Study on the self-diffusion coefficients of binary mixtures of supercritical water and H, CO, CO, CH confined in carbon nanotubes.

Nano-confined binary mixtures are prevalent in the chemical industry, geology, and energy sectors. Investigating their mass transfer behavior can enhance process intensification. This study examines the confined self-diffusion coefficients of binary mixtures of supercritical water (SCW) with H, CO, CO and CH in carbon nanotubes (CNT) using molecular dynamics (MD) simulations at temperatures of 673-973 K, a pressure of 25-28 MPa, solute molar concentrations of 0.01-0.3, and CNT diameters of 9.49-29.83 Å. We developed a novel machine learning (ML) clustering method to optimize abnormal MSD-t data, effectively extracting information and providing algorithmic enhancements for calculating the diffusion coefficient. We analyzed the effects of temperature, solute molar concentration, and CNT diameter on the confined self-diffusion coefficient and energy input. Results indicate that over 60 % of the solute energy input derives from the Lennard-Jones effect of the CNT wall. The confined self-diffusion coefficient of solutes increases linearly with temperature, saturates with increasing CNT diameter, and remains relatively constant with varying concentration. Finally, based on the unique relationship between CNTs and the confined self-diffusion coefficient, we developed a new mathematical model for prediction. The regression line exhibits an R value of 0.9789, offering a new method for predicting the properties of nano-confined fluids.