Correlation of H2 heat of adsorption and ethylene hydrogenation activity for supported Re@Pd overlayer catalysts

Article abstract of DOI:10.1016/j.jcat.2009.02.022

Alumina-supported bimetallic overlayer Pd on Re (Re@Pd) catalysts were synthesized using the directed deposition technique. Hydrogen chemisorption, TEM, EDS, XRD, and ethylene hydrogenation studies were used to characterize the catalysts and provide indic

Full text of DOI:10.1016/j.jcat.2009.02.022

Journal of Catalysis 263 (2009) 306–314
Contents lists available at ScienceDirect
Journal of Catalysis
www.elsevier.com/locate/jcat
Correlation of H2 heat of adsorption and ethylene hydrogenation activity for
supported Re@Pd overlayer catalysts
∗
Michael P. Latusek, Brett P. Spigarelli, Rebecca M. Heimerl, Joseph H. Holles
Department of Chemical Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, MI 49931, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Alumina-supported bimetallic overlayer Pd on Re (Re@Pd) catalysts were synthesized using the directed
deposition technique. Hydrogen chemisorption, TEM, EDS, XRD, and ethylene hydrogenation studies were
used to characterize the catalysts and provide indication of electronic modification of the Pd surface
layer due to the overlayer particle structure. First principles computation and single crystal studies of
Pd overlayers on Re in the literature have shown that electronic modification of the Pd overlayer is
observed and leads to decreased binding strength for chemisorbed species such as H2, C2H4, and CO.
Measured hydrogen chemisorption isotherms indicated that Pd was deposited on the Re and not as pure
isolated Pd particles. H2 heats of adsorption, as determined by chemisorption, indicated that the Re@Pd
overlayer catalysts were lower than either pure Pd or Re. The Re@Pd catalysts were slightly less active for
ethylene hydrogenation than pure Pd but displayed similar apparent activation energies and H2 and C2H4
reaction orders. A linear correlation between turnover frequency and maximum heat of H2 adsorption
was observed for the Pd and Re@Pd catalysts. This suggests an electronic modification of the Re@Pd
catalyst surface compared to Pd as predicted in the literature by first principles computational studies.
© 2009 Elsevier Inc. All rights reserved.
Received 12 December 2008
Revised 30 January 2009
Accepted 19 February 2009
Available online 13 March 2009
Keywords:
Ethylene hydrogenation
Hydrogen
Chemisorption
Overlayer catalysts
Pd
Re
Kinetics
1. Introduction
the Pt monolayer on W (PtML/W) system compared to either Pt or
W single crystal surfaces [15]. Pallassana and Neurock have studied
the PdML/Re system using first principles computation for hydro-
gen and ethylene adsorption [16–18]. They found that the heats of
hydrogen and ethylene adsorption are significantly lower on the
monolayer system compared to pure Pd or Re. They also exam-
ined this system for ethylene hydrogenation and found that the
activation barrier for ethylene hydrogenation changes very little
for PdML/Re (78 kJ/mol) compared to pure Pd (82 kJ/mol). How-
ever, the activation barrier for the reverse reaction (dehydrogena-
tion) increases significantly from 82 kJ/mol on Pd to 126 kJ/mol
on PdML/Re [17]. These works suggest that the PdML/Re system
may exhibit unique catalytic activity if a suitable technique for its
synthesis as a supported bimetallic overlayer catalyst can be devel-
oped.
There are several synthesis methods mentioned in the litera-
ture which have been used to selectively deposit one metal onto
a different base metal nanoparticle to obtain a core@shell struc-
ture. Some of these methods, such as polyol reduction, result in
reduced, unsupported particles which then require impregnation
onto a support in a later step [19]. Other methods are based on
electroless deposition, which can lead to the undesirable forma-
tion of three-dimensional islands before complete monolayer de-
position [20–23]. Thus there is still a need to develop methods
to synthesize bimetallic overlayer catalysts directly on the support
Bimetallic catalysts have long been an important area of cataly-
sis research. Certain combinations of metals are known to improve
activity, selectivity, or catalyst lifetime, such as Pt–Re for gasoline
production through reforming [1]. Several mechanisms to explain
the catalytic properties observed using bimetallic catalysts have
been proposed, including structure effects [2], bifunctional routes
[3], and electronic modification of surface properties [4].
Bimetallic overlayers, which consist of a monatomic metal layer
deposited on a different bulk metal, have been frequently studied
using single crystal and first principles computation techniques in
an attempt to better understand the surface properties of bimetal-
lic catalysts [5–12]. These overlayer systems have typically been
found to possess very different surface properties compared to
their component metals. The altered properties found in many
overlayer systems have been explained by an electronic interaction
that leads to a shift in the d-band center of an overlayer metal
when it is deposited on a suitable base metal [13,14].
Overlayer systems have also been shown to exhibit distinctive
reactivity properties compared to their pure components as well.
Cyclohexene decomposition activity has been found to be lower in
Corresponding author. Fax: +1 906 487 3213.
E-mail address: jhholles@mtu.edu (J.H. Holles).
*
0021-9517/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2009.02.022

M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
307
that can exhibit properties consistent with first principles predic-
tions.
In this work, high surface area supported Re@Pd (core@shell)
Table 1
Catalyst metal loadings from elemental analysis.
Catalyst
Re (wt%)
Pd (wt% )
overlayer catalysts were synthesized using the directed deposi-
tion technique. This technique is able to synthesize supported
bimetallic overlayer catalysts by directing the deposition of over-
layer metal onto pre-existing base metal particles using a surface
reaction. The intent of this work is to examine the particle struc-
ture of supported Re@Pd catalysts, determine the effect of that
structure using a probe reaction, relate the structure effects to
synthesis conditions, and finally, compare these results to compu-
tationally predicted values. Based on the existing first principles
computational research and the well-known use of ethylene hy-
drogenation as a probe reaction, it was decided to examine these
new overlayer catalysts using ethylene hydrogenation [24].
Re
1.26
1.26
1.26
1.26
–
–
Re@Pd HT
Re@Pd TD
Re@Pd NI
Pd
0.083
0.073
0.249
0.086
toluene. Finally, the catalysts were dried in an oven, calcined in air
◦
◦
at 400 C for 4 h, and reduced in 60 sccm H2 at 400 C for 4 h.
Three different Re@Pd overlayer catalysts were synthesized us-
ing the above technique modified in different ways. One catalyst
was synthesized exactly as described above (Re@Pd HT for high
temperature deposition of Pd). A second was performed also as de-
scribed above, but the complete synthesis procedure was repeated
on the same catalyst a total of three times (Re@Pd TD for triple
deposition). The last catalyst was performed without the water-
saturated helium and acetylacetone inhibition procedures (Re@Pd
NI for no inhibitor). This catalyst was expected to have Pd depo-
sition as an overlayer on Re particles, as well as on the alumina
support forming monometallic particles. Because all of the Re@Pd
catalysts were synthesized with the same Re base catalyst, di-
rect comparisons of particle size/structure between all the catalysts
are possible. The maximum expected palladium content for each
Re@Pd catalyst is 0.27 wt%, based on the amount added during
palladium deposition.
2. Experimental
2.1. Catalyst synthesis
The monometallic catalysts studied in this paper were syn-
thesized using standard techniques and were supported with
100 m2/g γ -alumina (Alfa Aesar, 99.97%). A 0.1 wt% Pd/Al2O3 con-
trol catalyst was prepared from a slurry of alumina and palladium
(II) acetylacetonate (Alfa Aesar) in toluene. A large batch of 2 wt%
Re/Al2O3 catalyst was used as a control and as a base for bimetallic
catalyst synthesis. This catalyst was prepared by incipient wetness
impregnation of alumina with a solution of ammonium perrhen-
ate (Alfa Aesar, 99.999%). Catalysts were dried in a oven for several
2.2. Sample characterization
◦
hours after synthesis, then calcined in air at 400 C for 4 h using
◦
a 3 C/min ramp rate. Finally, the catalysts were reduced in a 60
sccm H2 flow for 2–4 h using a 3 C/min ramp rate. The Pd/Al2O3
catalyst was reduced at 400 C, while literature reports suggested
rhenium required a temperature of 550 C to be thoroughly re-
Elemental analysis by Galbraith Laboratories, Inc. was used to
determine metal loadings for all catalysts synthesized in this study.
These loadings are reported in Table 1.
Chemisorption experiments were performed using a Micro-
meritics ASAP 2020 instrument. Samples were prepared for each
◦
◦
◦
duced [25,26].
◦
The supported overlayer catalysts described in this paper were
synthesized using the directed deposition technique. This tech-
nique was adapted from the refilling technique used for controlling
particle size in Pt/Al2O3 catalysts as described by Womes et al.
[27]. The synthesis procedure was started by reducing the previ-
ously synthesized base Re/Al2O3 catalyst as described above for the
monometallic catalyst. In order to inhibit overlayer metal precursor
adsorption by the support (to prevent formation of new particles),
water-saturated helium gas was then passed over the catalyst at
room temperature for 24 h. The desired Pd overlayer is deposited
onto the Re base via a hydrogenation reaction that decomposes the
palladium (II) acetylacetonate precursor:
analysis by reduction at 400 C in flowing H2. Isotherms were typ-
◦
◦
ically obtained at temperatures ranging from 35 C to 400 C and
pressures of 1 mTorr to 800 Torr. Reversible adsorption isotherms
were obtained after a 30 min evacuation at the experiment tem-
perature following completion of the total adsorption isotherm.
A leak rate of less than 2 mTorr/min was maintained for accuracy
in the low pressure region of the isotherms.
Reactivity experiments using the ethylene hydrogenation reac-
tion were performed in a 6 mm diameter tubular glass reactor.
The reactor was outfitted with an ethylene glycol/water cooling
◦
jacket that enabled precise temperature control between −10 C
◦
◦
◦
and 70 C. Typical temperatures were −10 C to 30 C. Reaction or-
◦
der determination experiments were performed at 10 C. Catalyst
Pd(acac)2 + 2 · Re − H → 2 · Re − Pd + 2 · acacH.
samples (typically ∼25 mg) were sieved to −100 mesh and di-
luted in 350 mg silica gel (Alfa Aesar, 70–150 mesh) before being
loaded into the glass reactor. Quartz wool was used to hold the
catalyst powder in place. Ethylene (99.9%) and hydrogen (99.999%)
were obtained from Praxair. Nitrogen (Linde) was used as the inert
make-up gas.
