Electrochimica Acta 139 (2014) 180189

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Inuence of the nature of the electrode material and process variables

on the kinetics of the chlorine evolution reaction. The case of
IrO2 -based electrocatalysts
Davide Rosestolato , Jacopo Fregoni, Sergio Ferro, Achille De Battisti
Dipartimento di Scienze Chimiche e Farmaceutiche, Universit degli Studi di Ferrara, via Fossato di Mortara 17, 44121, Ferrara, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Kinetic studies on the chlorine evolution reaction (ChlER) on oxide-based materials have been the sub-
Received 27 March 2014 ject of a number of papers since the seventies, following the introduction of DSAs (Dimensionally Stable
Received in revised form 1 July 2014 Anodes) in chlor-alkali plants. On the basis of experimental data, different pathways have been proposed
Accepted 2 July 2014
for the reaction over the years. Actually, specic experimental conditions and different approaches in
Available online 17 July 2014
sample preparation may lead to conicting explanations. In the present paper, the ChlER kinetics has
been studied at four electrode materials based on iridium and titanium oxides (with a 1:2 molar ratio).
Keywords:
Electrodes were synthetized at two temperatures (350 and 450 C) and by two different preparation
Chlorine Evolution Reaction
iridium oxide
methods: physical vapor deposition (rf-magnetron sputtering) and a conventional sol-gel technique,
adsorbed intermediates using special precursors developed in our laboratory. Both methodologies guarantee a high level of repro-
kinetics ducibility. As also observed by other authors, experimental data have shown a lack of linearity in Tafel
adsorption isotherm plots, high b slopes and reaction orders with respect to chloride 1, which have been justied on the basis
of a Volmer-Heyrovsky pathway, by considering a model proposed by Tilak and Conway in 1992. This
approach highlighted the role of the adsorbed intermediates, also at low overpotentials, for all electrode
materials. To analyze further the kinetics, Langmuir and Frumkin models for intermediates adsorption
were considered. Values for the lateral interaction parameter g were estimated, which ranged between
1 and 10, in all cases. Concerning the effect of pH, its inuence on the ChlER rate seems to be related only
with electrode surface modications, without any involvement of protons in the rate determining step of
the process. A slight inhibiting effect was assessed, by increasing the protons concentration. Eventually,
impedance spectroscopy analysis did not appear sensitive to intermediate adsorption, plausibly because
of the low variation of the coverage within the Tafel region; a poorly resolved contribution related to
porosity was found in the case of samples prepared at 350 C.
2014 Elsevier Ltd. All rights reserved.

1. Introduction research, mainly addressed to DSA-coating synthesis and charac-

The introduction of mixed-oxide lm-electrodes (the so-called
DSAs, Dimensionally Stable Anodes) in the chlor-alkali technology
has represented a fundamental progress in industrial electrochem-
istry, leading to enormous energy saving and cleaner processes,
through a continuous progress in cell and electrode engineering.
The impressive industrial impact of the original invention [15],
further enhanced by the extension of DSA applications to the oxy-
gen evolution reaction (relevant to aqueous electrometallurgy and
hydrogen production), has obviously prompted the fundamental
Corresponding author. Tel.: +39 0532 455166.
E-mail addresses: rssdvd@unife.it,
davide.rosestolato@gmail.com (D. Rosestolato).
terization.
In fact, papers have been published on the subject as early as
1972-1974, already quite well focused on preparation technique
and microstructural properties of RuO2 coatings [6,7]. The further
development of research has been more and more concerned with
the latter themes, while what could have been the main target, the
chlorine evolution reaction (ChlER), has been relatively less studied.
On the other hand, the rst of the two themes is quite rich and com-
plex. The methodology of preparation of active coatings, with the
wide variety of parameters involved (i.e., the quality and quantity
of precursors, the choice of the path leading from precursor mix-
ture to nal mixed-oxide coating), has attracted, and attracts so far,
a number of research works. In comparison, the study of a facile
electrochemical reaction, namely the electrochemical oxidation of
chlorides, has been considered a bit obvious.

