“This document is the Accepted Manuscript version of a Published Work that appeared in final form in
Mater. Today, copyright ©2018 Elsevier B.V.after peer review and technical editing by the publisher.
To access the final edited and published work see:

https://www.sciencedirect.com/science/article/pii/S2468606918301060

Light-driven water oxidation using hybrid photosensitizer-decorated Co 3O4
nanoparticles
Jonathan De Tovara, Nuria Romeroa, Sergey Denisovb, Roger Bofilla, Carolina GimbertSuriñachc, Diana Ciuculescu-Pradinesd, Samuel Drouetd, Antoni Llobeta,c, Pierre Lecantee,
Vincent Colliered, Zoraida Freixaf, Nathan McClenaghanb, Catherine Amiensd, Jordi GarcíaAntóna,*, Karine Philippotd,*, and Xavier Salaa,*
a

Departament de Química, Facultat de Ciències
Universitat Autònoma de Barcelona
08193 Bellaterra, Catalonia (Spain)
b

Institut des Sciences Moléculaires, Université Bordeaux / CNRS
351 Cours de la Libération
33405 Talence Cedex (France)
c

Institut Català d’Investigació Química (ICIQ)
Barcelona Institute of Science and Technology (BIST)
Av. Països Catalans 16
43007 Tarragona, Catalonia (Spain)
d

LCC-CNRS, Université de Toulouse, CNRS, UPS
205, route de Narbonne
F-31077 Toulouse (France)
E-mail:
e

CNRS, CEMES (Centre d'Elaboration de Matériaux et d'Etudes Structurales)
29 rue J. Marvig
F-31055 Toulouse (France)
f

Department of Applied Chemistry, Faculty of Chemistry
University of the Basque Country (UPV-EHU)
20080 San Sebastián (Spain) &
IKERBASQUE, Basque Foundation for Science
Bilbao (Spain)
* Corresponding authors:
E-mail adresses: Jordi.GarciaAnton@uab.es (J. García-Antón), Xavier.Sala@uab.cat (X. Sala),
Karine.Philippot@lcc-toulouse.fr (K. Philippot)
1

Keywords: Co3O4
photocatalysis

nanoparticles,

water oxidation,

2

dyad systems,

electrocatalysis,

ABSTRACT
Cobalt nanoparticles (NPs) have been prepared by hydrogenation of the organometallic complex [Co(3C8H13)(4-C8H12)] in 1-heptanol in the absence of any other stabilizer and then transformed to Co3O4 NPs using
mild oxidative reaction conditions. After deposition onto glassy carbon rotating disk electrodes, the electrocatalytic
performance of the Co3O4 NPs in water oxidation has been tested in 1M NaOH. The activity has been benchmarked
with that of state-of-the-art Co3O4 NPs through electrochemically-active surface area (ECSA) and specific current
density measurements. Furthermore, the covalent grafting of photosensitive polypyridyl-based RuII complexes
onto the surface of Co3O4 NPs afforded hybrid nanostructured materials able to photo-oxidize water into O2, while
steady-state and time-resolved spectroscopic measurements gave some further insight into kinetics and pertinent
reaction steps following excitation. These first-row transition metal oxide hybrid nanocatalysts display better
catalytic performance than simple mixtures of non-grafted photosensitizers and Co3O4 NPs, thus evidencing the
advantage of the direct coupling between the two entities for the photo-induced water oxidation reaction.
elements[12,13] have been successfully employed in
photocatalytic WO. Transition from homogeneous to
colloidal/heterogeneous species takes place
recurrently during WO catalysis[ 14 , 15 , 16 ],
particularly when first row transition metals and/or
easily oxidizable ligands are employed[17]. Despite
the high activity often shown by these in situ formed
species, their size, composition and reactivity are
poorly controlled. Therefore, the ex situ synthesis of
metal nanoparticle (NP) catalysts with wellcontrolled size and surface properties might allow a
better tuning of their catalytic performance. In this
regard, the WO photocatalytic performance of
colloidal dye (PS)-decorated IrOx suspensions has
been reported by Mallouk and co-workers[ 7,8,18].
In these hybrid dyads, the PS serves both as NP
stabilizer and light-harvester, thus helping to
overcome the thermodynamically unfavorable
barrier associated with water oxidation. Also, Frei
and collaborators have reported that the use of
electron conducting p-oligo(phenylenevinylene)
molecular wires covalently bonded to the Co3O4 core
of Co3O4/SiO2 core/shell NPs boost hole injection
from oxidized PS molecules either present in
solution[ 19,20] or electrostatically adsorbed onto the
SiO2 shell[21] to the Co3O4 core. However, to our
knowledge, no examples of WOCs are known so far
in which the PS is covalently bound to abundant first
row transition metal oxide NPs, which would reduce
the global cost of the water splitting process.
Recent literature data thus underline the interest
of a rational design of hybrid PS-NP catalysts for the
WO reaction. Within this context, the organometallic
approach, which is recognized to be highly efficient
to prepare NPs with high control over the size,
composition and surface properties[22], may offer
new opportunities in the synthesis of hybrid PS-NP
materials with covalent bonding between the PS and
the NP surface. In this way, well-controlled hybrid
catalytic materials would be accessible. Eventually,
they would permit studies of the influence of the
direct link between the PS and the NP surface on
their reactivity, adjusting the dyads to reach better
catalytic performances. With this idea in mind, we
decided to investigate the grafting of polypyridyl-

1. Introduction
Non-renewable fossil fuels are still nowadays the
main energy source used by mankind. However,
their fast depletion due to the constant increase of the
global energy consumption and their relationship
with worrying levels of greenhouse gases and
climate change make the development of less
polluting and renewable energy sources a central
topic for the scientific community. In this context,
the development of sunlight-driven water-splitting
procedures is gaining momentum[ 1 ]. The water
splitting reaction, in which both oxygen and
hydrogen gas are generated in the anode and cathode,
respectively, represents an attractive method for
obtaining energy in the form of the highly energetic
H–H chemical bond[2], as long as the energy used to
produce the hydrogen gas is renewable.
Nevertheless, the anodic oxidation of water into O2
is a thermodynamically uphill, mechanistically
complex and kinetically slow process, in which four
electrons have to be removed from two water
molecules and an O=O double bond has to be
formed[3]. With the aim of improving the kinetics of
this half reaction and inspired by Nature’s
photosystem II, which is responsible for oxygen
formation during photosynthesis, a range of
photocatalytic systems has been developed during
the past years[4,5]. Thus, several homogeneous and
heterogeneous water oxidation catalysts (WOCs)
have been tested in the presence of a photosensitizer
(PS), a chromophore able to harvest photons and
harness their energy to transfer electrons[ 6 ].
However, the catalytic performance of these systems
is often limited by the insufficient rate of electron
transfer from the WOC to the PS and the undesired
back-electron transfer phenomena between the PS
and the WOC[6,7]. In this context, covalently-bound
molecular PS-WOC dyad systems have proven to be
more efficient in photocatalytic WO since the rate of
electron transfer from the WOC to the PS is
significantly faster[8,9]. Thus, molecular PS-WOC
dyad systems based on second and third row
transition elements[ 9,10,11] and abundant first row

3

RuII complexes acting as the PS at the surface of
Co3O4 NPs prepared by organometallic chemistry as
a strategy to obtain advanced WO photocatalysts.
Therefore, herein we report the synthesis and full
characterization of novel hybrid PS-NP materials
obtained by covalently anchoring light-harvesting
[Ru(bpy)3]2+
derived-complexes
(bpy = 2,2’bipyridine) to preformed abundant first row cobalt
oxide NPs (Co3O4 NPs), and their use as catalysts in
the photoinduced oxidation of water. Their catalytic
performance is compared with that of colloidal
mixtures of non-bonded PS and Co3O4 NPs, and
Co3O4 NPs alone. Finally, flash photolysis/transient
absorption methods as well as detailed
photochemical analyses are used to gain further
insight into the specific processes and rates
associated with each component of the catalytic
system and thoroughly discussed.

controlled surface chemistry, the reducing agent and
metal precursor should be carefully chosen. One of
the best precursors in this regard is the
organometallic complex [Co(3-C8H13)(4-C8H12)],
which upon hydrogenation releases only cobalt
atoms and cyclooctane, a non-coordinating
molecule[ 24 ]. This synthetic strategy avoids
competition between the stabilizing agent introduced
and reaction by-products observed in other cases.
The NPs produced by this way display high
reactivity towards air[ 25 , 26 ], even at low
temperature, which is a prerequisite to keep the
morphological parameters unchanged during the
oxidation process and reach Co3O4 NPs of controlled
average size and size distribution. To facilitate the
eventual grafting of the polypyridyl RuII complexes
on the surface of the NPs, the use of strongly
coordinating stabilizers like carboxylic acids should
be avoided. For this reason and based on our former
work on Ru NPs, 1-heptanol was chosen as solvent
and stabilizer for the preparation of Co NPs[27].
Briefly, the synthesis of Co NPs was performed by
decomposition of a 1-heptanol solution of the
organometallic precursor [Co(3-C8H13)(4-C8H12)]
under 3 bar of H2 at room temperature. TEM
(Transmission Electron Microscopy) analysis
carried out from the so-obtained crude dark-brown
colloidal solution revealed the formation of a
monodisperse population of spherical NPs of mean
size 3.0 ± 0.1 nm (Fig. 1). These particles are welldispersed on the TEM grid and display a narrow size
distribution with a standard deviation below 5% of
the mean size.

