Machine Learning Assisted Optimization of Nanoscale MOSFET Parameters Using TCAD Simulation

Notice

This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.

Year : 2026 | Volume : 13 | 01 | Page :
    By

    Amit Pandhare,

  • Aditya Kumbhar,

  • Vaibhav Godase,

  1. Student, Department of Electronics and Telecommunication Engineering, SKN Sinhgad College of Engineering, Pandharpur, Maharashtra, India
  2. student, Department of Electronics and Telecommunication Engineering, SKN Sinhgad College of Engineering, Pandharpur, Maharashtra, India
  3. Assistant Professor, Department of Electronics and Telecommunication Engineering, SKN Sinhgad College of Engineering, Pandharpur, Maharashtra, India

Abstract

This paper presents a machine learning (ML) assisted framework for the multi-objective optimization of nanoscale bulk n-channel metal-oxide-semiconductor field-effect transistors (nMOSFETs) with a 10 nm physical gate length, high-k HfO₂ gate dielectric, and TiN metal gate. Technology computer-aided design (TCAD) simulations employing drift-diffusion transport, Shockley-Read-Hall recombination, Lombardi mobility degradation, and density- gradient quantum correction models are used to generate a parametric dataset of 2,400 device configurations spanning gate length (L₝), equivalent oxide thickness (EOT), channel doping (Nₐʰ), and source/drain peak doping (Nₛᵈ). Key s of merit—threshold voltage (Vₜʰ), subthreshold swing (SS), drain-induced barrier lowering (DIBL), and I₀ₙ/I₀ᵄᵄ ratio—are extracted from simulated Iᵈ-Vᵍ and Iᵈ-Vᵈ characteristics. An artificial neural network (ANN) and a random forest (RF) regressor are trained on this dataset. The ANN achieves a coefficient of determination R² of 0.9931 for Vₜʰ prediction and 0.9887 for SS, with root-mean-square errors of 4.2 mV and 1.8 mV/decade, respectively. The RF model attains comparable accuracy with superior interpretability via feature importance scores, identifying EOT and Nₐʰ as the dominant parameters governing electrostatic integrity. Bayesian-guided surrogate optimization using the trained ANN converges to a Pareto-optimal design with Vₜʰ = 0.38 V, SS = 71 mV/dec, DIBL = 42 mV/V, and I₀ₙ/I₀ᵄᵄ = 1.6 × 10⁷ in under 90 seconds—a 340× speedup relative to exhaustive TCAD sweeps. The methodology is scalable to FinFET and gate-all-around (GAA) architectures and provides a robust, physics-informed pathway for compact device co- optimization.

Keywords: MOSFET optimization, TCAD simulation, machine learning, artificial neural network, random forest, subthreshold swing, DIBL, high-k dielectric, and device parameter extraction.

How to cite this article:
Amit Pandhare, Aditya Kumbhar, Vaibhav Godase. Machine Learning Assisted Optimization of Nanoscale MOSFET Parameters Using TCAD Simulation. Journal of Microelectronics and Solid State Devices. 2026; 13(01):-.
How to cite this URL:
Amit Pandhare, Aditya Kumbhar, Vaibhav Godase. Machine Learning Assisted Optimization of Nanoscale MOSFET Parameters Using TCAD Simulation. Journal of Microelectronics and Solid State Devices. 2026; 13(01):-. Available from: https://journals.stmjournals.com/jomsd/article=2026/view=238972


