OXIDE SUPPORTED METAL THIN FILM CATALYST.PDF

INTRODUCTION

In recent years, metal nanoparticles and thin films supported on oxides have become

fundamental components of many devices as their small dimensions present

structures with new chemical and physical properties, often enhancing the reactivity

of these surfaces relative to their bulk counterparts. Numerous theoretical and experimental

studies show that the metal particle size and shape as well as direct adsorbate

interactions with the oxide support can each play a key role in enhancing the reactivity

of these surfaces.  Further investigations imply that the support material mayalso influence the metal’s intrinsic reactivity at these surfaces.  However, these

studies were not able to isolate this effect from those of particle size, shape, and direct

adsorbate-support interactions.

It was not until recently that Chen and Goodman probed the influence of the

oxide support material on the intrinsic properties at the metal surface. By covering

a titania support with one or two flat atomic layers of gold they eliminated, direct

adsorbate–support interactions as well as particle size and shape effects. Their results

definitively showed that the electronic properties at the metal surface changed due to

charge transfer between the support and the metal. Furthermore, their comparison of

one- and two-layer films highlighted the dependence of these effects on the thickness

of the metal slab.

These studies indicate that the charge transfer at the metal–oxide interface alters

the electronic structure of the metal thin film, which in turn affects the adsorption of

molecules to these surfaces. Understanding the effect that an oxide support has on

molecular adsorption can give insight into how local environmental factors control

the reactivity at the metal surface, presenting new avenues for tuning the properties

of metal thin films and nanoparticles. Coupled with the knowledge of how particle

size and shape modify the metal’s electronic properties, these results can be used

to predict how local structure and environment influence the reactivity at the metal surface.

  The Theory of Molecular Adsorption on Metal Surfaces

While there is no complete theory of surface reactivity, an understanding of how

reactant, intermediate, and product adsorbates interact with a surface often gives insight

into the catalytic properties of a metal. Quantum mechanical theories show that as long

as the perturbation due to the interacting systems is small, the interaction of two isolated

systems can be estimated using second order perturbation theory:

E= ( Vij )2 divided by  εi - ε j.

,

where εi and ε j are the eigenvalues of the unperturbed isolated systems, and Vi j

represents the coupling matrix for these interactions. When a molecule interacts

with a metal surface, the interaction involves both a transfer of electrons from

the molecule’s highest occupied molecular orbital (HOMO) to the metal surface

(direct bonding) and a shift of electrons from the metal surface to the molecule’s

lowest unoccupied molecular orbital (LUMO)(back bonding). For transition metal

surfaces, the exchange of electrons is achieved through interactions with the metal d

orbitals.

Using perturbation theory, Hammer and Nørskov developed a model for predicting

molecular adsorption trends on the surfaces of transition metals (HN model). They used

density functional theory (DFT) to show that molecular chemisorption energies could

be predicted solely by considering interactions of a molecule’s HOMO and LUMO with

the center of the total d-band density of states (DOS) of the metal.  

  METHODOLOGY

We examined the deposition of Pt thin films onto the strongly polar O-terminated

(OT) α-alumina surface. Studies have shown that this surface is stabilized by the

adsorption of hydrogen or a supported metal.  We used a slab geometry with an

in-plane.

  

Unit cell and periodic boundary conditions to model the alumina

surface. The layer stacking can be represented by the formula (Al-O3-Al)4-

Al-O3-(Pt3)n, where n is the number of Pt layers. The interfacial Pt atoms are in

the energetically preferred registry, directly above the surface O atoms, forming a Pt

(111) film. All calculations were performed using density functional theory (DFT)19,20

with the generalized gradient approximation21 for the exchange-correlation functional,

as implemented in the dacapo code,22 with a plane-wave cutoff of 30 Ry, ultrasoft

pseudopotentials23 and a 2 × 2 × 1 Monkhorst–Pack24k-point mesh. All slab calculations

were performed with at least 12 °A of vacuum between periodic images in the

[0001] direction. Total energies were tested to ensure that there were no surface interactions

through the slab or the vacuum. The theoretical α-Al2O3 in-plane lattice constant

of 4.798 °A was used (experimental=4.759 °A25). In order to eliminate strain effects, all

adsorption energies were compared to unsupported Pt (111) calculations at the same in plane

geometry as the supported metal. (This corresponds to the experimental Pt (111)

and occupied dxz and dyz states for the nPt/Al2O3 system. All energies are in eV. 

dz2 + dxz + dyz

 . For our simulations, we fixed the ions in the bottom

two alumina layers to their theoretical positions, relaxed the third layer perpendicular

to the surface, and fully relaxed the remaining alumina layers, Pt layers and adsorbates

until the force on each atom was less than 0.01 eV/A° .We corrected for known DFT CO

chemisorption errors using the extrapolation method of Mason and coworkers 

and estimated the error bar for computed chemisorption energies to be 0.01 eV.

  CO CHEMISORPTION TO ALUMINA-SUPPORTED PT THIN FILMS

  chemisorption energy for the adsorption of CO onto

the Pt/OT α-alumina system as a function of metal layers, relative to Pt (111). For

the monolayer of metal on the surface there is an enhancement of the CO top site

binding energy relative to Pt (111). On the other hand, the second layer of Pt/OT shows

a dramatic decrease in the top site chemisorption energy. For three layers of Pt on

this surface, the chemisorption energy oscillates above the Pt (111) energy, eventually

returning to the Pt (111) value for n > 4.

