Machine learning methods for aerosol synthesis of single-walled carbon nanotubes

This work is devoted to the strategy towards optimal development of multiparametric process single-walled carbon nanotube (SWCNT) synthesis. Here, we examine implementation machine learning techniques and discuss features dataset size density for aerosol chemical vapor deposition method with a complex source. We employ 369 points, comprising synthesis parameters (catalyst amount, temperature, feed sources) corresponding characteristics (yield, quality, structure, optoelectrical figure merit). Assessing performance six methods on dataset, demonstrate Artificial Neural Network be most suitable approach predict outcome processes. show that even 250 points inhomogeneous distribution input enough reach an acceptable Network, wherein error likely arise from experimental inaccuracy hidden uncontrolled variables. believe our will contribute selection appropriate regression algorithm controlled further autonomous system “on-demand” SWCNT production.