This document is the accepted manuscript version of a published work that appeared in final form in Nature Chemistry, 2018,
10, 24-30, DOI: 10.1038/NCHEM.2874

ARTICLE
PUBLISHED ONLINE: 30 OCTOBER 2017

Polyoxometalate electrocatalysts based on earthabundant metals for efficient water oxidation in
acidic media
Marta Blasco-Ahicart, Joaquín Soriano-López, Jorge J. Carbó, Josep M. Poblet and
J. R. Galan-Mascaros
Water splitting is a promising approach to the efficient and cost-effective production of renewable fuels, but water
oxidation remains a bottleneck in its technological development because it largely relies on noble-metal catalysts. Although
inexpensive transition-metal oxides are competitive water oxidation catalysts in alkaline media, they cannot compete with
noble metals in acidic media, in which hydrogen production is easier and faster. Here, we report a water oxidation catalyst
based on earth-abundant metals that performs well in acidic conditions. Specifically, we report the enhanced catalytic
activity of insoluble salts of polyoxometalates with caesium or barium counter-cations for oxygen evolution. In particular,
the barium salt of a cobalt-phosphotungstate polyanion outperforms the state-of-the-art IrO2 catalyst even at pH < 1, with
an overpotential of 189 mV at 1 mA cm–2. In addition, we find that a carbon-paste conducting support with a hydrocarbon
binder can improve the stability of metal-oxide catalysts in acidic media by providing a hydrophobic environment.

W

ater is vastly preferred as a source for hydrogen—one of
the most promising green fuels for the establishment of
a sustainable and environmentally friendly energy
cycle1. This closed cycle would work using renewable energy to
split water into oxygen and hydrogen, and recombining them
back into water via fuel cells, for example, to extract the stored
energy. However, water-splitting technologies are still far from
being economically competitive for solar fuel production. This is
a major issue to be overcome in order to support a societal
change in the energy paradigm, in particular when very cheap
hydrogen is currently available from steam reforming2. Therefore,
tremendous research effort is being made in the field to find
viable solutions to reduce solar fuel production costs, with most
being focused on electrochemical processes. Finding cost-effective,
earth-abundant, efficient and robust catalysts for water splitting
is essential.
Recently, low-cost catalysts for the hydrogen evolution reaction
(HER) have been described for multiple working conditions and a
large pH range3–5. However, the oxygen evolution reaction (OER)
remains a more difficult problem, especially in acidic media6.
Water oxidation is a complex four-electron redox process that produces protons and oxygen, and typically requires very high overpotentials η. Earth-abundant and inexpensive water oxidation catalysts
(WOCs) that match all these requirements have been found for
alkaline or neutral media applications. However, only noble-metal
oxides such as RuO2 or IrO2 fulfil the technological requirements
in acidic conditions7, and RuO2 suffers from rapid deactivation8.
Acidic solutions appear to be preferred as media to carry out
water electrolysis. Hydrogen production is much easier at lower
pH, where the proton (as reagent) concentration is higher9,10.
Alkaline electrolysis has additional drawbacks, such as more difficult
product separation or carbonation. The lack of a low-cost alternative
WOC to IrO2 in acidic media remains a major challenge in the field.

1

Transition-metal oxides have been studied for several decades as
WOCs, with excellent results11–18. Unfortunately, all lose their
remarkable activity on lowering the pH19,20. Only cobalt oxides
appear to retain interesting activity at neutral pH, with the help of
ancillary electrolytes21,22. At higher proton concentrations, catalytic
performance is insufficient and corrosion becomes a major issue.
Under these acidic conditions, manganese oxides, for example,
support only very low current densities, and doping or electrochemical treatments are being studied to improve their stability and
performance23,24. Nanostructured cobalt oxides can sustain good
current densities, but for short periods of time due to corrosion25.
However, the latter may be incorporated into the hydrogen
production scheme to improve overall yields26. Unfortunately,
neither of these clever strategies can compete with the performance
of IrO2.
As an alternative to oxides, cobalt-containing polyoxometalates
(Co-POMs) have shown to be promising homogeneous earthabundant WOCs in close-to-neutral pH27–30. Although their activity
has been subject to close scrutiny31–33, there is no doubt that they are
genuine molecular WOCs under controlled working conditions34–37.
Interestingly, we have demonstrated that Co-POMs maintain their
catalytic activity in heterogeneous conditions when supported on
modified electrodes as insoluble caesium salts38.
Here, we disclose the excellent catalytic activity of waterinsoluble Co-POM salts in acidic media. Modified electrodes incorporating barium salts of Co-POMs promote OER during water
electrolysis experiments, with superior performance to the stateof-the-art IrO2 in sulfuric acid solution (1 M). Lower onset potentials, higher current densities at lower overpotentials and Faradaic
oxygen evolution demonstrate that barium Co-POM salts are a
real alternative to the corresponding noble-metal champions.
This is the first time that a competitive activity has been found in
a low-cost, earth-abundant material, opening unprecedented

Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology (BIST), Av. Països Catalans 16, E-43007 Tarragona,
Spain. 2 Departament de Química Física i Inorgànica, Universitat Rovira i Virgili, Marcel.lí Domingo 1, E-43007 Tarragona, Spain. 3 ICREA, Passeig Lluis
Companys, 23, E-08010 Barcelona, Spain. †These authors contributed equally to this work. *e-mail: jrgalan@iciq.es

b

c

10

8

CP (blank)
10% Cs[Co-POM]
20% Cs[Co-POM]
40% Cs[Co-POM]

6

4

2

0
0.10

0.20

0.30
η (V)

