Significance of small voltage in impedance

spectroscopy measurements on electrolytic

cells
Cite as: J. Appl. Phys. 98, 113703 (2005); https://doi.org/10.1063/1.2137444
Submitted: 01 June 2005 • Accepted: 19 October 2005 • Published Online: 05 December 2005

G. Barbero, A. L. Alexe-Ionescu and I. Lelidis

ARTICLES YOU MAY BE INTERESTED IN

Impedance spectroscopy of an electrolytic cell limited by ohmic electrodes

Journal of Applied Physics 101, 054102 (2007); https://doi.org/10.1063/1.2709531

Theory of small-signal ac response of a dielectric liquid containing two groups of ions

Applied Physics Letters 92, 172908 (2008); https://doi.org/10.1063/1.2908044

Interdigitated electrodes as impedance and capacitance biosensors: A review

AIP Conference Proceedings 1885, 020276 (2017); https://doi.org/10.1063/1.5002470

J. Appl. Phys. 98, 113703 (2005); https://doi.org/10.1063/1.2137444 98, 113703

© 2005 American Institute of Physics.

 JOURNAL OF APPLIED PHYSICS 98, 113703 共2005兲

Significance of small voltage in impedance spectroscopy measurements

on electrolytic cells
G. Barbero
Dipartimento di Fisica del Politecnico and I. N. F. M., Corso Duca degli Abruzzi, 24-10129 Torino, Italy
A. L. Alexe-Ionescu
Departamentul de Fizica, Universitatea “Politehnica” din Bucuresti, Splaiul Independentei 313,
77206 Bucharest, Romania
I. Lelidis
Department de Physique, Laboratoire de Physique de la Matière Condensée, Université de Picardie,
33 rue Saint-Leu, 80039 Amiens, France
共Received 1 June 2005; accepted 19 October 2005; published online 5 December 2005兲
We investigate, theoretically, for what amplitude of the applied voltage to an electrolytic cell the
concept of impedance is meaningful. The analysis is performed by means of a continuum model, by
assuming the electrodes perfectly blocking. We show that, in the low-frequency range, the
electrolytic cell behaves as a linear system only if the amplitude of the measurement voltage is small
with respect to the thermal voltage VT = kBT / q, where kBT is the thermal energy, and q is the
modulus of the electrical charge of the ions, assumed identical except for the sign of the charge. On
the contrary, for large frequency, we prove that the amplitude of the applied signal has to be small
with respect to a critical voltage that is frequency dependent. The same kind of analysis is presented
for the case in which the diffusion coefficients of the positive ions is different from that for negative
ions, and for the case where surface adsorption takes place. © 2005 American Institute of Physics.
关DOI: 10.1063/1.2137444兴

I. INTRODUCTION z = ± d / 2, and it is assumed that in thermodynamical equilib-

rium the liquid contains a density N of ions of positive and
The impedance spectroscopy technique is used to char-
negative sign, uniformly distributed. The ions are assumed to
acterize the electrical properties of materials and their inter-
be identical in all the aspects, except for the sign of the
faces with the electrodes.1 According to this technique, an
electrical charge. In particular, they have the same mobility
external voltage is applied to the sample, and the electrical
current detected. In the limit of small applied voltage, the
␮ p = ␮m = ␮, where p and m indicate positive and negative
ions. The surfaces limiting the sample are also assumed iden-
electrical current is proportional to the applied voltage. In
tical, with the same adsorption energy with respect to the two
this framework, the electrical impedance defined by the ratio
types of ions, in order to avoid the problems connected with
voltage/current is independent of the amplitude of the ap-
the selective ion adsorption.4,5 In this situation the liquid is
plied voltage. Our aim in this paper is to study for what
globally and locally neutral. The presence of an external
amplitude of the applied voltage an electrolytic cell behaves
electric voltage produces a perturbation of the distribution of
as a linear system, and hence the concept of electrical im-
the ions in the liquid, in the sense that it remains globally
pedance is meaningful. Our analysis will be performed by
neutral, but now it is locally charged.6 In the following we
means of a continuum model, in the sense of Ref. 2. The
suppose that the sample is submitted to an external sinu-
paper is organized as follows. In this Introduction we enu-
soidal voltage of amplitude V0 and frequency f = ␻ / 共2␲兲. By
merate the fundamental hypotheses performed to obtain an
analytical solution for the fundamental equations of the con- indicating with n p and nm the densities of the two kinds of
tinuum model for the electrolytic cell. In Sec. II the case in ions, we have n p共z , t兲 = nm共z , t兲 = N, for V0 = 0, and n p共z , t兲
which the positive and negative ions have the same mobility ⫽ nm共z , t兲, for V0 ⫽ 0.
is considered. The generalization of the model to the case in
which the mobility of the positive ions is different from the
one of the positive ions is discussed in Sec. III. The case in II. CASE IN WHICH THE IONS HAVE THE SAME
which the surfaces adsorb ions is presented in Sec. IV. The DIFFUSION COEFFICIENT: DP = DM = D
results of the paper are discussed in Sec. V, whereas Sec. VI
is devoted to the conclusions. The fundamental equations of the problem are the equa-
The influence of the diffuse layer of the ionic charge on tion of continuity,
the impedance spectroscopy of a cell of liquid has been ana-
lyzed in Ref. 3. In the theoretical investigation presented in ⳵ nr ⳵ jr
=− , 共1兲
Ref. 3, it is considered a slab of thickness d filled by an ⳵t ⳵z
isotropic liquid. The z axis of the Cartesian reference frame
used in the description is normal to the bounding surfaces at where r = p, m, and the equation of Poisson,

