Interfacial properties of an Ir/TiO2 system and their relevance in crotonaldehyde hydrogenation

Article abstract of DOI:10.1006/jcat.2002.3566

Titania-supported iridium catalysts were prereduced in a hydrogen flow at 473 K (LT) and 723 K (HT). Metal particle sizes determined by H₂ chemisorption and by direct observation of metal particles by transmission electron microscopy were quite

Full text of DOI:10.1006/jcat.2002.3566

Journal of Catalysis 208, 229–237 (2002)
doi:10.1006/jcat.2002.3566, available online at http://www.idealibrary.com on
Interfacial Properties of an Ir/TiO System and Their Relevance
2
in Crotonaldehyde Hydrogenation
∗
∗
,1
P. Reyes, M. C. Aguirre, I. Meli a´ n-Cabrera,† M. L o´ pez Granados,† and J. L. G. Fierro†
∗
Facultad de Ciencias Qu ´ı micas, Universidad de Concepci o´ n, Casilla 160-C, Concepci o´ n, Chile; and †Instituto de
Cat a´ lisis y Petroleoqu ´ı mica, CSIC, Campus UAM, Cantoblanco, 28049 Madrid, Spain
Received November 21, 2001; revised February 15, 2002; accepted February 19, 2002
drogenation of α,β-unsaturated aldehydes. These reactions
Titania-supported iridium catalysts were prereduced in a hy- lead to two primary reaction products: the saturated alde-
drogen flow at 473 K (LT) and 723 K (HT). Metal particle sizes hyde and the unsaturated alcohol (see Fig. 1). For small
determined by H2 chemisorption and by direct observation of metal
particles by transmission electron microscopy were quite similar
for LT treatment, and close to 3.0 nm. For the HT case, a large
difference in particle size between both techniques is obtained, as
molecules like acrolein and crotonaldehyde the reaction to
the saturated aldehydes is favored both by kinetics and by
thermodynamics considerations (1–3), while for larger α,β-
unsaturated aldehydes, steric restrictions imposed by the
a consequence of the H2 chemisorption suppression (SMSI effect).
substituents on the C=C double bond might influence the
Photoelectron spectra revealed that iridium is in different oxida-
δ+
◦
product selectivity.
tion states, with a contribution of 19% Ir and 81% Ir species
δ+
In the presence of most of the conventional group VIII
metal hydrogenation catalysts, α,β-unsaturated aldehydes
for the LT sample and only a slight increase in Ir species (26%)
for the HT catalyst. Further insights into the surface structures
developed by LT and HT treatments were derived from the cata- are hydrogenated predominantly to saturated aldehydes
lytic performance in the vapor-phase hydrogenation of crotonalde- by reduction of the C=C group or to saturated alcohols
hyde. Activity, expressed as turnover frequency, was more than by reduction of both C=C and CHO groups. Therefore,
one order of magnitude higher for the HT catalyst than for its LT
counterpart. The interfacial metal–TiOx moieties, created upon re-
duction treatment, appeared to be responsible for the increase in
activity and in selectivity to crotyl alcohol, via the formation of a
it is desirable to find catalysts which may control the
intramolecular selectivity by preferentially hydrogenation
of the C=O group while keeping the olefinic double bond
intact (Fig. 1, reaction 1 vs 2). Additionally, in order to
[
–C=O. . . surface] σ-bonded complex (detected by in situ DRIFTS
prevent consecutive hydrogenation to the saturated alco-
hol (reactions 3 and 4) and the isomerization of the allylic
alcohol (reaction 5), the catalyst has to suppress these re-
action pathways. Different metals and supports have been
−1
as a band at 1650 cm ) stabilized at the metal–TiOx interface. HT
treatment enhances the metal–TiOx contact, which results in an im-
provement in catalyst performance. The catalysts deactivate slowly
with the time of reaction. Two reasons are advanced to explain the
catalyst deactivation: (i) the formation of a strongly chemisorbed used to study the selective hydrogenation of these kinds of
−1
asymmetric carboxylate (band at 1740 cm ); and (ii) the formation unsaturated compounds. Thus, metals such as Co, Pt, Ru,
of heavy products with conjugated C=O and C=C bonds (band Rh, Os, and Ir have been assayed, showing great differences
−
1
at 1720 cm ). Both complexes are formed at the expense of the
in activity and selectivity. To improve selectivity to the
unsaturated alcohol, metal-supported catalysts have been
modified in different ways: by alloying (1, 2), by adding pro-
moters (3, 4), by using strong interactive support (5–8), and
by inducing electronic effects and providing appropriate
morphologies in the active sites (9, 10). Despite the number
of investigations carried out in this field, there is no agree-
ment as to which effect may have a more significant impact
on selectivity toward the unsaturated alcohol, because
usually two or more effects are acting simultaneously.
Particularly important is the catalytic behavior exhibited
by catalysts supported on reducible oxides, which is respon-
sible for the strong metal support interaction (SMSI) effect.
Dandekar and Vannice (11) reported significant differences
in both activity and selectivity to crotyl alcohol during hy-
drogenation of crotonaldehyde on Pt-supported catalysts.
σ-complex and progressively block the active sites responsible for
crotonaldehyde hydrogenation.
