Efficient Models for Predicting Temperature-Dependent Henry’s Constants and Adsorption Selectivities for Diverse Collections of Molecules in Metal–Organic Frameworks

Abstract

Adsorption-based separations using metal–organic frameworks (MOFs) are a promising alternative to traditional energy-intensive separation process. Machine learning (ML) methods have been applied to predict large collections of adsorption isotherms in MOFs. Previous ML models, however, focus only on predicting single-component adsorption isotherms of a small number of molecules at a single temperature and lack accuracy in the dilute limit. Here we describe a useful strategy for predicting Henry’s constants and heats of adsorption for a diverse set of molecules in large collections of MOFs. To achieve this, a data set containing 21,195 MOF–molecule pairs with 45 adsorbates in 471 MOFs is generated, and a set of 135 descriptors combining energy and chemical information is developed. Robust ML models are developed to predict Henry’s constants and heats of adsorption after removing physically unfavorable adsorption pairs. The adsorption selectivity of near-azeotropic mixtures at two temperatures (300 and 373 K) is predicted with acceptable accuracy by using the predicted Henry’s constants and heats of adsorption. The ability to make temperature-dependent predictions is important for many practical separation applications. Our work sheds light on important challenges and opportunities for developing accurate models predicting adsorption properties for diverse collection of adsorbates and adsorbents.

Incorporating Flexibility Effects into Metal–Organic Framework Adsorption Simulations Using Different Models

“…To the best of our knowledge, all machine learning models to date aiming to predict adsorption properties of MOFs have been trained on simulation data from rigid structures. Examples exist in which information from one (or a small number of molecules) is used to enable predictions for a range of other adsorbing molecules. , Thus, one interesting direction for future work may be to augment the training data for machine learning models with carefully selected adsorption–relaxation simulation to capture the impacts of framework flexibility.…”

“…Examples exist in which information from one (or a small number of molecules) is used to enable predictions for a range of other adsorbing molecules. 69,70 Thus, one interesting direction for future work may be to augment the training data for machine learning models with carefully selected adsorption−relaxation simulation to capture the impacts of framework flexibility.…”

Minimal set of crystallographic descriptors for sorption properties in hypothetical Metal Organic Frameworks: Role in sequential learning optimization

“…The most challenging aspect of this task consists in the access to the entire isosteric field of each candidate MOF-water pair for estimating the engineering figure of merit of interest. Clearly, for a large number of MOFs candidates, this is challenging and time consuming both computationally (typically, only the Henry low-coverage regime is reported in literature works [56,57]) and experimentally [58]. In this section, we specifically focus on such challenging aspect.…”

Computational Design of MOF-Based Electronic Noses for Dilute Gas Species Detection: Application to Kidney Disease Detection

“…Fortunately, for disease detection by breath, the biomarker gases of interest are typically present only in trace quantities, and so their adsorption in MOFs obeys Henry’s law (adsorbed concentration is a linear function of the concentration in the bulk mixture outside of the MOF) . In this case, it is sufficient to determine only the corresponding Henry’s law coefficients for each gas/MOF pair to be able to predict the amount of each trace gas absorbed by the MOFs, thus drastically reducing computational demand.…”

“…Fortunately, for disease detection by breath, the biomarker gases of interest are typically present only in trace quantities, and so their adsorption in MOFs obeys Henry's law (adsorbed concentration is a linear function of the concentration in the bulk mixture outside of the MOF). 37 In this case, it is sufficient to determine only the corresponding Henry's law coefficients for each gas/MOF pair to be able to predict the amount of each trace gas absorbed by the MOFs, thus drastically reducing computational demand. In this work, we evaluated a modified form of Henry's coefficients, which we call combined linear adsorption coefficients (CLACs), which quantify not only the amount of trace gas species adsorbed but also the amount of air displaced by the trace gas, as a function of the trace gas concentration.…”