Professional Documents
Culture Documents
View Article Online / Journal Homepage / Table of Contents for this issue
Movement of solute between the bulk of the liquid phase and the
electrode/solution interface is an essential feature of all electrode re-
actions ; dissolved reactants must be supplied to the interface or products
must be removed. This movement can occur in three ways: (i) by
diffusion ; (ii) by convection (i.e. movement of the liquid containing the
solute) ; (iii) by ionic migration. Migration and convection cannot
account for the whole of the material required or produced by the reaction,
and the co-operation of diffusion is always necessary ; the concentrations
at the interface must therefore differ from those in the bulk of the solution.
Such concentration differences are responsible for concentration over-
potential 1 and limiting currents, and have an important influence on the
kinetics of electrode reactions generally.
These effects are usually discussed in terms of the Nernst diffusion
layer hypothesis (see next section), which can hardly be considered a
satisfactory theoretical treatment of the subject. I n this paper a different
method is adopted ; the variables involved in the problem are found by
inspection of the fundamental equations and certain properties of the
relations between these variables are then derived by dimensional analysis.
Further information can be obtained by considering experimental and
theoretical results relating to the closely analogous subject of heat transfer.
It is necessary to restrict the discussion to systems in which the
transport number, T,, of the ion or molecule concerned is constant ; in
other cases, calculation of the effect of migration becomes too complicated.
T , is constant, or approximately constant, for :
(a) non-ionised solutes or ions in a large excess of some other electro-
lyte ( T , = o ; no migration) ;
(b) solutions containing only two ions (i.e. a single salt).
No more need be said about (a); in ( b ) , migration is responsible for a
definite fraction of the total transfer of solute, and the remainder is
transferred by diffusion and convection, which proceed as if no migration
were occurring.l* The transfer rate, j , defined below, always refers to
this latter fraction, due to diffusion and convection and not to migration.
The same transfer rate is thus relevant to non-electrolytic reactions a t
a solid/liquid interface.
It is sometimes possible t o ensure that the liquid containing the elec-
trode is stationary, so that there is no c o n ~ e c t i o n . ~But the liquid is
usually in motion, owing to (i) intentional stirring (forced convection),
Agar and Bowden, Proc. Roy. Soc. A , 1938,169,206.
a See Bridgman, Dimensional Analysis (Newhaven, 1931).
Roseburgh and Lash Miller, J . Physic. Ckern., 1910,14, 816; see also Levich
(ref.1l).
4e.g. Sand, Phil. Mug., 1901,I, 45.
View Article Online
J. N. AGAR 27
(ii) gravitational motion due to variations in density associated with
Published on 01 January 1947. Downloaded by University of Massachusetts - Amherst on 06/10/2014 12:27:23.
But all such solutions are a t present limited to cases where the geometry
of the cell and electrode is simple, and restrictions may also be imposed
on the size of the electrode. In practice, the design of cells and electrodes
is governed by factors other than geometrical simplicity, and electro-
chemical experiments are often carried out under conditions which, from
the hydrodynamical point of view, are impossibly complicated.
3
where is the mean value of j over the whole electrode (mean transfer
rate), and the same equation is used without modification for non-planar
electrodes.
5 Rayleigh, Natuve, 1915,95, 66. McAdams, Heat Transmission (McGraw-
Hill, 1942). ten Bosch, Wci'rmeiibertragung (Springer, 1936).
6 e.g. Shenvood, Absorption and Extraction (McGraw-Hill, 1937).
7 King and Howard, I n d . Eng. Chew., 1937, 29, 75.
SHixson, ibid., 1944, 36, 488. Hixson and Baum, ibid., 19411 33, 478;
1942,34, 120.
9 Uchida, J . SOC.Chein. I n d . .Ja?an, B , 1933,36,416 ; 1934,37,456. Uchida
and Nakayama, ibid., 1933, 36, 635.
See ten Bosch (ref.5) and Goldstein (ref.18),ch. 14.
l1 Levich, Acta Physicochinz., 1942, 17,257.
l2 Levich, ibid., 1944, 19, 117.
Is Levich, ibid., 1944, 19, 133.
I4Nernst, 2. physik. Chem., 1904,47,52.
View Article Online
j- = I ( I - T,)
AzF
where I is the total current, A the electrode area, F the Faraday, z the
number of Faradays required per mole. reacting and T , the transport
number of the reacting ion.
I n reality the transition from diffusion alone a t the electrode surface,
t o convection alone in the bulk must take place continuously ; diffusion
and convection will be equally important a t some distance 6 from the
surface, where 6 is of the same order as 6. Any factor which increases
convection (e.g. increase of liquid velocity) will diminish 6 ; an increase
of the diffusion coefficient will enlarge the zone in which diffusion is
predominant and will thus increase 6. Similar changes in 6 are to be
expected.16
The experimental values of 6 are around 0.05 cm. for vertical surfaces
in unstirred solutions (natural convection) and 0.005 cm. or less in stirred
solutions, according t o the rate of stirring.ls It thus appears that forced
convection due t o the ordinary laboratory methods of stirring is much
greater than natural convection ; the latter can therefore be neglected in
stirred solutions, and the two types of convection can be considered
separately.
