ELECTROCATALYSIS IN THE ANODIC EVOLUTION OF

OXYGEN AND CHLORINE?

S. TRASAW

Department of Physical Chemistry and Electroohemistry, University of Milan. Via Venezian 21, 20133
Milan, Italy

(Receiwed 15 May 1984)

Abstract- Requisites for electrode materials to be suitable for technological applications are outlined and
discussed. Oxides with metallic or quasi-metallic conductivity meet these requirements best. Most of these
electrodes are prepared by thermal procedures. It is shown that the temperature of preparation affects the
catalytic activity through the surface area and the chemical composition (non-stoichiometry). The co-
variation of these parameters is best followed in situ by voltammetric curves and point of zero charge
measurements. Examples are given for pure RuO,, IrO,. CosO, and 1rQ-t RuOz mixtures. Kinetic and
mechanistic detailsare discussed for 0s evolution on RuO,,IrO, and CosOI and for Cl, evolution on RuO,
and Co30,. Finally, the electrocatalytic properties of different oxides are correlated with the energy change
involved in the lower * higher valency state transition. Experimental data for both O2 and Cl* evolution can
thus be organized into a “volcano” curve enabling predictive interpolations to be made.

The practical importance of such electrode processes Table 1. Electrodes for technological applications
as the anodic evolution of chlorine and oxygen can requisites
hardly be overemphasized. Cl, production, one of the
largest technological application of electrochemistry, (1) High surface area
is based on the direct anodic discharge ofchloride ions. (2) High electrical conduction
O2 evolution is an unavoidable side reaction in many (3) Good electrocatalytic properties
(41 Long-term mechanical and chemical stability at
anodic processes; moreover, it is the most common the support/active layer and at the active
anodic reaction coupled with most of the cathodic layer/solution interface
processes occurring in aqueous solutions (water elec- (5) Minimized gas bubble problems
trolysis, metal electrowinning, etc.). In any of these Enhanced selectivity
cases the appropriate selection of the electrode ma- ;:; Availability and low cost
terials is a prerequisite for the optimization of the (8) Health safety
electrolysis parameters.
The main requisites needed for an electrode material
choice is based on a compromise between activity,
to be useful for technological applications are listed in stability and cost[5]. While the prototype of this class
Table 1. The suitability of a material for practical
of oxides has been RuO,[6] which is most active but
problems in a result of the combination of all the
fairly expensive, the applied research is now in the
features mentioned in the table, but points (2) and (4) direction of replacing very good but expensive elec-
are those usually prevailing as large current densities
trocatalysts with cheaper materials whose electrocata-
are to be used. Nevertheless, fundamental research
lytic properties are perhaps not so excellent[7].
insists on the study of point (3) because this involves
Accordingly, an excellent electrocatalyst whose per-
the largest body of conceptual interdisciplinary
formances deteriorate with time is technologically less
implications. interesting than a material of lower electrocatalytic
The requisites outlined in Table 1 are best met by
qualities but of much better long-term stability. It may
transition metal oxides with metallic or quasi-metallic thus happen that in some applications under particu-
conductivity[l]. The discovery[2] that the perform- larly severe conditions, such as water electrolysis in
ances of some precious metal oxides are far better than SPE cells, a Poorer and much more expensive elec-
those of the corresponding precious metals has pro- trocatalyst such as IrOz is mixed with RuO, to
voked a technological jump in the large electrolytic
improve the stability of the latter[8].
industry, particularly in the chlor-alkali cell tech- Strictly, the term “electrocatalysis” implies the in-
noIogy[3]. The use of these oxide electrodes has later fluence of the electrode material on the rate of the
been extended to water electrolysis[4]. electrode reaction (primary effects, in Parsons de-
It is exceptional that all the requisites in Table 1 can fmition)[9] ie the kineticand mechanistic effects of the
be optimized in one electrocatalyst. More usuahy, the bonds formed by reactants, products and/or inter-
mediates with the electrode surface. This term has
+ Introductory lecture given al the Symposium on however acquired a wider meaning on a practical basis:
“Electrccatalysis: Practice, Theory and Further if a material shows a lower overpotential because it is
Development”, I>16 September 1983, Neunkirchen a. mainly a better conductor, this is as a whole regarded
Brandt, F. R. G. as a better electrocatalyst. Although this usage has a
I503
1504 S.T~~shln

Table 2. Oxides used as electrodes

Rut&-type RuO,, IrO,, MnO,, Pb02 Mixed oxides with

PtO,, OsO,, ReO, SnO,, Ti02, Fe203
Doped oxides
sno,, MnO,, Pbo,
Spine]-type CO,% Doped oxides
MCo,O, M = Ni, Mn, Cr. .. co304
Fe304
MFe,O, M = Ni, Co, Mn . ..
Perovskite-type NiM20, M = La, Pr, Nd
M; _,Sr,MO, Mn, Co, Fe, Ni X = 0; 1
j”=
M’ = La, Nd LaCoO,, LaNiO,
SrFeO,
Other oxides NiO,, PdO, M,WO,, Bi?M,O, M = Rh, Ru
Li,_,Pt,O,, SrPd,O,, Tl,O

