| Summary: | 碩士 === 國立成功大學 === 工程科學系碩博士班 === 98 === With the characteristics of miniature in size, good electric and thermal performance, low manufacture costs as well as low failure rate, the QFN has been paid attention gradually in the market. The stacked die QFN package is adopted to analyze the effects of the design parameters. Then the reliability analysis under the thermal cycle and the sensitivity analysis on the design parameters are conducted to enhance the fatigue life of the package and the stability of products.
ANSYS10.0 software with the Global/Local method is applied for analysis. Based on the JEDEC code, the stacked die QFN is subjected by a thermal cycle of -40°C~125°C. The viscoplastic property of solder paste is assumed to be Anand’s model. Darveaux’s energy concept is employed to analyze of the strain energy density distribution of the paste solder. Accordingly, an average value is treated as the reliability index for evaluating the package
Prior to the process of the optimal design on reliability of stacked die QFN package, one-factor-at-a-time analysis is conducted to investigate the feasibility of the ANSYS simulation model. Afterwards, the most significant factors are chosen by the fractional factorial design method and a regression model of the response surface is set up in which the genetic algorithm is introduced to obtained the optimal combination of parameters. It shows that the reliability of the QFN package can be effectively improved along with the reduction of the die size, thickness of PCB, CTE of PCB as well as the increase of CTE of mold compound. Finally, the interval genetic algorithm (IGA) is applied to analyze each parameter’s sensitivity to the reliability index. They can be ranked from the largest to the smallest as follows: CTE of mold compound, thickness of PCB, CTE of PCB and die size. It is expected to improve the accuracy of sensitive parameters in the manufacturing process so that the stability of the product quality can be ensured.
|
|---|
