Plasma-physics-informed explainable artificial intelligence for geometry-dependent etch-profile response in fluorocarbon-based high-aspect-ratio plasma etching

Robust control of high-aspect-ratio (HAR) plasma etching requires understanding of how plasma conditions shape the etch profile, which is set by ion and neutral transport into the feature and their reactions at its surfaces. These processes are difficult to resolve individually, so data-driven approaches are widely used; however, they typically relate the plasma state to average outcomes or selected profile metrics and do not indicate how strongly a geometric deviation in one feature region affects the downstream response. Here, we develop a plasma-physics-informed profile-analysis framework that uses explainable artificial intelligence to extract this geometry sensitivity from etch profiles. The profile is represented as a depth-ordered sequence of critical-dimension values, and a transformer-based self-attention model reconstructs it. Integrated gradients quantifies the contribution of a geometric deviation in one feature region to the response of another, interpreted with a Langmuir-type surface-reaction description and optical emission spectroscopy plasma indicators. The reconstruction model was developed using etch-profile data extracted from cross-sectional scanning electron microscopy images under the 20 gas-variation conditions for oxide-nitride-oxide trenches etched in a dual-frequency capacitively coupled plasma. The extracted sidewall-to-etch-front sensitivity was enhanced at an intermediate etching-to-passivation balance that coincided with a change in the etch-depth trend and the onset of polymer accumulation, and the measured feature-to-feature etch-depth variation showed similar condition dependence. These trends were qualitatively interpreted using the Langmuir-type surface-reaction framework. The resulting sensitivity provides a criterion beyond average-profile metrics for evaluating the process window for HAR etch control.

Authors

Institutions

Publication Details

Journal
Journal of Vacuum Science & Technology A Vacuum Surfaces and Films
Published
2026-11-19
DOI
https://doi.org/10.1116/6.0005860
Primary Topic
Low-Pressure Plasma Processing and Diagnostics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Plasma-physics-informed explainable artificial intelligence for geometry-dependent etch-profile response in fluorocarbon-based high-aspect-ratio plasma etching

Ji-Won Kwon, Seolhye Park, Taejun Park, Jihoon Park et al.
Journal of Vacuum Science & Technology A Vacuum Surfaces and Films
Low-Pressure Plasma Processing and Diagnostics
article

Plasma-physics-informed explainable artificial intelligence for geometry-dependent etch-profile response in fluorocarbon-based high-aspect-ratio plasma etching

Ji-Won Kwon, Seolhye Park, Taejun Park, Jihoon Park, Gon-Ho Kim
article en

Abstract

Robust control of high-aspect-ratio (HAR) plasma etching requires understanding of how plasma conditions shape the etch profile, which is set by ion and neutral transport into the feature and their reactions at its surfaces. These processes are difficult to resolve individually, so data-driven approaches are widely used; however, they typically relate the plasma state to average outcomes or selected profile metrics and do not indicate how strongly a geometric deviation in one feature region affects the downstream response. Here, we develop a plasma-physics-informed profile-analysis framework that uses explainable artificial intelligence to extract this geometry sensitivity from etch profiles. The profile is represented as a depth-ordered sequence of critical-dimension values, and a transformer-based self-attention model reconstructs it. Integrated gradients quantifies the contribution of a geometric deviation in one feature region to the response of another, interpreted with a Langmuir-type surface-reaction description and optical emission spectroscopy plasma indicators. The reconstruction model was developed using etch-profile data extracted from cross-sectional scanning electron microscopy images under the 20 gas-variation conditions for oxide-nitride-oxide trenches etched in a dual-frequency capacitively coupled plasma. The extracted sidewall-to-etch-front sensitivity was enhanced at an intermediate etching-to-passivation balance that coincided with a change in the etch-depth trend and the onset of polymer accumulation, and the measured feature-to-feature etch-depth variation showed similar condition dependence. These trends were qualitatively interpreted using the Langmuir-type surface-reaction framework. The resulting sensitivity provides a criterion beyond average-profile metrics for evaluating the process window for HAR etch control.

Journal of Vacuum Science & Technology A Vacuum Surfaces and FilmsVol. 45(1)
Seoul National University (KR)
Low-Pressure Plasma Processing and Diagnostics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon