Hydrogenation of crotonaldehyde over Sn/Pt(111) alloy model catalysts

Article abstract of DOI:10.1006/jcat.2001.3452

Gas-phase hydrogenation of crotonaldehyde (CH₃CH=CHCHO) was studied over the (2 × 2) Sn/Pt(111) and (√3 × √3) R30° Sn/Pt(111) surface alloys which were used as model catalysts. The influence of the alloy structure, hydrogen pressure, and temperature on activity and selectivity toward 2-butenol (CH₃CH=CHCH₂OH) formation was investigated. The results were compared to those obtained for the pure Pt(111) surface. All catalysts were characterized prior to use by X-ray photoelectron spectroscopy and low-energy electron diffraction to ensure proper alloy formation. The hydrogenation activity was about two times higher for the bimetallic Pt-Sn catalysts compared to that for Pt(111); however, little change in selectivity was observed. Butyraldehyde was formed as the main product in all cases. Therefore, alloy formation improves hydrogenation activity toward C=C and C=O functional groups and cannot solely explain the improvement in selectivity toward the formation of the unsaturated alcohol that has often been reported for supported Pt-Sn catalysts versus pure Pt catalysts. Both alloys investigated had similar catalytic activity and selectivity. An attempt was made to create oxidic tin species on the Pt(111) surface, because these species have been suggested to improve selectivity to the unsaturated alcohol product. However, the oxidic species formed by heating the alloys in 5 Torr O₂ for 5 min at 600 K were not stable under reaction conditions and were reduced to form a surface with low catalytic activity, probably consisting of metallic Sn rafts over the Pt(111) surface.

Full text of DOI:10.1006/jcat.2001.3452

Journal of Catalysis 205, 278–288 (2002)
doi:10.1006/jcat.2001.3452, available online at http://www.idealibrary.com on
Hydrogenation of Crotonaldehyde over Sn/Pt(111) Alloy Model Catalysts
Dmitri I. Jerdev, A. Olivas, and Bruce E. Koel1
Department of Chemistry, University of Southern California, Los Angeles, California 90089-0482
Received May 22, 2001; revised September 20, 2001; accepted September 28, 2001
In this work, we investigated Pt–Sn surface alloys as
==
3)
model, unsupported catalysts for the gas-phase hydrogena-
Gas-phase hydrogenation of crotonaldehyde (CH3CH
√
√
==
CHCHO) was studied over the (2 2) Sn/Pt(111) and (
3
tion of crotonaldehyde (2-butenal, CH3CH CHCHO). We
R30 Sn/Pt(111) surface alloys which were used as model catalysts.
The influence of the alloy structure, hydrogen pressure, and tem-
compared the activity and selectivity of these alloys to those
of similar Pt(111) model catalysts (4) and Pt–Sn supported
catalysts studied elsewhere (1). This approach allowed us
to address some issues in the mechanism over supported
Pt–Sn catalysts that may result from alloy formation. Using
thesesurfacealloysinsteadofsupportedbimetalliccatalysts
makes it possible to exclude the influence of several factors,
such as metal–support interaction, incomplete reduction of
the metal phase(s), size of the active metal particles, and
sintering, which are known to strongly affect catalytic be-
havior (5).
The selective hydrogenation of , -unsaturated alde-
hydes to the corresponding unsaturated alcohols has in-
dustrial applications. The alcohols obtained are important
intermediates in the production of perfumes, flavorings,
and pharmaceuticals. The challenge is to selectively hydro-
==
perature on activity and selectivity toward 2-butenol (CH3CH
CHCH2OH) formation was investigated. The results were com-
pared to those obtained for the pure Pt(111) surface. All catalysts
were characterized prior to use by X-ray photoelectron spectroscopy
and low-energy electron diffraction to ensure proper alloy forma-
tion. The hydrogenation activity was about two times higher for the
bimetallic Pt–Sn catalysts compared to that for Pt(111); however,
little change in selectivity was observed. Butyraldehyde was
formed as the main product in all cases. Therefore, alloy formation
==
==
improves hydrogenation activity toward C C and C O functional
groups and cannot solely explain the improvement in selectivity
toward the formation of the unsaturated alcohol that has often been
reported for supported Pt–Sn catalysts versus pure Pt catalysts.
Both alloys investigated had similar catalytic activity and selectiv-
ity. An attempt was made to create oxidic tin species on the Pt(111)
surface, because these species have been suggested to improve
selectivity to the unsaturated alcohol product. However, the oxidic
species formed by heating the alloys in 5 Torr O2 for 5 min at 600 K
were not stable under reaction conditions and were reduced to
form a surface with low catalytic activity, probably consisting of
==
==
genate the C O bond leaving the C C bond intact. The
analysis of thermodynamics shows that hydrogenation of
==
==
the C C bond is more favorable than that of C O hydro-
genation, forcing this reaction to be run at low temperatures
using kinetic control. However, the addition of hydrogen
to the C C bond occurs faster than that to the C O bond
over all known catalysts for the gas-phase hydrogenation of
unsaturated aldehydes (6). Research done using pure met-
als revealed that the activity rather than the selectivity of
the catalyst was more strongly affected by using different
metals (6). Most metals hydrogenate both double bonds to
some extent, except Pd which is known to be very selective
c
metallic Sn rafts over the Pt(111) surface.
2002 Elsevier Science
==
==
INTRODUCTION
Pt–Sn bimetallic catalysts are known for their perfor-
mance and stability in catalytic hydrocarbon transforma-
tions. These supported Pt–Sn catalysts have been studied
for hydrogenation–dehydrogenation (1), hydrogenolysis
(2), and recently for selective oxidation of hydrocarbons
(3). In general, Pt bimetallic catalysts exhibit increased
stability against coke formation and catalytic behavior
different from that of the pure Pt catalysts. Observa-
tions that alloys are formed in some catalyst systems
and the possibility that alloys are important phases in
other systems make alloys interesting targets for investiga-
tion.
==
with respect to the C C bond (7). Currently, the only way
to achieve selective hydrogenation to form the unsaturated
alcohol with high yield ( 99%) is by reduction using boron
or metal hydrides (NaBH4 or NaAlH4) (8). This method is
quite valuable in a laboratory but too expensive to carry out
on an industrial scale. Therefore, it is of interest to create
a selective heterogeneous catalyst that would result in an
efficient process for this reaction.
Promoters have been used to improve the selectiv-
ity of metal catalysts for the previously mentioned pro-
cesses. Three groups of promoters are commonly used (9):
alkali metal compounds, transition metal compounds, and
1 To whom correspondence should be addressed. E-mail: koel@
chem1.usc.edu.
278
0021-9517/02 $35.00
c
2002 Elsevier Science
All rights reserved.

