Atomic-Scale Mechanisms of SiO$_2$ Plasma-Enhanced Chemical Vapor Deposition Revealed by Molecular Dynamics with a Machine-Learning Interatomic Potential

Abstract

Plasma-enhanced chemical vapor deposition (PECVD) of silicon dioxide (SiO$_2$) is widely used for low-temperature fabrication of dielectric thin films, yet its atomic-scale growth mechanisms remain incompletely understood. In this work, we investigate SiO$_2$ PECVD using silane and N$_2$O as source gases via molecular dynamics simulations driven by a machine-learning interatomic potential. By systematically varying the oxidant-to-silane-derived species ratio $r$, we elucidate the evolution of film stoichiometry, density, and hydrogen content. Formation of the Si-O-Si network primarily proceeds via oxidation of surface Si-H groups to form Si-OH species, followed by condensation of neighboring Si-OH groups that produces H$_2$O as the dominant byproduct. At low $r$, H$_2$ formation via reactions between Si-H and Si-OH groups also contributes to the network formation. Increasing oxidant supply promotes the network formation through oxidation of residual Si-H species, suppressing hydrogen incorporation and leading to saturation of the Si/O ratio. Rapid chemisorption of silane-derived species, together with steric hindrance from pre-deposited species, results in localized growth and surface roughness. We further show that high-kinetic-energy plasma species can etch SiO$_2$ films, which potentially limits growth rates and enhances surface roughness under high RF-power conditions. These results provide atomic-scale insight into PECVD growth and guidance for optimizing film composition and quality.

Figures (10)

• Figure 1: PECVD simulation protocol consisting of amorphous bulk generation, surface equilibration, deposition, and annealing steps. One cycle of the SiO$_2$ PECVD simulation includes equilibration and deposition at 800 K, followed by annealing at 1700 K. For each impact, a uniform random number $u \in [0,1)$ is generated to stochastically select the incident species, thereby statistically realizing the target ratio $r = \mathrm{O}/\mathrm{SiH}_2\mathrm{O}$ over multiple incidences. After the $n^\mathrm{th}$ deposition step (the determination of $n$ is described in the main text), the slab undergoes annealing, after which the next cycle begins.
• Figure 2: Procedure of iterative fine-tuning. In Iteration 1, deposition simulations at $r = 0$ and $\infty$ are performed using 7net-0 to generate Fine-tuning set 1. In Iteration 2, deposition simulations at $r = 0$, 0.5, 2, and $\infty$ are carried out using the model fine-tuned on Fine-tuning set 1 to produce Fine-tuning set 2. In Iteration 3, post-deposition annealing simulations at the same $r$ values are conducted using the model fine-tuned on the combined Fine-tuning sets 1 and 2, yielding Fine-tuning set 3. The final fine-tuned model is obtained by training on the cumulative dataset comprising Fine-tuning sets 1--3. The sampled configurations include both slab snapshots and gas-phase species extracted from the MD trajectories. Fine-tuning is performed by retraining 7net-0 on the cumulative fine-tuning dataset at each iteration.
• Figure 3: Validation of FT-MLIP. Parity plots comparing DFT and MLIP total energies (left) and Cartesian force components $F_i$ ($i \in \{x, y, z\}$) (right) for deposition trajectories at (a) $r = 0$ and (b) $\infty$. (c) Comparison of DFT and MLIP reaction energies ($E_\mathrm{rxn}$) extracted from MD trajectories, including deposition and annealing stages at $r = 0$, 0.5, 2, and $\infty$.
• Figure 4: (a) Structural evolution during successive deposition cycles, showing progressive film thickening and the emergence of island-like growth (dotted circle). The supercell is expanded twice along the lateral direction for clarity. (b) Schematic illustration of how steric hindrance and prompt chemisorption of plasma species promote localized growth over time.
• Figure 5: Atomic structures before deposition (left) and after six deposition cycles (right) at $r = 6$. The black, green, and blue lines denote the boundaries of the substrate, deposited bulk region, and surface region, respectively.