Faculty Advisor: William Phillip,

Diafiltration Cascades for Critical Mineral Separations to Support a Clean Energy Transition

Research Significance: The renewable energy solutions required to promote widespread electrification and meet sustainability goals are increasing the demand for critical minerals. Unfortunately, current mineral processing techniques (e.g., liquid-liquid extraction) are energy intensive and utilize large quantities of hazardous chemicals. These requirements are at odds with the sustainability goals driving increased demand for critical minerals. Membrane-based separations offer opportunities to replace environmentally damaging unit operations with distributed, fit-for-purpose processes that enable circular mineral economies.

Research Scope: The objective of this research is to inform the design of diafiltration cascades capable of simultaneously separating and purifying lithium from Salar brines. In particular, these separations necessitate that membranes function within complex multicomponent environments. However, traditional experiments aimed at elucidating multicomponent interactions are time and resource intensive and are often incomplete. Here, automated high-throughput experiments will be utilized to identify the onset and concentration dependencies of these multibody interactions in multicomponent solutions.

Background: Current mineral processing technologies reduce the sustainability impact of the green technologies they inspire. Lithium recently found in the volcanic sedimentary at McDermitt caldera (Nevada, USA) is one of the largest reserves in the world. [1] Yet, securing domestic critical mineral supply chains will require processes that conform to increasingly stringent environmental and human rights regulations. The extraction and purification of critical minerals has changed little over the past century, leaving in place processes that are inefficient and environmentally damaging. One example is the processes used to obtain lithium from Salt Lake brines. [2] First, the brines are pumped out of aquafers and into large evaporating ponds. As water evaporates, the concentration of salts increases. Competing ions (e.g., potassium, sodium, and magnesium) precipitate, simultaneously purifying and enriching the concentration of lithium. After the solution reaches ~6% lithium, reagents are added to remove magnesium and calcium impurities. Subsequently, the addition of Na2CO3, followed by filtration of the precipitate, produces Li2CO3. Depending on the end-use application, liquid-liquid extraction, a process that uses large volumes of hazardous solvents, is used to remove contaminants and achieve higher purities. [2] One opportunity to improve this process lies in using membranes capable of fractionating brines into multiple product (e.g., lithium) and by-product (e.g., potassium) streams, ultimately decreasing the use of energy intensive reagents and removing the need for liquid-liquid extraction steps.

Diafiltration cascades simultaneously recover and purify solute. Membranes separate solutes by modulating their relative flux. Typically, low molecular weight solutes permeate across the membrane preferentially, and high molecular weight solutes are retained. While filtration processes are well suited for solute-solvent separations (e.g., concentrating a high value product in the retentate, or recovering solvent as the permeate in desalination), solute-solute separations are difficult to accomplish. In particular, the solvent permeates through the membrane far more rapidly than the time necessary for solute-solute separations to occur. This drawback has been addressed through the design of diafiltration modules that systematically dose solvent (i.e., the diafiltrate) over the length of the membrane. [3,4] Previously, I mentored an undergraduate researcher to develop equations describing the concentration of solute over the length of the module for a size-selective membrane. [4] After experimentally validating the model, simulations showed that cleverly engineering the connection between multiple modules could be leveraged to fractionate similarly sized solutes or purify high value products. [4]

While these staged diafiltration modules can achieve recoveries and purities equal to current processes, commercial membranes suggested for lithium recovery applications do not conform to the assumptions built within the current models. Specifically, models need to be adapted to (1) capture the feed dependent performances of membranes that separate solute based on charge and (2) incorporate the multi-body interactions that arise in complex feed streams. [5] As such, this proposal aims to increase the information gained from membrane characterization experiments and use these techniques to inform process models and materials development.

Research Objectives

Objective 1: Implement an in-situ permeate conductivity probe to increase the information provided by diafiltration experiments. I previously reported on the design and implementation of an automated, dynamic diafiltration device that was developed using tailor-made 3D printed parts in conjunction with commercially available test cells. [3] Working in collaboration with the Dowling group, I was able to show this single stage diafiltration device could systematically change the concentration of the retentate with respect to time. While the current apparatus provides snapshots of membrane performance with respect to the feed concentration, continuous readings of the permeate concentration, analogous to those of the retentate, are necessary to connect these points. This limitation can be overcome by utilizing 3D printing technology to place a conductivity probe immediately after the membrane. Preliminary results show a probe incorporated into the membrane holder works properly when calibrated off-line. During experiments, the formation of bubbles remains a challenge. This objective will use surface modification techniques (e.g., ozone/plasma treatment) to prevent bubbles from sticking to the probe surface.

Objective 2: Establish material-property relationships with diafiltration experiments. Commercially available nanofiltration membranes for Li separations rely on electrostatic interactions between fixed charges on the membrane surface and ions dissolved in solution to drive a separation. [5] Namely, co-ions (i.e., ions with the same charge as the fixed charge) are repelled while counter ions are attracted to the membrane, consequently promoting, or hindering ion partitioning, respectively. As a result, performance depends on the charge density of the membrane and the ionic strength of the feed solution.

Dynamic diafiltration experiments will elucidate the relationship between membrane charge and performance. Using commercial membranes, I will rapidly explore membrane performance for salt concentrations relevant to lithium brines found in the United States (e.g., in Clayton Valley Nevada: 0–1.5 M Mg2+, 1–25 mM Li+) using diafiltration experiments. The experiments will provide relationships that describe membrane properties (i.e., the sieving coefficient, S=c_permeate/c_retentate ) as a function of the feed concentration. The concentration profiles of the retentate and permeate will be obtained from the in-situ conductivity probes at five second intervals. The resulting relationships will be embedded into the diafiltration cascade models developed previously. While not my focus, future research can use this information to reverse engineer membranes and inform their molecular design.

Objective 3: Create high quality data sets to elucidate transport in multi-component feed streams. Solute transport is complicated by multi-component interactions. [5] For instance, charged membranes reject single component, monovalent ion solutions at ~80%. Yet, when trace amounts of the monovalent ion are present in multicomponent feeds, the rejection decreases. [5] There exists a lack of information and systematic studies that describe the ion concentrations and monovalent to divalent ion ratios that lead to this phenomenon. Therefore, diafiltration experiments using lithium chloride and magnesium chloride as model solutes will be used to probe the onset of this phenomenon. Specifically, diafiltration experiments will be conducted with a pure LiCl feed and a high (i.e., 1.5 M) MgCl2 diafiltrate. In this design, the concentration of MgCl2 will increase systematically over the length of the experiment. By measuring the LiCl rejection with respect to the ionic strength and monovalent: divalent ion ratio, I can begin elucidating the solution conditions at which transport of LiCl is affected by the presence of other ions in the solution. Ultimately, the information obtained from the diafiltration experiments, paired with physics-based models will be used to inform how membrane materials and processes are affected by multicomponent mixtures.

(1) Benson et al., Science Advances, 2023. (2) Meng et al., Miner. Process. Extr. Metall. Rev., 2019. (3) Ouimet et al. J. Membr. Sci., 2022. (4) Kilmartin et al., Ind. Eng. Chem. Res., 2022. (5) Foo et al., Environ. Sci. Technol., 2023.