A Physics-Chemistry-Informed Neural Network (PCINN) for Real-Time Spatial-ALD Coverage Prediction and Reliable Kinetics Inversion

2026-07-31

AI Summary

This work addresses the challenge in spatial atomic layer deposition (SALD) where conventional high-fidelity computational fluid dynamics (CFD) simulations are computationally prohibitive, while analytical models fail to capture transport phenomena such as gas curtains, leading to inaccurate and inefficient predictions of surface coverage and kinetic parameter inference. To overcome this, the authors propose a physics- and chemistry-informed neural network (PCINN) that hard-codes Arrhenius-based surface kinetics into a trainable chemical layer and employs a minimal network to map process conditions to near-wall precursor concentrations. The resulting invertible hybrid surrogate model features a single-scalar bottleneck and achieves high-fidelity coverage prediction (R²_log = 0.998) using only 30 training samples spanning four orders of magnitude in coverage, with an inference time of 7 ms per prediction—accelerating simulation by 5×10⁴ compared to CFD. Furthermore, the study introduces a comprehensive identifiability analysis based on Fisher information and profile likelihood to delineate parameter estimation boundaries and proposes a slope-deviation–based diagnostic for detecting unmodeled surface heterogeneity.

Abstract

Spatial atomic layer deposition (SALD) is a leading atmospheric-pressure, high-throughput route to industrial ALD, but design and control are limited by the cost of predicting surface coverage: high-fidelity CFD is far too slow for operating-window scans, while analytic models miss transport modulation such as the gas curtain. We present a physics-chemistry-informed neural network (PCINN), a hybrid surrogate with CFD-level accuracy at real-time speed: a query returns coverage in about 7 ms, roughly 5x10^4 times faster than a CFD solve, reaching a test R^2_log = 0.998 (leave-one-out R^2_raw = 0.974) from only 30 training cases spanning four orders of magnitude in coverage. The architecture is not a black box: a small network learns only the operating-condition to near-wall concentration closure, while the known surface kinetics is a hard-coded, trainable chemistry layer integrated along the substrate trajectory. This single-scalar bottleneck keeps it accurate under sparse data, interpretable and invertible. We add a full identifiability analysis (Fisher information, profile likelihood). The adsorption energy E_ads and desorption rate k_des are robustly identifiable; k_ads is not separately identifiable at a single temperature (only k_ads*c_wall is). Across four temperatures the prefactor nu and E_ads bind along a weakly identifiable degeneracy valley of slope 0.065 eV/decade, derived analytically as k_B T_eff ln(10) and turned into a reliability diagnostic: a seven-chemistry mismatch matrix shows it is invariant under any single-Arrhenius mismatch and shifts only when a second thermally activated process appears, so a slope departure flags unmodelled site heterogeneity. Data come from simulation with known ground truth inverted by the same kinetic form, so the study verifies pipeline self-consistency and the identifiability boundary, not real parameters.

Problem

• spatial atomic layer deposition
• surface coverage prediction
• kinetics inversion
• identifiability analysis
• real-time simulation

Innovation

• Physics-Chemistry-Informed Neural Network
• Spatial Atomic Layer Deposition
• Kinetic Parameter Identifiability
• Real-Time Surrogate Modeling
• Interpretable Inversion