Journal of The Electrochemical Society, 146 (3) 977-983 (1999) 977

S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

Corrosion-Induced Activation of SnO2 Films Promoted by the

Electrochemical Oxidation of Chloride Ions
H. Cachet,z F. Zenia,* and M. Froment**
UPR 15 du CNRS, Physique des Liquides et Electrochimie, Université Pierre et Marie Curie, 75252 Paris cedex 05, France

At potentials more positive than the flatband potential, tin dioxide electrodes are normally passivated with respect to charge trans-
fer because of an electron-depleted surface. Nevertheless, such a passivated state can be lost by the occurrence of a corrosion
process. In the potential region 0.9-1.3 V/saturated calomel electrode close to the thermodynamic potential for chloride oxidation,
sprayed SnO2 thin film electrodes present a loss of mass, determined in situ by quartz crystal microbalance experiments. This phe-
nomenon is characterized by a peak-shaped corrosion rate curve, the amplitude of which depends on the pH, the chloride concen-
tration, the presence of metallic traces, and the type of the dopant. From data obtained at pH 8, and in natural seawater, it is appar-
ent that the corrosion rate is (i) amplified by the presence of metallic traces, and (ii) strongly dependent on the Cl2 content, becom-
ing almost negligible for [Cl2] > 0.3 M. These results are tentatively explained by considering the electrochemical production of
OH• and Cl• radicals, able to break Sn-O surface bonds by electron capture (chemical steps), in competition with chlorine and oxy-
gen evolution. The corrosion process is also responsible for a positive shift of bandedges due to charged corrosion intermediates,
allowing the SnO2 electrode to become active with respect to anodic charge transfer reactions.
© 1999 The Electrochemical Society. S0013-4651(98)03-009-2. All rights reserved.

Manuscript submitted March 2, 1998; revised manuscript received November 9, 1998. This was Paper 410 presented at the Paris,
France, Meeting of the Society, August 31-Septembere 5, 1997.

