www.elsevier.

nl/locate/jelechem

Journal of Electroanalytical Chemistry 481 (2000) 222 – 226

On the ideal polarisability of electrodes displaying cpe-type

capacitance dispersion
A. Sadkowski *
Institute of Physical Chemistry of the Polish Academy of Sciences, Kasprzaka 44 /52, 01 -224 Warsaw, Poland

Received 4 January 1999; received in revised form 10 November 1999; accepted 26 November 1999

Abstract

The relevance of the concept of ideal polarisability to solid electrodes displaying constant phase angle behaviour (constant phase
element (cpe)) is reconsidered in the context of the recent dispute in this Journal [Zoltowski, 443 (1998) 149; Lang and Heusler,
457 (1998) 257]. It is shown that the cpe as a model of the electrochemical interface is ‘unphysical’ due, first of all to
unattainability of the steady state after potential or current perturbation. It is also concluded that the thermodynamic equations
governing the equilibria of an ideally polarisable electrode cannot be applied to an electrode modelled by cpe because the notions
of the charge and double layer capacitance (Cd) are irrelevant in the framework of the cpe model. As a consequence, procedures
of calculation of Cd from impedance data modelled by cpe are unjustified. © 2000 Elsevier Science S.A. All rights reserved.

Keywords: Electrochemical impedance; Solid electrodes; Double layer capacitance; Ideal polarisability; Capacitance dispersion; Constant phase
element

1. Introduction ability and interfacial capacitance of electrodes. The

question concerns the relevance of thermodynamics to
Recently Lang and Heusler [1] have discussed the solid electrodes with their impedance notoriously dis-
energetics of electrochemical interfaces exhibiting con- playing the ‘frequency dispersion’ of the interfacial
stant phase angle behaviour (so-called frequency disper- capacitance even in the absence of the faradaic reac-
sion of the interfacial capacitance modelled by constant tions, i.e. without charge being transferred through the
phase element (cpe)). These authors have contested the interface. This kind of frequency dispersion is charac-
statement of Zoltowski [2], based on purely formal terised by a harmonic response (impedance, admittance
considerations, that the cpe behaviour is in conflict with or, collectively [4], immittance) that exhibits a phase
the condition of the ideal polarisability. Lang and angle different from 90°, the value characteristic for the
Heusler invoked [1] the descriptive part of the definition capacitance. The most spectacular fact is the non-in-
of ideal polarisability recommended by IUPAC [3] teger power dependence of the interfacial immittance
(…no charged component is common to both phases on frequency, which is often displayed over a broad
adjoining the interphase) without reference to its math- frequency range and which suggests a unique physical
ematical consequences, i.e. the electrocapillary and origin for cpe behaviour [5].
Lippmann equations. On the other hand, the approach It was stated in Ref. [1] that ‘‘an electrode can be an
presented in Ref. [2] was based on purely formal con- ‘ideally polarisable electrode’, defined by the impossibil-
siderations of electrochemical immittance data, and ity of charge transfer through the interface, even if its
therefore the scope of the ideal polarisability concept in impedance spectrum measured in a certain frequency
Refs. [1,2] seems to be slightly different. Nevertheless, range can be described only by a model including a
the controversy raises the important question of proper double layer impedance of cpe character’’. With this,
definition and clear understanding of the ideal polaris- the authors suggest that the experimentally established
cpe-type response has to be valid only in a limited, even
* Fax: + 48-22-6325276. if fairly broad, frequency range, and outside of this
E-mail address: [email protected] (A. Sadkowski) range the electrode without charge transfer through the

0022-0728/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 0 2 2 - 0 7 2 8 ( 9 9 ) 0 0 4 8 0 - 5
 A. Sadkowski / Journal of Electroanalytical Chemistry 481 (2000) 222–226 223

interface has to behave properly, i.e. its frequency Q(t)=DEY0t 1 − n/G(1 − n) (5)
response has to be purely capacitive.
Analogously the energy response (electric work sup-
plied) is:
2. Physically unrealizable cpe A(t)= DE 2Y0t 1 − n/G(1 − n) (6)

