State-sensitive monitoring of gold nanoparticle sites on titania and the interaction of the positive Au site with O2 by Au Lα1-selecting X-ray absorption fine structure

Article abstract of DOI:10.1016/j.ica.2007.09.027

The heterogeneity of gold sites in various Au/TiO₂ catalysts was studied by means of state-sensitive Au L₃-edge X-ray absorption fine structure (XAFS) combined with high energy-resolution X-ray fluorescence spectrometry. A series of Au/TiO₂ catalysts were prepared via deposition-precipitation method on anatase-type or mesoporous (amorphous) TiO₂ added with NaOH (lower Au loading) or urea (higher Au loading). The mean Au particle size ranged between 29 and 87 ? based on high-resolution TEM (transmission electron microscope) measurements. The Au Lα₁ emission peak energy for Au/mesoporous-TiO₂ in air and Au/anatase-TiO₂ in CO (5%) corresponded to Au⁰ state. The emission peak energy for Au/anatase-TiO₂ in air shifted toward that of Au^I state. For relatively greater Au particles (average 87 ?) dispersed on mesporous TiO₂, the major valence state discriminated by Au Lα₁-selecting XANES (X-ray absorption near-edge structure) spectrum tuned to Au Lα₁ emission peak top was Au⁰, but the Au^δ- state could be successfully monitored by Au Lα₁-selecting XANES tuned to the emission energy at 9707.6 eV, of which population was relatively small compared to the case of smaller Au particles (average 29 ?) on anatase-type TiO₂. On the other hand, negative charge transfer from Au 5d to support was demonstrated in Au^δ+-state sensitive XANES tuned to 9718.3-9718.7 eV. The Au^δ+-state sensitive XANES spectra resembled theoretically generated XANES for interface Au^δ+ sites model on TiO₂ in contact with surface Ti sites. Further charge transfer was demonstrated from Au to adsorbed O₂ for Au/anatase-TiO₂ catalyst.

Full text of DOI:10.1016/j.ica.2007.09.027

Available online at www.sciencedirect.com
Inorganica Chimica Acta 361 (2008) 1149–1156
www.elsevier.com/locate/ica
State-sensitive monitoring of gold nanoparticle sites on titania
and the interaction of the positive Au site with O by
2
q
Au La -selecting X-ray absorption fine structure
1
a,
b
b
b
*
Yasuo Izumi , Diaa Mosbah Obaid , Kazushi Konishi , Dilshad Masih ,
c
c
c
c,
*
Masafumi Takagaki , Yasuko Terada , Hajime Tanida , Tomoya Uruga
a
Graduate School of Science, Chiba University, Yayoi 1-33, Inage-ku, Chiba 263-8522, Japan
Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, Nagatsuta 4259, Midori-ku, Yokohama 226-8502, Japan
b
c
Japan Synchrotron Radiation Research Institute, Kohto 1-1-1, Sayo 679-5198, Japan
Received 17 April 2007; received in revised form 14 September 2007; accepted 16 September 2007
Available online 29 October 2007
Dedicated to Edward Solomon.
Abstract
The heterogeneity of gold sites in various Au/TiO
structure (XAFS) combined with high energy-resolution X-ray fluorescence spectrometry. A series of Au/TiO
2
catalysts was studied by means of state-sensitive Au L
3
-edge X-ray absorption fine
catalysts were prepared
2
via deposition-precipitation method on anatase-type or mesoporous (amorphous) TiO added with NaOH (lower Au loading) or urea
2
˚
(
higher Au loading). The mean Au particle size ranged between 29 and 87 A based on high-resolution TEM (transmission electron micro-
1 2 2
scope) measurements. The Au La emission peak energy for Au/mesoporous–TiO in air and Au/anatase–TiO in CO (5%) corresponded
0
I
to Au state. The emission peak energy for Au/anatase–TiO in air shifted toward that of Au state. For relatively greater Au particles
2
˚
(
average 87 A) dispersed on mesporous TiO
near-edge structure) spectrum tuned to Au La
2
, the major valence state discriminated by Au La
1
-selecting XANES (X-ray absorption
0
dꢀ
1
emission peak top was Au , but the Au state could be successfully monitored by
Au La -selecting XANES tuned to the emission energy at 9707.6 eV, of which population was relatively small compared to the case
1
˚
of smaller Au particles (average 29 A) on anatase-type TiO
2
. On the other hand, negative charge transfer from Au 5d to support was
d+
d+
demonstrated in Au -state sensitive XANES tuned to 9718.3–9718.7 eV. The Au -state sensitive XANES spectra resembled theoret-
ically generated XANES for interface Au sites model on TiO
strated from Au to adsorbed O for Au/anatase–TiO catalyst.
d+
2
in contact with surface Ti sites. Further charge transfer was demon-
2
2
ꢀ
2007 Elsevier B.V. All rights reserved.
Keywords: Gold; Titanium oxide; X-ray absorption fine structure; Fluorescence spectrometry; Chemical state
1
. Introduction
Various possibilities have been proposed as the active sites:
anionic Au cluster site, perimeter site, and lower-coordina-
tion step-edge site [1,2]. However, it is difficult to observe
each site one by one among Au nano-particles dispersed
over porous, complex surface of TiO2.
Gold is chemically inert, however, it was found to
become active when dispersed as nano-particles of size
3
˚
0 A or smaller in low temperature CO oxidation [1,2].
In order to overcome the heterogeneity problem of Au
sites, we applied state-sensitive XAFS spectroscopy [3,4]
q
Part 23 in the series.
Corresponding authors. Tel./fax: +81 43 290 3696.
E-mail addresses: yizumi@faculty.chiba-u.jp (Y. Izumi), urugat@
to a series of Au/TiO samples. Due to the differences of
*
2
deposition-precipitation (DP) conditions of Au compound
with NaOH or urea, reaction time of DP, and crystalline
spring8.or.jp (T. Uruga).
0
020-1693/$ - see front matter ꢀ 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2007.09.027

1150
Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
phase of TiO support (anatase/rutile or amorphous), par-
4–16 h in the absence of light [11]. Then, the filtered pow-
der was processed similar to the preparation using NaOH
and finally calcined at 573 K in air (Table 1d–g). Thus,
2
ticle size and loading amount of Au were systematically
controlled. Catalytically active Au nano-particle sites pop-
˚
ulated in narrow size distribution centered at 29 A were
the obtained Au/TiO catalysts are denoted ‘‘Au/(type of
2
monitored state-sensitively in CO or O atmosphere to
TiO )-(alkaline used)-(DP time)-(calcined temperature)’’,
2
2
extract activation sites for the low-temperature oxidation
reaction.
e.g. Au/P25-urea-16-573.
The Au/TiO sample disks were sealed in on-reaction
2
In this spectroscopic method, better-resolved, steeper
XANES spectra were obtained to remove the core-hole
lifetime width (C) of K levels for Cu [4,5] and Sn [4] and
cell for X-ray measurements purged with dry air or with
the mixture of CO (5%) and argon (95%). The window
was polyethylene naphthalate film of 16 lm in thickness
(Q51-16, Teijin).
L levels for Dy [6], Pt [7], and Pb [4,8] by selecting the
3
Ka and La emission, respectively.
1
1
2.2. Characterization
2
. Experimental
Transmission electron microscope images were mea-
2
.1. Samples
sured using field-emission-type TEM (JEM-2010F, JEOL)
with an accelerating voltage of 200 kV at the Center of
Advanced Materials Analysis, Tokyo Institute of Technol-
ogy (Dr. A. Genseki). The powder was dispersed in carbon
tetrachloride using ultrasonic and mounted on carbon-
coated Cu grid.
The X-ray measurements were performed at 290 K at
the undulator beamline 37XU of SPring-8 (Sayo, Japan).
The storage ring energy was 8.0 GeV operated in top-up
mode at the ring current of 99–97 mA. A Si(111) mono-
chromator and Rh-coated mirror were used. The undulator
gap value was optimized to maximize the X-ray beam flux
at each data point. To stabilize the X-ray beam position on
the surface of the sample, a feedback mechanism using
MOSTAB (monochromator stabilization) [12] was applied.
No position change was detected within the range of 50 lm
throughout the beamtime.
+
I
ꢀ
2
K [Au (CN) ] (Wako) was thoroughly mixed with
boron nitride (BN) to be 1.0 wt% Au and pressed into a
disk of 20 mm in diameter. Au metal foil was controlled
to be the thickness of 6 lm and powders of Au Cl-
I
III
[
P(C H ) ] (>99.9%, Aldrich), Au (OH) (Wako), and
6
5 3
3
III
2
Au O (Wako) were thoroughly mixed with BN and
pressed into a disk of 20 mm in diameter, to make the
Au L absorption edge jump to be 1.
3
3
The mesoporous TiO sample was synthesized in aque-
2
ous solution using dodecylamine [9] given by Professor
H. Yoshitake at Yokohama National University. The spe-
cific surface area (SA) was 1000 m g and wormhole-like
pore size distributed in narrow range centered at 30 A. The
Au/TiO2 catalyst was prepared by mixing 1 g of TiO2
2
ꢀ1
˚
(
Degussa P25; anatase 70% and rutile 30%) or mesoporous
III
TiO with 50 ml of aqueous solution of HAu Cl Æ 4H O
A homemade Rowland-type fluorescence spectrometer
was used to analyze Au La1 emission equipped with
2
4
2
(
3
Wako, 1.3 mM) (Table 1a–c). The mixture was stirred at
28 K for 4.5–7 h. The pH of the solution was maintained
˚
Johansson-type Ge(555) crystal (d = 0.65321 A, Saint
at 8 by adequately adding NaOH [10]. The powder was fil-
tered and washed several times with de-ionized, distilled
water. Dried powders were calcined at 523–673 K in air.
Gobain). The detail of the spectrometer was described in
Refs. [4,13,14]. An ionization chamber was purged with
the mixture of nitrogen (30%) and helium (70%) in front
of the sample. The opening of slit 0 in front of the ioniza-
tion chamber was 0.25–0.5 mm (horizontal) · 0.5–1 mm
(vertical). The sample was placed in a plane near horizon-
tal. The plane was tilted by 6ꢁ toward the incident X-ray
The Au/TiO catalyst was also prepared by mixing 1 g of
2
TiO2 (P25) with 100 ml of aqueous solution of HAu-
Cl Æ 4H O (4.2 or 0.51 mM) by adding 42 or 5.1 mM of
4
2
urea, respectively. The mixture was stirred at 353 K for
Table 1
a
2
Preparation conditions via deposition-precipitation (DP) method and the property of various Au/TiO samples
Entry Alkaline TiO
2
pH
DP reaction
temperature (K)
DP reaction time Calcined
Au loading
(wt%)
Average Au particle
size (A)
b
˚
(h)
temperature (K)
a
b
c
d
e
f
NaOH
NaOH
NaOH
urea
urea
urea
mesoporous
mesoporous
P25
P25
P25
8
8
8
328
328
328
7
4.5
7
4
16
4
523
673
673
573
573
573
573
0.21
0.21
0.09
7.1
6.8
0.97
0.93
87 ± 23
78 ± 30
32 ± 21
37 ± 10
35 ± 19
29 ± 10
29 ± 18
3 ! 8 353
3 ! 8 353
4 ! 8 353
4 ! 8 353
P25
P25
g
urea
16
a
Gold precursor was HAuCl
Evaluated based on the Au La emission and Au L
4
Æ 4H
2
O.
b
1
3
absorption edge intensity.

Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
1151
beam and by 7ꢁ toward the Ge crystal [4,13,14]. A receiving
3. Results
slit 1 was placed at 100 mm from the sample. The opening
was 8 mm(h) · (0.5–4) mm(v). The sample, Ge(555) crys-
tal, and slit 2 in front of the NaI(Tl) scintillation counter
3.1. TEM observation of Au/TiO catalysts
2
(
2
SC) were controlled on Rowland circle (the radius
20 mm). The opening of slit 2 was 8 mm(h) · 2 mm(v).
High-resolution TEM images were observed for a series
of Au/TiO catalysts (Fig. 1). The average Au particle size
2
The sample and SC were covered with a lead plate housing
of 1–3 mm thickness except for the incident X-ray beam
entrance and the exit of fluorescence X-rays toward the
direction of Ge crystal.
and the standard deviation are listed in Table 1. The aver-
˚
age Au particle size was in the range 29–37 A supported on
anatase-type TiO (c–g) versus relatively greater particles
2
˚
size in the range 78–87 A supported on mesoporous
With the excitation energy set to 11943.1 eV, Au La1
emission (M ! L ; Bragg angle h = 77.700ꢁ) spectra
TiO (a,b). The Au particle size distribution was especially
narrow (standard deviation 10–19 A) in the cases of d–g
prepared using urea. Observed Au particle sizes (Table 1)
were basically consistent with values reported in Refs.
[10,11].
2
˚
5
3
B
were measured. The scan step was ꢁ0.9 eV and the accu-
mulation time was 60 s for a data point. The fluorescence
spectrometer was tuned to each emission energy and the
Au L -edge XANES spectra were measured. The scan steps
The degree of Au loading was reflected as the popula-
tion of spherical darker dots (Au) dispersed on amor-
phous TiO (a,b) or anatase-type TiO crystalline (c–g)
3
were ꢁ7.5, ꢁ0.5, and 2.4 eV in the pre-edge, edge, and
post-edge energy regions, respectively. The accumulation
2
2
time was 40–100 s for a data point. The Au La fluores-
(Fig. 1). The population of Au dots was smaller when
NaOH was used during the DP reaction (a–c). The pop-
ulation of Au dots was greater when urea was used (d–g)
and the population seems proportional to the initial con-
centration of HAuCl Æ 4H O:4.2 mM for Fig. 1d and e
1
cence and Au L -edge energy values were calibrated to
3
9
713.3 and 11921.2 eV, respectively, for the spectra of Au
metal [15]. The energy positions of the monochromator
and the fluorescence spectrometer were reproduced within
4
2
±
0.1 and ± 0.2 eV, respectively. Some of the XANES spec-
and 0.51 mM for Fig. 1f and g. The Au loading amount
tra for standard Au samples were also measured in trans-
mission mode at beamline NW10A of Photon Factory
Advanced Ring at KEK (Tsukuba, Japan). The obtained
data resolution was essentially the same as that measured
at beamline 37XU at SPring-8 in the transmission mode.
evaluated based on the Au La emission and Au L edge
1
3
jump intensity (Table 1) was in accord with the trend of
Au nano-particles population detectable in the TEM
images.
3.2. X-ray spectroscopy for standard Au compounds
2
.3. XAFS data analysis
The Au La emission spectra were measured for Au
1
The XAFS data analysis was performed using XDAP
+
I
ꢀ
2
metal and K [Au (CN) ] . The C values for Au L and
3
version 2.2.7 (XAFS Services International) based on
works of M. Vaarkamp, H. Linders, and D. Koningsber-
ger. The pre-edge background was approximated by the
M5 levels are 5.41 [17] and 2.18 eV [19], respectively.
Based on the FWHM (full width at the half maximum)
value of the Au La1 peak for Au metal (6.5 eV), the
energy resolution of fluorescence spectrometer was esti-
mated to 2.9 eV including the contribution of beamline
2
modified Victoreen function C /E + C /E + C . The back-
2
1
0
ground of post-edge oscillations was approximated by a
smoothing spline function calculated by the following
equation for the number of data points N.
at 9.7 keV. The chemical shift of Au La emission peaks
1
+
I
ꢀ
was ꢀ0.2 eV for K [Au (CN) ] compared to that for
2
N
2
X
ðlx ꢀ BG Þ
Au metal (Table 2). Because the shift value was compara-
ble to the reproducibility of energy position of the fluores-
cence spectrometer, the chemical shift was experimentally
checked three times and the shift was reproduced and
significant.
i
i
6
smoothing factor
2
i
expðꢀ0:075k Þ
i¼1
For Au cluster site models (Ti-top and O-top) supported
on anatase-type TiO (001) surface, Au L -edge XANES
2
3
spectra were theoretically generated using FEFF 8.2 [16]
in self-consistent field and full multiple scattering calcula-
tion modes. The Hedin–Lundqvist self-energy was used
for energy-dependent exchange-correlation potential. The
calculated potential was corrected by adding a constant
shift to ‘‘pure imaginary optical potential’’. Because the
By tuning the fluorescence spectrometer to each emis-
sion peak top, Au La
tra were measured for Au metal and K [Au (CN)
-selecting Au L -edge XANES spec-
3
1
+
I
ꢀ
]
2
(Fig. 2). Satisfying the condition that the energy resolution
to monitor the X-ray fluorescence was smaller than that C
for Au L level [4,18], absorption edge became steeper and
3
Au La -selecting XANES was measured under the condi-
tion that the energy resolution to detect the X-ray fluores-
the peaks at 11923.3, 11935.3, 11941.3, and 11948 eV
depicted in Fig. 2a and those at 11926.8 and 11948 eV
depicted in Fig. 2b were sharper and more resolved com-
pared to the corresponding spectra measured in transmis-
sion mode (dotted line).
1
cence (2.9 eV) was smaller than the C value of Au L level
3
(
5.41 eV) [17], the constant energy shift value was set to
negative (ꢀ1.0 eV) [4,18].

1152
Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
2
Fig. 1. High-resolution TEM images for Au/TiO catalysts. The images (a–g) correspond to catalyst samples listed in Table 1, entries a–g, respectively.
Table 2
The Au La
1 3 2
emission peak and Au L absorption edge positions for Au standard compounds and Au/TiO catalysts in air or 5% CO by tuning the
secondary fluorescence spectrometer to each energy
Sample
Purged gas
Au La
1
peak energy (eV)
Tune energy (eV)
3
Au L edge energy (eV)
Au metal
K [Au(CN)
AuCl[P(C H ) ]
6 5 3
Au/mesoporous–TiO
9713.3
9713.1
9713.3
9713.1
11921.2
11923.5
11923.0
11917.1
11920.8
11924.5
11921.1
11921.1
11916.5
11921.1
11927.9
11916.4
11921.1
11925.0
+
ꢀ
2
]
a
2
–NaOH-7-523
air
9713.3
9707.6
9
9
713.0
718.7
Au/mesoporous–TiO
Au/P25–NaOH-7-673
Au/P25–urea-16-573 (0.93 wt% Au)
2
–NaOH-4.5-673
air
air
air
9713.4
9713.0
9713.1
9713.2
9713.0
9707.8
9
9
713.1
718.3
CO (5%) + Ar (95%)
9713.4
9708.1
9
9
713.4
718.6
a
Measured in transmission mode.
3
.3. X-ray spectroscopy for Au/TiO catalysts
2
as for Au metal (Table 2). The peak shifted to 9713.0–
713.1 eV for Au/P25–NaOH-7-673 and Au/P25–urea-16-
573 both in air. The energy corresponds to the value for
9
The Au La emission spectra for Au/TiO catalysts are
1
2
+
I
ꢀ
depicted in Fig. 3. The peak top energy (9713.3–
713.4 eV) for Au/mesoporous–TiO –NaOH-7-523 and
K [Au (CN) ] (Table 2).
2
9
Next, the Au La emission spectra were measured for
2
1
Au/mesoporous–TiO –NaOH-4.5-673 in air and Au/P25–
Au/mesoporous–TiO –NaOH-7-523 in air by progressively
2
2
urea-16-573 (0.93 wt%) in 5% CO was essentially the same
changing the excitation energy (Eexc) on going from

Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
1153
Fig. 2. Au L
spectrometer to Au La
3
-edge XANES spectra measured by tuning the fluorescence
peak top (solid line, 1) and in transmission mode
1
+
I
ꢀ
(dotted line, 2) for Au metal (a) and K [Au (CN) ] (b).
2
1
1943.1 eV to 11919.1 eV [3]. In the Eexc range between
1
1943.1 and 11922.0 eV, the emission peak energy and
intensity negligibly changed. When the Eexc value decreased
to 11919.1 eV, the Au La1 emission peak shifted to
9
711.0 eV and the intensity decreased to 66% [3].
The fluorescence spectrometer was tuned to each peak
top (9713.0–9713.4 eV, the arrows in Fig. 3) and Au La1-
Fig. 3. Au La
1 2
emission spectra measured for Au/mesoporous–TiO –
selecting Au L -edge XANES spectra were measured for
3
NaOH-7-523 in air (a) and Au/P25–urea-16-573 (0.93 wt% Au) in air (b)
and in 5% CO (c). The arrows are tune energy in the Au L
measurements in Fig. 4.
Au/TiO catalysts listed in Table 1a–c and g. The absorp-
2
3
-edge XAFS
tion edge was at 11920.8–11921.1 eV (Table 2) similar to
the value for Au metal (11921.2 eV). The sharp, resolved
XANES pattern for these Au/TiO catalysts (Fig. 4a-(2),
2
b-(2), and c-(2)) was similar to Au La -selecting XANES
b-(1), and c-(1)). The post-edge pattern looked to shift by
1
for Au metal (Fig. 2a-(1)).
ꢀ3.7 to ꢀ4.7 eV compared to the spectra tuned to Au
Then, tune energy was set lower than Au La peak top,
La peak top.
1
1
corresponding to peak intensity of 1/3 (9707.6–9708.1 eV,
Subsequently, the fluorescence spectrometer was tuned
to the energy corresponding to peak intensity of 1/3, on
higher-energy side (9718.3–9718.7 eV, the arrows in
the arrows in Fig. 3) for Au/mesoporous–TiO –NaOH-7-
2
5
23 in air and Au/P25-urea-16-573 (0.93 wt% Au) in air
and in 5% CO. The Au L edge shift compared to the data
Fig. 3). The Au L edge shift compared to the data tuned
3
3
tuned to Au La peak top was ꢀ3.7 eV for Au/mesopor-
to Au La peak top was +6.8 eV for Au/P25–urea-16-
1
1
ous–TiO NaOH-7-523, smaller than shifts for Au/P25–
573 in air, greater than shifts for Au/mesoporous–TiO2–
NaOH-7-523 in air and Au/P25–urea-16-573 in 5% CO
(+3.7 to +3.9 eV) (Table 2). The intensity of first intense
peak at 11930.0–11931.2 eV above absorption edge was
in the order for spectra of Au/P25–urea-16-573 in air
2
urea-16-573 in air and in 5% CO (ꢀ4.6 to ꢀ4.7 eV) (Table
2
). The first intense peak at 11923.9–11924.2 eV above
absorption edge observed in Fig. 4a-(2), b-(2), and c-(2)
became a weak shoulder feature at 11922 eV (Fig. 4a-(1),

1154
Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
(Fig. 4b-(3)) > Au/P25–urea-16-573 in 5% CO (Fig. 4c-(3))
>
Au/mesoporous–TiO –NaOH-7-523 in air (Fig. 4a-(3)).
2
This resonance peak was reported to be more intense as
the Au 5d states were less occupied [20]. The post-edge pat-
tern looked to shift by +3.7 to +6.8 eV compared to the
spectra tuned to Au La peak top.
1
Au La -detecting EXAFS (extended X-ray absorption
1
fine structure) spectra for Au/mesoporous–TiO –NaOH-
2
7-523 (0.93 wt%) tuned to 9707.6 and 9718.7 eV (Fig. 3a,
arrows) were very similar to each other in the wave-number
ꢀ
1
˚
region 2–9 A in contrast to the typical difference of state-
sensitive spectra in the XANES region (Fig. 4a). A peak
˚
due to Au–Au bonding appeared at 2.3 A (phase shift
uncorrected) in the Fourier transform of EXAFS for Au/
mesoporous–TiO –NaOH-7-523 tuned to 9707.6 and
2
9718.7 eV and Au/P25–urea-16-573 (0.93 wt% Au) in air
tuned to 9713.1 eV (Fig. 3b, arrow), however, a peak due
to Au–O bonding [21,22] was unclear/unresolved in the
Fourier transform for all the EXAFS spectra.
4. Discussion
State-sensitive Au L -edge XANES spectra tuned to Au
3
La1 peak (9713.0–9713.4 eV) for Au/TiO2 catalysts
Fig. 4a-(2), b-(2), and c-(2)) resembled Au La -selecting
(
1
0
XANES for Au metal (Fig. 2a-(1)). The Au state among
Au/TiO catalysts was discriminated with this range of
2
tune energy.
When the fluorescence spectrometer was tuned to
9707.6–9708.1 eV, the first intense peak above the edge
became a weak shoulder at 11922 eV (Fig. 4a-(1), b-(1),
and c-(1)) suggesting greater Au 5d occupation [20,21].
Two types of Au cluster models were considered located
either on Ti or on O atom of anatase-type TiO (001) sur-
2
face (Fig. 5a and b, respectively). The spectral pattern,
especially in the absorption edge region, was similar to
the theoretical XANES generated by FEFF for O-top
interfacial Au sites (Figs. 4d-(6) and 5b). The Au L -edge
3
XANES spectra tuned to 9710.4–9710.6 eV for Au/meso-
Fig. 4. Au La
TiO –NaOH-7-523 in air (a) and Au/P25–urea-16-573 (0.93 wt% Au) in
air (b) and in 5% CO (c). Tune energy was 9707.6, 9713.0, and 9718.7 eV
1
-selecting XANES spectra measured for Au/mesoporous–
2
(
(
(
a), 9707.8, 9713.1, and 9718.3 eV (b), and 9708.1, 9713.4, and 9718.6 eV
I
3 6 5 3
c). (d) Au L -edge XANES spectra for gold metal (1), Au Cl[P(C H ) ]
III
III
2), Au (OH)
3
3
(3), Au2 O (4), and interfacial Au sites located on Ti and
O
atoms (5 and 6, respectively). 1–4: Experimental, measured in
Fig. 5. Au cluster models on anatase-type TiO (001) surface terminated
2
transmission mode. 5, 6: generated for model sites in Fig. 5a (Ti-top)
and b (O-top), respectively, using FEFF [16].
with Ti atoms (Ti-top; a) and O atoms (O-top; b).

Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
1155
porous–TiO –NaOH-7-523 in air [3] and Au/P25–urea-16-
The pattern of positively-shifted XANES spectra
(Fig. 4b-(3)) due to the effects of negative charge transfer
2
5
73 (0.93 wt% Au) in 5% CO (not shown) resembled the
theoretical data of Fig. 4d-(6) even more. Thus, negative
charge transfer from interfacial O atoms to Au is suggested
from Au to support and/or O was similar to the spectrum
2
for Au metal (Fig. 2A1) shifted by +6.7 eV rather than
+
I
ꢀ
I
I
[
23–25] at the interface of Au/TiO catalysts, successfully
ones for K [Au (CN) ] Au (Fig. 2b-(1)), Au Cl[P-
2
2
III
III
discriminated by tuning the spectrometer to 9707.6–
(C H ) ], Au (OH) , or Au O (Fig. 4d-(2)–(4)). Thus,
6 5 3 3 2 3
9
710.6 eV.
the Au states discriminated by tuning the spectrometer to
9718.3 eV for Au/P25–urea-16-573 in air were for elec-
tron-deficient Au nano-particle sites.
Among the Au La -selecting XANES spectra tuned to
1
9
707.6–9708.1 eV, the Au L absorption edge shift com-
3
pared to XANES spectra tuned to Au La peak top was
1
smaller for Au/mesoporous–TiO –NaOH-7-523 in air
Acknowledgements
2
(
ꢀ3.7 eV) versus ꢀ4.6 to ꢀ4.7 eV for Au/P25-urea-16-573
in air and in 5% CO (Table 2). This difference can be
explained due to the difference of average Au nano-particle
size: 87 and 29 A, respectively. As the Au nano-particle size
This work was supported from the Grant-in-Aid for Sci-
entific Research (Y.I., C17550073) and the Research Foun-
dation for Opto-Science and Technology (Y.I. 2005–2006).
The X-ray measurements were performed under the
approvals of the SPring-8 Program Review Committee
(2006A1156 and 2006B1486) and Photon Factory Proposal
Review Committee (2005G220).
˚
increases, the effect of charge transfer at the particle-sup-
port interface becomes relatively small per Au particle
and thus the population of Au state was greater. Instead,
this difference may be explained based on the report that
smaller Au particles were relatively more negatively
charged [21].
0
References
First intense peak above the edge was observed at
1923.9–11924.2 eV by tuning the fluorescence spectrome-
ter to peak top and at 11930.0–11931.2 eV by tuning it to
718.3–9718.7 eV (Fig. 4). An intense peak appeared at
1925.7 eV for theoretically generated XANES for Ti-top
[
[
[
1] A. Cho, Science 299 (2003) 1684.
2] A.T. Bell, Science 299 (2003) 1688.
1
3] Y. Izumi, D. Masih, J.P. Candy, H. Yoshitake, Y. Terada, H. Tanida,
T. Uruga, X-ray absorption fine structure 13th international confer-
ence, in: B. Hedman, P. Pianetta, (Eds.), AIP Conference Proceedings
882 (2007) 588–590.
4] Y. Izumi, H. Nagamori, T. Kiyotaki, D. Masih, T. Minato, E. Roisin,
J.P. Candy, H. Tanida, T. Uruga, Anal. Chem. 77 (21) (2005) 6969.
5] Y. Izumi, H. Nagamori, Bull. Chem. Soc. Jpn. 73 (2000) 1581.
6] K. H a¨ m a¨ l a¨ inen, D.P. Siddons, J.B. Hastings, L.E. Berman, Phys.
Rev. Lett. 67 (1991) 2850.
9
1
interfacial Au sites (Figs. 4d-(5) and 5a) [24]. Thus, nega-
tive charge transfer was suggested from Au 5d to support
[
[
9
22,26] for Au states discriminated by tuning to 9718.3–
718.7 eV. The charge transferred from Au 5d to reduce
[
[
3
+
to Ti states at the interface (Fig. 5a).
Among the Au La -selecting XANES spectra tuned to
718.3–9718.7 eV, the Au L absorption edge shift com-
3
[7] F.W.F. de Groot, Top. Catal. 10 (2000) 179.
1
[
[
8] Y. Izumi, T. Kiyotaki, T. Minato, Y. Seida, Anal. Chem. 74 (15)
2002) 3819.
9] H. Yoshitake, T. Sugihara, T. Tatsumi, Chem. Mater. 14 (3) (2002)
023.
9
(
pared to XANES spectra tuned to Au La peak top was
1
greater for Au/P25–urea-16-573 in air (+6.8 eV) versus
1
+
3.7 to +3.9 eV for the same catalyst in 5% CO and Au/
[10] G.R. Bamwenda, S. Tsubota, T. Nakamura, M. Haruta, Catal. Lett.
mesoporous–TiO –NaOH-7-523 in air. This difference
demonstrates that the negative charge transfer from Au
44 (1997) 83.
2
[
[
11] R. Zanella, L. Delannoy, C. Louis, Appl. Catal. A 291 (2005) 62.
12] A. Krolzig, G. Materlik, M. Swars, J. Zegenhagen, Nucl. Instr. Meth.
to O is only detected for smaller Au particles (mean size
2
2
19 (2) (1984) 430.
˚
2
5
9 A, Fig. 5a), not for Au/mesoporous–TiO –NaOH-7-
2
[
13] Y. Izumi, F. Kiyotaki, H. Nagamori, T. Minato, J. Electro. Spectr.
I
˚
23 (mean size 87 A). The Au species were detected by
Relat. Phenom. 119 (2/3) (2001) 193.
M o¨ ssbauer and X-ray photoelectron spectroscopy [26]
and suggested to be active for CO oxidation. Negative
[14] Y. Izumi, H. Oyanagi, H. Nagamori, Bull. Chem. Soc. Jpn. 73 (9)
2000) 2017.
(
[
[
15] J.A. Bearden, Rev. Mod. Phys. 39 (1967) 78.
16] A.L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Phys. Rev. B
charge transfer from Au to O was supported for Au nano-
2
particle model on rutile-type TiO (110) by ultraviolet pho-
2
5
8 (1998) 7565.
toelectron spectroscopy [27] and density functional theory
[17] M.O. Krause, J.H. Oliver, J. Phys. Chem. Ref. Data 10 (1979) 329.
(
DFT) calculations [28]. The effect of O was suggested
[18] Y. Izumi, D. Masih, E. Roisin, J.P. Candy, H. Tanida, T. Uruga,
Mater. Lett. 61 (2007) 3833.
2
based on the difference XANES spectrum measured in
transmission mode before/after the introduction of O to
Au/TiO [20]. The effect of O was directly demonstrated
by state-sensitive XANES in this work (Fig. 4b-(3)).
[19] G. Zschornack, Handbook of X-ray data, Springer, Berlin-Heidel-
berg, 2007.
2
2
2
[
20] N. Weiher, A.M. Beesley, N. Tsapatsaris, L. Delannoy, C. Louis, J.A.
van Bokhoven, S.L.M. Schroeder, J. Am. Chem. Soc. 129 (8) (2007)
2240.
d+
The Au site to interact with O may be at the interface
2
[
21] J.T. Miller, A.J. Kropf, Y. Zha, J.R. Regalbuto, L. Delannoy, C.
Louis, E. Bus, J.A. van Bokhoven, J. Catal. 240 (2006) 222.
22] E.D. Park, J.S. Lee, J. Catal. 186 (1999) 1.
based on the similarity of spectral pattern between Fig. 4b-
(
3) (state-sensitive XANES) and D-5 (theoretical XANES
[
[
d+
generated for interface Au site model, Fig. 5a), as also
proposed based on on-reaction EXAFS measurements
23] D.B. Akokekar, G. Foran, S.K. Bhargava, J. Synchrotron Radiat. 11
(
2004) 284.
[
29] and DFT calculations [30].
[24] M.S. Chen, D.W. Goodman, Science 306 (2004) 252.

1156
Y. Izumi et al. / Inorganica Chimica Acta 361 (2008) 1149–1156
[
[
25] D.W. Goodman, Catal. Lett. 99 (1–2) (2005) 1.
26] D. Boyd, S. Golunski, G.R. Hearne, T. Magadzu, K. Mallick, M.C.
Raphulu, A. Venugopal, M.S. Scurrell, Appl. Catal. A 292 (2005)
[28] L.M. Molina, M.D. Rasmussen, B. Hammer, J. Chem. Phys. 120 (16)
(2004) 7673.
[29] S.H. Overbury, V. Schwartz, D.R. Mullins, W. Yan, S. Dai, J. Catal.
241 (2006) 56.
76.
[
27] T. Okazawa, M. Kohyama, Y. Kido, Surf. Sci. 600 (2006) 4430.
[30] L.M. Molina, B. Hammer, Appl. Catal. A 291 (2005) 21.