readily measured. In addition, mercury is
distinguished by its lorn chemical activity and high hydrogen overvoltage. The
latter characteristic makes it possible to work a t potentials which would other-
wise be objectionable because of a rapid evolution of hydrogen from the reduc-
tion of the solvent. Kearly everything one desires to know about the electrical
double layer is ascertainable with mercury surfaces if it is ascertainable a t all.
11. T H E IDEAL POLARIZED ELECTRODE
For an understanding of electrocapillary phenomena, an understanding of the

ideal polarized electrode is prerequisite. This type of electrode is observed only
with high-overvoltage metals, and then only in approximation. It is a type very
different from the reversible electrodes with which one deals in thermodynamic
discussions of galvanic cells, yet it is reversible and a t equilibrium.' Its impor-
tance rests upon the fact that it can be treated by exact thermodynamic means.
The rigorous definition of an ideal polarized electrode is as follows (46) : It is
a system composed of a metal in contact with an electrolytic solution for which,
a t equilibrium, the concentration of every charged component is finite in one
phase only. As applied to electrons, this statement is to be taken to mean that
there exists in the solution phase no finite amount of any component which can
take up or give up finite amounts of electricity from the metallic phase.
This definition calls for some amplification. I n the first place it is necessary
to state what are regarded as charged components. The metallic phase is
thought of as composed of two charged components, electrons and metallic ions.
The solution phase is thought of as composed of neutral solvent molecules and
of ions of whatever sort may be present. These ions are the charged components
of the phase. If dissolved hydrogen is present and if its equilibrium with
hydronium ions is to be considered, this is regarded as consisting of protons and
electrons, both of which are charged components. Hydrogen ions are regarded
as hydrated protons. Dissolved oxygen is thought of as oxide ions plus elec-
trons. Hydroxyl ions are similarly thought of as hydrated oxide ions. I n
general, if two substances are present which can be converted into one another
1 The latter point is not always conceded. It is discussed in what follows.
THEORY O F ELECTIZOCAPILLARITY 445
by the addition or removal of charge, these two forms are regarded as forms of
the same basic component, the more reduced form of which is composed of elec-
trons plus the other. The reason for this procedure concerning the choice of
components is that it makes it easier to develop the thermodynamic theory of
electrocapillarity in a rigorous manner.
The definition of an ideal polarized electrode is chosen so that passage of charge
across the interface shall not take place spontaneously a t equilibrium. More
exactly stated, when the potential difference between the phases is altered
slightly, no finite amount of charge must cross the interface of an ideal polarized
electrode during the reestablishment of equilibrium. This is a corollary of the
definition, because if charge did cross the interface, finite amounts of some
charged component would then be present in both phases, contradicting the
requirement of the definition. Experimentally, the ideal polarized electrode is
an electrode which behaves like an electrical condenser without leakage. Its
capacity arises from the fact that charges may approach or recede from the
interface, though they do not cross it.
This definition of an ideal polarized electrode is too strict ever to be fully
realized in the laboratory, but close approximations to it are possible, and, what
is more important, the deviations which are unavoidable are of negligible effect,
as will presently be shown.
An example will serve to illustrate the definition and also to bring out addi-
tional points of importance. Consider a clean mercury electrode in an aqueous
1 M potassium chloride solution (completely deaerated) into which is immersed
a normal calomel electrode. Between the mercury in the calomel electrode and
the clean mercury electrode there is inserted a potentiometer and, optionally, a
galvanometer. The clean mercury electrode is attached to the negative end of
the potentiometer. No continuous current flows through the circuit if the
potential is not made too large. This system may have any number of conditions
of equilibrium, depending upon the setting of the potentiometer. For definite-
ness, consider that - E , the potentiometer reading, is 0.556 volt, in which case q,
the electronic charge on the mercury surface, will be zero,5as we shall show later.
Under these circumstances the following reactions are all conceivable, any one
of which would result in transferring charged particles from one phase to the
other in violation of our definition of an ideal polarized electrode.
Reaction Equilibrium concentration
2Hg + Hg$+ + 2e- cHg:+ = moles per liter
+
K+(aq> e- + K (in Hg) Ng = moles per mole
+
2C1- --+ Clz 2e- pcl, = 10-~* atm.
2H20 + +
2e- --+ H2 20H- pH2 = 4 X atm.
If the electrode is ideally polarized, all of these reactions must produce only
negligible quantities of their products at equilibrium. The calculation of these
quantities is straightforward and gives the results indicated on the right. NK is
6 This does not imply that the potential difference between the phases is zero. a oiirtstion
to be discussed below.
the mole fraction of potassium in mercury, assuming a perfect solution. The

first three concentrations are negligible, as demanded by the definition, but the
pressure of hydrogen required for equilibrium is not. An appreciable amount
of charge would therefore move from the mercury to the solution in the course
of building up this equilibrium concentration.
As a practical matter, the high hydrogen overvoltage of mercury causes the
attainment of complete equilibrium to be very slow compared to the rate of
attainment of ionic equilibrium within the double layer.' In all such cases,
which are extremely common in thermodynamic practice, the slow reaction may
be ignored. Occasionally the secondary effects need to be considered. Thus
the current flow accompanying the process may have electrical effects of an un-
desired nature, or the products of the reaction may build up to significant con-
centrations. But these are practical matters which can be minimized by care-
ful design.
I
At lower values of E I the equilibrium partial pressure of hydrogen goes down,
and values of E can be found for which all of the equilibrium concentrations are
negligible. At higher potentials no complications are observed in practice until
E = -2.0 volts or thereabouts, although the equilibrium partial pressure of
hydrogen is then about lo2' atm. The hydrogen ovewoltage of mercury, of
course, makes this possible.
Other slow reactions may be experienced because of imperfections of the ex-
perimental arrangement. Thus it is inconceivable that the mercurous-ion con-
centration could be kept down to moles per liter, since the solubility of
calomel allows it to rise to about lo-'* moles per liter. More important, par-
tides of colloidal calomel may find their way to the surface and react according
to the equation
Hg2Cl~ + 2e- + 2Hg + 2C1-
thus transferring charge across the interface. This reaction is again of no con-
sequence, since it does not affect the equilibrium of the double layer.
There was a wholly erroneous opinion held a t one time to the effect that the
properties of the double layer a t a mercury interface were fixed by the concen-
tration of the mercurous ion (the so-called potential-determining ion). Since
the latter is present a t equilibrium in amounts so small as to be meaningless
except in a statistical sense, and since in any case the concentration of mercurous
ions will generally exceed the equilibrium value because of experimental imper-
fections, it should be evident that the concentration of this ion is of no impor-
tance provided it is small. The old practice of adding mercurous ions to systems
to be investigated was unnecessary and escaped being harmful only because the
diffusion of such ions to the interface was slow compared to the rate of establish-
ment of equilibrium within the double layer. Those mercurous ions which
reached the interface were promptly removed by electrolysis.
We may now define electrocapillarity as the study of ideal polarized electrodes
6 Equilibrium is attained in a matter of microseconds or milliseconds, depending upon
the geometrical distribution of the electrodes (see Section IX).
THEORY O F ELECTROCAPILLARITY 447
and of electrodes which behave like ideal polarized electrodes. The term has
sometimes been used to include the field of electrokinetics, but this practice is
not recommended by the writer.
The potential of an ideal polarized electrode is fixed by an external source of
E.M.F., or if there be no such, then by the charge which happens to be present
a t the interface. In the latter case the potential of a stationary electrode will
rapidly drift to that of a calomel electrode (assuming the solution is a chloride),
owing to the action of colloidal calomel or other reducible matter.
If an isolated reservoir of mercury is allowed to flow in a fine stream through
a deaerated solution of an inert electrolyte, it will very rapidly come to a poten-
tial of zero electronic charge (19, 53, 73). This is so because each droplet of
mercury which falls carries with it a large part’ of any residual charge which
might have been present on the isolated reservoir initially. If negative ions are
adsorbed on the mercury after electronic equilibrium is attained, positive ions
are carried along in equivalent amounts, leaving the surface uncharged as far as
electrons are concerned. Even slow-dropping mercury (3 sec. per drop) comes
to a state of zero charge if oxygen and other reducible substances are very rigor-
ously excluded (42, 61, 62). The potential of zero charge is also the potential
of the electrocapillary maximum, as is well known and as we shall subsequently
prove.
It was thought a t one time that the potential of the isolated reservoir in the
experiment described above must be identical with that of the solution, it being
assumed that zero charge corresponded to zero potential difference. Such an
assumption is easily proved false by comparing different solutions (84),but it is
apparent anyway that zero charge does not necessarily imply zero potential
difference. For example, a layer of oriented dipoles will produce a difference of
potential on its two sides, although its net charge is zero. Such layers of oriented
dipoles are very common,8and it is not a t all certain that even water molecules
are not oriented a t a mercury surface, producing a potential difference which
cannot be measured.
111. T H E ELECTROCAPILLARY CURVE AM) A DEFINITION O F T H E

POTENTIAL SCALES EMPLOYED
It has been known for a long time that the interfacial tension of mercury in
an ideal polarized electrode varies with the potential difference imposed across
the interface. Numerical data are given by many authors (22, 25, 54, 58, 59,
67, 68,70,93), the work of Gouy being the most extensive and of high accuracy
(36, 38). Curves representing this variation are called electrocapillary curves,
and are often almost parabolic in shape (63) (figure 1). Nevertheless one makes
7 The capacity of the droplet relative to the solution is much greater than the self-capac-
ity of the reservoir; hence virtually all of any excess charge accumulates on the surface of
the mercury droplet.
8 An adsorbed anion, together with a non-adsorbed cation, may be regarded as a dipole.
It is in this sense that layers of oriented dipoles are most common, although layers of ori-
ented polar molecules are also common enough.
448 DAVID C. GRAHAhlE
a serious error in assuming them to be exactly so, for the deviations from para-
bolic form are thermodynamically related to the interesting properties of the
double layer. Because these deviations are small, other methods of obtaining
the same information are more commonly employed (27,42,78,79).
The right-hand end of the curves in figure 1 and in all of the curves to follow
corresponds to the largest negative potential imposed upon the mercury from
the external source. A point corresponding t o a potential more negative than
the electrocapillary maximum is said to lie on the cathodic branch of the curve in
question and to represent negative polarization. Potentials more positive than
the electrocapillary maximum correspond to the anodic branch and represent
420 -
J320-
5
4
5 280-
260 -
QB 0.6 0.4 0.2 0 - 0.2 , I I
-0.4 -Q6
1 ,
-0.8
, ,
- 10 - L2
, , , , I
-1.4
RAT 10N A L POTENT I A L, V: I N V 0 LT S
FIG.1. Interfacial tension of mercury in contact with aqueous solutions of the salts
named. T = 18°C. Abscissas are measured relative to a “rational” scale in which the
potential difference between the mercury and a capillary-inactive electrolyte is arbitrarily
set equal t o zero at the electrocapillary maximum.
positive polarization. (As used by other authors, positive po!arization often

means that E , the observed potential, is positive.)
Until the thermodynamic theory of electrocapillarity has been discussed, there
is not much to be learned from the electrocapillary curves in figure 1. There are
two features of these curves worthy of remark a t this point, however. The
shape and position of the curves do not depend significantly upon the cation
employed, except that hydronium ion and very large cations, such as tetramethyl-
ammonium ion, give somewhat different results. The differences observed with
hydronium ion are probably attributable to the effects of that ion on the anions
present.
Regardless of the anion present, the curves become nearly coincident on the
right-hand side corresponding to negative polarization. This is understandable,
since the negative charge of the mercury repels anions under these circumstances,
making close contact of mercury and anion rare.
Them are severnl kinds of electrical potential difference with which one has
to deal in discussing the electrical double layer. The “abso1ute’’ or “true”
potential difference betn-ecn dissimilar phases is an undefined concept, and no
attempt will be made to use it in this paper. The question is further discussed
in Section VIII, however.
The potential of a point within a macroscopic cavity in an isotropic conductor
is a definite quantity which can in principle be mensured.’ It will be called the
cavity potential and is equal to what is sometimes called the electrostatic poten-
tial of the medium (1)’the potential of a point “just outside” the surface. The
only reason for introducing a new terra is that the term “electrostatic potential”
I II:
REFERENCE
ELECTRODE
I I
FIG.2. Schematic representation of the system to which the equations refer
has been so often misused that it has largely lost its meaning. The term “cavity
potential”, on the other hand, can hardly be misunderstood.
The cavity potential diflerence between metals in contact is the Volta or con-
tact potential. It is discussed in Section VIII.
Consider the system illustrated in figure 2. I and I1 are wires of the same
metallic substance, say copper. a is the metallic phase of an ideal polarized
electrode in contact with the electrolytic solution p . The reference electrode
may be supposed to contain the electrolyte 0, so that liquid-junction potentials
will be absent. Designating by $I, $I1, $“, and $’ the cavity potentials of the
several phases, it is evident that
E + (IC.” - $3 + (\la - IC.”) + (IC.“ - $’I = 0 (1)
Actuall>+,one can only measure differences of potential, but this is a n unimportant dis-
tinction, sime differences are all one ever really needs t o know.
450 DAVID C. ORAHAME
The potentiometer reading, E , is regarded as $ I - #I1. It will most often be

negative in electrocapillary measurements.
I n the discussions to follow, cy, the metallic phase, will not be altered in com-
position (a more general case is discussed by Koenig (60)), so that $” - # I is a
constant and we may write:
d($’ - $‘) = -dE - d($“ - 4) (2)
This equation is used below in the derivation of an important thermodynamic
result.
I n the discussion of the kinetic theory of the double layer which is to be pre-
sented it is necessary to depart from the strict thermodynamic definition of poten-
tials. Even so, it is not necessary to talk about “true” potential differences of
dissimilar phases, except that liquid-junction potentials of aqueous solutions
will be regarded as defined. The exact sense in which this term is used is stated
in Section VIII. It is to be noted that liquid-junction potentials do not enter
into the thermodynamic treatment, however.
One of the most convenient points of reference in discussing the potential
difference between mercury and solutions is the potential of the electrocapillary
maximum, the point of zero electronic charge. This point of reference may be
determined simply and accurately by means of the mercury stream technique
described in Section 11. Potentials measured relative to this potential will be
represented by E, and negative values of E will correspond to negative polariza-
tion. Because so many authors have spoken of the potential defined in this way
as if it mere a t least an approximation to the “true” potential difference between
the phases of an ideal polarized electrode, it is probably well to repeat that no
such idea is here entertained.
Very often a normal calomel electrode is employed in measuring electro-
capillary curves and the like. If the solution in question is not potassium chlo-
ride of the same concentration, this has the disadvantage of introducing liquid-
junction potentials. One can eliminate the uncertainty by reporting the results
in terms of E, the potential relative to the electrocapillary maximum. Very
often one wishes to know also, however, how the potential of the electrocapillary
maximum varies with concentration or from electrolyte to electrolyte. At pres-
ent this information does not serve any exact thermodynamic purpose, so that
the introduction of small liquid-junction potentials in this case cannot be criti-
cized, particularly since it is unavoidable. The directly observed potential
(called the electron potential in Section VIII) of the metallic phase of an ideal
polarized electrode relative to the mercury in a normal calomel electrode connected
with the solution phase of the polarized electrode will be called G. The connec-
tion of the calomel electrode to the solution is supposed to be made in a manner
which eliminates liquid-junction potentials. I n practice this means that a cor-
rection will be made as well as possible for whatever liquid-junction potentials
are present.
The value of E a t the electrocapillary maximum of a number of electrolytes
is given in column 3 of table 1. The departure from a constant value is attrib-
uted to an adsorption of anions by the mercury. From independent evidence

(equation 21, vide infra), one knows that adsorption is small or negligible for
alkali carbonates, sulfates, and hydroxides. There is also good reason to believe
that alkali fluorides are not adsorbed, although the necessary data for proving
TABLE 1
Properties of the electrical double layer a t the potential o j the electrocapillary m a x i m u m
(36,&, 90)
,.MAX
ELECIBOLYTE CONCENlPATION G""" SALT
Y oollr ooltr cc coul.rcm.'

