The optimal substrate choice for effective biosensors based on THz metasurfaces

Abstract

This article explores the impact of substrate choice on the sensitivity of sensors that utilize metallic terahertz metasurfaces as the actuating element. While terahertz metasurfaces represent a rapidly evolving field, fundamental research remains essential and highly impactful. A critical component of any metasurface is the dielectric substrate on which it is fabricated - a factor that holds significance across all spectral ranges. This work focuses specifically on metallic terahertz metasurfaces operating in transmission mode. It provides an overview of various substrates (Ge, Si, SiO₂, TPX) and discusses design principles for such metasurfaces, including magnetic and electric dipole surface plasmon resonances, Fano resonances, and quasi-bound states in the continuum. We initially simulated the structures using COMSOL Multiphysics software and then confirmed the results experimentally by detecting various concentrations of bovine serum albumin. Our study systematically examines how real-metal modeling and substrate selection influence the Q-factor and sensing performance, in contrast to earlier research that either fixed the substrate type or modeled the metal as a perfect electric conductor. This dual approach provides valuable guidance for designing high-Q, low-loss terahertz metasurface biosensors. Our results demonstrate that for all types of terahertz metasurfaces operating in transmission mode, using a low-refractive-index substrate enhances sensor sensitivity, making it the preferred choice for sensing applications. This opens new prospects for the design of high-sensitivity biosensors.

Keywords: BSA; Biosensing; Metasurface; Simulations; Substrate; Terahertz.

Fig. 1

Schematics of unit cell of the structures studied: (a) Fano resonance structure; (b) purely SRR structure; (c) asymmetrical SRR (ASRR) structure. The Fano resonance structures (a) are characterized by specific dimensions: both Ax and Ay measure 40 μm, W₁ is 8 μm, W₂ is 6 μm, L is 24 μm, H is 22 μm, and the gap width is denoted as S = 1.5 μm. The SRR (b) and ASRR (c) structures have the same dimensions. Just in ASRR (c) there is an additional gap of 1.5 μm shifted by the distance d = 5 μm; (d) cross structure has the dimension W = 11 μm, L = 65 μm, Ax = Ay = 80 μm, d = 10 μm and g = 6 μm.

Fig. 2

Optical photographs of the Fano resonance structures with the gap of 1.5 μm fabricated on two different substrates: (a) TPX (in Namarski contrast) and (b) silicon. (c) The schematics of the metallic planar Fano structure on a TPX substrate.

Fig. 3

The results of simulations of the transmission with 10 μm layer of dielectric of a given refractive index for THz MS displaying Fano resonance (Fig. 1a) for two kind of substrates: TPX (a) and silicon (b).

Fig. 4

The results of simulations of the transmission with 10 μm layer of dielectric of a given refractive index for THz MS with SRR design (Fig. 1b) for two kind of substrates: TPX (a) and silicon (b).

Fig. 5

The results of simulations of the transmission with 10 μm layer of dielectric of a given refractive index for THz MS with ASRR design (Fig. 1c) for two kind of substrates: TPX (a) and silicon (b).

Fig. 6

The results of simulations of the transmission with 10 μm layer of dielectric of a given refractive index for THz MS with cross design (Fig. 1d) for two kind of substrates: TPX (a) and silicon (b).

Fig. 7

The results of simulations of the electric field intensity and current distribution for all designs studied: (a) Fano resonance structure, (b) SRR structure, (c) ASRR structure, (d) cross structure.

Fig. 8

The shift in frequency of resonance peak (Δf) as a function of refractive index change (Δn) for various substrates, for various structures: (a) Fano resonance, (b) SRR, (c) ASRR and (d) cross. Lines depicts linear fits for simulation results (points).

Fig. 9

The comparison of the experimental results (red line) with the simulation (n = 1) for two models of metal conductivity: the Drude-Lorentz model and the PEC model. The comparison is shown for the Fano resonance structures (all with W₁ = 6 μm) featuring a gap of 1.5 μm (as in Fig. 2), fabricated on two different substrates: (a) TPX (see Fig. 2a) and (b) silicon (see Fig. 2b).

Fig. 10

(a) Transmission spectra for four samples on TPX with varying BSA concentrations. (b) Fano resonance shift as a function of BSA concentration, derived from the data in panel (a). The errors in panel (b) from fitting the data in panel (a) do not exceed 5 GHz (our TDS resolution) and are smaller than the size of the squares marking the experimental points.