FULL PAPER
Multimetallic Oxynitrides Nanoparticles for a New Generation of
Photocatalysts
Tingke Rao
a
, Maria Luisa Saladino
b
, Yuanxing Fang
c
, Xinchen Wang
c
, Cristina Giordano
*,a
Abstract: A versatile synthetic strategy for the preparation of multi-
metallic oxy-nitrides has been designed and here exemplarily
discussed considering the preparation of nanoscaled zinc-gallium
oxy-nitrides and zinc-gallium-indium oxy-nitrides, two important
photo-catalysts of new generation, proven to be active in key energy
related processes from pollutant decomposition to overall water
splitting. The synthesis presented here allows the preparation of small
nanoparticles (less than 20 nm in average diameter), well-defined in
size and shape, yet highly crystalline and with the highest surface
area reported so far (up to ~80 m
2
/g). X-rays diffraction study shows
that the final materials is not a mixture of the single oxides but a
distinctive compound. The photocatalytic properties of the oxy-nitrides
have been tested toward the decomposition of an organic dye (as a
model reaction for the decomposition of air pollutants), showing better
photocatalytic performances than the corresponding pure phases
(reaction constant 0.22 h
-1
), while almost no reaction was observed in
absence of catalyst or in the dark. Photocatalysts have been also
tested for H
2
evolution (semi-reaction of the water splitting process)
with results comparable to the best literature values but leaving more
room for further improvement.
Introduction
In a world in constant evolution, scientists have the delicate issue
to provide suitable materials to fulfil the contemporary necessities.
This is especially important in energy related matter, where valid
alternatives to current systems must still be found. In the search
for novel materials two main strategies can be pursued: improving
already known materials or exploring novel possibilities. In this
respect, nanosized materials represent an eclectic world, where
one can draw on and where old and new materials still await to
be (re)discovered. Thanks to their high surface area and tuneable
properties depending on their size, nanomaterials can be
considered new functional systems. Among these, the so-called
metal oxy-nitrides (MON) have been receiving increasing
attention after the discovering that their electronic and optical
properties can be adjusted by changing the band gap. Thus,
oxides possess wide band gaps and many of them are insulators,
while the band gaps of metal nitrides are narrower and the
materials usually show metallic behaviour. The band gaps of
MON have intermediate values, nicely placed between pure MO
x
and MN and tunable by N-loading contents. Domen et al. have
shown that the incorporation of N in TaON based systems can
deeply change their optical and electronic properties, where a
systematic increase of the N loading in the crystal structure can
reduce the band gap
[1]
. Since then, more complex compositions
have been explored, going from simple mono-metallic
compounds to designed multi-metallic systems. Properties of the
metal oxy-nitrides arise from the simultaneous presence in the
crystal lattice of nitride and oxide ions, where the substitution of
O
2-
with N
3-
induces the substitution of M
2+
with M
3+
and possible
anions vacancies formation, to keep neutrality
[2]
.
Thanks to this peculiar structure, where the orbitals 2p orbitals of
N and/or O overlap the s, p, d orbitals of the metals causing the
dispersion of the valence and conduction bands
[3]
, MON are
photo-responsive and are thus considered photocatalysts of new
generation, where the electronic structure, including band gap,
can be adjusted by tuning the ratio of the compositions.
Photo-catalysts are semiconductor materials able to be activated
by exposure of suitable wavelength, usually in the UV-vis range.
The formation of a hole/electron pair, upon light absorption, aids
the processing of a reaction of interest. Domen et al.
[4]
have nicely
shown that GaN:ZnO solid solutions are active photo-catalysts for
overall water splitting, a key processes that might alleviate the
energy crisis by producing high potential molecules, such as H
2
,
in a clean way. More recently, also Domen group
[5]
, has shown
that ZnGaON in ZnO:ZnGaON nanowire-array-on-a-film photo-
anode could bring to stable and efficient sunlight water splitting
without the assistance of any co-catalyst. By adding zinc and
oxygen dopants, the quaternary nanoparticles possess a band
gap of 2.6-2.8 eV
[4]
, which is narrower than the band gap of pure
GaN nanoparticle (3.4 eV)
[4]
or ZnO (3.25 eV) and strongly
depends on the zinc/gallium ratio (increasing by decreasing the
metals’ ratio)
[4]-[6]
.