The arguments of Lang and Heusler [1] and also It is evident that for 0B nB1 both the charge (Eq. (5))
earlier of Zoltowski [2] were based on the consideration and the energy (Eq. (6)) diverge to infinity, and this
of the ‘energetics’ of the charging/discharging of the divergence suffices to consider the cpe as ‘physically
interface and invoked the concept of energy dissipation unrealisable’ at long times or low frequencies. A similar
in the case of cpe-type dispersion. The authors of Ref. derivation for current-step perturbations has been pub-
[1] have noted that, aside from the charge transfer lished already [7] without, however, regarding it as
through the interface, the energy dissipation can also evidence of ‘unphysical’ behaviour of cpe. The first
originate in other irreversible processes not commonly explicit assertion of this unphysical character can possi-
accounted for in electrochemistry, such as non-elastic bly be attributed to Macdonald (cf. Ref. [4] p. 91). Any
deformation etc. However, apart from energetics, the additional arguments about inconsistency of cpe mod-
cpe model used in electrochemistry also raises a prob- elling of the real electrodes are redundant as they are
lem with the charge balance, and this is possibly even mere consequences of this non-steady state feature of
more important in electrochemistry and easier to verify the cpe immittance. This obviously refers also to the
experimentally. It can be shown by simple calculation statement in Ref. [2] of the cpe violating the condition
of the Laplace-domain responses [6] that electrodes of of ideal polarisability based on the fitting of various
cpe-type require infinite charge and infinite energy to be impedance models to the ‘synthetic’ data presenting cpe
transferred from one steady state to another. In this capacitance dispersion.
sense, cpe-type electrodes represent intrinsically non-
steady state systems and are ‘physically unrealisable’ [4]
in the same way as the Warburg impedance for infinite 3. Ideal polarisability and capacitance versus cpe
diffusion. In view of the importance of this assertion dilemma
and of its being evidently overlooked in the cited litera-
ture, we recall the following formulae. Generally, there are two sources of the current at the
The admittance of the cpe in the Laplace (complex electrochemical interface [8] as a response to a potential
frequency) domain is: pulse: (1) transfer of the charge carriers through the
interface and (2) displacement of the charges near the
Ycpe =Y0s n (1)
interface. The first, referred to as the faradaic process,
where s is the complex angular frequency or variable in is characterised by the inseparable linkage of charge
the Laplace domain; in the frequency domain it is transfer to the transport of components common to
simply iv, where i = − 1, v is angular frequency = both phases (reactants). Hence its exclusion, according
2pf; f is the frequency in Hz; Y0 is the pre-exponential to IUPAC recommendation, requires that no compo-
parameter, the analogue of the capacitance for the ideal nent be common for both phases. The second current
(n= 1.0) case; n is the exponent, usually fractional, source consists of the ‘polarisation’ or non-faradaic
characteristic for the cpe behaviour. The deviation of n process, and it is not necessarily linked to the transport
from 1.0 is a measure of the departure from ideal of reagents. It is generally acknowledged that charge
capacitance behaviour. displacements are very fast (electrons in metals, ions
The current response for the potential perturbation and solvation shells in the electrolyte), their characteris-
(still in the Laplace domain) is: tic frequencies exceeding the frequency limit of typical
electrochemical experiments by several orders of magni-
I(s) =E(s)Y(s) (2) tude (B 1 MHz). Therefore, on the typical timescale of
For a potential step E(s) =DE/s this is: electrochemical experiments, the displacement processes
may be considered reversible (dissipation-less), and the
I(s) =DEY0s n − 1 (3) only relaxation may be due to external factors such as
unavoidable impeding of the displacement current by
In the time domain the current is:
transport in adjacent phases, its simplest manifestation
I(t) =DEY0t − n/G(1 −n) (4) being the electrolyte (ohmic) resistance RV.
The thermodynamics of the ideally polarisable inter-
where G is Euler’s gamma function. face is epitomised by the Gibbs adsorption isotherm:
Integration of the current gives the charge response
to the potential pulse: − dg= s M dE +Gi dmi (7)
224 A. Sadkowski / Journal of Electroanalytical Chemistry 481 (2000) 222–226

