Application of bat algorithm in carbon nanotubes growing process parameters optimization

Abstract

Since their discovery, carbon nanotubes (CNTs) have become a subject of intense research for their potential use in various applications. Chemical vapor deposition (CVD) process is the most common method used to grow CNTs. However, the growing process suffers many difficulties in finding the optimal process parameters. Applying computational intelligence methods is a possible solution for optimization problems to reduce using conventional methods and experimental runs. In this work, a combination between bat algorithm (BA) and response surface methodology (RSM) is proposed to solve CNTs process parameters optimization problem. The study aims to maximize the CNTs yield percentage for mass production in two different datasets. BA search process is based on the objective function developed by RSM which represents the prediction mathematical model of growing process parameters. The optimized parameters from datasets are reaction temperature, reaction time, catalyst weight, and methane partial pressure. The algorithm search process was conducted with parameters tuning at different setting values to improve the algorithm’s performance and CNTs yield value. Different evaluation metrics were applied to compare the experimental results. The results have shown that BA has an efficient search performance and obtained better CNTs yield result than RSM in one of the datasets with 21% improvement of CNTs yield value. Besides, BA has shown a fast and stable convergence. Finally, the result was validated and found reliable to be used in real laboratory experiments.