# Quantum-state specific scattering of molecules from surfaces

Kai Golibrzuch

Doctoral thesis
Date of Examination:
2014-09-12

Date of issue:
2014-10-15

Advisor:
Wodtke, Alec Prof. Dr.

Referee:
Wodtke, Alec Prof. Dr.

Referee:
Schwarzer, Dirk Prof. Dr.

Referee:
Hasselbrink, Eckart Prof. Dr.

Persistent Address: http://hdl.handle.net/11858/00-1735-0000-0023-9904-3

## Files in this item

Name:
Thesis_Kai Golibrzuch_webopt.pdf

Size:
15,6 MB

Format:
PDF

Description:
Dissertation

The following license files are associated with this item:

## Abstract

### English

In my work, I investigated the quantum-state resolved scattering of three different diatomic molecules (NO, CO, N_{2}) from different surfaces, including Au(111) and Pt(111). I focused on measurements of the energy transfer between the various degrees of freedom available using both state-of-the-art and new methods developed in the course of this work. I strove to investigate a few simple model systems with the goal of discovering generally valid rules for the coupling between different degrees of freedom of these simple model systems.

As a first system, I investigated vibrationally inelastic scattering of nitric oxide (NO) from a single crystal Au(111) surface, a system that has been extensively studied in the past and is thought to be well understood. I measured absolute vibrational excitation probabilities for v = 0→1, 2, 3 scattering as a function of surface temperature and incidence translational energy and compared the results to first-principles independent electron surface-hopping (IESH) theory as well as to an empirical state-to-state kinetic rate model. The excitation probabilities of NO(v =1, 2, 3) increase with surface temperature (T

_{S}) in an Arrhenius-like fashion under all conditions of my work. For each final vibrational state, I find that the Arrhenius activation energy is equal to the vibrational energy required for excitation which shows that the NO vibrational energy is taken from a thermal bath. Narrow angular distributions and early, narrow arrival time profiles indicate a direct scattering mechanism leading to fast recoiling molecules. The experimental observations allow for the conclusion that excitation into all vibrational states occurs upon coupling of the NO vibration to electron-hole pairs (EHPs) of the metal surface and that adiabatic (mechanical) coupling to phonons or translation is negligible. The comparison to predictions of first-principles IESH theory reveals quantitative agreement for v = 0→1 excitation but the theoretical predictions slightly underestimate the probabilities for v = 0→2 excitation and clearly underestimate v = 0→3 excitation. A detailed comparison of the excitation mechanisms reveals that this disagreement for scattering into final vibrational states v

_{f}> 1 results from an underestimation of overtone excitations in the scattering process.

Further failures of the current implementation of the IESH model appear in a comparison to measurements of incidence energy (E

_{I}) dependent NO(v = 3→1, 2, 3) relaxation probabilities. The experiments show that the probabilities for vibrational relaxation increase with incidence energy while the IESH simulations predict the opposite trend. A detailed trajectory analysis reveals that the theoretical model predicts a large fraction of multi-bounce collisions that increases with decreasing E

_{I}. A selection of only single-bounce collisions improves the E

_{I}dependence but still does not reproduce the experimental observations. The single bounce results predict relaxation probabilities that do not depend on E

_{I}. My results indicate that the overestimation of multi-bounce collision in the IESH model is probably related to a corrugated potential energy surface (PES) because multi-bounce artifacts occur also for simple adiabatic calculations on the ground-state PES. The failure might be directly related to a failure of the density-functional theory (DFT) calculations from which the PES was obtained.

As a final study on the NO/Au(111) system, I performed state-to-state time-of-flight experiments on scattering of incident NO(v = 2, 3) from Au(111) into different final vibrational and rotational states at various incidence energies. For the first time, my data shows that vibrationally inelastic scattering of NO from a metal surface can influence the final translational energy (E

_{f}) of the scattered molecules. I find that vibrational excitation leads to a decrease of E

_{f}while vibrational relaxation increases E

_{f}. The amount of vibrational energy that couples to the translational motion (T↔V coupling) depends on incidence energy as well as on surface temperature. I speculate that the T↔V coupling results from an EHP mediated energy transfer mechanism in which vibrational energy is first released (taken) into (from) EHPs which then couple to the translation motion. Furthermore, I observe that the dependence of E

