Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Communication

www.advenergymat.de

Highly Efficient Sunlight-Driven Seawater Splitting in


a Photoelectrochemical Cell with Chlorine Evolved at
Nanostructured WO3 Photoanode and Hydrogen Stored
as Hydride within Metallic Cathode
Michal Jadwiszczak, Katarzyna Jakubow-Piotrowska, Piotr Kedzierzawski,
Krzysztof Bienkowski,* and Jan Augustynski*

semiconductor photocatalytic particles,


A seawater splitting photoelectrochemical cell featuring a nanostructured pure water is generally used and both
tungsten trioxide photoanode that exhibits very high and stable photocurrents reduction and oxidation reaction prod-
producing chlorine with average 70% Faradaic efficiency is described. Fabrica- ucts, H2 and O2, are formed in the same
cell compartment and should be sub-
tion of the WO3 electrodes on fluorine-doped tin oxide substrates involves a sequently separated.[2,3] On the other
simple solution-based method and sequential layer-by-layer deposition with hand, employing a photoelectrolysis cell
a progressively adjusted amount of structure-directing agent in the precursor with separated anodic and cathodic com-
and a two-step annealing. Such a procedure allows tailoring of thick, highly partments imposes use of a supporting
porous, structurally stable WO3 films with a large internal photoactive surface electrolyte with ionic conductivity large
enough to avoid excessive Ohmic losses
area optimizing utilization of visible light wavelengths by the photoanode. With
within the cell.[4,6] The choice of the appro-
the application of an anodic potential of 0.76 V versus Ag/AgCl reference elec- priate electrolyte is even more important
trode (0.4 V below the thermodynamic Cl2/Cl− potential) in synthetic seawater, when relatively thick nanoporous film
the designed WO3 photoanodes irradiated with simulated 1 sun AM 1.5G photoelectrodes with high internal photo-
light reach currents exceeding 4.5 mA cm−2. Photocurrents close to 5 mA cm−2 active surface area are employed, where
are attained in the case of fresh water splitting using 1 m methane–sulfonic too low conductivity of the electrolyte may
adversely affect the amount of collected
acid supporting electrolyte with oxygen evolved at the WO3 photoanode. The photocurrent due to uneven current dis-
amount of formed hydrogen is determined by discharging the palladium sheet tribution across the semiconductor film.[7]
electrode employed as a cathode. Collection of hydrogen in the form of a Consequently, in addition to pure water,
hydride opens, more generally, the prospect of subsequently using such mate- also the chemicals required to prepare
rials as anodes in batteries employing oxygen reduction cathodes. electrolyte for the photoelectrolysis cell
will potentially constitute a nonnegligible
part of the operational cost of larger scale
Photocatalytic water splitting is widely investigated as a pro­­­ photoelectrochemical (PEC) water splitting devices. The fact
mising means for sustainable fuel generation using sunlight.[1–5] that cannot at all be neglected—extensive utilization for elec-
When the splitting process is conducted via suspensions of trolysis of fresh water would put heavy pressure on vital water
resources.
Since the seawater is a free and widely abundant electro-
M. Jadwiszczak, K. Jakubow-Piotrowska lyte, there were various attempts to use it in the PEC[8–13] and
Centre of New Technologies conventional electrochemical[14] devices to produce hydrogen.
University of Warsaw From the practical viewpoint, the photoelectrolysis of seawater
S. Banacha 2c, 02-097 Warsaw, Poland
requires at first identification of photoanode materials stable
Dr. P. Kedzierzawski
Institute of Physical Chemistry of Polish Academy of Sciences
in the presence of chloride ions under highly oxidizing condi-
01-224 Warsaw, Poland tions. Only few among works reported on the PEC seawater
Dr. K. Bienkowski, Prof. J. Augustynski splitting describe experiments conducted under visible light
Centre of New Technologies irradiation. An electrode consisting of molybdenum-doped bis-
University of Warsaw muth vanadate (Mo-BiVO4) reached, under simulated AM 1.5G
S. Banacha 2c, 02-097 Warsaw, Poland (100 mW cm−2) illumination, a photocurrent of 2.2 mA cm−2
E-mail: k.bienkowski@cent.uw.edu.pl; jan.augustynski@unige.ch
at 1 V versus RHE (reversible hydrogen electrode).[9] Loading
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/aenm.201903213.
the Mo-BiVO4 electrode with precious metal RhO2 catalyst
was effective in limiting to ≈10% the drop of the photocurrent
DOI: 10.1002/aenm.201903213 over a 5 h long stability test. Analysis of the photoelectrolysis

