Beyond the BET Analysis: The Surface Area Prediction of Nanoporous Materials Using a Machine Learning Method

Abstract

Surface areas of porous materials such as metal−organic frameworks (MOFs) are commonly characterized using the Brunauer− Emmett−Teller (BET) method. However, it has been shown that the BET method does not always provide an accurate surface area estimation, especially for large-surface area MOFs. In this work, we propose, for the first time, a data-driven approach to accurately predict the surface area of MOFs. Machine learning is employed to train models based on adsorption isotherm features of more than 300 diverse structures to predict a benchmark measure of the surface area known as the true monolayer area. We demonstrate that the ML-based methods can predict true monolayer areas significantly better than the BET method, showing great promise for their potential as a more accurate alternative to the BET method in the structural characterization of porous materials.

Monte Carlo simulations for water adsorption in porous materials: Best practices and new insights

“…The choices for adsorbent materials span several classes of materials including, but not limited to, zeolites, 4 composites of hygroscopic salts in gels 13,14 or activated carbons, 15 and metalorganic frameworks (MOFs). 16 MOFs have particularly drawn a great deal of interest because of their attractive properties such as large surface areas 17,18 and the tunability of their structural topologies and chemical functionalities. [19][20][21][22][23][24] To date, tens of thousands of MOFs have been reported.…”

Synthesis of covalent organic frameworks using sustainable solvents and machine learning

“…Thus, classical ab initio DFT calculations would require overly complex methods to quantify the properties of the COFs. 26 To overcome the issues with solvent dependency, we used QSPR computational tools to predict both the surface area and to verify if the resultant COF can be synthesized in the crystalline form. We hypothesized that by determining the structure of the solvent and the structure of the COF, a predictive relationship could be drawn while other parameters can be kept constant.…”

MOFSimplify, machine learning models with extracted stability data of three thousand metal–organic frameworks

“…However, a lack of systematic naming 53 , 54 in MOF chemistry (e.g., HKUST-1 and Cu-BTC are the same MOF) has limited the use of NLP-based named entity recognition for the design of new MOFs. While NLP has worked well for identifying MOF properties such as surface area through their unique units 55 , 56 , human interpretation of the structure name is required to relate extracted properties back to the original structure 55 , 56 . Due to challenges in mapping MOF names to structures 53 , 54 , coupled with the lack of unique units or measurements for stability assessments, no efforts have collated data on MOF stability.…”