Metal–Organic Frameworks for Water Harvesting: Machine Learning-Based Prediction and Rapid Screening

Abstract

Atmospheric water harvesting based on metal− organic frameworks (MOFs) is an emerging technology to potentially mitigate water scarcity. Because of the tremendously large number of existing MOFs, it is challenging to find suitable candidates. In this context, a data-driven approach to identify topperforming MOFs represents an important direction. Herein, we develop a machine learning (ML) method to predict water adsorption in MOFs and screen out top-performing MOFs for water harvesting. First, experimental water adsorption isotherms in MOFs are collected and water adsorption properties are extracted. Quantitative structure−property relationships are analyzed in terms of pore structure and framework chemistry, providing taskspecific design principles. Then, ML models are trained and interpreted to predict water adsorption properties by using structural and chemical features, as well as operating conditions as descriptors. The transferability of the ML models is validated by out-of-sample predictions in seven newly reported MOFs. Finally, the ML models are applied to screen ∼8000 "Computation-Ready, Experimental" (CoRE) MOFs. Top-performing candidates are identified including 149 MOFs with the maximum adsorption capacity ≥35 mmol/g, 39 MOFs with working capacity ≥10 mmol/g in a relative pressure window 0.1−0.3, and 139 MOFs with working capacity ≥8.7 mmol/g in a relative pressure window 0.6−0.9. The developed ML-based method would advance task-oriented design and rapid discovery of reticular materials for energy and environmental applications.

Leveraging Machine Learning for Metal–Organic Frameworks: A Perspective

“…In addition to pore geometry, framework chemistry (i.e., metal, linker, and functionality; Figure a) is also crucial. The recent ML studies by Jiang and co-workers demonstrated that the incorporation of chemical descriptors including atom types and densities in MOFs would improve the ML prediction accuracy for C 3 H 8 /C 3 H 6 separation and water adsorption. , A commonly used descriptor of MOF chemistry is the revised autocorrelation functions (RACs) . The RACs possess two key advantages: (1) decomposing a MOF into subgraphs of respective metal clusters, organic linkers, and functional groups, hence capturing subtle chemical details and (2) considering MOF hierarchy that couples atomic-level properties (i.e., atom identity, connectivity, Pauling electronegativity, covalent radii, nuclear charge, and polarizability) closely with subgraphs, thus expressing chemical descriptors across the atomic and molecular scales coherently.…”

“…As illustrated in Figure a for C 3 H 8 /C 3 H 6 separation, Tang et al found that the RF models trained upon CoRE MOFs exhibited good generalizability to CSD MOFs due to the large similarity between the two databases; however, less satisfactory prediction performance was observed when extending the RF models to hypothetical MOFs, which lack chemical similarity to CoRE MOFs . Starting from experimentally measured water adsorption isotherms in 285 MOFs, Zhang et al developed ML models for atmospheric water harvesting; the transferability of the ML models was validated by out-of-sample predictions in newly reported MOFs, and finally the ML models were applied to screen ∼8,000 CoRE MOFs and identify top-performing candidates . For gas diffusion in MOFs, Krokidas et al benchmarked a ML model to predict the diffusivities of 12 different gases in 72 variants of zeolitic-imidazolate frameworks (ZIFs) .…”

Recyclable and Degradable Biomass-Based Water Vapor Sorbents for Efficient Atmospheric Water Harvesting

“…Water vapor sorbent plays a key role in the solar-driven SAWH for extracting atmospheric water from the surrounding air and producing freshwater under solar irradiation, even at low relative humidity (RH) environments in the desert . Advanced sorbents, such as zeolites, porous carbon, metal–organic frameworks (MOFs), − covalent organic frameworks (COFs), − polymers, − hygroscopic salts, and hybrid sorbents, − have been developed in recent years. In addition, the applications of water vapor sorbents have been extended much beyond freshwater production, including, but not limited to, heat transformation, − dehumidification, − and power generation. , As the need for SAWH sorbents grows, the production of sorbents is expected to increase sharply in the future.…”

Insights into Isotherm Step and Pore-Filling Mechanisms for Water Adsorption in Covalent Organic Frameworks

“…However, this can be made amenable with computational tools. 18 Recently, Zhang et al 19 developed a machine learning (ML) model that could identify top-performing MOFs as AWHs from the CoRE MOF database, trained on 344 experimental data of 285 MOFs. As mentioned earlier, the location of the isotherm step is a crucial factor in identifying COFs as AWHs.…”