3D Spiky Needle‐Clustered Ag@Au Plasmonic Nanoarchitecture for Highly Sensitive and Machine Learning‐Assisted Detection of Multiple Hazardous Molecules

Abstract

Abstract To develop a field applicable hazardous molecular detection system, highly sensitive and multiplex detection capability is required for practical utilization. Here, a paper‐based 3D spiky needle‐clustered gold grown on silver (Ag@Au) plasmonic nanoarchitecture (3D‐SNCP) is fabricated through whole solution process. The developed substrate is investigated by scanning electron microscopy (SEM), transmission electron microscopy (TEM) and X‐ray diffraction (XRD) to find out morphological development mechanism. Also, finite‐domain time difference (FDTD) simulation is conducted for the observation of electromagnetic field (E‐field) distribution. After surface‐enhanced Raman scattering (SERS) characterization, the 3D‐SNCP is utilized for ultra‐sensitive and multiplex hazardous molecular detection, such as bipyridine pesticides including paraquat (PQ), diquat (DQ), and difenzoquat (DIF). Then, each of pesticide molecular Raman signals are trained by a machine learning technique of multinomial logistic regression (MLR), followed by multiplex classificationf of blank, PQ, DQ, DIF, and four mixture types of each pesticide, spiked in real agricultural matrix. The developed 3D‐SNCP substrate combined with the machine learning method successfully verifies the multiple pesticides and it is expected to be applied for various hazardous molecular detection in much complicated matrix environments.

Keywords

• machine‐learning
• nanoarchitecture
• pesticide detection
• plasmonic materials
• surface‐enhanced Raman scattering