-
3c = D div grad c - v grad c
3t
. . (za)
where t is time, D is the diffusion coefficient, and u, ZI and w are the x ,
y and z components of the velocity v a t the point x , y , z . The first term
on the right represents the effect of diffusion, which is assumed t o obey
Ficks law as in a stationary liquid ; the second term gives the change in
concentration due t o convection,
As the equation is linear in c, we can replace c by the concentration
difference 8 = c - c , ; hence :
where j has the same meaning as in (I) and n represents distance from the
electrode measured normally t o the surface.
l6 King, J . Amer. Chem. SOL, 1935,57, 828.
I6Nernst and Merriam, 2. physik. Chem., 1905, 53, 235. Brunner, ibid.,
1904,47,. 56 ; and see the review of the subject by Glasstone and Hickling,
Electrolytzc Oxidation and Reduction (Chapman and Hall, 1935).
View Article Online
J. N. AGAR 29
It will be assumed that c, has the same value at all points on the elec-
Published on 01 January 1947. Downloaded by University of Massachusetts - Amherst on 06/10/2014 12:27:23.
-+
3U
3% + =o
where g is the acceleration of gravity and g(p - p,) is the excess force
resulting from variation in p, the x axis being supposed vertical.
In the present case, the variation in density arises from variations in
concentration, and if only one solute is concerned, we may put
(8)
J. N. AGAR 31
of j , I/, V , I, D and (c, - c,,), and the analogous variables which occur in
Published on 01 January 1947. Downloaded by University of Massachusetts - Amherst on 06/10/2014 12:27:23.
Reynolds number : Re =-
uz
V
Nusselts number :
Prancltls number : Pr
D
=
where 0 = (c, - c o ) .
The nomenclature follows the usage of hydrodynamics and heat
transfer ; the nature of the corresponding thermal quantities will be clear
from Table 1. The relation between the six variables can thus be reduced
to :
Nu = F(Re, Pr) . * (10)
The form of the function F depends only on the shape of the apparatus.
(2) Natural Convection.-The characteristic velocity U does not
appear in natural convection problems, but the quantities g and a have
t o be included. They need not be considered separately, since only their
product (gu) occurs in the equations (see (91); the dimensions of (gm)
are given in Table I. From the six variablesj, I, V , D ,0 and (g.), three
dimensionless quantities can be formed ; the usual choice is Nzt and PY
as in forced convection, together with :
z3ga0
Grashofs number : GV = -
v2 *
We thus have :
NU = G(Gr, PY) . (11)-
where the form of the function G depends only on the shape of the apparatus.
Under certain conditions this equation can be simplified.lD If the
velocity of motion is small and steady, the acceleration of any element
of the fluid (i.e. the left hand side of ( 9 ) ) becomes negligible ; the pressure
gradient and gravity are balanced by viscous forces alone. (9) can now
be rewritten so that v appears only in the pressure terms, which are of
no further interest, and in a coefficient ( g m ) / v; in the final solution v
must therefore occur only in this form. This condition is satisfied by
introducing Gr and PY into the function G as their product GY . Pr, so that :
nuinbered I, 2, 3, etc., are additive, the term gas in (9) must be replaced
by g(cr,8,+ +
a,O, . . .), and there must be equations similar to (3),
(4) and ( 5 ) for each solute, involving its concentration and its diffusion
coefficient (Dl, D,,etc.). It does not seem possible to derive simple
expressions analogous to (11) and (IZ), but approximate values of Nu
will be given by
J. N. AGAR 33
There is little direct information about the value of b, which determines
Published on 01 January 1947. Downloaded by University of Massachusetts - Amherst on 06/10/2014 12:27:23.
using the mean velocity of flow (cc. per sec. per sq. cm. cross section)
and the diameter as the characteristic velocity, U , and length, 1. Similar
equations have been proposed for transverse flow past wires and pipes.
It should be possible to calculate diffusion layer thicknesses from equations
such as (17) (cf. King and Howard 7), and a rough calculation, using (17)
for a tube I cm. in diameter with U = IOO cm. per sec. and values of v
and D given in Table I1 leads to Sr fi 0 . 0 0 2 cm., which is of the order
commonly observed, although there do not appear to be any recorded
measurements of S for flow through tubes. Uchida has found, more-
over, that the rate of attack of copper tubes by ammonia solution is
proportional to the 0.79 power of the velocity of flow, and has drawn
attention t o the close agreement with the exponent 0.8 in (17). The rate
of attack is probably controlled by the rate a t which dissolved oxygen
reaches the surface of the metal.
Laminar flow along a flat electrode has been treated theoretically by
LevichJ1l and, with rather different approximations, by Eucken.22 The
general equation for convection in this case, if Pr > I,* has the form : 23
Nu = const. (Re)+(Pr)*.
