3
36
L.E. Murillo et al. / Journal of Catalysis 268 (2009) 335–342
example, recent results reported by our research group have sug-
gested that the bimetallic Pt–Ni–Pt(1 1 1) surface, with Ni atoms
residing in the second layer of the Pt(1 1 1) substrate, promoted
the selective hydrogenation of acrolein toward the corresponding
unsaturated alcohol under ultra-high vacuum (UHV) conditions
current study bimetallic surfaces were prepared by evaporating
approximately one ML of 3d transition metal (Ni, Co, and Cu) on
the Pt(1 1 1) surface by maintaining the crystal at 300 and 600 K
to prepare the surface 3d–Pt–Pt(1 1 1) and subsurface Pt–3d–
Pt(1 1 1) structures, respectively. The evaporative 3d metal doser
consisted of a metal wire (at least 99.99% purity) wrapped around
a resistively heated tungsten wire. This metal filament was enclosed
in a stainless steel cylinder with an opening of ꢀ1 cm in diameter.
[
18,19]. The combined experimental and DFT results suggested
that the presence of weakly adsorbed acrolein through a di- -C–
r
O configuration appeared to be responsible for this desired hydro-
genation pathway [18].
During the metal evaporation, the UHV pressure remained below
À10
In the current paper we extend our studies to other 3d/Pt(1 1 1)
bimetallic surfaces, including Co and Cu, to determine the general
trend in the hydrogenation activity and selectivity of acrolein on
subsurface Pt–3d–Pt(1 1 1) and surface 3d–Pt–Pt(1 1 1) structures.
In addition to experimental studies using TPD and HREELS, DFT cal-
culations are also performed to correlate the hydrogenation activ-
ity with the binding energy and adsorption geometry of acrolein on
these bimetallic surfaces.
5 Â 10
Torr. The metal coverage was estimated using AES by
monitoring the 3d(LMM)/Pt(241 eV) AES peak-to-peak ratio.
Self-consistent periodic slab calculations were performed by
density functional theory (DFT) using the code VASP (Vienna ab
initio Simulation Package) [26]. A plane-wave basis set with a cut-
off energy of 396 eV was used to solve the Kohn–Sham equations.
The PW91 functional was utilized to describe the exchange corre-
lation term. Vanderbilt ultrasoft pseudopotential was used to de-
scribe the core electrons and the nuclei of the atoms, as
described previously [27,28]. The different electronic energies
were calculated using a 3 Â 3 Â 1 k-point grid mesh. The bimetallic
surfaces were modeled using 3 Â 3 super cells of four layers of
thickness, which were separated by a vacuum region equivalent
in thickness to six metal layers to prevent electronic interaction
between slabs. The binding energy of acrolein was calculated with
one acrolein molecule adsorbed per unit cell, with the top two me-
tal layers allowed to relax. Calculations for gas-phase acrolein and
adsorbate–metal systems were carried out spin-unpolarized, as de-
scribed previously [18].
2
. Experimental and DFT methods
A two-level stainless steel UHV chamber with a base pressure
À10
less than 1 Â 10
Torr was used to carry out the TPD experiments.
This UHV chamber has been described elsewhere [20]. In brief,
bimetallic surfaces were prepared using a Pt(1 1 1) single crystal
(
Metal Crystals and Oxides, Ltd., Cambridge) as a substrate for the
different 3d transition metals. This crystal was spot welded directly
to two tantalum posts for resistive heating and thermal contacts for
cooling with liquid nitrogen. The Pt(1 1 1) surface was prepared by
cycles of sputtering, oxygen treatment, and annealing, as described
previously [21]. The cleanliness of the surface was checked by Au-
ger electron spectroscopy (AES). After dosing acrolein or hydro-
gen/acrolein at ꢀ100 K, the TPD experiments were performed
with the surface placed at a distance of ꢀ5 mm from the opening
of the random flux shield of the mass spectrometer. A heating rate
of 3 K/s was used to a maximum of 800 K while collecting 10
masses simultaneously. The TPD yields were estimated by using
the procedure reported by Ko et al. and the sensitivity factors rela-
tive to CO as explained in previous studies [19,22].
