The escalating atmospheric CO2 concentration demands sustainable and scalable carbon capture technologies. Deep eutectic solvents (DESs), especially those based on ionic liquids (ILs), offer a green and tunable alternative to conventional absorbents. However, their virtually infinite chemical space makes exhaustive experimental screening impossible. Artificial intelligence (AI) has emerged as a promising tool to bypass trial-and-error approaches by enabling high-throughput virtual screening and the discovery of structure–property relationships. This review provides a systematic analysis of recent advances from 2020 to 2025 in machine learning (ML) driven design of DESs for CO2 capture. A comprehensive database of CO2 absorption data for DESs reported during 2008–2025 is compiled, covering a wide range of chemical compositions and operating conditions. Feature engineering strategies are surveyed, ranging from basic molecular descriptors to advanced quantum chemical and COSMO-RS derived features. A multi-dimensional evaluation framework is introduced that assesses model performance not only in terms of predictive accuracy but also in terms of generalization capability and interpretability. The evolution of ML models from linear and kernel methods to ensemble trees and neural networks is critically discussed, with ensemble tree models (e.g., CatBoost and XGBoost) highlighted as the dominant approaches due to their generally high predictive accuracy and interpretability. Finally, key challenges such as data sparsity, model extrapolation, and integration with process simulation are identified, and future directions are proposed, including active learning platforms, physics-informed models, and generative design across scales. This review offers a systematic technical reference and a methodology to guide the intelligent design of next-generation absorbents for CO2 capture.

The widespread adoption of alternative plasticizers (APs) has raised global concerns over their environmental and health impacts. An alternative assessment is required to evaluate the safety of potential alternatives for hazardous chemicals. This study reports a comprehensive multicriteria alternative assessment (MCAA) framework that integrates 15 hazard end points across human health hazards, ecotoxicity, and environmental fate. By utilizing a hierarchical protocol to harmonize multitiered data from authoritative global sources, the MCAA assigns definitive safety ratings from Level 1 (safer) to Level 4 (more hazardous). Using this MCAA framework, we evaluated 522 currently used plasticizers, including 296 APs. The results revealed that the plasticizer percentages in Levels 1, 2, 3, and 4 were 4.60%, 12.07%, 38.12%, and 45.21%, respectively. Time-trend analysis of newly registered plasticizers (2000-2024) revealed a progressive shift toward safer alternatives, with the proportion of Levels 1 and 2 compounds increasing from 0% to 66.67%. Mobility and aquatic toxicities were identified as the key end points with the highest contributions to the overall hazard, highlighting that the ecological impacts were historically overlooked compared with the human health. The MCAA framework provides a scientifically rigorous protocol to support the environmental and health risk management of APs.

Covalent organic frameworks (COFs)-confined ionic liquids (ILs) membranes hold great promise for gas separation, yet their performance is often limited by IL aggregation within the pores. In this work, sulfonic acid-functionalized COF-confined IL (IL@TpPa-SO3H) membranes were designed to improve the dispersion of IL and enhance CO2/N2 separation through a combination of molecular dynamics (MD) simulations, Grand Canonical Monte Carlo simulations, and adsorption experimental methods. Charge density analysis reveals that the electrostatic interactions between the negatively charged sulfonic acid groups of TpPa-SO3H and the cations of IL promote a more homogeneous distribution of IL. Evaluation of cations with different alkyl chain lengths indicates that the 1-Butyl-3-methylimidazolium tetrafluoroborate ([C4mim][BF4])@TpPa-SO3H exhibits the best separation performance, achieving a CO2 permeance of 9.96 × 103 GPU and no N2 permeated across the membrane during the MD simulations with a composition of 50 vol% CO2:50 vol% N2. Radial number density and the orientational distribution analyses demonstrate that confining [C4mim][BF4] within TpPa-SO3H induces an ordered arrangement of cations and anions. This ordered structure enhances CO2 adsorption and creates shorter-range, more directional transport channels, leading to high CO2 permeance. Meanwhile, a synergy between the IL's stronger CO2 affinity and its confinement-induced pore size reduction enhances CO2/N2 selectivity.