B = Sb, Nb, Ta

practical justification, it is extremely misleading to mix Table 3. Factors of electrocatalysis at oxides

up different concepts. This paper will be restricted to
the analysis of purely catalytic properties of conduct- - Chemical nature of the catalyst
ing oxides. - Morphology (dispersed state, crystal size, crystallinity,
Fundamental .investigations on this topic have lattice distorsion, . . .)
- Non-stoichiometry (ionic defects, electronic defects,
grown exponentially in the past decade[l, lo], and
solid-state redox properties, . . .)
have spread to a larger and larger number of simple - Magnetic properties
and mixed oxides. Table 2 summarizes most of the - Band structure of the oxide
systems that have been studied thus far, but the list is - Surface electronic structure
incomplete because not all patent literature has been - Geometric factors
taken into account. Those on which the research work - Crystal-field stabilization energy
is presently more active are RuOz (of course), Ru02 - Synergetic effects (mixed and doped oxides]
-+ IrO, (SPE water electrolysis)[S], NiCo,O, (0,
evolution from alkaline solution)[4], CO,O~ (the
same, and also as possible anode in brine electrolysis to latter is the route preferred7 to prepare electrodes for
replace RuO,)[7], some perovskites[l 11. technological applications. The temperature and the
Oxides are seldom used alone in practical appli- procedure of preparation therefore turn out to be
cations[l, lo]. Usually they are more or less heavily additional variables which are very often neglected.
doped or mixed with less active oxides of higher Varying these parameters enables only structural fac-
chemical stability. In this context, also non- tors to be modified while the chemical ones are kept (to
electrocatalytic oxides are of great interest, notably some extent) constant. This paper is focussed on the
TiOZ and SnO,. effect of the conditions of thermal preparation on the
The essential aim of electrocatalysis as a science is to electrocatalytic properties of some of the most in-
establish a predictive basis for the design and the vestigated oxides (RuOZ, IrO,, Co&&, NiCo,O,) for
optimization of catalysts. Predictions can be for- Cl, and O2 evolution. The essential aim is to outline
mulated only if the factors which are responsible for the present understanding of the behaviour of this
the electrocatalytic properties are identified. This is the class of electrode materials, and to illustrate some of
fundamental goal of research. Various factors can be the most intriguing aspects which have received yet no
envisaged to govern the catalytic activity of oxide conclusive explanation.
electrodes. Singling out which is more important (if Oxide electrodes are usually prepared by thermal
any) in the various cases is the most difficult task of decomposition of suitable salts on a conducting sup-
research. port, mainly Ti, ‘although other preparation pro-
Electronic and geometric factors are usually recog- cedures have also been devised[l]. Figure 1 shows the
nized to govern catalysis. Those listed in Table 3 can be activity of IrOz for O3 evolution as a function of the
distinguished into chemical (based on the compo- temperature of preparation. Thereaction rate is seen to
sition) and structural (mainly based on the mor- drop as the temperature of preparation increases. This
phology) factors. The common way in which these is a common outcome, as Fig. 2 shows: also the activity
factors are investigated is by following the variation in of Co,O, for Clz evolution decreases steeply with the
the reaction rate as the nature of the catalyst is temperature of preparation.
changed. Oxides can be prepared anodically or therm-
ally. The stability of anodically grown oxides is as a + If in situ activation is used, the anodic oxide is necessarily
rule poorer[12] than that of oxides prepared by the active material [13-151. The question of stability, under
thermal decomposition of appropriate precursors. The similar circumstances, rests on different concepts.
 Anodic evolution of oxygen and chlorine 1505

I
‘00
h .,
400
T/ “C
800

Fig. 3. Variation of the BET surface area of ColO, powder

as a function of the calcination temperature.

powder with the temperature of calcination[ 161 para-

Fig. 1. Electrocatalytic activity in 0, evolution of Ti/IrO, llels that of the electrocatalytic activity. This is however
electrodes prepared by thermal decomposition of (0) iso-
not sufficient to conclude that all the variation in Figs 1
propaool and (0) aqueous sofutions of IrCt, at various
temperatures. and 2 is due to surface area effects.
The most direct way of investigating the surface
propertiesof oxide electrodes is to record the voltam-
metric curve in a potential range where neither Oz nor
H, are evolved[17, IS]. Figure 4 shows a typical voltam-
metric curve of Co,O,/Ti electrodes in alkaline sal-
0
utions. Although some of the charge spent to change
the electrode potential may be involved in bulk
modifications[4], conditions can be chosen ensuring
that only surface processes are being investigated. The
charge obtained by integration of the voltammetric
curves (henceforth denoted with q*) is thus that spent
“: in surface modifications and can be assumed to be
s proportional to the surface area exposed to the
1
\ solution. If each surface site undergoing oxidation
0

Fig. 2. Electrocatalytic activity of (0) Ti/RuO,/Co,O, and

(I) Ti/CosOn electrodes in Cl, evolution from 5 mol dm ’
NaCl +O.Ol moldm-’ HCI solution as a function of the
calcination temperature..

A number of parameters can actually change as the

calcination temperature is varied. The most outstand-
ing is certainly the surface area of the electrode. The
active layer consists ultimately of a sintered powder.
Depending on temperature, the crystallite size may
vary and so does the surface area. This is not easy to
prove directly with electrodes but it can be straightfor- Fig. 4. Voltammetric curves in KOH solution of (---)
wardly demonstrated with powders. Figure 3 shows Ti/Co304 and (-) Ti/RuOJCo,O, electrodes calcined at
that the variation of the BET surface area of Co,O.+ 500°C.
 S. TRA~AI-TI

nation is a necessary, but not sufticient reason, to

conclude that all of the activity variation is due to
surface area effects. Other parameters may co-vary
with temperature. The most important one is non-
stoichiometry[20]. IrOl and RuOa layers have been
analysed[21] by Rutherford backscattering spectro-
scopy (RBS) and the results are shown in Fig. 6. It can
be seen that both oxides are non-stoichiometric, but in
opposite directions. RuO, is oxygen deficient
(RuOz 6), whereas IrO, shows some excess of oxygen
(IrOz + a). Co,04 is known[22] to be metal deficient
with higher valency cations balancing the metal
vacancies.
Residues of the thermal decomposition route may
be present in the lattice of some of these oxides. Tn
particular RuOz (and IrO,) are prepared by thermal
decomposition of RuC13 {and IrCls): some Cl remains
in the bulk[23] (it is unclear if trapped or in the lattice
positions replacing oxygen) and the Cl to Ru ratio
decreases with the calcination temperature in a way
which again parallels the pattern shown in Fig. I. This
is clearly illustrated in Fig. 7. Thus, the experimental
results shown so far cannot say unambiguously which
Fig. 5. Voltammetric charge as a function of the calcination of the various factors is responsible for the variation of
temperature of Ti/IrO, electrodes prepared from (0) iso- the catalytic activity of oxides.
propanol and (0) aqueous solutions of IrCl,. Plots of the activity against charge q*, the elec-
trochemically active surface area, are usefut to distin-
(and reduction) is assumed to be an active site for the guish between surface area effects and other possible
anodic reaction, then q* can be envisaged to be a factors. If only the surface area changes with the
measure of the electrochemically active surface area. It calcination temperature, then a linear plot of unit slope
is obvious that this parameter cannot be used to should be observed. Otherwise, the experimental
compare the surface area of different oxides because points should deviate from this simple linear depen-
the density of chargeable surface sites may be (and as a dence. It has been shown elsewhere[20] that in the case
rule is) different for different oxides. Figure 5 shows of oxygen evolution on RuO, electrodes from acid
that the charge q* of IrO,/Ti electrodes changes with solutions, the reaction order with respect to the surface
the preparation temperature[19] in a way resembling active sites may be higher than one for highly defective
the pattern of variation of both the BET surface area layers. This indicates that, as the calcination tempera-
and the elcctrocatalytic activity. Even more intriguing ture is varied, also the nature of the surface active sites
is the fact that for a given calcination temperature, the changes along with the morphology of the layer.
value of q* depends on the nature of the solvent on The Tafel slope is another experimental parameter
which the precursor is dissolved. This may probably be which is sensitive to the structure of the surface.
related to differences in crystallization kinetics. However, the Tafel slope is an intensive parameter and
The way q* changes with the temperature of calci- is only sensitive to the reaction mechanism. Therefore,

f/k&

Fig. 6. Rutherford backscattering spectra of Ti/RuO, and Ti/IrOz elezrrodes prepared at 400°C.
 Anodic evolution of oxygen and chlorine 1507

sensitive to any variation in the energetic and struc-

tural surface parameters[26].
The experimental techniques to measure the point of
zero charge have been discussed elsewhere[24]. Figure
8 shows that the point ofzero charge of RuOz changes
appreciably with the calcination temperaturet271. This
agrees with and corroborates the experimental evi-
dence[fO] that the reaction mechanism is sensitive to
the preparation procedure and is related to a smooth
change in non-stoichiometry. This result again
stresses the fact that the calcination temperature
provokes real electrocatalytic effects. Further results
have shown that the nature of the support may also
influence the point of zero charge, but this is a more
complex effect which has been described
eIsewhere[28].
Co,O, is as a rule prepared by thermal decompo-
sition of Co(NO& The point of zero charge as shown
in Fig. 9, is in this case insensitive to the calcination
temperature up to about 5OO”C, then it starts to
Fig. 7. Chlorine content of Ru02 layers prepared by thermal
decomposition of RuC13 at various temperatures. (0.m) increase steadily. The situation is different from RuO,
Different samples. because no impurities remain in the lattice of Co,O,.
The chemical nature of the oxide does not change with
the calcination temperature and neither does the pzc+
constancy in Tafel slope does not imply absence of
Since CoJO, is known to be. non-stoichiometric by
electrocatalytic effects. But a variation in Tafel slope is
excess of oxygen, the constancy in the pzc would imply
conclusive evidence that something has changed on the
that some reversibility is probably operative at the least
electrode surface. Oxygen evolution on RuO, takes
at the surface. Oxygen which is released at higher
place with a Tafel slope of 30 mV at high q* values,
temperatures might be taken up again on cooling very
increasing to more than 50mV as q* decreases[20].
slowly. This aspect deserves further experimental
Thus, a modification in the surface concentration (and
investigation.
probably in the structure) of the active sites provokes a
variation in the rate determining step because the
surface-intermediate interactions also change.
The acid-base properties of the surface of oxides are
extremely sensitive to any change in the nature of
surface coordination complexes[24]. Their investi-
gation can give evidence to small changes in the
structure of the surface active sites. Oxide surfaces are
covered in solution by a “carpet” of OH groups which
can behave, for instance, as weak Bronsted acids

