Journal of Electroanalytical Chemistry 500 (2001) 52 – 61

www.elsevier.nl/locate/jelechem

Analysis of diffuse-layer effects on time-dependent interfacial

kinetics
Antoine Bonnefont a,*, Françoise Argoul a, Martin Z. Bazant b
a
Centre de Recherche Paul Pascal, A6enue Schweitzer, 33600 Pessac, France
b
Department of Mathematics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA

Received 6 June 2000; received in revised form 29 September 2000; accepted 1 October 2000

Dedicated to Professor R. Parsons on the occasion of his retirement from the position of the Editor in Chief of the Journal of
Electroanalytical Chemistry and in recognition of many contributions to electrochemistry

Abstract

We investigate the subtle effects of the diffuse charged layer on interfacial kinetics by solving the governing equations for ion
transport (Nernst–Planck) with realistic boundary conditions representing reaction kinetics (Butler– Volmer) and compact-layer
capacitance (Stern) in the asymptotic limit m= uD/L“0, where uD is the Debye screening length and L is the distance between the
working and counter electrodes. Using the methods of singular perturbation theory, we derive the leading-order steady-state
response to a nonzero applied current in the case of the oxidation of a neutral species into cations, without any supporting
electrolyte. In certain parameter regimes, the theory predicts a reaction-limited current smaller than the classical diffusion-limited
current; this over potential effect is not due to ohmic drop effects in the bulk of the cell but rather to antagonist processes involved
in the surface charge transfer and diffuse layer charging respectively. We demonstrate that the charging of diffuse charge, since
it is intimately coupled to the surface reaction and cannot be considered independently, plays a fundamental role in nonequi-
librium surface reactions when the transport of one of the reacting species is coupled to the total interfacial response of the
compact and diffuse layers. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Dynamic diffuse-layer effect; Time-dependent interfacial kinetics; Stern layer; Nernst– Planck continuum model; Asymptotic analysis

1. Introduction two types of models have been used: either (i) the
diffuse layer is treated as a continuum retaining the
The influence of diffuse-layer structure on electro- essential features of the mean concentration and electric
chemical surface reactions has been studied for many field profiles [3–10] or (ii) the interfacial double layer is
years since the pioneering work of Frumkin [1]. The treated as an idealized set of nested layers (e.g. the
double layer is typically modeled as a stationary outer inner and outer Helmholtz planes) consisting of discrete
‘diffuse layer’ (or ‘Debye layer’) of charged species in molecular and/or ionic species [11 –17]. When diffuse-
transport equilibrium and a inner ‘compact layer’ (or charge effects are incorporated into either type of
‘Stern layer’) where the electroactive species can react at model, the diffuse layer is assumed to have a stationary,
an effective electrostatic potential br (reaction plane) equilibrium distribution of charges. The dynamical ef-
which is different from the electrode potential be by an fect of the diffuse layer on charge transfer kinetics has
amount Dbs =be −br, the Stern-layer voltage. In such only rarely been considered since it was first mentioned
models the only effect of the diffuse layer on interfacial by Levich forty years ago [18,19].
kinetics is to modify the activation energy and effec- The present study is motivated by the idea that the
tively rescale the interfacial concentration of the react- dynamical charging of the diffuse layer could also have
ing species [2–15]. Within this theoretical approach, an important effect on surface reaction kinetics in weak
electrolytes (whose Debye screening length significantly
* Corresponding author. Fax: +33-5-56845600. exceeds the effective width of the compact layer). In

0022-0728/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S0022-0728(00)00470-8
 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61 53

this work, we adopt the classical Nernst – Planck ‘mean- C+ ze − “ N (1)

field’ continuum model for ion transport, without con-
in the presence of an anion A, which aids in charge
sidering the possible steric effects due to finite ion sizes
neutralization. For analytical convenience, we also as-
[20]. We do, however, allow for the compact Stern layer
sume that the electrolyte is symmetric, zC = −zA =z.
to have a different permittivity mS from the bulk solvent
The continuum fields representing the state of the
mb as described in Ref. [21], and we also consider
solution are the cation, anion and neutral molecule
nonlinear reaction kinetics given by the Butler – Volmer
concentrations (moles/volume) CC(X,t), CA(X,t) and
equation without any approximations (e.g.
CN(X,t), respectively, and the electrostatic potential
linearization).
b(X,t). (Note that we use a lower case t for dimen-
There have been few attempts in the literature aimed
sional time, to distinguish it from the absolute tempera-
at characterizing the dynamics of the diffuse layer and
ture T.) Although we allow for time-dependent electric
its interplay with surface reactions, and existing studies
fields and currents, it can be shown that magnetic
have been mostly numerical in various approximations
effects are negligible at the time scales under consider-
[10,18,22,23]. Here we illustrate the power of asymp-
ation so the electric field is given by E= − #Xb. In the
totic analysis to provide analytical predictions of the
dilute solution approximation, the average concentra-
statics and dynamics of realistic interfacial double lay-
tions evolve according to the Nernst –Planck equations
ers in response to dc. Asymptotic analysis has rarely
describing diffusion and electromigration
been applied to electrochemical systems, and previous
studies have focused only on static cases with oversim- #tCC = D#2XCC + v#X (CC#Xb) (2a)
plified boundary conditions [6,7,24 – 27]. Recently, more
realistic boundary conditions have been proposed to #tCA = D#2XCA − v#X (CA#Xb) (2b)
model the interplay between the diffuse and compact #tCN = D# CN 2
(2c)
X
layers, and asymptotic analysis has provided insight
into the static response of a binary cell near the classi- where the diffusion coefficients and electrostatic mobili-
cal diffusion-limited and reaction-limited currents [21]. ties of all species are assumed to be constant and equal
In this work, the same realistic equations and boundary to D and v= zFD/RT respectively. The electrostatic
conditions are used to model the dc response of a potential is given by Poisson’s equation
three-species electrochemical cell (composed of a binary − ‡b#2Xb= zF(CC − CA) (3)
electrolyte plus a neutral solute molecule). To our
knowledge, this is the first asymptotic analysis of a where ‡b is the permittivity of the bulk solvent (as-
time –dependent, nonequilibrium electrochemical sys- sumed to be constant outside the compact layer) and F
tem. is the Faraday constant.
The article is organized as follows. In Section 2, we The conduction current density due to the electromi-
describe the electrochemical system and present the gration of charged species is given by
model equations for ion transport and surface reac-
Jc = − zF[D#X (CC − CA)+ v(CC − CA)#Xb] (4)
tions. In Section 3 leading-order asymptotic approxi-
mations of steady solutions to these equations are Therefore, the total current density is given by
derived in the case of an applied current.
J(t)=Jc(X,t)− ‡b#t #Xb(X,t) (5)
which is the sum of the conduction and displacement
2. Description of the model equations current densities. Note that J is uniform across the cell
since #X J= 0 and therefore J is the experimentally
2.1. Transport equations controlled (or measured) current density in a time-de-
pendent situation, not Jc.
For simplicity, we consider an electrochemical cell
with a planar geometry, possessing one electrode at 2.2. Boundary conditions
X = 0 (working electrode) and another at X =L (coun-
ter electrode), and hence we ignore the possible effect of At each electrode, X= 0 and X =L, there are three
convection on ion transport. (Note that we use upper- boundary conditions on the ionic fluxes expressing
case letters to denote quantities with dimensions to mass conservation in the electrochemical surface reac-
distinguish them from the corresponding dimensionless tion Eq. (1)
quantities introduced below, which are denoted by
lower-case letters.) In order to limit our study to a D#XCC + vCC#Xb= − JF/zF (6a)
small number of species, we analyze the case of a cation
D#XCA − vCA#Xb= 0 (6b)
C which reacts at the electrodes to produce a neutral
molecule N according to the electrochemical reaction D#XCN = JF/zF (6c)
54 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61

(We adopt the sign convention that a cathodic current uS = ‡S/CS (9)
is negative.) In the mean-field approximation, the
defines an effective width for the Stern layer. We do not
Faradaic current density JF at each electrode (which
claim that uS corresponds to any well-defined molecular
contributes to the surface reaction, unlike the displace-
distance, but only that it captures the combined effect
ment current density) is related to the local concentra-
of the compact layer capacitance and permittivity. In
tions and potential through the Butler – Volmer kinetic
our numerical calculations below, we assume that the
equation (applied at X =0 and X = L)
permittivity of the Stern layer is ‡S = 10‡O, compared
JF = − KOCC exp(−hOzFDbS/RT) with the value ‡b = 80‡O for a water solvent, consistent
+KRCN exp(hRzFDbS/RT) (7) with previous studies [29]. From the equation uS =‡S/
CS and the literature data on CS  80 mF cm − 2 [29] an
where KO and KR are the oxidation and reduction estimate of uS  1 A, can be computed. We also assume
kinetic rate constants, respectively, CO and CN are the that uS is independent of the local ionic concentrations.
interfacial concentrations, hO and hR are the transfer
coefficients (which are set equal to 1/2 below) and
2.3. Dimensionless equations and boundary conditions
DKS = Ke −Kr is the voltage drop across the Stern com-
pact layer mentioned above. Following Frumkin [1], we The first step in any asymptotic analysis is to scale all
imagine applying Eq. (7) at the outer Helmholtz plane, quantities appropriately and identify the relevant di-
or any other convenient molecular distance which acts mensionless groups. Here we scale length to the elec-
as the edge of the continuum region. Note we do not trode separation L, time to the diffusion time across the
apply the Butler –Volmer equation across the entire cell L 2/D, potential to the thermal voltage zF/RT,
interface (including both the compact and diffuse lay- concentrations to the mean anion concentration C*
ers) as is frequently done.
No other boundary conditions are typically men-
tioned in electrochemistry textbooks because none are
C* =
1 & L
CA(X)dX (10)
L 0
needed when the common assumption of electroneutral-
and current to the classical diffusion-limited current of
ity is made [28]. In this work, however, since we treat
Nernst
diffuse charge explicitly we need another boundary
condition on the potential. Ignorance of this boundary 4zFDC*
condition is typically hidden in the ‘zeta-potential’ n JDL = (11)
L
(the potential drop across the diffuse layer DƒD) or
equivalently the total charge in the diffuse layer. In The dimensionless variables at these scales are
colloidal science these quantities can be assumed to be x=X/L, ~= tD/L 2, j=J/JL,
constant properties of a surface, but in electrochem-
istry, the zeta potential of an electrode should be deter- ci = Ci/C*(i= C,A,N), ƒ= zFb/RT (12)
mined self-consistently from the microscopic We also introduce the mean concentration of charged
electrochemical boundary conditions. We will see that species c(x, t) and the charge density z(x, t)
the zeta potential of a working electrode can vary
widely with electrochemical conditions, especially in c =1/2(cC + cA) (13a)
time-dependent situations.
z= 1/2(cC − cA) (13b)
A general expression for the missing boundary condi-
tion has recently been proposed based on the Grahame In terms of these variables, the Nernst –Planck and
model of the interface, which contains the Stern model Poisson equations take the dimensionless forms (using
as a special case [21]. Here we adopt the Stern model, the Einstein relation v = zFD/RT)
which postulates a constant capacitance CS for the
compact layer #~c= #2Xc+ #X (z#Xƒ) (14a)