NaF . . . . . . . . . . . . . . . 1.o -0.472 0.008
0.1 -0.472 0.008
0.01 -0.480 0.000
0.001 -0.482 -0.002
NaCl . . . . . . . . . . . . 1.0 -0.556 -0.076 3.6

0.3 -0.524 -0.044 1.8
0.1 -0.505 -0.025 1.1
KBr . . . . . . . . . . . . . 1.o -0.65 -0.17 10.6

0.1 -0.58 -0.10 3.5
0.01 -0.54 -0.06 0.6
K I ..................... 1.o -0.82 -0.34 15.2

0.1 -0.72 -0.24 9.1
0.01 -0.66 -0.18 4.8
0.001 -0.59 -0.11 1.3
NaCNS . . . . . . . . . . . . . . . 1.o -0.72 -0.24 14 .O

0.1 -0.62 -0.14 6.3
0.01 -0.59 -0.11 1.3
KiCOs. ............... 0.5 -0.48 0.00 -2.2
NaOH . . . . . . . . . . . . . . . . 1.o -0.48 0.00 Small
NanSOd.. . . . . . . . . . . . . . . . 0.5 -0.48 0.00 Small
KNOs . . . . . . . . . . . . . . . . . . 1.o -0.56 -0.08 5.5

0.1 -0.52 -0.04 2.0
0.01 -0.52 (?) 0)
this conclusively are not yet available. It is found that values of -&"^" all lie
close together in the neighborhood of 0.47-0.48 volt for these salts.
The 'Lbest"value for &"^" in the absence of adsorption of anions and cations
is close to -0.480 volt.'' If it were not for possible water orientation at the
interface and unknown electrical effects a t the mercury surface itself (such as an
10 Heretofore a value of -0.52 or -0.50 volt has been commonly used.
“overshoot” of electrons, for instance), one would possibly be justified in thinking

of this as a “true” potential difference between the phases of a normal calomel
electrode. What it is, rather, is the “rational” potential difference between these
phases,-rational because it accounts for all the factors which are susceptible
to measurement and ignores those which are neither susceptible to measurement
nor significant in the interpretation of results. The rational potential difference,
$‘, is defined by the equation:
3.‘ = & + 0.480 volt (3)

It is not a thermodynamic quantity because liquid-junction potentials enter into
its measurement.
There is at least one circumstance in which the rational potential must be used
with care. In non-aqueous systems, the orientation of solvent molecules affects
the observed potentials (23). All that can be observed is the difference between
the effects of water orientation and of another solvent’s orientation. Fnimkin’s
results (23) indicate that such solvents as methanol, ethanol, and acetone are
oriented with their positive (hydrocarbon) ends toward the mercury, a t least
more so than water. This result hardly helps one even to guess, however, how
water molecules are oriented a t a mercury surface, if indeed they are. Use of
the rational potential as defined by equation 3 with non-aqueous systems
amounts to assuming that any shifts in the value of & a t the electrocapillary
maximum in the absence of ion adsorption are to be attributed to orientation
of the non-aqueous solvent molecules at the interface.
IV. THE THERMODYKAMIC THEORY OF ELECTROCAPILLARITY (23,25, 33,

41,46,53,60,68, 77,86,89)
The thermodynamic theory of electrocapillarity may be derived from the Gibbs
adsorption equation (Gibbs’ equation 508, reference 33) , according to which
du = -I’,dT - ZI’j dpj (4)
In this equation u is the interfacial tension of two immiscible phases, p i is the
chemical potential (partial molal free energy a t constant temperature and pres-
sure) of a component Xi, I‘i is the excess of that component, in moles per unit
area of interface, over that which would be present in the system if each phase
were of uniform composition, with the composition of the interior of the phase,
up to a plane drawn parallel to but not necessarily coincident with the physical
interface. The actual position of this plane is arbitrary except for the dictates
of convenience. Curved surlaces may be considered essentially plane until the
thickness of the region of inhomogeneity approaches the radius of curvature of
the interface (51). The summation in equation 4 extends over all the inde-
pendent components of the system, which were always thought of as neutral
substances.” r8is a quantity expressing the excess of entropy in the system
11 This statement applies explicitly only to the derivation of Gibbs’ equation 508 (our
equation 4).
THEORY O F ELECTBOCAPILLARITY 453
in a manner exactly analogous to the definition of ri. A simple but rigorous

derivation of equation 4 is given by N. E(.Adam (2).
The adaptation of equation 4 to the problems of electrocapillarity has been
subject to much dispute. The difficulty has been partly associated with a
failure to understand the properties of the ideal polarized electrode, which is
continually confused with non-polarizable electrodes which form the more usual
subject of thermodynamic discussions of electrodes. A further difficulty has
arisen because of the fact that polarized electrodes are seldom actually a t equi-
librium, although this is purely a practical matter and influences the experi-
mental results almost not a t all. Finally one must inquire closely into the
validity of using equation 4 with charged components serving as independent
components.
The early treatments of this problem are characterized by what the present
writer regards as an unnecessarily obscure point of view. A typical exposition
is that of Frumkin (23). The electrons are omitted from the summation and
the “potential-determining” ion, Me“, the ionic component of the metallic
phase, is regarded as a component of the solution, although its equilibrium con-
centration is vanishingly small. The potential difference between the phases is
then related to the concentration of this ion, and the electronic charge is regarded
as represented by an excess or deficiency of this ion a t the interface. It is not
surprising that this theory has not been generally endorsed, although a very care-
ful analysis shows that it is basically correct and leads to the correct results.
A n apparently rigorous treatment is given by Gouy (41), who does not employ
the Gibbs adsorption theorem but proceeds from a thermodynamic analysis of
the whole system, including the reference electrode and the external source of
potential. The difficulty of following his analysis has apparently impeded its
general acceptance, although it appears that if it were couched in modern termi-
nology, it would be no more difficult than those treatments which employ the
Gibbs equation.
A relatively simple approach is that of Koenig (60), who achieved the correct
results by means of two rather improbable assumptions. He assumed, first,
that equilibrium does not subsist a t the interface and secondly, that there exists
a t the interface a barrier impermeable to charged particles. In spite of these
assumptions, Koenig has given a complete and in other respects very satisfactory
derivation of the equations of electrocapillarity.
Grahame and Whitney (46) have shown how it is possible to achieve Koenig’s
results without making his assumptions. Charged particles fail to cross the
interface, not because of the presence of any barrier but because of a lack of any
tendency to do so. At equilibrium, each charged particle is content to stay on
its own side of the fence. The exceptions are those charged particles which react
so slowly that their presence leaves the double layer virtually undisturbed.
Because of the differences in the initial assumptions, there is a slight difference
in the interpretation of the symbols employed by Koenig and by Grahame and
Whitney, although the equations are identical in form. In the treatment by
the latter, the symbols have the same meanings as already defined for the Gibbs
equation.
The bzsic equation, applicable to any ideal polarized electrode at constant

temperature and pressure and derived from equation 4 (46 or GO), is
da q d(@ - $a) - midpi (5)
or by equation 2
da = - 4 d E - qd(#" - 4) - m idps (6)
where q is the surface charge density of electricityI2 on the metallic phase.
Equation G is restricted through the use of equation 2 to systems in which t,he
composition of the metallic phase remains unaltered.
The componefits of the system over which the summations in equations 5
and G extend are the chargsd components and the neutral molecules which do not
dissociate appreciably. In principle the summation includes the components of
both phases, but since we are considering that the composition of the metallic
phase is to remain unaltered, the chemical potentials of the components of that
phase, including the electrons, are unchanged and dpi = 0 for each such com-
ponent. Thus in effect the summation includes only the components of the non-
metallic phase.
The definition of q used in the derivation of equation 5 is
B
q = - FC r,zi
where the summation is limited to the ions of the non-metallic phase. q is
therefore equal to the total charge of the ions of the double layer with signs re-
versed. F is the Faraday. z, is the valence of Xi, including sign.
It might appear a t first sight that the value of q would depend upon the posi-
tion of the surface of reference relative to which the values of r, are reckoned.
This is easily proved to be not so. Let the surface of reference be placed a t A
and let N: and N! be the numbers of equivalents of cations and of anions, re-
spectively, which would be present in the system if the phases were uniform with
the compositions of their interiors right up to the surface A. Let N+ and N-
be the numbers of equivalents of cations and of anions which actually are present.
Then from the definition of q
q = -F(N+ - N): + F(N- - N)!
Because the interior of the phase is electrically neutral
N: = NA
and therefore
q = -F(N+ - N-)
This last equation shows that q does not depend upon the position of the sur-
face of reference. If the extent of the interface is extended by 1 sq. cm., keeping
12 By this we mean to include only electronic charge. Adsorbed ions are explicitly ex-
cluded from the definition of q.
the intensive variables constant, the net work required is IS and the total charge
which flows through the external circuit connecting the phases is q. This makes
it plain why IS and q do not depend upon the position of the surface of reference.
On the other hand, the amount of one of the components which must be added
to maintain the composition constant is not so unambiguously specified, since the
concentration may be increased by removing appropriate amounts of the othzr
components or by adding the component in question. Specifying that the sur-
face of reference is to be fixed by setting r for some one component equal to zero
amounts to specifying that the composition is to be adjusted, after expanding
the surface, by adding or removing the other components only.
An interesting question arises concerning the interpretation of p i , the chernical
potential, when the component in question is charged. This is the familiar
problem of individual ionic activity coefficients, and it turns out, as always, that
the particular combinations which arise in any actual experiment are determi-
nate. As an example we may refer ahead to equation 17, where the chemical
potential of the salt finally replaces the chemical potentials of the individual
ionic types.
The first and simplest application of equation 6 is to an ideal polarized elec-
trode a t constant composition (designated by the subscript p ) . Each term in
the summation is zero, and so also is d($'I - @). Therefore
This is the well-known Lippmann equation (67, 68). It states that the slope of
the electrocapillary curve is equal to the electric charge density of the metallic
surface. Experiment,al results verifying this and the equations to follow will be
given in the next section.
It follows from equation 7 that at the potential of the electrocapillary maxi-
mum p = 0, as stated previously. It may be noted that the reference electrode
employed in the application of equation 7 need not be one which can be joined
to the solution without liquid junctions, since on the assumption merely that the
liquid-junction potential does not change during an experiment carried out a t
constant composition,
d E = d E = dlC/' = dG
and equation 7 is valid regardless of the kind of potential considered. The rate
of change of the slope of the electrocapillary curve is, from equation 7,
(d2a/dE2), = -(dp/dE), = C
where C is the diferential capacity of the double layer. The distinction between
the differential capacity, -dq/dE, and all other kinds of capacity is of particular
importance because of the fact that p is not proportional to E. For the sake of
clarity we may introduce a new type of capacity, K , designated as the static or
integral capacity, and defined as
K =: -q/E (9)
456 DAVID C. GEAHAME
This is an appropriate definition, since q = 0 when E = 0, but unlike ordinary

condensers, the ratio of q to E in the double layer is not constant. In figure 3,
sho-Yving q as a function of E for some typical cases, the differential capacity is the
slope of the curve at a given point, but the integral capacity is the slope of the
chord drawn from the origin to the point. Except in special cases, the differen-
2 oc
l ? t
-
I‘
~ t
.I.
IYAzLVj
# <
r-
P , , , , , * * , , ,
0.6 0.4 0.2 -0.2 -0.4 -0.6 -0.8 -1.0 . 2

0
P O T E N T I A L RELATIVE T O E.C.MAX. (VOLTS)
FIG.3. Electronic charge on mercury surface in contact with uninormal aqueous solu-
tions of .the salts named. T = 25°C. Curves would coincide at the right if “rational”
. . ..
potential scale had been chosen.
tiel capacity is the more significant quantity, but the integral capacity is very
useful as an intermediate function in the making of calculations and as a means
of representing data.
If one measures the capacity of a condenser or of the electrical double layer
by means of an impedance bridge, it is always the differential capacity which is
obtained. Unlike ordinary condensers, the capacity of the double layer varies
with the D.C. potential imposed across it. A number of plots of the differential
capacity as a function of the voltage are given in the next section.
The charge q can be obtained from differential capacity curves by graphical

integration, since from equation 8
E
q = - l CdE
The limits are as shown because q = 0 when E = 0. By a further integration

(of q with respect to E) one can obtain the shape of the electrocapillary curve
(equation 7), but the constant of integration cannot in this instance be ascer-
tained other than by direct measurement of the interfacial tension a t some one
point.
The integral capacity is easily calculated from the differential capacity through
equations 10 and 9. The differential capacity can be obtained from the integral
capacity through the following equation (obtained from equations 8 and 9) :
C = K +E(%)
It will be noted that a t the electrocapillary maximum C = K.
We turn now to a consideration of the effects of varying the composition of the
electrolyte. For these cases it becomes necessary to specify the nature of the
reference electrode more closely. Liquid-junction potentials are to be avoided,
which requires that the reference electrode must change as the electrolyte
changes. There are two possible types of electrodes which may then be consi-
dered, those which are simply reversible to one of the anions and those which are
simply reversible to one of the cations. The observed electromotive force E,
as read on the potentiometer, will be designated by or ET in the two types,
respectively. If the reference electrode is simply reversible to one of the anions,
for example, equation 6 will read:
du = - q dET - Zr,dpi - ad($" - $') (12)
From this point on it mill be supposed in this section that only one anionic
type and one cationic type are present. The surface of reference with respect to
which the r,'s are reckoned will be so chosen that rsOLYENT = 0. This latter
choice is discussed further below. The presence of only a single anionic and a
single cationic type makes the subscript j superfluous. Equation 12 now reads
(if the reference electrode is reversible to the anion)
d u = -4 dE- - F+dp+ - I'-dp- - 4 d(+" - 4) (13)
+
where the and - subscripts refer to cations and anions. From the definition
of chemical potentials it follows that
dp = v-dp- +- v+dp+ (14)
where p is the chemical potential of the neutral salt and Y+ and v- are the num-
bers of cations and anions, respectively, formed by the dissociation of one mole-
cule of salt.
The potential I'$ - $' varies with the chemical potential of the anion, if the
reference electrode is simply reversible to the anion, according to the equation

zJi'd(+" - @) = dp- (15)
From the definition of q (or from the electroneutrality requirement)
q + +
z+r+F 2 2 ' - F = 0 (16)
When equations 14, 15, and 16 are introduced into equation 13, and noting
that z+/z- = -v-/v+ , the following simple and important result is obtained:
If the reference electrode is reversible to the cation, one obtains by an exactly

