Acetic acid production from renewable processes such as biomass hydrolysis and electrochemical reduction of CO₂ exhibits low concentrations, which make downstream separation challenging. We measured the vapor–liquid equilibria of the binary systems acetic acid + 2-methyltetrahydrofuran (2-MTHF), methyl-t-butyl ether (MTBE) + acetic acid, and the ternary liquid–liquid equilibria of the system 2-MTHF + AA + water, fitted the data to the UNIQUAC-HOC and NRTL models, designed a hybrid extraction-distillation process for acetic acid separation with 2-MTHF, and evaluated its economics and compared with that of three other commonly used solvents (i.e., ethyl acetate, MTBE, and methyl propyl ketone). The lowest and highest costs of separation were observed for MTBE and MPK, while 2-MTHF and EA showed similar performance. The cost of separation increased exponentially as the feed concentration decreased, and renewable processes should aim for at least 5 wt % acetic acid in the feed to allow economically feasible separation.

Experimentally determining interfacial tension (IFT) for compositions relevant to CO₂ transport is challenging. We address this using molecular dynamics (MD) simulations and perturbed-chain statistical associating fluid theory (PC-SAFT) equation of state with classical density functional theory. We compute phase equilibria and interfacial properties of pure CO₂ and CO₂–CH₄, CO₂–Ar, CO₂–N₂, and CO₂–H₂ mixtures at 220–273 K. Both approaches accurately estimate CO₂ phase equilibria and IFTs. For binary mixtures, phase equilibria computed using PC-SAFT agree well with experiments when k_ij ≠ 0. IFTs computed from PC-SAFT depend strongly on k_ij, while MD simulations systematically overpredict IFTs. The IFT decreases with increasing pressure, least pronouncedly for H₂-containing mixtures. Binary mixtures exhibit interfacial enrichment of the light boiling component, decreasing with increasing temperature and pressure. Semiempirical Parachor and Winterfeld–Scriven–Davis models capture IFT–pressure trends with mixture-dependent accuracy. These results improve predictions of metastable limits and provide key insights for fast-transient multiphase CO₂ flow modeling.

Grotthuss transfer is responsible for a large increase in the self-diffusion of hydroxide and hydronium ions in aqueous solutions compared to similarly sized ions. Recent advances in machine-learning molecular dynamics have shown some success in capturing this process. In the present work, we show that classical molecular dynamics combined with experimentally measured electrical conductivities can also be used to determine self-diffusion coefficients and the lifetimes of hydroxide and hydronium ions in aqueous KOH, NaOH, and HCl solutions. This was tested and validated across a wide range of concentrations at 25 and 60 °C. The approach relies on augmenting classically computed trajectories with a biased random walk, which together accounts for both vehicular transport and Grotthuss transfer. The concentration and temperature dependence of this random walk are calibrated by comparing simulated electrical conductivities with available experimental electrical conductivity data. The computed self-diffusion coefficients match measurements at infinite dilution and results from machine learning molecular dynamics. Ion lifetimes reported by machine learning and ab initio molecular dynamics studies depend strongly on the precise definition of what constitutes a Grotthuss transfer event. Our approach for calculating ion lifetimes does not have this drawback. We also show that our self-diffusion coefficients and electrical conductivities are insensitive to the precise definition of what constitutes a Grotthuss transfer event.

We present a simulation-based framework to characterize the solvation and aqueous-phase reactivity of nitric oxide (NO) and nitrogen dioxide (NO2) in water. Using Continuous Fractional Component Monte Carlo (CFCMC) simulations, we compute Henry coefficients and chemical potentials of NO and NO2, while molecular dynamics (MD) simulations provide diffusion coefficients for NO. The results for NO are quantitatively in agreement with the experimental data when using the Saji force field. For NO2, we model the chemical equilibrium involving hydrolysis and acid-base reactions that generate HNO2, HNO3, NO2-, NO3-, and H3O+. By combining the chemical potentials obtained via CFCMC with a thermodynamic equilibrium model, we resolve the temperature- and pressure-dependent speciation and pH of the system. The model captures a transition from nitrous to nitric species with increasing temperature and predicts ionic distributions and pH shifts under varying NOx gas fluxes. This work provides a transferable methodology to connect molecular simulations with chemical speciation in reactive aqueous systems.