DbS =
!
− ‡S#Xb/CS at X = 0
(8)
#~z= #2Xz+ #X (c#Xƒ) (14b)
‡S#Xb/CS at X = L #~cN = #2XcN (14c)

where ‡S is an effective permittivity for the Stern layer. − ‡ # ƒ= z

2 2
X (14d)
Note that we do not consider, in this model, any where the dimensionless group ‡= uD/L indicates that
chemical adsorption effect. In this work, we model the natural scale for charge screening is the Debye
experiments in which the current J(t) is prescribed and length uD given by
the voltage difference between the working electrode
and the reference electrode DKtot =Ke −Kref is deter- ‡bRT
u 2D = (15)
mined by solving the equations. Note that 2F 2C*
 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61 55

rather than the electrode separation L. In this work we is identically zero, but this ‘macroscopic intuition’
consider the typical case in which ‡ is much smaller (which is valid in the limit ‡ “0) is flawed whenever
than unity (for a cell of typical length L = 1 cm and diffuse charge is being treated explicitly (‡\ 0), as we
uD B l00 nm, ‡ B 10 − 5), which is the basis for our do here, because there is a total charge (of order ‡) in
asymptotic analysis. Since ‡ multiplies the highest the diffuse layers which exists to satisfy the nonlinear
derivative in the equations, it is a singular perturbation and asymmetric reaction conditions at the electrodes
which must be treated using the method of boundary [21].
layer theory [30]. Note that the expression for the The fact that q(t) can vary in time does not contra-
dimensionless current (which is uniform across the cell dict the conservation of law for charge because dq/dt is
since #j/#x = 0) related by Eq. (14b) to the Faradaic current densities at
the electrodes
j(t) = jc(x,t)+jd(x,t)
where
(16)
#~q=
& 1
#~z(x,t)dx =2[ jF(0,~)− jF(1,~)] (22)
0
1
jc = − (#xz + c#xƒ) (17a) rather than the total current densities (which are equal).
2 Since the total current density j= jc + jd is uniform
1 across the cell, any changes in the total charge of the
jd = − ‡ 2#~ #xƒ(x,t) (17b) cell are caused by temporary differences between the
2
displacement currents at the two electrodes.
involves the perturbation parameter multiplying the
displacement current density jd which indicates that it
becomes important only at high frequencies (and/or 3. Steady state response to a dc current
small length scales).
The flux boundary conditions at x =0 and x =1 take 3.1. The Smyrl–Newman equation
the dimensionless forms
#xc +z#xƒ = − 2jF (18a) In this section, we derive leading-order steady-state
solutions of Eqs. (14), (18) –(20) in response to a con-
#xz+c#xƒ = − 2jF (18b) stant applied dc current density j, which are valid after
transients have died away, following the analysis of
#xcN =4jF (18c)
Ref. [21]. Integrating the steady Nernst –Planck equa-
where the dimensionless Faradaic current at each elec- tions once, we arrive at a system of ordinary differential
trode is given by equations

jF = − kO(c+ z) exp( −hODƒS) +kRcN exp(hRDƒS) #xc+ z#xƒ= − 2j (23a)