with its immediate consequence, the Lippmann equa- dispersion does not exist or, at least, it is not so
tion [8–11]: pronounced as for solid electrodes. Cpe-type behaviour
was almost exclusively attributed to the roughness of
− (#g/#E)T,m =s M (8)
the solid surfaces [15], and it was believed that in the
where g is the superficial work [12], E the electrode absence of faradaic reactions minimisation of the
potential, Gi the surface excess of species i and mi its roughness should suffice to restore the capacitive be-
chemical potential. The charge of the metal s M repre- haviour of solid electrodes. The emphasis on the ‘chem-
sents the ‘charge displacement’ or polarisation of the ical’ aspects of ideal polarisability in Ref. [3] (…no
interface and for the whole system to be electrically common components, no charge transfer reaction etc.)
neutral, it has to be compensated by the charge s S of can possibly be attributed to its descent from Gra-
the electrolyte: hame’s definition [16,17], which in electrochemistry
overshadowed the earlier and more physical approach
sM+sS=0 (9)
such as that credited to Planck [18] (state of the elec-
For solid electrodes, an additional term is included in trode determined by the charge supplied) who intro-
the Gibbs adsorption equation (7) to account for the duced the notion of ‘perfectly polarisable electrodes’’
contribution of elastic strain to the superficial work (6ollkommen polariesierbaren Elektroden) adopted later
[3,12]. As a consequence, an additional term represent- also by Frumkin and his co-workers in their ‘thermody-
ing the charge displacement related to the reversible namic theory of hydrogen and oxygen adsorbing elec-
mechanical work appears in the Lippmann equation [3]. trodes’ [19–21]. A review of this, now almost forgotten,
This term is, however, either irrelevant for liquid elec- history can be found e.g. in Frumkin’s monograph [22].
trodes, which by their very nature are not subjected to The definition of the ideally polarisable electrode by
internal stresses or, as was recently claimed authorita- Grahame [16,17] as one for which electrode reactions
tively, it is negligible in relation to s M for solid elec- are absent (‘no common charged components…’), al-
trodes [13,14]. It has to be noted here that inclusion of though focused on chemical aspects of the interface, is
the additional ‘irreversible charge’ related to the devia- equivalent to Planck’s definition [18] if it is assumed
tion from mechanical equilibrium between the inter- that without charge transfer across the interface, it can
phase and the bulk of the solid (as was suggested in be charged/discharged only reversibly as was stated
Ref. [30]) has to be considered unjustified on the basis explicitly: ‘Experimentally, the ideal polarized electrode
that it violates the strictly equilibrium character of the is an electrode which behaves like an electric condenser
Gibbs adsorption isotherm and Lippmann equation. without leakage’ [17]. This assumption is also implicit
The only charge variable relevant to the definition of in the IUPAC recommendation [3]. In most textbooks
ideal polarisability is the charge related to the reversible dealing with the ideal polarised electrode, it is defined
displacements of components of the double layer re- as an interphase behaving reversibly according to the
gion, and the definition of the ideal polarisability has to Gibbs and Lippmann equations (see e.g. part I, chapter
be seen inseparably from the Lippmann equation ac- II [11], chapter 16, p. 217 [23], p. 18 [24] or chapter 3.4
cording to the statement in Ref. [3] (equation 4.3.8 ): [25]), and therefore these equations have to be consid-
‘the Lippmann equation leads to a unique charge for a ered as inseparable attributes of ideal polarisability.
ideally polarisable electrode’. They lead to a unique charge–potential relation and
The charge defined by the Lippmann equation is a consequently define the capacitance, which is the
function of, among other variables, the electrode derivative of such a relation. In Bard and Faulkner’s
potential textbook [26] referred to in Ref. [1] the definition of the
ideal polarized electrode (one on which ‘no charge
s M = s M(T, mi, E) (10)
transfer across the metal solution interface can oc-
and hence the interfacial capacitance defined as cur…’) is also followed immediately (p. 7) by the state-
ment that on the ideal polarized electrode ‘…the
Cd = #s M/#E (11)
behaviour of the electrode solution interface is
is a thermodynamic variable inherent in the concept of analogous to that of a capacitor.’
ideal polarisability in the same way as the charge itself. No unique charge as a function of the electrode
At the time of formulation of the IUPAC recommen- potential can be attributed to an electrochemical inter-
dation [3], the problem of the frequency dispersion of face conforming to cpe characteristics. Therefore, such
the interfacial capacitance of solid electrodes was not an interface has to be considered as not conforming to
fully appreciated, and it was often considered an exper- the postulate of ideal polarisability. Consequently, the
imental error. This is understandable in view of the calculation of Cd from cpe impedance as proposed e.g.
overwhelming majority of experiments on double layer by Brugg et al. (equation (5) in Ref. [27]) and applied,
thermodynamics having been done on liquid mercury e.g. Ref. [28] or [29] has to be considered as an un-
electrodes for which the problem of the capacitance justified procedure. The criticism of this procedure ex-
 A. Sadkowski / Journal of Electroanalytical Chemistry 481 (2000) 222–226 225

