A Multiscale Modelling Approach to Large-Scale Screening of Porous Materials

Most of industrial separation processes are significantly energy-intensive. These processes account for 10 - 15% of total global energy consumption1. Less energy-consuming separation technologies are also being developed which can separate molecules based on their geometry, size and chemical properties by adsorption in a porous material. Energy efficiency of an adsorption process crucially depends on the selection of the appropriate porous material. Given the number of possible structures within MOF, COF, PAF, zeolite and other classes of the recently discovered materials, clearly this selection cannot be based on the experimental assessment and some efficient computational screening is required. So far, the computational screening strategies have been predominantly based on molecular simulations. This approach is associated with two major challenges. Firstly, accurate molecular simulation forcefields are not necessarily available for all candidate materials under consideration. Secondly, even if all the predictions from molecular simulations were accurate, performance of a material in a real process also depends on the process configuration and cycle parameters. As a result, published studies, although having paved the way for computational approaches in general, have not yet led to a viable adsorption technology2-6.
Here we present a novel multi-scale approach, where ranking of promising porous materials is based on their performance in the actual process. For this, equilibrium and transport data obtained from accurate molecular simulations is fed into process modelling and optimization toolbox. As a case study we consider a small set of materials representing MOF, ZIF and zeolite classes, in application to post-combustion carbon capture. Accuracy of the process simulation tools is validated using the in-house Dual-Piston Pressure Swing Adsorption (DP-PSA)7 system. For the actual carbon capture process we consider variants of the Skarstrom cycle and the final ranking is obtained from the analysis of the “recovery vs purity” and “energy penalty vs productivity” behaviour for each material.
We highlight the differences between the performance of a material in the optimized and a sub-optimal cycle and explore the influence of the accuracy of the molecular simulation predictions on the final ranking of the materials based on the process performance.

(1) Sholl, D. S. L., R. P. Nature 2016, 532, 435.
(2) Yazaydin, A. Ö.; Snurr, R. Q.; Park, T.-H.; Koh, K.; Liu, J.; LeVan, M. D.; Benin, A. I.; Jakubczak, P.; Lanuza, M.; Galloway, D. B.; Low, J. J.; Willis, R. R. J. Am. Chem. Soc. 2009, 131, 18198.
(3) Wilmer, C. E.; Leaf, M.; Lee, C. Y.; Farha, O. K.; Hauser, B. G.; Hupp, J. T.; Snurr, R. Q. Nat Chem 2012, 4, 83.
(4) Lin, L.-C.; Berger, A. H.; Martin, R. L.; Kim, J.; Swisher, J. A.; Jariwala, K.; Rycroft, C. H.; Bhown, A. S.; Deem, M. W.; Haranczyk, M.; Smit, B. Nat Mater 2012, 11, 633.
(5) Kim, J.; Martin, R. L.; Rübel, O.; Haranczyk, M.; Smit, B. J. Chem. Theory Comput. 2012, 8, 1684.
(6) Kim, J.; Abouelnasr, M.; Lin, L.-C.; Smit, B. J. Am. Chem. Soc. 2013, 135, 7545.
(7) Friedrich, D.; Ferrari, M.-C.; Brandani, S. Ind. Eng. Chem. Res. 2013, 52, 8897.