Electrochimica Acta 50 (2005) 53545369

Deactivation of copper electrode in electrochemical

reduction of CO2
Y. Hori ,1 , H. Konishi, T. Futamura, A. Murata, O. Koga 1 , H. Sakurai, K. Oguma
Department of Applied Chemistry, Faculty of Engineering, Chiba University, Inage-ku, Chiba 263-8522, Japan
Received 22 September 2004; received in revised form 27 February 2005; accepted 11 March 2005
Available online 7 April 2005

Abstract
Metallic Cu electrode can electrochemically reduce CO2 to CH4 , C2 H4 and alcohols with high yields as revealed by the present authors.
Many workers reported that formation of CH4 and C2 H4 rapidly diminishes during electrolysis of CO2 reduction. This paper shows that
such deactivation of Cu electrode is reproduced with electrolyte solutions prepared from reagents used by these workers. Deactivated Cu
electrodes recovered the electrocatalytic activity for CO2 reduction by anodic polarization at 0.05 V versus she in agreement with the
previous reports. Features of the deactivation depend greatly on the individual chemical reagents. Purification of the electrolyte solution by
preelectrolysis with a Pt black electrode effectively prevents the deactivation of Cu electrode. Anode stripping voltammetry of Cu electrodes,
which were deactivated during electrolysis of CO2 reduction, showed anodic oxidation peaks at ca. 0.1 or 0.56 V versus she. The severer the
deactivation of the Cu electrode was, the higher electric charge of the anodic peak was observed. It is presumed that some impurity heavy metal,
originally contained in the electrolyte, is deposited on the Cu electrode during the CO2 reduction, poisoning the electrocatalytic activity. On
the basis of the potential of the anodic peaks, Fe2+ and Zn2+ are assumed to be the major contaminants, which cause the deactivation of the Cu
electrode. Deliberate addition of Fe2+ or Zn2+ to the electrolyte solutions purified by preelectrolysis exactly reproduced the deactivation of a Cu
electrode in CO2 reduction. The amount of the deposited Fe or Zn on the electrode was below the monolayer coverage. Electrothermal atomic
absorption spectrometry (etaas) showed that Fe originally contained in the electrolyte solution is effectively removed by the preelectrolysis
of the solution. Mechanistic difference is discussed between Fe and Zn in the deterioration of the electrocatalytic property of Cu electrode
in the CO2 reduction. The concentration of the impurity substances originally contained in the chemical reagents as Fe or Zn is estimated to
be far below the standard of the impurity levels guaranteed by the manufacturers. Presence of trimethylamine in the electrolyte solution also
severely poisons a Cu electrode in the CO2 reduction. It was concluded that the deactivation of Cu electrode in CO2 reduction is not caused
by adsorption of the products or the intermediates produced in CO2 reduction.
2005 Elsevier Ltd. All rights reserved.
Keywords: CO2 ; Electrochemical reduction; Cu; Deactivation; Poisoning

1. Introduction
1.1. Deactivation of copper electrode in
electrochemical reduction of CO2 reported in the
previous works
CO2 is electrochemically reduced to CH4 , C2 H4 and alcohols in high yields at Cu electrode in aqueous electrolytes as

Corresponding author. Tel.: +81 3 3630 4086; fax: +81 3 3630 4086.
E-mail address: y-hori@white.plala.or.jp (Y. Hori).
ISE member.

0013-4686/$ see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2005.03.015

firstly revealed by the present authors [13]. With regard to

this reaction, many workers reported poisoning or deactivation of the copper electrode. Most of them described that
the formation of CH4 and C2 H4 at Cu electrode diminishes
rapidly and the hydrogen evolution prevails in 1030 min after the start of the CO2 electrolysis.
DeWulf et al. [4] studied CO2 reduction at high purity
copper foil electrodes, employing 0.5 M KHCO3 as the electrolyte solution prepared from an ACS reagent grade, MCB
Manufacturing Chemists Inc. and Milli Q water without
further purification. The current density in their study was
3035 mA cm2 at 2.00 V versus standard calomel elec-

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

trode (sce). The total Faradaic efficiency of CH4 and C2 H4

reached 65% after 20 min of electrolysis, and gradually decayed to 0% after 120 min. Meanwhile the electrode surface
was blackened, and they detected graphitic carbon on the
electrode surface by X-ray photoelectron spectroscopy (xps)
and Auger electron spectroscopy (aes) measurements. They
presumed that some products or intermediates formed in the
CO2 reduction are adsorbed on the electrode surface, suppressing the reaction. They also pointed out a possibility that
the poisoning or deactivation could be derived from deposition on the electrode surface of impurities originally contained in the electrolyte solution.
Wasmus et al. [5] reported CO2 reduction studied with on
line electrochemical mass spectrometry. The measurements
were performed in 0.5 M KHCO3 (Fluka p.a.) solution prepared from Millipore water without further purification. They
showed that CH4 and C2 H4 formation at a bulk copper electrode decayed during successive cathodic scans to 2.0 V
versus sce or less for 10 min with the electrode blackened
meanwhile. The activity of the electrode was regenerated by
anodic polarization at 0.15 V versus sce.
Kyriacou and Anagnostopoulos [6] reported deactivation
in the CO2 reduction using high purity copper foil electrodes
in 0.5 M KHCO3 (Aldrich, reagent grade) solution prepared
from doubly distilled water. In their electrolysis measurements at 1.85 V versus sce, the current efficiencies of the
products reached maxima (CH4 31%, C2 H4 10%, CO 3%),
and diminished to less than 5% in 10 min. Shiratuchi et al.
[7] reported poisoning of a Cu electrode in CO2 reduction
in 0.1 M KHCO3 . They could suppress the poisoning of the
electrode by applying anodic polarization pulses to the Cu
electrode periodically during the cathodic reduction of CO2 .
Jermann and Augustynski [8] observed severe deactivation
of a Cu electrode. They used 0.5 M NaHCO3 and KHCO3
(pro analysi, Merck) as the electrolyte solutions prepared
from doubly distilled water. Friebe et al. [9], Lee and Tak
[10], and Yano et al. [11] also reported the deactivation of Cu
electrode in electrochemical reduction of CO2 . Some authors
investigated deactivation of Au or Ag metal electrode in CO2
reduction. The behaviors of these metal electrodes are similar
to those of Cu electrode.
Frese referred in his monograph [12] to the deactivation
possibly derived from impurity metals deposited on the surface of the Cu electrode after electrolytic reduction of CO2 in
0.5 M KHCO3 . He suggested that main impurity substances
are Fe and Zn.
Kyriacou and Anagnostopoulos [6], and Jermann and Augustynski [8] attempted to examine the reason of the deactivation. Both of them preelectrolyzed the electrolyte solution
with platinum electrodes, and found little improvement in
suppression of the deactivation. They analyzed the surface of
the used electrode by energy dispersive spectroscopy (eds) [6]
or xps [8], and did not detect impurity metallic traces on the
electrode surface. Both of them concluded that deposition of
metallic impurities of the solution is not the main cause of the
electrode poisoning. They presumed that the deactivation