Adv. Energy Mater. 2019, 1903213 1903213 (1 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

products showed formation of a mixture of oxygen and chlo- of the photoelectrolyzer with a photovoltaic (PV) cell to form a
rine/hypochlorite (Cl2/HClO).[9] However, as demonstrated tandem device.[27]
by a recent extensive investigation,[15] a BiVO4 photoanode Besides long-term stability, other important criteria are to
undergoes in the electrolyte partial anodic dissolution (the loss be fulfilled contributing to an efficient operation of a semicon-
of V5+ ions from the lattice) that occurs in parallel with photo- ductor film photoelectrode. In particular, indirect nature of the
oxidation of water, the process that can be alleviated—but not optical transition in WO3 requires use of relatively thick films
suppressed by the use of catalysts. Another study, employing a to allow absorption of visible blue wavelengths close to the
titanium oxide (TiO2) nanorod electrode, coated with polymeric band edge. In principle, well-designed nanoporous electrodes
graphitic carbon nitride (g-C3N4) and subsequently decorated provide a large internal photoactive surface area in contact
with electrodeposited cobalt phosphate-oxide (Co-Pi) electro- with the electrolyte and, in the case of WO3, a relatively long
catalyst, demonstrated under similar standard conditions, i.e., hole diffusion length (≈150 nm)[24,28] may prevent bulk elec-
AM 1.5G illumination at 1.23 VRHE, stable photocurrent of tron–hole recombination within the nanoparticles (NPs) that
1.6 mA cm−2.[12] No chlorine formation along seawater photo- form the film. To allow even photocurrent distribution across
electrolysis was mentioned, that is apparently consistent with the NS film, the conductivity of the electrolyte filling the pores
the use in that case of the Co-Pi oxygen-evolution catalyst. should be large enough to avoid excessive Ohmic drops.[7] We
Recently, unassisted solar light-driven seawater splitting with note that the amount of generated current collected at the elec-
solar-to-hydrogen (STH) efficiency of 1.9 % has been dem- trode substrate depends not only on the activity of the catalytic
onstrated on p-type gallium nitride/indium gallium nitride sites, where positive holes react with the solution species, but
(p-GaN/InGaN) nanowire arrays.[13] The authors assigned the also on the electron mobility that is relatively large (in the range
increased rate of H2 evolution observed in the experiments con- of 6.5–12 cm2 V−1 s−1 ) in the WO3[29,30] and on the crystalline
ducted in synthetic seawater, compared to pure water, to better quality of interparticle contacts.[31]
conductivity of the reaction medium and to the oxidation of In an attempt to match in particular the penetration depth
chloride ions suggested to act as effective intermediates in the of blue light photons close to the band edge, we prepared
process of oxygen evolution. relatively thick (close to 3 µm) NS WO3 films with modulated
Another semiconductor that absorbs a portion of visible porosity. As detailed in the Supporting Information, such films
solar spectrum and is able to produce chlorine through pho- are processed by a sequential layer-by-layer deposition (of up to
tooxidation of Cl− ions in acidic media is an n-type tungsten 8–9 consecutive individual layers) of the colloidal precursor and
trioxide.[16–20] With a bandgap that varies within 2.5–2.7 eV annealing under flowing oxygen. Decomposing the precursor
depending on its crystalline structure,[21–24] the nanocrystalline and, subsequently, sintering the formed WO3 NPs in oxi-
WO3 absorbs light in the blue region up to 500 nm.[25] Due to dizing atmosphere are essential to avoid formation within the
the photocurrent onset potential of about 0.45 VRHE, to operate film of an excessive amount of oxygen vacancy defects that may
as a water splitting photoanode the WO3 requires assistance of act as electron traps.[32] Such formed WO3 films annealed in O2
an external voltage bias. On the other hand, the higher valence atmosphere exhibit yellow color consistent with an absorption
band (VB) edge of WO3 located above 2.9 VRHE[26] is by far posi- maximum at 390–400 nm. The porosity of the film and the size
tive enough to allow the photogenerated holes to drive a variety of formed NPs are actually controlled by the amount of organic
of oxidation reactions. Efficient photooxidation of water to templates-polyethylene glycols (PEG 300 and/or PEG 600)
form oxygen, producing stable photocurrents, has been dem- added to the WO3 precursor tungstic acid solution and by mod-
onstrated using methane–sulfonic acid supporting electrolyte.[7] ulating the annealing temperature applied to particular layers.
However, the latter case appears as an exception since several In order to reconcile the required optical WO3 film thickness
other anions of the employed supporting electrolytes (sulfates, with the possibly even photocurrent penetration down to the
perchlorates, and phosphates) undergo photooxidation at the back contact we prepared the first three to four layers from the
WO3 electrode leading to the formation in solution of peroxo precursor containing larger amount of PEG (0.25 w/w WO3/
species occurring in parallel with (or instead of) oxygen evolu- PEG ratio). Such increase of the amount of template in the
tion.[16–19,7] In such cases, peroxides tend also to build on the film precursor results in formation of smaller WO3 NPs and
WO3 electrode surface resulting in the progressive drop of larger porosity within the film. This choice stemmed from the
the photocurrent.[16,18,7] Interestingly, such “passivation” of the fact that one of the consequences of the layer-by-layer ­deposition
WO3 electrode is impeded when the supporting acidic electro- procedure is more important sintering of the first deposited
lyte contains even a relatively small amount of chloride ions.[16] layers that undergo several annealing sequences. We have also
In this article, we show that tailoring the morphology and used a two-step calcination procedure, typically 510 (±10) °C for
optimizing crystallinity of semitransparent nanostructured (NS) the first four layers and subsequently 550 (±10) °C, to avoid too
WO3 film electrodes allows attaining very high efficiency of vis- rapid burning of organic components of the precursor in the
ible light-driven seawater splitting. As shown by the detailed first deposited layers that might affect the film adherence to the
PEC characterization below, the designed NS WO3 photoanodes FTO substrate. We also note that the employed organic addi-
exhibit excellent charge separation, transport and collection effi- tives, such as PEG, act as stabilizers of the precursor tungstic
ciency, which combined with large photoactive surface area lead acid colloidal solution[33] and, subsequently, favor binding of
to high and stable seawater splitting saturation photocurrents the WO3 NPs within the film.
attained at 0.5–0.6 V above the onset potential. We note that the Figure 1a shows the cross-sectional scanning electron
operating potential point at which the photocurrent approaches microscopic (SEM) image of a ≈3 µm thick mesoporous WO3
saturation is particularly important in view of the combination film with some pores reaching apparently the FTO substrate.

Adv. Energy Mater. 2019, 1903213 1903213 (2 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 1. a) SEM cross-section and b) top-down images of a NS WO3 film c) TEM image of WO3 removed from the substrate. d) Raman spectrum of
the WO3 film annealed at 550 °C.

The film consists of a network of individual particles with and of NPs agglomerates, it is evident that a large part of the
dimensions ranging from 20 to 30 nm and of agglomerates reaction sites is located within the pores inside the film. While
including in some cases several NPs (cf. top-down surface SEM in the case when water is photo-oxidized into O2 this generally
image in Figure 1b). X-ray diffraction patterns (Figure S1 in leads to an initial drop of the photocurrent due to the formed
the Supporting Information) indicate monoclinic crystal struc- gas bubbles filling the pores within the WO3 electrode, we
ture of the WO3 film with three main sharp peaks character- observed that this problem becomes less important when the
istic of the preferential (002), (020), and (200) orientation of the main photoanodic product is chlorine soluble in water. Such
­crystallites.[31] We note the highest intensity of the (200) peak is indeed the case for the herein reported splitting of seawater
corresponding, according to the theoretical calculations,[34] to where, as detailed later, we observe dominant chlorine forma-
the WO3 facet at which the charge-transfer reactions require the tion and stable photocurrents. The additional advantage of the
lowest overpotential. The crystallinity of the synthesized WO3 Cl2 generation occurring within the pores in the WO3 electrode
is confirmed by the transmission electron microscopy (TEM) is a local acidification of the solution due to chlorine dispropor-
image of the particles removed from the film (Figure 1c). The tionation leading to the formation of hypochlorous acid and of
Raman spectrum in Figure 1d exhibits main features typical of hydrogen ions.[36]
monoclinic structure of WO3, with prominent bands located
Cl 2 + 2H2 O → HClO + Cl – + H3 O+ (1)
close to 715 and 805 cm−1 and weaker at around 270 and
325 cm−1.[31] We also characterized the WO3 films by X-ray This reaction allows preserve electrode stability over photo-
photoelectron spectroscopy (XPS). The Na 1s XPS signal indi- electrolysis of initially neutral electrolytes (such as seawater).
cated presence of around 1 at% of sodium expressed as Na/W Actually, the dissolution of Cl2, under arbitrary chosen rela-
atom ratio consistent with the optimum film doping that was tively low partial pressure of 0.067 atm. in a 0.5 m aq. NaCl
achieved by adding 0.4 at% of Na+ to the tungstic acid precursor (corresponding approximately to its typical concentration
containing a residue of Na2WO4.[35] On the other hand, the only in seawater) results already in a solution pH of around 2.5
detectable impurity on the WO3 film surface was adventitious (cf. Section SIV in the Supporting Information). We recall,
carbon (Figure S2 in the Supporting Information). in this connection, that according to the Pourbaix diagram[36]
Since the formed NS films consist of networks of several tens WO3 is thermodynamically stable in acidic solutions of pH
of superimposed WO3 particles with (20–30 nm) d ­ imensions lower than 4.

Adv. Energy Mater. 2019, 1903213 1903213 (3 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

The important advantage of the acidic environment established


within the pores and close to the photoanode is also to impede
the decomposition of hypochlorous acid to form chlorates and
oxygen. This process requires, in fact, the presence of both the
HClO and hypochlorite OCl− ions,[37] practically absent from an
acidic solution of pH 2, due to the fact that HClO is a weak acid
of pKa = 7.5 (cf. Section S IV in the Supporting Information).
To preserve acidic environment in the anodic compartment,
we carried out the seawater splitting experiments in a two-
compartment photoelectrolysis cell featuring sintered glass
diaphragm and the palladium sheet cathode (Figure 2a). We
illuminated the WO3 photoanode from the electrolyte side or in
some cases through the FTO substrate. In Figure 2b is shown
linear sweep voltammetry photocurrent density–potential (j–E)
plot for an around 3 µm thick WO3 electrode recorded under
simulated AM 1.5G solar irradiation (100 mW cm−2) in a syn-
thetic seawater of initial pH 6.5. The WO3 photoanode attains
here a very large saturation photocurrent of 4.78 mA cm−2 at
0.88 V versus Ag/AgCl. We note a negative shift of the onset
potential due to almost neutral initial pH of seawater. Interest-
ingly, we recorded the same saturation photocurrent (uncor-
rected for the light losses within the FTO layer on glass) at a
larger potential of 1.1 V when we illuminated the WO3 electrode
from the back side. The maximum photocurrents recorded for
a series of WO3 film electrodes prepared in a similar way were
quite reproducible and varied within 4.6–4.8 mA cm−2. The
WO3 photoanode that reached in the seawater the maximum
photocurrent of 4.78 mA cm−2 and served subsequently for
numerous other measurements, tested 5 months later, deliv-
ered a saturation photocurrent of 4.64 mA cm−2—only 3%
lower than the initial value. Measurements carried out under
chopped illumination, (Figure S3 in the Supporting Informa-
tion) that reveal presence of dark currents in the initial part of
the j–E plot, show only very small photocurrent transients asso-
ciated with light-on and light-off indicative of fast charge car-
rier mobility and good electric contact between the WO3 NPs.
The latter observation is consistent with the incident photon-
to-current conversion efficiency (IPCE) plot (Figure 2c) that
exhibits a maximum of around 85% over the 390–410 nm wave-
length range. Such measurements were also performed under
electrode illumination from the back side and were corrected
for the light absorption/reflection by the FTO layer on glass
(Figure 2c green trace). Importantly, we observe in that case
larger IPCEs, particularly at shorter wavelengths corresponding
to the slab of the WO3 film close to the FTO substrate, and an
IPCE of 94% reached at 400 nm in the region of optical absorp-
tion maximum. The latter result suggests that the electrolyte
may penetrate the pores in the designed ≈3 µm thick WO3
film close to the FTO substrate. We assign the drop of IPCEs,
observed for the wavelengths close to the band edge mainly to
Figure 2. a) Schematic diagram of the seawater splitting PEC cell. The
the still limited optical thickness of the WO3 light absorber. current of CO2 gas passing through the seawater represents a possible
We also note that those photocurrents, measured at the wave- option for maintaining slightly acidic pH in the cathodic cell compart-
lengths above 450 nm passing through the monochromator ment. b) Photoanodic current versus potential (j–E) plots for a ≈3 µm
(Figure S4 in the Supporting Information), are probably too thick WO3 electrode recorded in synthetic seawater under AM 1.5G
weak to saturate majority of electron traps present within the (100 mW cm−2) illumination from the electrolyte side (black curve) and
from the back side—through the FTO substrate (green trace). We note
relatively thick WO3 film that can thus operate as electron–hole
that the latter results have not been corrected for the light losses within
recombination centers. We expect the latter effect to become the FTO substrate. c) Corresponding IPCE spectra; the IPCE values
less important when we irradiate the WO3 electrode with much recorded under back side illumination (green plot) were corrected for the
more intense, simulated AM 1.5G sunlight. light absorption/reflection by the FTO layer on glass.

Adv. Energy Mater. 2019, 1903213 1903213 (4 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

Figure 3. a) PEC water oxidation current versus potential curve (recorded under simulated AM 1.5G solar light) and b) the IPCE spectrum for a 3 µm
thick WO3 photoanode both measured in a 1 m methane-sulfonic acid electrolyte.

