Mask Data Generation for Semiconductor Pattern Division
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for dividing patterns into multiple masks for semiconductor manufacturing are time-consuming, especially when dealing with two-dimensional layouts, and have not been effectively applied to one-dimensional layout patterns, such as hole or cut patterns, due to the need for extensive calculations and complex conflict graph analysis.
Innovation Solution
A mask data generation method that converts pattern data into map data, dividing regions into sections and setting unique mask information for each section based on constraint conditions, reducing unnecessary calculations and enabling faster pattern division by associating each pattern element with a grid, thereby minimizing the number of masks required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional pattern division methods using conflict graphs and mathematical programming are applied to one-dimensional layout patterns, then pattern division can be achieved, but calculation time becomes excessively long
Solution Approach 1:
The patent segments the calculation process by dividing pattern elements into groups based on their spatial relationships. Instead of calculating distances between all pairs of pattern elements, the method divides them into first group elements (requiring distance calculation) and second group elements (exempt from distance calculation), thereby reducing overall computation time while maintaining division accuracy
Solution Approach 2:
The patent performs preliminary classification of pattern elements into different groups before the main pattern division process. By pre-identifying which elements require distance calculation and which do not, the method prepares the data structure in advance to avoid unnecessary computations during the conflict graph generation and mathematical programming phases
2Reliability
If distance calculations are performed for all pattern elements in one-dimensional layouts, then accurate conflict graphs can be generated, but the computational complexity increases significantly
Solution Approach 1:
The patent applies local quality by treating different pattern elements differently based on their specific characteristics and positions. First group elements (which may cause resolution violations) undergo distance calculation, while second group elements (which do not affect resolution) are excluded from distance calculation, optimizing the conflict graph generation process
Data Source
AI summary
A mask data generation method includes obtaining data of a pattern including a plurality of pattern elements, dividing a region of the pattern into a plurality of sections so that each pattern element is arranged in each section by using the obtained data of the pattern and generating map data including information indicative of presence or absence of the pattern element in each section, setting one piece of mask individual information out of a plurality pieces of mask individual information for each section including the pattern element by using a constraint condition, which inhibits setting of same mask individual information in a constraint region including one section and surrounding sections thereof, and the map data, and generating the data of the plurality of masks corresponding to the plurality pieces of mask individual information by using the set mask individual information.