Content uploaded by Alexander Kraft
Author content
All content in this area was uploaded by Alexander Kraft on Jun 11, 2018
Content may be subject to copyright.
Electrochemical water disinfection.
Part II: Hypochlorite production from potable water, chlorine consumption
and the problem of calcareous deposits
A. KRAFT1*, M. BLASCHKE1, D. KREYSIG2, B. SANDT2, F. SCHRO
ÈDER2and J. RENNAU2
1Gerus mbH, Ostendstrasse 1, 12459 Berlin, Germany;
2AQUA Butzke-Werke AG, Parkstrasse 1±5, 14974 Ludwigsfelde, Germany
(*author for correspondence, fax: +49 30 53073856)
Received 15 July 1998; accepted in revised form 19 January 1999
Key words: chlorine evolution, hypochlorite, iridium oxide, platinum, water disinfection
Abstract
The electrolytic production of hypochlorite from tap water in a ¯ow-through reactor system is investigated using
stacked platinum or iridium oxide coated titanium sheet or expanded metal electrodes. The in¯uence of fast chlorine
consumption and polarity reversal on the hypochlorite production rate was determined along with the dependence
of the hypochlorite production rate on temperature, ¯ow through velocity and current density. It was found that in
most cases, the hypochlorite production rate was higher on iridium oxide compared to platinum electrodes. An
increase in the ¯ow-through velocity leads to an increased hypochlorite production rate while the hypochlorite
production rate falls with increasing temperature.
1. Introduction
Electrochemical disinfection is one possible method for
water treatment [1±6]. In this process, oxidizing and
bactericidal substances are not added to the water; they
are produced from naturally occurring substances in the
water by electrolysis without the requirement for addi-
tional chemicals. The oxidizing substances which are be
produced are short-lived, free radical species such as Oá,
OHáand more stable substances such as HClO, ClOÿ,
H2O2,O
3
, MnO2ÿ
4,S
2
O
2ÿ
8amongst others [3]. How-
ever, the most important electrochemically produced
components for the disinfection process are hypo-
chlorous acid and hypochlorite [2]. In electrochemical
disinfection, the main reaction is the electrolysis of
water; hydrogen is produced at the cathode and oxygen
at the anode; the anodic production of hypochlorite is a
side reaction.
In Part I we reported on electrolytic hypochlorite
production in very dilute chloride solutions [6]. These
experiments were performed by minimizing the chlorine
consumption by using deionized water, pure chemicals
and PTFE materials for the experimental setup, and by
performing a preelectrolysis process prior to each
experiment. The present paper is also concerned with
electrolytic hypochlorite production from very dilute
solutions, but without the requirement to minimize
chlorine consumption. The aim being to probe the
practical use of electrochemical disinfection for drinking
water. Comparing the results of the present paper with
our previously published results [6] gives helpful hints
for the exact adaptation of electrochemical disinfection
to the amount of and the properties of the water that is
to be disinfected.
2. Experimental details
Experiments were performed using a ¯ow-through
reactor in the water laboratory of AQUA Butzke±
Werke AG (Ludwigsfelde near Berlin). The reactor
(Gerus mbH) comprised of a polybutene pipe with an
electrode stack (Figure 1), the reactor pipe having an
inner diameter of 50 mm and a length of 230 mm. The
electrode stack was made up of a number of equidistant
monopolar plate electrodes. The following variations of
the electrode stack were tested: (i) electrode substrates:
titanium sheets or titanium expanded metal; (ii) elec-
Journal of Applied Electrochemistry 29: 895±902, 1999. 895
Ó1999 Kluwer Academic Publishers. Printed in the Netherlands.
trode coatings: platinum (thickness 2 lm) or iridium
oxide (IrO2±Ta2O5with 8 g Ir mÿ2; molar ratio of Ir to
Ta: 70 to 30).
The geometrical area of a sheet electrode was
30 mm 133 mm while the expanded metal electrode
had a geometric area of 30 mm 110 mm. A standard
electrode stack consisted of 12 plates of sheet electrodes
(electrode distance 2 mm) or 22 plates of expanded
metal electrodes (electrode distance 1 mm). With an
applied current of 10 A, this con®guration gave an
approximate current density of 20 mA cmÿ2for both
electrode stacks. For the calculation of the current
density on the expanded metal electrodes, a factor of
0.75 for the area was used. All electrodes were produced
by Metakem GmbH (Usingen, Germany).
The electrolytic water disinfection experiments were
performed by the continuous pumping through the
reactor of potable water from the Ludwigsfelde munic-
ipality at various ¯ow velocities. Water was only
passaged once through the reactor. The water temper-
ature varied between 10 and 60 C and in all experi-
ments, the chloride concentration was 70 mg lÿ1
(1.97 mmol lÿ1), conductivity was 0:9 mS cmÿ1, pH
was 7.3, and the Ca2and Mg2concentrations were
130 and 10 mg lÿ1, respectively. The power source was a
20 V, 100 A recti®er (Munk).
The hypochlorite concentration was measured in a
bypass situated directly after the electrolysis cell using an
amperometric active chlorine sensor (Iotronic GmbH);
this hypochlorite concentration was taken to be the
so-called active chlorine level. Active chlorine is the sum
of the three different dissolved chlorine species (Cl2,
HClO, and ClOÿ) which are present in the water [6, 7].
3. Results and discussion
3.1. Active chlorine consumption
Figure 2 shows the dependence of the active chlorine
production rate on current density at a ¯ow-through
velocity of 120 l hÿ1at two different temperatures.
Iridium oxide coated titanium sheets were used as
electrodes. A comparison of the results in Figure 2 with
the previously published results using minimized chlo-
rine consumption (see Figure 5 in [6]) shows that in both
cases, there is a linear increase in the active chlorine
production rate with increasing current density. How-
ever, with minimized chlorine consumption this straight
line begins near 0 mA cmÿ2[6], whereas in the present
study the linear increase starts at much higher current
densities; that is, 11 mA cmÿ2at 30 C and at
Fig. 1. Schematic view of the electrochemical reactor used in the
present study. Key: (1) reactor pipe, (2) rubber O-ring sealing,
(3) electrode stack and (4) current feeder.
Fig. 2. Dependence of the active chlorine production rate on current
density at 30 and 60 C on IrO2sheet electrodes (¯ow rate 120 l hÿ1).
896
23 mA cmÿ2at 60 C. This suggests that there is a
parallel shift in the nearly linear increase of the active
chlorine production rate with current density. The
degree of shift increases with increasing temperature.
This parallel shift can be explained by the enhanced
active chlorine consumption and, in particular, by the
fast chlorine consumption. The active chlorine con-
sumption is the result of reactions with components in
the water and with the walls of the water distribution
system. Brie¯y, two types of active chlorine consump-
tion can be distinguished [8, 9]. First, there is a fast
chlorine consumption due to fast reactions with certain
substances in the water. Secondly, there is a kinetically
slower chlorine consumption due to reactions with other
dissolved substances, with suspended particles, and with
the pipe walls of the plumbing system. The latter is a
function of pipe diameter, ¯ow through velocity, nature
of the pipe material and of the amount and nature of the
deposits on the pipe wall [9].
An example of fast chlorine consumption is the
reaction with dissolved iron species. It was found in
the present investigation, that the concentration of
dissolved iron in electrolysed water was reduced from
approximately 0.1 mg lÿ1before entering the electro-
chemical reactor, to less than 0.006 mg lÿ1directly on
exit from the reactor. This is below the detection limit of
the method used. After a few weeks of continuous
electrolysis, a brown staining on the reactor walls in the
area of the electrode stack and on the pipes behind the
reactor was visible. This staining was probably unsol-
uble Fe(III) substances, and it can be assumed that
dissolved Fe(II) was oxidized by active chlorine to Fe(III)
which precipitated on the reactor walls.
This fast chlorine consumption is completed before
the measurement of the active chlorine concentration.
As a result, we were unable to measure the real
produced active chlorine concentration but rather mea-
sured the difference between the produced active chlo-
rine concentration and that which removed by the fast
chlorine consumption. Therefore, the fast chlorine
consumption can be expressed as an equivalent of (lost)
active chlorine concentration. This connection is indi-
cated in the equation:
cactive chlorine;produced cactive chlorine;measured
cactive chlorine;fast consumption 1
The equivalent concentration of fast chlorine consump-
tion depends on the substances dissolved in the water
and on the water temperature. The higher the water
temperature, the higher the fast chlorine consumption.
This is because a temperature increase generally leads to
an increase in reaction rate. In Figure 3, a schematic
representation of the relationship between the active
chlorine production, fast chlorine consumption and the
measured active chlorine production is shown. If this
theory is used to calculate the hypothetical fast chlorine
consumption from the parallel shift of the two curves
shown in Figure 2, values of about 2.5 mg lÿ1for 30 C
and 6 mg lÿ1for 60 C are obtained.
