6 Conclusions
Today, molecular simulation is considered a powerful tool of the research community in general and of the zeolite community in particular. Experimental works are
complemented with atomistic information provided by these techniques. Furthermore, molecular simulation play a key role in the understanding of processes at a
molecular level. For this purpose, it is essential to have good models, force fields,
and methods able to reproduce experimental results and to make realistic predictions.
Over the last few years, there have been substantial improvements in the three
mentioned aspects; however, some challenges still remain.
Regarding the modeling, the development of the definitive water model might be
a chimera; nevertheless refining and testing current models are a valuable task which
is still necessary to improve the predictions of molecular simulation. Another
challenge is related to the role of the extra framework present in the zeolites. New
approaches and better adjustment of cation models need to be developed in order to
gain further understanding on cation mobility in the zeolites, as well as their
influence on the adsorption, diffusion, and separation processes. At the same time,
a better description of the interaction with the zeolite atoms and guest gases is
required, which leads to the need of more accurate force fields to describe the
interaction between the cation and the zeolite and between the cation and the host
gases, with a particular interest for the cation-hydrogen interaction. Although there
are many validated force fields (both universal and for specific systems), there is still
a need for research. A key feature is flexibility, and clearly better force fields would
be able to properly describe the flexibility of the frameworks and the effect of the
flexibility on the adsorption and transport properties. Current methods have proven
to be efficient in predicting adsorption isotherms, Henry coefficients, isosteric heats
of adsorption, cation distributions, preferred adsorption sites, and diffusion coefficients. Lately, great efforts have been put in developing methods that are suitable to
correctly describe and predict chemical reactions by using molecular simulation.
Commonly, the huge amount of conditions (such as temperature, pressure,
composition, defects, etc.) that are of interest in a molecular simulation or in the
screening studies over all zeolites makes their analysis for specific processes
unaffordable, not only experimentally but also via simulation techniques. In those
cases, the use of machine learning in big data context is emerging as a very powerful
tool, and it is reasonable to think that will have a big development in the next few
years.
Acknowledgments The authors thank Salvador R. G. Balestra for his help with Figure 1and Sneha
R. Bajpe, Ana Martin-Calvo, and Patrick J. Merkling for a critical reading of the manuscript.
74
J. J. Gutiérrez-Sevillano and S. Calero
Today, molecular simulation is considered a powerful tool of the research community in general and of the zeolite community in particular. Experimental works are
complemented with atomistic information provided by these techniques. Furthermore, molecular simulation play a key role in the understanding of processes at a
molecular level. For this purpose, it is essential to have good models, force fields,
and methods able to reproduce experimental results and to make realistic predictions.
Over the last few years, there have been substantial improvements in the three
mentioned aspects; however, some challenges still remain.
Regarding the modeling, the development of the definitive water model might be
a chimera; nevertheless refining and testing current models are a valuable task which
is still necessary to improve the predictions of molecular simulation. Another
challenge is related to the role of the extra framework present in the zeolites. New
approaches and better adjustment of cation models need to be developed in order to
gain further understanding on cation mobility in the zeolites, as well as their
influence on the adsorption, diffusion, and separation processes. At the same time,
a better description of the interaction with the zeolite atoms and guest gases is
required, which leads to the need of more accurate force fields to describe the
interaction between the cation and the zeolite and between the cation and the host
gases, with a particular interest for the cation-hydrogen interaction. Although there
are many validated force fields (both universal and for specific systems), there is still
a need for research. A key feature is flexibility, and clearly better force fields would
be able to properly describe the flexibility of the frameworks and the effect of the
flexibility on the adsorption and transport properties. Current methods have proven
to be efficient in predicting adsorption isotherms, Henry coefficients, isosteric heats
of adsorption, cation distributions, preferred adsorption sites, and diffusion coefficients. Lately, great efforts have been put in developing methods that are suitable to
correctly describe and predict chemical reactions by using molecular simulation.
Commonly, the huge amount of conditions (such as temperature, pressure,
composition, defects, etc.) that are of interest in a molecular simulation or in the
screening studies over all zeolites makes their analysis for specific processes
unaffordable, not only experimentally but also via simulation techniques. In those
cases, the use of machine learning in big data context is emerging as a very powerful
tool, and it is reasonable to think that will have a big development in the next few
years.
Acknowledgments The authors thank Salvador R. G. Balestra for his help with Figure 1and Sneha
R. Bajpe, Ana Martin-Calvo, and Patrick J. Merkling for a critical reading of the manuscript.
74
J. J. Gutiérrez-Sevillano and S. Calero