5355

was derived from adsorption of organic intermediate formed

in CO2 reduction [6], or formation of a substance related with
oxide or hydroxide formed on the copper electrode [8].
1.2. Possible causes of the deactivation of Cu
electrode
The electrochemical reduction of CO2 is an interfacial
reaction, and must naturally be sensitive to the state of the
electrode surface. Thus one could presume that the deactivation of Cu electrode may be derived from the following
causes.
(1) Heavy metal impurities contained in reagent chemicals
at extremely low level are introduced to the electrolyte
solution. Such heavy metal ions in the electrolyte solution will be cathodically reduced and deposited at
the electrode during the CO2 reduction. Metal electrodes other than Cu do not have electrocatalytic activity for hydrocarbon formation in CO2 reduction. Thus,
any deposited heavy metals except Cu will deteriorate
the electrocatalytic activity of Cu electrode for CO2
reduction.
(2) Very small amount of organic substances are possibly
contained in water, such as surface active reagents. They
are sometimes hard to remove even by distillation [13];
these substances may be adsorbed on the electrode during
the CO2 reduction, subsequently deactivating the electrode.
(3) In accordance with some previous workers mentioned
above, some intermediate species or products formed
during the CO2 reduction, adsorbed on the electrode,
might poison the subsequent reaction.
As described in the preceding section, the deactivation
or poisoning has a common concept that the formation of
CH4 or C2 H4 is depressed during the CO2 reduction. However, it is noticeable that the time elapsed till appearance
of the deactivation scatters widely, depending greatly on the
research groups. The appearance of the surface after the electrolysis depends on the individual research groups. Some
workers reported surface blackening and others not. These
widely scattered facts strongly suggest that the poisoning
or deactivation may be involved with experimental procedure of each research group, not from any intrinsic factors,
such as adsorption of intermediate species or products formed
during CO2 reduction.
1.3. Purpose of this article
It is well-known that trace amount of impurities interferes
with the surface process in electrochemical reactions on the
electrodes. Many workers studied hydrogen evolution reaction (her) on various metal electrodes since Tafels work.
They attempted to obtain the correlation between the hydrogen overvoltage and the properties of the electrode material.
During the course of these works, widely scattered voltage

5356

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

current relations were published by different workers. Kuhn

reviewed the previous works, and decided to lay down criteria to select the experimental data for their discussions [14].
They chose the data by only those workers who had used
preelectrolysis for the purification of the electrolyte solution.
Bockris described the importance of preelectrolysis of the
electrolyte solutions in the study of electrode process in detail on the basis of his own studies in long years as well as
the preceding studies [15,16].
We showed in our early paper that an In electrode loses
the electrocatalytic activity in CO2 reduction to formic acid
during prolonged electrolyses [17]. We pointed out that the
electrode was poisoned by some electrodeposition of heavy
metal impurities initially contained in the electrolyte solution. We employed electrolyte solutions purified by preelectrolysis for all the studies of CO2 reduction since then,
and showed that preelectrolysis of the electrolyte solution
is effective to prevent such deactivation in CO2 reduction
[13,18,19]. We took it for granted that the electrolyte solutions should be properly purified in any study of interfacial
electrode process. We have thus affirmed that the electrolyte
solutions for electrochemical reduction of CO2 need to be
purified by preelectrolysis to remove heavy metal impurities
possibly contained in chemical reagents [1,3,19]. Nevertheless, some recent papers still mention deactivation of Cu
electrode without referring to purification of the electrolyte
solution.
This article shows the results of CO2 reduction with aqueous electrolyte solutions of KHCO3 supplied from various
sources, used in the previous works. The deactivation of
Cu electrode takes place more or less in all the cases, and
the features of the deactivation of Cu electrode depend on
the reagent used for the electrolyte solution. The deactivation is efficiently suppressed if the electrolyte solution is
appropriately purified. Publication of this article attempts to
prevent further possible confusions in future with regard to
the electrochemical reduction of CO2 .

2. Experimental
2.1. Electrochemical technique
A two-compartment electrolysis cell was used with the
catholyte separated from the anolyte with a sheet of cation exchange membrane (Nafion N117). An electrodeposited polycrystal copper sheet (purity 99.999%) was cut into an electrode (14.1 mm 5.1 mm 1.4 mm), polished with emery
paper (No. 1500), and electrolytically polished in conc.
H3 PO4 . The electrode was kept cathodic when dipped in the
electrolyte and during the following processes. An Ag/AgCl
reference electrode was used in this study, and all the potentials are given with respect to she in this article.
The CO2 reduction was carried out with constant current
electrolysis for convenience of analytical procedure. We compared some results obtained by constant current electrolyses

with those with controlled potential electrolyses, and no important difference was found between the two electrolysis
methods. The current density was 5 mA cm2 in the present
electrolysis measurements. CO2 was bubbled continuously
through the catholyte at ca. 100 ml min1 during the electrolyses. The gas samples were taken from the effluent from
the electrolytic cell every 10 min by a gas sampling valve and
analyzed by gas chromatographs.
We showed in our previous papers that the total Faradaic
balance holds for the products of the CO2 reduction with
exceptions of copper single crystal electrodes of some specified crystal orientations [20]. In the present study, we analyzed only gaseous products for the purpose of presenting
time courses of the deactivation process in the CO2 reduction. Other experimental details are found in our previous
papers.
2.2. Purity of the electrolyte solutions and the
experimental procedure of the preelectrolysis
Small amounts of heavy metal impurities are contained in
chemical reagents of the best quality according to the manufacturers, such as Aldrich, Fluka, Merck, Wako Jun-yaku, and
Nacalai Tesque. For example, KHCO3 may contain 25 ppm
of heavy metals as Pb in the maximum, and 5 ppm of Fe
in the maximum. Electrolyte solution of 0.1 M KHCO3 , frequently used in CO2 reduction, may thus contain maximum
of 0.020.05 ppm of heavy metals and 0.05 ppm of Fe, corresponding to maximum of 1.2 M of heavy metals and Fe in
total. If an electrode is cathodically polarized, an appreciable fraction of the electrode surface will be covered by these
metals in a short term; the electrocatalytic activity of the electrodes is thus severely deteriorated as is discussed later. Such
a low level of heavy metals in solution can be detected only
by highly sophisticated atomic spectroscopy.
The preelectrolysis as a procedure for purification
of electrolyte solution consists of scavenging the solution
with an cathodically polarized large-area electrode. Platinum
black is usually used as the cathode [15]. Otherwise, the
electrodeposited impurities might easily be dissolved into
the electrolyte solution, when the electrode is removed from
the electrolysis cell after the preelectrolysis. Electrochemical measurements can be conducted in the same electrolysis
cell that is used for the preelectrolysis. Since the purified
electrolyte is not transferred from the cell to any other container, no contamination of the electrolyte can happen. The
operation can be easily carried out in an ordinary chemical
laboratory without any sophisticated device.
The catholyte KHCO3 aqueous solutions (ca. 100 ml)
was prepared from reagent chemicals and ultrapure water
supplied from Milli Q low TOC (Millipore). The KHCO3
reagents from several manufacturers are used for the measurements; A.C.S. reagent by Aldrich, puriss p.a. ACS
by Fluka, GR for analysis by Merck, GR by Wako Junyaku, and GR by Nacalai Tesque. We carried out the electrochemical reduction of CO2 using these electrolyte so-

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

3. Results and discussion

Table 1
Temperature program of the graphite furnace
Step
Drying
Ashing 1
Ashing 2
Atomization
Cleaning

Temperature ( C)
70100
1100
1800
2700
2800

5357

Ramp (s)

Hold (s)

40
0
0
0
0

0
5
10
5
6

lutions with and without preelectrolysis for comparison.

