AI4Plasma II: Geometry-Aware Surrogate Modeling for Variable Electrode Gap GEC Plasmas
Organizations: Theory Lab, Huawei Leibniz Research Center, Shatin, Hong Kong SAR, People’s Republic of China · School of Aeronautic Science and Engineering, Beihang University, Beijing 100191, China · Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, Shatin, N.T. 999077, Hong Kong SAR, China · School of Mathematical Sciences, Peking University, Beijing 100871, China ·
Abstract
Industrial plasma process development requires broad exploration of operating and geometric parameters. Geometric parameters, particularly electrode spacing, affect discharge structure and processing uniformity. Exploring these parameter spaces through repeated high-fidelity simulations can be computationally expensive. Surrogate models can offer rapid predictions, but geometric variation complicates their application because the computational domain and discretization may change across samples. In this work, six geometry-aware surrogate models based on prescribed and learned geometry mappings, extended-domain representation, continuous coordinates, and geometry-native discretizations are compared in terms of their data requirements for accurate predictions of full plasma fields and ion density profiles near the wafer. The simulation dataset spans pressure, voltage, frequency, and electrode gap, with each reference case requiring an average simulation cost of 5.32 CPU h. To achieve a 2% mean error across the four plasma fields, prescribed-mapping Geo-FNO and Transolver require 100 training cases, compared with 250-1000 for the other models. Predicting ion density profiles near the wafer is more demanding and reveals larger differences among models, with prescribed-mapping Geo-FNO requiring the fewest training cases for the profile accuracy targets considered. To further reduce the data required for accurate predictions near the wafer, we introduce charge-separation weighting into the training loss, which reduces the prescribed-mapping Geo-FNO's minimum tested training requirement for a 5% profile error from 1000 to 100 cases. These results support prescribed-mapping Geo-FNO as a practical starting point for the variable-gap problem studied here when a suitable geometry mapping is available, while Transolver achieves the highest prediction accuracy with sufficient training data.
Figures & tables
| Parameter | Symbol | Interval | Unit |
|---|---|---|---|
| Gas pressure | [0.10, 1.00] | ||
| RF voltage amplitude | [100, 300] | ||
| Driving frequency | [13.56, 67.8] | ||
| Electrode separation | [2.0, 6.0] |
| Model | (%) | (%) | (%) | (%) | Mean (%) |
|---|---|---|---|---|---|
| Geometry mapping | |||||
| Geo-FNO-A | |||||
| Geo-FNO-L | |||||
| Extended-domain representation | |||||
| eDAFNO | |||||
| Continuous-coordinate representation | |||||
| Model | |||
|---|---|---|---|
| Geo-FNO-A | 50 | 100 | 1000 |
| Geo-FNO-L | 250 | 250 | NR |
| eDAFNO | 100 | 250 | NR |
| C-MLP | 500 | 1000 | NR |
| PCNO | 100 | 250 | 1000 |
| Transolver | 50 | 100 | 1000 |
| Axial midplane | Wafer-adjacent | |||||
|---|---|---|---|---|---|---|
| Model | (%) | (%) | (pp) | (%) | (%) | (pp) |
| Geo-FNO-A | ||||||
| Geo-FNO-L | ||||||
| eDAFNO | ||||||
| C-MLP | ||||||
| PCNO | ||||||
| Model | ||
|---|---|---|
| Geo-FNO-A | 100 | 1000 |
| Geo-FNO-L | 1000 | NR |
| eDAFNO | NR | NR |
| C-MLP | NR | NR |
| PCNO | NR | NR |
| Transolver | 500 | 1000 |
| Model | Four-field error (%) | Wafer-adjacent profile error (%) | Four-field error (%) | Wafer-adjacent profile error (%) |
|---|---|---|---|---|
| Geo-FNO-A | ||||
| C-MLP | ||||
| PCNO | ||||
| Transolver | ||||
| Accuracy target | Recommended model | Supporting evidence |
|---|---|---|
| Four-field mean error | Geo-FNO-A | Reaches the target with , as does Transolver, but yields a lower mean error ( versus ). |
| Four-field mean error | Transolver | Reaches the target with , as do Geo-FNO-A and PCNO, and yields the lowest mean error ( ). |
| Wafer-adjacent profile error | Geo-FNO-A | Reaches the target with , compared with for Transolver. |
| Wafer-adjacent profile error | Transolver | Reaches the target with , as does Geo-FNO-A, but yields a lower profile error ( versus ). |
Appendix figures & tables7 assets
Supplementary material from the paper’s appendix.
Appendix
| Model | Architectural configuration |
|---|---|
| Geometry mapping | |
| Geo-FNO-A | Residual FNO; 4 Fourier blocks; feature width 36; 16 axial and 8 radial Fourier modes; padding ; group normalization; GELU activation; trainable parameters: 1.334 M. |
| Geo-FNO-L | Geo-FNO with learned IPHI mapping; latent grid ; 2 geometric and 3 standard Fourier layers; feature width 32; 16 axial and 8 radial Fourier modes; GELU activation; trainable parameters: 1.349 M. |
| Extended-domain representation | |
| eDAFNO | Uniform grid: ; lifting input ; incorporated into every eDAFNO block; ; 4 explicit eDAFNO blocks; latent width 32; 16 axial and 8 radial Fourier modes; padding ; GELU activation; trainable parameters: 1.054 M. |
| Coordinate-based continuous representation | |
| Model | Four-field error (%) | Wafer-adjacent error (%) | ||
|---|---|---|---|---|
| Original Prescribed mapping | Change | Original Prescribed mapping | Change | |
| eDAFNO | ||||
| C-MLP | ||||
| PCNO | ||||
| Transolver | ||||
| Model | Four-field error (%) | Wafer-adjacent error (%) | ||
|---|---|---|---|---|
| Original Without gap input | Change | Original Without gap input | Change | |
| Geo-FNO-A | ||||
| Geo-FNO-L | ||||
| eDAFNO | ||||
| C-MLP | ||||
| PCNO | ||||
| Model | Four-field error (%) | Wafer-adjacent error (%) | ||
|---|---|---|---|---|
| Original Reduced resolution | Change | Original Reduced resolution | Change | |
| Geo-FNO-A | ||||
| Geo-FNO-L | ||||
| eDAFNO | ||||
| C-MLP | ||||
| PCNO | ||||
Surrogate Modeling
Surrogate modeling builds faster or more interpretable approximations of expensive simulations and black-box models.
See all 253 papers