CuO-SiO2 sol-gel catalysts: Characterization and catalytic properties for NO reduction

Article abstract of DOI:10.1006/jcat.1999.2578

CuO-SiO₂ catalysts were prepared by the sol-gel technique and calcined at two temperatures, 673 and 1073 K. Characterization of the catalysts was achieved by BET, FTIR, TPR, dissociative N₂O adsorption, XPS, HRTEM, and reactivity toward NO + H₂. As a function of the calcination temperature, characterization techniques showed: textural (BET), morphological and structural (HRTEM), reducibility (TPR, XPS), dispersion, and catalytic reactivity (NO + H₂) differences between samples. Both samples showed remarkable stability during on-stream experiments. The microstructure of the catalysts as a consequence of the preparation method allows the understanding of the catalytic behavior of calcined and reduced samples as well as the stability toward sintering.

Full text of DOI:10.1006/jcat.1999.2578

Journal of Catalysis 187, 1–14 (1999)
Article ID jcat.1999.2578, available online at http://www.idealibrary.com on
CuO–SiO Sol–Gel Catalysts: Characterization and Catalytic
2
Properties for NO Reduction
1
�
�
G. D ´ı az, R. P e´ rez-Hern a´ ndez, A. G o´ mez-Cort e´ s, M. Benaissa, R. Mariscal, and J. L. G. Fierro
�
Depto. F ´ı sica Qu ´ı mica, Instituto de F ´ı sica UNAM, P.O. Box 20-364, Mexico 01000; and Instituto de Cat a´ lisis
y Petroleoqu ´ı mica, CSIC, Cantoblanco, 28049 Madrid, Spain
Received August 27, 1998; revised May 28, 1999; accepted May 28, 1999
method stabilizes copper species on silica. The Cu 2 clus-
CuO–SiO2 catalysts were prepared by the sol–gel technique ters crystallize to bulk CuO above 973 K, whereas bulk CuO
and calcined at two temperatures, 673 and 1073 K. Characteri- is already obtained at 773 K when copper is impregnated
+
zation of the catalysts was achieved by BET, FTIR, TPR, dissocia-
on silica. The various copper surface species are character-
ized by different reducibility properties. De Jong et al. (4)
observed that the particle size and shape of copper parti-
cles deposited on silica can be modified by repeated redox
tive N2O adsorption, XPS, HRTEM, and reactivity toward NO +
H2. As a function of the calcination temperature, characteriza-
tion techniques showed: textural (BET), morphological and struc-
tural (HRTEM), reducibility (TPR, XPS), dispersion, and catalytic
cycles. van der Grift et al. (5) noted that depending on the
reactivity (NO + H2) differences between samples. Both samples
severity of the reduction treatment, the structure of the
copper particles on silica suffered changes from an atomi-
cally rough surface to a stepped one and finally to a densely
showed remarkable stability during on-stream experiments. The
microstructure of the catalysts as a consequence of the preparation
method allows the understanding of the catalytic behavior of cal-
cined and reduced samples as well as the stability toward sintering. packed surface. These surface changesare reversible aslong
�
c 1999 Academic Press
as the particle size is not increased.
Key Words: CuO–SiO2 sol–gel; characterization by physical and
chemical techniques; NO reduction.
Stabilization ofthe active phase in supported catalystsde-
pends on the support and the preparation method. Among
the preparation methods for copper catalysts, the sol–gel
technique offers some advantages. With this preparation
method, copper species are effectively incorporated into
the ionic oxide network (6) and may lead to more stable
catalyststhan those prepared byconventionalmethods. The
sol–gel polymerization of metal alkoxides in the presence of
inorganic salts leads to the formation of ceramic oxide ma-
terials that exhibit important microstructure properties (7).
Accordingly, this work was undertaken with the aim
of exploring the capability of the sol–gel method for the
preparation of CuO–SiO2 catalysts and to see whether such
a method can yield copper species stabilized by the car-
rier. Characterization of the catalyst structures by several
physical and chemical techniques, including BET measure-
ments, infrared spectroscopy, temperature-programmed re-
duction, dissociative N2O adsorption, X-ray photoelectron
spectroscopy, and high resolution transmission electron mi-
croscopy, has been used to correlate the microstructure of
the solids with their catalytic performance in the NO reduc-
tion by hydrogen.
1
. INTRODUCTION
Nitrogen oxide reduction by catalytic means is one of the
most important reactions under research due to its impor-
tance in the control of atmospheric pollution. Currently,
the catalytic converters in automobiles contain Rh, Pt, and
Pd supported on a ceramic monolith (1); nevertheless the
high demand of rhodium in contrast to its small supply
in nature has led to increased efforts to find alternative
more abundant materials and less expensive replacements.
One of these options is copper-based catalysts. The selec-
tive catalytic reduction of NO over Cu-containing zeolites
has been the subject of intense research in the past years,
driven by the expectations of large-scale applications for
the control of NOx (2).
The catalytic behavior of supported copper oxide is influ-
enced by the nature of the support, the dispersion of copper,
and the surface changes occurring during the catalytic re-
action and redox cycles. The nature of the surface copper
species depends on the loading and calcination tempera-
tures. Shimokawabe et al. (3) reported for silica-supported
copper catalysts prepared by ionic exchange the presence of
2
. EXPERIMENTAL
2
.1. Catalyst Preparation
+2
isolated copper ions and highly dispersed Cu clusters for
Samples were prepared by the sol–gel method from
samples calcined between 673 and 873 K. The ion exchange Si(OC2H5)4 (TEOS) and Cu(CH3COO)2H2O precursors
1
0021-9517/99 $30.00
Copyright �c 1999 by Academic Press
All rights of reproduction in any form reserved.

