Fingerprint Spectral Inversion-Enabled Terahertz Metasurface for Multidimensional Detection of Trace Antibiotics

Antibiotic abuse-induced residues pose a severe threat to human health, yet traditional detection methods suffer from large sample consumption and complicated pretreatment, which fail to meet practical needs. Terahertz (THz) spectroscopy, with its molecular fingerprinting capability, holds great promise for drug detection. Here, we propose a THz metasurface for accurate qualitative and quantitative analysis of antibiotics. First, quantum chemical calculations reveal the spectral response of antibiotics in the THz band and the formation mechanism of characteristic fingerprint peaks. Then, leveraging the metasurface resonance spectral dip induced by molecular fingerprint peaks, we extract the specific information via spectral inversion. After that, a joint feature space is constructed by combining resonant frequency shift and differential absorption intensity, and the characteristic frequency is further integrated to form a multidimensional detection system. As a proof of concept, the metasurface realizes qualitative and quantitative detection of two antibiotics with a limit of detection as low as 1.11 μg/mm2 and effectively identifies trace norfloxacin in complex mouse blood matrices. Combined with machine learning, the metasurface efficiently identifies multiple antibiotics on our experimental dataset. This work presents a label-free approach enabling direct qualitative and quantitative detection of trace antibiotics via a single THz metasurface, offering new insight for trace identification of drug molecules, with potential application prospects in biomedical detection applications.