analogous procedure :
da -qdE+
r- dp
--
V-
(18)
These equations and also those which follow may be put in forms more suitable
for computation by the introduction of the relation
dp = vRT d In(mr(v~vl-)"'} = vRT dlnm*y = vRT d l n m y (19)
The nomenclature is essentially that of Lewis and Randall. y is the mean ac-
tivity coefficient, m is the molality, and mAis the mean molality, defined by the
relation mi = m(v;+vL-)"'. v is the sum of v+ and v- .
At constant composition equations 17 and 18 reduce to the Lippmann equa-
tion. At the electrocapillary maximum, where q = 0, one has
d#As/dp = - ryx/V+= - r:As/v- (20)
The symbol d#^"/dp means that the differentiation occurs at q = 0.
The quantities on the right are equal to rtiz, the number of moles of neutral
salt adsorbed per unit area of interface a t the electrocapillary maximum. There-
fore
d#^"/dp = - rtET (21)
Values of IZ
' E; computed from this equation are listed in table 1. Unfor-
tunately the accuracy with which Pax can be ascertained is not as great as one
would wish for this purpose. The equation itself, however, suffers from no
known restrictions other than those appropriate to the Gibbs adsorption equa-
tion itself.
At constant E- or E+ equations 16 and 17 give
- v+(du,/dp)B- = r+and - v-(du/dp),+ = I'- (22)
These results indicate that r+or r- can be obtained a t any point on the electro-
capillary curve by measuring the rate of change of interfacial tension with com-
position a t constant imposed potential, E+ or E-. (The potentiometer reading
is the thing to be held constant, not any ill-defined quantity.) It has not been
very generally realized that re and r- could be calculated separately without
the introduction of non-thermodynamic assumptions. Some results obtained
with these equations will be presented in the next section.
Very convenient alternate forms of equation 22 are obtained by dividing

through by equation 7 (with E- or E' for E)
- v+q(dE-/dp). = r+ (23)
and
- v-q(dE+/dp). = r- (24)
In obtaining these results the identity (du/dp) E(dp/dE),(dE/du), = - 1 has
been used. Differentiation of equation 22 gives:
- v+(a2a/apaE-) = ar+/aE- and - v-(a2a/apaE+) = ar-/aE+ (25)
Since the order of differentiation is immaterial, the left-hand members may be
rewritten, with the aid of the Lippmann equation (7), as
v+(dq/dp) I- = (dI'+/dE-), and v-(dq/dp) E+ = (dI'-/dE+), (26)''
Graphical integration of the left-hand members of these equations with respect
to E- or E+ gives I'+as a function of E' except for the constant of integration.
Since q can be obtained from capacity data (equation lo), I'+ and r- can be ob-
tained from capacity data, except for the constant of integration. The latter
can be evaluated through the use of equation 20 or 21, but this requires accurate
interfacial tension data which may not be available. Alternative methods of
fair accuracy but not of strictly thermodynamic character will be discussed later.
The evaluation of I?+ and I'- for electrocapillary systems of many kinds is an
important task so far barely begun. The chief difficulty has been the lack of
accurate experimental data and unclear ideas concerning the meaning of the
thermodynamic equations.
We may now consider the reasons for the assumption made earlier that
rBOLVENT = 0. Basically, the justification is that of simplicity. The equations
take their simplest form when this assumption is made. But nothing is lost
thereby, since it is only a stoichiometric calculation to find the amount by which
any I' changes for a shift in the position of the surface of reference through a
given distance. Moreover, this is all the information which can be obtained by
thermodynamic means, since if one chooses to set = 0, the only convenient
alternative, one obtains equations having exactly the same physical ~ignificance.'~
As a matter of convenience, I?+ and I'- will frequently be expressed in units of
charge per unit area in spite of the fact that their definitions are in terms of
moles per unit area. It is believed that this inconsistencywill cause no confusion.
13 These equations could have been obtained by inspection from equations 17 and 18 by
noting t h a t du is an exact differential. Equations 26 are the reciprocity relations satisfied
by the line integral of an exact differential.
"More explicitly, one obtains the equations given by Koenig (60). An example will
Berve to illustrate the truth of the assertion that these equations are equivalent to those
here derived. Instead of our equation 21, Koenig gives
V. VERIFICATION AND APPLICATION O F THE THERMODYNAMIC

THEORY O F ELECTROCAPILLARITY
The two principal kinds of equations developed in the thermodynamic theory
of electrocapillarity are those which deal with variations of electrical potential
and those which deal with changes of composition. The first class has been
extensively verified, the second not a t all. There is no serious doubt about the
correctness of either class, however.
A three-way check of the first class of equations is shown in figure 4. For con-
I
M/2 NA~SO,
c
8
4
-I
a
20 @FROM DIFFERENTIAL CAPACITY 0
0
e FROM ELECTROCAPILLARY CURVES
w
I- 0DIRECT MEASUREMENT
z
IO I I ! I I ! I ! I I I I
0.8 0.6 0.4 0.2 0 -0.2

-0.4 -0.6 -0.8 -10 -1.2 -1.4
POTENTIAL RELATIVE TO E.C. M A X . (VOLTS)
FIG.4 . Three-my checkof the Lippmann equation. Crossed circles computed from data
by Gouy (36).
venience the integral capacity is plotted. The three types of points represent
(a) an evaluation of K from differential capacity data using equation 10, ( b )
an evaluation of K from interfacial tension data using equation 7, and ( c ) an
evaluation of q from direct measurements of the charge carried by falling mer-
cury droplets of known surface area. In the case cited, all three types of data
were known with exceptional accuracy, There are no arbitrary constants in
these data. Other verifications of the first class of equations are given by Frum-
kin (231, by Proskurnin and Frumkin (78), by Borissova and Proskurnin (9),
by Craxford (14), and by Grahame (42). Proskurnin and Frumkin pointed
where rio is the mole fraction of solvent and NB is the mole fraction of solute within the
solution. But this equation can be derived from equation 21, for if rsoLVENT = 1/ a t some
arbitrarily chosen interface and zero at the interface t o which equation 21 refers, then
will change by exactly ~ f l y / when
~ i ~ the surface of reference is moved from one posi-
tion to the other. When this is added t o equation 21, Koenig’s equation results. It is to
be noted that dub<Ax/dphas a fixed value independent of the position of the surface of
reference.
out in 1935 (78) that contamination of the metallic surface by traces of capillary-
active substances had rendered earlier measurements of the differential capacity
so inaccurate that the demands of equation 8 had seemed not to be satisfied.
This difficulty is overcome nowadays either by extraordinary care in the puri-
fication of solutions and metallic surfaces (9) or by the use of a dropping elec-
trode technique (42, 43).
Differential capacity data for representative types of the solutions which have
been investigated are given in figures 5-6.15 Other data of the same kind are
given by Gouy (36), by the Russian group of workers (9, 27, 78, 87,88), and by
Grahame (42, 43).
The characteristics of the differential capacity curves in figures 5-6 depend
very greatly upon the anion and almost not a t all upon the cation, except the
hydronium ion, provided it is of fairly low molecular weight. This was to be
expected, of course, from the fact that the electrocapillary curves shorn such
behavior. An explanation of many of the characteristics of these curves will be
evident from the mathematical development of the kinetic theory to follow,
although not all of the details are fully understood. For instance, there is still
some difficulty in explaining the difference between the behavior of hydroxides
and fluorides under negative polarization, since both of these anions have large
hydration energies (7) and would be expected to be as much alike as two unival-
ent cations. The “humps” which appear near the electrocapillary maximum of
most curves are not correctly predicted by any theory and are believed by the
present author to arise from the mutual electrostatic repulsion of ions in the
double layer in directions parallel to the interface.
The second type of thermodynamic equation of electrocapillarity deals with
the effects of varying the composition. I n figure 7 there are presented data by
Gouy on the interfacial tension a t the electrocapillary maximum of several elec-
trolytes (36). Many more data are contained in the original paper. According
to equation 21, the slope of these curves measures the amount of electrolyte ad-
sorbed a t the interface. Values calculated in this way are included in table 1.
There has been no experimental verification of these results by independent means,
although such verification is theoretically possible. There is no real doubt,
however, concerning the validity of the equation from which they are derived.
16 These data by the author disagree somewhat with similar data by Vorsina and Frumkin
(87,88). The disagreement is not serious except at small negative values of E with dilute
solutions. I n plots of q u s . E one observes two roughly linear regions connected by a non-
linear region near E = 0 (6; 10, p. 74; 13, 76). Since the slope of these curves gives the
differential capacity, the impression has grown up that there are two “characteristic”
values of capacity corresponding to positive and negative polarization, respectively. This
impression is not borne out by the differential capacity curves themselves, which do not
approach a constant value over any considerable range of potentials with positive polariza-
tion. The appearance of the q us. E curves is somewhat misleading, since very considerable
changes of slope may go unnoticed or pass as experimental error. Even with negative
polarization the differential capacity does not approach a constant value but has a flat
minimum.
; : :of' I
Values have been plotted against #LAX in figure 8. A few discordant
values have been omitted (0.01 M sodium thiocyanate; 0.1 M sodium chloride;
I " " ' " " " ' " I
0 0'8 ' 0'4 ' o '-0'4 '-0'8 ' -1'2 ' - 1 ' 6 ' -
POTENTIAL RELATIVE TO ELECTROCAPILLARY MAXIMUM
FIG.5 . Differential capacity of the electrical double layer between mercury and aqueous
solutions of the salts named. T = 25°C.
0.1 M potassium nitrate). Except for the omitted points, the curves at a given
concentration are consistent enough to serve as a starting point for estimating
I'riGfrom #LAX. The latter is so much easier t o measure that in many instances
the values taken from the curve are likely to be more accurate than the measured
values. Thus for 0.01 M sodium thiocyanate and 0.1 M sodium chloride the
a
W
a
Ilo 0 8 04 0 -40 4 -08 -12 -16 L
POTENTIAL RELATIVE TO ELECTROCAPILLARY MAXIMUM
FIG.6. Differential capacity of the electrical double layer between mercury and aqueous
solutions of t,he salts named. T = 25°C.
observed values are probably in error. On the other hand, all of the observed
values for nitrates are a little higher than the curve would suggest, a result which
is correlated with the dissimilarity of the cathodic branches of the differential
capacity curves of halides and nitrates.
J
440- I I
P
K$430-
z
&
420-
2
0
cn
z
W 410-
I-
-I
-B
2V 400- e 0.1
I1c 0 0.1
W
t;
L
- 390- I
-5 -4 -3 -2 -I 0
LOG,O my
FIG.7. Maximum interfacial tension of mercury in contact with aqueous solutions of
the salts named. T = 18°C. Data by Gouy ( 3 6 ) . The slope measures the amount of salt
adsorbed (equations 19 and 21). (Additional points not shown were used in the construc-
tion of the figure.)
I I I I I I I 1 I I I I I I I 1L,.,'
1 ' i
t
I4 -
12 -
CI
510-
ii- J
3 -
s8
FIG.8. Salt adsorption on a mercury surface at the potential of the electrocapillary

maximum. Lines connect points of same concentration. Ordinates represent shift of the
potential of the electrocapillary maximum caused by the salt. Th'e dotted line shows
calculated values for unimolar solutions based upon the erroneous assumption that $ 0 = $i.
464
Curves of I?+ vs. E- can be obtained in two ways, through equation 22" or
through equation 26. I n the latter case only the shape of the curve is ascer-
tained, the constant of integration being unknown. If the curves are made to
coincide a t one point, the agreement at other points constitutes a partial check
on the validity of equations 22 and 26. Such a check is shown in figure 9,
FIG.9. Components of charge of the double layer between mercury and aqueous 0.3M
sodium chloride solution. T = 25°C. Open circles computed from data by Gouy (36).
is total charge. r +is charge attributable t o cations in the double layer; 7 5 is charge
attributable to anions repelled from the diffuse double layer; 7: is charge attributable t o
anions adsorbed on mercury surface. r- is the sum of these two. Potentials measured
relative to 0.3 M sodium chloride-calomel electrode.
where I?+ (expressed in p coul./cm?) has been calculated for 0.3 ill aqueous so-
dium chloride by means of each of the two equations mentioned. The agree-
ment obtained has more t o do with the accuracy of the experiments than with the
accuracy of the equations, however, since equation 26 is derived from equation
1 6 Other values of r +and r- obtained through the use of equation 22 are given by Iofa,
Frumltin, UstinskiI, and Eiman (54, 5 5 ) .

22. Again it would be possible to verify these results by direct analytical

means," but such verification has not been attempted."l
I n figure 9 there are also plotted l?- and - q (labeled 7 in conformity with the
nomenclature of the next section) obtained from equations 26 and 10, respec-
tively, and also two other quantities, 7: and &, to be defined later. These are
components of I?- and are obtained by non-thermodynamic means as explained
below.
Figure 10 gives similar data for another concentration of sodium chloride.
These data are not of the highest possible accuracy, and will probably need some
revision when recently discovered techniques (43, 79) have been more widely
applied.
I'+ shows the interesting property of being positive even on positive polariza-
tion. On positive polarization, the mercury surface holds anions not only by
coulombic forces but also by covalent bonding. The resulting binding is known
as a specific adsorption because it is specific to the various anions. As a result
of specific adsorption, more anions are held on the mercury surface than corre-
sponds to the positive charge on the mercury. Consequently the net charge of
the two layers is negative and cations are attracted. The specific adsorption
observed a t the electrocapillary maximum is only a special case of this phe-
nomenon.
We turn now to the application of the above results to problems which can-
not be settled by thermodynamic means alone. The assumptions we are about
t o make are in better agreement with the experimental data than any others of
which we know. Part of the evidence has been given by Whitney and Grahame
(90) and is repeated in Section VI in somewhat different form.
Ions which are held to the mercury surface by covalent bonds are certainly
not separated from the surface of the metal by water molecules. Ions which are
held to the mercury surface by van der Waals forces are also assumed not to be
separated from the surface by solvent molecules. Ions which are not held to
the mercury surface by covalent bonds or by van der Waals forces will be as-
sumed to be separated from that surface by a solvent sheath, at least when the
solvent is water. This amounts to assuming that the work needed to displace
the solvent on the side of the ion nearest the interface is not negligible compared
to the work needed to move the ion up to the nearest point of approach not in-
17 One could allow a stream of mercury to flow through a column of solution while main-
taining the potential a t a fixed value and measuring the current flow and the time. An
analysis of the top and bottom iicompartments'l would give the amount of neutral salt
transferred. Knowing the integral capacity as a function of E, one could compute q and
therefore the area of the mercury surface. From the total charge transferred and the
known transference numbers, one could then compute the amounts of each ion moved from
one compartment to the other by transference. These data would be sufficient to deter-
.
mine r +and r- The difficulty, of course, lies in the difficulty of avoiding convection cur-
rents during the experiment. The experiment is interesting primarily in that it shows that
r+ and r- are measurable magnitudes apart from considerations of interfacial tension.
18 Such measurements as have been made have been done with non-polarized electrodes
(66, 74, 75, 83).

volving such solvent displacement. The locus of the electrical centers of a layer
of adsorbed ions, whether adsorbed by covalent bonds or van der Waals forces or
both, will be called the inner Helmholtz plane. The locus of the electrical centers
of hydrated or solvated ions in contact with the mercury surface will be called
the outer Helmholtz plane. It is assumed that the outer Helmholtz plane is the
Same for all ions, even for anions, although the latter are generally more easily
FIG.10. Components of charge of the double layer between mercury and aqueous 1 M
sodium chloride solutions. T = 25OC. See legend of figure 9. Potentials measured rela-
tive t o 1.0 M sodium chloride-calomel electrode.
dehydrated. It should be noted that we do not speak of a layer of ions a t the

outer Helmholtz plane, since the forces which act there are long-range forces and
serve only t o produce an ionic atmosphere. Except in the case of very large
adsorbed ions, it is to be expected that the outer Helmholtz plane will lie farther
from the surface than the inner.
The great similarity of the electrocapillary properties of low-molecular-
weight cations leads one to believe that they do not populate the inner Helm-
holtz plane, where their various radii would lead to varying results. If this
conclusion be accepted (and additional evidence for it will be presented a t the

end of Section VI), it follows that r+is determined solely by the condition of the
double layer from the outer Helmholtz plane out into the solution. Therefore
if oqe compares two points having the same value of I?+ on a curve such as that
in figure 9, one expects that the potential of the outer Helmholtz plane will be
the same in the two cases and that a "map" of the double layer showing concen-
trations and potentials would be the same for the two cases from this point out.
Moreover, even on comparing different solutions of the same valence type and
of the same concentration, if the cation is the same or behaves the same, the dif-
ference of anion should make little difference, and a "map" of the double layer
for a condition corresponding to the same value of r+should be the same for
both. This principle will be subject to experimental verification when more
data of sufficient accuracy became available. Meanwhile it may be used as a
guide in the absence of such information.
The differential capacity of the region between the mercury surface and the
inner Helmholtz plane will be designated as Ci and defined by the equation
C i = dq/d($' - $') (27)
$t is the potential of the inner Helmholtz plane. Its meaning is discussed in
Section VII. The integral capacity of this region will be designated by K' and
defined by the equation
K' = Q/($' - gi) (28)
From these two equations it follows that""
+'-$A
Ki = -
Q
1
- !$I Ci d(gr 4) - (29)
This equation is used below for the evaluation of K'.