(19)
#xz+ c#xƒ= − 2j (23b)
where kO = KOL/4D and kR =KRL/4D are the dimen-
#xcN = 4j (23c)
sionless kinetic constants and DƒS =zFDƒS/RT. The
dimensionless Stern boundary conditions − ‡ 2#2xƒ= z (23d)

DƒS =
! −lD‡#xƒ at X = 0
(20)
By subtracting the first two equations we obtain
lD‡#xƒ at X =L #x (c− z)= (c −z)#xƒ (24)

involve a second dimensionless group lD =uS/uD, which is easily integrated to show that (at steady-state)
which is the ratio of the effective widths of the compact the anions are in Boltzmann thermal equilibrium
and diffuse parts of the interfacial double layer. The throughout the cell
Gouy –Chapman model (with no compact layer) corre- cA = c− z8exp(ƒ)=exp(− zAFb/RT) (25)
sponds to the limit lD =0 while the Helmholtz model
(with no diffuse layer) corresponds to the limit lD = . since they do not react at the electrodes.
In this work, we consider the full range of values for Our analysis begins by combining the three coupled
lD, but always require ‡1. equations Eqs. (23a), (23b) and (23d) into a single
It is ubiquitous in electrochemistry to assume that equation for the potential, which is decoupled from the
the total charge in the cell fourth equation Eq. (23c) for the neutral species con-

q(t) =
& 1
z(x,t)dx (21)
centration (except through the boundary conditions).
Substituting Eq. (23d) into Eq. (23a) and integrating we
0 obtain
56 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61

‡ 2
c(x) =co −2jx+ (#xƒ)2 (26) layer in the special case lD = 0, but it is more easily
2
generalized to the time –dependent case. We also study
where co is an integration constant to be determined by for the first time the novel nonlinear effects of the
the boundary conditions. If we substitute this expres- coupled Stern and Butler –Volmer boundary conditions
sion and Eq. (23d) into Eq. (23b) we obtain a third when lD \ 0, which include the possibility of a reac-


order nonlinear equation for the potential [21] tion-limited current.
1 n
‡ 2 #3xƒ − (#xƒ)3 +(2jx −co)#xƒ =2j (27) 3.2. Leading order solution of the outer problem
2
A similar equation was first derived by Smyrl and In the outer region, we set ‡=0 in Eq. (27) to derive
Newman for the electric field of a rotating disk elec- the leading order approximations
trode near the diffusion-limited current [7]. Once this
equation is solved for the potential ƒ, the concentration #xƒ= 2j/(2jx − co)+ O(‡) (28a)
c and charge density z are computed from Eqs. (26)
c(x)=co − 2jx+ O(‡) (28b)
and (23d), respectively.
The Smyrl –Newman Eq. (27) cannot be solved ana- z(x)= O(‡ 2) (28c)
lytically in terms of elementary functions, but due to
the singular perturbations (terms involving ‡), it is even from Eqs. (26) and (23d), where the integration con-
difficult to solve numerically. Both of these difficulties stant co can be evaluated using the conservation of
can be conveniently resolved with asymptotic analysis anions over the cell 10cA(x,t)dx =1 at zeroth order in ‡.
[6,7,30]. The idea is to construct a uniformly valid
cO = 1+ j (29)
approximation for all x (in the limit ‡“0) by adding
the leading order solution to the ‘outer problem’ in the since q(t)= O(‡). Note that the outer region is neutral
neutral bulk region ‡x  1 −‡ to the leading order (z=0) at leading order, as is commonly assumed in
solutions to the ‘inner problems’ in the charged diffuse electrochemistry. The bulk potential profile at leading
layers 0Bx‡ and 1 − ‡x B 1 and subtracting the order is obtained by a simple integration
overlapping parts through asymptotic matching, which
is valid well below the diffusion-limited current density ƒ(x)−ƒe = Dƒi + log [(co − 2jx)/co]+ O(‡) (30)
j =1. As emphasized by Newman [6], the asymptotic where Dƒi = ƒe − ƒ(0) is the total (leading-order) po-
approach used here reduces to the familiar Gouy – tential drop across the interface consisting of contribu-
Chapman (Poisson – Boltzmann) theory of the diffuse tions from the diffuse (Debye) and compact (Stern)
layers
Dƒi = DƒD + DƒS (31)
which must be calculated by asymptotic matching with
the inner problem. (Note again that DƒD is the familiar
zeta potential.) The equation for the neutral species
concentration is also easily integrated, writing the con-
servation over the cell of the sum of the cation and
neutral species numbers at zeroth order in ‡.
cN = k− 2j(1−2x)+ O(‡) (32)
where k is the averaged neutral species concentration.
Typical bulk concentration, electric field and poten-
tial profiles are shown in Fig. 1. (We set k= 1 in all the
numerical results presented here.) The potential profile
ƒ is slightly curved (logarithmic dependence on x), and
the electric field #xƒ is not constant since the migration
term is a nonlinear function of the concentration c
(which varies linearly in the bulk when j" 0). The
profiles of ƒ and #xƒ corresponding to opposite values
of j (e.g. 0.4 and − 0.4) are not symmetric across the
Fig. 1. Steady-state profiles of ƒ (a), dƒ/dy (b), c (c), and cN (d) at
leading order in the ‘outer’ (bulk electroneutral) region for j = 0.4, 0 x-axis because the product of the reduction of cations is
and − 0.4 and hO = 1. hO = hR = 1/2. It should be noted that all neutral. On the other hand, the profiles of c and cN are
quantities plotted in Figs. 1–6 are dimensionless. symmetric when the current density is reversed.
 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61 57

