Photocatalytic nonoxidative coupling of methane on gallium oxide and silica-supported gallium oxide

Article abstract of DOI:10.1016/j.jcat.2008.05.022

Both unsupported and silica-supported gallium oxide were found to promote the nonoxidative coupling of methane (NOCM) to produce ethane and hydrogen upon photoirradiation at around room temperature. The unsupported gallium oxide demonstrated high methane

Full text of DOI:10.1016/j.jcat.2008.05.022

Journal of Catalysis 257 (2008) 396–402
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Photocatalytic nonoxidative coupling of methane on gallium oxide and
silica-supported gallium oxide
Leny Yuliati a, Tadashi Hattori a, Hideaki Itoh b, Hisao Yoshida a,∗
a
Department of Applied Chemistry, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
Division of Environmental Research, EcoTopia Science Institute, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Both unsupported and silica-supported gallium oxide were found to promote the nonoxidative coupling of
methane (NOCM) to produce ethane and hydrogen upon photoirradiation at around room temperature.
The unsupported gallium oxide demonstrated high methane conversion and high ethane selectivity in
hydrocarbon products; however, the amount of hydrogen produced was not equivalent to the amount
of ethane produced, because other reactions, such as consecutive coupling and coke/carbon formation,
also would proceed. The gallium content on the silica-supported gallium oxide affected both the local
structure of gallium species and the photocatalytic property. The sample with the lowest loadings of Ga
(0.1 mol%), on which the gallium oxide existed mainly as highly dispersed tetrahedral species, exhibited
high specific activity per Ga atom and high selectivity for NOCM, where ethane and hydrogen were
produced in almost equal amounts. The gallium oxide clusters or nanoparticles on the silica-supported
samples with Ga loadings of 0.5–10 mol% exhibited low selectivity, likely due to the variety of the active
sites.
Received 22 January 2008
Revised 29 April 2008
Accepted 20 May 2008
Available online 20 June 2008
Keywords:
Photocatalytic nonoxidative coupling of
methane
Gallium oxide photocatalyst
Silica-supported gallium oxide
Local structure
© 2008 Elsevier Inc. All rights reserved.
1. Introduction
silica-based photocatalysts [1–6] and supported ceria photocata-
lysts [7]. A general trend observed on these photocatalysts is that
the highly dispersed metal oxide species on the support or the
surface defects on silica act as the photocatalytic active sites for
NOCM. It is noteworthy that the methane conversion in these
photocatalytic systems was much higher than that at the ther-
modynamic equilibrium. This is one merit of the photocatalytic
system. But its use is still very far from the application. There-
fore, further development of an efficient photocatalyst is highly
desirable. On the other hand, in contrast to the highly dispersed
species photocatalysts, the famous titanium oxide semiconductor
photocatalyst has been confirmed to be unsuitable for this reac-
tion due to its low activity and/or low stability under reductive
condition [1].
The use of gallium oxide as a catalyst has been increasing in
recent years. Gallium oxide has been reported to be a good cata-
lyst for many types of reactions, including aromatization of alka-
nes/alkenes [8–10], alkylation and isomerization of aromatics [8,
11–13], methanol conversion [14–16], selective catalytic reduction
(SCR) of NO [17–20], and the Diels–Alder reaction [21], where the
gallium oxide was usually dispersed on silica and/or alumina sup-
ports or incorporated into zeolites. But even though gallium oxide
is one of the semiconductors, its photocatalytic activity has been
reported only for water-splitting [22,23] and organic decomposi-
tion [24,25].
Being the major component of natural gas as well as a poten-
tial carbon source, methane utilization remains highly desirable.
Methane also has attracted much attention as a potential hydro-
gen source, having the highest H/C ratio among the hydrocarbons.
However, since methane is a stable molecule, effective conversion
to other useful molecules and hydrogen is not easy.
Nonoxidative coupling of methane (NOCM) is a possible reac-
tion for obtaining both ethane and hydrogen,
−1
2CH4 → C2H6 + H2,
ꢀG0298 K = 68.6 kJ mol
.
(1)
The positive and high value of Gibbs free energy, ꢀG0, demon-
strates that this reaction is unfavorable and difficult at low temper-
atures (e.g., room temperature). However, promoting this reaction
selectively with a high yield at high temperatures also is difficult,
because other reactions, such as methane decomposition to carbon
or coke, can proceed more easily. Moreover, carbon or coke forma-
tion will lead to rapid deactivation of the catalyst.
We have proposed an alternative pathway for converting
methane through NOCM at low temperatures with the aid of pho-
toenergy and photocatalysts. We have found many kinds of photo-
catalysts that promote NOCM around room temperature, including
Recently, we briefly reported on the potential ability of gallium
oxide as a photocatalyst for NOCM and dry reforming of methane
Corresponding author. Fax: +81 52 789 3178.
E-mail address: yoshidah@apchem.nagoya-u.ac.jp (H. Yoshida).
*
0021-9517/$ – see front matter © 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2008.05.022

L. Yuliati et al. / Journal of Catalysis 257 (2008) 396–402
397
(DRM) [26]. In the present study, we prepared unsupported gal-
lium oxides and a series of silica-supported gallium oxide with
various amounts of gallium oxide, and investigated the structure
and the photocatalytic properties for NOCM.
the value was corrected considering the absorption band (mainly
<270 nm) of Ga2O3-K.