´
DIAZ ET AL.
2
according to (8). The CuO content was 5.0 mol% . TEOS X-ray radiation source (h� = 1253.6 eV) and a hemispher-
was first partially hydrolyzed at reflux temperature us- ical electron analyzer. The powder samples were pressed
ing NH4OH as the catalyst and ethanol as the solvent. into small stainless steel cylinders and then were outgassed
The molar ratio TEOS/C2H5OH/H2O/NH4OH was at this in the pretreatment chamber of the spectrometer up to
�
9
time 1 :4 :1 :0.33 and there was Cu (CH3COO)2H2O in 10 Torr prior to transfer to the analysis chamber. The Si 2p
the right amount to get the nominal CuO content. After peak with a BE of 103.4 eV was chosen as an internal refer-
1
0 min at reflux temperature, 3 mol of H2O was added to ence (10). This reference was in all cases in good agreement
complete hydrolysis and reflux was continued for another with the BE of the C 1s peak, arising from contamination,
5
0 min. The resultant mixture was aged for 24 h and the at 284.9 eV. This reference gave an accuracy of � 0.1 eV.
residual liquid was then removed by decanting. The xero-
gel (CSSX) was obtained after heating at 353 K for 24 h.
Aliquots of the xerogel were calcined in air at 673 and
2
.3. Catalytic Activity Measurements
The reduction of nitric oxide was studied in the tem-
perature range 373–873 K using a commercial flow sys-
tem (RIG-100, In Situ Research Instruments Inc.) work-
1
073 K for 72 h. Each aliquot was heated at a rate of 5 K/
min from room temperature to 373 K where it was held
for 1 h. The temperature was then increased at the same
rate to reach the calcination temperature (673 or 1073 K).
Hereafter, catalysts will be referred as CSS4 and CSS8 for
samples calcined at 673 and 1073 K, respectively.
ing at atmospheric pressure. The NO/H2 molar ratio was
� 1
1
and 0.5 and the GHSV = 20,000 h . The reactant mix-
ture (balance He 98 vol% ) was fed to the microreactor
at room temperature after the system was purged in He.
The reactor temperature was then increased to reach 373 K
and catalytic activity measurements initiated. For each tem-
perature that was considered, the catalytic activity determi-
nations were done at the steady state. Total activity was fol-
lowed by the NO consumption as a function of the reaction
temperature according to % �NO = {[[NO]in � [NO]out]/
2
.2. Catalyst Characterization
The chemical transformations of the catalyst precursors
brought about by calcination temperature were followed
by FTIR (from treated materials) using a Nicolet-5SX spec-
trometer. Standard procedures were used for sample prepa-
ration; the sample was diluted in KBr (1/1000), crushed in
an agate mortar, and pressed into thin disks. Spectra were
recorded in the 4000 to 300 cm range working with a res-
olution of 4 cm
The TPR experiments and the surface area determina-
tion were done in a multitask characterization unit RIG-
1
[
NO]in}(100), where [NO]in and [NO]out are the NO concen-
tration at the inlet and at the exit of the reactor, respectively.
P
�
1
The selectivity in mol% was calculated as % Si = [Ci/ Ci]
�
1
.
(
% �NO).
The reaction products were analyzed by gas chromatog-
raphy using TCD. A 4-m packed Chromosorb 103 column
was used to separate products coming from the NO + H2 re-
action: N2 and NO were separated at 303 K, and N2O, NH3,
and H2O at 403 K. Calibration for quantitative analysis was
performed using pure gases. Estimated error in activity and
selectivity data is in the range � 5% .
00 from In-Situ Research Instruments Inc. For TPR runs,
experimental conditions were chosen according to the lit-
erature (9). The sample (0.1 g) was placed in the reactor,
purged with He at room temperature, and then heated at
a rate of 10 K/min in the presence of a 5% H2/Ar gas mix-
3
ture (30 cm /min). Total surface area was determined by
the single-point method using a 30% N2/He gas mixture.
3
. RESULTS
The number of surface Cu sites for TON evaluation was
determined by dissociative N2O adsorption at 363 K using Surface Area (BET)
the procedure described by van der Grift et al. (5). The cata-
BET surface areas of the xerogel and the CSS4 and CSS8
samples are compiled in Table 1. Calcination of the xero-
gel (CSSX) at moderately low temperature (673 K) led to
the development of a high surface area in the sample. The
lysts were first reduced at 773 K for 1 h. Afterward, they
were flushed with Ar and temperature was decreased to
3
63 K. Then, N2O was admitted to the reactor at this tem-
3
perature and allowed to flow (50 cm /min) for 45 min. Fi-
nally, the samples were again flushed with Ar and cooled to
room temperature in order to start a TPR run.
TABLE 1
BET Surface Area of CuO–SiO2 Catalysts
The HRTEM microstructure characterization of the
solids was achieved in a JEOL 4000EX electron microscope
and electron micrographs were recorded at a magnification
of 600,000� . Samples were supported on holey carbon grids
by simply grinding the specimen between two glass plates
and bringing the powder in contact with the grid.
2
Sample
Surface area (m /g)
XCSS
CSS4
CSS8
245
312
155
X-ray photoelectron spectra were obtained using a VG
ESCALAB 200R spectrometer equipped with a MgK�
Note. XCSS, xerogel; CSS4, xerogel calcined
at 673 K; CSS8, xerogel calcined at 1073 K.

NO REDUCTION ON CuO2–SiO2 SOL–GEL CATALYSTS
3
FIG. 1. Evolution of the solids as a function of thermal treatments. FTIR spectra of the xerogel (CSSX), the xerogel calcined at 673 K (CSS4), and
the xerogel calcined at 1073 K (CSS8).
surface area of the calcined sample was even higher (by are displayed in Fig. 1. The spectra showed absorption
�
1
about 25% ) than the surface area ofthe xerogel. The surface bands at approximately 1100, 800, and 470 cm , which are
area of silica gels often increases when organic residuals assigned to the different vibration modes of the Si–O bonds
(
precursors and/or nonhydrolyzed alkoxides) are removed in the amorphous SiO2. In the xerogel spectrum (CSSX), an
�
1
during calcination (11). On the other hand, a decrease in the absorption band at 1403 cm is associated with vibrations
surface area was observed as the calcination temperature of C–H bonds probably coming from residual organic com-
increased up to 1073 K.
pounds or incomplete hydrolysis of the alkoxides. The same
holds for the absorption band at around 1540 cm coming
from the asymmetric stretching mode of vibration of the
�
1
Fourier Transform Infrared Spectroscopy (FTIR)
The background infrared spectra of the xerogel (CSSX) –COO bond (12a). These absorption bands decreased
and the catalysts calcined at 673 (CSS4) and 1073 K (CSS8), for the solid calcined at 673 K (CSS4) and practically