2. Results and Discussion
2.1. Synthesis and characterization of the
nanomaterials
Since the direct synthesis of ultrafine Co3O4 NPs
is a challenging process[23], we designed a threestep strategy to obtain the target PS-Co3O4 NPs
hybrids: 1) synthesis of Co NPs with an average size
of a few nanometers and narrow size distribution; 2)
oxidation of these NPs into the target Co3O4
nanomaterial while keeping their morphological
features unchanged, and 3) grafting of RuIIpolypyridyl complexes with anchoring groups on the
surface of the Co3O4 NPs. A literature survey
revealed that to reach Co NPs with an average size
of a few nanometers, narrow size distribution and

4

Fig. 1. From left to right, synthesis and TEM images of Co, Co3O4 and PS-Co3O4 NPs (PS in the example shown
is PS1). Bottom left: Schematic drawing of [Ru(bpy)3]2+ (PS0) and modified [Ru(bpy)3]2+ complexes with 2 or 4
phosphonic acid coordinating pending groups (PS1 and PS2, respectively) used as photosensitizers.
heptanol, indicating the absence of extended
aggregation during the oxidation process. The
Recovery of the NPs from the 1-heptanol
average size of the final NPs was estimated to be 3.0
colloidal solution was performed by application of a
± 0.2 nm (Fig. 1). Taking into account the change in
magnet on the reactor walls (magnetic filtration).
material density (8.9 g·cm-3 for bulk Co compared to
Then, successive washings with pentane followed by
6.11 g·cm-3 for bulk Co3O4) one could expect the
drying of the obtained solid under vacuum afforded
volume of the NP to double upon oxidation, leading
Co NPs under the form of a black fine powder, which
to an expected diameter value of around 3.9 nm. This
was used for further characterizations. WAXS
apparent absence of volume change was already
(Wide-Angle X-Ray Scattering) analysis of the
reported for the study of the mild oxidation of Co
sample confirmed the presence of Co NPs in the α
NPs embedded in a polymer matrix[25]. This can be
(hcp) and ε (metastable cubic) crystalline structures
related to the lower density of the Co oxide NPs,
as evidenced by the good match observed between
which makes the determination of their exact
experimental data and the combination of reference
diameter from TEM images less precise and may
patterns (PDF 01-080-6668 and PDF 04-017-5578,
induce its systematic underestimation. Oxidation of
respectively, Fig. 2) and with a coherence length of
the CoNPs into Co3O4 could be confirmed by
ca. 2.5 nm (Fig. S1). The Co content in the sample
WAXS, XPS (X-Ray Photoelectron Spectroscopy)
determined by ICP-OES (35.35%) suggests an
analyses and by IR spectroscopy.
empirical
formula
(1-heptanol)0.9Co1
(see
Supporting Information) which points to more than
WAXS measurements evidenced the presence of
one 1-heptanol molecule per surface Co atom,
nanoparticles displaying a coherence length of ca.
suggesting the formation of interacting multilayers
3.0 nm (Fig. S1) and a crystalline structure
around the NPs.
corresponding to the Co3O4 phase as shown by the
good match observed between experimental data and
the reference pattern (PDF 04-005-4386) (Fig. 2).
The perfect agreement between the coherence
length and the mean size determined by TEM points
at first sight towards well crystallized nanoparticles.
This is surprising given the mild conditions used
during the oxidation of the Co NPs and would rather
suggest that the diameter determined from the
analysis of the TEM images is underestimated. In
XPS (Fig. S2), the Co3O4 NPs show two main peaks
at ca. 780.0 and 795.5 eV, corresponding to the Co
2p3/2 and Co 2p1/2 components, respectively,
accompanied by two broad satellite peaks with very
low intensity at higher energies (ca. 788 and 805
eV). According to the literature data, the presence of
satellite peaks are an indication of the presence of
unpaired electrons in the sample, i.e. the presence of
CoII (d7) atoms[28], while both the positions and the
intensities of all 4 types of bands observed in the
XPS spectrum clearly match the data already
reported for the mixed CoII-CoIII oxide
Co3O4[28,29].
IR spectroscopy revealed the presence of typical
Fig. 2. WAXS analysis of Co NPs (top) and Co3O4
Co-O stretching bands centered at 666 and 575 cm-1
NPs (bottom) in comparison with Co  / Co  and
[30] within the Co3O4 structure (green line, Fig. 3)
Co3O4 phase diagrams for Co NPs and Co3O4 NPs,
confirming the crystallinity and structure of the
respectively.
nanomaterial. Poorly defined absorptions in the
1600-1400 cm-1 region could correspond to adsorbed
Complete conversion of the Co NPs previously
water and carbonate molecules. All together these
described into Co3O4 NPs was achieved in soft
results show the transformation of preformed Co
reaction conditions by simple exposure of the
NPs into crystalline Co3O4 NPs while preventing
particles in the solid state under ambient air for 6
aggregation and coalescence of the NPs thanks to the
days. The NPs are well dispersed on the TEM images
use of soft reaction conditions.
recorded after dissolution of the powder in 1-

5

the 1465-1394 cm-1 region, indicative of the
presence of the bipyridine backbone of the PS. No
free P=O nor P-OH bands are visible, but there is a
band at 1061 cm-1 that can be attributed to -P(O-Co)
functionalization[35] (by comparison with the 1016
cm-1 value found for -P(O-Zn)3), thus supporting the
grafting of PS1 to the surface of Co3O4 NPs and the
use of all possible anchoring points. Additionally,
neither -CH3 nor -CH2- stretching bands are clearly
observed at ca. 2900 cm-1 (such as those seen for
Co3O4 NPs (red spectrum) in Fig. 3), indicating that
a partial replacement of the 1-heptanol molecules by
the PS1 molecules at the NP surface may have taken
place and/or that part of the 1-heptanol initially
present could have been released from the NPs
surface during dialysis. Therefore, we cannot
exclude the possible presence of residual 1-heptanol
molecules on the surface of the hybrid NPs. TEM
analysis from the aqueous colloidal dispersion of the
hybrid material revealed the presence of nanoobjects with homogeneous shape (spherical) that
display a mean size of 3.1 ± 0.3 nm (Fig. 1). The
results show that the initial size and morphology of
the Co3O4 NPs are maintained after anchoring the
PS1 at their surface. The presence of Ru only in the
vicinity of the NPs was evidenced through STEMEDX analysis, indicating a successful purification
process (Fig. S4). The PS2-Co3O4 NP hybrid
material was characterized accordingly, with very
similar results to the PS1-Co3O4 hybrid material
(Fig. S3, Fig. S5 and Fig. S6). Based on ICP-OES
analysis, incorporation of 19 PS1 and 32 PS2
complexes per Co3O4 NP is estimated (see
Supporting Information for further details). The
higher grafting density obtained in the case of the
PS2-Co3O4 hybrid could be related to the statistically
more favored interaction between the surface and the
chelating biphosphonate bipyridine ligands.
To sum-up this section, easily obtained Co NPs
could be transformed into Co3O4 NPs in mild
reaction conditions while preserving their
morphology. These results highlight: 1) the
efficiency of the synthetic approach, where 1heptanol acts as both solvent and stabilizer for the
preparation of small and size-controlled Co NPs; 2)
the possibility to access small Co3O4 NPs by a soft
oxidative treatment with a preserved morphology,
and 3) the possibility to graft RuII-polypyridyl
complexes at the Co3O4 NPs surface, despite the
coverage of the surface by 1-heptanol molecules, to
get hybrid nanostructured materials with multiple
functionalities, namely PS and catalyst. The catalytic
properties of the obtained Co3O4 NPs and hybrid PSCo3O4 NPs materials are described hereafter.
2.2. Electrocatalytic behavior in water oxidation
catalysis
The electrocatalytic performance of the prepared
ultrafine Co3O4 NPs was studied in 1M NaOH
solution after their deposition onto a glassy carbon

