Algorithms for DFM in electronic design automation

Guo, Daifeng

Permalink

https://hdl.handle.net/2142/104852 Copy

Description

Title

Algorithms for DFM in electronic design automation

Author(s)

Guo, Daifeng

Issue Date

2019-04-17

Director of Research (if dissertation) or Advisor (if thesis)

Wong, Martin D.F.

Doctoral Committee Chair(s)

Wong, Martin D.F.

Committee Member(s)

• Chen, Deming
• Vasudevan, Shobha
• Chiang, Charles

Department of Study

Electrical & Computer Eng

Discipline

Electrical & Computer Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Algorithm
• Optimization
• Design for Manufacturing
• Electronic Design Automation
• Computer-Aided Design

Abstract

As the dimension of features in integrated circuits (IC) keeps shrinking to fulfill Moore’s law, the manufacturing process has no choice but confronting the limit of physics at the expense of design flexibility. On the other hand, IC designs inevitably becomes more complex to meet the increasing demand of computational power. To close this gap, design for manufacturing (DFM) becomes the key to enable an easy and low-cost IC fabrication. Therefore, efficient electronic design automation (EDA) algorithms must be developed for DFM to address the design constraints and help the designers to better facilitate the manufacture process. As the core of manufacturing ICs, conventional lithography systems (193i) reach their limit for the 22 nm technology node and beyond. Consequently, several advanced lithography techniques are proposed, such as multiple patterning lithography (MPL), extreme ultra-violet lithography (EUV), electron beam (E-beam), and block copolymer directed self-assembly (DSA); however, DFM algorithms are essential for them to achieve better printability of a design. In this dissertation, we focus on analyzing the compatibility of designs and various advanced lithography techniques, and develop efficient algorithms to enable the manufacturing. We first explore E-Beam, one of the promising candidates for IC fabrication beyond the 10 nm technology node. To address its low throughput issue, the character projection technique has been proposed, and its stencil planning can be optimized with an awareness of overlapping characters. 2D stencil planning is proved NP-Hard. With the assumption of standard cells, the 2D problem can be partitioned into 1D row ordering subproblems; however, it is also considered hard, and no efficient optimal solution has been provided so far. We propose a polynomial time optimal algorithm to solve the 1D row ordering problem, which serves as the major subroutine for the entire stencil planning problem. Technical proofs and experimental results verify that our algorithm is efficient and indeed optimal. As the most popular and practical lithography technique, MPL utilizes multiple exposures to print a single layout and thus allows placement of features within the minimum distance. Therefore, a feasible decomposition of the layout is a must to adopt MPL, and it is usually formulated as a graph k-coloring problem, which is computationally difficult for k > 2. We study the k-colorability of rectangular and diagonal grid graphs as induced subgraphs of a rectangular or diagonal grid respectively, since it has direct applications in printing contact/via layouts. It remains an open question on how hard it is to color grid graphs due to their regularity and sparsity. In this dissertation, we conduct a complete analysis of the k-coloring problems on rectangular and diagonal grid graphs, and particularly the NP-completeness of 3-coloring on a diagonal grid graph is proved. In practice, we propose an exact 3-coloring algorithm for those graphs and conduct experiments to verify its performance and effectiveness. Besides, we also develop an efficient algorithm for model based MPL, because it is more expensive but accurate than the rule based decomposition. As one of the alternative lithography techniques, block copolymer directed self-assembly (DSA) is studied. It has emerged as a low-cost, high- throughput option in the pursuit of alternatives to traditional optical lithography. However, issues of defectivity have hampered DSA’s viability for large-scale patterning. Recent studies have shown the copolymer fill level to be a crucial factor in defectivity, as template overfill can result in malformed DSA structures and poor LCDU after etching. For this reason, the use of sub-DSA resolution assist features (SDRAFs) as a method of evening out template density has been demonstrated. In this dissertation, we propose an algorithm to place SDRAFs in random logic contact/via layouts. By adopting this SDRAF placement scheme, we can significantly improve the density unevenness and the resources used are also optimized. We also apply our knowledge in coloring grid graphs to the problem of group-and-coloring in DSA-MPL hybrid lithography. We derive a solution to group-3-coloring and prove the NP-completeness of grouping-2-coloring.

Graduation Semester

2019-05

Type of Resource

text

Permalink

http://hdl.handle.net/2142/104852

Copyright and License Information

Copyright 2019 Daifeng Guo

Owning Collections