Graphene oxide (GO) holds great promise for fabricating high-flux separation membranes, yet its inherent interlayer defects often compromise precise molecular sieving, particularly for challenging separations such as selective ammonia (NH3) capture and purification. In this work, a functionalized ionic liquid ([DBEAH][NTf2]) with protic hydrogen and hydroxyl groups as NH3-interactive sites is designed and confined within GO interlayers via hydrogen bonding and electrostatic interactions, thereby constructing tailored pathways for selective NH3 transport. The incorporated [DBEAH][NTf2] not only modulates the interlayer spacing of GO but also modifies the nanochannels to create an NH3-affinitive environment for precise recognition and separation of NH3. The resulting IL-confined GO membrane achieves an outstanding combination of high NH3 permeance of 682.17 GPU and ultrahigh ideal NH3/N2 selectivity of 1488.41 in single-gas tests. Under mixed-gas conditions (50/50 vol%, NH3/N2), the membrane maintains a high NH3 permeance of 493.82 GPU and an NH3/N2 separation factor of 248.34. Theoretical simulations further confirm that the confined [DBEAH][NTf2] facilitates rapid and selective transport of NH3 while effectively blocking N2, providing a mechanistic understanding of the enhanced separation performance. This study offers a feasible strategy for designing 2D membranes with simultaneously high permeance, selectivity, and stability for NH3 separation applications.

Ionic liquids (ILs) have exhibited great application potential in many fields due to their unique properties. Molecular dynamics (MD) simulation has been widely employed to investigate their microscopic structure. However, classical molecular dynamics simulations struggle to accurately describe the complex interactions in ILs using the existing parameterized force fields. Recently, the MD simulations based on machine learning force fields (MLFFs) trained by first-principles calculations have attracted considerable attentions due to their abilities to balance computational accuracy and efficiency. Herein, we report the Bayesian-based MLFFs which can be successfully applied in IL systems and accelerate MD simulation. The calculated atomic forces, structures, and vibrational behaviors were validated to match the accuracy of firstprinciples calculations. Properties of the imidazolium-based ILs, including density, self-diffusion coefficients, viscosity, and radial distribution functions were predicted at the extended scales. Z-bonds that describe the unique structures in ILs were analyzed and the influences of Cpositions, temperature, and solvent H 2 O on Z-bonding configurations were systematically investigated. Our results confirmed that MLFFs presented the strong feasibility to investigate the large and complex systems, especially to predict structures and properties of the ILs. And the procedure described for MLFFs provides valuable guidance for researchers who are studying ILs. (c) 2025 Institute of Process Engineering, Chinese Academy of Sciences. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

An extensive database of CO2 capacity of functional ionic liquids (ILs) was compiled containing 2482 data points of 231 ILs during 2002-2025. Interpretable Gradient Boosting Decision Tree (GBDT) machine learning (ML) models were established, including XGBoost, CatBoost, GBoost, and LightGBM, using three types of descriptors: group contribution (GC), molecular descriptors (MD), and their hybrid (MD&GC). MD&GC-based CatBoost model achieved the highest predictive accuracy (test R2 = 0.9616), good generalization capability, and minimal overfitting. SHAP analysis revealed that pressure, temperature, and molecular structures, specifically compact anions (low BCUT2D_MRHI) and delocalized cations (high MinEStateIndex), are the most important features, forming a micro-environment with weak ion networks and high anion availability for CO2 absorption. By using an up-to-date database, applying novel MD&GC descriptors, and GBDT + SHAP combination for prediction and mechanistic analysis, this work developed a novel strategy to accelerate the design of high-performance absorbents for CO2 capture.