pressed also in Ref. [30] p. 177 is therefore entirely thors, e.g. [33,37,38] (the lower the conductivity, the
appropriate, as is also the opinion presented recently by lower the cpe exponent, i.e. higher deviation from ideal
Kerner and Pajkossy [31] that: ‘... any attempt to capacitive behaviour) demonstrate the role of the cou-
determine exact double layer capacitances and to ex- pling between the interfacial capacitance and the elec-
tract quantities of thermodynamic significance… can- trolyte resistance. These experimental facts prompt us
not be successful’. The inconsistency inherent in the to believe that the origin of the cpe behaviour of solid
procedure in question is apparent as it involves the electrodes is the surface non-uniformity on the meso-
whole ohmic resistance RV in the formula for Cd, (see scopic scale. The non-uniformity can be brought about,
equation (5) in Ref. [27]). The electrolyte resistance RV e.g. by structuring of the adsorbate layers or of the
is the external parameter with respect to the interface, interfacial solvent [33,34] with different local charge
and it can be varied freely by the experimenter by densities and capacities. The term ‘mesoscopic scale’
changing the position of the reference electrode probe means that the size of the distinct structures formed on
without any relationship with the immittance of the the surface exceeds greatly that of the elementary unit
interface itself. (micro-scale) and at the same time it is definitely below
As stated already, an electrode with cpe characteris- the dimension of the electrode (macro-scale). The meso-
tics has to be considered as ‘unphysical’ for more scale structuring of the interface explains the coupling
fundamental reasons than the mere dissipative nature of the local capacities (Ci ) and adjoining resistive paths
of its immittance at low frequency. The problem dis- (Ri ) inseparable in macro-scale experiments, i.e. those
cussed here and in Refs. [1,2] might therefore be consid- involving the whole electrode without local specificity.
ered as purely academic if it were not for the fact that For this coupling to occur, the sizes of the structured
the experiments show that solid electrodes notoriously regions of different charge density have to be compara-
display this ‘unphysical’ cpe behaviour over a broad ble with the thickness of the electrolyte layer contribut-
frequency range. The present author’s experience [32– ing to the ohmic resistance, which in practice is the
34] indicates that cpe-type dispersion, observed also on distance from the reference electrode probe to the inter-
face. The coupling makes the averaging of the capaci-
well-defined monocrystalline surfaces, is discernible in
tance alone over the interface unfeasible, because such
the whole experimentally accessible frequency range,
averaging inevitably incorporates parts of local
i.e. down to 10 − 2 Hz. Lower frequencies are hardly
resistances.
attainable on ‘polarisable’ electrodes due to decreasing
Cpe behaviour may therefore be simply the manifes-
signal-to-noise ratios because the capacitive current
tation of an inherent non-uniformity of solid surfaces
(signal) decreases roughly proportionally to decreasing
and of our inability to measure the true local capacities
frequency in contrast to other components of the mea-
without the intervention of local resistivities. Even if
sured current (noise). To the author’s knowledge, it was
locally the interface is ideally polarisable (and most
also not reported in the literature that cpe-type capaci- probably it is) and is featured by local charge densities
tance dispersion is observed only in certain frequency and local capacities governed (locally!) by the Lipp-
ranges limited by lower and higher cut-offs beyond mann equation, the condition of ideal polarisability is
which the normal behaviour is restored. Rather the broken globally, i.e. for the electrode as a whole. This is
opposite was reported, such as deviation from the due to the impossibility of averaging charges and ca-
capacitive behaviour increasing at low frequencies (e.g. pacities distributed over the surface without involving
figures 5 and 7 in Ref. [5] or figure 1 in Ref. [31]). of local contributions of electrolyte resistance. This
interpretation is in fact equivalent to the uneven charg-
ing of the interface already postulated by Scheider [37].
4. On the nature of the cpe behaviour on ‘smooth’ The spectacular differences between almost perfectly
electrodes capacitive Au(111) in the broad potential range in
HClO4 solutions and Au(210) displaying under the
The crucial problem remains of the origin of ‘un- same conditions considerable cpe-type dispersion [27]
physical’ cpe behaviour of solid electrodes, especially has to be related to the almost perfect isotropy of the
those with well-defined surface structure and negligible first and considerable anisotropy of the latter. Direc-
roughness [33,34]. The interpretation accepted for tions of the steps and edges abound on the (210) face
rough electrodes [35,36] based on self-similar (fractal) and determine the lines of non-uniform charge density
geometry with the higher cut-off length of the order of (one-dimensional structures). It was shown recently
macroscopic dimensions, is evidently inappropriate for that even randomly distributed one- or two-dimen-
macroscopically smooth and homogeneous surfaces [5]. sional structures may give rise to scale invariance typi-
It was shown [32–34] that crystal structure, electrolyte cal for fractal structures [42]. Hence, even the surface
composition and electrode potential all influence the structured in this simple one-dimensional way may
cpe exponent of ‘smooth’ electrodes. Distinct effects of respond in accordance with fractional power laws typi-
the electrolyte conductivity reported by numerous au- cal for cpe.
226 A. Sadkowski / Journal of Electroanalytical Chemistry 481 (2000) 222–226

