Comparative studies of the N2O/H2, N2O/CO, H2/O2 and CO/O2 reactions on supported gold catalysts: Effect of the addition of various oxides

Article abstract of DOI:10.1016/S0021-9517(03)00185-4

The reduction of N₂O with H₂ and CO was studied on Au/TiO₂, Au/Al₂O₃, Au/M^IO_x/M^IIO_x/Al ₂O₃, and Au/MO_x/Al₂O₃ (M^I, M^II, M + Li, Rb, Mg, Co, Mn, Ce, La, Ti). All the catalyst exhibited a high catalytic activity even at low temperatures. The highest conversion was reached over Au/TiO₂-reductive pretreatment at 75°-175°C. The Al₂O₃-supported preoxidized catalysts exhibited a better performance than the TiO₂-supported preoxidized catalysts for the H₂ + O₂ reaction. For the H₂ + N₂O reaction, the differences in behavior of the TiO₂ and Al₂O₃-supported catalysts were relatively small. Over Au-based catalysts, experiments of N₂O in the absence of any reducing gas in the system did not reveal any decomposition products. A significant synergistic effects was also found for multi-component catalysts containing both an alkali metal oxide and a TMO, in this case CoO_x. However, the enhancement in activity was not as big as in the case of ceria and an alkali metal oxide.

Full text of DOI:10.1016/S0021-9517(03)00185-4

Journal of Catalysis 219 (2003) 197–205
www.elsevier.com/locate/jcat
Comparative studies of the N2O/H2, N2O/CO, H2/O2 and CO/O2
reactions on supported gold catalysts:
effect of the addition of various oxides
A.C. Gluhoi,a M.A.P. Dekkers,b and B.E. Nieuwenhuys a,∗
a
Leiden Institute of Chemistry, Department of Heterogeneous Catalysis and Surface Chemistry, PO Box 9502, 2300 RA Leiden, The Netherlands
b
Fluor Daniel BV, PO Box 4354, 2003 EJ Haarlem, The Netherlands
Received 23 December 2002; revised 31 March 2003; accepted 8 April 2003
Abstract
The reduction of N O with H and CO has been investigated on Au/TiO , Au/Al O , Au/M O /M O /Al O , and Au/MO /Al O
2 3
I
II
x
x
x
2
2
2
2
3
2
3
I
II
(M , M , M = Li, Rb, Mg, Co, Mn, Ce, La, Ti). In addition, for comparison, the oxidation reactions of H and CO with O have been
2
2
studied on some selected catalysts. All the samples are highly active at low temperatures for all the reactions. The oxides added to Au/Al O
2
3
improve the catalytic performance of gold. Interesting and synergistic effects were observed when a mixture of metal oxides is added to
Au/Al O , e.g., Rb O (Li O) and CeO . Possible mechanisms are proposed.
x
2
3
2
2
2003 Elsevier Inc. All rights reserved.
Keywords: Gold; Nitrous oxide; Carbon monoxide; Hydrogen; Oxygen; Additives; Transition metal oxides; Alkali metal oxides
1. Introduction
gold catalysts. However, the few available data show that
Au/TiO2 and Au/CoOx/TiO2 are promising catalysts for
CO/N2O reaction [21]. That reaction has been studied in de-
tail on Rh catalysts [11]. N2O reduction by CO on Rh/Al2O3
has been studied by McCabe and Wong [22] and the authors
proposed a reaction mechanism, which involves dissociative
adsorption of N2O. They suggested that the observed lower
activity for the CO/N2O reaction compared to the CO/O2 or
CO/NO reactions is caused by slower dissociative adsorp-
tion of N2O relative to O2 and NO. N2O dissociation might
be the rate-determining step, as was also concluded by some
other authors [23].
In the present paper the main results described concern
N2O reduction by hydrogen on several gold-based cata-
lysts (Au/Al2O3, Au/TiO2, Au/MIOx/MIIOx/Al2O3, and
Au/MOx/Al2O3, with MI, MII, M = Li, Rb, Mg, Co, Mn,
Ce, La, Ti). Additional information was achieved by compar-
ing the behavior of Au/Al2O3 and Au/TiO2 catalysts in the
H2/O2, N2O/H2, and N2O/CO reactions. The main purpose
was to study whether the same factors that are crucial for CO
oxidation also hold for N2O reduction (e.g., gold particle
size and oxidic additives). For CO oxidation on multicom-
ponent gold catalysts, Au/MIOx/MIIOx/Al2O3, it has been
suggested that MIOx (MI, transition metal) plays an impor-
The great interest in gold catalysis in the last years [1–3]
is due to the recently found high catalytic activity of gold-
based catalysts for a number of important reactions such
as CO oxidation [2,4,5], WGS reaction [6,7], and hydro-
genation of unsaturated hydrocarbons [8,9]. Various factors
can influence the catalytic activity of gold-supported cata-
lysts and the most important ones are: (a) the gold particle
size which must be in the nanometer range [3,10], and (b)
the nature of the support or additives. Many studies demon-
strated that gold becomes very active if it is supported on, or
promoted by, transition-metal oxides [3,11–15]. In view of
the more stringent legislation regarding NOx emission from
vehicles, gold-based catalysts were also tested for NO re-
duction by propene in the presence of oxygen [16], by H2
[17–19], and by CO [20]. Nitrous oxide is a principal prod-
uct of NO reduction and is regarded as an undesired and
harmful gas in automotive exhaust gases. There are not many
papers dealing with N2O decomposition and reduction on
*
Corresponding author.
E-mail address: Nieuwe_B@chem.leidenuniv.nl
(B.E. Nieuwenhuys).
0021-9517/$ – see front matter
2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00185-4

198
A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
tant role in O2 activation, via a Mars and Van Krevelen-type
mechanism and MIIOx (MII-alkaline earth metal) stabilizes
the gold particle size against sintering [3,24]. A beneficial
effect of MOx addition to Au/Al2O3 was also observed for
CH4 oxidation [25]. More recently, our studies were focused
on the effect of oxides added to gold-based catalysts on to-
tal oxidation of propene [26]. It turned out that the addition
of MIOx/MIIOx to Au/Al2O3 results in a drastic increase in
catalytic activity of gold catalysts.
The results for the N2O/H2 reaction presented here show
that the catalytic activity of Au/Al2O3 increases dramati-
cally by the addition of MOx or MIOx/MIIOx (MI and MII,
transition metals or alkali metals). The most active catalysts
are Au/Rb2O/CeOx/Al2O3 and Au/Li2O/CeOx/Al2O3.
BET surface areas of the catalysts were measured by N2
physisorption at −196 ◦C using an automatic Qsurf M1 an-
alyzer. Before each measurement the catalyst was heated in
helium at 200 ◦C for 2 h in order to remove impurities ad-
sorbed on the surface.
To obtain information about the identity of the metal/oxi-
de phases present on the catalyst, several techniques have
been used, including XRD, UV–vis, TEM, and EDX. X-ray
diffraction (XRD) was performed using a Philips Goniome-
ter PW 1050/25 diffractometer equipped with a PW Cu
2103/00 X-ray tube operating at 50 kV and 40 mA. The av-
erage gold particle size was estimated from XRD line broad-
ening by using the Scherrer equation [28]. The spectra were
recorded between 2θ = 20◦and 2θ = 60◦.
Diffuse-reflectance UV–vis spectroscopy was conducted
with a commercial unit (Perkin-Elmer Spectrometer, Lambda
900) over air-exposed samples. The scan was made between
200 and 2000 nm with a band pass of 1 nm.
2. Experimental
Transmission electron microscopy (TEM) was used to ex-
amine in more detail the particle size distribution for some
of the samples tested. TEM experiments were performed us-
ing a Philips CM30UT microscope equipped with a LaB6
filament as the source of electrons operated at 300 kV. The
samples were mounted on a micro-grid carbon polymer sup-
ported on a copper grid by placing a few droplets of a sus-
pension of ground catalyst in ethanol on the grid, followed
by drying under ambient conditions. Maximum resolution
was 0.5 nm at 500k magnification. Elemental analysis was
performed using energy dispersive X-ray analysis (EDX).
2.1. Catalyst preparation
γ -Al2O3 (Engelhard Al-4172 P, SBET = 275m2 g−1, V p
ca. 2.8 ml g−1) and TiO2 (Eurotitania 1, SBET = 37 m2 g−1
)
were used as supports. Supported MOx/Al2O3 and MIOx/
MIIOx/Al2O3 (with M, MI, MII as defined above) were
prepared via pore volume impregnation. Dried γ -Al2O3
was placed in a flask under vacuum. Desired amounts of
M(NO3)x · yH2O or MI(NO3)x · yH2O and MII(NO3)n ·
mH2O were dissolved in the corresponding quantity of dem-
ineralized water and were added at room temperature to the
support. The precursor was further dried in air at 80 ◦C for
at least 16 h. Subsequently, the sample was calcined in O2
flow at 350 ◦C for 2 h.
2.3. Activity measurements
The reduction and oxidation reactions were carried out
in a lab-scale fixed-bed reactor in which typically 200 mg
catalyst was used. Prior to measurements, the catalysts were
either reduced in 4 vol% H2 or oxidized in 4 vol% O2, both
in He balance, at 300 ◦C, and kept at that temperature for
30 min. The feed gases were controlled by mass flow con-
Au was added to Al2O3, MOx/Al2O3, MIOx/MIIOx/
Al2O3, or TiO2 via homogeneous deposition precipitation
(HDP) using urea as precipitating agent [27]. An aqueous
solution of HAuCl4 · 3H2O (99.999%, Aldrich) was used
as gold precursor. The suspension was heated at ca. 70 ◦C
under vigorous stirring in order to decompose the urea. In
this way the pH of the solution gradually increased up to 8.
The suspension was then cooled, filtered, and washed several
times with demineralized water in order to remove chlo-
rine ions. Finally, the samples were dried overnight in air
at 80 ◦C and, subsequently, calcined in O2 flow at 300 ◦C
for 2 h. The supported gold catalysts had a theoretical Au
loading of 5 wt% and an Au:M atomic ratio of 1:5. The
intended Au:MI:MII atomic ratio was 1:5:5 in the case of
Au/MIOx/MIIOx/Al2O3.
trollers (Bronkhorst) and set to a total flow of 30 ml min−1
.
All the gases were 4 vol% in He and different ratios were
used: for N2O/CO = 0.5 and 2, for N2O/H2 = 0.2 and
0.5, for CO/O2 = 2 and 1, and for O2/H2 = 1. The reac-
tion was started after stabilization at room temperature for at
least 30 min. Most of the experiments were performed in a
temperature programmed way: the temperature was slowly
increased or decreased (5 ◦C min−1) between room temper-
ature and 300 ◦C. The temperature ramp of 5 ◦C min−1 was
considered to be sufficiently slow to reach at every point
a pseudo-steady state. Each experiment consisted of two
heating cycles and two cooling cycles, in order to moni-
tor possible hysteresis and catalyst deactivation processes.
The effluent stream of the N2O/H2 reaction was analyzed
by a quadrupole mass spectrometer (Balzers) by monitoring
every minute the m/e values of 44 (N2O), 32 (O2), 30 (NO),
28 (N2), 18 (H2O), and 17 (NH3). Fragments of N2O and
H2 interfere with the signals of NO, N2, and NH3 and the
2.2. Catalyst characterization
The gold loading was verified by atomic absorption
spectroscopy (AAS) using a Perkin-Elmer 3100 with an
air/acetylene flame. For that purpose, the catalysts were
dissolved in aqua regia and the solution was diluted with
demineralized water before the analysis was performed.

A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
199
Table 1
Table 2
Fragmentation factors applied for N O/H reaction
Characterization of gold-based catalysts
2
2
a
b
d
Gas
Fragment
NO
Interfering with
NO
F
Sample
Au loading
(wt%)
S
d
BET
2
Au
Au
(m /g)
(nm)
(nm)
N O
2
0.68
0.27
0.27
N O
N
N
Al O
2
–
275
260
222
224
265
218
278
262
207
247
279
275
38
–
–
2
2
2
3
H O
2
OH
NH
3
Au/Al O
4.7
4.0
4.0
4.5
4.5
4.0
4.6
4.0
4.6
4.2
4.5
4.9
4.7
4.3 0.1
1.4 0.3 14.5 0.5
4.0 02 7.1 0.6
1.1 0.1 8.0 0.6
2.4 0.4 9.9 0.4
3.3 0.1 4.0 0.3
1.9 0.5 7.2 0.6
3.8 0.1 6.1 0.8
5.9 07
2
3
Au/MnO /Al O
Au/MgO/Al O
Au/MnO /MgO/Al O
Au/CeO /Al O
x
2
3
F (fragmentation factor) = relative signal ratio to original gas.
2
3
x
2
3
x
fragmentation factors listed in Table 1 were used for cor-
rections. These factors were calibrated by using pure gas
N2O in helium and for the water signal its signal intensity
measured at a conversion of 100% for the N2O/H2 reaction
over a Pt/Al2O3 catalyst was used. In addition to these cor-
rections, subtraction of the background signal was applied.
The N2O/CO, H2/O2, and CO/O2 reactions were moni-
tored by using a gas chromatograph (GC-2002, Chrompack)
equipped with a Molsieve 5 Å column, capable of detecting
CO, NO, H2, O2, and N2 and a Haysep A column, capa-
ble of detecting CO2. The effluent flow was analyzed every
minute.
2
3
Au/Li O/Al O
2
2 3
Au/Li O/CeO /Al O
x
2
2
3
Au/CoO /Al O
Au/CoO /Li O/Al O
2 3
x
2
3
2.0 0.2
3
0.2
x
2
Au/Rb O/Al O
7.0 0.5 11.5 0.4
1.3 0.3 7.4 0.3
3.9 0.2 6.5 0.1
1.8 0.1 5.1 0.5
2
2 3
Au/Rb O/CeO /Al O
x
2
2
3
Au/TiO
Au/TiO /Al O
2
243
x
2
3
a
Au
Average size of gold particles for fresh sample d
and for spent sam-
b
ple d (XRD measurements).
Au
3. Results and discussion
3.1. Catalyst characterization
AAS measurements show that the catalyst preparation
method resulted in a gold loading close to the target load-
ing of 5 wt% (see Table 2). In the same table are also quoted
the surface areas of the samples determined by BET. For
most of the catalysts these values did not differ significantly
from those of the bare supports (275 m2 g−1 for γ -Al2O3
and 37 m2 g−1 for Eurotitania).
To identify crystalline phases present in the catalysts,
XRD analysis was performed. All the samples showed
the diffraction peaks that correspond to metallic gold. For
several catalysts the diffraction lines corresponding to the
metal oxide additives were found. However, for some sam-
ples the formal oxidation state of these oxides could not
be identified unambiguously. Presumably, these oxides are
present as mixture of oxides. From XRD line broadening
the average Au particle size was determined (see Table 2).
Since the peak width at half-maximum intensity was used
to determine the average particle size, smaller crystallites
were not taken into consideration. This is the limitation
of XRD to estimate the average particle size. In addition,
the average crystallite size does not provide any infor-
mation about the particle size distribution. Table 2 shows
the beneficial role of Li oxide on the gold particle size
as is demonstrated by comparing the values of dAa u and
dAbu of Au/Al2O3, Au/CeOx/Al2O3, and Au/CoOx/Al2O3
with those of Au/Li2O/Al2O3, Au/Li2O/CeOx/Al2O3,
and Au/CoOx/Li2O/Al2O3. Similar beneficial effects were
reported by Grisel et al. [27] and Bethke et al. [29] for alkali-
earth metal oxides.
Fig. 1. HRTEM image of Au/CeO /Al O .
x
2
3
HRTEM was used to investigate the particle-size distrib-
ution. A typical TEM micrograph of the Au/CeOx/Al2O3
catalyst is presented in Fig. 1. The gold particles are uni-
formly distributed on the support, a general characteristic
of all the samples studied by HRTEM. Hence, the prepa-
ration method used proved to be a suitable one to obtain

200
A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
Fig. 2. Size distribution of Au particles supported on CeO /Al O (black) and Li O/CeO /Al O (gray).
x
x
2
3
2
2 3
Fig. 4. N O reduction versus temperature during N O–H reaction over
2
2
2
(ꢀ) prereduced Au/TiO , (ꢁ) prereduced Au/Al O , (♦) preoxidized
2
2 3
Au/TiO , and (ꢂ) preoxidized Au/Al O . N O/H = 0.2.
2
2
3
2
2
Fig. 3. UV–vis spectra for fresh and spent catalysts (N O/H reaction, re-
actant ratio 0.5).
2
2
3.2. Comparative studies of N2O reduction by hydrogen or
CO and H2 and CO oxidation by O2 over Au/Al2O3 and
Au/TiO2 catalysts
gold particles uniformly distributed on the oxide support.
The mean gold particle size for Au/CeOx/Al2O3 was es-
timated to be ∼ 4 nm (HRTEM). The distribution of the par-
ticle size for Au/CeOx/Al2O3 and Au/Li2O/CeOx/Al2O3,
as determined by HRTEM, is shown in Fig. 2. There is a
good agreement between the results of XRD and HRTEM
measurements.
UV–vis spectra for some selected catalysts are presented
in Fig. 3. All the spectra shown were obtained after subtrac-
tion of the support contribution. The most prominent feature
of these samples is an absorbance band at around 550 nm,
which is typically of the plasmon resonance peak of metal-
lic gold [30]. There is no indication that gold is present in
an ionic form. The spent catalysts show similar absorbance
spectra as the fresh catalysts.
Decomposition of N2O in the absence of a reducing gas
was not observed.
The presentation of the results of N2O reduction with
H2 will be divided into: (i) N2O reduction with H2 on
Au/Al2O3 and Au/TiO2 measured following either reduc-
tive or oxidative pretreatment using a reactant ratio of 0.2
and (ii) N2O reduction with hydrogen on Au/MIOx/MIIOx/
Al2O3 and Au/MOx/Al2O3 (MI, MII, M = Li, Rb, Mg, Co,
Mn, Ce, La, Ti) by using a reactant ratio of 0.5. The latter
results will be described in Section 3.3.
The conversion of nitrous oxide over Au/TiO2 and
Au/Al2O3 with different pretreatments is depicted in Fig. 4
(N2O/H2 = 0.2). The figure shows the conversion versus

A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
201
Fig. 7. CO oxidation versus temperature during CO–O reaction over
2
Fig. 5. CO oxidation versus temperature during N O–CO reaction over
2
(ꢀ) prereduced Au/TiO , (ꢁ) prereduced Au/Al O , (♦) preoxidized
2
2 3
(ꢀ) prereduced Au/TiO , (ꢁ) prereduced Au/Al O , (♦) preoxidized
2
2 3
Au/TiO , and (ꢂ) preoxidized Au/Al O . CO/O = 2.
2
2
3
2
Au/TiO , and (ꢂ) preoxidized Au/Al O . N O/CO = 0.5.
2
2
3
2
Fig. 6. CO oxidation versus temperature during N O–CO reaction over
2
(ꢀ) prereduced Au/TiO , (ꢁ) prereduced Au/Al O , (♦) preoxidized
Fig. 8. CO oxidation versus temperature during CO–O reaction over
2
2
2 3
Au/TiO , and (ꢂ) preoxidized Au/Al O . N O/CO = 2.
(ꢀ) prereduced Au/TiO , (ꢁ) prereduced Au/Al O , (♦) preoxidized
2
2
3
2
2
2 3
Au/TiO , and (ꢂ) preoxidized Au/Al O . CO/O = 1.
2
2
3
2
temperature as observed during the cooling stage. However,
these results were almost similar to those observed at the
heating stage. All the catalysts exhibited a high catalytic ac-
tivity even at low temperatures. However, differences in be-
havior of the catalysts were found. In the temperature range
between 75 and 175 ◦C the highest conversion is reached
over Au/TiO2-reductive pretreatment. The only detected
products were N2 and H2O. NH3 and NO were not detected.
N2O/CO reaction has been studied over the same cat-
alysts, using two different reactant ratios, with either an
excess of CO (N2O/CO = 0.5) or an excess of N2O
(N2O/CO = 2). The conversion of the samples versus tem-
perature during the cooling stage is presented in Fig. 5 for
the N2O/CO ratio of 0.5 and in Fig. 6 for the ratio 2.
As can be seen from Fig. 5, all the samples are already
active at room temperature and full N2O conversion (as in-
dicated by a CO conversion of 50%) was reached around
100 ◦C. There are no significant differences between ei-
ther preoxidized or prereduced samples, or Au/Al2O3 and
Au/TiO2.
samples were able to fully convert CO at temperatures below
200 ◦C. The sequence of increasing conversion for the inves-
tigated samples was similar to that found for a N2O/CO ratio
of 0.5: preoxidized Au/Al2O3 < prereduced Au/Al2O3 <
prereduced Au/TiO2 < preoxidized Au/TiO2. During re-
action the only detected products were CO2 and N2. NO
formation was not observed.
The two types of catalysts used in this comparative study
were quite active for CO oxidation by O2, as shown in Figs. 7
and 8 [31].
Fig. 9 presents the results of the activity experiments
for H2 oxidation with O2 carried out on the same set of
Au/Al2O3 and Au/TiO2 catalysts. The figure shows the hy-
drogen conversion during the cooling stage; the oxygen con-
version follows a similar pattern. The applied H2/O2 ratio
was 1 (excess oxygen). All the samples are already active
at room temperature, but an interesting feature should be
noted: whereas there is no significant difference in catalytic
performance of Au/TiO2 and Au/Al2O3 following reduc-
tive pretreatment in the whole temperature range studied, the
difference between the samples after oxidative pretreatment
is significant. An oxidative pretreatment applied to Au/TiO2
causes an important decrease in its capability to convert H2.
When excess N2O was used (N2O/CO = 2), the differ-
ences in activity patterns between Au/TiO2 and Au/Al2O3
and oxidative and reductive pretreatment are larger than for
N2O/CO = 0.5. The results are depicted in Fig. 6. All the

202
A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
also has been suggested that an active ensemble of Au–OH
and metallic Au atoms may play a key role [36 and the refer-
ences therein]. Density functional theory calculations point
to the role of special reaction geometries available at nano-
sized gold particles in combination with an enhanced ability
of low coordinated gold atoms to interact with molecules
from the surroundings. O2 dissociation appears to be ex-
tremely facile on these gold nanoparticles [37]. However,
DFT calculations by Liu et al. suggest that O2 dissocia-
tion on gold-particles cannot occur and that the oxidation
reactions occur directly via O2 [38]. In the opinion of the
present authors, the mechanism on gold-based catalysts may
be similar to that on the platinum group metal (PGM) cata-
lysts [11]. In this light the elementary steps for CO oxidation
are
Fig. 9. H oxidation versus temperature during H –O reaction over
2
2
2
(ꢀ) prereduced Au/TiO , (ꢁ) prereduced Au/Al O , (♦) preoxidized
2
2 3
Au/TiO , and (ꢂ) preoxidized Au/Al O . H /O = 1.
2
2
3
2
2
CO → COads
O2 → 2Oads
,
(1)
(2)
(3)
The general trend is that the TiO2-supported catalysts are
more active than the Al2O3-supported ones for the oxida-
tion reactions of CO: CO+O2 and CO+N2O. Interestingly,
however, the Al2O3-supported preoxidized catalysts exhibit
a better performance than the TiO2-supported preoxidized
catalysts for the H2 +O2 reaction. For the H2 +N2O reaction
the differences in behavior of the TiO2 and Al2O3-supported
catalysts are relatively small.
Prereduced Al2O3-supported catalysts are significantly
more active than the preoxidized ones. For the TiO2-sup-
ported catalysts the effect of pretreatment is smaller.
It should be noted that the effect of pretreatment is ob-
served for several subsequent runs. However, after many
runs (> 5) the effect of pretreatment disappears and the per-
formance of the catalyst shows a behavior intermediate of
that of the prereduced and preoxidized catalyst depending
on the reactant ratio used.
,
Oads + COads → CO2,
and for the H2/O2 reaction [41]:
H2 → 2Hads
O2 → 2Oads
,
,
(4)
(5)
(6)
2Hads + Oads → H2O (via OH ads).
In this view, gold nanoparticles have the ability to dissoci-
ate O2 and because of its low heat of adsorption, adsorbed
oxygen will be very reactive. Hydrogen is not dissocia-
tively adsorbed on gold bulk surfaces [39] especially at low
temperatures, but the dissociation might proceed on gold
nanoparticles. Some recent results obtained in our labora-
tory support this proposal. It was found that the rate of the
hydrogen–deuterium exchange reaction,
A summary of T50% (temperature needed for a conversion
of 50%) for all the reactions discussed above is presented in
Table 3.
Although a large number of studies dealing with gold
catalysis have been reported recently, the nature of active
sites and the reaction mechanisms still remain unclear. How-
ever, several mechanisms have been proposed in the litera-
ture, in an attempt to shed more light on catalysis by gold.
Suggestions include a special role of sites at the metal-
support interface [14,32,33], the availability of low coordi-
nated surface atoms [34], and quantum size effects [35]. It
H2 + D2 ꢃ 2HD,
(7)
is much faster over Au/Al2O3 catalysts than over Al2O3 [40].
The transition metal oxides and ceria may provide an addi-
tional route to supply Oads and Hads
Reduction of N2O with CO and hydrogen occurs already
at room temperature, both Au/Al2O3 and Au/TiO2 exhibit-
ing high catalytic activity. Since only N2 is formed as a
.
Table 3
Comparison of the performance of Au/Al O and Au/TiO catalysts for the four reactions described in this paper
2
3
2
◦
◦
( C) Au/TiO
50% 2
Reaction
Ratio
T
( C) Au/Al O
T
50%
2
3
Preoxidized
Prereduced
Preoxidized
Prereduced
CO + O
2
1
1
2
0.5
5
2
100
60
40
<50
66
100
110
60
50
40
<50
<50
80
<20
<20
60
<50
<50
80
<20
<20
70
<50
<50
100
2
H + O
2
2
CO + N O
2
H + N O
2
2
80

A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
203
nitrogen-containing product, the reduction of N2O with CO
and hydrogen may proceed via a Langmuir–Hinshelwood
mechanism, between Oads formed as a result of N2O dis-
sociation and COads or Hads on gold:
N2O → N2Oads
,
(8)
(9)
N2Oads → N2 + Oads
,
CO → COads
;
(10)
or
H2 → 2Hads
,
(11)
(12)
Fig. 10. N O reduction versus temperature during N O–H reaction
2
2
2
over (ꢀ) Au/Al O , (ꢁ) Au/CeO /Al O , (ꢄ) Au/Li O/Al O ,
Oads + COads → CO2;
or
x
2
3
2
3
2
2 3
(ꢀ) Au/Li O/CeO /Al O , (ꢂ) Au/CoO /Al O , and (") Au/CoO /
x
x
x
2
2
3
2 3
Li O/Al O . N O/H = 0.5.
2
2
3
2
2
2Hads + Oads → H2O.
(13)
Table 4
Additional experiments were carried out on Au/Al2O3 and
Au/TiO2 in order to detect possible N2O decomposition in
absence of CO or H2. Without CO or hydrogen in the system,
decomposition of N2O was not observed. Possible reasons
may include a buildup of oxygen atoms on the surface that
inhibit the reaction [41,42], or the absence of a promoting
effect of CO and H2 on the dissociation of N2O. The Au–
N2O bond is very weak and desorption proceeds rapidly.
Catalytic activity of gold-supported catalysts for N O/H reaction (reactant
2
2
ratio N O/H = 0.5)
2
2
◦
◦
( C)
95%
Catalyst
T
( C)
T
50%
Au/Al O
111
265
2
3
Au/Li O/CeO /Al O
52
75
50
160
56
64
57
82
69
103
140
110
179
>265
122
145
115
265
205
118
205
160
145
225
265
206
–
x
2
2
3
Au/Li O/Al O
2
2 3
Au/Rb O/CeO /Al O
x
2
2
3
Au/Rb O/Al O
2
2
3
3
Au/CeO /Al O
x
2
Au/CoO /Li O/Al O
x
2
2
3
3.3. N2O reduction with H2 over multicomponent catalysts
Au/CoO /Al O
x
2
3
Au/TiO /Al O
2
2 3
Au/MnO /MgO/Al O
x
2
3
For the applied ratio N2O/H2 0.5 several gold-based cat-
alysts were tested and the effect of different additives was
studied. The catalytic activities of some selected catalysts
are shown in Fig. 10, where the conversion observed at the
first cooling branch is depicted. During the other stages the
conversion patterns were similar. Difference in T50% during
heating and cooling stages did not exceed 30 ◦C.
Au/MgO/Al O
2
3
Au/MnO /Al O
x
2
3
Au/La O /Al O
2
3
2
3
Au/MoO /Al O
x
2
3
Li O/CeO /Al O
2 3
–
x
2
(at low temperature), N2 (at intermediate temperature), and
NH3 (at high temperature) [31,43]. It was already pointed
out before that N2O reduction with hydrogen produces N2
as the only nitrogen-containing product. The N–N bond-
ing (474 kJ mol−1) in the N2O molecule is much stronger
than that of N–O (161 kJ mol−1). Therefore, if a N2O mole-
cule dissociates, a reasonable assumption is that it will break
through its N–O bond (14), rather than dissociation to NOads
and Nads (15) [41]:
Fig. 10 shows the performance of various gold-based cat-
alysts, viz. Au/Al2O3, Au/CeOx/Al2O3, Au/CoOx/Al2O3,
Au/Li2O/Al2O3, Au/Li2O/CeOx/Al2O3, and Au/CoOx/
Li2O/Al2O3, tested in a N2O/H2 = 0.5 reaction. Clearly,
a large synergistic effect is observed for the multicomponent
catalysts. In the temperature range studied (up to 265 ◦C)
the supports did not reveal N2O conversion higher than 5%.
Hence, the measured catalytic activity is due to the presence
of gold and not because of the intrinsic activity of the sup-
port itself.
N2O → N2 + Oads
,
(14)
All the data regarding T50% and T95% conversion for the
various catalysts are summarized in Table 4.
N2O → NOads + Nads
.
(15)
N2O is a thermodynamically unstable molecule. How-
ever, the thermal decomposition reaction under homoge-
neous conditions does occur only above 600 ◦C. Over gold-
based catalysts, experiments of N2O in the absence of any
reducing gas in the system did not reveal any decomposi-
tion products. Supported gold catalysts convert nitric oxide
in the presence of hydrogen to the following products: N2O
The adsorbed oxygen then reacts with hydrogen to form
water. The experimental results obtained for the H2/O2 re-
action support this mechanism. Since decomposition of N2O
in the absence of H2 did not reveal any product formation, it
is proposed that the presence of hydrogen enhances the de-
composition of nitrous oxide.

204
A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
The mechanism of the beneficial effects of the additives
mance in reduction of nitrous oxide with hydrogen or car-
bon monoxide and in carbon monoxide oxidation, but is not
so important for hydrogen oxidation. The active catalysts
mainly contain gold in the metallic form.
During N2O reduction with H2 or CO the only prod-
ucts obtained are N2 and H2O or CO2, respectively. It is
suggested that the role of the partly reducible metal oxide
additive is to contribute to the formation of new active sites
and increase N2O dissociation. On the other hand, alkali
or alkali-earth metal oxides stabilize gold particles against
sintering. An increasing basicity results in an increased pro-
moting effect.
(TMO, ceria or alkali metal oxides) is still under debate.
Many researchers agree that the gold–TMO interface may
play a significant role [10,33,39,44]. The present results
prove that reduction of N2O to N2 and Oads is enhanced
in the presence of a TMO or ceria, or, more general, ox-
ides with a high capability to store and release oxygen (see
Fig. 10). The presence of CeOx or a TMO might create new
sites for N2O dissociation at the gold/ceria or TMO inter-
face. A striking result is that by addition of both an alkali
metal oxide (Li2O or Rb2O) and CeOx or a TMO, a dra-
matic increase in the catalyst performance is obtained. This
can be explained by taking into consideration the already
discussed effect of ceria on catalytic activity, in combination
with the beneficial effect of an (earth) alkali metal oxide on
the performance of gold-catalysts. The basicity of the alkali
metal oxide increases from Li2O to Rb2O. To the best of
our knowledge, this is the first time that such a large syner-
gistic effect was found for gold catalysts by adding both an
alkali metal oxide and a transition metal oxide. The effect
of alkali or alkali–earth metal oxide on Pt/Al2O3 was ex-
tensively studied by Yentekakis and co-workers [45–47] for
NO reduction by propene in the presence of oxygen. In case
of Pt/Al2O3, the promoting effect of Rb2O is the biggest
one. The effect of basicity on the catalytic activity was also
observed in a hydrogen/water isotopic exchange reaction on
nickel-based catalysts [48]. For gold-supported catalysts, the
combined actions of Rb2O (Li2O) and CeOx are very effi-
cient: sintering of gold particles is prevented by the presence
of the promoter Rb2O (Li2O) and the intrinsic catalytic ac-
tivity is enhanced by the presence of the cocatalyst CeOx.
The lower catalytic activity of Au/Rb2O/Al2O3 compared
with Au/Li2O/Al2O3 can be explained by the difference
in mean diameter of gold particles (as shown by XRD and
HRTEM) of these two samples.
A significant synergistic effect is also found for multi-
component catalysts containing both an alkali metal oxide
and a TMO, in this case CoOx (Fig. 10). However, the en-
hancement in activity is not as big as in the case of ceria and
an alkali metal oxide.
Table 3 also presents another set of three catalysts, viz.
Au/MnOx/Al2O3, Au/MgO/Al2O3, and Au/MnOx/MgO/
Al2O3. As expected, a synergistic effect is again observed.
The slightly lower conversion of N2O on this catalyst com-
pared with the multicomponent Au/CeOx/Rb2O/Al2O3 can
be twofold: MgO’s basicity and the redox cycle which
involves MnOx. However, Au/MnOx/MgO/Al2O3 was
found to be very efficient also in total CO oxidation and
selective oxidation of CO in hydrogen atmosphere [3].
Acknowledgments
The authors thank Dr. P.J. Kooyman of the National Cen-
tre for High Resolution Electron Microscopy, Delft Univer-
sity of Technology, Delft, The Netherlands, for performing
the HRTEM measurements. This work has been performed
under the auspices of NIOK, the Netherlands Institute for
Catalysis Research.
References
[1] M. Haruta, CATTECH 6 (2002) 102.
[2] G.C. Bond, D. Thompson, Gold Bull. 33 (2000) 41.
[3] R.J.H. Grisel, K.J. Weststrate, A.C. Gluhoi, B.E. Nieuwenhuys, Gold
Bull. 35 (2002) 39.
[4] M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, J. Catal. 115 (1989)
301.
[5] M. Haruta, T. Kobayashi, H. Sano, N. Yamada, Chem. Lett. 405
(1987).
[6] H. Sakurai, A. Ueda, T. Kobayashi, M. Haruta, J. Chem. Soc., Chem.
Commun. (1997) 271.
[7] D. Andreeva, V. Idakiev, T. Tabakova, A. Andreev, J. Catal. 158 (1996)
354.
[8] P. Claus, A. Brückner, C. Mohr, M. Hofmeister, J. Am. Chem. Soc. 122
(2000) 11430.
[9] C. Mohr, M. Hofmeister, M. Lucas, P. Claus, Chem. Eng. Technol. 23
(2000) 4.
[10] M. Haruta, Catal. Surv. Jpn. 1 (1997) 61.
[11] B.E. Nieuwenhuys, Adv. Catal. 44 (1999), and references therein.
[12] M. Haruta, S. Tsubota, T. Kobayashi, H. Kageyama, M. Genet, B. Del-
mon, J. Catal. 144 (1993) 175.
[13] M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, J. Catal. 115 (1989)
301.
[14] S.D. Lin, M. Bollinger, M. Vannice, Catal. Lett. 17 (1993) 245.
[15] G. Srinivas, J. Wright, C. Bai, R. Cook, Stud. Surf. Sci. Catal. 101
(1996) 427.
[16] A. Ueda, T. Oshima, M. Haruta, Appl. Catal. B 12 (1997) 81.
[17] T. Salama, R. Osnishi, T. Shido, M. Ichikawa, J. Catal. 162 (1996) 169.
[18] T. Salama, R. Ohnishi, M. Ichikawa, Chem. Commun. (1997) 105.
[19] M.A.P. Dekkers, M.J. Lippits, B.E. Nieuwenhuys, Catal. Today 54
(1999) 381.
[20] T. Salama, R. Ohnishi, M. Ichikawa, Faraday Trans. 92 (1996) 301.
[21] N.W. Cant, N. Ossipoff, Catal. Today 36 (1997) 125.
[22] R.W. McCabe, C. Wong, J. Catal. 121 (1990) 422.
[23] D.N. Belton, S.J. Schmeig, J. Catal. 144 (1993) 9.
[24] R.J.H. Grisel, B.E. Nieuwenhuys, J. Catal. 199 (2001) 48.
4. Conclusions
Gold-based catalysts are active at low temperature for
N2O/CO, H2/O2, CO/O2, and N2O/H2 reactions. The na-
ture of the additives is crucial to obtain a good perfor-

A.C. Gluhoi et al. / Journal of Catalysis 219 (2003) 197–205
205
[25] R.J.H. Grisel, J.J. Slyconish, B.E. Nieuwenhuys, Top. Catal. 1–4
(2001) 425.
[26] A.C. Gluhoi, B.E. Nieuwenhuys, N. Bogdanchikova, to be published.
[27] R.J.H. Grisel, P.J. Kooyman, B.E. Nieuwenhuys, J. Catal. 191 (2000)
430.
[28] P. Scherrer, Nachr. K. Ges. Wiss., Göttingen (1918) 98.
[29] G.K. Bethke, H.H. Kung, Appl. Catal. A 194 (2000) 43.
[30] Y.S. Su, M.Y. Lee, S.D. Lin, Catal. Lett. 57 (1999) 49.
[31] M.A.P. Dekkers, PhD thesis, Leiden University, 2000.
[32] M. Haruta, Catal. Today 29 (1997) 153.
[33] R.J.H. Grisel, B.E. Nieuwenhuys, Catal. Today 64 (2001) 69.
[34] F. Boccuzzi, G. Cerrato, F. Pinna, G. Strukul, J. Phys. Chem. B 102
(1998) 5733.
[38] Z.-P. Liu, P. Hu, A. Alavi, J. Am. Chem. Soc. 124 (2002) 14770.
[39] B.E. Nieuwenhuys, in: R.W. Joyner, R.A. van Santen (Eds.), Elemen-
tary Reaction Steps in Heterogeneous Catalysis, Kluwer Academic,
Dordrecht, 1993, p. 155.
[40] H.S. Vreeburg, J.W. Bakker, B.E. Nieuwenhuys, to be published.
[41] S.A. Carabineiro, B.E. Nieuwenhuys, Surf. Sci. 495 (2001) 1.
[42] J. Ma, N.M. Rodriguez, M.A. Vannice, R.T.K. Baker, Top. Catal. 10
(2000) 27.
[43] A.C. Gluhoi, B.E. Nieuwenhuys, to be published.
[44] R.J.H. Grisel, C.J. Weststrate, A. Goosens, M.W.J. Craje, A.M. van
der Kraan, B.E. Nieuwenhuys, Catal. Today 72 (2002) 223.
[45] I.V. Yentekakis, M. Konsolakis, R.M. Lambert, N. Macleod, L. Nal-
bantian, Appl. Catal. B 22 (1999) 123.
[35] M. Valden, X. Lai, D.W. Goodman, Science 281 (1998) 1647.
[36] C.K. Cosello, J.H. Yang, H.-Y. Law, Y. Wang, J.-N. Lin, L.D. Marks,
M.C. Kung, H.H. Kung, Appl. Catal. A 243 (2003) 15.
[37] N. Lopez, J.K. Norskov, J. Am. Chem. Soc. 124 (2002) 11262.
[46] M. Konsolakis, I.V. Yentekakis, J. Catal. 198 (2001) 142.
[47] M. Konsolakis, I.V. Yentekakis, Appl. Catal. B 29 (2001) 103.
[48] A.C. Gluhoi, P. Marginean, D. Lupu, E. Indrea, A.R. Biris, Appl. Catal.
A 232 (2002) 121.