Photo-electrochemical oxidation of organic C1 molecules over WO3 films in aqueous electrolyte: Competition between water oxidation and C1 oxidation

Article abstract of DOI:10.1002/cssc.201500800

To better understand organic-molecule-assisted photo-electrochemical water splitting, photo-electrochemistry and on-line mass spectrometry measurements are used to investigate the photo-electrochemical oxidation of the C1 molecules methanol, formaldehyde, and formic acid over WO₃ film anodes in aqueous solution and its competition with O₂ evolution from water oxidation O₂⁺ and CO₂⁺ ion currents show that water oxidation is strongly suppressed by the organic species. Photo-electro-oxidation of formic acid is dominated by formation of CO₂, whereas incomplete oxidation of formaldehyde and methanol prevails, with the selectivity for CO₂ formation increasing with increasing potential and light intensity. The mechanistic implications for the photo-electro-oxidation of the organic molecules and its competition with water oxidation, which could be derived from this novel approach, are discussed. Go Organic! The addition of organic molecules to an aqueous electrolyte and their photo-electro-oxidation generally yields higher photocurrents and can improve H₂ evolution in photo-electrochemical cells. The mechanisms of the reactions during photo-electro-oxidation of organic C1 molecules at WO₃ photoanodes and their contribution to the photocurrent were studied systematically by combined photo-electrochemistry and mass spectrometry measurements.

Full text of DOI:10.1002/cssc.201500800

DOI: 10.1002/cssc.201500800
Full Papers
Photo-electrochemical Oxidation of Organic C1 Molecules
over WO Films in Aqueous Electrolyte: Competition
3
Between Water Oxidation and C1 Oxidation
Robert Reichert, Christian Zambrzycki, Zenonas Jusys, and R. Jürgen Behm*
[a]
[b]
[a]
[a]
To better understand organic-molecule-assisted photo-electro-
chemical water splitting, photo-electrochemistry and on-line
mass spectrometry measurements are used to investigate the
photo-electrochemical oxidation of the C1 molecules metha-
nol, formaldehyde, and formic acid over WO3 film anodes in
electro-oxidation of formic acid is dominated by formation of
CO , whereas incomplete oxidation of formaldehyde and meth-
2
anol prevails, with the selectivity for CO formation increasing
2
with increasing potential and light intensity. The mechanistic
implications for the photo-electro-oxidation of the organic
molecules and its competition with water oxidation, which
could be derived from this novel approach, are discussed.
aqueous solution and its competition with O evolution from
2
+
+
water oxidation O2 and CO2 ion currents show that water
oxidation is strongly suppressed by the organic species. Photo-
Introduction
Since the early discovery of photo-electrochemical water split-
aqueous electrolytes and its competition with O2 evolution,
which results from the oxidation of water that can occur in
[1]
ting on TiO by Fujishima and Honda in 1972, the field of
2
photo-electrochemistry, and in particular, photo-electrocatalysis
has attracted considerable interest, motivated by the idea to
parallel. Compared to the large band gap material TiO (band
2
gap 3.2 eV), which allows only a few percent of the solar spec-
[
2,3]
use solar energy for obtaining valuable fuels such as H2.
trum to be used, WO has the advantage of a smaller band
3
This is usually achieved by photo-electrochemical water split-
ting into H and O or, in another approach, by water splitting
gap of 2.7 eV. The photo-electrocatalytic activity and selectivity
were determined by measuring the photocurrent and simulta-
neous detection of the main products CO and O using on-
2
2
in combination with the oxidation of sacrificial organic mole-
cules where the latter reaction frequently involves the so
called “photocurrent doubling”. The latter effect was proposed
to lead to higher solar efficiencies compared to the water split-
2
2
line mass spectrometry. The measurements were performed
under well-defined electrolyte transport conditions using
a thin-layer flow cell set-up, which has been introduced previ-
[
7]
ting process and hence to higher amounts of clean H pro-
ously. This novel approach allows us to i) correlate the activi-
ty, selectivity, and illumination intensity under reaction condi-
tions, where not only educt and product transport, but also
transport of reaction intermediates to or away from the photo-
electrode is well defined, and ii) differentiate between contri-
butions to the photocurrent resulting from water oxidation (O2
evolution) and from oxidation of the organic C1 molecules to
2
[4]
duced at the cathode. This could be an attractive alternative
to the production of clean H by electrocatalytic oxidation of
2
organic molecules on noble metal based catalysts in proton
[5,6]
exchange membrane electrolysis cells (PEMECs).
A precondition for possible application of such concepts
and their systematic optimization is a fundamental under-
standing of the elementary processes occurring during photo-
electro-oxidation of small organic molecules in aqueous elec-
trolytes. In the present study, we investigate the photo-electro-
oxidation of the C1 molecules methanol, formaldehyde, and
CO . Neither of these had been possible in previous studies
2
(see below).
Before presenting and discussing our results, we will briefly
summarize previous findings on the photo-electrochemistry of
formic acid on electrodeposited WO film electrodes in acidic
WO and on the photo-electro-oxidation of C1 molecules rele-
3
3
vant for this study. The photo-electrochemical properties of
WO , and specifically its water splitting activity, have been in-
3
[
a] R. Reichert, Dr. Z. Jusys, Prof. Dr. R. J. Behm
Institute of Surface Chemistry and Catalysis
Ulm University
vestigated by a number of different groups. In 1976 Hodes
et al. studied both oxidized tungsten metal electrodes and
spray deposited WO films in H SO as the supporting electro-
8
9069 Ulm (Germany)
3
2
4
E-mail: juergen.behm@uni-ulm.de
lyte and found that WO was equally stable against photocor-
3
[
b] C. Zambrzycki
rosion as TiO , which is an important requirement for the use
2
Institute of Chemical Engineering
Ulm University
[8]
as photocatalyst material. Furthermore, they reported a shift
in the onset potential for O evolution from about 2.0 V in the
8
9069 Ulm (Germany)
2
dark to 0.6 V under illumination. At about the same time,
Butler et al. studied the photoelectrolysis of water in different
Supporting Information for this article is available on the WWW under
http://dx.doi.org/10.1002/cssc.201500800.
ChemSusChem 2015, 8, 3677 – 3687
3677
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
electrolyte solutions and under illumination with light of differ-
and also observed both photocurrent doubling and a shift in
the respective onset potential compared to photo-electro-oxi-
dation of the pure supporting electrolyte. Based on kinetic
modelling, they concluded that formic acid is oxidized mainly
through a direct hole-transfer mechanism and suggested that
this requires strong interaction of formic acid with the WO3
surface. They could confirm the molecular adsorption of formic
ent wavelengths on WO single crystal electrodes. They found
3
a linear relationship between photocurrent and incident light
intensity as well as a Nernstian behavior of the flat-band po-
[9,10]
tential upon changing the pH of the electrolyte.
In the
same year, Gissler and Memming investigated the photo-elec-
trochemical behavior of sputtered and thermally grown WO3
layers and detected an additional increase of the photocurrent
in the visible light region (at around 430 nm) for annealed or
thermally grown WO layers compared to sputtered WO layers
acid on the nanostructured WO surface by IR spectroscopy.
3
Overall, these previous studies provide important data on
photo-electrochemical oxidation of organic C1 molecules and
3
3
without further heat treatment. They attributed this increase
to the formation of states with energies below the conduction
its competition with O evolution. A more general picture of
2
the competition between water oxidation and organic mole-
cule oxidation and of the reaction pathways including the se-
lectivities for the formation of different products, and finally
a quantitative analysis of the role of bias potential and light in-
tensity are still missing. Furthermore, previous studies have
been performed in stagnant electrolyte, where effects resulting
from or related to electrolyte transport cannot easily be sepa-
rated from reaction kinetics. These questions are topic of the
present study.