3.3. Leading order solution of the inner problem which is essentially the Gouy –Chapman solution to the
Poisson –Boltzmann equation, although our boundary
The leading order approximation for the diffuse layer conditions (specifying K) are substantially different
potential at the x = 0 electrode is obtained by changing from the classical theory, which corresponds to the
variables to the inner coordinate y =x/‡ in Eq. (27) limit lD “ 0. To clarify this connection, we let „=
Dƒi + ƒ, and observe (following some algebra) that the
1
− #3yƒ+ (#yƒ)3 +co#yƒ =O(‡) (33) solution Eq. (40) satisfies
2
#y„= 2 cosinh(„/2) (41)
where we ignore the O(‡) terms. Multiplying by 2#2yƒ
and integrating (using the matching conditions and
#yƒ, #2yƒ “0 as y“ ) we obtain
#2y„= co sinh(„) (42)
# ƒ= − #yƒ
2
y co + (#yƒ) /4
2
(34) which is the Poisson –Boltzmann equation. (Changing
which is a first-order separable equation for #yƒ. The the current density simply changes the bulk concentra-
next integration can be performed with a hyperbolic tion co = 1+j ). The Smyrl –Newman Eq. (27) is more
substitution #yƒ = 9 cocsch(u) (where csch(u)= 1/ general than the Poisson –Boltzmann equation, but the
sinh(u)), which yields the trivial equation #yu = co former is equivalent to the latter below the diffusion-
whose solution is u= co(y + K) for some constant K. limited current [21]. Combining Eqs. (36) and (41), we
Therefore, the electric field in the diffuse layer is obtain a relation between the Stern and diffuse layer
voltages
#yƒ= 92 cocsch( co(y + K)) + O(‡) (35)
DƒS = 2lD cosinh(DƒD/2) (43)
where K is related to the Stern-layer voltage through
the boundary condition at y = 0 and from Eq. (40) the constant K can be related to the
diffuse layer voltage
DƒS = 2lD cocsch( coK) (36)
tanh(DƒD/4)= e − coK
(44)
The upper sign refers to potentials below the potential
of zero charge (‘pzc.’) with DƒD, DƒS B0, and the To determine K we must solve the transcendental sys-
lower sign to potentials above the pzc. tem Eqs. (36) and (39) numerically.
Transforming to the inner coordinate in Eqs. (26)
and (23d) 3.4. Application of the electrochemical boundary
conditions
1
c = co + (#yƒ)2 + O(‡) (37a)
2 The behavior of K as a function of the ratio of the
z= − # ƒ + O(‡)
2
(37b) kinetics constants Rk = kR/kO and the current density j
y
is reported in Fig. 2 (with kR = 1 held constant). From
and substituting Eq. (35) for the electric field, we obtain Eq. (36) it is clear that large values of K correspond to
the leading order concentration and charge density small values of DƒS when j= 0, i.e. the symmetric case
profiles co = 1. The introduction of a nonzero value of j breaks
c = co[1+2csch2( co(y + K))] +O(‡) (38a) the symmetry of the K versus log (Rk ) curve. The plot
of K versus j in Fig. 2(b) illustrates the dissymmetry of
z = 9 2cocsch( co(y+ K)) coth( co(y + K)) + O(‡) the redox couple, since when the current is positive the
(38b) oxidation of neutral species produces positively charged
The constant of integration K, or equivalently the Stern cations which accumulate in the diffuse layer, thus
layer voltage DƒS, is determined by solving a transcen- hindering the negative charging of the diffuse-layer. In
dental equation provided by the Butler – Volmer kinetic Eq. (2) we also display the Stern and total interfacial
boundary condition at y =0 (see below) voltages, DƒS and Dƒi = DƒS + DƒD. We observe that
when j increases beyond 0.3, DƒS and Dƒi reach arbi-
j = − kO(c+z) exp( −hODƒS) + kR(k −2j) exp(hRDƒS) trarily large values to compensate the production of
(39) cations by oxidation (see below). For j= 0, the plots of
DƒS (Fig. 2(c)) and Dƒi (Fig. 2(e)) are symmetric with
where c(0)+ z(0)=cC(0) is a (generalized) Frumkin
respect to Rk = 1. This behavior is confirmed by the
correction [28].
plot of the different spatial profiles c, z, #yƒ and ƒ, for
The expression for the electric field Eq. (35) can be
two different values of j, in Figs. 3 and 4.
integrated analytically to obtain the diffuse layer poten-
In Fig. 3 we focus on the case of zero current. When
tial profile at leading order
the kinetic constants are equal, Rk = 1, this corresponds
ƒe − ƒ=Dƒi +4tanh − 1(e − co(y + K)
) (40) to the pzc., but when Rk " 1, the system builds a
58 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61

