Machine Learning‐Driven 3D Plasmonic Cavity‐in‐Cavity Surface‐Enhanced Raman Scattering Platform with Triple Synergistic Enhancement Toward Label‐Free Detection of Antibiotics in Milk

Abstract

The surface‐enhanced Raman scattering (SERS) technique with ultrahigh sensitivity has gained attention to meet the increasing demands for food safety analysis. The integration of machine learning and SERS facilitates the practical applicability of sensing devices. In this study, a machine learning‐driven 3D plasmonic cavity‐in‐cavity (CIC) SERS platform is proposed for sensitive and quantitative detection of antibiotics. The platform is prepared by transferring truncated concave nanocubes (NCs) to an obconical‐shaped template surface. Owing to the triple synergistic enhancement effect, the highly ordered 3D CIC arrays improve the simulated electromagnetic field intensity and experimental SERS activity, demonstrating a 33.1‐fold enhancement compared to a typical system consisting of Au NCs deposited on a flat substrate. The integration of machine learning and Raman spectroscopy eliminates subjective judgments on the concentration of detectors using a single feature peak and achieves accurate identification. The machine learning‐driven CIC SERS platform is capable of detecting ampicillin traces in milk with a detection limit of 0.1 ppm, facilitating quantitative analysis of different concentrations of ampicillin. Therefore, the proposed platform has potential applications in food safety monitoring, health care, and environmental sampling.

Fingerprinting and Differentiation of Landfill Leachate and Domestic Sewage in Surface Water with Machine Learning Based Surface Enhanced Raman Spectroscopy

“…On the other hand, the supervised learning cannot be directly performed, due to SERS spectra as multidimensional data with a larger number of variables than observations. Therefore, PCA was first conducted to extract key parameters for supervised learning. , The first five principal components with a cumulative contribution of 90% were chosen (Figure S8). As shown in Table , the correct rate of SVM, LDA, BC, and KNN were above 85%, which indicated that they could achieve accurate differentiation for LL@LW and DS@LW.…”

“…Therefore, PCA was first conducted to extract key parameters for supervised learning. 43,44 The first five principal components with a cumulative contribution of 90% were chosen (Figure S8). As shown in Table 3, the correct rate of SVM, LDA, BC, and KNN were above 85%, which indicated that they could achieve accurate differentiation for LL@LW and DS@LW.…”

SERS as a Probe of Surface Chemistry Enabled by Surface-Accessible Plasmonic Nanomaterials

“…The surface-accessible colloidal nanomaterials can be combined with a range of substrate materials to form composite materials with new functionalities and enhanced stability that can benefit SERS applications. The most straightforward way to construct composite materials containing nanoparticles is to physically deposit the colloidal nanomaterials onto solid substrate materials. , This method is particularly effective for transferring 2-dimensional Ag and Au interfacial assemblies onto hydrophilic solid materials, such as glass or metal. Since the surface-accessible Ag and Au nanoparticles are typically hydrophilic, they adhere to the surface of hydrophilic materials so they can be transferred onto solid supports via a simple dip-coating process. , Importantly, after the dip-coating process, the nanoparticle layer is held to the surface of the substrate material by favorable van der Waals interactions and does not show obvious shedding over time or when dipped repeatedly into solvents .…”

Interfacial Self-Assembly of Surfactant-Free Au Nanoparticles as a Clean Surface-Enhanced Raman Scattering Substrate for Quantitative Detection of As5+ in Combination with Convolutional Neural Networks

“…In recent years, surface-enhanced Raman scattering (SERS) spectroscopy is considered as a promising technique for the determination of trace heavy metal ions because of its special fingerprinting ability, high sensitivity, fast detection speed, and simple operation. − The bottom-up approach is a simple and cost-effective strategy for fabricating SERS substrates, which mainly relies on the assembly of nanoparticle (NP) suspensions at the interface. ,− To date, using strong adsorbent surfactants to guide particle growth has produced a variety of noble metal NPs with different sizes, morphologies, compositions, or structures. On the other hand, the surfactants facilitate the separation of nanoparticles and ensure the stability of the NPs. − Although the morphology of the synthesized NPs can be easily manipulated using surfactant molecules, their presence is problematic for the application of plasmonic sensing. − Surfactant molecules bound to the surface of NPs act as physical and chemical barriers that restrict access of the analyte to the hot spot region, resulting in reduced Raman intensity of the analyte. , However, it remains a significant challenge to avoid the influence of surfactants on the adsorption of detection molecules by converting surfactant-free NPs to stable and reliable SERS substrates.…”