[
11]
band. They also detected O as a product of the water pho-
2
toelectrolysis using mass spectrometric analysis. Later, Desilves-
tro and Grätzel published a detailed study on the photo-elec-
tro-oxidation of water on bare polycrystalline WO and WO3
3
modified with Au or RuO overlayers, applying various in situ
2
and ex situ techniques for qualitative and quantitative analysis
of the reaction products, such as a Clark electrode for O de-
2
2
À
tection and UV spectroscopy and iodometry for S O
detec-
2
8
[
12]
tion.
Subsequent studies on the photo-electrochemical
In the following section we will present and discuss poten-
tiodynamic on-line differential electrochemical mass spectrom-
etry (DEMS) measurements of the photo-electro-oxidation of
the C1 molecules performed in dilute aqueous solutions and,
for comparison, in pure base electrolyte. In the subsequent
two sections we focus on the influence of the bias potential
(at constant light intensity) and of the light intensity (at con-
stant bias potential) on the selectivity of the photo-electro-oxi-
dation reactions, and the resulting product formation. Finally,
we will discuss the implications of these results for the mecha-
nistic understanding of the photo-electro-oxidation processes.
A brief description of the experimental set-up and procedures
water oxidation on nanostructured WO3 focused on reaction
parameters such as the light intensity, temperature, or the
[
4,13]
type of electrolyte.
The role of different fabrication meth-
ods of nanostructured WO photoanodes, their operation for
3
application in photo-electrochemical water splitting, and their
[14]
performances were reviewed by Liu et al. Recently Ng et al.
reported that calcination at 4008C leads to an optimum water
splitting performance of WO photoanodes prepared by anodi-
3
zation of tungsten foils, which they attributed to an optimal
[15]
combination of crystallinity and surface area.
The photo-electrochemical oxidation of organic molecules
generally leads to higher photocurrents, which in turn results
and the preparation and characterization of the WO thin film
3
in a higher H production at the cathode compared to photo-
electrodes will be given at the very end.
2
electrochemical water splitting in the absence of the respective
[
16]
molecule. Focusing on photo-electro-oxidation of C1 mole-
cules on WO electrodes, Santato et al. demonstrated that the Results and Discussion
3
photo-electrocatalytic oxidation of methanol on WO in acidic
3
Characterization of the WO photoelectrodes
3
solution leads to photocurrent doubling and a negative shift
of the photocurrent onset potential compared to reaction in
The calcined WO photoelectrodes were characterized by cyclic
3
[17]
pure supporting electrolyte. The authors concluded that the
oxidation of methanol is the dominant process compared to
voltammetry in 0.5m H SO supporting electrolyte in the range
2
4
À1
between À0.2 and 1.3 V (scan rate 10 mVs ). The resulting
RHE
the photogeneration of O . Based on a photocurrent efficiency
voltammogram is shown in Figure 1. The negative current with
the characteristic peak at around 0.22 V was previously as-
cribed to the intercalation and reduction of protons adsorbed
2
value of 190% at approximately 400 nm, which demonstrates
almost perfect photocurrent doubling, they concluded that the
photo-electrochemical oxidation of methanol is even able to
compete with electron–hole recombination. Solarska et al. re-
ported similar results for the photo-electrocatalytic oxidation
of the organic molecules methanol, formaldehyde, and formic
on the WO surface, while the peak at around À0.1 V results
3
+
from hydrogen intercalation from H O present in the bulk
3
[
20,21]
+
electrolyte.
The intercalation of H into the WO is accom-
3
panied by a color change from colorless to blue, known as an
[18]
[21]
acid on mesoporous WO films. From the CO content in the
electrochromic effect. The subsequent de-intercalation in the
3
2
effluent gas, which was measured by gas chromatography,
they calculated fractional current efficiencies and their varia-
tion with time and concluded that methanol is oxidized to CO2
via a three-step mechanism, involving first the formation of
formaldehyde and then formic acid. Later Monllor-Satoca et al.
studied the photo-electrochemical oxidation of formic acid on
anodic scan results in a positive current with a maximum at
0.28 V, which is in good agreement with the characteristics ob-
[
21]
served by Bohnke et al. The higher charge of the cathodic
peak is attributed to H evolution. XRD measurements of the
2
calcined films identified them as WO with a tetragonal crystal
3
structure. The respective diffractogram is shown in the Sup-
porting Information (Figure S1).
[
19]
nanostructured, electrodeposited WO3 thin-film electrodes,
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3678
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 1. Cyclic voltammogram of an electrodeposited WO
after calcination (electrolyte: 0.5m H SO
3
photoelectrode
; potential scan rate: 10 mVs ).
À1
2
4
Figure 3. XPS spectra recorded on an electrodeposited and calcined WO
3
photoelectrode; a) survey spectra, b, c) detail spectra of the W4f (b) and the
O1s (c) regions (individual fit contributions are indicated by black lines).
Figure S3. Figure 3a shows an X-ray photoelectron spectrosco-
py (XPS) survey spectrum recorded on the electrodeposited
WO film showing the expected W4f signal at about 37 eV and
3
the O1s signal at about 530 eV. Furthermore, the Auger signals
of OKLL (976 eV) and CKLL (1225 eV) appear in the high binding
[
23]
energy region, in good agreement with previous findings.
Detailed spectra of the W4f and O1s signals are shown in Fig-
ure 3b and 3c, respectively. The W4f7/2 and W4f5/2 peaks are
located at binding energies (BEs) of 35.8 and 37.9 eV, respec-
[
24,25]
tively. Signals at these BEs have been assigned to WO3.
Figure 2. SEM top-view images of a) the nanocrystalline WO
nation at 4508C, b) the uncovered FTO surface, and c) a magnified image of
the WO surface.
3
film after calci-
The O1s signal shown in Figure 3c was fitted to two peaks
with BEs of 530.4 and 531.2 eV. The O1s peak at 530.4 eV has
3
[26]
previously been assigned to WO3.
The broad peak with
a maximum at 531.2 eV is attributed to C=O species such as
[
27]
Figure 2a shows an SEM overview image of a thin WO film,
adsorbed carbonates.
Peaks at similar BEs were observed
3
which was deposited for 5 min at 0.2 VRHE and annealed for
also in reference measurements on a commercial WO powder.
3
À1
9
0 min at 4508C (38Cmin ramping rate). The film shows
Quantitative analysis of the W4f and the O1s peak intensities
resulted in a tungsten/oxygen ratio of 1:3, in full agreement
with the expected formal stoichiometry of WO3.
a rather smooth, slightly “hilly” surface topography. As evident
from the SEM image in Figure 2b, the surface of the underly-
ing substrate (fluorine-doped tin oxide, FTO) differs significant-
ly from the structure of the film surface, indicating that the
Photo-electrochemical potentiodynamic oxidation of water
and of the C1 molecules
FTO substrate is essentially completely covered by a WO film.
