The Effect of Potassium and Tin on the Hydrogenation of Ethylene and Dehydrogenation of Cyclohexane over Pt(111)

Article abstract of DOI:10.1006/jcat.1998.2122

Stable ordered structures of Sn on Pt(111) (maximum coverage of 0.33 monolayers) were used as model catalysts to test the effect of tin and potassium on the hydrogenation of ethylene at 300 K and dehydrogenation of cyclohexane at 573 K and at pressures of 15 Torr of hydrocarbon and 100 Torr of H₂. Co-adsorption of tin and potassium on Pt(111) resulted in the direct interaction between potassium, tin, and platinum as verified by temperature-programmed desorption of CO. Tin deposition yielded a maximum in the turnover rate as a function of Sn coverage for ethylene hydrogenation and cyclohexane dehydrogenation with maxima at about 0.2 monolayers of tin and a turnover rate 75% higher than on clean Pt(111). This enhancement was explained by a lower rate of deactivation as tin was added. In contrast, the addition of potassium to Pt and Pt/Sn produced only a monotonic decrease in cyclohexane dehydrogenation. In an industrial system, where a higher tin coverage is used, interaction of tin with potassium may form an effective site blocker which could lower deactivation rates.

Full text of DOI:10.1006/jcat.1998.2122

JOURNAL OF CATALYSIS 178, 66–75 (1998)
ARTICLE NO. CA982122
The Effect of Potassium and Tin on the Hydrogenation of Ethylene
and Dehydrogenation of Cyclohexane over Pt(111)
Yong-Ki Park,1 Fabio H. Ribeiro,2 and Gabor A. Somorjai
Materials Sciences Division, Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, California, 94720
and Department of Chemistry, University of California, Berkeley, California 94720
Received October 16, 1997; revised April 8, 1998; accepted April 28, 1998
cation, the challenge has been how to produce a catalyst
that will be stable and selective at the high temperature
and low hydrogen partial pressure required to shift the
equilibrium from the hydrocarbon toward the olefin. Thus,
catalytic systems which can increase dehydrogenation se-
lectivity and reduce coke formation at high temperature
are highly desirable. Catalyst systems that employ Pt/Sn on
a neutral support have been reported to exhibit high dehy-
drogenation selectivity and catalyst stability for dehydro-
genation of light paraffins at elevated temperatures (1–3).
In addition to the role of Sn, the dehydrogenation selectiv-
ity and deactivation stability are increased by the addition
of potassium to a bimetallic Sn/Pt/silica catalyst (4, 5). In
particular, Cortright and Dumesic (5) have reported that a
1 :1 :2.7 Pt/Sn/K/SiO2 catalyst could operate with minimum
deactivation for many hours at 773 K with no H2 added
to the feed. These authors explained the synergistic effect
of tin on the basis of a decrease in the size of surface Pt
ensembles, suppressing hydrogenolysis and isomerization.
Potassium (in the oxidized form) would also act as a block-
ing agent. Through microcalorimetric studies of ethylene,
they also suggested that addition of Sn inhibits the forma-
tion of highly dehydrogenated surface species, leading to
an increase in isobutene selectivity and a decrease in coke
formation.
Stable ordered structures of Sn on Pt(111) (maximum coverage
of 0.33 monolayers) were used as model catalysts to test the effect of
tin and potassium on the hydrogenation of ethylene at 300 K and de-
hydrogenation of cyclohexane at 573 K and at pressures of 15 Torr
of hydrocarbon and 100 Torr of H2. Co-adsorption of tin and potas-
sium on Pt(111) resulted in the direct interaction between potas-
sium, tin, and platinum as verified by temperature-programmed
desorption ofCO. Tin deposition yielded a maximum in theturnover
rate as a function of Sn coverage for ethylene hydrogenation and
cyclohexane dehydrogenation with maxima at about 0.2 monolay-
ers of tin and a turnover rate 75% higher than on clean Pt(111). This
enhancement was explained by a lower rate of deactivation as tin
was added. In contrast, the addition of potassium to Pt and Pt/Sn
produced only a monotonic decrease in cyclohexane dehydrogena-
tion. In an industrial system, where a higher tin coverage is used,
interaction of tin with potassium may form an effective site blocker
c
which could lower deactivation rates.
ꢀ 1998 Academic Press
Key Words: Pt(111); CO TPD on Pt(111)/Sn/K; hydrogenation
of ethylene on Pt(111)/Sn/K; dehydrogenation of cyclohexane on
Pt(111)/Sn/K.
1. INTRODUCTION
Catalytic dehydrogenation processes are of increasing
importance because of a growing demand for olefins such
as propylene and isobutene. Isobutene is in particularly
high demand as a feedstock for the production of oxy-
genated hydrocarbons required in reformulated gasoline,
such as methyl-tert-butyl ether (MTBE), ethyl-tert-butyl-
ether (ETBE), and tert-butyl alcohol (TBA). Supported
platinum and platinum alloyed with other metals have long
been used as dehydrogenation catalysts in catalytic reform-
ing and in the production of olefins. For the latter appli-
The objective of this work was to use ordered Sn struc-
tures, discovered by Paffet and Windham (6), as model cata-
lysts to explore the effect of tin and potassium on hydro-
genation and dehydrogenation reactions. These structures
are obtained when Sn is deposited on Pt(111) forming a
√
√
p(2 ꢁ 2) structure (2Sn = 0.25) and a ( 3 ꢁ 3)R30ꢂ struc-
ture (2Sn = 0.33). The advantage of this system is that it is
ordered and thus the position of Sn is known. Its disadvan-
tage is that the maximum Sn coverage is much lower than
on systems of industrial importance.
1 Present address: Industrial Catalysis Research Team, Korea Research
Institute of Chemical Technology, Jang-dong 100, Yusung-Gu, Taejon 305-
606, Korea.
The ordered structures of Sn on Pt(111) have been shown
to inhibit the decomposition of ethylene (7), benzene (8),
cyclohexene (9), and 1,3-cyclohexadiene (10) when these
molecules were deposited on the surfaces and heated un-
der ultrahigh vacuum. Catalytic studies on Sn/Pt(111) at
2 Corresponding author. Current address: Worcester Polytechnic Insti-
tute, Department of Chemical Engineering, Worcester, MA 01609-2280.
E-mail: fabio@wpi.edu.
66
0021-9517/98 $25.00
c
Copyright ꢀ 1998 by Academic Press
All rights of reproduction in any form reserved.

HYDROGENATION AND DEHYDROGENATION
67
pressures close to one atmosphere have also been made measurements were made using a chromel–alumel (type K)
(11–13). The effect of potassium on hydrocarbon adsorp- thermocouple spot welded to the rear side of crystal. The
tion and reaction on Pt(111) (14–16) as well as the effect surface was cleaned in a vacuum by repeated cycles of Ar+
of cesium (17, 18) have also been studied. However, no sputtering, followed by oxidation at 800 K and subsequent
study is available for the influence of both tin and potassium annealing at 1073 K until no impurities could be detected
on Pt(111) for hydrogenation and dehydrogenation cata- by AES and a sharp (1 ꢁ 1) Pt(111) LEED pattern was
lytic reactions. For this reason, we studied a series of cata- observed.
lysts for the hydrogenation of ethylene and dehydrogena-
Carbon monoxide TPD was carried in a line-of-sight
tion of cyclohexane: Pt(111), K/Pt(111), Sn/Pt(111), and mode with the mass spectrometer equipped with a cone-
K/Sn/Pt(111). Although the Sn coverage on these model shaped nozzle that moved to 2 mm from the surface of the
systems can reach a maximum Sn coverage of 0.33 mono- crystal. Carbon monoxide gas (Matheson, Research Grade)
layers and some of the reported catalysts in the literature was dosed onto the crystal through a leak valve. During
have higher nominal tin coverage, we expected to verify TPD experiments an Eurotherm controller model 818 with
a trend for the beneficial effect of Sn and K on increas- a TCR programmable DC supply from Electric Measure-
ing rates and decreasing deactivation for dehydrogenation ments, Inc. was used to ramp the temperature at a rate of
reactions on these model catalysts. As a consequence, the 15 K sꢃ1
results here should not be expected to reproduce the per- Bimetallic surfaces were prepared by depositing tin on
.
formance of industrial systems but to show the expected Pt(111) in UHV with a pulsed metal vapor vacuum arc
trend. We found that the effect of Sn seems to be one of plasma gun (MEVVA). The MEVVA source has been de-
decreasing the initial carbon buildup on the catalysts and scribed in detail elsewhere (22, 23). Tin was deposited onto
potassium seems to act only as a site blocker. This role of Pt(111) at room temperature, followed by annealing to
potassium seems to be in conjunction with tin as we found 1000 K. The amount of adsorbed tin was controlled by vary-
– –
that Sn K Pt forms a “surface alloy.” Due to the large size ing the number of pulses from the evaporation source. Tin
of the potassium ion and its anchoring interaction with Sn, coverage was calibrated through AES and CO TPD.
–
–
–
–
Sn K may act in a similar way to Re S on the Pt Re S sys-
Potassium deposition was achieved through a deposition
tem used in reforming as an effective site blocker. This last system composed of a SAES getter source and a beam flag.
role has been proposed to decrease catalyst coking (19, 20) The source was heated with a 5.4 amperes current and a
in reforming reactions. The addition of Sn and K does not different coverage could be achieved by varying the expo-
make the catalyst more active it only makes it less prone to sure time. The sample was kept at room temperature during
deactivation.
deposition, followed by annealing to 600 K.
Hereafter all reported tin and potassium coverage will
be expressed in monolayers (ML), defined as the atomic
ratio of potassium or tin to the initial number of Pt surface
atoms.
2. EXPERIMENTAL METHODS
Experiments were carried out in an ultra-high vacuum
(UHV) chamber equipped with a high pressure reactor. Af-
ter bakeout the chamber routinely attained a base pressure
of 5 ꢁ 10ꢃ10 Torr (15). The analytical tools on the chamber
are Auger electron spectroscopy (AES), low energy elec-
tron diffraction (LEED), mass spectrometry (UTI-100C)
for temperature programmed desorption (TPD) as well as
facilities for gas dosing and metal deposition. The high pres-
sure batch reactor, which can be operated in the pressure
range of 1 to 1000 Torr, is partly inside the UHV chamber.
The reactor is engaged when the sample is enclosed by a
hydraulically operated moving tube that seals the gases by
means of a knife edged flange on a copper gasket (21). The
reactor volume was 650 cm3.
After tin and/or potassium deposition and associated
surface compositional analysis the crystal was introduced
into the high pressure reaction cell for hydrogenation/
dehydrogenation reactions. A typical reaction was carried
out as follows: (1) Ethylene hydrogenation; after introduc-
tion of ethylene (20 Torr), hydrogen (100 Torr), and he-
lium (640 Torr) at room temperature the circulation pump
was started to ensure the mixing of the reactant gases. (2)
Cyclohexane dehydrogenation; after introduction of cyclo-
hexane (15 Torr) and hydrogen (100 Torr) at room tem-
perature the circulation pump was activated for a sufficient
amount of time period to ensure the mixing of the reac-
tant gases, followed by sample heating to 573 K. The reac-
The Pt(111) was about 1 mm thick and about 1 cm2, and tion products were analyzed by direct injection through a
was cut from bulk material (single crystal rods, Goodfellow) gas valve into a gas chromatograph (Hewlett-Packard GC
with 0.5ꢂ precision and polished with standard techniques model 5720A and integrator model 3392) equipped with a
(only one side was polished). The Pt(111) crystal mounting flame ionization detector. After each reaction the reactor
was accomplished via Pt heating wires spot welded to the cell was pumped down with a diffusion pump before the cell
crystal edge, enabling the surface to be resistively heated was opened to the UHV chamber for postreaction surface
to 1200 K at a heating rate of up to 15 K sꢃ1. Temperature analysis.