A band gap energy of 2.13 eV was achieved for a Zn-rich Ga
oxynitride with Zn/Ga = 0.9/0.110. The Zn
x
Ga
1-x
O
x
N
1-x
solid
solution with x=0.20 exhibited a band gap energy of around 2.43
eV.
[6]
The band gap energy was smaller than those for GaN and
ZnO, and has been interpreted as an enhanced valence-band
maximum caused by a repulsion of N-2p and Zn-3d orbitals
[4]
.
One of the ways of shifting the photo-catalytic activity from the UV
to the visible region is by narrowing the band gap, which can be
obtained by doping In into the ZnGaON crystal structure. The
band gaps decrease from GaN to InN with the cation atomic
conductive d orbital energies decreasing.
[7]
Due to the InN smaller
band gap (0.7 eV)
[8]
, the zinc-gallium-indium oxynitride (ZGIN)
[a] Tingke Rao, Dr. Cristina Giordano
School of Biological and Chemical Science
Chemistry Department, Queen Mary University of London, Mile End
Road, London E1 4NS, United Kingdom
E-mail: c.giordano@qmul.ac.uk
[b] Prof. Maria Luisa Saladino
Dipartimento Scienze e Tecnologie Biologiche, Chimiche e
Farmaceutiche - STEBICEF and INSTM UdR - Palermo, Università
di Palermo, Viale delle Scienze pad. 17, Palermo, I-90128 Italy
[c] Prof. Xinchen Wang, Prof Yuanxing Fang, State Key Laboratory of
Photocatalysis on Energy and Environment, College of Chemistry,
Fuzhou University, Fuzhou 350002, P. R. China
Supporting information for this article is given via a link at the end of
the document.
FULL PAPER
nanoparticle is expected to have smaller band gap and thus a
greater photo-catalytic activity.
Tuning the band gap is especially important as it enables Zn-Ga-
oxynitride nanoparticle to harvest the most of incident visible light
in photo-catalytic reactions, while the introduction of Zn increases
the photo-activity, probably by reducing the number of defects (i.e.
the probability of electron-hole recombination) and by increasing
the crystallinity of the final material
[4]
. Therefore, high crystallinity
of the final material is a key property for an improved activity. The
other challenge to face by preparing Zn- and In-based material is
to avoid their volatilization, which also reduces photo-activity.
[9]
Currently, the ZnO:GaN solid solutions are always prepared by
nitridation of the mixed oxide commonly via co-precipitation and
more recently via sol-gel based process, however so far always
under NH
3
flow as nitrification agent.
[9], [10]
Used starting materials are the single or mixed oxides
[11], [12], [13],
[14]
, hydroxide
[15]
, gel-like precursors
[16]
, Zn/Ga/CO
3
layered
double hydroxides
[17]
or ammonolysis of amorphous precursors
[18]
.
Even when urea was used as nitrogen source, the synthesis
required the preparation of hydroxides, which were treated at
900 °C under ammonia flow
[19]
. To enhance the Zn content a
topotactic transformation was employed but under dangerous
reaction conditions (ammonia in the presence of oxygen)
[20]
. In
both cases, the final product was morphologically not very well-
defined.
Less rich is the list of papers reporting the synthesis of ZGION
nanoparticles and it includes the work of Kamata et al.
[9]
and
Miyaake et al.
[21]
both using nitridation of the metal oxide under
ammonia flow and in both cases with nanoparticles not well-
defined or poorly crystalline or too big. However, NH
3
is toxic and
most of the proposed synthesis are not suitable for large scale
production.
In the present work we prepare multi-metallic oxynitrides in an
easier way, via the so-called Urea-Glass-Route (UGR), which
allows the incorporation of more than one metal into the oxy-
nitrides structure, does not involve the prior preparation of the
pure oxides or nitrides and does not require the use of ammonia.