-M-OH-t-M-0-+H+.

In terms of Gutmann’s concepts of electron donor-

(1)
4

Lj I
/

xm 400
11
500
r/x
600
‘1
700 800

acceptor interaction[26], a highly electronegative

metal ion draws electrons from the surface OH group Point of zero charge of RuO, powder as a function of
the temperature of decomposition of RuCl,.
with a shortening of the M-O bond but a lengthening
of the O-H bond. The weakening of the O-H bond
results in an enhanced acidity of the surface. An
electropositive metal cation tends to push back elec-
trons in to the OH group with a weakening of the
M-OH bond. These sites may thus behave as weak
(Bronsted) bases

I 1
M-H-+-M+OH- (2a)
t I
or

I I
-M-OH + Hz0 + -M-OH:+ OH-. (2b)
I
The point of zero charge is defined as the pH at which
the oxide surface carries no net electrical charge. It is a Fig. 9. Point of zero charge of Co,O, powder as a function
quantity characteristic of the given oxide and is of the temperature of decomposition of Co (NO,),.
1508 S. TRASA-ITI

imply simply a difference in surface area. Electrodes

prepared by thermal decomposition of non-aqueous
solutions of I&l, deviate from the straight line of unit
slope showing an extra enhancement of the reaction
rate which is presumably related to a different nature
of the surface sites. Higher surface area layers are
presumably more defective. The smaller crystaliites
probably exhibit enhanced edge effects.
The point of zero charge is a very useful quantity to
investigate the surface composition of mixed oxides. It
has been shown that RuO, +TiO, mixed oxides
exhibit adsorption patterns which may imply syn-
ergetic effects[30]. RuO, + IrOz mixed oxides are
under active investigation because of their interesting
performances in SPE water electrolysis cell@].
Enhanced stability of RuO, in the presence of IrOz is
Fig. 10. Electrecatalytic activity of Co,O, electrodes in 0,
evolution from alkaline sotution as a function of the ‘Met-
claimed. Figure 12 shows preliminary data of point of
trochemical active surface area”. zero charge as a function of the composition. Since the
pzr of IrO, is very acid (ca. 0.5)[13], the IrO,-rich
oxides are difficult to study. The pzc is seen to decrease
Figure 10 shows the variation of’ the activity of with Ir02 content more slowly than expected on the
Co,O_JTi electrodes for oxygen evolution as a func- basis of a regular behaviour. This may be due to
tion of q *, The linear dependence of unit slope which is surface segregation of RuOz during the preparation.
observed indicates that no enhancement of the reac- Spectroscopic analysis in another laboratory showed
tion rate takes place beyond surface area effects. This that IrO, can in fact segregate to the surface but the
agrees with the results of’the pzc measurements. The preparation route was different[32] and this is an
steady increase of the pzc at temperatures higher than essential difference. The other possibility is that
about 500°C is probably due to the surface de- cooperative effects modify the simple Iinear rule.
composition of CosO, into Co0 + oxygen[29). Since Further experimental investigation is thus mandatory.
CosO, cannot be deposited on Ti in the high tempera- Chlorine evolution appears to be somewhat less
ture range, the related catalytic effects are difficult to sensitive to the details of the surface structure[33]: to
study. some extent it can be defined as a “facile” reaction[l].
That the morphology of the oxide layer is of Figure 13 shows that the activity of Coj04 for Cl1
paramount importance in imparting the electrocata- evolution does not deviate from the straight line of unit
lytic properties is clearly evident in Fig. ll where the slope when plotted against the voltammetric charge
dependence of the activity of IrO, for oxygen evol- q*. The same has been shown[33] to occur with RuOr
ution is plotted against the surface parameter q*. The electrodes although the high porosity of these active
figure reports the results for the two sets of electrodes layers brings about the problem that part of the inner
already shown in Fig. 5. These results substantiate the surface is not accessible to Cl- ions[34]. Under similar
idea that the different values of q * for two electrodes in circumstances the catalytic activity does not increase
the two sets prepared at the same temperature do not further with q *. Despite its simpler energetic situation,
the anodic reaction of Cl* formation exhibits a number
of different problems. Since it proceeds at high rate on
the more active electrodes (particularly RuOz-based
4 electrodes)[33] complications arise due to supersatu-