Because of a large overvoltage for oxygen evolution, highly ic reactions for chlorine and oxygen evolution, and the electrode
doped tin dioxide electrodes are used as anodes for wastewater treat- activation.
ment1-5 or for biofilm destruction by chlorine evolution.6 At poten- Experimental
tials positive with respect to the flatband potential, there is a large
Conductive SnO2 films were prepared by the spray pyrolysis
potential window in which organic pollutants can be oxidized or
technique starting from a 0.2 M SnCl4 solution in methanol. The
chlorine to be evolved. With respect to these applications, there are
deposition temperature was fixed at 5008C. Doping was achieved by
some points to be considered. First, in the anodic regime, SnO2 is
adding either NH4F (F/Sn 5 70%) or SbCl3 (Sb/Sn 5 1 and 4%) to
normally depleted and would rather form a passivated interface
the spray solution. The carrier concentration was 4.0, 8.5, and 5.0 3
which would be unfavorable for efficient charge transfer. The high-
1020 cm23, respectively.9 Mass change measurements were per-
er the pH, the more passivated the interface.7 Second, halogen and
formed by spraying the doped SnO2 layers onto 6 MHz AT-cut
hydroxide ions are also oxidized in the potential window of interest,
piezoelectric quartz disks, to be used as mass sensors with a home-
and it is important to monitor whether these reactions have some
made EQCM. The SnO2 layers (400 nm thick) were deposited either
influence on the working conditions of the SnO2 anode. In the case
on the keyhole shaped, gold quartz exciting electrode (SnO2-Au), as
of tin-doped indium oxide layers [indium-tin oxide (ITO)], marked
supplied, or the same but covered by a sputtered titanium layer
changes in electrochemical properties were observed after strong
(SnO2-Ti), or directly on the quartz crystal itself (SnO2-Q), after
anodic polarization corresponding to chlorine or oxygen evolution.8
removing all metallic traces on the quartz surface chemically. In the
Third, the influence of the dopant used for degenerating tin dioxide
latter case, the quality factor of the resonator in air was found above
has to be analyzed.
10000, suitable for EQCM experiments.
In a previous study, it was shown by in situ electrochemical quartz
Electrochemical measurements were performed using a three-
crystal microbalance (EQCM) experiments that SnO2 anodes under-
electrode configuration with a potentiostatic control. Potentials were
go corrosion phenomena in the presence of hydroxide and chloride
referred to the SCE reference electrode. Test solutions were either
ions.9 Localized corrosion figures were observed with transmission
natural seawater (pH 8.2) or 1 M (xLiCl 1 (1 2 x) LiClO4) for x 5
and scanning electron microscopy (SEM) techniques. These prelimi-
0, 0.1, 0.3, 0.5, 1, and, 3 M LiCl, in a pH 8 buffer solution (Merck-
nary investigations were mainly carried out in a pH 8 buffer solution
Titrasol, containing 0.04 M chloride). The electrode capacitance, Cp,
containing 0.04 M chloride ions, as a function of the dopant (F, Sb).
was determined by measuring the interfacial impedance at a fixed
The most striking result was that a loss of mass occurs only in a nar-
frequency by using a transfer function analyzer (Voltech TF2000).
row range of potentials [0.9 to 1.3 V/saturated calomel electrode
Mott-Schottky plots C p22 (V) were recorded by varying the applied
(SCE)], in the vicinity of the thermodynamic potential for chloride
potential between 0 and 1.75 V/SCE by steps of 50 mV.
oxidation. A similar corrosion behavior was evidenced for sprayed
ITO layers in contact with dilute HCl solutions.10 Results