10

Current density (mA cm–2)

Current density (mA cm–2)

a

0.40

8

6

4

2

0
0.10

0.20

0.30

0.40

η (V)

Figure 1 | Electrochemical behaviour of Co-POM/CP electrodes. a, Molecular structure of the [Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]16– cluster. Co, pink;
O, red; WO6 , grey octahedra; PO4 , black tetrahedra. b, Linear sweep voltammetry in H2SO4 (1 M) solution with a Cs[Co-POM]/CP blend working electrode
for different catalyst contents. The increment in current indicates the onset of electrocatalytic water oxidation. CP: blank experiment with a carbon-paste-only
electrode. c, Same experiments with a Ba[Co-POM]/CP blend electrode, showing the lower overpotentials needed to reach even higher currents. This indicates
a better electrocatalytic energy efficiency overall, because current is proportional to oxygen evolution.

opportunities to develop technologically and economically viable
acid water electrolysis.

accessibility to active sites, electrode geometry, electronic interface
and so on. To put our Co-POM results into context, we carried
out the same experiments using an IrO2/CP blend to mimic
Results and discussion
identical working conditions. As shown in Fig. 2a, no IrO2/CP
Electrochemistry. Due to the high level of interest in acid water blend can match the catalytic activity of the Co-POM electrodes at
oxidation catalysis, we decided to test the activity, in acidic low overpotentials with the same catalyst content by weight. The Ba
media, of the caesium salt of the [Co9(H2O)6(OH)3(HPO4)2 [Co-POM]/CP blend exceeds the performance of the noble-metal
(PW9O34)3]16– cluster (Co9, Fig. 1a), which was obtained by oxide in the entire potential range studied. Lower potentials are
simple metathesis of the corresponding potassium salt as Cs15K needed to reach the same current densities (Table 1), including a
[Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]·28H2O
(Cs[Co-POM]). remarkable 246 mV difference for the onset potential of water
We blended this salt with commercial carbon paste (CP, formed oxidation and 97 mV to reach a current density of 10 mA cm–2.
from carbon black and an organic oil binder) to obtain a The superior performance of Ba[Co-POM]/CP is even more
modified electrode to be used as the anode in water electrolysis evident if we take into account the total number of possible active
experiments in H2SO4 solution (1 M), without any additional sites (Supplementary section ‘Active sites calculation’) in each
electrolyte. The three-electrode configuration was completed with modified electrode, given the different molecular weight. For
a Pt mesh as counter-electrode and a Ag/AgCl reference electrode. instance, with the same catalyst weight, IrO2/CP blends contain at
We investigated the electrochemical activity of Cs[Co-POM]/CP least five times more active sites than the corresponding Ba[Coblends using linear sweep voltammetry (LSV). The current POM]/CP blend. Thus, Ba[Co-POM] easily outperforms the
densities showed a rapid increase deviating from the flat response kinetics of the best state-of-the-art OER catalyst in acidic media, per
of the CP, indicating the appearance of catalytic currents weight and per mol, while being obtained from inexpensive starting
(Fig. 1b). We optimized the catalyst content following the LSV materials (barium, cobalt, phosphate and tungstate), representing
response. Catalytic performance shows optimum activity for the first competitive catalyst of its kind. Such lower overpotentials
blends at 30–40% content (in weight) with a minimum onset would improve the voltage efficiency of a water-splitting device by
potential of 1.42 V vs NHE (η = 196 mV). Blends with more than a large margin, maximizing energy efficiency.
50% content became too brittle for accurate measurements, so we
We also studied the Tafel plots for these catalysts (Fig. 2b), with
limited the maximum catalyst content to 40% for all further studies.
steady-state current density values in the 0.0 < η < 0.4 range. The
We carried out analogous electrochemical studies substituting Tafel slopes were 66 mV dec–1 (IrO2/CP), 98 mV dec–1 (Cs[Co–1
+
Cs with Ba2+. This heavy alkaline earth dication forms the water- POM]/CP) and 97 mV dec (Ba[Co-POM]/CP). The Tafel slope
insoluble salt Ba8[Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]·55H2O depends exclusively on the rate-determining step of an electroche(Ba[Co-POM]). To our surprise, Ba[Co-POM]/CP electrodes mical process. The IrO2/CP slope is typical of this catalyst40, indicatshowed a significantly better electrocatalytic performance. The ing chemically controlled kinetics. The Co-POM slopes are higher,
LSV data show onset potentials as low as η = 88 mV and higher indicating competition between a chemical and an electron-transfer
current densities in the potential range studied (Fig. 1c and limiting step. This is not surprising given the insulating nature of the
Supplementary Fig. 1). This huge improvement in voltage efficiency Co-POM salts when compared with the high conductivity of IrO2.
suggests a strong influence of the counter-cation in electrocatalysis Even with a higher Tafel slope, the Ba[Co-POM]/CP blends reach
kinetics, because both salts contain the same active Co-POM. The higher currents in a large potential range because of their lower
origin of this positive effect could be related to better electron trans- onset potentials. This allows an overall lower energy consumption
fer kinetics. However, without further experimental evidence, we to be sustained and thus a higher energy efficiency. Extrapolating
cannot discount that it is related to higher hydration, a different the Tafel plots (only valid if the rate-determining step of the mechanism does not change with potential), IrO2 would match the persurface area/interface or lower crystallinity.
formance of Ba[Co-POM] at overpotentials greater than 0.7 V. The
Comparative electrochemistry. It is difficult to compare OER identical slope for both Co-POMs suggests the same oxygen evolcatalytic activities39, because many parameters affect the overall ution mechanism is taking place in both cases, as expected,
performance, including particle size, active surface area, because both Co-POMs contain the same active sites.

