Applied Catalysis B: Environmental 36 (2002) 261–277

Combinatorial preparation and high-throughput catalytic tests

of multi-component deNOx catalysts
M. Richter∗ , M. Langpape, S. Kolf, G. Grubert, R. Eckelt,
J. Radnik, M. Schneider, M.-M. Pohl, R. Fricke
Institut für Angewandte Chemie, Berlin-Adlershof e.V., Richard-Willstätter-Str. 12, D-12489 Berlin, Germany
Received 24 March 2001; received in revised form 13 May 2001; accepted 30 September 2001
Dedicated to Prof. Dr. B. Lücke on the occasion of his 65 birthday.

Abstract
A quaternary catalyst library of 56 samples comprising all combinations of four elements, viz. Ag, Co, Cu, In, with six equally
spaced atomic fraction increments from 0 to 1 was prepared by impregnation of a proprietary mesoporous alumina support.
Catalytic properties of the library were tested in the selective catalytic reduction (SCR) of NOx by propane under lean conditions
in the temperature range 400–500 ◦ C. The catalytic data acquired by a parallel 64-channel microreactor system with automated
time-of-flight mass spectrometric analysis have been evaluated regarding selectivity–compositional relationships, synergistic
effects for NOx conversion, and efficiency of propane utilization. Full conversion of NOx is achieved over Ag–Co combinations
at 450 ◦ C with N2 selectivities of more than 90% and reductant utilization of 20% in a feed of 1500 ppm NO, 1500 ppm propane
and 5 vol.% O2 (space velocity of 36,000 cm3 gcat −1 h−1 ). For the single-component catalysts Ag/Al2 O3 , Co/Al2 O3 , Cu/Al2 O3 ,
and In/Al2 O3 , the state of the elements on the mesoporous alumina was characterized by X-ray diffraction (XRD), X-ray
photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM). Cobalt forms a spinel-like cobalt aluminate
phase whereas copper and indium are present as oxides with small sizes not detectable by XRD. Silver occurs in both metallic
state and as Ag2 O, and forms Agn clusters of at least two different sizes, predominantly with diameters of about 30 nm. The
conclusions are consistent with the reducibility of the single-component catalysts samples by H2 . Surface area measurements
and pore size distributions revealed reasonable modifications of the textural properties. The main pore size of the alumina
support is decreased from 7 to ca. 5 nm after loading of the active components. © 2002 Elsevier Science B.V. All rights reserved.
Keywords: SCR of NOx by propane; Combinatorial heterogeneous catalysis; Noble metal-free catalysts; Catalyst library; Multi-channel
reactor system

1. Introduction worldwide [1–3]. These multi-functional catalyst

formulations contain precious metals for the three-way
The development of deNOx automotive catalysts function (hydrocarbon, CO and NOx abatement) at
which are active in excess of air (corresponding to stoichiometric or fuel-rich conditions, supplemented
lean operating conditions of combustion engines at by alkali- or alkaline-earth components for NOx stor-
λ > 1) is a primary target of current research work age under lean conditions [4–6]. During the NOx
storage lean cycle, NO is oxidized at the noble metal
∗ Corresponding author. component of the catalyst to form NO2 which then
E-mail address: mr@aca-berlin.de (M. Richter). reacts with the storage component forming nitrate.

0926-3373/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 6 - 3 3 7 3 ( 0 1 ) 0 0 2 9 0 - 9
262 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

For regeneration, the engine is shortly operated at even under fuel-rich conditions (1.575 vol.% O2 in the
a stoichiometric air/fuel ratio [7]. Concerning NOx feed) whereas Pd gains activity at higher temperatures.
abatement from diesel engines one strategy is the ad- Indium and sodium have no oxidation activity at all.
dition of hydrocarbon reductants in the exhaust gas. The approach we have chosen differs in several
The main component of the catalyst aimed at low points from that one reported. (i) The working temper-
temperature operation with hydrocarbon reductant ature of the catalysts aimed at was from 400 to 500 ◦ C.
in passenger cars is platinum supported on alumina. This means that precious metal components are not
High temperature heavy duty applications need no- suitable as constituents of the catalyst. (ii) The cata-
ble metal-free catalysts preferentially. The concept of lysts should operate under lean conditions, i.e. with
selective catalytic reduction (SCR) of NOx requires oxygen excess. (iii) A mesoporous aluminum oxide
reasonable amounts of hydrocarbons in the exhaust should be taken as support. Due to the ordered struc-
gas. Normally, the hydrocarbon concentration at the ture and a higher surface area of this specific alumina
engine outlet is low and not sufficient for effective support, deposition of the active components should
NOx conversion. As a consequence, hydrocarbons lead to a highly dispersed coverage. The active ele-
have to be supplied by adding secondary fuel into ments chosen were Ag, Co, Cu and In as the most
the exhaust pipe. Lab-scale investigations with model prominent examples with deNOx activities at medium
exhaust gases most frequently use propane, propene and high reaction temperatures. A quaternary cata-
or mixtures of the two hydrocarbons. Elements from lyst library of 56 samples comprising all combinations
the periodic table groups 8 (Co), 11 (Cu, Ag, Au), 13 of the four elements with six equally spaced atomic
(Ga, In) and 14 (Sn) supported on alumina have been fraction increments from 0 to 1 was prepared by auto-
reported to be active for operation temperatures be- mated impregnation of a proprietary mesoporous alu-
tween 350 and 500 ◦ C under fuel-lean conditions for mina support. Catalytic properties of the library were
NOx conversion with propane/propene and methane tested in the selective catalytic reduction of NOx by
[8–15]. Investigations focus on single-component cat- propane under lean conditions in the temperature range
alysts whereas nearly no attempts have been described 400–500 ◦ C by a parallel 64-channel microreactor sys-
to involve multi-component catalyst formulations by tem with automated time-of-flight mass spectrometric
combining several of the active elements on the alu- analysis The analytic equipment allowed the balancing
mina support. Not least one reason is that the number of C- and N-containing components and enabled the
of possible combinations if considering the loading calculation of selectivity of the NOx conversion to N2 .
as additional variable is enormously high, even with Single-component catalyst preparation were charac-
a limited number of active components. Progress in terized by N2 adsorption, X-ray diffraction (XRD),
high-throughput testing with parallel reactors and transmission electron microscopy (TEM), temper-
short analysis times rendered possible the screening ature-programmed H2 reduction, and X-ray photo-
of a large number of catalysts within short times electron spectroscopy (XPS).
[16]. One application of combinatorial catalysis to the
selective reduction of NO has been reported by Krantz
et al. [17]. Starting with four components, a set of 2. Experimental
catalysts has been prepared whose composition re-
sulted from a combination of the elements Pt, Rh, In 2.1. The alumina support
and Na supported on ␥-alumina. It turned out that NO
conversions by propene of 70% are achievable in a Preparation of alumina followed the sol–gel
narrow temperature range between 300 and 400 ◦ C for method described in [18] by controlled hydrolysis
compositions containing 6–8 wt.% Na, 0.4–0.6 wt.% of aluminum tri-sec-butylate, C12 H27 AlO3 (Merck-
Pt, and 0.2–0.4 wt.% Pd. Active catalysts are pre- Schuchardt, Germany) in the presence of structure
dominantly indium-free. C3 H6 consumption and CO2 directing agents in non-aqueous solution. The pre-
formation in parallel confirm that Pt is very active cipitated solid was dried and calcined at 600 ◦ C for
for hydrocarbon oxidation at low temperatures (50% 2 h in air. The fine powder was subsequently pressed,
conversion at 300 ◦ C, 100% conversion at 350 ◦ C) crushed and a fraction between 200 and 500 ␮m has
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 263