By electrochemical impedance measurements, it was also shown Figure 1a shows the change in mass of a SnO2:Sb (1%)-Ti elec-
that the conduction band energy level is shifted toward more positive trode when cycling the potential between 0.3 and 1.7 V/SCE in the
potential because of corrosion. The surface potential barrier which 0.1 M LiCl 1 0.9 M LiClO4 solution at pH 8. A loss of mass only
controls electron transfer is reduced allowing the SnO2 electrode to occurs at potentials more positive than 1.0 V/SCE during the posi-
be activated for anodic electron transfer reactions, as chlorine or tive scan. Except for the first cycle for which it is larger, the mass
oxygen evolution. loss per cycle is reproducible. The same behavior is observed in the
In this paper, we address the influence of chloride concentration buffer solution at very low Cl2 concentration, in agreement with our
on the corrosion rate in a slightly basic medium (pH 8) and that of previous results for tin oxide films deposited on gold sublayers.9 The
the conditions of preparation of the tin dioxide layers, sprayed onto corresponding current-potential (I-V) curve is depicted in Fig. 1b. At
either metal or glass substrates. A kinetic model is proposed to potentials lower than 1.0 V/SCE, the current is zero. During the scan
account for the coupling between the corrosion reactions, the anod- at increasing potentials, in connection with the loss of mass, a cur-
rent hump is observed between 1.1 and 1.3 V/SCE, followed by a
* Electrochemical Society Active Member.
** Electrochemical Society Fellow. second wave around 1.45 V/SCE and after that by the current wall
z E-mail: huc@ccr.jussieu.fr due to oxygen evolution. The effect of increasing chloride concen-

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 978 Journal of The Electrochemical Society, 146 (3) 977-983 (1999)
S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

tration is illustrated by Fig. 2a and b. For 0.5 M LiCl, the change in

mass per cycle is much lower than for 0.1 M LiCl (Fig. 1a) and pre-
sents a quasi-reversible character. Note that there is a small gain in
mass during the scan from 0.8 to 1.2 V/SCE and a loss of mass from
1.2 to 1.7 V/SCE. The latter can be correlated with the current wave
clearly visible in Fig. 2b, which can be assigned to chlorine evolu-
tion. During the reverse scan, the initial mass is partially recovered.
At low chloride concentration, 0.1 M or less, it is possible to obtain
the corrosion rate directly by numerical differentiation of the mass-
potential curve, as done previously.9 This procedure is illustrated by
Fig. 3 and 4 for the 1 M LiClO4 and 0.1 M LiCl 1 0.9 M LiClO4
solutions, respectively. These figures give evidence for the peak-
shaped profile of the corrosion rate. For the 1 M LiClO4 solution,
just one contribution, centered at 1.18 V/SCE, is observed. A more
complicated behavior is depicted in the presence of 0.1 M LiCl, with
a first peak around 1.2 V and a second one around 1.45 V/SCE. In
stationary conditions (except the first cycle), the height of the first
peak decreases at increasing chloride concentration.
The differentiation procedure cannot be applied efficiently to the
data of Fig. 2a, because of the too small change in mass. In such a
situation, the corrosion rate can be estimated by determining the
overall loss of mass after a number of potential cycles divided by the
elapsed time. This procedure is the only one applicable for most of
the data presented here, at high chloride concentration and/or in the
case of the SnO2-Q electrodes. Figures 5a and b show the behavior
of SnO2:Sb (1%)-Q electrodes in 1 M LiClO4. On quartz substrate,