The potential at which r+is a minimum, the rY;"-potential, is one of special
interest." At this point a small change in the imposed over-all potential causes
no change in the value of I?+ and therefore no change in the potential of the outer
Helmholtz plane, +'. The potential of the inner Helmholtz plane, $', must also
remain unchanged, since it controls $'. Therefore the differential capacity of
the whole double layer is equal to the differential capacity of the region between
the mercury surface and the inner Helmholtz plane, Ci. This capacity is nearly
the same a t the rT"-potentials of each of the sodium chloride concentrations
studied. It lies between 32 and 34 pflcm.' and its constancy leads one to sup-
pose that it is constant under a wide range of conditions. If it is constant, it fol-
lows from equation 29 that it is equal to the integral capacity of the same region.
Thus
C' = K' (30)
18. The limits of integration are set by noting froin elementary electrostatics that when
q = o , y = Ip.
19 Note that FYIN is the minimum value of r+,whereas rYaxis the value of J?+ at
the electrocapillary maximum. The subscript or superscript MAX always refers to the
electrocapillary maximum in this paper.
THEORY OF ELECTROCAPILLAKITY 469
This equation is important in that it gives a means of calculating K' and there-
fore # i , the potential of the inner Helmholtz plane, from experimentally observ-
able results. The results will be accurate near the I';l"-potential and probably
also elsewhere.
Equation 28 is one of the Stern equations (85), of which we shall speak later.
There is one important difference, however, in that il.i is not here regarded as iden-
tical with \Lo. Stern recognized the desirability of distinguishing between these
quantities, but underestimated the necessity for doing so. At the electrocapil-
lary maximum of 0.3 M sodium chloride, for instance, +i= -0.014 volt, whereas
$' = -0.022 volt according to a later estimate which is probably not greatly
in error.
It is evident from figures 9 and 10 that the adsorptive forces are becoming
stronger as the mercury becomes more highly charged positively.2o It has not
always been recognized that this is not simply the result of an increased coulom-
bic attraction in a simple sense but represents a strengthening of the covalent
bond by which the anions are held to the mercury surface. This result is to be
expected on any theory of chemical binding. The positive charge enhances the
stability of the electron-pair which binds the atoms. This same phenonxnon is
apparent in the chemical behavior of simple incrganic salts. When two such
salts are compared, differing only in the charge on the cation, the more highly
charged cation binds the anions much more strongly. Comparing ferrous and
ferric chlorides, for instance, the latter dissociates much less than one would
anticipate from interionic attraction theory. Ferrous and ferric hydroxides
display the same behavior, the greater insolubility of the latter corresponding
to far greater bond strength. Salts of tin, lead, mercury, manganese, and cobalt
exhibit this behavior in marked degree. Salts of copper appear to constitute
an exception, for which sc)me reason can perhaps be found. The nitrates, per-
chlorates, and bisulfatcs of the metals do not show the phenomenon, no doubt
because these ions are not bound to metallic ions by covalent bonds to any
marked extent. It is interesting to note that nitrate, perchlorate, and bisulfate
ions show little or no specific adsorption on mercury of the type associated with
covalent-bond formation until the mercury is strongly positively charged. The
fluoride ion, which behaves differently from all of these ions, is probably unique
because its strong electronegativity hinders the formation of covalent bonds with
cations to a large degree.
It would be desirable to be able to set down rigid rules for the experinental
recognition of the presence or absence of adsorption of ions. This we cannot do
in general, since the concept is not a purely thermodynamic one. Certain non-
thermodynamic considerations are possible, however. We restrict our attention
first to the electrocapillary maximum.
One is tcnipted to say that if I'z/z
= 0 a t the electrocapillary maximum,
ion adsorption is absent, a d this is nearly true. But it is foiind that in very
*o The increasing values of r +show this, since r +would not even be positive if adsorptive
forces were absent. The calculations leading t o figure 15 (see Section 1'11) give a quantita-
tive measure of the effect.
470 DAVID C. GRAFiAME
concentrated solutions of salts whose ions are believed to be unadsorbed, I’rfi

becomes negative (table 1 or figure 7). This effect may be thought of as being
caused by a small constant adsorption of solvent which results in the displace-
ment of larger amounts of solute in concentrated than in dilute solutions. Other
explanations which could be given are essentially equivalent to this. Because of
this effect, r:fi is negative in the absence of ionic adsorption. Positive values
of ;::’I indicate adsorption, but zero or negative values do not guarantee its
absence.
At potentials different from that of the electrocapillary maximum, the situ-
ation is likewise complicated. If I‘+ is positive on positive polarization, anion
adsorption is indicated. If r- is negative on negative polarization, cation ad-
sorption is indicated. If r+is negative on positive polarization, anion adsorp-
tion is probably absent. If r- is positive on negative polarization, cation ad-
sorption is probably absent.
When more data are available for capillary-inactive electrolytes such as so-
dium fluoride, it will probably be possible to say with more exactitude what
I’+ and must be in any given case if adsorption is to be deemed absent.
The term “specific adsorption’’ is used in cases where it seems probable that
covalent-bond formation is responsible for an observed adsorption. Otherwise
the less explicit term “adsorption” is used. Iodide ion is almost certainly held
to the mercury surface by covalent bonds, an effect which also limits the ioniz-
ability of mercurous and mercuric iodides. There is a considerable correlation
between the specific adsorbability of anions on mercury and the ionizability (or
solubility) of the corresponding mercury salts, but because of the exceptions
and special cases, this correlation cannot be made very convincing without ex-
tensive discussion. The nitrates constitute the most glaring discrepancy,since
nitrate ion shows greater adsorbability than chloride ion. van der Waals (dis-
persion) forces suggest themselves at once in this and similar cases (e.g., per-
chlorates), but the situation is by no means clear.
VI. THE KINETIC THEORY OF THE DIFFUSE DOUBLE
LAYER (8, 11, 16, 20, 21, 39, 40, 72)
The kinetic theory of the diffuse double layer deals with that part of the double
layer which lies a t or beyond the outer Helmholtz plane. At distances closer to
the interface than this, the assumptions break down and other methods of treat-
ment are necessary. These other methods are discussed in the next section.
There are three equations upon which the theory to be developed rests, and
since it is possible to proceed from these without further approximations or as-
sumptions directly to experimentally significant results, it is evident that any
faults in the theory must be faults of the original equations. The three equations
to which we refer are: ( I ) the Poisson equation for a system whose potential
varies in only one direction
dZ#/dx2 = - h p / D D o (31)
(2) the Boltzmann equation
ni = nOie-wi/kT
and (S) an equation expressing the work w i needed to bring an ion of type X,
from the interior of the solution to the point in question:
wi = Z,E$ (33)
I n these equations rl, is the electrical potential (relative to the interior of the
solution) a t a point within the double layer (discussed below) ; p is the electric
density (charge per unit volume) a t the point in question; x is the distance of this
point from the surface of discontinuity, positive values of x corresponding to
points within the solution; D is the dielectric constant of the solution; Do is 457
times the permittivity of free space (discussed below) ; ni is the number of ions
of type X, per unit of volume a t a point whose potential is such that an amount
of work w, is needed to bring an ion of this type from the interior of the solution
t o the point in question; noLis the number of ions of type X I per unit of volume
within the body of the solution; k is the Boltzmann constant; T is the absolute
temperature; z I is the valence number including sign; and z I e is the charge of an
ion X,. Most of these quantities call for further comment.
The potential a t a point within the double layer is not a wholly unambiguous
quantity, particularly when the “point” in question is really a region as large as
the solvated ion. For the present the best solution to the difficulty is to regard
equation 33 as a definition of $. This procedure is not a complete solution to the
problem, however, for it assumes that all of the work toI is electrical work, and
it assumes that the same value of $ will be obtained whatever ion is used for the
test ion. The other kinds of work neglected by this procedure are ( I ) the work
involved in displacing polar solvent molecules from a region of high field strength
(8), ( 2 ) the work involved in pushing all of the other ions in the region a little
closer together (the crowding effect), and (9) the work involved in distorting or
displacing the solvent sheath of a solvated ion as a result of a too close approach
of the ion to the metallic surface. The first-named effect is almost certainly
very small compared to other errors in the theory. The second effect has been
considered in a quantitative manner by Bikerman (8), but it is unfortunately
true that the introduction of factors intended to account for this effect in the fun-
damental equations leads to a differential equation which cannot be integrated
without assuming that the potential is very small, so small that the results have
a very limited range of validity. Within this range, however, the crowding
effect is almost certainly of minor consequence. Even if the mathematical
difficulties could be overcome, it is still not certain how the effects of croxding
should be introduced. One may treat the problem as a volume effect, the ions
having a lesser probability of entering a region already partly filled, or one may
treat the problem as an electrostatic effect, computing the work required to make
a place for the test ion. Both types of calculation are subject to great uncer-
tainties. It will be shown presently that the double layer is not more crowded
than a moderately concentrated solution, so that the error introduced by the
neglect of crowding is probably not so serious as to make the entire treatment
fictitious. Nevertheless, the neglect of cron-ding is likely to prove to be the most
serious defect of the kinetic theories of the diffuse double layer now in vogue.
The third effect listed above, the work required to displace the solvent sheath
of an ion on the side of the metallic phase, is certainly not small enough to justify
its neglect in aqueous solutions.21 The equations to be developed in this section
may not be applied, therefore, to ions whose centers lie closer to the interface
than the outer Helmholtz plane. This is a point which has sometimes been
overlooked or ignored.
The dielectric constant D is dimensionless. We shall use the value 78.49 for
mater at 25°C. (17). The propriety of using the dielectric constant of the
solvent in equation 31 and in the equations to be derived from it is discussed
below.
The constant DO is introduced in order to avoid the confusion which accom-
panies attempts to make the dielectric constant other than a dimensionless
constant. A discussion of this problem is given by Wood (94) and also by Gug-
genheim (50). DOmay be defined by the equation
where f is the force acting between charges q1 and q2 in a vacuum a t a distance T .

I n practical units DOhas the value 109/c2 = 1.112 x lo-'' cou1.-volt-'-cm.-'
A similar quantity, KO, called the permittivity of free space, has been introduced
by Harnwell (52), who gives a clear discussion of the reasons for the introduction
of such a constant. We hare chosen to introduce DO, equal to 4 ~ ~in0 order , to
make our equations more closely resemble those which have appeared in previous
treatments of the kinetic theory of the electrical double layer. Thus DDo in
our treatment is identical with what has been simply called D heretofore. It is
suggested that Do be called the diabattivity of free space, from 6 t a / % ~ d t assable.
The largest value of I?+ ordinarily observed is about 20 p coul./cm.'2p This is
the value found on extreme cathodic polarization of a mercury surface in 0.3 M
aqueous sodium chloride, for instance (figure 9). Under these conditions the
cations will be expected to lie in a very compact layer, essentially a monolayer,
although still formally a diffuse layer. The average spacing of the ions of the
double layer will then be about 9.0 A. from center to center. This calculation
includes an approximate correction for the ions normally present in that region
from the solution and also takes into account the fact that there is a deficiency
of anions amounting to about 3 p coul./crn.2 in the double layer under these
conditions. For cumparison, the average spacing of ions in a 1.0 M solution of a
z-z valent electrolyte is about 10.6 h.,only a little greater than the figure named
above. Thcse considerations offer some justification for the assumption that the
neglect of crowding effects does not invalidate the entire treatment. At the same
time, they indicate that the results to be obtained cannot be accepted without
some reservation.
The assumption is made through the use of equation 31 that the potential is a
function only of 2. This assumption has been questioned (88, p. 262) on the
41 If any one of the ionic types present in the solution a t an appreciable concentration has
a large solvation energy this statement is valid, for the development requires that equa-
tion 33 be valid for every ionic type in the solution.
THEOXY OF EL~CTl~OCdPILLdRITY 473
grounds that the electric charge density a t a given value of x is by n o means con-
stant on a microscopic scale. On the other hand, if the potential be defind by
means of equation 33, as suggested, then since the work t u , has a definite value
for each value of 2 , the value of the corrsspcnding potentia!s will suffer from no
uncertainty on this account.
It has become evident in recent years (e.s;., reference 27) that the values of $
encountered in the diffuse double layer are a good deal srnallcr than had com-
monly been supposed, and that the same is true cf d$/dx, the field strength. It
is the latter consideration which makes it reasonable to use the dielectric constant
of the solvent in the calculations which follo~v,n-hile the f o m e r avoids a diffi-
culty associated with the use of equations 32 and 33. Thus a t a potential of 0.3
volt, once considered easily attainable in unimoiar solutions, eqGations 32 and
33 lead to the result that the concentration of the more abundant ion, assumed
univalent, is about lo5 times greater than in the interior of the solution. This
is manifestly impossible in all but very dilute solutions, and if such potentials
were actually possible in unimolar solutions, then the theory would necessarily
be very greatly in error. It may appear that image forces are neglected in the
following treatment. This is not so, since the effects of the “image” charges in
the metallic surface are taken account of implicitly by giving a non-zero value
to $ a t one boundary (see reference 72). Combination of equations 32 and 33
gives:
ni - nOle-zit#lkT
(34)
The charge density a t any point is the sum of the charge densities of the in-
dividual ionic types, or
p = L: ngie = z nO,z,ee-’iffilkT (35)
Substituting this in equation 31 gives the fundamental differential equation:
This can be integrated once by introducing the identity
which after substitut,ion and integration gives :
The constant of integration has been evaluated by noting that in the interior
of the solution d$/dx = 0 and $ = 0.
For many purposes it is not necessary to integrate this equation further.
According to Gauss’ law (which may here be regarded as an integrated form of
equation 31)
d$/dx = 4.rrvd/DDo (38)
where
According to equation 39 qd is the surface charge density, the total charge in a

column of liquid of unit cross section extending from the plane in question (the
one to which d$/dx refers) on into the body of the solution where $ = 0. Equa-
tion 38 is only valid when the dielectric constant is constant fromx = x tox = 0 0 .
Since this assumption was already made in integrating equation 36, no new
approximations are involved in the use of equation 38. The superscript d in
qd refers to the diguse double layer.
Combining equations 37 and 38 gives qd as a function of $.
td = & p F Z n e i(e--rieS.lkT - 1) (40)