For comparison with the seawater, we also performed peroxide layer on WO3 may be quenched by chloride ions, it is
measurements in a 0.5 m NaCl solution acidified to pH 2 that important to recall a strong oxidant activity of the peroxo com-
gave comparable results (see Figure S5 in the Supporting plexes formed by early transition metals in their highest oxi-
Information). dation state: tungsten (VI), molybdenum (VI), vanadium (V),
To further evaluate the PEC performance of the designed and rhenium (VII). In particular, these complexes are able to
≈3 µm thick WO3 films, we also carried out measurements in a oxidize halides.[41] The latter might be the reason for which the
1 m methane–sulfonic acid supporting electrolyte where oxygen photooxidation at WO3 of chloride anions to form Cl2 may be
is formed at the photoanode.[7,35] As shown in Figure 3a, at perceived as a self-activated reaction.
1.23 VRHE) even larger, than in the seawater, saturation photo- In Figure 4a, is represented result of WO3 photoanode sta-
current of 4.97 mA cm−2 is attained; for several similar tested bility test performed in a 0.5 m NaCl solution acidified to pH 2.
WO3 photoanodes the maximum photocurrents exceeded regu- The evolution of the photocurrent in function of time was followed
larly 4.8 mA cm−2. Like in the case of seawater splitting experi- at an imposed potential of 0.76 V versus Ag/AgCl. We assign a
ments (cf. Figure S3 in the Supporting Information), we note moderate decline of the photocurrents to the accumulation of gas
here presence of dark currents at low anodic bias (cf. j–E trace bubbles (principally O2 cogenerated with Cl2) within the pores of
recorded under chopped illumination in Figure S6 in the Sup- the WO3 photoanode causing reduction of the internal photoactive
porting Information) that we assign to release of protons inserted surface area. The latter electrode used subsequently in a series of
into the WO3 lattice,[38] under electrode illumination at open cir- other experiments exhibited practically unchanged j–E behavior.
cuit before recording the j–E plot. In fact, the amount of such dark As already mentioned, the changes in the saturation photocur-
current was proportional to the period of time the WO3 electrode rents after several months of the WO3 photoanode utilization did
spent illuminated at open circuit, before starting anodic polariza- not exceed 3%. For comparison we show in Figure 4b result of
tion. However, the dark current contribution does no more affect a similar stability test conducted with a much thinner (≈0.5 µm
the photocurrents observed halfway from the saturation region. thick) WO3 photoanode. Consistently with its much lower internal
As shown in Figure 3b, consistently with the j–E plot, the WO3 surface area, the change of the photocurrent along the 20 h elec-
electrode exhibits in the 1 m CH3SO3H electrolyte high IPCEs trolysis is here less pronounced confirming good chemical and
with a maximum of 87% observed over 390–400 nm wavelengths. mechanical stability of the WO3 electrode.
As already mentioned, the photoanodic behavior of WO3 is The amounts of chlorine formed in the anodic cell compart-
in general dominated by oxidation of anions of the supporting ment over photoelectrolysis of a 0.5 m NaCl solution or seawater
electrolyte that, for example, in the case of HSO4− ions leads acidified to pH 2 were determined by spectrophotometry using
to the formation of peroxodisulfate species,[39,16] and is most N,N-diethyl-p-phenylenediamine sulfate (DPD) and/or methyl
often accompanied by the build-up of a surface peroxide layer orange reagents (see Section S V in the Supporting Information).
resulting in a reversible deactivation of the photoanode (with During the experiments, the cell was closed hermetically with
more or less rapid, progressive drop of the photocurrent).[16–19,7] the two compartments separated by a Nafion membrane; the
An important exception we identified previously,[7] is the photo- latter explains use in this case of slightly acidified electrolyte to
electrolysis conducted in aqueous solutions of CH3SO3H that preserve the ionic contact between the photoanode and cathode.
produces at the WO3 electrode stable O2 evolution currents, due The results of analyses summarized in Figure 4c show on
apparently to very high reactivity of the CH3SO3· radical spe- average a 70% Faradaic efficiency of chlorine formation at the
cies with water.[40] On the other hand, experiments performed WO3 photoanode occurring, apparently, in parallel with oxygen
with addition of even small amounts of chloride ions to solu- evolution:
tions containing larger concentrations of oxy-anions (ClO4−, 2Cl – → Cl 2 + 2e – (2)
SO42−) showed the absence of deactivation of the anodically
1
polarized WO3 photoelectrode. To understand the way a surface 2H2 O → O2 + 2H3 O+ + 2e – (3)
2

Adv. Energy Mater. 2019, 1903213 1903213 (5 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

