CMP Simulation Hot Spot Prediction via Grid Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing chemical mechanical polishing (CMP) simulation methods often lead to inaccurate predictions of surface profiles on semiconductor wafers, resulting in issues like short circuits, graphic bridging, and deviations in resistance and capacitance values due to incomplete consideration of grid geometry characteristics and long-range correlations between neighboring grids.
Innovation Solution
A CMP simulation method that partitions the chip pattern layout into targeting grids, calculates grid geometry characteristics, generates shifted grids, computes weighted average grid geometry characteristics, and modifies grid geometry characteristics based on identified hot spots to optimize the simulation, thereby improving the accuracy of hot spot prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing CMP simulation methods are used to predict surface profile, then the prediction process can be completed, but the prediction accuracy is insufficient leading to inaccurate hot spot identification
Solution Approach 1:
The chip pattern layout is divided into multiple grids, and each grid is further divided into multiple sub-grids. This segmentation allows for more detailed and accurate calculation of geometric characteristics, enabling precise prediction of hot spots while maintaining computational feasibility.
Solution Approach 2:
The patent introduces a weighted average dimension by considering not only the grid itself but also its neighboring grids with different weights. This adds a spatial dimension to the calculation, capturing long-range correlations between neighboring grids that significantly improves hot spot prediction accuracy.
2Measurement precision
If grid geometry characteristics are calculated without considering neighboring grids, then the calculation process is simple, but the prediction accuracy deteriorates due to inability to capture long-range correlations
Solution Approach 1:
Different weights are assigned to different neighboring grids based on their spatial relationship and correlation with the target grid. This local quality approach allows the system to capture long-range correlations where needed while maintaining simplicity in calculations where long-range effects are minimal.
Solution Approach 2:
The patent pre-calculates and stores geometric characteristics for all grids and sub-grids before the main simulation process. This preliminary action reduces the computational burden during simulation by avoiding redundant calculations, thus managing complexity while maintaining high accuracy.
3Measurement precision
If multiple simulation iterations are performed to optimize grid geometry, then hot spot prediction accuracy improves, but the simulation time increases
Solution Approach 1:
The patent implements a feedback mechanism where the simulation results are used to optimize the grid geometry characteristics. By using the predicted hot spots to adjust the geometric parameters, the system converges to more accurate results faster, reducing the number of iterations needed and thus the total simulation time.
Solution Approach 2:
The patent dynamically adjusts geometric parameters such as grid size, weight factors, and neighborhood radius based on the simulation progress and results. This adaptive parameter change allows the system to optimize accuracy when needed while maintaining speed during initial iterations, balancing time and precision.
Data Source
AI summary
A CMP simulation method includes inputting a chip pattern layout including a plurality of graphic patterns, partitioning the chip pattern layout into targeting grids including a plurality of surrounding grids, and then calculating grid geometry characteristics of the targeting grids. The method also includes generating shifted grids by shifting the targeting grids, calculating weighted average grid geometry characteristics of the targeting grids and the shifted grids, and locating first hot spots on the chip pattern layout by performing a CMP simulation based on the grid geometry characteristics of the targeting grids. Further, the method includes generating optimized grid geometry characteristics by modifying the grid geometry characteristics of the targeting grids based on the weighted average grid geometry characteristics and the defined first hot spots, and then locating second hot spots on the chip pattern layout by performing the CMP simulation based on the optimized grid geometry characteristics.