Based on this reaction, adsorbed H2 on the base Re particles
is necessary to achieve the overlayer deposition. A final treatment
◦
in flowing H2 was performed for 1 h at 100 C after the previous
support passivation procedure to provide the source of adsorbed
◦
hydrogen. The 100 C treatment temperature corresponds to the
Mass flow controllers (FMA-700 Series from Omega) were used
to control gas flow rates. Total flow rates of 100–1500 mL/min
were possible with this setup, with 750 mL/min being the stan-
dard flow rate for reactivity experiments. Ethylene concentrations
between 0.35 vol% and 10 vol% were used during ethylene reaction
order determination experiments, with 3 vol% ethylene being stan-
dard. Hydrogen concentrations between 10 vol% and 50 vol% were
used during hydrogen reaction order determination experiments,
with 20 vol% hydrogen being standard. At the standard conditions,
reactor pressure was approximately 3 psig. Reactor effluent was
analyzed by GC/MS (Thermo-Finnigan Trace GC/Trace DSQ) using a
temperature of maximum irreversible H2 adsorption by the Re base
metal, which was determined from H2 chemisorption experiments.
To ensure support adsorption of the overlayer metal precur-
sor is as low as possible, acetylacetone was also used as an in-
hibitor. This was achieved by transferring the catalyst to a beaker
of toluene and acetylacetone (Sigma-Aldrich, 99+%) after the H2
treatment. After 15 min a solution of toluene and Pd (II) acety-
lacetonate was added to begin the overlayer deposition reaction.
◦
The reaction was maintained at 60 C, with the volatile reactor
contents maintained in the flask using a water-cooled condenser.
After two hours, the catalysts were filtered and washed with fresh

308
M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
Fig. 1. Measured turnover frequency as a function of total flow rate at 303 K.
◦
Fig. 2. Log-scale isotherms for Re, Pd, and Re@Pd TD catalysts at 50 C.
6-port gas sampling valve. A 10 m Varian PLOT column was used
for chromatographic separation of ethane and ethylene.
3. Results and discussion
Transport limitation experiments were carried out to determine
the proper conditions for performing reactivity experiments. The
transport limitation experiments were performed by measuring
turnover frequency at constant temperature and inlet concentra-
tions, but varying total flow rates. The results of one of these
experiments performed at 303 K are shown in Fig. 1. The mea-
sured turnover frequency, which is independent of total flow rate
when transport limitations are absent, was observed to increase
below 500 mL/min total flow rate. This flow rate corresponded to
approximately 2% conversion of ethylene, therefore indicating that
conversion should be maintained below 2% in order to prevent
heat transfer limitations from this fast and exothermic reaction
from impacting reactivity results. Typical conversions for the re-
sults presented in this paper were 0.03–1.5%.
Transmission election microscopy (TEM) and energy dispersive
X-ray spectroscopy (EDS) was performed on a JEOL JEM-4000FX
microscope operating at 200 kV. Carbon-coated nylon grids (SPI
Supplies) were used to reduce the copper background spectrum,
which was found to interfere with analysis of rhenium during EDS
measurements.
3.1. Hydrogen chemisorption
◦
Hydrogen chemisorption isotherms at 50 C for the Pd, Re and
Re@Pd TD catalysts are shown in Fig. 2. The isotherms include
both reversible and irreversible adsorption and reveal important
adsorption characteristics for each catalyst. The pure Pd catalyst
shows that most low pressure adsorption occurs below 0.01 Torr,
while the pure Re catalyst shows that most low pressure adsorp-
tion occurs between 0.1–2 Torr. Although the Re@Pd TD catalyst
contains a similar amount of Pd as the pure Pd catalyst (see Ta-
ble 1 for metal loadings), the isotherm from the Re@Pd TD cat-
alyst does not show a linear combination of the two pure metal
isotherms. Rather, this isotherm maintains the isotherm features
and adsorption quantities of the pure Re catalyst. The isotherm for
the overlayer catalyst thus indicates that no particles of pure Pd
were formed during the synthesis process and that Pd was pri-
marily deposited on existing Re sites active for H2 chemisorption.
Although not shown here, the isotherm for the Re@Pd HT cata-
lyst was very similar to that of Re@Pd TD. The isotherm for Re@Pd
NI demonstrated significant H2 uptake below 0.01 Torr as in the

M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
309
Fig. 3. Isosteric heat of H2 adsorption on all studied catalysts as a function of volume adsorbed.
pure Pd catalyst, along with adsorption between 0.1–2 Torr. The
3.2. TEM/EDS results
low pressure uptake in this catalyst suggests formation of pure Pd
particles, which is consistent with the synthesis procedure for this
catalyst which did not utilize inhibitors to prevent formation of
monometallic Pd particles on the support.
TEM imaging was unable to observe metal particles on any
of the synthesized catalysts, suggesting that the deposited metals
are highly dispersed on the alumina surface. Rhenium is known
to form highly dispersed phases on alumina even at 1 wt% con-
centrations, making it generally difficult to observe particles with
TEM [32]. No particle observation for the Re@Pd catalysts also in-
dicates that palladium has been deposited on rhenium particles as
desired or as a highly dispersed phase.