http://dx.doi.org/10.1016/j.electacta.2014.07.037
0013-4686/ 2014 Elsevier Ltd. All rights reserved.
 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189
Nevertheless, in close temporal continuity with papers on the
kinetics of chloride discharge at graphite [8,9], the rst reports
and papers on chloride discharge [1012] at Ti-supported RuO2 -
IrO2 -TiO2 electrodes appeared essentially in conjunction with
rst fundamental papers on DSA preparation and characteri-
zation. The apparent simplicity of the reaction has favored
more or less straightforward application of mechanisms of typical
gas-phase reactions catalyzed by solid surfaces, like the Langmuir-
Hinshelwood and Eley-Rideal ones, to chloride discharge. Their
electrochemical equivalents, Volmer-Tafel and Volmer-Heyrovsky
mechanisms, had previously found application in the formally sim-
ilar case of hydrogen evolution reaction (HER), further justifying
such extension. However, the study of the literature of this period,
reviewed by Trasatti at the ending of the 80s [13], clearly shows all
the difculty of a mechanistic hypothesis for the reaction. A basic
problem is that, quite often, results have been obtained under dif-
ferent conditions: consequently, the lack of a sufciently rich and
homogenous experimental database makes generalizing attempts
quite difcult. A considerable number of papers deeply discussed
in ref. [13] have been devoted to ChlER from concentrated chloride
solutions (e.g. 5 M), under different acidities and sometimes at high
temperatures (around 60-80 C). In solutions with a so high ionic
strength, assuming solvation numbers of 6 for both, Na+ and Cl
[14], the number of free molecules of solvent is negligible. Even
though these features of the working solution resemble those of the
industrial chlor-alkali process, and because of that may supply use-
ful information for the latter, they denitely make the system more
complex than required by a classical study in electrochemical kinet-
ics. One more complication, unknown for the sister hydrogen
evolution reaction, is the electrode material. While in the case of
HER the electrode material is a metal, exposing polycrystalline sur-
face or a well-dened single-crystal face, for which reproducibility
of features is guaranteed, in the case of ChlER, the electrode mate-
rial is an oxide, or better an oxide mixture (for reasons of stability),
one of the components being RuO2 , or IrO2 , or a combination of the
two oxides. As well established by the literature, the electrochem-
ical performance of these electrodes is conditioned by a number of
variables hardly reproducible, sometimes even within one and the
same laboratory. Benchmark electrodes are not easily available
either.
From a strictly mechanistic point of view, the fact that
inevitably chloride discharge has to happen at metal-oxide,
oxygen-terminated surfaces (theoretical studies on single-crystal
faces of IrO2 and RuO2 have conrmed the existence of a surface
fully covered by oxygen under typical chlorine evolution condi-
tions, see e.g. refs. [15,16]), implies that the nature of adsorbed
reaction intermediate(s) and activated complexes have to reect
the chemistry of Ru-O, Ir-O, Cl-O and do not exclude a priori
also Ru-Cl and Ir-Cl chemistry. The not small number of reac-
tion mechanisms proposed, despite stoichiometric simplicity and
the substantial difculty to adapt mechanisms to experimental
kinetic evidence, is also related with this inescapable obstacle.
A recent paper by Over [17] has thoroughly discussed these
aspects, proposing unifying views of heterogeneous catalysis and
electrocatalysis.
In the present paper, a comparison has been made of the electro-
catalytic properties of electrode materials of the same composition:
33 mol% IrO2 - 66 mol% TiO2 , synthesized by both a sol-gel method
and Physical Vapor Deposition (PVD) (magnetron reactive sputter-
ing). The active mixed-oxide coatings were deposited on Ti metal
at two different temperatures: 350 and 450 C. The oversimpli-
cation of the electrodes composition, compared with industrial
benchmarks, has been chosen in order to try to isolate the contribu-
tion of components. The generalized higher stability of IrO2 -based
electrodes and consequent higher reproducibility of experimental
results has also been taken into consideration. On the basis of the
181
experimental acquired evidence, mechanistic considerations have
been developed.
2. Experimental
Titanium supports with a geometric surface area of 4 cm2 were
purchased from Titanium Consulting & Trading (Italy) and prelim-
inary electro-polished, as described in ref. [18]. Oxide lms were
grown by a sol-gel method (SG) [19] and through physical vapor
deposition (PVD). Electrode samples were synthesized at 350 and
450 C; further details on the preparation of samples have been
given in ref. [20]. In the case of sol-gel samples, 0.76 M Ir-HAC and
2 M Ti-HAC solutions were mixed in such a way to obtain an Ir/Ti
molar ratio close to 1:2. The precursor solution was deposited on
the Ti support by brushing; each precursor layer was rst dried
in oven at a temperature of 100 C for 5 minutes, then calcined at
350 or 450 C, for 10 minutes. The nal mixed-oxide loading (about
1 m of thickness) was attained with three successive precursor
depositions. Sputter depositions were performed in a custom-built
rf-magnetron sputtering system (13.56 MHz), equipped with three
independent 2 sources and an rf-biased sample holder. The base
pressure of the chamber was about 2 1010 bar; targets materials
were 2 diameter Ir (99.9%) and Ti (99.995%) metallic disks. Before
starting the depositions, sources were conditioned for 15 min. at
the operating conditions. Pure (99.9995%) Ar and a 20% mixture of
O2 in Ar gases were used at a working pressure of 6 106 bar.
Power applied at sources were of 25 and 300 W for Ir and Ti
targets, respectively; lms have been prepared with a thickness
of 1 m.
For the sake of brevity, throughout this paper, sol-gel samples
will be indicated as SG350 and SG450, while the acronyms PV350
and PV450 will be used in the case of sputtered specimens.
The surface morphology of the samples was investigated by
Atomic Force Microscopy (AFM), making use of a NT-MDT micro-
scope equipped with a NSG10 golden silicon tip. Analyses were
carried out in semi-contact mode; the tapping frequency was var-
ied from 0.8 to 0.9 Hz, depending on the sample analyzed, and was
adjusted before each test for optimizing the resolution of obtained
images.
The electrochemical in situ characterization of the samples
was carried out by the traditional cyclic voltammetric technique,
using an Autolab PGSTAT20 digital apparatus, equipped with a
SCANGEN module that allows the applications of linear sweeps.
A three-electrode cell with a Luggin capillary was used, and mea-
surements were carried out in different solutions in dependence
of the experiment. A double junction, saturated calomel elec-
trode (SCE), with an intermediate NaNO3 saturated solution, and
a large-area Pt mesh were chosen as reference and counter elec-
trodes, respectively. Experiments were carried out by varying the
scan rate from 10 to 1000 mV s1 (10, 25, 50, 100, 200, 300,
400, 500, 600, 700, 800, 900, 1000 mV s1 ) in 1 M perchloric
acid.
A limited surface area (1 cm2 ) of each sample was used for
electrochemical experiments; the remaining electrode surface was
protected by wrapping it with a Teon tape.
The investigation of the ChlER involved different chloride solu-
tions, saturated with Cl2 ; a pH <2 was employed for minimizing
the parasitic contribution due to oxygen evolution reaction (OER)
[21,22]; gaseous chlorine was electrochemically synthesized in a
secondary cell, and bubbled into the working cell by means of a sili-
cone pipe connected to a glass frit. A triple-body measuring cell was
adopted, in which the working and counter electrode (WE and CE,
respectively) compartments were divided by a Naon membrane.
The two compartments were kept at 25 C through a circulating
bath thermostat. The reference electrode (RE) was placed in a
182 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189
Fig. 1. (A) Cyclic voltammetric curves for different samples investigated, collected at 100 mV s1 in 1 M HClO4 ; (B) dependence of anodic and cathodic peak currents against
scan rate for SG350; (C) AFM image of a portion of surface of SG350 (50 50 m); (D) analogous AFM image for PV450.
separate compartment, and connected to the WE through a Luggin
capillary arranged downstream of a liquid junction bridge (for
increasing the impedance of the reference circuit).
The concentrations of Cl explored were 0.05, 0.1, 0.5 and
1 M, in 0.01 M HClO4 , and a constant ionic strength of 1 mol l1
was guaranteed by adding proper amounts of sodium perchlo-
rate. In the working electrode compartment, the pH of the system
remained stable to 1.8, during all the time of testing. After a
pre-treatment by cyclic voltammetry in 1 M HClO4 , quasi-steady
polarization curves were recorded in the potential range 1.50 -
1.05 V (vs. SCE). The polarization was started from the higher
potential, by scanning the interval at a rate of 0.5 mV s1 . In addi-
tion, steady-state polarization curves were collected by starting
from the more anodic potential and decreasing it with a step of
5 mV, applying a step polarization time of 5 minutes (required
to reach the steady state). The stability of the lm was checked
doing CVs in 1 M HClO4 before and after each series of polarization