b

10
10% IrO2
40% IrO2
40% Cs[Co-POM]
40% Ba[Co-POM]

8

66 mV dec–1

0.4

114 mV dec–1

6

η (V)

Current density (mA cm–2)

a

98 mV dec–1

0.3

Ba[Co-POM]
Cs[Co-POM]
IrO2
Co3O4
IrO2
IrOx/SrIrO3

0.2
97 mV dec–1

2
0.1

0.1

0.2

0.3

0.4

–5

0.5

–4

η (V)
15

d
40% Co3O4
20% Co3O4
10% Co3O4
40% Ba[Co-POM]
40% Cs[Co-POM]

10

j (mA cm–2) per µmol of metal

Current density (mA cm–2)

c

–3

–2

log j (log A cm–2)

5

0

1.0
30% Ba[Co-POM]
30% Cs[Co-POM]
30% IrO2
30% Co3O4

0.8

0.6

0.4

0.2

0.0
0.1

0.2

0.3

0.4

0.5

η (V)

0.0

0.1

0.2

0.3

0.4

η (V)

Figure 2 | Comparative electrochemical behaviour of catalyst/CP electrodes in acidic media, highlighting the superior activity of Co-POM modified
electrodes for oxygen evolution. Ba[Co-POM]/CP reaches higher currents at lower overpotentials in the entire range studied. When compared with state-ofthe-art nanostructured IrO2-based catalysts, Ba[Co-POM]/CP remains a competitive non-noble alternative, in particular at low overpotentials. a, Linear
sweep voltammetry in H2SO4 (1 M) solution with an IrO2/CP blend working electrode for different catalyst content, compared with Cs[Co-POM]/CP and
Ba[Co-POM]/CP electrodes. b, Tafel region for 30% catalyst/CP blends from steady-state chronopotentiometry experiments in H2SO4 (1 M) solution.
Orange and blue triangles are taken from the literature, and represent the state of the art for IrO2 and IrOx /SrIrO3 , respectively41,42. c, Linear sweep
voltammetry in H2SO4 (1 M) solution with a Co3O4/CP blend working electrode for different catalyst content, compared with Cs[Co-POM]/CP and
Ba[Co-POM]/CP electrodes. d, Normalized linear sweep voltammetry in H2SO4 (1 M) solution for 30% catalyst/CP blends with respect to the total active
metal content (Co or Ir), showcasing the superior Ba[Co-POM] activity per mol.

We need to mention that the IrO2 activity we found in our CP
blends is not comparable with state-of-the-art IrO2 preparations41.
Additionally, Jaramillo et al. reported how nanostructuring of
IrOx/SrIrO3 blends enhances their already high catalytic activity42.
Nevertheless, our Ba[Co-POM]/CP electrodes exhibit better catalytic activity at low overpotentials than the best IrO2 results available41,42 and are still competitive at moderate current densities up
to 10 mA cm–2 (Fig. 2b).
An important issue with Co-POMs is the possibility to evolve
into the corresponding CoOx species (a competitive WOC)
during water oxidation. Indeed, CoOx has been identified as the
major active phase in water oxidation catalysis when starting
with Co-POMs under certain experimental conditions31,32. We
carried out the same experiments starting with Co3O4 to gather
additional evidence on the possible participation of traces of
adventitious CoOx. We observed that the water oxidation activity
of Co3O4/CP electrodes is clearly better than what has been typically reported for this WOC in acid26. This suggests that the use
of a hydrophobic binder and an appropriate conducting support
could extend the use of classic metal oxides into acidic media.

Although Co2+ is a competent WOC20, we can rule out the participation of any homogeneous species during these experiments,
because post-electrocatalytic elemental analysis data (from inductively coupled plasma optical emission spectrometry, ICP-OES)
confirmed the absence of any metal in solution (Supplementary
Table 1).
The LSV data for the Co3O4/CP blends (Fig. 2c and Table 1) show
an electrocatalytic performance close to that of Cs[Co-POM]/CP by
weight, reaching a very low onset potential for high catalyst contents.
The Tafel plots indicate a different rate-determining step, with significantly higher slopes over 110 mV dec–1 (Fig. 2b). These oxide
blends thus cannot match the Ba[Co-POM]/CP modified electrodes
when reaching significant current densities (over 1 mA cm–2). As
described already, it is worth mentioning that these comparisons
by weight are unfair to the Co-POM catalysts. At the same weight,
the number of active sites in an oxide electrode is about one order
of magnitude higher due to its lower molecular weight and higher
density of active sites. If we normalize the current density data
taking into account the total number of Co centres, the activity of
both Co-POM/CPs is remarkably better (Fig. 2d).

NATARTICLES

cat/CP (%)

Cs[Co-POM]

Ba[Co-POM]

IrO2

90
196
305
386

ηonset (mV)
40
30
20
10

196
236
246
363

88
103
164
280

2.0

1.0
0.5
0.0
0

–2

η (mV) @ j = 1 mA cm
40
30
20
10

306
316
324
395

189
244
240
314

379
388
416
447

221
311
357
412

b

361
384
404
434

458
480
>500
>500

0.1
0.2
0.3
0.2

0.2
0.1
0.2
0.2

12

16

20

24

0.3

410
446
454
485

0.2

0.1

pH
40
30
20
10

8

0.4

j (mA cm−2)

466
>500
>500
>500

4

t (h)

η (mV) @ j = 10 mA cm–2
40
30
20
10

Ba[Co-POM]
Ba[Co-POM]
Cs[Co-POM]
IrO2
Co3O4
Co–Pi
NiFe

1.5

Co3O4

334
335
363
393

a

j (mA cm−2)

Table 1 | Comparison of linear sweep voltammetry data for
Co-POM/CP electrodes with the corresponding iridium and
cobalt oxide blends as a function of catalyst content.

