FTIR, UV-Vis, and HRTEM Study of Au/ZrO2 Catalyst: Reduced Reactivity in the CO-O2 Reaction of Electron-Deficient Gold Sites Present on the Used Samples

Article abstract of DOI:10.1021/jp980890t

A Fourier transform infrared (FTIR), UV-vis, and high-resolution transmission electron microscopy (HRTEM) study of fresh and used Au/ZrO2 catalyst is reported. FTIR data show that on a fresh sample CO is adsorbed on regular gold metallic sites, while on the used sample the adsorption sites are bidimensional gold clusters made positive by interaction with the support. NIR-UV-vis reflectance spectra show that the gold plasmonic peak appears broader and the ZrO2 absorption edge weaker on the used sample with respect to the fresh one. HRTEM indicates the disappearance of the gold metallic particles in the used catalysts. Moreover, a strong reduction in the CO oxidation activity has been evidenced on the used sample and ascribed to the positivization of gold and/or to the decrease of the number of highly uncoordinated gold sites, consistently with the spreading of gold at the support surface.

Full text of DOI:10.1021/jp980890t

© Copyright 1998 by the American Chemical Society
VOLUME 102, NUMBER 30, JULY 23, 1998
LETTERS
FTIR, UV-Vis, and HRTEM Study of Au/ZrO2 Catalyst: Reduced Reactivity in the
CO-O2 Reaction of Electron-Deficient Gold Sites Present on the Used Samples
F. Boccuzzi,*,† G. Cerrato,† F. Pinna,‡ and G. Strukul‡
Dipartimento di Chimica I. F. M., UniVersita` di Torino, Torino, Italy, and Dipartimento di Chimica, UniVersita`
di Venezia, Venezia, Italy
ReceiVed: January 22, 1998; In Final Form: April 27, 1998
A Fourier transform infrared (FTIR), UV-vis, and high-resolution transmission electron microscopy (HRTEM)
study of fresh and used Au/ZrO2 catalyst is reported. FTIR data show that on a fresh sample CO is adsorbed
on regular gold metallic sites, while on the used sample the adsorption sites are bidimensional gold clusters
made positive by interaction with the support. NIR-UV-vis reflectance spectra show that the gold plasmonic
peak appears broader and the ZrO2 absorption edge weaker on the used sample with respect to the fresh one.
HRTEM indicates the disappearance of the gold metallic particles in the used catalysts. Moreover, a strong
reduction in the CO oxidation activity has been evidenced on the used sample and ascribed to the positivization
of gold and/or to the decrease of the number of highly uncoordinated gold sites, consistently with the spreading
of gold at the support surface.
1. Introduction
close contact.5 No clear explanation has been found to date
for the different catalytic properties observed in samples
containing gold particles of similar size, supported on a variety
of supports,6 or for the deactivation observed on some supports
such as ZrO2.7,8 The problem is probably that, up to now, there
is not enough information at the atomic level on the metal-
oxide interfaces, on the electronic character of the metal atoms
present at these interfaces, and on how the contact with the oxide
interface can be influenced by physical and chemical treatments.
Provided that small particles must be considered a priori as out
of equilibrium, modifications of the interfacial and surface
energies can occur depending on the gas-phase composition and
temperature. In this respect, different phenomena have been
evidenced such as particle coalescence, migration, spreading,
breaking, encapsulation, alloying, and dissolution, depending
on the nature of the metal, on the surface stoichiometry of the
oxide, and on the gas composition. As for gold samples, since
gold is the least reactive and the noblest of all metals, only
coalescence phenomena are generally expected. In this work,
we report some results that evidence the occurrence of irrevers-
ible effects on the properties of an Au/ZrO2 sample upon
Gold did not attract much attention in heterogeneous catalysis
until 10 years ago because of its chemical inertness and because
of the difficulty to obtain highly dispersed samples. In the last
10 years, several methods have been described by Haruta et
al.1 for the preparation of small gold particles. The same authors
have shown that gold is remarkably active for low-temperature
CO oxidation when it is well dispersed on some metal oxides
or hydroxides of alkaline earth metals.2 Gold catalysts are now
being studied intensively also for their low-temperature activity
in the water gas-shift reaction3 and for other reactions such as
the catalytic combustion of hydrocarbons, NO reduction,
hydrogenation of carbon oxides, etc.4
As for the CO oxidation mechanism, on the basis of FTIR
studies of this reaction on Au/ZnO and Au/TiO2, it was
concluded that the reaction occurs mainly on small gold
particles, where oxygen and carbon monoxide are adsorbed in
* Address correspondence to Prof. Flora Boccuzzi, Dipartimento di
Chimica I. F. M., Via P. Giuria 7, 10125 Torino, Italy. Fax: +39011
6707855. E-mail: Boccuzzi@ch.unito.it.
† Universita` di Torino.
‡ Universita` di Venezia.
S1089-5647(98)00890-6 CCC: $15.00 © 1998 American Chemical Society
Published on Web 07/03/1998

5734 J. Phys. Chem. B, Vol. 102, No. 30, 1998
Letters
reduction and oxidation treatments under mild conditions that,
quite unexpectedly, give rise to electropositive gold surface sites.
2. Experimental Section
The catalysts were prepared by deposition-precipitation on
ZrO2 obtained by hydrolysis of zirconium oxycloride with
ammonia. The support was suspended in an aqueous solution
of HAuCl4, followed by neutralization with Na2CO3 (pH 7.5)
and aging in suspension at 343 K for 2 h. After drying at 383
K, the samples were calcined at 673 K for 2 h. Metal chemical
analysis was performed by atomic absorption spectroscopy. Gold
loading is 1.72 wt %.
FTIR spectra were recorded with a 1760 Perkin-Elmer FTIR
spectrometer at 298 K. The powders were pressed in pellets
and subjected, directly in the IR cell, to different thermal and
chemical treatments. The sample simply pretreated in oxygen
up to 673 K will be termed “fresh”, the sample heated first in
oxygen at 673 K, contacted with 15 mbar of hydrogen for 30
min and with 15 mbar of oxygen for 30 min at 353 K, and
finally treated in oxygen at 673 K, will be termed “used”.
HRTEM images were made with a JEOL 2000 EX electron
microscope, equipped with a top entry stage and a LaB6
filament. The powder was ultrasonically dispersed in n-heptane,
and the suspension was deposited on a copper grid, coated with
a porous carbon film. The residual vacuum at the specimen
region was approximately 1 × 10-6 mbar.
UV-vis near-infrared (NIR) diffuse reflectance experiments
have been performed with a Varian CARY5 spectrometer.
3. Results and Discussion
In Figure 1 the FTIR spectra of CO adsorbed on fresh and
used Au/ZrO2 catalysts are displayed: in part a the spectra
obtained by interaction with 10 mbar of CO on the two samples
are compared, in parts b and c the evolution of the spectra when
the equilibrium CO pressure decreases from 10 to 0.05 mbar
on the fresh and used samples is reported, respectively.
Two main bands are evident in both spectra reported in Figure
1a. The first one, at 2193 cm-1 on the fresh sample and
assigned to the CO adsorbed on Zr4+ surface sites, exhibits only
a very small frequency shift in the used sample while the other
one, at 2108 cm-1 on the fresh sample and assigned to CO
adsorbed on Au0 surface sites, is abruptly shifted up to 2140
cm-1 on the used one. The large difference in the intensity
ratio of the bands at 2193 and 2140 cm-1 in the spectra taken
under 10 mbar of CO, reported in Figure 1a and 1c can be
explained by taking in account that the spectra shown in Figure
1c have been taken at the end of the CO-O2 interaction, shown
below, and that, after this reaction, a large fraction of the surface
Zr4+ ions is masked by carbonate species.
Looking at the spectra reported in Figure 1b and 1c another
difference can be observed: on the fresh sample (part b) the
intensities of both absorptions smoothly decrease in parallel by
decreasing the equilibrium pressure and both absorption blue-
shift, while on the used sample (part c) the band at 2140 cm-1
appears as significantly more resistant to outgassing than the
2108 cm-1 one and exhibits, by decreasing the coverage, a red
shift. A recent paper9 concerning the FTIR study of CO
adsorbed on the Au (332) surface reported bands in the 2100-
2120 cm-1 range with blue-shifting due to decreasing coverage,
similar to the behavior exhibited by our fresh sample, by some
polycrystalline gold films studied by Dumas,10 and by gold small
particles on other supports.11 These studies clearly show that
metallic gold exhibits a behavior similar to that of metallic
copper. From the wide literature concerning the CO chemi-
Figure 1. FTIR absorbance spectra of CO adsorbed on fresh and used
Au/ZrO2 samples. (a) Curve 1, 10 mbar on fresh sample; curve 2, 10
mbar on used sample. (b) Spectra of CO adsorbed on the fresh sample
taken at decreasing CO equilibrium pressures, from 10 to 0.05 mbar
(curves 1-5). (c) Spectra of CO adsorbed on the used sample taken at
decreasing CO equilibrium pressures, from 10 to 0.05 mbar (curves
1-8).
sorption on monocrystalline and dispersed copper samples, it
is well known that the IR band of CO adsorbed on oxidized
copper sites is usually blue-shifted with respect to that of reduced
samples and it is less reversible upon outgassing.12 Bands
significantly blue-shifted with respect to those observed on
extended surfaces have been detected also in very small copper
clusters or copper ultrathin films deposited on different sub-
strates.13 They have been interpreted as due to the CO
chemisorption on copper sites where there is a charge polariza-
tion away from CO. Similar polarization effects occur also
when Cu atoms and/or two-dimensional Cu clusters are depos-
ited at the surface of an oxidic support.14 On this basis it can

Letters
J. Phys. Chem. B, Vol. 102, No. 30, 1998 5735
Figure 3. NIR-UV-vis reflectance spectra of fresh (curve 1) and
used (curve 2) samples.
by the chemical treatments: in fact, both shape and average
diameter of the support’s particles almost coincide with the fresh
catalysts. On the other hand, no “as is” Au particles are
evident: individual metallic particles, smaller and with less
evident lattice fringes, are very seldom detected as that indicated
by the arrow in the Figure 2b; this particle shows also a larger
contact area with the support. In these conditions it is possible
to transfer an electron from the metal to the support. The
electron transfer causes usually little effect on the surface
properties of the metal particles; however, if the metal particles
are very small, an effect could be observed. Moreover, the
formation of an intermetallic Au-Zr compound or a solid
solution is possible even if the spacing of the lattice fringes of
the particle shown in Figure 2b, ≈ 2.9 Å, does not coincide
exactly with those reported in the Au-Zr ASTM files.16
Figure 3 shows the NIR-UV-vis reflectance spectra of the
fresh and used samples. The first peak at ≈18 000 cm-1 can
be assigned to the plasmonic oscillation mode of nanosized gold
particles, and the edge at ≈42 000 cm-1 to the fundamental
absorption of ZrO2 microcrystals. Two main effects are evident
from the spectra of the used catalysts: the plasmonic peak
appears broader and the absorption edge seems to be weaker.
Recently many papers17 have been published on the optical
properties of nanocrystalline gold samples, concerning the effect
of the particle size, the shape of the particle, and the effect of
the dielectric properties of the surrounding material. It has been
shown that by decreasing the particle size a broadening of the
plasmonic peak is observed.18 Also, changes in the shape of
the metallic particles, such as a flattening, can produce similar
effects on the shape of the plasmonic peak. In our case, it can
be hypothesized that these changes may occur as a consequence
of alternate treatment in reducing and oxidizing atmosphere.
Small particles must be considered as in/out equilibrium
condition; this nonequilibrium condition can be modified by
changing the gaseous atmospheres.19 The decrease of the
fundamental ZrO2 absorption, associated with the positivization
of the gold sites observed on our used samples, could be
interpreted as an indication that a spreading of gold occurs at
the surface of the support. Gold has a high absorption
coefficient; by increasing the spreading of gold a larger fraction
of the oxide cannot interact with the radiation.
Figure 2. HRTEM micrographs of fresh (a) and used (b) Au/ZrO2
samples.
be hypothesized that the band at 2140 cm-1 observed on the
used Au/ZrO2 sample is due to CO adsorption on Au bidimen-
sional clusters made positive by interaction with the support.
Recently,15 a CO absorption band has been detected at a very
similar frequency and with a similar resistance to outgassing
on a Au/HY sample. This band was assigned to CO adsorbed
on Auδ+ sites produced on gold particles by interaction with
the acidic protons inside the framework of HY zeolite. For the
Au/HY sample, a shoulder at 2110-2120 cm-1 is also present,
assigned to CO adsorbed on neutral gold sites exposed at the
surface of big crystallites, probably located on the external
zeolite surface. In our case the 2108 cm-1 band is almost
completely lacking; however, the surface hydroxyls of the oxidic
support do not appear to be involved in the Auδ+ sites formation,
in fact, the spectra of both the fresh and the used sample (not
shown for sake of brevity) are almost identical.
To find an explanation for these findings, in Figure 2 the
electron micrographs of the fresh (Figure 2a) and used (Figure
2b) sample are shown. Figure 2a refers to the Au/ZrO2 fresh
catalyst: there is a high incidence of fringe patterns, typical of
the ZrO2 monoclinic modification and metallic particles, show-
ing a roundish shape of about 10-15 nm average diameter (the
statistics have been made on different images, not reported for
sake of brevity). The almost hexagonal particle reported in the
inset to Figure 2a, by inspection of the fringes, can be ascribed
to a gold particle exposing mainly (111) faces. In Figure 2b a
micrograph relative to a used Au/ZrO2 catalyst is reported. On
one hand, it can be said that the ZrO2 matrix is not influenced
Finally, we discuss the modification in the catalytic activity
of the gold sites in the CO-O2 reaction: in Figure 4 a
comparison between CO/O2 experiments made on a fresh and
on a used Au/ZrO2 catalyst is presented. The CO/O2 interaction
on the fresh sample (Figure 4a) produces, during the first
minutes of contact, the shift and the intensification of the band

5736 J. Phys. Chem. B, Vol. 102, No. 30, 1998
Letters
1000 cm-1 range (not shown for sake of brevity). The same
interaction in the used sample (Figure 4b) produces quite
different effects: immediately an erosion of the main band from
the low-frequency side, at ∼2100 cm-1, and a decrease of the
band assigned to CO adsorbed on Zr4+ sites are observed. At
increasing contact times the 2132 cm-1 band reduces slightly
in intensity. Molecular CO2 is not detected, and some bands
gradually increase with the contact time in the carbonate-like
region (Figure 4c) assigned to bidentate carbonate and to
bicarbonate species;20 however, the intensities of these bands
are significantly lower than those on the fresh samples.
Probably, on the used samples, where positive gold sites are
already present before oxygen contact, the oxygen does not
activate CO for carbonate-like species formation at the border-
line of the bidimensional metallic particles with the support.
From the data discussed it appears evident that on the used
sample only a small fraction of the exposed gold sites, those
associated with the weak component at ∼2100 cm-1, exhibits
the usual reactivity toward oxygen while the electron-deficient
two-dimensional metallic islands, which are the vast majority
of the exposed gold sites on the used samples, appear signifi-
cantly less reactive than the neutral gold sites exposed on three-
dimensional particles of the new sample. Electronic and/or
structural reasons can be at the origin of the observed differ-
ences. The positive gold sites give rise to stronger Au-CO
bonds, and this fact can cause a reduction of the oxidation
activity. The role of the oxidation state and that of the CO on
Cu, Cu2O, and CuO bond strength in the CO oxidation reaction
has been discussed by Jernigan and Somorjai.21 Moreover, the
coordinative unsaturation of the surface atoms can play a role.
The three-dimensional and almost spherical particles seen by
HRTEM on the fresh sample expose a large amount of step
and corner sites, where CO and oxygen can be adsorbed in close
contact, while flat, bidimensional particles expose mainly highly
coordinated sites.
References and Notes
(1) Haruta, M. Catal. Today 1997, 36, 153 and references therein.
(2) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet M.
J.; Delmon, B. J. Catal. 1993, 144, 175.
(3) Andreeva, D.; Idakiev, V.; Tabakova, T.; Andreev, A. J. Catal.
1996, 158, 354.
(4) Sakurai, H.; Haruta, M. Appl. Catal. A 1995, 127, 93.
(5) Boccuzzi, F.; Chiorino, A.; Tsubota, S.;. Haruta, M. J. Phys. Chem.
1996, 100, 3625.
Figure 4. FTIR absorbance spectra of CO-O2 interactions on Au/
(6) Haruta, M. Catal. SurVeys Jpn. 1997, 1, 61.
ZrO2 samples. (a) Fresh sample, spectral range 2400-2000 cm-1
:
(7) Baiker, A.; Kilo, M.; Maciejewski, M.; Menzi S.; Wakaun, A. In
New Frontiers in Catalysis; Guczi, L., et al., Eds.; Elsevier Science
Publishers: Amsterdam, 1993; pp 1257.
dashed curve, 10 mbar of CO; full line, spectrum taken 10 min after
the admission of O2. (b) Used sample, spectral range 2400-2000 cm-1
:
dashed curve, 10 mbar of CO; full lines,1-4 spectra taken at increasing
times after the admission of O2 (2, 6, 8, 10 min). (c) Used sample,
spectral range 4000-1000 cm-1: dashed curve, 10 mbar of CO; full
lines,1-3 spectra taken at increasing contact times (2, 6, 10 min).
(8) Boccuzzi, F.; Guglielminotti, E.; Pinna, F.; Strukul, G. Surf. Sci.
1997, 377-379, 728.
(9) Ruggiero, C.; Hollins, P. Surf. Sci. 1997, 377-79, 583.
(10) Dumas, P.; Tobin, R. G.; Richards, P. L. Surf. Sci. 1986, 171, 579.
(11) France, J.; Hollins, P. J. Electron Spectrosc. Relat. Phenom. 1993,
64/65, 251 and ref 5.
(12) Hollins P. Surf. Sci. Rep. 1992, 16, 53 and references therein.
(13) Hoffman, F. M.; Paul, J. J Chem. Phys. 1997, 86, 2990; 87, 1957.
(14) Campbell, C. T. Surf. Sci. Rep. 1997, 27, 1. Xu, X.; Vesecky, S.
M.; Goodman, D. W. Science 1992, 258, 788.
(15) Guillemot, D.; Boresvkov, V. Yu.; Kazansky, V. B.; Polisset-Thfoin,
M.; Fraissard, J. J. Chem. Soc., Faraday Trans. 1997, 93, 3587.
(16) JCPDS 4-784; 10-299; 16-46; 18-580; 28-442.
(17) (a) Alvarez, M. M.; Khoury, T. J.; Schaaff, T. G.; Shafigullin, N.
M.; Vezmar, I.; Whetten, R. L. J. Phys. Chem. B 1997, 101, 3706. (b)
Logunov, S. L.; Ahmadi, T. S.; El-Sayed, M. A.; Khoury, J. T.; Whetten,
R. L. J. Phys. Chem. B 1997, 101, 3713.
(18) van der Zande, B. M. I.; Bohmer, M. R.; Fokkink, L. G. J.;
Schonenberger, C. J. Phys. Chem. B 1997, 101, 852.
(19) Somorjai, G. A. Catal. Lett. 1992, 12, 17 and references therein.
(20) Guglielminotti, E. Langmuir 1990, 6, 5.
initially observed at 2108-2116 cm-1. Many of the observed
features are similar to those previously reported for other gold-
containing samples;5 in particular, the shift of the main band
and the shoulder at higher frequency were ascribed to the
presence of uncoordinated gold sites able to chemisorb oxygen
and carbon monoxide at the same time. The intensification of
the CO absorption band, not observed on the previously
examined samples in similar conditions,5 can be related to a
transient increase in the local CO pressure produced by the
oxygen inlet. This effect was not detected on the previously
examined samples because of their higher reaction rate due to
the smaller size of the gold particles. The band related to CO
adsorbed on Zr4+ sites decreases in intensity, and molecular
CO2 is detected (band at 2351 cm-1). Moreover, on the fresh
sample, very strong bands grow up in the carbonate-like 1700-
(21) Jernigan, G. G.; Somorjai, G. A. J. Catal. 1994, 147, 567.