▎ 摘　　要

Pore density (PPI) plays an important role in membrane performance; however, its effect on separation factor is always ignored and its quantitative relationship to membrane flux has not been established so far. Herein, for the first time, we explored the effect of PPI on the separations of water/ethanol and methanol/ethanol mixtures through graphene oxide (GO) membranes by molecular dynamical simulations. The fluxes of water (J(w)) and methanol (J(m)) are found to linearly increase with PPI owing to the enhanced diffusivity and the reduced permeation barrier. Their mathematical relationships are quantified as J(w) = 805.2 x PPI + 10.7 and J(m) = 263.8 x PPI -7.8 (PPI >= 0.11 nm(-2)). While the corresponding separation factors are sacrificed by large PPIs because the decreasing per-pore coverage of the pre-selected molecules vanishes the ethanol-resistant region at the membrane interface. Therefore, to achieve good separation performance, the moderate PPIs of 0.22 nm(-2) and 0.11 nm(-2) are well-judged for that GO membrane with 2.4 angstrom sized apertures to maintain not only considerable membrane fluxes but also a 100 wt% water content in permeate for water/ethanol separation and a high methanol separation factor of 18.0 for methanol/ethanol separation, respectively. This computational study quantitatively uncovers the crucial role of PPI in membrane separations and provides a theoretical basis to design high-performance GO membranes for ethanol dehydration and methanol recovery.