Graphene oxide nanosheets provide excellent gas separation performance. However, the tortuous nanochannels lead to low gas permeability. In this study, a novel strategy for helium (He) separation is proposed, using an ionic-liquid-modulated nanoporous graphene oxide (ILM-NPGO) membrane that combines nanopore-drilled GO layers with confined ILs, investigated via molecular dynamics simulations. The results show that both He and CH4 can easily permeate through NPGO without selectivity. Confining ILs within NPGO significantly improves selectivity, but a 70 % volume loading is required to achieve optimal performance. A comparative analysis of membranes based on three different cation chain lengths and five anions reveals that the [C6mim][SCN]-NPGO membrane shows great potential for He separation, achieving a He/CH4 selectivity of 48 and He permeance of 2.01 x 104 GPU at a layer spacing of 0.4 nm. The cations and anions of ILs play a synergistic role in enhancing the gas separation performance. The long-chain [C6mim]+ cations dynamically regulate the pore size by distributing around the pores due to their strong interaction with NPGO, effectively hindering CH4 permeation. Meanwhile, the small and linear [SCN]- anions reduce the free volume of the membrane and interact weakly with CH4, resulting in improved selectivity. This work could guide the development of novel membranes for efficient He separation.

Bubble formation during water electrolysis can significantly hinder reaction efficiency by decreasing active area and obstructing mass transfer. In this work, using molecular simulations, the nanobubble nucleation on macro/nanoelectrodes and its impact on reaction rates were investigated. As the electrode surface changes from hydrophobic to hydrophilic, regimes of gas layers, surface nanobubbles, bulk nanobubbles, and no nanobubbles are observed sequentially, along with a volcano-shaped relationship between wettability and current density (i.e., actual reaction rate). The optimal wettability balances the reactant concentration in the reaction zone and the surface bubble formation suppression. Applying appropriately elevated electrode potentials can further enhance reaction rates when no surface bubble forms, while excessive potentials cause surface nanobubble nucleation and reduced reaction rate. Nanoelectrodes exhibit higher reaction rates than macroelectrodes with the same wettability, owing to the increased reaction probability at electrode edges. These findings highlight the importance of optimizing electrode wettability, along with using suitable electrode potentials and nanostructured electrodes, to improve water electrolysis.

Covalent organic frameworks (COFs) are promising materials for gas separation due to their excellent stability and tunable structure. However, their inherently large pore sizes often limit their application in the separation of small gas molecules, such as CO2 and N-2. In this study, we regulate the pore size by confining imidazolium-based ionic liquids (ILs) within the TpPa-1 COF using molecular dynamic simulation. A systematic investigation was conducted to explore the effects of IL loading and anion type on the CO2/N-2 separation. It was found that high IL loading effectively reduces the COF pore size and enhances the dispersion of ILs within COF, resulting in a stratified IL structure, where cations primarily occupy the corners and centers of the COF pores, while anions are staggered near the cations. Due to differences in interactions, CO2 is adsorbed near the COF pore walls, while N-2 is dispersed in the ILs, which promotes faster diffusion of N-2 compared to CO2. The size and shape of anions influence their distribution, as well as the interaction and transport of gases. Among the ILs considered, the [Bmim][BF4]/COF membrane is identified as the optimal candidate for CO2/N-2 separation.

PIEZO1 is a mechanically activated cation channel that undergoes force-induced activation and inactivation. However, its distinct structural states remain undefined. Here, we employed an open-prone PIEZO1-S2472E mutant to capture an intermediate open structure. Compared with the curved and flattened structures of PIEZO1, the S2472E-Intermediate structure displays partially flattened blades, a downward and rotational motion of the top cap, and a spring-like compression of the linker connecting the cap to the pore-lining inner helix. These conformational changes open the cap gate and the hydrophobic transmembrane gate, whereas the intracellular lateral plug gate remains closed. The flattened structure of PIEZO1 with an up-state cap and closed cap gate might represent an inactivated state. Molecular dynamics (MD) simulations of ion conduction support the closed, intermediate open, and inactivated structural states. Mutagenesis and electrophysiological studies identified key domains and residues critical for the mechanical activation of PIEZO1. These studies collectively define the distinct structural states and gating transitions of PIEZO1.