2.3. Characterizations
Brunauer–Emmett–Teller (BET) specific surface areas of the
samples were calculated from the adsorption of N2 at 77 K. Be-
fore measurement, the sample was treated at 673 K for 30 min
in a flow of He. The amount of N2 adsorbed was determined by
GC TCD. Powder X-ray diffraction (XRD) patterns were recorded on
a Rigaku RINT 2500 diffractometer using CuKα radiation (50 kV,
100 mA).
DR UV–visible spectra were recorded at room temperature on
a JASCO V-550 spectrophotometer equipped with an integrating
sphere covered with BaSO4 using a specially designed in situ cell.
BaSO4 was used as the reference. Before the measurement, the
sample was treated in 13.3 kPa of oxygen atmosphere at 1073 K
for 1 h and evacuated at 1073 K for 1 h. Then the sample was
transferred to the optical part of the cell without being exposed to
the atmosphere.
2. Experimental
2.1. Materials
Two samples of gallium oxide were used as the semiconduc-
tor photocatalysts. One sample, designated Ga2O3-K, was obtained
commercially from Kishida Chemicals (>99.99%). The other, des-
ignated Ga2O3-P, was prepared by the precipitation method as
follows. Ammonia solution was added dropwise to an aqueous so-
lution of Ga(NO3)3 under continuous stirring at room temperature
until the additional precipitate was no longer formed. The precip-
itate was dried at 383 K overnight and then calcined at 1073 K
for 4 h. X-ray diffraction (XRD) patterns and diffuse reflectance
ultraviolet–visible (DR UV–visible) spectra confirmed that both gal-
lium oxide samples were of β-phase.
Phosphorescence spectra were recorded at 77
K on a Hi-
Amorphous silica as the catalyst support was prepared by a sol–
gel method from Si(OEt)4 [27]. After hydrolysis of the Si(OEt)4, the
obtained gel was dried in an oven overnight at 393 K. The dried
sample was powdered and then calcined in a flow of air at 773 K
for 5 h.
Silica-supported gallium oxide samples were prepared by a con-
ventional impregnation method. The silica (2 g) was impregnated
with an aqueous solution of Ga(NO3)3 (25 ml) and heated under
stirring to dryness on a hot plate at 393 K. The sample was fur-
ther dried in the oven overnight at 393 K, and then calcined at
773 K for 5 h. The samples were designated Ga(x)/SiO2, where x
represents the loading amount of Ga on silica (mol%), defined as
x = nGa/(nGa + nSi) × 100%.
tachi F-4500 fluorescence spectrophotometer with an attach-
ment for phosphorescence measurement, using a UV-cut filter
(λtransmittance > 330 nm) to remove the scattered light from the
light source. The sample was pretreated with 13.3 kPa of oxygen
at 1073 K in a specially designed in situ cell, followed by evacu-
ation for 1 h at 1073 K. Then the sample was transferred to the
optical part without being exposed to air.
Ga K -edge X-ray absorption near-edge structure (XANES) spec-
tra were recorded at room temperature in the transmission mode
for high-loading samples (Ga > 1 mol%) or in the fluorescence
mode for low-loading samples (Ga ꢀ 1 mol%) at the BL-10B station
of Photon Factory, High-Energy Accelerator Research Organization
(KEK-PF), Tsukuba, Japan, with a Si(311) channel-cut monochroma-
tor [28]. The sample was pretreated with oxygen for 1 h at 1073 K,
followed by evacuation for 1 h at 1073 K, and then sealed with a
polyethylene film in a dry atmosphere.
2.2. Photocatalytic activity test
Photocatalytic reactions were carried out similarly as in the
previous studies [1–7]. The silica-supported gallium oxide sample
(0.2 g) was spread on the bottom (14 cm2) of a closed quartz re-
actor (30 cm3). For the unsupported gallium oxide samples with a
low specific volume, water (ca. 1 ml) was added to help the sam-
ple spread over the reactor bottom, and then the reactor was dried
in the oven at 383 K overnight. Before the reaction, the sample
was usually treated in 13.3 kPa of oxygen atmosphere at 1073 K
for 1 h, and then evacuated at 1073 K for 1 h. In some cases, the
pretreatment was carried out at 773 K. After cooling to room tem-
perature, methane (200 μmol) was introduced to the reactor. The
photoreaction was carried out at room temperature under photoir-
radiation typically for 3 h. A 300-W Xe lamp, emitting UV and
visible light, was used as the light source. The light intensity mea-
sured at wavelengths of 220–300 nm was ca. 10 mW cm−2. The
gaseous products were collected and analyzed by gas chromatog-
raphy (GC). The amount of hydrogen produced was measured with
a thermal conductivity detector (TCD), and the amounts of hydro-
carbons produced were measured with a flame ionization detector
(FID). The adsorbed products were thermally desorbed, collected,
and then analyzed by GC FID.
The apparent quantum yield in the range of 220–270 nm
on Ga2O3-K was roughly estimated using the equation QY(%) =
Ne/Np ×100, where Ne is the number of the reacted electrons un-
der UV light irradiation and Np is the estimated number of the
incident photon in the range of 220–270 nm. Ne was determined
from the amount of produced hydrogen, assuming that the hydro-
gen was produced by two electrons as follows: 2H+ + 2e− → H2.
Np was estimated from the value measured by a Si photodiode
(Topcon, UVR-2 with UD-25) in the range of 220–300 nm, where
Simulation of the Ga K -edge XANES spectrum was done simi-
larly as in a previous study [29], based on the assumption that the
spectrum comprised two components: gallium in tetrahedral coor-
dination [Ga(Td)] and gallium in octahedral coordination [Ga(Oh)].