Figure 2. Behavior of a SnO2:Sb (1%)-Ti electrode in 0.5 M LiCl 1 0.5 M

LiClO4 at pH 8: (a) Electrogravimetric response showing a quasi-reversible
character; (b) current-potential curve. Scan rate: 3 mV?s21. Mark (F) indi-
cates the starting point.

the current starts slowly at 1.2 V/SCE, without a defined structure

(Fig. 5b). The loss of mass is significant only at potentials more pos-
itive than 1.4 V/SCE (Fig. 5a). At large Cl2 concentrations ($1 M),

Figure 1. (a) Electrogravimetric behavior and (b) current-potential curve of

SnO2:Sb (1%)-Ti in 0.1 M LiCl 1 0.9 M LiClO4 at pH 8. Scan rate: 3 mV
s21. The two resolved current waves before the oxygen wall originate from Figure 3. Corrosion rate profile obtained by numerical differentiation of the
radical formation and chlorine evolution, respectively. The mark (F) corre- electrogravimetric response for a SnO2:Sb (1%)-Ti electrode in 1 M LiClO4
sponds to the starting point for the potential scan. at pH 8. Scan rate: 5 mV s21.

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 146 (3) 977-983 (1999) 979
S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

Figure 4. Corrosion rate profile as in Fig. 3 for SnO2:Sb (1%)-Ti but in 0.1 M
LiCl 1 0.9 M LiClO4 at pH 8. Scan rate: 3 mV s21. The peak height is much
higher during the first potential scan.

a gain of mass is always observed, followed by the loss of mass asso-

ciated with the current wave for chlorine evolution (Fig. 6a and b).
The change in mass is reduced from 0.5 to 0.15 mg cm22 when the
Sb content is increased from 1 to 4% in 1 M LiCl.

Figure 6. Behavior of SnO2:Sb (1%)-Q in concentrated LiCl solutions at

pH 8: (a) Electrogravimetric response in 1 M LiCl showing a quasi-reversible
character associated with a very small mass change (scan rate: 1 mV s21); (b)
current density-potential curve in 3 M LiCl at pH 8 (scan rate: 3 mV s21); the
current wave originating from chlorine evolution is clearly resolved. (F) indi-
cates the starting point.

