Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advancing SEM Based Nano-Scale Defect Analysis in Semiconductor Manufacturing for Advanced IC Nodes

Bappaditya Dey, Matthias Monden, Victor Blanco, Sandip Halder, Stefan De Gendt

TL;DR

The paper addresses the challenge of detecting and segmenting nano-scale semiconductor defects at advanced nodes, where ground-truth pixel-wise masks are hard to obtain. It introduces ADCDS, a two-stage pipeline that uses Deformable DETR for defect detection and BoxSnake for box-supervised segmentation to generate pixel-precise defect masks without requiring dense pixel annotations. On real SEM datasets (ADI and AEI) focusing on Line-Space defects, ADCDS achieves $mAP@IoU=0.5$ of $72.19$ for detection and $78.86$ for segmentation on ADI, and $90.38$ and $95.48$ on AEI, respectively, with reasonable inference times and robust qualitative results. The framework reduces labeling burden, demonstrates first nano-scale segmentation on ADI data, and paves the way for unsupervised/weakly supervised defect analytics to enhance process control in high-volume manufacturing.

Abstract

In this research, we introduce a unified end-to-end Automated Defect Classification-Detection-Segmentation (ADCDS) framework for classifying, detecting, and segmenting multiple instances of semiconductor defects for advanced nodes. This framework consists of two modules: (a) a defect detection module, followed by (b) a defect segmentation module. The defect detection module employs Deformable DETR to aid in the classification and detection of nano-scale defects, while the segmentation module utilizes BoxSnake. BoxSnake facilitates box-supervised instance segmentation of nano-scale defects, supported by the former module. This simplifies the process by eliminating the laborious requirement for ground-truth pixel-wise mask annotation by human experts, which is typically associated with training conventional segmentation models. We have evaluated the performance of our ADCDS framework using two distinct process datasets from real wafers, as ADI and AEI, specifically focusing on Line-space patterns. We have demonstrated the applicability and significance of our proposed methodology, particularly in the nano-scale segmentation and generation of binary defect masks, using the challenging ADI SEM dataset where ground-truth pixelwise segmentation annotations were unavailable. Furthermore, we have presented a comparative analysis of our proposed framework against previous approaches to demonstrate its effectiveness. Our proposed framework achieved an overall mAP@IoU0.5 of 72.19 for detection and 78.86 for segmentation on the ADI dataset. Similarly, for the AEI dataset, these metrics were 90.38 for detection and 95.48 for segmentation. Thus, our proposed framework effectively fulfils the requirements of advanced defect analysis while addressing significant constraints.

Advancing SEM Based Nano-Scale Defect Analysis in Semiconductor Manufacturing for Advanced IC Nodes

TL;DR

The paper addresses the challenge of detecting and segmenting nano-scale semiconductor defects at advanced nodes, where ground-truth pixel-wise masks are hard to obtain. It introduces ADCDS, a two-stage pipeline that uses Deformable DETR for defect detection and BoxSnake for box-supervised segmentation to generate pixel-precise defect masks without requiring dense pixel annotations. On real SEM datasets (ADI and AEI) focusing on Line-Space defects, ADCDS achieves of for detection and for segmentation on ADI, and and on AEI, respectively, with reasonable inference times and robust qualitative results. The framework reduces labeling burden, demonstrates first nano-scale segmentation on ADI data, and paves the way for unsupervised/weakly supervised defect analytics to enhance process control in high-volume manufacturing.

Abstract

In this research, we introduce a unified end-to-end Automated Defect Classification-Detection-Segmentation (ADCDS) framework for classifying, detecting, and segmenting multiple instances of semiconductor defects for advanced nodes. This framework consists of two modules: (a) a defect detection module, followed by (b) a defect segmentation module. The defect detection module employs Deformable DETR to aid in the classification and detection of nano-scale defects, while the segmentation module utilizes BoxSnake. BoxSnake facilitates box-supervised instance segmentation of nano-scale defects, supported by the former module. This simplifies the process by eliminating the laborious requirement for ground-truth pixel-wise mask annotation by human experts, which is typically associated with training conventional segmentation models. We have evaluated the performance of our ADCDS framework using two distinct process datasets from real wafers, as ADI and AEI, specifically focusing on Line-space patterns. We have demonstrated the applicability and significance of our proposed methodology, particularly in the nano-scale segmentation and generation of binary defect masks, using the challenging ADI SEM dataset where ground-truth pixelwise segmentation annotations were unavailable. Furthermore, we have presented a comparative analysis of our proposed framework against previous approaches to demonstrate its effectiveness. Our proposed framework achieved an overall mAP@IoU0.5 of 72.19 for detection and 78.86 for segmentation on the ADI dataset. Similarly, for the AEI dataset, these metrics were 90.38 for detection and 95.48 for segmentation. Thus, our proposed framework effectively fulfils the requirements of advanced defect analysis while addressing significant constraints.
Paper Structure (18 sections, 5 figures, 7 tables)

This paper contains 18 sections, 5 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Proposed ADCDS framework
  4. Experiments
  5. Datasets
  6. Evaluation criteria
  7. Training
  8. Defect instance detection.
  9. Defect instance segmentation.
  10. Results
  11. Proposed ADCDS framework
  12. Defect detection module.
  13. Defect segmentation module.
  14. Discussion
  15. Proposed ADCDS framework
  16. ...and 3 more sections

Figures (5)

  • Figure 1: Schematic of the proposed ADCDS framework
  • Figure 2: Defect examples from ADI (a) and AEI (b) dataset
  • Figure 3: mAP @ IoU 0.5, train, and test loss with Deformable DETR architectures on the (a) ADI and (b) AEI SEM datasets
  • Figure 4: Defect detection results with proposed ADCDS framework (Deformable DETR based defect detection module) on ADI (a) and AEI (b) SEM dataset
  • Figure 5: Defect mask segmentation and generation results with proposed ADCDS framework (BoxSnake based defect segmentation module) on ADI (a) and AEI (b) SEM datasets.