Au@Ag double-ended nano-pencils combined with data-enhanced methods for quantitative detection of psychotropic drugs

The development of efficient and accurate methods for the detection of psychotropic drugs is essential for the prevention of drug abuse. In this study, we designed and prepared Au@Ag double-ended nano-pencils (DENPs) combined with data enhancement algorithms for quantitative detection of psychotropic drugs. Au@Ag DENPs was prepared by overgrowing Ag shells on the surface of Au bipyramids, which not only have two tip corners (the lightning rod effect), but also generates strong hotspots region at the gap of the neighboring nanoparticles (plasma resonance effect). The Au@Ag DENPs outperform the Au bipyramids in both theoretical simulations and experimental SERS performance, with an experimental enhancement factor as high as 11.4. This SERS substrate was applied to the detection of promethazine hydrochloride with a detection limit as low as 0.1 ppb. In order to overcome the limitations of machine learning algorithms to analyze a small amount of SERS spectral data, multiple data enhancement algorithms were used to expand the SERS spectra. It was shown that the generative adversarial networks (GAN) data enhancement algorithm combined with the principal component analysis-partial least squares (PCA-PLS) algorithm realized the accurate detection of promethazine hydrochloride, with the coefficient of determination between the predicted concentration and the actual concentration as high as 0.978, and the root-mean-square error as low as 47.56. Therefore, Au@Ag DENPs substrate combined with the data enhancement algorithm is promising to be applied in the rapid on-site detection of psychotropic drugs.