This document is the Accepted Manuscript version of a Published Work that appeared in
final form in Royal Society of Chemistry © peer review and technical editing by the
publisher. To access the final edited and published work see
https://pubs.rsc.org/en/content/articlelanding/2018/cp/c8cp03987a#!divAbstract

Synergistic interplay of Zn and Rh-Cr promoters on Ga2O3 based
photocatalysts for water splitting
Marta Borges Ordoño,a Shunsaku Yasumura,a Pieter Glatzel,b and Atsushi Urakawa*a
Photocatalytic water splitting activity of a wide-bandgap material, Ga2O3, is greatly boosted with the addition of Zn and RhCr co-catalyst at optimum loadings. To date, however, the exact roles of the co-catalysts and particularly the origin of their
synergistic functions are not clarified. Herein, we present how the optimum Zn loading on Ga2O3 leads to creation of
ZnGa2O4/Ga2O3 heterojunction favorable for charge separation through the information on the occupied and unoccupied
electronic states of Zn and Ga elucidated by X-ray absorption and emission spectroscopic methods. The function of Rh-Cr as
electron sink and reduction site was proven by photocatalytic experiments using an electron scavanger (Ag+) and by learning
where Ag deposites and its effects on the photocatalytic activity. Finally, perturbation of Zn electronic structure by photo
activation was evidenced by modulation exciatation X-ray absorption spectroscopy. Importantly, Rh-Cr markedly enhanced
the level of the perturbation, serving as a proof of direct communication and synergy between the electronic states of Zn,
present in ZnGa2O4, and Rh-Cr deposited on Ga2O3.

Introduction
Photocatalytic water splitting has been extensively studied
over the past years as a sustainable path to harvest sunlight to
convert the energy contained in photons into the form of
chemical energy such as hydrogen.1-3 TiO2 is the most widely
investigated photocatalyst material because of its abundance,
non-toxicity, and low cost;4 however, the low H2 productivity
achieved by TiO2 has motivated researchers to look for
alternative semiconductor materials.5-7 Among them, Ga2O3
exhibits excellent activity for pure water splitting under UV-light
due to its wide bandgap (4.6 eV) sufficient for the formation of
hydrogen as well as oxygen.8 Nevertheless, bare metal oxides
generally lack surface sites to efficiently activate the redox
reactions and often show poor photocatalytic water splitting
performance. The most common strategy to improve the
catalytic activity is the introduction of co-catalyst(s) to promote
electron trapping ability near the catalyst surface, extending the
life-time of electrons and holes generated upon light excitation
and facilitating the redox reactions of adsorbed species at
surface active sites of or near the co-catalyst.9 Noble metals
such as Pt, Ru, and Rh,8 and combinations of metal oxides such
as NiO,10 RuO2,11 and Rh2-yCryO3 (or simple written as Rh-Cr)12
are commonly reported as effective co-catalysts when optimally
loaded on the photocatalysts. Particularly for Ga-based oxide
materials, rhodium and chromium mixed oxide was found
greatly beneficial for overall water splitting as demonstrated for
the visible-light active (Ga1-xZnx)(N1-xOx) material.11,13 More
recently, Rh2-yCryO3 deposited on Zn-doped Ga2O3 was found to
exhibit one of the highest water splitting activity reported to
date (21.0 mmol of H2 h-1 and 10.5 mmol O2 h-1) under UV-light
irradiation.14,15
Promotion of photocatalytic functions by co-catalyst(s)
generally involves the formation of additional energy states

which facilitate charge carriers (electron-hole) separation by
electron trap or donation from the co-catalyst to the
semiconductor material.16 For the most active Rh-Cr and Zn codoped Ga2O3 material, the function of Rh-Cr is generally
explained as an electron sink whereas the role of Zn is not
clear,17 although we have recently shown that the
recombination of charges is greatly delayed by Zn promotion,
thus enhancing the life-time of active redox sites and
consequently the photocatalytic activity.18 However, the origin
of the promotional effects and detailed insights into the
electronic structure of the materials, particularly the synergistic
functions of Rh-Cr and Zn co-promotion at an optimum loading
of co-catalysts (especially Zn),14, 17 are not known. Since the
unique combination of the constituting elements of Rh-Cr/ZnGa2O3 catalyst affords outstanding photocatalytic performance
in pure water splitting by a few orders of magnitude higher than
those generally reported,14, 18 it is imperative to precisely
understand the promotional and synergistic effects of Rh-Cr and
Zn towards the rational photocatalyst design for hydrogen
production from water with sunlight.
Reflecting this background, herein we uncovered the effects
of Zn and Rh-Cr promotion through understanding their impacts
on the electronic structure of Ga2O3-based photocatalysts.
Element selective X-ray absorption and emission spectroscopies
(XAS and XES) were employed to learn about the structures of
unoccupied (related to conduction band) and occupied (related
to valance band) electronic states of Zn and Ga within the
catalyst materials, respectively.19-21 In contrast to the integral
information expressed as bandgap commonly gained measured
by UV-Vis diffuse reflectance spectroscopy (UV-Vis DRS),22, 23
the information from XAS-XES uniquely shows how the
electronic structures of atoms in a material are altered24 and,
together with additional structural insights gained by electron
microscopy and X-ray diffraction, the exact functions of Zn-Ga

heterojunction favourable for charge separation could be
elucidated. On the other hand, the function of Rh-Cr as
reduction and H2-evolution site was verified by a control
experiment using Ag+ as electron scavenger as well as by
monitoring the change in the electronic structure of Cr by XAS.
Finally, the direct involvement of Rh-Cr component on the
charge separation in Zn element was evidenced by means of
XAS combined with a modulation excitation technique,25 firmly
explaining the electronic interactions of the two promoters and
their concerted synergistic functions resulting in the drastic
boost of the photocatalytic water splitting activity of Ga2O3.

