A first-principles study on water flow through single-walled carbon nanotubes using artificial neural network method.

Ab initio molecular dynamics simulations are done to elucidate the electronic structure and properties of water/single-walled carbon nanotubes (SWCNTs) systems. The artificial neural network (ANN) approach and statistical methods are then used to model and analyze these properties. The ANN method substantially speeds up the ab initio electronic structure calculations and has superior accuracy in mimicking the results of such calculations. We aim to understand the effects of CNT chirality, temperature, and CNT flexibility on the water diffusion inside SWCNTs. In this regard, the CNT is fixed implies that the position of CNT is kept constant during the diffusion process. Statistical analysis of results shows that there is a nonmonotonic variation of diffusion length of water with respect to the CNT chirality. However, an increase in temperature and rigid CNTs accelerate the water diffusion.