Graph Neural Networks for Carbon Dioxide Adsorption Prediction in Aluminium-Exchanged Zeolites

TL;DR

A model which is 4 to 5 orders of magnitude faster at adsorption properties compared to molecular simulations is proposed and it is shown that the model can be used for identifying adsorption sites and is evaluated for generating novel zeolite configurations by using it in combination with a genetic algorithm.

Abstract

The ability to efficiently predict adsorption properties of zeolites can be of large benefit in accelerating the design process of novel materials. The existing configuration space for these materials is wide, while existing molecular simulation methods are computationally expensive. In this work, we propose a model which is 4 to 5 orders of magnitude faster at adsorption properties compared to molecular simulations. To validate the model, we generated datasets containing various aluminium configurations for the MOR, MFI, RHO and ITW zeolites along with their heat of adsorptions and Henry coefficients for CO$_2$, obtained from Monte Carlo simulations. The predictions obtained from the Machine Learning model are in agreement with the values obtained from the Monte Carlo simulations, confirming that the model can be used for property prediction. Furthermore, we show that the model can be used for identifying adsorption sites. Finally, we evaluate the capability of our model for generating novel zeolite configurations by using it in combination with a genetic algorithm.

Figures (7)

• Figure 1: Pipeline comparison between calculating heat of adsorption using Monte Carlo simulations and our proposed ML method. Black arrows represent the property prediction process, while red arrows represent the zeolite generation process.
• Figure 2: Heat of adsorption and Henry coefficient distribution for different topologies. Vertical bars indicate the 95% confidence interval for each amount of aluminium substitutions substitutions.
• Figure 3: Zeolite structures used in this work. MOR, RHO and ITW are visualised along the z-axis, while MFI is visualised along the y-axis. Top row: Zeolite structures visualised using iRASPA dubbeldam2018iraspa. Bottom row: Zeolite graph representation for ML. Circles and squares represent T-atom nodes and pore nodes, while solid edges are drawn between T-atoms and dotted edges between T-atoms and pores. Edges/Nodes of the same color and type are symmetric, and thus share parameters. Note that for MFI we only visualised the top layer of atoms.
• Figure 4: Heat of adsorption and Henry coefficient for different topologies predicted by MC and ML on the test set. Vertical bars indicate the 95% confidence interval for each amount of substitutions.
• Figure 5: Heat of adsorption and Henry coefficient for all topologies predicted by MC and ML on the test set.