0021-8979/2005/98共11兲/113703/5/$22.50 98, 113703-1 © 2005 American Institute of Physics

113703-2 Barbero, Alexe-Ionescu, and Lelidis J. Appl. Phys. 98, 113703 共2005兲

⳵ 2V q
2 = − 共n p − nm兲, 共2兲
⳵z ␧
where q is the electrical charge of the ions, and jr the density
of currents of positive 共r = p兲 and negative 共r = m兲 ions given
by

jr = − D 冉 ⳵ nr
±
q ⳵V
nr
⳵ z k BT ⳵ z
,冊 共3兲

with + for r = p and − for r = m.3 Since the electrodes are

supposed perfectly blocking, we have the following bound-
ary conditions on jr:
jr共±d/2,t兲 = 0.
The other boundary conditions of the problem are connected
with the imposed difference of potential that we assume of
the type
V共±d/2,t兲 = ± 共V0/2兲exp共i␻t兲.
As discussed in Ref. 3, the equation of continuity is nonlin-
ear. Hence, for arbitrary applied voltage the cell does not
behave as a linear system. Only in the case in which in the
presence of the external voltage the actual ionic densities,
nr = N + ␦nr共z , t兲, are such that ␦nr共z , t兲 Ⰶ N, the system be-
共4兲
FIG. 1. Frequency dependence of the critical voltage U for an electrolytic
cell. In the low frequency region U共␻ → 0兲 = VT关1 + 共1 / 2兲共␭ d␻ / 2D兲2兴. On
the contrary, in the high-frequency range, U共␻ → ⬁兲 = VT冑␻d2 / D. The curve
共5兲 is drawn for q = 1.6⫻ 10−19 A s 共monovalent ions兲, N = 4.2⫻ 1020 m−3, ␧
= 6.7⫻ ␧0, T = 300 K, d = 25 ␮m, S = 2 ⫻ 10−4 m2, and D p = Dm = 8.2
⫻ 10−12 m2 / s. The logarithms are in base 10 and ␻ is in rad/ s.
␻ cosh共␤d/2兲
c=i V0 , 共9兲
2D 共1/␭ ␤兲sinh共␤d/2兲 + i共␻d/2D兲cosh共␤d/2兲
2

haves in a linear manner, and the concept of electrical im- as reported in Ref. 8
pedance is meaningful. In this framework Eq. 共1兲 and Eq. We can now investigate for what amplitude of the ap-
共2兲, by taking into account Eq. 共3兲, can be rewritten as plied signal the condition ␦nr共z , t兲 Ⰶ N is verified. Since
⳵ ␦nr共z,t兲
⳵t
=−D 冉
⳵2␦nr共z,t兲 qN ⳵2V共z,t兲
⳵ z2
±
k BT ⳵ z 2
, 冊 ␦nr共z , t兲 = ␩r共z兲exp共i␻t兲, the maximum variation of ␦nr共z , t兲
is 兩␦nr共z , t兲兩 = 2兩␩共z兲兩 艋 兩p0 sinh共␤d / 2兲兩. Consequently, the
condition we are looking for is, by taking into account 共9兲,
⳵2V共z,t兲
⳵z 2
q
= − 关␦n p共z,t兲 − ␦nm共z,t兲兴,
␧
共6兲 V0 Ⰶ U = VT 冏 1
␭␤
+i
2 2
␻d
2D␤
coth ␤
d
2
冉 冊冏 , 共10兲