The heat of hydrogen adsorption for each catalyst was calcu-
lated from isotherms obtained at multiple temperatures using the
Clausius–Clapeyron equation, the use of which has been well doc-
umented in the literature [28–30]. Resulting values for heat of
adsorption as a function of the quantity of adsorbed hydrogen are
presented in Fig. 3. Results for heat of adsorption on Pd of approx-
imately 60 kJ/mol at low coverage compare well with literature
results. Chou et al. measure a value of 67 kJ/mol for hydrogen ad-
sorption on highly dispersed Pd/Al2O3 [30], while the value from
first principles computation for single crystal Pd(111) was deter-
mined to be 78 kJ/mol at 100% coverage of fcc sites [17]. Re heats
of adsorption were fairly constant at approximately 40 kJ/mol for
all coverages examined. No values for heat of hydrogen adsorption
on Re at high coverage have been found in the literature, but ini-
tial (i.e., low coverage) heats have been reported to be in the range
of 128–139 kJ/mol for unsupported rhenium [18,31].
The Re@Pd catalysts all show decreased adsorption heats com-
pared to the pure Pd catalyst and the base Re catalyst from which
they were synthesized from, in agreement with theoretical pre-
dictions. The maximum values calculated for each catalyst were
37, 28, and 23 kJ/mol for the Re@Pd NI, HT, and TD catalysts,
respectively. The heat of adsorption on the Re@Pd NI catalyst is
higher than the other Re@Pd catalysts, which may be due to the
presence of isolated Pd particles as indicated by chemisorption
isotherms. At 100% coverage of fcc sites, Pallassana and Neurock
predicted the hydrogen heat of adsorption to decrease from an
exothermic −78 kJ/mol to an endothermic +2 kJ/mol on Pd(111)
and PdML/Re(0001), respectively [17]. Similarly, we observe that
hydrogen heat of adsorption decreases from 63 kJ/mol on Pd to
15 kJ/mol on Re@Pd TD at 0.3 cm3/g (STP) hydrogen coverage. The
significantly decreased heats of adsorption shown in Fig. 3 are con-
sistent with the formation of a Pd overlayer on Re particles as
desired. The different metal loadings and synthesis methods also
yield a range of maximum adsorption heats which, together with
ethylene hydrogenation activity, may be used to determine struc-
ture/property relationships for the catalysts.
Since metal particles could not be observed directly, EDS was
also performed to look for regions on the alumina particles where
Pd or Re peak intensities were higher or lower than the average in-
tensity for the entire particle. No Pd peaks were observed on any
of the Re@Pd catalysts, likely owing to the very low loading of Pd
on these catalysts. In contrast, Re peaks were observed on all of
the Re@Pd catalysts, but its intensity was approximately constant
regardless of the location or diameter of the electron beam on an
alumina particle. Fig. 4 shows the results from a sample EDS ex-
periment on the Re@Pd TD catalyst, where the electron beam was
first spread over an entire alumina particle (wide beam area), then
condensed to illuminate a small region of the same particle (con-
densed beam area). The resultant rhenium peaks for these very
different sized areas was little changed, suggesting a uniform dis-
tribution of very small Re particles on the alumina surface.
Estimated Re dispersion determined from chemisorption at
◦
100 C was 34.6%, which gives an average particle size of 3.9 nm
assuming spherical particles, the Re atomic cross-sectional area is
0.065 nm2, and the Re density is 21.04 g/cm3 [33]. This particle
size may be large enough to observe using TEM if the disper-
sion is accurately reflected by hydrogen chemisorption. However,
the data of Chadzynski and Kubicka show that the conditions of
H2 isotherm measurement can significantly affect dispersion es-
timates. Under similar analysis conditions as those of this work,
their results showed that irreversible H2 adsorption for a 1 wt%
◦
Re/Al2O3 catalyst at 100–200 C produced a dispersion of only
approximately 40%, while the dispersion determined under other
conditions and by TEM was 90% [33]. Their results suggest that
the true particle dispersion for the Re and Re@Pd catalysts is likely
much larger than the 34.6% value determined from chemisorption,
decreasing the estimated particle size enough that detection by
TEM may be expected to be very difficult.

310
M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
Fig. 4. Recorded EDS spectra for Re@Pd TD catalyst showing normalized metal counts as a function of energy.
Fig. 5. Re and Pd catalyst Arrhenius-type plot for ethylene hydrogenation (3% ethylene, 20% hydrogen, balance nitrogen).
Diffraction patterns were also obtained for the Re@Pd NI cat-
3.3. Ethylene hydrogenation results
alyst on the TEM with the electron beam expanded over an en-
tire alumina particle. Prior to obtaining the diffraction patterns,
the alumina particles were verified by EDS to contain rhenium.
The diffraction patterns for this catalyst were compared against
diffraction patterns obtained from a sample of the pure γ -alumina
support. Both samples displayed ring-diffraction patterns. Imaging
software was used to measure the radius of each displayed ring
pattern. These values were indexed against the expected atomic
spacings for γ -alumina, rhenium, and palladium. Three diffraction
patterns were obtained on different Re@Pd NI particles. All of the
rings observed coincided with the expected atomic spacings for
alumina except in one case, where a single ring was observed that
was found to closely match the 1.1948 Å rhenium d-spacing. This
finding, along with the lack of observed Re diffraction rings in the
other patterns, simply highlights the low concentrations of metals
on the alumina support for these catalysts, while also suggesting
the presence of very small rhenium particles due to the observed
ring pattern for rhenium.
