A terahertz metamaterial absorber with independently tunable absorbance and bandwidth based on BP neural network optimization

Abstract

A broadband metamaterial absorber (MA) is proposed in this paper, whose parameters and absorption spectra are optimized and predicted by the back propagation (BP) neural network (NN). This MA realizes the independent modulation of absorbance and bandwidth using vanadium dioxide (VO₂) and graphene. When the conductivity of VO₂ is 2 × 10⁵ S m⁻¹ with the E_f of graphene at 1 eV, the MA can achieve more than 95% absorbance within 2.16–6.23 THz. Moreover, the absorbance modulation can be realized by temperature control with a modulation depth of 61.48%, and bandwidth modulation can be realized by voltage control with a modulation depth of 37.35%. The proposed MA allows for the modulation of both absorbance and bandwidth, addressing the limitations of modulation dimensions and presenting a new design approach for flexibly tunable MAs. Furthermore, by the use of the BP NN, the optimization of the structure can be achieved more efficiently.