The preelectrolysis of the catholyte was conducted using a
20 mm 30 mm Pt black cathode at 2.5 105 A cm2 under purified Ar atmosphere for more than 16 h. The electrolyte solution was mildly stirred by a magnetic stirrer with
ca. 220 rpm. The CO2 reduction was conducted immediately
after the preelectrolysis treatment in the same electrolytic
cell.
2.3. Analysis of the electrolyte solutions
In order to verify the effect of the preelectrolysis, we analyzed the electrolyte solutions by electrothermal atomic absorption spectrometry (etaas). The sample solutions were
treated in a clean box, CYH-2 manufactured by Yamato Kagaku Co. Ltd. The etaas measurements were conducted with a Hitachi Z-5000 atomic absorption spectrometer with a polarized Zeeman-effect background corrector. An autosampler was used to inject sample solutions
to a pyrolytic graphite furnace. Prior to atomization, the
sample solutions were dried and subsequently treated at
high temperature for ashing. Potassium, present in the electrolyte solution in high concentration, was eliminated from
the sample by this procedure; the interference of potassium with the measurements of Fe was successfully prevented. The temperature program of the furnace is given in
Table 1.

3.1. Time courses of CO2 reduction at a Cu electrode in

electrolyte solutions from various sources without
purication
Figs. 13 illustrate time courses of the electrode potential and the current efficiencies of various gaseous products
in CO2 reduction at a constant current density 5 mA cm2 .
Aqueous electrolytes 0.3 M KHCO3 were employed without
any purification. The reagents KHCO3 used in the present
work include those employed by the previous workers who
reported the deactivation of Cu electrodes in CO2 reduction.
The current efficiencies of the gaseous products from CO2 ,
typically observed with CH4 , rise at the initial stage, and subsequently decayed simultaneously accompanied with H2 increase. In our experimental equipment, the gas sample was
taken from the effluent gas from the electrolysis cell by a
sampling valve as mentioned above. A delay of the detection
of the gaseous products occurs due to the dead space from
the electrolysis cell to the gas sampling valve. Thus, the current efficiencies of gaseous products are relatively low at the
early stage of the electrolysis, i.e. at 510 min as shown in
Figs. 13.
Fig. 1 presents a result from the reagents from Fluka. After
20 min or more, the deactivation of Cu electrode in CO2 reduction takes place in agreement with the descriptions by the
previous workers [46,711]. The current efficiency of CH4
decreases gradually, accompanied with very low current efficiency of C2 H4 . Hydrogen evolution, relatively high in the
current efficiency from the initial stage of the electrolysis,
prevails with the elapsed time. We conducted identical measurements with the reagents from Nacalai Tesque and Aldrich
as well. The features of the time courses are similar to those
in Fig. 1 with small differences of the rate of the deactivation.
Fig. 2 shows a result from the reagent from Wako. The current efficiency of CH4 remains higher than 60% for 70 min,

Fig. 1. A time course of the current efficiencies of the gaseous products of electrochemical reduction of CO2 for 120 min with an electrolyte solution of 0.3 M
KHCO3 prepared from the chemical reagent by Fluka without preelectrolysis. The figure in the right hand side shows a time course of repeated CO2 reduction
after a series of procedures of the CO2 reduction for 120 min, a cyclic voltammetry between 0.9 and 0.2 V vs. she, and an anodic polarization at 0.05 V for
5 min.

5358

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

Fig. 2. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 prepared from the chemical reagent by Wako without preelectrolysis.
The right hand side shows a time course of repeated CO2 reduction after an anode stripping voltammetry and an anodic polarization at 0.05 V for 5 min. All
the procedures are identical with those in Fig. 1.

dropping slightly thereafter. C2 H4 formation is much higher

than the result given in Fig. 1. Hydrogen evolution is low,
growing slowly during the electrolysis. Fig. 3 shows a time
course with Merck reagent. The result resembles Fig. 2. The
degradation of CH4 and C2 H4 formation is slower than those
shown in Fig. 1, and the hydrogen evolution is lower.
The time courses show remarkable features which depend
on the individual reagent used. The rate of the deactivation
will be discussed later in more detail.
Immediately after the electrolyses of CO2 reduction for
120 min, cyclic voltammetric measurements were performed
starting from 0.9 to 0.2 V. Then, the electrode was anodically polarized at 0.05 V for 5 min, where the Cu electrode is not anodically oxidized. The electrolyses were repeated with the current density 5 mA cm2 for 40 min. The
results are given in the right hand side of Figs. 13. CH4 formation again increased with hydrogen formation depressed.
The electrode activity is evidently recovered by the anodic

polarization, in agreement with the reports by other workers

[5,710].
We studied CO2 reduction with electrolyte solutions of
higher concentration. Fig. 4 shows a result with 0.5 M
KHCO3 from Aldrich. The electrode was poisoned at a very
early stage of the electrolysis as shown by low current efficiency of CH4 . The CH4 formation was prohibited after
60 min of electrolysis, and hydrogen formation prevailed. The
deactivation is thus significantly accelerated by the increase
of the electrolyte concentration.
The appearance of any electrode did not change, giving
metallic luster after the electrolysis.
Votammograms obtained immediately after the electrolyses for 120 min of CO2 reduction are presented in Figs. 58
with the electrolyte solutions prepared from the reagents of
Fluka, Wako, Merck and Aldrich, respectively. Fig. 5 clearly
shows that the first anodic trace gives higher current at 0.1 V
than the second trace. Fig. 6 gives a slightly higher anodic

Fig. 3. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 prepared from the chemical reagent by Merck without preelectrolysis.
The right hand side shows a time course of repeated CO2 reduction after an anode stripping voltammetry and an anodic polarization at 0.05 V for 5 min. All
the procedures are identical with those in Fig. 1.

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

5359

Fig. 4. A time course of CO2 reduction with an electrolyte solution of 0.5 M KHCO3 prepared from the chemical reagent by Aldrich without preelectrolysis.
The right hand side shows a time course of repeated CO2 reduction after an anode stripping voltammetry and an anodic polarization at 0.05 V for 5 min. All
the procedures are identical with those in Fig. 1.

trace than the second one at 0.2 V with a small peak at

0.56 V. Fig. 7 shows a clear anodic peak at 0.56 V with
negligibly small increase of the anodic current at 0.1 V.
Fig. 8 shows an enhanced anodic current between 0.4 and
0.0 V, together with another anodic peak at 0.6 V in the
second scan.
The anodic charge obtained from the difference between
the first and the second trace may be attributed to some substance which is anodically oxidized in the anode stripping
procedure. A comparison of these voltammograms indicates
that higher anode stripping electric charge at 0.2 to 0.1 V
is related to more accelerated deactivation in CO2 reduction,
as discussed later.