Each component was expressed with the sum of one arctangent
curve, T (x), for continuum absorption and one combination of
Gaussian and Lorentzian curves, C(x), for the white line. The curve-
fitting analysis for the XANES spectra, F(x), was carried out using
the following equations [30]:
F(x) = T (x) + C(x),
(2)
T (x) = TTd(x) + TOh(x)
ꢀ
ꢃ
ꢁ
ꢂ
1
1
x − ETd
−1
= h
+
tan
2
π
B
ꢀ
ꢃ
ꢁ
ꢂ
1
2
1
x − EOh
−1
+ (1 − h)
+
tan
,
(3)
π
B
C(x) = CTd(x) + COh(x)
ꢄ
ꢅ
ꢁ
ꢂ
2
hTd
x − ETd
wTd
=
exp (M − 1)(ln 2)
2
1 + M((x − ETd)/wTd
)
ꢄ
ꢅ
ꢁ
ꢂ
2
hOh
x − EOh
+
exp (M − 1)(ln 2)
,
2
1 + M((x − EOh)/wOh
)
wOh
(4)
where x represents X-ray energy (eV); h, hTd, and hOh represent
height; B is a constant fixed at 0.7 eV; ETd and EOh are the cen-

398
L. Yuliati et al. / Journal of Catalysis 257 (2008) 396–402
Table 1
Results of photocatalytic methane conversion over gallium oxide and silica-supported gallium oxidea
−2
Entry
Sample
BET specific
surface area
Yield of hydrocarbon products (10
Gaseous products
Adsorbed productsc
C%b)
Sd (%)
Yield of H2 (μmol)
Experimental
Total yield
Expectede
−1
)
(m2
g
C2H6
C3H8
C2H4
C2H6
C3H6
C3H8
1
Ga2O3-K
Ga2O3-P
Ga2O3-K
SiO2
Ga(0.1)/SiO2
Ga(0.5)/SiO2
Ga(1)/SiO2
Ga(2)/SiO2
Ga(5)/SiO2
Ga(10)/SiO2
Ga(0.1)/SiO2
Ga2O3-K
2.1
18
2.1
554
499
495
485
436
402
389
499
16
13
40
1.0
0.3
5.9
tr.
n.d.
0.1
0.1
0.1
0.1
0.1
1.2
0.1
n.d.
0.1
0.4
0.2
n.d.
0.1
0.1
0.1
0.2
0.3
0.3
0.1
0.3
0.1
0.1
0.5
0.2
n.d.
tr.
0.1
0.1
0.2
0.5
0.5
tr.
n.d.
n.d.
0.3
n.d.
n.d.
0.2
0.2
0.2
0.2
0.3
n.d.
0.2
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
tr.
0.1
0.1
n.d.
n.d.
n.d.
17
14
96
95
86
93
94
86
88
90
92
90
94
90
73
0.11
0.07
0.87
0.18
0.15
0.54
2
3f
4
5
6
7
8
9
10
11f
12h
13h
49g
0.4
2.4
3.0
4.1
6.0
7.2
7.0
16
0.4
2.3
2.5
3.5
5.2
6.2
5.8
14
n.d.
n.d.
0.04
0.03
n.d.
0.04
0.03
0.17
0.09
n.d.
<0.01
0.03
0.03
0.04
0.06
0.08
0.08
0.16
2.1
499
3.8
0.2
0.1
n.d.
4.5
0.3
0.05
<0.01
Ga(0.1)/SiO2
a
Standard condition: reaction temperature was ca. 310 K, sample was 0.2 g, initial methane was 200 μmol, irradiation time was 3 h. Before reaction, the sample was
pretreated in oxygen atmosphere at 1073 K for 1 h, followed by evacuation at room temperature for 1 h. H2 was analyzed by TCD (detection limit was 0.03 μmol), hydrocarbons
were analyzed by FID (experimental error, <4%).
b
Based on the initial amount of methane.
The products were obtained after desorption procedure at 573 K for 15 min.
c
d
Selectivity of ethane (S = yield of ethane/total yield × 100%).
e
Calculated from hydrocarbons products.
f
Irradiation time was 24 h.
g
−2
−2
−2
C%) were also observed as gaseous products.
C2H4 (1.0 × 10
C%), C3H6 (0.4 × 10
C%) and C4H10 (1.7 × 10
h
Pretreatment prior the reaction was carried out at 773 K. n.d. = not detected. tr. = trace.
ter position of each band (eV); wTd and wOh are half width at
half maximum (eV), and M is the ratio of Gaussian to Lorentzian
functions, fixed at 0.85 in this study. The height of each T (x) corre-
sponds to the relative ratio of each gallium species in the sample.
The center position of each band, the half-width at half-maximum,
and the peak area of each C(x) were determined to simulate the
spectrum, and the quantitative ratio of each gallium species was
estimated from the best-fitting analysis.
conditions during the reaction. This stability is an important factor
required to promote the photocatalytic NOCM. Thus, we empha-
size that the gallium oxide is the first semiconductor photocatalyst
found to be suitable for promoting NOCM photocatalytically.