References

  1.  Yeap G, Lin SS, Chen YM, Shang HL, Wang PW, Lin HC, Peng YC, Sheu JY, Wang M, Chen X, Yang BR. 5nm cmos production technology platform featuring full-fledged euv, and high mobility channel finfets with densest 0.021 µm 2 sram cells for mobile soc and high performance computing applications. In2019 IEEE International Electron Devices Meeting (IEDM) 2019 Dec 7 (pp. 36-7). IEEE.
  2. Ghoshhajra R, Biswas K, Sarkar A. A review on machine learning approaches for predicting the effect of device parameters on performance of nanoscale MOSFETs. 2021 Devices for Integrated Circuit (DevIC). 2021 May 19:489-93.
  3. Mistry K, Allen C, Auth C, Beattie B, Bergstrom D, Bost M, Brazier M, Buehler M, Cappellani A, Chau R, Choi CH. A 45nm logic technology with high-k+ metal gate transistors, strained silicon, 9 Cu interconnect layers, 193nm dry patterning, and 100% Pb-free packaging. In2007 IEEE international electron devices meeting 2007 Dec 10 (pp. 247-250). IEEE.
  4. Fan G, Low KL. Physics-integrated machine learning for efficient design and optimization of a nanoscale carbon nanotube field-effect transistor. ECS Journal of Solid State Science and Technology. 2023 Sep 1;12(9):091005.
  5. Ehteshamuddin M, Sheelvardhan K, Kumar A, Guglani S, Roy S, Dasgupta A. Machine learning-assisted multiobjective optimization of advanced node gate-all-around transistor for logic and RF applications. IEEE Transactions on Electron Devices. 2023 Dec 27;71(2):976-82.
  6. Nagel L, Pederson DO. SPICE (simulation program with integrated circuit emphasis).April. 1973.
  7. Forrester A, Sobester A, Keane A. Engineering design via surrogate modelling: a practical guide. John Wiley & Sons; 2008 Sep 15.
  8. Raissi M, Perdikaris P, Karniadakis GE. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational physics. 2019 Feb 1;378:686-707.
  9. Frank DJ, Dennard RH, Nowak E, Solomon PM, Taur Y, Wong HS. Device scaling limits of Si MOSFETs and their application dependencies. Proceedings of the IEEE. 2001 Mar 31;89(3):259-88.
  10. Van Dort MJ, Woerlee PH, Walker AJ. A simple model for quantisation effects in heavily- doped silicon MOSFETs at inversion conditions. Solid-State Electronics. 1994 Mar 1;37(3):411- 4.
  11. McKay MD, Beckman RJ, Conover WJ. A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics. 2000 Feb 1;42(1):55-61.
  12. Ioffe S, Szegedy C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. InInternational conference on machine learning 2015 Jun 1 (pp. 448- 456). pmlr.
  13. Kingma DP, Ba J. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980. 2014 Dec 22.
  14. Breiman L. Random Forests Mach. Learn., 45 (1), 5–32. ed. 2001.
  15. Shahriari B, Swersky K, Wang Z, Adams RP, De Freitas N. Taking the human out of the loop: A review of Bayesian optimization. Proceedings of the IEEE. 2015 Dec 10;104(1):148-75.
  16. Godase V, Pawar P, Nagane S, Kumbhar S. Automatic railway horn system using node MCU. Journal of Control & Instrumentation. 2024 Jan;15(1).
  17. Godase V, Godase J. Diet prediction and feature importance of gut microbiome using machine learning. Evolution in Electrical and Electronic Engineering. 2024 Nov 6;5(2):214-9.
  18. Godase V. A Comprehensive Review on Scalable Arduino Radar Platform for Real-time Object Detection and Mapping. Journal of Microprocessor and Microcontroller Research e- ISSN. 2025 May 16:3048-6637.
  19. Raut P, Panda DK, Goyal AK. A Comprehensive Review on Next-Generation Modelling and Optimization for Semiconductor Devices. IEEE Access. 2025 Jul 10.
  20. Lyu Y, Chen W, Zheng M, Yin B, Li J, Cai L. Machine learning-assisted device modeling with process variations for advanced technology. IEEE Journal of the Electron Devices Society. 2023 May 18;11:303-10.
  21. Mamun KM, Pala N, Shawkat MS. A comprehensive review of machine learning approaches for semiconductor device modeling and simulation. IEEE Access. 2025 Sep 3.
  22. Tsang CI, Pu H, Chen J. Multiscale simulation and machine learning facilitated design of two-dimensional nanomaterials-based tunnel field-effect transistors: A review. APL Machine Learning. 2025 Mar 1;3(1).
  23. Geng M, Guo J, Sun Y, Gao D, Ni D. Accurate and Efficient Process Modeling and Inverse Optimization for Trench Metal Oxide Semiconductor Field Effect Transistors: A Machine Learning Proxy Approach. Processes. 2025 May 16;13(5):1544.
  24. Dharmireddy A, Parri S, Swethna N. MOSFET Device Modeling using Machine Learning. JOURNAL OF TECHNICAL EDUCATION. 2025:1.
Ahead of Print Subscription Review Article
Volume 13
01
Received 25/02/2026
Accepted 27/02/2026
Published 20/03/2026
Publication Time 23 Days
78

Login


My IP

PlumX Metrics