Alumina-supported Cu-Ag catalysts for ammonia oxidation to nitrogen at low temperature

Article abstract of DOI:10.1006/jcat.2001.3470

Selective catalytic oxidation (SCO) of ammonia to nitrogen in water could be a solution to pollution caused by different ammonia emission sources such as the SCR process. Alumina-supported Cu-Ag catalysts prepared and screened for their ability to catalyze SCR and SCO reactions were characterized using transmission electron microscopy, energy-dispersive X-ray spectroscopy, low-energy ion-scattering spectroscopy, XPS, and AES to determine their physicochemical compositions and these were compared with catalytic performance. The addition of copper to alumina-supported Cu-Ag catalysts, by co-incipient wetness impregnation, increased the selectivity to nitrogen during the catalytic oxidation of ammonia at 250°C without significant decrease in activity relative to that of Ag/alumina alone. Selectivity increased from 80 to ～ 95% at 100% ammonia conversion at the optimum Ag/Cu weight ratio (between 1:1 and 3:1). No increase in activity was observed for mechanical mixtures of Cu/alumina and Ag/alumina catalysts. The promotional effect of copper could be explained by a bifunctional mechanism in which the Ag component mainly catalyzes ammonia oxidation to NO, the first step of this reaction, and the copper catalyzes the SCO of NO to nitrogen, thus reducing N₂O formation on silver.

Full text of DOI:10.1006/jcat.2001.3470

Journal of Catalysis 206, 60–70 (2002)
doi:10.1006/jcat.2001.3470, available online at http://www.idealibrary.com on
Alumina-Supported Cu–Ag Catalysts for Ammonia Oxidation
to Nitrogen at Low Temperature
∗
∗
∗
∗
Lu Gang,∗,1 B. G. Anderson, J. van Grondelle, R. A. van Santen, W. J. H. van Gennip,
∗
J. W. Niemantsverdriet, P. J. Kooyman,† A. Knoester,‡ and H. H. Brongersma‡
∗
Schuit Institute of Catalysis, Laboratory of Inorganic Chemistry and Catalysis, Eindhoven University of Technology, P.O. Box 513, 5600 MB,
Eindhoven, The Netherlands; †National Centre for High Resolution Electron Microscopy, Delft University of Technology, Rotterdamsweg 137, 2628 AL
Delft, The Netherlands; and ‡Calipso B. V., Eindhoven University of Technology, P.O. Box 513, 5600 MB, Eindhoven, The Netherlands
Received June 11, 2001; revised November 2, 2001; accepted November 13, 2001
catalytic reduction (SCR) process, soda production, and
agricultural sources (1–12).
The addition of copper to alumina-supported silver catalysts, by
co-incipient wetness impregnation, increased the selectivity to ni-
trogen during the catalytic oxidation of ammonia at 250◦C without
significant decrease in activity relative to that of Ag/alumina alone.
An increase in selectivity from 80 to circa 95% was observed at 100%
ammonia conversion at the optimum Ag/Cu weight ratio (between
1 : 1 and 3 : 1). No increase in activity was observed for mechani-
cal mixtures of Cu/alumina and Ag/alumina catalysts. Based on
transmission electron microscopy (TEM), energy-dispersive X-ray
spectroscopy (EDX), low-energy ion-scattering spectroscopy, X-ray
photoelectron spectroscopy, and Auger analysis, the catalysts con-
sisted of monolayers of copper oxide on alumina upon which silver
particles sat. TEM images and EDX analysis showed that silver
particles, with a very large size distribution, existed on all catalysts.
EDX also revealed the presence of very small (<1 nm) silver dis-
persed in copper everywhere on the alumina surface. Thus intimate
contact between copper and silver on alumina-supported Cu–Ag
catalysts existed. No indication of the formation of CuAl2O4 or
of Cu–Ag phases was observed. The promotional effect of copper
can be explained by a bifunctional mechanism in which the silver
component mainly catalyzes ammonia oxidation to NO, the first
step of this reaction, and the copper catalyzes the selective cata-
lytic reduction of NO to nitrogen, thus reducing N2O formation on
c
Most studies on ammonia oxidation focus on the high-
temperature (>800 K) process that selectively produces
NO because of its industrial importance. However the low-
temperature reaction that mainly produces N2 and N2O
is becoming more important because of the environmen-
tal aspect. Various catalysts of different types have been
tested for the low-temperature ammonia oxidation reac-
tion: biological catalysts (1, 2), metal oxide catalysts (3–6),
ion-exchanged zeolites (7–9), and metallic catalysts (10–
12). When the various types of catalysts are compared it
turns out that the metallic catalysts such as Pt, Ir, and
Ag are the most active but the least selective. Significant
amounts of N2O are produced on these catalysts. The metal
oxide catalysts such as CuO, V2O5, and MoO3 show very
high selectivity to nitrogen, but the reaction temperature
needed is too high to be matched with some industrial
applications.
We previously reported that silver was very active for
ammonia oxidation (13). Alumina-supported silver cata-
lysts were superior even to noble metal catalysts in both
activity and selectivity to nitrogen. However the nitrogen
selectivity of the silver-based catalysts (circa 80%) is not
sufficient, as N2O and NO are still coproduced. As these
compounds are even more toxic than NH3, selectivity to
nitrogen of near 100% is required. Intermediate species
and reaction pathways for ammonia oxidation on an unsup-
ported silver powder catalyst were studied by temperature-
programmed desorption, temperature-programmed reduc-
tion, Fourier transform–Raman, and transient as well as
steady-state ammonia oxidation experiments (14). NO was
found to be the main reaction intermediate that produced
N2O as well as N2. NO could even be formed at room
temperature. Its subsequent oxidation to NOx led to ad-
sorption on the silver surface, blocking the active sites for
oxygen adsorption. The pathway for ammonia oxidation at
low temperature was found to consist of a two-step con-
secutive reaction. Ammonia was first oxidized to NO. This
silver.
Key Words: ammonia; oxidation; silver; copper; supported
catalyst.
ꢀ 2002 Elsevier Science (USA)
1. INTRODUCTION
The increasing problem of air pollution by N-containing
compounds, such as NO, NO2, N2O, and NH3, has led to
more stringent emissions control. The removal of ammonia
from waste streams is becoming an increasingly important
issue. Selective catalytic oxidation (SCO) of ammonia to
nitrogen and water could be a solution to pollution caused
by various ammonia emission sources, such as the selective
1 To whom correspondence should be addressed. Fax: +31 40 245 5054.
E-mail: l.gang@chem.tue.nl.
60
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

ALUMINA-SUPPORTED Cu–Ag CATALYSTS FOR AMMONIA OXIDATION
61
reaction step was very fast on silver. At moderate temper- on a combination of these two analysis techniques (balance
atures (below 300◦C) NO could then be removed either as of 90 10% for all the measurements).
N2O or as N2 through a surface SCR reaction, the second
High-resolution transmission electron microscopy mea-
step of the reaction mechanism. At even higher temper- surements were performed using a Philips CM30/T at the
atures. NO could directly desorb as one of the products. National Centre for HREM at the Delft University of Tech-
Apparently the ability of silver-based catalysts to perform nology. Samples were mounted on a microgrid carbon poly-
the SCR reaction was closely related to its ability to oxidize mer supported on a copper grid, followed by drying at ambi-
ammonia.
ent conditions. EDX analysis was performed using a LINK
As ammonia oxidation to nitrogen consists of two reac- EDX system. This enabled us to obtain information about
tion steps and as it is difficult to find a catalyst that acceler- the relative Ag/Cu ratios in selected areas of the sample.
ates both of these, a composite catalyst system made of two
X-ray powder diffraction (XRD) patterns were recorded
or more catalyst components, each of which catalyzes only on a Rigaku Geigerflex X-ray powder diffractometer us-
a part of the reaction, appeared to be a good alternative for ing Cu Kα radiation. Prior to the experiment, the catalysts
improving the performance of the overall reaction. Hence were ground and pressed into a sample holder containing
a bifunctional catalytic system was sought.
The bifunctional catalyst required must first contain a ground subtraction was not applied.
vaseline. The applied scanning speed was 1◦ per min. Back-
component active for ammonia oxidation at low tempera-
X-ray photoelectron spectroscopy (XPS) experiments
tures. The coproduction of N2O can be avoided, or at least were performed on a VG Ionex system modified with a VG
further reduced, by the second component, which performs Mg/Al Kα X-ray source and a VG Clam II analyzer. The
the SCR of the intermediate NO species. As stated above, powdered catalyst particles were pressed into indium foil.
Ag is capable of performing the first step. Copper oxide To correct for the energy shift caused by charging, all peaks
appeared to be an excellent choice for the second com- were corrected by setting the C 1s peak of adsorbed hydro-
ponent, as it is known to be both active and selective in carbons to a binding energy of 285.0 eV. The peaks were fit-
the selective catalytic reduction of NO by ammonia. Sup- ted using the XPSPEAK version 4.0 peak-fitting program.
ported copper oxide catalysts are commonly reported in the All peaks were fitted by using a linear background.
literature for this purpose (25). In addition, as previously
Low-energy ion-scattering (LEIS)experimentswereper-
mentioned, CuO yields a high selectivity to nitrogen in the formed in co-operation with Calipso b.v. at the Eindhoven
SCO reaction. Alumina-supported Cu–Ag catalysts were University of Technology. The newly developed ERISS sys-
thus prepared and screened for their ability to catalyze both tem was used as the ion-scattering apparatus. In this appa-
the SCO and the SCR reactions. These materials were also ratus ion doses can be reduced to such low levels that the
characterized by several spectroscopic techniques in order damage to the surface is negligible, thus enabling the per-
to determine their physicochemical compositions and these formance of static LEIS (15–20). To prevent charging of
were compared with catalytic performance. The results are insulating materials, a neutralizer was available to spray
discussed in this article.
low-energy electrons onto the sample. Samples were pre-
pared by pressing the ground sample with a load of 1600 kg
into a tantalum cup (φ, 1 cm) and subsequent loading into
the pretreatment chamber that was connected to the main
UHV chamber. All the samples were pretreated at 300◦C in
2. EXPERIMENTAL
Ag/Al2O3, Cu/Al2O3, and AgCu/Al2O3 samples with dif- 214 mbar of oxygen for 20 min. Most LEIS measurements
ferent Ag and Cu loadings were prepared by incipient wet- were performed using 5-keV Ne+ ions; these provided a
ness (co-)impregnation. The precursors for these catalysts good mass resolution for the heavy elements (Ag and Cu).
were AgNO3 and Cu(NO3)2 distributed over the γ -Al2O3 A number of samples were also analyzed using 3-keV4He+
support (Akzo/Ketjen 000 1.5E; surface area, 205 m2/g). ions. This yielded a good sensitivity for the light elements
The samples were calcined in air at 500◦C for 24 h before (C, O, Al, Si). In most cases, the samples were mechanically
testing.
scanned over an area 1×1 mm2 in size. During the measure-
Catalytic activity measurements were carried out in a ments, the pressure in the main chamber was determined
quartz, fixed-bed reactor (4-mm internal diameter). The by the noble gas used in the ion source and was typically
amount of catalyst used was about 0.2 g (250 to 425–µm around 1×10−8 mbar. The base pressure of the system was
particles). Ammonia, oxygen, and helium flow rates were in the 10−10 region. Analysis of the outermost surface layer
controlledbymassflowmeters. NH3, NO, andNO2 werean- and at different depths was performed on both the catalysts
alyzed by a ThIs Analytical model 17C chemiluminescence and on the gamma-alumina carrier. Depth analysis was per-
NH3 analyzer. Other substances such as N2O, N2, and H2O formed by sputtering the samples with Ne+ ions or He+
4
were analyzed by a quadruple mass spectrometer (Baltzers ions, thus removing up to 8 monolayers (ML) with Ne+ and
OmniStar). Thenitrogenmassbalancewascalculatedbased up to 0.8 ML with 4He+.

62
GANG ET AL.
TABLE 1
3. RESULTS
3.1. Catalytic Ammonia Oxidation Tests
SCO (NH3 + O2) Reaction on Different Catalysts
NH3
N2
Figure 1 shows the dependence of both ammonia conver-
sion level and selectivity to nitrogen on the catalyst com-
position at 250◦C. It can be seen that the conversion of
ammonia increases with an increase in copper weight ratio
in the catalyst, but that nitrogen selectivity changes in the
opposite direction. It has previously been shown that selec-
tivity is a very weak function of the ammonia conversion
level (23). The optimum Ag/Cu weight ratio seems to be in
the range between 1 and 3. Both activity and selectivity to
Temperature conversion selectivity
Catalyst
10 wt% Cu/Al2O3
(◦C)
(%)
(%)
250
300
350
12
90
100
97
96
90
10 wt% Ag/Al2O3
200
250
300
11
98
100
88
86
83
Ag/Al2O3 + Cu/Al2O3
200
250
300
8
96
100
89
87
82
nitrogen are high in this range. Apparently, nitrogen selec- (Ag/Al2O3 : Cu/Al2O3 = 3 : 1)
tivity can be greatly improved by addition of small amounts
of copper to the silver catalyst system, with only a small loss
in catalyst activity.
Note. Reaction conditions: NH3 = 1.14%; O2 = 8.21 vol%; flow rate =
74.7 N ml/min; catalyst weight = 0.2 g.
For comparison, a mechanical mixture of 10 wt% Ag/
Al2O3 and 10 wt% Cu/Al2O3 catalyst with a weight ratio
of 3/1 was also tested in the SCO reaction. The results
are listed in Table 1. It can be seen that a mechanically
mixed silver and copper catalyst behaves similarly to the
silver catalyst itself. Neither activity nor selectivity to nitro-
gen was improved by mechanical mixing of two different
catalysts.
Figure 2 shows the conversion of ammonia and the selec-
tivity to various products on an alumina-supported 7.5 wt%
Ag–2.5 wt% Cu catalyst as a function of temperature. Be-
tween 200 and 300◦C the catalyst converted 100% of the
ammonia with a selectivity to nitrogen of circa 95%. Above
300◦C, more NOx is produced. The catalyst was quite sta-
ble. There was no significant deactivation during a one-day
test under these experimental conditions. However, the se-
lectivity did change somewhat with time (see Fig. 3). It
can be seen from Fig. 3 that the selectivity to nitrogen in-
creased with reaction time during the first 40 min on stream,
whereafter it remained unchanged. This is mainly caused by
adsorbed NOx , N2Ox species produced during reaction
which can lower the surface oxygen coverage, as discussed
previously (14).
Selective catalytic reduction (SCR) of NO by ammonia
was also tested on various copper- and silver-based cata-
lysts to demonstrate the relationship between the SCR re-
action and the ammonia oxidation reaction. Table 2 shows
a comparison of the data obtained for the SCR reaction
on different catalysts. It can be seen that the activity for
the SCR reaction on the alumina-supported Cu–Ag cata-
lyst was much higher than that on the alumina-supported
silver catalyst. The SCR reaction was also more selective
to nitrogen on the alumina-supported Cu–Ag catalyst as
compared with alumina-supported silver catalysts.
3.2. XRD and XPS
Figure 4 shows the XRD diffractograms of 5 wt% Cu/
Al2O3, 5 wt% Ag/Al2O3, and 7.5 wt% Ag–2.5 wt% Cu/
FIG. 1. The effect of Ag/Cu weight ratio on ammonia conversion and
on selectivity to nitrogen for Ag–, Cu–, and AgCu/γ –Al2O3 catalysts.
Reaction conditions: NH3, 1.14 vol%; O2, 8.21 vol%; flow rate = 74.7 N
ml/min; catalyst weight = 0.2 g; T = 250◦C.
FIG. 2. Ammonia oxidation on 7.5 wt% Ag–2.5 wt% Cu/Al2O3 cata-
lyst at various temperatures. Reaction conditions: NH3 = 1000 ppm; O2 =
10 vol%; total flow rate = 50 N ml/min; catalyst weight = 0.1 g.

ALUMINA-SUPPORTED Cu–Ag CATALYSTS FOR AMMONIA OXIDATION
63
FIG. 3. Selectivity change with time on 7.5 wt% Ag–2.5 wt% Cu/
Al2O3 catalyst. Reaction conditions: NH3 = 1000 ppm; O2 = 10 vol%; total
flow rate = 50 N ml/min; catalyst weight = 0.1 g.
FIG. 4. XRD diffractograms of alumina-supported Cu–Ag catalysts
Al2O3 catalysts. All the catalysts were calcined at 500◦C in
air for 24 h before measuring. No XRD diffraction lines at-
tributable to crystallized copper metal or other copper com-
pounds were observed on the copper-containing samples.
However, diffraction lines attributable to AgO appeared in
the XRD spectra of both silver-containing samples. Low-
intensity diffraction lines due to metallic silver were also
observed on the 7.5 wt% Ag–2.5 wt% Cu catalyst.
calcined at 500◦C.
in the region 370–330 eV. As shown, the peaks are very
broad, but the kinetic energy of the Auger electrons is about
350 eV for all three samples. By comparison of these re-
sults with the literature (22) it can be concluded that silver
is not in a metallic state, but in a state between Ag+1 and
Ag+2
.
Four samples of Ag and Cu with different loadings of
metal were characterized by XPS. The spectra in Fig. 5a
show the Ag 3d region measured by XPS. Two peaks ob-
served are due to removal of electrons from the 3d3/2 and
3d5/2 levels. The more intense peak due to removal of 3d5/2
core level electrons had a binding energy of 368.3 eV. Un-
fortunately, according to literature data (21), the binding
energies of the 3d electrons in Ag0, Ag+1, or Ag+2 are all
very similar. Hence it is impossible to ascertain the oxi-
dation state of silver from this XPS peak alone. However,
more information can be obtained by measuring the Auger
bands to construct a Wagner plot (26). Figure 6 shows the
Auger spectra obtained on three silver-containing catalysts
Figure 7 shows XPS spectra in the Cu 2p region (925–
960 eV). The most intense peak occurs at about 932.5 eV.
Comparing this result with literature data (21) it seems that
Cu is in an oxidization state of +1 if copper is present as
copper oxide since the 2p electrons in Cu+2 have a much
higher binding energy (933.6 eV). Auger bands in the Cu
region are shown in Fig. 8. The kinetic energies are circa
914 eV, much lower than that of CuO (918.1 eV) and Cu2O
(916.2 eV). The UV-spectra of our γ -alumina-supported
copper catalysts (9) and extended X-ray absorption fine
structure studies on other γ -alumina-supported copper
catalysts (24) clearly showed that copper was in an oxi-
dization state of +2. The shift of the binding energy may be
TABLE 2
SCR (NO + NH3) Reaction on Different Silver-Based Catalysts
Temperature NO conversion N2 selectivity
Catalyst
(◦C)
(%)
(%)
CuAg/Al2O3
(7.5% Ag–2.5% Cu)
150
200
250
27
61
98
98
95
96
10 wt% Cu/Al2O3
10 wt% Ag/Al2O3
150
200
250
53
97
99
98
97
98
200
250
300
5
14
25
67
74
72
Note. Reaction conditions: NO = 500 ppm; NH3 = 1000 ppm; O2 =
8 vol%; flow rate = 100 N ml/min; catalyst weight = 0.2 g.
FIG. 5. Ag 3d XPS spectra of different Cu–Ag/Al2O3 catalysts.

64
GANG ET AL.
FIG. 6. Auger bands in Ag region for different Cu–Ag/Al2O3 cata-
FIG. 8. Auger bands in Cu region for different Cu–Ag/Al2O3 cata-
lysts.
lysts.
Subsequent quenching prevented the formation of micro-
crystallites. The XPS spectra of the Ag 3d and Cu 2p did not
show any difference from the pure metals after polishing.
Any direct interaction between copper and silver, either
through formation of an alloy or at boundaries, is thus very
unlikely.
caused by the interaction of copper with the support. New
copper species other than copper oxide are formed due to
this interaction. A second possibility is an interaction be-
tween Ag and Cu.
In order to be able to exclude any interaction between
silver and copper two solid solutions were made of 5 wt%
copper in silver and 5 wt% silver in copper. A special pro-
cedure was needed to fabricate these solid solutions, as at
low temperatures these metals do not mix at all and will
phase-separate into microcrystallites when cooling down
from the melt. Such rough samples were prepared by melt-
ing the appropriate amounts together in an alumina cru-
cible by induction in a high-frequency furnace, after which
the samples were placed in two evacuated quartz capsules.
These were heated for 96 h at a steady temperature of
800◦C, at which temperature solid solutions of these com-
positions are possible, according to the phase diagram (27).
3.3. LEIS Measurements
The samples measured with LEIS are listed in Table 3.
In addition to the Ag/Cu/γ -alumina catalysts, we analyzed
a bare γ -alumina carrier material and Ag- and Cu-foil ref-
erence samples. Furthermore, in order to analyze the in-
fluence of surface roughness on the results we measured
both a quartz sample and a silica carrier (Grace 332). Al-
though these samples had the same elemental composition,
quartz had a very smooth surface whereas the carrier mate-
rial exhibits a much larger surface roughness (specific area
of 235 m2/g). Finally, an “ultra-pure” CuO powder sample
was analyzed for quantification purposes. Figures 9 and 10
FIG. 9. 5 keV Ne+ spectra of Cu/γ -alumina (5 wt% Cu) at different
FIG. 7. Cu 2p XPS spectra of different Cu–Ag/Al2O3 catalysts.
depths.

ALUMINA-SUPPORTED Cu–Ag CATALYSTS FOR AMMONIA OXIDATION
65
TABLE 3
Samples Analyzed by LEIS and Their Pretreatments
Ag
Cu
Pretreatment
Ne+
4He+
Sample
(wt%)
(wt%)
(in LEIS reaction chamber)
(5 keV)
(3 keV)
Cu/γ –alumina
Ag/γ –alumina
Ag/Cu/γ –alum.
Ag/Cu/γ –alum.
Ag/Cu/γ –alum.
Ag/Cu/γ –alum.
γ –Alumina
Reference Ag
Reference Cu
Quartz
5.0
214 mbar O2, 300◦C, 20 min
213 mbar O2, 300◦C, 20 min
212 mbar O2, 300◦C, 20 min
201 mbar O2, 300◦C, 20 min
208 mbar O2, 300◦C, 20 min
210 mbar O2, 300◦C, 20 min
216 mbar O2, 300◦C, 20 min
Sputter-cleaned
x
x
x
x
x
x
x
x
x
10.0
10.0
2.5
5.0
9.0
2.5
7.5
5.0
1.0
x
x
x
x
x
100
x
x
100
Sputter-cleaned
Silica
Vacuum, 300◦C, 74 min
CuO
x
show the LEIS spectra of 5 wt% Cu/γ -alumina and 10 wt%
Going deeper, down to 3–5 MLs, the respective Cu sig-
Ag/γ -alumina samples as obtained with 5-keV Ne+ ions. nals decreased in a similar fashion for all copper-containing
At the outermost atom layers of the samples significant Ag samples. Concerning Ag, we observed a relatively large-
and Cu signals were detected.
intensity drop for the sample shown in Fig. 11. The 3-keV
The removal of the first (fraction of a) monolayer (ML) 4He+ spectrum recorded for the 10 wt% Ag–2.5 wt% Cu/
results in increased Ag and Cu signals. After a few MLs Al2O3 catalyst also shows the F peak originating from the
the intensities returned to a magnitude on the order of bare carrier (Fig. 12).
that of the outermost surface (Cu/γ -alumina) or slightly
The LEIS signals measured at 1-ML depth, i.e., after re-
lower (Ag/γ -alumina). Most likely, the first ML removed moval of surface contamination, contain information on
consisted of contamination that had been deposited on the the Ag and CuO dispersions. In order to obtain a (semi)
surface during the in situ calcination. The subsequent de- quantitative picture of these dispersions, we have per-
crease in the signals suggests that prolonged sputtering re- formed some calculations (see Table 4). Starting with the
sulted in smaller “accessible” silver and copper surfaces. Cu signals, the results indicate that the CuO species are
Figure 11 shows one of the 5-keV Ne+ spectra recorded for present as thin platelets with a thickness of roughly 1 ML.
the bimetallic catalyst (10 wt% Ag–2.5 wt% Cu/Al2O3). Between the samples, only minor differences are observed.
In all cases, the removal of the first (fraction of a) ML Obviously, at increasing Cu loading, relatively more (or
resulted in increased intensities for both Ag and Cu. Again, more extended) rather than thicker CuO platelets are
we assume that we are dealing with surface contamina- formed. Based on surface-tension considerations, Ag is ex-
tion that, most likely, originated from the calcination of the pected to form distinct particles on the γ -alumina sup-
samples.
port. Assuming (a) that these particles are spherical and
(b) that only their upper halves contribute to the LEIS
FIG. 10. 5-keVNe+ spectraofAg/γ -alumina(10wt%Ag)atdifferent
depths.
FIG. 11. 5-keV Ne+ spectra of Cu/Ag/γ -alumina (2.5 wt% Cu,
10 wt% Ag) at different depths.

66
GANG ET AL.
FIG. 12. 3-keV 4He+ spectra of Cu/Ag/γ -alumina (2.5 wt% Cu,
10 wt% Ag).
signals, we calculate (from the “1-ML signals”) that the
average particle size of Ag ranges between 2.5 and 5.5 nm.
Note that the numbers increase with an increase in Ag
loading.
FIG. 13. TEM images of 10 wt% Ag/Al2O3 catalyst.
3.4. TEM and EDX Measurement
Single-component alumina-supported Cu and Ag cata-
lysts with different loading were examined using transmis-
sion electron microscopy (TEM). No metal oxide parti-
cles were observed on the alumina surface of the 5 wt%
Cu/Al2O3 and 10 wt% Cu/Al2O3 catalysts. Particles with
diameters ranging between 5 and 10 nm were observed on
the 15 wt% Cu/Al2O3 catalyst. The TEM images of the
5 wt% Ag/Al2O3 and 10 wt% Ag/Al2O3 catalyst seem to be
similar. The silver particle size distribution was very broad.
The smallest particles were about 1 nm in diameter; the
largest particles found on the images were about 50 nm in
diameter (see Fig. 13).
Two samples of mixed Cu–Ag/alumina catalysts were ex-
amined using TEM and EDX elemental analysis. Both sam-
ples contained a total metal loading of 10 wt%. One sample
contained 7.5 wt% Ag and 2.5 wt% Cu. The other sample
contained 2.5 wt% Ag and 7.5 wt% Cu.
taining black particles and the other containing green par-
ticles. These two phases were mechanically separated (by
hand) and were analyzed separately. Figure 14 shows a
representative TEM image that was obtained for the frac-
tion containing black particles. EDX analysis was per-
formed at several different locations on the sample (see
numbered arrows). Data points 1–4 were chosen to analyze
areas containing dark particles. Data point 5 measured an
area in which no dark particles were present. As the rela-
tive amounts of copper and silver are of importance these
are shown in Table 5. Clearly the areas that include parti-
cles visible by TEM (points 1–4) contain very high amounts
of silver relative to copper (circa 95:5). By contrast the
area containing no visible particles (point 5, and others not
shown here) contains relatively more copper than silver
(circa 60:40).
Figure 15 shows a TEM image that was obtained for the
macroscopic green particles of this Ag-rich sample. The
7.5 wt% Ag–2.5 wt% Cu/Al2O3. This sample clearly
consisted of two different macroscopic phases: one con-
TABLE 4
Calculation of CuO Platelets and Ag Particles Dispersion from LEIS Measurements
CuO platelets
(depth: 1 ML)
Ag particles
(depth: 1 ML)
Catalysts
Cu
(wt%)
Ag
(wt%)
Ag/Cu
(at/at)
Cu LEIS
signal (cnts/nC)
Cu visible atoms
CuO platelets
(thickness, MLs)
Ag LEIS
signal (cnts/nC)
Ag visible atoms
Ag particles
(av. size, nm)
(atoms/cm2 × 103)
(atoms/cm2 × 103)
1
2.5
5
9
10
5
5.3
2.4
0.6
0.2
231
352
895
5.4
8.2
20.8
28.9
0.8
1.2
1.0
1.0
728
645
510
312
4.1
3.6
2.8
1.7
4.1
5.2
3.2
2.6
7.5
2.5
1242

ALUMINA-SUPPORTED Cu–Ag CATALYSTS FOR AMMONIA OXIDATION
67
TABLE 5
TABLE 6
Relative Amounts of Copper and Silver Measured by EDX on
Relative Amounts of Copper and Silver Measured by EDX on
Different Areas of 7.5 wt% Ag–2.5 wt% Cu/Al2O3 (Black-Particle Different Areas of 7.5 wt% Ag–2.5 wt% Cu/Al2O3 (Green-Particle
Fraction) Fraction)
Data point Data point
% atom Cu
% atom Ag
% atom Cu
% atom Ag
1
2
3
4
6.90
3.82
3.52
5.97
59.70
93.10
96.18
96.48
94.03
40.30
1
2
3
4
5
6
7
1.02
1.10
2.10
1.10
1.68
0.86
73.40
98.98
98.90
97.90
98.90
98.32
99.14
26.60
5
(area without dark particles)
Note. Data points refer to numbered arrows in Fig. 14.
(area without dark particles)
Note. Data points refer to numbered arrows in Fig. 15.
relative amounts of copper and silver (by EDX analysis)
are listed in Table 6. Similar to the black-colored particle
fraction, thisphasealsocontainedawidedistributionofpar-
ticle size (up to 50 nm). These particles (data points 1–6)
were slightly richer in silver relative to copper than were the
black particles. Again although data point 7 shows no par-
ticles visible by TEM, both copper and silver were present,
copper again being the more abundant.
7.5 wt% Cu–2.5 wt% Ag/Al2O3. Figure 16 shows a rep-
resentative TEM image that was obtained for this Cu-rich
sample. Again particles of a large distribution were ob-
served. Table 7 shows the relative amounts of copper and
silver measured at various locations. Again the darker parti-
cles visible by TEM (points 1–7) contain a larger abundance
of silver. However in several locations the relative amount
of copper to silver is higher than was observed in the pre-
vious samples. Data point 8 again shows that copper and a
smaller amount of silver are present even in an area without
visible dark particles.
FIG. 15. TEM image of green particles of 7.5% Ag–2.5% Cu/Al2O3
catalyst (the arrows point to the area where the EDX measurement is
performed).
FIG. 14. TEM image of black particles of 7.5% Ag–2.5% Cu/Al2O3
catalyst (the arrow point to the area where the EDX measurement is
performed).
FIG. 16. TEM image of green particles of 2.5% Ag–7.5% Cu/Al2O3
catalyst (the arrows point to the area where the EDX measurement is
performed).

68
GANG ET AL.
TABLE 7
analysis of Cu/Al2O3 catalysts revealed that no peaks due
to either metallic copper, copper oxide, or other copper
compounds were present up to loadings as high as 10 wt%.
TEM images of these catalysts also showed no visible par-
ticles. EDX analysis showed that copper was present on all
areas of the alumina catalyst.
Based on the absence of distinct Cu particles in the
TEM micrographs, it has been suggested that Cu might
have reacted with (surface layers of) γ -alumina and formed
CuAl2O4. However, the Cu atoms of this “normal spinel”
component are known to occupy tetrahedral, i.e., sub-
surface, positions. Consequently, we would not expect
CuAl2O4 to yield the large copper LEIS signals we actu-
ally have obtained. Former LEIS experiments performed
on normal spinels yielded much smaller signals. Further-
more, small Cu (oxide) species may simply exhibit insuffi-
cient contrast (in TEM) when imaged on γ -alumina. The
XRDresultsalsoshownodiffractionlineofCuAl2O4 phase
Relative Amounts of Copper and Silver Measured by EDX
on Different Areas of 2.5 wt% Ag–7.5 wt% Cu/Al2O3
Data point
% atom Cu
% atom Ag
1
2
3
4
5
6
7
8
3.26
4.56
6.61
96.74
95.44
93.39
95.38
87.11
88.15
86.25
16.72
4.62
12.89
11.85
13.75
83.28
(area without dark particles)
Note. Data points refer to numbered arrows in Fig. 16.
4. DISCUSSION
Ammonia oxidation experiments have shown that silver for all the copper-containing samples. For these reasons, we
alone supported on alumina has a very high ammonia ox- judge the occurrence of spinels to be rather unlikely. Ex-
idation activity but a low selectivity to nitrogen. Alumina- cluding the presence of spinels, Cu may be present either in
supported copper catalyst has a very high nitrogen selec- metallic or in oxidic form. Following surface-tension con-
tivity but activity is low at low temperature. As discussed siderations, metallic Cu species would form distinct parti-
previously the aim of this study was to determine if alumina- cles whereas CuO species would wet the γ -alumina surface
supported binary Cu–Ag catalysts yield a higher selectivity much better. The preparation of the catalysts (through wet
to nitrogen without decrease in activity. The results show impregnation of a mixture of Ag and Cu nitrates, followed
that this is indeed the case. Addition of Cu to Ag/Al2O3 by a reduction step) is expected to yield CuO species rather
catalysts, the incipient wetness coimpregnation, showed an than metallic Cu particles. Furthermore, the catalysts were
increase in N2 selectivity from 87 to circa 95% at 250◦C calcined prior to being analyzed with LEIS analysis. For
while activity was not significantly reduced at Ag/Cu weight these reasons, it is very likely that we are dealing with a
ratios greater than unity. An optimum ratio of between 1 CuO species. Although the expected form of copper oxide
and 3 was observed. The activity and selectivity to nitrogen is Cu(II) oxide, XPS analysis of Cu/Al2O3 catalyst employ-
for the SCR of NO by NH3 were also increased follow- ing Wagnen plots revealed that copper is in an oxidation
ing the addition of Cu. This is consistent with the previous state of +1, thus suggesting the presence of Cu2O.
observation that materials capable of catalyzing the SCO
Analysis of 5 and 10 wt% Ag/Al2O3 catalysts by TEM
of NH3 are also able to catalyze the SCR of NO by NH3. and EDX showed that Ag was always present as particles
This is due to the fact that the former reaction proceeds on the alumina surface. This is consistent with expectations
through an NO intermediate. Once formed the NO can be based on surface tension between Ag and Al2O3. TEM re-
selectively reduced by NH3 to nitrogen rather than being vealed that a very wide particle distribution was always
further oxidized to N2O.
present (1–50 nm). XPS analysis of Ag/Al2O3 employing
The fact is that a mechanical mixture of Cu/Al2O3 and Wagnen plots showed that Ag was present in an oxidation
Ag/Al2O3 catalysts, with an overall Ag/Cu ratio of 3 : 1, did state between +1 and +2. XRD analysis also showed the
not produce an increase in selectivity to nitrogen, as ob- presence of AgO besides metallic silver. This is surprising,
served in the catalysts prepared by incipient wetness coim- as Ag2O is very unstable and AgO decomposes at 200◦C,
pregnation. The above result indicates that either close so that following calcination at 500◦C metallic silver was
physical proximity or intimate chemical interaction be- expected.
tween the silver and the copper components is necessary
TEM analysis of binary Cu–Ag catalysts revealed that a
in order to increase selectivity. This can be understood by very large particle distribution was present (1–50 nm). EDX
the fact that the intermediate NO cannot desorb from the analysis revealed that the particles contained very large
catalyst surface at temperatures below 300◦C, as shown pre- amounts of Ag relative to Cu. Areas in which no particles
viously (9). Thus at low temperatures, transport of NO be- were present contained relatively more copper but still con-
tween Ag and Cu must occur by surface diffusion.
tained some silver. This suggests again that Cu was highly
In order to further understand the physicochemical na- dispersed and that the particles were of Ag. Silver parti-
ture of the Cu–Ag/Al2O3 catalysts the results of the vari- cles of very small diameters (<1 nm) must also be present.
ous spectroscopic techniques must now be examined. XRD The observations by TEM are consistent with the results

ALUMINA-SUPPORTED Cu–Ag CATALYSTS FOR AMMONIA OXIDATION
69
of LEIS. A model generated suggested that copper oxide silver in the areas without dark particles. This indicates that
was highly dispersed, as a monolayer, and that the silver a fraction of silver is dispersed in the form of platelets or
particles sit on top of this layer. The average Ag particle very small particles that cannot be observed by TEM. This
size for all silver-containing samples determined by LEIS fraction of silver may be most interesting from a catalytic
was between 2.5 and 5.5 nm. The simple model predicts that point of view. Unfortunately we still cannot unambiguously
the silver particle-size distribution is quite narrow and does determine which fraction of silver plays the major role in
not vary a lot among all the silver-containing catalysts. This ammonia oxidation at this stage.
distribution is much smaller than was observed by TEM.
Multifunctional catalysis is not a new concept, but a cata-
This probably arises from the fact that LEIS measures the lyst design deliberately based on the principle of multi-
silver content of the entire surface, thus including the small functionality will be necessary in order to develop high-
particles invisible to TEM.
performance catalysts which can cope with such difficult
XPS results show that the valence states of copper and problems as those in ammonia oxidation. This study only
silver are the same for single-component and bicomponent gives one example of such a design. Actually there are more
catalysts. The TEM images show very broad Ag particle- alternatives available. Besides silver, many noble metals,
size distributions for all these catalysts. There is no indi- such as Pt and Ir, are very active for ammonia oxidation
cation for the formation of a new Cu–Ag phase. Indeed, with low selectivity to nitrogen due to the low efficiency
as shown by the phase diagram (27), Ag and Cu have very for SCR reaction. There are also many good SCR catalysts,
lowsolubilities together at temperaturesbelow780◦C. They such as vanadium oxide and iron oxide. Opportunities still
tend to phase-separate at lower temperatures. There is no exist for preparing high-performance bifunctional catalysts
stable region in which a Cu–Ag alloy is formed.
from these catalytic components.
As described previously, the ammonia oxidation on silver
is a two-step reaction. Ammonia is first oxidized to NO
and then the adsorbed NO is further reacted with NHx
to produce N2. In a side reaction, NO can be oxidized to
N2O. The addition of copper to a silver catalyst enhances
the selectivity of the SCR reaction. The promotional effect
of copper can thus be well explained by a mechanism of
bifunctional catalyst.
Since the NO produced in the first step of ammonia ox-
idation cannot desorb at low temperature, the copper and
silver components should be in intimate contact with each
other so that the NO produced on the silver surface can
migrate to the copper surface easily. Otherwise, NO pro-
duced on the silver cannot reach the copper surface and
the chances are greater that N2O is produced rather than
nitrogen. The EDX results showed that copper and silver
existed together everywhere on the alumina surface. This
proved that there was an intimate contact between copper
and silver on alumina-supported Cu–Ag catalysts. Our ex-
periments show that the performance of the catalyst, which
is a physical mixture of 10 wt% Ag/Al2O3 and 10 wt%
Cu/Al2O3, behaves similarly to the 10 wt% Ag/Al2O3 cata-
lyst alone. This again confirms the importance of intimate
contact between copper and silver.
5. CONCLUSIONS
Alumina-supported silver catalysts are very active for
ammonia oxidation but N2O is also coproduced on these
catalysts. The addition of copper to these silver-based cata-
lysts can greatly improved the catalytic selectivity to nitro-
gen with a negligible loss in activity for ammonia oxidation
within the optimum Ag/Cu weight ratio range of between
1 and 3. XPS results show that the valence states of copper
and silver are the same for single-component and bicom-
ponent catalysts. LEIS and XPS measurements show no
indication of formation of a new Cu–Ag phase. Such Cu–
Ag alloys have been shown earlier to be thermodynamically
unstable.
TEM images show the very broad particle-size distribu-
tions for all these catalysts. EDX results showed that cop-
per and silver existed together everywhere on the alumina
surface; thus intimate contact between copper and silver on
alumina-supported Cu–Ag catalysts exists. All results seem
to be consistent with a model that suggests that the catalyst
structure consists of a highly dispersed (monolayer) of cop-
per oxide on alumina upon which silver particles of a very
wide particle-size distribution sit.
In summary, copper was highly dispersed in all catalysts
analyzed by LEIS and TEM. Most likely, the catalysts con-
tain copper oxides platelets with a thickness of roughly
1 ML. At increasing copper loading, relatively more (or
more extended) rather than thicker platelets are formed.
However, silver has a broad particle-size distribution. The
average size of the Ag species ranges between 2.5 and
5.5 nm, measured from LEIS. These numbers increase with
an increase in silver loading. From TEM images silver is
present as relatively large particles, with sizes ranging from
The promotional effect of copper can be explained by
a bifunctional mechanism, in which the silver component
mainly catalyzes ammonia oxidation to NO, the first step of
this reaction, and the copper catalyzes the SCR of NO to
nitrogen, thus reducing N2O formation on silver.
ACKNOWLEDGMENTS
The authors kindly thank Mr. C. D. de Haan, National Centre for High
Resolution Electron Microscopy, Delft University of Technology, The
1 to 50 nm. The EDX results also show the existence of Netherlands, for assistance with the EDX measurements. The preparation

70
GANG ET AL.
of the Cu(Ag) and Ag(Cu) solid solutions by Ir. M. J. H. van Dal is greatly
appreciated. Financial support was provided by the Netherlands Technol-
15. Hellings, G. J. A., Ottevanger, H., Boelens, S. W., Knibbeler, C. L.
C. M., and Brongersma, H. H., Surf. Sci. 162, 913 (1985).
ogy Foundation (STW) and the Netherlands Organization for Scientific 16. Knibbeler, C. L. C. M., Hellings, G. J. A., Maaskamp, H. J.,
Research (NWO).
Ottevanger, H., and Brongersma, H. H., Rev. Sci. Instrum. 58, 125
(1987).
17. Ackermans, P. A. J., van der Meulen, P. F. H. M., Ottevanger, H.,
van Straten, F. E., and Brongersma, H. H., Nucl. Instrum. Methods
Phys. Res. Sect. B 35, 541 (1988).
REFERENCES
1. van Loosdrecht, M. C. M., and Heijnen, S. J., TIBTECH 11, 117
(1993).
2. Cole, J., TIBTECH 11 368 (1993).
18. van Kessel, O., Brongersma, H. H., Ho¨lscher, J. G. A., van Welzenis,
R. G., Sengers, E. G. F., and Janssen, F. J. J. G., Nucl. Instrum. Methods
Phys. Res. Sect. B 64, 593 (1992).
3. Amblard, M., Burch, R., and Southward, B. W. L., Appl. Catal. B 22(3), 19. Jacobs, J. P., Maltha, A., Reintjes, J. G. H., Drimal, J., Ponec, V., and
L159 (1999).
Brongersma, H. H., J. Catal. 147, 133 (1994).
20. Fullarton, J. C., Jacobs, J.-P., van Benthem, H. E., Kilner, J. A.,
Brongersma, H. H., Scanlon, P. J., and Steele, B. C. H., Ionics 1, 51
(1995).
21. Moulder, J. F., Stickle, W. F., Sobol, P. E., and Bomben, K. D., in
“Handbook of X-Ray Photoelectron Spectroscopy” (J. Chastain, Ed.),
pp. 231–234. Perkin–Elmer, Eden Prairie, MI, 1992.
22. Briggs, D., and Seah, M. P. (Eds.), “Practical Surface Analysis by
Auger and X-Ray Photoelectron Spectroscopy.” Wiley, New York,
1983.
4. Biermann, J. J. P., Ph.D. thesis. University of Twente, Twente, 1990.
5. de Boer, M., Huisman, H. M., Mos, R. J. M., Leliveld, R. G., Dillen,
A. J., and Geus, J. W., Catal. Today 17, 189 (1993).
6. Wollner, A., and Lange, F., Appl. Catal. A 94, 181 (1993).
7. Li, Y., and Armor, J. N., Appl. Catal. B 13, 131 (1997).
8. Sazonova, N. N., Simakov, A. V., and Veringa, H., React. Kinet. Catal.
Lett. 57(1), 71 (1996).
9. Gang, L., van Grondelle, J., Anderson, B. G., and van Santen, R. A.,
J. Catal. 186, 100 (1999).
10. Ostermaier, J. J., Katzer, J. R., and Manoque, W. H., J. Catal. 41, 277
(1976).
23. Gang, L., van Grondelle, J., Anderson, B. G., and van Santen, R. A.,
Catal. Today 61, 179 (2000).
11. II’chenko, N. I., Russ. Chem. Rev. 45, 1119 (1976).
12. van den Broek, A. C. M., Ph.D. thesis. Technical University of
Eindhoven, Eindhoven, 1998.
24. Friedman, R. M., Freeman, J. J., and Lytle, F. W., J. Catal. 55, 10
(1978).
25. Armor, J. N., Appl. Catal. 1, 221 (1992).
13. Gang, L., van Grondelle, J., Anderson, B. G., and van Santen, R. A.,
Submitted for publication.
26. Wagner, C. D., Gale, L. H., and Raymond, R. H., Anal. Chem. 51(4),
466 (1979).
14. Gang, L., van Grondelle, J., Anderson, B. G., and van Santen, R. A.,
J. Catal. 199, 107 (2001).
27. Hansen, P. M., “Constitution of Binary Alloys.” McGraw–Hill,
New York, 1958.