0.1
0.2
0.1
0.1

0.2
0.2
0.2
0.2

14
12
9
5

19
14
8
4

0.0
8

12

16

20

24

20

24

t (h)

cat (total mg)
13
11
8
4

11
9
7
4

The exact pH of each 1 M H2SO4 solution was confirmed by direct measurements. Cat, catalyst.

electrochemical data. There is a significant difference in the LSV
data: the appearance of an additional precatalytic event in the
Co3O4 electrodes (Supplementary Fig. 1), which is absent for the
Co-POM catalysts.
Active surface area. To check if the different catalytic activity is only
due to a different active surface area in the as-prepared materials, we
carried out some additional characterization. First, we determined
the BET surface area for all four catalysts (Supplementary Fig. 2).
The results do not show significant differences, with very small
BET areas for all materials, as expected, from 2 m2 g−1 for IrO2 and
2 −1
2 −1
Cs[Co-POM], 11 m g for Ba[Co-POM] and 28 m g for Co3O4.
We also analysed the double-layer capacitance (Cdl ) of the electrodes, which is expected to be linearly proportional to the effective
active surface area (Supplementary Fig. 2). Although we cannot
extract exact values due to the unknown behaviour of the catalyst
and conducting support, we can easily estimate the relative surface
areas for all electrodes. The results show how Cdl is two orders of
magnitude higher for the Ba[Co-POM]/CP electrodes. This difference cannot be related to the BET surface area. These results indicate that the surface of the Ba[Co-POM] catalyst is intrinsically
much more active towards water oxidation, which must be caused
by the crystalline surface structure of this material and perhaps by
the presence of the Ba2+ cations in close proximity.
Oxygen evolution and post-catalytic analysis. Oxygen evolution
showed Faradaic production
(>99%) for both Co-POM/CP
electrodes, indicating a negligible contribution to the total
measured currents from any other redox process (Supplementary

30

j (mA cm–2) per mmol of metal

40
30
20
10

20

5
4
3
2
1
4

8

12

16
t (h)

Figure 3 | Stability and benchmarking of catalyst/CP electrodes.
Chronoamperometry data in H2SO4 (1 M) solution at a constant anodic
overpotential of 250 mV with 30% catalyst/CP blends. a, Over a period of
24 h, the Ba[Co-POM] exhibits superior currents and a robust response
after 4 h. The initial fast decay during the first hours is associated with a
capacitance event and is not due to catalyst fatigue. This was confirmed by
an intermittent analogous process (empty circles), where this trend is
reproduced. Arrows indicate initial time for each cycle. b, Detail of current
densities up to 24 h. c, Normalized data according to the total active metal
content (Co, Ir or Ni/Fe), highlighting the superior activity of [Co-POM] per
active site.

Fig. 3). Oxygen evolution in the case of cobalt oxide is not
Faradaic (∼60–70%, Supplementary Fig. 4), providing additional
evidence of a different electrochemical mechanism. Co3O4 is
probably partially catalysing the oxidation of the CP component.
After 2 h electrolysis at j = 1 mA cm–2, with an estimated turnover number (TON) of >10 (Supplementary section ‘TON estimation’), we collected ex situ Raman and X-ray photoelectron
spectroscopy (XPS) data (Supplementary Figs 5–7). These surfacesensitive characterization techniques are particularly useful for

detecting even traces of possible additional species generated in situ
on a catalyst surface43. Comparative analyses do not show any significant differences between fresh and used materials, suggesting
no change in the catalyst during oxygen production. The Raman
spectra are undistinguishable. The XPS spectra are also identical
at the W, P and Co edges. The O1s edge spectrum shows increased
intensities for the higher energy peaks, which can be assigned to
partial protonation of oxo sites. Thus, although we cannot rule
out the possible participation of other species, the simplest explanation at present is that the [Co-POM] structure should be close
to the true active species directly participating in the catalytic
cycle. No additional contribution can be identified in the post-catalytic characterization. Infrared and X-ray powder diffraction data
also support the bulk stability of the Co-POMs (Supplementary
Figs 8 and 9). Furthermore, we found no leaching of the active
species during working conditions, as confirmed by elemental
analysis (ICP-OES) in H2SO4 (1 M) mother liquor after the corresponding 2 h electrolysis (Supplementary Table 1). Only low contents of counter-cations
were detected for Cs[Co-POM],
corresponding to a total loss of ∼30% of the initial caesium and potassium content. We assigned this leaching to cation exchange by
protons due to the highly acidic media, in good agreement with
the XPS data. Barium leaching was not detected. We also carried
out ICP-OES analysis of several fractions of the recovered catalyst
(Supplementary Table 2). Although the carbon/binder content
makes it difficult to obtain good-quality data, the results yielded
consistent P/Co/W ratios for the used catalysts.
Long-term stability. The possibility to work at low overpotentials is
essential for an energy-efficient power-to-fuels platform, because
the extra potential needed to run the reaction represents a major
energy loss. The excellent energy efficiency offered by our Ba[CoPOM] catalyst is also matched by good long-term catalytic
stability. This was confirmed by chronoamperometric assays at
η = 250 mV (Fig. 3a). For these experiments the electrodes were
covered with a Nafion layer and with a frit to avoid fragmentation
of the CP electrodes, which are mechanically unstable and tend to
be expelled from the electrode pocket into the solution. The Ba
[Co-POM]/CP electrodes showed a very high initial current
density of >2 mA cm–2. This current slowly decreases for the first
hours to reach a stable current of ∼0.36 mA cm–2 (t = 4 h), which
is maintained for over 24 h without further significant changes
(∼0.35 mA cm–2 at t = 24 h), except those due to gas bubbles
formation and diffusion. We assigned the initial decay to charge
localization processes. We carried out an analogous experiment
including periodic 20 min stops after each hour of electrolysis. In
this case, the initial current density curve was identical in
successive cycles (Fig. 3a and Supplementary Fig. 10). This result
is not compatible with catalyst deactivation. The most plausible
explanation is an increase in the resistance of the blend due to
charge localization, what reduces the overall current density. At
open-circuit potential, charges delocalize back to the initial state.
The stable current density values demonstrated for over one day
by the Ba[Co-POM]/CP electrodes are unique among catalysts in
the same conditions for the same catalyst weight (Fig. 3b). For the
sake of comparison, we included IrO2 , Co3O4 and also two of the
most efficient catalysts in alkaline and neutral conditions:
Ni0.9Fe0.1Ox (NiFe)12 and CoPi21. Obviously, none of these catalysts
was optimized for our working conditions, but these experiments
were worth conducting to further assess the distinct behaviour of
the POMs. Only the CoPi/CP initially showed a current density
close to that of Ba[Co-POM]/CP (in the first hours), but continuous
and irreversible deactivation brought its activity down to the level of
the NiFe/CP. After 24 h, Ba[Co-POM]/CP promoted current densities three times higher than NiFe/CP and CoPi/CP, four times
higher than Co3O4/CP, and ten times higher than IrO2/CP, even