3
At higher magnification, flaky structures of about hundred
nanometers in diameter separated by thin cracks can be seen
Figure 4 shows the photocurrents, ion currents of O at m/z 32
2
on the WO surface (see Figure 2c). These flaky structures are
and CO2 at m/z 44, which were recorded simultaneously
3
À1
formed only after the calcination of the film, which converts
during a linear potential sweep (scan rate 10 mVs ) of the
the as-deposited amorphous WO into a more crystalline
WO3 film electrode in pure base electrolyte and in solutions
containing the C1 molecules formic acid, formaldehyde or
methanol under UV light illumination. For measurements in
pure supporting electrolyte (0.5m H SO ), the photocurrent
3
[
22]
form,
indicating that they are part of the WO3 film. We
cannot rule out that the tiny cracks reach down to the sub-
strate, but their contribution to the entire surface is marginal.
The smooth surface topography was confirmed by AFM meas-
urements, which yielded an areal roughness of 31Æ1 nm on
micrometer size areas (15 mm”15 mm) (Figure S2).
2
4
sets in at approximately 0.6 V, in good agreement with previ-
[
8,18,19]
ous studies on the photo-electrochemistry of WO3.
In sol-
utions containing C1 molecules, the photocurrents (Figure 4a)
show lower onset potentials (see inset of Figure 4a) as well as
higher maximum photocurrents, where the exact values
depend on the organic molecule present in the electrolyte. For
the oxidation of formic acid, the photocurrent onset is shifted
SEM-EDX (energy dispersive X-ray spectroscopy) measure-
ments were used to determine the film composition, which is
composed of 23.5% tungsten and 76.5% oxygen, as expected
for WO stoichiometry. A characteristic spectrum is shown in
3
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3679
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
cerned from voltammetric measurements. This is possible by
on-line mass spectrometry combined with photo-electrochemi-
[
7]
cal measurements, which provide additional information on
the nature and amount of volatile reaction products formed
during the different photo-electrochemical oxidation reactions.
In the experiments, the ion currents of O (Figure 4b) and CO2
2
(
Figure 4c) were monitored.
O2 formation in the supporting electrolyte starts at about
0
.8 V (inset in Figure 4b), approximately 0.2 V higher than the
photocurrent onset potential, (inset in Figure 4a). Apparently,
processes other than O evolution contribute to the photocur-
2
rent and dominate at low bias potentials. Possible reactions
that can occur in parallel to O evolution from water oxidation
2
have been reported to be hydrogen peroxide (H O ) formation
2
2
resulting from incomplete water oxidation as well as persulfate
2
À
(
S2O8 ) formation from the oxidation of sulfate in the support-
[
22,29]
ing electrolyte.
During formic acid oxidation, O formation is also observed
2
Figure 4. Potentiodynamic photo-electrochemical oxidation of 10 mm
HCOOH, HCHO, and CH OH on a WO photocatalyst: a) overall photocurrent,
b) O ion current (m/z 32) and c) CO
trolyte: 0.5m H SO
at potentials above 0.8 V, whereas the CO ion current rises si-
2
3
3
multaneously with the photocurrent (Figure 4c). Furthermore,
2
2
ion current (m/z 44) (supporting elec-
À1
the O evolution in the formic acid containing electrolyte is
2
4
; potential scan rate: 10 mVs ).
2
suppressed to about 50% of that obtained in pure supporting
to a lower value of about 0.5 V, in good agreement with litera-
electrolyte, even though the concentration of water
(
[
19]
À1
ture reports.
ꢀ55 molL ) is more than three orders of magnitude higher
À1
The higher photocurrent during the photo-electrochemical
oxidation of small organic molecules compared to that ob-
tained in pure supporting electrolyte is generally explained by
the so-called “current doubling” mechanism introduced by
than that of formic acid (0.01 molL ). This fits well with the re-
sults of a recent study on formic acid photo-electro-oxidation
on TiO (P25, Evonik) film electrodes, where we found the O2
2
ion current to be reduced to about one third in an electrolyte
[28]
[7]
Morrison and Freund. After generation of an electron–hole
pair [Eq. (1)], formic acid is oxidized by the photogenerated
containing 50 mm formic acid. The suppressed O evolution
2
indicates a competition between water and formic acid for
free adsorption sites or, more likely, for photogenerated holes
+
hole (h ) to yield a highly reactive adsorbed reaction inter-
mediate [HCOOC]ad [Eq. (2)], which in a second step decompos-
es and injects an electron into the conduction band (CB) of
WO forming the final product CO [Eq. (3)].
on the WO photocatalyst. Monllor-Satoca et al., who studied
3
the photo-electrochemical oxidation of formic acid on WO in
3
aqueous solution by infrared spectroscopy in an attenuated
total reflection (ATR) configuration, found that formic acid ad-
sorbs specifically as a non-dissociated neutral molecule on the
3
2
h Á n ! h þ eÀ
þ
ðCBÞ
ð1Þ
ð2Þ
ð3Þ
[
19]
HCOOHad þ h ! ½HCOOCŠ þ Hþ
HCOOCŠad ! CO þ H þ e
þ
WO surface. Strong adsorption on the photocatalyst surface,
3
ad
often suggested as enhancing the photo-oxidation rate, was
also proposed for the photo-electrochemical oxidation of
þ
À
½
2
ðCBÞ
[
30–32]
formic acid on TiO2.
Hence, excitation by one photon results in the formation of
In the presence of formaldehyde, the photocurrent sets in at
an even lower potential (ca. 0.45 V) compared to the reaction
in the electrolyte containing formic acid. Clearly, the photo-
electrochemical oxidation of formaldehyde is even faster than
the oxidation of formic acid. Furthermore, the maximum pho-
tocurrent for formaldehyde oxidation, reached at a potential of
around 1.4 V, is about 30% higher than that obtained in formic
acid solution. In analogy to the formation of the adsorbed re-
one hole and two electrons, the first one from the photoexcita-
tion, the second one from the injection by the [HCOOC]ad spe-
cies.
It should be noted that although the term “photocurrent
doubling” implies a doubling of the photocurrent (at identical
light intensity) compared to the value in supporting electro-
lyte, the exact value depends on the type and concentration
of the organic molecule present in the electrolyte, because the
reaction probabilities of water and the different C1 molecules
with a photogenerated hole differ from each other, resulting in
different recombination probabilities.
[
19]
action intermediate [HCOOC]ad in formic acid solution,
we
assume that an adsorbed formyl species is formed as a reaction
intermediate, which under injection of an electron into the
conduction band of WO reacts further to formic acid, accord-
3
Because the photocurrent can stem from different processes
such as the oxidation of water, the complete or incomplete ox-
ing to Equations (4) and (5).
HCHOad þ h ! ½HCOCŠ þ Hþ
þ
idation of organic molecules to either CO or byproducts (for
ð4Þ
ð5Þ
2
ad
formaldehyde and methanol), the individual contributions of
these reactions to the overall photocurrent cannot be dis-
þ
À
½
HCOCŠad þ H O ! HCOOH þ H þ e
2
ðCBÞ
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3680
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
The formic acid formed could then either be further oxidized
and formic acid, prevail. The stepwise oxidation of formalde-
hyde to formic acid, followed by the oxidation of the resulting
to CO (according to Equations (1)–(3)) or escape from the cell
2
without further oxidation.
formic acid to CO or a similar reaction of methanol via formal-
2
A similar trend of the maximum photocurrents for the differ-
ent C1 molecules was also observed by Solarska et al. in
dehyde and formic acid to CO , according to
2
CH
OH ! HCHO þ 2 H þ 2 eÀ
þ
ð6Þ
ð7Þ
ð8Þ
a photo-electrochemical study using colloidal deposited WO
3
3
[18]
films under visible light illumination. The O ion current, re-
2
þ
À
HCHO þ H O ! HCOOH þ 2 H þ 2 e
2
sulting from the competing water oxidation, is almost com-
pletely suppressed in the presence of formaldehyde, indicating
that the presence of formaldehyde and its photo-electrochemi-
þ
À
HCOOH ! CO þ 2 H þ 2 e
2
cal oxidation hinder the O evolution process much more effi-
2
fully agrees with previous findings in studies on methanol
(formaldehyde) electro-oxidation on noble metal electrodes/
catalysts. Therefore, formaldehyde and formic acid are also ex-
pected as reactive side products in the present study. Mecha-
nistic consequences will be discussed below, after the quanti-
tative evaluation of the effects of bias potential (at constant
light intensity) and light intensity (at constant bias potential).
ciently than formic acid. Alternatively, O formed during water
2
oxidation could be consumed in an oxidation reaction of form-
[33]
aldehyde to formic acid. Although this cannot be ruled out,
we consider this possibility to be unlikely. Surprisingly, CO for-
2
mation from formaldehyde starts only at 0.6 V, which is
1
50 mV higher than the photocurrent onset potential, in con-
trast to the behavior observed for formic acid photo-oxidation.