Experimental section
Photocatalysts preparation
Zn-modified Ga2O3 catalysts were prepared by the
impregnation method followed by a thermal treatment. First, βGa2O3 (≥99.99% trace metal basis, Sigma-Aldrich) raw material
was calcined at 773 K for 3 h. The β-Ga2O3 powder was then
impregnated with an aqueous solution of Zn(NO3)2·6H2O
(99.999% trace metal basis, Sigma-Aldrich) and calcined at 1123
K for 6 h, the Zn loading was varied from 0 to 70 wt% (based on
the weight of Zn with respect to that of Ga2O3). Zn-modified
Ga2O3 was then impregnated with Rh (0.5 wt%) and Cr (0.75
wt%) w.r.t Zn-Ga2O3 with an aqueous solution containing RhCl3
and Cr(NO3)3·9H2O precursors.26 The as-prepared materials
were finally calcined at 623 K for 1 h. All calcination treatments
were performed under the flow of synthetic air.
Reaction setup
Photocatalytic activity measurements were performed at
room temperature using a home-made continuous reaction
system.18 In brief, the setup consisted of a quartz reactor
connected to a gas supply. The reaction products were
monitored by an online MS (Pfeiffer Vacuum, Omnistar GSD
320) and the product quantification was verified by a GC
(Agilent Technologies, micro-GC 490). Powder catalyst (20 mg,
0.6 g L-1) was dispersed in ultrapure water (35 mL, Milli-Q). 0.01
M AgNO3 aqueous solution was used for electron-trapped
oxygen evolution studies. The suspension was kept under
stirring and continuous N2 gas flow (4.5 mL min-1) during the
reaction. The catalyst solution was purged with N2 for 1 h prior
to irradiation with a 400 W high-pressure Hg lamp (UV-Technik,
without optical filter) at 8 cm distance from the reactor to avoid
excessive heat-up and water evaporation. The Ga2O3-based
materials are known to show photocatalytic activity with only
UV light (<275 nm) and the test was performed without an
optical filter.18 The reactor temperature was monitored to be
about 353 K during the reaction. Repeated light on/off cycles for
1.5 h of each on/off phase were studied for overall water
splitting reaction. For the evaluation of O2 evolution reaction in
the presence of silver nitrate, the catalyst in solution was
continuously irradiated for 4 h.
Materials characterization

Powder X-ray diffraction (PXRD) were performed on a
Bruker D8 Advance power diffractometer with a vertical 2thetatheta goniometer in transmission configuration with Cu Kα
radiation (λ=1.5406 Å). Maud software was used to perform
phase quantification of the synthesised materials.27
Instrumental broadening was firstly corrected to perform
Rietveld refinement and phase quantification. XAS and valanceto-core (VtC)-XES measurements were performed at ID26 of the
ESRF (Grenoble, France). The incoming energy was selected by
the (111) reflection of a pair of cryogenically cooled Si crystals
and the beam footprint on the sample was 1.0 x 0.2 mm2. Higher
harmonics were suppressed by three Si mirrors working at
2.5mrad. An emission spectrometer with four spherically bent
(r=50mm, R=1000mm) analyser crystals in vertical Johann
geometry was employed to select the K fluorescence lines of
Cr, Zn and Ga using the reflections Ge (333), Si (555) and Ge
(555), respectively. HERFD-XANES were recorded on the
maximum of the K1,3 line (9572.60 eV for Zn and 10265.37 eV
for Ga; SI, Fig. S3. HERFD and total fluorescence detections are
compared in Fig. S4) and by varying the incident energy from
10.36 to 10.46 keV (Ga), 9.65 to 9.75 keV (Zn) and 5.98 to 6.08
(Cr) in continuous scan mode. The spectra were not corrected
for incident beam self-absorption (over-absorption) effects.
Linear combination fitting was performed by using additional
fitting parameters that account for the spectral distortion
arising from self-absorption.
XES was measured for Ga and Zn K-edges at a fixed incident
energy of 10.50 and 9.80 keV, respectively. VtC-XES were
acquired from 10.28 to 10.40 keV (Ga) and 9.59 to 9.70 keV (Zn)
with a step size of 0.30 eV. All samples and reference materials
were measured ex situ in the form of pellet (13 mm diameter),
containing 60 mg of material and 60 mg of cellulose. For in situ
irradiation experiments, pellets (13 mm diameter) containing
200 mg of pure material were transiently irradiated with DH2000 (Ocean Optics) UV-Vis-NIR deuterium-halogen light source
while measuring 10 spectra under dark and then under light
conditions. The light on-off cycles were repeated 11 times for
the material without Rh-Cr (220 spectra in total; 120 s for each
spectral acquisition) and 7 times for the samples containing RhCr (140 spectra in total), and the subtle differences of the
spectral features were studied by modulation excitation
methodology explained in the following section. Athena was
used to normalized the XANES spectra,28 and for all the samples
E0 was set individually to the corresponding maximum position
of the first derivative. Linear combination fitting (LCF) was
performed on the normalized XANES spectra and two standards
ZnO and ZnGa2O4 (Zn K-edge) and ZnGa2O4 and Ga2O3 (Ga Kedge) were used for the fitting in the range of 9.65-9.69 keV (Zn)
and 10.36-10.40 keV (Ga).29
Modulation excitation spectroscopy (MES)
MES allows precise kinetic studies and sensitivity boosting
of signals arising from transient processes.25, 30, 31 In this study
the latter advantage (typically 2-3 order of magnitude
sensitivity boost) was exploited. The mathematical core of MES
is phase sensitive detection (PSD, also called demodulation) as
shown in Eq. (1), which converts time-domain response A(t) to
a phase-domain response Ak(φkPSD) with almost full reduction of

the static noise and signals unaffected by the external
perturbation (here light on and off).25, 30
ଶ

்

Ak(φkPSD) = ‫׬‬଴ ‫ܣ‬ሺ‫ݐ‬ሻ sin(kωt+φkPSD)dt
்

(1)

The PSD analysis was performed at demodulation index k = 1.