HYDROGENATION OF CROTONALDEHYDE OVER Sn/Pt(111)
279
nontransition metal additives. The last group has the most
The procedure used to obtain the Pt–Sn alloys was sim-
prominent effect on catalytic behavior. Because tin belongs ilar to that described by Paffet and Windham (10). Sn was
to this group and is often involved in a number of bimetal- evaporated onto the Pt(111) surface at 300 K in UHV. This
lic catalysts, we investigated the influence of alloyed Sn on was followed by annealing the Pt(111) crystal to 1000 K for
catalytic hydrogenation of crotonaldehyde over a Pt(111) 10 s. To ensure proper alloy formation, both LEED and
surface.
XPS analyses were done on each surface. Prior to catalytic
experiments, the alloy of interest was formed on both sides
of the crystal. Special attention was given to ensure that
the alloy formed on each of two sides of the crystal was
EXPERIMENTAL PROCEDURES
All experiments were conducted in the apparatus shown as similar as possible. LEED appeared the same, and XPS
in Fig. 1, which consists of an ultrahigh vacuum (UHV) revealed differences in the Pt(4f)/Sn(3d) ratio of less than
chamber and an attached, atmospheric pressure batch re- 10% in all cases.
actor. The UHV chamber was pumped by an ion pump
The following procedure was used to obtain a clean
(240 L/s) and a cryo-pump (1500 L/s) and had a base pres- Pt(111) surface. The crystal was sputtered by 1-keV Ar+
10
sure of 1 10
Torr. The chamber was equipped with ions at room temperature for 1 h and subsequently for 5 min
7
a quadrupole mass spectrometer (QMS) for residual gas at 1000 K with simultaneous exposure to 1 10 Torr of
analysis and temperature-programmed desorption (TPD), oxygen; it was then annealed at 1000 K for a few minutes
low-energy electron diffraction (LEED) optics, a 180-mm in UHV. Long sputtering times were used to remove all
sphericalcapacitoranalyzer(SCA)forAugerelectronspec- Sn from the near-surface region. After this cleaning pro-
troscopy (AES), X-ray photoelectron spectroscopy (XPS), cedure, the amount of Sn and carbon on the surface, along
low-energy ion scattering spectroscopy (LEISS), a differen- with most other common contaminants, was below the XPS
tially pumped, 5-keV ion gun, several multichannel-array detection limit.
gas dosers connected to precision leak valves, and sev-
The UHV chamber was connected directly through a
eral metal evaporators. XPS measurements were made 4.5-in. gate valve to a stainless-steel batch reactor with a
by using a dual-anode X-ray source with a Mg anode volume of 480 cm3 that could be pressurized up to 1-atm
(MgK = 1253.6 eV).
pressure. The composition of the gas in the reactor cell was
The Pt crystal had two polished sides, each oriented along analyzed with a Hewlett–Packard 5890 gas chromatograph
the (111) crystal plane and it was spot-welded between two equipped with a flame ionization detector (FID). An evac-
tungsten wires (0.8 mm in diameter) wrapped with Pt foil at uated loop was used to sample the reaction mixture, and an
the spot of contact with the Pt crystal. The crystal could be AT-1000 (0.53-mm-diameter, 30-m-long) capillary column
resistively heated between 300 and 1200 K, as monitored was used for product separation.
by a chromel–alumel thermocouple that was spot-welded
to the side of the crystal.
The Pt crystal was transferred between the UHV analy-
sis chamber and the high-pressure reactor by means of a
FIG. 1. Schematic drawing of the experimental apparatus.

