Terahertz waves are electromagnetic waves with frequencies of 0.1-10.0 THz. They have the characteristics of wide bandwidth,strong penetration,and low photon energy. Notably,the energy levels of the terahertz spectrum correspond to the rotational and vibrational energy levels of several biological macromolecules. Therefore,the terahertz spectroscopy technology can be used to study the properties of biomolecules,such as their molecular structures and their molecular interactions with the surrounding environment. Because the wavelength of a terahertz wave (0.03-3.00 mm) is not in the same order of magnitude as the characteristic size of common biological macromolecules,it is difficult to produce sufficient interaction between the trace levels of biological macromolecules contained in the sample and the terahertz wave,and the weak change in the terahertz spectral line is difficult to capture. Metamaterial is an artificial material whose electromagnetic properties can be manipulated artificially. By combining terahertz spectroscopy technology with metamaterials,the small disturbance to the metamaterial caused by trace-level objects can result in significant changes in the spectra of metamaterials,and thus make high-sensitivity detection of biomacromolecules possible. In this study,a resonant terahertz metamaterial is used to enhance the interaction between terahertz wave and the determinand,and a bovine serum albumin (BSA) solution is selected as the analyte. An efficient BSA sensor is constructed using terahertz spectroscopy technology. The relationship between the concentration of the BSA solution and the resonance frequency offset of the sensor is analyzed,and the limit of detection of the sensor is examined.

To achieve rapid and highly sensitive detection of BSA solution,the unit structure designed by Sengupta et al. is adopted. Because of the C4 symmetry of the structure,the metamaterial-based sensor is insensitive to polarization. The electromagnetic properties of the metamaterial are simulated by full-wave numerical simulation. According to the simulation results,the above structure has resonance absorption near 0.8 THz. When the refractive index of the material at the surface of the metamaterial changes,the resonance frequency of the metamaterial shifts. By combining the terahertz spectrum technology,the relationship between the resonance frequency shift of the metamaterial and the concentration of determinand is established. The schematic of the experimental setup is shown in Fig. 1. First,the transmission spectrum of the metamaterial without a determinand is measured with a terahertz spectrometer and used as the background. Second,quantitative BSA is dissolved in distilled water to prepare a certain concentration of BSA solution,and the 20 μL solution is collected through a microsampler and dropped onto the metamaterial. The metamaterial is heated at 70 ℃ for 10 min. The transmission spectra of the metamaterials are measured using a terahertz spectrometer to provide a sample group. During the experiment,to reduce error,the linear weighted average processing is carried out on the data of the adjacent points in a single scanning. The final data for each sample is the average of the three scanning results. A second-order Gaussian fitting method is used to fit the transmission spectral data of the background and sample groups nonlinearly,and the resonance frequency shift of the metamaterial is obtained by comparing the fitted curves.

A resonance absorption peak of approximately 0.8 THz for the metamaterial without a determinand is found (Fig.2). According to the fitting curves,a BSA solution with a volume mass of 2 mg/mL can induce a 10.43-GHz redshift in the resonant frequency of the metamaterial. Using the control variable method,several groups of solutions with different volume mass values (2,3,4,and 8 mg/mL) are configured,and their terahertz transmission spectra are measured and compared with the transmission spectrum of the bare metamaterial (Fig.3). With a gradual increase in the concentration of the determinand,the resonance frequency of the metamaterial gradually moves in the lower frequency direction (Table 1). When the solution volume mass is in the range of 2-8 mg/mL,a linear relationship exists between the resonance frequency shift of the metamaterial and the volume mass of the determinand (Fig.4). The concentration of the BSA solution is determined according to a linear relationship. When the concentration of the BSA solution is reduced continuously and when the volume mass is decreased to 0.3 mg/mL,the frequency offset of the resonance chip is 0.87 GHz. When the volume mass is further decreased to 0.2 mg/mL,the resonance shift is less than 0.04 GHz,which is significantly lower than the sampling step size of the spectrometer (Fig.5 and Table 2). Furthermore,0.3 mg/mL exhibites the lowest detection limit. By optimizing the experimental environment,considering the change in transmittance and resonance frequency,and shortening the frequency step size,the detection sensitivity can be further improved,the limit of detection can be reduced,and the sensor specificity can be enhanced.

Based on metamaterials and terahertz spectroscopy, an efficient sensor for biological macromolecules is constructed. The experimental results show that the addition of determinand can cause a significant shift in the terahertz transmission spectrum and the variation in the resonant frequency of the sensor increases with an increase in the concentration of the determinand. When the volume mass of the BSA solution is in the range of 2-8 mg/mL, the offset of the resonant frequency is linearly correlated with the volume mass, demonstrating the potential of the proposed sensor for detecting the substance concentration. In addition, the limit of detection is 0.3 mg/mL. In conclusion, the proposed BSA sensor has high sensitivity, simple operation, and high efficiency. Subsequently, by further optimizing the materials and parameters of metamaterials, comprehensively utilizing the position and amplitude information of the resonance peak, and improving the test environment, it should be possible to obtain biological macromolecule sensors with higher specificity and sensitivity.