Arrhenius-type plots for Pd/Al2O3 and Re/Al2O3 catalysts are
shown in Fig. 5 for the ethylene hydrogenation reaction. Reaction
rates on palladium are several orders of magnitude higher than
on rhenium. The Arrhenius-type plots for Pd and Re@Pd catalysts
shown in Fig. 6 indicate that Re@Pd catalysts have activities for
the ethylene hydrogenation reaction very similar to pure Pd cata-
lysts. Due to the much higher activity of bimetallic Re@Pd catalysts
compared to Re, it stands to reason that this reactivity can be
attributed entirely to the presence of palladium. For this reason,
turnover frequencies for all palladium-containing catalysts were
calculated based on measured palladium content from elemental
analysis (see Table 1 for metal loadings). Dispersion values used
in turnover frequency calculations were 0.5 for the Pure Pd cata-
◦
lyst (determined from H2 chemisorption at 35 C) and unity for all
Re@Pd catalysts.
The Arrhenius-type plots of Fig.
6 graphically indicate the
relative reaction rates and apparent activation energies for the
palladium-containing catalysts. Table 2 lists the calculated acti-

M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
311
Fig. 6. Pd and Re@Pd catalyst Arrhenius-type plot for ethylene hydrogenation (3% ethylene, 20% hydrogen, balance nitrogen).
Table 2
nism which contains both competitive and noncompetitive hydro-
gen adsorption sites [36,37]. No similar experimental mechanism
studies were found for Pd. The modified Horiuti–Polanyi mecha-
nism is listed below [36,37]:
Measured catalyst activation energies and reaction orders.
Catalyst
Apparent activation
energy, kJ/mol
Reaction order
H2
C2H4
10 wt% Re/Al2O3
0.1 wt% Pd/Al2O3
Re@Pd HT
Re@Pd TD
Re@Pd NI
103 23
1.0
1.0
1.1
1.0
1.0
−0.01
−0.11
−0.12
−0.12
−0.12
76.1 4.2
78.3 1.5
72.4 1.9
73.9 3.0
1. H2 + 2S ↔ 2HS;
ꢀ
ꢀ
2. HS + S ↔ HS + S;
∗
∗
3. H2 + 2 ↔ 2H ;
ꢀ
∗
ꢀ
4. H + S ↔ HS + ∗;
5. C2H4 + 2 ↔ ∗C2H4
∗
∗
;
ꢀ
ꢀ
vation energies and associated uncertainties based on Fig. 6. The
Re@Pd catalysts show only small differences in reaction rates and
apparent activation energies. Based on the uncertainties, some of
the activation energies are indistinguishable, while others are dis-
tinct. Regardless, with the exception of Re, they are all very similar
and indicate that the overlayer catalyst structure has not resulted
in a mechanism change. The pure Pd catalyst shows an activation
energy in the same range as the Re@Pd catalysts but approxi-
mately 3 times larger reaction rates. The Re@Pd NI catalyst exhibits
higher activity more similar to the pure Pd catalyst than the other
two Re@Pd catalysts. This may be due to the presence of a small
amount of isolated Pd as indicated by the chemisorption isotherms.
Apparent activation energies for ethylene hydrogenation on pal-
ladium have been widely reported in the literature to be between
21 and 50 kJ/mol, while the activation energies measured here are
approximately 75 kJ/mol [34]. The reason for this discrepancy is
not immediately apparent considering the great lengths gone to
eliminating heat and mass transfer effects during measurements.
No activation energy values for ethylene hydrogenation on rhe-
nium were found in the literature, likely owing to its poor activity
for this reaction.
Hydrogen and ethylene reaction orders were determined by
holding the flow rate of one reactant constant while varying the
other and adjusting the nitrogen flow rate to maintain constant
total flow rate. Reaction order plots for hydrogen and ethylene
can be found in Figs. 7 and 8, with resulting values listed in
Table 2. All catalysts were found to show a first order depen-
dence on hydrogen and a slightly negative order in ethylene. At
ethylene concentrations above ∼10%, the reaction rate is indepen-
dent (zero order) of ethylene concentration. Similar reaction orders
were determined for ethylene hydrogenation studies on Pd using
first-principles based Monte Carlo simulations [35].
∗
ꢀ
6. C2H4∗ + HS ↔ ∗C2H5∗ + S ;
7. C2H5∗ + HS ↔ C2H6 + 2 + S .
∗
∗
ꢀ
The S sites are noncompetitive hydrogen adsorption sites that
are inaccessible for ethylene adsorption (explained by the small
size of hydrogen), while the sites are competitive sites that are
accessible to both hydrogen and ethylene. Goddard et al. included
hydrogen activation steps (S sites) to explain isotopic product dis-
∗
ꢀ
tribution observations [36]. They also determined that the ethylene
order was zero above 75 Torr ethylene pressures and slightly neg-
ative at lower pressures, similar to the observations from Fig. 8
here [36,37]. Hansen and Neurock also demonstrated using Monte
Carlo simulations that Pd reaction orders similar to those reported
here would result if lateral interactions were included, thus not
requiring separate ethylene and hydrogen adsorption sites [35].
