Artificial intelligence aided design for heterogeneous integration system in display

The laminated heterogeneous integration system in display (HISID) takes full advantages of the display board, and the multi-interactive function units (e.g., display units, storage units, etc.) are integrated into the display substrate. The HISID model is regarded as a promising candidate and potential architecture for higher density and performance products, and it is proposed with the advantages of intelligence, light-weight, high bandwidth, ultra-low latency, mobility. However, the thermal reliability challenges of the HISID model are driven by the mismatch of coefficient of thermal expansion (CTE) and mechanical properties among silicon, solder joint and substrate, etc. With the demands of shorter development time and lower cost of electronic products, the finite element method (FEM) is usually applied to the reliability design, where the stress distribution along heterogeneous interfaces and the warpage deformation are calculated. Nevertheless, the computational convergence, contact definition, mesh quality and postprocessing of FEM are also needed to considered in engineering application, especially the parameters study of material properties, geometry and thermal environment are needed to investigate through case-by-case method. To catch up with these technology tendencies and requirements, a comprehensive analysis with artificial intelligence aided design method is utilized for the electronic packaging for the HISID model, and the thermal reliability behaviors are optimized. The variation trend of warpage deformation and interfacial stress along with the entire temperature change of arbitrary thermal cycling are predicted from artificial neural network (ANN), and the computing time is decreasing from hours to seconds. Thus, the artificial neural network (ANN) is one useful tool to solve FEM problems with limited inputs, and then it is further applied to more complex and time-consuming heterogeneous integration system.