November_EDFA_Digital ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 23 NO . 4 14 METHODS TO ENABLE FAULT ISOLATION ON 2.5D MOLDED INTERPOSER PACKAGES Douglas Hunt 1 , Daniel Bader 1 , Pascal Limbecker 2 , and Heiko Barth 2 1 GlobalFoundries Inc., Essex Junction, Vermont 2 GlobalFoundries Inc., Dresden, Germany EDFAAO (2021) 4:14-17 1537-0755/$19.00 ©ASM International ® INTRODUCTION To meet the demands of high speed and high band- width communication between logic die and stacked memory modules, heterogeneous 2.5D packaging incor- porating a passive silicon interposer has emerged as a standard solution. Asmore dies need tobe interconnected on one module, the packages have become larger, often exceeding 60 x 60mm in size. This poses challenges to the assemblyof the components, interposer, andorganic lami- nate to create a reliable package. One emerging enhance- ment is to introduce a wafer-level interposer overmold process [1] (Fig. 1). The overmold allows for planarization of the top die surfaces, despite original stacked die height differences, allowing for a common heat spreader solu- tion. It also stabilizes the die to interposer interconnects around the edges and corners, and protects the die from moisture and corrosion. The presence of the overmold becomes a real chal- lenge during failure analysis of 2.5D modules. Removing the overmold without damaging the metallization on top of the interposer or laminate is difficult but neces- sary in order to perform any subsequent fault isolation. Localization techniques such as electron beam absorbed current (EBAC) and optical beam induced resistance change (OBIRCH) need to have the area of interest delay- ered close to the surfacewhere the defectmay be present. The large module size is also a challenge to configure in a system for EBAC analysis. This article will discuss the failure analysis challenges posed by large overmolded 2.5D packages, and introduce amicrowave-induced plasma (MIP) process to remove the overmold, as well as techniques to handle large packages for EBAC analysis. CASE STUDY DETAILS The 2.5D package used in this case study had an organic laminate, silicon interposer composedof through- silicon vias (TSVs) andpassivewiring levels, a large test site die in the center, and four smaller dies surrounding it. A metal stiffener ring was also attached to the outer edge of Fig. 1 Diagram of a typical 2.5D overmolded interposer module.