Active sites in sol-gel prepared silica-alumina for photoinduced non-oxidative methane coupling

Article abstract of DOI:10.1039/b111101a

The photoinduced non-oxidative methane coupling at room temperature was studied using a series of silica-alumina samples prepared by a sol-gel method to clarify the photoactive sites of silica-alumina. All the silica-alumina samples showed the maximum absorbance ～ 230 nm and the intensity varied with Al content. The absorption intensity increased with an increase of Al content up to ～ 20 mole %, then gradually decreased. All the prepared samples of silica, silica-alumina, and alumina exhibited activity for the photoinduced non-oxidative methane coupling around room temperature. Only ethane was obtained as the gaseous product under this condition, except for the sample of 16.7 mole % Al. On the 16.7 mole % Al silica-alumina sample, which showed the maximum conversion, a trace amount of propane was also produced. Ethylene was detected as a thermally desorption product on some samples of high Al contents. A good correlation was revealed between the phosphorescence intensity and the photoactivity for this reaction among the low Al content samples < 20 mole %. The aggregated species, e.g., Al oxide clusters and crystallites in silica-alumina, produced ethane and ethylene. The highly dispersed and isolated tetrahedral AlO₄ species in silica matrix would be the highly active sites for ethane formation. Thus, the aggregated Al oxide species would be the active sites for producing ethane and ethylene with low activity.

Full text of DOI:10.1039/b111101a

View Article Online / Journal Homepage / Table of Contents for this issue
Active sites in sol–gel prepared silica-alumina for photoinduced
non-oxidative methane coupling
Hisao Yoshida,* Norimitsu Matsushita, Yuko Kato and Tadashi Hattori
Department of Applied Chemistry, Graduate School of Engineering, Nagoya University, Nagoya,
464-8603, Japan. E-mail: yoshidah@apchem.nagoya-u.ac.jp; Fax: +81 52 789 3193
Received 6th December 2001, Accepted 12th February 2002
First published as an Advance Article on the web 22nd April 2002
The photoinduced non-oxidative methane coupling at room temperature was examined by using a series of
silica-alumina samples prepared by a sol–gel method, in order to clarify the photoactive sites of silica-alumina.
The maximum yield of ethane, the main product in this reaction condition, was obtained on the sample
containing around 20 mol% of Al. The variation of ethane yield was consistent with the variation of UV-
absorption intensity. Ethene as the minor product, which is obtained by thermal desorption after
photoreaction, was produced on the samples of larger Al content. The aluminium sites in the low Al content
samples exhibited higher specific activity and higher ethane selectivity than those in high Al content samples.
The silica-alumina sample was characterized by using N2-adsorption, XRD, XPS and Al K-edge XANES. In
addition to UV spectroscopy, phosphorescence emission spectroscopy was also applied to clarify the
photoactivation properties of silica-alumina. It is concluded that the highly dispersed and isolated tetrahedral
AlO4 species in silica matrix, which exhibit the fine structural phosphorescence spectra centred at 520 nm, would
be the highly active sites for ethane formation. On the other hand, the aggregated Al oxide species would be the
active sites for producing both ethane and ethene with low activity.
showed the highest yield among them, the product yield in
Introduction
the gaseous phase was low and heating or admission of water
vapor were required to collect the adsorbed coupling pro-
ducts.6
In order to make an efficient utilization of natural gas, we
must consider transforming it into more valuable chemicals.
Higher hydrocarbons are more useful for chemical industry.
The oxidative coupling of methane (OCM) is one of the effec-
tive reactions producing higher hydrocarbons and has been
extensively studied worldwide over the past two decades. The
OCM reaction has the thermodynamical advantage in compar-
ison with the direct coupling reaction of methane without oxi-
dant. However, no catalysts could reach the principal criteria
for industrial application of OCM.1 It is quite difficult to
obtain the coupling products in high yield, because the oxida-
tion of coupling products to COx proceeds more selectively
than the coupling reaction. The limits of the process have been
essentially indicated.2 Developments of other processes are
required.
Photoinduced reaction is one of the most available reactions.
The system employing photoinduced reaction has some advan-
tages: It works even at room temperature and it might employ
solar energy so that it is environmentally benign. In addition,
even the thermodynamically difficult reaction may possibly
proceed in photoinduced system. Actually, several researchers
have tried to activate methane by employing photo-
induced system.
Recent developments in the photocatalysis systems led us to
recognize that the employment of transition metal oxides is
not the sole solution. Some kinds of photoreactions were
successively discovered on silica and silica-based materials,
e.g., photoisomerization of butene,7 photometathesis of
alkenes,8–12 photooxidation of alkenes,13–17 and photooxida-
tion of CO.18 In addition, silica-alumina was found to exhibit
a
characteristic phosphorescence emission spectrum,19
which predicted that silica-alumina might function as a photo-
catalyst.