´
DIAZ ET AL.
4
disappeared for the solid calcined at 1073 K (CSS8). The
broad absorption band around 3450 cm is due to the over-
TABLE 3
�
1
Binding Energies, Auger Parameter, and Surface Atomic
Ratios of Cu/SiO2 Samples
�
lapping of the stretching mode of OH groups (11) and
�
1
adsorbed water. Similarly, the band near 1630 cm is as-
sociated with the bending mode of OH groups of molecu-
larly adsorbed water (8). It must be emphasized that sam-
ple preparation, including that of the hygroscopic KBr, may
largely contribute to these bands. Two bands are particu-
larly sensitive to the thermal treatment of the sample, the
A
Si 2p
(eV)
Cu 2p3/2
(eV)
�
Pretreatment
CSS4, vac
(eV)
Cu/Si at
0.0170
103.4
103.4
103.4
103.4
103.4
103.4
934.0(74)
—
—
936.2(26)
CSS4, H2a
933.5(81)
934.9(19)
0.0103
0.0087
0.0105
0.0072
0.0064
�
1
one at 560 and the other at 960 cm (stretching), associated
with the presence of nonbonded oxygens Si–O (12b). As
�
b
CSS4, NO + H2
932.8(100)
1851.7
1848.7
the calcination temperature increased, shift of some of the
absorption bands toward high frequencies and the disap- CSS8, vac
933.4(72)
935.6(28)
—
�
1
pearance of others (560 and 960 cm ) was observed, sug-
gesting a rearrangement of the structure of the silica matrix. CSS8, H2a
The vibration of CuO bonds that appear at 575, 500, and
933.1(76)
—
934.6(24)
�
1
b
4
60 cm (13) cannot be observed due to the presence of a
CSS8, NO + H2
932.8(100)
1851.4
1849.1
�
1
broad band at 460 cm from the support. In all spectra a
shoulder around 600 cm was observed and this could be
an indication of Cu (II)–O species as reported in (8).
�
1
Note. Values in parentheses are peak percentages.
Reduction at 523 K for 1 h.
Pretreatment at 723 K for 2 h.
a
b
Temperature-Programmed Reduction (TPR)
The raw TPR profiles and the deconvoluted profiles of
samples CSS4 and CSS8 are shown in Fig. 2. The sample
calcined at 673 K began its reduction at a lower tempera-
ture than the catalyst calcined at 1073 K. The CSS4 sample
showed a well defined peak with two contributions after this
distinct feature. The reduction profile of sample CSS8 also
showed three peaks but with a different distribution with re-
spect to the CSS4 sample. The characteristic temperatures
are summarized in Table 2. The hydrogen consumption in-
dicates incomplete reduction of the copper oxide (Table 2).
Encapsulation of copper is possible with the sol–gel prepa-
ration; therefore, the percentage of reduction may reflect
this situation.
spectra it is clear that the Cu 2p3/2 profile changed drasti-
cally upon reduction with hydrogen at 523 K. The calcined
and outgassed samples showed the principal Cu 2p3/2 line
somewhat above 933.4 eV with a shoulder on the high BE
+
2
side, which is characteristic of Cu species (14a–14c). The
+
2
Cu ions are also easily identified by the satellite peak,
placed around 943 eV, whose origin is complex (15a–15d).
Considering the asymmetry of the Cu 2p3/2 envelope, peaks
of calcined and reduced samples have been deconvoluted
into two contributions centered around 933.5 eV and some-
what above 934.6 eV. The disappearance of the satellite line
and the simultaneous shift of the principal Cu 2p3/2 peak to
lower BE upon H2-reduction at 523 K, showed that after
+
0
this treatment the copper species are reduced Cu or Cu ,
X-ray Photoelectron Spectroscopy (XPS)
2+
but in no case is it Cu . Judging from the shape and area
of the Cu 2p satellites, some photoreduction of CuO phase
might occur during the analysis of the outgassed samples
and more specifically for sample CSS4 in which the inten-
sity of the satellite is very small.
The surface atomic ratios Cu/Si, also summarized in
Table 3, have been calculated from peak areas, normalized
by photoelectron cross sections and mean free paths (16,
The chemical state of the catalyst components and the
relative concentration at the surface has been evaluated by
XPS in both calcined and H2-reduced samples. The binding
energies (BE) of core electrons of the most abundant ele-
ments are compiled in Table 3. For the sake of clarity the
Cu 2p3/2 core-level spectra of outgassed and H2-reduced
CSS4 and CSS8 samples are displayed in Fig. 3. From these
1
7). Cu/Si ratios are larger for the outgassed catalyst CSS4
and decreases by 37% upon reduction, suggesting aggre-
gation of reduced copper species. Unexpectedly, the ratio
Cu/Si was lower for sample CSS8 than for its CSS4 counter-
part. This result, together with the dark color and a broad
band around 350 nm in its diffuse reflectance spectrum,
indicates that CuO clusters are developed in the calcined
catalyst. Further sintering of copper species seems to occur
in the reduced CSS8 sample, as evidenced by the decrease
of the Cu/Si ratio in this sample.
TABLE 2
Characteristic Reduction Temperatures and Extent
of Reduction of CuO–SiO2 Sol–Gel Catalysts
Peak 1
Tmax (K)
Peak 2
Tmax (K)
Peak 3
Tmax (K)
%
Sample
Reduction
CSS4
CSS8
520
550
610
600
705
680
94
91

NO REDUCTION ON CuO2–SiO2 SOL–GEL CATALYSTS
5
FIG. 2. Thermo-programmed reduction of samples calcined at 673 K (CSS4) and 1073 K (CSS8). Deconvoluted profiles.
+
0
In order to get an insight of the nature of copper species Since the BE values of Cu and Cu are almost identical,
+
0
developed on the catalyst surface during on-stream opera- the distinction between the Cu and the Cu species present
tion, both CSS4 and CSS8 samples were pretreated under on the catalyst surface is feasible only through examination
a NO :H2 = 1 : 1 gas mixture at 723 K for 2 h within the of the Auger parameter (18a, 18b). The modified Auger pa-
pretreatment chamber of the XPS spectrometer and then rameter is defined by: �A = h� + (CuLMM � Cu2p3/2), where
analyzed. The BE of Cu 2p3/2 core level of both samples at �A represents the difference between the kinetic energy of
9
32.6–932.8 eV and the absence of satellite lines are con- the CuLMM Auger electron and the Cu 2p3/2 photoelectron,
+
0
clusive evidence that copper became reduced (Cu or Cu ). and h� is the energy of the incident photon. The observation