A hydrogen order that shifted from 0.5 at 248 K to unity at
336 K was also noted [36,37]. Based on the similar reaction or-
der results, it appears that the ethylene hydrogenation mechanisms
previously shown in the literature also apply to the Pd, Re, and
Re@Pd catalysts synthesized for this work. There is no indication
that the deposition of Pd on Re has resulted in a mechanism
change.
Aside from the Re@Pd NI catalyst, chemisorption isotherms in-
dicate that palladium content in the Re@Pd catalysts is not present
in significant amounts as isolated Pd particles, instead forming
bimetallic particles that present adsorption behavior similar to rhe-
nium. While it is possible that Re may re-segregate to the surface
as an oxide during the calcination step of the synthesis proce-
dure, the sample was subsequently reduced in H2. Christoffersen
et al. have shown that when metallic Pd and Re are present to-
gether, Pd will show a strong surface segregation [14]. In addition,
turnover frequencies, which were calculated based on Pd content
alone, are comparable across all of the Pd and Re@Pd catalysts.
Goddard et al. have shown previously that ethylene hydrogena-
tion on platinum proceeds via a modified Horiuti–Polanyi mecha-

312
M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
Fig. 7. Hydrogen reaction order plot for ethylene hydrogenation (3% ethylene, 10–50% hydrogen, balance nitrogen).
Fig. 8. Ethylene reaction order plot for ethylene hydrogenation (0.5–15% ethylene, 20% hydrogen, balance nitrogen).
These results suggest the existence of a palladium/rhenium inter-
action consistent with a Pd overlayer deposited on rhenium for the
Re@Pd catalysts. Heat of adsorption values calculated from hydro-
gen chemisorption isotherms also indicate that the Re@Pd catalysts
bind hydrogen more weakly than rhenium or palladium, in agree-
ment with single crystal and first principles computational studies.
While these results suggest that the directed deposition synthesis
technique has been successfully used to create the desired over-
layer catalysts, there is an apparent discrepancy between the ob-
served reactivity and hydrogen adsorption properties. Chemisorp-
tion results suggest that hydrogen adsorption behavior on Re@Pd
catalysts is similar to rhenium, but with notably weaker binding
strength. In contrast, reactivity results suggest that the Re@Pd sur-
face is little changed from pure Pd except for a slightly decreased
activity.
pends on the rates of hydrogen adsorption and desorption, as well
as the rate of reaction. A decrease in hydrogen adsorption strength
will increase the desorption rate, decreasing surface coverage and
hence decreasing the overall reaction rate as was observed. This
correlation between hydrogen adsorption strength and reaction
rate can be seen in Fig. 9, which plots the maximum heat of hydro-
gen adsorption determined from the chemisorption results against
the turnover frequency measured for each palladium-containing
catalyst at 273 K. The apparent relationship from this data set is
consistent with the suggestion that ethylene hydrogenation on a
palladium surface at the conditions of this work is affected by the
availability of adsorbed hydrogen.
Direct literature reactivity comparisons with this work are dif-
ficult. Most of the literature studies have examined only hydro-
gen and alkene (ethylene, 1-butene, 1-hexene, and cyclohexene)
temperature programmed desorption (TPD) or temperature pro-
grammed reaction (TPR) of an alkene with a predosed hydrogen
surface (see Chen et al. for a summary [38]). The TPD studies show
hydrogen and alkenes are less strongly bound as the d-band cen-
ter moves further away from the Fermi level. This is in agreement
The slightly decreased activity can be explained by considering
the reaction mechanism and the consequences of decreased hy-
drogen heat of adsorption. Because ethylene hydrogenation is first
order in hydrogen and slightly negative in ethylene, the reaction
rate is dependent on the hydrogen surface coverage, which de-

M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
313
Fig. 9. Correlation between hydrogen heat of adsorption and ethylene hydrogenation turnover frequency for studied Re@Pd and Pd catalysts.
with our hydrogen studies. Perhaps the best comparison of these
studies with ours is the work by Humbert and Chen involving hy-
drogen and cyclohexene [39]. They developed a plot showing TPR
hydrogenation activity first increasing and then decreasing as the
cyclohexene binding energy decreased (or as the center of the d-
band moved closer to the Fermi level). Thus depending on the
situation, hydrogenation activity can indeed decrease as the hydro-
gen binding energy decreases. The high initial hydrogen coverages
resulting from the predosing by Humbert and Chen may also par-
allel the possible reaction conditions where the hydrogen partial
pressure is increased to increase the hydrogen surface coverage.