4.0 -

Fig. 11. Eleetrocatalytic activity of Ti/lrO, electrodes in O2

evolution from alkaline solution as a function of the “elec-
trochemitil active surface area”. Prepared by decomposition Fig. 12. Point of zero charge of 110, -t RuO, mixed oxides
of IrC13 in (0) isopropanol and (0) aqueous solution. as a function of the content of Ir02.
 Anodic evolution of oxygen and chlorine

Fig. 13. Exchange current of the Cl2 reaction at Co30+

electrodes as a function of the “electrochemical active surface
area”.

ration of chlorine gas in the solution layer adjacent to

the electrode surface[35]. The Tafel slope for a Fig. 14. Reaction rate of Cl, evolution as a function of the
reaction limited by the diffusion of the gaseous activity of chloride in r M N&l+ 0.01 M HCI solutions at
product away from the interfaceE36, 371 is. 30 mV Coa04 electrodes prepared at various temperatures.
which is eaual to the Tafel sloue of the chemical
recombination (Cl + Cl) step. In addition, reaching the
solubility limit of Cl, at the interface at high current
0
densities further distorts the experimental polarization
curve[38] with possible simulations of Tafel slopes of
lower values than RT/ZF. This greatly complicates the
identification of the reaction mechanism.
/

I ,’
/ /

The effect of pH is even more dramatic since it is in ,&’ l -

I
.26OT
fact predicted neither on a thermodynamic nor on a Y ,F’ ,’
5 / I’
kinetic basis. pH effects were observed by Arikado et 7 7’ /
a@393 with RuO, electrodes and were ascribed to Ci- > -1
,
J’ ..
,a. .
.
-300
containing surface complex formation. However, these E
,=
/’
,fX
,’

authors gave no kinetic significance to their obser- ,,’ -’

vations. In particular, they derived the reaction order I’
>’

from i0 us I?,,, plots where E, is the equilibrium /’

,’
potential at constant PC,, and variable acl-. pH effects
were on the contrary ruled out by Denton et aI.[40] for
RuO, + TiO, electrodes.
pH effects are however to be expected on the basis of
the hydroxylated nature of oxides. Simply double layer
effects as proposed by 3ondar et al.[41] are unlikely Fig. 15. Reaction rate of Cl2 evolution at Co,Oc electrodes
because the ionic strength is always very high. Figure prepared at two temperatures as a function of the activity of
14 shows the variation of the activity of Co304 chloride in c M NaCl + 0.01 M HCI + 1 M NaC104 solutions.
for Cl, evolution as a function of the bulk activity of
Cl-. Similar results were also obtained with RuO, by
Krishtalik and co-workers[42]. The most immediate determination of the kinetic parameters for Cl, evol-
interpretation is that the reaction order determination ution at oxides. Involvement of the pH in the kinetics
is affected by self-retarding of the reaction due to of anodic Cl2 formation implies simultaneous dis-
specific adsorption of Cl-[43]. Figure 15 shows that if charge of Cl- and oxygenated particles. This has also
an excess of inert electrolyte is added to the solution, a been suggested by Burke and O’Neill[45],
reaction order of one can be measured in the range of Augunstinsky ei a[.[461 and by Denton et a/.[401 who
low Cl- concentrations. Again the results could be have however proposed that H,O discharge should
explained in terms of ClO, displacing specifically follow the Cl- discharge and be rate determining step
adsorbed Cl- ions and pushing them back into the (on account of the fact that they did not observe any
diffuse layer. pH effect and that no “traditional” mechanism could
The data in Figs 14 and 15 were however collected at explain the observed reaction order of one). More
constant Ht concentration (ca. lo-’ ma1 dmm3). This recently, Harrison et a!.[471 have put forward the idea
does not correspond to constant pH in Fig. 14, whereas that HClO may be an intermediate in the anodic
it does in Fig. 1.5 in the low Cl- concentration range. oxidation of Cl- to Cl, but again no pH effects are
That the pH has kinetic effects on Clz evolution at mentioned.
oxides has been stressed by Krishtalik[44]. Therefore, The examples discussed above show that oxides are
the non-constancy of the solution pH invalidates the usually covered with surface complexes (oxyhydroxy-
1510 S. TRASA~