ꢀc 2002 Elsevier Science (USA)
Key Words: iridium; titania; crotonaldehyde; hydrogenation;
XPS; DRIFTS; surface analysis.
1
. INTRODUCTION
Theproductionoffinechemicalsandchemicalintermedi-
ates routinely involves the selective catalytic hydrogenation
of certain chemical bonds to obtain the desired products.
One kind of reaction important to fine chemicals is the hy-
1
To whom correspondence should be addressed. Fax: +34-91-585-4760.
E-mail: jlgfierro@icp.csic.es.
229
0
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.
ꢀ

2
30
REYES ET AL.
was reduced in situ at LT and HT conditions and then out-
gassed for 4 h at 773 K.
X-ray diffraction was carried out in Rigaku diffractome-
ter, using a filter of Ni and Cu Kα radiation (40 kV and
20 mA). Prior to the experiments, the samples were re-
duced at 473 and 723 K. The samples were scanned at a
◦
◦
rate of 3 /min over a Bragg’s angle range of 3–120 . Trans-
mission electron microscopy (TEM) was used for direct ob-
servation of supported iridium particles. These experiments
were performed in a JEOL JEM-1200 EX-II System. The
samples were prepared by the extractive replica procedure.
Diffuse-reflectance infrared spectra (DRIFTS) were col-
FIG. 1. Reaction network for crotanaldehyde hydrogenation.
In fact, in catalysts in which little or no metal-support inter-
actions should exist, specific activities show similar values. lected on a Nicolet-510 FT-IR spectrometer working at a
−
1
Conversely, Pt/TiO2 (HT) catalysts, which exhibit the SMSI resolution of 4 cm (256 scans) using a Harrick HVC-DRP
effect, show much higher values. One of the characteristics environmentally controlled cell. About 50 mg of the pow-
of this state is a decrease in H2 chemisorption capacity pro- dered sample was packed in a sample holder and different
duced by physical blockage of the metal surface caused by gases were passed through the sample. First, the sample was
the migration of TiOx species during the HT step (12–14). reduced in 50 ml/min of pure H2 at 473 (LT) or 723 K (HT)
Even more critical than the differences in activity is the for 2 h (heating rate 5 K/min). Then it was cooled in H2 at
change in selectivity. Thus, the selectivity to crotyl alcohol 373 K. Subsequently, He flow was switched to 50 ml/min
increases from zero over Pt/SiO2 and Pt/η-Al2O3 to 37% and the DRIFT spectrum of the sample was recorded and
over Pt/TiO2. In all of these samples, the metal was highly taken as background. All further recorded spectra were ref-
dispersed. If the reaction is performed on poorly dispersed erenced to this background. Then the flow was switched to
catalysts, selectivity to the unsaturated alcohol may reach a reaction mixture, and DRIFT spectra of the sample un-
higher values (8, 15).
der the reaction conditions and after different amounts of
Accordingly, this work was undertaken with the aim of time onstream were recorded. The reaction mixture was
modifying the interface between iridium particles and tita- obtained by bubbling H2 through a saturator containing
nia substrate by reduction in hydrogen at low (LT) and high crotonaldehyde and maintained in an ice bath at 273 K. A
(
HT) temperatures, which results in different activity and number of spectra were recorded after different amounts
selectivitypatternsduringthehydrogenationofcrotonalde- of time under the reaction conditions. It must be said that
hyde. For this purpose, the surface properties of the cata- recording 256 scans takes ca. 3.5 min; therefore the spec-
lyst, subjected to two different thermal treatments, were trum is the average of former scans during such a record-
evaluated by H2 chemisorption, TEM, XPS and in situ dif- ing time. After some time onstream, the reaction mixture
fuse reflectance FT-IR spectroscopy. This latter technique was switched again and He was flowed through the sample.
provided useful information dealing with possible reaction Several spectra were then collected at different times and
intermediates and catalyst deactivation in the hydrogena- temperatures, while keeping the sample under a He flow.
tion of crotonaldehyde.
The high noise detected in the lower wavenumbers of the
spectra is a consequence of the strong absorption of the
support in this region.
2
. EXPERIMENTAL
Photoelectron spectra (XPS) were recorded using an
An iridium–titania catalyst was prepared by impregna- Escalab 200R spectrometer provided with a hemispherical
2
tion of a TiO2 (P25-Degussa, SBET = 72 m /g) at 298 K with electron analyzer, operated in a constant pass energy mode,
an appropriate amount of aqueous solution of H2IrCl6 to and a Mg Kα X-ray radiation source (hν = 1253.6 eV).
