Scholarlink

<p>Wafer probe testing is essential for evaluating the reliability of microelectronic devices. As device dimensions continue to shrink, the mechanical stress at the probe tip increases the risk of failure occurring in the brittle dielectric layer. Existing studies usually simplify the plastic behavior of bond pads in finite element (FE) models, making it difficult to accurately predict the stress in the dielectric layer. In this study, the FE model for wafer probe testing was improved by implanting the true plastic properties of Al-(4 wt.%)Cu bond pads obtained from nanoindentation inverse identification. Specifically, a MATLAB and ABAQUS co-simulation method is employed to inversely obtain the plastic properties of Al-(4 wt.%)Cu bond pads from the nanoindentation load-displacement curve. By comparing different gradient-based optimization algorithms, the Levenberg-Marquardt (LM) algorithm was found to be the most suitable for balancing computational efficiency and accuracy in inverse identification. Moreover, a hybrid optimization strategy combining Gauss-Newton and LM algorithms is proposed to mitigate sensitivity to initial values. The results show that the improved model of wafer probe testing achieves high accuracy in stress prediction. This study provides a high-fidelity approach for mechanical reliability assessment of wafer probe testing.</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

Mechanical modeling of wafer probe testing based on nanoindentation inverse identification

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Dongyang Hou,
  • Ting Lv,
  • Yuhang Ouyang,
  • Shunyong Jiang,
  • Fang Dong,
  • Sheng Liu

摘要

Wafer probe testing is essential for evaluating the reliability of microelectronic devices. As device dimensions continue to shrink, the mechanical stress at the probe tip increases the risk of failure occurring in the brittle dielectric layer. Existing studies usually simplify the plastic behavior of bond pads in finite element (FE) models, making it difficult to accurately predict the stress in the dielectric layer. In this study, the FE model for wafer probe testing was improved by implanting the true plastic properties of Al-(4 wt.%)Cu bond pads obtained from nanoindentation inverse identification. Specifically, a MATLAB and ABAQUS co-simulation method is employed to inversely obtain the plastic properties of Al-(4 wt.%)Cu bond pads from the nanoindentation load-displacement curve. By comparing different gradient-based optimization algorithms, the Levenberg-Marquardt (LM) algorithm was found to be the most suitable for balancing computational efficiency and accuracy in inverse identification. Moreover, a hybrid optimization strategy combining Gauss-Newton and LM algorithms is proposed to mitigate sensitivity to initial values. The results show that the improved model of wafer probe testing achieves high accuracy in stress prediction. This study provides a high-fidelity approach for mechanical reliability assessment of wafer probe testing.