Characterization and catalytic performance of designed surfaces

Article abstract of DOI:10.1016/S1381-1169(00)00043-1

Designed surfaces with new and unique structures provide an opportunity to develop efficient, catalytic, molecularly organized surfaces that can lead to the ultimate catalyst technologies in the microscopic, mesoscopic, and macroscopic regions. To elucidate on these surfaces, a study investigated surface catalytic reactions assisted by gas phase molecules, which is a new aspect of catalysis on designed Co/TiO₂, Co/Al₂O₃, Co/SiO₂ surfaces. Results showed that the catalytic NO-CO reaction on the attached Co/Al₂O₃ catalyst occurred through the dinitrosyls as reaction intermediates whose conformation and amount are influenced by the gas phase CO molecules. There were no CO molecules at the catalyst surface during the reactions. The gas phase CO molecules promoted the reactivity of the dinitrosyls on the trigonally distorted tetrahedral Co²⁺ ions attached on Al₂O₃ 23 times. A new attached Mo₂/TiO₂ layer/SiO₂ catalyst was prepared by taking advantage of the reaction of Mo₂(η³-C₃H₅)₄ with surface OH groups of the TiO₂ layer followed by successive treatments with H₂ and O₂. The catalytic behavior of the Mo₂/TiO₂ layer/SiO₂ catalyst resembled that of the Mo₂/SiO₂ catalyst, but the former showed a higher activity than the latter for ethanol dehydrogenation. A new kinetic aspect of molecules on terraces and steps on a TiO₂(110)-(1x1) surface by STM was observed, which may have relevance to oxide catalysis.

Full text of DOI:10.1016/S1381-1169(00)00043-1

Journal of Molecular Catalysis A: Chemical 158 Ž2000. 67–83
www.elsevier.comrlocatermolcata
Characterization and catalytic performance of designed
surfaces
1
2
3
T. Kawaguchi , N. Ichikuni , A. Yamaguchi, T. Shido, H. Onishi ,
)
K. Fukui, Y. Iwasawa
Department of Chemistry, Graduate School of Science, The UniÕersity of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Abstract
This paper attempts to extend our previous studies on the chemical design and characterization of metal atomsrensem-
bles attached at oxide surfaces and on the in-situ observation of adsorbed molecules on oxide model surfaces. This paper
presents a new aspect of catalysis at designed CorTiO , CorAl O , CorSiO surfaces characterized by FT-IR and
2
2
3
2
EXAFS, namely ‘‘Surface catalytic reactions assisted by gas phase molecules’’, in which the amount and reactivity of
reaction intermediate can be promoted and the structure of reaction intermediate can also be controlled by gas phase
molecules undetectable at the surface. This paper also reports the application of one-atomic-layer TiO on SiO as a unique
2
2
support for an Mo dimer complex to prepare a supported Mo-dimer catalyst. Finally, a new aspect of regulation of a catalytic
reaction by monoatomic-height steps on TiO Ž110. relevant to oxide-layer catalysis characterized by STM is also reported.
2
q 2000 Elsevier Science B.V. All rights reserved.
Keywords: Catalytic reaction mechanisms; Design of active structures; Characterization by EXAFS, FT-IR and STM; Supported Co and Mo
catalysts; TiO Ž110. single crystal; NO reduction; Ethanol oxidation; HCOOD dehydrogenation
2
1
. Introduction
oxide single crystals as model surfaces, provide
an opportunity for the development of efficient
catalytic molecularly-organized surfaces en route
to the ultimate catalyst technologies in the mi-
Designed surfaces with new and distinct
structures, prepared stepwise in a controllable
manner by using organic and inorganic metal
complexes and clusters as precursors andror
croscopic
Ž
atomicrmolecular
.
, mesoscopic, and
macroscopic
Ž
mean-field
.
regions w1–8x. This
class of catalysts and model surfaces can also
provide a fundamental implication on the
molecular-level mechanism and essential con-
trolling factors for catalysis at surfaces. The
longer term challenge to surface chemistry of
oxide-supported metal catalysts is to address
important issues in activity and selectivity in
catalytic reactions, in particular the principles of
tuning of metal reactivity and activating reac-
)
Corresponding author. Fax: q81-3-5800-6892.
E-mail address: iwasawa@chem.s.u-tokyo.ac.jp ŽY. Iwasawa..
1
Present address: Research Center for Characterization and
Analytical Science, Kao, 1334 Minato, Wakayama 640-8580,
Japan.
2
Present address: Department of Applied Chemistry, Faculty of
Engineering, Chiba University, Yayoi, Inage-ku, Chiba 263-0022,
Japan.
3
Present address: KAST, KSP East-404, Sakado, Takatsu-ku,
Kawasaki 213-0012, Japan.
1
381-1169r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S1381-1169Ž00.00043-1

6
8
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
tion intermediates at oxide surfaces. The re-
quirements and design of definite ensemble
sizesrstructures and reaction fieldsrenviron-
ments represent important but as yet few ad-
dressed challenge to the field. The role of gas
phase molecules in surface catalytic reactions
remained to be solved to understand catalytic
phenomena at surfaces under reaction condi-
tions w9x. Recently, molecular-level catalyst
preparation has become realistic on the basis of
modern physical techniques and accumulated
knowledge of oxide surfaces, and a clear exam-
ple of the surface catalytic reaction through
strongly adsorbed intermediates assisted by
weakly adsorbed molecules present only in the
presence of the gas phase molecules has been
faces via attaching a metal complex with appro-
priate ligands to oxide surfaces, and subse-
quently chemical transformation of the ligands
of the attached complex to reactive ones.
The present study reports the role of gas
phase molecules in surface catalytic reactions as
a new aspect of catalysis on CorAl O ,
2
3
CorTiO and CorSiO , which are prepared by
2
2
chemical vapor deposition
Co2
CO.8 on Al O , TiO and SiO , respec-
CVD reactions of
Ž .
Ž
2
3
2
2
tively, involving the reaction of surface Co
species with surface OH groups as a key step
for the catalyst preparation. This paper also
reports the preparation of Mo dimers attached
on one-atomic-layer TiO on SiO by taking
2
2
3
advantage of a selective reaction of Mo2
C H
Ž
h y
uncovered on a designed NbrSiO catalyst
.
with surface OH groups of the TiO2
2
3
5 4
w1,10,11x.
layer followed by oxidation with O . The cat-
2
The use of molecularly-designed catalysts and
model surfaces seems to be prerequisite to ex-
amining the mechanism and concept of cataly-
sis. There are several approaches to chemical
design of active metal sites and ensembles on
oxide surfaces by metal–organic and –inorganic
complexes and clusters. It has been demon-
strated that stable and active structures at cata-
lyst surfaces can be prepared by reaction and
strong interaction of inorganic and organic
metal-complex precursors with oxides such as
SiO , Al O , TiO , MgO, La O , zeolites, etc.,
alytic property of the obtained Mo rTiO
2
2
layerrSiO sample in ethanol oxidation was
2
examined, and compared to those of Mo rSiO
2
2
and Mo rTiO . One-atomic-layer oxides have
2
2
been demonstrated to provide a new class of
support for metals and metal oxides and also
new catalysis w5,23–27x. The scanning tunneling
microscopy STM study on TiO 110 as a
Ž . Ž .
2
model catalyst also reveals a new suppressive
effect of monoatomic-height steps on catalytic
formic acid dehydrogenation.
2
2
3
2
2
3
as proved by physical techniques w1,2,12x. In the
catalysts obtained by this method, metal atoms
are attached to oxide surfaces via metal–oxygen
bonding at metal–oxide interfaces. The attached
metal atoms are often highly dispersed as
monomers or distributed forming ensemble
structures such as dimers and clusters by the
direct bondings to oxide surfaces, depending on
the oxidation state of attached metals, the acid-
2
. Experimental
2
.1. Catalyst preparation
2.1.1. Attached Co catalysts
A CorTiO catalyst was prepared in a simi-
2
lar manner to that for CorAl O and CorSiO
2
3
2
reported in literature w28–30x. Co
Ž
CO
.
was
, SiO2
by a
2
8
attached on Al O
Ž
Degussa Alon C
, and TiO2 Degussa P25
CVD method. Al O , SiO and TiO2 were
.
2
3
ityrbasicity
Ž
intrinsic and impurity-induced
.
of
Ž
Aerosil 200
.
Ž
.
oxide surfaces, the distribution of hydroxyl
groups at oxide surfaces, the nature and behav-
ior of precursors, the loading of attached metals,
the preparation methods and conditions, the post
treatment conditions, etc. w1–23x. Catalytically-
2
3
2
pretreated for 2 h at 573 K before use as
supports for Co2
CO.8. Supporting Co2 CO.8
on the pretreated Al O led to the formation of
Ž
Ž
2
3
Co4
Ž
CO
.
as shown in Fig. 1. The species with
12
active species may be synthesized at oxide sur-
a Co`Co bond distance of 0.254 nm, which is

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
69
probed by in-situ EXAFS analysis in the pres-
ence of O w31x.
2
The attached CorSiO catalyst was also pre-
2
pared in a similar manner w28–30x. However,
the EXAFS analysis under any condition re-
vealed no Co`Co bonding, which indicates that
2
q
Co ions are dispersed in an isolated manner
at the SiO surface, as monomers rather than an
2
2
q
ensemble structure composed of four Co ions.
The attached CorTiO catalyst was also pre-
2
pared in a similar manner to the case of the
attached CorAl O catalyst. This catalyst has
2
3
been not characterized in detail. The Co atoms
are suggested to be attached to the TiO surface
2
in a bivalent state with Co`O bond at 0.205 nm
w32x.
2
.1.2. Mo rone-atomic-layer TiO rSiO
2
2
2
The one-atomic-layer TiO on SiO was pre-
2
2
pared by using the reaction between Ti
OC H . ŽSoekawa Chemical
and OH groups
Fuji Davison 952; surface
according to literature w33x.
vapor was deposited on the SiO2
Ž
i-
.
3
7 4
of the SiO surface
area, 300 m rg
Ti i-OC H
Ž
2
2
.
Ž
.
7 4
3
Fig. 1. Preparation steps for the attached CorAl O , CorTiO
2
3
2
surface that was pretreated at 473 K for 2 h in
air by a CVD method to react with the OH
and CorSiO catalysts.
2
larger than 0.249 nm for C `Co4ŽCO.12, was
groups of SiO at 353 K for 2 h. The unreacted
3
v
2
very sensitive to O to be oxidized to a new
Ti
Ž
i-OC H
.
was removed by evacuation at
2
3
7 4
species wCoOx rAl O , accompanied with the
473 K. After the attaching reaction, the obtained
sample was exposed to water vapor at room
4
2
3
disappearance of the direct Co`Co bond and
appearance of a new Co`O bond at 0.205 nm,
which is shorter than the Co`O bond in CoO
temperature
i-OC H ligands, followed by the calcination at
RT to decompose the remaining
Ž .
3
7
crystal
Ž
0.2133 nm
.
w28,29x. When heated at 513
773 K for 2 h. The above procedure was re-
K under vacuum, the wCoOx rAl O reacted
peated twice to prepare the TiO monolayers on
4
2
3
2
with surface OH groups of Al O , as indicated
SiO2.
2
3
by the decrease in the intensity of the n
Ž
OH
.
The Mo attachment on the TiO layerrSiO
2
2
y
1
band centered at 3550 cm , to form a
was performed in a similar manner in the case
2
q
wCo x rAl O catalyst
Ž
denoted as an attached
CorAl O catalyst hereinafter accompanied by
the formation of H O. The diffuse reflectance
of Mo rSiO w34x and Mo rAl O w35x by
4
2
3
2
2
2
2
3
3
3
.
using Mo
Ž
h -C H
.
. Mo
Ž
h -C H
.
was
2
3
2
3
5 4
2
3
5 4
purified by two recrystallization in strictly wa-
2
2
q
UVrVIS spectrum indicates that the Co ions
ter-free pentane at 193 K under high purity Ar
3
are situated in a trigonally distorted T position.
Ž
99.99995%
C H
.
. The reaction between Mo2
Ž
h -
d
The bonding scheme is simply illustrated as a
.
and surface OH groups of the TiO2
3
5 4
2
q
bivalent Co structure with two Co`O bonds
at 0.197 nm, characterized by EXAFS in Fig. 1.
The existence of the neighboring Co ions is
layerrSiO in pentane at 273 K was completed
2
within 20 min. The amount of molybdenum
attached on the support was determined from

7
0
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
the initial
attachment
Ž
before attachment
.
.
and the final
Ž
after
.
5 4
Ž
dead volume: 170 cm3. equipped with a gas
chromatograph using a Porapak Q column.
3
concentrations of Mo2
Ž
h -C H
3
in pentane by colorimetry w34x. The Mo loading
was 1.1 wt.%. The surface allyl-type complexes
thus obtained were further evacuated at 350 K
2.3. FT-IR measurements
6
q
The FT-IR spectra for the NO`CO catalytic
for 1 h and converted to Mo species by H2
reduction at 823 K and subsequent O oxidation
system were measured in an IR cell with two
2
at 773 K. The preparation steps are shown in
Fig. 2.
CaF windows, which were combined in a closed
2
circulating system, on a JEOL KIR 7000 spec-
trometer. The spectra were recorded as differ-
2
.2. Catalytic reactions
y1
ence spectra with a 2 cm resolution and about
0-s time acquisition by a double-beam method,
5
Catalytic NO`CO reactions were carried out
where the spectra for a sample cell with a
catalyst wafer were subtracted by the spectra for
the gas phase in a reference cell without catalyst
wafer which was also combined in the closed
in a closed circulating system
Ž
dead volume:
equipped with a gas chromatograph
Shimadzu GC-8A using a 5A Molecular Sieve
3
2
05 cm
.
Ž
.
column and a Unibeads C column in a parallel
manner. The reactions of NO alone were also
conducted in the same system.
The catalytic ethanol oxidation reactions were
also carried out in a closed circulating system
circulating system. FT-IR spectra for the
3
Mo2
Ž
h -C H
.
complex on the TiO2
3
5 4
layerrSiO sample at different temperatures
2
were measured in an IR cell with two NaCl
windows which were combined in a closed
circulating system, on a JASCO FTIR 7000.
The spectra were recorded by a single-beam
method, where the difference spectra were taken
from the spectra for the Mo complexrTiO
2
2
2
layerrSiO sample and the TiO layerrSiO
2
2
sample.
2
.4. XAFS measurements
X-ray absorption fine structure
Ž
XAFS
.
spec-
tra at Mo and Ti K-edges were measured at RT
in a transmission mode at the beam lines 10B
Ž
for Mo
Institute for Material Structure Science
IMSS. ŽProposal No.: 93G060
. Synchrotron ra-
diation emitted from a 2.5-GeV storage ring
was monochromatized by an Si 311 channel
cut monochromator and a Si 111 sagittal focus-
.
and 7C
Ž
for Ti
.
of Photon Factory in
Ž
PF-
.
Ž
.
Ž
.
ing double-crystal monochromator, respectively
for Mo and Ti XAFS measurements. The sam-
ple was transferred to glass cells with two Kap-
ton windows connected to a closed circulating
system without contacting air. The detail of the
EXAFS analysis has been described in our pre-
vious papers w17,33x.
Fig. 2. Preparation steps for the attached Mo rTiO layer rSiO2
2
2
catalyst.

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
71
2
.5. STM obserÕation of TiO (110) surfaces
the Nb-ethoxide intermediate
Ž
X
.
was stable
2
below 600 K and could decompose to a product,
ethene only above 600 K. In con-
qH O. Ž
trast, in the coexistence of another reactant
Formic acid
TiO2 110
in a UHV-compatible scanning tunneling micro-
scope JEOL-JSTM4500VT with an electro-
Ž
HCOOD
.
decomposition on a
Ž
P
.
2
Ž
.
single-crystal surface was examined
X
molecule
vated to decompose to a different product, ac-
etaldehyde even at low temperatures
A the intermediate X was acti-
Ž . Ž .
Ž
.
chemically etched tungsten tip. Constant current
topography was continuously imaged at a rate
of 33 s per frame and recorded in video. A
Ž
Q
qH2. Ž .
like 473 K. The reaction rate was remarkably
promoted and the reaction path was switched
polished TiO2Ž110
.
wafer of 6.5=1=0.25
Earth Chemicals, Japan gave a sharp
LEED pattern after repeated cycles of
3 keV, 0.3 mA and vacuum
over from the dehydration
Ž
X
P to the dehy-
.
3
mm
Ž
.
drogenation assisted by weakly ad-
Ž
X
Q
.
X
Ž
1=1
.
sorbed molecule A w9x.
q
Ar sputtering
Ž
.
Recently, we have found an example of sur-
face catalytic NO`CO reactions assisted by gas
annealing at 900 K. Research grade HCOOD
gas was purified through trap–thaw cycles w36x.
phase CO molecules, which are undetectable at
the attached Co catalyst surfaces w2,37x. Slow
N O formation from NO alone in a closed
2
3
. Results and discussion
circulating system was observed above 350 K
on the attached CorAl O catalyst. The N O
2
3
2
3
.1. Surface catalytic NO`CO reactions as-
formation at 463 K was still slow and the
sisted by gas phase CO molecules
amount of N O formed in the first 15 min of
2
reaction was only ca. 2% of total Co quantity
dispersed at the catalyst surface. The catalytic
reduction of NO in the coexistence of CO to
form N O and CO by the equation 2NOqCO
A typical mechanism for heterogeneous cat-
alytic reactions is simply illustrated in Fig. 3
where the reaction intermediate is trans-
formed to the product by surface unimolec-
a ,
Ž .
X
Ž .
2
2
Ž
P
.
N OqCO , proceeded rapidly compared to
2
2
ular decomposition; that is, a stoichiometric re-
action step proceeds without aid of other
molecules. In contrast to the simple expectation,
we have found evidence that the reaction inter-
mediate of an important catalytic reaction can
be profoundly influenced by the ambient gas
w9x. In the catalytic dehydrogenation reaction of
the reaction of NO alone. The initial rate of the
NO`CO reaction was 23 times larger than that
for the NO reaction at 463 K. If CO alone was
admitted to the fresh catalyst, no CO2 was
produced, indicating that there existed no reac-
tive oxygen species on the fresh CorAl O
2
3
catalyst. The activity of the CorAl O catalyst
2
3
ethanol on the Nb monomer catalyst
Ž
NbrSiO2
.
,
was much higher than that of a conventional
impregnation CorAl O catalyst prepared by
2
3
an impregnation method using an aqueous solu-
tion of Co nitrate as shown in Fig. 4. The
products of the catalytic NO`CO reaction in
the temperature range 353–463 K were N O
2
and CO . The activation energy was 14 kJ
2
y
1
mol for the CorAl O catalyst, while it was
2
3
y
1
much larger
Ž
46 kJ mol
.
for the impregnation
Fig. 3. Surface catalytic reactions assisted by gas-phase molecules;
Ža. surface catalytic reaction ŽA P. via intermediate X; Žb.
surface catalytic reaction ŽA P. promoted by reactant molecule
CorAl O catalyst. In the more elevated tem-
2
3
peratures N O was further converted to N ,
X
2
2
or coexisting molecule A, where the enhancement of the reaction
where N and CO were main products by the
equations 2NOqCO N OqCO and N O
rate for the formation of P or switchover of the reaction path from
2
2
X
P formation is caused by A.
2
2
2

7
2
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
intensity of this peak did not increase in the
pressure range 0.25 Pa–4.0 kPa. On the other
hand, the reaction rate was of 1.6th order with
respect to NO pressure in the same pressure
range. Hence, the small linear NO species seems
not to contribute to the catalytic NO reduction
significantly.
The relative intensity of the symmetric
y
1
stretching peak at 1831 cm and the antisym-
y
1
metric stretching peak at 1776 cm
changed
with exposure time and also by evacuation. The
bond angle
a Co atom
Ž
u
.
between the two NO ligands on
angle: NO`Co`NO.2can be calcu-
ur2 , where
I and I represent the antisymmetric stretching
Ž
lated by the equation, I rI stan
Ž
.
as
s
as
s
peak intensity
peak intensity, respectively w2,38x. The bond
Ias and the symmetric stretching
Ž .
Fig. 4. NO`CO reactions on the attached CorAl O Ža.,
2
3
Ž .
angles u are plotted against exposure time in
CorTiO Žb., and CorSiO Žc. catalysts and an impregnation
2
2
Fig. 6
proportional to I rsin
although n
nNO
a . The amount of adsorbed NO is
Ž . Ž .
CorAl O catalyst Žd.; NOs4.0 kPa, COs4.0 kPa, reaction
2
3
2
or I rcos2Žur2
temperatures463 K. The activation energies for the catalysts are
Ž
ur2
.
.
,
as
s
also shown.
is not an absolute value for the
NO
amount of adsorbed NO. The calculated values
of nNO are plotted against time in Fig. 6 . The
change in nNO coincided with the change in the
qCO N q CO . In this article, the first
b
Ž .
2
2
reaction step, 2NOqCO N OqCO , has
2
2
been studied from the mechanistic interest. The
amount of adsorbed NO determined gravimetri-
catalytic activity of the CorTiO catalyst was a
cally in a separate experiment. Fig. 6
reveals that there are two kinds of dinitrosyls
species A and species B ; the dinitrosyl with a
a and b
Ž . Ž .
2
little lower than that of the CorAl O catalyst.
2
3
2
q
The CorSiO catalyst with an isolated Co
Ž
.
2
structure showed a much lower activity than the
CorAl O and CorTiO catalysts as shown in
2
3
2
Fig. 4.
The surface catalytic NO`CO reaction on the
most active CorAl O catalyst among the ex-
2
3
amined catalysts was monitored by IR spec-
troscopy and the gas phase products were ana-
lyzed by gas chromatography. Note that CO
molecules, which promote the surface NO con-
version, were volumetrically and spectroscopi-
cally undetectable as adsorbates on the
CorAl O catalyst surface w2x. FT-IR spectrum
2
3
for adsorbed NO molecules in Fig. 5 exhibits
y
1
two peaks at 1831 and 1776 cm , which are
assigned to the dinitrosyl geminal NO species
adsorbed on a Co atom Co
NO. . w2x. Besides
the dinitrosyl peaks, a small peak at 1821 cm
is assigned to a linear NO species but the
Ž
.
Ž
Ž
2
Fig. 5. Peak deconvolution of FT-IR spectrum for NO adsorbed
y
1
on CorAl O ; the spectrum at 120 min after NO Ž4.0 kPa.
2
3
exposure at 298 K.

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
73
small bond angle species A is rapidly formed
Ž .
upon NO adsorption and easily desorbed by
evacuation at RT, while the dinitrosyl with a
large bond angle species B is slowly formed
Ž .
and not desorbed by evacuation at RT. We
estimated the ratio of species A to species B to
be 1:2.1 at RT. From this ratio and the results
that the bond angle u of species B was 1078 and
the bond angle for a mixture of species A and
species B at 120 min of Fig. 6 was 978, we can
estimate the u for species A at RT to be 808 w2x.
Fig. 7 shows the effects of CO on the bond
angle and amount of adsorbed NO at 323, 373,
4
23, 448 and 463 K w2x. The amount of the
dinitrosyls decreased with increasing tempera-
ture, while the bond angle u increased with an
increase of temperature. When CO was intro-
duced to the system, the bond angle u of the
surface dinitrosyls increased. The degree of the
angle expanding was different at temperatures
and larger at higher temperatures, and at 323 K,
almost zero effect was observed. It was also
Fig. 7. Time profiles and CO effects of the bond angle and
amount of dinitrosyls on CorAl2O3 at 323–463 K; NOs1.3
kPa, COs1.3 kPa.
found that the amount of the dinitrosyls in-
creased by CO admission. Again, the degree of
the increase was larger at higher temperatures.
Thus, the effects of the ambient CO were large
at the temperatures where the catalytic NO`CO
reactions proceeded significantly.
The present finding of the u extension and
the nNO increase promoted by the gas phase CO
molecules undetectable at the catalyst surfaces
is hardly explained by physical collision of
impinging CO molecules. We do not know the
exact reason for the observed events at present,
but this phenomenon may involve a local struc-
2
q
tural change of the Co –Al O interface by
2
3
CO impinging under the catalytic NO`CO reac-
tion conditions w32x.
3
.2. Performance of Mo TiO layerrSiO for
r
2
2
2
ethanol oxidation
Fig. 6. Time profiles of the bond angle and amount of dinitrosyls
It may be worthy to briefly summarize the
structures of Mo-oxide species supported on
on CorAl O at 298 K. The catalyst was exposed to NO of 1.25
2
3
kPa at time zero and evacuated at 120 min.

7
4
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
SiO2 and TiO2 by different precursors and
preparation conditions w3,5,30,39–42x.
with K MoO solutions led to the formation of
2 4
2
y
monomolybdate and Mo O
dimers with a
2
7
The interaction of ammonium heptamolyb-
date often used as a precursor with SiO2 is
generally weak. There is always controversy for
the structures of prepared Mo-oxide species in
weakly interacting systems presented thus far.
The weak interaction resulted in the formation
low symmetry. Thus, highly polymerized species
like MoO , octamolybdate, or heptamolybdate
3
are increasingly less favorable relative to
monomeric or probably dimeric species.
Instead of aqueous impregnation preparations
with ammonium heptamolybdate, preparation
procedures via non-aqueous organometallic Mo
precursors suppress the formation of crystalline
of bulk MoO , even for Mo loadings as low as
3
0
.4 wt.% w43x, while catalysts with loadings as
high as ca.6 wt.% were reported to contain no
MoO , which starts forming even at very low
3
crystalline MoO w44x. The pH of the impregnat-
Mo loadings in the impregnation method. Yer-
3
ing solution or at the silica surface may play a
role during the formation of the surface species.
makov et al. w8,47x prepared Mo catalysts by
3
taking advantage of reaction of Mo h -allyl
6
y
At pH-2, heptamolybdate
Ž
Mo O24.
and
anions are major
complex with surface OH groups of SiO and
7
2
4
y
6q
octamolybdate
Ž
Mo O26.
proposed tetrahedrally-coordinated Mo
ions
8
species, where the SiO surface is charged posi-
chemically attached to SiO2 through two
2
Mo O Si bonds. Iwasawa et al. 3,5,17,41
x
`
`
w
tively, leading to coulombic interaction to pro-
duce polymolybdate species. At 2-pH-6, the
heptamolybdate anions species are in equilib-
characterized the surface Mo species by many
physical techniques such as EXAFS, diffuse
reflectance UVrVIS, photoluminescence, etc,.
and presented definite structural models such as
2
y
rium with monomolybdate anion
Ž
MoO4
.
,
where the SiO surface is negatively charged,
2
2
y
leading to repulsive interaction between the Mo
species and the surface, therefore unfavorable
for the formation of isolated Mo species at the
surface. The low surface pH of SiO2 itself
under ambient conditions also results in the
formation of silicomolybdic acid or more likely
isolated or paired bidentate MoO4 or tetrahe-
dral dimers joined by bridging oxygen, depend-
ing on the kind of SiO supports and precursors
2
3
3
Ž
Mo h -C H or Mo2
Ž . Žh -C H . ., Mo load-
3 5 4 3 5 4
ing, and Na impurity level in the support. Che et
al. w48x reported a preparation of catalysts from
8
y
6q
of trimolybdic clusters
Ž
Mo O
.
at low Mo
MoCl to possess Mo monomeric dispersion.
3
13
5
loadings on SiO . Thus, on SiO , pH values
Monodispersed molybdenum oxide was charac-
2
2
5
q
high enough for monomolybdate formation are
usually not reached; consequently, under hy-
drated conditions, only polymeric species are
detected.
Alkali impurities are known to largely affect
the properties of catalysts. It can be assumed
that the effects of alkali ions are related to
structural changes of the surface compounds or
the active centers on the catalyst surface. The
terized by a high percentage of Mo
as de-
tected by ESR and by the absence of the inter-
5
q
6q
valence transition
Ž
Mo
Mo
.
around l000
nm in the UVrVIS spectra.
Using a non-aqueous preparation method with
3
5
Mo2
Ž
h -C H
.
and
Ž
h -C H5.Mo2
Ž
CO.4 as
3
5 4
5
precursors, Roark et al. w49x also concluded
from Raman spectra that microcrystalline MoO
3
was not formed on SiO as long as the number
2
structures of alkali-doped MorSiO catalysts,
of Mo atoms deposited per unit surface re-
mained below the density of silanol groups.
High dispersion and high loading of Mo can be
2
2
y
2
4
y
e.g. octamolybdate, Mo O , and MoO
2
7
2
7
y
species, K Mo O , Mo O dimers,
K MoO , were characterized by a combination
6
7
24
2
reached on SiO via non-aqueous preparation
2
4
2
of Raman, EXAFS, and X-ray absorption near-
edge structure XANES
w45x and correlated
with catalytic performance w46x; impregnation
routes.
Ž
.
The XAFS results revealed spreading or
break-up of the polymolybdate clusters on SiO2

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
75
upon dehydration. It was proposed from Raman
which was interpreted as being due to
monomolybdates. The observed frequency is
rather high for isolated monomolybdates and
alternatively suggests the presence of dimers
w55,56x. Raman bands w57x were detected at
y
1
1
that the polymolybdate cluster
changed into isolated monooxo sites
Ž
944 cm
986 cm
.
y
Ž
.
by calcination. However, Desikan et al. w50x
y
1
observed the 955 cm peak, which is assigned
to a dioxo structure in a distorted tetrahedral
symmetry and not a monooxo structure.
Williams et al. w43x concluded from the Raman
y
1
993–998 cm for MorTiO samples irrespec-
2
tive of loading, and together with XANES data,
assigned to an isolated, highly distorted tetrahe-
dral monooxo Mo species, but contradicted ear-
lier literature w58–61x.
Thus, it seems very hard to prepare a definite
molecular structure of Mo-oxide species on SiO2
data that in the dehydrated phase, a structure
6
q
containing isolated Mo
ion was formed on
SiO . In view of the large frequency differences
2
of the Mo5O band found by Williams et al.
w43x, Desikan et al. w50x, and de Boer et al. w51x,
it is likely that the different surface species were
prepared and detected by these groups.
Thus, a variety of structures of Mo species on
SiO2 have been demonstrated depending on
precursors, preparation conditions, post-treat-
ments, the atmosphere, Mo loadings, population
of surface OH groups, alkali impurity of the
support, etc., due to a weak interaction between
and also on TiO by a conventional impregna-
2
tion method using aqueous media. Instead, at
least at low loadings of Mo, we have prepared
well-defined Mo structures by using organo-
metallic precursors under controlled and limited
conditions and provided a new approach for
fundamental research for understanding the ori-
gin and mechanism of catalysis w3–5x. In the
present study we have prepared a new Mo
2
Mo species and SiO surface.
dimerrTiO layerrSiO catalyst by taking ad-
2
2
2
In contrast to SiO support, TiO support can
vantage of the reaction of the TiO layer on
2
2
2
interact more strongly with Mo species. Ng and
SiO , and found a higher catalytic activity than
2
Gulari w52x concluded that tetrahedral molybdate
a Mo rSiO catalyst and a higher selectivity
2
2
species were abundant and coexisting with octa-
hedrally coordinated polymeric molybdates at
submonolayer coverages. They concluded that
for acetaldehyde formation than a Mo rTiO
2
2
catalyst.
3
We can assume that Mo2
Ž
h -C H
.
with a
5 4
3
6q
Mo
ions bind to TiO surface strongly and
Mo`Mo bond distance of 0.2183 nm reacts
with the TiO2 layer supported on SiO2 in a
similar manner to that for the Mo rSiO cata-
2
uniformly until monolayer coverage is reached.
Bond et al. w53x demonstrated the presence of
Mo monomers at ca. 4 wt.% loadings, which is
2
2
3
lyst, where Mo h -C H reacted exclusively
Ž .
2 3 5 4
followed by MoO at higher loadings. They
with surface OH groups of SiO . The number of
3
2
also reported that no highly dispersed mono-
the allyl ligands on an Mo atom in the obtained
layer was formed on TiO2 anatase .
Ž .
surface complex was determined by temperature
Ž .
programmed reduction TPR up to 823 K under
The structure of the surface Mo-oxide species
was described as a highly distorted MoO octa-
16.2 kPa of H after evacuation of the surface
complex at 353 K for 1 h. The total amount
Ž .
carbon base of the desorbed products such as
propane, propene, ethane, ethene, etc., was 2.1
6
2
hedron by comparing the Mo`O stretching fre-
y
1
quency
Ž
948–965 cm
.
in the catalyst with
2
y
that in reference compounds like MoO ,
4
Ž
NH4.2 Mo O , K Mo O , Ba CaMoO , etc.
C per Mo2.
2
7
2
2
7
Ž
2
6
y
1
w54x. Bands at 954 cm
assigned to Mo5O
Fig. 8 shows the IR spectra for the incipient
attached Mo complex on the TiO layerrSiO ,
the sample after evacuation at RT for 10 min,
and the sample after evacuation at 353 K for 1
y
1
stretch
stretch
.
.
and 875 cm
indicated the presence of hepta or octa-
Ž
assigned to Mo`O`Mo
2
2
2
y
1
molyubdates, whereas the band at 934 cm
was observed for a low-loaded sample 1 wt.%
y
1
Ž
.
,
h. There are two peaks at 1627 and 1450 cm .

7
6
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
ter of the surface Mo complex on the one-
2
atomic TiO layer resembles that of the com-
2
plex on TiO bulk rather than that of the com-
2
plex on SiO2.
The surface Mo-allyl complexes were re-
duced with H at 823 K, followed by carefully
2
exposing to a small amount of O equivalent to
2
the amount of Mo atoms in the sample at RT,
and then they were oxidized to produce hexava-
lent Mo species with O at 773 K. The obtained
2
6
q
Mo ions in the Mo rTiO layerrSiO were
characterized by XANES. The XANES spec-
trum exhibited a distinct pre-edge peak, as
shown in Fig. 9, which indicates that the Mo
ions are situated in a tetrahedral symmetry.
2
2
2
6
q
Fig. 8. FT-IR spectra for the Mo -allyl complexes attached on the
2
TiO2 layerrSiO ŽA. and on KBr ŽB.. Ža. The incipient attached
2
Fig. 10 shows performance of the attached
Mo2 complex on TiO layerrSiO ; Žb. evacuation of Ža. at room
2
2
Mo rTiO layerrSiO catalyst for ethanol ox-
2
2
2
temperature for 10 min; Žc. evacuation of Žb. at 353 K for 1 h.
idative-dehydrogenation at 433 K in the absence
of O . The selectivity determined from the ini-
2
tial rate reached up to more than 99%. The
activity for acetaldehyde formation was about
twice that of the previous attached Mo rSiO
The former peak frequency is referred to C5C
stretching, and hence, the peak is assigned to
2
2
the s-allyl ligand. The latter peak is assigned to
catalyst w41x. The formation of acetaldehyde,
accompanied with water formation, decreased
with reaction time and eventually stopped in the
3
p-allyl ligand because the Mo2
Ž
h -C H
.
3
5 4
complex on a KBr disk showed a peak at 1457
y
1
and 1510 cm
for terminal-type p-allyl and
absence of O as shown in Fig. 10. The satu-
2
bridge-type p-allyl ligands, respectively. The
p-allyl ligand on an Rh atom in the attached
rated amount of acetaldehyde produced was half
of the amount of Mo atoms loaded on the TiO2
RhrTiO catalyst exhibited a peak at 1462
2
layerrSiO . This stoichiometric feature is simi-
y
1
2
cm
w39,40x. In consequence, the desorption
lar to that observed on the attached Mo rSiO
2
2
process of the allyl ligands in the Mo com-
2
plexrTiO layerrSiO sample is illustrated as
2
2
follows.
3
Mo2
Ž
h -C H
.
attached on SiO exhibited
3
5 4
2
y
1
the peaks at 1627 and 1503 cm , while the
complex attached on TiO2 bulk showed the
peaks at 1627 and 1450 cm . Thus, the charac-
Fig. 9. XANES spectrum for the attached Mo rTiO layerrSiO2
2
2
y
1
catalyst.

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
77
ethanol oxidation to produce acetaldehyde pro-
ceeded catalytically without stopping. The acti-
vation energy for the ethanol oxidation in the
presence of O2 on the attached Mo rTiO
2
2
y
1
layerrSiO was 71 kJ mol . The activation
2
energy for the ethanol oxidation on Mo rSiO
2
2
y1
was 53 kJ mol w41x.
To examine the dynamic behavior of the
active Mo atoms during ethanol oxidation, we
have measured the EXAFS spectra for the
Mo rTiO layerrSiO catalyst before and after
2
2
2
exposing to ethanol at 433 K. Fig. 11 a shows
Ž .
3
the Fourier transform of the k -weighted EX-
AFS data for the catalyst before ethanol dehy-
drogenation in the absence of O , which depicts
2
two distinct peaks in the region of 0.1–0.2 nm.
These two peaks are straightforwardly assigned
to Mo`O judging from the bond distances. This
was further confirmed by performing a two-
3
wave curve fitting for the k x
Ž
k
.
oscillation
Fig. 10. Reaction profiles for the ethanol dehydrogenation on
obtained by the inverse Fourier transform of the
region of the two peaks. The best-fit result is
listed in Table 1. The Mo5O double bond was
Mo rTiO layerrSiO , Mo rTiO , Mo rSiO , and TiO
2
2
2
2
2
2
2
2
layerrSiO catalysts at 433 K in the absence of O ; cat.: 0.25 g,
2
2
PŽC H OH.: 2.7 kPa.
2
5
observed at 0.169 nm
and the Mo`O single bond was
CN : 0.63
observed at 0.212 nm CN: 0.30 . These are
comparable to 0.170 nm CN: 0.9 and 0.210
nm CN: 0.9
Ž
coordination number
Ž
.
.
catalyst, where an Mo dimer unit acted as a
reaction site for the conversion of ethanol to
acetaldehyde w17,41x. We think that the Mo
Ž
.
.
Ž
Ž
.
though the CN for the Mo rTiO
2
2
dimer in the attached Mo rTiO layerrSiO
layerrSiO catalyst were smaller than those for
2
2
2
2
catalyst also works as an active unit. The
the Mo rSiO w17x. The CNs of the surface
2
2
Mo`Mo interaction was observed during the
bondings determined by EXAFS are often
smaller than the real CN expected from the
surface structure when the structure is highly
distorted. No direct Mo`Mo bonding was found
course of the ethanol oxidation by EXAFS as
discussed hereinafter. When a Mo rTiO cata-
2
2
lyst was employed as a catalyst, the amount of
acetaldehyde equivalent to the Mo quantity in
the Mo rTiO catalyst was produced under the
for the fresh Mo rTiO layerrSiO catalyst.
2
2
2
2
2
identical reaction condition. Hence, on the
Mo rTiO catalyst, an Mo monomer may be
2
2
regarded as an active site. From this point of
view, the behavior of the Mo rTiO layerrSiO
2
2
2
catalyst resembles that of the Mo rSiO cata-
2
2
lyst.
When O was admitted to the reaction sys-
2
tem after the saturation of the acetaldehyde
formation, the reaction began again. If both
Fig. 11. Mo K-edge EXAFS Fourier transforms for the Mo rTiO
2
2
layerrSiO catalysts before Ža. and after Žb. exposure to 0.93 kPa
2
ethanol and O were admitted to the system, the
of ethanol at 433 K for 7 h in the absence of O .
2
2

7
8
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
Table 1
distance probably by a bridging oxygen. The
bond distances are similar to those for the
Mo rSiO catalyst after exposing to ethanol at
EXAFS bet-fit resutls for the Mo rTiO layerrSiO catalysts
2
2
2
before ŽA. and after ŽB. ethanol dehydrogenation at 433 K in the
absence of O2
2
2
Reaction condition: PŽC H OH.s0.9 kPa, 7 h, 433 K.
The quality of the parameters determined by EXAFS analysis
were estimated by the residual factors.
Ž
, 0.194 nm CN: 1.7
. Ž .
2
5
433 K; 0.171 nm
and 0.261 nm CN: 0.5
Mo Mo, respectively. The Mo species Fig.
CN:1.0
Ž
.
for Mo5O, Mo`O and
`
11 b
Ž
a
y1
nm.b
2
Sample Bond
Distance Žnm. CN D E ŽeV. DW Ž10
o
Ž .. was oxidized to the original Mo species
2
A
B
Mo`O 0.169
Mo`O 0.212
Ti`O
0.184
0.6 y0.3
0.3 y15.9
2.9 y9.9
0.7 y9.9
0.034
0.032
0.060
0.074
Ž
Fig. 11
K. In the Mo rSiO system, the catalytic
Ž .. again by oxidation with O at 773
a
2
2
2
Ti`Ti 0.293
ethanol oxidation has been demonstrated to pro-
ceed in conjunction with the reversible struc-
tural oscillation of the active sites, lateral change
by ca. 0.04 nm and vertical change by ca. 0.01
nm. Similarly, the Mo`O bonds markedly
Mo`O 0.169
Mo`O 0.201
Mo`Mo 0.267
0.8
1.6
0.3
2.3 y9.4
0.8 y4.6
14.3
5.1
7.7
0.048
0.083
0.028
0.026
0.075
Ti`O
0.184
Ti`Ti 0.298
changed from 0.212 nm to 0.201 nm and the
a
Coordination number.
Debye Waller factor.
`
Mo Mo bonding feature remarkably changed
from no bonding state to a strong bonding state
with Mo`Mo bond at 0.267 nm in the course of
b
Fig. 11
the EXAFS data of the catalyst after exposing to
.93 kPa of ethanol at 433 K for 7 h. A peak
Ž
b
.
shows the Fourier transform for
the dehydrogenation of ethanol. Consequently,
the catalytic ethanol oxidation on the Mo rTiO
2
2
0
layerrSiO catalyst proceeds in conjunction
2
assigned for Mo5O bond was observed to-
gether with a shoulder assigned for Mo`O bond
in the region of 0.1–0.2 nm. Besides Mo`O
with breaking and formation of Mo`Mo bond
as follows.
bonds, a new peak around 0.23 nm
uncorrected appeared. This may be assigned to
Mo`Mo bond formed by reduction of the Mo
Ž
phase shift
.
6
q
state to a lower valent state. In fact, when the
catalyst was exposed to ethanol at 433 K, ac-
etaldehyde and water were produced, indicating
6
q
the reduction of the Mo
ions. The result of
6
q
Fig. 10 expects the transformation of Mo
5
q
structure to Mo
structure because of two-
This structural change may be associated with
the structure change at the interface. To exam-
electron reduction per an Mo dimer. The de-
tailed analysis of the EXAFS data by three-wave
ine the structural change of the TiO layer, we
2
Ž
Mo`O, Mo`O and Mo`Mo
.
fitting over the
have performed the EXAFS analysis at Ti K-
edge. Fig. 12 shows the Fourier transform of the
Ti K-edge EXAFS data for the Mo rTiO
region including all the three peaks was per-
formed. The best-fit result is shown in Table 1.
The bond distances were determined to be 0.169
2
2
layerrSiO catalyst before
Ž
a
.
and after
Ž
b
.
2
Ž
0
CN: 0.8
.
, 0.201
Ž
CN: 1.6
.
and 0.267 nm
Ž
CN:
exposing to ethanol at 433 K. However, both
Fourier transforms were similar to each other.
The EXAFS analysis by a curve-fitting tech-
nique is shown in Table 1. The Ti`O and
Ti`Ti bond distances for the fresh catalyst were
for Mo5O, Mo`O and Mo`Mo, respec-
.3
.
tively. Since the Mo`Mo bond of the fresh
catalyst was not observed at Rt, the appearance
of a Mo`Mo bond in the EXAFS result of the
catalyst exposed to ethanol at 433 K suggests
that the Mo`Mo bond was fixed at a definite
determined to be 0.184 and 0.293 nm, respec-
tively, while those for the ethanol-exposed cata-

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
79
Individual atoms that constitute a sputter-an-
nealed TiO2 110 1=1 surface were resolved
Ž
.
-
Ž
.
by STM w64x and atomic force microscopy oper-
ated in a frequency modulated non-contact mode
Ž
NC-AFM
. w65x, as shown in Fig. 13. Each Ti
atoms with five-fold coordination in the bright
rows of 0.65-nm separation with each other
Fig. 12. Ti K-edge EXAFS Fourier transforms for the Mo rTiO
w
x
2
2
parallel to the 001 direction were clearly ob-
served at a regular interval of 0.30 nm by STM.
Atomically resolved NC-AFM image repro-
layerrSiO catalysts before Ža. and after Žb. exposure to 0.93 kPa
2
of ethanol at 433 K for 7 h in the absence of O .
2
duced the 1=1 unit, 0.65=0.30 nm of the
Ž .
alignment of the bridge oxygen atoms. Several
dark sites on the bright lines in the NC-AFM
image should be vacancies of the bridge oxygen
atoms. The periodicity of these images repro-
duces an ideal surface truncation of rutile
lyst were 0.184 and 0.298 nm, respectively. The
Ti–Ti bond distance seems to increase by 0.005
nm. However, it would not be valid to discuss
on the structural change from this little change
in the Ti`Ti bond distance.
TiO2
When the TiO2
was exposed to a formic acid ambient of 1.3=
Ž
110
.
single crystal.
In this study, structural change in the outer-
Ž
110 surface heated at 450 K
.
most surface, that is the one-atomic TiO layer,
2
y
6
to which the active Mo atoms are chemically
attached, was not observed, but we suppose that
the interface structure may change to assist a
structural change in supported metal sites during
catalytic reactions.
10 Pa for 10 min, small particles were formed
on the terraces as imaged by STM Fig. 14
Ž
a .
Ž ..
The exposed surface was cooled to RT in the
3
.3. Long-range depletion effect of steps on
TiO (110) for catalytic HCOOD dehydrogena-
2
tion
The great potential of scanning probe mi-
Ž .
croscopy SPM to directly observe the surface
structures and reactions at a high resolution has
been demonstrated on metal oxide surfaces
w1,62x. The most promising ability of STM ex-
pected in relation to catalysis research is to
discriminate and visualize reaction events oc-
curring at different sites on a surface. Recently
the site-specific adsorption of pyridine pro-
moted on four-fold coordinated Ti centers at the
steps with particular azimuth orientations on
TiO2
We have found the long-range suppressive
effect of single-atom height steps at a TiO2 110
surface on the formic acid HCOOD dehydro-
genation reaction which takes place over ter-
races far from the steps.
110
Ž . has been visualized by STM w63x.
Ž
.
Ž
.
Fig. 13. STM Ža. and NC-AFM Žb. images and a structure model
2
for a TiO Ž110.-Ž1=1. surface; STM: V s1.0 V, 12=12 nm ;
2
s
2
AFM: f s270 kHz, A sca. 30 nm, D f s80 Hz, 9.4=9.4 nm .
o
o

8
0
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
cles were located at the on-top position of the
bridge oxygen atoms. Thus we assume that a
CO molecule adsorbed at the surface on-top
2
site of the bridge oxygen atom forming carbon-
2
y
ate
The STM image of the particles in Fig. 14 a
Ž .
CO3 group makes the product particle.
Ž .
seems elongated, not spherical. The similar
elongated feature was observed for carbonate
2
y
Ž
CO3
surface exposed to CO w71x. The formic acid
produced on an Na-deposited TiO2 110
. Ž .
2
dehydrogenation on TiO
bimolecular reaction mechanism w67x. Thus, the
present observation of CO
Ž
110
.
takes place by a
2
2
3
y
provides the fol-
lowing reaction mechanism.
The local number density depended on the
distance from the step. Few particles appeared
in the proximity of the step, whereas they were
produced at random on the terrace far from the
step. The near-step area on the terrace looks a
depletion belt completely covered with formate
ions, as a result.
Fig. 14. Ža. STM image after the reaction in 1.3=10y6 Pa of
HCOOD at 450 K for 10 min, followed by evacuation while
cooling to room temperature; V sq2.0 V, I s0.10 nA, 25=25
s
t
2
y6
nm . Žb. STM image during the catalytic reaction of 1.6=10
Pa of HCOOD at 420 K. The image was recorded at 230 s after
In-situ STM observation was tried during the
particle formation reaction in order to provide
the information on location and transport of
formate ions as an intermediate of the reaction.
exposure to HCOOD at 420 K; V sq1.7 V, I s0.20 nA,
s
t
2
2
0=20 nm .
presence of formic acid vapor and the atmo-
sphere was evacuated to saturate the surface
A TiO2
posed to formic acid vapor of 1.6=10 Pa in
the microscope chamber at ts0. Fig. 14
110 surface heated at 420 K was ex-
Ž .
y
6
with formate ions in a
Ž
2=1
.
ordered structure.
b
Ž .
The STM topography was determined at RT.
The smallest protrusions presented in a gray
Ž .
scale and arranged in a 2=1 order are the
formate ions w66x. Many blips larger and brighter
than the formate were observed on the terraces.
In the temperature range 420–500 K, the dehy-
drogenation reaction of formic acid to form H2
recorded at 420 K shows an image of the ex-
posed surface observed at ts230 s. The num-
ber of formate ions imaged as small bright dots
increased with exposure time. The number den-
sity of formates was often reduced in the prox-
imity of steps. We can assume that the non-uni-
form distribution of the product particles results
from the non-uniform distribution of formate
intermediates w36x.
and CO on TiO2Ž110
.
proceeded catalytically
Ž
the activation energy: 15 kJ
2
and selectively
y
1
mol
.
w67x. It was determined from a relative
These results indicate that the particle forma-
tion reaction, which occurred on the terrace far
from the step, was suppressed in nanometer-
position to the formates on two Ti atoms in a
bridge form w66,68–70x that the product parti-

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
81
proximity of single-atom height steps on
TiO2 110 . This may be a novel phenomenon.
The inverse effect, step-induced enhancement of
a reaction, has long been demonstrated w72x. The
suppressive effect by single-atom height steps
the dinitrosyls as reaction intermediates X
Ž .
Ž
.
whose conformation and amount are affected by
the gas phase CO molecules. There are no CO
molecules at the catalyst surface during the
catalytic reactions. The reactivity of the dinitro-
2
q
on TiO2
Ž
110
.
ranges over eight lattice units, 2.4
syls on the trigonally distorted tetrahedral Co
ions attached on Al O is promoted 23 times by
nm. It is quite important to know how an atom-
size surface singularity like the step affects the
reactivity of nanometer-proximity. This phe-
nomenon may be a result of diffusion-derived
picture. The formate ion may be consumed very
fast at the step site by recombinative desorption
or decomposition reaction, as is postulated
in the step-induced enhancement of reaction.
The local number density of the formate ion
has to reduce near the step, since adsorption–
desorption equilibrium and surface diffusion are
coupled to determine the local population of
adsorbed species. We have simulated the phe-
nomenon assuming a fast desorption from the
step and the boundary condition of us0 at
2
3
the gas phase CO molecules. In this reaction
X
mechanism, impinging CO molecules A are
Ž .
spectators at the catalyst surface, but they change
the conformation and amount of the active
Ž .
species X , and also behave as reactant. It
should be noted that molecules undetectable at
surface can give profound effects on the surface
structure of reaction sites and promote remark-
ably the reactivity of surface species. The
changes of the amount and conformation of the
dinitrosyls by the gas phase CO molecules on
the attached CorTiO catalyst were much
2
smaller and unclear at present. The phenomenon
was not observed with the attached CorSiO
2
2
q
xs0
Ž
u: particle coverage; x: distance from
catalyst with isolated Co ions.
the step end
tion w36x.
.
and obtained the following equa-
2 A new attached Mo rTiO layerrSiO
Ž .
2
2
2
catalyst was prepared by taking advantage of
3
the reaction of Mo2
Ž
h -C H
.
with surface
3
5 4
D
kh
OH groups of the TiO layer followed by suc-
lmax
s
s
Lc
2
) )
k
2kd
d
cessive treatments with H and O . The surface
2
2
Mo structure was characterized by FT-IR, TPR,
EXAFS and XANES. The character of the sur-
face Mo complex on the TiO layer resembled
where l
is maximum depletion length, and
max
D and k represent a diffusion constant and a
d
rate constant for desorption. k is a rate con-
2
2
h
that of the Mo complex on TiO bulk. The Mo
stant for hopping migration of formate ions and
2
2
dimer on the TiO layerrSiO worked as active
Lc is the lattice constant of TiO2
Ž
110
.
. When
2
2
site for ethanol dehydrogenation, whereas on
the Mo rTiO catalyst the Mo monomer be-
the ratio k rk is 100, the calculated lmax
h
d
reproduces the observed depletion-zone length
of 2.4 nm.
2
2
haved as an active unit. In this sense, the cat-
alytic behavior of the Mo rTiO layerrSiO
The present finding may predict that the TiO2
clusters with -2.4 nm size show a negligible
activity for the catalytic formic acid dehydro-
genation, while those with much larger than 2.4
nm size may work as an active catalyst.
2
2
2
catalyst resembled that of the Mo rSiO cata-
2
2
lyst, but the Mo rTiO layerrSiO catalyst
2
2
2
showed a higher activity than the Mo rSiO
2
2
catalyst for the ethanol dehydrogenation.
Ž
3
.
We have also found a new kinetic aspect
of molecules on terraces and steps on a
110 1=1 surface by STM, which may
TiO2
Ž
.
-
Ž
.
4
. Summary
The catalytic NO`CO reaction on the
be relevant to oxide catalysis. When the TiO2
1
Ž .
surface heated at 400–450 K was exposed to a
y
6
attached CorAl O catalyst proceeds through
formic acid ambient of 1–2=10
Pa, small
2
3

8
2
T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
w14x B. Pugin, F. Spindler and M. M u¨ ller, EP 496700-A1 Ž1991..
w15x Union Carbide, US Patent 5342907A Ž1994..
particles were formed on terraces. Post-reaction
STM observation revealed that the particle for-
mation was strongly suppressed in the vicinity
of single-atom height steps. The suppressive
effect of step ranged over 2.4 nm into terrace.
w16x S. Sato, Y. Iwasawa, H. Kuroda, Chem. Lett. Ž1982. 1101.
w
1
7 Y. Iwasawa, K. Asakura, H. Ishii, H. Kuroda, Z. Phys.
x
Chem. N. F. 144 Ž1985. 105.
w18x N. Ichikuni, Y. Iwasawa, in: Proc. 10th Int. Congr. Catal.,
Budapest, 1993, p. 477.
Formate ions possible reaction intermediate
Ž .
w19x K. Asakura, K.K. Bando, Y. Iwasawa, H. Arakawa, K. Isobe,
were imaged in-situ during formic acid expo-
sure at the reaction temperature. The local den-
sity of the formates and hence the product dis-
tribution, which were very low near the steps,
were simulated by a model. The produced parti-
J. Am. Chem. Soc. 112 Ž1990. 9096.
w20x K.K. Bando, K. Asakura, H. Arakawa, K. Isobe, Y. Iwasawa,
J. Phys. Chem. 100 Ž1996. 13636.
w21x Y. Izumi, T. Chihara, H. Yamazaki, Y. Iwasawa, J. Phys.
Chem. 98 Ž1994. 594.
w22x Y. Izumi, Y. Iwasawa, Chemtech 24 Ž1994. 20.
w23x K. Asakura, M. Aoki, Y. Iwasawa, Catal. Lett. 1 Ž1988. 395.
2
y
cles were suggested to be carbonates CO3 on
w
w
2
25
4
x
x
K. Asakura, Y. Iwasawa, Chem. Lett. 1988 633.
Ž
.
the bridge oxygen atoms of the TiO2
face.
Ž
110
.
sur-
K. Asakura, Y. Iwasawa, Chem. Lett. 1986 859.
Ž .
w26x M. Shirai, K. Asakura, Y. Iwasawa, J. Phys. Chem. 95
Ž1991. 9999.
w27x Y.-C. Xie, Y.-Q. Tang, Adv. Catal. 37 Ž1990. 1.
w28x M. Yamada, Y. Iwasawa, Nikkashi Ž1984. 1042.
w29x Y. Iwasawa, M. Yamada, Y. Sato, H. Kuroda, J. Mol. Catal.
Acknowledgements
2
3 Ž1984. 95.
w30x Y. Iwasawa, Supported catalysts from chemical vapor depo-
This work has been supported by Core Re-
search for Evolutional Science and Technology
sition and related techniques, in: G. Ertl, H. Kn o¨ zinger, J.
Weitkamp ŽEds.., Handbook of Heterogeneous Catalysis Vol.
2
Wiley-VCH, 1997, p. 853.
Ž
CREST
JST
.
of the Japan Science and Technology
w31x K. Asakura, Y. Iwasawa, J. Phys. Chem. 93 Ž1989. 4213.
w32x A. Yamaguch, T. Shido, T. Sasaki, K. Asakura, Y. Iwasawa,
Proc. 12th Int. Congr. Catal. Ž2000. to be published.
Ž
.
.
w33x K. Asakura, J. Inukai, Y. Iwasawa, J. Phys. Chem. 96 Ž1992.
8
29.
References
w34x Y. Iwasawa, M. Yamagishi, J. Catal. 82 Ž1983. 373.
w35x Y. Iwasawa, Y. Sato, H. Kuroda, J. Catal. 82 Ž1983. 289.
w1x Y. Iwasawa, in: Stud. Surf. Sci. Catal., J.W. Hightower,
W.N. Delgass, E. Iglesia, A.T. Bell ŽEds.., Proc. 11th Int.
Congr. Catal., Baltimore Vol. 101 Elsevier, Amsterdam,
36 Y. Iwasawa, H. Onishi, K. Fukui, S. Suzuki, T. Sasaki,
w x
Faraday Discus. 114 Ž1999. in press.
w37x Y. Iwasawa, in: Abstract of 7th Japan–China–USA Sympo-
1
996, p. 21.
sium on Catalysis, Tokyo, 1995, p. 49.
w2x A. Yamaguchi, K. Asakura, Y. Iwasawa, J. Mol. Catal.
w38x P.S. Braterman, in: Metal Carbonyl Spectra, Academic Press,
ŽSpecial Issue for SHHC, Sounthampton, 1998., 146 Ž1999.
1
975, p. 43.
6
5.
w39x Y. Iwasawa, H. Sato, Chem. Lett. Ž1985. 507.
w3x Y. Iwasawa, Adv. Catal. 35 Ž1987. 187.
w
40x Y. Iwasawa, K. Asakura, in: Homogeneous and Heteroge-
w4x Y. Iwasawa, Catal. Today 18 Ž1993. 21.
neous Catalysis, VNU Science Press, Utrecht, 1986, p. 75.
w5x Y. Iwasawa ŽEd.., Tailored Metal Catalysts, Reidel,
w41x Y. Iwasawa, H. Tanaka, in: Proc. 8th Int. Congr. Catal.,
Dortrecht, 1987.
Berlin, Vol. IV 1984, p. 381.
w6x Y. Iwasawa, Elementary reaction steps in heterogeneous
w
2 G. Mestl, T.K.K. Srinivasan, Catal. Rev. 40 1998 451.
x Ž .
4
catalysis, in: R.W. Joyner, R.A. van Santen ŽEds.., NATO
w43x C.C. Williams, J.G. Ekerdt, J.M. Jehng, F.D. Hardcastle,
ASI, Ser. C Vol. 398 1993, p. 287.
A.M. Turek, I.E. Wachs, J. Phys. Chem. 95 Ž1991. 8781.
w7x B.C. Gates, L. Guczi, H. Kn o¨ zinger ŽEds.., Metal Cluster in
w44x
Y. Barbeaux, A.R. Elamrani, E. Payen, L. Gengembre, J.P.
Catalysis, Elsevier, Amsterdam, 1986.
Bonnelle, B. Grzybowska, Appl. Catal. 44 Ž1988. 117.
w8x Yu.I. Yermakov, B.N. Kuznetsov, V.A. Zhakarov, Catalysis
w45x N.F.D.
Verbruggen, L.M.J. Von Hippel, G. Mestl, B.
Lengeler, H. Knozinger, Langmuir 10 Ž1994. 3073.
by Supported Complexes, Elsevier, Amsterdam, 1981.
w9x Y. Iwasawa, Acc. Chem. Res. 30 Ž1997. 103.
w10x M. Nishimura, K. Asakura, Y. Iwasawa, J. Chem. Soc.,
Chem. Commun. Ž1986. 1660.
¨
w
46x
x
N.F.D. Verbruggen, H. Kn o¨ zinger, Langmuir 10 1994 3148.
Ž
.
w
4
7
Y.I. Yermakov, Catal. Rev. — Sci. Eng. 13 1976 77.
Ž
.
w48x M. Che, C. Louis, J.M. Tatibouet, Polyhedron 5 Ž1986. 123.
¨
w11x M. Nishimura, K. Asakura, Y. Iwasawa, in: Proc. 9th Int.
w
4
9
x
R.D. Roark, S.D. Kohler, J.G. Ekerdt, D.S. Kim, I.E. Wachs,
Congr. Catal., Calgary Vol. 4 1988, p. 1566.
w12x J.M. Thomas, J. Mol. Catal. A: Chem. 115 Ž1997. 371.
w13x B. Pugin, F. Spindler, M. M u¨ ller, EP 496699-A1 Ž1991..
Catal. Lett. 16 Ž1992. 77.
w50
x
A.N. Desikan, L. Huang, S.T. Oyama, J. Phys. Chem. 95
Ž
1
991 10050.
.

T. Kawaguchi et al.rJournal of Molecular Catalysis A: Chemical 158 (2000) 67–83
83
w51x M. de Boer, A.J. van Dillen, D.C. Koningsberger, J.W. Geus,
M.A. Vuurman, I.E. Wachs, Catal. Lett. 11 Ž1990. 227.
w52x K.Y.S. Ng, E. Gulari, J. Catal. 92 Ž1985. 340.
w63x S. Suzuki, Y. Yamaguchi, H. Onishi, K. Fukui, T. Sasaki, Y.
Iwasawa, Catal. Lett. 50 Ž1998. 117.
w64x H. Onishi, K. Fukui, Y. Iwasawa, Bull. Chem. Soc. Jpn. 68
w53x G.C. Bond, S. Flamerz, L. van Wijk, Catal. Today 1 Ž1987.
Ž1995. 2447.
2
29.
w65x K. Fukui, H. Onishi, Y. Iwasawa, Phys. Rev. Lett. 79 Ž1997.
w54x D.S. Kim, Y. Kurusu, I.E. Wachs, F.D. Hardcastle, K.
4202.
Segawa, J. Catal. 120 Ž1989. 325.
w66x H. Onishi, Y. Iwasawa, Chem. Phys. Lett. 226 Ž1994. 111.
w67x H. Onishi, T. Aruga, Y. Iwasawa, J. Catal. 146 Ž1994. 557.
w68x Q. Guo, I. Cocks, E.M. Williams, J. Chem. Phys. 106 Ž1997.
w55x A.W. Armour, M.G.B. Dreu, P.C.H. Mitchell, J. Chem. Soc.,
Dalton Trans. 1975 Ž1975. 1493.
w56x H.J. von Becher, Z. Anorg. Allg. Chem. 474 Ž1981. 63.
2924.
w57x H. Hu, I.E. Wachs, S.R. Bare, J. Phys. Chem. 99 Ž1995.
w69x S. Thevuthasan, G.S. Herman, Y.J. Kim, S.A. Chambers,
1
0897.
w58x T. Machej, J. Haber, A.M. Turek, I.E. Wachs, Appl. Catal.
0 Ž1991. 115.
w59x F.D. Hardcastle, I.E. Wachs, J. Raman Spectrosc. 21 Ž1990.
83.
C.H.F. Peden, Z. Wang, R.X. Ynzunza, E.D. Tober, J.
Morais, C.S. Fadley, Surf. Sci. 401 Ž1998. 261.
w70x B.E. Hayden, A. King, M.A. Newton, J. Phys. Chem. B 103
Ž1999. 203.
7
6
w71x H. Onishi, T. Aruga, C. Egawa, Y. Iwasawa, J. Chem. Soc.,
w60x M.A. Vuurman, I.E. Wachs, J. Phys. Chem. 96 Ž1992. 5008.
w61x M.A. Ba n˜ ares, H. Hu, I.E. Wachs, J. Catal. 150 Ž1994. 407.
w62x Y. Iwasawa, Catal. Surveys. Jpn. 1 Ž1997. 3.
Faraday Trans. I 85 Ž1989. 2597.
w72x G.A. Somorjai, Introduction to Surface Science and Cataly-
sis, Wiley, New York, 1994.