Shape-selective adsorption in zeolites plays a pivotal role in catalytic hydroisomerization of long-chain alkanes, a key process in producing sustainable aviation fuels from Fischer–Tropsch products. Accurately predicting adsorption behavior for the large number of alkane isomers in different zeolite frameworks is computationally intensive. To address this, we have developed a machine learning framework that rapidly and accurately predicts Henry coefficients of linear (C1–C30) and branched (C4–C20) alkanes in one-dimensional zeolites. Using descriptors based on chain length, branching patterns, and molecular graphs, we evaluate multiple ML models, including Random Forest, XGBoost, CatBoost, TabPFN, and D-MPNN in MTT-, MTW-, MRE-, and AFI-type zeolites. TabPFN and D-MPNN offer the highest predictive accuracy. Active learning further boosts model performance by efficiently selecting diverse and structurally informative isomers. We also uncover activity cliffs, where small changes in molecular structure lead to sharp variations in adsorption, and demonstrate that targeted oversampling of these cases improves model robustness. Finally, we combine the ML-predicted Henry coefficients with gas-phase thermodynamics to compute reaction equilibrium distributions for C16 hydroisomerization. This integrated, data-driven approach enables efficient screening and design of shape-selective zeolite catalysts, thereby reducing the need for costly simulations

We study important aspects of shape selectivity effects of zeolites for hydroisomerization of linear alkanes, which produces a myriad of isomers, particularly for long chain hydrocarbons. To investigate the conditions for achieving an optimal yield of branched hydrocarbons, it is important to understand the role of chemical equilibrium in these reversible reactions. We conduct an extensive analysis of shape selectivity effects of different zeolites for the hydroisomerization of C₇ and C₈ isomers at chemical reaction equilibrium conditions. The reaction ensemble Monte Carlo method, coupled with grand-canonical Monte Carlo simulations, is commonly used for computing reaction equilibrium of heterogeneous reactions. The computational demands become prohibitive for a large number of reactions. We used a faster alternative in which reaction equilibrium is obtained by imposing chemical equilibrium in the gas phase and phase equilibrium between the gas phase components and the adsorbed phase counterparts. This effectively mimics the chemical equilibrium distribution in the adsorbed phase. Using Henry’s law at infinite dilution and mixture adsorption isotherm models at elevated pressures, we calculate the adsorbed loadings in the zeolites. This study shows that zeolites with cage or channel-like structures exhibit significant differences in selectivity for alkane isomers. We also observe a minimal impact of pressure on the gas-phase equilibrium of these reactions at typical experimental reaction temperatures 400 − 700 K . This study marks initial strides in understanding the reaction product distribution for long-chain alkanes.

Linear regression (LR) is used to predict thermochemical properties of alkanes at temperatures (0–1000) K to study chemical reaction equilibria inside zeolites. The thermochemical properties of C1 until C10 isomers reported by Scott are used as training data sets in the LR model which is used to predict these properties for alkanes longer than C10 isomers. Second-order groups are used as independent variables which account for the interactions between the neighboring groups of atoms. This model accurately predicts Gibbs free energies, enthalpies, Gibbs free energies of formation, and enthalpies of formation for alkanes which exceeds the chemical accuracy of 1 kcal/mol and outperforms the group contribution methods developed by Benson et al., Joback and Reid, and Constantinou and Gani. Predictions from our model are used to compute the reaction equilibrium distribution of hydroisomerization of C10 and C14 isomers in MTW-type zeolite. Calculation of reaction equilibrium distribution inside zeolites also requires Henry coefficients of the isomers which can be computed using classical force field-based molecular simulations using the RASPA2 software for which we created an automated workflow. The reaction equilibrium distribution for C10 isomers obtained using the LR model and the training data set for this model are in very good agreement. The tools developed in this study will enable the computational study of hydroisomerization of long-chain alkanes (>C10).