when containing a much lower total number of active sites, as discussed above. If we normalize the corresponding activity to the
total metal content, the electrocatalytic activity of Ba[Co-POM]/
CP is six times better than Cs[Co-POM]/CP and thirty times
better than any other oxide (Fig. 3c), indicating a much faster turnover frequency per active site. In terms of oxygen evolution, Ba[CoPOM]/CP produces a total of 47 TON, clearly superior to the other
catalysts: 8 (Cs[Co-POM]), 0.3 (CoPi) or 0.1 (IrO2).
At higher current densities (10 mA cm–2), Ba[Co-POM]/CP still
offers better performance (Supplementary Fig. 13), although the
instability of the CP electrodes, as mentioned above, did not allow
us to extend these studies for longer periods. Indeed, CP is suitable
for fundamental studies, but it is not stable under OER conditions.
Before the CP electrodes break down, after ∼20 min, Ba[Co-POM]/
CP is still competitive with IrO2 , suggesting the viability of this
catalyst even at relatively high currents.

Conclusion
Oxygen evolution in strongly acidic conditions with earth-abundant
catalysts is currently a major technological challenge. Our results
show the excellent and unparalleled performance of Ba[Co-POM]
for electrocatalytic water oxidation under these conditions. The significant influence of barium dications in the activity of the
[Co-POM], when compared with the Cs+ derivative, suggests that
this strategy could also be useful in enhancing the performance of
other water oxidation catalysts. POMs offer an intrinsic advantage
in this regard, because their polyanionic nature makes it easy to
selectively incorporate the desired ancillary counter-cations close
to their active sites, without structurally or chemically affecting
their appropriate environment.
We also need to highlight the non-innocent role of the CP component. Part of the better performance of the Co-POM/CP electrodes
could arise from a better matching between the CP support and the
Co-POM catalysts when compared with IrO2. Indeed, an overall
hydrophobic environment is known to enhance POM redox activity
and stability44,45. According to our data, this also happens to be true
for oxides, because the OER activity of the Co3O4/CP, CoPi/CP and
NiFe/CP electrodes is clearly superior to what has typically been
reported for these oxides in acid. This suggests that the use of an insulating hydrocarbon binder and an appropriate conducting support
could extend the use of classic metal oxides into acidic media OER
catalysis. Such corrosion-protecting strategies have already been
suggested by Schaak and co-authors25. Therefore, tuning the hydrophobic/hydrophilic character of composite electrodes may open
fascinating opportunities for low-cost acid water electrolysis.

Methods
Synthesis. Materials and detailed methods are further described in the
Supplementary Information. Na8K8[Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]·xH2O
(1) was obtained following a method described in the literature46. Cs15K
[Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]·28H2O (Cs[Co-POM]) and
Ba8[Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]·55H2O (Ba[Co-POM]) were prepared
by metathesis with an excess of CsCl or BaCl2. The presence of the
[Co9(H2O)6(OH)3(HPO4)2(PW9O34)3]16– anion in Cs[Co-POM] and Ba[Co-POM]
was confirmed by its signature infrared spectra (Supplementary Fig. 8). In POM
chemistry, the infrared bands represent a reliable fingerprint for the molecular
structure. Cation content was estimated by ICP-OES and energy-dispersive X-ray
spectroscopy (EDX). Water content was determined by thermogravimetric analysis
(TGA) (Supplementary Figs 11 and 12). Cs[Co-POM].28H2O Mw = 10,107.15.
Elemental analyses, Calc.: Cs, 19.73; K, 0.40; Co, 5.25; P, 1.53; W, 49,11. Exp.: Cs,
19.51; K, 0.52; Co, 5.40; P, 1.39; W, 49.30. Ba[Co-POM].55H2O Mw = 9659.40.
Elemental analyses, Calc.: Ba, 11.37; K, 0.00; Co, 5.49; P, 1.60; W, 51.39. Exp.: Ba,
11.51; K, <0.1; Co, 5.35; P, 1.58; W, 51.06.
Electrochemistry. Electrochemical experiments were performed with a Biologic SP150 potentiostat. Ohmic drop was compensated using the positive feedback
compensation implemented in the instrument. All experiments were performed
with a three-electrode configuration using H2SO4 (1 M) as electrolyte solution,
employing a platinum mesh counter electrode, a Ag/AgCl (KCl 3.5 M) reference
electrode and a CP working electrode (surface area = 0.07 cm2). The CP mixtures