´
DIAZ ET AL.
6
High-Resolution Electron Microscopy HRTEM
Results concerning the HREM characterization of the
catalysts have been published elsewhere (19). Figure 5
shows low magnification electron micrographs of the cal-
cined samples. From the images it is evident that the SiO2
particles increased their size as the calcination temperature
increased. The microstructure of the CSS4 catalyst, exam-
ined by high-resolution electron microscopy is shown in
Fig. 6. Small crystallites in a structureless form can be ob-
served on the surface ofthe amorphoussilica. The mean size
of these structures is 4 nm. A totally different picture is of-
fered by the CSS8 sample (Fig. 7). In this case, apparently
only the amorphous structure of the SiO2 was observed.
This would mean that the copper is either atomically dis-
persed or formed into clusters of such a size that they could
not be detected by HRTEM. Another possibility is linked
to the reaction of CuO and the support when heating un-
der oxidizing atmospheres, leading to the formation of a
structureless or amorphous silicate compound.
The microstructure characterization of the catalysts after
reaction (NO + H2, NO/H2 = 0.5, 373–873 K) showed dif-
ferent evolution. Figure 8 shows a typical image of the cata-
lyst calcined at 673 K after a catalytic run. Crystallites with
well-defined structures are evident. The mean particle size
is in the range of 10–20 nm. Figure 9 shows an image of the
CSS8 catalyst after reaction. Segregation of small crystals
on the edges of the silica support were identified. The mean
size of these crystallites is in the range 7–9 nm. HRTEM
images showed that copper oxide changes its morphology
and crystal structure, depending on the treatment that the
catalyst has undergone (calcination and redox treatments
during reaction).
FIG. 3. Cu 2p3/2 core-level spectra of samples CSS4 and CSS8 sub-
jected to outgassing and reduction treatments.
of two components for samples CSS4 and CSS8 (Table 3 and
+
0
Fig. 4) allowed the conclusion that Cu and Cu species are
simultaneously present on the surface of the catalysts ex-
posed to the reactant mixture NO + H2.
The HRTEM images were digitized and Fourier trans-
form patterns were obtained in order to measure with rel-
atively good precision the interplanar distances in particles
and to make the phase assignment. The results are shown
in Table 4 where the color of the sample at each stage of the
analysis has been included. Measurements showed that in
the CSS4 sample, copper is present as CuO crystallites in a
structureless form. Furthermore, this sample has the blue-
green color characteristic of the presence of cupric ions
+
2
Cu incorporated into silica (20). Thissuggeststhat besides
the small crystallites, copper oxide is either highly dispersed
or forms clusters of such a size that could not be detected.
Under a “reducingatmosphere,” i.e., under stream, the sam-
ple first showed a reddish-purple color which changed to
dark gray after few minutes of exposure to air. This result
is interpreted as a change of valence state from a mixed
0
1+
2+
(
Cu + Cu ) valence to a main valence of copper Cu
.
Finally, the main crystalline form observed corresponds to
CuO (Table 4).
For sample CSS8 no traces of crystalline copper were
detected. After reaction, segregation of crystallites to the
borders of the silica particles was shown, demonstrating
the influence of the redox treatments (catalytic reaction)
FIG. 4. CuLMM Auger parameter for samples CSS4 and CSS8 exposed
to a NO + H2 = 1 : 1 (molar ratio) at 723 K for 2 h within the pretreatment
chamber of the spectrometer.

NO REDUCTION ON CuO2–SiO2 SOL–GEL CATALYSTS
7
FIG. 6. (a) HRTEM micrographs of CSS4 catalyst. Supported small
CuO crystallites are indicated by arrowheads. (b) High magnified CuO
crystallite and (c) itscorrespondingFFT. (Reprinted from (19) M. Benaissa
and G. Diaz, Microsc. Res. Tech. 40, 49 (1998). Copyright �c 1998 Wiley–
Liss. Reprinted by permission of Wiley–Liss, Inc., a subsidiary of John
Wiley & Sons, Inc.)
on incorporated copper ions. This result suggests that these
incorporated ions enhance in “reducing atmospheres” the
ready formation of such crystallites. This phenomenon is
also thought to occur in the CSS4sample, which explainsthe
FIG. 5. Low magnification micrographs of samples calcined at (a) 673
(
CSS4) and (b) 1073 K (CSS8). (Reprinted from (19) M. Benaissa and
FIG. 7. (a) Typical HRTEM micrograph of CSS8 catalyst. (b) At the
G. Diaz, Microsc. Res. Tech. 40, 49 (1998). Copyright �c 1998 Wiley–Liss. inset FFT. (Reprinted from (19) M. Benaissa and G. Diaz, Microsc. Res.
Reprinted by permission of Wiley–Liss, Inc., a subsidiary of John Wiley &
Sons, Inc.)
Tech. 40, 49(1998). Copyright �c 1998Wiley–Liss. Reprinted bypermission
of Wiley–Liss, Inc., a subsidiary of John Wiley & Sons, Inc.)

