Triple Patterning Lithography Mask Assignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current double patterning lithography techniques face challenges in decomposing complex two-dimensional patterns and implementing stitch insertion at the full-chip level, leading to costly layout modifications and inefficiencies in feature density balancing.
Innovation Solution
A method for triple patterning lithography that uses multiple steps of double-patterning decomposition to address different types of spacing violations, identifying candidate stitch locations and employing three-way coloring to minimize conflicts and stitches, while maximizing exposure overlap and density balancing across masks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional double patterning lithography is used to decompose two-dimensional patterns, then feature density can be enhanced, but layout decomposition becomes very complex and stitch insertion locations may be missed leading to costly layout modifications
Solution Approach 1:
The patent segments the complex two-dimensional pattern decomposition problem into multiple one-dimensional decomposition steps. By performing decomposition along different directions (e.g., horizontal then vertical, or vertical then horizontal) in separate passes, the algorithm reduces the overall complexity while systematically identifying all candidate stitch locations without missing any potential solutions.
Solution Approach 2:
The patent transforms the two-dimensional pattern decomposition challenge into a series of one-dimensional problems by projecting patterns onto different axes. This dimensional reduction allows the use of simpler 1D decomposition algorithms while still achieving the desired 2D pattern separation, and enables comprehensive stitch location identification by examining conflicts from multiple directional perspectives.
2Reliability
If stitch insertion is considered during layout decomposition, then decomposition conflicts can be removed, but polygons are broken into multiple pieces requiring full-chip level implementation
Solution Approach 1:
The patent performs preliminary identification of all candidate stitch locations during the decomposition process itself, before final mask assignment. By systematically examining all possible stitch positions across different decomposition directions and storing them for later evaluation, the method prepares comprehensive conflict resolution options upfront, avoiding the need for complex post-processing or full-chip re-implementation.
Solution Approach 2:
The patent incorporates feedback mechanisms where decomposition conflicts are identified, candidate stitches are generated, and their effectiveness is evaluated iteratively. The system uses feedback from conflict analysis to refine stitch location selections and adjust decomposition strategies, enabling conflict resolution without requiring complete full-chip re-decomposition while maintaining high reliability.
3Adaptability or versatility
If multiple decomposition directions are examined to identify all candidate stitch locations, then stitch capability is fully utilized, but computational time and processing resources increase
Solution Approach 1:
The patent segments the exhaustive multi-directional decomposition search into independent passes along different axes. Each pass processes one dimension completely and identifies its candidate stitches, then the results are combined. This segmentation allows parallel processing of different directional analyses and avoids the inefficiency of truly exhaustive sequential search, reducing overall computational time while maintaining comprehensive stitch location identification.
Solution Approach 2:
The patent implements a practical balance by examining a sufficient number of decomposition directions (enough to capture all meaningful stitch opportunities) without performing truly exhaustive analysis of every possible angle or configuration. By identifying the key directional passes that yield the majority of valuable stitch locations, the method achieves near-complete stitch capability utilization with acceptable computational overhead.
Data Source
AI summary
A mechanism is provided for mask assignment for triple patterning lithography. The mechanism identifies tip-to-tip (TT), tip-to-side (TS), and side-to-side (SS) conflicting parts by design rule dependent projection. The mechanism finds stitch location for TT, TS, and SS conflicts separately. The mechanism colors TT, TS, and SS conflicting parts with mask0/mask1, mask0/mask2, mask1/mask2 coloring cycle with each type colored separately. The mechanism uses existing infrastructure of two-way coloring. As a first objective, the mechanism attempts to minimize conflicts. As a second objective, the mechanism attempts to minimize the number of stitches by assigning the two sides of stitches to the same mask. Once coloring of all conflicting parts is done, the mechanism colors non-conflicting parts to maximize minimum overlap of exposures and to use both colors if two sides are different colors and one color if both sides are the same color.