Machine-learning-supported analysis of synergistic extraction systems towards enhanced selectivity of lithium extraction from brines

Abstract

The development of technologies concerned with extraction and separation processes aimed at the sustainable production of rare metals, as well as at metal recycling, is in high demand. This study presents machine-learning-supported analysis of the experimental data on synergistic binary extraction systems for selective extraction of lithium from brines. We consider the narrow class of extraction systems that combine β-diketonate ligands (4,4,4-trifluoro-1-phenyl-1,3-butanedione (HBTA), 2-thenoyl-trifluoroacetone (HTTA), 1-heptyl-3-phenyl-1,3-propanedione (LIX54), 2,2-dimethyl-6,6,7,7,8,8-heptafluoro-3,5-octanedione (HFDOD)) and neutral organophosphorus ligands, such as trioctylphosphine oxide (TOPO), tributyl phosphate (TBP), triphenylphosphine oxide (TPPO) and trialkylphosphine oxide (TRPO). In this study, an analysis of the literature on the synergistic systems published to date was provided. This analysis has allowed distillation of the common characteristics of the formed hydrogen-bond-supported associates for the binary systems investigated to date. These results readily fit into the theory of eutectics and the chemistry of solvation processes. Currently, an urgent goal of the experimental research in this field is the optimization of processes that allow the selective extraction of lithium from brines of various compositions, including brines containing both alkali and alkaline earth metals. The benefits of liquid–liquid extraction and separation methods, which are concerned with the capacity of the corresponding systems to extract target metals from diluted media, require a deep understanding of the processes occurring at the interface of the two immiscible liquid phases, as well as in both the aqueous and organic phases themselves. This allows the recommendation of appropriate compositions of binary systems, along with the corresponding technological parameters of extraction and separation for certain brine compositions, using machine learning.