Rapid and Accurate Machine Learning Recognition of High Performing Metal Organic Frameworks for CO2 Capture

详细信息    查看全文

• 作者：Michael Fernandez ; Peter G. Boyd ; Thomas D. Daff ; Mohammad Zein Aghaji ; Tom K. Woo
• 刊名：The Journal of Physical Chemistry Letters
• 出版年：2014
• 出版时间：September 4, 2014
• 年：2014
• 卷：5
• 期：17
• 页码：3056-3060
• 全文大小：281K
• ISSN：1948-7185

文摘

In this work, we have developed quantitative structure鈥損roperty relationship (QSPR) models using advanced machine learning algorithms that can rapidly and accurately recognize high-performing metal organic framework (MOF) materials for CO₂ capture. More specifically, QSPR classifiers have been developed that can, in a fraction of a section, identify candidate MOFs with enhanced CO₂ adsorption capacity (>1 mmol/g at 0.15 bar and >4 mmol/g at 1 bar). The models were tested on a large set of 292鈥?50 MOFs that were not part of the training set. The QSPR classifier could recover 945 of the top 1000 MOFs in the test set while flagging only 10% of the whole library for compute intensive screening. Thus, using the machine learning classifiers as part of a high-throughput screening protocol would result in an order of magnitude reduction in compute time and allow intractably large structure libraries and search spaces to be screened.

Keywords:

virtual screening; nanoporous materials; MOFs; QSPR models; computer aided design; gas adsorption