Let
A=
DDokTmi
27
For a z-z valent electrolyte equation 40 can be simplified to:
d
q = - A d e x p (ze$/kT) + exp (- ze$/kT) - 2 (41)
= - A [exp (ze$/2kT) - exp (- ze$/2kT)] (42)
= - 2A sinh za$/2kT (43)
= - 11.72 sinh 19.46 z$ p coul./cm.* (44)
The negative sign is chosen because qd is positive when $ is negative and vice
versa. Equation 44 is evaluated for aqueous solutions a t 25°C. c i is the con-
centration in moles per liter and $ is in volts.
Up to this point in the treatment of the kinetic theory of the electrical double
layer the equations refer to values of $ anywhere within the diffuse double layer.
If one substitutes for $ in equation 44 the potential of the outer Helmholtz plane,
designated by $", one obtains the value of qd for the entire diffuse double layer.
From here on the symbol qd will sometimes have this special meaning. It will
be clear from the context which meaning is intended in any given case.
There are two kinds of capacity of the diguse double layer to be considered,
the differential capacity Cd and the integral capacity K d . These are defined in
terms of the potential of the outer Helmholtz plane, $', as follows:
Kd = - qd/$' and Cd = - dqd/d$O (45)
Kd = sinh ze$"/2kT (46)

P
Cd = ?
!! cosh zc$"/2kT (47)
kT
= 2 2 8 . 5 ~ cosh
6 ~ 19.46 z$' pf/cm.2 a t 25°C. (48)
At low values of $", the hyperbolic eosin.. is almost unity and both Cd and K d
(to which it is then equal) are constant, but this constancy does not extend to
values of Go comparable to those found in practice. Table 2 shows how K d and
Cd compare and also how they vary with qO. Values in parentheses correspond
to values of beyond what can easily be achieved in practice. It will be noted
that the maximum attainable value of $" goes up with decreasing concentration,
while the maximum value of Cd decreases slowly. Even in M solution, how-
ever, this maximum value is very large (- 300 pf/cm.2), which means, according
to equations to be deve!oped later (52 and 53), that the capacity of the diffuse
double layer exerts little influence on the observed over-all capacity at large posi-
tive or negative polarizing potentials. This fact is in some respects fortunate,
since it is exactly under these conditions that the postulates of the theory are
TABLE 2
Calculated differential and integral capacities, Cd and K d , o f the dLffuse double layer for
z-z valent electrolytes in aqueous solution al 55'C.
Capacities are in pf/cm?
CONCENTRA-
CHARGE x POTEWIAL OF OUTER HELMHOLTZ PLANE, 3+ba
TION
0.0 v 0.02 v 0.04 v 0.06v 0.08 v 0.10 v 0.12 v
~-
228 235 252 285 (334 1 (403)
228 246 302 403 (568) (820)
Kd. . .. . . . 72.2 74.3 79.9 90.0 105 128 159

CJ.. . . . . . . 72.2 77.8 95.5 127 180 259 378
0.12 v 0.14 v 0.16 v I 0.18 v 0.20 v 0.22 v 0.24 v
Kd. . . ..., 50.2 82.9

O.O1 \f C d . . . . . . . .
63.9 109 (263)
120 176 259 383 (1233)
15.9 20.2 26.2 34.4 83.2

37.8 55.6 82.2 121 390
most unreliable. Even a very large error in the theory, percentagewise, will lead
to only small errors in the expected properties of the double layer.
K d and Cd are sometimes mistaken for the capacities of the whole double layer,
which is quite erroneous because the potential qO,to which the calculations refer,
is very different from any potential one might think to use for a calculation of
the over-all capacity. Moreover it is not true that d$' = dE, which seems to
be the assumption sometimes made.
The relation between the capacity of the whole double layer, in the absence of
adsorbed ions, and the capacity of the diguse double layer is found as follows:
Let C" and KO be the differential and integral capacities of the region from the
mercury surface to the outer Helmholtz plane. Then from electrostatic con-
siderations alone (or as a definition of K")
- qd = KO($' - $') (49)
476 DSVID C. GRAIIAME
Remembering that KO is not necessarily a constant, one writes also, as a defi-

nition of C":
- dqd = Cod($' - $') (50)
If ionic adsorption is assumed absent,**the charge on the diffuse double layer,
T ~ is, identical (except for sign) with the charge on the metallic surface, q, and
tJr is identical with E. Therefore from equations 8 and 9:
- qd = KI)' and - dqd = C d$r (51)
Combining equations 45, 49, 50, and 51 one has
K = K"Kd/(K" K d ) + (52)
and
c = C°Cd/(CO+ C d ) (53)
Therefore the integral capacity of the Tyhole double layer is equal to the capacity
of two condensers of capacities KO and K d connected in series; an analogous state-
ment applies to the differential capacity.
The reasonableness of this result becomes apparent if one imagines two layers
of charge placed a t the outer Helmholtz plane, one equal to the charge on the
mercury surface, the other of the same magnitude but of opposite sign. Such
a system is readily seen to have a capacity given by equation 52 or 53.
For the theoretical evaluation of K and C, one needs to know K" and C",
17;hich are not given by the theory and represent an as yet unsolved theoretical
problem.
To judge from the appearance of the differential capacity curves (figure 5),
fluorides show less ionic adsorption at ordinary potentials than any other salts.23
This is presumed to be a consequence of the fact that the fluoride ion is hy-
drated (7) and that fluoride ion does not readily form covalent bonds with mer-
cury. For the experimental evaluation of KO and C", therefore, solutions of
fluorides would seem to be particularly suitable. Concentrated solutions are
more suitable than dilute, since K dand Cd are then so large that their effect on
K and C is minimized (equations 52 and 53). As a test of the theory so far de-
veloped the writer has calculated K o and Cofrom differential capacity data ob-
tained with 1 M sodium fluoride, using the relations 52 and 53 together with
equations 46 and 47 which had little effect. These values of K" and Coare shown
in figure 11. It was assumed, then, that KO and C" depended only upon q (or
ad)>,and with the aid of this assumption and the equations named, the differential
22 Adsorbed ions populate the inner Helmholtz plane. Hence the absence of ionic a d -
sorption insures that all of the ionic charge is accounted for in the diffuse double layer.
Since cation adsorption is assumed negligible in any case, the values of r +already found
and plotted in figures 9 and 10 represent the contribution of the cations to 7d. This quantity
will later be designated as 7:. The sum of 7: and (plotted in the figures) then gives td.
23 This fact will be subject t o esperimcntal verification when more accurate capacity data
become available.
THEOflY OF ELECTROCAPILLlRITY 477
capacity C of more dilute solutions of sodium fluoride was calculated. The

results are presented in figures 12, 13, and 14.24
Although the agreement is not perfect, it is better than one would expect in
view of the uncertain character of the theory of the diffuse double layer. Gen-
erally, the observed capacities on the cathodic side are lower than the calculated.
The significance of this result cannot yet be stated with any assurance.
The minimum capacity a t the electrocapillary maximum of dilute solutions
arises from the greater average distance of the ions of the difiuse layer under these
conditions. The maxima a t small cathodic potentialsz5arise from the fact that
301 I I I I I I I I I i I I I I
1 ~ " " " " " " ' l

.a 4 0 -4 -.8 -L2 -1.6 - 2.0
POTENTIAL RELATIVE PO E . C . M A X . (VOLTS)
FIG.11. Integral and differential capacities of the non-diffuse part of the double layer,
calculated from data for 1 M aqueous sodium fluoride. This capacity is in series with the
capacity of the diffuse double layer (table 2).
Co is decreasing while Cd is increasing. The calculations are not extended to

large values of positive polarization, because it was felt that the assumed absence
of ion adsorption would not then be justified. The conclusion which we draw
from these results is that the theory of the diffuse double layer is accurate enough
to be useful in spite of the dubious character of the assumptions upon which it is
based.
14 The values here given are of a preliminary nature, since the experimental data were
not of the best.
2 1 Vorsina and Frumkin (88) do not observe the effect except in the presence of polyvalent
cations (see footnote 15).
34 , I , I I , , I I , , , I ' I I ' I I -
30 - -
N 0.1 NAF
5 -
226 -
?
t
>-22
!z
-
u
a
n
-
5 16-
l4 016 ' 014 ' Oi2 ' 6 ' -6.2 ' -d.4 ' -0:s -Oh ' -IlO '
FIG.12. Calculated and observed differential capacity of mercury in contact with 0.1 J4
sodium fluoride. T = 25°C. Experimental data obtained with 1.0 M sodium fluoride enter
into the calculated values.
1 1 , , 1 , , , 1 1 1 , , , , , I
24-
lo- ' 0.'2 ' 0

'-012 '-014 '-016 '-0:8 '-1:O '-112 ' -1.14
P O T E N T I A L R E L A T I V E TO E . C . M A X CVOLTS)
FIG.13. Calculated and observed differential capacity of mercury in contact with 0.01 M
sodium fluoride. T = 25°C. Experimental data obtained with 1.0 M sodium fluoride enter
into these calculations.
The variation of K" and C" with E (figure 11) is probably due in part to the
effect of crowding a t the interface. Usually it is explained that anions, because
of their greater polarizability, are more compressible, hence move closer-to the
interface, and therefore have a greater intrinsic capacity in a monolayer. -Such
an effect would cause the values of K" and C" to vary as anions replace cations in
the double layer. But this explanation is hardly applicable to the fluoride ion,
whose polarizability is about the same as that of a typical cation. Moreover,
it ignores the fact that the polarizability of the cation plays no r81e. Thus po-
tassium and sodium ions give almost the same values of Coand K O , in spite of
having very different polarizabilities. It is probably true, however, that large
polarizability of anions makes for high specific adsorption, which in turn produces
large over-all capacity C. This effect presumably accounts for the large increase
of capacity always observed on strong positive polarization.
POTENTIAL R E L A T I V E TO E.C.M A X . (VOLTS)

FIG.14. Calculated and observed differential capacity of mercury in contact with 0.01M
sodium fluoride. T = 25°C. Experimental data obtained with 1.0M sodium fluoride enter
into these calculations.
Values of KO and C" depend very strikingly upon the anion present, even in
the absence of specific adsorption. Thus the hydroxide ion, which is hydrated
and not specifically adsorbed under negative polarization, differs greatly in its
electrocapillary properties from the fluoride ion, which it would be expected to
resemble. The investigation of problems such as this remains for the future.
It is now desired to find the manner in which $ varies with x . Combining
equations 43 and 38 gives:
It should be recalled that this equation is restricted, through use of equation 43,
to symmetrical valence types. It can be integrated through the identity
/ csch 8 d8 = In tanh (55)

whereupon
The constant of integration is 2,. It is the va!ue of x for which, if the differen-
tial equation were valid.at all values of x, tanh ze+/4kT would become unity
and )I would become infinite. x - x, is therefore the distance of a point from
this hypothetical plane of infinite potential, a distance designated by x'. Dif-
ferences in x' are physically significant, although x' itself is not.
A new quantity, K , is defined as
= 3.28 X lo7z 2 / & cm.-' a t 25°C.

and equation 56 takes the more compact form
x' = --K1 In tanh
or
9 = f4kT
- tanh-' e-Kz' (59)
%e
Equation 59 gives $ as a function of 2'. For small values of e-"' (large values
of x'),
tanh-b - 0 (0 << 1) (60)
This result shows that a t sufficiently large values of x', 9 is an exponential

~ 2.11 X lO-*/z2/;
function of 2' with a half-thickness of In 2 / or cm. a t 25°C.
For 1-1 electrolytes at unimolar concentration, the half-thickness is therefore
of the order of magnitude of the ionic radius of the non-solvated ion. This small
half-thickness points up the difficulty of defining potentials within the double
layer. At lower ionic concentrations the half-thickness is greater, and the
theory may be presumed to be more reliable.
The maximum value of the potential gradient may be calculated from equa-
tion 38. Numerical substitution shows that in aqueous solution at 25°C.
-= 0.1614 X 106qdvolts/cm.
d9
dx
where qd is expressed in' p coul./cm?
qd seldom exceeds 20 p coul./cm.', a value which corresponds to about 3 X 10'

volts/cm. This is not as large a field strength as previous calculations (of much
greater complexity) had indicated (72). It is not certain whether or not the
dielectric constant of water is greatly different from its usual value a t this field
strength. Malsch (69, 70,71; see also 15 and 17, p. 358) finds the dielectric con-
stant of water to be down 0.7 per cent a t 250,000 volts/cm., the percentage
change being proportional to the square of the field strength. By extrapola-
tion, this would lead one to expect a very low dielectric constant a t 3 X IO6
volts/cm., but such a long extrapolation is obviously not to be given much weight.
The whole argument is a little pointless in any case, since the practice of treating
the electrical double layer as a continuum is presumably unsound for high
values of t d . One carries cut the calculations primarily to obtain orders of
magnitude and to investigate the reasonableness of the mathematical
treatment.
The calculated maximum field strength is almost independent of the concen-
tration and of ionic charge, because the maximum value of vdis found to be almost
independent of these.
It is often desirable to calculate the contributim to the charge of the diffuse
layer made by the cations or anions individually. We define these quantities,
termed q$ and 75, through the equations:
p + and p - are the charge densities of cations and anions, respectively, both of
which are assumed to have valences of absolute value z. The sum of q$ and qd- is
d
v -
m m
+ qd_ = (P+ + P-1 dx = P da: = v

d
(65)
Equation 35 gives for p +

p + = nOizEe-'
where \k = ze#/kT. From equation 63
d = nOize
q+ / (e-' - 1) d s (67)
I n this equation and in those which immediately follow, the limits of integration
are from x = x to x = 00. Taking the square root of the square of the quan-
tity in parentheses gives:
Introducing equations 41 and 38 in succession gives:
This can be integrated to give:

