Pre-Silicon Accurate SPICE Modeling of Trench MOSFETs via Advanced TCAD Simulations and Dynamic Validation

Area of Science:

• Electrical Engineering
• Semiconductor Device Physics
• Computational Electronics

Background:

• Traditional power MOSFET SPICE model extraction relies heavily on experimental silicon data, often requiring multiple costly prototyping iterations.
• Optimizing device performance and ensuring accurate electrical parameters before fabrication is a significant challenge in semiconductor design.
• Existing simulation methods may lack the fidelity needed for precise dynamic characteristic prediction without physical validation.

Purpose of the Study:

• To present a novel, fully virtual methodology for extracting accurate SPICE models of power MOSFETs.
• To enable designers to optimize device performance and extract electrical parameters exclusively from TCAD simulations.
• To bypass the necessity of initial silicon prototyping, thereby reducing development time and costs.

Main Methods:

• Utilized advanced TCAD (Technology Computer-Aided Design) tools to generate a realistic power MOSFET device structure.
• Obtained key electrical characteristics directly from TCAD simulations.
• Performed precise SPICE model extraction and macromodel integration based on simulated data.
• Dynamically validated the extracted SPICE model using a gate-charge test in both TCAD and SPICE environments.

Main Results:

• Achieved excellent agreement between TCAD and SPICE simulations for the gate-charge test, with less than 2% error for Q_gs and Q_gd.
• Demonstrated the capability to generate highly faithful device simulations prior to hardware fabrication.
• Successfully extracted accurate electrical parameters for power MOSFETs, including those for an e-fuse application.

Conclusions:

• The proposed virtual flow effectively bypasses the need for initial silicon prototyping in power MOSFET development.
• This innovative approach significantly accelerates the design process and reduces costs associated with prototyping and design re-spins.
• The methodology enhances the accuracy of transient and dynamic characteristics, crucial for MOSFETs in specific applications like e-fuses.