Machine learning for the regulation strategy and mechanism of the integrated growth of carbon nanotube arrays

Abstract

Carbon nanotube (CNT) arrays are an attractive material, and achieving their low-cost and controllable growth is a difficult task. In this study, a strategy for the integrated growth of CNT arrays is proposed to control the deep fusion of catalyst preparation and CNT growth with full parameters. The experimental results of more than ten kinds of combinations controlling the reaction process from three aspects of substrate roughness, infiltration time and preheating temperature are described in detail. The proposed multi-objective full-parameter particle swarm optimization algorithm is used for adaptive iteration, and a multi-factor optimal CNT diameter and dispersion mathematical model is established. The error between the experimental diameter of the CNT arrays and the calculated value of the optimal mathematical model is 5.8%. Finally, inspired by plants growing towards the sun, the reaction-energy (entropy) growth mechanism is proposed, which provides a basis for the synergistic regulation of reaction and energy. The integrated growth of the CNT arrays provides an important reference for accurate, energy-saving and efficient prediction and regulation of the morphology and performance of the CNT arrays. Meanwhile, it also has great potential application value in electronic devices (such as carbon-based chips).