´
DIAZ ET AL.
8
size of the crystallites; i.e., in addition to the spread of the
deposited crystallites, incorporated copper ions segregate
to the surface and migrate to join the deposited out-
side crystallites. The segregated crystals in sample CSS8
0
were identified as a mixture of Cu and Cu2O along with
some areas showing structural defects (Fig. 9). These small
areas are thought to correspond to defective Cu2O struc-
tures. A defect structure of Cu2O, which corresponds to
CuO0.67(Cu3O2), iswellaccepted to occur either in thin films
(
21) or in small particles (22) as an intermediate and sug-
gests that the observed defects may correspond to a stable
copper (I) oxide defect structure.
Catalytic Activity (NO + H2)
(
a) Calcined samples. The overall activity of CSS4 and
CSS8 catalysts for NO reduction by hydrogen at a molar ra-
tio NO/H2 = 0.5 is shown in Fig. 10. At these experimental
FIG. 8. (a) HRTEM micrograph of CSS4 catalyst after reaction.
(
(
NO + H2, T = 298–973 K). A crystallite is shown by the arrowhead.
b) Highly magnified image of this area. (c) FFT corresponding to the
top part (indicated by double arrowhead) of the image shown in (b).
(
(
Reprinted from (19) M. Benaissa and G. Diaz, Microsc. Res. Tech. 40, 49
1998). Copyright �c 1998 Wiley–Liss. Reprinted by permission of Wiley–
Liss, Inc., a subsidiary of John Wiley & Sons, Inc.)
TABLE 4
Estimated Phases Determined by HREM in Calcined Samples
and after Reaction
dexp (nm)
Catalyst
(1 = � 0.005 nm)
� (nm)
Phase
CuO
Sample color
Blue-green
CSS4
Reflect. 1: 0.357
Reflect. 1: 0.205
4
CSS4R
10–20
Cu + CuOx
Reddish
purple
↓
↓
Reflect. 2: 0.191
air
↓
a
Dark gray
Cu + CuO
FIG. 9. (a) A typical HRTEM micrograph of CSS8 catalyst after reac-
tion (NO + H2, T = 298–973K). A segregation ofcopper phase isobserved.
CSS8
CSS8R
—
0.202
—
7–9
Glassy
Dark gray
Reddish
purple
Cu + Cu2O
(
b) FFT of the top right part of the crystallite (indicated by double arrow-
head in (a)). A small area showing structural defects is indicated by the
single arrowhead in (a). (Reprinted from (19) M. Benaissa and G. Diaz,
Microsc. Res. Tech. 40, 49 (1998). Copyright �c 1998 Wiley–Liss. Reprinted
by permission of Wiley–Liss, Inc., a subsidiary of John Wiley & Sons, Inc.)
Note. CSS4R, CSS8R, samples after reaction (NO + H2, NO/H2 = 0.5,
2
98–873 K). Reflect. 1, and 2; reflections in FFT.
a
HREM analyzed.

NO REDUCTION ON CuO2–SiO2 SOL–GEL CATALYSTS
9
TABLE 5
Reactivity of CuO–SiO2 Samples: NO + H2, NO/H2 = 1
Catalyst
Te (K)
T50 (K)
�873 (% )
CSS4
CSS8
623
573
788
708
77
76
Note. �873 (% ), conversion of NO at 873 K.
and 13 kcal/mol for CSS8 and CSS4 catalysts, respectively.
Finally, the TOF calculated at 623 K is 2.5 times higher for
CSS8 catalyst than for the CSS4 sample.
(
b) Reduced samples. The reactivity of prereduced
FIG. 10. Overall activity as a function of reaction temperature
NO + H2, NO/H2 = 0.5) of CSS4 and CSS8 catalysts.
samples was also investigated. For this purpose calcined
samples were subjected to a temperature programmed re-
duction from 298 to 673 K prior to the catalytic run (NO/
H2 = 0.5). The notation for samples will be RCSS4 and
RCSS8 for CSS4 and CSS8 catalysts, respectively. Figure 12
shows the behavior of the RCSS4 catalyst. Prereduction of
the CSS4 sample led to an increase of activity at low tem-
peratures (<573 K) and a decrease of activity at higher
temperatures. On the contrary, the same pretreatment per-
formed in the CSS8 catalyst did not have a major influence
in the catalytic behavior of the sample; RCSS8 catalyst is
slightly more active at low temperatures and slightly less ac-
tive at higher temperatures compared to the calcined sam-
ple (Fig. 13). At equimolar NO/H2 ratio (not presented),
RCSS4 catalyst showed a substantially lower activity in the
whole range of temperature compared to the calcined sam-
ple. For RCSS8 catalyst, a decrease in activity was also ob-
served (� 15% in the worst case).
(
conditions only slight differences were observed for the NO
conversion between CSS4 and CSS8 catalysts. Total conver-
sion of NO is achieved at 723 K in both samples. When a
molar ratio NO/H2 = 1 was used, the total activity of the
samples decreased and more pronounced differences in re-
activity between samples were observed (Fig. 11). The cata-
lyst calcined at 1073 K is more active than the catalyst cal-
cined at lower temperature, as can be observed from data
in Table 5 where the light-off temperature (Te), the temper-
ature for 50% conversion of NO (T50), and the conversion
at 873 K are presented. The dispersion of the copper phase
calculated from dissociative N2O adsorption, the turnover
frequencies (TOF), and the apparent activation energies
calculated in the low conversion region (up to 623 K) are
compiled in Table 6. Catalyst CSS8 showed a moderately
higher dispersion than CSS4 sample. The apparent activa-
(c) Time on-stream. Time on stream experiments were
tion energy for these catalysts in the range 548–623 K is 11 performed in order to estimate the stability of these
FIG. 11. Overall activity as a function of reaction temperature (NO + H2, NO/H2 = 1) of CSS4 and CSS8 catalysts.