get 0.5 wt% of the metal. Then, the sample was dried at Prior to analysis the samples were reduced in the pretreat-
3
83 K for 24 h and finally calcined at 673 K for 4 h. For ment chamber of the spectrometer at 473 and 723 K for
the sake of simplicity the sample is referred to hereafter 1 h. After this treatment, samples were outgassed until
−
6
as Ir/TiO2. Prior to characterization or catalytic use the 10 mbarandthenintroducedintotheanalysischamber. A
−
9
catalyst was reduced in situ in flowing H2 at 473 (LT) or base pressure of 2×10 mbar was maintained during data
23 K (HT) for 2 h.
acquisition. The spectral regions of C 1s, Ti 2p, and Ir 4f lev-
Hydrogen chemisorption at 298 K was carried out in a els were recorded under high-resolution conditions and av-
7
volumetric system to evaluate the H2 uptake. These H2 up- eraged for a number of scans in order to obtain good signal-
takeswerethenusedtoestimatetheIrdispersion. Asurface to-noise ratios. The spectra obtained, once the background
stoichiometry of H : Ir = 1 : 1 was assumed in all calcula- was removed, were fitted to Lorentzian and Gaussian lines
tions. Before the chemisorption experiments, the sample to obtain the number of components, peak position, and

SURFACE PROPERTIES AND HYDROGENATION ACTIVITY OF AN Ir/TiO2 CATALYST
231
their areas. The adventitious C 1s line at 284.9 eV was used catalysts on TiO2. The extent of suppression is usually taken
as an internal standard. The surface Ir/Ti ratios were esti- as an indication of the development of metal-support inter-
mated from the integrated intensities of Ir (4f) and Ti (2p) action and is described as decoration of metal particles by
lines corrected by the atomic sensitivity factors (16).
TiOx species generated upon the reduction process. Obvi-
Catalytic experiments were carried out in a pulse reac- ously, it is expected that this process will occur to a higher
tor connected online to a GC. This procedure was chosen extent as temperature increases. The presence of chlorine
because steady-state measurements could not be reached ions may also produce a slight suppression in the hydrogen
due to strong catalyst deactivation. This deactivation pro- uptake. The presence of residual chlorine, reported previ-
cess has been observed in other catalysts, as has been well ously by Bastein et al. (20) and Orita et al. (21), was found
documented (17). In each experiment 50 mg of catalyst was to be responsible for the decrease in CO hydrogenation ac-
reduced in situ in flowing hydrogen (20 ml/min) in a pro- tivity on Rh/TiO2 and was explained in terms of a chloride-
grammed mode at 10 K/min up to 473 (LT) or 723 K (HT) assisted TiOx migration toward metal particle.
for 2 h. Then, the sample was cooled to the reaction tem-
An X-ray diffraction pattern of the Degussa P25 TiO2
perature (323, 343, and 373 K). The catalytic activity was sample used as support showed the characteristic diffrac-
measured by injecting 1.0 µl of the organic reactant. Blank tion lines of both rutile and anatase phases. An 85% anatase
experiments showed no catalytic activity due to the support and 15% rutile has been calculated. This latter phase is the
in these conditions.
most stable but complete transformation of anatase to ru-
tile takes place at about 1300 K. However, this may occur
at lower temperatures in the presence of foreign elements
which can catalyze this transformation (22). The activated
catalysts did not show any change in support crystallinity,
in agreement with previous findings which have shown that
for M/TiO2 catalysts a minimum of 2 wt% metal is required
to enhance the anatase-to-rutile transformation (22).
The chemical state of iridium at the catalyst surface was
determined by photoelectron spectroscopy. For the Ir/TiO2
3
. RESULTS AND DISCUSSION
The metal particle sizes obtained from TEM and the H/Ir
atomic ratios are given in Table 1. The H/Ir value is much
higher (ca. 10 times) for the catalyst reduced at lower tem-
perature (LT), which indicates a higher metal exposure of
LT catalyst. Such a difference is not related to changes in
metal dispersion, as is shown by TEM results. The particle
size values of the LT catalyst are close to those previously
reported for other Ir/TiO2 catalysts prepared under com-
parable conditions (18). However, a slight but reliable dif-
ference is observed between both values, probably due to
extra H2 consumption and maybe ascribed to a H2 spillover
effect at the chemisorption temperature. TEM analysis re-
vealed only a slight increase in the particle size with the
reduction temperature. This apparent discrepancy between
chemisorption and TEM data is explained on the basis of
the significant decrease in the hydrogen uptake observed in
the catalyst reduced at higher temperature (HT). This fact,
first identified by Tauster et al. (19) is a distinctive feature
of the so-called SMSI catalysts. Nowadays, it is well estab-
lished that chemisorption suppression after HT reduction is
a common characteristic of all group VIII metal-supported
(
LT) sample, the binding energy for the most intense Ir 4f7/2
◦
level was found at 61.0 eV for the major component (Ir )
and at 62.7 eV for the less intense component (Ir ) (23)
(
δ+
δ+
Table 1). A detectable increase in the proportion of Ir /Ir
for the HT treatment together with a slight decrease in the
lower binding energy component of the Ir 4f7/2 peak was
found. Figures 2A and 2B display the Ir 4f core-level spectra
for the Ir/TiO2 (LT) and Ir/TiO2 (HT) samples, respectively.