68
PARK, RIBEIRO, AND SOMORJAI
as verified by LEED. The relationship between tin AES
signal and exposure, expressed in terms of the number
of pulses from the plasma source, was obtained using the
AES uptake curves of both tin and platinum. The deflec-
tion points in Sn(430 eV) and Pt(64 eV) AES signals corre-
sponding to successive monolayers could be seen clearly as
the coverage evolved indicating layer-by-layer growth. Af-
ter three equivalent monolayers, the Sn AES peak intensity
was constant and it was not possible to use AES to monitor
further Sn adsorption. When the crystal was annealed at
1000 K following tin deposition in excess of 0.5 ML (i.e.,
after 5 pulses of Sn) most of the tin dissolved into the bulk
or evaporated leaving a constant tin coverage on the sur-
face equivalent to 0.33 ML of surface atoms (Fig. 2). At this
√
√
coverage a ( 3 ꢁ 3)R30ꢂ LEED structure was formed
corresponding to a Pt2Sn surface alloy (6). Upon annealing
to 1000 K with an initial Sn coverage of less than 0.33 ML,
the LEED pattern gradually changed from a p(1 ꢁ 1) to the
√
√
more dense p(2 ꢁ 2) and finallyto the ( 3 ꢁ 3)R30ꢂ struc-
√
√
ture. Even if excess tin was deposited, the ( 3 ꢁ 3)R30ꢂ
LEED structure was always recovered following anneal-
ing to 1000 K. These results agree with these obtained by
Paffett and Windham (6).
The potassium coverage on Pt(111) was also calibrated
using AES uptake curves analogous to those employed for
the Sn/Pt(111) system. Potassium deposition was controlled
by varying the exposure time from the evaporating source.
The potassium coverage was obtained using the uptake
curve shown in Fig. 3, which clearly shows the deflection
point corresponding to a single monolayer coverage. How-
ever, as the surface coverage increased above a monolayer
successive deflection points could not be distinguished.
FIG. 1. Accumulation plots on Pt(111): (A) hydrogenation of ethy-
lene at 20 Torr of ethylene, 100 Torr of H2, 640 Torr of He, and reaction
temperature of 302 K; (B) dehydrogenation of cyclohexane on Pt at 15 Torr
of cyclohexane, 100 Torr of H2, and reaction temperature of 573 K.
Research purity gases (Matheson) ethylene, isobutane,
and hydrogen were used without further purification. Cy-
clohexane (Fluka, puriss) wasused followingseveralfreeze-
pump-thaw cycles for purification. Turnover rate (TOR) is
the number of molecules produced per surface Pt atoms
per second. The number of Pt atoms are the ones measured
before reaction. The TOR reported in the results section
represent reaction rates obtained after 30 min of reaction
time. Typical accumulation plots for the ethylene and cy-
clohexane reactions are shown in Fig. 1.
3. RESULTS AND DISCUSSION
3.1. Characterization of K/Sn/Pt(111)
3.1.1. Calibration of Surface Composition of Sn and K
on Pt(111) by AES, LEED, and CO TPD
FIG. 2. Tin Auger uptake curves on Pt(111) surface. Sn(430 eV) and
Pt(64 eV) peak intensities versus the number of Sn pulses with annealing
at 1000 K after Sn deposition at 298 K. Ordered structures, as observed by
LEED, are also shown.
The amount of tin deposited on Pt(111) was calibrated by
AES measurements and by the ordered structures formed,

HYDROGENATION AND DEHYDROGENATION
69
FIG. 4. Saturation CO TPD spectra from Sn/Pt(111) surface follow-
ing annealing at 1000 K after Sn deposition at 298 K. The coverage was
calibrated from Auger signal intensities. CO exposure at 1 ꢁ 10ꢃ7 Torr,
250 K for 5 min. The heating rate was 15 K sꢃ1
.
FIG. 3. Potassium Auger uptake curves on Pt(111) surface. K(252 eV)
and Pt(64 eV) peak intensities versus exposing time with the crystal held
at 298 K.
assumed that the stoichiometry of CO to Pt is 2 to 1 on
Pt(111) and that on defects (steps) the stoichiometry of CO
to Pt is 1 to 1 as discussed before (26), then curve “c” in
Fig. 4 predicts that at a coverage of tin of about 0.13 mono-
layers (adequate to cover all the defect sites) the ratio of
CO areas for curves “a” and “c” should be as indeed ob-
served in Fig. 5. The problem with this interpretation is that
a much higher desorption temperature peak is expected for
the step defects (about 550 K) and that about 15% of the
Pt(111) surface atoms should be at step sites. Paffett et al.
