Photomask Layout Decomposition for Optical Proximity Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing optical proximity correction (OPC) techniques for photomask design in semiconductor manufacturing often result in domain mismatches (DMMs) when neighboring chip domains converge to different solutions at their boundaries, leading to manufacturable issues and significant delays in the full-chip output.
Innovation Solution
The approach involves partitioning photomask design layouts into computational domains using a pathfinding algorithm with buffer regions and preferred boundary locations, avoiding features and regions on boundary edges, and employing a weighted graph for input into the A* algorithm to optimize domain boundaries and reduce DMMs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If photomask design layouts are partitioned into spatial sections for parallel processing, then computational efficiency is improved, but domain mismatches occur at boundaries when neighboring domains converge to different solutions
Solution Approach 1:
The photomask design layout is divided into multiple spatial sections or domains that can be processed independently in parallel. Each domain is assigned to a separate computing resource, enabling simultaneous OPC processing while maintaining the ability to handle complex full-chip designs through distributed computation
Solution Approach 2:
Buffer regions are introduced as intermediary zones between adjacent computational domains. These buffer regions contain features and data that are shared between neighboring domains, ensuring that when domains are processed independently and then reassembled, the boundary regions maintain consistency and avoid domain mismatches
2Manufacturing precision
If buffer regions are added to prevent domain mismatches, then manufacturing precision is improved, but device complexity increases
Solution Approach 1:
The buffer regions are systematically divided and assigned to adjacent domains in a structured manner. Each domain receives its designated buffer region as part of its computational boundary, allowing the complexity to be managed through organized segmentation rather than unstructured additions
Solution Approach 2:
Buffer regions are shared between adjacent domains, meaning the same physical region serves multiple purposes for neighboring computational units. This merging approach reduces overall complexity compared to creating separate buffer zones for each domain, as the buffer data is reused across domain boundaries
Data Source
AI summary
Methods for layout decomposition of photolithographic masks are provided. The decomposition creates domains that can be sent to independent computing resources for optimization. A first partition is created for the photolithographic mask design. Buffer regions are created around photolithographic features and a search distance is selected. The buffer regions and search distance are used in a pathfinding algorithm to determine new boundaries for new domains. The methods can be stored, for example, on at least one machine-readable storage medium as non-transitory instructions.