Advanced Materials Laboratory, Nissan Research Center, NISSAN MOTOR CO., LTD., 1 Natsushima-cho, Yokosuka-shi, Kanagawa 237-8523, JapanDepartment of Applied Physics and Chemistry, The University of Electro-Communications, 1-5-1 Chofugaoka, Chofu, Tokyo 182-8585, JapanDepartment of Engineering Science, The University of Electro-Communications, 1-5-1 Chofugaoka, Chofu, Tokyo 182-8585, JapanPRESTO, Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan
r t i c l e i n f o
rticle history:eceived 2 June 2012eceived in revised form 30 August 2012ccepted 9 September 2012vailable online 2 October 2012
eywords:
a b s t r a c t
In order to decrease the consumption of precious metals used in the catalytic converters used in auto-mobiles, we studied the relationship between the catalytic activity of Pt/alumina (Pt/Al2O3) and therelaxation process of photoexcited electrons. Firstly, we studied the relationship between the size of thePt particles in Pt/Al2O3 and catalytic performance. Secondly, the relationship between the size of the Ptparticles in Pt/Al2O3 and the decay time of the excited electrons was studied using an improved transientgrating (TG) technique. The results showed that faster decay of the excited electrons leads to greater
latinumluminum oxidehotoexcited electronransient grating techniquearrier dynamicslectron transfer
oxidation rates. The decay time obtained with the improved TG technique gives an indication of the timethat the exited electrons take to return to the ground state. According to studies utilizing FT-IR, one of theprocesses necessary for quickly generating CO2 with Pt is that the electron in the Pt O bond moves to thePt side and that the Pt+ becomes Pt metal. Thus, the decay time obtained with the improved TG techniquecorresponds to the process whereby Pt+ returns to Pt metal. Thus, we found that the consumption ofprecious metals can be reduced by increasing the speed of the decay of the excited electrons.
. Introduction
The exhaust emissions from automobile engines contain a mix-ure of hydrocarbons (HC), carbon monoxide (CO), and nitrogenxides (NOx). A catalyst is installed within the vehicle exhaustystem, and this converts the emissions into harmless materialswater, nitrogen, and carbon dioxide) via various chemical reac-ions. The catalytic converter contains a support material coatedith precious metallic particles such as Pt, Pd, and Rh. Because therice of these metals is high, research is continually being done toecrease their consumption in catalytic converters by as much asossible.
It is said that the catalytic activity is dependent on the size ofhe precious metal particles. In the case of Pt/Al2O3, there was
report which stated that the reaction rate for CO oxidations fast, and that the activation energy decreases with increasing
article size [1]. Generally, the coordination number of the sur-ace atoms depends on the particle size. The ratio of surface atomsrranged on a planar surface to interior atoms increases with
∗ Corresponding author at: The University of Electro-Communications, Depart-ent of Applied Physics and Chemistry, 1-5-1 Chofugaoka, Chofu, Tokyo, 182-8585
increasing particle size, whereas this ratio for atoms arrangedaround corners and edges decreases [2]. However, the ratio ofatoms arranged on planar Pt surfaces is different, and the reactionrates for CO oxidation are known to be different [3–6]. In addition tothe coordination number of the surface atoms being different whenthe size of the particles is different, the ratio of atoms on the surfaceto the number of internal atoms also changes. Fig. 1 shows a con-ceptual picture of the different particle sizes. Generally, the ratioof the surface area to total volume depends on the particle size.The surface to volume ratio is lower in Fig. 1(a) than in Fig. 1(b).In this state the dispersion is low. According to studies using FT-IR,when oxygen is introduced onto CO adsorbed on a Pt/Al2O3 surface,the generation of CO2 is observed. It is known that the genera-tion time for CO2 is fast when the dispersion of Pt in Pt/Al2O3 islow [7].
On the other hand, because the electrons change state when thestructure changes, it is believed that the electrons contribute to thecatalytic activity. We surmised that the reason that the catalyticactivity is different when the size of the particles is different, asdescribed in past reports, is because the lifetime of the excited elec-trons is different. Given this background, the purpose of this study
is to study the relationship between the lifetime of the excited elec-trons and the catalytic activity, and our aim is to obtain informationthat will make it possible to decrease the consumption of preciousmetals.
ig. 1. Conceptual representation of different size particles: (a) large size; (b) smallize.
In this paper, we first executed a propane oxidation reactionnd a CO oxidation reaction to test the catalytic activity to confirmhether the catalysis is different for particles of different size. Sec-
ndly, we report on the ultrafast carrier dynamics of Pt/Al2O3 filmsharacterized using an improved TG technique [8–10].
. Experimental
.1. Relationship between the size of the Pt particles in Pt/Al2O3nd catalytic performance
.1.1. Preparation of the catalystsAn r-Al2O3 support (Sasol North America; PURALOX®; BET sur-
ace area, 200 m2/g) was used for the support material. Pt/Al2O3atalysts were prepared by impregnating the Al2O3 with four kindsf colloidal Pt solution (Nippon Sheet Glass Co. Ltd.) for 1 h. The dif-erence between these four colloidal solutions is the size of the Ptarticles in them. The particle sizes were less than 2 nm, 2–3 nm,–5 nm, and 5–8 nm, respectively. The samples were dried at 80 ◦Cor 12 h and then calcined in air at 723 K for 30 min. The Pt loadingas 0.03 wt%.
.1.2. CO oxidation and C H oxidation reactions
3 8Fig. 2 shows the set-up for the catalytic reaction. Oxidation of
O gas was carried out in a fixed-bed flow reaction system at atmo-pheric pressure using 200 mg of the catalyst at a total flow rate
He gas
CO/He gas
O2/He gas
C3H8/He gas
Gas analyzer
(Q-Mass)
Quartz reactor tube
Furnace
Catalyst
Fig. 2. Set-up for catalytic reaction.
nce 263 (2012) 230–235 231
of 50 cm3/min (STP). The feed stream contained 0.4% CO, 0.2% O2,and it was balanced with helium. The temperature was controlledusing an external heater. The catalyst was inserted uniformly acrossthe silica tube. The effluent gas was analyzed using an online Q-mass spectrometer. The activity was observed for 30 min with eachsample.
Similarly, oxidation of C3H8 gas was carried out in the fixed-bed flow reaction system at atmospheric pressure using 200 mgof Pt/Al2O3 catalyst at a total flow rate of 50 cm3/min (STP). Thefeed stream contained 0.4% C3H8, 2.0% O2, and was balanced withhelium. The temperature was again controlled by the externalheater, and the catalyst was inserted uniformly across the silicatube. The effluent gas was analyzed using the Q-mass spectrometer,and the activity with each sample was observed for 30 min.
2.1.3. Evaluation of catalytic performance by conversionOne method for evaluating catalytic performance is to consider
the total conversion, paying particular attention to the decrease inthe amount of gas. For example, the CO conversion was calculatedbased on reaction (1).
CO + 1/2O2 = CO2 (1)
The CO conversion was calculated from the intensity of the CO2measured using the Q-mass spectrometer. The intensity of CO2when all of the CO gas has reacted is defined as representing 100%CO conversion. Likewise, the intensity of CO2 when none of the COgas has reacted is considered to represent 0% CO conversion. Usingthis relationship, the degree of CO conversion was calculated fromthe intensity of CO2 in the gas that had passed over the catalyst.
Determination of the size of the Pt particles in the Pt/Al2O3was carried out using transmission electron microscopy (TEM). Thediameters of 100 or more Pt particles were measured, and the aver-age size of the particles was determined.
2.1.4. Evaluation of catalytic performance by the reaction rateper unit surface area
Another method for evaluating catalytic performance is one thatlooks at the decrease in the amount of gas at the active site byconsidering the reaction rate with respect to surface area. Becausethe number of surface atoms changes depending on the size of the Ptparticles, we used the reaction rate per unit surface area to evaluatethe catalytic performance.
Firstly, the reaction rate was determined using the conversionmethod described in Section 2.1.3 for CO concentration. This wasdetermined for CO to be as follows;
R = COconv × COconc
100 × F/V(2)
where R is the reaction rate, COconv is the conversion described inSection 2.1.3, COconc is the concentration of CO, F is the flow rate ofthe gas, and V is the volume corresponding to 1 mol of CO.
Secondly, the surface area was calculated from the particlediameter.
S = 4�(
r
2
)2(3)
where r is the mean particle size observed from TEM and S is thesurface area of the Pt. Finally, the reaction rate per surface area wascalculated from R/S.
2.2. Relationship between the size of the Pt particles in Pt/Al2O3and the decay time of the excited electrons
2.2.1. Porous Pt/Al2O3 filmsTo measure the decay time of the excited electrons, it was nec-
essary to coat the Pt/Al2O3 powder used in the catalytic reaction
2 ce Science 263 (2012) 230–235
opufiiaaswa
2
sc(aamrfrlp
pFambsttmt
iTic(mltemw
ihbiTskitsptig
(ib
32 J. Ito et al. / Applied Surfa
nto glass to create a thin film. In order to do this, the powder wasrepared by crushing with beads in a mill (Hosokawa Micron Ltd.)ntil it consisted of particles of less than 300 nm. Porous Pt/Al2O3lms were prepared with these powders using a method described
n a previous paper [12]. A polyethylene glycol (PEG) binder wasdded to create a Pt/Al2O3/water paste. The resulting pastes werepplied onto glass slides using Scotch tape as both a frame and apacer. Excess paste was removed using a glass rod. The Al2O3 filmsere dried in air at room temperature for 10 min, and then heated
t 723 K for 30 min.
.2.2. Improved TG techniqueThe transient grating (TG) method is a well-established laser
pectroscopy technique involving four wave mixing [11,13]. In theonventional TG method, two time-coincident short laser pulsespump beams) with the same wavelength and intensity intersectt an angle in a sample to generate an optical interference patternt the intersection. The interaction between the light field and theaterial results in a spatially-periodic modulation of the complex
efractive index, which works like a transient diffraction gratingor a third laser pulse (probe beam) incident on the photoexcitedegion. Then, by measuring the time dependence of the diffractedight from the probe beam, the dynamics of the transient gratingroduced in the sample can be monitored.
Although this technique provides valuable information, itresents some technical difficulties for general researchers [13].irst, the three beams must overlap at a small spot, typically withinspot diameter of less than 100 �m, on the sample, and each beamust be temporally controlled. This is very difficult for pulsed laser
eams, especially for those with pulse widths of ∼100 fs. Secondly,ince the diffracted beam, i.e., the signal, is quite weak, it is difficulto locate it during the measurements. Thirdly, for a solid sample,he surface must be optically smooth. It is almost impossible to
easure a sample with a rough surface using the conventional TGechnique.
In 2003, Katayama and co-workers [8–10] proposed anmproved TG technique (it was also called a lens-free heterodyneG (LF-HD-TG) or a near field heterodyne TG (NF-HD-TG) techniquen some papers), which overcomes the difficulties that exist in theonventional TG technique. The improved TG technique features1) simple and compact optical equipment and easy optical align-
ent, and (2) high stability of phase due to the short optical pathengths of the probe and reference beams. Because it can be appliedo transmission and reflection-type measurements, this method isxtremely versatile. Thus, it is possible that it can be applied toany different kinds of sample, for example, opaque solids, solidsith rough surfaces, transparent solids and liquids.
The principle of the improved TG technique has been explainedn detail in previous papers, and is therefore only described brieflyere [8–10]. Unlike the conventional TG technique, only one pumpeam and one probe beam without focusing are needed in the
mproved TG technique. Fig. 3 shows the set-up for the improvedG technique. The pump beam is directed toward the transmis-ion grating. Then, the spatial intensity profile of the pump beam isnown to have an interference pattern with a grating spacing sim-lar to that of the transmission grating close to the other side of theransmission grating. When a sample is brought near the transmis-ion grating surface, it can be excited by the optical interferenceattern. The refractive index of the sample changes according tohe intensity profile of the pump light and the induced refractivendex profile functions as a different type of transiently-generatedrating.
The probe beam is diffracted both by the transmission gratingcalled the reference light) and the transiently generated grat-ng (called the signal light). In principle, the two diffracted lighteams progress in the same direction and, therefore, interfere.
Fig. 3. Set-up for the improved TG technique.
Measurements are made with a detector positioned at a visiblediffraction spot of the reference beam.
In the improved TG technique used for studying ultrafast carrierdynamics, the laser source was a titanium/sapphire laser (CPA-2010, Clark-MXR Inc.) with a wavelength of 775 nm, a repetitionrate of 1 kHz, and a pulse width of 150 fs. The light was separatedinto two parts. One part was used as a probe pulse and the otherwas used to pump an optical parametric amplifier (OPA) (a TOAPSfrom Quantronix) to generate light pulses with a wavelength tun-able from 290 nm to 3 �m, which was used as pump light for theTG measurement. In this experiment, the probe pulse wavelengthwas 775 nm and the pump pulse wavelength was 470 nm (2.6 eV).
3. Results and discussion
3.1. Relationship between the Pt particle size in Pt/Al2O3 andcatalytic performance
The dependence of the conversion on temperature needs to bemeasured in order to decide the temperature at which the reactionspeed for the various catalysts should be compared. Fig. 4 showsthe dependences of the conversion of C3H8 and CO gases on tem-perature using catalysts of Pt/Al2O3 with Pt particles of differentsize. Generally, the reaction rate is compared under low conver-sion conditions, because the amount of reactive gas supplied canbe considered to be almost constant when the conversion ratio islow. The temperatures selected to compare the reaction rates were548 K for C3H8 oxidation and 498 K for CO oxidation.
The average size of the Pt particles in Pt/Al2O3 was calculatedfrom TEM measurements. For the four kinds of Pt/Al2O3 powderthat we used, the average sizes of the Pt particles were 2.3 nm,6.8 nm, 7.3 nm, and 12.0 nm.
As for the amount of Pt contained in the four kinds of powder,the relative number of Pt atoms at the surface decreases as the sizeof the particles increases, and the relative number of Pt atoms inthe bulk increases. Fig. 1 shows the large and small particles in twodimensions. For the same number of atoms, the ratio of internalatoms is higher than the ratio of the surface atoms for the larger
particle. Fig. 1(a) shows the state in which the dispersion is low.Reactions that are catalyzed on the surface are usually comparedin terms of reaction rate per unit surface area. To compare the cat-alytic activity per unit surface area, Fig. 5 shows the dependences
J. Ito et al. / Applied Surface Science 263 (2012) 230–235 233
0
10
20
30
40
50
60
70
80
90
100
500 550 600 650
Pt ave.: 7.3nm
Pt ave.: 6.8nm
Pt ave.: 12.0nm
Pt ave.: 2.3nm
C3H
8co
nver
sion
(%
)
temp. (K)
(a)
450
CO
10
20
30
40
50
440 460 480 500 520
CO
con
ver
sion
(%
)
temp. (K)
Pt ave.: 2.3nm
Pt ave.: 12.0nm
Pt ave.: 7.3nm
Pt ave.: 6.8nm
0
(b)
Fig. 4. Dependences of C3H8 conversion and CO gas conversion on temperature using Pt/Al2O3 as a catalyst with Pt particles of different size. Reaction type: (a) C3H8 gas andO2 gas; (b) CO gas and O2 gas.
0
20
40
60
80
100
0 5 10 15 20
C3H
8re
act
ion
sp
eed
at
548 K
(mm
ol
sec-1
m
2-1
-Pt)
(a)
0
20
40
60
80
100
0 5 10 15
K4
98
at
spee
dre
act
ion
CO
(b)
(mm
ol
sec-1
m
2-1
-Pt)
F article
otofiCdpbtts
3t
fpaTPtlisi
�n(t) =2n0m3ω2
pε0, (4)
where Ne(t) is the photoexcited electron density, me, the effec-tive mass of the electron, ωp, the radial probe frequency, e, the
0.0
0.2
0.4
0.6
0.8
1.0
Sig
na
l in
ten
sity
(a
rb
. u
nit
s)
Pt size: 2.3nm
Pt size: 6.8nm
Pt size: 7.3nm
1.2
Ave. Pt size (nm)
ig. 5. Dependences of C3H8 and CO reaction rates per unit surface area on the Pt p
f the reaction rates per unit surface area on the size of the Pt par-icles in Pt/Al2O3 for C3H8 and CO. In each case, for the oxidationf propane and the oxidation of CO, the larger the particle size, theaster the reaction rate per unit surface area. The phenomena shownn Fig. 5 are similar to that reported as being due to a decrease in theO oxidation reaction activation energy with increasing Pt particleiameter [1]. Therefore, we consider that this result has reproducedast research. In the case of C3H8 oxidation using the catalyst, it haseen said that the oxidation of CO is the last process in the genera-ion of CO2 [14]. Therefore, we believe that it is correct to say thathe dependence of the C3H8 reaction rate per unit surface area isimilar to that for CO.
.2. Relationship between the size of Pt particles in Pt/Al2O3 andhe decay time of the excited electrons
Fig. 6 shows the signal produced by the improved TG techniqueor porous Pt/Al2O3 films. The signals are normalized at the peakositions. As shown in Fig. 6, the TG signal pulse is very short (<3 ps)nd then a small baseline signal remains over a longer time scale.he signals shown in Fig. 6 arise from optical absorption in the Pt int/Al2O3 since no TG signal was observed for an Al2O3 film underhe same conditions. The improved TG signal on a time scale of
ess than 100 ps corresponds to the photoexcited carrier dynam-cs. This has resulted from the change in refractive index of theample surface induced by the generation of photoexcited carriers,.e., a population grating. We assume that the change in extinction
Ave. Pt size (nm)
size in Pt/Al2O3. Reaction type: (a) C3H8 gas and O2 gas; (b) CO gas and O2 gas.
coefficient is negligible at the probe wavelength. The change inrefractive index upon photoexcitation of the electrons can beapproximated by the Drude model for the optical properties ofquasi-free carriers in metal, and is given by [15]
−Ne(t)e2
0 2 4 6 8 10
Delay time (ps)
Fig. 6. Signals obtained using the improved TG technique for porous Pt/Al2O3 films.
234 J. Ito et al. / Applied Surface Science 263 (2012) 230–235
0.6
0.8
0
0.2
0.4
1.0
1.2
1.4
0 2 4 6 8 10 12
Dec
ay t
ime
(ps)
Ave.Pt size (nm)
Ft
fnit
Pt
O CO
Pt
O COOCO
Pt
(a) adsorption (b) activation (c) desorption
Al2O3 Al2O3Al2O3
Fg
ig. 7. Dependence of the decay time obtained by the improved TG technique onhe Pt particle size in Pt/Al2O3.
undamental electron charge, ε0, the permittivity of free-space, and0, the refractive index. According to METALS-Drude’s theory, the
mportant feature of the population grating signal is that the pho-oexcited electron densities contribute to the signal. Therefore, in
Fig. 8. Schematic showing the excitation and decay processes of an electro
Decay time (ps)
0
20
40
60
80
100
0 1.00.2 0.4 0.6 0.8
(a)
C3H
8re
act
ion
sp
eed
at
548 K
(mm
ol
sec-1
m
2-1
-Pt)
ig. 9. Dependences of the C3H8 and CO reaction rates per unit surface area on the decayas; (b) CO gas and O2 gas.
Fig. 10. Reaction between CO and O2 on Pt according to the Langmuir–Hinshelwoodmechanism.
the case of Pt in Pt/Al2O3, we believe that the carrier is an electronat room temperature. Thus, we consider that the fast decay pro-cess for Pt/Al2O3 films mainly reflects a decrease in the numbers ofphotoexcited electrons due to relaxation to the ground state.
Fig. 7 shows the dependence of the decay time obtained by theimproved TG technique on the size of the Pt particles in Pt/Al2O3.The decay time is defined as the time for the signal to fall to halfthe intensity at the peak position. The larger the size of the Pt par-ticle, the faster the decay process. No TG signal was observed inPt/Al2O3 when the particle size was 12 nm. It was believed that the
signal could not be detected because the decay time was 0.1 ps orless, which is less than the temporal resolution (0.15 ps) of our TGset-up. It is known that the energy level of an electron becomesdiscrete when the size of the particle becomes small (quantum
n by pulsed excitation light used in the improved TG measurement.
Decay time (ps)
0
20
40
60
80
100
0 0.2 0.4 0.6 0.8 1.0
(b)
K498
at
spee
dre
act
ion
CO (m
mol
sec-1
m
2-1
-Pt)
time obtained with the improved TG technique. Reaction type: (a) C3H8 gas and O2
J. Ito et al. / Applied Surface Scie
Pt
Oe-
Pt
O
e-
Pt
COO
(a) adsorption (b) activation (c) desorption
Al2O3 Al2O3Al2O3
FL
sitito
3u
pTa2
pTtf
iIdrion
taitowoePitetomLts
i
[
[
[[
[[
of Chemical Physics 90 (1989) 1253.
ig. 11. Movement of electron about a Pt O unit according to theangmuir–Hinshelwood mechanism.
ize effects). When the energy levels of the electrons occur at setntervals, the electron can take only definite energy states, such ashose of the fundamental and doubled vibrations [16]. Therefore, its believed that, in this case, some time is required for a steady stateo be attained compared with larger sizes which have continuouslyrbiting particles.
.3. Relationship between the decay time and reaction rate pernit surface area of Pt
Fig. 8 shows a schematic explanation of the excitation and decayrocesses of an electron by the pulsed light used in the improvedG measurement. According to simulation of the electron densitynd energy level diagram for Pt, a pump beam at 470 nm (about.6 eV) corresponds to excitation in the conduction band [17].
Fig. 9 shows the dependence of the C3H8 and CO reaction rateser unit surface area on the decay time obtained by the improvedG technique. The faster the decay of the excited electron, the fasterhe oxidation rate. We considered the relationship of Fig. 9 to be asollows.
According to studies using FT-IR, when oxygen is introducednto adsorbed CO on Pt/Al2O3, the generation of CO2 is observed.t is known that the generation time of the CO2 is fast when theispersion of Pt in Pt/Al2O3 is low [7]. Hence, we presume that theeason that the generation time of CO2 is faster for larger particless due to an effect whereby the electron states with a larger numberf the terrace sites are more important than those with a smallerumber of surface atoms.
Fig. 10 shows the reaction between CO and O2 on Pt accordingo the Langmuir–Hinshelwood’s mechanism, in which an oxygentom and a CO atom are first adsorbed on the Pt surface. Accord-ng to their paper, the generation rate of CO2 is not influenced byhe partial pressure of CO, but that the higher the partial pressuref O2, the faster the generation rate of CO2 [1]. In order to explainhy a higher partial pressure of oxygen leads to faster generation
f CO2, it is believed that the bonding of Pt and oxygen can be moreasily broken under these conditions. That is, it is thought that thet O bond can be broken more easily because the dispersion of Pts low (because the size of the Pt particle is large), so the genera-ion of CO2 is easier. Next, we surmised that the relationship of thelectrons is as follows. The speed with which the electron moveso the Pt when the Pt O changes into metallic Pt can be thoughtf as an index for the strength of the Pt O bond. Fig. 11 shows theovement of the electron about the Pt O bond according to the
angmuir–Hinshelwood mechanism. It is assumed that the elec-
ron shifts more easily to the Pt when the Pt O unit is more easilyevered.
Therefore, when the decay time of the electron obtained by themproved TG method is fast (as shown in Fig. 8), the Pt O unit
[
[
nce 263 (2012) 230–235 235
shown in Fig. 11 is more easily cut, and the generation of CO2, asshown in Fig. 10, is fast.
As a result, we have been able to explain why faster decay ofthe excited electrons leads to faster reaction speeds. Therefore, webelieve that the tendency shown in Fig. 9 is explicable.
4. Conclusion
In summary, we have studied the relationship between the Ptparticle size in Pt/Al2O3 and catalytic performance. We also stud-ied the relationship between the Pt particle size in Pt/Al2O3 andthe decay time of excited electrons using an improved TG tech-nique. Consequently, we identified a tendency that the faster decayof excited electrons leads to a greater oxidation rate. Next, we con-sidered the possible mechanism for this tendency. From studiesusing FT-IR, one of the functions identified as being necessary forgenerating CO2 at a high rate has been thought to be that the elec-tron in the Pt O bond moves to the Pt side and the Pt+ changes toPt metal. The decay time obtained by the improved TG techniquegives an indication of the time that the excited electrons take toreturn to the ground state. Therefore, the decay time correspondsto the process whereby Pt+ becomes Pt metal. Thus, we were ableto explain the tendency for faster decay of the excited electronsto yield greater oxidation rates. These experimental results haveprovided us with information that will enable us to decrease theconsumption of precious metals by increasing the rate at whichexcited electrons decay.
Acknowledgements
The authors thank Prof. Y. Sekine from Waseda University andProf. K. Yamamoto from Nagoya University for discussions. Theauthors also thank the NISSAN MOTOR CO., LTD. for providing thechance to undertake this research. Part of this work was supportedby a New Energy and Industrial Technology Development Organi-zation of the Japanese Government (project code: P08023).
References
[1] B. Atalik, D. Uner, Journal of Catalysis 241 (2006) 268–275.[2] S. Schimpf, M. Lucas, C. Mohr, U. Rodemerck, A. Brückner, J. Radnik, H. Hofmeis-
ter, P. Claus, Catalysis Today 72 (2002) 63.[3] J.L. Gland, M.R. McClellan, F.R. McFeely, Journal of Chemical Physics 79 (1983)
6349.[4] A. Szabo, M.A. Henderson, J.T. Yates, Journal of Chemical Physics 96 (1992) 6191.[5] J.Z. Xu, J.T. Yates, Journal of Chemical Physics 99 (1993) 725.[6] Irene E. Beck, V.I. Bukhtiyarov, I.Y. Pak, Journal of Catalysis 268 (2009) 60–67.[7] S. Derrouiche, D. Bianchi, Journal of Catalysis 230 (2005) 359–374.[8] K. Katayama, M. Yamaguchi, T. Sawada, Applied Physics Letters 82
(2003) 2775.[9] M. Yamaguchi, K. Katayama, T. Sawada, Chemical Physics Letters 377 (2003)
589.10] K. Katayama, M. Yamaguchi, T. Sawada, Journal of Applied Physics 94 (2003)
Berlin, 1986.12] Q. Shen, T. Toyoda, Thin Solid Films 438–439 (2003) 167.13] A. Harata, Q. Shen, T. Sawada, Annual Review of Physical Chemistry 50 (1999)