or chloro-complexes) whose exact nature is on the surface should be calculated from the heat of forma-
other hand not well known. Anodic reactions are tion of the solid phase. However, correlations using
usually envisaged to occur with breakage of similar data have been found unsatisfactoryC54, 553.
cation-&and bonds and formation of new bonds. It is The reason is to be sought in the fact that, if the
usual to look at heterogeneous catalysis from the adsorption of oxygen on bare metals simulates the
viewpoint of co-ordination chemistry[48-501. The formation of the corresponding bulk oxide[57], ad-
same approach can reasonably be extended to eiec- sorption of oxygen on a bulk oxide is expected to
trochemistry where the interaction of the solid surface simulate the formation of the higher oxide:
with a condensed phase increases the probability of
MO,+O-+ MO,+ ,. (4)
formation of co-ordinately saturated surface groups.
The main and most difficult task of electrocatalytic Any step of the reaction of oxygen evolution can be
studies is to organize experimental kinetic data in a envisaged as a transition of the underlying oxide to a
rational fashion. Best comparison among different higher state. Thus, for instance, for the primary water
electrode materials is made in terms of overpotential at discharge:
constant current density. Irrespective of the detailed
origin of the phenomenon, electrocatalysis can be -M-OH + H,O -+ -M ,,
I /OH +H++e. (5)
expressed, to a first approximation, in terms of bond
strength between the electrode surface and reaction
1 OH
intermediatesc51, 521. Since no in siru electrochemical This reaction can also be viewed[53] as a transition
data are available for oxides, the evaluation of the between two different configurations of the surface
electrocatalytic activity must first rely on experimental coordination complex or better still between two
data taken from neighbouring fields. Considering different complexes (the coordination number of the
oxygen evolution, bonds between surface metal ions surface hydroxy-complex is increased by one). In this
and oxygenated species can be envisaged to be formed context, the parameter which is expected to parallel the
and broken during the anodic reaction. The analogue energy change occurring during the electrochemical
in the gas phase can be found in the adsorption of reaction is the enthalpy of transition of the oxide from
oxygen on oxides. The kinetic features of this reaction a lower to a higher oxidation state[5S]. Figure 17
can be investigated by homomolecular isotopic oxygen shows that the activity for oxygen evolution changes
exchange: with the nature of the catalyst according to a volcano-
shaped curve. Oxides which are oxidized with difficulty
‘802+1602*2180’60. (3) are not very active because intermediates adsorb too
Data have been reported by Boreskov et ul.[53] and by weakly. Water discharge is the rate determining step in
Winter[54]. Figure 16 shows a plot of overpotential this case. Oxides which are oxidized too easily are also
for oxygen evolution against the activation energy for not very active because intermediates are adsorbed too
isotopic oxygen exchange[55]. Despite the scatter of strongly. Removal of oxygenated species is rate de-
data, the two reactivity scales can be seen to run termining for these electrodes.
parallel which suggests that the same energetic par- Figure 17 corroborates the old Sabatier’s principle
ameters are responsible for the observed catalytic of catalysis. It is probably only a first approximation
phenomena. Some exception is given by CoaO,. The since the plot is not at constant Tafel slope (constant
reason for this might be related to its semiconducting mechanism) nor at constant surface area, and the
characteristics but this explanation is probabiy not possibility that for some of the oxides the Tafel slope
conclusive. may be governed by the space charge behaviour
Some authors[53,56] have suggested that the bond [59, 603 is not taken explicitly into consideration.
strength of oxygen atoms adsorbed on the oxide Nevertheless the approach works satisfactorily and the

0.8

c “r

Fig. 16. Electrocatalytic activity in O2 evolution at various Fig. 17. Electrocatalytic activity in 0, evolution at various
oxide electrodes as a function of the activation energy for the oxide electrodes as a function of the enthalpy of the lower
homomokular isotopic O5 exchange in the gas phase. (0) + higher oxide transition. (0) Alkaline and (a) acid solutions
Aikaline and (0) acidic solulions are indicated. are indicated.
 Anodic evolution of oxygen and chlorine 1511

