Improved Carbon Nanotube Growth Via Autonomous Jump Regression Targeting of Discontinuous Behavior

Abstract

In supported catalyst carbon nanotube (CNT) growth, the reduction of an oxidized metal catalyst enables growth, but reduction also accelerates catalyst deactivation via Ostwald ripening, agglomeration, and other effects.Here, we conducted autonomous experiments guided by a hypothesis-driven machine learning planner based on a novel jump regression algorithm.This planner iteratively builds a model to identify discontinuities in the response surface, such as those created by a material phase change, and selects further experiments to improve the fit and reduce uncertainty in its model.This led us to identify input conditions resulting in the greatest CNT yields as a function of the driving forces of catalyst reduction in a fraction of the time and cost of conventional experimental approaches.We identified discontinuous jumps in CNT growth resulting in the largest observed yields in narrow and distinct regions of thermodynamic space for different thicknesses of iron catalyst.We also observed the longest growth lifetimes and a greater degree of diameter control at both jumps.We believe that conducting CNT growth at the jumps identified in thermodynamic space optimizes catalyst activity, since the relatively oxidizing environment inhibits Ostwald ripening-induced deactivation, keeping catalyst nanoparticles smaller and more numerous.This work demonstrates the strong interplay between catalyst morphology and reactor conditions in CNT synthesis, identifies optimal thermodynamic conditions for the production of small diameter single-walled CNTs, establishes a thermodynamic framework for understanding other metal catalysts in CNT growth, and demonstrates the capability of iterative, hypothesis-driven autonomous experimentation to greatly accelerate materials science.

Keywords: carbon nanotube synthesis, autonomous research, ARES, catalyst control, oxidation/reduction