Experimental Methods

3.2.1 High-pressure-compatible UHV reaction system

The experimental system consisted of a UHV base chamber and a high pressure reaction cell, as shown in Figure 3.1. The base pressure of the system was 2x10-9 Torr. The high pressure cell (volume 1.25 liter) capable of handling a range of pressures from UHV to ambient conditions was separated from the base chamber by a gate valve (MDC 300004), which allows the base chamber to retain UHV conditions while the high pressure cell is pressurized. The two chambers were pumped either separately or together.

The base chamber was constantly pumped by a turbomolecular pump (Varian Turbo V 250, Macro Torr), which could also pump the high pressure cell when the gate valve was open. The upper chamber was equipped with a separate turbomolecular pump (Varian Turbo V 70LP, Macro Torr). During experiments, gases were introduced to either or both chambers, via a leak valve in the base chamber and a gate valve in the upper chamber. A three-dimensional translational stage (MDC PSMA-1512) equipped with compressible and flexible cooper bellows allowed for fine position control and provided an in situ

sample transfer mechanism between the UHV chamber and high pressure cell. The base chamber included an Ar-ion sputter gun (Omicron ISE 5) for sample cleaning and a quadrupole mass spectrometer (MS) (Hiden HAL IV RC) for residual gas analysis.

Figure 3.1: High-pressure compatible UHV reaction system: (1) Liquid nitrogen feed line for cooling; (2) Thermocouple feed through; (3) Copper feed through for heating; (4) Rotation stage (θ=0-360º); (5) Transitional stage (X, Y, Z); (6) Mass spectrometer; (7) Ion gun; (8) KBr window; (9) Sample, e.g. single crystal Pt(100); (10) Gate valve; (11) Pumping system (Turbo pump & roughing pump).

The model catalyst studied was vertically mounted inside the chamber. Two tantalum wires were spot welded to the back of the sample and used to suspend the sample from a copper sample holder that was electrically isolated from the rest of the internal system. Insulated copper ribbon connecting external electrical feedthroughs with the sample holder allowed the sample to be resistively heated up to 1500 K using a DC power supply (Lambda) by passing current through the tantalum wires. A K-type thermocouple was also spot-welded to the back of the sample for temperature measurement. Heating and cooling were computer controlled within ±0.5 K. A key advantage of the setup is that the thermocouple was in direct contact with the sample, providing exact measurement of the sample temperature. Similar systems often estimate sample temperature by using a thermocouple connected near the sample, for example to the sample holder, or by pyrometer.

As a benchmark test of the PM-IRAS system, serial CO-TPD experiments were performed once the Pt(100) was cleaned using the method described in Chapter 2. Typically, the clean sample was exposed to pure CO at 10-6 Torr condition for certain amount of time (1 Langmuir = 10-6 Torr * 1s) at 300 K, and then the system was pumped down to the base pressure of 10-9 _{Torr prior the IR and TPD measurements. After the} PM-IRAS measurement in the upper chamber, the sample was transferred to the bottom chamber where the MS was located for TPD measurements. The sample was heated from 300 K to 700 K with a ramping rate of 3 K/s, and the desorbed molecules were monitored with MS. High pressure studies of CO adsorption and dissociation were all performed in the upper high pressure cell where the IR spectra were taken during the experiments.

3.2.2 Dipole-dipole coupling model

The vibrational frequency of adsorbed molecules on metal surface is known to be influenced by the mechanical coupling of the molecule to the substrate, the dipole- coupling effect due to the dipole-dipole, dipole-self-image and dipole-other-image interactions, and the chemical effects due to the formation of chemisorption bond [69]. Frustrated translation of CO molecules, usually referred to as the anharmonic coupling of the molecule with the substrate, can provide the dephasing mode during heating and cause a red shift of the vibrational band due to rapid energy exchange between the adsorbates and the substrate [121, 122]. In addition, several studies have shown that adsorption of CO on metal surfaces can be explained by the Blyholder model [123, 124], in which CO molecules bond to the metal via the 5σ CO orbital, with simultaneous backdonation of electrons from the metal into the 2π* CO orbitals. Since 2π* orbitals are antibonding with respect to the CO bond, the backdonation leads to a weakening of C-O bond, thus a lower CO stretching frequency. Recently, Curulla et al. [125] used the ab initio cluster model approach to study adsorption of CO on Pt(100) and found that the dipole-coupling effect, which is dependent on surface coverage, is the main cause of frequency shift in CO stretching vibrational modes.

In this work, a dipole-coupling model was applied to examine the coverage- dependent frequency shift of the absorption band. The relationship between vibrational frequency and surface coverage is shown in Equations 3.1 and 3.2 [99, 126]:

0 0 2 1 1            e v s     (3.1)

5 2 3 3 0 ˆ 2 12 ˆ 2 1 1 z d r d z d r r ij ij ij j i        (3.2)

As shown in Equation 3.1, the frequency of surface adsorbates, ω, with respect to the singleton frequency of an isolated CO molecule on the surface, ωs, is a function of

the adsorbate coverage,  , and the dipole sum Σ0 at a complete coverage. Here, αe and αv,

are fitting parameters associated with the electronic and vibronic polarizabilities of the adsorbed molecules. The dipole sum Σ0 can be calculated according to Equation 3.2, in which rij is the distance between dipoles in the plane of the surface and d is the distance

between the center of the molecular dipole and the image plane of the surface. In this work, the distance d was set to 1Å according to the literature [99, 100]. Based on previous work, the singleton frequency for CO adsorbed on atop sites on Pt(100) was selected to be 2067 cm-1 [107, 125]. The electronic polarizability was chosen to be 2.54 Å3 following the estimation by Scheffler [127]. The vibronic polarizability was set to 0.21 Å3, which was consistent with estimations used in previous CO dipole-coupling models [100].

Dipole coupling simulations were performed using an algorithm programmed in Python that utilized the Numpy random module, and represented a Pt surface as a 60 x 60 array with periodic boundary conditions. The array spacing was equal to the lattice spacing of the Pt(100) surface and CO molecules were only allowed to adsorb onto atop sites. The initial CO molecule was placed randomly on the surface. After the first CO molecule adsorbed, the next molecules were placed either next to or away from existing molecules. Coverage was defined as the ratio of the number of adsorbed CO molecules to the total number of adsorption sites. The dipole sum, Σ0 and the band position of the

system were calculated and recorded for different  values. This process was repeated multiple times for both random and island adsorbate geometries.