As can be seen in Fig. 2, there is overlapping of the strong
Ti 3s peak with the Ir 4f doublet. Fortunately, the whole
spectrum was accurately decomposed into five lines: two
for the Ir 4f7/2, another two for the Ir 4f5/2, and one for the
Ti 3s level. At low temperature, the small metal particles
are in intimate contact with the TiO2, making it possible
to produce electronic transfer between the support and the
δ+
metal and generating Ir species. Additionally, a decrease
in the Ir/Ti atomic ratio as reduction temperature increases
may be expected due to a migration of TiOx entities on the
surface of the metal crystals producing a decoration of the
metallic particles. However, no changes were detected in
Ir/Ti ratios (Table 1). This may be understood by consider-
ing that the XPS signal averages the contribution of several
layers of atoms, including atoms below the metal surface
and support atoms. Although the photoelectron process is
TABLE 1
H/Ir Ratios Obtained from H2 Chemisorption, Particle Size Eva-
luation by TEM, Binding Energies (BE) of Ir 4f7/2 Core Electrons,
and Ir/Ti Surface Atomic Ratios for a 0.5 wt% Ir/TiO2 Catalyst
D (nm)
Ir 4f7/2
BE (eV)
Catalyst
H/Ir
Chem
1.9
TEM
2.8
Ir/Ti
Ir/TiO2-LT
0.570
61.0 (81)
0.021 produced to a depth on the order of several tens of mi-
crons, only those photoelectrons produced until depths of
6
2.7 (19)
60.4 (74)
2.7 (26)
Ir/TiO2-HT
0.055
20.0
3.2
0.020
approximately 5 nm are reached may be ejected without
losing kinetic energy through inelastic scattering processes.
6

2
32
REYES ET AL.
TABLE 2
Catalytic Activity, TOF (s 1), and Selectivity to Crotylalco-
hol (CROL), Butyraldehyde (BUHO), and Butylalcohol (BUOH)
during Crotonaldehyde Hydrogenation at Different Temperatures
−
(Last Pulse) on a 0.5 wt% Ir/TiO2 Catalyst
Selectivity (mol%)
Temp.
(K)
Activity
TOF
−
1
−1
Catalyst
(µmol/s g
)
(s
)
CROL BUHO BUOH
cat
Ir/TiO2-LT
323
0.121
0.242
0.604
0.242
0.483
1.812
0.008
0.016
0.041
0.168
0.337
1.260
32
21
11
34
28
13
67
77
84
65
64
60
1
2
5
1
8
3
3
43
73
Ir/TiO2-HT
323
3
3
43
73
27
FIG. 2. Ir 4f7/2 core levels for reduced Ir/TiO2 catalysts at low (LT)
and high (HT) temperatures.
δ+
◦
Similar arguments may be given to account for the Ir /Ir
δ+
ratio. The presence of Ir species may account for the in-
crease in the selectivity to crotyl alcohol, discussed below.
Hydrogenation of crotonaldehyde was carried out in a
pulse reactor at different temperatures, in the range 323–
3
73 K. This alternative allows the initial catalytic activity
to be studied without much interference to the deactiva-
tion processes. The obtained products were mainly croty-
lalcohol (CROL), butyraldehyde (BUHO), butylalcohol
(
BUOH), CO, and products of hydrogenolysis and poly-
merization. Table 2 summarizes the activity and selectivity
of the prepared Ir/TiO2catalysts at the last pulse, whereas
FIG. 3. (A) Product distribution (mole) vs pulse numbers in Ir/TiO2-
HT at 343 K: ꢀ, CROL; ꢁ, BUHO; and ꢂ, BUOH. (B) Selectivity to
Fig. 3A shows the evolution of the conversion level and the CROH against catalyst activity for both LT and HT treatments.

SURFACE PROPERTIES AND HYDROGENATION ACTIVITY OF AN Ir/TiO2 CATALYST
233
selectivity to the different products with the number of
pulses for the HT sample at 343 K.
A significant drop in conversion as the number of pulses
increases is observed. In parallel with this change, an en-
hancement in selectivity to CROL and a decrease in selec-
tivitytoBUHOtakesplace. ThedatacompiledinTable2in-
dicate that activity, expressed as micromoles converted per
second per gram of catalyst, is twice that for the HT catalyst
compared to its LT counterpart. However, more-significant
differences were found in the TOF activity, which reaches
values more than 20 times higher for the catalyst reduced at
high temperature. In both catalysts, the saturated aldehyde
is the main product (selectivity higher than 60%) and the
selectivity to the unsaturated alcohol is almost zero. How-
ever the selectivity to CROL is substantially improved for
the HT case, as indicated in Fig. 3B. The selectivity toward
CROH for the same conversion was higher for the HT than
for the LT treatment. This indicates that many more sites
selective for hydrogenation of the C=O bond are present
in the HT state. This behavior may be understood by con-
sidering that a higher number of Ir/TiOx interfacial sites
are involved (as indicated by the suppression of H2 uptake
shownintheHTsample), inspiteofthefactthatunderthese
conditions a lower amount of metal surface is exposed to
the reactant molecules. As Vannice has pointed out (24a),
the creation of these new active sites in the interfacial re-
gion may explain not only the change in activity but also the
change in selectivity. In fact, these sites involved partially
reduced TiOx moieties (or O-vacancies) which coordinate
to the oxygen atom in the C=O group to account for a more
FIG. 4. DRIFT spectra of Ir/TiO2 (HT) during crotonaldehyde hy-
drogenation at 373 K: (a) after 5 min onstream; (b) at 15 min; (c) at 30 min;
d) at 60 min; (e) at 90 min; (f) at 120 min; and (g) after 5 min under helium
flush at 373 K.
