Modeling CO₂ Adsorption in Zeolites Using DFT-Derived Charges: Comparing System-Specific and Generic Models

Abstract

For most guest molecules, electrostatic interactions have a non-negligible impact on the adsorption properties of microporous materials, such as zeolites or metal–organic frameworks. In force-field-based simulations of adsorption, partial charges located at atomic sites are most commonly used to account for electrostatics. These charges are either derived empirically or obtained from electronic structure calculations. In previous work addressing the adsorption of CO2 in all-silica zeolites, we have used a first-principles approach to derive system-specific charges from density functional theory (DFT) calculations. While this approach has been shown to perform very well, it has the drawback that it requires a separate DFT calculation for every system. In this work, we develop a set of generic charges that reproduces interaction energies from dispersion-corrected DFT calculations equally well as the initial, system-specific set. The performance of this set of charges is then assessed using grand-canonical Monte Carlo simulations of CO2 adsorption and CO2/N2 mixture adsorption for a total of 24 zeolite frameworks. The results are compared to analogous simulations using the system-specific charges. While the qualitative features are reproduced very well, the quantitative deviations are often non-negligible, amounting to more than 20% in a number of cases. Therefore, the generic charge set can be recommended for screening studies, but system-specific charges should be employed for more detailed investigations.

Synthesis Strategies for Ultrastable Zeolite GIS Polymorphs as Sorbents for Selective Separations

“…Interestingly, the strongest CO 2 binding occurs on the theoretical Si‐P1 structure. Our value for the CO 2 binding energy at 0 K of −0.48 eV (Table S3, Supporting Information) agrees rather well with the value of −0.35 eV reported by Fischer and Bell, which was obtained from comparable DFT methods . Assuming that the selectivity for CO 2 /N 2 adsorption can be approximated by the ratio of the corresponding adsorption equilibrium constants, , then our results for Si‐P1 yield a ratio of approximately 10, which is again in reasonable agreement with the previous work.…”

“…In addition, the Na + ion in Na‐P1 carries a positive charge of +0.86| e |. According to Fischer and Bell, their results showed a strong sensitivity to the quality of partial charge assignment, which offers an explanation as to why the predicted CO 2 uptake cannot be realized in Al‐containing GIS frameworks with stronger polar character.…”

“…Model structures of Na‐P1 and Na‐P2 were simulated with seven and six heteroatom substitutions per unit cell (16 T sites), respectively, to approximate the Si/Al ratios experimentally determined by EDS analysis of as‐synthesized samples (recall that Si/Al≈1.3 for Na‐P1 and Si/Al≈1.7 for Na‐P2). Table lists the Gibbs free binding energies of different probe molecules on both Al‐containing GIS structures (in order of increasing kinetic diameter) with one molecule adsorbed per active site (i.e., the maximum theoretical uptake capacity), and a single molecule in the unit cell of Si‐P1 for comparison with the work by Fischer and Bell …”

Advances in theory and their application within the field of zeolite chemistry

“…Recent years have also seen efforts to improve the accuracy of force fields used in the Monte Carlo simulations. 248 The direct comparison between GCMC simulations using the two sets of charges showed qualitative agreement in terms of comparing the isotherms and related data for different zeolites, but, quantitatively, there were discrepancies of between 10 and 30% in calculated datasets, with the generic charge model tending to underestimate the strength of CO 2 adsorption. Inevitably the simplicity of the model is a compromise with the speed and capacity of the modeling technique.…”

A DFT-D study of the interaction of methane, carbon monoxide, and nitrogen with cation-exchanged SAPO-34

“…While more sophisticated methods to include dispersion corrections in DFT give more accurate results in some cases [34,38], this method (also termed DFT-D for simplicity) was chosen as a reasonable compromise between accuracy and computational cost. The same method has been applied successfully in previous studies of light gases adsorbed in zeolites and related materials, both by our group and by others [40,[55][56][57][58][59]. Spin-polarised calculations were performed for Fe-SAPO-34.…”