Entropies for alkane isomers longer than C10 are computed using our recently developed linear regression model for thermochemical properties which is based on second-order group contributions. The computed entropies show excellent agreement with experimental data and data from Scott’s tables which are obtained from a statistical mechanics-based correlation. Entropy production and heat input are calculated for the hydroisomerization of C7 isomers in various zeolites (FAU-, ITQ-29-, BEA-, MEL-, MFI-, MTW-, and MRE-types) at 500 K at chemical equilibrium. Small variations in these properties are observed because of the differences in reaction equilibrium distributions for these zeolites. The effect of chain length on heat input and entropy production is also studied for the hydroisomerization of C7, C8, C10, and C14 isomers in MTW-type zeolite at 500 K. For longer chains, both heat input and entropy production increase. Enthalpies and absolute entropies of C7 hydroisomerization reaction products in MTW-type zeolite increase with higher temperatures. These findings highlight the accuracy of our linear regression model in computing entropies for alkanes and provide insight for designing and optimizing zeolite-catalyzed hydroisomerization processes.

We present the RUPTURA code (https://github.com/iraspa/ruptura) as a free and open-source software package (MIT license) for (1) the simulation of gas adsorption breakthrough curves, (2) mixture prediction using methods like the Ideal Adsorption Solution Theory (IAST), segregated-IAST and explicit isotherm models, and (3) fitting of isotherm models on computed or measured adsorption isotherm data. The combination with the RASPA software enables computation of breakthrough curves directly from adsorption simulations in the grand-canonical ensemble. RUPTURA and RASPA have similar input styles. IAST is implemented near machine precision but we also provide several explicit mixture prediction methods that are non-iterative and potentially faster than IAST. The code supports a wide variety of isotherm models like Langmuir, Anti-Langmuir, BET, Henry, Freundlich, Sips, Langmuir-Freundlich, Redlich-Peterson, Toth, Unilan, O'Brian & Myers, Asymptotic Temkin, and Bingel & Walton. The isotherm model parameters can easily be obtained by the fitting module. Breakthrough plots and animations of the column properties are automatically generated. In addition to highlighting the code, we also review all the developed techniques from literature for mixture prediction, breakthrough simulations, and isotherm model fitting, and provide a tutorial discussing the workflows.

Ideal Adsorbed Solution Theory (IAST) is a common method for modelling mixture adsorption isotherms based on pure component isotherms. When the adsorbent has distinct adsorption sites, the segregated version of IAST (SIAST) provides improved adsorbed loadings compared to IAST. We have adopted the concept of SIAST and applied it to an explicit isotherm model which takes into account the different sizes of the adsorbates: the so called Segregated Explicit Isotherm (SEI). The purpose of SEI is to have an explicit adsorption model that can consider both size-effects of the co-adsorbed molecules and surface heterogeneities. In sharp contrast to IAST and SIAST, no iterative scheme is required in case of SEI, which leads to much faster simulations. A comparative study has been performed to analyse the adsorption isotherms calculated using these three methods. The adsorbed loadings predicted by SEI and SIAST are in excellent agreement with the Grand-Canonical Monte Carlo (GCMC) simulation data. The loadings estimated by IAST show considerable deviations from the GCMC data at high pressures. Breakthrough curve modelling is used to compare the effects of these three models at dynamic conditions. The explicit model (SEI) leads to the fastest simulation run time, followed by SIAST.

A reactor model that deconvolutes thermodynamics of adsorption of hydrocarbon in the pores of zeolite Beta, obtained by Configurational-bias Monte Carlo simulations, from intrinsic, intraporous kinetics of hydroisomerization and hydrocracking reactions, provides a good quantitative description of all significant reactions in the kinetic network for interconversion and cracking of different heptane isomers. Activation enthalpies obtained for intraporous reactions follow the expected order according to the carbenium ion formalism: methyl shift< ethyl shift < isom(B) ∼ crack(B2) < crack(B1) < crack(C) ∼ crack(D) < crack(E) and apparently within each isomerization class, in terms of carbenium ions formally involved: sec → tert < sec → sec ∼ tert → tert < tert → sec. except for the ethyl shift reaction forming 3-ethylpentane. Cracking happens primarily through 2,4-dimethylpentane (type B2), regardless of the initial reactant. The model can be subsequently used to separate the effect of pore structure on selective adsorption and on intraporous reaction kinetics. Zeolite Beta will serve as a base case for a comparison of different zeolite structures.