Fig. 5. Cyclic voltammogram with the scan rate of 100 mV s1 between

0.9 and 0.2 V vs. she after CO2 reduction with an electrolyte solution
of 0.3 M KHCO3 prepared from the chemical reagent by Fluka without
preelectrolysis, as presented in Fig. 1.

3.2. Electrochemical reduction of CO2 in electrolyte

solutions treated with preelectrolysis
Figs. 911 show time courses of CO2 reduction with electrolyte solutions purified by preelectrolysis. Fig. 9 may be
compared with Fig. 1; both of the time courses were obtained
with electrolyte solution 0.3 M KHCO3 from Fluka. Fig. 9
shows that the current efficiency of CH4 formation, much
higher than that in Fig. 1, remains above 40% at 120 min.
Hydrogen evolution is much lower than that in Fig. 1. The
deactivation of the electrode is obviously prevented, whereas
CH4 formation still declined slightly and hydrogen evolution grew slowly during the electrolysis. The anode stripping
voltammogram given in Fig. 9 shows a negligibly small an-

Fig. 6. Cyclic voltammogram with the scan rate of 100 mV s1 between

0.9 and 0.2 V vs. she after CO2 reduction with an electrolyte solution
of 0.3 M KHCO3 prepared from the chemical reagent by Wako without
preelectrolysis, as presented in Fig. 2.

5360

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

Fig. 7. Cyclic voltammogram with the scan rate of 100 mV s1 between

0.9 and 0.2 V vs. she after CO2 reduction with an electrolyte solution
of 0.3 M KHCO3 prepared from the chemical reagent by Merck without
preelectrolysis, as presented in Fig. 3.

Fig. 8. Cyclic voltammogram with the scan rate of 100 mV s1 between

0.9 and 0.2 V vs. she after CO2 reduction with an electrolyte solution
of 0.5 M KHCO3 prepared from the chemical reagent by Aldrich without
preelectrolysis, as in Fig. 4.

Fig. 9. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 prepared from the chemical reagent by Fluka with preelectrolysis. The
right hand side shows a time course of repeated CO2 reduction after an anode stripping and an anodic polarization. All the procedures except the preelectrolysis
are identical with those in Fig. 1. A cyclic voltammogram after the CO2 reduction with the scan rate of 100 mV s1 is also presented.

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

5361

Fig. 10. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 prepared from the chemical reagent by Wako with preelectrolysis. The
right hand side shows a time course of repeated CO2 reduction after an anode stripping and an anodic polarization. All the procedures except the preelectrolysis
are identical with those in Fig. 1. A cyclic voltammogram after the CO2 electrolysis with the scan rate of 100 mV s1 is also presented.

odic peak at 0.1 V where a small but clear anodic peak

appeared in Fig. 5.
Fig. 10 illustrates the results obtained using Wako 0.3 M
KHCO3 which was purified by preelectrolysis. Fig. 10 may
be compared with Fig. 2, which resulted from electrolyte solution prepared from Wako reagent without preelectrolysis.
The current efficiencies of CH4 and C2 H4 in Fig. 10 remained
nearly constant during the electrolysis. H2 formation, increasing slowly for 120 min in Fig. 2, is greatly suppressed during
the electrolysis shown in Fig. 10. The deactivation caused by
Wako reagent is not very severe even without preelectrolysis,
and the electrode activity is further extended by preelectrolysis of the electrolyte solution. The anode stripping voltammogram in Fig. 10 shows still a small anodic peak at 0.1 V
and a small one at 0.56 V.
A severe poisoning was observed with 0.5 M KHCO3
from Aldrich used without preelectrolysis as shown in Fig. 4.
Fig. 11 shows the result with 0.5 M KHCO3 from Aldrich
treated by preelectrolysis. CH4 formation was enhanced to
a great extent together with higher and stable production of
C2 H4 . H2 evolution is suppressed. The electrocatalytic activity in 0.5 M KHCO3 from Aldrich is remarkably improved by
preelectrolysis, whereas the deactivation of the electrode still

takes place as shown by the slow decline of CH4 formation

and the growth of H2 evolution during the electrolysis. The
voltammetric measurement shows that the large anodic peak
at 0.15 V in Fig. 8 becomes negligibly small as shown in
Fig. 11.
An identical measurement was carried out with an electrolyte solution prepared from Merck reagent and purified by
preelectrolysis; the deactivation was also suppressed. CH4
and C2 H4 formations were maintained high for longer time
and hydrogen evolution grew more slowly in comparison with
Fig. 3. The anodic peak at 0.56 V (Fig. 7) became negligibly
lower.
The experimental facts described above have verified that
purification of the electrolyte solution by preelectrolysis effectively suppresses the deactivation of Cu electrode in the
electrochemical reduction of CO2 .
Two research groups, Kyriacou and Anagnostopoulos [6],
and Jermann and Augustynski [8], conducted preelectrolysis
of the electrolyte solutions with Pt plate electrodes prior to
the CO2 reduction. Nevertheless, both of them found no improvement for suppression of the deactivation. Kyriacou and
Anagnostopoulos analyzed the Cu electrode surface by eds
after CO2 reduction, and Jermann and Augustynski employed

5362

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

Fig. 11. A time course of CO2 reduction with an electrolyte solution of 0.5 M KHCO3 prepared from the chemical reagent by Aldrich with preelectrolysis. The
right hand side shows a time course of repeated CO2 reduction after an anode stripping and an anodic polarization. All the procedures except the preelectrolysis
are identical with those in Fig. 1. A cyclic voltammogram after the CO2 reduction with the scan rate of 100 mV s1 is also presented.

xps for the same purpose. Both of the two groups detected
no impurity metals responsible to the deactivation of the Cu
electrode. However, they did not refer to the detection limit
and the reliability of the analysis of impurity metals on the
electrode at all. It is difficult to analyze any impurity substances far below monolayer coverage without an extreme
care. In addition to the analytical procedures, the details of
their experimental procedures are not presented in their papers; for example it is not clear whether they used a Pt black
electrode for the preelectrolysis. Thus we cannot discuss the
discrepancy between our results and theirs any more.
3.3. Deactivation caused by electrodeposition of
impurity substances on the electrode
3.3.1. A rough numerical estimation
As described above, the deactivated Cu electrodes are regenerated by anode stripping to 0.2 V followed by anode polarization at 0.05 V. The electric charge of the oxidation
peaks in the anode stripping is related with the decay rate
of CH4 formation. The deactivation is accelerated by the in-

crease of the concentration of the electrolyte. These experimental facts strongly suggest that the deactivation is caused
by some metallic impurity substances which are deposited on
the electrode surface from the electrolyte solution during CO2
reduction.
If the electrode is cathodically polarized in CO2 reduction
at the current density of 1 mA cm2 for example, electrodeposition of cationic impurity substances on the electrode will
proceed in a limiting current condition owing to the low concentration. A simple estimation of the limiting current density
il is given, with the number of the electrons involved with the
electrodeposition assumed to be 2,


D
(C)
(1)
il = (2)(F )

where F denotes the Faraday constant, D the diffusion constant, the diffusion layer thickness, and C the concentration
of the cationic impurity substance. C is taken constant during
the CO2 reduction for the sake of simplicity. Using numerical values for the variables, D 0.72 105 cm2 s1 for Fe2+
[21] for example, 0.0025 cm according to our measurement