_{f}on the final rotational energy (E

_{rot}) depends on incidence energy as well as on the final vibrational state. At higher E

_{I}, the decrease of E

_{f}with E

_{rot}is similar for all vibrational channels. With decreasing incidence energy,E

_{f}gradually becomes independent of E

_{rot}. This effect is more pronounced and occurs already at higher E

_{I}for vibrationally inelastic scattering. The mechanism for this observation remains unclear but the observations are in agreement with the expectation for dynamical steering effects or multi-bounce collisions that might become important at low E

_{I}. Nevertheless, the data can act as an ideal benchmark for future new or improved theoretical models, which have to treat nonadiabatic as well as adiabatic interactions of the NO molecules with the Au(111) surface correctly in order to obtain reasonable agreement.

As next model systems, I investigated the scattering of CO molecules from Au(111) and Pt(111). The experiments on CO/Au(111) scattering involve measurements of v = 0→1 excitation probabilities as well as measurements of v = 2→2, 1 branching ratios. In both cases, I find that the probabilities for vibrational (de-)excitation first decrease with increasing E

_{I}but then increase again for E

_{I}> 0.4 eV. Overall, the absolute excitation probabilities are about a factor of three lower than observed for NO/Au(111). The results on v = 0→1 excitation are partly in agreement with expectations for trapping followed by desorption at low E

_{I}if one assumes complete equilibration with the surface. However, the time-of-flight distributions for scattering of incident CO(v

_{i}= 2) show that the assumption of complete equilibration with the surface in trapping-desorption is probably not valid in this system. The experimental data shows that incident CO(v

_{i}= 2) molecules can be trapped at the surface but are desorbed in v

_{f}= 1, 2 prior to complete equilibration. This observation is direct evidence for vibrationally hot molecules, often referred to as hot precursors, at the surface. The experiments raise the question about the vibrational lifetime of CO adsorbed on Au(111) and whether it is similar to observed picosecond lifetimes found for CO/Pt(111) or CO/Cu(100).

I further measured CO(v = 0→1) excitation probabilities in scattering from Pt(111). The CO/Pt(111) system exhibits broad angular distributions and T

_{S}dependent arrival time distributions. The excitation probabilities agree with the thermal expectation and reflect complete equilibration with the surface. The data supports vibrational excitation occurring due to trapping followed by desorption after equilibration with the surface. This is further supported by measured speed distributions for desorbing/ recoiling CO(v

_{f}= 0, 1), which only show significant direct scattering for v = 0→0 scattering.

Furthermore, I used a new velocity selected residence time technique to investigate the desorption kinetics of CO from Pt(111) in real-time with microsecond resolution. I measured the time dependent flux of molecules leaving the surface at well-defined final velocity, s

_{f}, as a function of surface temperature. I compare the results of previous studies to the experimental data using a simple first-order kinetic rate model. The comparison demonstrates the capability of the method to judge the reliability of previous results; it is very sensitive to the choice of the kinetic parameters. Furthermore, the experimental data shows clear deviations of the experimental data from the first-order desorption kinetics reported previously. By comparison to a kinetic model involving surface diffusion and adsorption at step sites, I am able to assign the two processes to direct desorption from terraces and to step-to-terrace diffusion followed by desorption from the terrace sites. Finally, I derive a binding energy of E

_{0}= 1.43-1.51 eV for CO adsorption at Pt(111) terraces using transition-state theory; the value is in agreement with recent heat of adsorption measurements.

As a last system, I measured vibrational excitation probabilities for N

_{2}scattering from Pt(111) at various incidence energies ranging from 0.1-1.1 eV. I find again an Arrhenius-like dependence of the v = 0→1 excitation probability on the temperature of the surface with an activation energy equal to the vibrational energy uptake. The Arrhenius prefactors increase with increasing incidence energy with zero threshold and are about one order of magnitude lower than for NO/Au(111). Narrow and T

_{S}independent angular and time-of-flight distributions clearly indicate a direct scattering mechanism. Consequently, the experimental results exhibit all possible fingerprints of nonadiabatic V-EHP coupling for a molecule-surface system in which the very low electron affinity of the gas phase molecules seems to make electron transfer processes very unlikely.

**Keywords:**surface scattering; nonadiabatic; desorption kinetics; CO scattering; NO scattering; Nitrogen scattering; electron transfer