selectiveactivationoftheseorganicmolecule(8, 24a). How-
ever, with respect to the active form involved in selective
hydrogenation of the carbonyl group, it has been shown
that the production of the allyl alcohol starts only after
(
an induction period (24b). The modification of the cata- the point of view of the highly deactivated state. At the C–H
lyst surface by the reaction mixture has been considered stretching vibration region, bands at 2965, 2930, 2882, and
−
1
−1
responsible for the creation of this in situ-created center. 2718 cm predominate. The 2718-cm band is assigned
In terms of our system, this phenomenon is not observed to ν(C–H) of the aldehydic group, which suggests that ad-
maybe due to the large deactivation upon crotonaldehyde sorbate containing an aldehyde functional group is present
pulses. Further support for this theory is supplied by the at the surface of the catalyst. They cannot be assigned ei-
DRIFT experiments presented below.
ther to a reaction intermediate or to gas-phase products
Figure 4 shows the DRIFT spectra obtained under re- since band intensity remains when the reaction mixture is
action and after switching to a He flow for Ir/TiO2-HT. switched to a He flow. They must arise from side products or
Figure 5 displays the result of subtracting the DRIFT spec- spectators of the reaction which are strongly chemisorbed
trum of the sample after 5 min under the reaction mixture at the catalyst surface (their intensity is not much altered
from the spectrum of the sample after more than 5 min after flowing He at 473 K). Their concentration at the cata-
onstream. On the other hand Fig. 6 shows the result of sub- lyst surface increases with time onstream, as can be seen in
tracting the DRIFT spectrum of the sample after 120 min Fig. 5, which means that the surface is progressively covered
under the reaction mixture from the spectra obtained af- by such compounds.
Unequivocal assignation of 2965-, 2930-, and 2882-cm 1
−
ter different times onstream. These last figures are very
helpful for seeing bands appearing and disappearing un- bands is complicated by the fact that many of the feasible
der reaction from different points of view. Figure 5 gives molecules show similar features. They cannot be assigned
relevance to the alteration of the band intensity with re- to C–H vibrations of CH groups of weakly adsorbed cro-
x
spect to the fresh state and Fig. 6 gives information from tonaldehyde because its bands are placed at quite different

2
34
REYES ET AL.
−
1
1
545 (s), and 1454 (s) cm are clearly detected in Fig. 4a.
Wide and featureless bands centered at 1410 (s) and
−
1
1
1
265 (s) cm are also present. The strong feature at
−
1
653 cm is assigned to crotonaldehyde O σ-bonded to
coordinatively unsaturated metal cations (Lewis acid sites)
in metal oxide surfaces (29). The σ complex is formed be-
tween the unshared electron pair on the antibonding n(σ)
orbital localized on the oxygen atom of the carbonyl group
and the metal cation (30). This adsorption mode has also
been described as a donating-on-top η1 adsorbed species
−
1
(
31, 32). Such interaction results in a 50- to 70-cm red-
shift of the carbonyl ν(C=O) frequency with respect to
gas-phase crotonaldehyde. This band has been observed
when α,β-unsaturated aldehydes are chemisorbed on vari-
ous metals supported on metal oxides (11, 32, 33).
This adsorbate has been said to be the intermediate
for the selective reduction of the C=O group in α,β-
unsaturated aldehydes (31, 32). Dandekar and Vannice (11)
showed a direct relationship between the intensity of this
band and the formation of crotyl alcohol from crotonalde-
hyde and ascribed this band to a different complex: a di-σ
FIG. 5. DRIFT spectra of Ir/TiO2 (HT) during crotonaldehyde hy-
drogenation at 373 K after subtraction of 5-min onstream spectrum from
a) 15-, (b) 30-, (c) 60-, (d) 90-, and (e) 120-min spectra.
(
wavenumbers (25). Olefinic residues can also be discarded
because the C=C–H vibrations take place at wavenumbers
−1
higher than 3000 cm , as pointed out by Sheppard et al.
(
26). It can then be proposed that vibrations at 2965, 2930,
−1
and 2882 cm arise from saturated alkyl residues strongly
adsorbed on metal particles (26). However, the possibility
that alkyl chains of other residues strongly chemisorbed
at the oxide surface may also contribute to these bands
cannot be excluded. Actually Figs. 5 and 6 show different
vibration frequencies, which indicates that other residues
may be appearing under increasingly higher amounts of
time onstream. At lower wavenumbers a feature in the
ν(C=O) frequency region is detected (not shown in the fig-
ure). This feature did not undergo important changes either
under time onstream or after swithing to the He flow, which
means that it arises from strongly chemisorbed species. This
is attributed to CO adsorbed on Ir arising from the decar-
bonylation processes of crotonaldehyde. Such CO cannot
be discarded as a source of the deactivation (27, 28).
FIG. 6. DRIFT spectra of Ir/TiO2 (HT) during crotonaldehyde hy-
drogenation at 373 K after subtraction of 120-min onstream spectrum
A more complex pattern appears below this region. Af-
ter 5 min onstream bands at 1720 (m), 1690 (m), 1653 (s), from (a) 5-, (b) 15-, (c) 30-, (d) 60-, and (e) 90-min spectra.

