A feasible method to predict thin package actual warpage based on an FEM model integrated with empirical data

One of the current trends for advanced packages is increasingly thinner packaging for mobile devices. Thin packages increase warpage tremendously making control of package warpage very critical and challenging. Furthermore, a thin package is also much more sensitive to various factors including incoming conditions, processing conditions and nonlinear material behaviors. As a result, current warpage models based purely on the finite element method (FEM) have difficulty accurately predicting the warpage of packages in an actual production environment because the models are not able to capture all the complicated process issues as well as material nonlinearities. On the other hand, current advanced packages are mostly custom built with fast development cycles. Avoiding problems requires a feasible method to predict warpage during the package development stage to guide the design improvement. This paper presents a new feasible method to accurately predict package actual warpage in production based on a finite element model integrated with empirical shadow moiré warpage data. The unique aspect of this method uses existing empirical warpage data from a broad range of different package types and design parameters to estimate the actual reference temperature point as well as an initial warpage as inputs into the FEM warpage simulation model. The adjusted actual reference temperature and the initial warpage are used to account for any warpage caused by unknown factors in the assembly processes and materials which cannot be captured by the FEM model. Through the product development cycle, the semi-empirical warpage model is continually updated as more and more new empirical data are available making it more and more accurate for next-cycle predictions. This paper uses a Package-On-Package (POP) and a flip-chip Chip-Scale-Package (fcCSP) as test vehicles to illustrate this unique semi-empirical warpage prediction method.