The product selectivity for ethane in the total yield of hydrocar-
bons (S = ethane yield/total yield × 100%) was very high on both
Ga2O3 samples (ca. 95–96%). Other consecutive reactions, such as
consecutive coupling to produce longer-chain hydrocarbons and
dehydrogenation to produce alkenes from alkanes, proceeded very
slowly as minor reactions. These results imply that NOCM shown
in Eq. (1) should proceed selectively, at least in the first 3 h of
the irradiation; however, the amount of hydrogen produced was
lower than that calculated from the amounts of hydrocarbons pro-
duced on both unsupported samples after the reaction for 3 h. One
possible explanation for this finding is that some type of induction
period for hydrogen formation through NOCM may exist, likely due
to hydrogen absorption and/or chemisorption on the gallium ox-
ide surface or to the formation of surface gallium hydride species
[31–33]. On the other hand, when the reaction was carried out for
longer irradiation times, such as 24 h, excess hydrogen was ob-
tained, and the ethane selectivity decreased to 86% (entry 3). Thus,
the possibility that during NOCM, the consecutive reactions men-
tioned above will be further accelerated to increase the production
of hydrogen but decrease production of the detectable longer-chain
hydrocarbons should be considered.
The amorphous silica exhibited very low photoactivity (en-
try 4), as reported previously [1–7]. However, methane conversion
in the photocatalytic NOCM over the silica still was much higher
(ca. 10-fold higher) than the theoretical value calculated from the
thermodynamic equilibrium constant (0.0004%).
All of the silica-supported gallium oxide samples exhibited
higher yields of hydrocarbons than that of silica (entries 5–10).
Similar to the case of the unsupported gallium oxide, all of the
silica-supported samples produced ethane as the major product,
suggesting that NOCM shown in Eq. (1) was the main reaction.
Total yield increased with increasing Ga loading, indicating that
the photocatalytic active sites should be related to the Ga species;
however, the total yield became constant at loading amounts
>5 mol%, as also shown in Fig. 1. No clear relationship between
3. Results and discussion
3.1. Activity for photocatalytic NOCM
Table 1 presents the photocatalytic activities of the unsupported
gallium oxide and the silica-supported gallium oxide samples for
NOCM. The table confirms that no product was obtained when the
reaction was carried out in the dark and/or without photocata-
lyst. Both Ga2O3-K and Ga2O3-P showed high activity for NOCM
to produce ethane and hydrogen (entries 1 and 2). The methane
conversion was 0.17% on Ga2O3-K after photoirradiation for 3 h.
The yield of ethane in the equilibrium for NOCM was estimated
to be 0.0004% based on the thermodynamic equilibrium constant
(K = 3.6 × 10−12) at 314 K. Thus, it is clear that the gallium ox-
ide selectively promoted the forward reaction in the photocatalytic
NOCM. The apparent quantum yield for NOCM on Ga2O3-K (en-
try 1) was roughly estimated to be ca. 0.01% at wavelengths of
220–270 nm.
Even though the Ga2O3-K had a much lower BET specific sur-
face area, the sample exhibited slightly higher activity than the
Ga2O3-P with a higher specific surface area. XRD patterns revealed
that the Ga2O3-K sample had much higher crystallinity than the
Ga2O3-P sample. This result suggests that, instead of specific sur-
face area, high crystallinity is more important for obtaining high
activity for NOCM.
The active semiconductor for the photocatalytic NOCM has not
been reported to date. TiO2, a known semiconductor photocata-
lyst, is not suitable for the reaction, because it can be readily
reduced by hydrogen under photoirradiation [1]. However, the gal-
lium oxide semiconductor exhibited high stability under reductive

L. Yuliati et al. / Journal of Catalysis 257 (2008) 396–402
399
Fig. 1. Dependences of hydrocarbons yield (open circle) and ethane selectivity
(closed circle) on the Ga content of silica-supported gallium oxide samples.
the BET specific surface area and photoactivity of the samples
could be observed, because the major contribution to the surface
area should be from the silica support.
The highly dispersed metal oxide species on the supports, such
as silica and alumina, have been reported to show higher activity
than their corresponding unsupported metal oxides even if they
were semiconductors. Thus, this is the first case in which un-
supported samples have been shown to have higher activity than
silica-supported samples. The unsupported gallium oxide clearly
exhibited higher activity than the silica-supported gallium oxide;
however, Ga(0.1)/SiO2 had ca. 45-fold greater specific activity per
Ga atom compared with the bulk gallium oxide, along with the
highest selectivity for ethane formation (94%) among the silica-
supported gallium oxide samples studied (entry 5). Even when the
detection limits for the minor products were considered, the selec-
tivity was not lower than 92%. Furthermore, high ethane selectivity
(ca. 94%) was obtained even after 24 h of irradiation, similar to
that obtained after 3-h photoreaction (entries 5 and 11), confirm-
ing that NOCM would proceed selectively on Ga(0.1)/SiO2 for a
prolonged period. At a reaction time of 3 h, hydrogen was not de-
tected on Ga(0.1)/SiO2, likely due to the detection limit (estimated
as 0.03 μmol) of GC (entry 5). However, sufficient hydrogen was
obtained after 24 h of photoreaction (entry 11), with the amount of
hydrogen obtained almost the same as that expected from NOCM
in Eq. (1). Because the total yield of the hydrocarbon products
(0.16 C%) obtained on Ga(0.1)/SiO2 after 24 h was similar to that
obtained on Ga2O3-K (0.17 C%) (entries 1 and 11), we can compare
the selectivities on these samples. Ethane selectivity among hydro-
carbon products was similarly high on these samples. However, it
is notable that the amount of hydrogen produced was less than
that expected on the unsupported gallium oxide but almost equiv-
alent to that expected on Ga(0.1)/SiO2. These findings indicate that
Ga(0.1)/SiO2 exhibited higher selectivity for NOCM, producing both
ethane and hydrogen, compared with the unsupported gallium ox-
ide semiconductor photocatalyst.