q$ = A(e-*” - 1)
Similarly one has that
$. = -A(e“’ - 1) (73)
The signs in equations 72 and 73 are chosen as they were for equation 41. By
division one obtains the ratio of to qd, a ratio which depends only upon 9.
d d
- (e-”’ - l)/(e*’* - 1)
-
q+/q- = (74)
d d
For low values of $, q+/q- - 1. For large values of I # 1 , q$ or - q!. increased
indefinitely, whereas the other quantity tends to the limit
lim of q i = A (75)
= 5.864; p coul./cm? a t 25°C.
For a 0.3 144 solution of z-z valent electrolyte, this gives as a limit of qd the
value 3.2 p coul./cm.2 The observed limiting value of I’- in figure 9 is identical
with this result, although the agreement is doubtless fortuitous. The largest
observed value of I?- in unimolar sodium chloride is about 5.5 p coul./cm.*,
which may be compared with 5.86 p coul./cm.2 predicted by equation 75. This
agreement lends some support to the kinetic theory of the diffuse double layer,
although it is probable that these results make the theory look better than it
really is.
r- is composed of two parts representing the charge attributable to the ad-
sorption of anions, a!., and the charge attributable to repelled anions qd. Then
z-FI’- = q-i + q-
d
(76)
: will always be negative or zero, and qd will nearly always be positive or zero,
7
Equation 72 can be used to evaluate \k for a given experimental value of T&
(taken to be identical with r+). This value of \k can be used in equation 73 to
evaluate the corresponding value of qd. Since r- is also known experimentally,
one can evaluate qL by equation 76. This procedure has been carried out for
the evaluation of q: represented in figures 9 and 10. There is relatively little
uncertainty in this procedure, because qd is small at potentials where 7: is
appreciably different from zero.
Rice (81) has considered the nature of the diffuse double layer of electrons to
be expected within the metallic phase itself. The treatment is similar to that
here outlined for ions. By using a low value for the dielectric constant of
mercury, Rice obtained values for the capacity of the diffuse double layer in
mercury comparable to those observed for the whole double layer. It is more
usual to regard the dielectric constant of an electronic conductor as sensibly
infinite, however, in which case the calculated capacity of the diffuse double
layer in the metallic phase is also sensibly infinite, and its capscity, being in
series with the rest of the double layer, exerts no effect. This amounts to saying
that in classical electrostatics no appreciable part of the potential drop occurs
within the metallic phase itself. If it were otherwise, in fact, the nature and con-
c:ntraticn of the electrolyte would not be expected to exert so decisive an influ-
ence on the properties of the double layer.
If one assumes (erroneously) that 9" and Iti are identical, it becomes possible
to calculate r:z (or &, which is the same thing at the electrocapillary maxi-
mum) from the theory of the diffuse double layer. The results of the calculation
are in violent disagreement n-ith expcrixent and also with any reasonable con-
ception of the double layer. As an example, consider 0.1 M potassium iodide
at the electrocapillary maximum. Gi, which is identical with J.r at the electro-
capillary maximum, is observed to be -0.24 volt (table 1). If one assumes that
this is also the value of J.", then equation 72 gives for 9: a value of 192 p coul./cm.2
(an impossible result). This is to be compared with the experimental value of
9.1 p coul./cm.' At higher concentrations the disagreement is even worse and
thc calculated value even more impossible (4300 p coul./cm.' at 1 M). These
results illustrate the necessity of distinguishing between Gi and J.", and also
illustrate that J." must be the smaller of the two absolute magnitudes. It is
these considerations which have led us to assume throughout the non-thermo-
dynamic parts of this paper that the diffuse double layer does not extend to the
inner Helmholtz plane and that low-molecular-weight cations, which are not
specifically adsorbed, do not populate that plane.
The dotted line in figure 8 gives the values of calculated in this manner
for unimolar solutions. The disagreement reflects the fundamental nature of
the error implicit in the assumption that $" = 4'.
VII. T H E THEORY OF THE COMPACT DOUBLE LAYER (8, 86, 87, 92)
I n Section VI ionic adsorption has been assumed absent, meaning that no ions
come closer to the interface than the outer Helmholtz plane. This assumption
is probably satisfied to a gocd approximation on negative polarization of mer-
cury in solutions of capillary-inactive electrolytes and even in solutions of capil-
lary-active electrolytes when the active substance is an anion and when the
polarization is great enough to repel these anions from the interface. Under
most other conditions it is to be expected that an adsorption of ions occurs
and that these adsorbed ions lie closer to the interface than unadsorbed ions.
The layer of adsorbed ions is called the compact double layer or, better, the com-
pact part of the double layer. The double layer is now a triple layer, but it is
not generally so called.
Stern (85) worked out a theory of the compact part of the double layer based
484 DAVID C. GRAHARCE
upon the Boltzmann distribution law. In spite of some defects, that theory
still has much to recommend it. The central part of the theory will be here
presented together with an indication of its limitations and the manner in which
they may be removed.
Stern did not distinguish between what n e have called the inner and outer
Helmholtz layers, although he mentioned the possibility that such a distinction
might be necessary. In the discussion which follows, this distinction will
be made from the first.
Following Stern, let ni be the number of adsorbed ions per square centimeter,
noi the number of the same type of ion per cubic centimeter of solution, zi the
maximum number of ions that can be on 1 sq. cm. of surface, and zoi the maxi-
mum number of ions for which there is space in 1 cc. of solution. If the course
of a single ion is followed for some time, it will be found for a fraction of the time
ui a t the interface and for a fraction of time uoiin the solution. If no work were
required to move an ion from the interior of the solution to a place a t the inner
Helmholtz plane, ui/uoi would be the ratio of the number of free places on the
surface to those in the solution.
ui /uOi = (2' - ni)/(zoi - noi) (77)
If one considers not one ion but all the ions of a given type, it is evident that
i
u /uoi = ni/noi (78)
If, finally, the amount of work wi needed to move an ion from the interior of the
solution to a free place a t the interface is not zero, the right-hand term of equa-
tion 77 must be multiplied by the Boltzmann factor so that, with equation 78:
ni/noi = (zi - n')/(zoi - noi)e-Yi'kT (79)
Except in very concentrated solutions, no; is negligible relative to Z O ~ . With
this simplification, equation 79 can be solved for ni to give:
Stern identifies noi/zoi with the mole fraction, which is admittedly only an ap-
proximation a t best, and he also changes the number 1 in the numerator to 2.
This latter change is made in order that ni shall reach the limit zi/2 when wi has
large negative values, it being assumed that nearly equal numbers of cations
and anions will be adsorbed, in which case the monolayer cannot be more than
half filled with ions of a given type. The logic here is not compelling, particu-
larly if one distinguishes between the inner and outer Helmholtz planes, in which
case ions of only a single type need populate the inner Helmholtz plane.
Experimentally it is found that the compact part of the double layer is not
usually near to a complete monolayer. Thus a charge of 30 p coul./cm.* of chlo-
ride ions is about as large an adsorbed charge as one can conveniently deal with
experimentally, and represents about 25 per cent of a complete monolayer.
Under these circumstances the unity in the denominator of equation 80 is small

compared to (ZOi/nOi)ewi'kT, and one may write the equation in the form:
I n equation 82, T is the radius of the non-solvated ion. The quantity zi/zoi
has been set equal to 2r, because the thickness of the region in question is to the
thickness of a centimeter cube as 2r is to 1.
Equation 82 could have been derived directly by a consideration of the ratio
of the probability that an ion would be in the solution to the probability that it
would be in the interface. The numerical coefficient in equation 82 would then
be 1 or 2, depending upon whether the mean free path of the ions is large or small
compared to r. If the mean free path is short, spatial considerations are all that
matter and the coefficient is 2; if the mean free path is long, the fact that ions
can enter from only one side reduces the number of ions a t the interface by
one-half.
The work w i may be divided up formally into a chemical and an electrostatic
work term:
wi = L t ( # i - 4): (83)
I n this equation # i is the electrical potential of the inner Helmholtz plane as
before, and 4: is an adsorption potential of the anions. It is only under special
circumstances that this division has any physical significance, as explained below.
Equation 83 differs slightly in form from Stern's equation because we find it more
convenient to express $1 in units of electrical potential, but the physical meaning
is the same. Stern considered the possibility that both positive and negative
ions might be adsorbed simultaneously, but since this case is experimentally rare
and theoretically very complicated, we prefer to assume that only one type of
ion is adsorbed. For cation adsorption equation 83 mould be replaced by
w i = z+t(+' + 4): (83')
There is no general experimentally unambiguous procedure for dividing up
wi in this manner. The potential of the inner Helmholtz plane, #', may be calcu-
lated from equation 28 if one assumes that K i is a constant. Although K' does
appear to be nearly constant under the conditions for which it can be evaluated
(the ryN-potential), it is only on this assumption that Gi can be evaluated
in general. At the electrocapillary maximum, hoivever, # i is known regardless
of K i (it is equal to #r), and one can say in that case a t least that 4' is defined
unambiguously by equation 83 or 83'. The equations given in this section,
together with equations 43 and 28 and the identity
9 = -q = +$ (84)
constitute the essential parts of the Stern theory.
There are not sufficient equations in the Stern theory as here outlined to per-
mit the calculation of the properties of the double layer without recourse to ex-
perimental data. The most significant use of the theory vhich can be made a t
present is the calculation of 4: from equations 82 and 83 taken together with
experimental data like that shown in figure 9. Values of d.obtained in this
way are plotted in figure 15. The principal result of this calculation is the fact
that 4: varies with q. As the mercury becomes more positive, the chemical
binding energy becomes greater. As already explained, this result is attributable
0.32 1 1
0.28 i:
FIQ.15. Specific absorption potential, +:, for chloride ion on mercury as a function of q,
the electronic charge on the metallic surface. Computed from data presented in figure 9.
to the increased stability of the electron-pair in a covalent bond in the presence

of a positive charge.
I n the original Stern theory it was implied that in the absence of ionic adsorp-
tion, 4: and $5 were zero. This is not quite right in terms of our present con-
cepts, for the work required to move an ion from the int'erior of the solution to a
place on the interface mould not then be zero but would be positive, owing to the
work needed to displace the solvent sheath on the side of the interface. As a
+'
result of this, $1 is positive and T~ would be nearly zero when is zero, a result
which differs from the prediction of the original Stern theory but which agrees
with the physical picture we have here adopted, i.e., the inner Helmholtz layer
should not be populated appreciably in the absence of specific adsorption.
Although equation 82 was derived upon the assumption that the monolayer of
adsorbed ions was far from complete, it is possible to interpret w i in such a
manner that the equation retains its validity a t all ion densities. If wi is
regarded not merely as the energy needed to move the ion into a previously
vacant space but as the energy needed to make a vacancy and to move the ion
into it, it will be evident that ni cannot exceed a monolayer even according to
equation 82. In this sense equation 82 is no less reliable than equation 80.
Equation 82 may be combined with equation 83 and rewritten in terms of
charge to make it more suitable for calculation.
q' = 2z-~rmoiexp { -z-E(+~- &//cT) (85)
- { -38.92z-(+' - c$L)] p coul./cm? a t 25°C. if r = 2 A. (86)
= 3 . 8 6 ~ exp
I n these equations, as in those of preceding sections, z- is the valence number,

including sign, of the anion which is adsorbed. Cation adsorption is assumed
absent.
We conclude this section with figures designed to illustrate the structure of
the electrical double layer.
Figure 16 shows the potentials within the double layer calculated from equa-
tions 28, 43, 59, and 72, together with the experimental data represented in
figure 9. A distance of 3 A. from the interface has been assumed in plotting the
potential of the outer Helmholtz plane, but this number does not enter into the
calculations. It is only an accident that the I'yN-potential coincides with the
potential of the electrocapillary maximum in this case. Perhaps the most sig-
nificant feature of figure 16 is the fact that the potential of the inner Helmholtz
plane reaches a maximum, a result which might have been anticipated from the
fact that r+reaches a minimum.
Figures 17-19 show schematic representations of the electrical double layer
under various conditions of polarization. Large circles represent an excess of
an ion type assumed to be solvated. Dotted circles represent a deficiency of an
ion type. Small circles represent an excess of a non-solvated ion; minus or plus
signs not surrounded by circles represent electrons or a deficiency of electrons,
The manner in which the potential varies with distance is indicated in the box in
each figure. The magnitude of +'is exaggerated in order to bring out the char-
+
acter of the variation of with x in the diffuse double layer.
VIII. POTESTIAL DIFFERENCES BETWEEN DISSIMILAR PHASES (3, 47, 48, 49)
Progress in the study of the electrical double layer calls for a very clear under-
standing of the term "electrical potential difference" as applied to dissimilar
media. I n the discussion here given the author has r3frained from introducing
any new concepts. There is not much room for raal disagraemsnt in this subject
nowadays, except in regard to terminology and conventions. It is hoped that
the treatment here given will make it possible for future authors to discuss
the subject accurately without prolonged explanations.
0.4
0.3 g NACL
0.2
0-l-0.2 OUTER HELMHOLTZ PLANE

5
0
>
z
-J - 0.4
F“z
w
I-
O
a 0.6 - INNER HELMHOLTZ PLANE
1 I 1 I I
I I I
0 2 4 6 8 IO
DISTANCE IN ANGSTROM UNITS
FIQ.16. Potentials in the electrical double layer between mercury and aqueous 0.3 M
sodium chloride solutions a t 25°C. at various polarizing potentials. Values calculated from
data in figure 9. The potentials are “rational” in the sense defined in the text. Note
that the potential of the inner Helmholtz plane reaches a maximum as the polarizing poten-
tial is varied from one extreme to the other.
Guggenheim in 1929 wrote a classic paper (47) in which it was demonstrated

that absolute or “true” potential differences between dissimilar phases are un-
measurable because they are not defined. Moreover, it was pointed out that
THEORY OF ELECTRBCAPILLARITY 489
attempts to define an absolute potential difference are unnecessary, since it is

possible to describe all actual experimental observations in terms of potential
differences of the types already known. In a subsequent paper (48) Guggen-
heim allowed that it was possible to define new types of potential difference which
might prove to be convenient for some purposes, but the important point
remained that there was no necessity for doing so.
SOLUTION
;e-- ;
5
’.d
-OUTER HELMHOLTZ PLANE

