Au-decorated ZnO nanorod arrays for SERS-active substrates towards trace detection and classification of pentaerythritol tetranitrate

Abstract This work proposed a systematic fabrication process towards high-performance SERS-active substrates as the chemical sensors, based on a combination of the hydrothermal-synthesized ZnO nanorods (ZnO NRs) and the decorated Au nano-surface. The fabrication process was first focused on the preparation of the Au/ZnO NRs hybrids. The ZnO NR templates were grown on silicon (1 0 0) wafer substrates by the hydrothermal method. Then, the Au films were deposited on the ZnO NR templates by the DC pulse magnetron sputtering technique. The obtained SERS substrates were utilized towards trace detection and classifications of pentaerythritol tetranitrate (PETN), which is commonly used in industrial explosives and improvised explosive devices. The results from field-emission electron microscopy (FE-SEM) and transmission electron microscopy (TEM) indicated the changes in physical morphologies and crystal structure, respectively, based on the variation of the fabrication parameters. The FE-SEM results showed that well-vertical ZnO NRs demonstrated the length of the nanorods before and after the Au decorations. In addition, the hexagonal structures, before and after the Au decorations, were observed with the average diameter of 50 and 181 nm, the nanorod distances of 178 and 81 nm, respectively. The TEM results also confirmed the average size and d-spacing of the ZnO NRs and Au/ZnO NR hybrids, with validated elements/compounds from the EDX mapping technique. Furthermore, the sample surfaces were observed with Raman mapping spectroscopy towards uniformly distributed examination. Finally, the fabricated samples were investigated with trace concentration of PETN towards the limit of detection. The results that our SERS substrates could detect PETN of less than 1 µg/ml, which indicated excellent performance of the trace detection of the explosive substance. The SERS substrates were further used to successfully classify different sources of the PETN substances, with 10 mg/ml, based on the machine learning approach with the principal component analysis.