Furthermore, the maximum amount of CO formed is less than
2
1
0% of that formed from formic acid, while the highest photo-
Reaction selectivity as a function of bias potential at con-
stant light intensity
current of all three organic molecules is observed for formalde-
hyde photo-electro-oxidation. We assume that most of the
formaldehyde is oxidized to formic acid under these conditions
To quantitatively address the question of how much the selec-
tivities during the photo-electrochemical oxidation of the three
organic molecules and during water splitting in base electro-
lyte depend on the bias potential, we calculated the photocur-
rent efficiencies for the two products O and CO from the re-
rather than to CO (for a quantitative evaluation see below),
2
[
34–36]
similarly to observations for its electrocatalytic oxidation.
Finally, the photocurrent resulting from the oxidation of
methanol shows a similar shift of the photocurrent onset po-
tential as that in the formaldehyde-containing solution. The
evolution of the photocurrent with increasing potential is also
similar to that observed in formaldehyde containing solution,
only the maximum photocurrent is slightly lower, in agreement
2
2
spective ion currents in Figure 4 at different potentials.
Figure 5 shows the photocurrent efficiencies at different bias
potentials for the formation of O (Figure 5a: O current effi-
2
2
ciency, OCE) and CO (Figure 5b: CO current efficiency, CCE).
2
2
The OCE during the photo-electrochemical water splitting in
[18]
with previous findings. Since the molecular ions of methanol
the 0.5m H SO supporting electrolyte increases steadily with
2
4
and O have the same mass-to-charge ratio of m/z 32, a quanti-
2
potential, and is approximately 10% at 0.8 V, 35% at 1.2 V, and
2% at 1.46 V (Figure 5a). Hence, the contribution from O2
evolution to the overall photocurrent increases with bias po-
tative evaluation of the O evolution from the m/z 32 ion cur-
2
4
rent is not possible. The continuous decrease of the m/z 32 ion
current with increasing potential over the whole potential
range implies, however, that methanol is increasingly con-
sumed with increasing potential and that this increasing meth-
anol consumption is not compensated by O evolution, since
2
the latter would lead to a m/z 32 increase or at least a slower
decrease of the m/z 32 signal.
The m/z 44 ion current (CO formation from methanol oxida-
2
tion) shows a similar onset potential as that obtained during
formaldehyde oxidation (ꢀ0.6 V). The maximum CO2 forma-
tion rate from the photo-electrochemical oxidation of metha-
nol is only one third of that measured for formaldehyde oxida-
tion, and only about 2.5% of that measured during formic acid
oxidation. The combination of a very low CO formation rate
2
during the photo-electrochemical oxidation of methanol and
rather high overall photocurrents indicates that photo-oxida-
tion of methanol predominantly results in the formation of the
incomplete oxidation products formaldehyde and formic acid.
In total, the combined potentiodynamic photocurrent and
on-line DEMS measurements show that formic acid is directly
oxidized to CO , whereas for formaldehyde and methanol the
2
2 2
Figure 5. Current efficiencies of a) O (OCE) and b) CO (CCE) during the po-
formation of reactive side products (reaction intermediates),
most likely of the incomplete oxidation products formaldehyde
tentiodynamic, photo-electrochemical oxidation of 10 mm HCOOH, HCHO,
and CH
3
OH on a WO
3
photocatalyst (for calculation, see experimental).
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3681

Full Papers
tential, changing the selectivity of the reaction. A rather similar
value of 35% was recently reported by Hill and Choi for the
tion cell. The maximum CCE reached during the photo-electro-
chemical oxidation of formaldehyde is about one order of
magnitude lower than that obtained during photo-electro-
chemical oxidation of formic acid.
photocurrent efficiency for O at a comparable bias potential
2
(
0.9 V vs. Ag/AgCl) on electrodeposited WO films (0.1m
3
[22]
Na SO electrolyte, pH 3). Hydrogen peroxide (H O ) resulting
For the photo-electrochemical oxidation of methanol, the
OCE could not be determined because of the similar mass-to-
charge ratio of m/z 32 of the molecular ions of methanol and
2
4
2
2
2À
from incomplete water oxidation as well as persulfate (S O8
)
2
formed by sulfate oxidation in the supporting electrolyte have
been reported as side products, whose formation can affect
O . The CCE values between 1% (0.7 V) and 5% (1.4 V) reflect
2
[22,29]
the selectivity for the O evolution reaction (OER).
a very poor selectivity for the complete oxidation to CO in the
2
2
The OCE in the electrolyte containing 10 mm formic acid
also depends on the bias potential. As shown in Figure 5a,
methanol oxidation reaction, which is even lower than that ob-
tained for formaldehyde oxidation. The remaining 95% of the
photocurrent is expected to result from the formation of other
products, most likely the incompletely oxidized reaction inter-
mediates formaldehyde and formic acid. It has been suggested
that the rather low CCEs obtained for the oxidation of formal-
dehyde and methanol point to a reaction mechanism which in-
cludes sequential re-adsorption and further oxidation of the in-
even in the presence of formic acid, O evolution contributes
2
between 1 and 12% to the overall photocurrent. Nevertheless,
most of the photocurrent stems from the oxidation of formic
acid to CO , indicative of a preferential oxidation of formic acid
2
compared to the O evolution reaction from water, with CCEs
2
of about 90% in the potential range between 0.9 and 1.46 V
[
18]
(
Figure 5b). At bias potentials of 0.7 and 0.8 V the CCE is slight-
termediates formed. This will be discussed in more detail in
the next section, after the presentation of the effects induced
by varying the light intensity.
ly lower, with values of about 80%. This might be caused by
higher contributions from the formation of the water oxidation
2
À
side products H O or S O
8
at lower bias potentials, which
2
2
2
was also observed during water oxidation in pure supporting
electrolyte. The fact that at bias potentials above 0.8 V the
sum of the OCE and CCE during this reaction is about 100%
provides quantitative proof that only O and CO are formed
Reaction selectivity as a function of light intensity at con-
stant bias potential
To further address the question whether the amount of photo-
generated charge carriers can change the activity and selectivi-
ty of the respective reactions, as indicated by the bias-poten-
tial-dependent photocurrent efficiencies (Figure 5), we per-
formed systematic light chopping experiments at different illu-
mination intensities at a constant bias potential of 1.2 V to
vary the flux of the impinging photons and hence the concen-
tration of the photogenerated holes and electrons at the sur-
face. Figure 6 summarizes the temporal evolution of the pho-
tocurrents and ion currents of O and CO during light-on and
2
2
during the reaction and that the formation of unwanted side-
products, as generated in pure supporting electrolyte, is negli-
gible. Obviously, the facile photo-electrochemical oxidation of
formic acid does not only suppress the O evolution, but also
2
inhibits the formation of undesired side products such as hy-
drogen peroxide, which is regarded as a precursor for the for-
mation of surface peroxo species that lead to photocorrosion
[29]
and a loss of photoactivity. This could be of interest for tech-
nical applications, where side reactions that can reduce the
2
2
[29]
lifetime or activity of the WO photoanode can be avoided.
