关键词: 积层绝缘膜, 先进封装, 性能权衡, 人工智能, 协同设计

Abstract: Build-up dielectric films are critical dielectric materials in advanced packaging, where their performance directly determines the reliability and signal integrity of chip systems. However, simultaneously optimizing four key properties—high copper adhesion, a low coefficient of thermal expansion (CTE), low dielectric constant, and excellent processability—presents a significant challenge due to fundamental conflicts in their underlying physics and chemistry. This review first provides a systematic overview of conventional research strategies for build-up dielectric films, focusing on mainstream approaches in low-dielectric resin design, filler/resin interface engineering, and processing of highly-filled composites. Building on this analysis, we examine the core scientific issues that hinder synergistic performance improvements at the molecular and mesoscales: the inherent conflict between polarity requirements (for adhesion vs. dielectric properties) and the physical limits of filler loading versus viscosity. To overcome the limitations of traditional trial-and-error methods, this paper explains how artificial intelligence (AI) offers a powerful new approach to resolve these challenges through a multi-scale co-design framework. We also identify the key hurdles that must be addressed to successfully implement AI in this field. Finally, we provide an outlook on future trends and technological directions for the development of advanced build-up dielectric film materials.

Key words: Build-up film, advanced packaging, performance trade-off, artificial intelligence, co-design