Theoretical and experimental Raman study for mechanical stress in die-attach process

Abstract

We measured the stress in a Si chip mounted on a Cu plate using Raman spectroscopy and compared the results with those calculated by using the finite element method (FEM). The difference in the stress observed at the vicinity of the chip edge was approximately 30 MPa, and the simulated value was smaller than that observed using Raman spectroscopy. We found that the difference occurred owing to the absence of the “filet” structure generated from the epoxy adhesive, and we subsequently modified the simulation model. The modified FEM results were in good agreement with the measured results. These results indicated that the “filet” structure, which was not the main structure of die-bonding, played an important role in residual stress formation. Feedback using Raman measurements is crucial, and we evaluated the stress in electronic devices with high accuracy by combining FEM with Raman microprobe measurements. In cross-sectional measurements, we observed discrepancy between the results, and the difference in the vertical stress was found to be dependent on excitation laser wavelength. Except for the discrepancy near the backside, the Raman results were in good agreement with the FEM results. This indicated that Raman spectroscopy can evaluate the stress components in the Si cross-section, and full three-dimensional stress in the packaging devices can be determined by combining FEM with cross-sectional Raman measurements, even if the internal stress is relaxed via cutting.