基于浸润性图案化SERS基底的色素分子指纹探测技术研究

饮料色素滥用已成为影响食品安全的风险要素，严重危害了人民健康。色素的快速检测技术对饮料安全质量控制具有重要意义。表面增强拉曼光谱（Surface Enhanced Raman Spectroscopy，SERS）技术具有高效、无损等优势，在食品安全快速分析中受到广泛关注。高灵敏低波动的SERS基底构建以及智能光谱识别方法的开发是推进SERS技术实际应用的关键。本课题创新地提出浸润性图案化调控方法，藉由液滴的咖啡环效应及界面的浸润性差异，自发驱动贵金属纳米结构及待测分子富集于预先设计的图案化区域中，从而获得高性能SERS基底，提高了制样-检测的可重复性并降低了现场化操作的难度；提出基于卷积神经网络的色素特征提取识别方法，构建低数据资源条件、计及台间仪器差异的SERS指纹识别模型。预期提供一种灵敏、特异、智能化的色素快检支撑技术，为SERS技术在复杂食品基质中的非法物质微痕量检测提供科学依据。

Abuses of colorants are serious threats to food safety, which might lead to severe problems to human health. It is of critical importance to develop novel methods for rapid and accurate detection of trace amount of colorant additions to ensure food quality and safety. Extensive attentions have been paid to surface enhanced Raman spectroscopy (SERS) as a promising analysis method for efficient, non-invasive detection. The development of sensitive, uniform SERS substrates and intelligent spectral identification methods are the keys to practical applications of SERS technique. In this project, a wettability-patterned modification method is proposed to develop sensitive SERS substrates for detection of important colorants in drinks. The noble metal nanoparticles and analytes would be driven in the pre-designed patterns via the coffee-ring effect of droplet and the difference of interfaces wettability, which might improve the repeatability and efficiency of on-site operations. Moreover, an intelligent convolutional neural network (CNN) model considering the conditions of low data resources and instrumental differences would be constructed to recognize fingerprint information of colorants. This project will establish a sensitive, specific, and intelligent detection-identification methodology for colorants, and provide a theoretical basis for the applications of SERS on analysis of complex matrices.