Alauddin Ahmed

This repository contains the complete data and code for the accompanying paper. It includes the benchmark dataset for 37 molecular solvents: the pairwise miscibility matrix (592 miscible, 59 immiscible, and 15 undetermined classifications across 651 classified pairs) and the associated solvent property tables, in machine-readable CSV format. It also provides the claw-screening tool (clawscreen.py), which enumerates induced claws (K1,3 subgraphs) in a miscibility graph, together with the descriptor-model benchmarking scripts that reproduce the cross-validated AUC and average-precision results and generate all manuscript figures. All classifications derive from publicly available literature sources documented in the accompanying materials. Released under the MIT License.

This repository contains the complete data and code for the accompanying paper. It includes the benchmark dataset for 37 molecular solvents: the pairwise miscibility matrix (592 miscible, 59 immiscible, and 15 undetermined classifications across 651 classified pairs) and the associated solvent property tables, in machine-readable CSV format. It also provides the claw-screening tool (clawscreen.py), which enumerates induced claws (K1,3 subgraphs) in a miscibility graph, together with the descriptor-model benchmarking scripts that reproduce the cross-validated AUC and average-precision results and generate all manuscript figures. All classifications derive from publicly available literature sources documented in the accompanying materials. Released under the MIT License.

Solvent miscibility is commonly estimated using scalar polarity descriptors, including the Hildebrand solubility parameter, Snyder polarity index, and Godfrey miscibility number. On a compiled benchmark of 37 molecular solvents comprising 651 classified pairs, a single Hildebrand descriptor achieves a cross-validated AUC of 0.94. We show that this predictive success does not imply structural adequacy. The benchmark contains configurations where a single hub solvent, such as isopropanol, is miscible with three leaves ( n -hexane, glycerol, and propylene carbonate) that are mutually immiscible. This quartet forms an induced claw (K 1,3 ), a forbidden subgraph in any unit interval graph generated by a fixedthreshold scalar rule mapping solvents A and B to real-valued descriptor scores f (A) and f (B) , which predicts miscibility if their distance falls below a constant threshold τ (| f (A)− f (B)|< τ). By Roberts’ theorem, no such rule can represent this structure, regardless of its average predictive accuracy. The benchmark contains 55 induced claws across 8 immiscible triples, and these motifs persist under alternative classification boundaries for partially miscible pairs. A representation based on the dielectric constant and molar volume possesses sufficient geometric freedom to represent the witness claw, but this added dimensionality alone does not improve global predictive performance (ROC AUC = 0.830, compared with 0.834 for the dielectric constant alone). Incorporating hydrogen-bonding information instead raises the cross-validated ROC AUC to 0.948, while the three-dimensional Hansen representation achieves 0.959. These results illustrate that multidimensional descriptor spaces are not subject to the unit-interval obstruction that excludes one-dimensional fixed-threshold representations of graphs containing an induced claw.

, heat - would improve the efficiencies of numerous processes throughout multiple sectors of the global economy. Nevertheless, the development of these thermal storage devices remains at a relatively early stage. To engage more researchers in the development of these devices and to accelerate their commercialization, this review presents an introduction to the properties of thermal storage materials that absorb and release heat through thermochemical reactions. Thermochemical materials typically exhibit the largest energy densities among all approaches to material-based heat storage. Nevertheless, they suffer from limited reaction rates and poor cycle life. An additional challenge is the multiscale nature of the energy storage process, which ranges from atomistic interactions that govern the storage of heat through alteration of chemical bonds, to mesoscale processes that control the transport of mass and heat. Following an overview of general concepts related to thermal energy storage, emphasis is placed on describing properties relevant for low-temperature applications. These applications include domestic heat storage/amplification (hot water heating), adsorptive cooling (air conditioning), and heat-moisture recuperation. Subsequently, detailed introductions are provided to the mechanisms and materials relevant for the three primary approaches to low-temperature thermochemical storage, including: (i) absorption in solids (hydrates, ammoniates, and methanolates); (ii) adsorption in porous hosts (zeolites, metal-organic frameworks); and (iii) dilution in liquids. For each category, advantages and shortcomings of benchmark and emerging materials are discussed. Finally, challenges and opportunities are highlighted for research aimed at developing optimal materials for thermochemical energy storage.