plot possesses some useful predictive character. selectivity for chlorine evolution should not depend
Conceptually, the approach does not conflict with appreciably on the electrode material. Experimental
other views according to which the catalytic activity is data for chlorine evolution from concentrated brines
expressed in terms of (1) matching between the poten- appear to support this conclusion[66,67], but the
tial for oxygen evolution and the potential of oxide situation may differ at other pH values and low
further oxidation[61] or (2) variation in crystal field chlorine concentrations because of the interference
stabilization energy for the transition of the surface between the two processes.
complexes between two configurations[58].
The task of rationalizing the experimental data for Acknowl~d~~~nt-Financial support for this work from the
Cl2 evolution is in principle harder since no useful data Italian National Research Council (CNR) is gratefully
exist for adsorption of chlorine from the gas phase acknowledged.
onlo oxides. There is however ample evidence in
catalysis that the M-0 bond strength can organize the
catalytic activity ofdifferent catalystsalso in thecase of REFERENCES
reactions where oxygen is not involved[62]. This
I. Eleclrodes ojConduche h4efaIfic Oxides Parts A and B
suggests that the reactivity scale (relative activity)
(Edited by S. Trasatti). Elsevier, Amsterdam (1980 and
depends only on the electronic structure of the sohd 1981).
adsorbent. This is true, eg in the case of metals for 2. H. B. Beer. J. electrochem. Sot 127. 303C (19801.
which the heats of adsorption of different gases are 3. Modern Chlor-alkali Technology (Edited by C. Jackson),
linearly related[52,63]. Vol. 2. Ellis Horwood, Chichester (1983).
Also in the case of Cl adsorption, the formation of 4. ial J. M. SedIak, R. J. Lawrence and J. F. Enos. Int. J.
the surface hydroxychloride corresponds to a stoichio- &drogen En&y 6, 159 (1981); (b) Hydro& BS an
metric oxidation of the metal cation. Energy Currier (Edited by G. Imarisio and A. S. Strub).
Reidel, Dordrecht (1983).
M’O, + Cl -+ M’ + ‘QCl. 5. Ya. M. Kolotyrkin, V. V. Losev, D. M. Shub and Yu. E.
(6) Roginskaya, Elektrokhitiyo IS, 291 (1979).
If Cl is adsorbed on the oxide surface during chlorine 6. S. Trasatti and G. B-nca, J. &crroanaL Chem. 29.
evolution, the oxide in fact passes to a higher valency Appl. 1 (1971).
state: 7. (a) D. L. Caldwell and M. J. Hazelrigg, in Modern Chlor-
alkali Technolonv Edited by M. 0. Coulterl, D. 121. Ellis
I /OH Horwood, Chit&&r (1986); (b) A. Tvaru&, 341h ISE
-M-OH + Cl- --t -M (7) Meeting, Erlangen, Abstract III-7 (1983).
1 ‘Cl +p* 8. R. Kiitz, Electrochim. Acta in preparation.
9. R. Parsons, Surj: Sci. 2, 418 (1964).
10. D, M. Novak, B. V. Tilak and B. E. Conway, in Modern
The concepts applied to the oxygen evolution elec- Aspects ofElec!roc/~mis~y, Vol. 14 (Edited by J. O’M.
trocatalysis can thus be extended to chlorine evolution, Bockris, B. E. Conway and R. E. White), p. 195. Plenum
including the calculation of the change in the crystal Press, New York (1982).
field stabilization energy for the transition from one 11. J. O’M. Bockris and T. Otagawa, J. phys. Chem. 87,296O
surface complex to another[64,65]. (1983).
A more direct comparison between electrocatalysis 12. M. M. Peeherskii, V. V. Gorodetskii, N. Ya. Bun6 and V.
V. Losev, EIektrokhimiya IS, 415 (1982).
for oxygen and electrocatalysis for chlorine evolution
13. D. N. Buckley and L. 6. Burke, J. &e&anal. Chem. 52,
can be carried out by plotting the overpotential for the 433 (I 9741.
former reaction against the overpotential for the latter. 14. C. Iwakura, A. Asada and H. Tamura, Denki Kagaku 44,
Figure 18 shows that, despite the small number of 639 (19761.
oxides for which similar data are available, there is 15. H. tiendt’and V. PI&k, Extended Abstracts Eiectrorhem.
teasonable evidence that the two reactivity scales run Sot. Meeting, 8>1, 1070 (1983).
parallel. The slope of the straight line through the 16. R. Garavagiia, C. M. Mari, S. Trasatti and C. De
points is in fact unity. This result suggests that the Asmundis, Surf. Technol. 19, 197 (1983).
17. L. Burke and 0. Murphy, J. electrochem. Sac. 96, 19
11979).
1.9- 18. b. V. Kokoulina, T. V. Ivanova, Yu. I. Krasovitskaya, 2. I.
Kudryavtseva and L. I. Krishtalik, Elektrokhimiyu 13,
1511 (1977).
19. S. Ardizzone, A. Carugati and S. Trasatti, J. eleciroana!.
Chem. 126,287 (1981).
20. G. Lodi, E. Sivieri, A. Dehaltisti and S. Trasatti, J. app[_
E!ectrochem. 8, 135 (1978).
21. C. Battaglin, A. Camera, P. Mazzoldi, G. Lodi, P.
Bonora, A. Daghetti and S. Trasatti, J. electroanal. Chem.
135. 313 (19821.
22. I. D. Beloia, V: V. Shalaginov, B. Sh. Galyamor, Yu. E.
Roginskaya and D. M. Shub. Zh. neorg. Khim. 23. 161
(1978).
23. G. Lodi, C. Bighi and C. De Asmundis, Mower. Chem 1,
177 (1976).
24. A. Daghetti, G. Lndi and S. Trasatti, Muter. Chem. Phys.
Eg. 18. Plot of the potential for Cl, agamst the potential TOT 8, 1 (1983).
0, evolution at the same current density for a number of 25. V. Gutmann, The Donor-Acceptor Approach io
oxides. (0) Alkaline and (m) acidic solutions are indicated. Molecular Interactions. Plenum Press, New York (1978).
1512 S. TRASA~I