(27) studied the adsorption of CO on the Sn/Pt(111) struc-
tures and found the same broad peak for CO TPD and the
same peak narrowing when Sn is added. They could rule
out, however, by HREELS the possibility that the peak
The CO TPD results for Sn/Pt(111), K/Pt(111), and K/
Sn/Pt(111) are discussed in turn. In general, the bonding of
CO to metal atoms involves a synergistic electron transfer
from the highest occupied molecular orbitals of CO (5ꢀ)
to the metal and, in turn, metal electrons are backdonated
into the lowest unoccupied molecular orbital (2ꢁꢄ) of CO
(24, 25). Therefore, the presence of other metals on the Pt
surface may change the bonding characteristic of CO. Any
ad-atom which withdraws electrons from Pt will cause a
decrease in the backdonation of metal electrons into the
2ꢁꢄ CO and consequently weaken the metal–carbon bond
while electron donation into Pt will produce the opposite
effect. This effect will manifest itself as a change in the CO
binding energy reflected in its desorption profile. It should
be noted that other molecules (like hydrocarbons) may not
interact with the surface in the same manner as CO and
thus the TPD results obtained for CO cannot be extrap-
olated readily for other systems. There may also be other
factors involved in displacing the CO desorption peak like
the blocking of the preferred CO adsorption sites by an-
other adsorbant (e.g., Sn and K) or the titration of defects
on the crystal surface by Sn.
Figure 4 shows a series of saturation CO TPD spectra on
Sn/Pt(111). The desorption peak is quite broad for a first-
order desorption on Pt(111) but its shape changes as tin is
added. For ꢂSn < 0.33 the shift of the CO desorption peak
to lower temperatures (425 to 365 K) with increasing tin
coverage can indicate that on Pt(111) adsorbed Sn reduces
–
–
the Pt CO bond strength because of a Pt Sn interaction, or
that CO cannot bind on the same way because of geometric
constraints (its preferred adsorption sites are now occupied
by tin), or that there were defects on the crystal that were
FIG. 5. Peak area of CO TPD versus Sn coverage. The area was ob-
titrated by tin. This last possibility is discussed next. If it is tained by integrating the CO TPD spectra of Fig. 4.

70
PARK, RIBEIRO, AND SOMORJAI
narrowing is caused by different CO chemisorption modes
or by significantly different ratios of bridge versus atop ad-
sorption. The peak narrowing may be a combination of dif-
ferent sites (less bridging sites available) and the titration
of surface defects.
In the case of adsorbed Sn, the CO TPD exhibited a linear
decrease in the peak area up to ꢂSn = 0.33, indicating that
CO adsorption does not occur on surface tin atoms (Fig. 5).
For all Sn exposures corresponding to greater than five Sn
pulses from the plasma source, Fig. 5 shows that the CO
desorption peak area remains unchanged, indicating that
a constant amount of tin always remains on the Pt(111)
surface after annealing to 1000 K. The CO TPD results
indicate a Sn coverage of 0.5 ML while the AES/LEED
results suggest 0.33 ML coverage; tin seems to be blocking
more than one CO site per Sn atom on the surface. Note
that Paffet et al. (27) found that the number of Pt atoms
counted by CO TPD blocking was higher than predicted on
a Pt(111)/Sn surface while Haner et al. (28) found that CO
could count perfectly the number of Pt atoms on the surface
FIG. 7. Saturation CO TPD spectra from Sn/K/Pt(111). CO exposure
at 1 ꢁ 10ꢃ7 Torr, 250 K for 5 min. The heating rate was 15 K sꢃ1
.
by the potassium. As the potassium coverage increased, the
desorption peak area at 640 K increased and that at 425 K
decreased gradually. However, above one monolayer cov-
erage (ꢂK > 1.0), almost all of the surface is potassium cov-
ered, as evidenced by the AES results, and consequently, no
CO adsorption sites remain. This result can be interpreted
on the basis that only exposed surface Pt atoms adsorb CO
molecules while co-adsorbed potassium acts as an effec-
tive site blocker for CO adsorption. This is reflected in the
monotonic decrease in CO TPD peak area with increasing
K coverage (Fig. 6), indicating that, similar to the effect of
Sn, CO does not adsorb on K. The apparent increase in CO
binding energy induced by K co-adsorption is also consis-
tent with results obtained by Crowell et al. (14) who found
that the adsorption energy of CO is increased from 25 to
36 kcal molꢃ1 on potassium adsorbed Pt(111).
–
for single crystals made of Pt Sn alloys. This discrepancy
can be resolved if it is assumed that there are defects on the
surface which adsorb with different stoichiometry. In view
of the variation of counting Pt sites by CO TPD, we will use
the AES results helped by LEED for the calculation of the
amount of Pt on the surface. These values calculated from
AES data are adequate for the purpose of this work.
Compared to tin, potassium on Pt(111) produced a bi-
modal CO desorption profile with a peak centered at ap-
proximately 425 and 640 K (Fig. 6), in contrast to a single
peak at 425 K on the clean surface. This 215 K increase in
desorption temperature, evident in the higher desorption
peak, is a result of the electron donating ability of potas-
–
sium, strengthening the Pt CO bond. These two peaks re-
veal two environments for the CO: pure Pt and Pt modified
Having considered the independent interaction of Sn and
K with Pt(111), the interaction ofco-adsorbed tin and potas-
sium with Pt(111) was studied next. The above CO TPD re-
sults over Sn/Pt(111) and K/Pt(111) systems reveal that the
modification induced by tin and potassium have opposite
trends on the Pt(111) surface. As shown in Fig. 7, when they
are co-adsorbed on Pt(111), the TPD profile resembles that
of CO adsorbed on clean Pt(111). This result implies that
the effect of tin is compensated for by the effect of potas-
sium. The other possibility is that the interaction of tin and
potassium with platinum are independent of one another.
However, this scenario would have yielded three different
environments such as Pt, K/Pt, and Sn/Pt with three distinct
CO desorption peaks at 425, 365, and 640 K, respectively.
