Crotonaldehyde hydrogenation on Pt/TiO2 and Ni/TiO2 SMSI catalysts

Article abstract of DOI:10.1006/jcat.1999.2419

A kinetic and DRIFTS (diffuse reflectance FTIR) investigation of crotonaldehyde adsorption and hydrogenation was conducted over TiO₂-supported Pt and Ni with the intent of gaining insight into the adsorption modes of molecules with carbonyl groups on these catalysts in the SMSI and non-SMSI states. Significant enhancement in selectivity toward crotyl alcohol was observed with each catalyst after reduction at 773 K. DRIFT spectra under reaction conditions identified crotonaldehyde species strongly adsorbed through the C=C bond and weakly coordinated through both the C=C and the C=O bonds on these catalysts after reduction at 573 K, which gave a peak at 1693 cm^-1. After reduction at 773 K, an additional adsorbed species with a strong band at 1660 cm^-1, indicating a significant interaction between the carbonyl group and the surface, was observed, which is presumed to be stabilized at interfacial Pt-TiO_x and Ni-TiO_x sites. A decrease in the surface coverage of this species paralleled a drop in selectivity to crotyl alcohol with time on stream. After reduction at 573 K, decarbonylation occurred during the initial few minutes on stream to create adsorbed CO on Pt/TiO₂ in addition to carbon deposition, but these reactions were significantly suppressed after reduction at 773 K, presumably due to a TiO_x overlayer which covers part of the Pt surface and breaks up the ensembles of Pt atoms required for these reactions.

Full text of DOI:10.1006/jcat.1999.2419

Journal of Catalysis 183, 344–354 (1999)
Article ID jcat.1999.2419, available online at http://www.idealibrary.com on
Crotonaldehyde Hydrogenation on Pt/TiO and Ni/TiO SMSI Catalysts
2
2
1
2
Ajit Dandekar and M. Albert Vannice
Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802
Received October 5, 1998; revised January 4, 1999; accepted January 13, 1999
dinate the oxygen atom in the C == O group to account for
A kinetic and DRIFTS (diffuse reflectance FTIR) investigation the selective activation of the carbonyl bonds in these or-
of crotonaldehyde adsorption and hydrogenation was conducted ganic molecules (1, 2, 8). The genesis of this strong metal–
over TiO2-supported Pt and Ni with the intent of gaining insight
into the adsorption modes of molecules with carbonyl groups on
these catalysts in the SMSI and non-SMSI states. Significant en-
hancement in selectivity toward crotyl alcohol was observed with
each catalyst after reduction at 773 K. DRIFT spectra under reac-
the metal (1, 2, 4). Although several spectroscopic investi-
tion conditions identified crotonaldehyde species strongly adsorbed
through the C == C bond and weakly coordinated through both the
support interaction (SMSI) behavior has been linked to a
high-temperature reduction in the presence of a metal ca-
pable of dissociating H2, thus facilitating the creation of
TiOx species which can migrate and partially “decorate”
gations of TiO2-supported Group VIII metals exist which
have identified features characteristic of these interfacial
sites (4), no report of an infrared study could be found in
C == C and theC == Obondson thesecatalystsafter reduction at 573K,
�
1
which gave a peak at 1693 cm . After reduction at 773 K, an addi-
�
1
tional adsorbed species with a strong band at 1660 cm , indicating the published literature which identifies the differences in
a significant interaction between the carbonyl group and the sur- the adsorption modes of molecules with carbonyl groups
face, was observed, which is presumed to be stabilized at interfacial on these catalysts in the SMSI and the non-SMSI state.
Pt–TiOx and Ni–TiOx sites. A decrease in the surface coverage of
The applicability of DRIFTS (diffuse reflectance infrared
Fourier transform spectroscopy) to obtain in situ spectra
during this reaction on supported Cu catalysts has recently
been described (19–21). This paper describes a similar ki-
this species paralleled a drop in selectivity to crotyl alcohol with
time on stream. After reduction at 573 K, decarbonylation occurred
during the initial few minutes on stream to create adsorbed CO on
Pt/TiO2 in addition to carbon deposition, but these reactions were
netic and DRIFTS investigation of crotonaldehyde hydro-
significantly suppressed after reduction at 773 K, presumably due
genation over TiO2-supported platinum and nickelcatalysts
after reduction at either 573 or 773 K.
to a TiOx overlayer which covers part of the Pt surface and breaks
up the ensembles of Pt atoms required for these reactions.
�c 1999
Academic Press
EXPERIMENTAL
INTRODUCTION
The TiO2 support used in this study was a 60/120 mesh cut
of Degussa P25 (80% anatase, 20% rutile, 50 m /g). Nom-
inal 1 wt% Pt or Ni catalysts were prepared by an incipi-
2
The unique catalytic properties exhibited by TiO2-sup-
ported Group VIII metals after a high temperature reduc-
tion (HTR) at 773 K have been clearly demonstrated in
a number of reactions during the past two decades (1–4).
In particular, overwhelming evidence exists for the promo-
tional role of TiO2 in enhancing the activities and altering
the selectivities in reactions between H2 and molecules con-
taining carbonyl bonds, such as CO (5–8), acetone (9), cro-
tonaldehyde (10–14), cinnamaldehyde (15), benzaldehyde
q
q
ent wetness technique using H2PtCl6 6H2O or Ni(NO3)2
H2O (Sigma Chemical Co.) dissolved in distilled deionized
6
water. The TiO2 was calcined in dry air for 2 h at 773 K prior
to impregnation, and the impregnated catalysts were dried
at 393 K overnight before storage in a desiccator. Prior to
any characterization, the catalyst samples were pretreated
in situ in flowing He for 1 h at either 573 or 773 K and then
reduced in flowing H2 at the same temperature for 1 h at a
(
16), acetophenone (17), and phenylacetaldehyde (18). A
�
1
typical space velocity of 20,000 h . The low-temperature
pretreatment at 573 K is 100 K higher than the typical LTR
step, while the high-temperature pretreatment (HTR) at
model has been proposed invoking the creation of special
sites at the metal–support interface involving partially re-
duced TiOx species (or oxygen vacancies) that can coor-
7
73 K is the same (3, 5–10). Samples with the latter pre-
1
treatment will be designated HTR. Total and reversible H2
uptakes on the pretreated catalyst samples were measured
at 300 K in a stainless steel volumetric adsorption system
Current address: Mobil Technology Company, Paulsboro, NJ 08066-
0
480.
2
To whom correspondence should be addressed.
344
0021-9517/99 $30.00
Copyright �c 1999 by Academic Press
All rights of reproduction in any form reserved.

CROTONALDEHYDE HYDROGENATION
345
giving a vacuum below 10 6 Torr at the sample. The H2 grams recorded for that particular catalyst (19–21). In an
99.999% purity, MG Ind.) and He (99.999% purity, MG effort to approximate the reaction conditions under which
Ind.) were flowed through molecular sieve traps (Supelco) the kinetic behavior was determined, the sample was then
and Oxytraps (Alltech Assoc.) for additional purification. heated to the desired reaction temperature and a flowing
�
(
The kinetics of vapor-phase crotonaldehyde hydrogena- mixture of370Torr H2 and 15Torr crotonaldehyde (balance
tion were determined at atmospheric pressure in the mi- Ar) was introduced into the reactor chamber. The mixture
croreactor system described in detail elsewhere (19–21). was obtained by bubbling 50% H2 in Ar (20 sccm) through
After loading typically 0.04–0.06 g of catalyst sample in crotonaldehyde maintained at 273 K in an ice bath. Al-
the reactor, it was subjected to reduction at either 573 or though the H2/crotonaldehyde ratio was about the same as
7
73 K under flowing H2 and the reactor was then cooled that in the microreactor, the partial pressures were around
to the desired reaction temperature. Flow rates of H2 and one-half those in the reactor so that no crotonaldehyde con-
the diluent He (99.999% purity, MG Ind.) were regulated densation would occur in the DRIFTS system. Spectra were
by mass flow controllers (Tylan, Model FC-260). A con- recorded for each catalyst as well as for pure TiO2 under
stant flow of vapor-phase crotonaldehyde (CH3–CH == CH– these reaction conditions at the desired reaction temper-
CH == O) was established by bubbling either H2 or He as ature. To obtain spectra of CO adsorbed at 300 K on the
the carrier gas through a saturator containing liquid croton- Pt/TiO2 samples, the pretreated catalyst was exposed to a
aldehyde (99.9+% purity, Aldrich). Standard conditionsfor mixture of 10% CO (99.999% purity, MG Ind.) and 90%
activity runs were 35 Torr crotonaldehyde with the balance Ar for 30 min before it was purged with pure Ar for 30 min.
H2 (ca. 700 Torr) and Arrhenius plots were obtained be- Interferograms were collected before and after purging the
tween 323 and 383 K. The data during the Arrhenius runs catalyst.
were obtained after waiting for a period of about 30 min at
Temperature-programmed oxidation (TPO) profiles for
each set of reaction conditions to allow the system to reach Pt/TiO2 samplesafter use in the crotonaldehyde hydrogena-
steady-state behavior. In addition, an ascending tempera- tion reaction were obtained in a U-shaped Pyrex microreac-
ture sequence was followed by a descending temperature tor system. The sample was placed between two quartz wool
3
sequence to check for deactivation. All kinetic measure- plugs and heated from 300 to 873 K in 30 cm /min (STP) of
ments were made at atmospheric pressure and under differ- a 10% O2/90% He mixture at a heating rate of 5 K/min. The
ential conditions as conversions were routinely maintained product analysis was done with a Perkin-Elmer Sigma 3 gas
below 15% .
chromatograph equipped with a Chromosorb 102 column
Details of the modified DRIFTS reactor system used in maintained at room temperature to separate O2 and CO2
the present study and the standard procedures employed at a sampling interval of 3 min.
to record the spectra have been provided earlier (19–21).
