Design of an ultra-broadband terahertz absorber based on a patterned graphene metasurface with machine learning

The development of patterned graphene metasurface absorbers (PGMAs) offers potential solutions for achieving light weight, thinness, wide absorption bandwidth, and tunable terahertz (THz) absorption properties. In order to optimize the absorption properties of PGMA, the absorption spectrum is usually used as an important evaluation metric, which can provide many important properties of PGMA such as absorption bandwidth, absorption magnitude, etc. However, analysis of the absorption spectra corresponding to a large number of variable structural parameters is required when designing the structure, which consumes a lot of resources, since the electromagnetic (EM) wave absorption in PGMA involves complex impedance matching and electric field excitation processes. To address this issue, this study proposes a machine learning (ML) approach based on the random forest (RF) algorithm to predict the absorption bandwidth and structural parameters for designing PGMA, reducing the need for unnecessary numerical simulation and spectra analysis time. With the RF model, a very large effective absorption bandwidth of 3.83 THz and a perfect absorption bandwidth of 2.52 THz are predicted with the R-2 of 0.938 and 0.907, and the forecast absolute percentage errors (APEs) are only 1.56% and 1.16%, respectively, which is much better than other classical ML algorithms. Furthermore, the proposed PGMA has the advantages of being thin, polarization insensitive, with a large stable incident angle of 60 degrees, and excellent electrical tuning capabilities. This study provides a feasible and effective approach for the sophisticated design of complex systems related to EM wave propagation of absorption, reflection, and transmission.