This is clearly in contrast to what is observed experimen-
tally in Fig. 7, where curve “d” is not composed of curve
“b” plus curve “c.” Thus, tin and potassium on Pt(111) form
a three-component surface alloy. A similar possibility was
also raised by Spiewak et al. (29) to explain their results
on Pt/Sn/K/SiO2 catalysts but note that on their case the
FIG. 6. Saturation CO TPD spectra from K/Pt(111) as a function of K
coverage. The coverage was calibrated from Auger signal intensities. CO
exposure at 1 ꢁ 10ꢃ7 Torr, 250 K for 5 min. The heating rate was 15 K sꢃ1
.

HYDROGENATION AND DEHYDROGENATION
71
potassium was present in the form of an oxide. These au- activity enhancement at ꢂSn = 0.07). For these experiments,
thors believe that the oxidized alkali interacts with tin form- the annealing to 1000 K leads to a more homogeneous dis-
ing a strong bond that anchors the alkali and provides fur- persion of tin on Pt(111) surface which seems to be neces-
ther blocking.
sary to attain the maximum increase in the rate for ethylene
In conclusion, the important facts from the CO TPD re- hydrogenation.
sults are that there seems to be a compound formation on
Windham et al. (16), Salmeron and Somorjai (33), and
the surface by the interaction of tin, potassium, and plat- Berlowitz et al. (34) have shown for the saturation cover-
inum and that the effect of Sn on Pt seems to be one of age of ethylene adsorbed on Pt(111) at 120 K that 60%
titrating high energy sites. The other effects seems to be of it desorbs molecularly upon heating, a small fraction
specific to CO because of the characteristic electronic struc- (<2% ) of ethane is produced, while the remainder of the
ture of this molecule and they may be different for other adsorbed molecules decompose to produce a graphitic car-
probe molecules like hydrocarbons. The relationship be- bon residue and H2 (g) during heating to 800 K. At a low
tween these surface modifications and hydrogenation and ethylene coverage, all of the ethylene molecules adsorbed
dehydrogenation are discussed in the next section.
on clean Pt(111) decompose and desorption is completely
irreversible. However, in the presence of Sn the ethylene
adsorption strength is reduced, and no dissociation of ethy-
lene occurs upon heating on the two Sn/Pt surface alloys
3.2. Hydrogenation and Dehydrogenation Reactions
√
√
3.2.1. Ethylene Hydrogenation
having the p(2 ꢁ 2) and ( 3 ꢁ 3)R30ꢂ structures (7). The
same is true for other molecules like benzene (8), cyclo-
hexene (9), and 1,3-cyclohexadiene (10). Also, the activa-
tion energy decreases with the Sn surface concentration.
Specifically, the activation energy of ethylene desorption
was found to be 17 kcal mol
Ethylene hydrogenation was carried out on tin covered
Pt(111) after two different sample pretreatments and the
results are shown in Fig. 8. The result for the Pt(111) with
no Sn is a factor of about 4 higher than observed previously
by Zaera and Somorjai (30) or reported in the literature
ꢃ1 on Pt(111), falling to 15.1 and
√
√
(31). The reason for the higher rates is not known, although 11.8 kcal molꢃ1 on the p(2 ꢁ 2) alloy and ( 3 ꢁ 3) alloys,
the most recent rates on Pt(111) and Pt foil reported on our respectively (7).
group (32) were identical to the ones reported here. When
The loweringofthe activation energyon the Sn/Pt surface
the sample was annealed at 1000 K after tin deposition, the could be due to a lower binding energy on Pt. Alternatively,
ethylene TOR showed a maximum at ꢂSn = 0.1, correspond- different adsorption sites may exist for ethylene adsorbed
√
√
ing to a 75% activity enhancement over clean Pt(111). In
contrast, when the sample was not annealed following tin
deposition, the activity enhancement was significantly less
pronounced than that of the annealed sample (only 35%
on Pt in the p(2 ꢁ 2) and ( 3 ꢁ 3) surfaces which have
different heats of adsorption, although the properties of
platinum itself has not been substantially modified by the
addition of Sn. The existence of a different binding energy
is supported by a previous study of C2H4/Bi/Pt(111), where
the heat of ethylene adsorption on Bi/Pt(111) was found
to be the same as ethylene on Pt(111) (35). Since Bi has
a relatively small effect on the electronic structure of Pt
and functions as a pure site blocking adatom on the Pt(111)
surface (35), any change in the heat of ethylene adsorp-
tion on the Sn/Pt alloys is most likely due to a modification
of surface Pt atoms, i.e. due to a ligand effect. This idea is
also supported in the present study by the observation that
a small amount of tin (ꢂSn = 0.1), when alloyed with plat-
inum, shifted CO desorption peak to lower temperature
(Fig. 4). Note, however, that at the maximum ethylene TOR
(ꢂSn = 0.1), there is no observed structural changes in the
LEED patterns from the clean p(1 ꢁ 1) Pt(111). The other
interpretation is that the effect of tin could be to titrate the
small amount of defects on the Pt surface. These defects
are the source of carbonaceous deposits and, if not titrated
before reaction, can decrease the overall rate by carbon
deposition. Because we cannot account for a rate decrease
due to carbon deposits (it is not measured; the amount of
Pt exposed before reaction is used as a base to calculate
turnover rates), a catalyst quickly poisoned by a higher
FIG. 8. Turnover rate for ethane production as a function of Sn cov-
erage on Pt(111). Reaction condition; ethylene = 20 Torr, H2 = 100 Torr,
He = 640 Torr, and reaction temperature = 302 K. The turnover rate is
based on the number of Pt atoms on the surface before reaction.

72
PARK, RIBEIRO, AND SOMORJAI
deposition rate of carbon would appear to have a lower
rate and this is the reason why the rates appear to increase
when tin is added. If tin were changing the binding energy
and thus accelerating the rates, a monotonic increase of rate
should be expected. Instead, the turnover rates actually de-
crease as more tin is added. Note that Paffett et al. (7) have
shown that the adsorption sites suitable for reversible ethy-
lene adsorption are more favorable on Sn/Pt(111) than on
Pt(111). Again, the milder Sn/Pt surface does not form the
carbonaceous precursors and appears as a catalyst with a
higher rate, because under reaction conditions it has more
Pt atoms exposed.
