Excellent absorption-dominant electromagnetic interference shielding performances of asymmetric gradient layered composite films exploited with assistance of machine learning.

Developing high-performance absorption-dominant electromagnetic interference (EMI) shielding composites is essential yet challenging for advanced high-power electronic devices to minimize the second EMI radiation. Traditional experiment-based approaches for shielding material exploitation usually require extensive fabrication and characterization procedures, leading to a long duration and high expense. Herein, machine learning was applied to assist in developing calcium alginate/sodium montmorillonite/CNT@FeCo/CNT (CA/MMT/CNT@FeCo/CNT, CMF/CMFC-x wt%/CMC-y wt%) EMI shielding composites with the asymmetrical gradient layered architecture, triggering the optimization of absorption-dominant EMI shielding properties and reducing experimental costs. The fabricated CMF/CMFC-48.4 wt%/CMC-43.9 wt% film with a small thickness (341.4 μm) exhibits the superior averaged total EMI shielding effectiveness (EMI SE) of 38.9 dB and optimal absorption coefficient (A) of 0.61, when electromagnetic waves (EMWs) are incident from CMF layer. Based on experimental data, the reflection shielding effectiveness (SE), absorption shielding effectiveness (SE), reflection coefficient (R), and A are utilized to train and test four different machine learning models. Polynomial Linear model (PL) possesses the best prediction accuracy and reliability with the root mean square error (RMSE) of SE and SE lower than 0.7022, and RMSE of R and A below 0.0361, suggesting that machine learning can effectively alleviate the experimental burden. Moreover, the composite film also features the acceptable mechanical properties and prominent fire resistance, which is vital for the practical application. This work provides a new idea for reducing experimental costs and accelerating the discovery of advanced absorption-dominant EMI shielding materials.