On Ga(0.5)/SiO2, the ethane selectivity was obviously low,
though the yield of hydrocarbons was higher than that on Ga(0.1)/
SiO2 (entry 6). But with further increases in Ga content above
0.5 mol%, both product yield and ethane selectivity tended to
increase (entries 7–10; also see Fig. 1). The samples contain-
ing medium loadings of Ga, such as Ga(0.5)/SiO2 and Ga(1)/SiO2,
produced similar amounts of hydrogen as those expected from
calculations based on the amounts of hydrocarbons produced
(entries 6 and 7). But on the samples with higher Ga loadings
(ꢁ1 mol%), less hydrogen than expected was produced (entries
7–10). Even when considering the detection limit for hydrogen
production (0.03 μmol), the difference between the amount of hy-
drogen produced and the amount expected from calculations based
on the amount of hydrocarbons produced increased from 0.01 to
0.05 μmol with increasing Ga loading (entries 7–10). This is likely
Fig. 2. DR UV–visible spectra of the silica-supported gallium oxide samples and un-
supported Ga2O3; (a) Ga(0.1)/SiO2, (b) Ga(0.5)/SiO2, (c) Ga(1)/SiO2, (d) Ga(2)/SiO2,
(e) Ga(5)/SiO2, (f) Ga(10)/SiO2, (g) Ga2O3-K, and (h) Ga2O3-P. Spectrum intensity of
(g) and (h) was multiplied by 1/3.
due to the presence of aggregated gallium oxide species on the
high-loading samples, which can absorb hydrogen, as mentioned
above. We confirm the presence of the gallium oxide nanoparticles
on high-loading samples in the next section.
The effect of the pretreatment temperature was examined on
Ga2O3-K and Ga(0.1)/SiO2. Both of the samples pretreated at higher
temperature (1073 K) exhibited much higher activity and higher
ethane selectivity than those pretreated at lower temperature
(773 K) (entries 1, 5, 12, and 13). These findings suggest that the
treated catalyst surface is important for the photocatalytic NOCM.
The generation of the active sites on both the unsupported and
silica-supported gallium oxide samples may be closely related to
the dehydroxylation of hydroxyl groups on the surface.
Under similar conditions, the silica-supported gallium ox-
ide samples exhibited similar activity as has been reported for
other silica-based photocatalysts [4], such as silica–alumina, silica-
supported zirconium oxide, and silica-supported magnesium oxide.
Each of these photocatalysts had the optimum loading amount of
metal oxide for high activity in NOCM; however, the activities of
silica-supported gallium oxide samples increased with increasing
gallium content, with even the bulk gallium oxide exhibiting pho-
tocatalytic activity.
3.2. Photoexcitation
Fig. 2 shows DR UV–visible spectra of the unsupported gallium
oxide samples. Both Ga2O3-K and Ga2O3-P exhibited a broad and
intense band at below 270 nm, with an estimated band gap of
ca. 4.7 eV for both samples. Even though a slight difference was
observed on these spectra, these bands can be assigned to β-phase
gallium oxide [24,25].
Fig. 2 also shows DR UV–visible spectra of the silica-supported
gallium oxide samples. The sample with the lowest loading
amount, Ga(0.1)/SiO2, exhibited a weak, broad band at around
250 nm (Fig. 2a), similar to the spectrum of SiO2 reported pre-
viously [2,3]. With increasing Ga loading (to 0.5 and 1 mol%), two
absorption bands centered at around 215 and 240 nm appeared
(Figs. 2b and 2c). For Ga(2)/SiO2, a broad, large absorption band
was observed at around 240 nm with a minor band at around
215 nm (Fig. 2d). For the Ga(5)/SiO2 and Ga(10)/SiO2 samples, only
one absorption band centered at 250 nm was observed (Figs. 2e
and 2f). These findings indicate that the structure and electronic

400
L. Yuliati et al. / Journal of Catalysis 257 (2008) 396–402
state of the gallium species clearly differed between the low-
loading and high-loading samples.
Judging from the band position close to that of the bulk gal-
lium oxide samples, the strong band centered at 250 nm should
be due to the absorption of gallium oxide nanoparticles or crystal-
lites on silica. On the other hand, the bands centered at 215 and
240 nm can be considered a characteristic of the highly dispersed
gallium oxide species on silica support. Because the lowest-loading
sample, Ga(0.1)/SiO2, did not exhibit a clear absorption band at
ca. 215 nm, the highly dispersed gallium oxide species would ex-
hibit an absorption band at 240 nm. Thus, the weak, broad band
at ca. 250 nm for Ga(0.1)/SiO2 would originate from the absorp-
tion bands of both the highly dispersed gallium oxide species at
240 nm and the defect on silica at 258 nm [2]. On the other hand,
the band at 215 nm would originate from the very small (e.g.,
2–5 nm) gallium oxide nanoparticles, as reported previously [34].
The shift of the band for gallium oxide nanoparticles from 215 to
250 nm with increasing Ga content from 0.5 to 10 mol% would
correspond to the increase in semiconductor nanoparticle size.