Recently, we have found out that the photoinduced non-oxi-
dative coupling of methane proceeds on silica-alumina around
room temperature.20 In this system C2+ products are obtained
in the gaseous phase without any forced desorption process,
and no oxygenated products are formed, including COx . In
addition, the stoichiometric formation of H2 was also
revealed.21 Hydrogen is being used as a fuel, and is thought
to be a clean energy source for the twenty-first century to meet
environmental requirements. This photoreaction is very attrac-
tive since this reaction is energetically up-hill type reaction
(DG > 0) where the total chemical potentials of products, such
as ethane and hydrogen, are higher than that of reactant,
methane:
For example, photoinduced oxidative methane coupling sys-
tem has already been investigated by using TiO2 at 373–473 K
in the presence of oxygen.3 However, the most important
problem on this system was high selectivity to CO2 in addi-
tion to low yield. To avoid complete oxidation, the photoindu-
ced non-oxidative coupling systems without using oxidant
molecules were examined by employing transition metal oxi-
2 CH4 ꢀ! C2H6 þ H2; DG ð298 KÞ ¼ 68:6 kJ molꢀ1
This means that the photoenergy is partially stored as chemical
potential in the products.
In the present work, we examined the activity on a series
of sol–gel prepared silica-alumina samples containing
5
des, such as V/SiO2 ,4 TiO2 and Mo/SiO2 .6 How-
ever, for example, in the system employing Mo/SiO2 , which
DOI: 10.1039/b111101a
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465
2459
This journal is # The Owner Societies 2002

View Article Online
various aluminium contents and discuss the active sites on the
silica-alumina for photoinduced non-oxidative methane cou-
pling.
Okazaki, Japan, with ring energy 750 MeV and stored current
of 130–200 mA. A two-crystal monochromator employing
YB66 crystals29 was used for recording Al K-edge XANES.
Data were collected in a total electron yield mode under high
vacuum (residual pressure <1 ꢁ 10ꢀ6 Torr) at room tempera-
ture. The sample was put on the first Cu–Be dynode of the
electron multiplier (ꢀ1.5 kV) by using an adhesive carbon
tape.
Experimental
Materials
Diffuse reflectance UV–vis spectra were recorded on a
JASCO V-570 equipped with an integrating sphere covered
with BaSO4 . Phosphorescence spectra were recorded at 77 K
with a Hitachi F-4500 fluorescence spectrophotometer using
UV filter (ltransmittance > 330 nm) to remove scattered light
from UV source (Xe lamp) and an attached apparatus to cut
fluorescence. The excitation wavelength employed was 300
nm. Phosphorescence from the optical cell was weak enough
to ignore in most cases. Before recording UV–visible absorp-
tion spectra and phosphorescence spectra, the sample (0.5 g)
was treated with 60 Torr (1 Torr ¼ 133 Pa) O2 for 1 h at
1073 K in a pretreatment part of an in situ cell, followed by
evacuation for 1 h at the same temperature. Then, the sample
was transferred to the optical part without exposing to air.
The samples of silica-alumina were prepared by the sol–gel
method.22 After a mixture of Si(OEt)4 and ethanol (mol ratio,
1/5) was stirred for 1 h, another mixture of ethanol, distilled
water, Al(NO3)3 and nitric acid (mol ratio, 5/50/x/0.02,
x ¼ 0.02 ꢀ 2) was added dropwise, followed by stirring for
1 h. Dehydration proceeded at 353 K with suction. When gela-
tion started, the mixture was dried without stirring, followed
by drying at 383 K for 12 h, grinding in air at room tempera-
ture and calcination in a flow of dry air (60 ml minꢀ1) at 773 K
for 5 h to obtain the silica-alumina powder samples. The Al
content (mol%) calculated from the amount of the reagents
used was defined as a molar ratio of the number of Al atoms
to the total number of Si and Al atoms in the silica-alumina;
NAl/(NSi + NAl) ꢁ 100.
Pure silica sample was prepared from Si(OEt)4 by the sol–gel
method in a water–ethanol mixture at boiling point followed
by grind and calcination.23 Pure alumina was prepared by cal-
cination of Al(NO3)3 . They were stored under ca. 80% humid-
ity at room temperature.
Results
Structure of the silica-alumina samples
Two reference samples were employed for Al K-edge
XANES study.24,25 MFI-type aluminosilicate (JRC-Z5-
25H,26 SiO2/Al2O3 ¼ 25, known as H-ZSM-5) containing Al
atoms in the zeolite framework, tetrahedral site, was supplied
from the Catalysis Society of Japan. y-Alumina, in which Al
atoms are equally distributed between the tetrahedral and
octahedral sites of spinel, was prepared by calcination of g-alu-
mina (JRC-ALO-426,27 supplied from the Catalysis Society of
Japan) at 1073 K for 12 h.
We prepared a series of silica-alumina containing 0–100% of
Al. Table 1 shows the specific surface areas of the silica-alu-
mina samples as synthesized (calcined) and those of the sam-
ples pretreated at 1073 K. All the samples had a specific
surface area larger than 140 m2 gꢀ1 after calcination at 773
K. Even after pretreatment at 1073 K, the samples maintained
a high surface area larger than 90 m2 gꢀ1. Although the pre-
treatment at high temperature reduced the specific surface area
of the samples to some extent, high temperatures such as
1073 K were employed for the sample pretreatment in this
study. This is because the yield of photoinduced methane cou-
pling increased with an increase of the pretreatment tempera-
ture in spite of the decrease of specific surface area.21 It was
suggested that surface dehydration was related to formation
of the active sites.21,30
The silica and the silica-alumina samples of lower than 50
mol% Al were basically amorphous structure as shown by
XRD (Fig. 1(a)–(c)). Other phases such as alumina, zeolite
and layered structures were not observed. As for the high Al
sample of 66.7 mol% Al (Fig. 1(d)), broad diffraction lines
were observed at 46ꢂ and 67ꢂ, which were assignable to bulk
g-Al2O3 . The broadness of the diffraction means that the alu-
Methane (99.95%) was purified by a vacuum evaporation
before every use.
Photoreaction test
The reaction test was carried out in a closed quartz reaction
vessel (82 cm3). The sample (1.0 g) was spread on the flat bot-
tom of the vessel (19.6 cm2). Before the reaction test, the sam-
ple was heated in air up to 1073 K and then evacuated.
Subsequently the sample was treated with 60 Torr oxygen (1
Torr ¼ 133.3 N mꢀ2) at 1073 K for 1 h, followed by evacua-
tion at the same temperature for 1 h and then by introduction
of methane at room temperature. The initial pressure of
methane (200 mmol) in the reactor was 45 Torr and no oxidant
molecule was introduced into the reaction system. The sample
was irradiated with a 250 W Xe lamp for 6 h. Under photoir-
radiation, the temperature of sample bed was measured to be
ca. 310 K. Products in the gaseous phase were analyzed by
GC. Then adsorbed products were thermally desorbed by heat-
ing (573 K, 15 min), collected, and analyzed.
Table 1 Al contents and specific surface areas of the synthesized sam-
ples
BET surface area/m2 gꢀ1
Al content/mol%a
As synthesized After pretreatment at 1073 K
0.0b
2.0
633
464
498
489
264
143
149
241
297
137
258
273
203
98
Characterizations
Specific surface areas (BET) were calculated from N2-adsorp-
tion at 77 K. Powder X-ray diffraction (XRD) patterns were
recorded on a Rigaku diffractometer RINT 1200 using a radia-
tion of Ni-filtered Cu-Ka radiation (40 kV, 20 mA). X-ray
photoelectron spectra (XPS) were recorded with a Shimadzu
ESCA-3300 spectrometer using Mg-Ka radiation (10 kV, 20
mA, 1253.6 eV) to calculate surface concentrations of Al and
Si estimated by the band area intensity of Al 2p and Si 2s,
respectively.
9.1
16.7
37.5
50.0
66.7
100c
93
92
a
Al content (mol%) calculated from the amount of the reagents used
was defined as a molar ratio of the number of Al atoms to the total
number of Si and Al atoms in the silica-alumina; NAl
b
(NSi + NAl) ꢁ 100. The silica sample. The alumina sample.
/
Soft X-ray absorption experiments were carried out on the
beam line 7A28 at UVSOR, Institute for Molecular Science,
c
2460
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465

View Article Online
Fig. 1 XRD patterns of the silica sample (a), the silica-alumina sam-
ples whose Al contents (mol%) was 25.0 (b), 50.0 (c), 66.7 (d), and the
alumina sample (e).
mina crystallites would have small size or low crystallinity. The
alumina sample (Fig. 1(e)) showed the diffraction pattern
assigned to g-Al2O3 .
The Al composition at surface layers of the silica-alumina
samples is estimated by XPS analysis and plotted as a function
of the Al content (Fig. 2). A good linear correlation was
obtained as shown, meaning that the Al ions/species were
homogeneously dispersed on the surface and bulk of the silica.
In the present paper, the Al content is employed as the Al sur-
face composition.
Al K-edge XANES is useful to determine the aluminium
coordinations (AlO4 tetrahedra and AlO6 octahedra) in oxide
environments.24,25,31–34 Fig. 3 shows the normalized Al K-edge
XANES spectra of the silica-alumina samples and the alumina
samples used for the photoreaction (Fig. 3(a)–(g)), together
with those of reference compounds. The white line peak at
1566 eV is assigned to the tetrahedral AlO4 compounds, while
the peaks at 1568 eV and at 1572 eV are assigned to octahedral
AlO6 compounds.31,32 However, since the peak at 1572 eV is
not suitable for quantitative analysis,24,25 the peaks at 1566
eV and at 1568 eV are employed for the estimation of the frac-
tion of the tetrahedral AlO4 species and the octahedral AlO6
species, respectively. The MFI-type zeolite having AlO4 species
shows a peak at 1566 eV (Fig. 3(h)), while the y-alumina,
which consists of the AlO4 and AlO6 species equally, shows
almost the same intensity of peaks at 1566 eV and 1568 eV
(Fig. 3(i)).24,25,35
Fig. 3 Normalized Al K-edge XANES spectra of the silica-alumina
samples used for the photoreaction, whose Al contents (mol%) was
2.0 (a), 9.1 (b), 16.7 (c), 37.5 (d), 50.0 (e), 66.7 (f), and those of the alu-
mina sample used for the photoreaction (g), the MFI-type zeolite hav-
ing AlO4 (h) and y-alumina which contains equal amount of AlO4 and
AlO6 (i).
higher energy with an increase of Al content, which is probably
due to the overlap of the band at 1568 eV (Fig. 3(c)–(f)). The
sample of 16.7 mol% Al seems to have a very small fraction of
AlO6 species and a dominant fraction of AlO4 (Fig. 3(c)), while
the samples of higher Al content have larger fractions of AlO6
(Fig. 3(d)–(f)). The alumina sample showed the obvious peak
at 1568 eV (Fig. 3(g)) together with the shoulder at 1566 eV.
By comparison with the spectra of y-alumina, there are almost
an even amount of AlO4 and AlO6 species.
The spectra of the silica-alumina samples gradually varied
with the Al content (Fig. 3(a)–(f)). This means that the local
structure of Al species in silica-alumina gradually varied with
the Al content. The low Al content samples of 2 and 9.1
mol% Al (Fig. 3(a), (b)) show a peak at around 1566 eV, which
is assigned to the AlO4 species. This peak was broadened to
Thus, the AlO4 species are predominant in the lower Al con-
tent silica-alumina samples, and the fraction of the AlO6 spe-
cies in the silica-alumina samples increased gradually with an
increase of Al content. In the high Al content samples and alu-
mina sample, the amount of the AlO4 species and the AlO6
species are almost even.
Photoabsorption and phosphorescence of the silica-alumina
samples
Fig. 4 shows the diffuse reflectance UV–vis spectra of the sam-
ples evacuated at 1073 K. All the silica-alumina samples
showed the maximum absorbance around 230 nm and the
intensity varied with Al content. The maximum absorption
intensity around 230 nm was plotted in Fig. 5. The absorption
intensity increased with an increase of Al content up to around
20 mol%, then gradually decreased. The intense photoabsorp-
tion should be related to a certain Si–O–Al sites in silica-alu-
mina, since silica and alumina themselves exhibited low
absorption as shown at 0% and 100% in Fig. 5, respectively.
It has been reported that the highly dispersed AlO4 species in
silica-alumina evacuated at high temperature such as 1073 K
exhibit a characteristic fine structural phosphorescence spec-
trum.19,21,30 Fig. 6 shows the phosphorescence emission spectra
Fig. 2 Relationship between the surface Al composition estimated by
XPS and the Al content calculated from the amount of reagents used.
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465
2461

View Article Online
Fig. 4 Diffuse reflectance UV–vis spectra of the silica (a) and the
silica-alumina samples whose Al content (mol%) was 2.0 (b), 9.1 (c),
16.7 (d), 37.5 (e), 50.0 (f), 66.7 (g), and that of the alumina sample
(h). The samples were pretreated at 1073 K.
Fig. 6 Normalized phosphorescence spectra of the silica sample (a),
the silica-alumina samples whose Al contents (mol%) was 2.0 (b), 9.1
(c), 16.7 (d), 37.5 (e), 50.0 (f), 66.7 (g), and the alumina sample (h) mea-
sured at 77 K. The excitation wavelength was 300 nm. The samples
were pretreated at 1073 K.
of the present samples evacuated at 1073 K. The intensity in
Fig. 6 was normalized by each appropriate factor to discuss
the line shape. The pure silica exhibited a broad band with a
maximum around 460 nm (Fig. 6(a)). On the spectra of the
silica-alumina samples of 2.0–37.5 mol% Al (Fig. 6(b)–(e)),
the fine structure was clearly observed as reported, indicating
that there are highly dispersed AlO4 species existing in the
silica matrix. The progressions show that photoexcitation
occurs a certain localized site, probably due to vibration of
M–O (M ¼ Al or Si) moiety related to AlO4 species. As the
Al content increases further, the fine structure became unclear,
and finally the alumina sample showed a broad band centered
around 470–480 nm (Fig. 6(h)). The spectra of the high Al con-
tent samples (Fig. 6(f) and 6(g)) resemble that of the alumina
sample (Fig. 6(h)).
coupling around room temperature. Only ethane was obtained
as the gaseous product under this condition, except for the
sample of 16.7 mol% Al; on the 16.7 mol% Al silica-alumina
sample which showed the maximum conversion, a trace
amount of propane was also produced. Ethene was detected
as a thermally desorption product on some samples of high
Al contents. Molecular hydrogen was produced stoichiometri-
cally within experimental error in this system, as reported else-
where.21
The yields of ethane and adsorbed ethene are shown in Fig.
8(a) and 8(b) as a function of the Al content. The reproducibil-
ity was confirmed on some representative points. The curve for
the ethane yield (Fig. 8(a)) shows a maximum at around 20
mol% Al. Note that the curve showing the ethane yield in
Fig. 8(a) resembles the curves showing the photoabsorption
intensity in Fig. 5 and the phosphorescence intensity of the
samples below 20 mol% Al in Fig. 7. On the other hand, the
ethene yield increased with Al content, especially over 20
mol% (Fig. 8(b)). The selectivity to ethane (Fig. 8(c)) was ca.
100% on the silica and silica-alumina samples below 20
The phosphorescence intensity (at 520 nm) of the samples
was plotted in Fig. 7. Among all samples, the four silica-alu-
mina samples of low Al content (2.0–37.5 mol% Al) showed
high intensity with the fine structure, and the maximum is
around 20 mol% Al.
Photoinduced methane coupling activity of the silica-alumina
samples
All these prepared samples, silica, silica-alumina and alumina,
exhibited activity for the photoinduced non-oxidative methane
Fig. 5 The maximum absorption intensity around 230 nm in the dif-
fuse reflectance UV–vis spectra shown in Fig. 4. As for the alumina
sample (the point of 100 mol%), the intensity at 230 nm was employed.
Fig. 7 The phosphorescence intensity at 520 nm of the samples evac-
uated at 1073 K. The excitation wavelength was 300 nm.
2462
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465

View Article Online
atomically substitute for silicon in silica tetrahedra network,
and the Al species should be AlO4 tetrahedra.
On the other hand, when the aluminium content increases,
the AlO4 species could no longer be isolated and small Al oxide
clusters and alumina crystallites that consist of both AlO4 and
AlO6 should be also formed in the sample. It is supposed that
this variation of Al species in the silica-alumina occurs mono-
tonously with the increasing of the Al content. Judging from
XANES spectrum (Fig. 3(c)), a very small fraction of AlO6
that most probably belongs to small Al oxide clusters seemed
to exist in the 16.7 mol% Al sample. However, it would be
recognized that the major species are still the highly dispersed
AlO4 species. The sample of 16.7 mol% Al exhibited the max-
imum intensity on the phosphorescence emission spectra with
the fine structure (Figs. 6 and 7). This means that there is
the maximum amount of the highly dispersed AlO4 species
among the series of the silica-alumina samples. The illustrated
model of this state is proposed in Fig. 9(b) though small Al
oxide cluster species in silica matrix as a very minor species
are not shown here.
As for the sample of 37.5 mol% Al, XANES spectrum (Fig.
3(d)) indicated the existence of both AlO4 and AlO6 species
more clearly than that of the lower Al content samples, while
the phosphorescence spectrum (Fig. 6) showed that a small
fraction of highly dispersed AlO4 species still exists. On the
37.5 mol% Al sample, it is suggested that a small fraction of
highly dispersed AlO4 species coexist with small Al oxide clus-
ters consisting of AlO4 and AlO6 as shown in Fig. 9(c).
On the high Al content samples such as 50 and 66.7 mol%
Al, the fraction of the AlO6 species became larger as clarified
by XANES spectra (Fig. 3(e) and 3(f)). Moreover, in these
samples, the line shapes of the phosphorescence spectra (Fig.
6(f) and 6(g)) were different from that of the low Al sample,
but resemble rather that of alumina (Fig. 6(h)), and the phos-
phorescence intensity was very low (Fig. 7). These findings
support the idea that the more aggregated Al oxide species
such as alumina crystallites became the major species in these
samples at the sacrifice of the highly dispersed AlO4 species
and small Al oxide clusters. In the case of the 66.7 mol% Al
sample, the g-alumina phase was detected by XRD (Fig. 3),
confirming that the alumina crystallites exist; this sample
would contain a larger amount/size of crystallites than the
sample of 50.0 mol% Al.
Fig. 8 The yield of ethane (a) and adsorbed ethene (b), and the
ethane selectivity (c) in photoinduced non-oxidative methane coupling
over a series of the silica sample, the silica-alumina samples and the
alumina sample.
mol% Al content, and decreased almost linearly with increas-
ing Al content over 20 mol% Al. On the alumina sample, the
selectivity was ca. 70%.
Discussion
Structural aspects of the aluminium species in silica-alumina
The silica-alumina samples below 50 mol% Al showed similar
XRD patterns to silica but no diffraction lines due to alumina
(Fig. 1(a)–(c)). With low Al content samples (2.0 and 9.1
mol%), XANES spectra indicate that the dominant aluminium
species were coordinated by four oxygen atoms (Fig. 3). Both
samples of 2.0 and 9.1 mol% Al showed the fine structural
phosphorescence spectra due to highly dispersed AlO4 species
(Fig. 6). These results indicate that the major aluminium spe-
cies in these low Al silica-alumina samples were highly dis-
persed AlO4 species. The proposed model is shown in
Fig. 9(a). The highly dispersed AlO4 species are proposed to
be isolated in the SiO4 tetrahedra network, or, in other words,
they are surrounded by SiO4 species as the first neighbor tetra-
hedral units. When Al content is low, the aluminium would
As a conclusion about the structure of the silica-alumina
samples, the variation in the amount of each Al species with
increasing the Al contents is summarized here. The amount
of the highly dispersed AlO4 species increased with an increase
of the Al content to reach a maximum in the sample of around
20 mol% Al, then decreased. The amount of small Al oxide
Fig. 9 Tentatively proposed models for some major AlO4 species in the silica-alumina samples of 2 and 9 mol% Al (a), 16.7 mol% Al (b), and 37.5
mol% Al (c). Highly dispersed AlO4 species are in the silica-alumina samples less than 16.7 mol% Al, while small Al oxide clusters consisting of
AlO4 and AlO6 exist in the high Al content samples.
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465
2463

View Article Online
clusters also gradually increased at least from around 20 mol%
Al at the sacrifice of the highly dispersed AlO4 species. Then, in
the high Al content samples, the highly dispersed AlO4 species
decreased further while the Al oxide clusters and crystallites
became major. In this way, although the variation of the spe-
cies should be monotonous, the amount of the highly dispersed
species shows a maximum, and then decreases with increasing
the Al content.
of existence of other active sites on the silica surface. Fig. 10
shows that the average specific activity of the Al sites in
silica-alumina samples decreased obviously with an increase
of the Al content. This means that the decrease of the product
yield on the samples of higher Al content than 20% Al would
not be only due to the decrease in the surface area but also
due to the decrease in the average specific activity of each Al
sites.
In addition to this variation of the average specific activity
of the Al sites (Fig. 10), the ethane yield showed the maximum
around 20 mol% Al, then decreased with increasing the Al con-
tent (Fig. 8(a)), while the selectivity of ethane also started to
decrease around 20 mol% Al (Fig. 8(c)). These variations on
the activity would be explained by the amount and perfor-
mance of the representative two Al species; one is the highly
dispersed AlO4 species (Fig. 9(a), 9(b)) and another is the
aggregated Al oxide species (major species in Fig. 9(c)).
The silica-alumina samples of low Al content less than 16.7
mol% have predominantly the highly dispersed AlO4 species.
In the qualitative view, the highly dispersed AlO4 species
exhibited high specific activity per surface aluminium (Fig.
10(a)) and high ethane selectivity (Fig. 8(c)). In the quantita-
tive view, the sample of 16.7 mol% Al, which has the largest
amount of this species, exhibited the highest ethane yield
(Fig. 8(a)) and the most intense UV absorption band (Fig.
4(b)–(d)). Thus the highly dispersed AlO4 species are effectively
excited by UV light and exhibit the high activity for the selec-
tive ethane formation.
The highly dispersed AlO4 species exhibited fine structural
phosphorescence spectra when they were excited by UV light.
The properties of these photoexcited sites were well studied by
monitoring the phosphorescence emission spectra,30 and it was
clarified that the photoexcited sites could activate the methane
molecules.21 In the present study, a good correlation was
revealed between the phosphorescence intensity and the
photoactivity for this reaction among the low Al content
samples less than 20 mol%. This strongly supports that the
phosphorescence sites act as the highly active sites for ethane
formation.
The photoabsorption and photoreaction activity of alumina
species
For the photoinduced reaction, photoexcitation is the essential
process. Though methane molecules absorb light in the
vacuum-UV region, such as 94, 120 and 128 nm wavelength,
they cannot absorb the light in the region of ultraviolet and visi-
ble light that come from the Xe-lamp through air. Without the
silica-alumina samples, methane could not be converted even
upon photoirradiation. As mentioned above, the silica-alumina
samples showed a photoabsorption band in the UV region
around 230 nm (Fig. 4), and the variation of the photoabsorp-
tion intensity with the Al content (Fig. 5) was in good agreement
with the activity for the photoinduced non-oxidative methane
coupling (Fig. 8(a)). These findings strongly suggest that the
surface photoabsorption sites on the silica-alumina would first
absorb the light around 230 nm to be excited, and then would
transfer the energy to the adsorbed methane molecule.
The pure silica (0 mol% Al) also exhibited activity for this
reaction. As mentioned above, the pure silica materials have
been reported to promote some photoinduced reactions.8–14
In those cases, the surface defects or surface activated sites
on silica are thought to be the active sites. Also in the photo-
induced non-oxidative methane coupling, these surface sites
might be active sites when there is no aluminium in the sample.
However, the activities on all silica-alumina samples were
higher than the pure silica. Thus, it is suggested that the major
active sites on silica-alumina are aluminium-related sites.
Assuming that only the Al sites are the active sites on silica-
alumina, the relative specific activity was estimated: The pro-
duct yield per hour was divided by the BET specific surface
area and by the Al composition. The values for ethane and
adsorbed ethene were plotted in Fig. 10. Here, although the
silica-alumina sample of 2 mol% Al (the plot for ethane was
not shown) showed too high a value alone (8 times larger than
the sample of 9.1 mol% Al), it might be overestimated because
As the Al content increased to more than 50 mol% Al, the
product selectivity for ethane decreased, and the formation
of adsorbed ethene increased, as shown in Fig. 8. This would
correspond to the formation of the small Al oxide clusters
and crystallites (AlO4 and AlO6 species). Thus, it is demon-
strated that these aggregated species, such as the Al oxide clus-
ters and crystallites, in silica-alumina produce both ethane and
ethene.
Although the aggregated alumina species such as clusters
and crystallites also absorbed photoenergy and actually pro-
moted the coupling of methane, the weak phosphorescence
spectra without the fine structure were clearly different from
those of the highly dispersed Al species. The aggregated Al
species would activate methane molecules upon photoirradia-
tion by a different photoexcitation mechanism to those on
the highly dispersed AlO4 species, resulting in the formation
of both ethane and ethene.
Conclusion
The silica-alumina samples in the entire range of Al concentra-
tion exhibit the activity for the photoinduced non-oxidative
methane coupling. The highly dispersed AlO4 species, which
are dominant in the low Al silica-alumina samples and exhibit
the fine structural phosphorescence spectra, are the highly
active sites for the selective ethane production. On the other
hand, the aggregated Al oxide species such as Al oxide clusters
and crystallites, which are major in the high Al silica-alumina,
are the active sites for the formation of both ethane and
ethene.
Fig. 10 The relative specific activity of aluminium sites for the for-
mation of ethane (a) and ethene (b) in photoinduced non-oxidative
methane coupling over a series of the silica-alumina samples. The rela-
tive specific activity is the yield per hour normalized by the specific sur-
face area and the Al composition. The plot of the sample of 2 mol% Al
for ethane was eliminated (see text).
2464
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465

View Article Online
14 H. Yoshida, T. Tanaka, M. Yamamoto, T. Yoshida, T. Funabiki
and S. Yoshida, J. Catal., 1997, 171, 351.
Acknowledgement
15 M. Anpo, C. Yun and K. Kubokawa, J. Catal., 1980, 61, 267.
16 Y. Kubokawa, M. Anpo and C. Yun, Proc. 7th Int. Congress
Catal. (B), 1980, 1171.
17 H. Yoshida, C. Murata, Y. Inaki and T. Hattori, Chem. Lett.,
1998, 1121.
18 A. Ogata, A. Kazukawa and M. Enyo, J. Phys. Chem., 1986, 90,
5201.
19 H. Yoshida, T. Tanaka, A. Satsuma, T. Hattori, T. Funabiki and
S. Yoshida, Chem. Commun., 1996, 1153.
20 Y. Kato, H. Yoshida and T. Hattori, Chem. Commun., 1998,
2389.
21 H. Yoshida, Y. Kato and T. Hattori, Stud. Surf. Sci. Catal., 2000,
130, 659.
22 M. Taira and M. Yamaki, J. Mater. Sci.: Mater. Med., 1995, 6,
197.
The authors thank Professor Atsushi Satsuma for valuable dis-
cussions. Dr Ken-ichi Shimizu and Dr Tomoko Yoshida are
acknowledged for their cooperation on the XAFS measure-
ments. X-ray absorption experiments were carried out as the
Joint Studies Program (1997) of UVSOR of the Institute for
Molecular Science. H. Y. acknowledges the Nippon Sheet
Glass Foundation for Materials Science and Engineering,
The Nitto Foundation, and Japan Chemical Innovation Insti-
tute (JCII) for the financial support. This work was also sup-
ported by a grant-in-aid for Exploratory Research from the
Ministry of Education, Culture, Sports, Science and Technol-
ogy (MEXT), Japan.
23 H. Yoshida, C. Murata and T. Hattori, J. Catal., 2000, 194,
364.
24 K. Shimizu, Y. Kato, T. Yoshida, H. Yoshida, A. Satsuma and
T. Hattori, Chem. Commun., 1999, 1681.
References
1
2
Y. Xu and L. Lin, Appl. Catal. A, 1999, 188, 53.
H. D. Gesser and N. R. Hunter, Catal. Today, 1998, 42, 183 and
references therein.
25 Y. Kato, K. Shimizu, N. Matsushita, T. Yoshida, H. Yoshida, A.
Satsuma and T. Hattori, Phys. Chem. Chem. Phys., 2001, 3, 1925.
26 T. Uchijima, Catalytic Science and Technology, ed. S. Yoshida, N.
Takazawa and Y. Ono, Kodansha, VCH, Tokyo, 1991, vol. 1,
p. 393.
27 Y. Murakami, Stud. Surf. Sci. Catal., 1983, 16, 775.
28 T. Murata, T. Matsukawa, S. Naoe, T. Horigome, O. Matsudo
and M. Watanabe, Rev. Sci. Instrum., 1992, 63, 1309.
29 T. Kinoshita, Y. Takata, T. Matsukawa, H. Aritani, S. Matsuo,
T. Yamamoto, M. Takahashi, H. Yoshida, T. Yoshida, Y. Ufuk-
tepe, K. G. Nath, S. Kimura, H. Kumigashira and Y. Kitajima,
J. Synchrotron Rad., 1998, 5, 726.
3
4
5
6
7
8
9
K. Okabe, K. Sayama, H. Kusama and H. Arakawa, Chem. Lett.,
1997, 457.
S. L. Kaliaguine, B. N. Shelimov and V. B. Kazansky, J. Catal.,
1978, 55, 384.
G. N. Kuzmin, M. V. Knatko and S. V. Kurganov, React Kinet.
Catal. Lett., 1983, 23, 313.
W. Hill, B. N. Shelimov and V. B. Kazansky, J. Chem. Soc., Fara-
day Trans. 1, 1987, 83, 2381.
A. Morikawa, M. Hattori, K. Yagi and K. Otsuka, Z. Phys.
Chem., N. F., 1977, 140, 309.
H. Yoshida, T. Tanaka, S. Matsuo, T. Funabiki and S. Yoshida,
J. Chem. Soc., Chem. Commun., 1995, 761.
T. Tanaka, S. Matsuo, T. Maeda, H. Yoshida, T. Funabiki and
S. Yoshida, Appl. Surf. Sci., 1997, 121/122, 296.
30 Y. Kato, H. Yoshida and T. Hattori, Phys. Chem. Chem. Phys.,
2000, 2, 4231.
31 D. A. McKeown, G. A. Waychunas and G. E. Brown, Jr., J. Non-
Cryst. Solids, 1985, 74, 349.
32 D. Cabaret, P. Sainctavit, P. Ildefonse and A.-M. Flank, J. Phys.:
Condens. Matter, 1996, 8, 3691.
10 H. Yoshida, K. Kimura, Y. Inaki and T. Hattori, Chem. Com-
mun., 1997, 129.
33 M. Fro¨ba and M. Tiemann, Chem. Mater., 1998, 10, 3475.
34 J. A. van Bokhoven, H. Sambe, D. E. Ramaker and D. C.
Koningsberger, J. Phys. Chem. B, 1999, 103, 7557.
35 Although the peak height at 1568 eV seems larger than that at
1566 eV, it is mainly due to the overlapping of the continuum
absorption of AlO4 species that appears at lower energy.
11 Y. Inaki, H. Yoshida, K. Kimura, S. Inagaki, Y. Fukushima and
T. Hattori, Phys. Chem. Chem. Phys., 2000, 2, 5293.
12 Y. Inaki, H. Yoshida and T. Hattori, J. Phys. Chem. B, 2000, 104,
10 304.
13 H. Yoshida, T. Tanaka, M. Yamamoto, T. Funabiki and
S. Yoshida, Chem. Commun., 1996, 2125.
Phys. Chem. Chem. Phys., 2002, 4, 2459–2465
2465