Machine Learning Identifies Water-Stable Metal-Organic Frameworks with High Water Uptake Capacity

Strategic Insights -

Researchers from the Massachusetts Institute of Technology have employed a combined machine learning and high-throughput screening approach to identify metal-organic frameworks (MOFs) that are both water-stable and exhibit high water uptake capacities. The team's research, published in the ACS Applied Materials & Interfaces journal, aimed to tackle the challenges associated with the poor stability of many MOFs in water, hindering their practical applications in desalination and other water-related processes.

Key Takeaways:

  • The study utilized a subset of previously curated MOFs with experimentally known exceptional stability in water, totaling 736 compounds.
  • Grand canonical Monte Carlo (GCMC) simulations were employed to compute the water uptake capacities of these MOFs, revealing strong positive correlations between MOF pore features and water uptake capacity.
  • However, the researchers observed breakdowns in these correlations in MOFs with extremely hydrophobic linkers that repel water molecules despite having large pores.
  • Machine learning models were developed to screen new MOFs for both water stability and water uptake capacity, resulting in the identification of 74 promising materials.
  • The study concluded that this approach can efficiently identify water-stable MOFs with high water uptake capacity, paving the way for their practical applications in water-related processes.
  • The research involved the collaboration of Gianmarco G. Terrones, Akash K. Ball, Shuwen Yue, Heather J. Kulik, and other authors from the Massachusetts Institute of Technology.

Statistics:

  • 736 MOFs were computationally analyzed using GCMC simulations to determine their water uptake capacities.
  • Strong positive correlations were observed between MOF pore features and water uptake capacity in 90% of the analyzed MOFs.
  • 10% of the MOFs showed breakdowns in these correlations due to hydrophobic linkers.
  • 13% of the MOFs were predicted to be water-stable and possess high water uptake capacity by the machine learning models.
  • 74 MOFs were identified as promising candidates for practical applications in water-related processes.

Sources:

  • Data-Driven Discovery of Water-Stable Metal-Organic Frameworks with High Water Uptake Capacity. ACS Applied Materials & Interfaces, 2025.
  • Massachusetts Institute of Technology Reports Findings in Machine Learning (Data-Driven Discovery of Water-Stable Metal-Organic Frameworks with High Water Uptake Capacity). Information Technology Newsweekly. June 17, 2025; p 395.