Fig. 3. Overlay of ATR-IR spectra of Co3O4 NPs
(red), free PS1 (black) and PS1-Co3O4 NPs hybrid
material (blue) and of IR spectrum of a KBr pellet of
Co3O4 NPs (green). The transmittance of each
sample has been shifted along the y-axis for
comparison purposes. The positions of the bands
described along the text have been added as dotted
lines.
The next step was the building of the hybrid
nanomaterials (PS-Co3O4 NPs) by grafting RuIIpolypyridyl complexes at the surface of the Co3O4
NPs. For this purpose, the [Ru(bpy)3]2+ complexes
bearing phosphonic acid pending groups (PS1 with
two and PS2 with four) were prepared following
literature procedures[31,32] (Fig. 1).
The phosphonic acid anchoring group is well
known to efficiently interact with metal oxide
surfaces[ 33 ]. The PS grafting was performed by
mixing a 1-heptanol colloidal dispersion of Co3O4
NPs with a methanol solution of the chosen complex
([RuII complex]/[Co3O4 NPs] = 0.24) and leaving the
obtained reaction mixture under vigorous stirring in
the dark for 4 days. The nanohybrid materials were
recovered by centrifugation and further purified
from unreacted Ru complexes by dialysis against
deionized water. This procedure was applied for PS1
and PS2 complexes giving rise to the PS1-Co3O4 and
PS2-Co3O4 hybrids, which could be recovered as
black powders after evaporation of water under
vacuum. From ICP-OES analysis, it can be inferred
that
these
nanomaterials
show
(1heptanol)0.80(PS1)0.09-Co3O4
and
(1heptanol)0.97(PS2)0.15-Co3O4 empirical formulas (see
Supporting Information). The attachment of PS1 and
PS2 complexes at the surface of the Co3O4 NPs was
attested by ATR-IR spectroscopy (Fig. 3 and Fig. S3,
respectively).
As shown in Fig. 3 (black spectrum) the PS1
complex shows two absorption bands at 1149 and
928 cm-1 corresponding to the free P=O and P-OH
units, respectively[34,35]. The PS1-Co3O4 hybrid
nanomaterial (Fig. 3, blue spectrum) shows bands in

6

rotating disk electrode (GC-RDE, Fig. S7). The
electrode was prepared by depositing three 5 L
drops from a THF dispersion (1 mg of Co3O4 NPs in
250 L of THF) of the NPs onto the glassy carbon
disk of the RDE. Fig. S7a shows the rotating disk
voltammetry (RDV) of the as-deposited Co3O4 NPs,
where a steep increase in intensity above 0.7 V vs.
NHE is visible. This increase in intensity is
attributed to the oxidation of water into molecular
oxygen[36,37,38], which in our case happens at an
onset overpotential () of ca. 0.29 V.
The electrocatalytic performance of our Co3O4
NPs deposited onto GC-RDE was compared with
that of other electrocatalysts following the
benchmarking methodology reported by Jaramillo et
al.[36,37,38]. Thus, the electrochemically-active
surface area (ECSA) of the Co3O4 modified GCRDE was estimated to be 0.175 cm2 from the
electrochemical double-layer capacitance (Cdl) by
measuring the non-Faradaic capacitive current
associated with double-layer charging from the scanrate dependence of cyclic voltammograms
(CVs)[ 39 , 40 ] over a 0.145-0.245 V vs. NHE
potential range (Fig. S9). The roughness factor (RF)
was calculated by dividing the estimated ECSA by
the geometric area of the electrode, and these factors
as well as those corresponding to the electrocatalytic
activity of our nanocatalyst are compared with those
of the state-of-the-art Co3O4 NPs in the same
electrolyte in Table 1. However, it is important to
note that the ECSA serves only as an approximate
guide for the determination of the RF, since the

accuracy of the data lies normally within an order of
magnitude[37]. Therefore, comparison with
literature data can only be analyzed in terms of
general trends. As depicted in Table 1, the Co3O4
modified GC-RDE catalyst reported herein (entry 1)
shows an onset overpotential (ηonset) of ca. 0.29 V vs.
NHE, which lies close to the reported values for
Co3O4 NPs in graphene[41,42] (entries 3 and 6) and
single walled carbon nanotubes (SWCNTs)[ 43 ]
(entry 7). To achieve a current density of 10 mA·cm2
, approximately the current density expected for a
10%
efficient
solar-to-fuel
conversion
device[37, 44 , 45 , 46 ], a η10mA/cm2 of 0.486 V is
required (entry 1, this work). This value is close to
the η10mA/cm2 reported for Co3O4 NPs of similar RF
by Jaramillo et al.[38] (entry 4) but higher than the
η10mA/cm2 values published for catalytic systems with
nearly two orders of magnitude higher RF[47,48]
(entries 2 and 5). Even so, it falls within the reported
area of interest for catalyst benchmarking[36,37,38].
Most interestingly, when the current at an η of 0.35
V (jg) is normalized by the ECSA[ 49 , 50 ], a
normalized current density (js) of 1.04 mA cm-2 is
obtained. This value is significantly higher than that
reported by Jaramillo et al. for their Co3O4-based
nanocatalyst of similar RF and η10mA/cm2 (entry 4) and
also for all other nanostructured metal oxide
electrocatalysts deposited onto GC-RDE[38]. Thus,
the Co3O4 modified GC-RDE reported herein is very
active at low η (0.35 V). The system becomes less
competitive at higher overpotentials due to the
relatively high slope of the Tafel plot above η = 0.35
V (approx. 100 mV·dec-1, Fig. S7b).

Table 1. Benchmarking parameters vs. NHE for Co3O4 NPs in 1 M NaOH and comparison with state-of-the-art
data in the same electrolyte.
Entry

Catalyst

ECSA
/cm2

RF

ηonset/V

η10mA/cm2, η10mA/cm2,
t =0/V
t=1h/V

jg,η=0.35 V
/mA·cm-2

js,η=0.35 V
/mA·cm-2

 (%)

Ref.

1

3.0 nm Co3O4 NPsa

0.175

2.5

0.29

0.486h

0.480h

2.6

1.04

95g

this work

2

10 nm Co3O4 NPsb

429

429

---

0.290

---

ca. 25

ca. 0.06

95

47

3

ca. 10 nm Co3O4 NPsc

---

---

0.24

0.313

0.313

ca. 30

---

---

41

4

ca. 70 nm Co3O4 NPsd

1.52

7.8

---

0.50

0.51

0.06

0.039

---

38

5

50 nm Co3O4 nanocubese

6.44

91.2 ---

0.28

---

ca. 70

ca. 0.77

---

48

6

4-8 nm Co3O4 NPsf

---

---

ca. 0.27 0.31

---

ca. 35

---

---

42

7

6 nm Co3O4 NPsg

---

---

ca. 0.28 0.593

---

ca. 1.8

---

---

43

a

Using a GC-RDE; b deposited on Ni foam; c in graphene nanocomposite; d deposited onto GC-RDE; e In N-doped
graphene nanocomposite; f in N-doped graphene deposited onto Ni foam; g in SWCNTs; h deposited onto an FTO
electrode.
FTO electrode was loaded with 15 L of a dispersion
of Co3O4 NPs (2 mg in 500 L of 1-heptanol) by
spin-coating, and afterwards a poly(methyl

Both stability and Faradaic efficiency () are key
parameters for a catalyst to be suitable for practical
applications in WO catalysis. To analyze them, an