The effect of potassium on ethylene hydrogenation was
not tested.
3.2.2. Cyclohexane Dehydrogenation
In dehydrogenation and hydrogenolysis reactions of
cyclohexane over Pt crystals (36) the relative rate of de-
hydrogenation is several orders of magnitude higher than
hydrogenolysis and isomerization, which results in only one
dehydrogenation product (that is, benzene). Because of this
specificity the reaction represents an ideal system to in-
vestigate dehydrogenation effects of tin and potassium on
Pt(111). Cyclohexane dehydrogenation was carried out at
15 Torr cyclohexane, 100 Torr H2 at 573 K with varied cov-
erage of tin and potassium.
FIG. 10. Amount of carbon on the surface as a function of Sn coverage
on Pt(111) after reaction with cyclohexane for about 1 h. Reaction con-
dition; cyclohexane = 15 Torr, H2 = 100 Torr, and reaction temperature =
573 K.
low surface concentrations of tin (ꢂSn = 0.2). At this point
a 70% enhancement in activity was measured, compared
to the clean Pt(111) (Fig. 9). The literature value for the
turnover rate on Pt(111) extrapolated to our conditions is
14 sꢃ1 (36). However, this value was measured at a lower
temperature (533 K) and, due to severe deactivation at the
higher temperature used in this work, a lower value for the
turnover rate is expected.
3.2.2.1. Sn/Pt(111). Figure 9 shows the effect of co-
adsorbed Sn on cyclohexane dehydrogenation. This figure
shows a trend similar to that observed in ethylene hydro-
genation (maximum activityat ꢂSn = 0.1), where the catalyst
exhibited maximum dehydrogenation activity at relatively
The Sn/Pt(111) system clearly showed a synergistic ef-
fect in both hydrogenation and dehydrogenation reactions.
It is clear that on Pt(111) a higher final rate for hydrogena-
tion and dehydrogenation activity is obtained when tin is
added at about 0.2 monolayers. To check the dependence of
catalyst structure for the cyclohexane dehydrogenation, re-
√
√
actions were carried out over p(2 ꢁ 2) and ( 3 ꢁ 3)R30ꢂ
Sn/Pt(111) alloys. The Sn/Pt(111) having these structures
did not show any special enhancement in activity, compared
to the primitive (1 ꢁ 1); in fact the turnover rate was lower
on these structures. The higher turnover rate enhancement
appeared at ꢂSn = 0.2, where there are no extra structures
on the surface except the p(1 ꢁ 1) structure. As already ad-
vanced in the earlier section, this volcano curve could be
due to a lower rate of deactivation on the catalysts that oc-
curs by the titration of platinum sites by tin or by a less
reactive platinum surface obtained by the addition of tin.
If this argument is valid, then the amount of carbon de-
posited should decrease as tin is added to the surface as
indeed shown in Fig. 10.
The fact that the rates decreased after the maximum and
did not stay constant as expected for a structure-insensitive
reaction cannot be explained easily. Cortright and Dumesic
(5) also found that the turnover rates for the dehydrogena-
tion of isobutane decreased by the addition of Sn. This
FIG. 9. Turnover rate for benzene production as a function of Sn cov-
erage on Pt(111). The turnover rate is based on the number of Pt atoms on
the surface before reaction. Reaction condition; cyclohexane = 15 Torr,
H2 = 100 Torr, and reaction temperature = 573 K.

HYDROGENATION AND DEHYDROGENATION
73
turnover rate decreases with the amount of potassium on
the surface (note that the turnover rate is already corrected
–
for blocking effects), the presence of potassium makes C H
bond-breakingmore difficult and reducesthe dehydrogena-
tion activity. Again, the addition of potassium seems to
make it difficult to form highly dehydrogenated species that
may deactivate the catalyst.
3.2.2.3. K/Sn/Pt(111). Additional dehydrogenation ex-
periments were carried out on the four catalyst series,
Pt(111), Sn/Pt(111), K/Pt(111), and K/Sn/Pt(111). The
turnover rate for these four catalysts are shown in
Fig. 12. Among these four catalysts, Sn/Pt(111) showed
the highest turnover rate. When potassium was added to
Sn(0.25)/Pt(111), the activity gradually decreased in an
analogous trend to the results observed for cyclohexane
dehydrogenation over K/Pt(111) (Fig. 11). Thus, regardless
of the presence or absence of tin, the addition of potassium
causes the turnover rates to decrease, with no apparent ad-
vantage for the use of potassium on Sn/Pt catalysts.
However, in contrast to our observations, there are re-
sults reported in the literature to support the idea that
the addition of potassium on Sn/Pt/supported catalysts in-
creases both dehydrogenation turnover rate and catalytic
stability (4, 5, 37–41). These results are actually in line with
our findings as will be explained in turn. On supported cata-
lysts the high surface area support may play a role in the
kinetics and in our studies there is no support present. It is
FIG. 11. Turnover rate for benzene production as a function of K
coverage on Pt(111). Reaction condition; cyclohexane = 15 Torr, H2 =
100 Torr, and reaction temperature = 573 K. The turnover rate is based on
the number of Pt atoms on the surface before reaction.
volcano-type plot was also found by Szanyi et al. (11) for the
reaction of n-butane hydrogenolysis. These authors found
that the TOR for the p(2 ꢁ 2) structure was higher than for
Pt(111) but then decreased again as more Sn was added to
√
√
form the ( 3ꢁ 3)R30ꢂ structure. A change in the binding
energy of reaction intermediates that will in turn affect the
turnover rate would be a possible explanation.