The situation of higher coverage and weaker hydrogen bonding
may then result in an activity increase for the overlayer catalysts
compared to pure Pd.
results in stronger binding of adsorbates [13,14,17]. Christoffersen
et al. used first principles computational techniques to predict the
d-band shifts for a number of bimetallic overlayer systems. In or-
der to achieve stronger adsorbate binding on a Pd overlayer, their
results point to the PdML/Ag and PdML/Au systems as capable of
providing this desired property [14]. However, their data also sug-
gests that Ag and Au will tend to surface segregate or alloy with
Pd, making the Pd overlayers on Ag and Au unstable for synthe-
sis and characterization efforts [14,40]. Single crystal work has also
indicated Au segregation or alloying in the Pd/Au system [8,41]. Fi-
nally, increasing metal–hydrogen bond strength too much may not
always be beneficial either. If the hydrogen is too strongly bound,
this can also limit reactivity. If the bond becomes too strong, the
resulting decrease in activity might manifest itself in a volcano plot
similar to that of Humbert and Chen [39].
If the activation barriers for hydrogenation were also affected
by the overlayer structure, significant changes in reaction rate or
reaction orders may be possible. First principles computation of
ethylene hydrogenation on PdML/Re and Pd instead indicate that
the forward activation barrier for ethylene hydrogenation to sur-
face ethyl (C2H5) is relatively unchanged (78 kJ/mol on PdML/Re vs
82 kJ/mol on Pd) [17]. These values compare very well with the ap-
parent activation energies obtained on the Pd and Re@Pd catalysts
studied here (72.4–78.3 kJ/mol). The agreement of our measured
values with the calculated values from the literature leaves open
the possibility that the apparent activation energy measured here
is the activation barrier for hydrogenation of adsorbed ethylene to
ethyl. This is possible since a first order hydrogen dependence does
not automatically imply that the reaction rate is limited by hydro-
gen adsorption. A first order hydrogen dependence is equally possi-
ble when the surface reaction is limiting, but hydrogen coverage is
low compared to ethylene as suggested in this case by the slightly
negative ethylene order that was observed. Only the reverse activa-
tion barrier was projected by computational studies to be altered
significantly (126 kJ/mol on PdML/Re vs 82 kJ/mol on Pd), but the
effect of this would only be observable under conditions where the
reaction is reversible. Therefore, the measured values for activation
energy and reaction orders are consistent with the predicted acti-
vation barriers for the reaction at the conditions of this work.
The correlation observed in Fig. 9 suggests that ethylene hy-
drogenation activity could be increased through electronic modi-
fication of a Pd overlayer that results in stronger H2 adsorption.
First principles computation research has shown that a decrease
in the d-band center of a metal surface will result in lower ad-
sorbate binding strengths, while an increase in the d-band center
Christoffersen et al. have also shown a linear relationship be-
tween calculated H2 adsorption energy and d-band center for a
number of transition metals and pseudomorphic overlayers (such
as PtML/Ru and PdML/Re) [14]. Our results above have further
demonstrated a correlation between H2 heats of adsorption and
ethylene hydrogenation activity. Thus, for ethylene hydrogenation
and the Re and Pd system, there appears to be a direct relation-
ship between activity and the calculated center of the d-band.
Research by Zhou et al. with Cu@Pt catalysts supported on alu-
mina for NO reduction showed that the Cu@Pt catalysts had equal
activity as pure Pt catalysts. However, they also found a signifi-
cantly increased selectivity for N2 formation on the Cu@Pt catalyst
compared to pure Pt [42]. This result, along with the results pre-
sented here, indicates that it may be possible that overlayer cata-
lysts will in general provide similar activities as catalysts consisting
of the pure overlayer metal, but with differing selectivity.
4. Conclusions
Hydrogen chemisorption and ethylene hydrogenation results
both suggest the formation of Pd overlayers on Re for the cata-
lysts studied here. Chemisorption isotherms and hydrogen heats
of adsorption show distinctly different hydrogen adsorption behav-
ior than would be expected for a physical mixture of Pd and Re
particles and are in agreement with previous literature first prin-
ciples computation and single crystal studies of PdML/Re [16–18].
Ethylene hydrogenation results suggest a highly dispersed Pd sur-
face layer that displays similar reactivity properties as pure Pd,
however, the Re@Pd catalysts show decreased TOF compared to

314
M.P. Latusek et al. / Journal of Catalysis 263 (2009) 306–314
pure Pd. A correlation between measured hydrogen heat of adsorp-
tion and TOF was also observed for Pd-containing catalysts.
All of these results are consistent with electronic modification
of Pd by the formation of Pd overlayers on Re, which was the
desired outcome from the use of the directed deposition synthe-
sis technique. Ethylene hydrogenation activity was observed to be
mildly inhibited for this particular overlayer system due to the de-
crease in surface hydrogen, which suggests that activity enhance-
ment may be observed if a Pd overlayer system was developed that
could bind hydrogen more strongly than pure Pd.
[11] G.F. Cabeza, N.J. Castellani, P. Legare, Comput. Mater. Sci. 17 (2000) 255.
[12] V. Pallassana, M. Neurock, G.W. Coulston, J. Phys. Chem. B 103 (1999) 8973.
[13] B. Hammer, J.K. Norskov, Surf. Sci. 343 (1995) 211.
[14] E. Christoffersen, P. Liu, A. Ruban, H.L. Skriver, J.K. Norskov, J. Catal. 199 (2001)
123.
[15] M.B. Zellner, J.G. Chen, J. Catal. 235 (2005) 393.
[16] V. Pallassana, M. Neurock, L.B. Hansen, B. Hammer, J.K. Norskov, Phys. Rev.
B 60 (8) (1999) 6146.
