DigiMOF: A Database of MOF Synthesis Information Generated via Text Mining

Abstract

The vastness of materials space, particularly that which is concerned with metal-organic frameworks (MOFs), creates the critical problem of performing efficient identification of promising materials for specific applications. Although high-throughput computational approaches, including the use of machine learning, have been useful in rapid screening and rational design of MOFs, they tend to neglect descriptors related to their synthesis. One way to improve the efficiency of MOF discovery is to data mine published MOF papers to extract the materials informatics knowledge contained within the journal articles. Here, by adapting the chemistry-aware natural language processing tool, ChemDataExtractor (CDE), we generated an open-source database of MOFs focused on their synthetic properties: the DigiMOF database. Using the CDE web scraping package alongside the Cambridge Structural Database (CSD) MOF subset, we automatically downloaded 43,281 unique MOF journal articles, extracted 15,501 unique MOF materials and text mined over 52,680 associated properties including synthesis method, solvent, organic linker, metal precursor, and topology. This centralised, structured database reveals the MOF synthetic data embedded within thousands of MOF publications. The DigiMOF database and associated software are publicly available for other researchers to conduct further analysis of alternative MOF production pathways and create additional parsers to search for other desirable properties.

Unveiling the Synthesis Patterns of Nanomaterials: A Text Mining and Meta-Analysis Approach with ZIF-8 as a Case Study

“…"ZIF-8nano," 79 confounding attempts for automated identification of reaction products using regular expressions. 33 While this issue will undoubtably be resolved by the adoption of transformer-based language models such as BERT 80 and GPT-4, 81 such models became available only recently and the scientific community, including our group, is in the process of probing their extension to scientific data mining. In fact, the current study highlighted a number of issues with the current structure and completeness of reported synthetic protocols, understanding of which will be very helpful in engineering and fine-tuning GPT-based models.…”

“…To produce a corpus of ZIF-8 synthesis protocols, we initially followed established methods to download collections of papers and identify synthesis protocols within them. 32,33 Synthesis papers were identified by searching the SCOPUS database using Elsevier's elsapy software (https://github.com/ElsevierDev/elsapy). Papers were identified using the search term "ZIF OR zeolitic imidazol* AND synthesis," returning 4198 results.…”

“…Once downloaded, synthesis paragraphs were identified using ChemDataExtractor2.1 10 according to previously developed protocols for identifying MOF synthesis methods. 32,33 In this procedure, chemical named entity recognition was performed using BERT 47 to identify potential reagents, and part-of-speech (POS) tagging was carried out on the remaining tokens to interpret sentence grammar. Chemical quantities were identified from the POS tags as CD-NN bigrams (phrases consisting of a cardinal number followed by a noun), and regex matching of the noun against a library of SI units.…”

“…6 To this end, large-scale post-hoc analyses of experimental MOF synthesis protocols have recently been developed. 32,33 These studies apply NLP to the underlying publications in the CSD MOF subset to interpret their synthesis protocols, identifying such details as solvents used, specific reagents, solvents, and reaction parameters. As a result, broad descriptive statistics about the synthesis strategies to produce MOFs have been developed, 33 and even predictive models to suggest synthesis parameters for novel MOF materials when given a hypothetical structure.…”

“…32,33 These studies apply NLP to the underlying publications in the CSD MOF subset to interpret their synthesis protocols, identifying such details as solvents used, specific reagents, solvents, and reaction parameters. As a result, broad descriptive statistics about the synthesis strategies to produce MOFs have been developed, 33 and even predictive models to suggest synthesis parameters for novel MOF materials when given a hypothetical structure. 32 While these approaches give an excellent overview of the field of MOFs in general, they are vulnerable to bias in the papers submitting to the CSD.…”