The electrogravimetric behavior of SnO2:Sb (1%)-Q and

SnO2:F-Q in natural seawater ([Cl2] 5 0.56 M; pH 5 8.2) is depict-
ed in Fig.7. In both cases, the electrode mass decreases during each
potential cycle, with a rate decreasing with time. Both electrodes
behave similarly. The mean corrosion rate initially reaches
1.3 ng s21 cm22 during the first 30 min, but goes down to
0.05 ng s21 cm22 after 2 h and 0.02 ng s21 cm22 after 4 h. The I-V
curves of Fig. 8, show that the SnO2:F-Q electrode is much more
passivated (current density lower than 40 mA/cm2 at 1.7 V/SCE)
than the SnO2:Sb (1%)-Q one (<500 mA/cm2 at 1.7 V). In fact, for
SnO2:F-Q, the chlorine wave is shifted to 2.7 V/SCE, which corre-
sponds to a band bending close to the bandgap width, and oxygen
evolution occurs at more than 3 V/SCE (see the inset in Fig. 8). To
check the influence of seawater, the electrogravimetric behavior of
SnO2:Sb (1%)-Q has been also studied in a 0.5 M LiCl 1 0.5 M
LiClO4 solution at pH 8 for a number of potential cycles between 0.3
and 1.7 V/SCE. The mass-time behavior is similar to the one in
Fig. 7; the only difference is that the current wave for chlorine evo-
lution is now well distinguishable on the contrary what happens in
natural seawater for which [Cl2] 5 0.56 M, very close to 0.5 M
Figure 5. Behavior of a SnO2:Sb (1%)-Q electrode in 1 M LiClO4 at pH 8: (Fig. 8). The difference may be attributed to the complex composi-
(a) Electrogravimetric response showing that a loss of mass only occurs at tion (of organic and inorganic components) of seawater.
potentials above 1.3 V/SCE; (b) Current density-potential curve; the current In Fig. 9, the values of the mean corrosion rate measured for var-
wave corresponding to chlorine evolution is not resolved. Scan rate: ious SnO2 electrodes, when cycling the potential between 0.3 and
1 mV s21. (F) indicates the starting point. 1.7 V/SCE, are given as a function of chloride concentration. The

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 980 Journal of The Electrochemical Society, 146 (3) 977-983 (1999)
S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

Figure 7. Electrogravimetric behavior in the case of SnO2:Sb (1%)-Q and Figure 9. Corrosion rate of various SnO2 electrodes as a function of chloride
SnO2:F-Q in natural seawater. Scan rate: 1 mV s21. Potential sweep was concentration. For seawater, the value corresponds to the final slope of the
between 0.3 and 1.7 V/SCE. mass-time curve in Fig. 7.

large values (>1 ng s21 cm22) for SnO2:Sb (1%)-Ti and SnO2-Au at Discussion
low chloride concentrations (<0.3 M) correspond to the maximum of The voltammetric behavior of SnO2 in contact with chloride con-
the corrosion rate peak (Fig. 3 and 4 and Ref. 9). Figure 10 shows a taining solutions at pH 8 is characterized in the anodic range by two
C p22 (V) Mott-Schottky diagram for SnO2:Sb (1%)-Q in contact current waves preceding oxygen evolution. The first one is only vis-
with a 0.5 M LiCl 1 0.5 M LiClO4 solution at pH 8, recorded at ible in the 1 M LiClO4 solution, at low Cl2 content. It was shown to
31 Hz because of the large values of the interfacial capacitance. A be correlated with the peak of the corrosion rate situated at
linear plot is obtained at potentials less positive than 1.0 V/SCE as 1.18 V/SCE. This correlation is particularly well evidenced for
expected for a n-type semiconductor in the depletion regime. From SnO2-Au and SnO2-Ti, whatever the type of doping, demonstrating
the slope, the charge carrier density is found to be 1.2 3 1021 cm23, that electrochemically produced species contribute to the corrosion
in very close agreement with Hall measurements (8.5 3 1020 cm23 process. In the potential window considered, these species are prob-
from Ref. 9). The flatband potential is VFB 5 20.74 V/SCE, indicat- ably OH• and Cl• radicals, resulting from the electro-oxidation of
ing that SnO2 is strongly depleted for chlorine and oxygen to be OH2 and Cl2 ions. By balancing the passed electrical charge and the
evolved, especially for F doping. Above 1 V/SCE, in the region lost mass, it was demonstrated that two electrons are consumed per
where corrosion occurs, the C p22 (V) is strongly disturbed. This re- Sn-O group lost by corrosion.9 It was proposed that breaking of sur-
sult suggests that the potential distribution through the interface is face bonds takes place through electron capture by the very oxidiz-
modified because of surface charge effects induced by corrosion. ing radical species.
They give rise to a positive band-edge shift of about 1.7 V if we The second current wave arises around 1.45 V/SCE, except in the
assume that the flatband potential for the corroded surface is VFB* 5 case of SnO2:F-Q for which it is rejected to 2.7 V/SCE. The ampli-
1 V/SCE (Fig. 10). tude of this wave is found proportional to the square root of the
potential sweep rate, and also proportional to the Cl2 bulk concen-
tration. This wave, controlled by diffusion of chloride ions, can be