3
. Results and discussion
3
3
.1. TPD of acrolein on Pt–3d–Pt(1 1 1) and 3d–Pt–Pt(1 1 1) surfaces
.1.1. Reaction products from Ni/Pt(1 1 1) surfaces
Fig. 1 shows the TPD spectra after the adsorption of 0.5 L acro-
lein, corresponding to a coverage near the saturation of the first
monolayer, on Pt–Ni–Pt(1 1 1), H/Pt–Ni–Pt(1 1 1) (representing
Pt–Ni–Pt(1 1 1) with ꢀ50% saturation coverage of pre-adsorbed
hydrogen), Ni–Pt–Pt(1 1 1), and Pt(1 1 1). The TPD results on these
surfaces have been described previously [18,19]; they are shown
here to provide a reference to compare with other Pt–3d–
Pt(1 1 1) and 3d–Pt–Pt(1 1 1) surfaces. In brief, masses characteris-
tic of different desorption products are compared in Fig. 1, includ-
The vibrational spectroscopic measurements were carried out
in a separate UHV chamber equipped with an LK-3000 double-pass
HREEL spectrometer for vibrational analysis, as described previ-
ously [23,24]. The intensity of the elastic peak was in the range be-
4
5
tween 3 Â 10 and 3 Â 10 counts per second (cps) with a spectral
À1
ing hydrogen (H
molecular desorption of acrolein (CH
panol (CH –CH –CH –OH, 60 amu), propanal (CH
8 amu), and 2-propenol (CH @CH–CH –OH, 31 amu). Consistent
2
,
2 amu), carbon monoxide (CO, 28 amu),
@CH–CH@O, 56 amu), 1-pro-
–CH –CH@O,
resolution between 30 and 40 cm full-width at half maximum
2
(
FWHM). The exposure of acrolein or hydrogen/acrolein was made
3
2
2
3
2
with the Pt(1 1 1) or bimetallic 3d–Pt(1 1 1) surface held at <120 K.
The initial spectrum was scanned at low temperature (110À120 K).
The adsorbed layer was then annealed to a specific temperature
with a linear rate of 3 K/s, held for 5 s, then cooled down to
5
2
2
with that observed by Zaera et al., acrolein undergoes mainly
decarbonylation on the Pt(1 1 1) surface [29]. No noticeable
desorption features are observed from 31, 58, or 60 amu from
Pt(1 1 1). In comparison, both propanal and 2-propenol are de-
tected at 272 K from the Pt–Ni–Pt(1 1 1) surface. When acrolein
is adsorbed on the hydrogen pre-dosed H/Pt–Ni–Pt(1 1 1) surface,
a small desorption peak of 1-propanol is observed at 268 K, and
the desorption of propanal at 193 and 272 K and 2-propenol at
ꢀ
120 K for data collection.
Acrolein (Alfa Aesar, 99.9% stabilized with 0.1% hydroquinone)
was purified by successive freeze–pump–thaw cycles prior to
use. The purity was verified in situ by mass spectrometry. Hydro-
gen, oxygen, and neon were all of research grade purity
(
99.999%) and were introduced into the UHV chamber without fur-
2
72 K is enhanced due to the presence of pre-adsorbed hydrogen.
ther purification. Doses are reported in Langmuirs (1 Langmuir
À10
The molecular desorption of acrolein is noticeable from both Pt–
Ni–Pt(1 1 1) and H/Pt–Ni–Pt(1 1 1). On the Ni–Pt–Pt(1 1 1) surface,
the molecular desorption of acrolein occurs at 190 K, and a desorp-
tion peak of propanal is observed at 302 K and a very weak peak of
(
L) = 1 Â 10
Torr s) and are not corrected for ion gauge sensitiv-
ities. Acrolein, hydrogen, and oxygen were dosed through direc-
dosing, the
tional dosing tubes with a diameter of ꢀ5 mm. For H
2
exposure of 0.5 L in the TPD chamber and 5 L in the HREELS cham-
ber resulted in ꢀ50% saturation coverage of hydrogen.
2
-propenol is detected at 304 K.
As reported earlier for the Ni/Pt(1 1 1) system, a surface Ni–Pt–
Pt(1 1 1) structure was prepared by depositing one monolayer
3.1.2. Reaction products from Co/Pt(1 1 1) surfaces
(
ML) of Ni at 300 K, while the subsurface Pt–Ni–Pt(1 1 1) structure
The TPD spectra after the exposure of 0.5 L acrolein on different
Co/Pt(1 1 1) surfaces are displayed in Fig. 2. The characterization of
was obtained by the deposition of one ML Ni at 600 K [25]. In the