Substrate Deformation Computation in Thermo-Compression Bonding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for mechanical analysis of thermo-compression bonding are time-consuming due to complex finite element computations, limiting the number of temperature snapshots and hindering rapid redesign of substrate structural layouts.
Innovation Solution
The implementation of Global Reduced Order Modeling, Zonal Reduced Order Modeling, and Layer-by-Layer finite element modeling methodologies to efficiently compute substrate deformations by reducing computational effort and enabling parallelization, using Proper Orthogonal Decomposition to determine temperature and displacement factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full finite element computations are used for mechanical analysis of thermo-compression bonding, then prediction accuracy is improved, but computational time increases significantly
Solution Approach 1:
The substrate structure is segmented into multiple layers, and the modeling approach is divided into global reduced order modeling, zonal reduced order modeling, and layer-by-layer finite element modeling. This segmentation allows computational resources to be focused on critical regions while using simplified models for other areas, thereby reducing overall computational time while maintaining prediction accuracy.
Solution Approach 2:
Reduced order models serve as intermediaries between full finite element computations and rapid predictions. These reduced order models are pre-computed from full FEM solutions and then used to rapidly predict substrate deformations for different temperature distributions, acting as a computational bridge that maintains accuracy while dramatically reducing computation time.
2Reliability
If full finite element computations are used for mechanical analysis, then model fidelity is improved, but memory requirements increase
Solution Approach 1:
The substrate is divided into multiple layers, allowing the modeling to be performed in a layer-by-layer manner. This segmentation reduces the memory footprint by processing smaller subsystems independently rather than loading the entire complex model into memory simultaneously, while still maintaining overall model fidelity through proper interface coupling.
Solution Approach 2:
The patent extracts and pre-computes reduced order models from full finite element solutions. These extracted reduced order models contain the essential mechanical behavior information but in a compact form that requires significantly less memory, while still providing accurate predictions when applied to different temperature distributions.
3Loss of information
If complex finite element computations are performed for each temperature snapshot, then analysis completeness is improved, but productivity decreases
Solution Approach 1:
Reduced order models are pre-computed from full finite element solutions before the actual analysis of multiple temperature snapshots. This preliminary action captures the essential mechanical behavior in advance, allowing subsequent predictions for different temperature distributions to be made rapidly without repeating full FEM computations, thus maintaining analysis completeness while dramatically improving productivity.
Solution Approach 2:
The patent changes the computational parameters by transitioning from full-order models to reduced-order models. This parameter change involves reducing the number of degrees of freedom in the computational model while preserving the essential physics, enabling rapid computation across multiple temperature snapshots without sacrificing the completeness of the thermal-mechanical analysis.
Data Source
AI summary
A system can apply a Global Reduced Order Modeling (ROM), a Zonal ROM, or a Layer-By-Layer analysis method to accelerate computations of a static displacement field of a substrate during modeling of thermo-compression bonding. The system fully takes into account a glass transition of dielectric substrates as well as nonhomogeneity of the various layers.