´
DIAZ ET AL.
1
0
TABLE 6
is related to the presence of this compound. At higher H2
concentration (NO/H2 = 0.5) ammonia concentration be-
Dispersion, Turnover Frequencies, and Apparent Activation
came important.
Energies CuO–SiO2 Catalysts, NO + H2, NO/H2 = 1
Dispersiona
(% )
TOF � 10
(s
2 b
Eac
(Kcal/mol)
4. DISCUSSION
�
1
Catalyst
)
The most significant features of these CuO–SiO2 sol–gel
catalysts evidenced by the various characterization tech-
niques may be summarized as follows: (i) Different species
are present in the catalysts as a function of the calcina-
tion temperature; i.e., the sample calcined at 1073 K had
copper species that reduced at a higher temperature than
those present in the sample calcined at 673 K. (ii) Small
crystallites identified as CuO were observed in the sample
calcined at 673 K. On the contrary, no CuO structure was
observed in the catalyst calcined at 1073 K. After the reac-
tion, sintering was observed in both catalysts but the extent
of this phenomenon was different for each sample. A mix-
ture of reduced and oxidized phases (in the sample calcined
at 673 K) and reduced phases (in the sample calcined at
CSS4
CSS8
23
32
0.72
1.83
13
11
a
Calculated from dissociative N2O adsorption.
TOF for NO reaction calculated at 623 K.
Determined in the 548–623 K range.
b
c
sol–gel catalysts (NO/H2 = 0.5, T = 673 K). Figure 14 shows
the behavior of the CSS4 and CSS8 catalysts compared
to samples prepared by the classical impregnation method
and identified as CSI4 and CSI8 (same copper precursor,
same CuO nominal content, and similar calcination proce-
dures were used for these preparations). Differences are
observed, the catalysts prepared by conventional methods
had lost about 50% of their ability to convert the NO af-
ter 6 h of time on-stream. On the contrary, sol–gel catalysts
were very stable. This behavior was also observed in the
case of an equimolar NO/H2 ratio. The sample calcined at
1
073 K) was determined after reaction. (iii) Catalyst reduc-
tion at 523 K led to the disappearance of satellite line and
the simultaneousshift ofthe principalCu 2p3/2 peak to lower
BE. Both facts are conclusive that copper species were re-
6
73 K apparently showed an induction period. The activity
increased in more than 1 h to reach a stable value that lasted
for the entire run.
+
0
2+
duced Cu or Cu , but in no case was it Cu . The sur-
face atomic ratios Cu/Si were larger for the outgassed CSS4
catalyst and markedly decreased upon H2-reduction, sug-
(
d) Selectivity. The products for the NO + H2 reaction gesting aggregation of reduced copper species. The Cu/Si
are N2, N2O, NH3, and H2O. At a NO/H2 = 1 molar ratio, ratio is even lower for sample CSS8 than for its CSS4 coun-
the selectivity plotted as (N2 + N2O) (mol% ) is shown in terpart, indicating that CuO clusters are developed in the
Fig. 15. The (N2 + N2O) concentration is around 70 mol% calcined catalyst. Further sintering occurs upon H2 reduc-
at 873 K. The curve followed nearly the same trend as that tion. (iv) Dissociative N2O adsorption showed a moderately
of the overall activity. Ammonia was not detected at signifi- higher dispersion of the copper phase in CSS8 catalyst. Re-
cant levels under this experimental condition, but the slight activity differences in the NO + H2 reaction were observed
decrease in (N2 + N2O) concentration at the end of the run for calcined and prereduced samples as a function of the
FIG. 12. Catalytic activity of calcined (CSS4) and prereduced (RCSS4) catalyst (NO + H2, NO/H2 = 0.5).

NO REDUCTION ON CuO2–SiO2 SOL–GEL CATALYSTS
11
FIG. 13. Catalytic activity of calcined (CSS8) and prereduced (RCSS8) catalyst (NO + H2, NO/H2 = 0.5).
calcination temperature. At a NO/H2 = 1 molar ratio, the been documented. Bond etal. (23) in a studyofthe influence
calculated TOF was 2.5 times higher for a catalyst calcined of the support on the reduction properties of copper based
at higher temperature. Remarkable stability of both sol–gel catalysts, mentioned that two kinds of copper species are
catalysts was observed compared to samples prepared by present in silica and alumina catalysts: CuO particles which
conventional methods.
reduce in a very narrow temperature range and dispersed
From all these results it seems clear that the preparation Cu(II) species which reduce at high temperatures. Gentry
method on the one hand and the calcination temperature of and Walsh (24), on the other hand, showed that copper ox-
the catalysts on the other are relevant to the microstructure ide supported on silica presented two copper species, dis-
and catalytic behavior of the samples.
persed CuO and copper silicate. TPR profiles showed that
The microstructure of the calcined catalysts (small CuO the first one reduces at lower temperatures than the second
crystals along with a vitreous structure in sample CSS4 and one and that the reduction temperature of the species in
only an amorphous structure in sample CSS8) may be cor- strong interaction with the support depends on the calci-
related with the reduction properties of the samples. The nation temperature. Guerreiro et al. (25), studying copper
effect of dispersion of CuO in the reduction of the phase has on silica catalysts prepared by ionic exchange, report the
FIG. 14. Time on-stream experiment showing the behavior of CSS4 and CSS8 sol–gel catalysts along with CSI4 and CSI8 impregnated samples.

´
1
2
DIAZ ET AL.
FIG. 15. Selectivity for (N2 + N2O) in mol% as a function of the reaction temperature (NO + H2, NO/H2 = 1).
presence of two reduction peaks in TPR studies. The max- ied. The reactive surface that is generated upon reduction
ima of these peaks are between 523 and 623 K and 873 and of the calcined sample and its evolution during the reac-
9
23 K, respectively. Deconvolution of the first one showed tion seem to be different for each sample. The nature of
two contributions. The first is the one-step reduction of low the active site and the mechanism associated to both the
+2
0
interacted copper species (Cu → Cu ) and the second is decomposition and the reduction of NO in Cu-based cata-
the partial reduction of highly dispersed and surface inter- lysts has been a matter of controversy. Copper species with
+
2
+1
0
1+
2+
acted species (Cu → Cu ). The second peak is related to various oxidation states (Cu , Cu , Cu ) have been in-
+1
0
the second step reduction of this species (Cu → Cu ).
voked to participate in the reaction mechanism (26). In the
The TPR profiles of the sol–gel catalysts in this work presence of a reductant (e.g., H2 or CO), this transforms
may be correlated with the occurrence of low- and high- copper ions in a reduced copper site; the dissociative ad-
interacted copper species as a function of the calcination sorption of NO produces N2 and oxygen atoms adsorbed
temperature. Reduction at lower temperature could be re- on the surface. The reductant regenerates the active site by
lated to the presence of small CuO crystallites as shown removing the adsorbed oxygen atoms (27). In Cu–ZSM-5
by the HRTEM characterization; reduction at higher tem- catalysts, the rate of NO decomposition has been related to
peratures could be related to copper species not detected the cuprous ion concentration (28).
by HRTEM. These copper species may be visualized as
According to this model, the prereduction of CSS4 sam-
“
stabilized” species resulting from the preparation method. ple led, as expected, to an increase of catalytic activity com-
UV–VIS–NIR characterization of samples prepared by the pared to that seen in the calcined sample. Furthermore, the
same experimental procedure pointed out the existence of behavior of the catalysts as a function of the hydrogen con-
two copper (II) species in the samples, one within the sil- centration is also an indication of the role of the reductant in
ica network and the other one as highly dispersed clusters the reaction mechanism (27). Nevertheless, for this sample,
(
8).
a decrease of activity at higher temperatures was also ob-
Dispersion of the copper phase calculated from dissocia- served. The explanation for this decay in activity can be re-
tive N2O adsorption followed the same trend suggested by lated to a loss of active surface area due to sintering. For the
the HREM examination of the catalysts (a better disper- CSS4 catalyst, the HRTEM showed an increase in particle
sion for CSS8 than for CSS4 catalyst). XPS measurements size after reaction (Fig. 8). This fact could be an indication
showed, on the contrary, that CSS4 catalyst was better dis- of the evolution of the active phase when subjected to ther-
persed. This apparent disagreement may have its origin in mal treatments as the catalytic reaction. The prereduction
the fact that XPS is a sensitive-surface technique (0–3 nm), of the sample and the reaction itself could, therefore, have
while chemisorption is related to the entire surface of the an important effect on the evolution of the surface of the
solid.
The occurrence of species of different nature (low and
small (CuO and/or Cu) crystallites present in the sample.
For the sample calcined at 1073 K, the prereduction of the
high interacted species) was evidenced when the catalytic catalyst had no major effect in the activity of the sample.
behavior of the calcined and prereduced samples was stud- HRTEM examination of the CSS8 sample after reaction