light-off sequences at different light intensities (6, 12, 20, and
3
À2
The current efficiencies during the photo-electrochemical
oxidation of formaldehyde are below 5% for the OCE, and also
the CCE shows rather low values of only between 2% at 0.7 V
43 mWcm ) in the pure supporting electrolyte and in electro-
lytes containing 10 mm of formic acid, formaldehyde, and
methanol. In all cases, at t=40 s, the photocurrents rise imme-
diately after the onset of illumination and reach the steady-
state value essentially instantaneously. These values increase
with the light intensity. Similar to the potentiodynamic meas-
urements, in these transient experiments the highest photo-
current is always reached in the formaldehyde-containing elec-
trolyte, followed by methanol-containing electrolyte, formic-
acid-containing electrolyte, and finally the pure supporting
electrolyte.
and 10% at 1.46 V. The low OCE means that the O evolution
2
is largely suppressed by the presence of formaldehyde and the
ongoing formaldehyde oxidation reaction, similar to the find-
ings for photo-electro-oxidation of formic acid. In that case it is
likely that the formation of the side products originating from
the supporting electrolyte (hydrogen peroxide and persulfate)
is essentially inhibited. The unexpectedly low values of the
CCE indicate that the formation of the incompletely oxidized
side product formic acid prevails for this reaction and contrib-
utes more than 80% of the overall photocurrent, similar to
what has been found for the electrochemical oxidation of
The mass spectrometric ion currents also show a steep in-
crease upon the onset of illumination, but then the increase
gets slower, and the saturation values are reached only after
approximately 20 s. Interestingly, the ion currents of the elec-
trolyte containing formic acid increase proportionally to the
photocurrents. For the other 3 electrolytes, the relative in-
crease of the ion currents with increasing light intensity differs
from that of the respective photocurrents. As an example, the
photocurrent in the formaldehyde containing electrolyte is
[35,36]
formaldehyde on nanostructured Pt electrodes.
The steady
increase of the CCE with bias potential, with a close to linear
relationship, points to potential-induced change in the reac-
tion selectivity. This is most likely related to the higher density
of photogenerated charge carriers, arising from the more effi-
[
37]
cient charge carrier separation at higher bias potentials.
These are required for the further photo-oxidation of the inter-
doubled when increasing the light intensity from 20 to
À2
mediate formic acid to CO before it can escape from the reac-
43 mWcm , whereas at the same time the CO ion current
2
2
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3682
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
havior of the photocurrents) is
a clear indication of a change in
product composition and hence
in selectivity of the respective re-
actions with increasing light in-
tensity.
These changes in the selectivi-
ty are evaluated quantitatively
by plotting the variation of the
photocurrents, the O or CO ion
2
2
currents, and the resulting O2
and CO2 current efficiencies
(
OCEs and CCEs) with increasing
light intensity (Figure 7). In all
cases the total photocurrent de-
pends essentially linearly on the
light intensity (Figure 7a, d),
which is a well-known phenom-
enon for semiconductor photo-
electrodes, indicating that the
number of absorbed photons/
photogenerated holes required
per electron of photocurrent is
constant and does not vary with
[
38,39]
light intensity.
The same is
true also for the O ion current
2
in pure supporting electrolyte
(
Figure 7b). In contrast, the CO2
Figure 6. Photo-electrochemical oxidation of 0.5m H
lower left) and 10 mm CH OH (lower right) on a WO
intensities; a), c), e), g) photocurrents, b), d), f), h) ion currents (m/z 32 or m/z 44) (supporting electrolyte: 0.5m
SO ; Ebias =1.2 V).
2
SO
4
(upper left), 10 mm HCOOH (upper right), 10 mm HCHO
ion currents in the solutions con-
taining C1 molecules (Figure 7e)
show a distinctly non-linear de-
pendence on the light intensity,
growing either less rapidly at
higher light intensities than ob-
tained for a linear correlation
(
3
3
photocatalyst under chopped illumination at different light
H
2
4
(
formic acid), or more rapidly at
higher light intensities (metha-
nol, formaldehyde).
The linear increase in the O2
ion current converts into a non-
linear increase of the OCE with
increasing light intensity. On the
other hand, the non-linear rela-
tion between CO2 ion current
and light intensity results in an
approximately linear increase of
the CCEs for the photo-electro-
chemical oxidation of formalde-
hyde and methanol (Figure 7 f).
This means that modifying the
light intensity (and hence modi-
fying the rate for photogenera-
tion of holes and electrons and
their steady-state concentrations
at the surface), while staying at
Figure 7. Dependence of the photocurrents a), d), the m/z 32 (b) and m/z 44 (e) ion currents, and the OCE (c) and
CCE (f) on the light intensity for the photo-electrochemical oxidation of 0.5m H SO (left column) and 10 mm
HCOOH, HCHO, and CH OH (right column) on a WO photocatalyst (supporting electrolyte: 0.5m H SO
bias =1.2 V). Lines are included to guide the eye.
2
4
3
3
2
4
;
E
grows by a factor of five. This deviation of the ion current from
a linear increase with light intensity (which differs from the be-
a constant bias potential of 1.2 V, leads to a distinct and ap-
proximately linear change in the reaction selectivity. This
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3683
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
means that during water oxidation the selectivity for the final
oxidation product O2 is increased at higher light intensities
and less H O is formed. According to this behavior, the selec-
40% at the highest light intensity used. Earlier studies on the
reverse reaction, the oxygen reduction reaction (ORR) on nano-
structured glassy-carbon-supported Pt model electrodes re-
ported that an increase of the density of Pt nanostructures re-
2
2
tivity for the final oxidation product CO2 is also increased
during oxidation of formaldehyde and methanol.
sulted in a higher selectivity for the ORR product H O, while at
2
The behavior of the CCE during photo-oxidation of formic
acid is also different. It shows constant values of about 95%
the same time it quenched the formation of the intermediate
[
42,43]
H O .
This trend agrees fully with the present findings of
2
2
before dropping to about 85% at a higher light intensity of
higher selectivities for complete oxidation to O rather than in-
2
À2
4
3 mWcm . This drop of the CCE at high light intensity is
complete oxidation to H O with higher steady-state concen-
2
2
caused by the increasing O evolution, which starts to contrib-
trations of photogenerated holes. Hence, the selectivity for O2
evolution can also be modified and controlled by varying the
light intensity at constant potential and continuous electrolyte
flow.
2
ute more to the photocurrent when a larger number of photo-
generated holes is available, leading to an increasingly success-
ful competition with formic acid in the reaction with photo-
generated holes. Clearly, the selectivity for CO2 formation is
largely independent of the light intensity, as anticipated for
In full agreement with this mechanism, the CCE for formic
acid photo-electro-oxidation does not change with increasing
bias potential, except for deviations in the very low potential
range, where contributions from incomplete water oxidation
are expected, and for high light intensities, where the increas-
a reaction in which only a single product, in this case CO , is
2
expected, and the competing water oxidation reaction lowers
the selectivity for CO only at high light intensities.