Results and discussion
Photocatalytic activity of Zn-Ga2O3 and Rh-Cr/Zn-Ga2O3
First, effects of Zn-loading amount as well as addition of RhCr promoters to Ga2O3 on photocatalytic water splitting activity
were systematically studied. Fig. 1 compares the productivity of
hydrogen and oxygen as a function of Zn-loading with/without
addition of Rh-Cr co-catalyst. Bare Ga2O3, without Zn and Rh-Cr
promoters, shows low H2 productivity (0.16 mmol gcat-1 h-1),
although this productivity level is comparable to that of wellknown Pt/TiO2 photocatalyst (0.10 mmol gcat-1 h-1).32, 33 At 2-4
wt% Zn-loading, the H2 productivity was enhanced reaching 1.1
mmol gcat-1 h-1. However, the H2 productivity drops at higher Znloadings (>20 wt%), as similarly reported by Wang et al. who
observed the drop in the activity from 3 to 4 Zn-atom% (which
would correspond to 1.1-1.4 wt% of Zn of this study).34
Remarkably, the addition of Rh-Cr boosts the water splitting
activity by almost one order of magnitude, while retaining the
activity trends induced by the varying amount of Zn. The highest
H2 productivity of 9.0 mmol gcat-1 h-1 observed for Rh-Cr/4 wt%
Zn-Ga2O3 was more than the sum of the Rh-Cr effect (3.3 mmol
gcat-1 h-1) and the Zn-effect (1.1 mmol gcat-1 h-1), confirming the
previously observed synergistic effects of the Zn and Rh-Cr
promoters.14, 18 Stoichiometric production of H2 and O2 was not
achieved with higher H2/O2 ratio possibly due to the formation
of peroxo-species,33, 35 although we could not identify such
species in the solution.

Fig. 1 H2 (green) and O2 (purple) productivity in the
photocatalytic water splitting reaction using pure water for xZnGa2O3 and Rh-Cr/xZn-Ga2O3 (where x = 0, 2, 4, 20, 50, and 70
wt% of Zn). H2/O2 productivity was averaged from 4 runs by 1.5
h UV-Vis irradiation cycles. O2 productivity for the materials
without Rh-Cr is not shown because of the low amount of O2

(below the detection limit of the GC) for confident
quantification.
Geometrical and electronic structure changes by Zn-loading
The structures of the Zn-promoted Ga2O3 materials without
Rh-Cr co-catalyst were studied by XRD, XAS, and XES to
elucidate the changes induced by the Zn addition. The effects of
Rh-Cr co-catalyst on the structure of the catalyst materials are
considered to be solely electronic ones due to the low loading
of Rh-Cr, as observed from the negligible influence on the
characterization results by the presence of Rh-Cr (not shown
here).
PXRD results (Fig. 2a) show that the Zn-Ga2O3 materials are
composed of β-Ga2O3 and ZnGa2O4 phases at low Zn
concentrations (2-4 wt%) in accordance with literature,34
whereas in this work the presence of ZnO phase is also
confirmed at high Zn-loading (≥20 wt%). The absence of the
shifts of the β-Ga2O3 reflection peaks at all Zn-loadings indicates
that Zn is not incorporated into the Ga2O3 lattice.36 Electron
microscopy and energy dispersive analysis indicate the
presence of Zn elements near the material surface (Fig. S1). This
suggests that ZnGa2O4 and ZnO are deposited on the surface of
Ga2O3 as expected from the impregnation and subsequent
calcination methods used to prepare the Zn-Ga2O3 materials.
Quantitative phase analysis (Fig. S2) confirms that up to 4 wt%
Zn-loading, Zn is present only as ZnGa2O4, while the amounts of
both ZnGa2O4 and ZnO, particularly the latter, increase as the
Zn-loading increases. Analysis of the HERFD-XANES spectra at
Zn and Ga K-edge (Fig. 2b) by linear combination fitting,
together with Zn K-edge VtC-XES (Fig. S5) and Zn K-edge EXAFS
(Fig. S6) point to the identical conclusion, clearly showing how
the Zn elements in the materials are present, either solely as
ZnGa2O4 (2, 4 wt% Zn) in the vicinity of the Ga atoms with a
creation of additional electronic states in the valance bands, or
as a mixture of ZnGa2O4 and ZnO (≥20 wt% Zn).

Fig. 2 a) PXRD patterns for Zn-modified Ga2O3 materials,
including reference materials ZnO, ZnGa2O4, and Ga2O3. The
dotted lines indicate Ga2O3 (35.2°), ZnGa2O4 (35.7°), and ZnO
(36.3°) phases. b) Linear combination fitting analysis of Zn and
Ga K-edge XANES spectra using ZnGa2O4 and ZnO, and ZnGa2O4
and Ga2O3 as standards to perform the fitting, respectively.
Furthermore, the integral electronic properties of Znmodified Ga2O3 materials are studied by UV-Vis DRS through
their optical properties (Fig. S7). The optical bandgap of Ga2O3
is barely modified by the Zn addition (from 4.58 to 4.55 eV).
Therefore, the catalytic activity trends at different Zn-loading
(Fig. 1) cannot be explained by the change in the bandgap. This
is in accordance with the report by Wang et al. where Zn-Ga2O3
materials were studied for CO2 reduction.36 This also implies
that the photon energy required to generate excitons (bound
electron-hole pairs) is not strong enough to separate electrons
and holes, which is required for photocatalytic reactions, and
the functions and interface of different components of the
photocatalyst facilitate the charge separation.
To gain precise insights into the electronic structures of the
Zn-promoted materials, XAS and VtC-XES were employed to
study unoccupied and occupied energy states, respectively, of
Zn and Ga atoms within the materials (Fig. S8).37 At both Zn Kedge (Fig. 3a) and Ga K-edge (Fig. 3b) the highest occupied
energy level (VtC-XES results) are virtually unaltered among the
reference materials. Nevertheless, it should be noted that
electronic orbital mixing from s orbitals of O atom and d orbitals
of Ga atom is observed at the VtC-XES Kβ” line of ZnGa2O4 (inset
in Fig. 3a; the electronic structure assignments based on DFT
calculations are shown in Fig. S5), affirming electronic