tests.
As regards the reaction order with respect to the proton, it
was evaluated basing on results obtained at two samples (SG350
and PV350), and using the same triple-body cell described before.
The low-eld approximation was considered for these determi-
nations; the initial test solution was 1 M NaCl in 0.01 M HClO4 ,
with a pH of 1.8. Then, the concentration of protons was mod-
ied by adding dropwise a concentrated NaOH solution (about
30 g l1 ), before each test, while measuring the obtained pH:
thanks to this approach, the volume of the solution in which
the working electrode is placed remains almost unchanged, as
well as its ionic strength. The range of explored overpotentials
() was of 20 mV around the equilibrium potential, with a
scan rate of 0.1 mV s1 , moving from the higher potential to the
lower.
Electrochemical Impedance Spectrometry (EIS) data were
obtained, under the different conditions discussed in the result
section, using a Solartron SI 1287 electrochemical interface cou-
pled with a Solartron SI 1260 impedance/gain-phase analyzer. For
these investigations, the cell was modied by removing the liq-
uid junction between WE and RE. A quasi-reference electrode was
used, constituted by an Ag wire covered with a thin layer of AgCl,
which was put directly inside the Luggin capillary, in the chloride-
containing WE electrolyte; its potential was evaluated before each
test by using a commercial double-junction saturated calomel elec-
trode.
All used chemicals were chosen with a high degree of purity.
3. Results and Discussion
The voltammetric analysis of electrodes has shown that the dif-
ferent samples are characterized by rather different voltammetric
charges, the latter being directly related to the surface morphology
and porosity degree. As shown in Fig. 1A, the SG350 sample exhibits
a much larger charge, compared with the other three, with sim-
ilar charge-storage capacity. By using two preparation methods,
as well as two preparation temperatures, it was thus possible to
obtain electrodes with different porosity. In particular, as exem-
plied in Figs. 1C and 1D (AFM images: 50 50 m of explored
surface), the PVD method allows to prepare more compact struc-
tures with respect to the sol-gel approach, which in turn produces
lms characterized by the typical cracked-mud surface. A more
detailed discussion on surface analysis has been provided in ref.
[20]. Despite the observed differences in morphology and also in
charge-storage-capacity, plots of jpeak vs. scan rate (as exemplied
in Fig. 1B for the SG350) highlight an extended linearity over the
whole range of explored scan rates (0.011.00 V s1 ). According
to largely accepted hypotheses, these voltammetric current values
contains contributions due to changes in oxidation state for the
iridium sites (i.e., IrIII IrIV ), which appear as peak signals at about
0.63 VSCE (anodic branch) and 0.58 VSCE (cathodic branch), respec-
tively. In consideration of the linear behavior shown in Fig. 1B, it
is reasonable to assume that the processes of charge/discharge of
electroactive sites involves only the near-surface-region and that
investigated samples behave as surface-modied electrodes.
As regards the chlorine evolution reaction, extensive investiga-
tion has been carried out by means of quasi-steady polarization
curves by collecting data at different chloride concentration for the
four electrode materials. The E/logj plots, obtained in 1 M NaCl solu-
tion (pH 2, T = 298 K), are shown in Fig. 2. Their main feature is
represented by the lack of linearity, which could be tentatively
sought only within restricted ranges of potentials. As a limiting
trend, in the lower part of the explored potential range, slopes
around 50-70 mV can be found. In any case, the generally accepted
requirement of linearity along at least one order of magnitude of j
is never met. Although substantially not commented, several cases
of non-linearity in Tafel plots can be found in literature, e.g. in the
papers proposed by Kuhn et al. [23], Burke et al. [24] and renburg
et al. [25].
The observed non-linearity of Tafel plots, in a potential range
in which the high approximation can be applied, prevents a
straightforward assignment of mechanism and rate-determining
 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189
Fig. 2. Tafel plots (derived from steady-state polarizations) for chlorine evolution
in 1 M NaCl/0.01M HClO4 , collected for () SG350, () SG450, () PV350 and ()
PV450; continuous lines represent tting of experimental points through second-
order-degree polynomial functions.
step, without further assumptions. The kinetics of the reaction is
clearly more complex than expectable from the simpler Volmer-
Tafel and Volmer-Heyrovsky schemes. The introduction of Cl+
adsorbed intermediates and oxychloro-adsorbates [26,27] does not
seem to improve the picture either.
In Fig. 3, E/logj curves obtained for a PV450 electrode at different
chloride concentrations, at constant pH and ionic strength, are
shown. Similar results were obtained at the other three electrodes.
From the above data, the reaction order with respect to Cl- , RCl ,
could be evaluated and results are reported in Table 1. The choice
of the lower and the higher potentials for determining log(j) at
different chloride concentrations is legitimated by the fact that
also current values collected at 1.22 V appear reasonably free
from contributions due to the backward reaction of chlorine
reduction: this evaluation was done by considering the whole
polarization curves, collected under quasi-steady conditions
(those accomplished for activating each electrode material). A
couple of examples, relating to data obtained for SG350 at the
Fig. 3. Tafel plots (derived from steady-state polarizations) (dot) for chlorine evo-
lution in xM NaCl/(1-x)M NaClO4 /0.01 M HClO4 , collected at PV450; continuous
lines represent tting of experimental points through second-order-degree poly-
nomial functions. Effect of the chloride concentration: NaCl () 0.05 M, () 0.10 M,
() 0.50 M and () 1.00 M.
183
Table 1
Reaction orders with respect to chloride, for all the samples studied, estimated at
different electrode potentials (vs. SCE).
1.22 V 1.24 V 1.26 V 1.28 V
SG350 1.17 1.04 0.92 0.79
SG450 1.00 0.95 0.93 0.92
PV350 1.25 1.20 1.23 1.29
PV450 0.64 0.62 0.83 0.52
lowest and highest concentrations of chloride, is shown in Fig. 4.
Reaction orders close to 1 are found in all cases, up to a lower limit
of 0.52 for PV450 electrode; as a general observation, for each
electrode, low R values are found at higher electrode potentials.
The above values contradict the expectations based on Volmer-
Tafel (V-T) and Volmer-Heyrovsky (V-H) mechanisms under their
simpler forms. On the other hand, they are also somewhat in
disagreement with the mechanism proposed by Krishtalik and
renburg [2628], involving a Cl(+1) intermediate (possibly an
oxochloride complex), which would contemplate a unitary value
for RCl [29,30].
Basing on Tafel slope and reaction order with respect to chlo-
ride, a straightforward assignment of mechanism does not seem
possible. Maintaining in rst approximation the two models, V-T
and V-H, the rst seems less applicable, no limiting current being
observed in j/ plots. Slopes close to 120 mV in the higher portion
of /logj curves, on the other hand, would suggest the V-H as more
realistic (see also Over [17]). The lack of extended linearity with
a slope close to 40 mV, and the reaction order centered around
1, suggest that data analysis has to be carried out along different
lines. In this context, the approach proposed by Tilak and Conway
in 1992 [31], introducing the effect of coverage by reaction inter-
mediates on Tafel plots and electrochemical reaction orders, was
considered. This treatment is a general one. In that paper, analytical
relations between Tafel slope (b) and reaction order (R) have been
deduced in connection with the two main types of mechanism, i.e.
the V-T and V-H pathways, which are the most common models
for explaining electrochemical reactions when only one interme-
diate specie is involved (e.g. H2 , Cl2 , etc.). According to the authors, a
discrepancy between experimentally measured kinetic parameters
and theoretical values is often found, when considering heteroge-
neous processes that involve the adsorption of intermediates. In
particular, a role is played by the quantity dln/dlnc, deriving from
Fig. 4. Current-potential curves (derived from quasi-steady polarizations) for chlo-
rine evolution in xM NaCl/(1-x)M NaClO4 /0.01 M HClO4 , collected at SG350. Effect
of the chloride concentration. The blue dash line points out the minimum potential
considered for evaluating the reaction order with respect to chloride.
184 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189
the absorption isotherm, which links the fractional coverage of the
reaction intermediate to the concentration of the reactant specie in
the bulk (Cl- , in our particular case).
The model was tested on polarization curves recorded at 25 C
in 1 M NaCl, for all the materials analyzed, and considering both
mechanisms (V-T and V-H) and two adsorption isotherms (Lang-
muir and Frumkin). To do that, Tafel plots were initially tted by
using a second-order polynomial function, as exemplied in Fig. 2
(continuous lines).
Since the Tafel slope is a derivative function (see Equation 1),
when considering a second-order polynomial, b cannot assume a
constant value.
 