In view of comparison, are also represented in Figure 4c


Faradaic efficiencies measured for another photoanode able
to produce chlorine—a film of rutile TiO2 formed by thermal
oxidation of metallic titanium substrate at 600 °C[42] as well as
those corresponding to the amounts of chlorine formed at the
bright platinum and FTO anodes (in the absence of illumina-
tion). Like the WO3 photoanode, the Pt electrode produced a
mixture of Cl2/HClO and oxygen with larger amount ≈60% of
chlorine species. For the WO3, we performed the photoelectro-
lyses at 0.6 V versus Ag/AgCl corresponding to the rising por-
tion of the j–E plot (Figure S5 in the Supporting Information).
On the other hand, the Pt anode was polarized at 1.5 V, ≈0.32 V
above the Cl2/Cl− equilibrium potential. Low chlorine forma-
tion Faradaic efficiencies were observed at the FTO used as an
anode; in that case, generation of any sizeable current required
high overpotentials.
The results of our analyses showing dominant formation of
chlorine occurring during photo-oxidation of simulated sea-
water at the WO3 electrode are consistent with the conclusions
of another PEC study performed in a 3 m NaCl electrolyte that
used photoanodes formed by deposition of commercial WO3
nanopowder on a conductive glass substrate.[20] By controlling
with a bipotentiostat the potential of a Pt “collector” microelec-
trode placed very close to the WO3 photoanode, the authors
detected cathodic currents in phase with the photocurrents gen-
erated at the WO3 submitted to blue light pulses. Such reduc-
tion current responses have been detected up to a potential
around 1.24 V versus standard hydrogen electrode (SHE), i.e.,
just below the Cl2/Cl− equilibrium potential.
Recently, photoelectrolysis of a yet more concentrated 4 m
NaCl solution acidified to pH 1 at a BiVO4 photoanode coated
with a thin (20 nm) layer of amorphous WO3 has been reported
to produce chlorine with up to 85% (on average 74%) Faradaic
efficiency.[43] The WO3 coating served as a corrosion protec-
tion of the BiVO4 absorber film and apparently as a catalyst
allowing Cl2 formation with maximum photocurrent density of
2.8 mA cm−2 attained under 1 sun illumination at 1.6 V versus
RHE (≈1.34 V vs Ag/AgCl). The latter potential is, however,
already 0.18 V above the Cl2/Cl− thermodynamic potential. In
comparison as shown in Figure 2b, the WO3 photoanodes used
in our experiments reached during the seawater splitting the
photocurrent densities of 4.5 mA cm−2 at 0.76 V versus Ag/
AgCl, i.e., ≈0.4 V below E (Cl2/Cl−).
Although the use of seawater as electrolyte in the PEC device
to produce chlorine and hydrogen is undoubtedly an attractive
option, it has also a relative drawback related to the absence
of the pH buffering capacity. The consequence of such situa-
tion is the local drift of the pH close to the cathode in alkaline
­direction occurring during electrolysis with the cathode poten-
Figure 4. Seawater splitting photocurrent–time curves recorded for a
larger (≈1.5 cm−2) a) 3 µm thick and b) 0.5 µm thick WO3 photoanodes
tial becoming more negative. The solution we employed to limit
illuminated with simulated AM 1.5G (70–100 mW cm−2) solar light in a this problem in our relatively short PEC measurements was a
0.5 m NaCl solution acidified to pH 2. c) Seven Faradaic efficiencies from slight acidification of seawater by addition of a small amount
spectrophotometric measurements of chlorine formed in seawater of of hydrochloric acid (to attain pH 2). Another possibility (not
pH 2 at a) WO3 photoanode at a potential of 0.6 V versus Ag/AgCl (average tested), that we suggest for longer PEC cell operation, might
value of 10 measurements with a maximum of 78%) b) WO3 photoanode consist in circulating through the cathode compartment the
at a potential of 0.75 V versus Ag/AgCl (average value of two measure-
seawater saturated with carbon dioxide. The latter approach is
ments with a maximum of 79%). c) FTO electrode at 1.5 VAg/AgCl. d) FTO
at 1.75 VAg/AgCl. e) TiO2 photoanode at 0.6 VAg/AgCl. f) TiO2 photoanode at made possible by the use of a palladium cathode with hydrogen
1 VAg/AgCl. g) Pt electrode at 1.5 VAg/AgCl. Apart from the WO3 photoanode, being stored in the form of hydride. We note that in the range
all other efficiencies were determined within single measurements. of potentials the palladium hydride is formed the reduction of

Adv. Energy Mater. 2019, 1903213 1903213 (6 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