It can be concluded that deviations of solid elec- [6] A. Sadkowski, Electrochim. Acta 38 (1993) 2051.
trodes from simple thermodynamic and kinetic laws [7] P.H. Bottelberghs, G.H.J. Broers, J. Electroanal. Chem. 67
(1976) 155.
involving lumped parameters and kinetics governed by [8] R. Parsons, in: C.W. Tobias, P. Delahay (Eds.), Advances in
ordinary differential equations still have no unquestion- Electrochemistry and Electrochemical Engineering, vol. 7, Inter-
able and universal explanation. This, however, should science, New York, 1970, p. 177.
not be excessively frustrating for electrochemists, be- [9] R. Parsons, in: J.O.M. Bockris, B.E. Conway (Eds.), Modern
cause the situation is similar in the physics of dielectric Aspects of Electrochemistry, vol. 1, Butterworths, London, 1954.
[10] R. Parsons, in: J.O.M. Bockris, R.E. White, B.E. Conway, E.
relaxation [4,39–42] and other types of relaxation pro- Yeager (Eds.), Comprehensive Treatise of Electrochemistry, vol.
cesses in non-uniform systems that are often governed 1, Plenum, New York, 1980, p. 1.
by ‘fractional power laws’. Numerous interpretations [11] P. Delahay, Double Layer and Electrode Kinetics, Wiley–Inter-
have been put forward, but not one of them seems to be science, New York, 1965.
generally accepted as universally explaining the real [12] R.G. Linford, Chem. Rev. 78 (1978) 81.
[13] J. Lipkowski, W. Schmickler, D.M. Kolb, R. Parsons, J. Elec-
nature of the low frequency dispersion and/or frac- troanal. Chem. 452 (1998) 193.
tional power law. The cpe behaviour of (non-ideally!) [14] R. Guidelli, J. Electroanal. Chem. 453 (1998) 69.
polarisable electrodes is only one of many instances of [15] R. de Levie, in: P. Delahay, C. Tobias (Eds.), Advances in
‘universal relaxation laws’ featuring fractional powers Electrochemistry and Electrochemical Engineering, vol. 6, Wiley,
of frequency or time established in relaxation experi- New York, 1967.
[16] D.C. Grahame, R.W. Whitney, J. Am. Chem. Soc. 64 (1942)
ments in the frequency or time domain, respectively. 1548.
[17] D.C. Grahame, Chem. Rev. 41 (1947) 441.
[18] M. Planck, Ann. Phys. 44 (1891) 385.
5. Conclusions [19] A.N. Frumkin, O.A. Petrii, Electrochim. Acta 15 (1970) 391.
[20] B.B. Damaskin, O.A. Petrii, Introduction to Electrochemical
Kinetics, Vysshaya Shkola (High School), Moscow, 1975.
Electrodes whose immittance spectra conform to the [21] B.B. Damaskin, O.A. Petry, Principles of Theoretical Electro-
cpe model do not comply with the condition of ideal chemistry (in Russian), Vysshaya Shkola (High School),
polarisability. This is because they cannot be assigned a Moscow, 1978.
unique charge, a thermodynamic variable inherent to [22] A.N. Frumkin, Zero Charge Potentials (in Russian), Nauka,
Moscow, 1979.
the ideal polarisability, represented mathematically by