The starting material is a gel-like precursor, where the
composition can be chosen as whished and, upon suitable heat
treatment under N
2
flow, leads to a homogeneous system of small
yet well-defined (in size and shape) nanoparticles, highly
crystalline and with very high surface area (up to ~80 m
2
/g). The
lower temperature used helps to hinder Zn and In volatilization.
For a better comparison, also pure ZnO and GaN have been
prepared with the same procedure. This is, at the best of our
knowledge, the first time that such versatility is achieved with one
synthetic path.
For simplicity, the wording “ZGON” and “ZGION”, will be used in
the following to indicate the simultaneous presence of Zn, Ga or
Zn, Ga, In atoms in the final oxy-nitride. The nominal ratio
between the metals will be always indicated as R
M
or R’
M
(R
M
=Ga/Zn or R’
M
=Ga/In).
Results and Discussion
The thermally treated samples have been studied via powder X-
rays diffraction (XRD). In Figure 1, the patterns of the metallic
binary and ternary samples are reported (ZGON, R=10 and R
M
=2;
ZGION, R=10, R
M
=2 and R’
M
=2) alongside those for pure ZnO
and GaN (also prepared via UGR). The pattern obtained from the
physical mixture (prepared grinding together the ZnO and GaN
powders) can be described by the two separated crystalline
phases, also reported for comparison. From this figure it can be
seen that both GaN and ZnO phase prepared via UGR nicely
match those expected from the database (see also figure SI.1).
This result is not surprising for GaN, which has been already
prepared via UGR
[22], [23]
but ZnO was prepared following this
route here for the first time. The usefulness of the UGR to prepare
metal oxide nanoparticles was also recently shown.
[24]
The bi- and
tri-metallic phase is also present as hexagonal (wurtzite type)
phase with different features corresponding to the (100), (002),
(101), (102), (110), (103), (112), and (201) crystallographic planes.
However, it differs from the pure GaN phases both in peaks
position and/or intensities. No XRD features corresponding to
ZnO phase and mixed phase are observed, confirming the
insertion of Zn in the GaN lattice.
The peaks of the multi-metallic phase are in fact observed
between those of the pure phases and are very different from
those of the physical mixture, where the overlapping of the ZnO
and GaN peaks can be easily noted. This finding clearly indicate
that the multi-metallic phases are not a physical mixture of the
pure oxide and nitride but form a distinct compound.
10 20 30 40 50 60 70
Intensity (a.u.)
2
θ
GaN ICCD 00-050-0792
ZnO ICCD 01-070-8072
Figure 1. XRD pattern of ZGON (R=10 and R
M
=2) and ZGION (R=10 and R
M
=2,
R
M
’=2) compared with the pattern of pure GaN and ZnO powder, also prepared
via UGR at 750°C. The pattern of the physical mixture powder (ZnO+GaN) is
also shown for comparison. The black vertical dotted lines in the inset are just
guidelines for eyes to show the peak shift.
We have observed that for lower R
M
(namely R
M
=1, i.e. equimolar
ratio between the metals) peaks position is similar to the one of
pure GaN (figure SI.2), while a higher R
M
(i.e. higher amount of
gallium) shifts to the peaks at lower angles (compare to those of
FULL PAPER
pure GaN). This was previously observed and attributed to the
inclusion in the lattice of bigger Zn
2+
ions (compared to the smaller
Ga
3+
ions) that leads to its expansion.
[25]
On the other hand, the
further incorporation of indium ions seems to have a minor effect
on the position of peaks, possibly due to the little In loading). XRD
patterns of ZGION samples prepared with different In loading are
reported in the SI (figure SI.3). In fact, ZGION peaks are observed
at lower angles compared to pure GaN as expected (expanded
structure due to incorporation of bigger ions, but higher angles
compared to ZGON, which is somehow surprising considering
that In
+3
is bigger than Ga
+3
and Zn
+2
(80, 62 and 74 pm,
respectively).
