Adsorption in nanoporous materials is pivotal for addressing global challenges in gas storage, separation, sensing, catalysis, and atmospheric water harvesting. Consequently, molecular simulations are essential for understanding adsorption mechanisms and accelerating material discovery. Key thermodynamic descriptors, such as adsorption isotherms, density distributions, and Henry constants, are particularly valuable for high-throughput screening and predicting separation performance. Recently, machine learning potentials (MLPs) have emerged as a powerful tool, offering near-ab-initio accuracy with high computational efficiency. While MLPs have been extensively applied in molecular dynamics simulations, their integration into Monte Carlo (MC) simulations for adsorption remains largely untapped. This limitation arises primarily because mainstream MC simulation codes are designed for empirical force fields and lacks native support for MLPs. In this work, we developed a flexible Python package, high-throughput vniversal learning-enabled utility for adsorption (HULU), to bridge this gap. We present the first demonstration of calculating full adsorption isotherms using state-of-the-art foundation MLPs (MACE-MATPES-PBE-0, NEP89, and ORB v3). Furthermore, we systematically benchmark these models against standard baselines, such as available experimental or density-functional-theory calculation data, and elucidate the microscopic origins of deviations in the simulation results. Ultimately, HULU paves the way for incorporating high-fidelity MLPs into high-throughput screening workflows, significantly enhancing the predictive design of nanoporous materials for energy and environmental applications.

A kind of two-dimensional (2D) metal-organic framework (MOF) material, Cu-meso-tetrakis (4-carboxyphenyl) porphine (Cu-TCPP) nanosheets with wrinkled and flat morphologies are used as building blocks to assemble membranes by vacuum filtration (VF) and electrophoretic deposition (EPD) as energy-efficient nanofiltration (NF) membranes to remove dyes from water. Since the nanosheets with wrinkled structure can provide additional water transport channels, thereby increasing the water permeance, in the premise of a high rejection (> 97.0%) for the dye brilliant blue G (BBG) (1.60 nm × 1.90 nm), the water permeance of the membrane assembled by the wrinkled nanosheets (~ 1170 nm) is about 4 times that of the membrane assembled by the flat nanosheets (~ 530 nm), reaching 16.39 L·m⁻²·h⁻¹·bar⁻¹. Additionally, the use of the relatively flat nanosheets and the membrane preparation method of electrophoretic deposition is more conducive to stack nanosheets orderly and reduce defects. Therefore, the water permeance of the membrane prepared by EPD (~ 1170 nm) with flat nanosheets is about twice that of the membrane prepared by VF (~ 530 nm), achieving 9.40 L·m⁻²·h⁻¹·bar⁻¹ with similar rejection (> 97.0%) of dye evans blue (EB) (3.10 nm × 1.20 nm). Furthermore, these membranes still exhibit good separation performance at high pressure of 0.6 MPa. Nanosheets with diverse structures and various membrane fabrication processes provide new directions for the separation performance optimization of 2D MOF materials for water purification.

Two-dimensional (2D) material-based membrane separation has attracted increasing attention due to its promising performance compared with traditional membranes. However, in-depth understanding of water transportation behavior in such confined nanochannels is still lacking, which hinders the development of 2D nanosheets membranes. Herein, we investigated water confined in graphene or MoS₂ nanochannels by molecular dynamics (MD) simulations and found water’s diffusivity always varied linearly with their mean square displacement along z direction ( $⟨$Δz² $⟩$) when system variables (e.g., water molecules’ number, channel height, nonbonded interaction parameter, and harmonic potential constraining water’s z-coordinate) changed. Such linear correlation applies to different water models and different force fields (FFs) of channel walls (e.g., different Lennard–Jones parameters or even flexible FF), no matter whether water molecules form 3-, 2-, or quasi-2-layer structure in the nanochannel. This indicates, though water molecules’ motion along z direction (z-fluctuation, confined within 1 nm) and that in xy plane (xy-diffusion) are entirely different, they are tightly coupled: Violent z-fluctuation would produce more transient void to facilitate xy-diffusion, which is to the sharp contrary of bulk water, where motions in x, y, and z directions are symmetric, but independent. Our work could help design high performance 2D nanochannels and discover more novel principles in nano-fluidics and membrane separation fields.