If it is not carefully accounted for, chlorine consump-
tion by the electrolysed water can lead to discrepancies
in the dependence of the active chlorine production on
temperature, current density and ¯ow through velocity.
For example, if the real active chlorine concentration is
constant but the ¯ow through velocity is increased, more
substances which cause fast chlorine consumption are
able to react with the active chlorine produced. Conse-
quently, an apparently lower active chlorine production
is measured.
Further comparison of the results of the present
investigation with our recent results [6] shows that there
are differences in the slope of the linear increase of
chlorine production on IrO2with current density. In the
present study, using a chloride concentration of
60 mg lÿ1, a slope of 30 mg Aÿ1hÿ1was measured
whereas the slope measured in [6] with a concentration
of 150 mg lÿ1chloride ions was 60 mg Aÿ1hÿ1. A
dependence of this slope on chloride concentration is
possible but was not investigated in the present study as
only the original tap water with a ®xed chloride
concentration was used for the experiments. This
phenomenon will be investigated further.
All the values for active chlorine production presented
in the current investigation are only measured values of
the active chlorine production rate. The real active
chlorine production rate can be calculated by adding the
hypothetical fast chlorine consumption to these values.
Due to variations in the water quality, drastic changes in
Fig. 3. Schematic representation of the connection between the real
active chlorine production rate, fast chlorine consumption and
measured active chlorine production rate.
897
the fast chlorine consumption and, consequently, in the
measured active chlorine production rate, can occur.
The measured active chlorine production rate can be
relatively constant over extended periods of time but can
also change on an hourly basis. This makes it dif®cult to
apply a method of electrochemical disinfection to water
treatment.
3.2. Calcareous deposits and polarity reversal for
their removal
During water electrolysis, a thin ®lm of calcareous
deposits is formed at the cathode surface. This ®lm
mainly consists of CaCO3and Mg(OH)2[6, 10±12]
being produced in the vicinity of the cathode surface as a
result of a local pH increase [13] due to the evolution of
hydrogen:
2 H2O2 eÿ!H22 OHÿ2
whereas the pH on the cathode surface is increased, the
pH on the anode surface is decreased due to the
evolution of oxygen:
2 H2O4 eÿ!O24 H3
This phenomenon can be used for the removal of
cathode scaling; for this, a regular reversal of electrode
polarity is necessary [14, 15].
Following polarity reversal, the former cathode with
calcareous deposits acts as an anode and, with con-
tinued electrolysis, a local pH decrease at the anode
surface occurs. This leads to dissolution of the scale
on the anode surface according to the following equa-
tions:
CaCO32 H!Ca2CO2H2O4
Mg(OH)22 H!Mg22 H2O5
The scale that is more distant from the anode surface is
unable to adhere and is, therefore, ¯ushed away by the
water ¯ow. By applying this method of polarity reversal,
it is possible to clean calcareous deposits from the
electrode. The time between two polarity reversals
should be in the range of 10±60 min. A strong outgas-
sing effect is visible for a few seconds following polarity
reversal on the electrode that was formerly the cathode.
This is probably due to the formation of CO2according
to Equation 4.
When the polarity reversal method is used for
electrode cleaning, it is necessary that all parts of the
titanium electrode substrates are coated with electro-
chemically active ®lms (e.g., IrO2or Pt as used in this
study) otherwise polarity reversal does not work. This is
due to the fact that titanium, as a valve metal, allows
current ¯ow only when used as a cathode. When used as
an anode, the titanium surface is passive and little or no
current ¯ows through it. Therefore, the resultant pH
decrease at the anode surface is not strong enough to
dissolve the scale.
If cathodically formed scale deposits are not com-
pletely removed, they continue to grow. The remaining
deposits act as nucleation centers for the growth of
calcareous scale. The higher the water temperature, the
greater the hardness of the water and the ¯ow velocity,
the greater the growth rate. The increase in ¯ow velocity
improves the mass transfer of Ca2and Mg2ions to
the surface. Eventually, this leads to a total blockage of
the electrode stack, as the gaps between the electrodes
become ®lled with scale [15].
Polarity reversal has a strong side effect on the activity
of electrode materials in active chlorine production.