According to the HN model, an increase in chemisorption energy should correlate

to an upward shift in the total d-band center of the metal. Contrary to this theory, the

monolayer of Pt/OT system shows stronger chemisorption but a large decrease (0.59 eV)

in the d-band center (−2.28 eV) relative to bulk Pt (111) (−1.69 eV). Furthermore, the

downward shift in the d-band center of the bilayer Pt/OT system (0.03 eV) is too small

to account for the reduction in chemisorption energy. To explain these deviations, it is

necessary to understand how CO binds to a metal surface and how the oxide support

alters the electronic properties of the metal surface to allow for changes in molecular

adsorption.

 

   .

Explanation with an example:  

A) Metallic Deposition on an Electronegative Support: Pt/OT

The absence of Al atoms on the OT α-alumina surface leaves the surface oxygen

ions electron deficient. As a result, the deposition of a metal onto this support will

result in the formation of hybrid orbitals between the metal d orbitals and the oxygen

p orbitals. Figure 2.1 shows the projected DOS of the surface Pt atoms for one and two

layers of Pt deposited onto an OT-alumina surface. Since the Pt atoms are deposited

directly on top of oxygen ions, hybrid orbitals are formed with metal d orbitals with

components perpendicular to the oxide surface (dz2 , dxz and dyz ). Each of these d states

participates in a higher energy anti-bonding state and a lower energy bonding state. For

the dz2 orbital, the higher energy state is unoccupied (∼1 eV above the Fermi level).

Forming this hybrid orbital increases the number of free d states available for direct

bonding with incoming adsorbate molecules. However, the newly formed dz2 bonding

orbital is quite low in energy (∼6 eV below the Fermi level). This hybrid orbital reduces

the number of occupied states near the Fermi level and results in a downward shift in

the center of the metal d-band. The newly formed dxz and dyz bonding hybrid orbitals

also decrease the number of filled states with energies just below the Fermi level and increase the number of states with energies below −5.0 eV, with little or no change in

the number of free states in these orbitals. Therefore, the formation of bonding hybrid

orbitals will weaken the back bonding of molecules to these states and shift the center

of the metal d-band downward. For a monolayer of Pt on Al2O3, the large enhancement

in CO chemisorption energy is due to the formation of strong direct bonds between the

filled CO 5σ and the unoccupied dz2 antibonding hybrid states.

The second layer of Pt on OT Al2O3 shows a smaller deviation from the HN model

than the monolayer. In this system, Pt atoms on the surface occupy hollow sites of the

interfacial Pt atoms, as in a fcc (111) surface. This allows the d orbitals of the interfacial

Pt atoms to directly interact with the dxz and dyz orbitals of the second-layer atoms,

forming metallic bands. Due to the influence of bonding at the metal–oxide interface,

the dxz and dyz orbitals exhibit a slight downward shift in the center of the metal bands

relative to Pt (111), implying that these orbitals will have only a slight change in their

back bonding relative to Pt (111). On the other hand, the dz2 states gain a fraction of an

electron in response to the decreased electron–electron repulsions in the Pt atom due to

the loss of electrons from the other d-orbitals. This gain in charge results in a dramatic

reduction in free dz2 states and shifts the dz2 DOS upward due to strong electron–

electron repulsions within this orbital. These changes suggest that the dz2 orbitals will

form much weaker direct bonds with adsorbed molecules as they will have fewer states

to accept electrons. Therefore, the large decrease in bonding at the metal surface is due

to the reduction in the number of unoccupied dz2 . For three or more layers of Pt, metal

charge screening effects prevail and the DOS of the surface atoms shift’s back to Pt

(111) values, thereby defining the length scale of the metal-support interactions.

Conclusion

In the above discussion , elaboration of how chemisorption trends for oxide-supported

metal systems can be correlated to shifts in the d-band center of a modified DOS

constructed from the occupied and unoccupied states that are involved in bonding with

the adsorbate molecule. While we presented specific results for Pt supported on an

OT alumina surface, similar investigations on Pd and Rh supported on the AlT and OT

surfaces indicate that these results may be generalized for transition metals deposited

on electronegative and electropositive supports. Furthermore, we use DFT and the

adsorption of CO to the Pt surface to demonstrate how the orbital-specific electronic

structure of a metal surface can be used to predict the strength of the metal’s interaction

with molecular adsorbates.We show that orbital-specific considerations are necessary

to predict bonding, as metal–oxide interfacial charge transfer can have significant

effects on the electronic structure of molecular orbitals which may result in shifts in

the total d-band center which are not related to the binding of a molecule to the metal

surface. The fact that these effects are greatly diminished at four or five layers indicates

that the metal–support interactions have a small characteristic length scale on the order

of the metal interlayer spacing. While it is known that low-coordinated sites greatly

enhance the reactivity of metal particles, these results demonstrate the importance of

metal-support charge transfer in defining the properties at the metal surface, offering

further support for the recent work of Chen and Goodman. In addition, our findings

suggest that increased reactivity at the perimeter of metal particles with diameters

<5 nm may be partially attributed to the strong metal–oxide coupling accessible at these boundaries. These concepts may be useful for applications as diverse as chemical

sensors, fuel cells, and photochemical reactions.