After loading the catalyst sample into the DRIFTS cell, it
was purged overnight with Ar (99.999% purity, MG Ind.) to
remove any moisture in the cell. The sample was then sub-
RESULTS AND DISCUSSION
Irreversible hydrogen uptakes on Pt/TiO2 and Ni/TiO2
jected to in situ pretreatment at the desired temperature and after reduction at 573 and 773 K are listed in Table 1 along
cooled to 300 K, at which point the first interferogram was with the fraction of available surface metal atoms (H/PtT
recorded. This was used as a background reference in the and H/NiT) and surface-weighted average Pt and Ni crystal-
fast Fourier transform analysis of all subsequent interfero- lite sizes when appropriate. The two former ratios represent
TABLE 1
Crotonaldehyde Hydrogenation over Pt/TiO2 and Ni/TiO2 PCROALD = 35 Torr, Balance H2, TRXN = 333 K
H2 uptakea
(�mol/g)
dcrys
(nm)
Activityb
(�mol/s/g cat)
TOF
b,c
EAPP
(kcal/mol)
Tred
(K)
�
Selectivityd
Catalyst
H/MT
(s 1)
0
0
1
1
.82% Pt/TiO2
.82% Pt/TiO2(HTR)
.3% Ni/TiO2
573
773
573
773
17.0
0.5
22.0
1.5
0.81
(0.019)
0.20
1.4e
—
5.1
3.04
4.75
1.51
1.64
0.089
4.7
0.034
0.55
5.6 � 0.5
6.4 � 0.5
2.8 � 0.3
3.6 � 0.5
15
69
7
f
.3% Ni/TiO2(HTR)
(0.013)
—
59
a
Irreversible uptake at 100 Torr at 300 K.
Measured after 30 min on stream.
Based on irreverible H2 uptake at 300 K.
Initial selectivity to crotyl alcohol.
Based on d = 1.13/D.
b
c
d
e
f
Based on d = 1.01/D.

3
46
DANDEKAR AND VANNICE
valid dispersion values (surface atom/total atom) only af-
ter reduction at 573 K. The drop in the H2 uptake on
Pt/TiO2(HTR) has been routinely observed before (3, 4, 7,
1
0) and has been shown by TEM not to be due to sintering
(
22), but rather the decrease is caused by partial coverage
of the Pt surface by migrating TiOx species (4, 10, 23). In
the case of Ni, although complete reduction may not be
achieved at 573 K (4, 5), suppression of H2 chemisorption
following HTR still occurs and again can be attributed to a
blocking of surface Ni sites by a TiOx overlayer (4, 5, 24).
Table 1 lists the steady-state activities, turnover frequen-
cies(based on the H uptakes), and apparent activation ener-
gies during crotonaldehyde hydrogenation on the Pt and Ni
catalysts, based on crotonaldehyde disappearance to form
crotyl alcohol, butyraldehyde, and n-butyl alcohol. The cor-
responding Arrhenius plots are shown in Fig. 1. No large
difference exists between the apparent activation energies
after reduction at either 573 or 773 K with either Pt or Ni,
although the EAPP values after HTR are higher in each case.
Also, the activation energyishigher for Pt than for Ni. Asre-
ported previously for crotonaldehyde hydrogenation over
Pt catalysts, markedly higher TOFs are obtained with the
�
1
Pt/TiO2(HTR) catalyst (10), and both the TOF (2.8 s at
18 K) and the activation energy (6.4 kcal/mol) agree well
3
�
1
with earlier values of 1.5 s and 7 kcal/mol, respectively
10). A large enhancement in TOF also occurred with the
(
Ni/TiO2(HTR) sample. Of equal importance, the activity
per gram metal (i.e., per g catalyst) increased after the HTR
step for each catalyst. The corresponding activity mainte-
nance profiles for Pt/TiO2 are shown in Figs. 2A and 2B
after reduction at 773 and 573 K, respectively. Similar plots
for Ni/TiO2 are shown in Figs. 3A and 3B. The activity was
FIG. 2. Activity profiles during crotonaldehyde hydrogenation over
Pt/TiO2 reduced at 573 and 773 K (HTR); TRXN = 333 K; PCROALD =
35 Torr.
calculated in two ways—either as the rate of appearance of
crotyl alcohol and butyraldehyde as products or as the rate
of disappearance of crotonaldehyde as a reactant. Whereas
no significant difference can be observed between the two
methods with Pt/TiO2(HTR), the rate of crotonaldehyde
disappearance is significantly higher than the rate of crotyl
alcohol and butyraldehyde appearance during the initial
period on stream over Pt/TiO2 after reduction at 573 K.
This suggests that a significant amount of the reactant is
consumed elsewhere and suggests the possibility of parallel
side reactions like dimerization and decarbonylation (11,
2
5) or perhaps the formation of coke precursors. In con-
trast, no significant difference exists between the rates for
Ni/TiO2 after either reduction temperature.