The separation of xylenes is one of the most important processes in the petrochemical industry. In this article, the competitive adsorption from a fluid-phase mixture of xylenes in zeolites is studied. Adsorption from both vapor and liquid phases is considered. Computations of adsorption of pure xylenes and a mixture of xylenes at chemical equilibrium in several zeolite types at 250 °C are performed by Monte Carlo simulations. It is observed that shape and size selectivity entropic effects are predominant for small one-dimensional systems. Entropic effects due to the efficient arrangement of xylenes become relevant for large one-dimensional systems. For zeolites with two intersecting channels, the selectivity is determined by a competition between enthalpic and entropic effects. Such effects are related to the orientation of the methyl groups of the xylenes. m-Xylene is preferentially adsorbed if xylenes fit tightly in the intersection of the channels. If the intersection is much larger than the adsorbed molecules, p-xylene is preferentially adsorbed. This study provides insight into how the zeolite topology can influence the competitive adsorption and selectivity of xylenes at reaction conditions. Different selectivities are observed when a vapor phase is adsorbed compared to the adsorption from a liquid phase. These insight have a direct impact on the design criteria for future applications of zeolites in the industry. MRE-type and AFI-type zeolites exclusively adsorb p-xylene and o-xylene from the mixture of xylenes in the liquid phase, respectively. These zeolite types show potential to be used as high-performing molecular sieves for xylene separation and catalysis.

Vapor-liquid equilibria of xylenes were computed using Monte Carlo simulations in the Gibbs ensemble. For binary mixtures, the predicted composition of the liquid phase is in agreement with experiments. The computed vapor phase densities of each isomer showed an effect on the predicted composition of the vapor phase of the mixture. Unfortunately, the Monte Carlo code used to calculate the vapor-liquid equilibrium contained an error in the acceptance rule for molecule transfer in multi-component mixtures. The Continuous Fractional Component (CFCMC) [1–3] algorithm in the Gibbs ensemble adds an extra molecule -the fractional molecule-per molecule type. In the CFCMC algorithm, trial moves are attempted to transform a fractional molecule of type a in box i to a whole molecule and, simultaneously, transform a molecule of type a in box j [Formula presented] i to a fractional molecule. These trial moves should be accepted according to Ref. [1]: [Formula presented] where [Formula presented] is the number of molecules of type a in box j, [Formula presented] is the number of molecules of type a in box i, and [Formula presented] is the energy difference between the old and the new configuration. Unfortunately, what was implemented (incorrectly) in the computer code is: [Formula presented] where [Formula presented] and [Formula presented] are the total number of molecules of all types in box i and j, respectively, and not only of type a. For the simulation of a single component, [Formula presented] and [Formula presented]. Therefore, the mistake in the acceptance rule only affects the simulations that include more than one component. As such, Figs. 1, Fig. 2, Fig. 3, Fig. 4, and Table 1 of the article are not affected by this error. This error was hard to detect for xylene isomers, as the boiling point of these isomers are close, and therefore [Formula presented] is close to [Formula presented]. The corrected code has been tested to yield the same results as the RASPA software [4,5] for various multi-component systems. The phase composition diagram using the correct code is shown in Fig. 5. The observations based on the incorrect results remain. The composition of the liquid phase is not affected by the choice of force field or method for electrostatic interactions. For all the force fields, the correct vapor phase composition calculated with the Ewald and the Wolf methods are closer than in the incorrect simulations. In the incorrect results, the largest difference between vapor phase composition calculated with the Wolf and the Ewald method was 0.089 [mol/mol]. In the correct results, this difference is reduced to 0.067 [mol/mol]. [Figure presented] The predicted composition of the liquid phase is in excellent agreement with the experimental data. The correct simulations of the vapor phase compositions are closer to the experimental data than the incorrect simulations. The simulations with the incorrect acceptance rules showed an azeotrope behaviour that is not present in the correct simulations of the phase compositions. [Figure presented] The phase composition of the binary mixture and the excess chemical potential are not related. This is consistent with the observations from the incorrect simulations. [Formula presented] The predicted composition of the liquid phase is in excellent agreement with the experimental data. The correct simulations of the vapor phase compositions are closer to the experimental data than the incorrect simulations. The simulations with the incorrect acceptance rules showed an azeotrope behaviour that is not present in the correct simulations. [Figure presented]