CO2 does not take place. Maintaining the electrolyte pH slightly Natural Sciences, Cardinal Stefan Wyszynski University for the Raman
acidic in the cathode compartment would, consequently, allow measurements and Dr. M. Pisarek from the Institute of Physical
maintaining the operating voltage of the PEC cell producing Chemistry PAS for the XPS measurements.
chlorine and hydrogen in the 1.3–1.4 V range.
In Section S VI and Figure S7 in the Supporting Informa-
tion are represented the curves of charging of the Pd cathode Conflict of Interest
during seawater splitting and, subsequently, of the hydrogen The authors declare no conflict of interest.
extraction conducted over three times longer period of time.
To avoid eventuality of the palladium corrosion, the extraction
of hydrogen was conducted in a 0.1 m NaOH solution. That
Keywords
experiment showed an extraction efficiency of 93% suggesting
that practically all hydrogen formed at the cathode of the PEC chlorine generation, hydride-forming metallic cathode, hydrogen
seawater splitting cell is in the form of hydride. generation, nanostructured tungsten trioxide, photoelectrochemical cell,
seawater splitting, WO3 photoanode
In summary, we have shown that a photoelectrolyzer that
uses a nanostructured WO3 photoanode combined with a pal- Received: September 30, 2019
ladium cathode allows efficient seawater splitting with the Revised: November 14, 2019
dominant chlorine production in the anodic compartment and Published online:
hydrogen generation at the cathode in the form of palladium
hydride. The principal benefits of such cell configuration are
largely reduced gas mixing and the possibility of moderating
change of the pH close to the cathode by circulating through [1] K. Sivula, R. van de Krol, Nat. Rev. Mater. 2016, 1, 15010.
the corresponding cell compartment slightly acidified sea- [2] K. Maeda, M. Higashi, D. Lu, R. Abe, K. Domen, J. Am. Chem. Soc.
water (the suggested means is here carbon dioxide). These 2010, 132, 5858.
[3] T. Hisatomi, J. Kubota, K. Domen, Chem. Soc. Rev. 2014, 43, 7520.
results open, more generally, the prospect of the direct storage
[4] M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi,
of hydrogen generated in a PEC water splitting device in the E. A. Santori, N. S. Lewis, Chem. Rev. 2010, 110, 6446.
form of a hydride that might be subsequently used as an [5] S. Wang, G. Liu, L. Wang, Chem. Rev. 2019, 119, 5192.
anode in a hydride-oxygen battery. There are numerous mul- [6] B. Seger, I. E. Castelli, P. C. K. Vesborg, K. W. Jacobsen, O. Hansen,
ticomponent metallic alloys investigated in view of application I. Chorkendorff, Energy Environ. Sci. 2014, 7, 2397.
as negative electrodes in the metal hydride-nickel oxide storage [7] R. Solarska, R. Jurczakowski, J. Augustynski, Nanoscale 2012, 4,
batteries[44,45] that might be a less expensive alternative to the 1553.
palladium cathode used in the present work. The described [8] S. M. Ji, H. Jun, J. S. Jang, H. C. Son, P. H. Borse, J. S. Lee,
NS WO3 photoanodes deliver under 1 sun illumination photo- J. Photochem. Photobiol., A 2007, 189, 141.
current densities above 4.5 mA cm−2 at a potential 0.4 V lower [9] W. Luo, Z. Yang, Z. Li, J. Zhang, J. Liu, Z. Zhao, Z. Wang, S. Yan,
T. Yu, Z. Zou, Energy Environ. Sci. 2011, 4, 4046.
than the thermodynamic chlorine evolution potential (with the
[10] Y. Li, Y. Xiang, S. Peng, X. Wang, L. Zhou, Electrochim. Acta 2013,
maximum photocurrent of 4.78 mA cm−2 recorded at 0.88 V vs 87, 794.
Ag/AgCl)—that outperform the earlier, including very recent, [11] C.-J. Chang, Z. Lee, C.-F. Wang, Int. J. Hydrogen Energy 2014, 39,
experimental results of PEC seawater splitting and of chlorine 20754.
generation from more concentrated NaCl solutions reported in [12] Y. Li, R. Wang, H. Li, X. Wei, J. Feng, K. Liu, Y. Dang, A. Zhou,
the literature. J. Phys. Chem. C 2015, 119, 20283.
The sunlight-driven photoelectrochemical seawater splitting [13] X. Guan, F. A. Chowdhury, N. Pant, L. Guo, L. Vayssieres, Z. Mi,
devices, operating on-site, might be particularly well suited to J. Phys. Chem. C 2018, 122, 13797.
the remote places where relatively small amounts of chlorine/ [14] N. A. A. Ghany, N. Kumagai, S. Meguro, K. Asami, K. Hashimoto,
hypochlorite are required for a variety of vital disinfecting pur- Electrochim. Acta 2002, 48, 21.
[15] D. K. Lee, K.-S. Choi, Nat. Energy 2018, 3, 53.
poses including purification of drinking water and where sea-
[16] J. Augustynski, R. Solarska, H. Hagemann, C. Santato, Proc. Soc.
water and sunlight are abundant. Photo-Opt. Instrum. Eng. 2006, 6340, U140.
[17] J. A. Seabold, K.-S. Choi, Chem. Mater. 2011, 23, 1105.
[18] J. C. Hill, K.-S. Choi, J. Phys. Chem. C 2012, 116, 7612.
Supporting Information [19] Q. Mi, A. Zhanaidarova, B. S. Brunschwig, H. B. Gray, N. S. Lewis,
Energy Environ. Sci. 2012, 5, 5694.
Supporting Information is available from the Wiley Online Library or [20] S. Ahmed, I. A. I. Hassan, H. Roy, F. Marken, J. Phys. Chem. C 2013,
from the author. 117, 7005.
[21] M. Gillet, K. Aguir, C. Lemire, E. Gillet, K. Schierbaum, Thin Solid
Films 2004, 467, 239.
[22] H. Zheng, J. Z. Ou, M. S. Strano, R. B. Kaner, A. Mitchell,
Acknowledgements K. Kalantar-zadeh, Adv. Funct. Mater. 2011, 21, 2175.
This work was supported by the MAESTRO grant No. UMO-2013/10/A/ [23] Y. Wang, W. Tian, C. Chen, W. Xu, L. Li, Adv. Funct. Mater. 2019, 29,
ST5/00245 awarded to J.A. by the Polish National Science Centre. 1809036.
K.B. acknowledges support from the SONATA Bis Grant 2017/26/E/ [24] M. A. Butler, J. Appl. Phys. 1977, 48, 1914.
ST5/01137 granted by the Polish National Science Centre. The [25] C. Santato, M. Ulmann, J. Augustynski, J. Phys. Chem. B 2001, 105,
authors thank Dr. D. Kurzydlowski from Faculty of Mathematics and 936.