E
b= (1)
log j c,T
Furthermore, for the reasons mentioned above, the Tafel slope
depends on the coverage of reaction intermediates, and the latter
is a function of the potential [15,16]; likewise, also the reaction
order depends on both, the coverage and the potential. In conclu-
sion, a given mechanism can be tested by using the general relation
resumed by Eq. 2.
R(E) b(E) = f ((E) ) (2)
If a Langmuir-type adsorption is assumed, for the Volmer-Tafel
(chemical recombination) mechanism one obtains:
2RT
R(E) b(E) = (3)
F especially at low overpotentials. Strong interactions among the
while in the case of the Volmer-Heyrovsky (electrochemical
desorption) path, the outcome is [31]:
RT 2 (E)
R(E) b(E) = (4)
F + 1 (E)
In the present work, calculations were made considering both
the chemical (V-T) and the electrochemical (V-H) desorption
mechanisms, but a convergence (i.e., the compliance with the math-
ematical relation described by Eq. 2 and made explicit by either Eq.
3 or Eq. 4) was found only in the case of the V-H pathway, which
was thus conrmed as the active mechanism in our cases.
Once the reaction mechanism was dened, the dependence of
coverage on overpotential was examined (as anticipated, polariza-
tion curves show a lack of linearity in the Tafel region, which is
plausibly due to a variation of coverage with the electrode poten-
tial), and both the Langmuir and Frumkin adsorption isotherm
models were taken into consideration. For the Langmuir case, the
relation arising between the Tafel slope and the coverage function
is [31]:
3 RT
(E) = 2.303 (5)
2 b F
where b is given by Eq. 1, and it depends on the coverage through
the electrode potential.
Table 1.
In the case of the Frumkin isotherm, it is easier to consider
the relation between the reaction order and the coverage function,
which is as follows [31]:
 
