The Born-Oppenheimer approximation (BOA) is fundamental to computational chemistry, assuming rapid electronic adjustment to nuclear motion. Its validity has been probed using hydrogen atom scattering from various surfaces. While adequate for insulators and reasonably accurate for metals (with electronic friction accounting for deviations), its behavior with semiconductors remains unclear. This study investigates hydrogen atom scattering from a Ge(111)c(2×8) semiconductor surface to examine BOA validity in this context. Semiconductors offer a unique opportunity to study a system intermediate between insulators and metals. Visible light, with its high-frequency electric fields (~10¹⁴⁻¹⁵ Hz), efficiently excites electrons across the bandgap in semiconductors. This raises the question: can atom-surface collisions, generating lower-frequency electric fields, also induce bandgap transitions? Existing electronic friction theories predict weak effects. Limited prior work, involving transient currents from Xe scattering, offered evidence of BOA failure but provided limited insight into collision dynamics. This study aims to directly test the BOA's limits in semiconductor surfaces using high-energy hydrogen atoms, resolving sub-picosecond excitation dynamics.

Previous studies using hydrogen atom scattering have examined BOA validity on various surfaces. For insulators like Xe, MD simulations accurately reproduced experimental energy losses, confirming BOA's applicability. Similar results were seen for semi-metals like graphene. However, energetic hydrogen atoms colliding with metal surfaces consistently excite electron-hole pairs (EHPs), demonstrating BOA failure. This failure, however, was successfully modeled using weak-coupling electronic friction approximations, suggesting a perturbative treatment suffices. Existing literature on semiconductor surfaces is limited. Transient currents observed during Xe atom scattering from semiconductors indicated BOA failure, but the underlying collision dynamics remained poorly understood. These past studies provide a context for the current research, highlighting the need for a more comprehensive investigation of BOA validity on semiconductor surfaces.

Experiments involved scattering high-energy hydrogen atoms from a reconstructed Ge(111)c(2 × 8) surface. Translational energy-loss measurements were performed to analyze excitation mechanisms. First-principles electronically adiabatic molecular dynamics (MD) simulations were conducted using a newly developed high-dimensional neural-network potential energy surface (NN-PES). These simulations allowed for a comparison between the experimental results and the predictions based on the BOA. The simulations included and excluded electronic friction effects to assess the contribution of non-adiabatic processes. The experiments and simulations explored a range of incident energies above and below the semiconductor's bandgap (0.49 eV) and different incident angles, providing a detailed picture of scattering dynamics.

Experiments revealed bimodal energy-loss distributions at incident energies above the bandgap. A low-energy-loss feature (adiabatic channel) was accurately reproduced by the electronically adiabatic MD simulations, similar to H atom scattering from Xe. However, a high-energy-loss feature (VB-CB channel) was observed experimentally but absent in the simulations, both with and without electronic friction. The onset of the VB-CB channel precisely corresponded to the semiconductor's bandgap energy (0.49 eV). The VB-CB channel’s intensity increased significantly with the hydrogen atom's incident energy, indicating a strong non-adiabatic interaction responsible for exciting electrons across the bandgap. At higher incident energies, the VB-CB channel accounted for about 90% of the scattering events, confirming BOA failure. Angular distribution analysis showed the VB-CB channel exhibits a much narrower angular distribution than the adiabatic channel, suggesting distinct scattering mechanisms. The average energy transfer in the adiabatic channel remained a small fraction (~10%) of the incident energy, while the VB-CB channel demonstrated a larger fraction at lower incident energies due to the bandgap energy constraint. These findings strongly suggest that hydrogen atom collisions with the semiconductor surface efficiently promote electrons from the valence band to the conduction band through a non-adiabatic process.

The findings demonstrate a clear instance of BOA failure in semiconductor surfaces, distinct from previously observed behavior in insulators and metals. While the adiabatic channel resembles H atom scattering from insulators like Xe (without BOA failure), the VB-CB channel represents a novel non-adiabatic mechanism. The efficient electron promotion across the bandgap caused by hydrogen atom collisions suggests new possibilities for manipulating electron behavior in semiconductors. The sub-picosecond interaction time inferred from the scattering angles further indicates a direct, highly efficient energy transfer from the incident hydrogen atom to the semiconductor's electrons. The absence of the VB-CB channel in the MD simulations highlights limitations of current theoretical models in capturing such highly non-adiabatic processes. Future research should focus on developing improved theoretical approaches to accurately simulate the dynamics of this non-adiabatic excitation mechanism.

This study reveals a highly efficient mechanism for promoting electrons across the bandgap of a semiconductor using hydrogen atom collisions. The observed bimodal energy-loss distributions and the discrepancy between experiment and adiabatic MD simulations confirm the failure of the Born-Oppenheimer approximation under these conditions. These findings open new avenues for exploring and manipulating electron dynamics in semiconductors. Future studies should investigate the detailed microscopic mechanisms of this VB-CB excitation process and explore potential applications of this phenomenon.

The study focused on a specific semiconductor surface (Ge(111)c(2 × 8)). The generalizability of the findings to other semiconductor materials and surface reconstructions remains to be determined. The experiments primarily focused on in-plane scattering; out-of-plane scattering contributions might influence the quantitative interpretation of branching ratios and sticking probabilities. The high-dimensional NN-PES used for simulations, although advanced, might still not fully capture all aspects of the complex atom-surface interactions. Further investigation is needed to explore the influence of surface temperature on the observed phenomena.