XRD patterns have been analysed by means of Rietveld
method.
[26], [27]
The results of cell parameter for the GaN phase
and crystallite size and lattice microstrain are reported in Table
SI.1. The trend of cell parameters cannot be taken into account
considering the crystallographic structure of GaN (wurtzitic)
[28], [29]
,
and interesting the cell parameter a seems to be affected more by
R
M
than R.
It worth to note the fact that no secondary phases or residual
products are observed, including the pure mono-metallic phases
or the normally observed spinel phases (ZnGa
2
O
4
). In order to
have information on the reaction mechanism, samples were
quenched at lower temperature and the corresponding XRD
patterns are reported in Figure 2 (for R
M
=2) and figure SI.4 (for
R
M
=1). These study clearly indicated that for T<650°C (regardless
of R
M
), the structure is closer to that of pure ZnO (although peaks
of a secondary phase are also observed), somehow suggesting
that the formation of the oxide phase takes place first and then
the incorporation of Ga ions follows in a second step and only at
temperature above 700 °C. Similar result has been observed by
Armetta et al.
[30]
for the formation of yttrium aluminum garnet
(YAG) nanoparticles prepared by co-precipitation in
microemulsion and calcining at high temperature by the Kirkendall
effect.
[31]
The formation of an oxide intermediate phase using
UGR has been also previously observed for the synthesis of
TiN.
[22]
10 20 30 40 50 60
T=750C
T=650C
Intensity (a.u.)
2
Θ
ZnO
GaN
T=550C
R
M
=2
Figure 2. XRD pattern of ZGON synthesized from R=10 and R
M
=2 and treated
at different temperature (550, 650 and 750°C). The pattern of ZnO and GaN
from the database are also shown for comparison as vertical lines.
The influence of urea/metal molar ratio (R) was also studied; this
ratio was in fact found to be decisive to control not only
composition but also crystallinity and morphology.
[32], [33]
The XRD
patterns of the bimetallic phase prepared with three different
urea/metal ratio and R
M
=2 (figure SI.5) shows that for lower ratios
(R ≤ 5) the typical wurtzite-like pattern is formed but alongside a
secondary phase. For R ≥ 10 the wurtzite-like phase becomes
more defined and the secondary phase is no longer observed.
Further increase in R (R ≥ 15) leads to sharper peaks (more
crystalline structure). It must be noted that the secondary phase
observed for R=5 is not the same phase observed in samples
quenched at lower temperature (see figure SI.6), which was
attributed to ZnGa
2
O
4
(ICDD 96-400-1768).
The effect of larger amounts of urea (higher R) seems to have
little impact on the final crystallite size (assumed to be spherical
and estimated by the Scherrer equation, see table 1) but
precursor’s composition has a decisive effect on the morphology
of the metallic binary phase and homogeneity. Figure 3 shows
SEM micrographs of samples prepared at different urea/metals
ratio. In each case, homogeneous system can be observed,
however, increasing R, seems to bring more defined and just
slightly bigger particles, which would explain the smaller surface
area found for the R=15 sample.
Figure 3. SEM micrographs of metallic binary phase synthesized at R
M
=2 and
urea ratio of: A) 8, B) 10, C) 12 and D) 15.
TEM analysis (figure 4 A-B) of the as prepared bimetallic phase
shown quasi-spherical particles but the sample was mainly ill-
defined in size and shape, and particles highly aggregated. To
overcome this drawback and also to increase the surface area of
the final material (i.e. maximizing its surface properties for use as
photo-catalyst), the use of additives was considered and, in
particular, the addition of small amount of NH
4
Cl was found to be
beneficial. Samples prepared with the addition of NH
4
Cl shown a
homogeneous texture, a lower degree of aggregation and a more
defined morphology (figure 4 C-D). In this figure TEM and SEM
A
B
C
D
1 μm
1 μm
1 μm
1 μm
30 35
T=750C
T=650C
Intensity (a.u.)