280
JERDEV, OLIVAS, AND KOEL
transfer rod passing through a differentially pumped as- The temperature of the reactor walls was kept at 350 K to
sembly of sliding seals. The reactor was sealed by the slid- avoid condensation of crotonaldehyde and other reaction
ing seal at one end and by the gate valve separating the products. This temperature determined the lower end of
high-pressure cell from the UHV chamber at the other end. the temperature range. The upper end of the temperature
The reaction cell could be evacuated with a turbomolecular range, 395 K, was determined by the high conversion rates
pump, and it took only a few minutes to bring the cell from encountered at short reaction times.
a reaction pressure of 5 Torr to a pressure low enough to
The reaction products detected were n-butane
transfer the sample into the UHV chamber while maintain- (CH3CH2CH2CH3), butyraldehyde (CH3CH2CH2CHO),
10
ing a background pressure of 8 10
the reaction cell were heated to 350 K during reaction mea- (2-butenol, CH3CH CHCH2OH). Under all conditions
surements to decrease the adsorption of crotonaldehyde. studied for Pt(111) and both Pt–Sn alloys, the main
Torr. The walls of n-butanol (CH3CH2CH2CH2OH), and crotylalcohol
==
Crotonaldehyde (Aldrich, 99+%) was purified by vac- hydrogenation product was butyraldehyde. The amount of
uum distillation prior to use. Hydrogen (Matheson, the unsaturated alcohol product crotylalcohol was at most
99.9999%) was used as supplied. GC analysis of the reac- 5 mol%.
√
√
tion mixture prior to introduction into the reactor showed
The activity and selectivity of the ( 3
3)R30 and
99.99 mol% crotonaldehyde with only a small amount (2 2) Sn/Pt(111) alloys were compared at 365 K using
(<0.01 mol%) of light hydrocarbons. We also performed the lowest and highest HC : H2 ratios (1 : 100 and 1 : 800)
blank experiments to ensure zero catalytic activity (below that we explored. These two alloys showed similar catalytic
1% of the least active catalyst studied) of the reactor in the performance, with a 10% higher overall activity for the (2
absence of a catalyst.
2) Sn/Pt(111) alloy and only a marginal difference in the
Catalytic results are described by selectivity, conversion, selectivity. Thus, we chose to present in this paper only the
and turnover frequencies (TOF) as defined in the following. results and conclusions that were specific for the (2 2)
The selectivity Si toward a given product was determined as Sn/Pt(111) alloy. However, these should also be applicable
√
√
to the ( 3
3)R30 Sn/Pt(111) alloy except for the small
ni
differences that were mentioned.
P
Si =
,
[1]
i ni
There was no evidence for surface carbon buildup under
any conditions on the alloy catalyst. XPS measurements
of the carbon coverage taken 20 s after the introduction
of the alloy catalyst into the reaction cell showed essentially
the same amount of carbon (2C = 0.5 monolayer (ML))
as that obtained after 30 min of reaction. This was a conse-
quence of the high H2 : HC ratio that was used, and allowed
us to consider that TOF values were not influenced by this
factor. The carbon coverage was calibrated with XPS by
using ethylene to produce a known coverage of ethylidyne
on Pt(111) at 300 K of 2C = 0.5 ML.
where ni is the mole fraction of product i in the reaction
mixture. The conversion C (mol%) is defined as
.
X
C =
1
nc
ni
100,
[2]
where nc is the mole fraction of crotonaldehyde in the
P
reaction mixture, and
ni is a sum over all components
in the reaction mixture. The TOF is defined as the number
of crotonaldehyde molecules that undergo hydrogenation
per surface atom per second according to
Figure 2 shows the composition of the reaction mixture
over the (2 2) Sn/Pt(111) alloy catalyst as a function of
time. The main product of the reaction was butyraldehyde
VR PCH NA (1 XCH
)
TOF =
,
[3]
R
T
t
2pt Spt
==
from hydrogenation of the C C bond, and only a small
where VR is the reactor volume (3.84 10 4 m3), PCH is the
initial pressure of crotonaldehyde in the reaction mixture
(Pa), NA is Avogadro’s number, XCH is the mol% of croton-
aldehyde in the reaction mixture, R is the gas constant, T is
the temperature of the reaction mixture (350 K), 2Pt is the
surface atomic density of Pt(111) (1.505 1015 atoms/cm2),
t is the reaction time (seconds), and S is the total surface
area of the crystal (1.7 cm2). No correction was made a pri-
ori for the reduced number of Pt surface atoms in the alloys.
portion of crotonaldehyde was converted to the unsatu-
rated alcohol 2-butenol from hydrogenation of the C O
==
bond. This was a general result for all sets of experimental
conditions that were explored.
The influence of temperature on the reaction product
yield was monitored at a fixed value of H2 : HC = 100 and
Ptotal = 5 Torr over a temperature range from 355 to 385 K.
These data are shown in Fig. 3. The rates of all monitored
==
==
processes, i.e., hydrogenation of C O and C C functional
==
groups and C O bond cleavage, increased with increas-
ing temperature. No new products were detected at this
temperature range. Using these data, we calculated an ap-
Catalytic hydrogenation of crotonaldehyde was carried parent activation energy for the hydrogenation of the C C
RESULTS
==
out for a range of temperatures between 350 and 395 K. double bond of Eapp = 8.6 kcal/mol at <10% conversion.