interaction and communication between Zn and Ga atoms in
ZnGa2O4.
Clear differences are observed for the energy levels of
unoccupied states at the Zn and Ga absorption edges. For the
Zn K-edge, at small Zn concentrations (2 and 4 wt%) the lowestunoccupied electronic states appear to be similar to those of
ZnGa2O4 (Fig. S8a). At increasing Zn-loading where ZnO phase is
more prominent, the absorption edges shift to lower energies,
corresponding to smaller valance-conduction bandgaps (3.1 eV
for ZnO, calculated from the difference between the maximum
of the 1st derivative of the emission spectrum and the
absorption edge as shown in Fig. S9). The bandgap of ZnO
derived from the XAS-XES analysis that reflects the Zn pprojected density of states agrees with that of the optical
bandgap (3.2 eV) determined by UV-Vis DRS. The small bandgap
of ZnO (Fig. 3 and Fig. S9), its increasing light absorption at
higher loading (Fig. S7) and the coexistence of ZnGa2O4 and
Ga2O3 phases (Fig. 2a) suggest that ZnO covers ZnGa2O4-Ga2O3
at high Zn loading. This leads to inefficient use of photons for
water splitting reactions (Fig. 1) because the ZnO conduction
band energy is very close to the one required for H2 evolution (0.2 eV compared to 0 eV, respectively),38 thus ZnO does not
have the appropriate energy for water reduction upon light
absorption.
On the other hand, the lowest-unoccupied energy states
(XANES onset) of Ga are shifted toward higher energies with
higher Zn content (Fig. S8b) as the amount of ZnGa2O4 increases
(Fig. 2b), and pure ZnGa2O4 shows the highest K-edge energy
(Fig. 3b, 4.42 eV as shown in Fig. S9). Thus, Ga in ZnGa2O4 has a
considerably higher edge energy by ca. 1 eV compared to Ga2O3,
implying a higher conduction band energy level of the former.
These indications of higher conduction band energy level of
ZnGa2O4, the slightly higher valance band energy by 0.050.01
eV (using the 1st moment of the K2,5 peak) determined for
ZnGa2O4 with respect to that of β-Ga2O3 (smaller than 1.5 eV
reported by valance band XPS),34 and the Zn-loading-dependent
coverage of ZnGa2O4 and ZnO over β-Ga2O3 surface point to the
important roles of ZnGa2O4/Ga2O3 heterojunction as depicted
in Fig. 4. Due to the similar optical bandgaps, excitons are
generated in/on both materials and the electrons will migrate
from ZnGa2O4 to Ga2O3 surface to catalyse reduction reactions
and conversely the holes migrate from Ga2O3 to ZnGa2O4 to
catalyse oxidation reactions. The creation of the heterojunction
explains the reported longer life-time of photogenerated
electrons and holes by transient spectroscopic methods.18, 34
Adequate amount of ZnGa2O4 should be present on Ga2O3 to
efficiently create the heterojunction, while sufficiently large
area of Ga2O3 should also be exposed to catalyse hydrogen
evolution reactions. On the other hand, upon ZnO formation on
the catalyst surface due to excessive Zn-loading, light is merely
absorbed by the material without energy harvesting for water
splitting (Fig. 4).

Role of Rh-Cr co-catalyst

Fig. 3 XAS and VtC-XES spectra for a) Zn K-edge and b) Ga Kedge. Pure ZnGa2O4 (black), ZnO (blue), and Ga2O3 (red).
Spectral intensity was normalized to the spectral area.

Fig. 4 Schematic of light absorption for Zn-Ga2O3 materials with
a) low Zn loadings (2-4 wt%) showing the electron and hole
migrations between the ZnGa2O4/Ga2O3 heterojunction, and b)
high Zn loadings (≥20 wt%).

As presented in Fig. 1, a small amount of Rh-Cr drastically
boosts the catalytic activity of Ga2O3 and Zn-Ga2O3 materials.
Promoting effects of metal co-catalysts in photocatalytic
reactions are generally attributed to their function as electron
trap and sink.9, 16, 39 To firmly understand and attribute the
function of Rh-Cr co-catalyst, water splitting activity and
structural changes of 4 wt% Zn-Ga2O3 with/without Rh-Cr were
studied using silver nitrate as electron scavenger.
Photocatalytic experiments in the presence of silver nitrate
(Fig. 5b) show fully suppressed H2 production as anticipated by
the efficient electron intercept by Ag+ at the location where free
electrons are concentrated and released. To our surprise, only
a small amount of O2 was detected for both materials despite
the well-documented function of Ag+ to enhance water
oxidation activity of photocatalysts.40, 41 Also, the O2
productivity with and without Rh-Cr co-catalyst was similar,
indicating disabled function of Rh-Cr in the presence of Ag+.
STEM (Fig. S10) and Cr K-edge XANES (Fig. S11) of the Rh-Cr/ZnGa2O3 material after the reaction in the presence of AgNO3
clearly show that Ag deposition occurs around the CrxOy
particles, and also the oxidation of Cr takes place (in comparison
to reference spectra the initial +3 oxidation number of Cr
increases, although it is not exactly +6, Fig. S11).26 Similarly to
Cr, small changes in the electronic state of Rh were indicated
after Ag deposition from the Rh K-edge XANES (Rh oxidation
number is slightly oxidized, although initially is not exactly 3+,
Fig. S12),26 although Rh and its spatial distribution could not be
studied by TEM due to the low Rh loading (0.5 wt%). Thus RhCr, particularly Cr, is the site where electrons are released for
reduction reactions and hence Ag is selectively deposited on Cr
oxide particles. The reduction reaction proceeds in a sacrificial
manner for Cr (i.e. Cr is self-oxidized) and thus Ag reduction is
efficient through the reducing power of CrxOy sites. This explains
the lowered ability of the catalyst for charge separation and
thus leading to less efficient electron trap after Ag deposition at
Rh-Cr sites. Furthermore, the lowered efficiency for charge
separation by ineffective function of Rh-Cr by Ag deposition
explains the suppressed oxygen formation in the presence of
Ag+ (Fig 5a vs. 5b).
Ag deposition under illumination (i.e. photodeposition) is
shown to be site selective and it is expected to uniquely
influence the water splitting activity. This was verified by
performing the photocatalytic water splitting tests in pure
water using the material after Ag photodeposition for 30 min
(20 wt% Ag determined by EDX) and using a comparable
material prepared by the impregnation method at the same Ag
loading (Fig. 5c and 5d). In all cases, lower activity is found in
comparison to the materials without Ag deposition (Fig. 5a). The
most important difference is observed for the Rh-Cr promoted
catalyst where Ag-impregnated catalyst showed relatively high
H2 productivity (4.9 mmol gcat-1 h-1) in contrast to the minor
activity of the Ag-photodeposited one (0.8 mmol gcat-1 h-1).
Microscopic study shows that the Ag deposition by the
impregnation method leads more homogenous distribution of
Ag and not site-selective (Fig. S13). Thus, the poisoning of the

active sites is not prominent. On the other hand, the catalytic
activity is largely suppressed when Ag selectively covers Rh-Cr
co-catalyst (speculated for Rh).