To enhance the efficiency of thermodynamic cycles in heat pumps and power plants, we explore a novel approach: replacing conventional inert pure fluids or mixtures with reactive fluids that undergo reversible chemical reactions. A key step towards the implementation of this concept is the development of a fully predictive framework for determining the thermodynamic properties of such reactive working fluids. In this context, the present work extends a semi-empirical methodology previously proposed by the authors, aiming to address the challenge introduced by newly developed reactive fluids for which experimental data are unavailable. The methodology presented in this work requires only the critical-point properties and acentric factor of the molecules participating in the chemical reaction. As in the earlier approach from the authors, it combines ab-initio quantum mechanics calculations to determine the ideal gas properties of each molecule, the a-thermal version of the “Peng-Robinson + EoS/a_res^E,γ mixing rules” equation of state and molecular Monte Carlo simulations to assess real fluid properties and enable cross-validation between methods. This work, however, applies a simplification to the force fields used in Monte Carlo simulations consisting in employing single-particle force fields instead of all-atom models. This strategy decreases the amount of experimental data required to parametrise the force field of each molecule contained in the reactive mixture, and allows the use of the same inputs in equation of state modelling and Monte Carlo simulations (i.e., molecular critical parameters). Indeed, this work proposes to calculate force field parameters using either the critical temperature and pressure, or the critical temperature and density of each molecule. The methodology is applied to two reactive systems, Al₂Br₆ ⇌ 2AlBr₃ and Al₂Cl₆ ⇌ 2AlCl₃. The results show that Monte Carlo predictions, although less accurate than those from the equation of state, remain acceptably close to experimental data, while the equation of state results demonstrate significantly higher accuracy.

Hydrogen can play a central role in a fossil-free energy economy, yet its implementation is hindered by the lack of safe, dense, and efficient storage methods. Hybrid H₂ physisorption-hydrate formation, which combines physisorption in porous materials with encapsulation in clathrate hydrates, presents a promising route, but the fundamental synergistic mechanisms remain largely elusive. Here, we perform microsecond-scale molecular dynamics simulations to study the hybrid H₂ storage process in the hydrophobic metal–organic framework ZIF-8 seeded with THF hydrate nanoparticles. The results indicate that ZIF-8 rapidly physisorbs H₂, while effectively excluding H₂O and THF. Our simulations reveal a dynamic, three-step hybrid storage pathway, i.e. , (1) ZIF-8 selectively adsorbs and enriches H₂ within its pores, creating a high local H₂ concentration; (2) The growing binary H₂-THF hydrate crystals selectively capture the H₂; (3) Transfer of H₂ from the ZIF-8 to the hydrate until the hydrogen source transfer reaches a dynamic equilibrium. This hybrid storage method results in a total H₂ storage capacity reaching 1.82 wt%, exceeding the storage capacity of either physisorption or THF-driven hydrate formation alone. These findings provide critical molecular-level insights, showing that coupling hydrophobic ZIF-8 with hydrate promoters is a highly effective strategy for developing next-generation H₂ storage methods.

We study the interactions of plasma-generated Reactive Oxygen and Nitrogen Species (RONS) with water due to their importance for applications in health and agriculture. Atomic oxygen, a key RONS, is produced by plasma in both its triplet ground state, O(³P), and its singlet excited state, O(¹D). Experimental studies indicate that when plasma interacts with water, atomic oxygen can remain sufficiently stable to enter the aqueous phase. Recent measurements show that ground-state oxygen atoms can persist for tens of microseconds and penetrate hundreds of micrometres into the aqueous phase. However, quantitative data on the solubility and diffusion of atomic oxygen remain scarce. This is likely due to limitations in experimental diagnostics and the challenges that the complex electronic structure of atomic oxygen presents to modeling approaches. To overcome these challenges, we developed state-specific force fields to model the interactions of O(³P) and O(¹D) with water to account for quantum-state-dependent interactions. Using these force fields, we provide the first estimates of temperature- and quantum-state-dependent self-diffusion and Henry coefficients of atomic oxygen in aqueous environments. Building upon these results, we propose a general framework to estimate the solubility and diffusion of other plasma-generated charge-neutral RONS in water by representing each species as a charge-neutral Lennard-Jones particle. The influence of particle size, solute–solvent interaction strength, and temperature on the transport and thermodynamic properties of RONS was systematically investigated. This approach enables the estimation of the Henry coefficients and the diffusion coefficients of RONS in water based on particle size, solute–solvent interactions, and temperature. These estimates provide key parameters for device-level plasma-liquid simulations and offer molecular-scale insight for interpreting experimental findings.