were prepared in a mortar by mixing amorphous CP and the desired material in the
corresponding ratio by weight. This modified CP mixture was used to fill the CP
electrode. Bulk water electrolysis and LSV were performed with an ALS RRDE-3A
set-up, using a CP rotating disk electrode (surface area = 0.07 cm2) at 1,600 and
500 r.p.m., respectively. All LSV experiments were carried out with a 1 mV s–1 scan
rate. Steady-state current data for Tafel analyses were obtained after the needed
waiting time for stabilization of the current at each overpotential (at least 10 min) in
an H-cell. For long-term electrolysis (over 1 h) the CP pocket was covered with a
Nafion ink (4 μl at 5%) and with a glass frit (P0, 1 mm thick) on top to avoid
mechanical losses of the CP into the solution.
Oxygen evolution was detected with an Ocean Optics NeoFOX oxygen sensing
system equipped with a FOXY probe. The FOXY probe was calibrated with a two
point calibration, fixing 0% O2 under N2 flow and 20.9% O2 in air. The experiments
were carried out in an H-cell, with the anode and cathode compartments separated
by a porous frit. The FOXY probe was inserted into the gas space of the anodic
compartment (VGAS SPACE ≈ 4.9 ml). The solution was completely deaerated by
purging with N2 before starting the experiment, for at least 1 h. N2 flow was removed
and a base line of 10 min was recorded before starting the chronoamperometry. The
mols of O2 generated during the electrochemical experiment were calculated with
the following equation, considering ideal gas behaviour:
nO 2 =

%O2 · PTOTAL · VGAS SPACE
100 × R × T

where %O2 is given by the FOXY probe, PTOTAL is 1 atm, VGAS SPACE (litres) is
measured for each experiment, R is 0.082 (atm l/K mol) and T is 298 K. The faradaic
oxygen production curve was calculated taking into account the charge data from the
chronoamperometry experiment as described in the following equation:
nO 2 =

Q
ne · F

where Q (C) is the charge passed through the system, ne = 4 is the number of
electrons needed to generate one molecule of O2, and F is the Faraday constant
(96,485 C mol–1).
pH was measured for each experiment using an 877 Titrino Plus pH-probe
(Metrohm), calibrated weekly. The pH value was used to calculate the
thermodynamic water oxidation potential (E0 2O/O 2) using the Nerst equation:
H
E0 2 O/O2 = 1.229 − 0.059 pH(V) vs NHE at 25 °C
H
The overpotential η was calculated by subtracting the thermodynamic water
oxidation potential to the applied potential Eapp:
0
η = E app − EH2 O/O2

All potentials were converted to the NHE reference scale using ENHE = EAg/AgCl +
0.205 (V). All current densities were calculated based on the geometrical surface area
of the electrodes. Onset potentials in Table 1 were estimated from the intersection
point between the tangent lines of the Faradaic (above j = 0.2 mA cm–2) and nonFaradaic (below η = −0.1 V) currents (Supplementary Fig. 1)
Recovery of the catalysts. After electrochemical experiments, the POM/CP blend
(40 mg) was suspended in acetone (30 ml) and sonicated for 5 min. The supernatant
liquid (containing carbon black and the organic oil binder) was decanted, retaining
the POM catalyst in the beaker. This procedure was repeated 10 times to obtain a
clean catalyst sample. The supernatant fraction was also collected for additional
analysis (Supplementary Table 2).
Data availability. Additional methods and materials characterization are available
in the Supplementary Information. Supplementary figures include electrochemical
(Supplementary Figs 1, 2, 10 and 13), oxygen evolution (Supplementary Figs 3 and
4), Raman spectroscopy (Supplementary Fig. 5), XPS (Supplementary Figs 6 and 7),
infrared spectroscopy (Supplementary Fig. 8), powder X-ray diffraction
(Supplementary Figs 9 and 10) and thermogravimetry (Supplementary Figs 11 and
12) data. Supplementary tables include elemental analyses (Supplementary Tables 1
and 2) and TON estimations (Supplementary Tables 3 and 4).

References
1. McKone, J. R., Lewis, N. S. & Gray, H. B. Will solar-driven water-splitting
devices see the light of day? Chem. Mater. 26, 407–414 (2014).
2. Ursua, A., Gandia, L. M. & Sanchis, P. Hydrogen production from
water electrolysis: current status and future trends. Proc. IEEE 100,
410–426 (2012).