Fig. 5 H2 (green) and O2 (purple) productivity of 4 wt% Zn-Ga2O3
with and without Rh-Cr in the photocatalytic water splitting
reaction; (a) in pure water, (b) in the presence of 0.01 M AgNO3,
(c) in pure water after 20 wt% Ag-photodeposition and (d) in
pure water after 20 wt% Ag-impregnation. The duration of
photocatalytic test was 1.5 h (averaged from 4 runs) except for
the case in the presence of AgNO3 (b) for which the test was
performed only for 30 min due to fast deactivation induced by
Ag deposition on the catalysts. The small difference in the
catalytic activity (a) and those shown in Fig. 1 is due to the
different batch of catalysts synthesized and possibly to the
experimental error of different photocatalytic tests.
Nevertheless, the difference is within the levels of common
tolerance (about 10%).
Synergistic effects of Rh-Cr and Zn
Fig. 5 also shows that O2 productivity is fully suppressed
when the function of Rh-Cr is disabled by the Ag deposition,
affirming that the charge separation facilitated by the role of
Rh-Cr as electron sink (reduction site) and the function of Zn
through the formation of ZnGa2O4/Ga2O3 heterojunction (Fig. 4)
are interlinked, as also evidenced by the synergistic effects of
Zn and Rh-Cr in the photocatalytic activity (Fig. 1). With the aim
to directly sense the influence of Rh-Cr co-catalyst on the
electronic structure of Zn atoms as a proof of the synergistic
interactions, here the unoccupied state electronic structures of
Zn are studied by XANES with in situ illumination.
As expected, the XANES difference spectra of Zn-Ga2O3 and
Rh-Cr/Zn-Ga2O3 between the dark and illuminated conditions
are very small, close to the level of noise even after signal to
noise (S/N) improvement by averaging the respective spectra
before subtraction (Fig. 6a). To drastically boost the sensitivity,
the experiments are performed periodically in light-on/off
cycles and the resulting spectra were averaged into one lighton/off cycle. Then the averaged spectra were treated by the
mathematical engine of modulation excitation spectroscopy,
phase sensitive detection (PSD) (Fig. 6b and Fig. S14).25 PSD

highlights only the intensity which varies at the same frequency
as the light on/off stimulus, thus effectively eliminates
uncorrelated noise and time-independent static signals. Only
the phase-domain spectra at φPSD = 0 are presented since this
would be equivalent to the difference spectra of the catalyst
under-light minus under-dark. The improvement in S/N is
evident, although the spectral features are very small. The
physical nature of the small spectral characteristics is verified by
the so-called in-phase angle analysis (Fig. S14-iii), clearly
showing distinct phase angles for each peak (continuously
shifting phase angles are due to peak overlaps as expected in
XANES, also serving as the proof of the physical nature of the
low intensity signals and not of noise).
Two important observations can be made from this lighton/off XANES study. First, the electronic structure of Zn atom,
which is present as ZnGa2O4 (vide supra), is perturbated by the
illumination. This is an indirect evidence of the charge
separation at ZnGa2O4 by which the electronic structure of Zn is
also influenced to vary. Another very striking observation is the
effect of Rh-Cr on the degree of the change. The degree of
change in the Zn electronic structure is about double by the
presence of Rh-Cr co-catalyst. Detailed spectral analysis (i.e.
how the Zn orbitals are altered) is beyond the scope of this
work; nevertheless, the clearly different spectral features (e.g.
where the maxima appear for the two materials) indicate the
electronic structure of Zn atom under illumination is different
with/without Rh-Cr, possibly due to the different degree of
charge separation. This is the first and a clear proof that the
charge separation facilitated by Rh-Cr communicates with the
electronic structure of Zn. Also, the results imply that the
physical vicinity of Rh-Cr is important and how and where Rh-Cr
are deposited (e.g. on Ga2O3, ZnGa2O4 or at their interface)
matter for the photocatalytic activity.

Conflicts of interest
There are no conflicts to declare.

Acknowledgements
We thank the Generalitat de Catalunya for financial support
through the CERCA Programme and recognition (2017 SGR
1633) and MINECO (CTQ2016-75499-R (AEI/FEDER-UE)) for
financial support and support through Severo Ochoa Excellence
Accreditation 2014-2018 (SEV-2013-0319). We acknowledge
the European Synchrotron Radiation Facility for provision of
synchrotron radiation facilities. We thank Belén Ballesteros
from Institut Català de Nanociència i Nanotecnologia (ICN2) for
the support in microscopy and EDX measurements.

References
1.
2.
3.

Fig. 6 a) Averaged Zn K-edge HERDF-XANES spectra, light off
subtracted from light on cycle, and b) phase-domain response
with k = 1 of 4 wt% Zn-modified Ga2O3 (grey), and Rh-Cr/4 wt%
Zn-Ga2O3 (brown).

Conclusions
Zn and Rh-Cr co-catalysts greatly influence the
photocatalytic water splitting activity of Ga2O3 and their
synergistic function is confirmed. At the optimum Zn-loading (24 wt%) the formation of the ZnGa2O4/Ga2O3 heterojunction,
with an appropriate exposure of both materials for the redox
reactions, is favoured facilitating the charge transfer and
separation between the two semiconductor materials and thus
enhancing the photocatalytic activity. At higher Zn-loading (≥20
wt%) ZnO is formed covering the ZnGa2O4 and Ga2O3 surface
and negatively influences the photocatalytic activity due to the
light absorption properties and water splitting activity of ZnO.
The role of Rh-Cr as electron trapping sites is evidenced by the
studies in the presence of Ag+, clearly showing the poisoning of
the Rh-Cr active sites after Ag deposition. By depositing Ag in a
selective (photodeposition) and non-selective (impregnation)
fashions and detailing the electronic structure change of Cr, the
function of Rh-Cr co-catalyst in water splitting reaction was
affirmed. Finally, the concerted actions between the electron
sink (Rh-Cr) and the ZnGa2O4/Ga2O3 heterojunction are proven
by in situ illumination modulation excitation XAS spectroscopy.
The subtle electronic structure change of Zn atoms in ZnGa2O4
under illumination is uncovered. The degree of the change is
boosted by the presence of Rh-Cr, serving as a direct proof of
communication and synergy of Zn and Rh-Cr co-catalysts on the
electronic structure level under photocatalytic conditions.

4.
5.
6.
7.
8.
9.
10.

11.

12.
13.
14.

15.

16.
17.
18.
19.

M. D. Hernandez-Alonso, F. Fresno, S. Suarez and J. M.
Coronado, Energ. Environ. Sci., 2009, 2, 1231-1257.
A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253-278.
K. Maeda and K. Domen, J. Phys. Chem. Lett., 2010, 1, 26552661.
A. Fujishima and K. Honda, Nature., 1972, 238, 37-38.
A. J. Bard and M. A. Fox, Accounts. Chem. Res., 1995, 28,
141-145.
A. L. Linsebigler, G. Lu and J. T. Yates, Chem. Rev., 1995, 95,
735-758.
K. Maeda and K. Domen, J. Phys. Chem. C., 2007, 111, 78517861.
F. E. Osterloh, Chem. Mater., 2008, 20, 35-54.
G. Liu, L. Wang, H. G. Yang, H.-M. Cheng and G. Q. Lu, J.
Mater. Chem., 2010, 20, 831-843.
T. Yanagida, Y. Sakata and H. Imamura, Photocatalytic
Decomposition of H2O into H2 and O2 over Ga2O3 Loaded
with NiO, 2004.
K. Maeda, K. Teramura, H. Masuda, T. Takata, N. Saito, Y.
Inoue and K. Domen, J. Phys. Chem. B., 2006, 110, 1310713112.
K. Maeda, K. Teramura, D. Lu, T. Takata, N. Saito, Y. Inoue
and K. Domen, Nature., 2006, 440, 295.
K. Maeda, K. Teramura, D. Lu, N. Saito, Y. Inoue and K.
Domen, Angew. Chem. Int. Edit., 2006, 45, 7806-7809.
Y. Sakata, Y. Matsuda, T. Nakagawa, R. Yasunaga, H.
Imamura and K. Teramura, ChemSusChem., 2011, 4, 181184.
Y. Sakata, T. Nakagawa, Y. Nagamatsu, Y. Matsuda, R.
Yasunaga, E. Nakao and H. Imamura, J. Catal., 2014, 310,
45-50.
K. Shimura, T. Yoshida and H. Yoshida, J. Phys. Chem. C.,
2010, 114, 11466-11474.
Y. Sakata, Y. Matsuda, T. Yanagida, K. Hirata, H. Imamura
and K. Teramura, Catal. Lett., 2008, 125, 22-26.
A. Bazzo and A. Urakawa, Catal. Sci. Technol., 2016, 6,
4243-4253.
P. Glatzel and U. Bergmann, Coordin. Chem. Rev., 2005,
249, 65-95.

20.

22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.

38.
39.
40.
41.

J. Szlachetko and J. Sá, CrystEngComm., 2013, 15, 25832587. 21. E. Gallo and P. Glatzel, Adv. Mater., 2014, 26,
7730-7746.
R. López and R. Gómez, J. Sol-Gel. Sci. Techn., 2012, 61, 17.
A. B. Murphy, Sol. Energ. Mat. Sol. C., 2007, 91, 1326-1337.
K. Wojtaszek, K. Tyrala, J. Czapla-Masztafiak, J. Sa and J.
Szlachetko, Chem. Phys. Lett., 2016, 664, 73-76.
A. Urakawa, T. Bürgi and A. Baiker, Chem. Ehg. Sci., 2008,
63, 4902-4909.
K. Maeda, K. Teramura, D. Lu, T. Takata, N. Saito, Y. Inoue
and K. Domen, J. Phys. Chem. B., 2006, 110, 13753-13758.
L. Lutterotti, R. Ceccato, R. Dal Maschio and E. Pagani,
Mater. Sci. Forum., 1998, 278-281, 87-92.
B. Ravel and M. Newville, J. Synchrotron. Radiat., 2005, 12,
537-541.
T. Ressler, J. Wong, J. Roos and I. L. Smith, Environ. Sci.
Technol., 2000, 34, 950-958.
D. Baurecht and U. P. Fringeli, Rev. Sci. Instrum., 2001, 72,
3782-3792.
A. Urakawa, T. Bürgi and A. Baiker, Chem. Phys., 2006, 324,
653-658.
X. Chen, S. Shen, L. Guo and S. S. Mao, Chem. Rev., 2010,
110, 6503-6570.
S. Tabata, H. Nishida, Y. Masaki and K. Tabata, Catal. Lett.,
1995, 34, 245-249.
X. Wang, S. Shen, S. Jin, J. Yang, M. Li, X. Wang, H. Han and
C. Li, Phys. Chem. Chem. Phys., 2013, 15, 19380-19386.
J. C. Hill and K.-S. Choi, J. Phys. Chem. C., 2012, 116, 76127620.
Z. Wang, K. Teramura, Z. Huang, S. Hosokawa, Y. Sakata
and T. Tanaka, Catal. Sci. Technol., 2016, 6, 1025-1032.
P. Glatzel, T. C. Weng, K. Kvashnina, J. Swarbrick, M. Sikora,
E. Gallo, N. Smolentsev and R. A. Mori, J. Electron.
Spectrosc., 2013, 188, 17-25.
T. Inoue, A. Fujishima, S. Konishi and K. Honda, Nature.,
1979, 277, 637.
J. Disdier, J.-M. Herrmann and P. Pichat, J. Chem. Soc.
Farad. T. 1., 1983, 79, 651-660.
K. Maeda, H. Hashiguchi, H. Masuda, R. Abe and K. Domen,
J. Phys. Chem. C., 2008, 112, 3447-3452.
B. Ohtani, S. Zhang, J. Handa, H. Kajiwara, S. Nishimoto and
T. Kagiya, J. Photoch. Photobio. A., 1992, 64, 223-230.

Supporting Information

Synergistic interplay of Zn and Rh-Cr promoters on
Ga2O3 based photocatalyst for water splitting

Marta Borges Ordoñoa, Shunsaku Yasumuraa, Pieter Glatzelb,
and Atsushi Urakawa*a
aInstitute

of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology,
Tarragona, 43007, Spain
bESRF

- The European Synchrotron, Grenoble, 38043, France
*E-mail: aurakawa@iciq.es

MATERIALS CHARACTERIZATION
UV-Vis DRS measurements were performed on a Shimadzu UV-2401PC spectrophotometer with
D2 and W lamps as light sources, and a photomultiplier detector. Solid samples were introduced in a round
holder with a fused silica window on an integrating sphere attachment ISR-240A from Shimadzu. BaSO4
was used as internal standard and spectra were collected between 240 and 800 nm. Band gap (Eg) was
calculated from the slope intersection of the normalized Kubelka-Munk absorption spectrum.
Scanning transmission electron microscopy (STEM) was measured in annular dark-filed imaging
mode (HAADF) in a high-resolution FEI Tecnai F20 STEM microscope at ICN2 (Barcelona). Chemical analysis
and chemical mapping were performed by energy dispersive X-ray spectroscopy (EDX) and electron
energy loss spectroscopy (EELS), respectively. 4 wt% Zn-Ga2O3 and Rh-Cr/4wt% Zn-Ga2O3 materials
prepared in ethanol solution and were deposited on the a TEM support grid.
Cr and Rh elements were studied at their corresponding K-edge energies 5.99 and 23.22 keV,
respectively. HERFD-XANES were measured from 5.98 to 6.08 keV (Cr) and 23.18 to 23.40 keV (Rh); with
a step size for Cr of 0.1 eV and 0.05 eV for Rh. The spectrometer was equipped with four spherical Ge
(333) crystal for Cr, and a diode IF3 with Cu+Ni filter was used for Rh detection. All the samples and
reference materials were measured ex situ in the form of pellet (13 mm diameter), containing 60 mg of
material and 60 mg of cellulose.

Fig. S1 (left) STEM-HAADF image of 4 wt% Zn-Ga2O3 material, with the transversal line indicating where
EDX was measured. (right) Elemental profiles of Ga (top) and Zn (bottom) along the transversal line.

Fig. S2 Quantitative phase analysis of the XRD data for xZn-modified Ga2O3 photocatalysts where x= 2, 4,
20, 50, and 70 wt% of Zn.

Table. S1 Relative shift calculated from Kβ lines using the center of mass for Zn-Kedge (top) and Ga-Kedge
(bottom).
Sample

Kβ

Center of mass
(eV)

Relative shift
(eV)

ZnO

1
2

9571.863
9571.863

0

1
2

9571.832
9571.832

0

3

1
2

9571.842
9571.844

0.002

3

1
2

9571.842
9571.842

0

3

1
2

9571.855
9571.852

0.003

3

1
2

9571.854
9571.859

0.005

3

1
2

9571.853
9571.854

0.001

Kβ

Center of mass
(eV)

Relative shift
(eV)

1

10264.09

2

10264.09

1

10264.07

2

10264.07

1

10264.11

2

10264.12

1

10264.11

2

10264.11

1

10264.13

2

10264.13

1

10264.12

2

10264.13

1

10264.06

ZnGa O
2

4

2 wt% Zn-Ga O
2

4 wt% Zn-Ga O
2

20 wt% Zn-Ga O
2

50 wt% Zn-Ga O
2

70 wt% Zn-Ga O
2

Sample

Ga2O3

0

ZnGa2O4

0

2 wt% Zn-Ga2O3

0.01

4 wt% Zn-Ga2O3

0

20 wt% Zn-Ga2O3

0

50 wt% Zn-Ga2O3

70 wt% Zn-Ga2O3

0.01

0

2

10264.06

Fig. S3 Comparison of successive scans of Kβ main line for Zn K-edge a) ZnO and b) ZnGa2O4 and Ga Kedge c) Ga2O3 and d) ZnGa2O4.

Fig. S4 XANES for Zn K-edge and Ga K-edge comparing HERFD scans (a and c) with total fluorescence scans
(b and d).
The performance of density functionals PBE and PBE0 was compared, but due to the high
computational cost of PBE0, PBE was applied to perform the cell optimization of the ZnO, Ga2O3, and
ZnGa2O4 structures. First, the atomic positions in the corresponding cells and lattice parameters were
optimized using PBE. After this, the energy calculations were performed by PBE0 to obtain more accurate
DOS with the atomic positions and lattice parameters calculated above.
Table S2. DOS calculations details
Package
CASTEP
Conditions of Cell optimization
Functional
Energy tolerance (eV/atom)
Max. force(eV/Å)
Max. stress (Gpa)
Max. displacement (Å)
Energy cutoff (eV)
K-mesh
SCF tolerance (eV/atom)

ZnO
PBE
1.00E-05
0.03
0.05
0.001
750
5x5x4
1.00E-06

Ga2O3
PBE
1.00E-05
0.03
0.05
0.001
800
5x5x2
1.00E-06

ZnGa2O4
PBE
1.00E-05
0.03
0.05
0.001
800
3x3x3
1.00E-06

ZnO
PBE0
600
4x4x2
2.00E-06

Ga2O3
PBE0
600
4x4x2
2.00E-06

ZnGa2O4
PBE0
600
3x3x3
2.00E-06

Conditions of SCF by PBE0
Functional
Energy cutoff (eV)
K-mesh
SCF tolerance (eV/atom)

The coverage or phase transformation of ZnGa2O4 into ZnO could be further determined by looking
deeper into the Zn K-edge VtC-XES region. Typically, VtC spectra has two lines corresponding to Kβ2,5
(located below the Fermi level) and Kβ” (cross over) at lower fluorescence energy.1 At low Zn
concentrations (2 and 4 wt%) the Kβ” line was split because of the interaction from the metal (Zn) with s
orbitals of O and d orbitals of Ga (Fig. S5), this result was supported by the density of states calculations
on the reference materials (details of calculation shown in Table S2). On the contrary, at high Zn loadings
where ZnO phase was detected the Kβ” line only contains the contribution of s orbitals from O, indicating
that first coordination shell of Zn atoms does not contain Ga atoms which might be attributed to the
reduced accessibility of Zn atoms to the Ga2O3 surface, in agreement with the EXAFS analysis (Fig. S6). In
addition, the third emission line due to multielectron transitions (ME) from the emitted and absorbed
photoelectrons is observed in the Zn K-edge VtC spectra (Fig. S5). The intensity of this ME region is located
above the Zn absorption edge energy and decreases with ZnO formation following the same trend that
the photocatalytic activity results. However, correlation with the catalytic results are not discussed here
because of the self-absorption effects that could also contribute to the ME intensity.2, 3

Fig. S5 Zn K-edge VtC-XES with normalized intensity to the spectral area (top). PDOS calculations for O,
Ga, and Zn where mainly s and p regions correspond to the Kβ” and Kβ2,5 lines, respectively. The dashed
blue line indicates the Fermi level, and the zoomed region corresponds to Kβ”.

Fig. S6 EXAFS analysis of low Zn content Ga2O3 (2 wt%) compared to ZnO (dotted line) reference material.
Self-absorption correction was performed on the ZnO EXAFS data by using the Fluo algorithm (Haskel,
1999).

Fig. S7 UV-Vis DR spectra of xZn-Ga2O3 materials with x = 2 (black), 4 (cyan), 20 (maroon), 50 (blue) and
70 (green) and of reference materials, ZnGa2O4 (red), Ga2O3 (dotted line) and ZnO (dashed line). Spectra
are shown in normalized Kubelka-Munk absorbance unit and the bandgaps were calculated from the slope
of the adsorption spectra. Zoomed plot shows the region highlighting light absorption by ZnO.

For both Zn K-edge (Fig. S8a) and at Ga K-edge (Fig. S8b), opposite trends are observed for
the unoccupied energy levels at the Zn and Ga absorption edges (inset plots, Fig. S8). For the Zn
K-edge, at small Zn concentrations (2 and 4 wt%) the lowest-unoccupied electronic sates appear
to be comparable to those of ZnGa2O4 (Fig. S8a), while at increasing Zn-loading where ZnO phase
is more prominent, the absorption edges shift to lower energies, corresponding to smaller
valance-conduction band gaps. This agrees with the UV-Vis DRS study (Fig. S7), at high Zn loading
(≥20 wt%) a weak absorption extending in the visible region appears clearly due to the presence
of ZnO, with the optical bandgap of 3.0-3.1 eV.
In contrast, the lowest-unoccupied energy states of Ga are shifted toward higher energies
at higher Zn content, pure ZnGa2O4 shows the highest lowest-unoccupied energy state (Fig. S8b).
The small shift of the unoccupied energy states of Ga at low Zn concentrations (e.g. energy
difference between Ga2O3 and 4 wt% Zn-Ga2O3 is 0.06 eV) is likely attributed to the minor
contribution of ZnGa2O4 phase for such Zn loadings.

Fig. S8 XAS and VtC-XES spectra for a) Zn K-edge and b) Ga K-edge. Pure ZnGa2O4 (dotted line),
ZnO (dashed line), and Ga2O3 (dot-dashed line). Spectral intensity was normalized to the spectral
area. Inset figures show the opposite energy changes of the absorption edge.

Fig. S9 First derivative of XAS and VtC-XES spectra for Zn K-edge (top) and Ga K-edge (bottom).
Red and yellow lines are positioned at the intersection between baseline at 0 intensity and the
slope of the first derivative of XAS, and the intersection from the baseline for XAS and XES was
used to calculate the difference between the XAS and XES spectra.

Fig. S10 STEM-HAADF image of Rh-Cr/4 wt% Zn-Ga2O3, and elemental profiles over the diagonal for Ag
(black), Cr (blue), and Ga (pink). Rh was not detected likely due to its low content.

Fig. S11 XANES spectra at Cr K-edge; Cr metal (dashed line), Rh-Cr/4 wt% Zn-Ga2O3 (cyan), and after silver
photodeposition (black).

Fig. S12 XANES spectra at Rh K-edge; Rh-Cr/4 wt% Zn-Ga2O3 (cyan), and after silver photodeposition
(black). Rh2O3 (red) and metallic Rh (dashed lines) were used as reference materials.

Fig. S13 STEM-HAADF images of 20 wt% Ag deposited on Rh-Cr/4wt% Zn-Ga2O3 after photodeposition
(left) and after impregnation (right).

Fig. S14 Zn-Kedge HERDF-XANES phase-domain spectra and in-phase angles analysis for a) 4wt% Zn-Ga2O3
and b) Rh-Cr/4wt% Zn-Ga2O3. The amplitude threshold to plot the in-phase angles was 0.0001.

References

1.
2.
3.

U. Bergmann, C. R. Horne, T. J. Collins, J. M. Workman and S. P. Cramer, Chem. Phys. Lett., 1999, 302, 119-124.
D. R. Mortensen, G. T. Seidler, J. J. Kas, N. Govind, C. P. Schwartz, S. Pemmaraju and D. G. Prendergast, Phys. Rev. B., 2017, 96,
125136.
R. A. Valenza, E. P. Jahrman, J. J. Kas and G. T. Seidler, Phys. Rev. A., 2017, 96, 032504.