7

*PSx+ + S2O82-  PS(x+1)+ + SO42- + SO4-. (Equation
2)

methacrylate) (PMMA) layer was added as “gluing”
material[ 51 ]. The generated FTO/Co3O4NPsPMMA electrode showed onset  of ca. 0.29 V (see
Fig. S10b), identical to that of the Co3O4 NPs on GCRDE (see Table 1), highlighting the negligible effect
of the support used. The FTO/Co3O4NPs-PMMA
electrode was then held at a constant current density
of 10 mA·cm-2 in a current-controlled experiment for
1h in 1M NaOH. As shown in Fig. S10a, the system
showed a stable operating potential, changing
negligibly from 10mA/cm2, t=0 h = 0.486 V to 10mA/cm2,
t=1 h = 0.480 V during 1 h (entry 1, Table 1).
Comparison of the CV polarization curves measured
before and after this current-controlled experiment
shows a slight increase in the observed current
density after catalytic turnover (Fig. S10b). This
global increase in current density is indicative of a
certain activation of the Co ions in the Co3O4 NPs.
An increase in the Co+3/+4/Co+2 population ratio, as
suggested by Frei and co-workers[52], and / or the
elimination of 1-heptanol molecules present at the
surface of the NPs increasing the number of exposed
active sites are plausible reasons for this behavior.
XPS analysis of the resulting electrode after
electrolysis shows the intact composition of the NPs
as Co3O4 (Fig. S11), thus pointing to the removal of
1-heptanol molecules from their surface under
catalytic conditions as the origin of the observed
activation process. Furthermore, the activation of the
NPs does not significantly affect the Tafel slopes
(Fig. S10c), which are very similar before and after
catalytic turnover. In addition, a Faradaic efficiency
of 95% was determined by quantifying the amount
of O2 generated during a bulk electrolysis (0.886 V
vs. NHE corresponding to an initial current density
of 10 mA·cm-2) using an O2-probe and dividing it by
the theoretical O2 amount calculated from the total
charge passed through the system (Fig. S12), thus
confirming the production of O2 as the only reaction
taking place.

PSx+ + SO4-.  PS(x+1)+ + SO42- (Equation 3)
The influence of PS concentration on the
turnover number (TON) has been studied with PS1
in
order
to
determine
the
best
[PS]/[Co3O4(heptanol)2.8 units] ratio to be used (Fig.
S16). The curve (TON vs. PS1 equiv) obtained
reaches a plateau at ca. 6.0 equiv of PS.
Consequently, a PS:Co3O4(heptanol)2.8 units ratio of
6.0:1.0 was chosen for the following photochemical
WO studies. As reported in Table 2 (entries 1-3), all
the tested PS (Fig. 1) could oxidize the Co3O4 NPs
under catalytic conditions. These results are
consistent with the electrochemical analysis of the
Co3O4 NPs shown above, since their onset potential
at pH 5.6 (ca. 1.1 V vs. NHE, Fig. S17) is lower than
that of the RuIII/RuII redox couple for all PS tested
(1.2-1.3 V vs. NHE, Fig. S18). TON and TOF
(turnover frequency, min-1) values were determined
from the estimated total number of NPs and of PS
molecules present on each sample (as presented in
the Supporting Information, these values are
underestimated; yet relative values can be obtained).
Thus, the TON per NP obtained for the catalytic
mixtures made of Co3O4 NPs plus PS0, PS1 or PS2
are all similar and within the range 453-604 (entries
1, 2, 3 in Table 2). Concerning the TOF per NP, it is
reduced by approximately half when doubling the
number of phosphonate groups present (67.9 vs. 36.2
min-1 for PS1 and PS2, respectively; entry 2 vs. entry
3). The TOF per PS is also reduced by a similar
amount (0.053 vs. 0.028 min-1 for PS1 and PS2,
respectively). One potential reason for the reduced
TOF values for the PS2 case could be that the
presence of four phosphonate binding groups could
allow the simultaneous binding of a single PS2 unit
onto two Co3O4 NPs, thus favoring their aggregation,
in contrast to the PS1 system. This hypothesis is
confirmed when comparing the TEM images of the
as-synthesized hybrid PS1-Co3O4 and PS2-Co3O4
systems (Fig. 1 and Fig. S5, respectively), where
higher aggregation and lower dispersion of the NPs
are obtained for PS2-Co3O4. At this point, it is also
worth mentioning that the limiting factors that may
stop the O2 evolution in such photocatalytic systems
are usually the degradation of the PS and the pH
decrease due to the release of protons during the
reaction[53,54,55]. However, it is interesting to note
that O2 evolution was not resumed under our
conditions after the addition of an extra aliquot of
PS. Thus, some kind of inhibition of the whole
photocatalytic system and not only PS degradation
takes place. This inhibition could be caused by the
progressive increase in the ionic strength of the
medium (sulfate ions are produced during catalysis),
as this can negatively affect the performance of the
NPs due to a reduction of the quenching efficiency
of the photoexcited PS (*PSx+) by peroxodisulfate

2.3. Photochemical water oxidation catalysis
The efficiency of the Co3O4 NPs as a
photocatalyst for WO was first evaluated in Na2SiF6NaHCO3 (0.02-0.04 M, pH 5.60) in the presence of
PS0 ([Ru(bpy)3]2+) and its phosphonate derivatives
PS1 or PS2 as photosensitizers (Fig. 1), using
sodium peroxodisulfate as the sacrificial electronacceptor (SEA) and liquid phase Hansatech-type
microsensors for measuring the evolved oxygen
(Fig. S13, Fig. S14 and Fig. S15). It is noteworthy
that the semiconducting Co3O4 NPs alone do not
behave as photocatalysts for this reaction. The
photoactive species responsible for the activation of
the Co3O4 NPs in WO is PS(x+1)+, which is generated
after a three-step process as follows:
PSx+ + h  *PSx+

(Equation 1)

8

(Equation 2)[56]. In addition, another phenomenon
that typically explains the reduced performance of
nanocatalysts with time in the presence of the
NaSiF6-NaHCO3 buffer is the hydrolysis of the
buffer to generate SiO2 particles, which can provoke
the adsorption of the cationic PS molecules onto
their surface, thus competing with the catalytic NPs
and reducing the global catalytic performance[18].

This last hypothesis has been confirmed by HREM
(High Resolution TEM) and STEM (Scanning TEM)
analyses of the recovered nanocatalyst after
photocatalytic turnover in the presence of PS1, in
which Co3O4 NP aggregates of ca. 50 nm attached to
a bigger Si-containing aggregate are observed (Fig.
S19).

Table 2. TON and TOF (min-1) per NP and per PS obtained as a function of PS nature in photochemical WO
measurements with single Co3O4 NPs and hybrid PS-Co3O4 NPs at pH 5.6.
Entry

System

PS:Co3O4 ratio

TON (O2/NP)

TOF min-1 (O2/NP)

TON (O2/PS)

TOF min-1 (O2/PS)

1

Co3O4 + PS0

6.0:1.0

453

49.4

0.35

0.038

2

Co3O4 + PS1

6.0:1.0

566

67.9

0.44

0.053

3

Co3O4 + PS2

6.0:1.0

604

36.2

0.47

0.028

4

Co3O4 + PS1

0.09:1.0

<1

-

< 0.1

-

5

Co3O4 + PS2

0.15:1.0

<1

-

< 0.1

-

6

PS1-Co3O4

0.09:1.0

5.4

0.90

0.28

0.046

7

PS2-Co3O4

0.15:1.0

82.0

2.05

2.53

0.063

(9.66·10-4 M) in the presence of PS1 (1.23·10-4 M)
and Na2S2O8 (7.85·10-2 M). Khaki line: Co3O4 NPs
(9.95·10-4 M) in the presence of PS2 (1.05·10-4 M)
and Na2S2O8 (8.08·10-2 M). Blue line: PS1-Co3O4
NPs (1.02·10-3 M) in the presence of Na2S2O8
(7.32·10-2 M). Purple line: PS2-Co3O4 NPs (1.09·103
M) in the presence of Na2S2O8 (7.87·10-2 M). All
measurements were performed in Na2SiF6-NaHCO3
(0.02-0.04 M, pH 5.60) buffer solution. Irradiation
provided by a Xe lamp equipped with a 400 nm cutoff filter and calibrated to 1 sun (100 mW cm-2). T =
25°C.
As shown in Table 2, the oxygen evolved by both
unbound systems is almost negligible (entries 4 and
5 and Fig. 4). This can be attributed to the kinetic
prevalence of the deactivation processes described
above (PS degradation, NP aggregation) competing
with oxygen evolution when very low concentrations
of unbound PS are used. Conversely, the same
PS / Co3O4 ratio is rather more active when bound
PS-Co3O4 hybrid systems are employed (entries 6
and 7). These results could be expected in part if
taking into account that in PS-NP dyad systems the
electron transfer between both entities is obviously
not limited by diffusion, as for the unbound systems.
Furthermore, comparison of entries 6 and 7 in Table
2 shows the rather superior activity of PS2-Co3O4
(TON and TOF per NP of 82 and 2.05 min-1,
respectively) versus PS1-Co3O4 (TON and TOF per
NP of 5.4 and 0.90 min-1, respectively). Thus, the
superior PS surface functionalization in PS2-Co3O4

Concerning the PS1-Co3O4 and PS2-Co3O4
hybrid materials, the number of PS per Co3O4 NP
unit were estimated to be 0.09 and 0.15, respectively
(see Supporting Information). Thus, in order to
compare the photocatalytic performance of these
hybrid dyads with that of the corresponding unbound
systems, similar PS / Co3O4 ratios were applied
under catalytic conditions for the two control
experiments (Fig. 4). Note that for experiments in
which little oxygen is evolved, we observe a
decrease of the signal inside the chamber when
exposed to light due to the reaction of the singlet
state of PS with residual oxygen traces[57].

