21
Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC‘13

Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Embed Size (px)

Citation preview

Page 1: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Linear Time Algorithm to Find All Relocation Positions for EUV Defect

Mitigation

Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong

ASPDAC‘13

Page 2: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Outline

• Introduction• Problem formulation• Algorithm• Experimental results • Conclusion

Page 3: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Introduction

• With the current VLSI technology shrinking to sub-20nm, EUV lithography has become a leading candidate to replace the 193nm technology.

• But, the absence of defect-free blanks in EUV mask fabrication, a methodology that is able to deal with defective blanks will be required

Page 4: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Introduction

Page 5: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Introduction

Page 6: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Problem formulation

• Maximize the number of valid dies within the exposure field.

Page 7: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Problem formulation

• Find the all feasible regions.

Page 8: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Algorithm

• Blank Region Partition– Blank Region with No Effective Defect– Blank Region with Single Effective Defect– Blank Region with Multiple Effective Defect

Page 9: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Blank Region Partition

• no part of the die can be shifted outside the exposure field• Each defect has a unique impact range

Page 10: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Blank Region with No Effective Defect

• The bottom left corner of the die is located within this region, no defect will locate within the die area, and hence all defect impact is completely mitigated.

Page 11: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Blank Region with Single Effective Defect

• Step 1. Impacted Feature Extraction• Step 2. Impacted Feature Shrinking• Step 3. Shrunk Die Area Rotation and Shift

Page 12: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Impacted Feature Extraction

• The width/height of 3’ is equal to the width/height of 3 plus the width/height of the defect.

• Impacted Feature: F2,F3

Page 13: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Impacted Feature Shrinking

• In the second step, the impacted features and the impacted die area 3 are shrunk by the size of the effective defect.

Page 14: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Shrunk Die Area Rotation and Shift

• Therefore, the shrunk features in 3’’ and the feasible regions in 3 are diagonally symmetric.

Page 15: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Blank Region with Multiple Effective Defect

• The objective is to find the feasible regions to locate the die where all defects are covered by features simultaneously.

• We first consider each defect separately• And the sets of feasible regions are

intersected to obtain the final feasible region

Page 16: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Time Complexity Analysis

• n: the number of features in the die.

Page 17: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Improved Strategy for Layout Intersection

Page 18: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Experimental result

• We implement our algorithm using C++ on a workstation with an Intel Xeon E5620 2.40GHz CPU and 36GB memory

• Then we carry out our experiments with a 11nm design

• The defects size form 50nm to 200nm• The size of the exposure field is 10.4cm by

13.2cm.

Page 19: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Experimental results

Page 20: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Experimental results

Page 21: Linear Time Algorithm to Find All Relocation Positions for EUV Defect Mitigation Yuelin Du, Hongbo Zhang, Qiang Ma and Martin D. F. Wong ASPDAC13

Conclusion

• In this paper we propose an efficient algorithm to find all layout relocation positions to place a valid die on a defective blank for defect mitigation.

• By simple parallelism, the efficiency can be further improved.