NO REDUCTION ON CuO2–SiO2 SOL–GEL CATALYSTS
13
showed, on the other hand, a lower sintering of the copper will be available during reaction. The strength of the ad-
phase compared to that seen in CSS4, which agrees with sorption bond of the NO depends on the copper oxidation
the Cu/Si ratios derived from XPS. The behavior of the state; thus, the ability of these species to activate the NO
CSS8 catalyst cannot then be related only to a sintering will be a function of their aggregation state and stability
0
1+
2+
phenomenon, and it seems that the reactive surface that under reaction. The equilibrium between Cu /Cu /Cu
is generated in situ during reaction in the calcined sample species and their occurrence during reaction, influenced by
is not very different from that generated in a prereduced the interaction with the support and the concentration of
sample. This could mean that copper species at the surface the reductant, will determine the behavior of the catalyst.
of both catalysts (calcined and reduced) undergo similar Supporting this, the Auger parameter determined on cata-
processes (redox and sintering) and this as a consequence lysts after reaction showed that the relative abundance of
0
1+
of the original microstructure of the catalyst.
reduced Cu /Cu species, present on the surface of the
Both samples prepared by the sol–gel method showed re- catalysts, is different in each sample (Fig. 4). Stabilization
markable stability in time on-stream when compared to the of active species through the support due to the preparation
catalysts prepared by impregnation. The sol–gel method of- method and to the calcination treatments has important im-
fers the possibility of stabilizing ions or structures of ions in plications on the activity of the catalyst and the resistance
the oxide network (6). In our case this fact is clearly shown of the active phase toward sintering.
by the catalytic activity of the CSS4 and CSS8 samples after
6
h on stream; i.e., the active phase in these catalysts is more
CONCLUDING REMARKS
stable toward sinteringthan in catalystsprepared bya classi-
cal method. In catalysts prepared by conventional methods,
In the present study, CuO–SiO2 catalysts prepared by
a severe loss of the active surface could easily explain their sol–gel and calcined at 673 and 1073 K were character-
behavior. It is known from the literature that the mobility of ized by BET, FTIR, TPR, dissociative N2O adsorption,
metallic copper in catalysts prepared by impregnation leads XPS, HRTEM techniques, and the reactivity of the sam-
to significant sintering when reduction treatments are per- ples was evaluated in the NO reduction by hydrogen in the
formed at 573 K (29). Indeed, HREM examination of cal- 373–873 K range. The following conclusions can be drawn:
cined catalysts prepared by impregnation showed already (i) FTIR shows the typical absorption bands corresponding
the presence of CuO crystallites with a particle size higher to the vibration modes of the SiO2. The surface area of the
than 15.0 nm. Particles of this size were never observed in solids decreases as the calcination temperature increases.
the catalysts prepared by the sol–gel method. On the other (ii) The preparation method and the calcination tempera-
hand, no clear interpretation can be given for the presence ture are relevant to the microstructure of the solids in such
of an induction period in catalyst CSS4.
a way that stabilization of dispersed copper species not de-
We can speculate that dispersed copper species stabi- tected by HRTEM is achieved for samples calcined at high
lized by the support, as a consequence of the preparation temperature along with small CuO crystallites in samples
method, are responsible for the catalytic behavior of the calcined at low temperature. The occurrence of these stabi-
samples when compared to catalysts prepared by conven- lized speciesare thought to be related to the reduction prop-
tional methods such as impregnation. As an alternative, the erties of the copper phase in the samples. Dissociative N2O
possibility of formation of a silicate compound from reac- adsorption shows a moderately better dispersion for cata-
tion of the CuO with the support in catalyst calcined at high lyst calcined at high temperature. (iii) Reduced catalysts
temperature has been mentioned before; nevertheless, the show after reduction treatments not only a disappearance
characterization techniques used in this work do not pro- of satellite line but also a shift of the principal Cu 2p3/2 peak
+
2
vide evidence of the presence of such a phase.
to lower BE. Both facts are conclusive that Cu species be-
+
0
Kohler et al. (22) suggested that the low deactivation rate came reduced to Cu and Cu . Measurements of the CuLMM
for Cu/SiO2 prepared by ion-exchange is related to a strong line in catalysts exposed to the NO + H2 mixture allow the
+
0
interaction between small copper particles and the ion ex- conclusion that Cu and Cu are simultaneously present on
changed Cu(I) present on the silica surface. On the other the surface of the catalysts exposed to the reactant mixture
+1
hand, stabilization of Cu species by the support has been NO + H2. In addition, the relative contribution of each re-
mentioned to be required for the NO decomposition on duced species on the surface of the catalyst is different. The
model Cu/Al2O3 (0001) catalysts (30).
Cu/Si ratio in the samples decreases upon H2 reduction, in-
Copper species with various aggregation states may co- dicating sintering of copper species. (iv) The catalysts were
exist at the surface of the sol–gel catalysts. Some of these active for NO reduction by hydrogen. Remarkable stabil-
species are stabilized by the support. The dispersion mea- ity of the sol–gel catalysts in time on-stream was observed
surements showed only moderate differences between both compared to catalysts prepared by conventional methods
catalysts. The redox behavior of these surface species will be (impregnation). (v) The observed differences could be an
different and therefore, various oxidation states of copper indication of the various properties of the Cu species while