3.2.2.2. K/Pt(111). Figure 11 shows that the TOR de-
creased linearly with potassium coverage. This linear de-
crease in activity over the whole coverage range implies
that the electron donation associated with potassium ad-
sorption on metals does not affect the dehydrogenation
of cyclohexane. Potassium itself is inactive in cyclohexane
dehydrogenation and, as a result, at monolayer coverage
K/Pt(111) showed almost no activity.
Based on the results of Zaera and Somorjai (15), the
interaction of n-hexane on K/Pt(111) is characterized by
a decrease in the sticking coefficient of n-hexane. In this
–
process, at least one C H bond-breaking event takes place,
followed by the formation of a carbon–metal bond between
the metal and alkyl fragment. This adsorption is accompa-
nied by charge transfer from the hydrocarbon to the plat-
inum, leaving the former with a carbonium ion character.
Potassium on the surface, acting as an electron donor, in-
–
hibits this charge transfer, making the C H bond-breaking
process less favorable. Accordingly, this effect should be
more accentuated as the alkyl fragment continues to de-
hydrogenate, Zaera and Somorjai have showed that the
presence of potassium on Pt(111) produced a steep rise in
the activation energy for hydrogen ꢃ elimination of the car-
FIG. 12. Turnover rate for benzene production as a function of K
and Sn coverage on Pt(111). Reaction condition; cyclohexane = 15 Torr,
H2 = 100 Torr, and reaction temperature = 573 K. The turnover rate is
bonaceous deposits. Based on these results, the fact that the based on the number of Pt atoms on the surface before reaction.

74
PARK, RIBEIRO, AND SOMORJAI
well known that the presence of acidic supports can result in same role as before and the initial increase in activity is due
the catalysis of undesired parallel reactions (cracking and to the titration of the highly active and coke-forming sites
polymerization) and for this reason, an increasing number which poison the catalyst before kinetic measurements can
of papers have been published recently on the use of acidic be made. The addition of Sn will then imply a lower amount
supports suitably modified by alkali metals (37, 38) since of carbon deposits as indeed observed (Fig. 10).
the alkali will neutralize the acidity. However, it seems that
at least the effect of potassium reported by Cortright and
Dumesic (4, 5, 42) is not related to that cause. These authors
found that in their supported catalyst system for isobutane
dehydrogenation, potassium is important for the selectivity
and stability of the catalyst. Our results suggest that potas-
sium does not modify drastically the catalytic properties of
Pt/Sn/K catalysts. It certainly does not increase the turnover
rate but it may provide for additional or a better site blocker
as suggested by Cortright and Dumesic (5). This is a plausi-
ble explanation for the effect of K since our CO TPD results
SUMMARY
The adsorption of tin on Pt(111) decreased the tendency
of the surface to deactivate by carbon deposition (coking).
This was verified directly by the decrease of carbon buildup
after reaction with cyclohexane. These results point to Sn
titrating the high activity, coke-forming defect sites present
on Pt(111) or impeding the formation of highly unsaturated
molecules by making the presence of large ensembles of Pt
unavailable. This decrease in carbon buildup explains the
increase in rates for hydrogenation of ethylene and dehy-
drogenation of cyclohexane when less than 0.2 monolayers
of Sn is added.
In the presence of both tin and potassium the CO TPD
suggests an interaction between K and Sn with the forma-
tion of a “surface alloy.” Adsorbed potassium decreased
the turnover rate of cyclohexane dehydrogenation on Pt
and Pt/Sn samples with a monotonic decrease in activity
with increasing potassium coverage. Potassium could pos-
sibly decrease the rate of deactivation by site blocking, es-
pecially because K has a large ionic radius, it interacts with
Sn, and there is an example in the literature of a similar ef-
–
suggest that the Sn and K are interacting and thus the Sn K
–
system would be similar to the Re S moiety formed on the
–
–
Pt Re S system, with K interacting with Sn and acting as
–
an effective site blocker (like S in the Re S system), mak-
ing it difficult to form carbonaceous deposits. The reason
we could not find the same enhancement for K and Sn as
reported by Cortright and Dumesic may be related to the
much higher Pt to Sn ratio in our studies. As explained be-
fore, we could not decrease the ratio below Pt/Sn = 3. The
√
√
number of contiguous Pt sites for the ( 3ꢁ 3)R30ꢂ struc-
ture is still very large for this catalyst to be significantly
better than Pt on its deactivation pattern. In the same way,
–
–
–
–
the blocking effect of K (by the K Sn interaction) cannot
fect of bulky surface compounds like Re S on the Pt Re S
system increasing the catalyst stability.
The Pt/Sn/K catalyst is a better catalyst for dehydrogena-
tion because it deactivates less, not because it has a higher
turnover rate.
be seen at this low dilution of Sn on Pt. The most resistant
catalysts to deactivation reported in the literature are the
ones with a low Pt/Sn ratio (Pt/Sn = 0.25). Thus, as in the
–
–
case of Pt Re S, for the effect of K to be seen in creating a
more stable catalyst the amount of Sn on the surface should
be high (not attainable in our case) and we cannot make a
direct comparison of our samples with the supported cata-
lysts. For this reason, it is possible to observe an increase in
turnover rate (actually a lower deactivation rate) when tin
and potassium are added simultaneously at a high coverage
as Cortright and Dumesic (5) have found.
ACKNOWLEDGMENTS
This work was supported by the Director, Office of Energy Research,
Office of Basic Energy Sciences, Materials Sciences Division, of the U.S.
Department of Energy under Contract DE-AC03-76SF00098. Y.-K. Park
thanks Dr. D. H. Fairbrother for his help.
