Metal-Dielectric Stack Damage Evaluation Utilizing Acoustic Emission Signal Analysis

L. Schubert,E. Zschech,Oliver Köchel,A. Clausner,J. Silomon,M. Stephan
Abstract:The mechanical reliability is of major importance for the development, qualification, and implementation of semiconductor devices. This is especially the case for demanding application cases like e.g. automotive, aerospace, and industrial. Mechanically critical aspects like the stability of metal-dielectric stacks, the so called “back end of line” (BEoL), have to be evaluated thoroughly before implementation. This evaluation is mainly conducted by inflicting (thermo-)mechanical damage to the respective samples and a consequent damage analysis. A new approach presented in this work contains the utilization of acoustic emission measurements during the damage infliction process to identify damage occurrence and obtain more information on the damage mode. This enables the identification of the most damage prone areas in the BEoL stack. The sample analyzed in this work is an unpackaged microchip bumped with Copper pillars (Cu-Pillars) as electrical connectors. These pillars are attached to the subjacent BEoL stack which provides the power supply and distribution of electrical signals of the chip. The application of shear force to the Cu-pillars utilizing a customized indenter tip enables the controlled infliction of damage into the BEoL stack. A sample holder was specifically designed for the experiments which facilitates the placement of an acoustic emission (AE) sensor below the sample. This enables the immediate identification and measurement of AE events during the damage infliction process. Previous works have already exhibited the potential of AE measurements for damage indication in semiconductor structures [1,2]. Acoustic data analysis techniques like cluster analysis can be utilized for damage evaluation, e.g. in composite materials, as well [3]. In this work, the possibility of utilizing this approach for the analysis of damages like delamination or cracking occurring in BEoL stacks is evaluated. This is done by correlating the occurring lateral forces and damage modes with the characteristics of the AE signals. The developed setup as well as first results regarding damage mode evaluation and categorization are provided in this work.
Materials Science,Engineering,Physics
What problem does this paper attempt to address?