Bubble formation in electrochemical system often hinders reaction efficiency by reducing active surface area and obstructing mass transfer, yet the mechanisms governing their nanoscale nucleation dynamics and impact remains unclear. In this study, we used molecular dynamics simulations to explore nanobubble nucleation and reaction rates during water electrolysis on planar- and nano-electrodes, with systematically tuning electrode wettability through water-electrode and gas-electrode interactions. We identified distinct nucleation regimes: gas layers, surface nanobubbles, bulk nanobubbles, and no nanobubbles, and revealed a volcano-shaped relationship between wettability and reaction rate, where optimal wettability strikes a balance between suppressing bubbles and ensuring sufficient reactant availability to maximize performance. Nanoelectrodes consistently exhibit higher current densities compared to planar electrodes with the same wettability, due to pronounced edge effects. Furthermore, moderate driving forces enhance reaction rates without triggering surface bubble formation, while excessive driving forces induce surface nanobubble nucleation, leading to suppressed reaction rates and complex dynamics driven by bubble growth and detachment. These findings highlight the importance of fine-tuning wettability and reaction driving forces to optimize gas-evolving electrochemical systems at the nanoscale and underscore the need for multiscale simulation frameworks integrating atomic-scale reaction kinetics, nanoscale bubble nucleation, and microscale bubble dynamics to fully understand bubble behavior and its impact on performance.

Carbon dioxide (CO2) separation plays a vital role in environmental protection and resource reuse. A potential approach to achieving satisfactory CO2 separation is combining the benefits of ionic liquids (ILs) and membranes. In this work, the IL-LDH nanosheets were obtained by functionalized IL modifying layered double hydroxides (LDH) and used to construct the Pebax/IL-LDH mixed matrix membranes (MMMs). The X-ray diffraction (XRD), Fourier transform infrared (FTIR) spectroscopy, and X-ray photoelectron spectroscopy (XPS) spectra analysis displayed that the IL was successfully grafted to the LDH nanosheets. The characterizations of MMMs showed that the introduction of IL-LDH reduces the crystallinity of the Pebax membrane, and there are hydrogen bond interactions between fillers and Pebax. The IL modification strengthens the membranes with better CO2 affinity and improves the interfacial compatibility between Pebax and LDH. Meanwhile, the CO2-philic nanochannels formed by the slits of IL-LDH nanosheets facilitate CO2 transport in MMMs. Such inherent properties of the IL and LDH conjunctly contribute to the improvement of the CO2 separation performance of membranes. The optimum Pebax/IL(20)-LDH MMM has a high CO2 permeability of similar to 388 barrer at 30 degrees C, which is approximately 2.9 times that of pure Pebax, and the CO2/CH4 selectivity does not change significantly. Besides, the Pebax/IL(20)-LDH MMM exhibited a notable increase in CO2 permeability as the working temperature increased, ultimately attaining similar to 576 barrer at 50 degrees C.

Achieving strong interaction with the targeted composition and constructing abundant transport channels is crucial to obtain the pervaporation (PV) membrane with high selectivity and flux. Here, three ionic liquids (ILs) were screened out based on their relative selectivity and capacity targeting for bioethanol dehydration by COSMO-RS. The interaction energies analysis between ILs, EtOH, and H2O suggests that the ILs can form strong hydrogen bonds with water and disrupt the hydrogen bond in the EtOH-H2O azeotropic mixture, which is beneficial for improving the selectivity. Furthermore, driven by the multiple hydrogen bonds, electrostatic interactions, and van der Waals forces, ILs could self-assemble with polyvinyl alcohol (PVA) to fabricate the PV membrane with well-ordered micelle nanostructure, as its structure was revealed by MD simulations. The formation of the ILs-PVA micelle dramatically influenced membrane surface morphology, roughness, and water contact angle, providing an extra transport channel for the membrane. The optimal membrane (at the cmc point) exhibited a superior ethanol dehydration separation factor of 1627, along with a flux of 684 g/m2h at 50 degrees C. It can be expected that this novel self-assembled ILs-PVA micelle nanostructure strategy will find promising applications in other azeotropic mixture separation processes, like ethanol-ethyl acetate, water-butanol, etc.