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

with similar hydraulic conditions [22], and C 1 M. il is thus

5.6 107 A cm2 . If the electrode is negatively polarized
for CO2 reduction for 10 min, the deposited impurity substance would amount to 334 C cm2 in the electric charge.
This value corresponds to a considerable fraction of the coverage of the electrode.
3.3.2. Reproduction of the deactivation by deliberate
addition of impurity cations in the electrolyte solution
Experimental reproduction of the deactivation is
demonstrated by deliberate addition of some cations in the
electrolyte solution. We conducted electrochemical reduction
of CO2 in electrolyte solutions purified by preelectrolysis and
added with some cations. A part of the similar results were
published in our previous communication [22].
On the basis of the anodic peaks at 0.1 and 0.56 V in
the anode stripping voltammograms (Figs. 58), one could
tentatively assume that the major impurity substances in the
electrolytes may be Fe and Zn. This assumption is in agreement with a previous mention by Frese [12]. According to
a thermodynamic consideration, the equilibrium potential of

5363

Fe/Fe2+ is 0.65 V at Fe2+ concentration 107 M, and that of

Zn/Zn2+ 0.97 V at Zn2+ concentration 107 M, respectively
[23]. If one takes into account underpotential deposition (upd)
on a Cu electrode, the potential of the anode stripping peak
may be shifted to the positive direction to a considerable extent, such as 0.40.6 V [24].
We employed KHCO3 solutions prepared from Wako and
Nacalai reagents and treated them with preelectrolysis. A
small aliquot of FeSO4 or ZnSO4 solution was added to thus
purified electrolyte solutions to a certain definite concentrations.
Fig. 12 demonstrates a time course of CO2 reduction at
a constant current density of 5 mA cm2 in 0.3 M KHCO3
with Fe2+ adjusted at 0.1 M. Degradation of the electrode
activity during the electrolysis is obviously shown; the formation of CH4 and C2 H4 drops quickly in 4050 min, whereas
H2 evolution prevails meanwhile. After the electrolysis of
CO2 reduction for 120 min, an anode stripping and subsequent anode polarization at 0.05 V were also conducted as
an identical procedure described previously. CO2 reduction
was repeated after these treatment. As shown in the right side

Fig. 12. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 deliberately added with 0.1 M Fe2+ . The electrolyte solution was
prepared from the chemical reagent by Wako and treated with preelectrolysis prior to the addition of 0.1 M Fe2+ . The right hand side shows a time course of
repeated CO2 reduction after an anode stripping and an anodic polarization. A cyclic voltammogram after the CO2 reduction with the scan rate of 100 mV s1
is also presented.

5364

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

Fig. 13. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 from the chemical reagent by Wako treated with preelectrolysis and
deliberately added with 0.05 M Fe2+ . Other points are identical with Fig. 12.

of Fig. 12, the electrode activity was recovered, but decayed

in a short period. Fig. 13 shows another time course with
Fe2+ at 0.05 M. The decay of the electrocatalytic activity
is obviously shown as well with much slower rate than that
shown in Fig. 12. The anode stripping voltammetry indicates
an anodic peak at 0.10 V (Figs. 12 and 13). The potential
of the anodic peak shown in Figs. 5, 6 and 8 agrees with that
shown in Figs. 12 and 13. Thus, the assumption is valid that
Fe2+ in the electrolyte may be a major contaminant causing
the deactivation of the Cu electrode shown in Figs. 1 and 2,
although one cannot exclude a possibility from other impurity
substances responsible to the deactivation.
Fig. 14 shows a time course of CO2 reduction at a constant
current density of 5 mA cm2 in 0.3 M KHCO3 with Zn2+
adjusted at 0.1 M. The anode stripping voltammogram is
also given. A degradation of the electrode takes place, and
the features of the time course resemble those in Fig. 3. The
anodic peak appears at 0.57 V nearly the same potential
shown in Fig. 7. The anode stripping process will oxidize Zn
deposited on the surface to Zn(OH)2 which remains partly
on the electrode surface due to low solubility in accordance
with the pHpotential diagram given in Pourbaix diagram

[25]; Zn(OH)2 on the electrode surface will be reduced again

to metallic Zn in the reverse cathodic scan. Thus, the second
and third anodic scans give smaller anodic peaks than the first
scan at almost the same potential.
As described above, the deactivation is obviously reproduced by deliberate addition of Fe2+ and Zn2+ to the electrolyte solution purified by preelectrolysis. The description
above evidences that purification of the electrolyte solution
by the preelectrolysis effectively suppresses the deactivation of Cu electrode. It is also evident that the deactivation
of Cu electrode in CO2 reduction is not caused by adsorption of the products or the intermediates produced in CO2
reduction.
3.3.3. Decay rate index
The deactivation of Cu electrode during the electrolysis is
represented by the decay of the rate of formation of CH4 and
C2 H4 as demonstrated in Figs. 14 in Faradaic efficiency. We
define the decay rate index (dri) as the reciprocal of the time
elapsed in which the rate of CH4 formation decreases to 1/2
the maximum rate (t1/2 ). The rate of CH4 formation should
be initially maximum when the electrode surface is not con-

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

5365

Fig. 14. A time course of CO2 reduction with an electrolyte solution of 0.3 M KHCO3 from the chemical reagent by Wako treated with preelectrolysis and
deliberately added with 0.1 M Zn2+ . Other points are identical with Fig. 12.

taminated, but the maximum actually appears at 1520 min

due to the presence of the dead space between the electrolysis
cell and the gas sampling valve for the gas chromatographs
as mentioned above. Table 2 presents the values of t1/2 and
dri for CO2 reduction in the electrolyte deliberately added
with Fe2+ or Zn2+ .
The dri value for Entry 2 is approximately twice that of
Entry 1, in agreement with the difference of the Fe2+ concentration added to the electrolyte. The rate of Fe deposition
on the electrode would be proportional to the concentration
of Fe2+ in the electrolyte, as shown by (1). The dri value for

Entry 3, conducted under identical experimental conditions

with Entry 2, is close to that for Entry 2. A small discrepancy
between the two may be derived from the hydraulic conditions not sufficiently reproducible in the present experimental
setup.
The time course for Entry 4 was similar to Fig. 4. No
maximum of CH4 formation rate was observed; the deactivation proceeds too fast and the maximum rate of the CH4
formation probably appeared before the first gas chromatographic measurement at 10 min. We took the rate at 10 min
as the maximum rate in this case. The real t1/2 value must be

Table 2
Decay rate index and the electric charge of the anodic peak in the anode stripping voltammetry in the CO2 reduction electrolysis in the electrolyte solutions
deliberately added with heavy metal ions
Entry No.

