BEoL stack robustness investigations utilizing Cu-pillar immobilization and micromechanical loading methods

Abstract

Under harsh application conditions in fields such as aerospace or automotive, it is crucial that the mechanical robustness of microchip components is ensured at any time. To obtain this, thorough and highly specialized testing approaches have to be deployed already in the design phase. The general strategy of these approaches is to induce (thermo-)mechanical load to a given system or component under well controlled conditions and assess the resulting effects. The testing methods should emulate the mechanical application conditions as closely as possible to identify the most damage prone areas of the system and to obtain an understanding of the occurring damage processes. In this work, the development and deployment of two different testing methods to evaluate mechanical BEoL (Back end of Line) stack robustness by inducing micromechanical load to superjacent Cu-pillars are introduced. In flip chip applications, Cu-pillars are immobilized by a solder connection as well as underfill material. Compared to previously developed methods, it was attempted to realize mechanical scenarios which are closer to assembly or application conditions. Therefore, the methods presented in this work can be regarded as subsequent developments of e.g. the Cu-pillar shear-off approach presented in Silomon et al. (2021). The details are discussed in more depth in Silomon et al. (2022). Specifically, a mechanical immobilization approach utilizing an indenter tip with a cavity was developed as well as a soldering approach utilizing a Cu indenter tip as a soldering bolt and a landing pad simultaneously. The approaches were deployed on a high-end microchip sample which has been introduced and investigated in Silomon et al. (2021).Two different mechanical loading conditions were emulated utilizing the developed approaches. These conditions are relevant at different points in the life cycle of a microchip and could not be induced utilizing previously developed methods. Oscillating load as an emulation of thermal expansion and contraction or vibration (application conditions) as well as tensile loading as an emulation of assembly conditions were induced. Applying mechanical load to individual Cu-pillars utilizing the different methods enabled the determination of the failure conditions and subsequent analysis led to the identification of the respective damage mode. Additionally, sub-critical experiments utilizing AE (acoustic emission) signal monitoring as a damage indicator were conducted to determine the exact location of damage initiation. This was achieved by subsequent damage assessment measurements utilizing optical microscopy, nXCT as well as SEM/FIB analyses. Applying this workflow, it was possible to describe the damage modes as well as to identify the most damage prone areas for the specific loading conditions induced by the two different developed mechanical loading methods.