26. G. A. Parks, Chem. Rev. 65, 177 (1965). 46. J. Augustynski, L. Balsenc and J. Hinden, J. electrochem.
27. P. Siviglia, A. Daghetti and S. Trasatti, Colfoids Sut$7, I5 sot. 125, 1093 (1978).
(1983). 47. J. A. Harrison. D. L. Caldwell and R. E. White.
28. G. Battaglin A. Camera, G. Lodi, E. Giorgi, A. Daghetti Elecrrochim. Ac;a 28, 1561 (1983).
and S. Trasatti, J. Chem. Sot. Faraday Trans. I, 80, 48. H. Noller, Acta chim. Acd. Sci. Hungaria 109,429 (1982).
913 (1984). 49. A. C. Balazs and K. H. Johnson, SUB Sci. 114; 197 il982j.
29. K. Koumoto and H. Yanagida, Jap. J. appl. Phys. 20,445 50. R. Mason and M. W. Roberts, Inorg. Chim. Acta 50. 53
(1981). (1981).
30. V. N. Andreev, V. E. Kazarinov, D. V. Kokoulinaand L. I. 51. R. Parsons. Trans. Faraday $0~. 54. 1053 (1958).
Krishtalik, Elektrokhimivo 14. 1278 (19781. 52. S. Trasatti. in Elertrocatnl& (Edited by Wl E. bGrady,
31. S. Ardizzbne. D. Lettieh and ‘S. Tra&i, j. electroanal. P. N. Rossand F. G. Will), p. 73. The Electrochem. Sot..
C-hem. 146,431 (1983). Pennington (1982). ‘. _
32. R. Hutchings, R. Kiitz, K. Muller and S. Stucki, Extended 53. G. K. Boreskov, V. V. Popovskii and V. A. Sazonov, in
Abstracts 34th ISE Meeting, Erlangen, Abstract 0305 Proc. Fourth lnr. Gong. Catalysis, Moscow, 1968, p. 439.
(1983). Akadkmiai Kiadb, Budapest (1971).
33. S. Ardizzone, A. Carugati, G. Lodi and S. Trasatti, J. 54. E. R. S. Winter, J. them Sot. A, 2889 (1968).
electrochem. Sot. 129, 1689 (1982). 55. S. Trasatti, J. electroar&. Chem. 111, 125 (1980).
34. L. I. Krishtalik, Elekrrokhimiya 15, 462 (1979). 56. P. Ruetschiand P. Delahay,J. chrm. Phys. 23,556 (1955).
35. I. V. Kadija, J. electrochem. Sot. 127, 599 (1980). 57. G. C. Bond, R. I. C. Rev. 3, 1 (1970).
36. L. J. J. Janssen, G. J. Visser and E. Barendrecht. 58. M. Inai, C. Iwakura and H. Tamura, Den& Kagnku 48,
Electrochim. Actn 28, 155 (1983). 173, 229 (1980).
37. M. M. Pecherskii, V. V. Gorodetskii, S. V. Evdokimov 59. J. W. Schultze, P. Kohl, M. Lohrengel and U. Stimmmg,
and V. V. Losev, Elektrokhimiva 17. 1087 (1981). Chem. Ing. Techn. 51, 643 (1979).
38. V. V. Losev, Elektrokhimiya l?, 73j (1981): 60. B. D. Cahan and C. T. Chen, J. electrochem. Sot. 129,700
39. T. Arikado, C. Iwakura and H. Tamura, ELctrochim. (1982).
Acta 23, 9 (1978). 61. A. C. C. Tseung and S. Jasem, Electrochim. Acfa 22, 31
40. D. A. Denton, J. A. Harrison and R. I. Knowles. (1977).
Electrochem. Acta 24, 521 (1978). 62. Yu. D. Pankratiev, React. Kinef.Catul. Lrtt. 20,255 (1982).
41. B. U. Bondar, A. A. Borisova and E. A. Kalinovskii. 63. J. J. Burton, Surf. Sci. 66, 647 (1977).
Elekrrokhimiya 10, 44 (1974). 64. M. lnai, C. Iwakura and H. Tamura, Eleczrochim. Acta
42. R. G. Erenburg, L. I. Krishtalik and I. P. Yaroshevskaya, 24, 993 (1979).
Elektrokhimiya 11, 1072 (1975). 65. T. Arikado, C. Iwakura and H. Tamura, Electrochim.
43. B. E. Conway and D. M. Novak, J. them. Sot. Faraday Acta 23, 799 (1978).
i-runs. 1 75, 2454 (1979). 66. H. H. Walter, H. Kaden and H. Hofmann, Chem. Tech. 29,
44. L. I. KrishtaIik, 34th [SE Meeting, Erlongen, Abstract 212 (1977).
III--6 (1983). 67. A. A. Uzbekov, V. G. Lambrev, I. F. Yazikov, N. N.
45. L. D. Burke and J. F. O’Neill, J. rlectroanol. Chem. 101, Rodin, L. M. Zabrodskaya, V. S. Klement’eva and Yu. M.
341 (1979). Vlodov, Efektrokhimiya 14. 1150 (1978).