Figure 8. Current density-potential curves for SnO2:Sb (1%)-Q and SnO2:F-

Q in natural seawater. Sb-doping produces a much more active surface for
chlorine and oxygen evolution than F doping. The inset shows the I-V curve Figure 10. Mott-Schottky plot for SnO2:Sb (1%)-Q in a 0.5 M LiCl 1 0.5 M
for the SnO2:F-Q electrode when scanning the potential up to 4 V/SCE. Scan LiClO4 solution at pH 8, recorded at 31 Hz. The flatband positions for the
rate: 1 mV s21. uncorroded surface (VFB) and for the corrroded surface (V *FB) are indicated.

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 146 (3) 977-983 (1999) 981
S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

assigned to the chlorine evolution reaction. Let us recall that at a

support of this proposal, accumulation of OH• radicals at the SnO2
basic pH chlorine is readily transformed into hypochlorous species.
surface was demonstrated during water discharge, on the contrary of
In literature, many papers were devoted to this very important reac-
other oxide materials as PtOx or IrO2. 4 The next step corresponds to
tion [Ref. 11, 12 and references therein]. Elementary steps proposed
a reversible charge separation, giving rise to a positively charged sur-
for chlorine evolution reaction involve discharge of chloride ion at a
face site X1 s, with the release of a Cl2 or a OH2 ion. This assump-
catalytic surface site, X, producing a surface intermediate X-Cl, fol-
tion is necessary to account for the positive band-edge shift taking
lowed by recombination or electrochemical desorption of the X-Cl
place during corrosion. Oxygen and chlorine evolutions are assumed
species. These steps can be written as follows
to occur through electro-oxidation of OH2 and Cl2 ions at the X-
X 1 Cl2 r X-Cl 1 e2 OH and X-Cl surface sites, respectively. Every surface intermediates
(X-OH, X-Cl, and X1 s) are characterized by a surface concentration,
2 X-Cl r Cl2 1 2X depending on reaction kinetics. Oxygen evolution is also expected to
or occur by direct recombination between OH• and OH2 species to
account for the oxygen current wall. The rate-determining corrosion
X-Cl 1 Cl2 r X 1 Cl2 1 e2 reaction is assumed to be a purely chemical one which would pro-
ceed by the combined attack of specific corrosion sites, defined as
Note that the small gain of mass observed in chloride-concen-
the X1 s charged surface sites, by OH•/Cl2 or Cl•/OH2 species. The
trated solutions is of the order of one or two monolayers of chlorine
breaking of a Sn–O bond is stabilized by the formation of new
atoms at the oxide surface, assuming a surface concentration of 1015
Sn–OH and Sn–O–Cl surface bonds. The final corrosion product
Cl atoms per cm2 as in Ref. 11. Furthermore, taking into account the
could be a variety of tin hydroxochloride compounds, as it was iden-
quasi-reversible change of mass during a potential cycle, the idea of
tified by reflexion high energy electron diffraction at pH 11.9 The
a surface intermediate of the X-Cl type is well supported by our
concept of surface bond breaking through electron capture by an oxi-
QCM data (see, for instance, Fig. 2a and 6a). The above proposal for
dizing agent has been considered as a corrosion mechanism, for
the overall chloride/chlorine reaction is taken into consideration in
instance in the case of photocorrosion of n-type III-V semiconduc-
the complete reaction scheme given in Fig. 11 to account for the
tors13 or that of ITO chemical etching by undissociated halogen
electrogravimetric behavior of SnO2 at anodic potentials. Two sym-
acids.14
metrical pathways are considered: (i) the hydroxide-oxygen pathway
and (ii) the chloride-chlorine pathway. They involve electrochemical At anodic potentials, the SnO2 surface is depleted and electrons
and chemical steps. For every case, the first step corresponds to the are repelled from the surface to the oxide bulk at a distance equal to
OH• and Cl• radical production, followed by an adsorption step lead- the space-charge region width. Because of the large carrier concen-
ing to the formation of X-OH and X-Cl surface intermediates. In tration, this distance is low, about 2-3 nm. As previously established,
charge transfer can occur at the level of the bulk Fermi level by tun-
neling through the space-charge region.7 Then, in the model, the
transfer rate constants are taken as proportional to a tunnel probabili-
ty, controlled by the surface potential barrier. The latter can be as large
as 2 V in the absence of corrosion and to be reduced at less than 0.5 V
if there is corrosion producing a positive band-edge shift. This is the
key point to understanding why the SnO2-electrolyte interface can be
blocked or active with respect to anodic charge transfer reactions.
Another important point in the model is that the potential distri-
bution between the semiconducting oxide and the double-layer re-
gion is taken into account, knowing that it is strongly influenced by
the corrosion reaction. A consequence is that the electron transfer
reactions are only activated by the potential drop through the double
layer and not by the total potential drop through the interface.15
Within all these assumptions, considering a first-order kinetics
for all the reactions, the proposed reaction scheme is able to account
for (i) the I-V behavior, (ii) the peak-shaped corrosion rate, (iii) the
tendency for the corrosion rate to decrease when increasing the bulk
chloride concentration, and (iv) a potential dependent mass variation
due to the change in surface concentration of adsorbed OH• and Cl•
species. Some details about the numerical calculation are given in
the Appendix. Figure 12 gives a comparison of experimental and
simulated corrosion rate data, according to the proposed model,
reproducing the marked increase at low chloride concentrations. It
must be remarked that at high corrosion rates, as is the case at low
chloride concentration, any change in mass is essentially due to the
release of tin atoms and is weakly influenced by the change in sur-
face hydration and/or adsorption of Cl and OH species.
The results presented here for SnO2:F-Q and SnO2:Sb (1%)-Q
point out a marked difference between the two types of doping, F or
Sb. The SnO2:F-Q electrode is completely passivated up to poten-
tials corresponding to a band bending close to the SnO2 bandgap
width. This is a strong indication for the absence of any electronic
level (surface states) in the gap of F-doped SnO2, contrary to Sb-
doped SnO2. It has been shown, by X-ray photoelectron spec-
troscopy (XPS),16 by XPS and Auger spectroscopy,17 and by analy-
Figure 11. Proposed reaction scheme describing the competition among tin
sis of transport properties,18 that antimony is present as Sb51 (in
dioxide corrosion, chlorine, and oxygen evolution, initiated by the electro- substitution of Sn41 in the lattice) and Sb31 in interstitial positions,
chemical production of OH• and Cl• radicals. largely in excess of the Sb51 donor form. Interstitial Sb31 species

