Separation of Rare-Earth Elements by Supported Liquid Membranes: Impacts of Soluble Iron, Aluminum, and pH in Low-Grade Feedstocks

Abstract

Low-grade residuals such as mine wastes and combustion ash are potential sources of critical metals such as rare-earth elements (REEs). Major challenges in the efficient recovery of REEs are the matrix interferences in the waste extracts that impede subsequent purification steps. This study evaluated feedstock matrix variables such as aqueous aluminum (Al), iron (Fe), and pH for their impact on neodymium (Nd) and erbium (Er) recovery flux by supported liquid membrane (SLM) separations, a type of liquid–liquid extraction method. We initially hypothesized that REE mass transfer would be lower at low [REE]/[Fe] and [REE]/[Al] molar ratios due to increased competition for chelation sites at the membrane interface. However, the results showed that the absolute Fe and Al concentrations, not the molar ratios, controlled Nd and Er mass transfer. The permeability coefficients of Nd and Er were most sensitive to the feedstock concentration of Fe3+ relative to Al3+ and Fe2+. The threshold Fe3+ concentration that resulted in reduced Nd and Er permeability was more than 100 times lower than the concentrations required for Al or Fe2+ to decrease REE permeability. REE recovery rates also increased with increasing pH of the feedstock. Separations performed with excess Fe3+ did not result in observable fouling at the membrane interface. Instead, the pH gradient across the membrane and the relative cation affinity for the chelator were the major drivers of mass transfer. These results provide insights for predicting REE mass transfer rates and SLM separation performance for extractions of low-grade feedstocks.

Recovery of Rare Earth Elements from Acid Mine Drainage with Supported Liquid Membranes: Impacts of Feedstock Composition for Extraction Performance

“…In contrast, increasing the pH of the AMD samples to pH 4 did not appreciably change P coefficients for Nd, the final product purity, or theoretical REE daily yields (Figure S10). This result is inconsistent with trends shown with synthetic feedstocks . One potential reason for this discrepancy is that the AMD pH values were adjusted immediately prior to the separation process, and the feed samples were not at equilibrium at the start of separation.…”

“…While the groupings are largely qualitative, the approximate cutoff values for each categorization variable are provided. Fe concentration was highlighted as a differentiating variable because of previous work , demonstrating the influence of Fe on REE flux and also because Fe levels spanned more than 2 orders of magnitude across all sites. For each set of experiments (as described in the Methods section), we tested at least one AMD sample from each site group.…”

“…SLM units also have a modular configuration that enables flexibility in process design, a feature that is especially relevant for geographically distributed REE resources like AMD. SLMs have been shown to selectively and efficiently separate REEs from a variety of synthetic mixtures and leachates of waste materials such e-waste and coal ash. − Previous work using SLMs to separate REEs from coal ash showed that a major rate-limiting step for REE mass transfer was the availability of cation binding sites at the feed–membrane interface and competition for these sites by other major cationic metals (e.g., Fe, Al, and Ca) . The pH gradient across the membrane has also been shown to increase metals mass transfer for SLM units that use DEHPA as the extractant. , …”

“…18 The pH gradient across the membrane has also been shown to increase metals mass transfer for SLM units that use DEHPA as the extractant. 21,22 The primary objective of this work was to determine the impact of AMD water quality on the extraction of REEs with SLMs. To the best of our knowledge, SLMs have not previously been tested for REE extraction from AMD.…”

“…Additional scanning electron microscopy with energy X-ray dispersive spectroscopic analysis of the types of membranes used in this study was performed in a previous work. 21 The membrane was then placed at the interface of two 250 mL glass vessels in the H-cell reactor. The AMD sample was placed in the feed side of the reactor, and a 5 mol L −1 HNO 3 stripping solution was placed in the product side.…”