2Θ
T=550C
FULL PAPER
images of the sample prepared with and without the assistance of
additives are reported.
Figure 4. SEM (A, C) and TEM (B, D) micrographs of ZGON (R
M
=2, R=10)
prepared at 750°C without (A, B) and with (C, D) the assistance of NH
4
Cl.
For this reason, unless otherwise specified, the samples
presented in the following were all prepared with the addition of
NH
4
Cl. To ascertain the simultaneous presence of both metals in
the same cluster, elemental mapping has been undertaken and
results reported in figure 5. The presence of Ga, Zn, N and O in
the original cluster (black and white image) can be seen from this
figure.
Figure 5. Elemental Mapping of ZGON with R
M
=2 and Urea:Zn ratio R=10.
In order to have further information on the formation mechanism,
IR investigation on the precursors at different R has been
conducted. From the IR peaks attribution, peaks at 2200 cm
-1
belongs to the -CH
3
groups, from residual solvent (ethanol), in fact
all samples have been dried in a vacuum oven overnight before
recording IR spectra. We assume that some ethanol molecules
were trapped in the metal-urea complex network as previously
observed.
[22]
In figure 6A the IR spectrum of Zn:Ga:Urea precursor at different
Urea/Metals ratio (R) and R
M
=2 is reported. In figure 6.B, the IR
spectrum of single metal:urea precursors are reported together
with the spectra of pure urea and the bimetallic phase, for
comparison.
Figure 6. A) IR spectra of Zn:Ga:Urea precursor at different Urea/Metals ratio
(R) and R
M
=2; B) IR spectra of single metal:urea precursors and spectra of
pure urea and the bimetallic phase reported for comparison.
A complete peak attribution is reported in Table SI. 2. Peaks near
1455, 1630 and 1670 cm
-1
corresponding to C-N, N-H and C=O
bonds are present. These gradually enhanced peaks meet with
the increasing of the urea ratio from 5, 10 and 15. For peaks near
3360, 3230 and 3360 cm
-1
, these are bonds corresponding to a
nitrate component in the urea nitrate sol-gel precursor.
Band gap calculation
The band gap of the synthesized ZGON was estimated via the
Tauc Plot
[34]
method, after dispersing the ZGON nanoparticles in
water. The band gap of the product with R= 5 (urea/zinc) is
calculated to be 2.3 eV, as in figure SI.11. Through the covalent
bonding and dispersion of conduction and valence bands, the
ZnO cooperated into GaN lattice reduces the band gap, which
leads to a yellow color, as shown in figure SI.12.
Dye decomposition tests
To test the photo-catalytic property of ZGON, the synthesized
powder nanoparticles were used in a model reaction to mimic the
decomposition of organic pollutants (the organic degradation
should give a period of stabilizing time to avoid the color reduction
causing by chemical/physical absorption). For the purpose, the
catalytic powder (20mg) was added to an aqueous solution of
Rhodamine B (RhB, 20 ml, 5ppm) and let under stirring under UV
light (250W). A significant color change was observed already
after 1 hour, while after 2 hours no color was visible (see figure
7.A). For comparison, the decomposition of RhB was also tested
in the presence of pure GaN but also in absence of any catalyst.
As shown in figure 7.B, the concentration of RhB is decreasing
with a faster rate when the bimetallic catalyst was added.
D
C
A
B
1 μm
1 μm
1 μm
FULL PAPER
0 1 2 3
0.2
0.4
0.6
0.8
1.0
Absence of Additives
GaN
ZnGaON
ZnGaInON
Relative Concentration (C/C
0
)
Time (Hour)
Figure 7. A) Concentration of Rhodamine B after UV exposure by measuring
every hour the peak intensity in the RhB solution after UV exposure; B)
Pictures of RhB solution samples after UV light exposure at different hours. C)
Rhodamin B molecular structure. The straight lines are only guidelines for
eyes.