Added
ion

Concentration
(M)

KHCO3 concentration
(M)

Electrolysis
time (min)

t1/2
(min)

dri
(min1 )

Electric charge
(C cm2 )

Peak potential
(V)

1
2
3
4
5

Fe2+
Fe2+
Fe2+
Fe2+
Zn2+

0.05
0.1
0.1
0.2
0.1

0.3
0.3
0.3
0.1
0.3

120
120
120
60
120

95.7
48.5
40.0
<18.9
88

0.010
0.021
0.025
>0.053
0.011

10.5
74.1
81.9
109
225

0.13
0.08
0.08
0.06
0.57

Source of KHCO3 used in the experiments. Entries 1, 2, 5: Wako. Entries 3, 4: Nacalai Tesque. All the electrolyte solutions were purified by preelectrolysis,
and deliberately added with Fe2+ or Zn2+ as designated in the table.

5366

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

shorter than the time at 1/2 the rate at 10 min. Thus, the dri
for Entry 4 is twice or more that of Entry 2 or 3, compatible
with the variation of the Fe2+ concentration.
Table 2 also presents a dri value for added Zn2+ concentration 0.1 M. The dri value is remarkably smaller than those
for the identical concentration of Fe2+ , responding to milder
deactivation of the Cu electrode.
3.3.4. Electric charge of the impurity metals deposited
on the electrode
We determined the electric charge of the anodically
stripped Fe or Zn deposited on the electrode during CO2
reduction in the electrolytes deliberately added with Fe2+
or Zn2+ . Integration of the anodic charge in Fig. 12, conducted between the first and second scanned traces, gives
74.1 C cm2 . Table 2 indicates the results determined in
the identical manner together with the potentials of the anodic peaks. Integration of the anodic peak of Zn in Fig. 14
was conducted between the first scanned trace and the base
line. Some data were obtained from cyclic voltammograms
not given in the present paper.
The dri is proportional to the Fe2+ concentration in the
electrolyte as mentioned above. Thus one can expect that the
electric charge of the anodic peak would also be proportional
to the Fe2+ concentration in the electrolyte, if the time elapsed
during CO2 reduction is constant. Nevertheless, the electric
charge of the anodic peak of Entry 1 in Table 2 is much less
than 1/2 that of Entry 2.
The diffusion coefficient of Zn2+ (0.703 105 cm2 s1
[21]) is close to that of Fe2+ given in the previous section.
Thus, the amount of the deposited Fe during the CO2 reduction would be close to that of the deposited Zn in accordance
with Eq. (1), if the measurement is carried out under an equal
concentration of Fe2+ and Zn2+ in a constant electrolysis time.
Table 2 shows that the electric charge given in Entry 2 or 3
for Fe2+ is ca. 1/3 that presented in Entry 5 for Zn2+ .
These facts suggest that the electric charge of the anodic peak at 0.1 V in the anode stripping voltammogram
in Fig. 12 does not corresponds to all the Fe deposited on the
electrode. A part of Fe, strongly bonded to the surface Cu
atoms, is not oxidized in a distinctive peak near 0.1 V. Such
Fe atoms may be oxidized at more positive potential, and the
anodic current will coalesce into that of the anodic oxidation
of the Cu electrode. Thus, it is impossible to measure accurately the total amount of the electric charge of the deposited
Fe on the Cu electrode by the present anode stripping voltammetry. A small anodic peak in the voltammogram in Fig. 13
suggests that most of the Fe deposited in an early stage would
be more strongly bonded to Cu atoms.
Fe, not oxidized in the anodic peak near 0.1 V, would be
oxidized to Fe(OH)2 or Fe(OH)3 at more positive potential.
These substances are little soluble in the pH range of the electrolyte according to Pourbaix diagram [25], and will not be
totally removed from the electrode surface in a way similar to
the voltammogram in Fig. 12. The subsequent prolonged polarization at 0.05 V will not totally remove the Fe remaining

on the electrode surface. As shown in Figs. 1214, the CO2

reduction is recovered by the anodic polarization at 0.05 V.
However, Fig. 12 shows that the CH4 formation after the anodic polarization is not perfectly recovered, and the current
efficiency of the CH4 formation is remarkably lower than that
in the initial part of the time course. Fig. 13 shows a similar
trend, but the CH4 formation after the anodic polarization is
relatively higher than that in Fig. 12, probably because the
amount of the deposited Fe is much less than that in Fig. 12.
Fig. 14, obtained in an electrolysis with added Zn2+ , shows
that the CH4 formation is entirely recovered after the anodic
polarization at 0.05 V, indicating that the deposited Zn is
thoroughly removed from the electrode surface.
3.3.5. Coverage of impurity metals on the Cu electrode
The amount of metals deposited on the electrode during
CO2 reduction may be roughly estimated in terms of coverage on the electrode using the electric charges of the anode
stripping voltammetry.
Surface atom density of Cu single crystals, calculated from
the atomic radii of Cu in metal (0.1278 nm [26]), is 1.77, 1.53,
and 1.08 in the unit of 1015 per 1 cm2 of the single crystal
surfaces of (1 1 1), (1 0 0) and (1 1 0), respectively. Let us
take tentatively the surface atom density of the polycrystal
Cu electrode in the present study as 1.5 1015 cm2 .
A Zn atom in the anode stripping will be involved with
two electrons, and the electric charge of the stripped Zn,
225 C cm2 as given in Entry 5 in Table 2, corresponds to
0.70 1015 cm2 in terms of the surface atom density. The
atomic radii of Zn (0.1332 nm [26]) is slightly larger than
that of Cu, and the coverage of Zn atoms on Cu surface may
be simply defined by the ratio of Zn atoms to Cu atoms on
the surface. The coverage of Zn atoms thus obtained amounts
approximately to 47% of the surface Cu atoms.
If an identical procedure is applied to Fe of Entry 2 or
3 in Table 2, one can estimate the coverage as 17%, taking
the atomic radii of Fe (0.1241 nm [26]). However, the real
coverage of Fe would be close to that of Zn, i.e. 47%, due
to nearly same values of the diffusion constants as discussed
above.
3.3.6. Comparison of Fe and Zn in the deactivation of
Cu electrode in CO2 reduction
The features of the deactivation of Cu electrode depend
greatly on the deposited substance. The dri value of Entry 1 in
Table 2, determined in CO2 reduction in the electrolyte added
with 0.05 M Fe2+ (Fig. 13), is approximately equal to that of
Entry 5 obtained with twice concentration of Zn2+ (0.1 M)
(Fig. 14). Thus, the poisoning effect of Fe with regard to dri
is twice that of Zn.
Zn atoms affects CO2 reduction to a moderate extent.
Fig. 14 shows that CH4 formation rate decays to ca. 1/2
the maximum rate in 120 min of CO2 reduction in 0.1 M
Zn2+ added electrolyte, while 47% of the surface Cu is covered with deposited Zn atoms. Meanwhile, CO formation increases to 20% of the total current efficiency, roughly equal to

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

5367

Table 3
dri of the CO2 reduction, and the potential and electric charge of the voltammetric peak derived from the impurity substances in the electrolyte solution
Entry No.

11
12
13
14
15
16

dri (min1 )

KHCO3 solution
Source

Concentration (M)

Nacalai
Fluka
Aldrich
Wako
Merck
Aldrich

0.3
0.3
0.3
0.3
0.3
0.5

Voltammetric peak

0.010
0.016
0.013
<0.008
<0.008
0.033

the fraction of CH4 lost during the CO2 reduction. This fact
obviously shows that Cu atoms not covered with Zn atoms
are still electrocatalytically active in hydrocarbon formation,
and Cu atoms covered with Zn atoms reduce CO2 to CO.
On the other hand, deposition of Fe severely prevents hydrocarbon formation on the Cu electrode, leading to suppression of CH4 formation accompanied with prevalent hydrogen
evolution as shown in Fig. 12. The electrocatalytic property
of Cu surface for hydrocarbon formation is entirely lost by
deposited Fe with the presumed coverage 47%, equivalent to
Zn as mentioned above.
We thus presume that Fe atoms interact electronically with
Cu atoms by forming a surface alloy, entirely deteriorating
the electrocatalytic nature in CO2 reduction, whereas Zn is
adsorbed on the electrode, merely blocking Cu atoms by a
steric effect.

Peak potential (V) vs. she

Electric charge (C cm2 )

0.05
0.05
0.05
0.05
0.56
0.15

15.0
37.6
27.3
7.7
56.3
150.5

solutions employed in CO2 reduction. The results are shown

in Table 4.
Table 4 also indicates another estimation of the initial concentration of impurity ions in the electrolyte solutions on the
basis of the electric charge and the potential of the peaks in
the anode stripping measurements. The anodic peaks near
0.1 V are broadened, and the peak potentials are not definitely determined. Thus the peak potentials between 0.05
and 0.15 V are assumed to be contributed from the anodic
dissolution of Fe from the Cu electrode, and the anodic peak
near 0.56 V was compared with Zn in the Entry 5 in Table 2.
The initial concentrations of Fe2+ and Zn2+ in the electrolytes
are thus estimated by interpolation of the data in reference to
the electric charges given in Table 2.
The results obtained by the two separate methods agree
pretty well with each other. The initial concentrations of the
impurity ions in the electrolyte solutions may lead to estimation of the amount of the impurity ions originally contained
in the chemical reagents. The original concentration of impurity ions is also tabulated in Table 4. We presumed that the
substance relevant to Entries 1114 and 16 is Fe2+ , and that
relevant to Entry 15 is Zn2+ . Entries 13 and 16, obtained in
the measurements with different electrolyte concentrations
from the same chemical reagent, give close values for the
original concentration of the impurity ions. The agreement
confirms validity of the present analysis. The concentrations
of the impurity ions in the chemical reagent are extremely
low, far below the levels guaranteed by the manufacturers.
We would like to emphasize that such low concentrations of
Fe2+ or Zn2+ from reagents of the best quality may affect the
electrocatalytic property in CO2 reduction.

3.4. Estimation of the concentration of impurity

substances in the chemical reagents
Table 3 tabulates the dri values determined from Figs. 14.
The data given in Entries 11 and 13 were obtained from identical measurements not presented in this paper. The manufacturers of the reagent are specified. The current efficiencies of
CH4 formation in Figs. 2 and 3 do not drop to 1/2 of the maximum rate during the electrolysis measurements, and thus
t1/2 is greater than 120 min; the dri value must be lower than
1/120 min1 . The dri values thus obtained were compared
with those in Table 2. Interpolation of the data in Table 3 in
reference to the dri data in Table 2 gives a rough estimation
of the initial concentration of impurity ions in the electrolyte

Table 4
Estimated concentrations of the presumed impurity substances in the electrolyte solution and in the chemical reagent
Entry No.

11
12
13
14
15
16

KHCO3

Presumed substance

Source

Concentration (M)

Presumed
substance

Concentration in the
electrolyte, from dri
(M)

Concentration in the
electrolyte, from
charge (M)

Concentration in the
reagent (ppm)

Nacalai
Fluka
Aldrich
Wako
Merck
Aldrich

0.3
0.3
0.3
0.3
0.3
0.5

Fe2+
Fe2+
Fe2+
Fe2+
Zn2+
Fe2+

0.05
0.07
0.06
<0.04
<0.07
0.13

0.05
0.07
0.06
<0.05
0.025
0.15

0.09
0.13
0.11
<0.08
0.05
0.16

Concentrations of the presumed substances were estimated on the basis of dri and the electric charge of the peak in the anode stripping voltammograms.

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

5368

Table 5
etaas measurements of Fe contained in 0.1 M KHCO3 with and without preelectrolysis treatment

Milli Q

watera

Absorbance

Standard deviation

Slope (ppb1 )

Correlation coefficient

0.0074

0.0021

0.1 M KHCO3 (Fluka)b,c

Preelectrolyzed
Not preelectrolyzed

0.0079
0.0180

0.0017
0.0013

0.0092
0.0098

0.9981
0.9998

0.1 M KHCO3 (Aldrich)b,c

Preelectrolyzed
Not preelectrolyzed

0.0085
0.0130

0.0013
0.0015

0.0092
0.0096

0.995
0.988

Thirteen measurements were carried out with five samples for Milli Q water.
Ten measurements were carried out with four samples not added with the Fe standard solution.
c Dependence of the absorbance on Fe concentration was measured with standard addition method. Two samples were taken from four different concentrations
up to 10 ppb. Three measurements were made for every sample.
b

3.5. etaas analysis of the electrolyte solutions

etaas technique was employed to demonstrate the effect
of the preelectrolysis of the electrolyte solution in order to
confirm the discussion above. We determined Fe in 0.1 M
KHCO3 prepared from the reagents manufactured by Fluka
and Aldrich before and after the preelectrolysis; Fe causes
severer deactivation of the electrode. The analytical results
were sufficiently reproducible. We obtained dependence of
the absorbance on the concentration of Fe by standard addition method. The results are shown in Table 5 together with
the number of samples taken from the solutions as well as
the number of the repeated measurements. The absorbance
is linear against the concentration of Fe in the sample solution with high correlation coefficients. The values of the
slope, the standard deviation and the correlation coefficients
are tabulated in Table 5.
The absorbances of both preelectrolyzed solutions, from
Fluka and Aldrich, agree with that of the Milli Q water within
the standard deviation as shown in Table 5. Therefore, we can
reasonably judge that Fe is not detected beyond the background in the preelectrolyzed solutions. On the other hand,
the absorbance of Fe in not preelectrolyzed solutions gives
definite values. The concentration of Fe in the solution from
Fluka may be estimated as follows:

3.6. Deactivation of a copper electrode caused by

trimethylamine
The electrocatalytic activity of the copper electrode may
be heavily damaged by the presence of some organic substances. We demonstrate the effect of trimethylamine as an
example. Manufacturers of ion exchange resin apply tertiary
amines such as trimethylamine to introduce quarternary ammonium as an anion exchanger group to the resin [27]. Thus
slight amount of trimethylamine may occasionally remain in
anion exchange resin, and is sometimes dissolved in water
treated with such anion exchange resins.
Electrolyses of CO2 reduction were carried out at a copper
electrode in 0.1 M KHCO3 added with trimethylamine. The
results are shown in Fig. 15 with the average current efficiencies of some gaseous products. The electrocatalytic activity
for formation of CH4 and C2 H4 apparently drops by addition
of small amount of trimethylamine, whereas the current efficiencies of the products did not change with time during the
electrolyses. Since amines are strongly adsorbed on metals
as is well known, the copper electrodes with trimethylamine
adsorbed on the surface lose their electrocatalytic activity to
CO2 reduction. The surface of the electrode was probably in

0.0180 0.0074
= 1.08 ppb
0.0098
The uncertainty is derived from the standard deviation as,
0.0021/0.0098 = 0.21 ppb
We can obtain the values for the Aldrich reagent likewise;
0.58 0.22 ppb.Using these values, we can estimate the concentration of Fe in the reagent; 0.11 0.02 ppm for Fluka,
and 0.06 0.02 ppm for Aldrich. These values are in reasonable agreement with those indicated in Table 4. Thus the etaas
analysis of the electrolyte solutions has established that the
preelectrolysis treatment can effectively remove Fe in electrolyte solutions.

