Professor Lee, Eun-Ho and his research team has proposed a new method for evaluating and optimizing thermal and mechanical properties in complex semiconductor package designs, and implemented it into a program. This research has attracted great attention from academia and industry as it provides a comprehensive way to analyze thermal and mechanical properties to improve performance, secure reliability, and reduce design costs of semiconductor packages.

Semiconductor package design has traditionally focused on electrical properties, but as highly integrated package designs evolve, thermal and mechanical properties are becoming increasingly important to ensure reliability. In recent years, the complexity of package patterns has increased significantly for applications such as chiplet structures, but it has been difficult in the field to determine the thermal and mechanical properties of all proposed designs due to the significant increase in design costs. Ph.D student Jeong-Hyeon Park and prof. Lee, Eun-Ho of the Department of Mechanical Engineering proposed a methodology to quickly obtain big data of thermal and mechanical properties of packages with complex patterns at low cost through numerical analysis and effectively analyze this big data through deep learning (see Figure 1). In addition, they collaborated with Samsung Electronics from 2021 to 2024 to verify the proposed methodology by applying it to Samsung Electronics' actual package blueprints. The verified methodology predicted thermal and mechanical properties in real time for new design drawings and created a property map to help Samsung Electronics with design (see Figure 2).

The platform has applied for a national (10-2022-0129656) and US patent (18/206,278) with Samsung Electronics, and two international papers were published in 2022 (IEEE ACCESS, JCR top 34%) and 2024 (Applied Mathematical Modelling, JCR top 9%). Mr. Park won the best paper at the Spring Conference of the Korean Society of Precision Engineering in 2024, and Professor Lee won the 34th Outstanding Science and Technology Paper Award. He is also scheduled to give an invited talk at an international conference on the IMPACT package in Taiwan in October 2024.

“This research provides an important tool for the integrated evaluation of the thermal and mechanical properties of semiconductor package designs, which will greatly improve the efficiency and reliability of package designs,” said Prof. Lee. His research team is currently working on a follow-up paper on a new thermal resistance network structure that can more effectively represent the thermal properties of semiconductor packages, and is collaborating with other universities and research institutes to expand the application of this platform. It is expected to set a new standard in semiconductor package design and optimization.

[Figure 1] Thermal-mechanical property training model developing algorithm

[Figure 2] AI based thermal-mechancial real time prediction program