Resonant states and nuclear dynamics in solid-state systems: the case of silicon-hydrogen bond dissociation

TL;DR

The paper tackles carrier-induced bond dissociation in solids, focusing on Si–H as a model system. It develops a non-adiabatic framework that extracts localized bonding/antibonding states from first-principles data using MLWFs and PAOs, builds potential-energy curves, and propagates a quantum proton on excited surfaces via the MGR model, then computes dissociation probabilities and quantum yields averaged over lifetimes. The results explain subthreshold desorption, threshold saturation, and the giant isotope effect observed in STM experiments, and extend to oxide-stress trap-generation data, showing that a single-electron injection into antibonding states suffices to describe hot-carrier degradation without invoking multi-electron mechanisms. The approach provides a first-principles, transferable description of strong electron–nuclear coupling in bond-breaking processes with implications for radiation damage, device reliability, and beyond-Si–H chemistry in solids.

Abstract

Bond breaking in the presence of highly energetic carriers is central to many important phenomena in physics and chemistry, including radiation damage, hot-carrier degradation, activation of dopant-hydrogen complexes in semiconductors, and photocatalysis. Describing these processes from first principles has remained an elusive goal. Here we introduce a comprehensive theoretical framework for the dissociation process, emphasizing the need for a non-adiabatic approach. We benchmark the results for the case of silicon-hydrogen bond dissocation, a primary process for hot-carrier degradation. Passivation of Si dangling bonds by hydrogen is vital in all Si devices because it eliminates electrically active mid-gap states; understanding the mechanism for dissociation of these bonds is therefore crucial for device technology. While the need for a non-adiabatic approach has been previously recognized, explicitly obtaining diabatic states for solid-state systems has been an outstanding challenge. We demonstrate how to obtain these states by applying a partitioning scheme to the Hamiltonian obtained from first-principles density functional theory. Our results demonstrate that bond dissociation can occur when electrons temporarily occupy the antibonding states, generating a highly repulsive excited-state potential that causes the hydrogen nuclear wavepacket to shift and propagate rapidly. Based on the Menzel-Gomer-Redhead (MGR) model, we show that after moving on this excited-state potential on femtosecond timescales, a portion of the nuclear wavepacket can continue to propagate even after the system relaxes back to the ground state, allowing us to determine the dissociation probability. Our results provide essential insights into the fundamental processes that drive carrier-induced bond breaking in general, and specifically elucidate hydrogen-related degradation in Si devices.

Figures (15)

• Figure 1: Schematic representation of (a) an intact Si-H bond and the corresponding band structure with no states in the band gap but resonant states in the valence band (VB) and conduction band (CB) corresponding to bonding $(\sigma)$ and antibonding $(\sigma^*)$ states; and (b) after dissociation, exposing an electronically active Si dangling bond and introducing a mid-gap state in the band structure. Isosurfaces of the Si-H antibonding state and the Si dangling bond are shown.
• Figure 2: Schematic representation of the proposed mechanism for Si-H bond dissociation. Historically, two primary mechanisms have been discussed: (a) electron excitation from the Si-H bonding state to the antibonding state, generating a repulsive potential that leads to dissociation, and (b) inelastic resonant scattering by electrons, resulting in vibrational excitations that accumulate and ultimately cause dissociation. (c) The electron injection mechanism presented in this work: injection into the Si-H antibonding state produces a repulsive potential and drives the dissociation process. Mechanisms (b) and (c) could likewise occur through the interaction of holes with bonding states.
• Figure 3: Schematic representation of the electronic Hamiltonian with its basis set and transformation process. The blue shaded region indicates the diagonal elements of the Hamiltonian, corresponding to the energy of the respective basis states. The areas shaded in grey represent the off-diagonal terms that can have non-zero electronic coupling values. The conversion from the tight-binding Hamiltonian to the Newns-Anderson Hamiltonian leaves the local state itself unchanged $(t_{a,a}=\epsilon_a)$, while modifying the subspace associated with the bulk states and their coupling to the local state.
• Figure 4: Schematic representation of the wavepacket propagation described within the MGR approach. The hydrogen nuclear wavefunction starts in the ground-state potential and undergoes excitation, transitioning to the excited-state potential. It propagates in this excited state for a period of time before eventually decaying back to the ground state.
• Figure 5: Wavefunctions of the bonding state (left) and antibonding state (right) of Si-H bonds obtained from the partitioning process.