Emitting NH3 into the atmosphere leads to significant air pollution, while NH3 itself serves as an essential component for fertilizers and refrigerants in industry. Thus, recovering and reusing NH3 is highly valuable. Ionic liquids (ILs) have shown great potential for NH3 capture, where the accurate prediction of solubility is a critical point for selecting ILs and designing a separation process. This work combined the Ionic Fragment Contribution (IFC) strategy with machine learning (ML) to develop four models (IFC-ML) to predict NH3 solubility in ILs. A dataset containing 785 solubility data points, covering 10 cations and 10 anions, was collected. From this dataset, the S1–S6 descriptors based on the IFC method were used as inputs for the ML models, together with temperature (T) and pressure (P). Among the models, the IFC-GBR model was recommended for predicting NH3 solubility in ILs due to its higher coefficient of determination (R2) of 0.9945 and lower mean squared error (MSE) of 0.0003 than the others. Additionally, in comparison with previous conductor-like screening model for real solvents (COSMO-RS) and extreme learning machine (ELM) methods, the IFC-GBR (gradient boosting regressor) method showed a more accurate prediction of the NH3 solubility in ILs over a wider range of temperatures and pressures, providing additional chemical insights into IL-NH3 system that cations played a more important role for NH3 solubility. These results highlighted the developed IFC-GBR model offered valuable insights for helping guide the process design of absorbing NH3 through IL-based technology.

The urgent need to mitigate anthropogenic CO2 emissions has driven the development of energy-efficient carbon capture systems. This study investigated a [N1111][Triz]-H2O hybrid solvent for CO2 capture using integrated experimental and computational approaches. A multiscale methodology combining thermodynamic analysis, phase equilibrium measurements, and molecular dynamics (MD) simulations was employed to elucidate the absorption mechanisms and the composition-property relationships. The thermodynamic analysis, incorporating Henry's law, the non-random two-liquid (NRTL) model for activity coefficients, the Redlich-Kwong equation, and reaction equilibrium constraints, accurately predicted the gas-liquid equilibrium (GLE) behavior, achieving an R2 of 99.1% and an average absolute relative deviation (AARD) of 7.76%. The [N1111][Triz]-H2O hybrid solvent exhibits exceptional CO2 absorption performance, with a capacity of 0.25 mol/mol (at 313.15 K and 0.025 MPa for wIL = 80%), attributed to synergistic physical-chemical interactions. MD simulations reveal the dynamic CO2 absorption process in [N1111][Triz]-H2O hybrid solvents: CO2 molecules preferentially accumulate at the gas-liquid interface before gradually diffusing into the bulk phase. Increasing the [N1111][Triz] content enhances CO2 absorption capacity by providing more interaction sites, while water modulates interfacial behavior and diffusion kinetics. This research provides in-depth insights into the absorption behaviors of [N1111][Triz]-H2O hybrid solvents for CO2, offering theoretical support for the development of efficient CO2 capture solvents and highlighting its potential for industrial implementation.

Covalent organic framework (COF) membranes show great potential for gas separation, but their large pore size make it hard to effectively separate gases with similar kinetic diameters. A promising strategy is to cover the surface of COF membranes with an ultrathin film of ionic liquids (ILs) to improve the selectivity of gas permeation. Here, we investigated the CO2/CH4 separation performance of composite membranes composed of COF (CTF-1) and 20 kinds of ILs, respectively, using molecular simulations. The [C4mim][PF6]/COF membrane is found to exhibit the best separation performance, with CO2 permeability up to 7.93×104 GPU and CO2/CH4 selectivity of 12.33 when the IL thickness is 8 Å. By studying the microstructure of the composite membrane, it was found that there is a relatively dispersed distribution of cations and anions on the COF surface, which is beneficial to the diffusion of gas molecules. The gas permeation process results reveal that CO2 rapidly achieves equilibrium in both the adsorption and dissolution layers simultaneously, thereby enhancing the permeation rate of gas molecules. The analysis of interaction energy and PMF indicates that the improved selectivity is due to the stronger interaction between IL and CO2 than that for CH4.