3. Staszak-Jirkovsky, J. et al. Design of active and stable Co-Mo-Sx chalcogels as
pH-universal catalysts for the hydrogen evolution reaction. Nat. Mater. 15,
197–203 (2016).
4. Andreiadis, E. S. et al. Molecular engineering of a cobalt based electrocatalytic
nanomaterial for H2 evolution under fully aqueous conditions. Nat. Chem. 5,
48–53 (2013).
5. Hinnemnann, B. et al. Biomimetic hydrogen evolution: MoS2 nanoparticles as
catalysts for hydrogen evolution. J. Am. Chem. Soc. 127, 5308–5309 (2005).
6. McCrory, C. C. L. et al. Benchmarking hydrogen evolving reaction and oxygen
evolving reaction electrocatalysts for solar water splitting devices. J. Am. Chem.
Soc. 137, 4347–4357 (2015).
7. Harriman, A., Pickering, I. J., Thomas, J. M. & Christensen, P. A. Metal-oxides as
heterogeneous catalysts for oxygen evolution under photochemical conditions.
J. Chem. Soc. Faraday Trans. 84, 2795–2806 (1988).
8. Sardar, K. et al. Water-splitting electrocatalysis in acid conditions using
ruthenate-iridate pyrochlores. Angew. Chem. Int. Ed. 53, 10960–10964 (2014).
9. Harriman, A. Prospects for conversion of solar energy into chemical fuels: the
concept of a solar fuels industry. Phil. Trans. R. Soc. A 371, 20110415 (2013).
10. Yu, E. H., Wang, X., Krewer, U., Li, L. & Scott, K. Direct oxidation alkaline fuel
cells: from materials to systems. Energy Environ. Sci. 5, 5668–5680 (2012).
11. Corrigan, D. A. The catalysis of the oxygen evolution reaction by iron impurities
in thin film nickel oxide electrodes. J. Electrochem. Soc. 134, 377–384 (1987).
12. Trotochaud, L., Ranney, J. K., Williams, K. N. & Boettcher, S. W. Solution-cast
metal oxide thin film electrocatalysts for oxygen evolution. J. Am. Chem. Soc.
134, 17253–17261 (2012).
13. Smith, R. D. L. et al. Photochemical route for accessing amorphous metal oxide
materials for water oxidation catalysis. Science 340, 60–63 (2013).
14. Gerken, J. B., Shaner, S. E., Massé, R. C., Porubsky, N. J. & Stahl, S. S. A survey of
diverse earth abundant oxygen evolution electrocatalysts showing enhanced
activity from Ni–Fe oxides containing a third metal. Energy Environ. Sci. 7,
2376–2382 (2014).
15. Galan-Mascaros, J. R. Water oxidation at electrodes modified with earthabundant transition-metal catalysts. ChemElectroChem 2, 37–50 (2015).
16. Suntivich, J., May, J., Gasteiger, H. A., Goodenough, J. B. & Shao-Horn, Y. A
perovskite oxide optimized for oxygen evolution catalysis from molecular orbital
principles. Science 334, 1383–1385 (2011).
17. Zhang, B. et al. Homogeneously dispersed multimetal oxygen-evolving catalysts.
Science 352, 333–337 (2016).
18. McCrory, C. C. L., Jung, S., Peters, J. C. & Jaramillo, T. F. Benchmarking
heterogeneous electrocatalysts for the oxygen evolution reaction. J. Am. Chem.
Soc. 135, 16977–16987 (2013).
19. Gorlin, M. et al. Tracking catalyst redox states and reaction dynamics in Ni–Fe
oxyhydroxide oxygen evolution reaction electrocatalysts: the role of catalyst
support and electrolyte pH. J. Am. Chem. Soc. 139, 2070–2082 (2017).
20. Gerken, J. B. et al. Electrochemical water oxidation with cobalt-based
electrocatalysts from pH 0–14: the thermodynamic basis for catalyst structure,
stability, and activity. J. Am. Chem. Soc. 133, 14431–14442 (2011).
21. Kanan, M. W. & Nocera, D. G. In situ formation of an oxygen-evolving
catalyst in neutral water containing phosphate and Co2+. Science 321,
1072–1075 (2008).
22. Surendranath, Y., Lutterman, D. A., Liu, Y. & Nocera, D. G. Nucleation, growth,
and repair of a cobalt-based oxygen evolving catalyst. J. Am. Chem. Soc. 134,
6326–6336 (2012).
23. Huynh, M., Bediako, D. K. & Nocera, D. G. A functionally stable manganese
oxide oxygen evolution catalyst in acid. J. Am. Chem. Soc. 136,
6002–6010 (2014).
24. Frydendal, R., Paoli, E. A., Chorkendorff, I., Rossmeisl, J. & Stephens, I. E. L.
Toward an active and stable catalyst for oxygen evolution in acidic media:
Ti-stabilized MnO2. Adv. Energy Mater. 5, 1500991 (2015).
25. Mondschein, J. S. et al. Crystalline cobalt oxide films for sustained
electrocatalytic oxygen evolution under strong acidic conditions. Chem. Mater.
29, 950–957 (2017).
26. Bloor, L. G., Molina, P. I., Symes, M. D. & Cronin, L. Low pH electrolytic water
splitting using earth-abundant metastable catalysts that self-assemble in situ.
J. Am. Chem. Soc. 136, 3304–3311 (2014).
27. Yin, Q. et al. A fast soluble carbon-free molecular water oxidation catalyst based
on abundant metals. Science 328, 342–345 (2010).
28. Lv, H. J. et al. An exceptionally fast homogeneous carbon-free cobalt-based
water oxidation catalyst. J. Am. Chem. Soc. 136, 9268–9271 (2014).
29. Lv, H. J. et al. Polyoxometalate water oxidation catalysts and the production of
green fuel. Chem. Soc. Rev. 41, 7572–7589 (2012).
30. Goberna-Ferron, S., Vigara, L., Soriano-Lopez, J. & Galan-Mascaros, J. R.
Identification of a nonanuclear {CoII 9} polyoxometalate cluster as a
homogeneous catalysts for water oxidation. Inorg. Chem. 51,
11707–11715 (2012).
31. Stracke, J. J. & Finke, R. C. Electrocatalytic water oxidation beginning with the
cobalt polyoxometalate [Co4(H2O)2(PW9O34)2]10–: identification of
heterogeneous CoOx as the dominant catalyst. J. Am. Chem. Soc. 133,
14872–14875 (2011).