been sieved out and used for catalyst preparation. Table 1

To simulate the influence of the impregnation pro- Catalyst samples of the 56 member library
cedure, a part of the support has been impregnated Catalyst no. Loading (atomic fractions)
with distilled water, dried and calcined again. The In Ag Co Cu
alumina support and the single-component catalysts
were characterized by N2 adsorption for their surface 01 0 0 1 0
02 0.2 0 0.8 0
areas, pore volumes and average pore diameters. 03 0 0.2 0.8 0
04 0.4 0 0.6 0
2.2. Preparation of the catalyst library 05 0.2 0.2 0.6 0
06 0 0.4 0.6 0
Ag, Co, Cu and In were chosen as active ele- 07 0.6 0 0.4 0
08 0.4 0.2 0.4 0
ments for reasons discussed in Section 1. Solutions 09 0.2 0.4 0.4 0
were prepared by solving AgNO3 , Co(NO3 )2 ·6H2 O, 10 0 0.6 0.4 0
Cu(NO3 )2 ·3H2 O and In(NO3 )3 ·6H2 O (all from 11 0.8 0 0.2 0
Aldrich, Germany) in distilled water to yield a 1 M 12 0.6 0.2 0.2 0
solution. The catalyst library comprises all possible 13 0.4 0.4 0.2 0
14 0.2 0.6 0.2 0
combinations of the four elements in six concentra- 15 0 0.8 0.2 0
tion increments taken as atomic fractions

xi in steps 16 1 0 0 0
of 0.2 from 0 to 1 with the limitation xi = 1. From 17 0.8 0.2 0 0
this approach, a library results comprising 56 samples 18 0.6 0.4 0 0
with different compositions. The maximum loading 19 0.4 0.6 0 0
was set to 0.5 mmol g−1 leading to single-component
20 0.2 0.8 0 0
21 0 1 0 0
catalyst samples of 5.00 wt.% Ag/Al2 O3 , 2.95 wt.% 22 0 0 0.8 0.2
Co/Al2 O3 , 3.20 wt.% Cu/Al2 O3 , and 5.52 wt.% 23 0.2 0 0.6 0.2
In/Al2 O3 . 24 0 0.2 0.6 0.2
The catalysts have been prepared by impregnation 25 0.4 0 0.4 0.2
26 0.2 0.2 0.4 0.2
of the mesoporous alumina support on the synthe- 27 0 0.4 0.4 0.2
sis robot SOPHAS (Zinsser Analytik, Germany) 28 0.6 0 0.2 0.2
with different quantities of the nitrate solutions. The 29 0.4 0.2 0.2 0.2
impregnation was performed as incipient wetness. 30 0.2 0.4 0.2 0.2
The capacity of the solid to absorb solution until com- 31 0 0.6 0.2 0.2
32 0.8 0 0 0.2
plete wetness was determined by preliminary mea- 33 0.6 0.2 0 0.2
surements and amounted to 1 cm3 g−1 . The volume of 34 0.4 0.4 0 0.2
solution was adjusted, if necessary, by dilution with 35 0.2 0.6 0 0.2
distilled water. After impregnation, the set of catalysts 36 0 0.8 0 0.2
was calcined in an oven at 550 ◦ C for 2 h. Catalyst 37 0 0 0.6 0.4
38 0.2 0 0.4 0.4
samples are listed in Table 1. The numbers given will 39 0 0.2 0.4 0.4
be used throughout the work for abbreviation. 40 0.4 0 0.2 0.4
41 0.2 0.2 0.2 0.4
2.3. Catalytic measurements 42 0 0.4 0.2 0.4
43 0.6 0 0 0.4
44 0.4 0.2 0 0.4
Catalytic experiments were carried out in a catalyst 45 0.2 0.4 0 0.4
testing system using a multi-channel reactor. The re- 46 0 0.6 0 0.4
actor is made of ceramic blocks containing 64 reaction 47 0 0 0.4 0.6
channels with a diameter of 5 mm in three concentric 48 0.2 0 0.2 0.6
circles. An inner and an outer heating ensured a good 49 0 0.2 0.2 0.6
50 0.4 0 0 0.6
isothermal zone in the whole ceramic block. The 51 0.2 0.2 0 0.6
reaction channels are comparable to classical fixed
264 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

Table 1 (Continued ) (±5%). The selectivity of the reaction can be ex-

pressed by different terms. N2 selectivity is understood
Catalyst no. Loading (atomic fractions)
as the percentage of NO converted to N2 , expressed
In Ag Co Cu as (SN = 100(cN2 /XNO ). Formation of undesired ni-
52 0 0.4 0 0.6 trous oxide is the main side reaction to be considered.
53 0 0 0.2 0.8 The extent of CO and propene formation is accounted
54 0.2 0 0 0.8 for by the percentage S CO = 100(cCO /Xpropane ) and
55 0 0.2 0 0.8 S propene = 100(cpropene /Xpropane ), respectively.
56 0 0 0 1
Synergistic effects will be judged to exist if experi-
mentally observed NOx conversions for multi-compo-
nent catalyst formulations deviate from hypothetical
bed reactors. They were operated at or slightly above
values calculated from results of the single-component
ambient pressure. Catalyst samples (mcat = 0.1 g,
catalysts assuming additive contributions to the over-
d p = 200–500 ␮m) are placed on stainless steel
all activity. Taking the experimentally determined
grids. A similar flow through the parallel reactors is
NO conversions of the single-component catalyst
achieved by a quartz powder bed in each channel giv-
samples, a NOx conversion X is calculated for
ing the same pressure drop across the reactor block.
all multi-component catalyst samples according to
All channels are permanently on stream, and they are
Eq. (1)
connected to the analytical instrument by a system of
multi-port valves that allows a sequential analysis of X = xAg XAg + xCo XCo + xCu XCu + xIn XIn (1)
the exit stream of each reaction channel. The on-line
where xi is the atomic fraction of the given element
analyses of the product stream are accomplished on
and Xi the conversion achieved by the given single-
a combination of a fast gas chromatography (GC)
component catalyst.
and a time-of-flight mass spectrometer (TOF-MS).
The utilization of propane for reduction of NO is of
The GC provides a partial separation of the differ-
interest, besides the competing oxidation to CO and
ent products which might be easily analyzed by the
CO2 in excess oxygen, because fuel saving is higher
TOF-MS. This combination enables the analysis of
with lower percentages of reductant oxidation. Oxi-
NO, N2 , N2 O, propane, propene, CO, CO2 five times
dation of propane by oxygen is given by the reaction
in less than 3 min. A complete analysis of the entire
stoichiometry
library of 64 catalysts required about 2.5 h. Fur-
ther technical details are described elsewhere [19]. C3 H8 + 0.5(3z + 4)O2 → 3COz + 4H2 O (2)
Reaction temperatures have been varied between 400
and 500 ◦ C. The reaction mixture consisting of NO considering CO and CO2 as oxidation products with
(1500 ppm), C3 H8 (1500 ppm), O2 (5%) and helium z = (nCO + 2nCO2 )/(nCO + nCO2 ). Formally, the
(balance) passed the catalysts at a total flow rate of SCR of NO by propane is equivalent to oxidation of
60 ml (STP) min−1 corresponding to a gaseous hourly propane by nitrogen monoxide (Eq. (3)).
space velocity (GHSV) of 36,000 cm3 gcat −1 h−1 .
C3 H8 + (3z + 4) NO
→ 0.5(3z + 4)N2 + 3COz + 4H2 O (3)
2.4. Data representation
This stoichiometry represents a borderline value
The NO conversion (in %) is given by XNO where the reaction between propane and nitrogen
= 100(cNO,in − cNO,out )/cNO,in where cNO,in and monoxide is accomplished without participation of
cNO,out are the concentrations of NO at the reactor gas phase oxygen. If z = 2, 1 mol of propane should
inlet and outlet, respectively. Correspondingly, the convert 10 mol of NO. However, preliminary exper-
overall propane conversion is formulated. On the basis iments showed that NO reduction without gas phase
of product concentrations a balance of N-containing oxygen does not proceed on the single-component
components and C-containing products has been cal- catalysts. Hence, at least 0.5 mol O2 is necessary
culated leading to values of 100% with good accuracy to enable initial oxidation of NO to NO2 (Eq. (4)).
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 265

NO2 as strong oxidizing agent converts propane, and (Altamira Instruments) equipped with a thermal con-
becomes simultaneously reduced to N2 (Eq. (5)). ductivity detector. For reduction, a feed of 5% H2
in Ar was used at a heating rate of 5 K min−1 . The
NO + 0.5O2 ↔ NO2 (4)
overall flow rate amounted to 25 cm3 min−1 . The con-
sumption of H2 was determined from the integrated
C3 H8 + 0.5(3z + 4)NO2 peak areas by calibration.
→ 0.25(3z + 4)N2 + 3COz + 4H2 O (5)
2.5.5. X-ray photoelectron spectroscopy
Linking both reaction steps after adjusting the XPS spectra of the single-component catalysts were
stoichiometric factors, Eq. (6) describes the overall obtained using an ESCALAB 220iXL spectrometer
process where 1 mol of propane is able to convert (VG Instruments) equipped with a monochromatic Al
between 3.5 mol NO (z = 1) and 5 mol NO (z = 2). K␣ X-ray source (1486.6 eV). The spectra were cor-
rected with respect to the Al 2p signal of the Al2 O3
C3 H8 + 0.5(3z + 4)NO + 0.25(3z + 4)O2
support at 74.4 eV. For determination of the binding
→ 0.25(3z + 4)N2 + 3COz + 4H2 O (6) energies and the peak intensity a curve fitting has
been performed by combination of Lorentz and Gauss
2.5. Accompanying catalyst characterization curves after a Shirley background subtraction. Peak
intensities were normalized utilizing the sensitivity
2.5.1. Surface areas and pore volumes factors of Scofield [22] and the transmission function
Surface areas and pore volumes of the support and of the instrument. The XPS intensity ratios IMe /IAl
the single-component catalyst samples were deter- were calculated from the Co 2p, In 3d, Ag 3d and the
mined by nitrogen adsorption on the ASAP 2010M Cu 2p normalized peak areas, related to the Al 2p nor-
facility (Micromeritics). The samples were kept at malized peak intensity. A surface concentration nMe
400 ◦ C under vacuum (0.13 Pa) for 4 h before starting (mol%) is calculated as nMe = 100(IMe /(I Me + I Al )).
N2 adsorption at liquid nitrogen temperature. Calcu-
lation of pore sizes followed the method of Barrett–
Joyner–Halenda (BJH) [20] according to implemented 3. Results and discussion
software routines using the Halsey thickness equation
[21] for relating the thickness of the adsorbed layer 3.1. Textural characterization
to the relative pressure. All calculations are based on
the desorption model. Textural information on the alumina support and the
single-component catalyst samples are summarized in
2.5.2. X-ray diffraction measurements Table 2. Nitrogen adsorption isotherms are presented
XRD powder diffraction measurements were car- in Fig. 1(a) indicating a moderate loss of surface area
ried out with a STOE powder diffraction system using caused by the preparation procedure. Largely, this
Cu K␣ radiation at room temperature. loss is already observed for the alumina after impreg-
nation with distilled water and analogous treatment
2.5.3. Transmission electron microscopy as applied for catalyst preparation. That means, the
Electron microscopic images of samples were taken observed textural modifications are only partly at-
by TEM applying the Philips CM20 microscope at tributable to a deposition of the active components.
200 kV (LaB6 ). Specimens were prepared by trans- This is reasonable considering the relatively low metal
ferring small quantities of the samples to microgrids loading. On the other side, a considerable loss of pore
coated by Lacey carbon films. EDX analysis was volume is observed, accompanied by lower average
performed using a PV 9900 spectrometer (EDAX). pore diameters. The pore size distribution shown in
Fig. 1(b) underlines the fact that the hydration and
2.5.4. Temperature-programmed reduction (TPR) calcination caused the major effect of the pore size
Temperature-programmed reduction was perfor- shift to smaller values, whereas the impregnation with
med applying the characterization system AMI-100 the active components lead to only minor changes.
266 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

Table 2
Surface characterization of the mesoporous alumina support and the single-component catalysts Ag/Al2 O3 , Co/Al2 O3 , Cu/Al2 O3 , and
In/Al2 O3
Catalyst Surface area (m2 g−1 ) Pore volume (cm3 g−1 ) Pore diameter (nm)

Al2 O3 (compacted) 291 0.78 8.6

Al2 O3 (impregnated-dried) 265 0.58 6.7
5.00 wt.% Ag/Al2 O3 239 0.45 5.9
2.95 wt.% Co/Al2 O3 224 0.38 5.2
3.20 wt.% Cu/Al2 O3 255 0.41 4.8
5.52 wt.% In/Al2 O3 241 0.41 5.2

The XRD pattern (Fig. 2) show no diffraction lines than for other samples. Because spinel-like cobalt alu-
attributable to crystalline bulk oxide phases of the minate CoAl2 O4 has its most intense diffraction line
supported metals nor to metallic silver in the case at 2Θ = 36.8 [23], a contribution from Co aluminate
of Ag/Al2 O3 . The amorphous character of the meso- might be responsible for the intensity enhancement.
porous alumina support before calcination was con-
firmed (not shown). But, after catalyst preparation, 3.2. Morphology
broad diffraction lines of low intensity at 2Θ = 45.9
and 37.6 indicate the presence of crystalline ␥-alumina. TEM images of the single-component catalysts
This points to some detrimental influence of the prepa- revealed nano-sized clusters in the case of Ag/Al2 O3
ration procedure on the amorphous alumina phase (Fig. 3) with predominant sizes of about 30 nm,
structure. It has to be noted that the diffraction line at besides a few larger aggregates of 90–100 nm. Accor-
2Θ = 37.6 is more intense in the case of Co/Al2 O3 ding to EDX analysis, the clusters consist of metallic

Fig. 1. (a) Nitrogen adsorption isotherms of the alumina support after calcination at 600 ◦ C: (䊏) as prepared, (䊉) after impregnation with
distilled water, drying and calcining, and of the single-component catalysts (×) Ag/Al2 O3 (5.00 wt.%), (+) Co/Al2 O3 (2.95 wt.%), (䉬)
Cu/Al2 O3 (3.20 wt.%), and (䊊) In/Al2 O3 (5.52 wt.%). (b) Pore size distribution.
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 267

Fig. 2. XRD diagrams of single-component catalysts (a) Co/Al2 O3 , (b) In/Al2 O3 , (c) Ag/Al2 O3 , and (d) Cu/Al2 O3 .

silver. On the other side, temperature-programmed besides nano-sized silver clusters. Consequently, the
reduction by H2 (see later) proved the existence of dispersion of silver after preparation is not uniform.
reducible silver species, either Ag2 O or Ag+ , repre- Clusters are also detectable by TEM for sample
senting nearly 50% of the overall loading. Therefore, In/Al2 O3 but, formation of indium oxide aggregates
silver should exist in highly dispersed Ag2 O and Ag+ , takes place during measurement. This implies a

Fig. 3. TEM micrograph of single-component catalyst Ag/Al2 O3 .

268 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

Cu/Al2 O3 . The reduction profile of Ag/Al2 O3 (no.

21) reveals three overlapping peaks centered at about
100–120, 275 and 650 ◦ C. The peak at relatively low
temperatures is attributable to the reduction of Ag2 O.
H2 consumption during TPR in the temperature range
of 500–700 ◦ C is indicative for reduction of stabi-
lized Ag+ ions to Ag0 , at least in the case of zeolite
supports [24,25].
Reduction of Co/Al2 O3 (sample no. 01) proceeds
in a non-specific way with faint peaks of H2 con-
sumption at about 575 and above 800 ◦ C. The formal
reduction degree amounts to only 10% (cf. Table 3).
It is known [26] that cobalt ion diffusion into the
structure of alumina leads to species that are difficult
Fig. 4. H2 consumption profiles of TPR measurements of to reduce. This would imply a prevalent incorporation
single-component catalysts Ag/Al2 O3 (no. 21), Cu/Al2 O3 (no. 56),
of the impregnated Co into the support matrix with
Co/Al2 O3 (no. 01) and In/Al2 O3 (no. 16). Conditions: 5 vol.%
H2 /Ar, flow rate 25 cm3 min−1 , heating rate 5 K min−1 , catalyst high dispersion of Co ions. This supports the conclu-
mass 0.1 g. Isothermal hold for 10 min at 800 ◦ C. sions drawn from XRD and XPS measurements that
a cobalt aluminate phase is formed after calcination.
mobility with subsequent aggregation caused by the Reduction of Cu/Al2 O3 (no. 56) is characterized
electron beam. The time to accomplish the electron by one broad H2 consumption centered at 350 ◦ C.
image is sufficient to change the surface state of the Formation of CuO (black) is not dominant because
indium oxide. For this reason, the TEM picture is the color of the Cu/Al2 O3 remained blue-greenish
not representative for the state of the indium after after calcination. Nevertheless, the consumption of
preparation and during catalysis. The TEM pictures of H2 corresponds to a complete accessibility of the Cu
Co/Al2 O3 and Cu/Al2 O3 gave no indication of clus- for reduction from Cu2+ to Cu0 .
ters or surface metal oxide phases. The reduction of In/Al2 O3 (no. 16) starts at about
175 ◦ C and is characterized by three overlapping
3.3. TPR measurements steps. Reduction is not completed even at 800 ◦ C.
This is significantly different from reduction profiles
Reduction profiles of the single-component cata- reported for indium impregnated on pyrogenic silica
lysts are presented in Fig. 4. After integration, the and for indium oxo cations ([InO]+ ) fixed to cationic
overall consumption of hydrogen was calculated positions of zeolites [27]. The overlapping processes
and compared with amounts required for reasonable cannot unambiguously be assigned to single reduc-
changes of the valence state of the active components tion steps. Formally, the hydrogen balance allows a
as summarized in Table 3. complete reduction of In2 O3 to In2 O where the com-
Qualitatively, it can be seen that the reduction is plex shape of the reduction profile indicates a certain
not completed at 800 ◦ C, with exception of that for heterogeneity of the impregnated indium species.

Table 3
Results of the TPR characterization for single-component catalyst formulations
Catalyst no. Catalyst Loading H2 consumed Reduction Valence change
(wt.%)a (mmol g−1 ) degree (%) assumed
21 Ag/Al2 O3 5.00 0.13 52 Ag+ → Ag0
01 Co/Al2 O3 2.95 0.05 10 Co2+ → Co0
56 Cu/Al2 O3 3.20 0.45 108 Cu2+ → Cu0
16 In/Al2 O3 5.52 0.56 112 In3+ → In+
a Corresponding to 0.50 mmol g−1 .
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 269

Fig. 5. XPS spectra of single-component catalysts Ag/Al2 O3 , Co/Al2 O3 , Cu/Al2 O3 and In/Al2 O3 ; sat. denotes satellites.

Considering the TEM results, a modification of the In The Ag 3d5/2 binding energy of sample Ag/Al2 O3
state during the TPR run cannot be excluded. at 368.0 eV has to be compared with binding ener-
gies of 368.3 and 367.5 eV for metallic silver and
3.4. XPS characterization Ag2 O, respectively [10,24,28]. Thermally, bulk Ag2 O
decomposes to Ag metal at higher temperatures
XPS spectra of the single-component catalysts are [28]. A comparison of binding energies of sample
presented in Fig. 5. Binding energies (BE), IMe /IAl Ag/Al2 O3 with the literature data favors metallic sil-
surface ratios, surface concentrations and assignments ver although the peak width published for metallic
are listed in Table 4. silver and silver oxide peaks are around 1 eV whereas

Table 4
Results of the XPS characterization for single-component catalyst formulations
Catalyst No. Peak BE (eV) IMe /IAl a nMe /n(Me+Al) (mol%) Valence state Compound

Ag/Al2 O3 21 Ag 3d5/2 368.0 0.0496 0.047 Ag0 /Ag+ Ag, Ag2 O

Co/Al2 O3 01 Co 2p3/2 780.8 0.0820 0.076 Co2+ CoAl2 O4
Cu/Al2 O3 56 Cu 2p3/2 933.9 0.0887 0.081 Cu2+ CuO
In/Al2 O3 16 In 3d5/2 445.3 0.2767 0.217 In3+ In2 O3
a The theoretical value at homogeneous metal distribution throughout the alumina bulk phase amounts to 0.025.
270 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

the experimental value for Ag/Al2 O3 amounts to that reported for indium in In2 O3 (444.4 eV) [29].
2.4 eV. The Ag surface concentration resulting from Indium at the catalyst surface should be located in
XPS analysis is near to the homogeneous bulk distri- an In2 O3 bulk phase environment, although no crys-
bution what is unlikely because of the impregnation talline In2 O3 could be detected by XRD. The In sur-
method applied for preparation. Moreover, location face concentration is the highest, compared with the
of silver at the alumina surface is documented by other single-component catalysts and implies that no
TEM results. Because the size of the Ag clusters is indium has penetrated into subsurface regions. TEM
larger than the penetration depth of the photo elec- investigations prove that aggregates are formed during
trons (30 nm versus 2 nm), it is concluded that not all acquisition of the micrograph. However, the initial
of the silver inside the clusters is detected by XPS. dispersion of In2 O3 after sample preparation should
This leads to the low IAg /IAl surface ratio. be higher and aggregation under the oxidizing cat-
Sample Co/Al2 O3 revealed a Co 2p3/2 binding alytic conditions at normal pressure is unlikely. The
energy of 780.8 eV that is higher than characteristic H2 -TPR profile showed full accessibility of indium
for Co3 O4 (780.0 eV) and CoO (780.1 eV) [28,29]. A and nearly quantitative reduction from In3+ to In+ .
binding energy of 781.9 eV is reported for cobalt alu-
minate [12,29]. Ji et al. [30] attributed the Co 2p peak 3.5. Catalytic results
at 782.0 eV of 10 wt.% Co impregnated on sol–gel
derived alumina to a CoAl2 O4 phase. Although ther- 3.5.1. Activities and selectivities
mal decomposition of Co nitrate preferably leads to The properties of the single-component composi-
Co3 O4 , the interaction of Co species with the alu- tions are summarized in Fig. 6(a–d). Ag/Al2 O3 has the
mina surface results in formation of bulk Co alumi- highest activity achieving full conversion at ca. 475 ◦ C,
nate by Co2+ diffusion into the alumina framework whereas In/Al2 O3 is the single-component catalyst
depending on the alumina reactivity and temperature with the lowest activity. The selectivity SN amounts to
of calcination [30]. 90% at minimum, i.e. the reaction proceeds selectively
The observed shift of binding energies to higher to N2 without noticeable N2 O formation. Also, a de-
values indicates the presence of surface Co2+ , proba- sirable low selectivity of CO formation is maintained
bly located by a cobalt aluminate-like phase. This is by the single-component catalyst formulations with
reconcilable with the poor reducibility stated by TPR, exception of In/Al2 O3 where SCO reaches a maximum
because reduction of Co2+ in CoAl2 O4 is more diffi- of 40% at low temperature. Although, at higher re-
cult than that of Co3 O4 due to the stronger interaction action temperatures, SCO decreases to approximately
with the support. 10% at 475 ◦ C. Interestingly, Co/Al2 O3 shows a high
The Cu 2p3/2 peak at 933.9 eV found for sample percentage of propene formation at low temperatures
Cu/A2 O3 corresponds to that one reported for bulk obviously resulting from oxidative dehydrogenation
CuO (934.1 eV) [29]. Additionally, satellite struc- of propane. Correspondingly, CO2 formation is the
tures typical for CuO were observed. The existence least observed for the single-component samples.
of Cu+ ions as well as metallic Cu can be excluded. The NOx conversion levels for all catalyst compo-
Moreover, there are no indications supporting the ex- sitions are presented in Fig. 7(a–d). The corners of the
istence of CuAl2 O4 which has XPS Cu 2p3/2 binding pyramid represent the single-component catalysts and
energies of 935 eV [23,31]. It is concluded that the the gradation is done in steps of 0.2 representing the
surface phase of sample Cu/Al2 O3 is equivalent to six concentration steps in atomic fractions from 0 to 1.
CuO. The CuO phase is X-ray amorphous, and should For all members of the catalyst library, no NO2
be finely dispersed because TEM could not prove the was detected in the exit stream. This might have sev-
existence of aggregates at the surface of the support. eral reasons. (i) The NO/NO2 equilibrium (Eq. (4)) is
Nevertheless, enrichment of Cu at the alumina surface shifted to NO2 dissociation with increasing tempera-
follows from the ICu /IAl surface ratio, comparable to tures. At maximum, 20% NO2 are present in a NO
the value observed for Co/Al2 O3 . (950 ppm)/O2 (2%) mixture at 450 ◦ C. This percentage
A binding energy of 445.3 eV was measured for decreases to approximately 10% at 500 ◦ C [13]. (ii)
In 3d5/2 of sample In/Al2 O3 . This value is close to Due to its intermediary character, NO2 , once formed,
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 271

Fig. 6. Catalytic results for single-component catalysts (a) Ag/Al2 O3 , (b) Co/Al2 O3 , (c) Cu/Al2 O3 , and (d) In/Al2 O3 in the 64-channel
reactor. NOx conversion, (䊉); propane conversion, (䊊); N2 selectivity, (䉱); propene selectivity, (×); CO selectivity, (䉬); N2 O selectivity,
(䉲). Feed 1500 ppm NO, 1500 ppm propane, 5 vol.% O2 , GHSV = 36,000 cm3 gcat −1 h−1 .

is further converted to N2 . (iii) To a certain extent, the also In-containing multi-component catalyst samples
lack of NO2 will be associated with the specific of reveal the highest CO selectivity. One example is
GC–MS analysis if NO2 is transformed into N2 and given in Table 5 for a reaction temperature of 425 ◦ C.
O2 inside the mass spectrometer. These 10 catalysts with the highest CO selectivities
It follows from Fig. 7(a–d) that active catalyst for- contain In but no or less Cu. If Cu is present in higher
mulations are rich in Ag and poor in Cu. Samples concentrations, CO will be oxidized to CO2 . On the
containing a high percentage of In are generally less other side, propene is observed on catalysts having a
active than others. The pure alumina support, calcined high percentage of Co but no Cu (cf. Table 6).
at 600 ◦ C, yields 8–10% NOx conversion at 500 ◦ C. It turned out that the optimum catalyst is composed
At lower temperatures, contribution of the support to by Co and Ag in approximately equal concentrations,
the deNOx activity of the catalysts is negligible. because at 425 ◦ C catalyst no. 06 (0.4 Ag, 0.6 Co) and
Selectivity values SN are near 100% for almost no. 10 (0.6 Ag, 0.4 Co) reveal the highest activities, ca.
all catalyst samples and decrease to a minimum of 70% NOx conversion, besides the single-component
90% over Ag-rich catalysts at higher reaction tem- catalyst samples Ag/Al2 O3 and Co/Al2 O3 .
peratures. Correspondingly, N2 O formation is always The relatively high concentration of propene
lower than 10%. As already follows from the catalytic observed for the Co/Al2 O3 single-component cata-
test results of single-component catalyst samples, the lyst is attributed to the low oxidation activity of the
SCR of NOx by propane over In/Al2 O3 is charac- CoAl2 O4 phase despite the large oxygen excess app-
terized by a high percentage of CO formation. Thus, lied. Therefore, propene is provided as an alternative
272 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

Fig. 7. NOx conversion (gray-scale) of the 56-membered catalyst library at (a) 400 ◦ C, (b) 425 ◦ C, (c) 450 ◦ C, and (d) 475 ◦ C. Feed
1500 ppm NO, 1500 ppm propane, 5 vol.% O2 , GHSV = 36,000 cm3 gcat −1 h−1 . Catalyst enumeration according to Table 1.
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 273

Table 5 tion (1000 ppm NO, 1000 ppm propene, 5 vol.% O2 )

Multi-component catalyst samples with highest CO selectivities at are fairly comparable to the conditions applied in the
425 ◦ C
present work, with exception of the higher space veloc-
Catalyst no. Composition (atomic fractions) SCO ity and the different reductant (propene). NOx conver-
Ag Co Cu In
(%)
sion levels of more than 90% at 450 ◦ C had been found
16 0 0 0 1.0 44.5
if the catalyst was calcined at appropriate high tem-
17 0.2 0 0 0.8 43.4 peratures (600 or 700 ◦ C). In comparison to the mea-
11 0 0.2 0 0.8 40.1 surements presented in this work, the NOx conversion
07 0 0.4 0 0.6 38.2 versus temperature profile seems to be shifted by 50 ◦ C
12 0.2 0.2 0 0.6 30.1 to lower temperatures. But this is attributable to the
33 0.2 0 0.2 0.6 29.1
08 0.2 0.4 0 0.4 28.2
propene reductant, because olefins are more reactive
30 0.4 0.2 0.2 0.2 26.5 than alkanes in SCR chemistry. The Ag-impregnated
35 0.6 0 0.2 0.2 26.4 Al2 O3 was found considerably less active.
04 0 0.6 0 0.4 25.7 A thorough study on Co/Al2 O3 deNOx catalysts
was published in Ref. [13] involving Al2 O3 sup-
ports of different provenience and varied Co contents
reductant, besides the propane fed. Involvement of from 2 to 5 wt.%. Catalytic results reported by Yan
Cu affects adversely the reaction because propane is et al. [13] for lean NOx reduction with propane over
preferentially oxidized to CO2 . Indium leads to inter- Co/Al2 O3 are in fair agreement with results reported
mediate formation of CO which may serve as further in this work, despite diverging space velocity (12,000
reductant for NOx . versus 36,000 cm3 gcat −1 h−1 ) and feed composition
Reference data for single-component catalysts are (1.7% H2 O versus no H2 O added). Yan et al. [13]
available, but direct comparison is difficult due to used a ␥-Al2 O3 prepared by a sol–gel method which,
varying reaction conditions and reductants. Seker et al. however, is different to the preparation route of alu-
[10] studied the SCR of NOx by propene on a 5 wt.% mina applied in the present investigation. The authors,
Ag/alumina catalyst prepared by various methods, by physicochemical characterization, got evidence for
e.g. co-precipitation, a one step sol–gel process, and the existence of four different Co species at 5 wt.%
impregnation of alumina by silver nitrate. The authors loading, Co2+ in CoAl2 O4 , small Co3 O4 particles
reported the highest activity for the sample prepared strong interacting with the support, large Co3 O4 par-
by a sol–gel process. This process is not comparable ticles, and dispersed surface Co2+ ions in octahedral
with the synthesis procedure of the alumina support coordination. The latter Co2+ species are proposed to
described in this work because Seker et al. [10] per- be the most active species. From TPR measurements
formed the hydrolysis of the aluminum alkoxides in with H2 , the authors concluded that tetrahedrally co-
the presence of silver nitrate in aqueous solution. ordinated Co2+ ions in CoAl2 O4 are not reducible
Nevertheless, Ag loading (5 wt.%) and feed composi- below 800 ◦ C. This is in line with TPR results of
Co/Al2 O3 shown in Fig. 4.
Results reported by Maunula et al. [15] included In,
Table 6 Ag and Co catalysts prepared on sol–gel alumina that
Multi-component catalyst samples with highest propene selectivi- consisted mainly of ␥-Al2 O3 after final calcination
ties at 425 ◦ C
at 600 ◦ C. The 3.3 wt.% In/Al2 O3 sample achieved
Catalyst no. Composition (atomic fractions) Spropene nearly 100% NOx conversion to N2 at 400 ◦ C with
(%)
Ag Co Cu In dry feed (1000 ppm NO, 1000 ppm propene, 10 vol.%
02 0 0.8 0 0.2 33.1 O2 ) at a space velocity of 20,000 cm3 gcat −1 h−1 . The
28 0 1.0 0 0 31.3 Co/Al2 O3 sample (1.75 wt.%) revealed intermedi-
03 0.2 0.8 0 0 27.7 ate activity (ca. 60% at 400 ◦ C) whereas Ag/Al2 O3
06 0.4 0.6 0 0 16.2 (3.14 wt.% Ag) had only 10% conversion of NOx to
20 0.8 0 0 0.2 2.3
N2 but the highest oxidation activity for propene. A
10 0.6 0.4 0 0 0.1
pre-calcination of the Ag/Al2 O3 sample at 800 ◦ C
274 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

improved the activity considerably. The authors sug- It is obvious that the majority of the multi-compo-
gested that finely dispersed silver oxide particles still nent catalyst samples is less active than expected from
existed on the sol–gel alumina calcined at 600 ◦ C the sum of single-component contributions because
but larger silver and silver aluminate particles were the difference (XNO,exp − XNO,cal ) is negative by
formed during high-temperature calcination sup- 60–70 points. On the other side, there are catalyst
pressing the total hydrocarbon oxidation, thus, en- samples where experimental NOx conversions are
hancing the NOx conversion. We got no evidence higher than expected from the sum of contributions
for the existence of a silver aluminate phase in the according to Eq. (1). It is striking that this effect is
single-component Ag/Al2 O3 catalyst. Nevertheless, mostly observed for sample numbers up to 20, that is
Ag/Al2 O3 was found highly active for the SCR of for Cu-free catalysts. Starting with catalyst number 22
NOx . The good performance may be attributed to the that contains 0.2 atomic fraction of Cu, a tremendous
predominance of nano-sized silver clusters. deterioration of the experimental NOx conversions are
recognizable, although with some exceptions which,
3.5.2. Synergistic effects however, do not exceed 10–15 points at the highest
To evaluate synergistic effects in NOx conversions, reaction temperature of 475 ◦ C. Further compositional
the data base was screened for the differences between Cu-free variants tend to show good NOx conversion if
experimentally observed NOx conversions and those they mainly contain Ag and Co. A synergistic effect if
calculated by Eq. (1). Results are shown for the 56 being associated with kinetic phenomena should rea-
catalysts in Fig. 8(a–d). sonably depend on the reaction temperature. This is

Fig. 8. Differences of experimental NOx conversions and pre-calculated conversions on the basis of results from single-component catalysts
Ag/Al2 O3 , Co/Al2 O3 , Cu/Al2 O3 and In/Al2 O3 at (a) 475 ◦ C, (b) 450 ◦ C, (c) 425 ◦ C, and (d) 400 ◦ C.
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 275

Table 7
NOx conversions of samples 20 and 21
Catalyst no. Composition (atomic factions) NOx conversion (%) at different temperatures (◦ C)

In Ag 400 425 450 475

20 0.2 0.8 28.4 72.0 100.0 98.8

21 0 1.0 27.1 66.0 95.5 92.8

the case, e.g. for sample 20 where deviations are +4.6, combinations of Co, Ag and In operate above the av-
+18.1, +21.3 and +20.6 at 400, 425, 450 and 475 ◦ C, erage 1:1 stoichiometry. Cu-containing catalysts keep
respectively. a 1:1 ratio at 425 ◦ C, cf. e.g. the single-component
Catalyst 20 is composed of 0.8 atomic fraction of catalyst, no. 56. But at higher reaction temperatures
Ag and 0.2 of In and its catalytic performance is almost (475 ◦ C, Fig. 9(b)), the Cu single-component catalyst
completely that of the pure silver single-component
catalyst formulation, catalyst no. 21 (Table 7).
Because the overall loading is fixed to a constant
value, the dependence of the reaction on the Ag load-
ing in the Ag/Al2 O3 formulation is not investigated.
Therefore, it cannot be decided whether the presence
of indium is necessary to keep the NOx conversion of
the 0.8 Ag/Al2 O3 catalyst near to the 1.0 Ag/Al2 O3
single-component catalyst.

3.5.3. NOx versus propane conversion

NOx versus propane conversion for the catalyst
library studied in this work are displayed in Fig. 9(a
and b) for 425 and 475 ◦ C, respectively. The dotted
lines indicate the 1:1 ratio between the conversions.
According to the assumed stoichiometry that 1 mol
of propane is able to reduce up to 5 mol of NOx , the
dotted lines means 20% utilization of propane for the
SCR of NOx and 80% of competing propane oxida-
tion. It has to be noted that an estimation of reductant
utilization often refers to the stoichiometry
2NO + 4O2 + C3 H8 → N2 + 3CO2 + 4H2 O (7)
leading to a value of 50% reductant utilization if NOx
and reductant conversions are equal. Without knowl-
edge how much oxygen in carbon oxides and water
originates from NO, various stoichiometries are plau-
sible. On the basis of the approach given, the pattern
of Fig. 9 implies an average utilization of the reductant
of about 20% with some exceptions in dependence on
Fig. 9. NOx vs. propane conversion for (a) 425 ◦ C, and (b) 475 ◦ C.
the reaction temperature. At the lower reaction temper-
Feed composition 1500 ppm NO, 1500 ppm C3 H8 , 5 vol.% O2 ,
ature (425 ◦ C), the single-component Co/Al2 O3 and space velocity = 36,000 cm3 gcat −1 h−1 . Enumeration according
Ag/Al2 O3 samples reveal higher reductant selectivi- to Table 1. Some numbers are omitted for clarity. Dotted lines
ties than 20% at high NOx conversions. Furthermore, indicate a 1:1 stoichiometry.
276 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277

reveals the lowest degree of propane utilization. The Ag) is not an optimum. Nevertheless, a combination
conclusion is drawn that copper does not provide a of silver and cobalt is beneficial for the reaction. This
balanced oxidation activity but preferentially oxidizes may partly be attributed to the ability of the CoAl2 O4
propane. Combining Cu with increasing amounts of phase to convert propane to propene which is the
Ag (catalyst nos. 55, 52, 46, 36) improves the utiliza- more favorite reductant in the deNOx process.
tion of propane at 475 ◦ C, accompanied by higher NOx The utilization of the propane for NOx reduction
activity. If the Ag content surmounts the Cu content seems to be confined to a stoichiometry of 1:1 notwith-
(catalysts nos. 46, 36), the behavior of the samples re- standing other stoichiometries that can be formulated.
sembles more that of the Ag/Al2 O3 single-component Correspondingly, the degree of reductant utilization
catalyst. Poor efficiency of NOx conversion is varies between 20 and 50%.
observed for In-containing catalyst formulations. To achieve further improvements of the deNOx
The single-component In/Al2 O3 catalyst (no. 16) and activity, one of the less promising components, say
the 0.8 In/0.2 Ag combination (catalyst no. 17) have copper, can be replaced by another element. It is ex-
the lowest NOx conversion of the catalyst library with pected that a tri-component catalyst formulation can
a propane utilization at the average 1:1 level. be identified with improved catalytic properties.
The main advantage of the performed stage II
screening (optimization mode) is the comparability to
4. Conclusions technical unit operations and the higher information
depth of the analyses providing various selectivity
Multi-component catalysts suitable for the SCR parameters relevant for the deNOx process.
of NOx by hydrocarbons under lean conditions have
been prepared by combination of non-platinum group
metals known for their activity as single-component Acknowledgements
supported catalysts. For four elements chosen, a
systematic combination with six concentration in- Financial support by the Senate of Berlin and the
crements resulted in a 56 member catalyst library, BMBF (FKZ 03C30120) is gratefully acknowledged.
that could be screened for the catalytic performance
by high-throughput testing facilities in a short time.
It was shown that synergistic effects are existent in References
the multi-component catalyst compositions yielding
high NOx conversions at 450 ◦ C and negligible N2 O
[1] E. Hums, Catal. Today (1998) 25.
[2] H. Klein, S. Lopp, E. Lox, M. Kawamani, M. Horiuchi, SAE
by-product. Technical Paper Series 1999-01-0109, International Congress
Up to the metal loading of 0.5 mmol g−1 , the alu- and Exposition, Detroit, MI, USA, March 1999.
mina support prevents formation of crystalline bulk [3] M. Shelef, R.W. McCabe, Catal. Today 62 (2000) 35.
oxides detectable by XRD although the binding en- [4] E. Fridell, M. Skoglundh, B. Westerberg, S. Johansson, G.
Smedler, J. Catal. 183 (1999) 196.
ergies found by XPS point to the existence of In [5] S. Matsumoto, Y. Ikeda, H. Suzuki, Appl. Catal. B: Environ.
and Cu in a bulk-like oxide environment equiva- 25 (2000) 115.
lent to In2 O3 and CuO, respectively. Cobalt forms a [6] H. Huang, R. Long, R. Yang, Energy Fuel 15 (2001) 205.
spinel-like structure, thus, becoming incorporated into [7] Y. Li, S. Roth, J. Dettling, T. Beutel, in: Proceedings Fifth
the alumina bulk. Silver is present in form of metallic International Congress on Catalysis and Automotive Pollution
Control, Brüssels, Belgium, Vol. I, 12–14 April 2000, p. 256.
nano-sized clusters and Ag2 O according to character- [8] H.-W. Jen, Catal. Today 42 (1998) 37.
ization data. The existence of isolated Ag+ ions could [9] K. Takagi, T. Kobayashi, H. Ohktia, T. Mizushima, N. Kakuta,
not unambiguously be verified. The findings are rec- A. Abe, K. Yoshida, Catal. Today 45 (1998) 123.
oncilable with literature data [32,33] that silver load- [10] E. Seker, J. Cavataio, E. Gulari, P. Lorpongpaiboon, S.
ing on alumina supports around 2 wt.% or lower leads Osuwan, Appl. Catal. A: General 183 (1999) 121.
[11] A. Keshavaraja, X. She, M. Flytzani-Stephanopoulos, Appl.
to stabilized Ag+ ions, whereas at higher loading, Catal. B: Environ. 27 (2000) L1.
the existence of Ag clusters is predominant. It can be [12] R. Bechara, D. Balloy, J.-Y. Dauphin, J. Grimblot, Chem.
estimated that the loading of 0.5 mmol g−1 (5.0 wt.% Mater. 11 (1999) 1703.
 M. Richter et al. / Applied Catalysis B: Environmental 36 (2002) 261–277 277

[13] J.Y. Yan, M.C. Kung, W.M.H. Sachtler, H.H. Kung, J. Catal. [24] K.A. Bethke, H.H. Kung, J. Catal. 172 (1997) 93.
172 (1997) 178. [25] H. Berndt, M. Richter, T. Gerlach, M. Baerns, J. Chem. Soc.,
[14] M. Ozawa, H. Toda, S. Suzuki, Appl. Catal. B: Environ. 8 Faraday Trans. 94 (1998) 2043.
(1996) 141. [26] S. Bessell, Appl. Catal. 96 (1993) 253.
[15] T. Maunula, Y. Kintaichi, M. Inaba, M. Haneda, K. Sato, H. [27] R.M. Mihalyi, H.K. Beyer, V. Mavrodinova, Ch. Minchev, Y.
Hamada, Appl. Catal. B: Environ. 15 (1998) 291. Neinska, Microporous Mesoporous Mater. 24 (1998) 143.
[16] S.M. Senkan, Angew. Chem. 113 (2001) 322. [28] G.B. Hoflund, Z.F. Hazos, G.N. Salaita, Phys. Rev. B 62
[17] K. Krantz, S. Ozturk, S. Senkan, Catal. Today 62 (2000) (2000) 11126.
281. [29] XPS International Fundamental Data Tables, www.xpsdata.
[18] R. Eckelt, M. Richter, H. Kosslick, R. Fricke, DE 19843241 org.
(1998). [30] L. Ji, J. Lin, H.C. Zeng, J. Phys. Chem. B 104 (2000) 1783.
[19] M. Langpape, E. Guillon, S. Kolf, O.V. Buyevskaya, M. [31] S.F. Tikhov, V.A. Sadykov, G.N. Kryukova, E.A. Paukshtis,
Baerns, Submitted for publication. V.V. Popovskii, T.G. Starostina, G.V. Kharlamov, V.F.
[20] E.P. Barrett, L.S. Joyner, P.P. Halenda, J. Am. Chem. Soc. Anufrienko, V.F. Polyboyarov, V.A. Razdobarov, N.N.
73 (1951) 373. Bulgakov, A.V. Kalinkin, J. Catal. 134 (1992) 506.
[21] G. Halsey, J. Chem. Phys. 16 (1943) 431. [32] T.E. Hoost, R.J. Kudla, K.M. Collins, M.S. Chattha, Appl.
[22] J.H. Scofield, J. Electron Spectrosc. 8 (1976) 129. Catal. B: Environ. 13 (1997) 59.
[23] K. Shimizu, H. Maeshima, H. Yoshida, A. Satsuma, T. Hattori, [33] K. Shimizu, J. Shibata, H. Yoshida, A. Satsuma, T. Hattori,
Phys. Chem. Chem. Phys. 2 (2000) 2435. Appl. Catal. B: Environ. 30 (2001) 151.