Die-to-die alignment for lithographic processing of reconstructed wafers

Mikhail Loktev, Sylvain I. Misat, Ralph Schiedon,Jeroen de Boeij,Michiel van der Stam, Pierre Sixt,Haidar Al Dujaili,Tristan Dewolf,Nacima Allouti,Laurent Pain, Cyril Vannuffel,Perceval Coudrain,Arnaud Garnier

2022 Smart Systems Integration (SSI)(2022)

引用 2|浏览28
暂无评分
摘要
In this work, we address one of the challenges of Fan-Out Wafer Level Packaging (FO-WLP), which is chip placement error, which occurs during the process of wafer reconstruction and molding. In a typical FO-WLP process, a wafer processed on a front-end tool is diced, and the dies are repositioned on a carrier with additional space created for fan-out structures. A dedicated alignment approach, which includes measurement of individual dies to adjust the settings per exposure, would result in a dramatic improvement of the overlay performance. This process further referred to as die-to-die alignment is generally known to have negative impact on throughput for IC manufacturers.The accuracy of die-to-die alignment on 200 mm wafers is evaluated experimentally in collaboration between Kulicke & Soffa and CEA LETI. The wafer layout consists of dies of different sizes (120 chips of 4 mm x 4 mm and 70 of 10 mm x 10 mm) for which intentional misalignments have been introduced with different amounts of translation (up to +/-50 micrometers) and rotation (up to +/-10 milliradians) in order to emulate typical errors found in reconstructed wafers.Alignment of the "back-end" layer with respect to existing patterns on the wafer was measured both at Kulicke & Soffa and at CEA LETI. Both measurements confirmed sub-micrometer accuracy of overlay between the structures on the reference wafer and the new layer exposed on LITEQ 500. Throughput in-line with current industrial standards was achieved. Data analysis shows that major improvements of the throughput can be achieved by optimizing the exposure process.
更多
查看译文
关键词
die-to-die,alignment,WLP,Fan-out,RDL,Lithography,LITEQ 500,Overlay
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要