To understand better this unexpected behavior ob-

served for j \ 0.3, we plot the influence of the ratio
uD/uS = l − 1
D on the potential drops of the diffuse layer
DƒD, The Stern layer DƒS and the whole double layer
Dƒi = DƒD + DƒS in Fig. 5. We notice that an exchange
of potential drop between the Stern layer and the
diffuse layer occurs when the size of the diffuse layer
becomes larger than uS. The potential drops in the
Stern layer and diffuse layer compensate exactly when
j= 0, whereas the potential drop in the diffuse layer
diverges when uD increases at a nonzero current larger
than 0.3.

3.5. Theoretical polarograms

For electrochemists, polarograms have great signifi-

cance. In the present study of stationary regimes in
finite size the total cells such a representation of current
density J versus voltage between the reference and
working electrodes Dƒtot = ƒe − ƒref. is shown in Fig. 6.

Fig. 2. (a) K versus Rk =kR/kO in linear1og representation, ( j =0).

(b) K versus j for Rk =1. (c) DƒS vs. Rk in lin-log representation,
( j=0). (d) DƒS vs. j for Rk = 1. (c) Dƒi vs. Rk in lin-log representa-
tion, ( j= 0), k=1. (d) Dƒi vs. j for Rk = 1. Two different ratios l are
illustrated in the six panels: circles l =0.01, triangles l =1. hO =
hR =1/2.

charged diffuse layer to equilibrate the oxidation and

reduction currents on each electrode. For comparison,
we have also added in (dashed lines) the stationary
Poisson –Boltzmann solution for j =0, which corre-
sponds to choosing K such that lD =0 in Eq. (40).
In Fig. 4, a positive current density j =0.3 has been
chosen, with symmetric kinetic constants (Rk =1). In
this case, the system builds a charged diffuse layer to
favor either reduction (negative Faradaic current den-
sity) or oxidation (positive Faradaic current density). In
each figure, two values of uD have been chosen, one
corresponding to a weak electrolyte concentration uD =
100uS, where the diffuse layer has a strong effect on the
charge distribution and potential profiles, and another
corresponding to a strong electrolyte concentration
with uD =uS, where there is a much more significant
potential drop in the Stern layer (and hence less influ- Fig. 3. Steady-state profiles of c (a), z (b), #yƒ (c), ƒ (d), cC (e) and
cA (f) at leading order in the ‘inner’ (diffuse) layer at y = 0 for j= 0,
ence from the diffuse layer). We note that for j =0.3
Rk =5, (kR =5 and kO =1), K= 0.972 for l =0.01 (circles) and
the charge and electric field profiles shown in Fig. 4 K =1.629 for l = 1 (triangles), For comparison the Poisson–Boltz-
start to diverge at y =0, particularly so at the small mann profiles are reported as dashed lines which correspond to the
value of lD =1/100. limit l =0.
 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61 59

The significance of j= 0.3 in our numerical results is

most easily understood in the limit of a very weak
electrolyte with uD  uS (uD “ 0) where it corresponds
to a ‘reaction-limited current density’ generally given by
kR k
jRL = (45)
1+ 2kR
whenever this value is smaller than the Nernst diffu-
sion-limited current density k/2, when cN(0)= 0. (Recall
that we have chosen kR = k= 1, so that jRL = 1/3 and
k/2 = 1/2.) Instead of being caused simply by transport
limitations, the reaction-limited curren density jRL is
determined by the electrochemical boundary condi-
tions. Note that when uD = 0 (no Stern layer), and the
Butler –Volmer Eq. (39) takes the simpler form
j= − kO(c+ z)+ kR(k−2j ) (46)
and in this case, a limitation of positive current density
occurs because the first term representing the cathodic
current density is negative (but small) while the positive
second term representing the anodic current density
decreases with applied current density as the concentra-
tion of neutral species is reduced (but not necessarily

Fig. 4. Steady-state profiles of c (a), z (b), #yƒ (c), ƒ (d), cC (e) and
cA (f) in the diffuse layer at leading order for j = 0.3, Rk =1,
K= 0.523 for l = 0.01 (circles) and K= 1.394 for l= 1 (triangles).