Adv. Energy Mater. 2019, 1903213 1903213 (7 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.advancedsciencenews.com www.advenergymat.de

[26] L. Weinhardt, M. Blum, M. Bär, C. Heske, B. Cole, B. Marsen, [38] B. W. Faughnan, R. S. Crandall, P. H. Heyman, RCA Rev. 1975, 36,
E. L. Miller, J. Phys. Chem. C 2008, 112, 3078. 177.
[27] J. Brillet, J.-H. Yum, M. Cornuz, T. Hisatomi, R. Solarska, [39] J J. Desilvestro, M. Graetzel, J. Electroanal. Chem. Interfacial Electro-
J. Augustynski, M. Graetzel, K. Sivula, Nat. Photonics 2012, 6, 824. chem. 1987, 238, 129.
[28] C. Santato, M. Ulmann, J. Augustynski, Adv. Mater. 2001, 13, 511. [40] C. J. Myall, D. Pletcher, J. Chem. Soc., Perkin Trans. 1 1975, 1,
[29] J. M. Berak, M. J. Sienko, J. Solid State Chem. 1970, 2, 109. 953.
[30] K. J. Patel, C. J. Panchal, V. A. Kheraj, M. S. Desai, Mater. Chem. [41] M. S. Reynolds, K. J. Babinski, M. C. Bouteneff, J. L. Brown,
Phys. 2009, 114, 475. R. E. Campbell, M. A. Cowan, M. R. Durwin, T. Foss, P. O’Brien,
[31] C. Santato, M. Odziemkowski, M. Ulmann, J. Augustynski, J. Am. H. R. Penn, Inorg. Chim. Acta 1997, 263, 225.
Chem. Soc. 2001, 123, 10639. [42] K. J. Hartig, N. Getoff, G. Nauer, Int. J. Hydrogen Energy 1983, 8,
[32] M. Gillet, C. Lemire, E. Gillet, K. Aguir, Surf. Sci. 2003, 532-535, 519. 603.
[33] E. Richardson, J. Inorg. Nucl. Chem. 1959, 12, 79. [43] A. M. Rassoolkhani, W. Cheng, J. Lee, A. McKee, J. Koonce, J. Coffel,
[34] A. Valdes, G. J. Kroes, J. Chem. Phys. 2009, 130, 114701. A. H. Ghanim, G. A. Aurand, C. S. Kim, W. I. Park, H. Jung,
[35] M. Sarnowska, K. Bienkowski, P. J. Barczuk, R. Solarska, S. Mubeen, Commun. Chem. 2019, 2, 57.
J. Augustynski, Adv. Energy Mater. 2016, 6, 1600526. [44] M. A. Fetcenko, S. R. Ovshinsky, B. Reichman, K. Young, C. Fierro,
[36] M. J. N. Pourbaix, Atlas D’Equilibres Electrochimiques, Gauthier-Villars, J. Koch, A. Zallen, W. Mays, T. Ouchi, J. Power Sources 2007, 165,
Paris 1963. 544.
[37] M. Busch, N. Simic, E. Ahlberg, Phys. Chem. Chem. Phys. 2019, 21, [45] Y. Zhang, Y. Ji, W. Zhang, F. Hu, Y. Qi, D. Zhao, Appl. Surf. Sci. 2019,
19342. 494, 170.

Adv. Energy Mater. 2019, 1903213 1903213 (8 of 8) © 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

You might also like