Even though no clear evidence yet exists, we believe that the
highly dispersed gallium species exist as isolated gallium oxide
monomers (<0.4 nm). This is supported by the photoluminescence
findings discussed later. As for the small and large nanoparticles,
estimating their sizes is difficult, but because the crystallite size of
Ga2O3-K was found to be 57 nm by XRD and the DR UV spectra of
the nanoparticles on silica clearly differed from that of Ga2O3-K,
the nanoparticles on the high-loading samples should be smaller
than 50 nm. Because nanoparticles of 2–5 nm have been reported
to exhibit an absorption edge at 225–264 nm [34], gallium oxide
clusters or small nanoparticles should exist on the medium-loading
samples. Based on these results, we can propose that the lowest-
loading sample (0.1 mol%) will have highly dispersed gallium oxide
monomer species (subnanometer), the medium-loading samples
(0.5–2 mol%) will have both highly dispersed gallium species and
very small gallium oxide nanoparticles (a few nanometers), and the
highest-loading samples (5–10 mol%) will have larger nanoparticles
(<50 nm) as the main species.
Fig. 3. Photoluminescence emission spectra of (a) Ga(0.1)/SiO2, (b) Ga(0.5)/SiO2, and
(c) Ga(2)/SiO2. The spectra were recorded at 77 K with the excitation wavelength at
250 nm.
Regarding band intensity, the unsupported gallium oxide sam-
ples clearly were of much greater intensity compared with the
silica-supported gallium oxide samples. The strong absorption
band for the unsupported gallium oxide means that photoexcited
electrons and holes can be formed with high efficiency under UV
irradiation. The absorption band for semiconductors is explained
by the electron transition from the valence band to the conduc-
tion band, that is, electron transfer from O2− to Ga3+ in gallium
oxide. In contrast, the bands observed for the silica-supported gal-
lium oxides, especially those with low loadings of Ga, were quite
different from those for the unsupported gallium oxides, corre-
sponding to the “quantum photocatalysts” [35,36]. These bands
can be assigned to localized photoexcitation at isolated sites, such
as monomeric gallium species on silica. The excitation process
occurs in the Si–O–Ga linkages, as was proposed for highly dis-
persed metal oxide on silica (i.e., silica-based photocatalysts) [4].
On the silica-supported samples with medium Ga loading hav-
ing both highly dispersed species and gallium oxide nanoparticles,
photoexcitation through the band gap excitation (similar to that on
the gallium oxide semiconductor) should be considered along with
photoexcitation of monomeric species in the Si–O–Ga linkages.
Photoluminescence spectroscopy has been proposed as a good
tool to provide significant information about the photoactive sur-
face sites, especially on the samples with low loadings of metal
oxide (<0.1 wt%) [37]. The silica-supported gallium oxide sam-
ples exhibited photoluminescence spectra, as shown in Fig. 3. The
relatively low-loading samples, Ga(0.1)/SiO2 and Ga(0.5)/SiO2, ex-
hibited phosphorescence spectra with a fine structure (Figs. 3a and
3b), similar to those of silica–alumina, silica-supported zirconium
Fig. 4. Ga K -edge XANES spectra of (a) Ga(0.1)/SiO2, (b) Ga(0.5)/SiO2, (c) Ga(1)/SiO2,
(d) Ga(2)/SiO2, (e) Ga(5)/SiO2, (f) Ga(10)/SiO2 samples, and (g) Ga2O3-K (β-Ga2O3).
oxide, and silica-supported magnesium oxide [3–6,38], although
the present ones were not so clear. Highly dispersed metal oxide
species on silica, Si–O–M linkages (M = metal oxide), have been
suggested to be the sites responsible for exhibiting the fine struc-
ture of the phosphorescence spectra; therefore, the fine structure
of the emission spectra would support the presence of highly dis-
persed gallium oxide species on silica, where photoexcitation with
the electron charge transfer is proposed to be localized to the Si–O
moiety, as has been suggested for silica-based photocatalysts [4].
On the other hand, the samples with higher Ga loadings exhib-
ited different-shaped phosphorescence spectra compared with the
low-loading samples. Ga(2)/SiO2 exhibited a broad emission band
centered around 465 nm without the fine structure (Fig. 3c), which
can be assigned to gallium oxide nanoparticles [34,39]. These find-
ings are in good agreement with the DR UV–visible spectra of the
samples.
3.3. Local structure
XANES can be used to determine the local structure of the
gallium species on the samples both qualitatively and quantita-
tively [20,29,40]. Fig. 4 shows Ga K -edge XANES spectra of the

L. Yuliati et al. / Journal of Catalysis 257 (2008) 396–402
401
Fig. 6. Dependence of Ga(Td) ratio on the Ga content.
Fig. 5. The best results in the curve fitting analysis for XANES spectra of (a) Ga2O3-K,
(b) Ga(1)/SiO2, and (c) Ga(0.1)/SiO2. The solid lines show the experimental spec-
tra. The symbols show the simulated spectra, which are composed of two arctan-
gent functions and two combinations of Gaussian and Lorentzian functions (broken
lines).
Fig. 7. Plot of ethane selectivity versus the ratio of Ga(Td) on the Ga2O3-K and the
silica-supported gallium oxide samples. The numbers show the Ga content (mol%)
on the silica-supported samples, and 100 shows the Ga2O3-K.
the ethane selectivity decreased with increasing Ga(Td) fraction. As
discussed above, these silica-supported samples with medium and
high Ga loadings contained gallium oxide nanoparticles; therefore,
these samples can be expected to behave somewhat similarly to
the unsupported gallium oxide samples, especially those with high
amounts of gallium oxide nanoparticles. But the photocatalytic
properties of the nanoparticles would vary with increases in the
Ga(Td) fraction. Our findings suggest that the Ga(Td)-rich nanopar-
ticles would exhibit greater activity for the consecutive reactions.
On the other hand, the plot for the lowest-loading sample hav-
ing the highly dispersed species, i.e., Ga(0.1)/SiO2, was out of line.
These results again confirm that the photoexcitation mechanism
for these two kinds of photocatalysts—nanoparticles of gallium
oxide (semiconductor photocatalyst) and highly dispersed tetrahe-
dral gallium oxide species (quantum photocatalyst)—clearly differ,
as was also supported by DR UV–visible and photoluminescence
spectra findings. The different excitation mechanisms explain the
differences in yield and product selectivity seen in the photocatal-
ytic NOCM.
unsupported and silica-supported gallium oxide samples. In the
unsupported β-phase gallium oxide, Ga species are known to exist
equally in both tetrahedral and octahedral coordinations (1:1 ra-
tio) [41]. The distinguishable peaks at 10,375.2 and 10,379.1 eV on
the XANES spectrum of β-phase gallium oxide have been assigned
to Ga(Td), and Ga(Oh), respectively [20,29].
The XANES spectra of the silica-supported gallium oxide sam-
ples suggest that the local structure of the gallium species differ
between the low-loading and high-loading samples. The highly dis-
persed gallium oxide species on the low-loading samples would be
in tetrahedral coordination, and the gallium oxide nanoparticles on
the high-loading samples would consist of tetrahedrally and octa-
hedrally coordinated gallium species.
To obtain the quantitative ratio between Ga(Td) and Ga(Oh),
deconvolution analysis of the spectra was carried out by the curve-
fitting method mentioned in Section 2. Fig. 5 shows simulated
spectra of some samples. Although the samples with low Ga load-
ings exhibited low spectrum/noise and spectrum/background ra-
tios, making the fitting analysis more difficult, and although unsat-
isfied parts remained for the 10,370–10,374 eV region and above
10,380 eV on the most of the spectra, the best fitting for each
spectrum was determined based on the main features of the target
spectrum, as suggested in the literature [42].
The dependence of the Ga local structure on Ga content is rep-
resented as the ratio of Ga(Td), as shown in Fig. 6. For comparison,
the figure also shows the value for the unsupported β-phase gal-
lium oxide (100 mol% of Ga content), in which the ratio of Ga(Td)
was confirmed to be ca. 50%. With increasing gallium content, the
ratio of Ga(Td) species decreased, and the value became close to
that of the bulk β-phase gallium oxide.
3.4. Active sites
Based on the foregoing results, we proposed the photocatalytic
active sites on each type of gallium oxide photocatalyst. The find-
ing of greater ethane selectivity on the lowest-loading sample,
Ga(0.1)/SiO2, compared with the medium- and high-loading sam-
ples suggests that the highly dispersed tetrahedral gallium oxide
species on silica acted as the photoactive sites selectively promot-
ing NOCM. Photoexcitation was localized on the isolated active
sites, selectively activating methane molecules. Because the active
sites were uniform and well dispersed, other reactions hardly oc-
curred, providing the high selectivity for NOCM.
Plotting ethane selectivity versus the Ga(Td) ratio of the sam-
ples revealed a good relationship between the unsupported gallium
oxide samples and the silica-supported ones with medium and
high Ga loadings (>0.5 mol%), as shown in Fig. 7. In these samples,
Together, the unsupported bulk gallium oxide particles, as well
as supported nanoparticles existing as major species on Ga(5)/SiO2
or Ga(10)/SiO2, function as a semiconductor photocatalyst. This

402
L. Yuliati et al. / Journal of Catalysis 257 (2008) 396–402
catalyst can be efficiently photoexcited, as suggested by the large
absorption band. The activity should be high. But the presence of
photoexcited electrons and holes on the various surface sites on
gallium oxide particles will diminish the reaction selectivity for
NOCM and enhance that for other reactions, such as consecutive
coupling, dehydrogenation, and coke/carbon formation.
The medium-loading samples, such as Ga(1)/SiO2, were found
to contain various photocatalytically active species, such as Ga(Td)-
rich gallium oxide small nanoparticles or clusters on silica. Thus,
some types of photoactive sites, such as coordinatively unsaturated
surface sites, would reduce the selectivity.
[2] L. Yuliati, M. Tsubota, A. Satsuma, H. Itoh, H. Yoshida, J. Catal. 238 (2006) 214.
[3] H. Yoshida, M.G. Chaskar, Y. Kato, T. Hattori, J. Photochem. Photobiol. A Chem.
160 (2003) 47.
[4] L. Yuliati, T. Hattori, H. Yoshida, Phys. Chem. Chem. Phys. 7 (2005) 195.
[5] H. Yoshida, N. Matsushita, Y. Kato, T. Hattori, Phys. Chem. Chem. Phys. 4 (2002)
2459.
[6] Y. Kato, H. Yoshida, T. Hattori, Chem. Commun. (1998) 2389.
[7] L. Yuliati, T. Hamajima, T. Hattori, H. Yoshida, Chem. Commun. (2005) 4824.
[8] G. Caeiro, R.H. Carvalho, X. Wang, M.A.N.D.A. Lemos, F. Lemos, M. Guisnet, F.R.
Ribeiro, J. Mol. Catal. A 255 (2006) 131.
[9] R.L. van Mao, J. Yao, L.A. Dufresne, R. Carli, Catal. Today 31 (1996) 247.
[10] P. Mériaudeau, C. Naccache, Catal. Today 31 (1996) 265.
[11] A. Raj, J.S. Reddy, R. Kumar, J. Catal. 138 (1992) 518.
[12] Y.K. Park, K.Y. Park, S.I. Woo, Catal. Lett. 26 (1994) 169.
[13] A. Corma, F. LLopis, P. Viruela, C. Zicovich-Wilson, J. Am. Chem. Soc. 116 (1994)
134.
4. Conclusions
Unsupported and silica-supported gallium oxide were found to
promote the photocatalytic NOCM to produce ethane and hydrogen
on photoirradiation at around room temperature. Because the con-
version exceeded the equilibrium at 314 K, it is obvious that these
gallium oxide photocatalysts can selectively promote the forward
reaction of the photocatalytic NOCM. This is the advantage of the
photocatalytic reaction.
[14] E. Lalik, X. Liu, J. Klinowski, J. Phys. Chem. 96 (1992) 805.
[15] V.R. Choudhary, A.K. Kinage, Zeolites 15 (1995) 732.
[16] M. Kang, C.-T. Lee, J. Mol. Catal. A 150 (1999) 213.
[17] E. Kikuchi, K. Yogo, Catal. Today 22 (1994) 73.
[18] Y. Li, J.N. Armor, J. Catal. 145 (1994) 1.
[19] J.N. Armor, Catal. Today 31 (1996) 191.
[20] K. Shimizu, M. Takamatsu, K. Nishi, H. Yoshida, A. Satsuma, T. Tanaka, S. Yoshi-
da, T. Hattori, J. Phys. Chem. B 103 (1999) 1542.
The unsupported gallium oxide as a semiconductor photocata-
lyst exhibited greater activity than the silica-supported gallium ox-
ide. On the other hand, the highly dispersed tetrahedral species as
the quantum photocatalyst on the silica-supported gallium oxide
sample of low Ga loading (e.g., 0.1 mol%) was responsible for the
high reaction selectivity. Future studies are expected to improve
the selectivity of the unsupported gallium oxide photocatalyst and
to design more active supported gallium oxide photocatalysts.
[21] A. Satsuma, Y. Segawa, H. Yoshida, T. Hattori, Appl. Catal. A 264 (2004) 229.
[22] A. Kudo, I. Mikami, J. Chem. Soc. Faraday Trans. 94 (1998) 2929.
[23] T. Yanagida, Y. Sakata, H. Imamura, Chem. Lett. 33 (2004) 726.
[24] Y. Hou, X. Wang, L. Wu, Z. Ding, X. Fu, Environ. Sci. Technol. 40 (2006) 5799.
[25] Y. Hou, L. Wu, X. Wang, Z. Ding, Z. Li, X. Fu, J. Catal. 250 (2007) 12.
[26] L. Yuliati, H. Itoh, H. Yoshida, Chem. Phys. Lett. 452 (2008) 178.
[27] H. Yoshida, C. Murata, T. Hattori, J. Catal. 194 (2000) 364.
[28] M. Nomura, A. Koyama, KEK Report 89-16 (1989) 1.
[29] K. Nishi, K. Shimizu, M. Takamatsu, H. Yoshida, A. Satsuma, T. Tanaka, S. Yoshi-
da, T. Hattori, J. Phys. Chem. B 102 (1998) 10190.
[30] Y. Kato, K. Shimizu, N. Matsushita, T. Yoshida, H. Yoshida, A. Satsuma, T. Hattori,
Phys. Chem. Chem. Phys. 3 (2001) 1925.
Acknowledgments
[31] E.A. Gonzales, P.V. Jasen, A. Juan, S.E. Collins, M.A. Baltanás, A.L. Bonivardi, Surf.
Sci. 575 (2005) 171.
[32] B. Xu, B. Zheng, W. Hua, Y. Yue, Z. Gao, J. Catal. 239 (2006) 470.
[33] E.A. Pidko, V.B. Kazansky, E.J.M. Hensen, R.A. van Santen, J. Catal. 240 (2006)
73.
[34] G. Sinha, D. Ganguli, S. Chaudhuri, J. Phys. Condens. Matter 18 (2006) 11167.
[35] H. Yoshida, Curr. Opin. Solid Mater. Sci. 7 (2003) 435.
[36] H. Yoshida, Catal. Surv. Asia 9 (2005) 1.
The X-ray absorption experiments at the Ga K -edge were per-
formed under the approval of the Photon Factory Program Advi-
sory Committee (proposal 2003G248). This work was partially sup-
ported by a Grant-in-Aid for Young Scientists (A) (16686045) and
a Grant-in-Aid for Scientific Research on Priority Areas (19028023,
“Chemistry of Concerto Catalysis”) from the Japanese Ministry of
Education, Culture, Sports, Science and Technology (to H.Y.). The
authors thank Katsuya Shimura for preparing the unsupported gal-
lium oxide and Naoki Hirabayashi for carrying out the preliminary
photoluminescence study.
[37] M. Anpo, M. Che, Adv. Catal. 44 (2000) 119.
[38] Y. Kato, H. Yoshida, T. Hattori, Phys. Chem. Chem. Phys. 2 (2000) 4231.
[39] X. Xiang, C.-B. Cao, H.-S. Zhu, J. Cryst. Growth 279 (2005) 122.
[40] K. Shimizu, M. Takamatsu, K. Nishi, H. Yoshida, A. Satsuma, T. Hattori, Chem.
Commun. (1996) 1827.
References
[41] S. Geller, J. Chem. Phys. 33 (1960) 676.
[42] S. Yoshida, T. Tanaka, in: Y. Iwasawa (Ed.), X-ray Absorption Fine Structure for
Catalysts and Surfaces, World Scientific, Singapore, 1996, p. 304.
[1] L. Yuliati, H. Itoh, H. Yoshida, Stud. Surf. Sci. Catal. 172 (2007) 457.