Fig. 4. Photocatalytic oxygen production by
different Co3O4-PS systems. Green line: Co3O4 NPs

9

(32 PS molecules per NP vs. the 19 molecules
present in PS1-Co3O4) enhances the kinetics of
oxygen evolution and better stabilizes the catalytic
system, increasing its durability under photocatalytic
conditions (purple line, Fig. 4). When the kinetics of
oxygen evolution are normalized by the PS
concentration, TOF values of both hybrid systems
get closer (rightmost column in Table 2, entries 6 and
7), thus confirming the relationship between the rate
of photocatalytic turnover and the degree of
functionalization of the NPs surface. In terms of
stability, the weakness of P-O-M bonds has been
extensively identified as a main deactivation
pathway of grafted molecular complexes and dyesensitized systems when employed as catalysts for
the oxidation of water[34,35,58]. Thus, the higher
number of anchoring groups present in PS2 (4
phosphonate groups vs. the 2 present in PS1) can
also contribute to the superior longevity of the PS2Co3O4 hybrid system.
To further study the fate of the hybrid dyad
nanocatalysts under photocatalytic WO conditions,
the crude reaction mixture after a 1h photocatalytic
test with PS1-Co3O4 was dialyzed against 2 L of
deionized water for 4 days followed by
centrifugation and air-drying. Some partial
aggregation was observed by TEM analysis (Fig.
S20), although less intense than for the non-anchored
system (Fig. S19), thus confirming the above
proposed protective role of the attached PS1
molecules against aggregation. Similar studies with
Co3O4 NPs stabilized by phosphonate-derived
ligands have also shown the preservation of the NP
size along light driven WO catalysis[ 59]. On the
other hand, IR spectroscopy showed the loss of PS1
(Fig. S21), since the intensity of the bpy bands at
1400-1500 cm-1 and the -P(O-Co) bands at ca. 1060
cm-1 decreases after photocatalysis. Also, ICP-OES
analyses evidence the decrease in the [Ru] / [Co]
ratio from 0.03 to 0.003 after catalysis. This loss in
PS can be not only due to its decomposition, which
is kinetically competitive with the oxidation of
water[55], but also to its decoordination from the
surface of the Co3O4 NPs, as commonly observed in
related systems[34,35,58].
In summary, the results shown in this section
highlight the benefits of the dyad approach where the
direct connection between the Co3O4 nanocatalyst
and the PS facilitates electron-transfer and stabilizes
the system against aggregation under turnover
conditions due to the protective effect of the PS at
the surface of the Co3O4 NPs. However, they also
emphasize the relative instability of the -P(O-Co)
bonds under turnover conditions and the need of
further research for developing more stable systems
with higher durability.

In order to rationalize the kinetics of light
induced oxygen evolution described above, and the
precise role and contribution of each component in
the catalytic systems, a series of steady-state and
time-resolved measurements were undertaken.
Indeed, photo-oxidation of different PS upon
addition of the electron acceptor peroxodisulfate,
according to Equation 2, was investigated by steadystate MLCT luminescence quenching and emission
lifetime changes of the inherently luminescent PS in
Na2SiF6-NaHCO3 0.02-0.04 M, pH 5.60 (Fig. S22,
see Supporting Information for experimental
details). Information on the nature of the quenching
processes (i.e. static vs dynamic) was sought via
Stern-Volmer (SV) plots, I0/I vs. [S2O82-] (where I0
and I are the emission intensity of the excited PS
(*PS) in the absence and presence of the quencher,
respectively) for all PS (alone and attached to Co3O4
NPs), and are shown in Fig. S23.
The SV plots for the PS1-Co3O4 and PS2-Co3O4
hybrid systems show a linear trend, while for the free
PS the SV plots significantly deviate from linearity.
These results are similar to those previously reported
by Musaev[60] and Bard[61] for [Ru(bpy)3]2+ (PS0
in our work) and the same quencher in a different
electrolyte. They showed that the SV plot is
described by a model that takes into account the
formation of ground-state ion pairs between the
emitter and the quencher. In this case, two different
quenching processes can occur: collisional or
dynamic quenching (bimolecular pathway) or a
static
or
complex
formation
quenching
(unimolecular pathway)[62,63]. Dynamic quenching
occurs when the excited photosensitizer collides
with the quencher, following the conventional SternVolmer behavior. However, in some cases the
photosensitizer can initially form a stable ion-pair
complex with the quencher, followed by a
photoexcitation of the whole system (Scheme S1).
The model that considers both quenching processes
at the same time gives rise to the following equation
for quenching emission[61]:
=

(

)(
(

)
)(

(

)

(Equation 4)

)

where Keq is the equilibrium constant of ion pair
formation, and the unquenched unimolecular decay
time and the bimolecular quenching constant of the
excited emitter in the free form are, respectively, τ0
and kq, while in ion pair state are τ’ and kq’. Under
conditions of negligible ion-pair formation,
Keq[S2O82-]<<1, that is, under dynamic (or
bimolecular) quenching, Equation 4 is simplified to
the Stern-Volmer equation:
=

=

= 1 + 𝑘 𝜏 [𝑆 𝑂 ] (Equation 5)

while for Keq[S2O82-]>>1, that is, under static (or
unimolecular) quenching

2.4. Photophysical studies

10

=

[

]

PS there is a deviation from linearity of the SV plots
at [S2O82-]<15 mM, which indicates the formation of
a ground-state ion pair between PS and S2O82-.
Assuming that kq and kq’ are similar[61], these SV
plots can be fitted by Equation 4, and the results of
all fittings are listed in Table 3. Also, the SV plots of
PS0, at [S2O82-] > 50 mM show a change in the
curvature, with an extrapolated intercept at the x axis
origin away from (0,1) (Fig. S23), which suggests
that adsorption of the cationic PS to negatively
charged silica particles (originated from Na2SiF6
hydrolysis)[18] or dynamic quenching processes are
taking place as a result of the increased ionic
strength[61,64,65].

(Equation 6)

and when 𝑘 𝜏 [𝑆 𝑂 ] in the denominator is
large compared to 1, then
=

+ 𝑘 𝜏 [𝑆 𝑂 ]

(Equation 7)

As shown in Fig. S23, data for PS1-Co3O4 and
PS2-Co3O4 fit well into Equation 5, suggesting that
the equilibrium constant corresponding to the
formation of an S2O82-/PS-Co3O4 ion pair is
negligible and most quenchers are not in the ion-pair
ground-state. Thus, the quenching proceeds through
a bimolecular pathway. On the contrary, for all free

11

Table 3. Results of kinetic analysis of the {PS*,S2O82-} and {Co3O4-PS*,S2O82-} systems at pH 5.6.
En
try

System

ΦO2a

ΦArb

Ar0,c
ns

O20,
d
ns

NPs0
,e ns

’,f
ns

0/’g

kq (kq’),h
mol-1·L·
s-1

Keq,i mol1
·L

kET,j s-1

kq[S2O82]k, s-1

kq[S2O82]l, s-1

1

PS0 no
buffer

0.028

0.042

550

366

N/A

108

5.09

3.9·108

6.93·102

7.4·106

1.17·108

7.41·106

2

PS0

0.021

0.032

558

364

550

55

10.22

4.39·108

2.46·102

1.65·107

1.32·108

8.34·106

5

PS1

0.026

0.031

501

420

512

109

4.59

3.17·107

1.29·102

7.17·106

9.52·106

6.02·105

6

PS2

0.023

0.029

429

340

425

312

1.37

7.50·107

2.81·101

8.72·105

2.25·107

1.43·106

7

PS1Co3O4

0.025

0.030

497

414

497

-

-

5.05·107

-

-

PS2Co3O4

0.022

8

9.60·105
1.52·107

0.030

412

327

412

-

-

a

1.94·107

-

-

5.82·106

3.69·105

Quantum yield in air-saturated buffer solutions in the absence of quencher; b quantum yield in argon-saturated
buffer solutions in the absence of quencher; c from lifetime measurements in the absence of S2O82- in an argonsaturated buffer solution; d from lifetime measurements in the absence of S2O82- in an air-saturated buffer solution;
e
from lifetime measurements in the absence of S2O82- in an argon-saturated buffer solution in the presence of
Co3O4 NPs; f calculated from intercept of SV plots (τ0/ τ’) and τ0; g intercept of SV plots at variable [S2O82-]; h from
slope of SV plots at variable [S2O82-]; i computational best fit to Equation 4 (see text); j Calculated from Equation
8 (see text); k Calculated for 0.3 M S2O82-; l Calculated for 0.019 M S2O82-.
Adding Co3O4 NPs to the buffered solutions
containing the different PS and measuring
Data gathered in Table 3 shows that the quantum
luminescence lifetimes in the presence of Ar (τNPs0)
yield Φ (the ratio between emitted photons and
allowed testing of significance of charge transfer
absorbed photons) in the presence of oxygen (ΦO2) is
between *PS and Co3O4 NPs (processes 3 and 7 in
lower than in the presence of Ar (ΦAr) because of the
Scheme 1). As shown in Table 3, no significant
luminescence quenching effect of oxygen in the
differences are observed between τAr0 and τNPs0.
former case. Accordingly, lifetimes in the presence
Thus, we can assume that no direct interactions occur
of O2 (τO20) are shorter than in the presence of Ar
between the *PS and free Co3O4 NPs.
(τAr0). Moreover, the similarity of ΦAr in the free and
hybrid systems (entries 5,6 vs. 7,8) indicates that no
direct electron transfer process exists between the
excited PS and the Co3O4 NPs in the hybrid systems.

12

Scheme 1. Scheme of pertinent processes leading to photoinduced water oxidation using prototype dye-sensitized
cobalt oxide NPs in a photoelectrochemical system along with main energy levels and relevant desired (in green)
and undesired (in red) electron transfer processes. (1) Dye photoexcitation; (2) electron injection/electron acceptor
reduction; (3) hole injection/catalyst oxidation; (4) water oxidation; (5) dye (radiative or nonradiative)
deexcitation; (6) electron-hole recombination to the oxidized dye; (7) oxidative dye excited-state quenching by the
catalyst; (8) electron-hole recombination to the oxidized catalyst. Note that for the sake of simplicity the position
of the different species along the y axis does not necessarily correspond to their energy level.
behavior at pH 5.6 is observed at pH 8.4 for PS0
since PS0 is not affected by pH.
On the other hand, the intrinsic radiative and nonradiative rate constants of {PS···S2O82-}* ion pairs
Transient absorption spectroscopy was also used
should be similar to those of the *PS due to the weak
to study the difference between the free PS1 and the
(a few kilocalories per mole) electrostatic interaction
PS1-Co3O4 hybrid systems. No significant
between PS and S2O82-[60]. Then, the photoinduced
differences were observed between the free PS and
unimolecular electron transfer (ET) rate (kET) can be
that attached to Co3O4 NPs (Fig. S26), and in both
cases the ground-state bleaching of PS1 at 452 nm
estimated according to Equation 8, where  is the
and the formation of the PS1* excited state near 360
observed lifetime and 0 is the unquenched lifetime.
nm could be observed (Fig. S27). Thus, direct charge
𝑘 = −
(Equation 8)
transfer from the Co3O4 NPs to PS1* is not observed
2in any system, as otherwise deduced from the
At 0.3 M concentration of S2O8 , the bimolecular
2comparison between τAr0 and τNPs0 (Table 3),
ET rates kq[S2O8 ] are faster than the unimolecular
meaning that PS1* remains excited until its reaction
ET rates (kET) for all systems at pH 5.6 (Table 3),
with the sacrificial electron acceptor S2O82- in
suggesting that at this concentration S2O82agreement with luminescence analyses, and in
deactivates the unimolecular quenching process
accordance with Scheme 1.
since it inhibits the formation of the ground-state
{PS···S2O82-} ion pairs[60]. However, under
catalytic conditions (0.019 M S2O82-, last column in
3. Conclusions
Table 3) the bimolecular quenching is basically
In conclusion, we have demonstrated that 1favored for PS2 (kq[S2O82-]  kET), whereas the
heptanol serves as both an effective solvent and
unimolecular quenching is preferential for PS0 and
stabilizing agent for the synthesis of ultra-small Co
PS1 (kET  kq[S2O82-]).
NPs (ca. 3 nm), preventing their aggregation. The Co
For all systems, their photo-oxidation by S2O82NPs have been oxidized into Co3O4 NPs by airhas also been studied at pH 8.4, in which PS1 and
exposure in mild conditions preserving their
PS2 are deprotonated (Table S1, Fig. S24 and Fig.
morphology and dispersion. When deposited at the
S25). Under these conditions we can see no
surface of a GC-RDE electrode and in 1M NaOH,
deviations from linearity for the Stern-Volmer plots,
these Co3O4 NPs electrocatalytically oxidize water
since the unimolecular ET mechanism is clearly
with an onset  of ca. 0.29 V and η10mA/cm2 of 0.486
disfavored now due to the repulsion that appears
V, showing ECSA normalized current densities (js)
between the negatively charged PS and S2O82- at
of 1.04 mA cm-2 at  = 0.35 V, a value that fairly
basic pH. On the other hand, the same Stern-Volmer
outperforms that of all benchmarked nanostructured
metal oxide electrocatalysts deposited onto GC-

13

RDE. Despite stable and showing 95% Faradaic
efficiency, the system is less competitive at higher
current densities due to its Tafel slope of ca. 100
mV·dec-1 at  > 0.35 V.
RuII photosensitizers displaying phosphonic acid
pending groups (PS1 with two and PS2 with four)
were attached to the surface of Co3O4 NPs, yielding
PS-Co3O4 hybrid systems with a different degree of
surface functionalization, namely incorporation of
ca. 19 PS1 and 32 PS2 complexes per Co3O4 NP.
The capacity of these dyad systems to photo-oxidize
water into oxygen using visible light and S2O82- as
sacrificial electron acceptor at pH 5.6 was evaluated
and the results compared with those of unbound
systems of the same components and concentrations.
The benefits of the dyad approach arise when
observing the inactivity of the unbound Co3O4/PS
systems with regards to the significant TON and
TOF values per NP (5.4 / 0.90 min-1 and 82 / 2.05
min-1) obtained for PS1-Co3O4 and PS2-Co3O4,
respectively. The better catalytic performance of the
latter over the former was attributed to the higher
surface functionalization of PS2-Co3O4, which
enhances the kinetics of WO and protects better the
catalytic entity under working conditions against
aggregation. These data stress the important role of
the direct connection between the PS and the
nanocatalyst by; 1) favoring their efficient electronic
communication that allows being kineticallycompetitive with the typical side deactivation
processes of light-driven WO and 2) minimizing
catalyst aggregation under turnover conditions
thanks to the protective / stabilizing effect of the PS
coordinated at the surface of the Co3O4 NPs.
Photophysical measurements show the determinant
roles of the sacrificial electron acceptor, diffusion
and ground state acceptor-PS complexes, and lack of
direct electron transfer between NPs and *PS.
In summary, this work opens the way towards
precisely defined first row PS-NP hybrid dyads by
means of the so-called organometallic approach as
synthetic methodology, which provides wellcontrolled and fully characterized cobalt, cobalt
oxide and RuPS-Co3O4 nanomaterials. The latter
species have proven capable to photo-oxidize water
into dioxygen at pH 5.6, being fairly superior to their
unbound Co3O4 NPs / PS counterparts under
identical conditions. Therefore, the fine tuning of
this system through the length and nature of the PSCo3O4 NPs connection is expected to lead to a better
understanding of the key parameters governing the
catalytic process and is already under way in our
laboratories.

nanoparticles were carried out using standard
Schlenk tubes, Fisher-Porter glassware and vacuum
line techniques or in a glove-box (Braun) under an
argon atmosphere. Reagents and solvents were
degassed before use via a multi-cycle freeze-pumpthaw
process.
The
(cyclooctadienyl)(1,5cyclooctadiene)cobalt(I) complex, [Co(3-C8H13)(
4-C8H12)], was purchased from NanomepsToulouse. 1-Heptanol, sodium persulfate, sodium
hydroxide, sodium hexafluorosilicate, and sodium
bicarbonate were acquired from Sigma-Aldrich.
Hydrogen and argon were purchased from Alphagaz.
1-Heptanol was dried over activated molecular
sieves (4 Å) prior to use and other reagents were
employed as received unless otherwise specified.
Solvents (THF, pentane, dichloromethane, diethyl
ether) were purified before use by filtration on
adequate alumina columns in a purification
apparatus (MBraun) and handled under argon
atmosphere.
4.2. Synthesis protocols
Photosensitizers: The photosensitizers used in
this work (see Fig. 1) were prepared according to
literature data[31,32] and obtained with Cl- as
counterion.
Co nanoparticles: [Co(3-C8H13)(4-C8H12)]
(120 mg, 0.43 mmol) as cobalt source and anhydrous
1-heptanol (20 mL) as both solvent and stabilizer
were mixed into a Fisher-Porter reactor under an
argon atmosphere inside a glove-box, leading to a
brownish solution. Then, the Fisher-Porter reactor
was pressurized with 3 bar of H2 and the reaction
mixture was kept under vigorous stirring overnight,
after which a dark colloidal dispersion was obtained.
Excess H2 was eliminated under vacuum. A TEM
grid was prepared under argon for TEM analysis of
the crude colloidal solution. The application of a
magnet on the reactor walls allowed to attract the Co
NPs as a solid and then to isolate them from 1heptanol, which was then removed via cannula. The
Co NPs were then washed with degassed anhydrous
pentane (4 x 20 mL) and dried under vacuum. ICPOES (wt%): Co (35.35%).
Co3O4 nanoparticles: Co3O4 NPs were prepared
by treatment of isolated Co NPs under ambient air at
room temperature during 6 days. Estimated Co
content: 31.33%.
PS-Co3O4 NP hybrids: A solution of PS1 (11 mg,
0.014 mmol, 0.24 eq) or PS2 (13 mg, 0.014 mmol,
0.24 eq) in methanol (0.4 mL) was added to a
colloidal dispersion of Co3O4 NPs (33 mg, 0.058
mmol of Co3O4(heptanol)2.8 units) in 1-heptanol (0.6
mL). The reaction mixture was kept under vigorous
stirring for 4 days in the dark. Then, precipitation of
the crude product was achieved by adding
isopropanol (1.5 mL) and diethyl ether (10 mL) and
centrifuging at 1000 rpm for 10 min. The obtained

4. Experimental Section
4.1. General
All procedures concerning the synthesis and
preparation of samples for characterization of Co

14

crude product and water (1 mL) were introduced in
a cellulose membrane bag for dialysis against
deionized water (2 L). Dialysis was pursued until the
external dialysis solution remained colorless. Then,
centrifugation allowed to recover a solid, which was
washed 3 times with a mixture of diethyl
ether / isopropanol (8:2, v/v) to remove water and 3
times again with diethyl ether before drying under
vacuum. The Ru/Co ratio of the obtained PS-Co3O4
NPs hybrids was calculated from ICP-OES
measurements. PS-Co3O4 NPs hybrids were stored
in the dark. ICP-OES (wt%): PS1-Co3O4, Ru
(2.23%), Co (43.60%); PS2-Co3O4, Ru (3.12%), Co
(35.22%). Recovered: 19 mg for PS1-Co3O4; 18 mg
for PS2-Co3O4.

10 min. After that, the FTO/Co3O4 NPs electrode
was dipped into a 0.5% wt PMMA dichloromethane
solution for a few seconds (<10 sec) and air-dried.
PMMA Coating. A poly(methyl methacrylate)
(PMMA) coating was formed onto the FTO
electrode supporting Co3O4 NPs by simply dipping
this electrode in dichloromethane (DCM) with 0.5 %
wt concentration of PMMA. After soaking the
electrode in the PMMA solution for a few seconds
(< 10 sec), the electrode was air-dried.
Acknowledgements
J. De T. acknowledges the Universitat Autònoma de
Barcelona for a PIF doctoral grant. Financial
supports were provided by MINECO / FEDER
(CTQ2015-64261-R and CTQ2015-65268-C2-1-P),
IDEX UNITI Emergence (UFTMIP: 2015-209-CIFD-DRD-127185) and CNRS. GDRI HC3A CNRS
action (Catalunya / Midi-Pyrénées) and CTP
regional action (Catalunya / CTP2013-00016, MidiPyrénées / n°13053026, Basque Country/CTP2013R03 and Région Aquitaine / n°13010761) are
gratefully acknowledged for exchange funding
between the partners. J.G.-A. acknowledges the
Serra Húnter Program.

4.3. Preparation of electrodes
RDE/Co3O4 and GC/Co3O4 NPs: 15 L of a
dispersion of Co3O4 NPs (1 mg) in THF (250 L)
were deposited onto the 0.07 cm2 glassy carbon (GC)
disk from the rotatory disk electrode (RDE) or onto
a GC electrode and let it dry under air.
FTO/Co3O4 NPs-PMMA: A dispersion of Co3O4
NPs (2 mg) in 1-heptanol (500 μL) was prepared.
Then, the NPs were deposited by spin-coating 15 μL
of this dispersion onto an FTO electrode followed by
evaporation of the solvent in a furnace at 100ºC for