We introduce OpenPyTEA , an open-source Python toolkit for conducting flexible and detailed techno-economic assessments (TEA) of chemical and energy systems. TEA is essential for evaluating economic feasibility in process design, yet commercial tools are “black-box” solutions with limited flexibility. Existing open-source options are usually process-specific, incomplete, or poorly documented, limiting reproducibility and cross-study comparisons. OpenPyTEA addresses these challenges by integrating equipment cost estimation, cash-flow analysis, and sensitivity and uncertainty methods into transparent and adaptable workflows. Its capabilities are demonstrated through a case study comparing hydrogen production via steam methane reforming, methane pyrolysis, and water electrolysis.

Electrochemical CO₂ reduction to CO offers a sustainable route for converting CO₂ into value-added chemicals and fuels. However, CO₂ streams derived from industrial sources often contain SO₂ impurities that severely poison conventional metal-based catalysts. Here, we report a nitrogen-doped carbon catalyst that exhibits pronounced tolerance and stability for CO₂-to-CO conversion in the presence of SO₂ (100–10,000 ppm). The catalyst maintains over 90% Faradaic efficiency toward CO during 8 h of electrolysis at −1.0 V vs RHE with 100 ppm of SO₂, whereas Ag foil electrodes undergo rapid deactivation. Density functional theory calculations combined with surface analyses indicate that weak SO₂ adsorption and the absence of stable sulfur accumulation on nitrogen-doped carbon strengthen its resistance to impurity-induced deactivation, in contrast to Ag catalysts that form Ag₂S. Gas-fed tests in a membrane electrode assembly (MEA) electrolyzer further confirm that nitrogen-doped carbon sustains high CO selectivity at elevated current densities, while Ag nanoparticles suffer irreversible sulfur poisoning. These results demonstrate that nitrogen-doped carbon is intrinsically resistant to SO₂-induced deactivation and highlight its potential as a robust catalyst for CO₂ electroreduction under impurity-containing conditions.

Knowledge of the effect of different organic molecules on spontaneous hydrate nucleation is crucial for understanding the formation of gas hydrates in marine reservoirs. Herein, microsecond MD simulations are conducted to investigate the spontaneous nucleation of CH₄ hydrates in oceanic sediments. The simulation results indicate that hydrate nucleation is influenced by the coupling effects of organic molecules, clay surfaces and salt ions, where organic molecules alter hydrate nucleation by modulating the diffusion fluctuation of CH₄ molecules via controlling the shape and size of CH₄ nanobubbles. Furthermore, CH₄ hydrates are primarily concentrated at a moderate distance away from the nanobubbles, with fewer hydrates located either close or at a more distant from the nanobubbles. In the region about 1.0 nm away from the nanobubbles, the hydrates become more unstable when closer to the nanobubbles, whereas hydrates have better stability when locating above 1.0 nm away from the nanobubbles. Different organic molecules exert distinct effects on spontaneous hydrate nucleation. Specifically, propanol adsorbed to the nanobubble surface kinetically promotes hydrate nucleation, exhibiting a distinct advantage over other organic molecules. These molecular insights expand the understanding of the formation of natural gas hydrate resources and help to effectively utilize this resource.

Hydroisomerization of alkane isomers is an important step in the manufacture of current kerosene and sustainable aviation fuels. Zeolites are used as acid catalysts in this process. It is therefore important to have predictions of the adsorption capacity or maximum loading of hydrocarbons in zeolites. Here, a cascade model using machine learning models is used to predict the maximum loading of alkane isomers in zeolites. The cascade is composed of a gradient-boosted tree classifier stage that predicts whether adsorption occurs and a regressor predicting the value of the maximum loading. The final data set consists of 45 different adsorbates (both linear and branched alkanes up to C₁₆) and 97 different zeolite structures, resulting in 4365 data points. Descriptors include information on the geometry and topology of zeolite channels as well as the shape and size of the adsorbates. Extra composite descriptors are also present to provide the physical basis for predictions. Multiple regressors of different natures are considered: support vector regressors, gradient-boosted trees, extreme gradient-boosted trees, and the TabPFN pretrained model. TabPFN yields the highest generalization performance and the lowest error. An interpretability analysis using SHAP reveals that the most influential descriptors are physically meaningful, highlighting steric and volumetric constraints as the primary factors controlling the prediction of q_max. It is shown that despite both the classifier and the regressor being insensitive to random splits in data, the regressor is prone to overfitting at low fractions of data withheld for testing. The cascade model is compared to an Artificial Neural Network for training and resource efficiency. Despite training being longer for the neural network, the final model is lighter in both memory and storage. This work is built on our previous research in predicting the Henry coefficients of long-chain alkanes in zeolites. Using this previous model and the findings of this work, one could construct the adsorption isotherm for any alkane, thus enabling the analysis of adsorption behavior of alkane mixtures using IAST.

