Samsung Advanced Institute of Technology

As semiconductor technology becomes rapidly advanced, semiconductor packaging materials demand critical characteristics in various aspects. Besides, the development of these new epoxy mold compounds (EMC) by batch reaction process requires time-consuming experimentation and the multi-step chemistry further reduces the efficiency of the synthetic process. We report a one-flow multi-step process for the synthesis of newly developed Np-C₄-Np, which comprises two mesogenic units connected with a flexible spacer, as a monomeric precursor of semiconductor packaging material. Graphical convolution neural network (GCNN), a deep learning model, predicts a common solvent for three-step reactions, thereby enabling serial esterification-deprotection-epoxidation integrated with in-situ multi-phasic separations that were accomplished within 76 min in flow over the long batch process (> 20 h excluding separation step). Eventually, the Np-C₄-Np is synthesized rapidly with higher space-time yield compared to the batch system, thus confirming benefits in terms of productivity. Moreover, we prevent the deterioration of electronic circuits during the semiconductor packaging process by lowering the molding temperature (126 °C) of EMC and quickly dissipating the heat generated from semiconductor chip by increasing the thermal conductivity (0.34 Wm^-1K^-1) of EMC.