[17] V. Pallassana, M. Neurock, J. Catal. 191 (2000) 301.
[18] V. Pallassana, M. Neurock, L.B. Hansen, J.K. Norskov, J. Chem. Phys. 112 (12)
(2000) 5435.
[19] S. Alayoglu, A.U. Nilekar, M. Mavrikakis, B. Eichhorn, Nat. Mater. 7 (2008) 333.
[20] F. Epron, F. Gauthard, C. Pineda, J. Barbier, J. Catal. 198 (2001) 309.
[21] F. Gauthard, F. Epron, J. Barbier, J. Catal. 220 (2003) 182.
[22] M.T. Schaal, A.Y. Metcalf, J.H. Montoya, J.P. Wilkinson, C.C. Stork, C.T. Williams,
J.R. Monnier, Catal. Today 123 (2007) 142.
Acknowledgments
[23] M.T. Schaal, A.C. Pickerell, C.T. Williams, J.R. Monnier, J. Catal. 254 (2008) 131.
[24] R.D. Cortright, S.A. Goddard, J.E. Rekoske, J.A. Dumesic, J. Catal. 127 (1991) 342.
[25] J. Okal, L. Kepinski, L. Krajczyk, W. Tylus, J. Catal. 219 (2003) 362.
[26] F. Solymosi, T. Bansagi, J. Phys. Chem. 96 (1992) 1349.
[27] M. Womes, T.C.F.L. Peltier, S. Morin, B. Didillon, Appl. Catal. A 283 (2005) 9.
[28] D. Shen, M. Bulow, F. Siperstein, M. Engelhard, A.L. Meyers, Adsorption 6 (4)
(2000) 275.
The authors gratefully acknowledge funding support from the
Center for Fundamental and Applied Research in Nanostructured
and Lightweight Materials at Michigan Technological University
sponsored by the National Renewable Energy Laboratory of the De-
partment of Energy through grant number DE-FG36-08GO88104.
[29] W. Ranke, Y. Joseph, Phys. Chem. Chem. Phys. 4 (2002) 2483.
[30] S.-C. Chou, S.-H. Lin, C.-T. Yeh, J. Chem. Soc. Faraday Trans. 91 (1995) 949.
[31] K.F. Poulter, J.A. Pryde, J. Phys. D Appl. Phys. 1 (2) (1968) 169.
[32] J. Okal, H. Kubicka, L. Kepinski, L. Krajczyk, Appl. Catal. A 162 (1997) 161.
[33] G.W. Chadzynski, H. Kubicka, Thermochim. Acta 158 (1990) 353.
[34] H. Zea, K. Lester, A.K. Datye, E. Rightor, R. Gulotty, W. Waterman, M. Smith,
Appl. Catal. A 282 (2005) 237.
References
[1] J.P.M. Treichel, in: J.I. Kroschwitz, M. Howe-Grant (Eds.), Kirk-Othmer Encyclo-
pedia of Chemical Technology, vol. 21, John Wiley & Sons, New York, 1991,
p. 335.
[2] R. Massard, D. Uzio, C. Thomazeau, C. Pichon, J.L. Rousset, J.C. Bertolini, J.
Catal. 245 (2007) 133.
[35] E.W. Hansen, M. Neurock, J. Catal. 196 (2000) 241.
[3] M.E. Davis, R.J. Davis, Fundamentals of Chemical Reaction Engineering, 1st ed.,
McGraw–Hill, New York, 2003.
[36] S.A. Goddard, R.D. Cortright, J.A. Dumesic, J. Catal. 137 (1992) 186.
[37] J.E. Rekoske, R.D. Cortright, S.A. Goddard, S.B. Sharma, J.A. Dumesic, J. Phys.
Chem. 96 (1992) 1880.
[38] J.G. Chen, C.A. Menning, M.B. Zellner, Surf. Sci. Rep. 63 (2008) 201.
[39] M.P. Humbert, J.G. Chen, J. Catal. 257 (2008) 297.
[40] A.V. Ruban, H.L. Skriver, J.K. Norskov, Phys. Rev. B 59 (24) (1999) 15990.
[41] A. Jablonski, S.H. Overbury, G.A. Samorjai, Surf. Sci. 65 (1977) 578.
[42] S. Zhou, B. Varughese, B. Eichhorn, G. Jackson, K. McIlwrath, Angew. Chem. Int.
Ed. 44 (2005) 4539.
[4] W. Juszczyk, Z. Karpinski, Appl. Catal. A 206 (2001) 67.
[5] A. Sellidj, B.E. Koel, Phys. Rev. B 49 (12) (1994) 8367.
[6] R.A. Campbell, J.A. Rodriguez, D.W. Goodman, Phys. Rev. B 46 (11) (1992) 7077.
[7] R. Wu, A.J. Freeman, Phys. Rev. B 52 (16) (1995) 12419.
[8] B.E. Koel, A. Sellidj, M.T. Paffett, Phys. Rev. B 46 (12) (1992) 7846.
[9] N.A. Khan, H.H. Hwu, J.G. Chen, J. Catal. 205 (2) (2001) 259.
[10] H.H. Hwu, J.J. Eng, J.G. Chen, J. Am. Chem. Soc. 124 (4) (2002) 702.