Cross-Sectional Nanoindentation for Characterizing Adhesion Strength of Metallic Thin Films

Adhesion strength is a critical parameter that determines the mechanical robustness and reliability of metallic thin film systems in semiconductor devices. Quantifying adhesion strength is essential for optimizing manufacturing processes and understanding reliability issues under operational conditions. However, due to dimensional limitations, direct measurement and corresponding assessment of the thin film adhesion strength pose significant challenges. The Cross-Sectional Nanoindentation (CSN) technique was originally developed to quantify the adhesion strength of elastic brittle thin films on semiconductor substrates. While this technique is suitable for ductile thin film systems from an experimental perspective, the current assessment approach relies solely on elastic theory and does not account for plastic work dissipated in the thin film. Furthermore, state-of-the-art methods measure CSN-induced delamination only from the front view and model it in 2D with an axial symmetric assumption. However, observations from Roy et al.[93] suggest that CSN-induced delamination exhibits irregular axial symmetry geometry. This 2D axial symmetric assumption leads to an inaccurate estimation of both the delamination area and the corresponding adhesion strength. Thus, the objective of this work is to develop novel assessment approaches for the CSN technique that accurately characterize the adhesion strength of ductile metallic thin films in semiconductor devices. In this study, CSN measurements are performed on a typical metallic thin film system, specifically Al on SiO2/Si, which represents metallic thin film applications in semiconductor devices. By varying the thin film thickness and incorporating a Ti intermediate adhesion layer, the adhesion strength can be modified qualitatively. Using a Scanning Electron Microscope (SEM), CSN-induced delaminations are measured not only from the front view but also from the top view and cross-section view, allowing for the acquisition of detailed spatial information. The top view measurements, conducted after etching the metallic thin films, reveal the CSN-induced wedge fracture in the substrate, highlighting that solely relying on measurements from the front view is insufficient for accurately determining the adhesion strength. A characterization strategy is developed in this work, which can be hierarchically divided into two levels: the "material level" and the "interface level." At the material level, the selected Al thin film systems will be characterized experimentally using conventional nanoindentation. Additionally, corresponding Finite Element (FE) simulations and inverse optimization are conducted to determine the plasticity of the Al thin films. The determined material properties will subsequently be used to characterize adhesion strength and calculate the plastic work dissipated in the thin film during CSN-induced delamination. At the "interface level," two different approaches are developed to assess thin film adhesion strength from the CSN measurements: a quantitative analytic approach based on energy dissipation extracted from load-displacement curves, and a numerical approach utilizing FE modeling and Cohesive Zone Model (CZM). Both approaches yield consistent results, indicating that reducing the thin film thickness and adding an intermediate Ti layer can enhance adhesion strength. Quantitatively, reducing the Al thin film thickness from 1500 nm to 1000 nm strengthens adhesion by approximately 7 times according to the analytic approach and by 5.2 times according to the 3D FE approach. Adding a Ti intermediate layer improves adhesion strength by approximately 30 times and by 19.4 times according to both approaches, respectively. These results demonstrate the suitability of the CSN technique and the developed approaches for assessing differences in adhesion strength due to varied deposition parameters. Moreover, the 3D FE simulation facilitates the calculation of internal mechanical parameters such as mode mixity and plastic work dissipated during the CSN experiment. These findings indicate that the complex nature of the CSN-induced plastic fields and mode mixity distribution must be considered for an accurate assessment of thin film adhesion strength. In a case application, the CSN technique and the developed approaches are employed to assess changes in metallic thin film adhesion strength following Temperature Shock Test (TST) cycles. In contrast to reports in the literature regarding degradation, the adhesion strength of the aged thin film system (Al 1500 nm + 20 nm Ti) increases after TST cycles. This increase is likely due to the release of internal stresses resulting from cyclic thermomechanical loading and diffusion processes at high temperatures. Modeling the morphology development resulting from TST does not impact the assessment of adhesion strength using the developed approaches.