The initial selectivity to crotyl alcohol at 333 K over these
catalysts is also listed in Table 1, and a large difference ex-
ists between the two Pt/TiO2 samples or the two Ni/TiO2
samples. Considerably larger amounts of crotyl alcohol are
produced after HTR during the initial hydrogenation pe-
riod, with the selectivity reaching almost 70% for Pt. As
FIG. 1. Arrhenius plots for crotonaldehyde consumption during hy-
drogenation over Pt/TiO2 and Ni/TiO2 reduced at 573 or 773 K (HTR);
PCROALD = 35 Torr.

CROTONALDEHYDE HYDROGENATION
347
that deactivation occurs to some extent for all catalysts, but
is least noticeable with Pt/TiO2–HTR and most severe with
Pt/TiO2 reduced at 573 K; nevertheless, once steady state
is reached, the activities are relatively constant for signif-
icant periods of time. In contrast, the selectivity to crotyl
alcohol did not stabilize over any of the catalysts investi-
gated until very low values were reached, which indicates
that the Pt–TiOx and Ni–TiOx sites favoring crotyl alcohol
formation are deactivated faster than those responsible for
olefinic C == C bond hydrogenation.
Limited IR studies exist to which the current spectra can
be compared. Englisch et al. reported spectra taken dur-
ing this reaction over a Pt/SiO2 catalyst and identified cro-
tonaldehyde adsorbed on the SiO2 surface along with CO
chemisorbed on Pt (11). They attributed the latter, which
wascreated via a decarbonylation reaction, asbeingrespon-
sible for the observed deactivation. Previously, Waghray
and Blackmond had obtained IR spectra during hydro-
genation of 3-methyl-2 butenal over Ru/SiO2 catalysts,
FIG. 3. Activity profiles during crotonaldehyde hydrogenation over
Ni/TiO2 reduced at 573 and 773 K (HTR); TRXN = 333 K; PCROALD =
3
5 Torr.
before, this enhancement in carbonyl bond hydrogenation
is attributed to special sites created at the metal–TiO2 inter-
face by TiOx entities which can migrate and decorate the Pt
and Ni particles (1, 2, 10). The coordinatively unsaturated
+
3
+2
Ti (or Ti ) cations in these suboxide species can interact
with the terminal oxygen in crotonaldehyde to activate the
C == O bond and enhance its reactivity. The presence of ac-
tivated hydrogen from the Pt or Ni surface provides for the
facile hydrogenation of this group and would not require
H spillover over any significant distance (10, 26). There is
one report by Lercher and coworkers that an increase in
selectivity to crotyl alcohol occurs as Pt crystallites become
larger; however, past studies have consistently shown that
the HTR step to induce SMSI behavior produces no metal
sintering (4, 22). Therefore, the average Pt crystallite size
in the two Pt/TiO2 catalysts should be similar and it is un-
reasonable to attribute the enhanced selectivity to crotyl
alcohol over the Pt/TiO2 HTR sample to sintering. The cor-
responding selectivity maintenance profiles as a function of
FIG. 4. Selectivity toward crotyl alcohol obtained during crotonalde-
hyde hydrogenation over Pt/TiO2 and Ni/TiO2 reduced at 573 or 773 K
time on stream are shown in Fig. 4. Figures 2 and 3 indicate (HTR); TRXN = 333 K; PCROALD = 35 Torr; Conv. = 4–15% .

3
48
DANDEKAR AND VANNICE
FIG. 5. In situ DRIFTS spectra during crotonaldehyde hydrogenation over Pt/TiO2 reduced at 573 K as a function of time on stream; TRXN = 333 K.
Vapor-phase crotonaldehyde spectrum removed.)
(
and their principal observation was the growth of the peak the C == C bond in this case argues against the assignment
�
1
for chemisorbed CO, which they stated was the cause of of 1465 and 1382 cm bands to adsorbed crotonaldehyde.
�
1
activity loss (25). The addition of potassium inhibited this The strong bands at 2965, 2939, and 2878 cm lie in the
decarbonylation reaction. Neither study reported spectra frequency region where CHx stretches in hydrocarbons are
for TiO2-supported metals.