INNER HELMHOLTZ PLANE
-J III
4
I-
2
w
I-
x I::
FIG.17. Schematic representation of the electrical double layer at the potential of the
electrocapillary maximum. Small circles represent adsorbed ions. Large circles represent
solvated ions. Dotted circles represent “ghosts”, ions which would be present if the double
layer were not there.
In a third, seemingly little-known paper (49), Guggenheim suggested a defini-

tion of the activity coefficient of individual ionic species. The definition also
serves t o define the potential difference between dissimilar salt solutions in
contact, or between two concentrations of the same salt solution. There was no
contention that the suggested definition gave values of an “absolute” potential
difference except with infinitely dilute solutions, where the quantity was defined
anyway. The only advantage claimed was one of convenience and, all impor-
tant, the advantage of not conflicting in any way with established thermodynamic
usage. The basis of the definition was an assumption that the ratio of the indi-
vidual ionic activity coefficients in a salt solution is given by the Debye-Hiickel
limiting law. This much can be assumed without thermodynamic contradiction.
Its chief virtue is that it makes it possible to talk about individual ionic activities
and liquid-junction potentials without ambiguity. There is also a certain advan-
tage in that it makes it possible to calculate liquid-junction potentials with less
confusion of thought than had been the case formerly. If the idea is firmly
-INNER HELMHOLTZ PLANE
FIG.18. Schematic representation of the electrical double layer with negative polariza-
tion. Note absence of adsorbed ions and increased concentration of positive ions as com-
pared with figure 17. The concentration of [‘ghosts” is also increased.
grasped that the potential differences in question are not “absolute”, there can
be a great gain in clarity of exposition and of thought in Guggenheim’s proposal.
Where liquid-junction potentials have been mentioned or used in the pre-
ceding parts of this review, Guggenheim’s definition has been assumed.
The potential difference between immiscible phases is not defined by Guggen-
heim’s assumption. Again it is convenient to have a working definition, provided
one can keep clear of any suspicion of having proposed a definition of absolute
potential difference. The rational potential difference, $‘, defined in Section 111,
is an example of a useful working definition.
The potential difference between two dissimilar isotropic metals in contact

has been frequently misunderstood. There are two kinds of potential difference
between dissimilar metals in contact which are measurable, the Volta or contact
potential difference and what may be called the electron potential difference.
The former can be measured only with very clean surfaces; it is the work required
to move unit electric charge from the interior of a macroscopic cavity in one
metal to the interior of a macroscopic cavity in the other. The walls of the cavi-
ties are assumed to be clean, in which case the work involved is a definite quantity
which can be measured in a variety of ways (4). It is, for instance, the difference
z
W
I-
X+
FIG.19. Schem tic representation of the electrical double layer with positive polariza-
tion. Note presence of adsorbed anions. Diffuse double layer is identical with that
depicted in figure 18.
in the work functions of the metals, or minus the potential difference which must
be put in series with the wire connecting the two metals in order to permit their
clean surfaces to be moved toward or away from one another without a flow of
electric current through the connecting wire. A constant temperature through-
out the system is assumed.
The Volta potential difference between metals in contact is not affected by
contamination of the metals a t the point of junction, because any number of
electronic conductors can be inserted between two metals without altering the
electronic equilibrium of the metals. Thus as long as the contact is electronic
(not ionic), the Volta potential difference does not depend upon the cleanliness
of the junction. The surfaces at which the measurement is made, however,
must be clean.
The Volta potential difference is similar in magnitude to the E.M.F.established
when the two metals are made the electrodes in a galvanic cell containing solu-
tions of their salts. This is because the electron affinity of the uncharged metal
atoms is a major controlling factor in both. Zinc has a lower electron affinity
than copper; hence zinc in contact with copper loses electrons until it becomes
positively charged, the copper becoming negatively charged. Because of its
lesser electron affinity, zinc lies higher in the electromotive series than copperr2'
and its behavior in galvanic cells therefore depends upon one of the same factors
as does its Volta potential.
It was supposed a t one time that the similarity of Volta potentials and elec-
trode potentials indicated the presence of a film of moisture between metals in
contact. Today such a supposition appears unthinkable on any grounds, not
the least of which is that the sign of the potential difference is wrong.
The electron potential difference between two metals will be defined as the work
required to transfer an electron (not an idealized charge) from one phase to the
other. If the metals concerned are in contact and a t uniform temperature, this
work will evidently be The electron potential difference is what is meas-
ured by all voltmeters, potentiometers, and the like, including electrometers.
If the ends of a uniform bar of metal are maintained a t different temperatures,
the electron potential a t the two ends is not the same, as is evidenced by the exist-
ence of the thermoelectric effect. It is customary to explain this effect as arising
from differences in the contact potentials a t the two junction temperatures, but
this explanation is misleading since it ignores the facts (a) that contact (Le.,
cavity) potential differences must also be presumed to exist between two parts
of a uniform bar of metal a t different temperatures and (b) that what is measured
is not a contact potential but an electron potential difference. Although it is
undoubtedly true that contact potentials do depend upon temperature, the con-
nection between this fact and the thermoelectric effect is remote. It is much
more satisfactory to explain it as the difference in the electron potential differences
generated by dissimilar metals meintained a t the same terminal temperatures.
Since contact (Volta) potentials are a measure of the work required to move a
unit charge from a point in the vicinity of one surface to a point in the vicinity
of another, it is to be expected that all methods of measuring Volta potentials
should be extremely sensitive to changes in the cleanliness or character of the
surfaces in question, as is found to be the case. This fact is sometimes useful
in deciding what kind of potential a given experiment measures. As mentioned
above, however, the contact potential of electronic conductors does not of itself
depend upon the cleanliness of the surface. It is only in the measurement of
contact potentials that surface cleanliness becomes important.
*e The position of a metal in the electromotive series also depends upon the hydration
energy of the ion formed, but this effect is not dominant.
3' This statement assumed no continuous current flow from one phase to the other.
* THEORY OF ELECTROCAPILLARITY 493
The potential of the electrocapillary maximum, -G””“,observed with dilute
amalgams or with metals other than mercury leads to interesting results. Thus
an amalgam containing 0.004 per cent of cadmium gives with normal potassium
chloride a value of 0.625 volt (12) instead of the 0.56 volt observed with pure
mercury. Somewhat similar results are obtained with thallium amalgams in
sodium sulfate solutions (30),and it is noteworthy that metallic gallium gives a
value of about 0.9 volt in potassium chloride solutions (29). These variations
can be accounted for in part by specific adsorption of anions and in part, perhaps,
by reorientation of solvent molecules, but it is also true that the contact potential
which arises when the mercury in the reference electrode is joined to the other
metal composing the polarized electrode shifts the potential of the electrocapillary
maximum by an equal amount (26).
There is a long-standing argument concerning the question of where the
potential difference of a galvanic cell arises. The question can be answered
correctly once it is accurately stated. To the question-Where does the electron
potential difference ark?-there can be but one answer. It arises wholly within
the cell. Since the electron potential of the solution is undefined, one cannot be
more explicit. If one asks instead-Where does the cavity potential difference
(the electrostatic potential difference) arise?-the answer is likewise unam-
biguous. It arises partly a t the electrodes and partly a t the junction between
the metals which compose the electrodes. Since electromotive force is a form
of electron potential difference, no part of the E.M.F. of a cell can be said to arise
from the contact potential of the metallic electrodes.
Cavity potential differences between mercury and potassium chloride solutions
have been measured by Klein and Lange (57), the junction between the two
phases being made by means of a calomel electrode. I n this way it was hoped
to measure a “true” potential difference of the calomel electrode, but it is now
evident that what was measured was only a little different from the potential
of the electrocapillary maximum. The effects of the specific adsorption of anions
on mercury were surely eliminated, but not the effects of different ionic hydration
energies, which build up what are known as surface potentials (vide infra).
Moreover, and even more serious, the method does not eliminate the effects of
water orientation on the mercury, for water vapor was of necessity in contact
with the mercury. Even if all of these effects could have been eliminated, no
“true” potential difference would have been measured, since one knows nothing
about the potential gradients which may exist a t a clean surface of mercury.
This all amounts to saying again that “true” potential differences cannot be
measured because they are not defined.
The “true” potential difference between a metal and a solution has sometimes
been defined as the work required to move an ion of the metal from the interior
of the metal to the interior of the solution. It is not reasonable to call this the
“true” potential difference, for a t the very least one would have to correct for
the hydration energy of the ion in solution and the “solvent” effect of the metal
on the metallic ion. A metal in equilibrium with a solution of one of its salts
reaches a state such that no work is needed to move the ion of the metal from one
phase to the other. But no one would suggest that the true potential difference
in all such cases is zero.
Cavity potential differences have been measured for dissimilar solutions joined
by salt bridges (24, 32). These measurements yield “surface potentials”, a
good approximation of the work required to move a test ion out through one inter-
face and back through another. The second solution is so dilute that ionic
accumuIation at its interface may be ignored, and since the solvent effects on
the test ion presumably cancel, what one has is an approximate measure of the
work done in moving an idealized electric charge through any potential gradients
set up by the presence of ions in the air-solution double layer of the more con-
centrated solution. It is found that the hydration energies of the ions are the
controlling factor. Those ions (generally anions) which are least readily
hydrated seek positions in the surface, making the surface negative to the interior
of the solution. The order of the effectiveness of the anions in this regard is
similar to that of the Hofmeister series, being CNS- > Cloy > I- > NO; >
Br- > C1- > OH- > F- (32).
Much larger surface potentials are observed with surface layers of neutral but
polar molecules (1, p. 133). These layers of molecules also occur a t the elec-
trical double layer between mercury and salt solutions containing small amounts
of sparingly soluble organic substances (31, 34, 35, 37, 38). Although these
layers are most often monolayers, the experimental evidence of Gorodetskaya
and Frumkin (31, 35) indicates that polymolecular layers may form when the
surface layers are compressed. If the potential of the mercury is made strongly
positive or strongly negative, the attraction of the surface for ions exceeds its
attraction for the polar molecules, and the latter are displaced. This displace-
ment phenomenon sometimes occurs over i-i small range of potentials and gives
rise to sudden changes in the charge of the double layer. This leads to large
values of the differential capacity and more or less abrupt changes of slope of the
electrocapillary curves. This behavior is exhibited by solutions of heptyl alcohol
in an inert electrolyte, for instance, as is shown by figures 20 and 21. The capac-
ity is almost uninfluenced by the frequency of the current used in the measure-
ment, but the apparent electrical resistance of the adsorbed layer increases with
decreasing frequency. This phenomenon has been studied and explained by the
author as arising from the fact that the true resistance of the adsorbed layer is
not the directly measured resistance minus the resistance of the solution because
the resistance of the solution is in series with the capacity of the double layer,
whereas the resistance of the adsorbed layer is in parallel with its capacity (43).
IX. TIME LAG AND POLARIZATION RESISTANCE AT AN IDEAL POLARIZED ELECTRODE
It is frequently observed that the establishment of complete equilibrium a t an
interface may be a slow process requiring times of the order of seconds or more.
Thus it is well known that the surface tension of some types of solutions changes
with time after the formation of a fresh surface (5, 18). The E.M.F.of galvanic
cells frequently reaches equilibrium only slowly, and the attainment of stable
electrokinetic potentials may be a matter of days (95). I n the two latter cases
a slow step can be postulated which does not demand that the actual orientation
of ions in the double layer be slow. Thus the slow approach to a state of equi-
librium in a calomel cell is undoubtedly to be attributed to the fact that a solid
substance, mercurous chloride, is involved in the reaction by which equilibrium
is attained:
Hg&12 + 2e- + 2Hg + 2C1-
440-
I
q420-
v,
W
z
>
400 -
0
z
z
0
380 -
W
I-
-t
5360-
Q
ll.
a
W
I-
z340-
320 -
0 0.4 0.8 1.2 1.6 2.0

-E R E L A T I V E TO E - C A L E L E C T R O D E
FIG.20. Interfacial tension of mercury-aqueous sodium sulfate solutions with and with-
out the addition of n-heptyl alcohol. Data by Gouy (38).
This reaction can proceed only when a mercurous chloride molecule is lying next
to the mercury surface with the mercury ion favorably situated to receive the
charge. This reaction does not necessarily take place instantaneously and in the
amounts required by Faraday’s law when a small current is passed through a
calomel cell, since the capacity of the double layer is sufficient to supply a large
charge with only a small change of potential. The above reaction then slowly
restores the system to its equilibrium potential.
The slow attainment of equilibrium in the setting up of certain electrokinetic
potentials is most probably associated with a slow change in the character of the
surface (95). The slow change of surface tension exhibited by certain types of
solutions is not yet satisfactorily explained, but is probably not attributable to

time lag in the formation of the double layer, once the necessary particles arrive
at the surface.
An upper limit may be set to the time required to establish ionic equilibrium
in the double layer a t a mercury surface. According to measurements made
by the author (43), a 1 sq. cm. mercury surface in a solution of 1 M sodium
901 t I I I I I I I I I I I
0.4 0.8 1.2 1.6 2.0 24

-E RELATIVE TO E-CAL ELECTRODE
FIG.21. Differential capacity of double layer between mercury and aqueous sodium
sulfate solution with and without the addition of n-heptyl alcohol. The peaks do not occur
exactly a t the potentials of the breaks in the electrocapillary curve (figure 20),presumaLly
because the temperature (and therefore the solubility of heptyl alcohol) was difierent in
the two experiments.
chloride exhibits an apparent resistance, commonly termed a polarization resist-

ance, not in excess of 0.02 ohm and possibly considerably less. The capacity
of the interface under the conditions of this experiment was about 18 i.tf/cm?
The polarization resistance is in series with the capacity. The time constant of
such a combination is RC or 3.6 X lo-’ sec. Thus the time required for virtually
complete establishment of ionic equilibrium within the double layer is a micro-
second or less. In practice the attainment of equilibrium will be delayed by the
finite resistance of the solution, of course, but that is a separate matter. The
time constant of such delay is again given by RC, where R is now the total
resistance between the electrodes and C is the series capacity of the electrodes.
The polarization resistance exhibited by other types of electrodes is associated
with. the occurrence of chemical reactions a t the electrodes (43, 56, 90, 91).
These reactions cause changes of concentration of the reactants a t the interface,
and because of the diffusion of these reactants or because the reactions are not
rapidly reversible, the back E.M.F. does not remain in quadrature with the im-
posed alternating potential (82). A back E.M.F. not in quadrature with the
imposed E.M.F. manifests itself as an electrical resistance in series (or in parallel)
with the electrode. It has been shown both practically and theoretically (43,44)
that the changes of concentration which occur a t an ideal polarized electrode give
rise to diffusion effects too small to result in measurable polarization resistance.
The electrical double layer therefore behaves very nearly like a pure capacitance.
X. CONCLUSION
A complete discussion of the applications of the theory of the electrical double

layer would involve a discussion of electrokinetic phenomena, of the stability of
colloids, of the conductance and chemical potentials of electrolytic solutions, of
the behavior of clays and carbon black in suspension, of membrane equilibria, of
hydrogen overvoltage, of the theory of corrosion phenomena, and of other phe-
nomena not usually thought of as related to the double layer. Since the author
has not had first-hand experience with most of these subjects, and since in any
case excellent reviews of most of them are already available, they are omitted
from this review. There are a few points upon which additional comments by
the author may not be out of place, however.
At the surface of a non-metallic particle in a salt solution there exists an elec-
trical double layer no different in principle from those which we have discussed
above. It is convenient to regard the neutral surface of the particle as the coun-
terpart of the metallic surface. Adsorbed ions then produce a charge on the
particle, but since this is not an electronic charge, the particle is analogous to
the metallic surface at its elecfrocapillary maximum. Beyond this layer of
adsorbed ions lies the diffuse double layer, whose maximum potential J." is pre-
sumably to be regarded as the {-potential. At the isoelectric point this c-poten-
tial is zero, corresponding to a state in which the concentrations of the ions in tEe
solution are so adjusted that positive and negative ions are equally adsorbed or
unadsorbed (see Verwey (86a)).
Many attempts have been made to elucidate the structure of the double layer
from studies of hydrogen overvoltage phenomena. In the opinion of the author,
these phenomena are complicated by factors not yet well understood, with the
result that reliable information about the double layer is not obtained in this way.
There can be little doubt, however, that the high overvoltage of hydrogen on
mercury is occasioned by the large activation energy associated with the reaction
H30+ + e- +H + He0
since this reaction is highly endoenergetic (65). The fact that mercury is a
poor catalyst for the gas reaction
H+H+Hz
indicates that hydrogen atoms are not strongly bound to mercury, but does not
indicate that this latter reaction is slow when the necessary third body (which
1 1
0.1 g NACL
+
0.002 COCL,
-a
0 DIFFERENTIAL VI
- 0
I
'
C APACl T Y
6
0 DIFFUSION CURRENT -6
-
z
c
z
w
4a:
a:
3
V
io
I I I I I
0.4 0 -0.4 -0.8 -1.2 -1.6