Covalent organic frameworks (COFs), recognized for their potential in vehicular methane storage, typically require comprehensive and resource-intensive evaluations for capacity determination. Herein, we introduce a novel approach that synergizes symbolic regression with machine learning, augmented by descriptive statistics, systematic feature selection, and high-throughput computational techniques, to develop predictive equations for COF methane storage capacities. By strategically selecting a concise set of three to seven features based on their interpretability and measurability, we have developed a spectrum of symbolic regression-based equations to meet diverse precision requirements in forecasting COF methane capacities. The fidelity of these equations varies: an equation with three variables achieves a root mean square error (RMSE) of $10\phantom{\rule{4pt}{0ex}}\mathrm{c}{\mathrm{m}}^{3}$ (standard temperature and pressure, STP) per $\phantom{\rule{0.16em}{0ex}}\mathrm{c}{\mathrm{m}}^{3}$ and a mean absolute percentage error (MAPE) of 5%, while expanding to seven variables fine-tunes the RMSE to $8.1\phantom{\rule{4pt}{0ex}}\mathrm{c}{\mathrm{m}}^{3}$ (STP) per $\mathrm{c}{\mathrm{m}}^{3}$ and the MAPE to 4.2%. Our structured approach ensures accurate predictions that correspond to the resolution of available data. Upon applying our most precise equation to an extensive dataset of $449\phantom{\rule{0.16em}{0ex}}468$ COFs, we identified numerous COFs with exceptional methane storage capacities. The adaptability of our equations to integrate new data broadens their applicability beyond COFs, making them versatile tools for optimizing methane storage across a range of nanoporous materials.

A machine learning (ML) model is developed for predicting useable methane (CH 4 ) capacities in metal–organic frameworks (MOFs). The model applies to a wide variety of MOFs, including those with and without open metal sites, and predicts capacities for multiple pressure swing conditions. Despite its wider applicability, the model requires only 5 measurable structural features as input, yet achieves accuracies that surpass less-general models. Application of the model to a database of more than a million hypothetical MOFs identified several hundred whose capacities surpass that of the benchmark MOF, UMCM-152. Guided by the computational predictions, one of the promising candidates, UMCM-153, was synthesized and demonstrated to achieve superior volumetric capacity for CH 4 . Feature importance analyses reveal that pore volume and gravimetric surface area are the most important features for predicting CH 4 capacity in MOFs. Finally, a reverse ML model is demonstrated. This model predicts the set of elementary MOF structural properties needed to achieve a desired CH 4 capacity for a prescribed operating condition.

Evaluation of metal-organic frameworks (MOFs) for adsorbed natural gas (ANG) technology employs pure methane as a surrogate for natural gas (NG). This approximation is problematic, as it ignores the impact of other heavier hydrocarbons present in NG, such as ethane and propane, which generally have more favorable adsorption interactions with MOFs compared to methane. Herein, using quantitative Raman spectroscopic analysis and Monte Carlo calculations, we demonstrate the adsorption selectivity of high-performing MOFs, such as MOF-5, MOF-177, and SNU-70, for a methane and ethane mixture (95:5) that mimics the composition of NG. The impact of selectivity on the storage and deliverable capacities of these adsorbents during successive cycles of adsorption and desorption, simulating the filling and emptying of an ANG tank, is also demonstrated. The study reveals a gradual reduction in the storage performance of MOFs, particularly with smaller pore volumes, due to ethane accumulation over long-term cycling, until a steady state is reached with substantially degraded storage performance.