Machine Learning Derived TiO2 Embedded Frequency Selective Surface for EMI Shielding Applications

This work intends to offer the design, synthesis, and testing of a polarization-insensitive, angular stable, high shielding effectiveness (SE) nanoparticles embedded frequency selective surface (FSS) structure. Dielectric characterization of the titanium dioxide (
$\text {TiO}_{{2}}$
) nanoparticles with different particle sizes is investigated using the waveguide-based microwave measurement setup. SE of 8.4 dB is achieved with a 1.8 mm thick
$\text {TiO}_{{2}}$
layer. Further, an efficient equivalent circuit model (ECM) aided with the deep neural network (DNN) technique is used for material selection, parameter extraction, and understand the physical mechanism of the shielding structure. With the coupling of FSS with
$\text {TiO}_{{2}}$
nanoparticle substrate, the entire
${X}$
-band is shielded with a minimum SE of 19.52 dB and a mean SE of 26.9 dB. Additionally, angular stability up to 70° is achieved by cascading a compensation layer, and due to the symmetry of the offered FSS, the designed shielding structure is polarization insensitive. Results obtained from the ECM-backed DNN approach are verified first with full-wave simulation and then by fabrication of the prototype and corresponding measured results at both normal and oblique incidence. Good agreement among all verifies the potential of the aforesaid technique for
${X}$
-band shielding applications.