Asymmetric multistage composite foams for high-efficient absorption-dominant electromagnetic interference shielding exploited with aid of machine learning

High-efficient absorption-dominant electromagnetic interference (EMI) shielding materials with both ultra-high total EMI shielding effectiveness (SE T  > 100 dB) and absorption coefficient (A>0.95) as ideal “green” shielding materials are imminently demanded yet scarcely reported for minimizing secondary EMI radiation pollution. Herein, pyrolytic carbon nanotube (CNT)@FeCo/CNT/silver-coated tetra-needle-like zinc oxide whisker/poly ( l -lactic acid) (CNT@FeCo-700/CNT/T-ZnO@Ag/PLA) composite foams were successfully fabricated based on a conductivity-gradient modular design. The rational layout of the gradient conductive network structures of CNT@FeCo-700/PLA, CNT/PLA, and T-ZnO@Ag/PLA layers enables the “strong absorption-weak absorption-reflection-reabsorption” shielding mechanism. The optimized foam features an ultra-high average EMI SE T of 104.02 dB, including an absorption SE (SE A ) of 103.95 dB and a reflection SE (SE R ) of only 0.07 dB. Notably, its average absorption coefficient (A) reaches 0.984 (the maximum value of 0.998), surpassing most of the EMI shielding composites reported to date. By learning the experimental datasets of EMI SE T and A at different frequency, Fully Connected Neural Networks (FCNN) exhibits excellent prediction accuracy on unseen samples, with average Root Mean Square Error (RMSE) values of only 1.251 and 0.014 for EMI SE T and A, and average errors of less than 2.82 % and 1.52 % for EMI SE T and A, respectively, suggesting that it is highly applicable for this work and can effectively reduce the experimental costs. This work offers an innovative strategy for fabricating high-efficient absorption-dominant EMI shielding materials and reduce experimental consumption.