log j
RCl =
log cCl ,T
 
= 1 + (1 ) 1 + g / [1 + g (1 )] (6)
In order to obtain the real coverage function, a recursive algo-
rithm based on the previous equations was applied; by taking the
Langmuir isotherm as a guess-model, a rst value for the coverage
was estimated, starting from the experimental values of RCl , and a
Table 2
Lateral interaction parameter, g, estimated through the model proposed by Conway
[29].
Sample g
SG350 4.0
SG450 2.9
PV350 6.3
PV450 4.9
value for the lateral interaction factor, g, was eventually quantied.
Then, the latter was used for calculating a new coverage function
that, in turn, was used to obtain a more accurate value of g. By
repeating this mathematical iteration several times, convergence
between the coverage function and the value of g can be obtained;
the last step consisted in plotting the coverage functions obtained
by considering the two different hypotheses. The whole iteration
procedure was carried out by using an algorithm written for the
software Wolfram MathematicaTM . The different coverage func-
tions, calculated by starting from experimental data for the various
electrode materials, are shown in Fig. 5. Estimates for the g param-
eter, obtained through the above-described recursive procedure,
have been collected in Table 2.
Generally speaking, obtained values of g are signicant, in
agreement with the fact that some deviations can be appreciated
between the Langmuir- and the Frumkin-based functions in Fig. 5,
intermediates involved in the gas evolution reaction have been
found in particular for the PV350 specimen, probably because of
the conformation of electroactive sites at this electrode, which may
favor a lateral attraction between adsorbed chloride radicals. Look-
ing at Fig. 5, noteworthy is also the rapid increase of the coverage
related to the Frumkin isotherm: typically, when the overpoten-
tial is higher than a few tens of mV, the coverage function has
already arrived to a plateau, characterized by a high value, com-
prised between 0.9 and 1.0 (the surface saturation limit).
Further information on the behavior of the different samples has
been obtained through EIS; the performed analysis has revealed
two distinct situations, which are characterized by impedance
responses that depend on the temperature of preparation rather
than by the adopted deposition method. As exemplied in Fig. 6
for the sputtered electrodes, a single time constant is visible for
specimens prepared at 450 C, while two contributions not com-
pletely resolved appear in the case of lms synthesized at 350 C
(i.e., the Nyquist plot has a attened shape, Fig. 6A-top, while the
frequency part of the Bode representation is slightly asymmetric,
Fig. 6A-bottom).
The tting of data was carried out applying the CNLS (Complex
Non-Linear Least Square) method, through the Calculus-Modulus
algorithm; the choice of physically meaningful equivalent models
has been done on the basis of experimental results.
Results for samples prepared at 450 C have proven to be well
modeled by using the simplest circuit (Randles equivalent circuit
[32]), which comprises the resistance of the solution (Rs ) in series
with a single RC-mesh involving the charge transfer resistance
(related to the faradic process of chloride discharge, Rct ), in parallel
with the double-layer capacity (Cdl ). The latter, which includes also
the effects of adsorbed intermediates, is often replaced by a CPE
(Constant Phase Element), in order just to take into account several
non-ideal behaviors.
A verication of the goodness of the tting can be made
by plotting the Log Rct 1 vs. , and comparing the trend with
that of the corresponding Log j vs.  plot, determined step by
step, as exemplied in Fig. 7 for SG450 and PV450. The agree-
ment between the measured currents and the charge-transfer
 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189 185

Fig. 5. Coverage functions vs. overpotential, related to steady-state polarization curves collected in 1 M NaCl/0.01 M HClO4 (pH 1.8) at 25 C for (A) SG350, (B) SG450, (C)
PV350 and (D) PV450 samples.

Fig. 6. Impedance responses: Nyquist (top) and Bode (bottom) plots for (A) PV350 and (B) PV450 samples.
186 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189

Fig. 7. Tafel plots for samples prepared at 450 C: current vs. inverse of the charge transfer resistance.

Table 3 samples prepared at 350 C in 1 M NaCl solution, the problem of dis-

Double-layer capacities evaluated from modeling EIS data, through the Brugs model
criminating between the two models has requested a modication
[31].
of the experimental conditions.
Sample Cdl (Brugs model) (1 M Cdl (Brugs model) (1 M In fact, if the added mesh is due to porosity, the experimental
NaCl25 C) (mF cm2 ) HClO4 25 C) (mF cm2 )
results obtained should not vary when parameters such as the tem-
SG350 53.9 54.4 perature or the reactant concentration in the solution are changed,
SG450 13.2 - since the added contribution to the circuit represents a physical
PV350 9.0 9.7
and geometric characteristic of the electrode. This is not true in the
PV450 9.3 -
second case, where nested meshes are considered, because of the
kinetic nature of R1 C1 . Thus, the following experimental conditions
resistances, determined through the tting procedure, is rather were investigated, for the two samples:
good, conrming the validity of the chosen tting model.
In addition, from the CPE values, a rough estimation of the NaCl 1 M in HClO4 0.01 M, Cl2 sat.25 C (E: 1.381.56 V vs. SHE);
double-layer capacity can be obtained, by means of the Brugs rela- NaCl 1 M in HClO4 0.01 M, Cl2 sat.15 C (E: 1.381.56 V vs. SHE);
tion (Eq. 7) [33]: HClO4 1 M (E: 0.341.24 V vs. SHE);
1 1 NaCl 0.5 M in HClO4 0.01 M, Cl2 sat.25 C (E: 1.3641.544 V vs.
 1 SHE).
T = Cdl + (7)
Rs Rct