SURFACE PROPERTIES AND HYDROGENATION ACTIVITY OF AN Ir/TiO2 CATALYST
235
bond with the C end of the carbonyl group anchored on the nucleophilic attack of surface oxygen to the electrophilic
metal particle and the O end on the metal cationic sites. In C atom of the carbonyl group (29, 30). The blueshift of
any case the close contact between the hydrogenating site the ν(C=O) frequency with respect to the gas-phase cro-
(
noble metal particles) and the electrophilic sites (partially tonaldehyde vibration is related to the electron-donor na-
reduced metal oxides) is needed for the formation of the ture of the surface O sites that shortens the C=O bond.
selective intermediate species. Some support for this also It has been reported that the O σ-bonded of C=O group
δ+
comes from the slightly higher proportion of M in this is transformed to an asymmetric surface carboxylate com-
−1
sample (Table 1, Fig. 2), but the 1653-cm band can only plex when heated or when increasing the aldehyde–metal
be explained if TiOx suboxides are covering the Ir parti- oxide surface contact time (29, 30). Therefore, it can be pro-
cles. Because the H2 chemisorption results have clearly in- posed that the catalyst deactivation must be related to the
dicatedthattheIr/TiO2-HTcatalystisintheSMSIstate, this transformation of the active intermediate into an inactive
DRIFT band can be considered a probe of the metal cov- and strongly chemisorbed surface asymmetric carboxylate
ering by the TiOx species. When the reaction mixture was spectator while the catalyst is under the reaction conditions.
switched to a He flow, after 5 min of the He flush the C=O
H-bonded crotonaldehydeinteractingwith BrønstedOH
feature at 1653 cm disappeared, which demonstrates that sites of TiO2 support accounts for the 1690-cm band: this
−1
−1
this species only develops when gas-phase crotonaldehyde interaction weakens the carbonyl C=O bond and lowers
−
1
is circulating. Thus this disappearance provides additional by 20-30 cm the ν(C=O) frequency (25, 30) compared to
support for the assumption that the σ complex is an inter- the gas-phase value. This adsorbed crotonaldehyde species
mediate of the crotyl alcohol formation.
must show another less intense band at ca. 1640 ascribable
−1
The 1720-cm feature does not arise from miscancel- to the ν(C=C) bond (25) but it is obscured by the strong
−
1
lation of the crotonaldehyde bands because it is also de- one at 1653 cm (it appears as a shoulder of it) and is
tected after switching to the He flow, once crotonaldehyde only observable when crotonaldehyde is flushed (Fig. 4g).
−
1
has been removed from the gas phase (Fig. 4g). Its posi- After He switching the 1690-cm feature remains, which
tion points to a carbonyl-group-containing compound. It is indicates that this complex is strongly chemisorbed on the
strongly adsorbed at the surface because the band remains support surface and it is not involved in the reaction mecha-
after the He flush at 373 (Fig. 4g) and 474 K (not shown in nism. It must be said that a negative peak in the ν(O-H) fre-
−1
Fig. 4). Figure 5 indicates that the 1720-cm band develops quency of Ti–OH surface groups was observed (not shown
only during the beginning of the reaction since it is visible in Fig. 4). Negative absorption, in this case, means that such
in Figs. 5a–5c and then its intensity remains rather constant. OH groups are perturbed by interaction with any of the
This band is almost not visible in Fig. 6 because this figure molecules present under the reaction conditions. Anatase
stresses the point of view of the higher deactivated state. is assumed not to exhibit Brønsted acidity; however weak
According to the literaure, it can be proposed that croton- bases like benzene have been shown to interact with Ti–OH
aldehyde can undergo aldol condensation and oligomer- groups of anatase by H-bonds and to shift the stretching
ization processes (36, 37) on the catalyst surface, leaving modes of the O–H groups to lower frequencies (38). More-
heavyresiduescontainingC=OandC=Cfunctionalgroups over stronger acid O–H sites can be present at the surface
on the surface, which are only removable at high temper- of the TiO2 since impurities like S (sulfates) and P (phos-
atures and which must be taken into account as a source phates) are present in most commercial Ti oxides (39). Any
of the deactivation. The C=O groups in this structure are former phenomena can explain the negative feature but,
weakly affected by the surface and the vibration frequency in any case, these interactions are irrelevant to mechanism
−1
approached the frequency (1720 cm ) of the C=O group clarification since no intensity variations are detected when
of the gas-phase acrolein. switching to a He flow: they very likely arise from spectator–
Figures 5 and 6 allow the identification of a band at OH group interactions.
−
1
−1
1
740 cm that is barely visible in Fig. 4. Figure 5 shows
The 1545- and 1454-cm bands can be ascribed to sym-
−1
−
that the intensity of the 1740-cm feature increases with metric carboxylate (crotanoates or butanoates) (R–CO2
time onstream (see Fig. 5), parallel to the disappearance of group) (25, 30, 33, 34). It has been suggested that sym-
−1
the band at 1653 cm . Figure 6 confirms this phenomenon metric complexes are related to the lack of crotyl alcohol
but from the point of view of the highest deactivated state: formation while onstream (33), but considering that the
−
1
−1
positive intensity of the 1653-cm feature indicates that disappearance of the 1653-cm band is not related to any
during the lower amount of time onstream such a band was change in the carboxylate group frequencies, such a hypoth-
more intense than in the 120-min onstream spectrum. A esis can be discarded. Moreover the symmetric carboxylate
−1
band at 1653 cm has been found when α,β-unsaturated complex bands develop rapidly and the intensity reached
aldehydes are chemisorbed on a metal oxide surface with after 5 min onstream remains almost constant under time
electron-donor centers (oxygen ions): an asymmetric bu- onstream (see Figs. 5 and 6). They were not removed af-
tanoate chemisorbed on the TiO2 surface is formed by ter switching to a He flow, even after heating under He at

2
4
36
REYES ET AL.
