Machine Learning-Accelerated Discovery of High-Performance Insulating Electromagnetic Interference Shielding Composites for Integrated Electronics.

Abstract

Electrically insulating electromagnetic interference shielding composites can block electromagnetic waves by inducing currents within microcapacitors, offering a downsized package paradigm for integrated electronics. However, this contradiction makes it challenging to achieve high shielding performance within insulating systems, which is fundamentally constrained by the difficulty in dramatically strengthening induced current intensity as it is sensitively mapped by the intricate microcapacitor structures. Herein, for the first time, it is revealed that uniform fillers significantly boost current intensity within microcapacitors compared to that of widely-used random fillers by an integrated workflow combining machine learning and simulations, offering a targeted guideline for high-performance insulating shields. Guided by this, a microfluidic technique is utilized to produce uniform, monodisperse Gallium particles in high-throughput as fillers for insulating composites, which exhibit excellent shielding effectiveness (>90 dB, Ka-band) and thermal conductivity (3.7 ± 0.1 W m^-1 K^-1) at high resistivity (1.7 × 10¹² Ω·m), outperforming conventional non-uniform and conductive systems. Its superior near-field shielding and heat-dissipation abilities allow for directly tackling electromagnetic compatibility and overheating issues without short-circuit failures over a wide temperature range (-196-200 °C). Therefore, this work offers valuable insights into the manufacturing of high-performance insulating shields, potentially breaking the limitations of traditional low-throughput iterative experimentation and promoting microelectronics industry.