HYDROGENATION OF CROTONALDEHYDE OVER Sn/Pt(111)
281
FIG. 2. Composition of the reaction mixture over the (2 2) Sn/Pt(111) surface alloy for the initial reaction conditions: PH = 10 Torr, P2-butenal
=
2
2
5
10 Torr, and T = 355 K.
==
This was done by assuming that the rate of C C bond hy-
Figure 6 quantifies the influence of H2 partial pressure
==
drogenation was much higher than that of the C O bond. on the total conversion of crotonaldehyde. As shown in
==
This was a reasonable assumption because C C bond hy- Fig. 6, experiments performed under the same conditions
drogenation resulting in butyraldehyde formation was the on Pt(111) revealed that the order with respect to H2 was
predominant reaction pathway, as shown in Fig. 3.
Arrhenious plots for our results, as well as those for ref- which was 0.9. Because the rate of C C bond hydrogena-
==
0.5, significantly lower than that for both Pt–Sn alloys
==
erence data from supported catalysts [1, 11] and another tion was much higher than that of the C O bond, this order
==
Pt(111) model catalyst, are shown in Fig. 4. To produce was mostly related to C C bond hydrogenation.
these plots for the reference catalysts, we used the hydro-
genation rate reported at a specific temperature and ex-
tended this through the indicated temperature range us-
Influence of Oxygen Pretreatment
ing the reported activation energies. Unsupported catalysts
In an attempt to explore the role of metal cations or ox-
displayed higher overall activities than supported catalysts, idic species on the selectivity for this catalysis, we tried
although one cannot make too much of this difference, be- to produce such species by direct oxidation of the Sn–Pt
cause no correction was made for differences in the pres- alloy model catalysts. For these experiments, the (2 2)
sures of crotonaldehyde and hydrogen that were used. The Sn/Pt(111) alloy was prepared as usual, but it was then ex-
trends on the supported catalysts were similar to those posed to 1 Torr O2 at 600 K for 5 min in the high-pressure
in our data on the unsupported catalysts. Pt–Sn catalysts cell prior to use as a catalyst. This treatment produced SnOx
had higher activity and lower apparent activation engergies species at the surface (12), which we expected would then
than pure Pt catalysts for both supported and unsupported improve the selectivity toward formation of unsaturated
catalysts.
aldehyde based on previous proposals in the supported
The performance of the (2 2) Sn/Pt(111) model catalyst catalyst literature (13). In addition, we hoped that this
at 365 K was tested at a crotonaldehyde partial pressure of experiment would check how exposure of the catalyst to
2
5
10 Torr for several different HC : H2 ratios: 1 : 100, conditions similar to those used industrially for catalytic
1 : 200, 1 : 400, and 1 : 800. Figure 5 shows these results. The regeneration might influence catalytic properties.
rates of formation for all products were positively influ- XPS was used to monitor the amount of oxygen and
enced by H2 partial pressure except that for n-butane which oxidation states of Sn and Pt before and after oxidation
remained nearly constant. The existence of a critical hydro- and upon completion of catalytic reactions, as shown in
gen partial pressure, above which the reaction rate stops Fig. 7. The 0.8-eV shift of the Sn(3d) peak after the fresh
increasing with an increase in H2 partial pressure, was not (2 2) Sn/Pt(111) alloy was exposed to oxygen indicated
encountered in these measurements. This indicates that the the oxidation of Sn. The total oxygen coverage following
==
rates of all of the reactions, except that of C O bond cleav- this oxidation pretreatment was 1.2 ML. The determina-
age, are proportional to the surface hydrogen concentration tion of the Sn oxidation state (II or IV) using XPS is usu-
which is involved in the limiting steps of these reactions.
ally regarded as difficult, because nearly identical chemical

282
JERDEV, OLIVAS, AND KOEL
FIG. 3. Composition of the reaction mixture over the (2 2) Sn/Pt(111) surface alloy for several temperatures under the following conditions:
2
PH = 5 Torr and P2-butenal = 5 10 Torr.
2
shifts were reported (14) for those oxidation states. Our tunneling microscopy (STM) studies on these same Sn–Pt
XPS results for these surfaces (15) allowed us to exclude alloys (17) showed that oxidation takes place by abstraction
formation of SnO2 (Sn IV) under these conditions, and we of Sn from the alloy to form various SnOx adlayers at the
assigned the oxidized forms of Sn shown in Fig. 7b to the surface. Reduction of this adlayer forms Sn that remains
so-called “quasimetallic” form (16) of Sn (485.6 eV binding in an adlayer, possibly as Sn rafts, blocking most of the
energy (BE)) and SnO (486.15 eV BE). The total Sn frac- Pt(111) surface. This is because alloying of Sn on Pt(111)
tion that was oxidized in one form or the other was equal requires high temperatures (>600 K) (18). Because Sn
to 0.9.
itself is not a good catalyst, this results in a sample with
Table 1 shows that this oxygen treatment significantly low catalytic activity.
decreased the activity of the (2 2) Sn/Pt(111) alloy in
We were able to restore the activity of the preoxidized
subsequent catalytic reactions. The catalytic activity for this catalyst to that of the normally prepared surface alloy by
samplewaslowerthanthatforeitherthepurePt(111)orthe reduction in 1 Torr H2 at 600 K for 1 min. This result in-
(2 2) Sn/Pt(111) alloy. However, it is evident from Fig. 7c dictes that oxidation–reduction cycles at 600 K can form
that when the oxidized Pt–Sn alloy was exposed at 365 K to an alloy catalyst. Oxygen treatments are commonly used to
the reaction mixture containing hydrogen, Sn was comp- remove surface carbon deposits, but in the case of Pt–Sn
letely reduced to metallic form. Recent XPS and scanning bimetallic catalysts, this treatment also enriches the surface