The Brugs model is only valid when  is above 0.8-0.9, and this
condition must be veried, at least within a portion of the explored
potential window. It is worth observing that obtained values (see
Table 3) plausibly represent an overestimation of the double-layer
capacity of coatings: in fact, as the charging of these coatings is
conditioned by proton injection/ejection [34], it does involve con-
tributions from a few monolayers within a near-surface-region. The
contribution of the more external part, the one likely involved in
faradaic processes as Chl.E. and O.E. reactions, is expected to be
lower (as also witnessed by available data in Table 3). As a result,
the Cdl has to be considered as a measure of the catalytic dispersion
of the material, related to the real surface area of the lm (which is
however not fully accessible to chloride anions).
The experimental framework relating to samples prepared at
350 C is more complex to interpret; two different equivalent cir-
cuits can be considered: one is related to the porous electrode
(Fig. 8A), while the other takes into account the contributions due
to adsorbed intermediates (Fig. 8B). In the former, the additional
R1 C1 mesh is connected in series to the Rct Cdl , while, in the lat-
ter, the two meshes are nested. Analyzing the data collected for the
Both the concentration of chloride and the temperature were
varied under chlorine evolution conditions, and only one test was
carried out in HClO4 by working at lower potentials, in the same
conditions used for the estimation of the number of electroactive
sites through cyclic voltammetry [20].
An example of Nyquist and Bode plots, collected under differ-
ent experimental conditions but at constant overpotential, for the
PV350 sample, is shown in Fig. 9; in order to allow an easy com-
parison between data, the values of Z were shifted by subtracting
the solution resistance, Rs .
For both specimens (SG350 is not shown), a constant contribu-
tion at high frequencies is well visible, which does not seem to be
related to a kinetic-controlled process, but rather attributable to
geometric characteristics of the samples. Accordingly, the porous
electrode model seems suitable to describe the behavior experi-
mentally determined. A further proof (not shown) was obtained by
comparing the plots of the logarithm of the steady-state current (log
i) and of the logarithm of the inverse of the charge transfer resis-
tance (log Rct 1 ), obtained through the tting of EIS data, vs. the
overpotential, as previously done for samples prepared at 450 C.

Fig. 8. Equivalent circuit for the porous electrode (A) and for a system in which adsorbed intermediates are involved (B).
 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189 187

Fig. 9. Nyquist (A) and Bode (B, C) diagrams, collected under different experimental conditions, for PV350 (= 100 mV).

Moreover, by plotting Rct and R1 against the overpotential, it is
possible to appreciate that the latter remains practically constant
(Fig. 10A), in agreement with expectations based on the porosity,
which in fact characterizes the materials prepared at the lower
temperature, as evidenced by AFM and CV analysis [20].
The pore resistances, estimated for the different conditions of
temperature and concentration examined, were comparable and
comprised between 0 and 2  (Fig. 10B).
A very similar behavior was attested for SG350: also in this case,
the equivalent circuit of the porous electrode proved to be quite
satisfactory for modeling the system.
EIS experiments performed in perchloric acid conrmed the
contribution of the porosity, seen under ChlER conditions. The con-
sidered equivalent circuit was similar to that shown in Fig. 8A, apart
from the necessary changes: in fact, the faradaic mesh degenerates
in a single CPE element, which models the double-layer charg-
ing within the explored potential window. The latter was limited
between 0.1 and 1.1 VSCE : in this range, the electrode material
is characterized by a basic capacitive behavior, and a pseudo-
capacitive contribution adds to the former, when the potential
reaches 0.63 VSCE (changes in oxidation state of the electroac-
tive sites) [35]. Moreover, under these conditions, also the pore

Fig. 10. Dependence of resistances on overpotential, for PV350: (A) Rct and R1 (related to porosity); (B) R1 for the different conditions explored (under ChlER).
188 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189
Fig. 11. Reaction orders with respect to the proton, for (A) SG350 and (B) PV350. The plotted currents are the exchange currents ( = 0) obtained by polarization tests carried
out under low-eld approximation.
Table 4
Pore resistances estimated in NaCl and HClO4 solutions, for samples prepared at
350 C.
Sample Rpore (1 M NaCl25 C) () Rpore (1 M HClO4 25 C) ()
SG350 0.6 0.4
PV350 0.9 0.2
resistance can be estimated. Obtained values are lower than those
evaluated under ChlER, probably because of the different nature
of the chemical species involved: the dimensions of proton, with
respect to those of the chloride anion, justify the smaller pore
resistance measured, since their diffusion through the nano/micro-
porosity of the coating is easier and characterized by a lower
activation energy. To allow a comparison, data of pore resistance,
determined in NaCl and HClO4 solutions, have been collected in
Table 4. Moreover, by assuming again the Brugs model, also the
double-layer capacities were estimated (obtained results have been
collected in Table 3).
With reference to Fig. 9, it is worth mentioning that the attened
character of Nyquist diagram was noted also by Denton et al. [29],
who worked with RuO2 -TiO2 electrodes in solutions of KCl + HCl,
at different compositions; they also ascribed the feature to the lm
porosity, but without any sort of experimental proof. An Asiatic
research group has recently highlighted an analogous behavior, by
working with oxide- materials based on Ru and La [36]. The same
model proposed in this paragraph has been adopted, with the addi-
tion of a Warburg element in order to justify the presence of a at
portion in the curve (Nyquist diagram), also at very low frequencies,
which was attributed to a process controlled by diffusion.
In all cases, and in all probability, an RC contribution related to
the adsorption/desorption of intermediates exists, which cannot be
measured because of the not-so-high variation of d/d within the
Tafel region: according to this boundary condition, the R1 C1 mesh
degenerates in a purely resistive contribution, which adds to the
faradic resistance associated with the discharge of chloride.
Eventually, Fig. 11 shows the plots related to the reaction order
with respect to H+ (RH ), estimated at the open circuit potential
(OCP), for the sol-gel and sputtered samples prepared at 350 C;
values of -0.22 0.04 and -0.23 0.08, respectively, were estimated
by considering the rst portion of the graph, where log[H+ ] is lower
than -1. Our results agree with what has been mentioned in a paper
by renburg et al. [30]: the rate of the ChlER increases with the pH.
Rather different results were obtained and discussed by Janssen
[37,38], by Thomassen et al. [39] and by Tamura and coworkers
[40]; in particular, Janssen observed an increase of Tafel slope and a
sharp decrease of current for pH values higher than 3. A clear reason
for these contradictory observations has not been proposed yet, but
it could be related to effects of screening in the double-layer or to
something resembling a competition between two or more adsorp-
tion processes. Another anomalous outcome is represented by the
abrupt decrease in the logarithm of the exchange current when the
pH exceeds a particular value, which seems to be dependent on the
nature of the material; an interesting aspect is that this pH value is
ever comprised between 0 and 1. Since this result is in close agree-
ment with the pzc of IrO2 powders proposed by Ardizzone et al.
[41], it is possible that H+ are not directly involved in the mecha-
nism of chlorine evolution, but rather in a surface modication of
danglingOH groups. A protonated surface might affect negatively
the process of chlorine evolution, the effect being more incisive
at values below the plausible pzc of iridium oxide. When oxygen
atoms are completely engaged with hydrogen atoms (through the
formation ofOH2 + groups), interaction with chloride anions before
the formation of the radicals could be hindered at a certain extent.
Contrary to expectations, the phenomenon gives the impression of
not being connected to the net electrostatic charge of the electrode
surface: in fact, a positive net charge should favor the interaction
with chloride in solution.
4. Conclusions
Polarization experiments, carried out for investigating the kinet-
ics of the chlorine evolution reaction, have evidenced a lack of
linearity in the high-overpotential region of Tafel plots, for all sam-
ples; besides, average values of Tafel slope higher than 60 mV/dec
have been obtained (estimates based on tting of data with a rst-
order polynomial function).
As for the reaction order with respect to Cl- , a value almost uni-
tary (or less than 1), has been found, which is not compatible with
the commonly proposed mechanisms.
Following the approach proposed by Tilak and Conway in the
early 90s [31], which has been further developed in this work,
we can afrm that the reaction proceeds via a Volmer-Heyrovsky
pathway, involving high coverage values for the reaction interme-
diate. The Langmuir adsorption model has proved to be insufcient
to describe the state of the adsorbed intermediate, and signi-
cant improvements have been obtained by introducing the lateral
interaction parameter that characterizes the Frumkin isotherm.
Coverage values 0.9 were estimated, in the Tafel region, which
have been shown to be characterized by small variations, but large
enough to produce in some cases a curvature in the Tafel plots.
In addition, the observed inhibitory effect of the protons, on the
rate of the ChlER, can be possibly ascribed to modications of the
properties of the surface, and may represent an additional proof
to the hypothesis that the discharge of chloride anions takes place
 D. Rosestolato et al. / Electrochimica Acta 139 (2014) 180189
on a strongly oxidized surface, where the interaction between sites
and intermediates can occur only via surface-oxygen atoms.
The impedance spectroscopy analysis did not supply additional
useful information on the behavior of adsorbed intermediates,
plausibly because of the small variation of the coverage with the
overpotential, in the Tafel region. The faradic contribution of the
electrochemical discharge of chloride was visible in all cases, while
a second contribution could be found only for samples prepared at
350 C. The latter, which was assessed at the highest frequencies,
has proved to be due to geometrical factors, which are characteristic
of the electrodes prepared at the lowest calcination temperature.
However, because of the nature of samples analyzed, this feature
did not show to play a substantial role in the kinetics of the ChlER,
conrming the fact that the latter is practically a surface reaction,
which involves only the most external portion of the electrocata-
lyst.
Acknowledgements
The authors are indebted to Prof. Andrzej Lasia (Dpt. de Chimie,
Universit de Sherbrooke) for his teachings on impedance spec-
troscopy and for supervising an important part of the experimental
section, and to Prof. Giancarlo Battaglin (Dip. di Scienze Molecolari
e Nanosistemi, Universit Ca Foscari di Venezia) for the synthesis
of sputtered electrodes and fruitful discussion.
References
[1] H.B. Beer, Electrode having platinum metal oxide coating thereon, and method
of use thereof. US 3711385 (1973).
[2] H.B. Beer, Electrode and coating therefor. US 3632498 (1972).
[3] A. Scrutton, L. Denis, Electrodes for electrochemical processes, US 3773554
(1973).
[4] J.F. Gauger, J.M., Hinden, M. Katz, Electrode coating with platinum-group metal
catalyst and semi-conducting polymer, US 4402996 (1983).
[5] D. Galizzioli, F. Tandardini, S. Trasatti, Ruthenium dioxide: a new electrode
material. I. Behaviour in acid solutions of inert electrolytes, J. Appl. Electrochem.
4 (1974) 57.
[6] S. Pizzini, G. Buzzanca, C. Mari, L. Rossi, S. Torchio, Preparation, structure and
electrical properties of thick ruthenium dioxide lms, Materials Research Bul-
letin 7 (1972) 449.
[7] K.J. OLeary, T.J. Navin, Morphology of dimensionally stable anodes, Chlorine
Bicenten. Symp., Electrochem. Soc. Inc., Princeton, N.J., 1974, pp. 174186.
[8] H.J. Vanderbroele, Kinetic investigation of chlorine evolution in the electrolysis
of molten salts, Rev. Fac. Cienc. Quim., Univ. Nacl. La Plata 34 (1964) 215.
[9] L.I. Krishtalik, Z.A. Rotenberg, Overvoltage in the anodic evolution of chlorine
on graphite. II. Kinetic dependence in different regions of potentials, Zhurnal
Fizicheskoi Khimii 39 (1965) 907.
[10] S. Trasatti, G. Buzzanca, Ruthenium dioxide: a new interesting electrode mate-
rial. Solid state structure and electrochemical behavior, J. Electroanal. Chem.
29 (1971), App.1.
[11] G. Bianchi, Fundamental and applied aspects of the electrochemistry of chlo-
rine, J. Appl. Electrochem. 1 (1971) 231.
[12] G. Faita, G. Fiori, Anodic discharge of chloride ions on oxide electrodes, J. Appl.
Electrochem. 2 (1972) 31.
[13] S. Trasatti, Progress in the understanding of the mechanism of chlorine evolu-
tion at oxide electrodes, Electrochim. Acta 32 (1987) 369.
[14] H. Ohtaki, T. Radnai, Structure and dynamics of hydrated ions, Chem. Rev. 93
(1993) 1157.
[15] H.A. Hansen, I.C. Man, F. Studt, F. Abild-Pedersen, T. Bligaard, J. Rossmeisl,
Electrochemical chlorine evolution at rutile oxide (110) surfaces, Phys. Chem.
Chem. Phys. 12 (2010) 283.
189
[16] K.S. Exner, J. Anton, T. Jacob, H. Over, Chlorine evolution reaction on RuO2 (110):
ab initio atomistic thermodynamics study - Pourbaix diagrams, Electrochim.
Acta 120 (2014) 460.
[17] H. Over, Atomic scale insights into electrochemical versus gas phase oxidation
of HCl over RuO2 -based catalysts: A comparative review, Electrochim. Acta 93
(2013) 314.
[18] P. Barricelli, M., Boccardo, G. Maranelli, IT PD 2008A000104 (2008).
[19] A. Morozov, A. De Battisti, S., Ferro, G. Martelli, EP 1656471 B1 (2009).
[20] D. Rosestolato, S., Neodo, S., Ferro, G., Battaglin, V., Rigato, A. De Battisti, New
insights on structural and electrochemical properties of iridium oxide-based
electrocatalysts prepared by sol-gel and reactive sputtering deposition, sub-
mitted.
[21] S. Trasatti, G. Lodi, Electrodes of Conductive Metallic Oxides, Part B, 535, Else-
vier, 1981.
[22] S. Trasatti, Physical electrochemistry of ceramic oxides, Electrochim. Acta 36
(1991) 225.
[23] A.T. Kuhn, C.J. Mortimer, The kinetics of chlorine evolution and reduction on
titanium-supported metal oxides especially RuO2 and IrO2 , J. Electrochem. Soc.
120 (1973) 231.
[24] L.D. Burke, J.F. ONeill, Electrochromism in anodic iridium oxide lms: II. pH
effects on corrosion stability and the mechanism of coloration and bleaching,
J. Electroanal. Chem. 101 (1979) 341.
[25] R.G. renburg, L.I. Krishtalik, I.P. Yaroshevskaya, Mechanism of chlorine deposi-
tion on a ruthenium oxide-titanium oxide electrode, Elektrokhimiya 11 (1975)
1068.
[26] R.G. renburg, L.I. Krishtalik, I.P. Yaroshevskaya, Kinetics of the evolution and
ionization of chlorine on ruthenium oxide and ruthenium oxide-titanium oxide
electrodes, lektrokhimiya 11 (1975) 1072.
[27] R.G. renburg, L.I. Krishtalik, I.P. Yaroshevskaya, Separation and ionization of
chlorine on oxide ruthenium-titanium electrodes at increased current densi-
ties, lektrokhimiya 11 (1975) 1236.
[28] L.I. Krishtalik, Kinetics and mechanism of anodic chlorine and oxygen evolution
reactions on transition metal oxide electrodes, Electrochim. Acta 26 (1981) 329.
[29] D.A. Denton, J.A. Harrison, R.I. Knowles, Chlorine evolution and reduction on
RuO2 /TiO2 electrodes, Electrochim. Acta 24 (1979) 521.
[30] R.G. renburg, L.I. Krishtalik, N.P. Rogozhina, Effect of pH on the kinetics
of a chlorine reaction on ruthenium dioxide-titanium dioxide anodes, Elek-
trokhimiya 20 (1984) 1183.
[31] B.V. Tilak, B.E. Conway, Analytical relations between reaction orders and Tafel
slopes derivatives for electrocatalytic reactions involving chemisorbed inter-
mediates, Electrochim. Acta 37 (1992) 51.
[32] J.E.B. Randles, Kinetics of rapid electrode reactions, Discuss. Faraday Soc. 1
(1947) 11.
[33] G.J. Brug, A.L.G. Van den Eeden, M. Sluyters-Rehbach, J.H. Sluyers, The analy-
sis of electrode impedances complicated by the presence of a constant phase
element, J. Electroanal. Chem. 176 (1984) 275.
[34] S. Gottesfeld, J.D. McIntyre, Electrochromism in Anodic Iridium Oxide Films: II.
pH Effects on Corrosion Stability and the Mechanism of Coloration and Bleach-
ing, J. Electrochem. Soc. 126 (1979) 742.
[35] T.F.A. Lassali, J.F.C. Boodts, L.O.S. Bulhes, Faradaic impedance investigation of
the deactivation mechanism of Ir-based ceramic oxides containing TiO2 and
SnO2 , J. Appl. Electrochem. 30 (2000) 625.
[36] L. Ping, L. Quinfeng, X. Likun, X. Lili, S. Hongging, Electrochemical Impedance
Spectroscopy (EIS) Study of Ru-La Oxide Coatings in NaCl Solution, Acta Chimica
Sinica 70 (2012) 1166.
[37] L.J.J. Janssen, L.M.C. Starmans, J.G. Visser, E. Barandrecht, Mechanism of the
chlorine evolution on a ruthenium oxide/titanium oxide electrode and on a
ruthenium electrode, Electrochim. Acta 22 (1977) 1093.
[38] L.J.J. Janssen, Kinetics and mechanism of chlorine evolution on metal oxide elec-
trodes, in: C. Jackson (Ed.), Modern Chlor-Alkali Technology 271, Ellis Horwood,
Chichester, 1983.
[39] M. Thomassen, C. Karlsen, B. Brresen, R. Tunold, Kinetic investigation of the
chlorine reduction reaction on electrochemically oxidised ruthenium, Elec-
trochim. Acta 51 (2006) 2909.
[40] T. Arikado, C. Iwakura, H. Tamura, Some oxide catalysts for the anodic evolution
of chlorine: reaction mechanism and catalytic activity, Electrochim. Acta 23
(1978) 9.
[41] S. Ardizzone, S. Trasatti, Interfacial properties of oxides with technological
impact in electrochemistry, Advances in Colloid and Interface Science 64 (1996)
173.