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 982 Journal of The Electrochemical Society, 146 (3) 977-983 (1999)
S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

when cycling the potential, which are essentially related to the ad-
sorption/desorption of Cl and OH species.
The radical attack creates positively charged surface sites, shift-
ing the bandedges to largely positive potentials. Decreasing the
potential barrier makes charge transfer easier and SnO2 electrodes
become more active. Independently of the doping, the presence of
metallic traces favors radical production and consequently enhances
both corrosion and chlorine evolution reactions. Furthermore, it is
shown that SnO2 films are rather stable in natural seawater during
chlorine evolution, favoring this latter reaction to be used as a possi-
ble antifouling treatment of immersed optical systems.
Acknowledgments
This work was supported by IFREMER-Brest under contract 96
2 342 401. We would also like to thank G. Folcher for preparing the
tin dioxide specimens.
C.N.R.S. assisted in meeting the publication costs of this article.

Appendix
Figure 12. Simulated values of the corrosion rate (- - -) showing the marked The model given in Fig. 11 yields no analytical solution and has to be
increase at low chloride concentrations, with comparison to the experimental solved numerically by means of an iterative procedure, using for instance the
data for SnO2:Sb (1%)-Q (j). Mathcad software from Mathsoft. For calculation, first-order kinetics is
assumed for all the reactions. In the following, c1, c2, and c3 represent the sur-
face concentrations (in cm22) in X-OH, X-Cl, and X1 s, respectively, and
create (i) a large density of crystallographic defects which limit the
[OH•], [Cl•] the volume concentration of radical species electrochemically
carrier mobility, and (ii) a state density in the SnO2 gap, able to sus- produced at the surface. The rate constants are in s21, except ka1, ka2, kt1, and
tain interfacial electron transfer. Figure 5b shows that current starts kt2 which are in cm3 s21.
at potentials positive of 1.2 V/SCE, i.e., for a band bending of about Under stationary conditions, the concentrations can be written as a func-
2 V. It can be concluded that Sb-doping introduces surface states tion of the rate constants and the current densities i1 and i2 corresponding to
located 2 eV below the bottom of the conduction band. In the same the OH• and Cl• production respectively
way, for SnO2-Au and SnO2-Ti, the presence of foreign metal atoms
i1 / q
(Au or Ti), brought at the oxide surface during the fabrication [OH • ] 5 [A-1]
process, also create surface states. In agreement with the data on ka1 co 1 kt1 c3
SnO2-Au from Ref. 9, Fig. 1b shows that the anodic current for the
oxidation of chloride and hydroxide ions starts close to the thermo- i2 / q
[Cl • ] 5 [A-2]
dynamic potentials of both reactions, about 1 V/SCE.19 The absence ka 2 co 1 kt 2 c3
of overpotential would mean that these states are located in the upper
half of the SnO2 bandgap. ka1 co [OH • ] 1 kd1 c3
[c1 ] 5 [A-3]
SnO2 electrodes are originally passivated with respect to chlorine kc1 1 kox1
and oxygen evolution reactions, except in the presence of foreign
metal atoms (for instance, Sb, Au, or Ti) at the surface. In the latter ka 2 co [Cl • ] 1 kd 2 c3
[c2 ] 5
case, electrochemical oxidation of OH2 and Cl2 ions can develop kc2 1 kCl
[A-4]
close to the thermodynamic potential and give rise to the complex
reaction scheme in Fig. 11, coupling chlorine evolution, oxygen evo- c3 is solution of the following nonlinear equation
lution, and corrosion reactions. From capacitance data (Fig. 10), it is
shown that a positive band-edge shift is produced by corrosion: the kc1c1 1 kc2c2 2 (kd1 1 kd2 1 kt1[OH•] 1 kt2[Cl•])c3 5 0 [A-5]
surface potential barrier is reduced from almost 2 V to a few hun-
which has to be solved numerically. co represents the total number of avail-
dreds of millivolts and the transfer rate constants are increased by able sites at the oxide surface and q is the elementary charge.
about four orders of magnitude. Actually, electrode activation is The applied voltage, h, through the interface is the sum of two compo-
achieved because corrosion through the formation of positively nents, hsc, through the space-charge region in the oxide, and hh, through the
charged surface intermediates. SnO2:F-Q electrodes remain blocked double layer in the electrolyte. The change dhsc corresponding to a change
to a band bending of the order of the bandgap. In such a case, acti- dh of the total applied voltage h reads
vation may result from a different mechanism, like field-assisted
emission of electrons from the valence band. Ch (hsc )
dhsc 5 dh [A-6]
Csc (hsc ) 1 Ch
Conclusions
where Csc(h) is the space-charge layer capacitance of the oxide and Ch is the
At potentials positive of the flatband potential, the F-doped SnO2- double-layer capacitance. Every current density components are expressed as
aqueous solution interface is passivated in a potential window close to
the SnO2 bandgap width. The blocking character, even for a highly  F (h) 
doped material, arises from (i) the existence of a high surface poten- i(h) 5 Io exp 2 b  exp(bh) [A-7]
 Eoo 
tial barrier, and (ii) the absence of surface states as a relay for electron
transfer. Sb-doping, through the presence of interstitial Sb31 species, The last exponential term is the activation factor depending on the applied
produces electronic states close to the midgap allowing anodic charge potential, h. Electrons are assumed to be exchanged at the Fermi energy level
tranfer reactions to occur at a low rate. The electrogravimetric studies of the oxide in the bulk. The first exponential term stands for the control of
presented here demonstrate that chlorine evolution is coupled with a charge transfer by tunneling in the oxide. Eoo is a constant characteristic of the
corrosion mechanism involving the breaking of Sn–O surface bonds electronic properties of the oxide and Fb is the potential barrier seen by the
by electrochemically produced OH• and Cl• radicals. electrons. Fb changes with the applied potential and also because of surface
charging effects due to the formation of charged corrosion intermediates Xs 1
At low chloride concentration, the loss of mass is fully deter-
mined by the tin oxide dissolution. On the contrary, at large chloride c3
F b m11 5 F bo 1 ( dhsc ) m 2 q [A-8]
concentration, quasi-reversible, small mass changes are observed Ch