Figure 4 shows an example of the change in the active
chlorine production rate for iridium oxide and platinum
electrodes directly after a polarity reversal. On platinum
electrodes, the active chlorine production rate rises
sharply directly following polarity reversal. This can be
attributed to a change in the platinum surface during the
time the electrode was acting as a cathode. In this time,
the thin oxide ®lm of PtOx(platinum oxide) which was
formed when the electrode was the anode is reduced to
Pt (platinum) metal. Platinum metal has a much higher
activity for chlorine production (or rather a lower
activity for oxygen evolution) than the PtOxsurface.
Following polarity reversal, the electrode is working as
anode and produces oxygen and chlorine. Again, the
Fig. 4. Change in the active chlorine production rate with time (t)
following polarity reversal (polarity reversal at t0 min) at 60 C,
15 A (current density 34 mA cmÿ2), ¯ow rate of 120 l hÿ1on IrO2and
Pt coated titanium sheet electrodes.
898
surface becomes covered with a thin layer of oxide and
consequently the chlorine production rate decreases.
Most of the decrease in the active chlorine production
rate occurs in the ®rst 5 min. After 30 min, the active
chlorine production rate steadies to a constant value of
approximately 0.4the value obtained directly follow-
ing polarity reversal. Using platinum, the change in the
active chlorine production rate after current reversal is
reproducible.
With iridium oxide, the inverse effect compared to
platinum is observed. Directly after polarity reversal the
chlorine production rate is very low. The active chlorine
production rate slowly increases during anodic oxygen
and chlorine production but this increase is much slower
than the decrease in the active chlorine production rate
on platinum. In addition, even if an IrO2electrode has
only been used once as cathode, the value obtained for
the active chlorine production rate never matches that
obtained using a new IrO2electrode, which has never
been used as cathode. The amount of the decrease in the
active chlorine production rate depends upon the time
that the electrode was polarized as a cathode and on the
current density during this time. The longer the elec-
trode is used as cathode and the higher the cathodic
current density, the lower the active chlorine production
rate following polarity reversal. The increase in the
active chlorine production rate during the anodic cycle
also depends on the current density. The higher the
current density, the faster the rise in the active chlorine
production rate.
The lowering of the active chlorine production rate
by iridium oxide after cathodic polarization of the
electrode is probably due to the partial reduction of the
mixed oxides that comprise the electrode coating. This
leads to a composition that is less active for chlorine
evolution than the composition of a newly prepared
electrode. When polarized as an anode, the reduced
mixed oxides are reoxidized and therefore the chlorine
production rate increases. However, the initial com-
position of the iridium oxide electrode cannot be
reproduced.
Another negative side effect in the use of polarity
reversal is the diminution of the electrode lifetime. This
problem mainly concerns iridium oxide coated titanium
electrodes. For example, at a current density of
20 mA cmÿ2, a temperature of 40 C, and a regular
polarity reversal cycle of 15 min, the electrode lifetime
of iridium oxide coated electrodes is decreased to
approximately three months. Again, this diminution is
greater at higher than lower current densities. Under the
same conditions, the lifetime of a platinum-coated
electrode is more than twelve months (an experiment
is still in progress).
3.3. Dependence of hypochlorite production rate on
IrO2and Pt on current density, ¯ow through velocity
and temperature
All results presented in this section were produced by
using electrodes to which a regular polarity reversal
(30 min for each half cycle) was applied. To make the
results more comparable all measurements were per-
formed approximately 20 min after polarity reversal.
Previous results [6] with minimized chlorine consump-
tion indicated that IrO2is generally much more effective
than platinum in active chlorine production and that the
active chlorine production rate is lowered by an increase
in temperature. The chlorine production rate [mg hÿ1]
exhibits an almost linear increase with increasing current
density, whereas the active chlorine production rate
[mg Aÿ1hÿ1] is only slightly increased by increasing the
current density. The in¯uence of the ¯ow through
velocity was not investigated in this study. However, it
is generally assumed that by increasing the ¯ow through
velocity, the mass transfer to the electrode is enhanced;
this would lead to an increase in the active chlorine
production rate.
Figure 5 shows a comparison of the active chlorine
production rate by iridium oxide and platinum sheet
electrodes at 60 C. Following the parallel shift due to
the fast chlorine consumption, a linear increase in the
active chlorine production rate was measured. The
slopes for IrO2and Pt were 30 and 10 mg Aÿ1hÿ1,
respectively. Surprisingly, at lower current densities, a
platinum electrode is able to produce more active
chlorine than an iridium oxide electrode. Starting with
a current density of 37 mA cmÿ2, iridium oxide pro-
duces more chlorine than platinum. The differential
between the two materials increases further with in-
Fig. 5. Comparison of the active chlorine production rate on iridium
oxide and platinum sheet electrodes (60 C, 120 l hÿ1).