Fig. 15. Average current efficiencies of the gaseous products in CO2 reduction in 0.3 M KHCO3 from the chemical reagent by Wako in the presence
of various concentrations of trimethylamine. The electrolyte solution was
treated with preelectrolysis prior to the addition of trimethylamine.

Y. Hori et al. / Electrochimica Acta 50 (2005) 53545369

adsorption equilibrium before the electrolysis started, and the

current efficiencies of the products remained nearly constant
during the electrolyses.

5369

Acknowledgments
The authors are grateful to Professor M. Fujinami of Chiba
University for the discussion of the surface analysis of the
electrode.

4. Conclusion
Deactivation of a Cu electrode in electrochemical reduction of CO2 , reported by many workers, was reproduced using
electrolyte solutions prepared from chemical reagents used
by the previous workers. Deactivated Cu electrodes recovered the electrocatalytic activity for the electrochemical reduction of CO2 by anodic polarization at 0.05 V versus she,
in agreement with the previous reports. Features of the deactivation depend greatly on the individual chemical reagent.
Purification of the electrolyte solution by preelectrolysis with
a Pt black electrode effectively prevents the deactivation of
Cu electrode.
Anode stripping voltammetry was carried out with the
Cu electrodes poisoned in prolonged electrolysis of CO2
reduction, showing an anodic oxidation peak at ca. 0.1 or
0.56 V versus she. The severer the deactivation of the Cu
electrode was, the higher electric charge of the anodic peak
was observed. We thus presumed that some impurity heavy
metals, originally contained in the electrolyte, are deposited
during the CO2 reduction and poison the electrocatalytic
activity of the Cu electrode. Fe2+ and Zn2+ are assumed
to be the contaminants which cause the deactivation of the
Cu electrode on the basis of the potential of the anodic
peaks.
Deliberate addition of Fe2+ or Zn2+ to the electrolyte solution purified by preelectrolysis exactly reproduced the deactivation of a Cu electrode in the CO2 reduction. The amount
of the deposited Fe or Zn on the electrode was much below
the monolayer coverage. The concentration of the impurity
substances, originally contained in the chemical reagents as
Fe or Zn, was estimated far below the standard of the impurity
levels of the manufacturers.
etaas showed that Fe contained in the electrolyte solutions
is effectively removed by the preelectrolysis of the solution.
The amounts of Fe in the chemical reagents (KHCO3 ) as
determined by etaas agreed with those estimated from the
electrochemical measurements mentioned above.
Presence of trimethylamine in the electrolyte solution also
severely deteriorates the electrocatalytic property of the Cu
electrode.
We conclude that the deactivation of Cu electrode in CO2
reduction is not caused by adsorption on the electrode of the
products or the intermediates produced in CO2 reduction.
The deactivation can be effectively prevented by sufficient
purification of the electrolyte solution.

References
[1] Y. Hori, K. Kikuchi, S. Suzuki, Chem. Lett. (1985) 1695.
[2] Y. Hori, K. Kikuchi, S. Suzuki, Chem. Lett. (1986) 897.
[3] Y. Hori, A. Murata, R. Takahashi, J. Chem. Soc., Faraday Trans.
1.85 (1989) 2309.
[4] D.W. DeWulf, T. Jin, A.J. Bard, J. Electrochem. Soc. 136 (1989)
1686.
[5] S. Wasmus, E. Cattaneo, W. Vielstich, Electrochim. Acta 35 (1990)
771.
[6] G. Kyriacou, A. Anagnostopoulos, J. Electroanal. Chem. 322 (1992)
233.
[7] R. Shiratuchi, Y. Aikoh, G. Nogami, J. Electrochem. Soc. 140 (1993)
3479.
[8] B. Jermann, J. Augustynski, Electrochim. Acta 39 (1994) 1891.
[9] P. Friebe, P. Bogdanoff, N. Alonso-Vante, H. Tributsch, J. Catal. 168
(1997) 374.
[10] J. Lee, Y. Tak, Electrochim. Acta 46 (2001) 3015.
[11] H. Yano, F. Shirai, M. Nakayama, K. Ogura, J. Electroanal. Chem.
519 (2002) 93.
[12] K.W. Frese Jr., in: B.P. Sullivan, K. Krist, H.E. Guard (Eds.), Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, Elsevier,
Amsterdam, 1993, p. 191.
[13] D.T. Sawyer, A. Sobkowiak, J.L. Roberts Jr., Electrochemistry for
Chemists, 2nd ed., Wiley, 1995.
[14] A.T. Kuhn, in: H. Bloom, F. Gutmann (Eds.), Electrochemistry: The
Past Thirty and Next Thirty Years, Plenum Press, New York, 1977.
[15] J.OM. Bockris, S.U.M. Khan, Surface Electrochemistry: A Molecular Level Approach, Plenum Press, New York, 1993, p. 9.
[16] J.OM. Bockris, A.K.N. Reddy, Modern Electrochemistry, vol. 2,
Plenum Press, New York, 1970.
[17] Y. Hori, S. Suzuki, J. Res. Inst. Catal. Hokkaido Univ. 30 (1982)
81.
[18] A. Murata, Y. Hori, Abstracts of the Fall Meeting of the Electrochemical Society of Japan, Chiba, 1I03, September 29, 1990.
[19] Y. Hori, R. Takahashi, Y. Yoshinami, A. Murata, J. Phys. Chem. B
101 (1997) 7075.
[20] Y. Hori, I. Takahashi, O. Koga, N. Hoshi, J. Mol. Catal. 199 (2003)
39.
[21] D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics, 82nd
ed., CRC Press, 2001, Section 5, p. 95.
[22] Y. Hori, A. Murata, S. Ito, Chem. Lett. (1989) 1567.
[23] J.A. Dean (Ed.), Langes Handbook of Chemistry, 13th ed., McGrawHill, 1985, Section 6, p. 2.
[24] D.M. Kolb, in: H. Gerischer, C.W. Tobias (Eds.), Advances in Electrochemistry and Electrochemical Engineering, vol. 11, Wiley, New
York, 1978, p. 125.
[25] M. Pourbaix, Atlas DEquilibres Electrochimiques, Gauthier-Villars,
1963.
[26] J.A. Dean (Ed.), Langes Handbook of Chemistry, 13th ed., McGrawHill, 1985, Section 3, p. 122.
[27] F.W. Billmeyer Jr., Textbook of Polymer Science, Interscience Publishers, a division of John Wiley and Sons, 1962, p. 398.