HYDROGENATION OF CROTONALDEHYDE OVER Sn/Pt(111)
283
2
FIG. 4. Arrhenious plots for unsupported Pt(111) and Sn–Pt surface alloys (PH = 5 Torr and P2-butenal = 5 10 Torr) with reference data on
supported (1 wt% Pt)/C, (1 wt% Pt, 0.25 wt% Sn)/C, (0.7 wt% Pt)/SiO2 and unsupp2orted Pt(111) catalysts.
of the catalyst with Sn. The formation of alloy phases may differential equations; for example,
occur during the subsequent reduction or “activation’ stage.
It is useful to compare these results on a preoxidized cata-

dC0/dt = (k1 + k2 + k3) C0





lysttosimilarexperimentsperformedbyColomnaetal. (19)
on silica and alumina supported PtSn catalysts. They used
oxygen pretreatment to elucidate the role of oxidized tin
speciesinthecatalyticbehaviorofPtSn/SiO2 catalysts. They
reportedreducedactivityforoxidizedsamplescompared to
those not oxidized, and our observations are similar to
theirs. They also reported that the total amount of oxi-
dized tin was 83%, which is close to our value of 90%. In
addition, theyreportedsomeincreaseintheactivityofthese
catalysts when the oxidation was performed at elevated
temperatures (673 K). This treatment may form ordered
tin oxide phases, based on studies done in our group (20),
showing that tin oxide overlayers order into well-defined
structures on the Pt(111) surface at temperatures above
800 K. This phenomenon may occur on supported catalysts
at lower temperatures because of the various metal surface
planes exposed, smaller metal particle size, or higher oxy-
gen pressure, and it may account for the changes observed
in catalytic activity.
dC1/dt = k1 C0 k4 C1
dC2/dt = k2 C0 k5 C2
dC3/dt = k3 C0


[4]







dC4/dt = k5 C2 + k4 C1,
where C0, C1, C2, C3, and C4 are molar concentrations of
crotonaldehyde, butyraldehyde, 2-butenol, n-butane, and
n-butanol, respectively. At t = 0, C0 = 100% and the con-
centrations of all products were zero.
Within this model, crotonaldehyde can be hydrogenated
to butyraldehyde, 2-butenol, and to n-butane. The first two
products are considered to be intermediates that could
be hydrogenated further to form n-butanol. The solu-
tion to Eq. [4] provides values for these rate constants.
This system of equations was solved numerically using the
Euler method. The solutions to Eq. [4] at Ptotal = 10 Torr,
HC/H2 = 1 : 200, and T = 365 K are given as solid lines
along with the experimental data points in Fig. 2. All of
the rate constants determined in this manner are listed in
Table 2. Because the model curves describe the experimen-
tal behavior quite well, we believe that the model shown in
Fig. 8 is a reasonable approximation.
DISCUSSION
To obtain kinetic parameters for 2-butenal hydrogena-
tion reactions, we constructed a kinetic model according to the C O bond by the (2 2) Sn/Pt(111) catalyst is 18 times
the simplified reaction scheme shown in Fig. 8. All reaction slower than that of the C C bond. Also, the rate constants
This kinetic analysis shows that the hydrogenation rate of
==
==
pathways that were considered are shown as arrows, with for primary (from crotonaldehyde, k1) and secondary (from
==
a rate constant for each pathway shown above the arrow. 2-butenal, k5) C C bond hydrogenation are similar (1.66
This reaction network can be represented as a system of 10 1 versus1.05 10 1), whilerateconstantsfortheprimary

284
JERDEV, OLIVAS, AND KOEL
2
FIG. 5. Influence of the hydrogen pressure on the catalytic activity of the (2 2) Sn/Pt(111) surface for the initial conditions: P2-butenal = 5 10
Torr and T = 365 K.
==
(k2) and secondary (k4) C O bond hydrogenation differ by
TPD studies of crotonaldehyde adsorption on Pt(111)
a factor of 3 (9.0 10 3 versus 3.0 10 3). The presence of (21) revealed complete decomposition to yield only H2 and
==
==
the C C double bond promotes hydrogenation of the C O CO desorption products under UHV conditions. No hydro-
bond in crotonaldehyde over that in butyraldehyde which carbon desorption was observed. This indicates strong, irre-
==
hasnoC Cbond. Thiscouldresultfromeitheranincreased versible chemisorption. At submonolayer coverages of cro-
==
residence time for crotonaldehyde or possibly a change in tonaldehyde, the (C O) stretching mode was observed at
==
the reaction mechanism due to conjugation between C C nearly the same position as that for a condensed phase, mul-
==
and C O bonds.
tilayer sample, but it had a weak intensity. Jesus and Zaera
==
A discussion of the mechanism of crotonaldehyde hy- (21) concluded that the C O bond was oriented parallel to
drogenation on Pt–Sn alloys benefits from insights that the surface and interacted only weakly with it. It was con-
arise from basic studies of adsorption on Pt and Pt–Sn sur- cluded that crotonaldehyde was strongly chemisorbed on
faces. Acrolein and crotonaldehyde adsorption on Pt(111) the Pt(111) surface by nearly complete rehybridization of
was studied by Jesus and Zaera (21) by using reflection– the C C bond to give a peak at 1077 cm 1 and a formation
==
absorption infrared spectroscopy (RAIRS) and TPD. To of two Pt–C bonds with the surface, as shown in Fig. 9a.
our knowledge, no similar work using these molecules on Semiempirical extended Huckel calculations support this
Pt–Sn alloys has been done yet, but such studies are planned picture (22). A limiting case for di- -bonded adsorption
==
through rehybridization of the C O bond is shown in
in our laboratory.

