MOF-GRU: A MOFid-Aided Deep Learning Model for Predicting the Gas Separation Performance of Metal–Organic Frameworks

ACS Applied Materials & Interfaces, Год журнала: 2023, Номер 15(51), С. 59887 - 59894

Опубликована: Дек. 12, 2023

The remarkable versatility of metal–organic frameworks (MOFs) stems from their rich chemical information, leading to numerous successful applications. However, identifying optimal MOFs for specific tasks necessitates a thorough assessment attributes. Conventional machine learning approaches MOF prediction have relied on intricate and structural details, hampering rapid evaluations. Drawing inspiration recent advancements exemplified by Snurr et al., wherein text string was used represent (MOFid), we introduce MOFid-aided deep model, named the MOF-GRU model. This founded natural language processing principles utilizing gated recurrent unit architecture, leverages serialized representation forecast gas separation performance. Through focused study CH4/N2 separation, substantiate efficacy this approach. Comparative assessments against traditional techniques underscore our model's superior predictive accuracy its capacity handle extensive data sets adeptly. model remarkably uncovers latent structure–performance relationships with only sequences, obviating necessity three-dimensional (3D) information. Overall, judicious design empowers efficient utilization, thereby hastening discovery high-performance materials tailored

Язык: Английский

---

Accelerating Metal–Organic Framework Selection for Type III Porous Liquids by Synergizing Machine Learning and Molecular Simulation

ACS Applied Materials & Interfaces, Год журнала: 2023, Номер 15(48), С. 56253 - 56264

Опубликована: Ноя. 21, 2023

MOF-based type III porous liquids, comprising MOFs dissolved in a liquid solvent, have attracted increasing attention carbon capture. However, discovering appropriate to prepare liquids was still limited experiments, wasting time and energy. In this study, we used the density functional theory molecular dynamics simulation methods identify 4530 MOF candidates as core database based on idea of prohibiting pore occupancy by [DBU-PEG][NTf2] ionic liquid. Based high-throughput simulation, random forest machine learning models were first trained predict CO2 sorption CO2/N2 selectivity screen liquids. The feature importance inferred Shapley Additive Explanations (SHAP) interpretation, ranking top 5 descriptors for sorption/selectivity trade-off (TSN) gravimetric surface area (GSA) > porosity metal fraction size distribution (PSD, 3.5–4 Å). RICBEM predicted be one candidate preparing with capacity 20.87 mmol/g 16.75. experimental results showed that RICBEM-based successfully synthesized 2.21 63.2, best capture performance known date. Such screening method would advance cores solvents different applications addressing corresponding factors.

Язык: Английский

---

In Silico-Directed Design and Experimental Validation of an IL/UiO-66 Nanocomposite with Exceptional CO₂ Selectivity across a Wide Pressure Range

ACS Applied Nano Materials, Год журнала: 2024, Номер unknown

Опубликована: Сен. 13, 2024

Язык: Английский

---

Solvent‐Directed Construction of a Nanoporous Metal‐Organic Framework with Potential in Selective Adsorption and Separation of Gas Mixtures Studied by Grand Canonical Monte Carlo Simulations

ChemPlusChem, Год журнала: 2023, Номер 89(1)

Опубликована: Окт. 21, 2023

Abstract In this report, a microporous metal–organic framework of [Ca(TDC)(DMA)] n ( 1 ) and two‐dimensional coordination polymer [Ca(TDC)(DMF) 2 ] ), (TDC 2− =Thiophene‐2,5‐dicarboxylate, DMA=N, N'‐dimethylacetamide DMF=N, N'‐dimethylformamide) based on Ca(II) were designed by the effect solvent, X‐ray analysis was performed for single crystals . Then, compound synthesized in three different methods identified with set analyses. Compared to other adsorbents, MOFs are widely used field adsorption separation various gases due series distinctive features such as diverse adjustable structures pores dimensions, high porosity surface area regular distribution active sites. Therefore, ability uptake (CH 4 , CO C H 2, N several binary mixtures (CO /CH /N /H /C investigated using Grand Canonical Monte Carlo simulations. Volumetric gravimetric isotherms operating conditions, isosteric heat (q st chemical potential each thermodynamic state, snapshots during simulation process reported all cases. The results obtained from indicate that has capacity (16 mmol g −1 (12.5 (6.6 (5 CH (1.5 at bar. It also performs well separating mixtures, which can be attributed presence open metal sites nodes their electrostatic tendency interact containing higher quadrupole dipole moment compared components mixture.

Язык: Английский

---

Toward Greener Flow Assurance: Review of Experimental and Computational Methods in Designing and Screening Kinetic Hydrate Inhibitors

Energy & Fuels, Год журнала: 2024, Номер 38(18), С. 17191 - 17223

Опубликована: Сен. 6, 2024

Язык: Английский

---