Thermodynamic Multi-Field Coupling Optimization of Microsystem Based on Artificial Intelligence.

Area of Science:

• Microsystems Engineering
• Computational Mechanics
• Optimization Algorithms

Background:

• Microsystem design involves complex trade-offs between thermal, mechanical, and structural parameters.
• Optimizing these parameters is crucial for performance, reliability, and miniaturization.
• Existing methods can be computationally intensive, limiting design exploration.

Purpose of the Study:

• To develop an efficient multi-objective optimization method for microsystem design.
• To map the relationship between through-silicon via (TSV) structural parameters and performance objectives.
• To optimize temperature, stress, and thermal expansion deformation effectively.

Main Methods:

• Utilized particle swarm optimization (PSO) for multi-objective optimization.
• Employed finite element method (FEM) to establish design-performance relationships.
• Integrated neural networks for mapping design and performance parameters.
• Iteratively optimized TSV design parameters using PSO and validated with FEM.

Main Results:

• Achieved optimized values for peak temperature (97.90 °C), bump temperature (56.01 °C), TSV temperature (31.52 °C), maximum stress (247.4 MPa), and maximum thermal deformation (11.14 μm).
• Determined optimal TSV design parameters: radius (10.28 μm), pitch (65 μm), and SiO2 thickness (0.83 μm).
• Reduced single simulation time from 2 hours to 70 seconds, demonstrating significant efficiency gains.

Conclusions:

• The proposed PSO-based optimization method is efficient and reliable for microsystem design.
• The method significantly reduces computational time and resource requirements.
• This approach offers substantial improvements for the optimization design of microsystems, aiding future development.