2
These findings and those on the influence of the bias poten-
tial discussed earlier have important consequences for the
mechanism of the overall photo-oxidation reaction, in which
both photo-oxidation of water and photo-oxidation of the re-
spective C1 molecule is possible, and both reactions can pro-
duce (reactive) side products in addition to the fully oxidized
products O and CO .
ing O evolution starts contributing to the photocurrent more
2
strongly. The independence of the product distribution fully
agrees with a reaction mechanism where only one photogen-
erated hole is involved and the reaction with another hole
[
18,19,28]
does not play a role.
Although it may be tempting, there is no correlation be-
tween the absolute photocurrent and the number of reaction
steps involved. Therefore, the lower photocurrent observed for
methanol compared to formaldehyde cannot be simply ex-
plained by the fact that the photo-electro-oxidation of metha-
2
2
The distinct variation of the CCE and hence of the selectivity
observed for the photo-electro-oxidation of methanol and
formaldehyde closely resembles observations in the electroca-
talytic oxidation of these compounds on both carbon support-
ed Pt catalysts and on nanostructured Pt/glassy carbon electro-
nol to CO proceeds via two reaction intermediates, formalde-
2
hyde and formic acid, whereas for formaldehyde photo-elec-
[35,36,40,41]
des.
In these reactions, an increase of the Pt loading or
tro-oxidation to CO there is only one (formic acid). This is ob-
2
the Pt nanodisk coverage on the glassy carbon substrate, or al-
ternatively, a decrease of the flow rate, was found to result in
vious when comparing the photocurrents for photo-electro-ox-
idation of formic acid, which are much lower than those
obtained for formaldehyde at otherwise identical reaction con-
ditions.
a higher CO current efficiency. This was explained by an in-
2
creasing probability of re-adsorption and further oxidation of
the incompletely oxidized reaction intermediates to the com-
pletely oxidized and stable product CO2.
Finally, we want to comment on the fact that the change in
selectivity in organic molecule photo-electro-oxidation with in-
creasing light intensity (at constant bias potential) does not
cause deviations from the linear increase of the overall photo-
current with increasing light intensity. This is simply because
each additional photo-electro-oxidation step correlates to the
reaction with one hole and the injection of 2 electrons [see
Eqs. (1)–(5)]. Therefore, increasing the light intensity and hence
increasing the steady-state concentration of photogenerated
Following the above reaction scheme, we explain the in-
creasing CO formation (increasing CO current efficiencies) in
2
2
the photo-electrochemical oxidation of methanol and formal-
dehyde upon increasing the bias potential or with increasing
light intensity by an increasing probability of the incompletely
oxidized reaction intermediates to further react to the fully oxi-
dized product CO when a larger number of photogenerated
2
holes is available at the WO surface, rather than being trans-
holes at the WO surface may lead to changes in the selectivity,
3
3
ported out of the reaction cell. This means that the basic con-
cept of “desorption–re-adsorption–reaction” is also valid for
photo-electrochemical reactions that include the formation of
reaction intermediates that can re-adsorb and react further to-
but the current-doubling characteristics (injection of 2 elec-
trons upon reaction with one hole) will be maintained.
This would be different only in specific situations. A devia-
tion from the linear increase of the photocurrent with light in-
tensity would be expected, for example, if an increasing light
intensity would induce a change in selectivity, as described
before, and furthermore, if the competing reactions result in
a different number of electrons injected per reaction with one
photogenerated hole. This would be the case, for example, if
the increasing light intensity would modify the ratio of C1 oxi-
dation (2 electrons per reaction with 1 hole) to water oxidation
(1 electron per reaction with 1 hole). The fact that experimen-
wards more stable final products such as CO . The only differ-
2
ence is that the higher probability for the electrocatalytic reac-
tion upon re-adsorption is replaced by a higher probability of
reacting with a photogenerated hole at the surface.
Further support for the general validity of this mechanistic
model comes from the light intensity dependence of the OCE
during water oxidation in bare supporting electrolyte (Fig-
ure 7c). It increases from about 20% at the lowest to about
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3684
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tally there are no measurable deviations from a linear relation-
ship between the photocurrent and the light intensity confirms
that modifications in this ratio are below the detection limit.
This agrees fully with the observation that the presence of the
tion, is proposed to be responsible for the change in reaction
selectivity for O evolution during water oxidation and for CO2
2
formation during oxidation of formaldehyde and methanol. In-
creasing the steady-state concentration of photogenerated
holes at the anode surface by either increasing the bias poten-
tial (more efficient separation of charge carriers) or by modify-
ing the light intensity increases the chances for further oxida-
tion of adsorbed reaction intermediates, and thus increases
the selectivity for forming the completely oxidized reaction
product. Because the number of electrons generated per oxi-
dation step, that is, per reaction with a photogenerated hole,
is identical in the different photo-electro-oxidation steps of C1
molecules, the change in selectivity does not modify the over-
all photocurrent generated per reaction with a photogenerated
hole, which explains the linear correlation between light inten-
sity and measured overall photocurrent, despite the change in
selectivity.
C1 molecules strongly suppresses the O evolution. Changes in
2
the selectivity, on the other hand, may arise from changes in
the coverage of the reactants or from changes in the surface
concentration of photogenerated holes.
Conclusions
With the aim to better understand organic-molecule-assisted
photo-electrochemical water splitting, we systematically inves-
tigated the photo-electrochemical oxidation of the organic C1
molecules methanol, formaldehyde, and formic acid over elec-
trodeposited WO3 film anodes in aqueous solution and its
competition with O2 evolution from water oxidation. Using
a combination of a photo-electrochemical thin-layer flow cell
and on-line DEMS measurements, potentiodynamic and poten-
tiostatic transient experiments performed under defined mass
transport conditions in aqueous solutions of the C1 molecules
and, for comparison, in pure base electrolyte, and varying the
light intensity led to the following conclusions:
In summary, the present study offers a more detailed under-
standing of the processes important for organic molecule as-
sisted photo-electrochemical water splitting at semiconductor
photoanodes such as WO , specifically, the processes occurring
3
during photo-electrochemical oxidation of small organic mole-
cules in aqueous electrolyte under continuous electrolyte
transport, where photo-oxidation of water and of the respec-
tive organic molecule compete for photogenerated charge car-
riers and removal of incompletely oxidized reaction intermedi-
ates from the reaction cell may affect the reaction characteris-
tics. We show that the reaction selectivity can be modified by
changing the reaction parameters such as bias potential or
light intensity, which may have important consequences for
technical applications.
1
. Photo-electrochemical oxidation of water to O2 at the
WO photoanode requires the highest onset potential among
3
the molecules studied and shows a maximum O current effi-
2
ciency of about 40%, illustrating that under the present reac-
tion conditions O evolution is not the main process during
2
the anodic water oxidation process and the formation of side
2À
products such as H O or S O has to be considered as well.
2
2
2
8
2
. The higher photocurrents observed during photo-electro-
oxidation of the C1 molecules methanol, formaldehyde, and
formic acid in aqueous solutions are compatible with expecta-
tions based on the current-doubling effect. The deviations
from a duplication of the photocurrent on the other hand indi-
cate that the reaction probability per photogenerated hole at
the surface is different for water oxidation and C1 oxidation.