899
creasing current density. This is due to the difference in
the slopes for the active chlorine production rate as a
function of current density. The slope of the increase in
the active chlorine production rate on iridium oxide
using 70 mg lÿ1chloride and with polarity reversal is
30 mg Aÿ1hÿ1. This value is about half that measured
for iridium oxide using a chloride concentration of
150 mg lÿ1with no polarity reversal (60 mg Aÿ1hÿ1)
[6]. The slope for Pt is almost constant at both chloride
concentrations, with or without polarity reversal. Thus,
the change in the slope of the increase in the active
chlorine production rate by the iridium oxide electrode
as a function of current density, can also be explained by
the change in the activity of the iridium oxide electrode
material for chlorine production due to polarity rever-
sal.
Figure 6 shows the same values for the active chlorine
production rate as seen in Figure 5, but the values are
normalized for the current. Even at 60 C, with a high
chlorine consumption, the measured active chlorine
production rate (mg Aÿ1hÿ1) on Pt is higher when
compared to the values obtained using minimized
chlorine consumption at 23 C, a low ¯ow through
velocity and the same current density (see Figure 6 in
[6]). For example, in the present study a current density
of 60 mA cmÿ2yields a production rate of about
18 mg Aÿ1hÿ1, whereas in our previous study [6] a
production rate of only 10 mg Aÿ1hÿ1was measured.
This is due to the enhanced mass transfer as a result of
the much higher ¯ow-through velocity in the reactor
used in the current study.
Despite an increased mass transfer rate, the measure-
ment of the active chlorine production rate on IrO2at
60 C with high chlorine consumption does not reach
the values obtained using minimized chlorine consump-
tion at 23 C [6]. For example, at a current density of
60 mA cmÿ2, as in the present investigation, a produc-
tion rate of 35 mg Aÿ1hÿ1was measured, whereas in
our previous study the rate was 58 mg Aÿ1hÿ1. This is
due to the strong decrease in the active chlorine
production rate on IrO2due to the effects of polarity
reversal.
The temperature dependence of the active chlorine
production rate on IrO2at two different current
densities is shown in Figure 7. The active chlorine
production rate is decreased with increasing tempera-
ture. This is explained by a lowering of the real active
chlorine production rate and, more importantly, an
increase in chlorine consumption with temperature [6].
The increase in fast chlorine consumption is the dom-
inant factor when looking at the temperature depen-
dence of the active chlorine production rate.
In Figure 8, the dependence of the active chlorine
production rate on ¯ow through velocity and current
density is shown for IrO2sheet electrodes. The measured
active chlorine production rate slightly increases with
increased ¯ow-through velocity. Generally, a much
steeper increase in the production rate with an increased
¯ow-through velocity would be expected. That only a
slight increase is observed is due to the fact that
increasing ¯ow-through velocity also increases chlorine
consumption. This is because at higher ¯ow-through
velocities a greater number of substances which cause
active chlorine consumption pass through the reactor.
Therefore, by increasing the ¯ow-through velocity both
the real active chlorine production and the fast chlorine
consumption rate are increased. Depending on the
numbers involved, this can also lead to a decrease in
the measured active chlorine production rate (see
Equation 1).
Fig. 6. Dependence of the active chlorine production rate on current
density using iridium oxide and platinum sheet electrodes normalized
on the current (60 C, 120 l hÿ1).
Fig. 7. Dependence of the active chlorine production rate on temper-
ature for two dierent current densities (22 and 34 mA cmÿ2), using
IrO2sheet electrodes (120 l hÿ1).
900
Figure 9 compares the dependence of the active chlorine
production rate by IrO2sheet and expanded metal
electrodes on current density. In the lower current
density range, the active chlorine production rate is
higher with expanded metal electrodes than with sheet
electrodes. This is explained by the improved mass
transfer of chloride ions to the anode surface due to a
more turbulent ¯ow between the expanded metal elec-
trodes as compared to sheet electrodes. At higher
current densities, sheet electrodes produce more active
chlorine than expanded metal electrodes. This may be
due to the fact that at higher active chlorine concentra-
tions, the reduction of the produced active chlorine on
the cathode becomes more important. The more turbu-
lent ¯ow characteristics lead to higher reduction losses
on the expanded mesh electrodes as compared to the
sheet electrodes.