which are now linear. We look for solutions of Eqs. 共6兲 of the where VT = kBT / q is the thermal voltage. For a numerical
type ␦nr共z , t兲 = ␩r共z兲exp共i␻t兲 and V共z , t兲 = ␾共z兲exp共i␻t兲. By analysis of the analysis presented above, we suppose that q
substituting these trial functions in Eqs. 共6兲, we get = 1.6⫻ 10−19 A s 共monovalent ions兲, N = 4.2⫻ 1020 m−3, ␧

冉 冊
= 6.7⫻ ␧0, T = 300 K, d = 25 ␮m, and S = 2 ⫻ 10−4 m2.9 With
qN
i␻␩r共z兲 = − D ␩r⬙共z兲 ± ␾⬙共z兲 , these values for the physical parameters, relevant to a com-
k BT mercial 5CB nematic liquid crystal, the length of Debye is
␭ ⬃ 1.05⫻ 10−7 m. In Fig. 1 we show the frequency depen-
q
␾⬙共z兲 = − 关␩ p共z兲 − ␩m共z兲兴. 共7兲 dence of U by assuming D p = Dm = 8.2⫻ 10−12 m2 / s.9 As it is
␧ evident from this figure, in the limit of ␻ → 0, U tends to VT,
as expected. In fact, in this region the profile of the electric
Equations 共6兲 form a system of ordinary differential equa-
potential is given by the Poisson–Boltzmann equation, where
tions with constant coefficients. In the case under consider-
the thermal energy kBT is compared with the electrostatic
ation, where D p = Dm, the condition V共d / 2 , t兲 = −V共−d / 2 , t兲
energy qV. On the contrary, for large ␻ , U diverges as ␻1/2.
implies V共z , t兲 = −V共−z , t兲 and n p共z , t兲 = nm共−z , t兲. It follows By means of 共10兲 we get, in the limit of d Ⰷ ␭, the asymptotic
that ␩+共z兲 = ␩−共−z兲. A standard analysis gives expressions
␩r共z兲 = ± p0 sinh共␤z兲,

␾共z兲 = − 2共q/␧␤2兲p0 sinh共␤z兲 + cz, 共8兲

U共␻ → 0兲 = VT 1 + 冋 冉 冊 册1 ␭d
2 2D
2
␻2 ,

where ␤ = 共1 / ␭兲冑1 + i共␻ / D兲␭ , and ␭ = 冑␧kBT / 共2Nq 兲 is the

2 2 1 d
冑␻ .
U共␻ → ⬁兲 = VT 共11兲
length of Debye.7 The coefficients p0 and c appearing in Eqs. 2 冑 D
共8兲, are determined by means of the boundary conditions 共4兲
and 共5兲 and are found to be From the previous discussion, it follows that if the investi-
gations on the dielectric properties of the material have to be
Nq␤ 1 performed in the very low-frequency region, the amplitude
p0 = − V0 ,
2kBT 共1/␭ ␤兲sinh共␤d/2兲 + i共␻d/2D兲cosh共␤d/2兲
2
of the signal applied to the sample has to be small with
113703-3 Barbero, Alexe-Ionescu, and Lelidis J. Appl. Phys. 98, 113703 共2005兲

respect to kBT / q. On the contrary, for large ␻ this condition

is substituted by a new one, which depends on the frequency.

III. CASE IN WHICH THE IONS HAVE DIFFERENT

DIFFUSION COEFFICIENT: DP Å DM
The analysis presented previously can be generalized to
the situation in which the mobilities of the positive ions, D p,
is different from that of the negative ions, Dm. In this case, jr
are given by

j r = − Dr 冉 ⳵ nr
±
q ⳵V
nr
⳵ z k BT ⳵ z
. 冊 共12兲

The fundamental equations of the problem are still Eq. 共1兲

and Eq. 共2兲, which have to be solved with the boundary
conditions 共4兲 and 共5兲. By assuming again that ␦nr共z , t兲 Ⰶ N,
the profiles of the bulk densities of ions and of the electric
potential are given by
␩ p共z兲 = C1 sinh共␥1z兲 + C2 sinh共␥2z兲,
␩m共z兲 = k1C1 sinh共␥1z兲 + k2C2 sinh共␥2z兲,
FIG. 2. The same as in Fig. 1, when q = 1.6⫻ 10−19 A s 共monovalent ions兲,
N = 4.2⫻ 1020 m−3, ␧ = 6.7⫻ ␧0, T = 300 K, d = 25 ␮m, S = 2 ⫻ 10−4 m2, and
D p = 10Dm. The dotted line corresponds to the case D p = Dm = 8.2
⫻ 10−12 m2 / s. The continuous line correspond to the case D p = 10Dm = 8.2
⫻ 10−11 m2 / s. The logarithms are in base 10 and ␻ is in rad/ s.
共13兲

冉 冊
d␴r共±d/2,t兲 1
q 1 − k1 1 − k2 = knr共±d/2,t兲 − ␴r共±d/2,t兲, 共18兲
␾共z兲 = − C1 sinh共␥1z兲 + C2 sinh共␥2z兲 dt ␶
␧ ␥21 ␥22
where k and ␶ are the adsorption coefficient and the desorp-
+ Az, 共14兲 tion time, respectively.13 In this framework, the profiles of
respectively, as it has been shown in Ref. 10 The integration the ionic densities and of the electrical potential across the
constants C1 , C2, and A are given by the boundary condi- sample are still given by 共8兲, where p0 and c follow from the
tions 共4兲 and 共5兲. In 共13兲 and 共14兲 the parameters ␥1 , ␥2 , k1, boundary conditions 共5兲, 共17兲, and 共18兲, and are
and k2 are

␥1,2 = 冑冉 1 1 1
冊 冑冋 冉 1 1 1
冊册 2
1
+ 4 共15兲
冋 1+␤
k␶
1 + i␻␶
tanh 冉 冊册
␤d
2
p0 − i
Nq/共kBT兲
共␻/D␤兲cosh共␤d/2兲
c = 0,
2 + 2 ± 2 − 2
2 ᐉ p ᐉm 2 ᐉ p ᐉm
冉 冊
4␭
q ␤d d
and −2 sinh p0 + c = V0/2, 共19兲
␧␤2

冉 冊
1
k1,2 = − 2␭2 ␥1,2
2
− , 共16兲
ᐉ2p
where ᐉ␣2 = 2␭2 / 关1 + 2i共␻ / D␣兲␭2兴. By imposing again the
condition 2兩␩r共z兲兩 Ⰶ N, we obtain the critical voltage U such
that the system under investigation can be considered linear
when V0 Ⰶ U. The results of the numerical calculations are
reported in Fig. 2, by assuming for the physical parameters
of the sample again the ones reported above, and D p
= 10Dm = 8.2⫻ 10−11 m2 / s.11 Again, in the low-frequency
range U → VT, whereas for large ␻ , U increases as 冑␻,
according to 共11兲, with D p or Dm. Hence, the critical voltage
depends on the larger diffusion coefficient.
IV. CASE IN WHICH THE SURFACES ADSORB IONS
Finally, we can consider the case in which at the elec-
trodes there is adsorption of particles. In this case, by indi-
cating with ␴r共±d / 2 , t兲, the surface density of adsorbed par-
ticles, instead of the boundary conditions 共4兲 we have now,
d␴r共±d/2,t兲
jr共±d/2,t兲 = . 共17兲
dt
In the Langmuir’s approximation,
2 2
where N is now the bulk density of ions in the absence of an
external electrical voltage, as discussed in Ref 12. By impos-
ing, as in the previous cases, the condition 2兩␩r共z兲兩 Ⰶ N, we
obtain the critical voltage U such that the system under in-
vestigation can be considered linear when V0 Ⰶ U. The result
of the numerical calculations is reported in Fig. 3, by assum-
ing for the physical parameters of the sample again the ones
reported above, with D p = Dm = 8.2⫻ 10−12 m2 / s,9 k
= 10−6 m s−1 and ␶ = 0.1.13 As expected in the low-frequency
range U → VT. In the high-frequency range, U is still given
by 共11兲. The maximum deviation of the present U with re-
spect to the one relevant to the case in which the adsorption
is absent takes place for a frequency ␻ ⬃ 1 / ␶.
V. DISCUSSION
The important point in the analysis presented previously
is that in the problem under investigation there is an intrinsic
voltage U. For V0 Ⰶ U the system behaves in a linear manner,
and the concept of electrical impedance is useful. When the
condition V0 Ⰶ U is not satisfied, the system behave in a non-
linear way. This means that if V共±d / 2 , t兲 = ± 共V0 / 2兲exp共i␻t兲,
113703-4 Barbero, Alexe-Ionescu, and Lelidis J. Appl. Phys. 98, 113703 共2005兲