The precise cellular mechanisms underlying heightened proinflammatory cytokine production during coronavirus infection remain incompletely understood. Here we identify the envelope (E) protein in severe coronaviruses (SARS-CoV-2, SARS, or MERS) as a potent inducer of interleukin-1 release, intensifying lung inflammation through the activation of TMED10-mediated unconventional protein secretion (UcPS). In contrast, the E protein of mild coronaviruses (229E, HKU1, or OC43) demonstrates a less pronounced effect. The E protein of severe coronaviruses contains an SS/DS motif, which is not present in milder strains and facilitates interaction with TMED10. This interaction enhances TMED10-oligomerization, facilitating UcPS cargo translocation into the ER-Golgi intermediate compartment (ERGIC)—a pivotal step in interleukin-1 UcPS. Progesterone analogues were identified as compounds inhibiting E-enhanced release of proinflammatory factors and lung inflammation in a Mouse Hepatitis Virus (MHV) infection model. These findings elucidate a molecular mechanism driving coronavirus-induced hyperinflammation, proposing the E-TMED10 interaction as a potential therapeutic target to counteract the adverse effects of coronavirus-induced inflammation. It remains unclear why a heightened proinflammatory cytokine response is observed during coronavirus infection. Here, Liu et al show that the envelope protein of severe coronaviruses triggers hyperinflammation by activating TMED10-mediated release of inflammatory factors.

Composite membranes incorporating ionic liquids (ILs) within MXene demonstrate promising potential for CO2 separation. However, studies on the separation of CO2/CH4 using MXene-confined ILs membranes are limited, especially in terms of understanding the mechanisms at the molecular level. In this work, the system of CO2/CH4 in MXene-confined ILs membranes was studied by molecular dynamic simulations. The number density results reveal that MXene stratifies the ILs between the layers, with higher concentrations of ILs near MXene and lower concentrations in the middle layer. Notably, MXene has a greater impact on cations distribution compared to anions. As the layer spacing of MXene expands from 1.5 to 3 nm, the interaction between MXene and IL weakens, while that between the cations and anions strengthens. The confined ILs enhance gas solubility capability but impede gas diffusion. CO2 is distributed closer to anions, while CH4 tends to be closer to cations, with the distance between CH4 and cations decreasing as the layer spacing increases. Additionally, with the increase of layer distance, the proportion of confined ILs gradually decreases, and the gas diffusion coefficient gradually increases. Furthermore, compared to 1-Ethyl-3-methylimidazolium tetrafluoroborate ([EMIM][BF4]) and 1-Ethyl-3-methylimidazolium hexafluorophosphate ([EMIM][PF6]), MXene-confined 1-Ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TF2N]) is identified as the most effective for CO2/CH4 separation, owing to its superior CO2 solubility and highest diffusion selectivity.

Accurate prediction of the structurally diverse complementarity determining region heavy chain 3 (CDR-H3) loop structure remains a primary and long-standing challenge for antibody modeling. Here, we present the H3-OPT toolkit for predicting the 3D structures of monoclonal antibodies and nanobodies. H3-OPT combines the strengths of AlphaFold2 with a pre-trained protein language model, and provides a 2.24 Å average RMSD Cα between predicted and experimentally determined CDR-H3 loops, thus outperforming other current computational methods in our non-redundant high-quality dataset. The model was validated by experimentally solving three structures of anti-VEGF nanobodies predicted by H3-OPT. We examined the potential applications of H3-OPT through analyzing antibody surface properties and antibody-antigen interactions. This structural prediction tool can be used to optimize antibody-antigen binding, and to engineer therapeutic antibodies with biophysical properties for specialized drug administration route.