Pixel-Based DRC Neural Network for Curvilinear Layout Checking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional design rule checking (DRC) methods struggle with curvilinear designs, leading to false positives and performance issues due to reliance on one-dimensional measurements and integer-based data formats, which are inadequate for complex nano-scale geometries and manufacturing processes.
Innovation Solution
A pixel-based DRC method using a machine-trained neural network, such as a convolutional neural network, is employed to perform design rule checks for both rectilinear and curvilinear designs, trained through deep learning with data from traditional geometric, equation-based, or circle-tracing methods, and capable of rasterizing and converting back to the geometry domain for accurate and efficient rule checking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional one-dimensional measurement methods are used for DRC checking, then the method is simple to implement, but it produces false positives and cannot accurately handle curvilinear designs at nano-scale geometries
Solution Approach 1:
The patent transitions from traditional one-dimensional measurement methods to two-dimensional pixel-based analysis. By representing design geometries as pixel grids and applying convolutional kernels, the system captures spatial relationships and curvilinear features that 1D measurements cannot detect, thereby improving measurement precision for complex nano-scale designs.
Solution Approach 2:
The patent replaces traditional geometric measurement mechanisms with a machine learning-based pixel analysis system. Convolutional neural networks process the pixel representations of design geometries, substituting deterministic geometric calculations with probabilistic pattern recognition that better handles manufacturing variations and curvilinear shapes.
2Productivity
If curvilinear designs are fragmented into polygons for traditional DRC checking, then the design can be processed by existing tools, but it generates a huge number of false DRC errors and increases processing time
Solution Approach 1:
The patent creates a pixel-based copy or representation of the curvilinear design geometry rather than fragmenting it into polygons. This pixel grid representation preserves the continuous nature of curvilinear shapes while enabling efficient parallel processing, avoiding the false errors introduced by polygon approximation.
Solution Approach 2:
The patent transforms curvilinear geometry from vector/polygon representation into a pixel grid representation. This dimensional transformation allows the design to be processed as an image, enabling the use of convolutional operations that can evaluate curvilinear features directly without fragmentation, thereby improving both speed and accuracy.
3Reliability
If the number of design rules is increased to accommodate all possible polygon combinations, then design reliability is improved, but the complexity of keeping track of and debugging rules becomes extreme
Solution Approach 1:
The patent implements a universal pixel-based DRC checking system that can handle multiple design rules and geometry types through a single unified approach. The convolutional neural network framework processes various rule types (spacing, width, enclosure, etc.) using the same pixel analysis mechanism, eliminating the need for separate rule evaluation systems and reducing management complexity.
Data Source
AI summary
A method for performing pixel-based design rule checking (DRC) is described. This method is used to perform design rule checks for rectilinear and curvilinear designs. In some embodiments, the pixel-based approach is based on computational deep-learning. The pixel-based DRC method of some embodiments is more resilient to false positives than traditional geometric approaches, particularly for designs with curvilinear content, and the inference time remains constant, regardless of how many shapes exist in the design being checked, or how many polygon edges are needed to represent its curvature. The DRC method of some embodiments is implemented by highly parallel architectures (such as Graphics Processing Units (GPU) and Tensor Processing Units (TPU)) to improve processing throughput compared to traditional means.