.png){kind=icon}
.png){kind=icon}
Return
Machine-Learning-Enabled Ligand Screening for Cs/Sr Crystallizing Separation
.png){kind=icon}
PRE
AI
DOI：10.1021/acs.inorgchem.3c01564.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The reprocessing of spent nuclear fuel is critical forthe sustainabilityof the nuclear energy industry. However, several key separation processespresent challenges in this regard, calling for continuous researchinto next-generation separation materials. Herein, we propose a high-throughputscreening framework to improve efficiency in identifying potentialligands that selectively coordinate metal cations of interest in liquidwastes that considers multiple key chemical characteristics, includingaqueous solubility, pK (a), and coordinationbond length. Machine-learning models were designed for the fast andaccurate prediction of these characteristics by using graph convolutionand transfer-learning techniques. Suitable ligands for Cs/Sr crystallizingseparation were identified through the computational funnel,and several top-ranking, nontoxic, low-cost ligands were selectedfor experimental verification. In this work,a high-performance machine learning-enabledscreening framework for spent nuclear fuel reprocessing was proposed.In the case of Cs/Sr separation, the crystallization process was designedto incorporate key chemical properties, aiming to create a chemicallyinterpretable computational funnel. Promising ligands were screenedand validated experimentally to demonstrate the effectiveness of ourmethod. A combination of both computational and experimental verificationprovided a time-, resource-saving, and environment-friendly approachfor chemical design and discovery.
Keywords:
SOLUBILITY PREDICTION
AQUEOUS SOLUBILITY
SOLVENT-EXTRACTION
CHEMISTRY
ROUTES