Mask Pattern Determination Using Topology Types
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As feature density on semiconductor wafers increases, existing techniques struggle to determine suitable mask patterns for photolithography that minimize distortions and artifacts, especially when dimensions approach light wavelength sizes, and similar challenges occur in mask-less semiconductor processes.
Innovation Solution
A method is developed to determine mask patterns by defining different regions on a target pattern with varying weights, convergence criteria, and mask rules, using pixilated images and models of photolithographic processes, including photo-resist and etch models, to optimize pattern representation and printing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional OPC and RET techniques are used to improve resolution and process window, then photolithography printing accuracy is improved, but determining suitable mask patterns becomes increasingly difficult as feature density increases
Solution Approach 1:
The pattern is divided into multiple regions with different topology types (e.g., dense regions, isolated regions, corner regions). Each region is assigned different weights and parameters in the merit function, allowing differential optimization. This segmentation enables the system to handle high feature density by treating different areas with appropriate customization, resolving the contradiction between maintaining printing accuracy and managing mask pattern determination complexity.
Solution Approach 2:
Different regions of the pattern are assigned different weights and optimization parameters based on their local characteristics. For example, dense regions may receive higher weights to ensure adequate spacing, while isolated regions may receive lower weights. This local quality approach allows the system to optimize each region according to its specific requirements, improving overall printing accuracy without uniformly increasing complexity across the entire mask pattern.
2Quantity of substance
If feature dimensions are decreased to increase density on dies and wafers, then circuit density is improved, but wafer patterns deviate from photo-mask patterns with unwanted distortions and artifacts
Solution Approach 1:
The system performs preliminary distortion analysis and pre-distorts the mask pattern before fabrication. By identifying regions that are prone to distortions (such as dense regions with tight spacing) and pre-adjusting their dimensions in the mask design, the system compensates for expected lithographic distortions. This preliminary action allows smaller feature dimensions to be used for increased density while maintaining pattern fidelity through proactive correction of anticipated deviations.
Solution Approach 2:
The system dynamically adjusts mask pattern parameters (such as line widths, spacing, and corner radii) based on local topology characteristics. In high-density regions, parameters are modified to account for diffraction and proximity effects. This parameter adaptation allows the system to maintain manufacturing precision across varying feature densities, enabling increased circuit density without sacrificing pattern fidelity.
3Measurement precision
If different regions are treated with different weights and parameters in the merit function, then mask pattern determination accuracy is improved, but calculation complexity increases
Solution Approach 1:
The pattern is segmented into discrete topology-based regions, each with assigned weights and parameters. This segmentation transforms the complex global optimization problem into a series of more manageable regional optimizations. By processing regions independently with their specific parameters and then combining results, the system achieves high determination accuracy while controlling calculation complexity through modular processing.
Solution Approach 2:
The system uses parameter changes strategically by assigning different weights only to regions where they are most needed (e.g., dense regions with high sensitivity to spacing). In less critical regions, standard parameters are used, reducing the overall computational burden. This selective parameter application maintains high accuracy in critical areas while limiting the increase in overall calculation complexity.
Data Source
AI summary
A method for determining a mask pattern is described. During the method, a first mask pattern that includes a plurality of second regions corresponding to the first regions of the photo-mask is provided. Then, a second mask pattern is determined based on the first mask pattern and differences between a target pattern and an estimate of a wafer pattern that results from the photolithographic process that uses at least a portion of the first mask pattern. Note that the determining includes different treatment for different types of regions in the target pattern, and the second mask pattern and the target pattern include pixilated images.