Predictive framework for shape-selective separations in three-dimensional zeolites and metal-organic frameworks.

With the growing number of zeolites and metal-organic frameworks (MOFs) available, computational methods are needed to screen databases of structures to identify those most suitable for applications of interest. We have developed novel methods based on mathematical optimization to predict the shape selectivity of zeolites and MOFs in three dimensions by considering the energy costs of transport through possible pathways. Our approach is applied to databases of over 1800 microporous materials including zeolites, MOFs, zeolitic imidazolate frameworks, and hypothetical MOFs. New materials are identified for applications in gas separations (CO2/N2, CO2/CH4, and CO2/H2), air separation (O2/N2), and chemicals (propane/propylene, ethane/ethylene, styrene/ethylbenzene, and xylenes).

Cost-effective CO2 capture based on in silico screening of zeolites and process optimization.

A hierarchical computational approach is introduced that combines materials screening with process optimization. This approach leads to novel materials for cost-effective CO2 capture. Zeolites are screened using shape, size, and adsorption selectivities. Next, process optimization is introduced to generate a rank-ordered list based on total cost of capture and compression. We not only select the most cost-effective materials, but we also attain the optimal process conditions while satisfying purity, recovery, and other process constraints. The top ten zeolites (AHT, NAB, MVY, ABW, AWO, WEI, VNI, TON, OFF and ITW) can capture and compress CO2 to 150 bar from a mixture of 14% CO2 and 86% N2 at less than $30 per ton of CO2 captured. Several zeolites have moderate selectivities, yet they cost-effectively capture CO2 with 90% purity and 90% recovery using a 4-step adsorption process. Such nonintuitive selection demonstrates the necessity of combining materials-centric and process-centric viewpoints.

Estimation of diffusion anisotropy in microporous crystalline materials and optimization of crystal orientation in membranes.

The complex nature of the porous networks in microporous materials is primarily responsible for a high degree of intracrystalline diffusion anisotropy. Although this is a well-understood phenomenon, little attention has been paid in the literature with regards to classifying such anisotropy and elucidating its effect on the performance of membrane-based separation systems. In this paper, we develop a novel methodology to estimate full diffusion tensors based on the detailed description of the porous network geometry through our recent advances for the characterization of such networks. The proposed approach explicitly accounts for the tortuosity and complex connectivity of the porous framework, as well as for the variety of diffusion regimes that may be experienced by a guest molecule while it travels through the different localities of the crystal. Results on the diffusion of light gases in silicalite demonstrate good agreement with results from experiments and other computational techniques that have been reported in the literature. A comprehensive computational study involving 183 zeolite frameworks classifies these structures in terms of a number of anisotropy metrics. Finally, we utilize the computed diffusion tensors in a membrane optimization model that determines optimal crystal orientations. Application of the model in the context of separating carbon dioxide from nitrogen demonstrates that optimizing crystal orientation can offer significant benefit to membrane-based separation processes.

Stereochemically consistent reaction mapping and identification of multiple reaction mechanisms through integer linear optimization.

Reaction mappings are of fundamental importance to researchers studying the mechanisms of chemical reactions and analyzing biochemical pathways. We have developed an automated method based on integer linear optimization, ILP, to identify optimal reaction mappings that minimize the number of bond changes. An alternate objective function is also proposed that minimizes the number of bond order changes. In contrast to previous approaches, our method produces mappings that respect stereochemistry. We also show how to locate multiple reaction mappings efficiently and determine which of those mappings correspond to distinct reaction mechanisms by automatically detecting molecular symmetries. We demonstrate our techniques through a number of computational studies on the GRI-Mech, KEGG LIGAND, and BioPath databases. The computational studies indicate that 99% of the 8078 reactions tested can be addressed within 1 CPU hour. The proposed framework has been incorporated into the Web tool DREAM ( http://selene.princeton.edu/dream/ ), which is freely available to the scientific community.

Computational characterization of zeolite porous networks: an automated approach.

An automated method has been developed to fully characterize the three-dimensional structure of zeolite porous networks. The proposed optimization-based approach starts with the crystallographic coordinates of a structure and identifies all portals, channels, and cages in a unit cell, as well as their connectivity. We apply our algorithms to known zeolites, hypothetical zeolites, and zeolite-like structures and use the characterizations to calculate important quantities such as pore size distribution, accessible volume, surface area, and largest cavity and pore limiting diameters. We aggregate this data over many framework types to gain insights about zeolite selectivity. Finally, we develop a continuous-time Markov chain model to estimate the probability of occupancy of adsorption sites throughout the porous network. ZEOMICS, an online database of structure characterizations and web tool for the automated approach is freely available to the scientific community (http://helios.princeton.edu/zeomics/).