High-throughput synthesis of silver nanoplates and optimization of optical properties by machine learning

Abstract

High-throughput synthesis of silver nanoplates was performed by two-step reduction via the seed particles to clarify how the absorption spectrum is controlled by the concentrations of AgNO3 and citrate for the seed particle synthesis and that of the seed particles for the nanoplate synthesis. A wide variety of the concentrations, whose combination amounted to 486 conditions, was explored with an automated liquid handler, and the absorption spectrum was measured with a microplate reader. The spectrum was decomposed into two peaks, one around 400 nm from the spherical seed particles and the other at longer wavelength from the nanoplates. The nanoplate peak’s wavelength, width, and height were expressed as functions of the concentrations of AgNO3, citrate, and seed particles by non-linear support vector regression. This enabled us to predict optimum conditions for synthesizing nanoplates with high and narrow absorption peaks at desired wavelengths ranging widely from 450 to 900 nm.