The other difference in our studies was that the addi-
tion of Sn caused a volcano type curve, not reported for
supported catalysts. The reason for that is due to the dif-
ferent defect densities and deactivation mechanism on the
two systems. We can prepare pristine samples and preserve
them before reaction. These samples have very high ac-
tivity sites that will not be present on supported catalysts
because these sites are highly reactive and during the nor-
mal supported catalyst preparation they will be poisoned
before the reaction can start. These sites can, however, be
poisoned by the addition of Sn on the model catalysts. We
have seen before that the effect of Sn on a polycrystalline
Pt foil was to titrate the highly active and coke forming
REFERENCES
1. Miller, S. J., U.S. Patent 4,727,216 (1986).
2. Brinkmeyer, F. M., and Rohr, D. F., U.S. Patent 4,866,211 (1987).
3. Imai, T., and Hung, C. W., U.S. Patent 4,430,517 (1983).
4. Cortright, R. D., and Dumesic, J. A., J. Catal. 148, 771 (1994).
5. Cortright, R. D., and Dumesic, J. A., J. Catal. 157, 576 (1995).
6. Paffett, M. T., and Windham, R. G., Surf. Sci. 208, 34 (1989).
7. Paffett, M. T., Gebhard, S. C., Windham, R. G., and Koel, B. E., Surf.
Sci. 223, 449 (1989).
8. Xu, C., Tsai, Y. L., and Koel, B. E., J. Phys. Chem. 98, 585 (1994).
9. Xu, C., and Koel, B. E., Surf. Sci. 304, 249 (1994).
10. Peck, J. W., and Koel, B. E., J. Am. Chem. Soc. 118, 2708 (1996).
11. Szanyi, J., Anderson, S., and Paffett, M. T., J. Catal. 149, 438 (1994).
12. Szanyi, J., and Paffett, M. T., Catal. Lett. 29, 133 (1994).
centers before reaction (26). In the present case, Sn has the 13. Szanyi, J., and Paffett, M. T., J. Am. Chem. Soc. 117, 1034 (1995).

HYDROGENATION AND DEHYDROGENATION
75
14. Crowell, J. E., Garfunkel, E. L., and Somorjai, G. A., Surf. Sci. 121,
303 (1982).
29. Spiewak, B. E., Levin, P., Cortright, R. D., and Dumesic, J. A., J. Phys.
Chem. 100, 17260 (1996).
15. Zaera, F., and Somorjai, G. A., J. Catal. 84, 375 (1983).
16. Windham, R. G., Bartram, M. E., and Koel, B. E., J. Phys. Chem. 92,
2862 (1988).
17. Davidsen, J. M., Henn, F. C., Rowe, G. K., and Campbell, C. T., J. Phys.
Chem. 95, 6632 (1991).
18. Bussell, M. E., Henn, F. C., and Campbell, C. T., J. Phys. Chem. 96,
5978 (1992).
19. Sachtler, W. M. H., J. Mol. Catal. 25, 1 (1984).
20. Ribeiro, F. H., Bonivardi, A. L., Kim, C., and Somorjai, G. A., J. Catal.
150, 186 (1994).
30. Zaera, F., and Somorjai, G. A., J. Am. Chem. Soc. 106, 2288
(1984).
31. Ribeiro, F. H., Schach von Wittenau, A. E., Bartholomew, C. H., and
Somorjai, G. A., Catal. Rev. Sci. Eng. 39, 49 (1997).
32. Quinlan, M. A., Ph.D. thesis, University of California at Berkeley,
1996.
33. Salmeron, M., and Somorjai, G. A., J. Phys. Chem. 86, 341 (1982).
34. Berlowitz, P., Megris, C., Butt, J. B., and Kung, H. H., Langmuir 1,
206 (1985).
35. Windham, R. G., Koel, B. E., and Paffett, M. T., Langmuir 4, 1113
(1988).
36. Herz, R. K., Gillespie, W. D., Petersen, E. E., and Somorjai, G. A.,
J. Catal. 67, 371 (1981).
37. Gravelle-Rumeau-Maillot, M., Pitchon, V., Martin, G. A., and
Praliaud, H., Appl. Catal. A 98, 45 (1993).
38. de Miguel, S., Castro, A., Scelza, O., Fierro, J. L. G., and Soria, J.,
Catal. Lett. 36, 201 (1996).
39. Park, Y., and Price, G. L., Ind. Eng. Chem. Res. 31, 469 (1992).
40. Davis, R. J., and Derouane, E. G., J. Catal. 132, 269 (1991).
21. Kahn, O. R., Petersen, E. E., and Somorjai, G. A., J. Catal. 34, 294
(1974).
22. Kim, C., Ogletree, D. F., Salmeron, M. B., Godechot, Y., Somorjai,
G. A., and Brown, I. G., Appl. Surf. Sci. 59, 261 (1992).
23. Kim, C., and Somorjai, G. A., J. Catal. 134, 179 (1992).
24. Doyen, G., and Ertl, G., Surf. Sci. 43, 197 (1974).
25. Nieuwenhuys, B. E., Surf. Sci. 105, 505 (1981).
26. Fujikawa, T., Ribeiro, F. H., and Somorjai, G. A., J. Catal. 178, 58
(1998).
27. Paffett, M. T., Gebhard, S. C., Windham, R. G., and Koel, B. E., J. 41. Clarke, J. K. A., Bradley, M. I., Carvie, L. A. J., Craven, A. J., and
Phys. Chem. 94, 6831 (1990). Baird, T., J. Catal. 143, 122 (1993).
28. Haner, A. H., Ross, P. N., Bardi, U., and Atrei, A., J. Vac. Sci. Technol. 42. Cortright, R. D., and Dumesic, J. A., Appl. Catal. A 129, 101
A 10, 2718 (1992). (1995).