usually witnessed (27, 28) and could also correspond to ad-
In situ DRIFTS spectra acquired during crotonaldehyde sorbed crotonaldehyde; however, similar to the peaks at
�
1
hydrogenation at 333 K on Pt/TiO2 after reduction at 573 K 1465 and 1382 cm , the intensity of these peaks is much
are shown in Fig. 5 as a function of time on stream. The greater than that expected from the relatively weak peaks
�
1
reference spectrum is that of the pretreated catalyst just at 1693 and 1641 cm which correspond to vibrational fre-
prior to exposure to the reactant mixture. The vapor-phase quencies characteristic of adsorbed crotonaldehyde. Alter-
�
1
spectrum of crotonaldehyde was obtained and subtracted natively, bands around 2970, 2940, and 2880 cm have been
from these spectra. The spectrum obtained after 5 min ex- assigned previously to CHx stretches in propene adsorbed
hibits bands at 1382, 1465, 1641, 1693, 2067, 2878, 2939, on transition metals (28), and corresponding wagging defor-
�
1
� 1
� 1
and 2965 cm . The 1641 and 1693 cm bands can be as- mations have been observed at 1460 and 1380 cm . Con-
signed to the respective C == C and C == O frequencies of cro- sequently, it is reasonable to assign the bands at 2965, 2939,
�
1
tonaldehyde weakly coordinated on the catalyst surface, 2878, 1465, and 1382 cm to olefinic surface species formed
as indicated by the small red-shifts from the correspond- due to decarbonylation of crotonaldehyde. Bands around
ing vapor-phase frequencies. No bands attributable to a 1693 and 1641 cm were also observed in the spectrum of
strong coordination through the C == C bond appear to ex- crotonaldehyde adsorbed on pure TiO2 after reduction at
ist. Bands near 1460 and 1375 cm are characteristic of 573K (21);consequently, a contribution from these bandsin
the asymmetric and symmetric deformation of CH3 groups the Pt/TiO2 sample is likely. Similarly, the prominent band
in hydrocarbons perturbed by hydrogen bonding with hy- at 2067 cm with a broad tail lies in the region for CO
droxyl groups on the TiO2 surface (27). These bands have adsorbed on transition metals and cannot be attributed to
been observed before only for crotonaldehyde adsorbed adsorbed crotonaldehyde. In conjunction with the activity
strongly through the C == C bond on supported Cu catalysts maintenance profile for this catalyst (Fig. 3) and published
�
1
�
1
�
1
(
20), but the weak stabilization of crotonaldehyde on Pt via literature on Pt/SiO2 (11) and this band is attributed to

CROTONALDEHYDE HYDROGENATION
349
irreversibly adsorbed CO on Pt generated by crotonalde-
hyde decarbonylation over the catalyst during the initial
transient phase.
Several changes occur in the relative peak intensities as
the spectra evolve with time on stream. A prominent band
�
1
� 1
develops around 1525 cm and the 1452 cm band broad-
ens noticeably. The former band can be assigned to cro-
tonaldehyde adsorbed strongly on Pt via the olefinic C == C
bond (28). In the CHx stretching region, a weak peak grows
�
1
at 2732 cm which is characteristic of the aldehydic CH
stretch in crotonaldehyde adsorbed weakly via the C == O
bond (28), and three new bands emerge at 2922, 2862, and
�
1
3
022 cm . Whereas the former two are characteristic of
CHx stretches in adsorbed crotonaldehyde, the last band
has been detected before with supported Cu catalysts (20)
and is tentatively attributed to a new C–H bond formed by
the addition of one H atom to adsorbed crotonaldehyde. In
�
1
addition, the broad tail accompanying the 2067 cm band
�
1
grows into a distinct band at 2026 cm . The former can
be associated with linearly adsorbed CO on Pt, while the
latter is due to a dicarbonyl species stabilized on Pt (28). A
broad, weak band, characteristic of bridge-bonded CO on
�
1
Pt, is also detected at 1821 cm
.
All these bands analyzed together suggest that croton-
aldehyde adsorbs both weakly through the C == O and C == C
bonds and strongly through the C == C on Pt/TiO2 after re-
duction at 573 K, corresponding to the �CO, �CC, and di-
�
CC adsorption modes (27, 29), as shown in Fig. 6. Whereas
only the �CO and �CC modes are detected on the surface
during the initial transient period, the di-�CC adsorption
mode evolves as the system approches steady state. The ab-
sence of any detectable surface coverage of crotonaldehyde
adsorbed strongly through either of the two unsaturated
bonds during the initial transient phase may suggest either
the absence of such interactions or the complete consump-
tion of these more strongly coordinated species in the main
FIG. 6. Different possible adsorption modes of crotonaldehyde over
Pt/TiO2.
reaction as well as the parallel decarbonylation reaction, at steady state, is difficult to assess from the available data;
leading to the formation of CO and lower hydrocarbons. but, based on mechanistic analyses discussed elsewhere (1,
If indeed a species adsorbed in the di-�CO mode does exist 20, 30), the � (C–C) and � (C–O) modes seem to be favored
2
2
during the initial period on stream, it can explain the low but on Pt/TiO after reduction at 573 K.
2
finite initial selectivity toward crotyl alcohol; thus the latter
2
In contrast to Pt/TiO , no evidence for decarbonyla-
proposal is favored. The appearance of the di-�CC mode tion or any other side reaction in detected in the spectra
at steady state further supports this proposition. Although of Ni/TiO after reduction at 573 K, as shown in Fig. 7,
2
one cannot determine exactly which surface sites catalyze which is consistent with the kinetic data discussed earlier
each of these individual reactions from the available in- (Fig. 3). The spectrum obtained after 5 min is very similar
formation, it is possible that as the sites responsible for to that obtained after 120 min on stream, suggesting min-
the decarbonylation reaction as well as crotyl alcohol for- imal changes in the adsorption modes of crotonaldehyde.