73 K, which indicates that these complexes are strongly
chemisorbed and are not involved in the reaction.
Figure 7 shows the DRIFT spectra under reaction condi-
tions and after a He flush of the Ir/TiO2-LT sample. Figure 8
presents the results of subtracting the DRIFT spectrum of
the sample after 5 min under the reaction mixture from the
spectra of the sample after more than 5 min onstream. The
spectra are much simpler. Negative features (not shown) in
−
1
−1
the 3800- to 3600-cm region and at 2044 cm can be as-
signed to interaction of the spectators with the OH groups.
ν(C-H) frequencies were too weak and they could not be
detected. The 5-min spectrum is dominated by the band at
−
1
1
650 cm , which vanishes under a He flush, which is as-
cribable to crotonaldehyde O σ-bonded to coordinatively
unsaturated metal cations (Lewis acid sites) in metal oxide.
The intensity of this band is lower than that for Ir/TiO2-HT,
which can explain the lower activity and CrOH selectiv-
ity of the LT catalyst. Moreover, the weaker intensity of
this band can arise from a lower coverage of Ir particles by
TiOx moities with respect to the HT state. Such an effect
FIG. 8. DRIFT spectra of Ir/TiO2 (LT) during crotonaldehyde hy-
drogenation at 373 K after subtraction of 5-min onstream spectrum from
(
a) 15-, (b) 30-, (c) 60-, and (d) 90-min spectra.
was clearly detected by H2 chemisorption and was also sug-
gested in the Ir 4f core-level spectra (Fig. 2). It seems that a
−
1
DRIFT band at 1650 cm is a valuable probe for detecting
the coverage of metal particles by TiOx patches. No other
intense bands were detected although featureless and wide
low intense bands can be detected at lower wavenumbers.
−
1
After longer time onstream, the 1650-cm band be-
−
1
comes less intense whereas a new wide band at 1720 cm
predominates. The second band is ascribed above to the
formation of heavy products containing C=O and C=C
functionalities arising from crotonaldehyde aldol conden-
sation on TiO2 surface. This wide band seems to display a
−
1
shoulder at 1740 cm . Detection by subtraction (Fig. 8) is
not conclusive. This band was assigned to the formation of
asymmetric carboxylate as a result of the O lattice nucle-
ophilic attack on the carbonyl group. Figure 8 shows that the
intensity of the former band(s) grows whereas the band at
−
1
1
650 cm vanishes, which confirms that these complexes
FIG. 7. DRIFT spectra of Ir/TiO2 (LT) during crotonaldehyde hydro-
genation at 373 K: (a) after 5 min onstream; (b) at 15 min; (c) at 30 min;
d) at 90 min; (e) after 15 min under helium flush at 373 K; and (f) after
0 min under helium flush at 373 K.
can be involved in deactivation processes in LT catalyst.
On the other hand, it must be stressed that bands arising
from symmetric carboxylate complexes were not observed
(
3

SURFACE PROPERTIES AND HYDROGENATION ACTIVITY OF AN Ir/TiO2 CATALYST
237
in Fig. 6 and therefore they cannot be responsible for deac-
tivation, as has been suggested in previous works.
7. Coq, B., Kumbhar, P. S., Moreau, C., Moreau, P., and Figueras, F.,
J. Phys. Chem. 98, 10180 (1994).
8
9
. Vannice, M. A., and Sen, B., J. Catal. 115, 65 (1989).
. English, M., Jentys, A., and Lercher, J. A., J. Catal. 166, 25 (1997).
4
. CONCLUSIONS
1
1
0. Galvagno, S., Milone, C., Donato, A., Neri, G., and Pietropaolo, R.,
Catal. Lett. 18, 211 (1993).
1. Dandekar, A., and Vannice, M. A., J. Catal. 183, 344 (1999).
A titania-supported catalyst exhibits different surface
and catalytic properties after reduction at low or high tem- 12. Vannice, M. A., Catal. Today 12, 255 (1992).
1
1
3. Haller, G. L., and Resasco, D. E., Adv. Catal. 36, 173 (1989).
4. Munuera, G., Gonz a´ lez-Elipe, A. R., Espin o´ s, J. P., Conesa, J. C., Soria,
J., and Sanz, J., J. Phys. Chem. 91, 6625 (1987).
peratures. High-temperature reduction (HT) induces the
decoration of Ir particles generating metal–TiOX moieties,
which appear to be responsible for the increase in the ac-
tivity of crotonaldehyde hydrogenation and in selectivity
to crotyl alcohol. In situ DRIFTS studies have shown the
formation of a [–C=O. . . surface] σ-bonded complex, cha-
1
1
5. Vannice, M. A., Top. Catal. 4, 241 (1997).
6. Wagner, C. D., Davis, L. E., Zeller, M. V., Taylor, J. A., Raymond,
R. H., and Gale, L. H., Surf. Interface Anal. 3, 211 (1981).
7. Raab, G. C., and Lercher, J. A., Catal. Lett. 18, 99 (1993).
1
−1
18. Reyes, P., Aguirre, M. C., Pecchi, G., and Fierro, J. L. G., J. Mol. Catal.
64, 245 (2000).
