Characterization And Optimal Design Of Vanadium Dioxide Terahertz Metamaterials

Due to the unique electromagnetic properties of terahertz(THz)waves,such as low energy and coherence,they have broad potential applications in fields such as radar,communication,and non-destructive testing.However,most natural substances are not sensitive to THz radiation,and using metamaterials to achieve active control of THz waves has become a new research direction.For traditional metamaterials,once their electromagnetic properties are determined after preparation,this brings many limitations to the practical application of the device.Therefore,research on tunable metamaterials is of great significance for practical applications.This paper focuses on research on tunable THz metamaterials and data-driven optimization design,respectively.In this study,a switchable multifunctional THz metamaterial device based on vanadium dioxide(VO2)is proposed,and metamaterial structures with grating strip and asymmetric splitring resonators(SRRs)were designed,both of which can achieve tunable multifunctionality.VO2 material has phase transition characteristics and undergoes a dielectric-to-metal transition with temperature changes.Based on this,the metamaterial device with a grating strip structure achieves tunable multi-band perfect absorption and transmission electromagnetic properties.The metamaterial device with an asymmetric SRRs structure exhibits dual-band electromagnetic induced transparency(EIT)effects and multi-band perfect absorption electromagnetic properties.Meanwhile,due to the group delay characteristics of EIT effects,the device realizes tunable slow light effects.The dual-band EIT effect of the asymmetric SRRs structure was experimentally verified,and the experimental and simulation results matched well.The traditional design method based on computational electromagnetics and numerical simulations suffers from the problems of high design difficulty and long computation time.This paper proposes a data-driven research method based on deep learning,which significantly reduces the design cost and computational resources.The deep learning network model was used for in-depth research on the spectrum prediction and reverse on-demand design of VO2 THz metamaterials.The predicted spectra of the model output have good consistency with simulation results,and the metamaterial structures designed by reverse on-demand design meet the design specifications.On this basis,a small sample design task based on transfer learning algorithm was studied,which reduces the demand of deep learning model for a large amount of high-quality annotated data and realizes the tunable spectrum prediction of VO2 metamaterials.Compared with traditional deep learning research methods,this approach significantly reduces computational resources and design costs,and can quickly and accurately obtain the spectrum response of VO2 THz metamaterials.Therefore,this research provides a new idea for the design and application of THz functional devices.