mation get saturated by irreversibly adsorbed CO and/or Figure 7 exhibits absorption bands characteristic of weak
hydrocarbon fragments, the overall activity stabilizes, se- �-adsorption of crotonaldehyde through C == O and C == C
�
1
lectivity toward the alcohol drops, and the �CO, �CC, and bonds (1693 and 1636 cm ), a strong �-interaction via the
�
1
di-�CC adsorption modes of crotonaldehyde become visi- C == C bond (1543 cm ), and the terminal –CH deforma-
3
ble. Whether these individual �2(C–C) and �2(C–O) inter- tion modes perturbed due to hydrogen bonding with the
�
1
actions can occur simultaneously to yield an �4(C–C–C–O) TiO surface (1432 and 1376 cm ). In addition, bands char-
2
adsorption geometry, either during the transient phase or acteristic of CH stretches in adsorbed crotonaldehyde are
x

3
50
DANDEKAR AND VANNICE
FIG. 7. In situ DRIFTS spectra during crotonaldehyde hydrogenation over Ni/TiO2 reduced at 573 K as a function of time on stream; TRXN = 333
K. (Vapor-phase spectrum removed.)
�
1
seen at 2961, 2936, 2878, and 2729 cm along with a weak catalyst during the HTR step. Based on these arguments,
�
1
� 1
peak at 3020 cm , which is assigned to a new C–H bond the 1660 cm peak is assigned to crotonaldehyde adsorbed
formed after addition of one H atom to crotonaldehyde ad- more strongly via the C == O bond in a di-� mode on inter-
sorbed in one of the above forms. Similar to Pt/TiO2, these facial Pt–TiOx and Ni–TiOx sites created on these catalysts
�
CO, �CC, and di-�CC interactions can occur independently during HTR. In accordance with the model discussed in
or simultaneously on Ni/TiO2 to yield either the �2 or the detail elsewhere (1, 2, 4, 8–11), the oxygen in the carbonyl
+
3
�
4 adsorption geometry.
groups is envisioned to be coordinated with either Ti (or
+
2
The time-dependent spectra of Pt/TiO2(HTR) in Fig. 8 Ti ) cations at these interfacial sites with the C atom sta-
and those for Ni/TiO2(HTR) in Fig. 9 are significantly dif- bilized on a neighboring Pt site, thus leading to a di-�CO
�
1
1
ferent from those just discussed. The spectra obtained af- adsorption configuration. The shoulder at 1693 cm , as-
�
ter 5 min on stream in both cases exhibit one dominant signed to a �CO configuration, and that at 1630 cm , asso-
�
1
absorption band at 1660 cm with shoulders at 1693 and ciated with a �CC stabilization, indicate that normal weakly
1
�
1
630 cm . Whereas the latter two bands can be attributed adsorbed crotonaldehyde is also present; however, whether
to crotonaldehyde with a weak �-interaction involving the these configurations lead to discrete �2 or overlapping �4 ad-
C == O and C == C bonds, respectively (�CO and �CC modes), sorption geometries is not obvious, but previous analyses
�
1
the shape and position of the 1660 cm peak indicate that of �, �-unsaturated aldehydes adsorbed on catalysts in the
it can be attributed to a crotonaldehyde species with the SMSI state have favored the �4(C–C–C–O) adsorption ge-
C == O bond more strongly coordinated to the catalyst sur- ometry, with the O atom stabilized on the suboxide islands
face, as compared to the weak �-interaction which gives the (1, 10, 30). Concurrent with the decrease in the coverage
�
1
1
693 cm band in most other cases. Spectra for adsorbed of the di-�CO species with time on stream is the growth
�
1
butyraldehyde or crotyl alcohol do not provide peaks that of a strong band at 1548 cm and the strengthening of the
correspond to this 1660 cm peak (20, 28). Figures 8 and 9 1693 cm peak, which are characteristic of crotonaldehyde
�
1
� 1
indicate that the surface concentration of this species, which adsorbed in the di-�CC and �CO configurations, respectively.
is more strongly bonded through the C == O bond, decreases
As mentioned earlier, the variation in the relative cover-
with time on stream and parallels the drop in selectivity to ages of crotonaldehyde adsorbed in these different modes
crotyl alcohol (Fig. 4A and 4B). The fact that this species is synchronous with the shift in the intramolecular product
is detected only after HTR suggests that it corresponds to selectivity detected over these catalysts during crotonalde-
crotonaldehyde adsorbed on special sites created on the hyde hydrogenation (Fig. 4). This implies that the rate of

CROTONALDEHYDE HYDROGENATION
351
FIG. 8. In situ DRIFTS spectra during crotonaldehyde hydrogenation over Pt/TiO2 reduced at 773 K (HTR) as a function of time on stream;
TRXN = 333 K. (Vapor-phase spectrum removed.)
FIG. 9. In situ DRIFTS spectra during crotonaldehyde hydrogenation over Ni/TiO2 reduced at 773 K (HTR) as a function of time on stream;
TRXN = 333 K. (Vapor-phase spectrum removed.)