HYDROGENATION OF CROTONALDEHYDE OVER Sn/Pt(111)
285
2
FIG. 6. Reaction order plot for the (2 2) Sn/Pt(111) surface alloy and Pt(111) surfaces for P2-butenal = 5 10 Torr and T = 365 K.
==
==
Fig. 9b. However, carbonyl functional groups usually in- C C and C O double bonds are conjugated and oriented
teract weakly with Pt surfaces by forming a dative bond via in the plane parallel to the surface as shown in Fig. 9c.
the oxygen atom (23). We should not exclude a possible
The adsorption energy for crotonaldehyde on these sur-
role for -bonded crotonaldehyde molecules in which the faces may help to explain the different dependencies on
hydrogen pressure. We observed a positive order with re-
spect to hydrogen pressure on all surfaces, with values of 0.9
on the Pt–Sn alloys and 0.5 on Pt(111). For hydrogenation
==
of the C C bond, the rate should depend on both surface
hydrogen concentration and crotonaldehyde adsorption
energy. Previous studies of olefins (24) lead us to expect that
crotonaldehyde is more weakly bound to the alloys than to
Pt(111), andthusH2 pressureshouldhaveamoreimportant
effect on the rate on the alloys than on the Pt(111) surface. If
breaking the Pt–C bond is involved in the rate-determining
TABLE 1
Hydrogenation Activity of Model Catalysts for PH /PHC = 100,
2
PH = 5 Torr, and T = 365 K
2
Overall
Sn
Ea(app)
hydrogenation
1
Catalyst
coverage (kcal/mol) activity (TOF) s
Pt(111)
(2 2) Sn/Pt(111) alloy
0
0.25
0.33
10.8 (11.0 (4)) 0.13 (0.31 (4))
8.6
0.21
0.18
√
√
(
3
3)R30 Sn/Pt(111)
alloy
(2 2) alloy + oxidation
in 1 Torr O2 at 600 K
for 5 min
Oxidized (2 2) alloy +
reduction in 1 Torr of
H2 at 600 K for 1 min
0.05
0.25
0.25
FIG. 7. Sn(3d) peaks in XPS for alloys before (a) and after (b) oxi-
dation of the (2 2) Sn/Pt(111) alloy and after (c) the oxidized alloy was
used as a catalyst.
Note. Measured at P = 3 Pa, T = 330 K. The original TOF value of
5.6 10 was extrapolated to 365 K by using the reported activation
energy of 46 kJ/mol (4).
2
1
s

286
JERDEV, OLIVAS, AND KOEL
FIG. 8. Reaction network for 2-butenal hydrogenation.
==
==
step, this could also explain the lower apparent activation gies for C C and C O hydrogenation are very close (4),
energies on the alloys compared to those on Pt(111), which an increase in temperature will favor n-butane production.
are 8.6 kcal/mol for the two Pt–Sn alloys and 10.8 kcal/mol
for Pt(111).
Another feature observed in Figs. 3 and 5 was that there
were two distinct regimes in the rate of n-butane produc-
The full hydrogenation of 2-butenal to n-butane proceeds tion; i.e., all curves for the n-butane yield clearly displayed
via the cleavage of the C–OH bond, as shown in Fig. 9d. This a “break” or change in slope as a function of time. The
hydrogenation pathway, as we mentioned when discussing first regime was observed at low conversion when the most
Fig. 5, is the only one which does not depend on the hydro- abundant component in the reaction mixture was croton-
gen partial pressure. The overall process for producing bu- aldehyde which could be characterized by a fairly high rate
tane includes a number of elementary steps such as C–OH of n-butane production. The second regime was observed at
bond cleavage, hydrogenation of surface intermediates, and higher conversions and had a lower rate primarily because
desorption of the butane and water products. Surface sci- n-butane formation under these conditions occurred from
ence studies showed that desorption of butane (25) and butyraldehyde. This interpretation implies that adsorbed
==
water (26) was very fast under catalytic conditions. Among species containing a C C double bond are more suscepti-
the remaining steps, only C–OH cleavage does not directly ble to C–OH cleavage.
involve hydrogen, and we can conclude that this must be
Themaindifferenceincrotonaldehydehydrogenationon
the limiting step for the full hydrogenation process. The Pt–Sn surface alloy catalysts compared to that on Pt(111)
fact that hydrogenation of alkyl groups is not rate limiting is an increased activity for all hydrogenation reactions that
==
==
is consistent with expectations based on observations that were monitored. Both C O and C C hydrogenation reac-
such reactions are facile on Pt(111) and these Pt–Sn alloys. tions were promoted to an equal extent. One factor influ-
If we assume that C–OH cleavage is the limiting step for encing this increased activity is a change in the electronic
n-butane formation, then we can use the rate of n-butane state of Pt that is induced by alloying with Sn. This change
formation versus temperature data to calculate an activa- influences the bond strength between the adsorbed species
tion energy for C–OH bond cleavage. We determine a value and the surface. It was reported that such hydrogenation re-
of 11.2 kcal/mol on the (2 2) alloy. If the alloys follow the actions occured faster on surfaces with decreased electron
same trend as that on Pt(111), where the activation ener- density (27). While no significant shifts occur in the Pt core
levels due to formation of the surface alloys (28), there is a
reduction in the density of states at EF that is revealed in
TABLE 2
ultraviolet photoelectron spectroscopy (29). Furthermore,
recent scanning tunneling microscopy studies (30) reported
an “electronic contrast” for Pt atoms depending on the
number of Sn neighbors bonded to a Pt atom, and this gives
direct evidence for charge transfer from Pt to Sn to reduce
the electron density at Pt sites. This then presents a consis-
tent picture for explaining our catalytic results. Second, the
ensemble size of contiguous reactive Pt atoms at the surface
is changed by alloying with Sn. However, since hydrogena-
tion reactions do not require large ensembles, the reduced
ensemble size for the Pt–Sn alloys should not influence this
particular type of reaction.
Rate Constants for the Different Reaction Pathways Shown in
Fig. 7 on the (2 2) Sn/Pt(111) Alloy Catalyst for PH /PHC = 200,
2
PH = 10 Torr, and T= 365 K
2
1
Rate constant
TOF (s
)
Ea(app) (kcal/mol)
k1, n-Butenal
k2, 2-Butenol
k3, n-Butane
k4, n-Butanol
k5, n-Butanol
Overall TOF
0.166
0.009
0.010
0.003
0.105
0.29
8.6
11.2

