Atom-surface scattering dynamics

RAT Group

Hydrogen is the simplest open-shell atom, and understanding its surface dynamics has implications ranging from interstellar chemistry to maximizing the performance of neutral beam injectors at the International Thermonuclear Experimental Reactor (ITER). Due to its simplicity, H atom surface scattering is particularly attractive to make detailed comparisons between experiment and first-principles theories. Furthermore, due to its low mass, an electronically adiabatic picture predicts inefficient energy transfer to most solids. Hence, H atom interactions with solids can be particularly sensitive to failure of the Born-Oppenheimer approximation. Furthermore, hydrogen is an ideal candidate to test the influence of nuclear quantum effects and the validity of the classical approximation for nuclear motion.


H Streuung 460

We have designed a ground breaking new experimental apparatus where H-atoms are produced by photodissociation (HI + hν → H + I) of a diatomic molecule pre-cooled in a molecular beam expansion. This allows us to control the translational energy of H-atoms between, Ei= 0.1 and 7 eV, and to produce H-atoms with extremely narrow initial translational energy distributions, ΔEi < 2 meV. This is combined with the well known Rydberg atom tagging (RAT) method, which can read the final translational energy of the H-atom after collision with the surface with a similar energy precision and accuracy. This instrument is equipped with a 6-axis sample manipulator that is designed to control the temperature of the target crystal from 100K to 1500K. The detector used in the Rydberg atom tagging is rotatable around two rotation axes, allowing us to readily investigate in-plane and out-of-plane scattering.


Experimental setup of the RAT machine. In the source chamber, a hydrogen halide molecular beam is formed in a supersonic expansion from a pulsed nozzle, passes a skimmer, and is intersected by the photolysis laser before it hits a liquid nitrogen cooled beam dump. Part of the generated H atoms leave the source chamber through a second skimmer, pass two differential pumping stages (DS1 + DS2), and enter the scattering chamber where they hit the sample surface. The sample is mounted on a manipulator allowing the incidence angles and surface temperature to be varied. The scattered H atoms are excited to a metastable Rydberg state by the tagging lasers, pass an aperture defining the angular resolution of the detector, are field ionized after a 250 mm flight path, and detected by a MCP detector. The detector is mounted on a rotatable arm to enable variation of the scattering angles. Bünermann et al., RSI 89, 094101 (2018), Bünermann et al., JPC A 125, 3059 (2021)

H-atom adsorption on metal surfaces

The high adsorption probability of atomic hydrogen on metal surfaces has long puzzled scientist. For an atom to adsorb to a surface, its translational energy and the binding energy have to be transferred to the solid. However, the hydrogen atoms low mass limits the energy transfer to the lattice vibrations of the surface. With our experiments, we could show that a large amount of translational energy is transferred to electronic excitations of the metal, which explains the high adsorption probabilities. We compared the energy loss of H atoms scattered from a metal and an insulator surface. In an adiabatic picture, both surfaces should lead to a very similar energy loss. However, we observed a large difference between the two surfaces: In the case of the metal surface, the H atoms lost much more energy. Our observation suggests that in the metal case the kinetic energy of the hydrogen atoms can be transferred very efficiently to electronic excitations. In the case of the insulator surface, no low-energy electronic excitations are available due to the large band gap, making energy transfer to electronic excitations impossible. Molecular dynamics (MD) simulations further support our interpretation. We work in close collaboration with colleagues from theoretical chemistry, who in parallel develop theoretical models to describe the hydrogen atom interaction with metal surfaces. MD simulations can reproduce our experimental observations only when electronic excitations and all nuclear degrees of freedom are included.


H atom scattering from a metal and an insulator surface. A: Energy loss of 2 eV H atoms scattered from a metallic Au(111) surface (open squares) and an insulating Xe surface (filled squares). B: Comparison of the experimental results for Xe with adiabatic Molecular Dynamics (MD) simulations (blue line). C: Comparison of the experimental results for Au(111) with adiabatic MD simulations (blue line) and MD simulations with electronic friction (red line). Bünermann et al., Science 350, 1346 (2015), Herlt et al., JPC A 125, 5745 (2021)

To gain a better understanding of the underlying processes we studied the influence of different experimental parameters on the energy loss. We varied the incidence conditions of the H atoms: kinetic energy, polar and azimuthal incidence angles. We studied the isotope effect as well as six different fcc (111) metal surfaces. For silver, we studied the influence of the surface structure by comparing three different surface facets (111), (110) and (100). Surprisingly, we observe very similar energy loss distributions in all the cases. However, an important aspect that we have not yet investigated in detail is the influence of the surface temperature. Janke and Hertl et al. have theoretically studied the effect of the surface temperature on the energy loss spectra of the scattered H atoms. They found, that the spectra observed at room temperature are significantly broadened thermally. The simulations predict that the broad, structureless spectra at room temperature begin to show distinct features for temperatures below 100 K. This predicts that different metals, as well as different facets, will show different spectra at low temperatures. All of these differences are washed out at room temperature by thermal broadening. These observations make it highly desirable to perform experiments at low surface temperatures to confirm the prediction and to provide a much more rigorous test of the theoretical simulations. Therefore, one of our main goals is to modify the experimental setup so that we can perform scattering experiments at very low surface temperatures.