9. Tauster, S. J., Fung, S. C., and Garten, R. L., J. Am. Chem. Soc. 100,
70 (1978).
racterized by a band at 1650 cm , which became stabi-
lized at the metal–TiOx interface. HT treatment enhances
the metal–TiOx contact and improves the catalytic proper-
1
1
1
ties of the Ir/TiO2 catalyst. The catalytic performance in the 20. Bastein, A. G. T. M., van der Boogert, W. J., van der Lee, G., Luo, H.,
and Ponec, V., Appl. Catal. 29, 243 (1987).
1. Orita, H., Naito, S., and Tamaru, K., J. Phys. Chem. 89, 3066 (1985).
2. Resasco, D. E., Ph.D. thesis. YaleUniversity, NewHaven, Connecticut,
vapor-phase hydrogenation of crotonaldehyde showed that
2
2
turnover frequency was more than one order of magnitude
higherforHTcatalystthanforitsLTcounterpart. Deactiva-
tion of the catalysts takes place under onstream operation.
Two proposals are advanced to explain the catalyst deacti-
vation: (i) the formation of a strongly chemisorbed asym-
1983.
2
3. Briggs, D., and Seah, M. P., Eds., “Practical Surface Analysis:
Auger and X-Ray Photoelectron Spectroscopy.” Wiley, Chichester,
1990.
−1
24. (a) Vannice, M. A., J. Mol. Catal. 59, 165 (1990). (b) Ponec, V., Appl.
Catal. A 149, 27 (1997).
5. Rekoske, J. E., and Barteau, M. A., Langmuir 15, 2061 (1999).
6. Sheppard, N., and de la Cruz, C., Adv. Catal. 41, 1 (1996).
metric carboxylate (band at 1740 cm ); and (ii) the for-
mation of heavy products with conjugated C=O and C=C
2
2
−1
bonds (band at 1720 cm ). Both complexes are formed at
the expenses of the σ-complex and block the active and 27. de la Cruz, C., and Sheppard, N., J. Chem. Soc. Faraday Trans. 93, 3569
(
1997).
selective sites.
2
2
3
8. Coloma, F., Coronado, J. M., Rochester, C. H., and Anderson, J. A.,
Catal. Lett. 51, 155 (1998).
9. Serwicka, M., Black, E. J. B., and Goodenough, J. B., J. Catal. 106, 23
(
0. Popova, G. Ya., Davydov, A. A., Andrushkevich, T. V., and Zakharov,
I. I., Kinet. Catal. 36, 125 (1995).
1. Ponec, V., Appl. Catal. 149, 27 (1997).
2. Nishiyama, S., Hara, T., Tsuruya, S., and Masai, M., J. Phys. Chem.
103, 4431 (1999).
ACKNOWLEDGMENTS
1997).
The authors thank CONICYT (Chile) through FONDECYT Research
Grants 1980345 and 2990065 and CSIC-CONICYT bilateral cooperative
programme for financial support.
3
3
REFERENCES
33. Bailie, J. E., Rochester, C. H., and Hutchings, G. J., J. Chem. Soc.
1
2
3
. Raab, G. C., and Lercher, J. A., J. Mol. Catal. 75, 7 (1992).
. Bonnier, J. M., Damori, J. P., and Masson, J., Appl. Catal 42, 285 (1988).
. Marinelli, T. B. L. W., Nabuurs, S., and Ponec, V., J. Catal. 151, 431
Faraday Trans. 93, 4389 (1997).
34. Yee, A., Morrison, S. J., and Idriss, H., J. Catal. 186, 279 (1999).
35. Popova, G. Ya., Andrushkevich, T. V., Meshcheryakov, V. D., and
Davydov, A. A., Kinet. Catal. 31, 351 (1990).
(1995).
4
5
6
. Sep u´ lveda-Escribano, A., Coloma, F., and Rodriguez-Reinoso, F.,
J. Catal. 178, 649 (1998).
. Galvagno, S., Milone, C., Neri, G., Donato, A., and Pietropaolo, R.,
Stud. Surf. Sci. Catal. 78, 163 (1993).
36. Kubelkova, L., and Trifir o` , F., J. Catal. 26, 242 (1972).
37. Luo, S., and Falconer, J. L., Catal. Lett. 57, 89 (1999).
38. Busca, G., Saussey, H., Saur, O., Lavalley, J.-C., and Lorenzelli, V.,
Appl. Catal. 14, 245 (1985).
. Yoshitake, H., and Iwasawa, Y., J. Catal. 125, 227 (1990).
39. Hadjivanov, K. I., and Klissurski, D., Chem. Soc. Rev. 25, 61 (1996).