HYDROGENATION OF CROTONALDEHYDE OVER Sn/Pt(111)
287
FIG. 9. Reactive species expected to be present on the metal surface during hydrogenation of crotonaldehyde and possible reaction mechanisms.
It is useful to compare our newly obtained results for phases of specific composition and structure, thus defining
well-defined, unsupported Pt–Sn alloys to results from Pt the role that these phases play in supported catalysts.
and Pt–Sn catalysts supported on carbon black (1). An in-
crease in the overall activity and a decrease in the appar- influence hydrogenation selectivity of unsaturated aldehy-
==
As mentioned earlier, one of the major factors thought to
ent activation energy were observed on both unsupported des is a higher preference for adsorption via the C O bond
==
and supported Pt–Sn catalysts with respect to Pt catalysts. versus the C C bond. In the model proposed by Marinelli
The primary difference was that the TOF values reported and Ponec (31), an important factor facilitating adsorption
==
for the supported catalysts (1) were about 10 times lower via the C O bond was the presence of metal cations present
than those obtained in this study. Furthermore, the sup- in oxidic phases at the catalyst surface. These species could
ported catalyst with a 25% Sn loading had twice the overall be produced by partial oxidation of the primary catalytic
activity of a pure Pt supported catalyst, and we also ob- metal component, e.g., Pt (1), during catalyst pretreatment
served that the (2 2) Sn/Pt(111) catalyst, which had 25% and by oxidation of the more electropositive promoters
Sn atoms in the surface layer, had twice the overall activity such as Sn or Fe. Based on similar activities of reduced Pt–
ofthePt(111)catalyst. However, increasingtheSncoverage Sn and Pt–SnOx catalysts, it was concluded that the promo-
√
√
further to 33% for the ( 3
3)R30 Sn/Pt(111) alloy re- tion effect was a result of cations which survived reduction
sulted in a decrease of 10% in the TOF, while for supported or were formed upon initial contact with the aldehyde (32).
catalysts, the highest activity was observed for a catalyst The latter possibility was supported by the observation that
with 50% Sn loading. We also did not observe a signifi- the activity increased when CO pulses were added to the
cant change in selectivity between pure Pt and Pt–Sn model reaction mixture (31). Our results on the reduced Pt–Sn al-
catalysts, whereas supported Pt–Sn catalysts exhibited loy model catalyst show that alloy phases alone do not lead
==
improved selectivity for C O bond hydrogenation with in- to an increase in selectivity. To this extent, our results are
creased Sn loading. consistent with the explanation of Marinelli and Ponec (31)
These differences are revealing about the origin of the concerning supported catalysts.
improved performance of the supported catalysts because
The results after preoxidation of the (2 2) Sn/Pt(111)
the unsupported catalysts are quite different, and more de- alloy differ significantly from those obtained previously for
fined, from the supported catalysts. One difference is that freshly prepared Pt–Sn catalysts supported on carbon black
the morphology of the supported catalyst is obviously dif- (1) in which most of the Sn and a significant amount of
ferent from that of the model catalysts. The influence of the Pt were oxidized. While we found complete reduction
the support is an important factor that could affect the of SnOx on our model catalysts, only 11% of the total Sn
metal particle size, shape, and electron density, and change loading was reduced on the supported catalyst even after
adsorption energies, geometries, and reaction mechanisms. thermal treatment in hydrogen (12 h in H2 flow at 760 Torr
“Spillover” effects that lead to communication between ad- at 623 K). Carbon black is an inert support, interacting only
sorbed species on the support and metal particles can also weakly with supported metal particles; thus, the observed
occur. Changing the Sn loading might and should affect difference in the extent of Sn reduction must be a result of
all of these interactions. Also, high Sn loadings may also the nature of the phase and domain size of SnOx formed
result in larger amounts of unreduced Sn present in the and/or of the chemical state of Pt. On carbon black some
catalyst that are not incorporated into Pt–Sn alloy phases. of the supported Pt prior to reduction was found to be in
Our unsupported catalysts reveal the importance of Pt–Sn oxidic form (i.e., Pt (II)). We did not observe any changes in

288
JERDEV, OLIVAS, AND KOEL
the Pt(4f) XPS peak position after our oxygen treatment,
indicating that oxidation under our conditions preserved
Pt in a metallic state. This decreased the thermal stabil-
ity of SnOx species (33) and increased their reducibility
by H2.
Further work is needed to clarify the influence of oxidic
species on selectivity. Because SnOx species formed by ox-
idation of Pt–Sn alloys are immediately reduced, we plan
work in the future with admetals forming more stable ox-
ides. The change from Sn to a more electropositive metal
such as Fe should increase the stability of oxidic species
and allow them to withstand reaction conditions that com-
pletely reduce SnOx .
REFERENCES
1. Coloma, F., Sepulveda-Escribano, A., Fierro, J. L. G., and Rodriguez-
Reinoso, F., Appl. Cat. A Gen. 136, 231 (1996).
2. Szanyi, J., Anderson, S., and Paffet, M. T., J. Catal. 149, 438 (1994).
3. Bodke, A. S., Olschki, D. A., Schmidt, L. D., and Ranzi, E., Science
285, 712 (1999).
4. Beccat, P., Bertolini, J. C., Gauthier, Y., Massardier, J., and Ruiz, P.,
J. Catal. 126, 451 (1990).
5. Balakrishnan, K., and Schwank, J., J. Catal. 68, 42 (1981).
6. Bond, G. C., “Catalysis of Metals.” Academic Press, London, 1962.
7. Rylander, P. N., in “Catalytic Hydrogenation in Organic Synthesis,”
p. 72. Academic Press, New York, 1979.
8. Ponec, V., Appl. Catal. A Gen. 149, 27 (1997).
9. Marinelly, T. B. L. W., Vleeming, J. H., and Ponec, V., in “Proceedings,
10th International Congress on Catalysis, Budapest, 1992” (L. Guczi,
F. Solymosi, and P. Tetenyi, Eds.), Vol. 2, p. 1211. Akade´miai Kiado´,
Budapest, 1993.
CONCLUSION
10. Paffet, M. T., and Windham, R. G., Surf. Sci. 34, 208 (1989).
Gas-phase hydrogenation of crotonaldehyde over 11. Vannice, M. A., and Sen, B., J. Catal. 115, 65 (1989).
12. Batzill, M., Beck, D. E., and Koel, B. E., Appl. Phys. Lett. 78, 2766
Pt(111) and two ordered surface alloys, a (2 2) Sn/Pt(111)
√
√
(2001).
and a ( 3
3)R30 Sn/Pt(111) structure, was studied at a
13. Gallezot, P., and Richard, D., Catal. Rev. 40, 81 (1998).
14. Weaver, J. F., Campbell, T. J., Hoflund, G. B., and Salaita, G. N.,
J. Electron. Spectrosc. Relat. Phenom. 106, 81 (2000).
temperature range from 355 to 395 K, a H2 pressure range
from 5 to 40 Torr, and a HC/H2 ratio range from 1 : 100
to 1 : 800. The Pt–Sn alloy catalysts displayed about twice 15. Jerdev, D., and Koel, B. E., submitted.
16. Rotermund, H. H., Penka, V., DeLouise, L. A., and Brundle, C. R.,
the activity of the Pt(111) model catalyst, with the activity
of the (2 2) Sn/Pt(111) about 10% higher than that for
J. Vac. Sci. Technol. 5, 1132 (1987).
√
√
17. Batzill, M., Beck, D., and Koel, B. E., submitted.
18. Paffet, M. T., and Windham, R. G., Surf. Sci. 208, 34 (1989).
19. Colomna, F., Llorca, J., Narcis, H., Ramirez de la Piscina, P., Rodrigez-
Reinoso, F., and Sepulvedo-Escibano, A., Phys. Chem. Chem. Phys. 2,
3063 (2000).
20. Batzill, M., Beck, D. E., Jerdev, D., and Koel, B. E., J. Vac. Sci. 19, 1037
(2001).
21. Jesus, J. C., and Zaera, F., Surf. Sci. 99, 430 (1999).
22. Delbecq, F., and Sautet, P., J. Catal. 152, 217 (1995).
the ( 3
3)R30 Sn/Pt(111) alloy. However, the selec-
tivity observed on all of the model catalysts was similar,
with butyraldehyde formed as the main product. We con-
clude that the improvement in selectivity toward formation
of the unsaturated alcohol that has often been reported for
Pt–Sn catalysts versus supported pure Pt catalysts cannot
be solely due to Pt–Sn alloy formation. Pt–Sn alloys that
were preoxidized at 600 K in O2 had a lower catalytic ac- 23. Marinelli, T. B. L. W., Nabuurs, S., and Ponec, V., J. Catal. 151, 431
(1995).
tivity than either of the Pt–Sn surface alloys or the Pt(111)
surface. We propose that this results from dealloying Sn to
form the SnOx adlayer during oxidation followed by re-
24. Tsai, Y. L., Chen, X., and Koel, B. E., Surf. Sci. 358, 37 (1997).
25. Salmeron, M., and Somorjai, G. A., J. Phys. Chem. 85, 3835 (1981).
26. Wagner, F. T., and Moylan, T. E., Surf. Sci. 191, 121 (1987).
duction of the oxidic Sn to an inactive metallic Sn adlayer
during low-temperature catalytic reaction. The catalytic ac-
tivity of the preoxidized alloys can be restored to that for
the as-prepared alloys by H2 reduction at a temperature
exceeding 600 K.
27. Blackmond, D. G., Oukaci, R., Blank, B., and Gallezot, P., J. Catal.
131, 401 (1991).
28. Paffett, M. T., and Windham, R. G., Surf. Sci. 208, 34 (1989).
29. Paffett, M. T., Gebhard, S. C., Windham, R. G., and Koel, B. E., Surf.
Sci. 223, 449 (1989).
30. Batzill, M., Beck, D. E., and Koel, B. E., Surf. Sci. 466, L821 (2000).
31. Marinelli, T. B. L. W., and Ponec, V., J. Catal. 156, 51 (1995).
32. Marinelli, T. B. L. W., Nabuurs, S., and Ponec, V., J. Catal. 151, 431
(1995).
ACKNOWLEDGMENT
This work was supported by the Division of Chemical Sciences, Office
of Basic Energy Sciences, U.S. Department of Energy.
33. Saliba, N. A., Tsai, Y. L., and Koel, B. E., J. Chem. Phys. 103, 1532
(1999).