Prediction length of carbon nanotubes in CVD method by artificial neural network

Most features of carbon nanotubes such as electrical, mechanical and thermal properties are depended on the length of them. Thereby, the applications of carbon nanotubes significantly developed by controlling this key factor. In this paper, we predict the length of carbon nanotubes in chemical vapor deposition (CVD) by using an artificial neural network. First, the effective parameters in CVD for synthesizing carbon nanotubes include the thickness of catalyst, temperature and time of heat treatment, rate of reactant gas; collected from various studies and they were determined as the input. Then, the length of carbon nanotube considered as the output of the artificial neural network. A Feed-forward backpropagation network was designed with 16 and 12 neurons in the first and second hidden layers, respectively. The predicted outcomes were very close to the experimental results, and the created model with 5.6% root mean square error was able to predict the length of carbon nanotubes. It is expected that the designed model can be helpful for researchers to adjust and regulate the suitable parameters among different effective variables in the CVD method. Furthermore, the result of the sensitivity analysis showed that the temperature and rate of reactant gas and thickness of catalyst have the highest impact on the length of carbon nanotubes, respectively.