´
DIAZ ET AL.
1
4
being in or out the support framework after the calcination 12. (a) Guglielminoti, E. W., Langmuir 6, 1455 (1990); (b) Decontignies,
M., Phalippou, J., and Zarzycki, J., J. Mater. Sci. 13, 2605 (1978).
3. Orel, B., Svegel, F., Bukovec, N., and Kosec, M., Bull. Soc. Chim. Fr.
process.
1
122 (1963).
ACKNOWLEDGMENTS
14. (a) Hermann, R. G., Klier, K., Simmons, G. W., Finn, P. B., Bulko, J. B.,
and Kobylinski, T. P., J. Catal. 56, 407 (1979); (b) Okamoto, Y., Fukino,
K., Imanaka, T., and Teranishi, S., J. Phys. Chem. 87, 3740 (1983);
(c) Garbassi, F., and Petrini, G., J. Catal. 90, 113 (1984).
15. (a) Kim, K. S., J. Electron Spectrosc. Relat. Phenom. 3, 217 (1974);
(b) van der Laan, G., Westra, C., Hass, C., and Sawatzky, G. A., Phys.
Rev. B 23, 4369 (1981); (c) Okada, K., and Kotani, A., J. Electron Spec-
trosc. Relat. Phenom. 86, 119 (1997); (d) Parmigiani, F., and Sangaletti,
L., J. Electron Spectrosc. Relat. Phenom. 66, 223 (1994).
Thanks to Dr. Margarita Viniegra and Dr. Sergio Fuentes for stimulat-
ing discussions, to Dr. Jos e´ Saniger for making available to us the FTIR
spectrometer, to Mr. Alfredo S a´ nchez for the prints, and to CONACYT
for funding.
REFERENCES
1
6. Scofield, J. H., J. Electron Spectrosc. Relat. Phenom. 8, 129 (1976).
17. Penn, D. R., J. Electron Spectrosc. Relat. Phenom. 9, 29 (1976).
. (a) Iwamoto, M., and Hamada, H., Catal. Today 17, 94 (1991); 18. (a) Fierro, J. L. G., Catal. Rev.-Sci. Eng. 34, 255 (1993); (b) Tavares
1
2
. Taylor, K. C., Catal. Rev. Sci. Eng. 35(4), 457 (1993).
(
b) Iwamoto, M., Stud. Surf. Sci. Catal. 54, 121 (1990).
. Shimokawage, M., Takezawa, N., and Kobayashi, H., Appl. Catal. 2,
79 (1982).
Figuereido, R., Martinez-Arias, A., Lopez Granados, M., and Fierro,
J. L. G., J. Catal. 178, 146 (1998).
19. Benaissa, M., and Diaz, G., Microsc. Res. Tech. 40, 49 (1998).
3
4
5
3
. De Jong, K. P., Geus, J. W., and Joziasse, J. Appl. Surf. Sci. 6, 273 20. Bae, B. S., and Weinberg, M. V., J. Non-Cryst. Solids 168, 223 (1994).
(
1980).
21. Wieder, H., and Czanderna, A. W., J. Phys. Chem. 66, 816 (1962);
Yanase, A., and Komiyama, H., Surf. Sci. 248, 11 (1991).
. van der Grift, C. J. G., Willers, A. F., Joghi, B. P. J., van Beijimen, J.,
de Boer, M., Versluijs-Helder, M., and Geus, J. W., J. Catal. 131, 178 22. Kohler, M. A., Curry-Hyde, H. E., Hughes, A. E., Sexton, B. A., and
1991). Cant, N. W., J. Catal. 108, 323 (1987).
. (a) Yan, J., Buckley, A. M., and Greenblatt, M., J. Non Cryst. Solids 23. Bond, G. C., Namijo, S. N., and Wakema, J. S., J. Mol. Catal. 64, 305
(
6
1
80, 180 (1995); (b) Gracinda Ferreira da Silva, M., and Fernandez
(1991).
Navarro, J. M., J. Non-Cryst. Solids 100, 447 (1988).
. Buckley, A. M., and Greenblatt, M., J. Non-Cryst. Solids 146, 97
24. Gentry, S. J., and Walsh, P. T., J. Chem. Soc. Faraday Trans. I 78, 1515
(1982).
7
8
9
(
1992).
25. Guerreiro, E. D., Gorriz, O. F., Rivarola, J. B., and Arrua, L. A., Appl.
Catal. A 165, 259 (1997).
. Cordoba, G., Arroyo, R., Fierro, J. L. G., and Viniegra, M., J. Solid
State Chem. 123, 93 (1996).
. Fierro, G., Lo Jocano, M., Inversi, M., Porta, P., Lavecchia, R., and
Cioci, F. L., J. Catal. 148, 709 (1994).
26. Centi, G., and Perathoner, S., Appl. Catal. A 132, 179 (1995).
27. Burch, R., and Scire, S., J. Phys. Chem. 94, 177 (1994).
28. Liu, D. J., and Robota, H. J., Catal. Lett. 21, 291 (1993).
29. Texter, J., Strome, D. H., Hermann, R. G., and Klier, K., J. Phys.
Chem. 81, 333 (1977).
1
1
0. Briggs, D., and Seah, M. P. (Eds.), “Practical Surface Analysis: Auger
and X-ray Photoelectron Spectroscopy” 2nd ed. Wiley, Chichester,
1
990.
1. Yamane, M., Aso, S., Okano, S., and Sakaino, T., J. Mater. Sci. 14, 607
1979).
30. Ishikawa, H., Nakamura, Y., Nakayama, T., Kurusu, C., Okada, O.,
and Yamaguchi, R., “15th Meeting of the North American Catalysis
Society.” Chicago, 1997.
(