Experimental Section
Preparation of the WO3 photoelectrodes: WO3 films were pre-
pared by electrodeposition from a peroxytungstate solution follow-
[44]
ing the procedure described by Luo and Hepel. Briefly, 2 g of
tungsten powder (12 mm mean particle diameter, Sigma–Aldrich)
were dissolved in 10 mL hydrogen peroxide solution (35%, Sigma–
Aldrich). After complete dissolution, the excess peroxide was de-
composed on a platinum sheet immersed into the solution until O2
formation had stopped. Afterwards, a mixture of 100 mL water
3
. The efficiency for O evolution (OCE) is drastically reduced
2
in the presence of the C1 molecules, despite a lower concen-
tration by 3 orders of magnitude compared to water.
4
. The oxidation of the C1 molecules is characterized by
a low CO current efficiency for the photo-electro-oxidation of
2
(
Millipore Milli-Q, 18.2 MWcm) and 30 mL isopropanol (>99.9%, Li-
Chrosolv, Merck) was added. The freshly prepared electrolyte
pH 1.6) was then used in a three-electrode electrochemical cell to
methanol and formaldehyde, while for formic acid it is about
9
0–95%. The former is interpreted as indicative of significant
(
formation of reaction intermediates, which are removed from
the reaction cell under the experimental conditions (continu-
ous electrolyte transport). The latter agrees with expectations
deposit circular WO films (Ø=7 mm) onto conductive glass elec-
3
trodes (20”30 mm, FTO, ꢀ7 W/sq, Sigma–Aldrich). The counter
electrode was an FTO glass cut, and a saturated calomel electrode
(SCE) was used as reference electrode. All potentials are referenced
to that of the reversible hydrogen electrode (RHE). The potential
was controlled by a programmable potentiostat (PAR 273A, Prince-
ton Applied Research). For deposition, the potential was stepped
from 0.345 to 0.200 VRHE and held there for 5 min before stepping
back. Finally, the WO films were calcined in a muffle furnace (HTC
1
a sufficient adhesion and to increase the crystallinity.
for a reaction where only one product, in this case CO , is
2
formed, independent of the experimental situation. The differ-
ence to 100% results from contributions from water oxidation.
5
. We suggest that the “desorption–re-adsorption–reaction”
concept, which had been introduced to explain changes in the
reaction selectivity upon varying the catalyst loading/electro-
lyte transport during electrocatalytic reactions involving the
formation of reactive reaction intermediates, is also valid for
photo-electrochemical reactions. Further photo-oxidation of re-
action intermediates, possibly after desorption and re-adsorp-
3
600, Strçhlein Instruments) at 4508C for 90 min in air to ensure
Characterization of the WO photoelectrodes: The electrochemi-
cal properties of the as prepared WO photoelectrodes were char-
3
3
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3685
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
acterized by cyclic voltammetry in a standard three-electrode con-
figuration with a saturated calomel reference electrode (SCE) and
suming that these were the only reaction products in the respec-
tive reactions. The calibration of the mass spectrometer allowed us
to calculate the current efficiency (CE) as the ratio between the Far-
adaic current stemming from one specific oxidation product and
the overall photocurrent according to the following equation: CE=
[(n·IMS)/K*]/I . Here, n is the number of exchanged electrons, I
an FTO counter electrode in 0.5m H SO4 (Suprapur, Merck) sup-
2
porting electrolyte. The potential was controlled by a potentiostat
(
AFRDE 5, Pine Instruments). The morphology of the electrodepos-
ited WO3 films before and after calcination was determined by
SEM using a Zeiss Leo 1540 microscope (magnifications 20000 and
Ph
MS
the ion current of the respective oxidation product, IPh the overall
2
00000, acceleration voltage 2 kV). The chemical surface composi-
photocurrent and K* the calibration constant for either O or CO .
2
2
tion of the films was characterized by XPS measurements, which
were performed with a Physical Electronics PHI 5800 Multi ESCA
system using monochromatic AlKa radiation. All spectra were ob-
tained at 458 electron emission and at 93.90 and 29.35 eV pass
energy for survey and detailed spectra, respectively. Peak deconvo-
lution was performed by using CasaXPS software, applying a Shirley
background correction and a symmetric peak fit. The BEs were cali-
brated by using the C1s signal at 284.8 eV as reference.
The time delay of approximately 6 seconds between the produc-
tion of volatile species in the flow cell and their mass spectrometric
detection was corrected for all measurements accordingly. The
supporting electrolyte (0.5m H SO ) was prepared from sulfuric
2
4
acid (Suprapur, Merck) and ultrapure water (Millipore Milli-Q,
18.2 MWcm). The electrolytes containing organic compounds were
prepared by adding defined amounts of the following C1 mole-
cules to the supporting electrolyte: methanol (>99.9%, LiChrosolv,
Merck), formaldehyde (aqueous solution of paraformaldehyde,
methanol-free, 16 wt%, Alfa Aesar), formic acid (>98%, pro analysi,
Merck). The electrolyte flow rate driven by the hydrostatic pressure
Photo-electrochemical measurements: All experiments were per-
formed with a setup combining a thin-layer photo-electrochemical
flow cell with on-line mass spectrometry detection, which was re-
À1
in the supply bottles was about 0.8 mLmin . All solutions were
[7]
cently developed by our group. The WO photoelectrodes were
deaerated in the supply bottles by constant purging with high
3
pressed against the cell body, separated by a 50 mm thick gasket
purity N before and during the measurements.
2
(
cut from FEP foil, fluorinated ethylene propylene, Bola) to form
a rectangular thin-layer channel with a volume of about 8 mL. One
inlet and one outlet capillary located in grooves across the rectan-
gular channel at its beginning and at its end ensured a laminar
Acknowledgements
[45]
flow over the whole channel width. Directly opposite from the
photoelectrode, a circular quartz glass window coplanar with the
cell body allowed the illumination of the photocatalyst. A platinum
foil counter electrode was located in a separate compartment and
connected to the electrolyte inlet port using a Luer connection. A
saturated calomel electrode (SCE) connected at the outlet of the
cell served as reference electrode. Photocurrent–potential measure-
ments were performed using a potentiostat (AFRDE 5, Pine Instru-
ments) and recorded using a computerized data acquisition
system. The photoanodes were illuminated by a 200 W Hg(Xe) arc
light source (LOT-QD) equipped with a home-made water filter,
a neutral density filter (10% transmission, LOT-QD), and steel wire
grids (mesh size=1.2”1.2 mm, bar width=0.4 mm) for varying
the light intensity as well as an ultraviolet irradiation transmitting
filter (UG5, Schott AG) and a focusing lens (LOT-QD). The light in-
tensity was measured with a power meter (Model 407A, Newport
Corporation). Experiments under chopped illumination and at con-
stant bias potential were carried out by manually opening or clos-
ing the light path between the flow cell and the focusing lens. The
membrane inlet to the quadrupole mass spectrometer (Pfeiffer
Vacuum, QMS 422) was established by connecting the outlet capil-
lary from the thin-layer photo-electrochemical cell to a second cir-
cular thin-layer compartment, which was equipped with a porous
Teflon membrane (Scimat, 60 mm thick, 50% porosity, 0.2 mm pore
We are grateful to A. Minkow (Institute of Micro- and Nanomate-
rials, Ulm University) and to T. Diemant (Institute of Surface
Chemistry and Catalysis, Ulm University) for the SEM, EDX, XRD,
and XPS measurements.
Keywords: organic
molecule
oxidation
·
photo-
electrochemistry · reaction mechanisms · water splitting · WO3
[1] A. Fujishima, K. Honda, Nature 1972, 238, 37.
[2] A. J. Bard, M. A. Fox, Acc. Chem. Res. 1995, 28, 141.
[
[
3] M. Grätzel, Nature 2001, 414, 338.