This includes leading order voltage drops calculated for

both double layers and the bulk electroneutral region.
It confirms again that for j positive larger than 0.3, a
limitation of the current density occurs which is very
different from diffusion limited current densities which
are usually encountered in electrochemical cells. The
standard current –potential curve (‘PB’, plotted as a
dotted line in Fig. 6) shows a limiting current density
equal to k/2 which occurs when cN “0 at the working
electrode surface. When j\ 0 the production of cations
increases due to oxidation of neutral species; to main-
tain the imposed current density, the system usually
increases DƒS to compensate this excess. When lD = uS/
uD becomes much smaller than 1, the potential drop in
the diffuse layer increases much more rapidly than the
potential drop in the Stern layer and there is an en-
hancement of the electric field in the diffuse layer which
tends to strip the cations from the anode. For a given
current density ( j\0.3 in this case) the interfacial
cation concentration tends to zero. This limitation of
Fig. 5. Dependence of the potential differences DƒD (diffuse layer),
the current however is not due to transport but rather DƒS (Stern layer), Dƒi =DƒD +DƒS (double layer) with −log l. The
the reaction kinetics coupled to the diffuse layer different parameters Rk and j corresponding to each plot are inserted
charging. in panel (a).
60 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61

currence of overpotential in these curves is attributed to

ohmic drop effects. We claim here that the diffuse layer
may have a greater role than ohmic drop in the situa-
tions where one (at least) of the electroactive species is
charged.

4. Conclusion

We have performed a leading-order asymptotic anal-

ysis of a realistic set of transport equations and
boundary conditions for a three-species electrochemical
cell in which cations are reduced into 4 neutral
molecules in the presence of anions, in both stationary
and time –dependent nonequilibrium situations. By
solving the full nonlinear problem at leading order
(numerically where necessary), we have made the first
systematic exploration of the influence of the parameter
l= uS/uD (introduced in Ref. [21] to describe the cou-
pling of the Stern and diffuse layers) on stationary and
time –dependent interfacial kinetics. The mathematical
model is able to predict very subtle effects related to the
coupling between diffuse charge and interfacial reac-
tions, such as the existence of both reaction-limited and
diffusion-limited currents. The strength and novelty of
these theoretical predictions is their firm analytical basis
starting from realistic transport equations and
boundary conditions.
Fig. 6. Polarograms obtained from the asymptotic analysis for (a) An interesting extension of this study would be to
kR =kO =1 and (b) kR =kO = 0.01.
electrochemical reactions involving the reduction of
anions (or the oxidation of cations). Such systems have
depleted) at the electrode. When uD \0 (however already been the subject of many investigations, both
small), the situation is quite different because the sys- theoretical and experimental, but most of these studies
tem can avoid the reaction limitation in Eq. (39) by have not considered simultaneously the dynamics of the
producing a large positive Stern-layer voltage DƒS diffusion and diffuse layers and their possible interplay

DƒS 
1
log
 j  (47)
during the interfacial reactions, as we have in this work.
Such situations would complicate the present analysis
hR kR(k− 2j) significantly because there would be more than two
charged species, and it appears that any analytical
which suppresses the negative cathodic current density solution of the inner problem at leading order would
and exponentially enhances the positive anodic current not be possible in such cases. However, as long as ‡ is
density enough to maintain the applied total current. small (uD  L), the mathematical machinery of asymp-
This also produces a large diffuse layer voltage for totic analysis can be used to derive simplified, well
j\ jRL according to Eq. (43), although this effect is behaved equations at leading order, which would be
reduced as lD is increased. straightforward at least to integrate numerically.
The actual limiting current density in the system is Even restricting ourselves to the simple electrochemi-
controlled by the smaller of the reaction-limited current cal system studied here, there are a number of interest-
density, Eq. (45), and the diffusion-limited current den- ing directions for future work. Higher order corrections
sity j =k/2. Note that for negative current densities in ‡ could be derived (with some numerical computa-
one observes a diffusion limiting current density at tion required). It would also be interesting to solve the
j = −k/2 which corresponds to the annulation of the full system of equations with ‡= O(1), where the
neutral species concentration at the counter electrode. asymptotic analysis breaks down. This limit, which
Therefore, the system never reaches the classical limit- corresponds to very small cells on the order of the
ing value j= − 1 which would be obtained with a Debye length, has increasing importance as microelec-
strongly supporting electrolyte (purely diffusing sys- trochemical systems reach smaller and smaller length
tem). This result must be compared to the analysis of scales. We have also applied this analysis to character-
quasi-steady state voltammograms [31] where the oc- izing the ac response of the cell and we hope to
 A. Bonnefont et al. / Journal of Electroanalytical Chemistry 500 (2001) 52–61 61

elaborate on this point in a future communication and DƒD dimensionless diffuse layer voltage
to compare this analysis to real experimental data. DƒS dimensionless Stern layer voltage