Following covalent bond formation of H on graphene

When the open-shell H-atom, the simplest example of a free-radical, collides with an unsaturated molecule it is possible that electronic re-hybridization occurs and a transient chemical bond will form. The bond energy being released creates an addition complex with enormous energy initially localized in the newly formed bond. This transient bond is intrinsically unstable with respect to re-dissociation; but energy flow from the newly formed bond to the rest of the molecule can delay re-dissociation or lead to adsorption.

H-atom scattering from graphene is an ideal model system in this context. For a C-H-bond to be formed, the delocalized electronic structure of graphene has to be locally destroyed, giving rise to an adsorption barrier. For low incidence energies, the H-atom is reflected from the barrier, but for energies exceeding the barrier, a transient C-H bond can be formed. In experiment, we observed that both cases lead to very different scattering distributions, enabling us to follow the energy flow from a newly formed C-H bond to the graphene layer.


H atom scattering from graphene. A: Structural change of graphene upon hydrogen atom adsorption. B: Cut of the multidimensional potential energy surface (PES) of a hydrogen atom on graphene. Hz denotes the distance of the H atom located above one carbon atom from the surface. Cz denotes the corresponding carbon atom z-coordinate. Three example trajectories are shown: (cyan) reflection from barrier, (gold) adsorption, (black) transient bond formation. C: Angular and energy resolved kinetic energy distribution of 2 eV hydrogen atoms scattered from graphene on Pt(111) for an incidence angle of 44° (normal energy En = 0.99 eV). Bünermann et al., Science 364, p. 379 (2019)

In the experiment, graphene is grown on a substrate, while in the theory it is assumed to be free-standing. Therefore, only qualitative agreement can be achieved, for a quantitative agreement, the substrate effect must be included in the theory. Experimentally, we approach this problem by varying the substrate and following the changes in the energy loss spectra with the intention to identify the influence of the substrate on the properties of graphene. Our first experiments were performed on graphene on Pt(111). Further substrates we plan to study are Ni(111), Ir(111) as well as graphite (HOPG).

Electronic excitation of semiconductor surfaces

It is well known that absorption of photons in the bulk of a semiconductor excites electrons from the valence to the conduction band. In our experiments, we have shown that a colliding atom can also efficiently promote electrons to the conduction band in a similar way in a purely surface-specific process. The probability to convert the translational energy of the H atom into electronic excitation of the solid increases dramatically with the incidence energy, as does the average excitation energy. The large excitation probability as well as the large energy loss cannot be explained by theory so far. Therefore, our observation stands as a challenge for new theories of electronically non-adiabatic surface chemistry.

Investigation of the isotope effect by replacing hydrogen with its heavier isotope deuterium revealed that the probability to excite an electron into the conduction band does not depend significantly on the isotope. This result is surprising, since non-adiabatic effects normally show a strong isotope effect. The absence of a significant isotope effect combined with the band structure calculations performed by Prof. Jiang suggest a site-specific non-adiabatic process.


H atom scattering from Ge(111)c(2x8). A: Energy loss of H atoms with various incidence energies scattered from Ge(111)c(2x8) compared to MD simulations (blue line). B: Surface structure of the Ge(111)c(2x8) reconstruction. C: Angular resolved kinetic energy distributions of H atom scattered from Ge(111)c(2x8) in comparison to MD simulations for two incidence energies and an incidence angle of 45°. Krüger et al., Nature Chemistry 15, p. 326 (2023), Krüger et al., Natural Sciences (2023)

Details of the underlying mechanism that facilitate electronic excitations in a hydrogen atom semiconductor surface collision are still lacking. We intend to study the influence of the surface electronic structure on the energy loss of the hydrogen atom to deepen our understanding. Depending on the material, facet, and reconstruction, surfaces of semiconductors can have diverse electronic structures. While the Ge(111)c(2x8) is semiconducting, the famous Si(111)(7x7) reconstruction is metallic. Another example is the Ge(100)(2x1) reconstruction, where different reconstructions with different electronic structures coexist. Partial hydrogenation of a surface can lift the reconstruction and thereby change the electronic structure of the surface. In summary, semiconductor surfaces provide an interesting test ground for non-adiabatic processes. Our current focus is to study the effect of doping on the hydrogen atom scattering dynamics from various silicon surfaces.

Controlling energy transfer by surfaces with tunable properties

Nowadays, new materials having unique properties that can be tailored by structure, composition, temperature and external fields can be synthesized using a large variety of methods. Such materials may have the potential to influence energy transfer processes on their surfaces and for example allow control over surface reactions. To identify the potential of this materials, we want to perform fundamental studies of the energy exchange on such surfaces.
We are interested in various classes of materials: thin film materials that have tunable electronic and photonic states, solids that show strong electron phonon coupling, and materials showing metal insulator phase transitions. The samples are either provided by colleges of the SFB 1073 or are synthesized in-situ in our apparatus.