15

References

[1] S. Berardi, S. Drouet, L. Francàs, C. Gimbert-Suriñach, M. Guttentag, C. Richmond, T. Stoll, A. Llobet, Chem.
Soc. Rev. 43 (2014) 7501-7519.
[2] R. Bofill, J. García-Antón, L. Escriche, X. Sala, A. Llobet, Water Oxidation, in: J. Reedijk, K. Poeppelmeier
(Eds.), Comprehensive Inorganic Chemistry, II, Elsevier, 2013, Vol 8, pp. 505-523.
[3] R. Bofill, J. García-Antón, L. Escriche, X. Sala, J. Photochem. Photobiol. B: Biol. 152 (2015) 71-81.
[4] N.S. Lewis, Science 351 (2016) aad19201.
[5] M.D. Kärkäs, O. Verho, E.V. Johnston, B. Akermark, Chem. Rev. 114 (2014) 11863-12001.
[6] B. Limburg, E. Bouwman, S. Bonnet, ACS Catal. 6 ( 2016) 5273–5284.
[7] W.J. Youngblood, S.-H.A. Lee, Y. Kobayashi, E.A. Hernandez-Pagan, P.G. Hoertz, T.A. Moore, A.L. Moore,
D. Gust, T.E. Mallouk, J. Am. Chem. Soc. 131( 2009) 926-927.
[8] J. Youngblood, S.H. Lee, K. Maeda, T.E. Mallouk, Acc. Chem. Res. 42 (2009) 1966-1973.
[9] D.L. Ashford, M.K. Gish, A.K. Vannucci, M.K. Brennaman, J.L. Templeton, J.M. Papanikolas, T.J. Meyer,
Chem. Rev. 115 (2015) 13006-13049.
[10] N. Kaveevivitchai, R. Chitta, R. Zong, M. El Ojaimi, R.P. Thummel, J. Am. Chem. Soc. 134 (2012) 1072110724.
[11] H. Li, F. Li, B. Zhang, X. Zhou, F. Yu, L. Sun, J. Am. Chem. Soc. 137 (2015) 4332-4335.
[12] A.M. Lopez, M. Natali, E. Pizzolato, C. Chiorboli, M. Bonchio, A. Sartorel, F. Scandola, Phys.Chem. Chem.
Phys. 16 (2014) 12000-12007.
[13] E.A. Karlsson, B.-L. Lee, R.-Z. Liao, T. Akermark, M.D. Kärkäs, V.S. Becerril, P.E.M. Siegbahn, X. Zou,
M. Abrahamsson, B. Akermark, ChemPlusChem 79 (2014) 936-950.
[14] R.K. Hocking, R. Brimblecombe, L.-Y. Chang, A. Singh, M.H. Cheah, C. Glover, W.H. Casey, L. Spiccia,
Nat. Chem. 3 (2011) 461-466.
[15] M.W. Kanan, D.G. Nocera, Science 321 (2008) 1072-1075.
[16] H.-Y. Wang, J. Liu, J. Zhu, S. Styring, S. Ott, A. Thapper, Phys. Chem. Chem. Phys. 16 (2014) 3661-3669.
[17] P. Garrido-Barrios, C. Gimbert-Suriñach, R. Matheu, X. Sala, A. Llobet, Chem. Soc. Rev. 46 (2017) 60886098.
[18] P.G. Hoertz, Y.-I. Kim, W.J. Youngblood, T.E. Mallouk, J. Phys. Chem. B 111 (2007) 6845-6856.
[19] H.S. Soo, A. Agiral, A. Bachmeier, H. Frei, J. Am. Chem. Soc. 134 (2012) 17104-17116.
[20] A. Agiral, H.S. Soo, H. Frei, Chem. Mater. 25 (2013) 2264 -2273.
[21] E. Edri, J.K. Cooper, I.D. Sharp, D.M. Guldi, H. Frei, J. Am. Chem. Soc. 139 (2017) 5458-5466.
[22] C. Amiens, B. Chaudret, D. Ciuculescu-Pradines, V. Collière, K. Fajerwerg, P. Fau, M. Kahn, A. Maisonnat,
K. Philippot, New J. Chem. 37 (2013) 3374-3401.
[23] M. Salavati-Niasari, A. Khansari, CR. Chimie 17 (2014) 352-358.
[24] J. Osuna, D. de Caro, C. Amiens, B. Chaudret, E. Snoeck, M. Respaud, J.-M. Broto, A. Fert, J. Phys. Chem.
100 (1996) 14571-14574.
[25] M. Verelst, T.O. Ely, C. Amiens, E. Snoeck, P. Lecante, A. Mosset, M. Respaud, J.-M. Broto, B. Chaudret,
Chem. Mater. 11 (1999) 2702-2708.
[26] C. Amiens, Faraday Discuss. 125 (2004) 293-309.
[27] K. Pelzer, K. Philippot, B. Chaudret, Z. Phys. Chem. 217 (2003) 1539-1547.
[28] M.C. Biesinger, B.P. Payne, A.P. Grosvenor, L.W.M. Lau, A.R. Gerson, R.St.C. Smart, Appl. Surf. Sci. 257
(2011) 2717-2730.
[29] J. Yang, H. Lu, W.N. Martens, R.L. Frost, J. Phys. Chem. C 114 (2010) 111-119.
[30] J. Jiang, L. Li, Materials Letters 61 (2007) 4894–4896.
[31] M.R. Norris, J.J. Concepcion, C.R.K. Glasson, Z. Fang, A.M. Lapides, D.L. Ashford, J.L. Templeton, T.J.
Meyer, Inorg. Chem. 52 (2013) 12492-12501.
[32] P. Jansa, O. Baszczyňski, E. Procházková, M. Dračínský, Z. Janeba, Green Chem. 14 (2012) 2282-2288.
[33] S.A. Paniagua, A.J. Giordano, O’Neil L. Smith, S. Barlow, H. Li, N.R. Armstrong, J.E. Pemberton, J.-L.
Brédas, D. Ginger, S.R. Marder, Chem. Rev. 12 (2016) 7117-7158.

16

[34] G. Guerrero, P.H. Mutin, A. Vioux, Chem. Mater. 13 (2001) 4367-4373.
[35] G. Guerrero, J.G. Alauzun, M. Granier, D. Laurencin, P.H. Mutin, Dalton Trans. 42 (2013) 12569-12585.
[36] C.C.L McCrory, S. Jung, I.M. Ferrer, S.M. Chatman, J.C. Peters, T.F. Jaramillo, J. Am. Chem. Soc. 137
(2015) 4347-4357.
[37] C.C.L. McCrory, S. Jung, J.C. Peters, T.F. Jaramillo, J. Am. Chem. Soc. 135 (2013) 16977-16987.
[38] S. Jung, C.C.L. McCrory, I.M. Ferrer, J.C. Peters, T.F. Jaramillo, J. Mater. Chem. A 4 (2016) 3068-3076.
[39] S. Trasatti, O.A. Petrii, Pure Appl. Chem. 63 (1991) 711-734.
[40] J.D. Benck, Z. Chen, L.Y. Kuritzky, A.J. Forman, T.F. Jaramillo, ACS Catal. 2 (2012) 1916-1923.
[41] Y. Zhao, S. Chen, B. Sun, D. Su, X. Huang, H. Liu, Y. Yan, K. Sun, G. Wang, Sci. Rep. 5 (2015) 7629, DOI:
10.1038/srep07629.
[42] Y. Liang, Y. Li, H. Wang, J. Zhou, J. Wang, T. Regier, H. Dai, Nat. Mater. 10 (2011) 780-786.
[43] J. Wu, Y. Xue, X. Yan, W. Yan, Q. Cheng, Y. Xie, Nano Res. 5 (2012) 521-530.
[44] M.G. Walter, E.L. Warren, J.R. McKone, S.W. Boettcher, Q. Mi, E.A. Santori, N.S. Lewis, Chem. Rev. 110
(2010) 6446-6473.
[45] M.F. Weber, M.J. Dignam, J. Electrochem. Soc. 131 (1984) 1258-1265.
[46] Y. Gorlin, T.F. Jaramillo, J. Am. Chem. Soc. 132 (2010) 13612-13614.
[47] N.H. Chou, P.N. Ross, A.T. Bell, T.D. Tilley, ChemSusChem 4 (2011) 1566-1569.
[48] S.K. Singh, V.M. Dhavale, S. Kurungot, ACS Appl. Mater. Interfaces 7 (2015) 442-451.
[49] H.A. Gasteiger, S.S. Kocha, B. Sompalli, F.T. Wagner, Appl. Catal. B 56 (2005) 9-35.
[50] J. Suntivich, K.J. May, H.A. Gasteiger, J.B. Goodenough, Y. Shao-Horn, Science 334 (2011) 1383-1385.
[51] K.-R. Wee, M.K. Brennaman, L. Alibabaei, B.H. Farnum, B. Sherman, A.M. Lapides, T.J. Meyer, J. Am.
Chem. Soc. 136 (2014) 13514-13517.
[52] M. Zhang, M. de Respinis, H. Frei, Nat. Chem. 6 (2014) 362-367.
[53] H.-C. Chen, D.G.H. Hetterscheid, R.M. Williams, J.I. van der Vlugt, J.N.H. Reek, A.M. Brouwer, Energy
Environ. Sci. 8 (2015) 975-982.
[54] P. Comte, M.K. Nazeeruddin, F.P. Rotzinger, A.J. Frank, M. Grätzel, J. Mol. Catal. 52 (1989) 63-84.
[55] P.K. Ghosh, B.S. Brunschwig, M. Chou, C. Creutz, N. Sutin, J. Am. Chem. Soc. 106 (1984) 4772-4783.
[56] A. Lewandowska-Andralojc, D.E. Polyansky, R. Zong, R.P. Thummel, E. Fujita, Phys. Chem. Chem. Phys.
15 (2013) 14058-14068.
[57] M.S. Baptista, J. Cadet, P.D. Mascio, A.A. Ghogare, A. Greer, M.R. Hamblin, C. Lorente, S.C. Nunez, M.S.
Ribeiro, A.H. Thomas, M. Vignoni, T.M. Yoshimura, Photochem. Photobiol. 93 (2017) 912-919.
[58] L. Francàs, C. Richmond, P. Garrido-Barros, N. Planas, S. Roeser, J. Benet-Buchholz, L. Escriche, X. Sala,
A. Llobet, Chem. Eur. J. 22 (2016) 5261-5268.
[59] I. Bazzan, A. Volpe, A. Dolbecq, M. Natali, A. Sartorel, P. Mialane, M. Bonchio, Catal. Today 290 (2017)
39-50.
[60] A.L. Kaledin, Z. Huang, Y.V. Geletii, T. Lian, C.L. Hill, D.G. Musaev, J. Phys. Chem. A 114 (2010) 73-80.
[61] H.S. White, W.G. Becker, A.J. Bard, J. Phys. Chem. 88 (1984) 1840-1846.
[62] D. Rehm, A. Weller, Bunsenges Ber. Phys. Chem. 73 (1969) 834.
[63] D. Rehm, A. Weller, Isr. J. Chem. 8 (1970) 259.
[64] F. Bolletta, M. Maestri, L. Moggi, J. Phys. Chem. 77 (1973) 861-862.
[65] A.J. Bard, M.A. Fox, Acc. Chem. Res. 28 (1995) 141-145.

17