Acknowledgements
References
We are very grateful to Armand Ajdari, Françis
Nadal and Jacek Lipkowski for stimulating discussions. [1] A. Frumkin, Z. Electrochem. 59 (1955) 807.
This work has been supported by the Centre National [2] P. Delahay, Chem. Rev. 41 (1947) 441.
[3] R. Parsons, in: J.O Bockris, B.E. Conway (Eds.), Modern As-
des Etudes Spatiales (CNES) under grants 97/CNES/
pects of Ectrochemistry, vol. 1, Butteworths, London, 1954, pp.
071/6850 and 793/98/CNES/7315. 103– 179.
[4] P. Delahay, Double Layer and Electrode Kinetics, Wiley-Inter-
science, New York, 1965.
Appendix A. List of symbols [5] P. Delahay, J. Phys. Chem. 70 (1966) 2373.
[6] J. Newman, Trans. Faraday Soc. 61 (1965) 2229.
[7] W.-H. Smyrl, J. Newman, Trans. Faraday Soc. 62 (1966) 207.
[8] R. de Levie, Adv. Electrochem. Eng. 6 (1967) 380.
[9] J.R. Macdonald, Trans. Faraday Soc. 66 (1970) 943.
D diffusion coefficient [10] J.D. Norton, H.S. White, S.W. Feldberg, J. Phys. Chem. 94
v mobility (1990) 6772.
zC, zA, z charge numbers zC =−zA =z [11] S. Levine, J. Colloid Int. Sci. 3 (1971) 619.
[12] W. Fawcett, J. Electroanal. Chem. 39 (1972) 474.
F Faraday constant
[13] B. Damaskin, V. Safonov, N. Federovich, J. Electroanal. Chem.
R universal gas constant 349 (1993) 1.
T absolute temperature [14] B. Damaskin, V. Safonov, N. Federovich, Elektrokhimiya
t time (translation) 29 (1993) 1124.
KO, KR kinetic constants [15] W.R. Fawcett, in: J. Lipkowski, P. Ross (Eds.), Double Layer
kO, kR dimensionless kinetic constants Effects in the Electrode Kinetics of Electron and Ion Transfer
Reactions, Wiley-VCH, New York, 1998, pp. 323– 371 Chapter
CS capacitance of the Stern layer
8.
‡b permittivity of the bulk solvent [16] W.A. Fawcett, J. Electroanal. Chem. 43 (1973) 175.
‡S effective permittivity of the Stern layer [17] L. Blum, Adv. Chem. Phys. 78 (1991) 171.
uS effective width of the Stern layer [18] B. Levich, Dokiady Akad. Nauk SSSR 67 (1949) 309.
uD Debye screening length [19] L. Gierst, in: E.B. Yeager (Ed.), Transactions of the Symposium
L distance between the electrodes (work- on Electrode Processes, Electrochemical Society Series, Wiley,
New York, 1961, p. 109.
ing and counter)
[20] I. Borukhov, D. Andelman, H. Orland, Phys. Rev. Lett. 79
lD = uS/uD (1997) 435.
‡ = uD/L [21] M. Bazant, Asymptotic analysis of diffuse charge and limiting
J current density current in binary electro chemical cells, unpublished.
JF Faradaic current [22] T. Brumeleve, R. Buck, J. Electroanal. Chem. 90 (1978) 1.
JDL diffusion-limited current [23] W.-D. Murphy, J.-A. Manzanares, S. Maté, H. Reiss, J. Phys.
Chem. 96 (1992) 9983.
JRL reaction-limited current [24] B.M. Grafov, A.A. Chernenko, Dokiady Akad. Nauk S.S.S.R.
CC, CA, CN concentrations of the species C, A and 146 (1962) 135.
N [25] A.A. Chernenko, Dokiady Akad. Nauk S.S.S.R. 153 (1963)
ƒ electric scalar potential 1963.
C* mean anion concentration [26] I. Rubinstein, L. Shtilman, J. Chem. Soc. Faraday Trans. II 75
(1979) 231.
cC, cA, cN dimensionless concentrations
[27] J. Henry, B. Luoro, Nonlinear Anal. 13 (1989) 787.
c dimensionless average concentration of [28] J. Newman, Electrochemical Systems, Prentice Hall, Englewood
charged species Cliff, New Jersey, 1991.
z dimensionless charge density [29] Electrified interfaces in physics, chemistry and biology, vol. 355
ƒ dimensionless electrical potential of NATO ASI Series C: Mathematical and Physical Sciences, in:
j dimensionless current density R. Guidelli (Ed.) Kiuwer Academic Publishers, Netherlands,
1992.
jF dimensionless Faradaic current density
[30] C.M. Bender, S.A. Orszag, Mathematical Methods for Scientists
jd dimensionless displacement current and Engineers, McGraw-Hill, New York, 1978.
density .
[31] C. Amatore, B. Fosset, J. Bartelt, M.R. Deakin, R.M. Wight-
Dƒi dimensionless double layer voltage man, J. Electroanal. Chem. 256 (1988) 255.