Methane pyrolysis is a promising route for low-emission hydrogen (H₂) production, with solid carbon as a potentially valuable byproduct. Despite this potential, the economic feasibility of Catalytic Methane Pyrolysis (CMP) with fluidized bed reactors (FBR) has been insufficiently studied. This study develops a conceptual CMP plant using two novel isothermal reactor models—based on continuous stirred-tank reactor (CSTR) and plug-flow reactor (PFR) assumptions—to represent the operational extremes of FBRs. Our reactor models incorporate reaction and catalyst deactivation kinetics from experiments with nickel-supported catalysts, and the framework enables process simulations that account for the catalyst rate required to sustain reactor activity. These models address the lack of proper reduced-order FBR models and the reliance on oversimplified assumptions in the literature. In the baseline scenario, the conceptual plant yields an LCOH ranging from $3.89 to $4.79 per kilogram, defining the expected cost bounds for an FBR-based CMP plant. At a H₂ selling price of $5.00 per kilogram, the process achieves favorable payback time and net present value. Monte Carlo and sensitivity analyses indicate that CMP remains cost-competitive under economic uncertainties. Increased carbon sales could make CMP more economical than steam methane reforming, while unsold byproducts may incur costly sequestration. Reactor heating assessment shows methane combustion with carbon capture minimizes both cost and emissions. Overall, this work demonstrates the economic potential of CMP for H₂ production and provides a practical modeling framework for process evaluation.

In the days prior to the Thermodynamics2024 conference in Delft (The Netherlands), the annual RASPA workshop/school took place at Delft University of Technology with 55 participants (both industry and academics) from all over the world. RASPA is a popular open-source molecular simulation software package for studying adsorption and diffusion in fluids and nanoporous materials, and it is especially popular in the metal-organic frameworks and zeolite communities. The main contributors to RASPA (and organisers of the RASPA workshop/school in Delft) are David Dubbeldam, Sofia Calero, Randall Q. Snurr, and Thijs J.H. Vlugt. In this short paper, we briefly explain the history of RASPA and the RASPA workshops, as well as our strategy to teach the workshop participants how to use RASPA for their specific research projects.

Adsorption simulations often assume a rigid framework, which can be exploited by replacing the expensive framework-adsorbate energy/force evaluation by interpolation of a precomputed energy grid. We present the implementation in RASPA3 of a triquintic interpolation algorithm by Boateng and Bradach and compare it to the tricubic algorithm of Lekien and Marsden. We extended the scheme to interpolation in fractional space to facilitate interpolation of non-rectangular frameworks and evaluated the accuracy. We find that the use of grids is advantageous for larger systems and/or large cutoffs, but generally the efficiency gains are modest (a factor of 2–5).

Porous materials such as zeolites and Metal-Organic Frameworks are widely used for molecular separations based on adsorption and enthalpy/entropy characteristics. Ideal adsorption solution theory (IAST) predicts mixture adsorption behaviour on the basis of pure component isotherms of adsorbents in porous media. Mixture data at all mole fractions are required for breakthrough simulations. The use of IAST avoids the expensive computations of mixtures with Monte Carlo methods. Matching outcomes from computational physics studies to experimentally measurable properties is the foundation of the materials design pipeline. Here, we report the regression of an Invertible Autoencoder (IAE) for the forward and backward mapping of pure and mixture isotherms. The invertible autoencoder is defined as a soft-invertible neural network, which can be used as mapping function. Pure component isotherms are modelled using a 3-site Langmuir-Freundlich model, with a broad range of equilibrium pressure and heterogeneity factors. A synthetic dataset is generated from pure component isotherms and mixture isotherms calculated with RUPTURA. The IAE predicts pure and mixture isotherms with high precision over a large fugacity range, for up to 6 components and 3-site isotherms. This work contributes to inverting the full design pipeline from physical gas separation to adsorbate design, enabling property-guided materials discovery.