In order to understand the absorption mechanism of RhB on the
catalyst surface, the catalyst powder has been recovered after
testing and an IR spectrum was recorded (figure SI.17). The
spectra of pure RhB and catalyst before testing are also reported
for comparison. The catalytic powder was thoroughly washed with
water before recording the IR spectrum. From this figure peaks at
1217 cm
-1
, 1364 cm
-1
and 1738 cm
-1
can be observed,
corresponding to CH
2
-OH and CH
2
-CH
3
and C=N bonds,
respectively. This indicates RhB is absorbed (physical absorption)
electrostatically (possibly via van der Waals forces) onto the
catalyst surface. Finally, the catalyst powder was washed with
acetone and IR spectrum recorded again (figure S.I. 17) showing
that all absorbed species were removed, thus ruling out any
covalent bond formation and enabling us to reuse the powder for
further testing. To prove that the removal of RhB takes place via
catalytic decomposition and to rule out simple adsorbtion process,
RhB was also tested in the dark, using the same synthetic
condition. Results showed significant changes neither in the
colour of the RhB solution (figure SI.18) nor in its structure, as
ascertained by UV-vis (figure SI.18) and FT-IR (figure SI.19)
spectra, recorded before and after the testing in the dark. Full RhB
decomposition mechanism by the oxy-nitride is still under
investigation. The stability of the catalysts was also proved by
XRD study in the material before and after testing (figure SI.20,
showing no significant differences.
Water splitting
Finally, the potential of the bi-metallic oxy-nitride was tested for
photo-chemical water splitting and results reported in figure 8. The
evolution of hydrogen was 16 μmol/h with using 50 mg
photocatalysts, similar to what observed previously.
[35]
But in the
present case with room for improvements as pointed out by other
authors, crystal imperfections strongly affect performances.
[19]
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0
10
20
30
40
50
Hydrogen Evolution (µmol)
Time (Hour)
ZnGaInON
ZnGaON
Figure 8. Hydrogen evolution by photocatalytic water splitting in the presence
of ZGON and ZGION nanoparticles. Pt-based co-catalyst was added to the
solution in form of 0.1 vol% chloroplatinic acid (platinum source).
From the figure above, ZGON nanoparticle shows a catalytic
efficiency with the hydrogen evolution rate of 12 μmol/hour, while
the rate is 16 μmol/hour for the ZGION catalyst. The reaction has
been repeated for 5 times and the error bar is near 0.3%,
indicating both high photo-activity and reusability of the
quaternary catalyst.
Conclusions
In summary, we have presented the synthesis of multi-metallic
metal(s) oxy-nitride systems via a facile route. By using the urea
glass route, we are able to control particle size and crystallinity.
The urea amount in particular has a decisive effect. ZGON and
ZGION nanoparticles with a band gap energy of ∼2.2 eV and ∼1.9
eV, respectively (depending on the urea and metal ratio), were
tested for the photo-catalytic decomposition of rhodamine B with
good performance (reaction constant of 0.22 h
-1
for ZCON and
0.29 h
-1
for ZGION.
Experimental Section
Gallium(III) nitrate hydrate (CAS 69365-72-6, crystalline, 99.9% trace
metals basis), zinc acetate (CAS 5970-45-6, 99.99% trace metals basis),
indium nitrate hydrate (CAS 207398-97-8, 99.99% trace metals basis) and
urea (CAS 57-13-6, ACS, Reag. Ph Eur) were purchased from Sigma-
Aldrich, Inc. Ethanol (603-002-00-5, Absolute, 99.8%) was purchased from
Honeywell Inc.
In a standard synthesis, the metal precursors are dissolved in ethanol, then
suitable amount of urea are added. The solution is stirred till homogeneity
and let for 24h to evaporate the excess of solvent. In this step a gel-like
precursor is formed alongside a complex between urea and metals.
[22]
The
resulting “jelly” phase was transferred into a crucible and thermally treated
up to 800°C for 5 hours, under nitrogen flow. After cooling down to room
temperature, a yellow/orange powder is obtained. The process is
exemplary shown in figure 9 for the preparation of ZGON.
A
0h
1h
3h
C
B