The separation of metal complexes in solution is essential for a wide range of applications, one of which is the recycling of nuclear fuels. Due to the complexity of lanthanide and actinide separations, experimental and computational data for metal mixture separation remains limited, which makes the deployment of powerful machine-learning approaches challenging. Therefore, improved accelerated tools, software, and approaches are needed. We introduce a high-throughput workflow that addresses this challenge and enables the efficient screening of thousands of candidate ligands for enhanced liquid-liquid extraction. Our approach combines (1) execution on leadership-computing infrastructure, (2) a hierarchical simulation approach from multiple levels of theory and (3) Bayesian optimization for efficient sampling. The workflow is implemented in the open-source framework pyiron (http://pyiron.org), which disentangles the technical complexity and the scientific complexity, enabling rapid prototyping and up-scaling of simulation protocols. <br/>Here, we demonstrate the workflow on the lanthanide separation by employing a tight binding model for the initial screening of metal-ligand binding energies. After the initial screening the quality of the predictions is improved by characterizing the most promising candidate ligands with density functional theory (DFT). As a result, this workflow can increase the screening throughput by two orders of magnitude, while maintaining DFT-level accuracy for promising candidates. While this approach is currently demonstrated for separation sciences,  it is flexible and can readily be extended to other complex chemical processes.

High-throughput screening of ligand-metal-complexes for chemical science with pyiron

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.