Using Machine Learning to Predict the Statistical Distribution of Metal Nanoparticles

Noble and transition metal nanoparticles (NPs) of sub-100 nm diameter range are widely explored for a variety of sensing and electronics applications including printed electronics. They are prepared using thin film dewetting process, which results in high density of NPs in the order of >10
²⁰
/cm
²
. This process is highly sensitive to process conditions and a universally applicable model is highly desirable to predict the NPs fabrication and distribution. This work presents a machine learning algorithm for the statistical distribution of NPs and resulting pattern with respect to various process conditions. To this end, 3-4 nm thick Au thin films were used to produce three different diameter distributions at temperature of 900°C for 15 minutes. The statistical distributions were simulated using Gaussian and Gamma functions to evaluate the pattern using numerical values. The statistical values of the particle distribution extracted from simulations matches with the one exhibited by the dewetting process. The presented study shows the potential of using machine learning tools in micro/nanofabrication and printed electronics.