We systematically study how the degree of framework flexibility affects the adsorption and diffusion of aromatics in MFI-type zeolites as computed by Monte Carlo simulations. It is observed that as the framework is more flexible, the zeolite structure is inherently changed. We have found that framework flexibility has a significant effect on the adsorption of aromatics in MFI-type zeolites, especially at high pressure. Framework flexibility allows the zeolite framework to accommodate to the presence of guest aromatic molecules. For very flexible zeolite frameworks, loadings up to two times larger than that in a rigid zeolite framework are obtained at a given pressure. We assessed the "flexible snapshot"method, which captures framework flexibility using independent snapshots of the framework. We have found that this method only works well when the loadings are low. This suggests that the effect of the guest molecules on the zeolite framework is important. Framework flexibility lowers the free-energy barriers between low energy states, increasing the rate of diffusion of aromatics in the straight channel of MFI-type zeolites for many orders of magnitude compared to a rigid zeolite framework. The simulations show that framework flexibility should not be neglected and that it significantly affects the diffusion and adsorption properties of aromatics in an MFI-type zeolite.

Computer simulations of adsorption of aromatics in zeolites are typically performed using rigid zeolite frameworks. However, adsorption isotherms for aromatics are very sensitive to small differences in the atomic positions of the zeolite (Chem. Phys. Lett., 1999, 308, 155-159). This article studies the effect of framework flexibility on the adsorption of aromatics in MFI-type zeolites computed by grand-canonical Monte Carlo simulations. New experimental data of adsorption of ethylbenzene in a MFI-type zeolite at 353 K is presented. The adsorption of n-heptane, ethylbenzene, and xylene isomers is computed in three MFI-type zeolite structures. It is observed that the intraframework interactions in flexible framework models induce small but important changes in the atom positions of the zeolite and hence in the adsorption isotherms. Framework flexibility is differently "rigid": flexible force fields produce a zeolite structure that vibrates around a new equilibrium configuration with limited capacity to accommodate to a bulky guest molecule. The vibration of the zeolite atoms only plays a role at high loadings, and the adsorption is mainly dependent on the average positions of the atoms. The simulations show that models for framework flexibility should not be blindly applied to zeolites and a general reconsideration of the parametrization schemes for such models is needed.

This article explores how well vapor-liquid equilibria of pure components and binary mixtures of xylenes can be predicted using different force fields in molecular simulations. The accuracy of the Wolf method and the Ewald summation is evaluated. Monte Carlo simulations in the Gibbs ensemble are performed at conditions comparable to experimental data, using four different force fields. Similar results using the Wolf and the Ewald methods can be obtained for the prediction of densities and the phase compositions of binary mixtures. With the Wolf method, up to 50% less CPU time is used compared to the Ewald method, at the cost of accuracy and additional parameter calibration. The densities of p-xylene and m-xylene can be well estimated using the TraPPE-UA and AUA force fields. The largest differences of VLE with experiments are observed for o-xylene. The p-xylene/o-xylene binary mixtures at 6.66 and 81.3 kPa are simulated, leading to an excellent agreement in the predictions of the composition of the liquid phase compared to experiments. The composition of the vapor phase is dominated by the properties of the component with the largest mole fraction in the liquid phase. The accuracy of the predictions of the phase composition are related to the quality of the density predictions of the pure component systems. The phase composition of the binary system of xylenes is very sensitive to slight differences in vapor phase density of each xylene isomer, and how well the differences are captured by the force fields.