32. Natali, M. et al. Is [Co4(H2O)2([α]-PW9O34)2]10– a genuine molecular catalyst in
photochemical water oxidation? Answers from time-resolved hole scavenging
experiments. Chem. Commun. 48, 8808–8810 (2012).
33. Stracke, J. J. & Finke, R. C. Water oxidation catalysis beginning with 2.5 μM
[Co4(H2O)2(PW9O34)2]10−: investigation of the true electrochemically driven
catalyst at ≥600 mV overpotential at a glassy carbon electrode. ACS Catal. 3,
1209–1219 (2013).
34. Vickers, J. W. et al. Differentiating homogeneous and heterogeneous water
oxidation catalysis: confirmation that [Co4(H2O)2(α-PW9O34)2]10– is a
molecular water oxidation catalyst. J. Am. Chem. Soc. 135, 14110–14118 (2013).
35. Stracke, J. J. & Finke, R. C. Water oxidation catalysis beginning with
Co4(H2O)2(PW9O34)2 10– when driven by the chemical oxidant ruthenium(III)
tris(2,2′-bipyridine): stoichiometry, kinetic, and mechanistic studies en route to
identifying the true catalyst. ACS Catal. 4, 79–89 (2014).
36. Natali, M. et al. Photo-assisted water oxidation by high-nuclearity cobalt-oxo
cores: tracing the catalyst fate during oxygen evolution turnover. Green Chem.
19, 2416–2426 (2017).
37. Goberna-Ferrón, S., Soriano-López, J., Galán-Mascarós, J. R. & Nyman, M.
Solution speciation and stability of cobalt-polyoxometalate water oxidation
catalysts by X-ray scattering. Eur. J. Inorg. Chem. 2015, 2833–2840 (2015).
38. Soriano-López, J. et al. Cobalt polyoxometalates as heterogeneous water
oxidation catalysts. Inorg. Chem. 52, 4753–4755 (2013).
39. Huynh, M., Shi, C., Billinge, S. J. L. & Nocera, D. G. Nature of activated
manganese oxide for oxygen evolution. J. Am. Chem. Soc. 137,
14887–14904 (2015).
40. Kushner-Lenhoff, M. N., Blakemore, J. D., Schley, N. D., Crabtree, R. H. &
Brudvig, G. W. Effects of aqueous buffers on electrocatalytic water oxidation
with an iridium oxide material electrodeposited in thin layers from an
organometallic precursor. Dalton Trans. 42, 3617–3622 (2013).
41. Ouattara, L., Fierro, S., Frey, O., Koudelka, M. & Comninellis, C. Electrochemical
comparison of IrO2 prepared by anodic oxidation of pure iridium and IrO2
prepared by thermal decomposition of H2IrCl6 precursor solution. J. Appl.
Electrochem. 39, 1361–1367 (2009).
42. Seitz, L. C. et al. A highly active and stable IrOx/SrIrO3 catalyst for the oxygen
evolution reaction. Science 353, 1011–1014 (2016).
43. Ahn, H. S. & Tilley, T. D. Electrocatalytic water oxidation at neutral pH by a
nanostructured Co(PO3)2 anode. Adv. Funct. Mater. 23, 227–233 (2013).

44. Carraro, M., Sandei, L., Sartorel, A., Scorrano, G. & Bonchio, M. Hybrid
polyoxotungstates as second-generation POM-based catalysts for microwaveassisted H2O2 activation. Org. Lett. 8, 3671–3674 (2006).
45. Berardi, S. et al. Polyoxometalate-based N-heterocyclic carbene (NHC)
complexes for palladium-mediated C–C coupling and chloroaryl dehalogenation
catalysis. Chem. Eur. J. 16, 10662–10666 (2010).
46. Galan-Mascaros, J. R., Gomez-Garcia, C. J., Borras, J. J. & Coronado, E. High
nuclearity magnetic clusters: magnetic properties of a nine cobalt cluster
encapsulated in a polyoxometalate [Co9(OH2)3(HO)3(HPO4)2(PW9O34)3]16–.
Adv. Mater. 6, 221–223 (1994).

Acknowledgements
This work was supported by the European Union (project ERC StG, grant CHEMCOMP,
no. 279313), the Spanish Ministerio de Economía y Competitividad (MINECO; through
projects CTQ2015-71287-R, CTQ2014-52774-P and the Severo Ochoa Excellence
Accreditation 2014-2018 SEV-2013-0319), the Generalitat de Catalunya (2014-SGR-797
and 2014SGR-199) and the CERCA Programme/Generalitat de Catalunya.
J.M.P. acknowledges the ICREA Foundation for an ICREA Academia award.
M.B.A. acknowledges the Generalitat Catalana (AGAUR) for a predoctoral fellowship.
The authors also thank Á. Reyes-Carmona for discussions.

Author contributions
J.R.G.-M. proposed the concept. J.R.G.-M., M.B.-A. and J.S.-L. designed the experiments.
M.B.-A. and J.S.-L. performed the experiments. All authors participated in data analysis
and co-wrote the manuscript.

Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Publisher’s note:
Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations. Correspondence and requests for materials should be addressed to
J.R.G.-M.

Competing financial interests
The authors declare no competing financial interests.