FIG. 3. The same as in Fig. 1, when the adsorption phenomenon is taken
into account, in the Langmuir’s approximation, by assuming q = 1.6
⫻ 10−19 A s 共monovalent ions兲, N = 4.2⫻ 1020 m−3, ␧ = 6.7⫻ ␧0, T = 300 K,
d = 25 ␮m, S = 2 ⫻ 10−4 m2, and D p = Dm = 8.2⫻ 10−12 m2 / s. In the low- and
high-frequency ranges the critical voltage U is still given by 共11兲. The maxi-
mum influence of the adsorption effect on U is localized around the fre-
quency ␻ ⬃ 1 / ␶. The continuous curve corresponds to the case where k
= 10−6 m s−1 and ␶ = 0.1 s, whereas the dotted line corresponds to the case in
which the adsorption phenomenon is absent. The logarithms are in base 10
and ␻ is in rad/ s.
␦nr共z , t兲, and V共z , t兲 are not harmonic functions with fre-
quency f = ␻ / 共2␲兲. In this case ␦nr共z , t兲 and V共z , t兲 are of the
type
q
␾2⬙共z兲 = 关␩ p,2共z兲 − ␩m,2共z兲兴, 共23兲
␧
with the boundary conditions at z = ± d / 2
␾2 = 0, ␩r,2 ± 共qN/kBT兲关␾2⬘ + 共␩r,1/N兲␾1⬘兴 = 0. 共24兲
In the same manner it is possible to obtain the bulk differen-
tial equations and the boundary conditions for all h. By
means of ␩r,h共z兲 and ␾h共z兲 it is possible to evaluate the elec-
trical current in the external circuit I共t兲, and hence to deter-
mine the electrical impedance of the cell defined by Z
= V0 exp共i␻t兲 / I共t兲. Since I共t兲 is not a harmonic function of t
with frequency f = ␻ / 共2␲兲, it follows that Z = Z共t兲. At this
point, knowing the accuracy of the used experimental setup,
it is possible to determine the maximum amplitude, VM , for
which the system under investigation does not behave in a
nonlinear manner. It is clear that this procedure is, in some
way, arbitrary, since it depends on the experimental setup.
On the contrary, our analysis, based on the condition ␦nr
Ⰶ N, proves the existence of an intrinsic voltage with which
it is necessary to compare the amplitude of the applied volt-
age.
In the discussion presented previously, we have consid-
ered, for simplicity, the case in which the diffusion coeffi-
cients of the positive and negative ions are the same, and the
adsorption phenomenon is absent. However, a simple inspec-
tion shows that it is valid also in the case where D p ⫽ Dm and
when the adsorption phenomenon is present.

␦nr共z,t兲 = 兺 ␩r,h共z兲exp共ih␻t兲,
h
V共z,t兲 = 兺 ␾h共z兲exp共ih␻t兲,
h
where ␩r,1共z兲 = ␩共z兲 and ␾1共z兲 = ␾共z兲, introduced previously.
It was possible to analyze the problem in a more techni-
cal way, according to the following scheme. By substituting
expansions 共20兲 into 共1兲 and 共2兲 and into the boundary con-
ditions 共4兲 and 共5兲 one obtains a cascade of differential equa-
tions and the relevant boundary conditions for ␩r,h共z兲 and
␾h共z兲. For h = 1 the equations are
VI. CONCLUSION
共20兲 By means of a continuum model we have analyzed what
means “small voltage” in the impedance spectroscopy mea-
surements on an electrolytic cell. Actually our analysis is of
some interest for all systems in which the ionic contribution
to the total current is not negligible, as the liquid crystals.
We have shown that, in the low-frequency range, the
applied voltage has to be small with respect to the thermal
voltage VT = kBT / q. In the high-frequency region, the applied
voltage has to be small with respect to the critical voltage

冉 冊
diverging as the square root of the frequency. Our analysis
qN
⬙ 共z兲 ±
i␻␩r,1共z兲 = D ␩r,1 ␾⬙共z兲 , has been performed by considering the case in which the
k BT 1 D p = Dm = D, as well as the case in which D p ⫽ Dm. We have
also considered the case in which the limiting surface of the
q liquid can absorb ions. The analysis presented in our paper is
␾1⬙共z兲 = − 关␩ p,1共z兲 − ␩m,1共z兲兴, 共21兲
␧ simple, however, it was not previously presented. The results
are rather important from an experimental point of view. In
where, as before, for r = p there is the sign +, and for r = m the
fact, as it has been stressed previously in this paper, the
sign −. Equations 共21兲 have to be solved with the boundary
impedance spectroscopy technique is meaningful only if it is
conditions at z = ± d / 2,
possible to define the electrical impedance of the cell under
␾1 = ± V0/2, ⬘ ± 共qN/kBT兲␾1 = 0,
␩r,1 共22兲 investigation. This implies that the system behaves as a lin-
ear system. If this condition is not satisfied, the analysis of
as we have done in Sec. II. For h = 2 one gets the experimental data is more complex, and it has to be done,

⬙ 共z兲 ±
2i␻␩r,2共z兲 = D ␩r,2 冉 qN
␾⬙共z兲
k BT 2
冊 taking into account the nonlinear character of the system
under investigation. In recent papers devoted to the imped-
ance spectroscopy analysis of different systems the applied
Dq voltage to the cell cannot be considered small in the fre-
± 关␩r,1共z兲␾1⬘共z兲兴,
k BT quency range where the measurements are performed.11,14–16
113703-5 Barbero, Alexe-Ionescu, and Lelidis J. Appl. Phys. 98, 113703 共2005兲

7
Consequently, the systems cannot be considered as linear J. Israelachvili, Intermolecular Forces 共Academic, London, 1985兲, Chap.
systems, and some of the conclusions derived in Refs. 11 and 12.
8
G. Barbero and A. L. Alexe-Ionescu, Liq. Cryst. 32, 943 共2005兲.
14–16 could be not correct. 9
A. Sawada, K. Tarumi, and S. Naemura, Jpn. J. Appl. Phys., Part 1 38,
1
1418 共1999兲.
J. Ross Macdonald, Impedance Spectroscopy 共Wiley, New York, 1987兲, 10
I. Lelidis and G. Barbero, Phys. Lett. A 343, 440 共2005兲.
Chap.1. 11
A. Sawada, K. Tarumi, and S. Naemura, Jpn. J. Appl. Phys., Part 1 38,
2
J. Jamnik and J. Maier, J. Electrochem. Soc. 146, 4183 共1999兲.
3 1423 共1999兲.
P. A. Cirkel, J. P. M. van der Ploeg, and G. J. M. Koper, Physica A 235, 12
G. Barbero, Phys. Rev. E 71, 062201 共2005兲.
269 共1997兲. 13
4
M. Scott, R. Paul, and K. V. I. S. Kalert, J. Colloid Interface Sci. 230, 377 B. Maximus, E. De Ley, A. De Meyere, and H. Pauwels, Ferroelectrics
共2000兲. 121, 103 共1991兲.
14
5
M. Scott, R. Paul, and K. V. I. S. Kalert, J. Colloid Interface Sci. 230, 388 A. D. Hollingsworth and D. A. Saville, J. Colloid Interface Sci. 257, 65
共2000兲. 共2003兲.
15
6
M. Z. Bazant, K. Thornton, and A. Ajadari, Phys. Rev. E 70, 021506 S. Murakami and H. Naito, Jpn. J. Appl. Phys., Part 1 36, 773 共1997兲.
16
共2004兲. S. Murakami, H. Iga, and H. Naito, J. Appl. Phys. 80, 6396 共1996兲.