Levich has also considered a rotating disc electrode; putting D = IO-~
cm.2 sec.-l and v = 1 0 - 2 cm.2 sec.-1 (see Table 11), he finds :
0'022
-
a= . cm.
drev. per sec.
Natural Convection.-A theoretical solution of the natural convection
equations for heat loss from vertical plates in air has been obtained by
Schmidt, Pohlhausen and Beckmann ; 2* when reduced t o the simplified
form ( IZ), i t becomes :
NU = G' (Gr . P Y )= 0.525 (Gr . P r ) ) = 0 - 5 2 5 -(z3fiB)f
(18)
and this equation should hold for other types of convection in other media.
N u is independent of the width of the plate provided i t is large compared
t o 6, and the relevant characteristic length to be used in (18) is the height.
A similar expression is given by Levich.12
Fig. I shows the theoretical relation between log Nu and log (Gr Pr) .
according t o (IS), and also according to data derived from heat transfer
measurements given by M ~ A d a r n s . ~ The two curves agree well from
Gr . Pr = 1 0 4 to Gr . Pr = 1oS. If the diameter is used as the character-
* Pr = v / D is of the order 1 0 3 for diffusion of electrolytes in aqueous solution
at room temperature. See Table 11.
21 Glasstone and Hickling, ref.16.
22 Eucken, 2. Elektrochem., 1932,38, 341.
23 See, e.g., Goldstein, ref.l*, ch. 14.
24 See ten Bosch (ref.5),p. 159.
B
View Article Online
1
cm.
-- --- -
(a) 1'0 40x10~ IO-~ IO-~ IO-~ 4x106 26.3 0.038
J. N. AGAR 35
SO that 6, should change appreciably with the concentration difference 0.
Published on 01 January 1947. Downloaded by University of Massachusetts - Amherst on 06/10/2014 12:27:23.
and
6, = k,O-n . -
(cf. (19))
I = k 2 j = k,8*+1
where I is the total current and R,, k , and k , are constants.* Assuming
further that the potential of the electrode is controlled by the concentra-
tion of only one solute and that diffusion potentials can be neglected, the
concentration overpotential, vC, is given by :
Summary.
I. The method of dimensional analysis is applied t o the diffusion and
convection processes which govern the transfer of a solute between an
electrode and the bulk of a solution, and the results thus obtained are
extended by consideration of experimental and theoretical work on the
closely analogous subject of heat transfer.
2 . The thickness of the diffusion layer at a vertical electrode in un-
stirred solution is calculated from experimental heat transfer data, in
satisfactory agreement with observed values.
3. It is shown that the thickness of the diffusion layer, 6, a t an electrode
in a stirred solution is independent of the concentration difference be-
tween the interface and the bulk of the solution. I n unstirred solutions,
some variation of 6 with concentration difference is t o be expected, and
the resulting changes in the relation between current and concentration
overpotential are briefly discussed.
R6sum6.
( I ) Lanalyse dimensionnelle est appliqube d la diffusion et & la con-
vection, qui rkgissent le transport dune substance dissoute entre une
blectrode et le corps de la solution; les rbsultats ainsi obtenus sont
&endus par comparaism avec des travaux expkrimentaux et thboriques
sur un sujet analogue, le transport de chaleur.
(2) Lkpaisseur de la couche de diffusion sur une klectrode verticale
en solution non agitke est calculbe B partir de valeurs expbrimentales
du transport de chaleur et se trouve en bon accord avec les rksultats
observbs.
(3) On prouve que lkpaisseur de la couche de diffusion, au voisinage
de lklectrode dans une solution agitke, est indkpendante de la diffkrence
de concentration entre linterface klectrode-solution et le corps de la
solution. Dans des solutions non agitkes, on doit sattendre B une cer-
taine variation en fonction de la diffkrence de concentration ; les change-
ments qui en rksultent dans la relation entre le courant et le survoltage
pax concentration sont brikvement discutks.
Zusammenfassung.
(I) Die den obergang von geloster Substanz zwischen einer Elektrode
und der Hauptmenge der Losung kontrollierenden Diffusions- und
Konvektionsprozesse werden mit Hilfe der Dimensionsmethode analysiert.
Die Erwagung von experimentellen und theoretischen Arbeiten auf dem
eng verwandten Gebiet der Warmeu bertragung ermoglicht eine Enveiterung
der so erhaltenen Ergebnisse.
(2) Die aus den Resultaten von W2rmeu bertragungsmessungen berech-
ne te Dicke der biff usionsschicht an einer vertikalen Elektrode in einer
ungeruhrten Losung stimmt gut mit den beobachteten Werten iiberein.
View Article Online
J. N. AGAR 37
(3) Es wird gezeigt, dass die Dicke der Diffusionsschicht an einer
Published on 01 January 1947. Downloaded by University of Massachusetts - Amherst on 06/10/2014 12:27:23.