Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

iPREFER: An Intelligent Parameter Extractor based on Features for BSIM-CMG Models

Zhiliang Peng, Yicheng Wang, Zhengwu Yuan, Xingsheng Wang

TL;DR

This work addresses the slow, data-hungry transition from TCAD device simulations to BSIM-CMG compact models within the DTCO workflow. It introduces iPREFER, which combines IV/CV curve feature extraction with a four-layer neural network to predict BSIM-CMG parameters from curve features, trained on large Monte Carlo SPICE datasets and validated against 5 nm nanosheet TCAD curves. The approach delivers high-precision parameter extraction, robustness to ±10% variations in LG and EOT, and outperforms prior AI-based methods while remaining applicable to other compact models. By accelerating the TCAD-to-compact-model transfer, iPREFER can significantly speed up process optimization and circuit performance prediction in advanced semiconductor technologies.

Abstract

This paper introduces an innovative parameter extraction method for BSIM-CMG compact models, seamlessly integrating curve feature extraction and machine learning techniques. This method offers a promising solution for bridging the division between TCAD and compact model, significantly contributing to the Design Technology Co-Optimization (DTCO) process. The key innovation lies in the development of an automated IV and CV curve feature extractor, which not only streamlines the analysis of device IV and CV curves but also enhances the consistency and efficiency of data processing. Validation on 5-nm nanosheet devices underscores the extractor's remarkable precision, with impressively low fitting errors of 0.42% for CV curves and 1.28% for IV curves. Furthermore, its adaptability to parameter variations, including those in Equivalent Oxide Thickness and Gate Length, solidifies its potential to revolutionize the TCAD-to-compact model transition. This universal BSIM-CMG model parameter extractor promises to improve the DTCO process, offering efficient process optimization and accurate simulations for semiconductor device performance prediction.

iPREFER: An Intelligent Parameter Extractor based on Features for BSIM-CMG Models

TL;DR

This work addresses the slow, data-hungry transition from TCAD device simulations to BSIM-CMG compact models within the DTCO workflow. It introduces iPREFER, which combines IV/CV curve feature extraction with a four-layer neural network to predict BSIM-CMG parameters from curve features, trained on large Monte Carlo SPICE datasets and validated against 5 nm nanosheet TCAD curves. The approach delivers high-precision parameter extraction, robustness to ±10% variations in LG and EOT, and outperforms prior AI-based methods while remaining applicable to other compact models. By accelerating the TCAD-to-compact-model transfer, iPREFER can significantly speed up process optimization and circuit performance prediction in advanced semiconductor technologies.

Abstract

This paper introduces an innovative parameter extraction method for BSIM-CMG compact models, seamlessly integrating curve feature extraction and machine learning techniques. This method offers a promising solution for bridging the division between TCAD and compact model, significantly contributing to the Design Technology Co-Optimization (DTCO) process. The key innovation lies in the development of an automated IV and CV curve feature extractor, which not only streamlines the analysis of device IV and CV curves but also enhances the consistency and efficiency of data processing. Validation on 5-nm nanosheet devices underscores the extractor's remarkable precision, with impressively low fitting errors of 0.42% for CV curves and 1.28% for IV curves. Furthermore, its adaptability to parameter variations, including those in Equivalent Oxide Thickness and Gate Length, solidifies its potential to revolutionize the TCAD-to-compact model transition. This universal BSIM-CMG model parameter extractor promises to improve the DTCO process, offering efficient process optimization and accurate simulations for semiconductor device performance prediction.
Paper Structure (15 sections, 8 figures, 3 tables)

This paper contains 15 sections, 8 figures, 3 tables.

Table of Contents

  1. Introduction
  2. PRELIMINARIES
  3. 5nm Nanosheet TCAD Device
  4. Overall Architecture of iPREFER
  5. iPREFER: Details and Structure
  6. Feature Extraction
  7. Parameter Extraction Network
  8. Testing Results by TCAD Data
  9. Testing Results of Baseline TCAD Data
  10. Testing Results of Variable TCAD Data
  11. Advancements of iPREFER
  12. Exceptional Parameter Extractor Performance
  13. Higher Accuracy and Generalization Capability
  14. Versatile Frameworks for Parameter Extraction
  15. Conclusion

Figures (8)

  • Figure 1: Simplified DTCO Workflow
  • Figure 2: Main Workflow Diagram of iPREFER
  • Figure 3: 5nm Nanosheet's 3D Structural Diagram and Cross-sectional view.
  • Figure 4: Diagram of iPREFER's Structure and Details: (a) Specific Structure of the Feature Extractor, (b) Structure Details of the Parameter Extraction Neural Network.
  • Figure 5: Model Testing Results Extracted by iPREFER from Baseline TCAD Data: (a) Cgg-Vgs, (b) Ids-Vgs at Vds=0.05V and 0.7V, (c) Gm-Vgs and Gm'-Vgs at Vds of 0.05V and 0.7V, and (d) Ids-Vgs at Vgs of 0.5V, 0.6V, and 0.7V.
  • ...and 3 more figures