3
52
DANDEKAR AND VANNICE
FIG. 10. DRIFTS spectra of CO adsorbed at 173 K on Pt/TiO2 reduced at: (A) 573 K and (B) 773 K (HTR).
formation of crotyl alcohol is proportional to the surface
coverage of crotonaldehyde adsorbed in the di-�CO config-
uration at Pt–TiOx and Ni–TiOx interfacial sites, leading to
the very high initial selectivity toward crotyl alcohol. Ev-
idence to support this hypothesis is provided by plotting
the rate of crotyl alcohol formation versus the concentra-
tion of this surface intermediate, as represented by the in-
�
1
tensity of the peak at 1660 cm , and the correlations for
Pt/TiO2 (HTR) and Ni/TiO2 (HTR) are shown in Fig. 12. An
FIG. 11. CO2 evolution profiles during temperature programmed oxi-
dation (TPO) of Pt/TiO2 catalysts used for crotonaldehyde hydrogenation;
FIG. 12. Correlation of the rate of crotyl alcohol formation with the
concentration of adsorbed crotonaldehyde species with a 1660 cm band.
�
1
�
= 5 K/min.

CROTONALDEHYDE HYDROGENATION
353
+
4
obvious dependence exists between crotyl alcohol forma- Ti sites, and observed with DRIFTS, or to unreacted cro-
tion and the surface concentration of this particular inter- tonaldehyde strongly bound to these surface Lewis acid
mediate. Although the relationship is not exactly linear, centers, or to heavier deposits formed due to other side re-
this may well be due to the fact that the rates were ob- actions like aldol condensation. In contrast, the HTR cata-
tained from the microreactor data, rather than from the lyst profile exhibits only a very weak band around 700 K,
sample in the DRIFTS cell which was operating under one- indicating significant suppression of virtually all side reac-
half the reactant pressures used in the microreactor. The tions leading to adsorbed CO as well as carbon deposition.
drop in the selectivity to crotyl alcohol, which parallels the This is consistent with the activity profile of this catalyst
decrease in the surface coverage of this species, suggests which showed agreement between the rates based either on
a reduction in the concentration of these interfacial sites the formation of detectable products or on crotonaldehyde
responsible for stabilizing the di-�CO mode. This could be consumption, and this behavior may be attributed to the
due to blockage ofthese siteswith irreversiblyadsorbed car- formation of the TiOx suboxide phase reducing the popula-
+
4
bonaceous species, reoxidation of these sites, reversal of the tion of surface Ti Lewis acid sites as well as an ensemble
SMSI effect due to water possibly formed under reaction effect caused by the TiOx overlayer which destroys Pt sites
conditions, or surface reconstruction leading to formation active for crotonaldehyde decarbonylation.
larger three-dimensional TiOx islands which decreases the
effective metal–support interfacial area. The precise reason
is unclear at this point.
SUMMARY
A kinetic and DRIFTS study of crotonaldehyde hydro-
genation was conducted over TiO2-supported Pt and Ni
with the intent of gaining insight into the adsorption modes
of carbonyl group-containing molecules on these catalysts
in the SMSI and the non-SMSI states. As reported before,
significant enhancement in the selectivity toward crotyl
alcohol was observed after reduction at 773 K. DRIFT
spectra under reaction conditions identified crotonalde-
An additional feature evident in Fig. 8 is the absence of
any significant amount of CO adsorbed on Pt/TiO2(HTR)
after exposure to crotonaldehyde. Only a very weak band
�
1
can be detected at 2022 cm whose intensity slightly in-
creases with time. Spectra of Pt/TiO2 and Pt/TiO2(HTR)
samples after exposure to 75 Torr CO at 300 K are shown in
Fig. 10. With the sample reduced at 573 K, a significant cov-
erage of CO on Pt is indicated by various linear monocar-
�
1
==
hyde species strongly adsorbed through the C C bond and
weakly coordinated through the C == C and C == O bonds
bonyl species with peaks at 2088, 2069, 2050, and 2027 cm
as well as bridged CO species giving a broad band around
�
1
on these catalysts after reduction at 573 K. After HTR,
an additional adsorbed species was observed with a C == O
1
836 cm (28, 31). This latter catalyst was the only one
showing extensive deactivation; however, decarbonylation
reactions to produce adsorbed CO have been invoked by
Waghray and Blackmond (25) and by Lercher and cowork-
ers (11) to explain such losses in activity. A significant de-
crease in the population oflinear carbonylsisobserved after
�
1
peak position at 1660 cm that implied a strong interac-
tion between the carbonyl group and the surface, which
is attributed to stabilization at interfacial Pt–TiOx and Ni–
TiOx sites. A decrease in the surface coverage of this species
paralleled a drop in selectivity to crotyl alcohol with time
on stream. After reduction at 573 K, decarbonylation oc-
curred during the initial few minutes on stream on Pt/TiO2,
but this reaction was significantly suppressed after HTR at
�
1
HTR, with a single peak at 2075 cm remaining and the
virtual elimination of the bridged species. This behavior has
been observed before and is characteristic of the blocking
of Pt sites by migrating TiOx species (4, 7, 31). In addition
to the manifestation of the above effect, the extremely low
intensity of Pt–CO bands on the Pt/TiO2(HTR) sample un-
der reaction conditions suggests a suppression of the decar-
bonylation reaction after HTR. The TPO profilesofPt/TiO2
samples used for crotonaldehyde hydrogenation, shown in
Fig. 11, support the above proposal. The CO2 evolution
profile for the catalyst which had been reduced at 573 K
prior to exposure to crotonaldehyde exhibits a distinct low
temperature peak around 400 K and a high temperature
doublet between 700 and 800 K. Whereas the former can be
attributed to CO deposited on Pt by decarbonylation of cro-
tonaldehyde, the latter peaks are characteristic of oxidation
of carbonaceous deposits formed during hydrogenation re-
actions (32). These could correspond to the hydrocarbon
fragments formed during crotonaldehyde adsorption and
decarbonylation over Pt and/or coordinatively unsaturated
7
73 K, presumably due an ensemble effect caused by the
TiOx overlayer which removes Pt sites active for croton-
aldehyde decarbonylation.
ACKNOWLEDGMENTS
Partial support for this study was provided by the Department of En-
ergy, Basic Energy Sciences Division, via Grant DE-FG02-84ER13276.
REFERENCES
1. Vannice, M. A., Catal. Today 12, 255 (1992).
. Vannice, M. A., J. Mol. Catal. 59, 165 (1990).
. Tauster, S. J., Fung, S. C., and Garten, R. L., J. Am. Chem. Soc. 100,
2
3
170 (1978).
4
5
. Haller, G. L., and Resasco, D. E., Adv. Catal. 36, 173 (1989).
. Vannice, M. A., and Garten, R. L., J. Catal. 56, 236 (1979).
6. Vannice, M. A., and Garten, R. L., J. Catal. 66, 242 (1980).

3
54
DANDEKAR AND VANNICE
7
. Vannice, M. A., and Twu, C. C., J. Catal. 82, 213 (1983).
20. Dandekar, A., Baker, R. T. K., and Vannice, M. A., J. Catal., in press.
21. Dandekar, A., Ph.D. dissertation, The Pennsylvania State University,
1998.
8. Vannice, M. A., and Sudhakar, C., J. Phys. Chem. 88, 2429 (1984).
. Sen, B., and Vannice, M. A., J. Catal. 113, 52 (1988).
9
1
1
1
0. Vannice, M. A., and Sen, B., J. Catal. 115, 65 (1989).
1. Englisch, M., Jentys, A., and Lercher, J. A., J. Catal. 166, 25 (1997).
2. Wismeijer, A. A., Kieboom, A. P. G., and van Bekkum, H., React.
Kinet. Catal. Lett. 29, 311 (1985).
3. Makouangou, R. M., Murzin, D. Y., Dauscher, A. E., and Touroude,
R. A., Ind. Eng. Chem. Res. 33, 1881 (1994).
4. Raab, C. G., and Lercher, J. A., Catal. Lett. 18, 99 (1993).
5. da Silva, A. B., Jordao, E., Mendes, M. J., and Fouilloux, P., Appl. Catal.
A 148, 253 (1997).
22. Baker, R. T. K., Prestridge, E. B., and Garten, R. L., J. Catal. 56, 390
(1979).
23. Tauster, S. J., Acc. Chem. Res. 20, 389 (1987).
24. Bradford, M. C. J., and Vannice, M. A., Appl. Catal. A 142, 73 (1996).
25. Waghray, A., and Blackmond, D., J. Phys. Chem. 97, 6002 (1993).
26. Neikam, W. C., and Vannice, M. A., J. Catal. 27, 207 (1972).
27. Colthup, N. B., Daly, L. H., and Wiberley, S. E., “Introduction to In-
frared and Raman Spectroscopy,” 3rd ed. Academic Press, San Diego,
CA, 1990.
1
1
1
1
1
6. Vannice, M. A., and Poondi, D., J. Catal. 169, 166 (1997).
7. Lin, S. D., Sanders, D. K., and Vannice, M. A., Appl. Catal. A 113, 59
28. Davydov, A. A., “Infrared Spectroscopy of Adsorbed Species on the
Surface of Transition Metal Oxides.” Wiley, London, 1990.
29. Delbecq, F., and Sautet, P., J. Catal. 152, 217 (1995).
30. Yoshitake, H., and Iwasawa, Y., J. Catal. 125, 227 (1990).
(
1994).
1
1
8. Poondi, D., and Vannice, M. A., J. Mol. Catal. A 124, 79 (1997).
9. Rao, R., Dandekar, A., Baker, R. T. K., and Vannice, M. A., J. Catal. 31. Vannice, M. A., and Twu, C. C., J. Catal. 79, 70 (1983).
71, 406 (1997). 32. Querini, C. A., and Fung, S. C., App. Catal. A 117, 53 (1994).
1