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 Journal of The Electrochemical Society, 146 (3) 977-983 (1999) 983
S0013-4651(98)03-009-2 CCC: $7.00 © The Electrochemical Society, Inc.

Equation A-8 expresses the barrier height at step m 1 1 as a function of step 7. J. Bruneaux, H. Cachet, M. Froment, and A. Messad, Electrochim. Acta, 39, 1251
m, and Fbo stands for the initial potential barrier for h 5 0. The total current, (1994).
8. A. Kraft, H. Henning, A. Herbst, and K. H. Heckner, J. Electroanal. Chem., 365,
itot, reads 191 (1994).
itot(h) 5 i1 1 i 2 1 qkox c1 1 qkox2[OH•] 1 qkClc2 9. H. Cachet, M. Froment, and F. Zenia, J. Electrochem., Soc., 143, 442 (1996).
[A-9]
1 10. G. Folcher, H. Cachet, M. Froment, and J. Bruneaux, Thin Solid Films, 301, 242
Finally, the corrosion rate, vcor, is expressed as (1997).
11. L. Tomcsanyi and A. De Battisti, Electrochim. Acta, 41, 2917 (1996).
[ ]
vcor 5 kt1 [OH • ] 1 kt 2 [Cl • ] c3 M (Sn – O) [A-10] 12. B. E. Conway and B. V. Tilak, in Advances in Catalysis, Vol. 38, p. 99, D. D. Eley,
H. Pines, and P. B. Weisz, Editors, Academic Press, San Diego (1992).
13. K. W. Freese Jr., M. J. Madou, and S. R. Morrison, J. Electrochem. Soc., 128, 1527
where M(Sn–O) represents the mass of a Sn–O bond. (1981).
14. J. E. A. M. van den Meerakker, P. C. Baarslag, and M. Scholten, J. Electrochem.
References Soc., 142, 2321 (1995).
1. Ch. Comninellis and A. Nerini, J. Appl. Electrochem., 25, 23 (1995). 15. P. Allongue and S. Blonkowski, J. Electroanal. Chem., 317, 77 (1991).
2. C. Iwakura, M. Inai, T. Uemura, and H. Tamura, Electrochim. Acta, 26, 579 (1981). 16. B. Correa-Lozano, Ch. Comninellis, and A. De Battisti, J. Electrochem. Soc., 143,
3. C. S. Fugivara, P. T. A. Sumodjo, A. A. Cardoso, and A. V. Benedetti, Analyst, 121, 203 (1996).
541 (1996). 17. H. Cachet, J. Stoch, and A. Kowal, p. 493, in EuropaCat-3, Book of Abstracts Vol.
4. Ch. Comninellis, Electrochim. Acta, 39, 1857 (1994). 2, J. Haber, K. Dyrek, and J. M. Lagan, Editors, Polish Academy of Sciences, Insti-
5. B. Correa-Lozano, Ch. Comninellis, and A. De Battisti, J. Appl. Electrochem., 27, tute of Catalysis and Surface Chemistry, Krakow, Poland (Sept 1997).
970 (1997). 18. A. Messad, J. Bruneaux, H. Cachet, and M. Froment, J. Mater. Sci., 29, 5095 (1994).
6. J. M. Roelens, C. Nury, M. Rumeau, J. Vedel, and M. Levart, Fr. Pat. 88 14625 19. A. J. Bard, R. Parsons, and J. Jordan, Standard Potentials in Aqueous Solution,
(1988). Chap. 4 and 5, IUPAC, Marcel Dekker, New York (1985).

Downloaded on 2016-10-07 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).