Reverse Designing Ferroelectric Capacitors with Machine Learning-based Compact Modeling
By: Diego Ferrer , Jack Hutchins , Revanth Koduru and more
Potential Business Impact:
Designs computer chips faster by guessing what's needed.
Machine learning-based compact models provide a rapid and efficient approach for estimating device behavior across multiple input parameter variations. In this study, we introduce two reverse-design algorithms that utilize these compact models to identify device parameters corresponding to desired electrical characteristics. The algorithms effectively determine parameter sets, such as layer thicknesses, required to achieve specific device performance criteria. Significantly, the proposed methods are uniquely enabled by machine learning-based compact modeling; alternative computationally intensive approaches, such as phase-field modeling, would impose impractical time constraints for iterative design processes. Our comparative analysis demonstrates a substantial reduction in computation time when employing machine learning-based compact models compared to traditional phase-field methods, underscoring a clear and substantial efficiency advantage. Additionally, the accuracy and computational efficiency of both reverse-design algorithms are evaluated and compared, highlighting the practical advantages of machine learning-based compact modeling approaches.
Similar Papers
Defect-Aware Physics-Based Compact Model for Ferroelectric nvCap: From TCAD Calibration to Circuit Co-Design
Emerging Technologies
Makes computer memory faster and use less power.
Fixed Point Neural Acceleration and Inverse Surrogate Model for Battery Parameter Identification
Machine Learning (CS)
Checks electric car batteries much faster and better.
Machine learning interatomic potential can infer electrical response
Materials Science
Predicts how materials react to electricity.