Optimizing silver nanowire synthesis: machine learning improves and predicts yield for a polyol, millifluidic flow reactor

This study highlights optimizing polyol reaction conditions to produce 100% silver nanowire (AgNW) yields (AgNWs count/all nanostructure count) using a millifluidic flow reactor (MFR). AgNWs of uniform length and diameter offer potentially low-cost, transparent, and flexible conductors. MFRs produce AgNWs with superior uniformity, yield, and concentration due to the reduced dimensions of the reaction environment. A statistical design of experiments (DoE) considering polyol reaction temperature and the three reagent concentrations optimized the process. The AgNWs are characterized by scanning electron microscopy (SEM) to calculate the yield of AgNWs per reaction. After completing the DoE, calculated yields are put into Minitab statistical software for analysis. Minitab discovered the optimal reaction conditions to be T = 170 °C, [AgNO3] = 0.177 M, [CuCl2] = 6.05 mM, and [PVP] = 0.224 M, with an R 2 value of 85%. Results of the DoE were imported into supervised decision tree (DT) and random forest (RF) machine learning (ML) algorithms. The DT and RF predicted yields of AgNWs given reaction temperature and reagent concentrations with 96.9% and 97.5% accuracy, respectively. The optimal polyol reaction conditions synthesized 100% AgNW yield with average concentrations of 16 mg/mL, lengths of 32 µm (σ ± 3.5 µm), diameters of 68 nm (σ ± 12 nm), and aspect ratios of 475. © 2023, King Abdulaziz City for Science and Technology.