Scholarlink

<p>In ion implantation process for semiconductor device fabrication, beam shape is one of important physical quantities that affect device characteristics, specifically dose uniformity and damage control. However, to control it, we must monitor shape accurately, and adjust multiple tuning parameters. This adjustment process is difficult to automate by rule-based programming because implanter requires tuning many parameters. Previously, we had developed beam size controlling system using machine learning model. This model is trained by using equipment parameters and predicts appropriate parameters to tune beam size. However, sometimes beam shape is different even if beam condition is same and the size is OK. To deal with it, we developed beam shape controlling system. We defined such beam shape parameters as “Mura value” and calculated these parameters by image processing. The value becomes indicator of dose uniformity and this system can detect abnormal beam shape and prevent non-uniformity implantation.</p> Graphical abstract <p>In ion implantation process for semiconductor device fabrication, beam shape is one of the important physical quantities that affect device characteristics, specifically dose uniformity and damage control. However, to control beam shape, we must monitor beam shape accurately, and adjust multiple beam tuning parameters according to ion implantation conditions. This adjustment process is difficult to automate by rule-based programming. Especially, medium current implanter requires tuning many parameters because the ranges of the energy and the beam current are so wide. Previously, we had developed beam size controlling system using machine learning model. This model is trained by using equipment parameters and predicts parameters to tune beam size. During mass production, if beam condition was same, beam size is almost same. But sometimes beam shape is different. Figure&#xa0;<InternalRef RefID="Fig1">1</InternalRef> shows the case. Regarding dose uniformity, double peak is worse than single peak. To deal with the above case, we developed beam shape controlling system. We defined such beam shape parameters as “Mura value” and calculated these parameters by image processing. The value is increasing as smaller beam size and number of peaks and become indicator of beam uniformity. By using this system, we can detect abnormal beam shape and prevent non-uniformity implantation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based beam shape controlling system on NISSIN medium current ion implanter

  • 期刊论文

  • 发表日期：

    2025年2月24日

查看全文
  • Shinya Takemura,
  • Masashi Imura,
  • Shigeki Sakai

摘要

In ion implantation process for semiconductor device fabrication, beam shape is one of important physical quantities that affect device characteristics, specifically dose uniformity and damage control. However, to control it, we must monitor shape accurately, and adjust multiple tuning parameters. This adjustment process is difficult to automate by rule-based programming because implanter requires tuning many parameters. Previously, we had developed beam size controlling system using machine learning model. This model is trained by using equipment parameters and predicts appropriate parameters to tune beam size. However, sometimes beam shape is different even if beam condition is same and the size is OK. To deal with it, we developed beam shape controlling system. We defined such beam shape parameters as “Mura value” and calculated these parameters by image processing. The value becomes indicator of dose uniformity and this system can detect abnormal beam shape and prevent non-uniformity implantation.

Graphical abstract

In ion implantation process for semiconductor device fabrication, beam shape is one of the important physical quantities that affect device characteristics, specifically dose uniformity and damage control. However, to control beam shape, we must monitor beam shape accurately, and adjust multiple beam tuning parameters according to ion implantation conditions. This adjustment process is difficult to automate by rule-based programming. Especially, medium current implanter requires tuning many parameters because the ranges of the energy and the beam current are so wide. Previously, we had developed beam size controlling system using machine learning model. This model is trained by using equipment parameters and predicts parameters to tune beam size. During mass production, if beam condition was same, beam size is almost same. But sometimes beam shape is different. Figure 1 shows the case. Regarding dose uniformity, double peak is worse than single peak. To deal with the above case, we developed beam shape controlling system. We defined such beam shape parameters as “Mura value” and calculated these parameters by image processing. The value is increasing as smaller beam size and number of peaks and become indicator of beam uniformity. By using this system, we can detect abnormal beam shape and prevent non-uniformity implantation.