Single AgNF@ZIF-8 nanoparticle for deep learning assisted SERS detection of gaseous molecule

Abstract

Random nanogaps between plasmonic nanoparticles formed by the surface tension of solution bring a great challenge in the application of Surface Enhanced Raman Spectroscopy (SERS) technique. Meanwhile, due to its natural nanogaps on the surface, single plasmonic nanoflower exhibits a unique charm for avoiding random gaps in the quantitative analysis of SERS. Herein, a single core-shell plasmonic nanoparticle consisting of an Ag nanoflower (AgNF) core and zeolite imidazolate frame-8 (ZIF-8) shell is synthesized for SERS detection of the gaseous molecule. Benefiting the porous shell of ZIF-8 to capture gaseous molecules and natural nanogaps on the surface of AgNF to enhance Raman intensity, SERS spectra of the gaseous molecule are successfully collected with high sensitivity at 10−6 M, which also exhibit high SERS stability with RSD at 2.29 % for single nanoparticle and 9.86 % for multiple single nanoparticles. In addition, two levels of automatic identification based on deep learning (DL) are also implemented in this work, whose data demonstrate artificial neural network (ANN) algorithm could identify gaseous glutaraldehyde (GA) as typical volatile organic compounds (VOCs) with high accuracy at 97.5 %. This single core-shell plasmonic MOF nanoparticle overcomes the limitations of traditional SERS detection in nanoparticle aggregating and realizes the detection of gaseous molecules, which enlarges the application of SERS technique.