4. Conclusions
The use of electrochemical disinfection for water treat-
ment has been investigated by measuring the active
chlorine production rate in potable water at different
temperatures, ¯ow-through velocities, current densities,
and by using different electrode materials.
Fast chlorine consumption has considerable in¯uence
on measured values of active chlorine production. To
adapt the method of electrochemical disinfection to the
amount, and quality, of the water that is to be
disinfected, the fast chlorine consumption of this water
must ®rst to be determined and taken into account. The
chloride content and chlorine consumption properties of
the water to be processed can vary considerably. The
size of the electrochemical reactor must be chosen in a
way that allows it to respond to a reduction in the
chloride concentration and to an increase in chlorine
consumption.
A major problem is the formation of calcareous
deposits on the cathode surface. These deposits can be
removed by regular polarity reversal of the electrodes.
Unfortunately, in the case of iridium oxide electrodes,
which in general are more effective than platinum elec-
trodes, this polarity reversal leads to a shortening of the
electrode lifetime and a reduction in the active chlorine
production rate. When electrodes to which polarity
reversal has been applied are used, platinum is more ef-
fective in active chlorine production in the lower current
Fig. 8. Dependence of the active chlorine production rate on ¯ow
through velocity for three dierent current densities (22, 34 and
45 mA cmÿ2), using IrO2sheet electrodes (60 C).
Fig. 9. Comparison of the active chlorine production rate on IrO2sheet and expanded metal electrodes, dependence on current density
(120 l hÿ1, 60 C).
901
density range, whereas at higher current densities,
iridium oxide is the more effective electrode material.
The improved mass transfer on expanded metal
electrodes leads to an increase in the active chlorine
production rate on the anode and in parallel, to an
increase in the chlorine reduction rate on the cathode.
The addition of these two processes produces the result
that at lower current densities and active chlorine
production rates, the measured active chlorine rate is
higher on expanded metal than on sheet electrodes.
However, at higher current densities and production
rates the sheet electrode stacks deliver more active
chlorine than expanded metal electrodes.
References
1. A. Reis, GIT Fachz. Lab. 20 (1976) 197±204 (in German).
2. G.E. Stoner, G.L. Cahen, Jr., M. Sachyani and E. Gileadi,
Bioelectrochem. Bioenerg. 9(1982) 229±243.
3. G. Patermarakis and E. Fountoukidis, Water Res. 24 (1990) 1491±
1496.
4. M. Scho
Èberl, Eur. Pat. EP 0 515 628 B1. (1991).
5. N. Kanekuni, N. Shono, M. Kiyohara, K. Tabata, S. Kono and
M. Hayakawa, Eur. Pat. Appl. EP 0 711 730 A1. (1995).
6. A. Kraft, M. Stadelmann, M. Blaschke, D. Kreysig, B. Sandt, F.
Schro
Èder and J. Rennau, J. Appl. Electrochem. 29 (1999) 861±868.
7. A.R. Despic, M.M. Jaksic and B.Z. Nikolic, J. Appl. Electrochem.
2(1972) 337±343.
8. B.M. Saunier and R.E. Sellek, J.A.W.W.A. 71 (1972) 297±332.
9. L. Kiene, W. Yu and Y. Levi, C.R. A.W.W.A., Annual Confer-
ence, San Antonio, (1993), pp. 503±511.
10. J.-F. Yan, T.V. Nguyen, R.E. White and R.B. Grif®n, J. Electro-
chem. Soc. 140 (1993) 733±742.
11. J.-F. Yan, R.E. White and R.B. Grif®n, J. Electrochem. Soc. 140
(1993) 1275±1280.
12. C. Gabrielli, M. Keddam, H. Perrot, A. Khalil, R. Rosset and M.
Zidoune, J. Appl. Electrochem. 26 (1996) 1125±1132.
13. C.-Y. Chan, K.H. Khoo, T.K. Lim and A.T. Kuhn, Surf. Technol.
15 (1982) 383±394.
14. A.T. Kuhn and R.B. Lartey, Chemie-Ing. Techn. 47 (1975) 129±35.
15. G.H. Kelsall, I. Chem. E. Symp. No.127, `Electrochemical Engi-
neering and the Environment', (1992) pp. 71±83.
902
