| Summary: | 碩士 === 國立中山大學 === 機械與機電工程學系研究所 === 104 === The thesis aims to enhance the reliability of the flip-chip ball grid array (FCBGA) package and coreless substrate by changing the material properties of dielectric material, thickness and copper coverage. The probability of failure of the package owing to warpage because of CTE mismatch is then significantly reduced.
The warpage of a FCBGA package under thermal loadings through both experiment and simulation is investigated. Shadow Moiré was applied to assess the warpage of a package at room temperature after finishing each assembly process. And we compared the influence of dielectric material, thickness and copper coverage on warpage of the package. Utilizing the finite element software ANSYS to build up models under thermal loadings we received the warpage of package and bare substrate and verified the results with experimental data to identify the significant degree for various factors.
The experimental data and numerical simulation results show the warpage of the package only applied dielectric material A in the substrate is not good. From the material analysis of the equivalent package model, we can decrease the warpage if we choose lower CTE, higher modulus and lower Tg of dielectric material, and their contributions are 22%, 13.5% and 18%, respectively. From the structure analysis of the equivalent package model, we find out the factor of improving warpage of package is the thickness of copper trace and the thickness of dielectric material, and their contributions are 30% and 1.5%, respectively. It shows the thickness of copper trace is more significant than that of dielectric material.
From the results of detailed bare substrate model, it is obvious to see that the factors of improving warpage of substrate are the thickness of dielectric material at chip side, the thickness of copper trace at chip side, the thickness of dielectric material at ball side and the thickness of copper trace at ball side. The contribution of the thickness of copper trace at ball side is 30% and the contributions of other layers are not significant. As for the analysis of copper coverage, it is better to find that copper coverage is 51% on the copper trace layer at chip side, 32% on the copper trace layer at ball side. And the influence of the copper coverage of copper trace layer at chip side is more outstanding than that of copper trace layer at ball side.
|
|---|
