molecules; and (c) methods, computational methods to calculate the physical quantities. The accuracy and validity of the results rely equally on all three components.
This section summarizes the most common models for adsorbates and adsorbents.
Force fields are addressed in Sect. 3, and the methods used for the simulation of
guest molecules in zeolites are described in Sect. 4.
In systems involving host-guest interactions, like zeolite-gas interactions, one can
use different strategies to model both the zeolite framework and the adsorbates. For
the zeolite framework, the simplest model is to consider the zeolite as rigid, where
each atom acts as a single interaction center placed at the crystallographic positions
reported by experiments. The atoms are considered “frozen,” neglecting interatomic
interactions between framework atoms. More complex models for the zeolites take
into account the flexibility of the framework as well. In such cases, potential energy
functions describing the bonding, bending, and torsion of the atoms are considered.
Although one could think that the increasing complexity of the model should
increase accuracy, several reports have shown that in most cases rigid and flexible
models lead to similar results when computing adsorption isotherms or adsorption
properties. However, when the focus is on understanding transport properties of
gases, the use of flexible framework models can make a significant difference [29–
31].
2.1 Models for Adsorbents
Using flexible models increases the computational cost of simulations significantly.
The most popular flexible models were developed in order to reproduce peaks of the
infrared spectra [32–35] in ensembles with fixed volume. However, such flexible
models usually failed to predict structural changes. To evaluate the degree of
accuracy of these force fields [32, 33, 35] in reproducing infrared spectra, BuenoPérez et al. [36] calculated the IR spectra of a large variety of zeolite frameworks in
the limit of pure silica composition and compared them with experimental reports.
They found that the force field developed by Nicholas et al. [33] reproduces best the
experimental IR spectra, but still it was not accurate enough to enable the identification of unknown frameworks based on the comparison of their experimental
spectra with simulated ones. The length of the unit cell of some frameworks can
vary upon adsorption, and therefore, flexible force fields for zeolites that change their
volume were needed. In a similar work, Balestra et al. [23] studied structural changes
in zeolite RHO and found that several of the framework models available in the
literature were unable to reproduce qualitatively a structural phase transition and
highlighted the role of the point charges in the model. The force field reported by
Nicholas et al. [33] predicted two stable structures for the pure silica zeolite RHO,
whereas the models proposed by Schröder et al. [34] and Sauer et al [35] retained
structural stability and reproduced cell lengths similarly to the experiment by
incorporating a core-shell structure for the first model and strong torsion energies
for the second, respectively. Nevertheless, it was discovered that these models can
Computational Approaches to Zeolite-Based Adsorption Processes
59
This section summarizes the most common models for adsorbates and adsorbents.
Force fields are addressed in Sect. 3, and the methods used for the simulation of
guest molecules in zeolites are described in Sect. 4.
In systems involving host-guest interactions, like zeolite-gas interactions, one can
use different strategies to model both the zeolite framework and the adsorbates. For
the zeolite framework, the simplest model is to consider the zeolite as rigid, where
each atom acts as a single interaction center placed at the crystallographic positions
reported by experiments. The atoms are considered “frozen,” neglecting interatomic
interactions between framework atoms. More complex models for the zeolites take
into account the flexibility of the framework as well. In such cases, potential energy
functions describing the bonding, bending, and torsion of the atoms are considered.
Although one could think that the increasing complexity of the model should
increase accuracy, several reports have shown that in most cases rigid and flexible
models lead to similar results when computing adsorption isotherms or adsorption
properties. However, when the focus is on understanding transport properties of
gases, the use of flexible framework models can make a significant difference [29–
31].
2.1 Models for Adsorbents
Using flexible models increases the computational cost of simulations significantly.
The most popular flexible models were developed in order to reproduce peaks of the
infrared spectra [32–35] in ensembles with fixed volume. However, such flexible
models usually failed to predict structural changes. To evaluate the degree of
accuracy of these force fields [32, 33, 35] in reproducing infrared spectra, BuenoPérez et al. [36] calculated the IR spectra of a large variety of zeolite frameworks in
the limit of pure silica composition and compared them with experimental reports.
They found that the force field developed by Nicholas et al. [33] reproduces best the
experimental IR spectra, but still it was not accurate enough to enable the identification of unknown frameworks based on the comparison of their experimental
spectra with simulated ones. The length of the unit cell of some frameworks can
vary upon adsorption, and therefore, flexible force fields for zeolites that change their
volume were needed. In a similar work, Balestra et al. [23] studied structural changes
in zeolite RHO and found that several of the framework models available in the
literature were unable to reproduce qualitatively a structural phase transition and
highlighted the role of the point charges in the model. The force field reported by
Nicholas et al. [33] predicted two stable structures for the pure silica zeolite RHO,
whereas the models proposed by Schröder et al. [34] and Sauer et al [35] retained
structural stability and reproduced cell lengths similarly to the experiment by
incorporating a core-shell structure for the first model and strong torsion energies
for the second, respectively. Nevertheless, it was discovered that these models can
Computational Approaches to Zeolite-Based Adsorption Processes
59