[23] W. Schmickler, Interfacial Electrochemistry, Oxford University
the Lippmann equation. Press, New York, 1996.
The cpe, although it often represents a very good [24] J. Goodisman, Electrochemistry: Theoretical Foundations, Wi-
model of the interface immittance for solid electrodes, ley, New York, 1987.
is ‘physically unrealisable’ because it implies that infi- [25] J. Koryta, J. Dvorak, V. Bohackova, Lehrbuch der Elektro-
chemie, Springer Verlag, Wien, 1975.
nite charge and infinite energy are required to attain the
[26] A.J. Bard, L.R. Faulkner, Electrochemical Methods, Wiley, New
new steady-state after potential perturbation. York, 1980.
The universality of cpe as a model of solid electrodes [27] G.J. Brug, A.L.G. van den Eeden, M. Sluyters-Rehbach, J.
results most probably from meso-scale structures on the Sluyters, J. Electroanal. Chem. 176 (1984) 275.
surface with diverse charge densities. The structuring, [28] W.R. Fawcett, M. Fedurco, Z. Kovacova, Z. Borkowska, J.
Electroanal. Chem. 368 (1994) 275.
whatever its physical origin, results in experimentally
[29] B.L. Kuzin, D.I. Bronin, Russ. J. Electrochem. 33 (1997) 572.
inseparable coupling of diverse, local capacities with [30] G. Lang, K.E. Heusler, J. Electroanal. Chem. 391 (1995) 169.
local electrolyte resistances. As a consequence, the non- [31] Z. Kerner, T. Pajkossy, J. Electroanal. Chem. 448 (1998) 139.
uniform electrode, even if ideally polarisable in a local [32] A. Motheo, J.R. Santos, A. Sadkowski, A. Hamelin, J. Elec-
sense, is seen globally as non-ideal, breaking the ther- troanal. Chem. 397 (1995) 331.
[33] A.J. Motheo, A. Sadkowski, R.S. Neves, J. Electroanal. Chem.
modynamic condition of ideal polarisability. Any at-
430 (1997) 253.
tempts to calculate (global) charge densities and double [34] A. Sadkowski, A.J. Motheo, R.S. Neves, J. Electroanal. Chem.
layer capacities inevitably fail for such electrodes due to 455 (1998) 105.
their failure to comply with the Lippmann equation. [35] L. Nyikos, T. Pajkossy, Electrochim. Acta 30 (1985) 1533.
[36] R. de Levie, J. Electroanal. Chem. 281 (1990) 1.
[37] W. Scheider, J. Phys. Chem. 79 (1975) 127.
[38] E.D. Bidoia, L.O.S. Bulhoes, R.C. Rocha-Filho, Electrochim.
References Acta 39 (1993) 763.
[39] A.K. Jonscher, Nature 267 (1977) 673.
[1] G. Lang, K.E. Heusler, J. Electroanal. Chem. 457 (1998) 257. [40] A.K. Jonscher, Dielectric Relaxation in Solids, Chelsea Dielec-
[2] P. Zoltowski, J. Electroanal. Chem. 443 (1998) 149. tric Press, London, 1983.
[3] S. Trasatti, R. Parsons, Pure Appl. Chem. 58 (1986) 437. [41] A.J. Jonscher, Universal Relaxation Law, Chelsea Dielectric
[4] J.R. Macdonald, Impedance Spectroscopy Emphasizing Solid Press, London, 1996.
Materials and Systems, Wiley, New York, 1987. [42] O. Malcai, D.A. Lidar, O. Biham, D. Avnir, Phys. Rev. E 56
[5] T. Pajkossy, J. Electroanal. Chem. 364 (1994) 111. (1997) 2817.