4] J. Augustynski, R. Solarska, H. Hagemann, C. Santato in Solar Hydrogen
and Nanotechnology (Ed.: L. Vayssieres), Proc. of SPIE, 2006, 6340,
6
3400J.
5] C. Lamy, A. Devadas, M. Simoes, C. Coutanceau, Electrochim. Acta 2012,
0, 112.
6] C. Lamy, T. Jaubert, S. Baranton, C. Coutanceau, J. Power Sources 2014,
245, 927.
[
[
6
[7] R. Reichert, Z. Jusys, R. J. Behm, Phys. Chem. Chem. Phys. 2014, 16,
25076.
[
[
8] G. Hodes, D. Cahen, J. Manassen, Nature 1976, 260, 312.
9] M. A. Butler, R. D. Nasby, R. K. Quinn, Solid State Commun. 1976, 19,
1011.
[
[
[
[
10] M. A. Butler, J. Appl. Phys. 1977, 48, 1914.
[46]
size) supported on a stainless steel frit. The electrolyte coming
from the photo-electrochemical cell flows over the porous mem-
11] W. Gissler, R. Memming, J. Electrochem. Soc. 1977, 124, 1710.
12] J. Desilvestro, M. Grätzel, J. Electroanal. Chem. 1987, 238, 129.
13] H. Wang, T. Lindgren, J. He, A. Hagfeldt, S. E. Lindquist, J. Phys. Chem. B
2000, 104, 5686.
[7]
brane, allowing the on-line detection of gaseous species. The
mass spectrometer was calibrated for both O and CO detection
2
2
by calculating the ratio K* between the corresponding ion current
IMS) and the Faradaic current (I ) according to the following equa-
[14] Q. Liu, H. Zhang, H. Zhong, S. Zhang, S. Chen, Electrochim. Acta 2012,
81, 313.
(
F
[15] C. Ng, Y. H. Ng, A. Iwase, R. Amal, ACS Appl. Mater. Interfaces 2013, 5,
tion: K*=(n·IMS)/I . Here n is the number of electrons necessary to
F
5269.
form one molecule of O (n=4) or CO (n=2). For calibration, po-
2
2
[16] L. Li, W. Guo, Y. Zhu, Y. Wu, ChemSusChem 2011, 4, 1475.
[17] C. Santato, M. Ulmann, J. Augustynski, J. Phys. Chem. B 2001, 105, 936.
[18] R. Solarska, C. Santato, C. Jorand-Sartoretti, M. Ulmann, J. Augustynski,
J. Appl. Electrochem. 2005, 35, 715.
tential step experiments were performed on a polycrystalline Pt
film electrode, which was prepared by Pt electrodeposition on FTO
[47]
and installed into the flow cell.
Measurements performed in
either pure supporting electrolyte (0.5m H SO ) or in supporting
2
4
[19] D. Monllor-Satoca, L. Borja, A. Rodes, R. Gomez, P. Salvador, ChemPhys-
Chem 2006, 7, 2540.
electrolyte containing 10 mm formic acid allowed the correlation
of the Faradaic currents to the respective products O or CO , as-
[20] M. Rezrazi, O. Bohnke, J. Pagetti, Displays 1987, 8, 119.
2
2
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3686
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
21] O. Bohnke, C. Bohnke, A. Donnadieu, D. Davazoglou, J. Appl. Electro-
chem. 1988, 18, 447.
[36] Y. E. Seidel, Z. Jusys, R. Lindstrçm, M. Stenfeldt, B. Kasemo, K. Krischer,
ChemPhysChem 2010, 11, 1405.
[
[
22] J. P. Hill, K. S. Choi, J. Phys. Chem. C 2012, 116, 7612.
23] J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben in Handbook of X-
ray Photoelectron Spectroscopy (Ed.: J. Chastain), PerkinElmer Corp.,
Eden Prairie/USA, 1992.
[37] H. Gerischer, Electrochim. Acta 1990, 35, 1677.
[38] H. Gerischer, J. Electrochem. Soc. 1966, 113, 1174.
[39] K. Rajeshwar in Encyclopedia of Electrochemistry,Vol. 6 (Eds.: A. J. Bard,
M. Stratmann, S. Licht), Wiley, 2002, pp. 1–51.
[
[
24] F. P. J. M. Kerkhof, J. A. Moulijn, A. Heeres, J. Electron Spectrosc. Relat.
Phenom. 1978, 14, 453.
25] A. Katrib, F. Hemming, P. Wehrer, L. Hilaire, G. Maire, J. Electron Spec-
trosc. Relat. Phenom. 1995, 76, 195.
26] T. L. Barr, J. Phys. Chem. 1978, 82, 1801.
27] S. D. Gardner, C. S. K. Singamsetty, G. L. Booth, G. R. He, C. U. Pittman,
Carbon 1995, 33, 587.
[40] Z. Jusys, J. Kaiser, R. J. Behm, Langmuir 2003, 19, 6759.
[41] Y. E. Seidel, A. Schneider, Z. Jusys, B. Wickman, B. Kasemo, R. J. Behm,
Langmuir 2010, 26, 3569.
[42] A. Schneider, L. Colmenares, Y. E. Seidel, Z. Jusys, B. Wickman, B.
Kasemo, R. J. Behm, Phys. Chem. Chem. Phys. 2008, 10, 1931.
[43] Y. E. Seidel, A. Schneider, Z. Jusys, B. Wickman, B. Kasemo, R. J. Behm,
Faraday Discuss. 2009, 140, 167.
[
[
[
[
[
28] S. R. Morrison, H. J. Freund, J. Chem. Phys. 1967, 47, 1543.
29] J. A. Seabold, K. S. Choi, Chem. Mater. 2011, 23, 1105.
30] T. Lana-Villarreal, R. Gómez, M. Neumann-Spallart, N. Alonso-Vante, P.
Salvador, J. Phys. Chem. B 2004, 108, 15172.
[44] J. Luo, M. Hepel, Electrochim. Acta 2001, 46, 2913.
[45] J. Fuhrmann, H. Zhao, E. Holzbecher, H. Langmach, M. Chojak, R. Hal-
seid, Z. Jusys, R. J. Behm, Phys. Chem. Chem. Phys. 2008, 10, 3784.
[46] S. Sun, M. Chojak-Halseid, M. Heinen, Z. Jusys, R. J. Behm, J. Power Sour-
ces 2009, 190, 2.
[31] T. Lana-Villarreal, J. M. Pꢁrez, R. Gómez, C. R. Chim. 2006, 9, 806.
[32] X.-Q. Gong, A. Selloni, A. Vittadini, J. Phys. Chem. B 2006, 110, 2804.
[33] W. J. McElroy, S. J. Waygood, J. Chem. Soc. Faraday Trans. 1991, 87,
[47] W. Zhao, Z. Jusys, R. J. Behm, Anal. Chem. 2012, 84, 5479.
1
513.
[
[
34] M. Heinen, Z. Jusys, R. J. Behm in Handbook of Fuel Cells, Vol. 5, (Eds.:
H. A. Gasteiger, W. Vielstich, H. Yokogawa), Wiley, Chichester 2009,
pp. 183–214.
35] R. Lindstrçm, Y. E. Seidel, Z. Jusys, M. Gustavsson, B. Kasemo, R. J. Behm,
J. Electroanal. Chem. 2010, 644, 90.
Received: June 15, 2015
Revised: July 31, 2015
Published online on September 18, 2015
ChemSusChem 2015, 8, 3677 – 3687
www.chemsuschem.org
3687
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim