Palladium and platinum-based catalysts in the catalytic reduction of nitrate in water: Effect of copper, silver, or gold addition

Article abstract of DOI:10.1016/S0021-9517(03)00252-5

Supported bimetallic palladium and platinum catalysts promoted by metals of group 11 (Cu, Ag, and Au) were prepared by control surface deposition and tested in the liquid-phase reduction of nitrates. Whereas bimetallic catalysts promoted by gold are totally inactive, copper or silver deposition leads to bimetallic catalysts active for nitrate reduction. The promoting effect of the second metal can be related to its redox properties, confirming that nitrate reduction occurs through a bifunctionnal mechanism following (i) a direct redox mechanism between promoter and nitrate and (ii) a catalytic reaction between hydrogen, chemisorbed on the noble metal, and intermediate nitrite. TEM experiments, TPR, and FTIR of chemisorbed CO studies of the different systems have been used to evidence the metal-metal interaction and the localization of the promoter. The characterization results have been correlated with the catalytic behavior of the materials.

Full text of DOI:10.1016/S0021-9517(03)00252-5

Journal of Catalysis 220 (2003) 182–191
www.elsevier.com/locate/jcat
Palladium and platinum-based catalysts in the catalytic reduction
of nitrate in water: effect of copper, silver, or gold addition
∗
Florence Gauthard, Florence Epron, and Jacques Barbier
Laboratoire de Catalyse en Chimie Organique (UMR6503), CNRS, Université de Poitiers, 40, Avenue du Recteur Pineau, 86022 Poitiers cedex, France
Received 25 February 2003; revised 23 May 2003; accepted 23 May 2003
Abstract
Supported bimetallic palladium and platinum catalysts promoted by metals of group 11 (Cu, Ag, and Au) were prepared by control surface
deposition and tested in the liquid-phase reduction of nitrates. Whereas bimetallic catalysts promoted by gold are totally inactive, copper
or silver deposition leads to bimetallic catalysts active for nitrate reduction. The promoting effect of the second metal can be related to
its redox properties, confirming that nitrate reduction occurs through a bifunctionnal mechanism following (i) a direct redox mechanism
between promoter and nitrate and (ii) a catalytic reaction between hydrogen, chemisorbed on the noble metal, and intermediate nitrite. TEM
experiments, TPR, and FTIR of chemisorbed CO studies of the different systems have been used to evidence the metal–metal interaction and
the localization of the promoter. The characterization results have been correlated with the catalytic behavior of the materials.

2003 Elsevier Inc. All rights reserved.
Keywords: Platinum or palladium based catalysts; Metals of group 11; Nitrate; Catalytic reduction
1
. Introduction
Alloying of metals may result in important changes in
their activity and selectivity in catalytic reactions. As far
as catalytic nitrate reduction is concerned, the changes are
experimentally well established but are still not understood.
Palladium catalysts have proved to be the most active and
selective for nitrite reduction. However, platinum-based cat-
alysts present an activity and a selectivity that are almost
satisfactory [26]. To reduce nitrate, it is necessary to activate
the precious metal by addition of a second metal of groups
Groundwater pollution by harmful nitrogen-containing
compounds is a growing concern everywhere in the world.
Nitrate is one of the most problematic and widespread of the
vast number of potential groundwater contaminants. Sources
of nitrate in groundwater can be considered in four cate-
gories: (i) natural sources, (ii) waste materials, (iii) row crop
agriculture, and (iv) irrigated agriculture [1].
1
1, 12, 13, or 14. Mostly, copper was used as a promoting
Biological and physicochemical treatments allow effec-
tive removal of nitrates but have several economical and eco-
logical disadvantages. Catalytic reactions constitute promis-
ing approaches for the destruction of pollutants in water.
Since the first paper of Tacke and Vorlop [2] on the use of
palladium–copper bimetallic catalysts for nitrate reduction,
numerous studies have been aimed at the development of
suitable catalysts for the selective reduction of nitrate into ni-
trogen gas [2–26]. Their application in various reactor types
such as suspended-bed reactor, monolithic reactor [20], hol-
low fibers [17], and membrane reactors [22] has been stud-
ied.
second metal [2–26], but other suitable promoters, such as
tin, indium, or zinc [5–7,17,19,21,22,25] have demonstrated
interesting performances. Whereas in all cases the second
metal promotes the activity, it leads to contrasting results for
the selectivity.
While many studies were carried out to enhance the
activity and the selectivity toward nitrogen formation, the
changes in the catalytic behavior as a function of the second
metal are still not fully understood and a general reaction
mechanism has not been proposed yet.
Recently, the behavior of platinum catalysts promoted by
copper deposition has been studied for nitrate reduction in
the liquid phase [26]. The proposed mechanism involves a
redox reaction between metallic copper and nitrate, leading
to intermediate nitrite or directly to nitrogen gas or ammo-
*
Corresponding author.
E-mail address: florence.epron.cognet@univ-poitiers.fr (F. Epron).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00252-5

F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
183
nium ion, and to an oxidized form of copper. In this way,
the role of the precious metal is to activate hydrogen, allow-
ing the reduction of oxidized copper to prevent deactivation.
Concerning intermediate nitrite, it can be reduced either on
copper according to a redox reaction or on precious metal by
a catalytic reaction.
With the aim of gaining a deeper understanding of the
reaction mechanism, we have studied the effect of the
promoter on the activity and selectivity of platinum and
palladium-based catalysts. For that purpose, metals of group
of a cationic exchange between the precursor salt and the
alumina surface possible at high pH. After evaporation of
water, catalysts were dried at 120 C, calcinated in flowing
◦
◦
air at 450 C for 4 h, and reduced by flowing hydrogen at
◦
◦
500 C for platinum and 300 C for palladium.
Monometallic catalysts Pt/γ -Al2O3 and Pd/γ -Al2O3 had
a metal loading of 3 wt% and 1.6 wt%, respectively. These
two different metal loadings correspond to equimolar quan-
tities of the precious metal. In the following, these cata-
lysts are identified by their respective notation: Pt3Al and
Pd1.6Al.
1
1 (Cu, Ag, Au) have been chosen.
Generally, bimetallic catalysts are prepared either (i) by
loading the support simultaneously with both metal precur-
sors (deposition–precipitation or impregnation) or (ii) by
modifying a parent monometallic catalyst by impregnating
the promoter precursor (successive impregnation). Under
these conditions, the promoter is randomly deposited on the
support and/or on the noble metal. Previous studies [26,27]
showed that controlled surface reactions (catalytic reduction
or direct redox reaction) constitute a way of catalyst prepa-
ration bringing the two metals in close contact, with a spe-
cific orientation of the second metal deposit on the first one,
which is different from classical methods described above.
Moreover, in the case of Pt-Cu/γ -Al2O3 bimetallic catalysts,
it was demonstrated that the preparation methods involved
different behaviors in nitrate abatement. Precisely, the prepa-
ration method by catalytic reduction led to the most active
catalysts compared to those resulting from classical methods
such as coimpregnation or successive impregnation [26].
This paper describes the performance for nitrate and ni-
trite removal and characteristics of various bimetallic cat-
alysts, supported on alumina, involving a metal of group
2.1.2. Bimetallic catalysts
Bimetallic catalysts were prepared by a controlled surface
reaction. In order to obtain great metal–metal interactions,
the second metal was deposited by a redox reaction occur-
ring between the hydrogen adsorbed on the prereduced noble
metal and the oxidized modifier.
A known amount of a monometallic catalyst was placed
in an atmospheric glass reactor. The reactor was flushed
with a countercurrent nitrogen flow for 15 min at room
temperature. Afterward the catalyst was reduced by a hy-
◦
drogen stream at a temperature rising to 400 C and then
cooled to room temperature. At the same time, the desired
amount of the modifier was predissolved in water and de-
gassed by bubbling with N2. The solution was then added
to the monometallic catalyst and maintained under stirring
by hydrogen flow for 2 h. The bimetallic catalyst was sep-
arated from the aqueous solution via a filter in the bottom
of the reactor, and washed. Therefore, the resulting material
◦
was dried overnight under a nitrogen stream at 70 C.
Each bimetallic catalyst, made up of a precious metal (M)
and a promoter (X), was identified by its specific atomic ratio
x (at.%) in the metallic phase corresponding to the number
of atoms of X divided by the total number of atoms (M+X).
The catalyst was noted MXxAl.
The experimental nominal composition of each catalyst
was verified by elementary analysis. Generally, the experi-
mental value gave an incertitude error not exceeding 5%.
1
1 and prepared by catalytic reduction. The character-
istics of mono and bimetallic particles were studied by
temperature-programmed reduction (TPR), infrared spec-
troscopy (FTIR) of chemisorbed CO, transmission electron
microscopy (TEM) coupled with energy dispersive spec-
trometry (EDS) and H2 chemisorption. They were tenta-
tively related to metal–metal interactions, resulting from the
preparation steps, and to the catalytic activity for nitrate re-
duction.
2.2. Characterization methods
2
.2.1. Metal accessibility
2
2
2
. Experimental
H2 chemisorption measurements were carried out using
a conventional volumetric apparatus equipped with a turbo-
molecular pump. Prereduced samples were reduced again in
flowing hydrogen at their reduction temperature. Then, they
were outgassed at this temperature until a vacuum below
1 Pa was achieved, in order to obtain metal free of hydro-
gen, and after that cooled to room temperature. Hydrogen
chemisorption was conducted at room temperature for plat-
.1. Catalyst preparation
.1.1. Monometallic catalysts
A powdered γ -alumina from Procatalyse (surface area
2
3
(
BET method), 216 m /g; pore volume, 0.55 m /g; isoelec-
tric point, 7) has been used as support. First, it was ground
and then sieved to retain particles with sizes between 0.04
and 0.08 mm.
Monometallic catalysts were prepared by an impregna-
tion method using aqueous solutions of the precursor salts:
Pd(NH3)4(NO3)2 and Pt(NH3)4(OH)2. This step consisted
◦
inum catalysts, whereas it was operated at 70 C for palla-
dium catalysts to avoid the formation of β-PdH [28,29]. The
linear portions of both the total and the reversible chemisorp-
tion isotherms were extrapolated to zero pressure to estimate
the corresponding uptakes. The fraction of exposed noble

184
F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
metal atom was taken as (H/M)irr, assuming that the surface
stoichiometry of the irreversibly adsorbed hydrogen is unity.
nitrite or nitrate, and of their selectivity toward ammonium
ions, determined at the end of the reaction.
2
.2.2. TEM measurements
2.4. Analysis
TEM measurements were carried out with a Philips
CM120 electron microscope operating at 120 kV with a
resolution of 0.35 nm. The powder was ultrasonically dis-
persed in ethanol, and the suspension was deposited on
an aluminum grid coated with a porous carbon film. The
particle-size distribution was obtained from TEM pictures
calculating the surface average particle diameter from dp =
The reaction components were analyzed by the HPLC
method. Nitrate and nitrite concentrations were determined
after separation on a Zorbax Eclipse XDB-C18 column us-
ing an UV detector at λ = 210 nm. Ammonium ions were
quantified using an Alltech Universal cation column coupled
with a conductivity detector. The acidic mobile phase (oxalic
acid) used provided the complete conversion of the ammonia
basic form into ammonium ions.
ꢀ
ꢀ
3
nid / nidi.
i
X-ray energy dispersive spectrometry (EDX) was per-
formed with the STEM mode of the microscope using a
Si-Li Super UTW detector.
3
. Results and discussion
2
.2.3. FTIR spectroscopy
Spectra were recorded on a Perkin-Elmer spectrometer
3
.1. Monometallic catalysts
−1
from Nicolet in the 4000–1000 cm range with a resolution
−1
set at 4 cm . Samples were pressed in order to obtain thin
disks which were directly introduced in the IR cell and sub-
jected to different thermal and chemical treatments, namely
outgassing and reduction. The spectrum of the reduced sam-
ple was used as the reference and the plot function was set
in absorbance mode.
The metal accessibility of the monometallic catalysts,
Pt/γ -Al2O3 and Pd/γ -Al2O3, was determined immediately
after preparation (fresh catalyst) and also after a blank test
(
i.e., under flowing hydrogen in water) in order to discrimi-
nate between the modifications induced by the preparation
conditions of the bimetallic catalyst and the actual effect
of the second metal. A sintering phenomenon occurred in
water under hydrogen flowing which was expressed by a
decrease of the metal accessibility determined by hydrogen
volumetric chemisorption: from 61 to 52% for Pt/γ -Al2O3
and from 80 to 13% for Pd/γ -Al2O3. So the catalytic per-
formances for nitrite reduction were determined after ageing
in water under flowing hydrogen. Catalytic properties and
particle size of the monometallic catalysts are summarized
in Table 1. These results demonstrate that the platinum ac-
tivity is lower and its selectivity toward ammonium ions is
higher compared to palladium. These results are quite in
agreement with the first investigations reported in the liter-
ature [3]. The lower selectivity toward ammonium ions of
monometallic palladium could be explained by the less no-
ble character of palladium compared with that of platinum,
that is, unfavorable to the deep reduction of nitrogen species
into ammonium ions. This selectivity difference could also
The catalysts were reduced in flowing hydrogen at the re-
◦
◦
duction temperature for 1 h (500 C for platinum and 300 C
for palladium) and evacuated at the same temperature for 1 h
in order to remove adsorbed hydrogen over the surface of
metal particles. Carbon monoxide was introduced in the cell
at room temperature by injecting pulses until saturation of
the metal surface. Then, the samples were evacuated at room
temperature for 1 h. The spectra presented in the present
study were obtained by difference between the absorbance
of the sample measured under vacuum before and after ad-
sorption of the probe molecule.
2
.3. Nitrate and nitrite reduction
Nitrate and nitrite reduction reactions were performed in
◦
a semibatch reactor, at atmospheric pressure and 10 or 25 C.
In a typical experiment, 200 mg of the powdered catalyst,
with a particle diameter between 40 and 80 µm, was placed
in 90 ml of ultrapure water (18.2 Mꢀ), purged with nitro-
gen and then with hydrogen. Afterward, 10 ml of a solution
Table 1
−1
Initial activity and final selectivity toward ammonium ions for nitrite
reduction, and particle size of platinum and palladium catalysts supported
on alumina
(
(
16 mmol L ) of nitrate (KNO3 or Mg(NO3)2) or nitrite
a
KNO2 or Ba(NO2)2) was introduced in the reactor to start
the reaction. The catalyst dispersion in the aqueous medium
Monometallic Metal loading
d
d
TOF
Selectivity
(mol%)
was achieved by the hydrogen flow (pH = 1 bar, flow rate =
2
b
c
−3 −1
catalyst
(wt%)
(Å)
(Å)
(10
s
)
−1
2
50 ml min ) through a porous glass located at the bottom
Pt3Al
3
16
72
19
59
29
230
67
39
of the reactor. To monitor the progress of the reaction, repre-
sentative aqueous samples were periodically withdrawn and
immediately separated by filtration and then analyzed as de-
scribed in Section 2.4.
Catalysts were compared as a function of their initial
activity, corresponding to the initial disappearance rate of
Pd1.6Al
a
1.6
◦
Nitrite source, Ba (NO ) ; reaction temperature, 25 C.
2
2
b
Average particle size after ageing in water under hydrogen flow deter-
mined by H chemisorption.
2
c
Average particle size after ageing in water under hydrogen flow deter-
mined by TEM.

F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
185
Table 2
Redox potentials of metals of group 11 and principal nitrogen species in-
volved in the reaction as intermediates or final products [30,31]
0
Redox systems
E
(V/ERH)
2
+
Cu /Cu
0.34
0.799
1.52
0.835
0.88
1.25
+
Ag /Ag
3
+
Au /Au
−
−
+
NO /NO
3
2
4
−
NO /NH
3
−
NO /N
3
2
be explained by the difference of particle size of the two
monometallic catalysts (see 3.2.2.2b).
3
3
.2. Bimetallic catalysts
.2.1. Pd-Au/γ-Al2O3 catalyst
Two Pd-Au/γ-Al2O3 catalysts were prepared from the
parent catalyst Pd1.6Al containing 10 and 50 at.% of gold in
the bimetallic phase. The catalytic experiments performed
in the presence of these materials demonstrated that Pd-
Au/γ -Al2O3 is inactive for nitrate reduction. This could be
explained by the inability of gold to reduce nitrate by a
direct redox process, according to the redox potentials of
Fig. 1. TPR profiles of γ -Al O -supported monometallic catalysts and their
2
3
corresponding bimetallic promoted by copper addition: (a) oxidized Pt3Al,
b) oxidized Pd1.6Al, (c) fresh PtCu50Al, (d) fresh PdCu50Al, and (e) oxi-
dized Cu1Al.
(
the Au3 /Au couple (Table 2) and of the possible nitrogen
species involved in the reaction. This hypothesis was con-
firmed by testing a monometallic 10 wt.% Au/γ -Al2O3 cat-
alyst after an in situ prereduction, which was totally inactive
for nitrate as well as for nitrite reduction under nitrogen or
hydrogen atmosphere. The inactivity toward nitrite reduction
under hydrogen could be surprising since gold has a higher
noble character than palladium, but it cannot chemisorb hy-
drogen at ambient temperature [32] that could account for
this result.
+
3
.2.2. Bimetallic catalysts promoted by copper or silver
addition
Contrary to gold, the addition of copper or silver to Pt3Al
and Pd1.6Al leads to active catalysts for nitrate and nitrite
reduction. In the following, Pt–Ag, Pt–Cu, Pd–Ag, and Pt–
Cu catalysts will be characterized by TPR, FTIR, TEM, and
XRD and then their activity and selectivity for nitrate and
nitrite reduction will be examined.
Fig. 2. TPR profiles of γ -Al O -supported monometallic and their cor-
responding bimetallic catalysts promoted by silver addition: (a) oxidized
Pt3Al, (b) oxidized Pd1.6Al, (c) fresh PtAg50Al, (d) fresh PdAg50Al, and
2
3
3
.2.2.1. Characterization Pt–Ag, Pt–Cu, Pd–Ag, and Pt–
Cu supported on alumina with a same atomic ratio of
promoter in the metal phase (50 at.%) were characterized
by TPR, FTIR of adsorbed CO, and TEM coupled with
EDS. Their respective notation are the following: PtAg50Al,
PtCu50Al, PdAg50Al, and PdCu50Al.
(e) oxidized Ag1.7Al.
The TPR profiles of the bimetallic catalysts were com-
pared to those of the corresponding monometallic catalysts
◦
3
.2.2.1.a. TPR. Temperature-programmed reduction pro-
determined after oxidation under air at 400 C (Figs. 1
files of bimetallic catalysts were established after prepara-
tion and exposition to air at ambient temperature (fresh cata-
lyst). Indeed, it was demonstrated that an oxidation at high
temperature of a bimetallic catalyst can lead to the segrega-
tion of the two metals [33].
and 2). The temperatures of the maximum hydrogen uptake
rate (TM) for the different metal oxides are listed in Table 3.
The TPR profiles of the oxidized Pt3Al (Fig. 1, a) and
Pd1.6Al (Fig. 1, b) catalysts show a hydrogen consumption
from ambient temperature and are characterized by a sin-

186
F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
Table 3
Temperature (TM) at the maximum of reduction rate for alumina-supported
monometallic catalysts
◦
Metal oxides
TM ( C)
PdOx
PtOx
CuO
∼ 25
70
310
Cu O
2
420
AgO
Ag O
2
∼ 100
∼ 200
gle peak. Concerning 1 wt% Cu/γ -Al2O3 catalyst, named
Cu1Al, (Fig. 1, e), its profile is characterized by a hy-
drogen consumption at higher temperature with two peaks
◦
at 300 and 400 C. These peaks could be respectively as-
signed to superficial Cu2O and CuO reduction to metallic
copper [33]. In the case of 1.7 wt% Ag/γ -Al2O3, named
Ag1.7Al, (Fig. 2, e), the TPR profile is composed of two
◦
peaks at 110 and 220 C, corresponding to the reduction of
AgO and Ag2O [34].
As far as the TPR profiles of bimetallic catalysts are con-
cerned (Figs. 1 and 2), they all present an important hydro-
gen consumption whereas they were not preoxidized at high
temperature, that underlines a nonnegligible oxidation of the
bimetallic catalysts at ambient temperature.
Fig. 3. FTIR spectra of CO adsorbed at room temperature onto monometal-
lic Pt3Al and Pd1.6Al and bimetallic PtAg50Al and PdAg50Al catalysts
(after saturation of the surface with CO followed by evacuation for 1 h at
room temperature): (a) Pd1.6Al, (b) PdAg50Al, (c) Pt3Al, (d) PtAg50Al.
On the TPR profile of the PtCu50Al catalyst (Fig. 1, c),
3
.2.2.1.b. FTIR. Infrared spectroscopy of chemisorbed
◦
an important hydrogen uptake around 150 C followed by a
CO was principally used in order to determine the loca-
tion of the promoter onto the noble metal particles. This
study was carried out by comparing IR spectra of freshly
prepared bimetallic catalysts and blank monometallic cata-
lysts, that means treated in water under hydrogen flow. On
Fig. 3 are seen the IR spectra of the bimetallic catalysts pro-
moted by silver and of the monometallic parent catalysts.
The IR spectrum recorded on the blank Pd1.6Al catalyst
(Fig. 3, a) shows two main peaks of adsorbed CO in the
◦
slight hydrogen uptake between 250 and 400 C is identified.
The first reduction peak situated between those of pure plat-
inum and copper catalysts indicates the reduction of mixed
Pt–Cu oxidized species where both metals are in close con-
tact. The peak at higher temperatures could be attributed to
oxidized copper species not interacting with the platinum.
From the TPR profile of PdCu50Al catalyst (Fig. 1, d),
one can observe an important hydrogen consumption char-
−1
−1
acterized by two maximum temperatures (at TM = 75 and
1750–2150 cm range. The band at ca. 2085 cm corre-
sponds to linearly bound adsorbed CO while the band at ca.
1950 cm is attributed to bridged CO [35,36]. Otherwise,
◦
1
4
35 C) and a wide hydrogen consumption between 200 and
◦
◦
−1
00 C. The first reduction peaks at 75 and 135 C are shifted
toward higher temperatures compared to monometallic pal-
ladium catalyst, demonstrating the presence of an interac-
tion between palladium and copper in the metallic particles.
the IR spectrum of the blank Pt3Al catalyst (Fig. 3, c) is com-
−1
posed of only one band at 2070 cm which is characteristic
of linear CO species [37], the bridged form of CO adsorbed
◦
−1
The hydrogen consumption from 200 C results from the re-
onto platinum (near 1850 cm ) being negligible [38]. The
duction of isolated copper particles as it has been already
observed for PtCu50Al catalyst.
comparison of the different IR spectra points out that the de-
position of silver onto the monometallic Pd1.6Al and Pt3Al
catalysts leads to a shift to lower wavenumbers of the stretch
frequency of CO adsorbed on the noble metal. This result
bears out that the silver deposit induces an effect of dilu-
tion of the particles of precious metal. In other words, it
indicates an interaction between the noble metal and the pro-
moter into the bimetallic catalysts. The infrared spectra of
the monometallic catalysts modified by the addition of cop-
per have the same profile that also underlines the presence
of interactions between both species of the metallic phase.
In accordance with previous works, the spectra corre-
sponding to the CO adsorption on Pd and Pt monometallic
catalysts can be decomposed into five elementary adsorp-
Considering now the TPR profiles of catalysts promoted
by silver PtAg50Al and PdAg50Al reported in Fig. 2, only
one peak of thermoreduction is identified, located at a higher
temperature than that of the corresponding pure noble cata-
lyst. Therefore, it may be assumed that this peak corresponds
to the coreduction of both metals.
All the TPR profiles of the different bimetallic catalysts
have proved the presence of metal–metal interaction. It leads
one to suppose that bimetallic entities are present in all these
bimetallic catalysts. This result is rather satisfying since the
preparation method favors the deposition of the promoter
preferentially onto the noble metal particles.

F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
187
tion bands in the case of palladium crystallites [39,40] and
only three elementary bands for platinum [41]. The carbonyl
bands of adsorbed CO on the noble metal were deconvo-
luted using spectra adjustment software (Peakfit). The de-
convolution of the νCO bands was realized assuming each
elementary adsorption band as a gaussian curve. Knowing
the surface of each elementary band forming the IR spec-
tra for all catalysts, the contribution to the total absorbance
of the two types of adsorption of CO (linear noted L and
bridged noted B) can be calculated for each palladium-based
catalyst. Whereas linear species L are mainly considered to
correspond to CO adsorbed on low coordinated palladium
atoms (edges, corners etc.), the bridged species B are situ-
ated on Pd(100) or Pd(110) and on Pd(111) [42–45]. Results,
reported in Table 4, show that for palladium-based catalysts
the proportion of linear species is increased when Cu or Ag
is added to the detriment of the bridged species compared
to the monometallic palladium-supported catalyst. This phe-
nomenon appears independently of the nature of the second
metal, although it is more pronounced in the case of sil-
ver. Moreover, this behavior is in accordance with the one
we could attempt considering the preparation method, the
Fig. 4. Schematic representation of the localization of the promoter on the
palladium or platinum particle according to the FTIR results.
where Hads is adsorbed hydrogen on the precious metal (Pt
n+
or Pd), M is cation of the second metal, and Mads is ad-
sorbed second metal.
Since the silver precursor is a monovalent cation whereas
the one of copper is bivalent, it can be assumed that the
preferential sites for the second metal deposit are different.
Indeed, copper requires two hydrogen atoms chemisorbed
on adjacent metallic sites to be deposited whereas the de-
position of silver presents no constraint. Therefore, copper
could be preferentially deposited on the terrace sites what-
ever the nature of the noble metal, as schematized on Fig. 4.
This type of deposit could lead to the creation of defects
on the surface of the noble metal, which would contribute
to the adsorption of CO as linear species instead of bridged
species. As far as silver is concerned, it could be deposited
on the terrace sites as well as on the edge and corner sites of
the crystallites of platinum or palladium (Fig. 4).
“
catalytic reduction,” which favors the deposition of the sec-
ond metal (Cu or Ag) onto the precious metal (Pd). Conse-
quently, the probability of the presence of two adjacent sites
able to adsorb carbon monoxide as bridged species is de-
creased. This indicates that the deposition of the promoter
could occur preferentially on the terrace sites.
3
.2.2.1.c. TEM and EDX. TEM pictures coupled with
For platinum-based catalysts only the linear species are
quantified. Results, reported in Table 4, demonstrate that
apparently copper induces a weak variation of the quan-
tity of sites suitable for absorption of carbon monoxide as
linear species corresponding to the edge and corner sites
of platinum crystallites. This effect is not observed for the
catalyst modified by silver, that could evidence different de-
position sites depending on the second metal. The method
used to deposit the second metal onto the noble metal by
catalytic reduction could account for this different behavior.
This method consists of a surface redox reaction between
chemisorbed hydrogen and the second metal according to
the overall reaction:
EDS were carried out to collect more information about the
homogeneity of the particles in terms of particle size and
of composition. The TEM pictures and particle-size distrib-
utions of the four different bimetallic catalysts (PtCu50Al,
PdCu50Al, PtAg50Al, and PdAg50Al) are illustrated in
Figs. 5 and 6, respectively.
If we refer to the particle size of the monometallic cata-
lysts (Table 1), we can see that the average particle sizes of
the monometallic parent catalysts are practically not modi-
fied by the addition of copper, whereas the deposition of sil-
ver leads to an important increase of the average diameters.
Moreover, the comparison of the particle-size distributions
(
Fig. 6) demonstrates that bimetallic catalysts containing sil-
ver are heterogeneous with sizes varying between 10 and
around 200 nm.
nHads + Xn+ → Xads + nH+,
Table 4
Surfaces (L and B) and ratio (L/(L+B)) of the IR bands corresponding to bridged and linearly bonded CO on monometallic catalysts (Pt and Pd) and bimetallic
catalysts (Pt-X and Pd-X with X = Ag or Cu)
Catalyst
Linear carbonyl species L
Bridged carbonyl species B
Surface ratio
Frequency (ν¯)
Surface (a.u.)
Frequency (ν¯)
Surface (a.u.)
L/(L+B)
0.089
0.286
0.800
Pd1.6Al
PdCu50Al
PdAg50Al
2083
2065
2043
0.34
0.36
1.12
1965
1981
1951
3.48
0.90
0.28
Pt3Al
PtCu50Al
PtAg50Al
2075
2060
2025
8.3
10.6
8.1
–
–
–
–
–
–
–
–
–

188
F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
Fig. 5. TEM pictures of (a) PtCu50Al, (b) PdCu50Al, (c) PtAg50Al, and (d) PdAg50Al.
EDX analysis has demonstrated that the composition of
With the aim of gaining a deeper understanding of this
type of catalytic system, the influence of the promoter con-
tent on the catalytic properties was studied for four types of
catalysts, Pt–Ag, Pt–Cu, Pd–Ag, and Pd–Cu supported on
alumina.
3.2.2.2.a. Effect of promoter content in the metal phase on
the activity. The promoter content in the bimetallic catalysts
was varied, maintaining constant the precious metal loading
around 80% of the analyzed individual particles corresponds
to the average composition of the whole catalyst, except for
PtAg50Al. For this catalyst, bimetallic particles were com-
posed either of 20 at.% or of 80 at.% of Ag, and some
monometallic silver particles were found isolated on the sup-
port.
(
0.15 mmol/gcat). The activities of the different bimetallic
3
.2.2.2. Nitrate reduction It is well known that copper is
catalysts are presented in Fig. 8 as a function of the promoter
proportion in the bimetallic phase. This figure shows that,
for both types of noble metal-based catalysts, the activity
for nitrate reduction appears to go through a maximum for
a promoter content between 33 and 50 at.%, depending on
the catalyst. The optimal composition and the corresponding
activity are reported in Table 5 as a function of the bimetal-
lic catalyst. Except for the bimetallic PdCuAl catalyst, for
which the maximum of activity is reached for 33 at.% of
copper, in agreement with results of the literature [3,26], the
optimal content of the promoter is 50 at.% in the bimetal-
lic phase. The maximum of activity of the four bimetallic
systems is quite the same, between 0.13 and 0.19 mmol of
nitrate converted per minute and gram of catalyst. This max-
imum of activity is reached when the optimal concentrations
of bimetallic entities, made up of a promoter, able to re-
duce nitrate, and of a noble metal, able to activate hydrogen
to maintain the promoter in the metallic state, are simul-
taneously present in the catalyst. In conclusion, the cata-
a good promoter for nitrate reduction whereas in the case of
silver, published investigations have reported a weak activ-
ity. In Fig. 7 are displayed the evolution of nitrate and inter-
mediate nitrite concentrations in the presence of PdCu50Al
and PdAg50Al. This figure underlines that, contrary to the
results of the literature, silver-promoted noble metal cat-
alysts can have a higher activity than classical copper-
promoted catalysts, when bimetallic catalysts are prepared
+
by catalytic reduction. As the redox potential of the Ag /Ag
couple is lower than that of nitrogen species involved in the
reaction, we can suggest that, according to our previous pro-
posed mechanism [26,46] metallic silver is able to reduce ni-
trate yielding to nitrite formation. As we have demonstrated
that silver is deposited on the noble metal, that is favored by
the preparation method, the enhancement of the activity of
the present Pd–Ag catalyst compared to the results of the lit-
erature could be explained by the ability of the noble metal
to maintain silver in the metallic state under hydrogen.

F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
189
Fig. 8. Initial activity for nitrate reduction as a function of the promoter
content in the metal phase (X/(M + X) at.%) for bimetallic catalysts pro-
moted by copper or silver: PtCuAl ("), PtAgAl (2), PdCuAl (F), and
◦
PdAgAl (Q) (T = 10 C, nitrate source Mg(NO ) ).
3
2
Table 5
Initial activity and final selectivity toward ammonium ions of bimetallic
catalysts (Pt-X and Pd-X with X = Ag or Cu) with the optimal composition
◦
for nitrate reduction (T = 10 C, nitrate source Mg(NO ) )
3
2
Catalyst
Promoter content
Activity
(mmol/(min gcat))
Selectivity
(%)
(at.%)
PtCuAl
PdCuAl
PtAgAl
PdAgAl
50
33
50
50
0.19
0.13
0.15
0.17
75
60
65
56
lysts leading to the higher activity are PtCu50Al, PdCu33Al,
PtAg50Al, and PdAg50Al.
3
.2.2.2.b. Selectivity toward ammonium ions. The se-
lectivity toward ammonium ion formation was determined
for the four optimized catalysts PtCu50Al, PdCu33Al,
PtAg50Al, and PdAg50Al. Results are reported in Table 5.
According to this table, bimetallic catalysts show high selec-
tivities toward ammonium ions, 56 to 75% of nitrate being
reduced into ammonium ions, depending on the noble metal
and the promoter.
Fig. 6. Histograms of particle sizes of (a) PtCu50Al, (b) PdCu50Al, (c)
PtAg50Al, and (d) PdAg50Al determined from TEM pictures.
Several parameters could explain (i) the high selectivities
and (ii) the slight differences in selectivities as a function of
the bimetallic catalysts. It has been demonstrated that the pH
in the vicinity of the active sites strongly affects the selectiv-
ity toward ammonium ions [6,12,47,48]. Consequently, as
hydroxide ions are produced during nitrate and nitrite reduc-
tion, the porosity of the support and the pH of the solution
during the reaction play a role of major importance in the
ammonium ion formation. In the present study, nitrate reduc-
tion was performed without pH control and the same value
of pH 11 was reached at the end of the reaction. Then, the
pH of the solution can explain the high selectivities toward
ammonium ions, observed for all the catalysts. On the other
hand, the same support, a γ -alumina with a BET surface of
2
3
2
15 m /g and a total pore volume of 0.55 cm /g, was used
Fig. 7. Nitrate (") and nitrite (Q) concentration vs time in the presence
◦
of PdCu50Al (—) or PdAg50Al (- - -) catalysts (T = 10 C, nitrate source
to prepare all the catalysts and it was verified that the poros-
ity is not affected by the addition of the metals. Then the
Mg(NO ) ).
3
2

190
F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
Table 6
silver is a good promoter for this reaction, bimetallic cata-
lysts having an activity of the same order as those containing
copper. The promoting effect of these metals can be related
to their ability to reduce nitrate according to a redox re-
action under the conditions of the reaction. Gold, which is
nobler than palladium or platinum, is not able to achieve this
process, contrary to silver and copper. Then, the proposed
mechanism for nitrate reduction onto Pt-Cu/γ -Al2O3 cata-
lysts [26] can be generalized to whole systems associating a
noble metal and an oxidizable promoter.
In conclusion, the present study confirms that nitrate re-
duction occurs according to a bifunctionnal mechanism in-
volving a direct redox mechanism on the promoter followed
by a classical catalytic reaction on the noble metal. A sys-
tem active for nitrate reduction is necessarily composed of
Initial activity of monometallic catalysts (Pt and Pd) and bimetallic catalysts
◦
(
Pt-X and Pd-X with X = Ag or Cu) for nitrite reduction (T = 10 C, nitrite
source Ba(NO ) )
2
2
Catalyst
Promoter content (at.%)
Activity
mmol/(min gcat))
(
Pt3Al
–
–
50
50
50
50
0.04
0.060
0.058
0.061
0.02
Pd1.6Al
PtCuAl
PdCuAl
PtAgAl
PdAgAl
0.032
differences of selectivities, as a function of the bimetallic
catalysts, cannot be related to the porosity. Other parameters
such as the particle size, the electronic, and/or by a geo-
metric effect induced by the promoter could explain these
changes.
(
i) one promoter easily oxidizable and reducible in the reac-
tion conditions and (ii) one noble metal able to chemisorb
hydrogen.
3
.2.2.3. Nitrite reduction Nitrite reduction was performed
in the presence of the four bimetallic systems Pt–Ag, Pt–
Cu, Pd–Ag, and Pt–Cu supported on alumina, containing the
same atomic ratio of the promoter (50 at.% in the bimetallic
phase). Their activity, compared with that of the monometal-
lic equivalents, is reported in Table 6. Contrary to the reduc-
tion of nitrate, the reduction of nitrite proceeds rapidly on
monometallic catalysts. The activity of bimetallic catalysts
demonstrates that the reaction rate is slightly increased by
the addition of copper, whereas it is strongly decreased by
the addition of silver. This difference of activity, probably
due to the inactivity of silver for nitrite reduction contrary
to copper, explains the higher intermediate nitrite concen-
tration during nitrate reduction in the case of silver as pro-
moter (see Fig. 8). Moreover, for a given promoter, modified
Pd/γ -Al2O3 catalysts show higher activity than modified
Pt/γ -Al2O3 catalysts. This is consistent with the activity of
the parent monometallic catalysts (Pd/γ -Al2O3 and Pt/γ -
Al2O3) for nitrite reduction.
Acknowledgment
F.G. gratefully acknowledges La Région Poitou-Charen-
tes for research fellowships.
References
[1] L.W. Canter, Nitrates in Groundwater, CRC Press, Boca Raton, FL,
1996.
[2] T. Tacke, K.-D. Vorlop, Dechema Biotechnology Conferences, Vol. 3,
VCH, Weinheim, 1989, p. 1007.
[3] S. Hörold, T. Tacke, K.D. Vorlop, Environ. Technol. 14 (1993) 931.
[4] S. Hörold, K.D. Vorlop, T. Tacke, M. Sell, Catal. Today 17 (1993) 21.
[5] U. Prüsse, S. Hörold, K.-D. Vorlop, Chem.-Ing.-Tech. 69 (1997) 93.
[
6] U. Prüsse, M. Hahnlein, J. Daum, K.D. Vorlop, Catal. Today 55 (2000)
9.
7] U. Prüsse, J. Daum, C. Bock, K.D. Vorlop, Stud. Surf. Sci. Catal. 130
2000) 2237.
[8] U. Prüsse, K.-D. Vorlop, J. Mol. Catal. 173 (2001) 313.
9] A. Pintar, J. Batista, J. Levec, T. Kajiuchi, Appl. Catal. B 11 (1996)
1.
7
[
(
[
8
[
[
[
[
[
[
10] A. Pintar, M. Setinc, J. Levec, J. Catal. 174 (1998) 72.
11] A. Pintar, J. Batista, J. Levec, Water Sci. Technol. 37 (1998) 177.
12] A. Pintar, J. Batista, Catal. Today 53 (1999) 35.
13] A. Pintar, J. Batista, J. Levec, Chem.-Eng.-Sci. 56 (2001) 1551.
14] A. Pintar, J. Batista, J. Levec, Catal. Today 66 (2001) 503.
15] F. Deganello, L.F. Liotta, A. Macaluso, A.M. Venezia, G. Deganello,
Appl. Catal. B 24 (2000) 265.
4
. Conclusion
The modification of a parent monometallic Pd/γ -Al2O3
or Pt/γ -Al2O3 catalyst by deposition of metal of group 11
Cu, Ag, Au) using the preparation method called catalytic
(
reduction leads to bimetallic catalysts with high metal–metal
interactions, as proved by TPR and TEM coupled with EDX
and FTIR. As a function of the promoter, the characteriza-
tion by FTIR of adsorbed CO and by TEM coupled with
EDX has revealed different morphologies. The addition of
copper leads to homogeneous bimetallic particles with a nar-
row size distribution, copper being deposited selectively on
terrace sites. On the contrary, the deposition of silver is made
at random on all types of sites.
[16] O.M. Ilinitch, L.V. Nosova, V.V. Gorodetskii, V.P. Ivanov, S.N.
Trukhan, E.N. Gribov, S.V. Bogdanov, F.P. Cuperus, J. Mol. Catal.
A 158 (2000) 237.
[
17] M. Hählein, U. Prüsse, J. Daum, V. Morawsky, M. Kröger, M.
Schröder, M. Schnabel, K.-D. Vorlop, in: B. Delmon, et al. (Eds.),
Preparation of Catalysts, Vol. VIII, Elsevier, Amsterdam, 1998, p. 99.
[18] G. Strukul, F. Pinna, M. Marella, L. Meregalli, M. Tomaselli, Catal.
Today 27 (1996) 209.
[19] G. Strukul, R. Gavagnin, F. Pinna, E. Modaferri, S. Perathoner, G.
Centi, M. Marella, M. Tomaselli, Catal. Today 55 (2000) 139.
[20] J. Wärna, I. Turunen, T. Salmi, T. Maunula, Chem. Eng. Sci. 49 (1994)
Gold deposition on a precious metal leads to bimetallic
catalysts inactive for nitrate reduction. On the other hand,
5
76.
[21] A.J. Lecloux, Catal. Today 53 (1999) 23.

F. Gauthard et al. / Journal of Catalysis 220 (2003) 182–191
191
[
22] K. Daub, G. Emig, M.-J. Chollier, M. Callant, R. Dittmeyer, Chem.-
Ing.-Sci. 54 (1999) 1577.
[35] R.P. Eischens, W.A. Pliskin, Adv. Catal. 10 (1958) 1.
[36] L.L. Sheu, Z. Karpinski, W.M.H. Sachtler, J. Phys. Chem. 93 (1989)
4890.
[23] H. Berndt, I. Mönnich, B. Lücke, M. Menzel, Appl. Catal. B 30 (2001)
1
11.
[37] N. Sheppard, Y.T. Nguyen, in: R.J.H. Clark, R.E. Hester (Eds.), In-
frared and Spectroscopy, Vol. 5, Books Demand, Ann Arbor, MI, 1978,
p. 67.
[38] A. Palazov, C.C. Chang, R.J. Kokes, J. Catal. 36 (1975) 338.
[39] C. Binet, A. Jadi, J.C. Lavalley, J. Chim. Phys. 86 (1989) 451.
[40] C. Binet, A. Jadi, J.C. Lavalley, J. Chem. Soc. Farad. Trans. 88 (1992)
2079.
[
[
24] Y. Yoshinaga, T. Akita, I. Mikami, T. Okuhara, J. Catal. 207 (2002) 37.
25] L. Lemaignen, C. Tong, V. Begon, R. Burch, D. Chadwick, Catal. To-
day 75 (2002) 43.
26] F. Epron, F. Gauthard, C. Pinéda, J. Barbier, J. Catal. 198 (2001) 309.
27] C. Micheaud, P. Marecot, M. Guerin, J. Barbier, Appl. Catal. A 171
[
[
(1998) 229.
[
28] V. Ragiani, R. Giannantonio, P. Magni, L. Lucarelli, G. Leofanti,
[41] R.G. Greenler, K.D. Burch, Surf. Sci. 152/153 (1985) 338.
[42] R.F. Hicks, Q.J. Yen, A.T. Bell, J. Catal. 89 (1984) 498.
[43] V. Pitchon, M. Primet, H. Praliaud, Appl. Catal. 62 (1990) 317.
[44] J. Evans, B.E. Hayden, G. Lu, Surf. Sci. 360 (1996) 61.
[45] L.F. Liotta, G.A. Martin, G. Deganello, J. Catal. 164 (1996) 322.
[46] F. Epron, F. Gauthard, J. Barbier, J. Catal. 206 (2002) 363.
[47] U. Prüsse, V. Morawsky, A. Dierich, A. Vaccaro, K.-D. Vorlop, in:
B. Delmon, et al. (Eds.), Preparation of Catalysts, Vol. VII, Elsevier,
Amsterdam, 1998, p. 137.
J. Catal. 146 (1994) 116.
[29] H. Berndt, J. Dembowski, K. Flick, Chem. Technol. 49 (1997) 12.
[30] M. Pourbaix, N. de Zoubov, Atlas d’équilibres électrochimiques à
◦
2
5 C, Gauthier-Villars, Paris, 1963.
[
31] J.T. Maloy, in: A.J. Bard, R. Parsons, J. Jordan (Eds.), Standard Poten-
tials in Aqueous Solution, Dekker, New York, 1985, p. 127.
32] R. Culver, J. Pritchard, F.C. Tompkins, Proc. 2nd Int. Congr. Surf.
Act. 2 (1957) 243.
[
[
[
33] F. Epron, F. Gauthard, J. Barbier, Appl. Catal. A 237 (2002) 253.
34] Q. Zhang, J. Li, X. Liu, Q. Zhu, Appl. Catal. A 197 (2000) 221.
[48] R. Gavagnin, L. Biasetto, F. Pinna, G. Strukul, Appl. Catal. B 38
(2002) 91.