POTENTIAL RELATIVE T O E. C. MAX.
(VOLTS)
Fro.22. Differential capacity observed in the presence of traces of reversibly reducible
substance (Cd++). Open circles show polarographic diflusion current for same solution.
may be mercury) is present, as it is during the reduction of hydronium ions. A

critical discussion of the theories and experimental data pertaining to hydrogen
overvoltage on mercury is given by Frumkin (28).
When the reduction of an ion can occur reversibly, the apparent capacity of
an electrode is enormously increased (42). An example of this effect is given
in figure 22. The capacity is increased in this manner because the reduced
form of the ion acts as a reservoir of charge. The concentration of the reduced
form of the ion is fixed either by the potential or by the rate of diffusion of the ions
to the surface, depending upon whether one is below or above the LLhalf-wave
potential.” When the potential of the metallic electrode is made so strongly
negative that the concentration of the reduced form of the ion is limited by dif-
fusion, changes of potential cause no further change in the concentration of the
reduced form of the ion, and the differential capacity becomes normal. This is a
particularly good illustration of an electrode which is not ideally polarized acting
like one which is.
The enhanced capacity caused by the reversible reduction of ions has been
called by the author (42) a “pseudo-capacity,” since it does not correspond to
any bound charge except in the sense that neutral metallic atoms represent bound
charge. This pseudo-capacity is observed on extreme cathodic polarization of
sodium, potassium, rubidium, ammonium, and tetramethylammonium salts and
presumably would be found with practically all cations. Hydronium ions do not
give this effect, however, which can only mean that hydronium ions are not
reduced reversibly a t a mercury cathode, a result in harmony with the supposi-
tion that the reduction of hydronium ions on mercury is a process requiring
activation energy. This must be regarded as strong support for the slow-
discharge theory of hydrogen overvoltage mentioned above.
The reduction of oxygen on mercury does not give rise to any pseudo-capacity
(42). This observation indicates that the first step in the reduction of oxygen
0 2 + 2H20 + 2e- --$ H202 + 20H-
is a slow step, a fact which was known independently from the fact that it exhibits
overvoltage. The absence of pseudo-capacity from the reduction of oxygen is
fortunate for the making of differential capacity measurements, since the
last traces of oxygen are difficult to remove from aqueous solutions at room
temperature.
The reduction of nitrate ion is irreversible by the criterion of the absence of
pseudo-capacity, a result which will occasion no surprise.
As an analytical tool the measurement of the pseudo-capacity a t a dropping-
mercury electrode might be found to possess the advantage of high sensitivity
in the presence of small amounts of substances which are reducible but not
reversibly reducible. The technique would be a modified form of polarography,
but its greater complication would limit its usefulness to cases where ordinary
polarographic methods fail. As a tool for investigating chemical reactions a t an
electrode, however, it can hardly be excelled.
References mentioned in the text and listed below are not comprehensive but
include only those best suited to illustrate the points under discussion. In nearly
every case the references cited give all of the essential information and provide a
key to the earlier literature. Two other recent and fairly extensive reviews of
electrocapillarity are those of Butler (10) and of Adam (3), to which the reader
is referred for information about topics not here discussed.
XI. REFERENCES
(1) ADAM:Physics and Chemistry of Surfaces, 3rd edition, p. 303. Oxford University
Press, London (1941).
(2) Reference 1, p. 107.
(3) Reference 1, p. 300 et sep.
(4) Reference 1, p. 308.
(5) Reference 1, p. 129 and references there cited.
(6) BARCLAY AND BUTLER:Trans. Faraday SOC.36, 128 (1940).
(7) BERNALAND FOWLER: J. Chem. Phys. 1, 515 (1933).
(8) BIKERMAN: Phil. Mag. 171 33, 384 (1942).
(9) BORISSOVA AND PROSKURNIN: Acta Physicochim. U.R.S.S. 4, 819 (1936).
(10) BUTLER:Electrocapillarzly. Chemical Publishing Co., Inc., New York (1940).
(11) CHAPMAN: Phil. Mag. 161 26, 475 (1913).
(12) CIRVESAND FRUiMKiN: J. Phys. Chem. 34, 74 (1930).
(13) CRAXFORD: Trans. Faraday SOC.36, 85 (1940).
(14) CRAXFORD: Dissertation, Oxford University, 1936; quoted in reference 13.
(15) DEBYE:Polar Molecules, p. 111. Chemical Catalog Co., Inc., New York (1929).
(16) DEBYE AND H ~ ~ C K EPhysik.
L: Z. 24, 185 (1923).
(17) DORSEY:Properties of Ordinary Water-Substance, p. 367. Reinhold Publishing
Corporation, New York (1940).
(18) Du Noiiu: Colloid Symposium Monograph 3, 25 (1925).
(19) ERDEY-GRUZ A N D SZARVAS: 2. physik. Chem. A177, 277 (1936).
(20) EVERSOLE AND BOARDMAN: J. Chem. Phys. 9, 798 (1941).
(21) EVERSOLEAND LAHR:J. Chem. Phys. 9, 530 (1941).
(22) FRUMKIN: Z.physik. Chem. 103, 43 (1923).
(23) FRUMKIN: Z.physik. Chem. 103, 55 (1923).
(24) FRUMKIN: Z. physik. Chem. 109, 34 (1924).
(25) FRUMKIN: Ergeb. exakt. Naturw. 7, 235 (1928).
(26) FRUMKIN: Colloid Symposium Annual 7, 89 (1930).
(27) FRUMKIN: Trans. Faraday SOC.36, 117 (1940).
(28) FRUMHIN: Acta Physicochim. U.R.S.S. 18, 23 (1943).
(29) FRUMKIN AND GORODETZKAYA: 2;. physik. Chem. 136, 215 (1928).
(30) FRUMKIN AND GORODETZKAYA: Z. physik. Chem. 136, 451 (1928).
(31) FRUMKIN, GORODETZKAYA, AND CHUGUNOV: Acta Physicochim. U.R.S.S. 1, 12 (1934).
(32) FRUMKIN, REICHSTEIN,AND KULVARSKAJA: Kolloid-Z.40,9 (1926).
(33) GIBRB:Collected Works, Vol. I, p. 336. Longmans, Green and Company, New York
(1928).
(34) GORODETZKAYA: Acta Physicochim. U.R.S.S. 12, 309 (1940).
(35) GORODETZKAYA AND FRUMKIN: Compt. rend. scad. sci. U.R.S.S. 18, 639 (1938).
(36) GOUY:Ann. chim. phys. [7] 29, 145 (1903).
(37) GOUY:Ann. chim. phys. I81 9, 75 (1906).
(38) GOUY:Ann. chim. phys. [SI 8, 291 (1906).
(39) GOUY:J. phys. radium [4] 9,457 (1910).
(40)GOUY:Compt. rend. 149, 654 (1910).
(41) GOUY:Ann. phys. [9] 7, 129 (1917).
(42) GRAHAME: J. Am. Chem. SOC.63, 1207 (1941).
(43) GRAHAME: J. Am. Chem. SOC.68, 301 (1946).
(44) GRAHAME: Document 2195 of the American Documentation Institute, 1719 N St.,
N.W., Washington, D. C.
(45) GRAHAME: Unpublished results.
(46) GRAHAME AND WHITNEY: J. Am. Chem. SOC.64, 1548 (1942).
(47) GUGGENHEIM: J. Phys. Chem. 33, 842 (1929).
THEORS OF ELECTROCAPILLARITY 501
(48) GUGGENHEIN: J. Phys. Chem. 34, 15-40 (1930).
(49) GUGGENHEIM: J. Phys. Chem. 34, 1758 (1930).
(50) GUGGENHEIM: Trans. Faraday SOC.36, 139 (1940).
(51) GUGGENHEIhl: Trans. Faraday SOC. 36, 397 (1940).
(52) HARNWELL: Principles of Electricity and Electromagnetism, p. 12. McGraw-Hill
Book Company, Inc., New York (1938).
(53) HELMHOLTZ: Wiss. Abhandl. physik. tech. Reichsanstalt I, p. 925 (1879).
(54) IOFA A N D FRUMKIN: Acta Physicochim. U.R.S.S. 10, 473 (1939).
(55) IOFA,~ S T I N S K I I ,AND EIhlAN: J. Phys. Chem. (U.S.S.R.) 13, 934 (1939).
(56) JONES AND CHRISTIAN: J. Am. Chem. SOC.67, 272 (1935).
(57)KLEINAND LANGE:Z. Elektrochem. 43, 570 (1937).
(58) KOENIG:2. physik. Chem. A164, 454 (1931).
(59) KOENIG:Z.physik. Chem. A167, 96 (1931).
(60) KOENIG:J. Phys. Chem. 38, 111,339(1934).
(61) KOLTHOFF A N D LINGANE: Chem. Rev. 24, 1 (1939).
(62) KOLTHOFF A N D LIKGANE: Polarography, p. 106. Interscience Publishers, Inc., New
York (1941).
R KRUMREICH: z. Elektrochem. 19, 617 (1913).
(63) K R ~ G EAND
(64) KSENOFONTOV, PRCEKURKIN, AND GORODETZKAYA: Acta Physicochim. U.R.S.S. 9,
39 (1938).
(65) LATIMER:Ozidation Pofentials,p. 29. Prentice-Hall Inc., Kew York (1938).
(66) LEWIS: Z.physik. Chem. 73, 129 (1910).
(67) LIPPMANN: J. phys. radium 121 2, 116 (1S83).
(68) LIPPMANN: Ann. chim. phys. [5] 6, 494 (1875).
(69) MALSCH:Ann. Physik 84, 841 (1927).
(70) MALSCH:Physik. %. 29, 770 (1928).
(71) MALSCH:Physik. Z.30, 837 (1929).
(72)MULLER:Cold Spring Harbor Symposia Quant. Biol. 1, 1 (1933).
(73) PASCHEN: Ann. Physik 43, 568 (1891).
(74) PATRICK: Z.physik. Chem. 86, 545 (1914).
(75) PATRICK A N D BACHMAN: Z. physik. Chem. 30, 134 (1926).
(76) PHILPOT: Phil. Mag. (7113, 775 (1932).
(77) PLAKCK: Ann. Physik 44, 413 (1891).
(78) PROSKURNIN AND FRUNKIN: Trans. Faraday SOC.31, 110 (1935).
(79) PROSKURNIN AND VORSINA:Compt. rend. acad. sci. U.R.S.S. 24, 915 (1939).
(80) QUINCKE:Pogg. Ann. 113. 513 (1861).
(81) RICE: Phys. Rev. 31, 1051 (1928).
(82) ROSEBRUGH AND LASHMILLER:J. Phys. Chem. 14, 816 (1910).
(83) SCHOFIELD: Phil. Mag. [7]1, 641 (1926).
(84) SMITHAND Moss: Phil. Mag. [SI16,478(1908).
(85)STERN:Z.Elektrochem. 30, 508 (1924).
(86) THOMSON: Application of Dynamics to Physics and Chemistry, p. 191. Macmillan
Company, New York (1888).
(86a) VERWEY:Chem. Rev. 16, 363 (1935).
(87) VORSINAAND FRUMKIN: Compt. rend. acad. sci. U.R.S.S. 24, 918 (1939).
(88)VORSINAAND FRUMKIN: Acta Physicochim. U.R.S.S. 18, 242 (1943).
(89) WARBURG: Ann. Physik. 41, 1 (1890).
(90)WHITNEYAND GRAHAXE:J. Chem. Phys. 9, 827 (1941).
(91) WIEN: Ann. Physik 68, 37 (1896).
(92) WIEN: Ann. Physik [4] 8, 372 (1902).
(93)WILD: Z.physik. Chem. 103, 1 (1923).
(94) WOOD:J. Am. Chem. SOC.68, 432 (1946).
(95) WOOD:J. Am. Chem. SOC.68, 437 (1946).

Electrical Double Layer and Electrocapillary

Uploaded by

Copyright:

Available Formats

Electrical Double Layer and Electrocapillary

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Electrical Double Layer and Electrocapillary

Uploaded by

Copyright:

Available Formats

This paper is in a collection of

"Historic Papers in Electrochemistry"

Electrochemical Science and Technology Information

I. Introduction.. ... ............................. 441

OL = the metallic phase of an ideal polarized electrode

r+(r-)= superficial density of cations (anions)

11. T H E IDEAL POLARIZED ELECTRODE

For an understanding of electrocapillary phenomena, an understanding of the

the mole fraction of potassium in mercury, assuming a perfect solution. The

111. T H E ELECTROCAPILLARY CURVE AM) A DEFINITION O F T H E

positive polarization. (As used by other authors, positive po!arization often

FIG.2. Schematic representation of the system to which the equations refer

The potentiometer reading, E , is regarded as $ I - #I1. It will most often be

uted to an adsorption of anions by the mercury. From independent evidence

Y oollr ooltr cc coul.rcm.'

NaCl . . . . . . . . . . . . 1.0 -0.556 -0.076 3.6

KBr . . . . . . . . . . . . . 1.o -0.65 -0.17 10.6

K I ..................... 1.o -0.82 -0.34 15.2

NaCNS . . . . . . . . . . . . . . . 1.o -0.72 -0.24 14 .O

KiCOs. ............... 0.5 -0.48 0.00 -2.2

NaOH . . . . . . . . . . . . . . . . 1.o -0.48 0.00 Small

NanSOd.. . . . . . . . . . . . . . . . 0.5 -0.48 0.00 Small

KNOs . . . . . . . . . . . . . . . . . . 1.o -0.56 -0.08 5.5

“overshoot” of electrons, for instance), one would possibly be justified in thinking

3.‘ = & + 0.480 volt (3)

IV. THE THERMODYKAMIC THEORY OF ELECTROCAPILLARITY (23,25, 33,

in a manner exactly analogous to the definition of ri. A simple but rigorous

The bzsic equation, applicable to any ideal polarized electrode at constant

This is an appropriate definition, since q = 0 when E = 0, but unlike ordinary

0.6 0.4 0.2 -0.2 -0.4 -0.6 -0.8 -1.0 . 2

The charge q can be obtained from differential capacity curves by graphical

The limits are as shown because q = 0 when E = 0. By a further integration

reference electrode is simply reversible to the anion, according to the equation

If the reference electrode is reversible to the cation, one obtains by an exactly

Very convenient alternate forms of equation 22 are obtained by dividing

V. VERIFICATION AND APPLICATION O F THE THERMODYNAMIC

0.8 0.6 0.4 0.2 0 -0.2

POTENTIAL RELATIVE TO ELECTROCAPILLARY MAXIMUM

Ilo 0 8 04 0 -40 4 -08 -12 -16 L

POTENTIAL RELATIVE TO ELECTROCAPILLARY MAXIMUM

FIG.8. Salt adsorption on a mercury surface at the potential of the electrocapillary

Frumltin, UstinskiI, and Eiman (54, 5 5 ) .

22. Again it would be possible to verify these results by direct analytical

(66, 74, 75, 83).

dehydrated. It should be noted that we do not speak of a layer of ions a t the

conclusion be accepted (and additional evidence for it will be presented a t the

This equation is used below for the evaluation of K'.

concentrated solutions of salts whose ions are believed to be unadsorbed, I’rfi

where f is the force acting between charges q1 and q2 in a vacuum a t a distance T .

This can be integrated once by introducing the identity

which after substitut,ion and integration gives :

According to equation 39 qd is the surface charge density, the total charge in a

td = & p F Z n e i(e--rieS.lkT - 1) (40)

Kd = sinh ze$"/2kT (46)

Kd. . .. . . . 72.2 74.3 79.9 90.0 105 128 159

Kd. . . ..., 50.2 82.9

15.9 20.2 26.2 34.4 83.2

Remembering that KO is not necessarily a constant, one writes also, as a defi-

capacity C of more dilute solutions of sodium fluoride was calculated. The

1 ~ " " " " " " ' l