Multidimensional ligand-assisted chromatographic method for the purification of rare earth elements and other metal ions from waste magnets

Abstract

A method for substantially recovering rare earth elements (REEs) from a magnet, comprising: first dissolving the magnet to obtain a solution containing Nd, Pr, and Dy; then equilibrating a first column with a Cu2+ solution to obtain a first equilibrated column; introducing the solution into the first equilibrated column; and introducing a ligand solution into the first equilibrated column to establish three bands of different solution compositions in the column, the three bands including a Dy / Nd mixed band, a first pure Nd band, and a Nd / Pr mixed band. Next, the steps of sending the Dy / Nd mixed band to a second column containing a Cu2+ solution, introducing a ligand solution into the second column to establish a pure Dy band and a second pure Nd band in the second column, and sending the Nd / Pr mixed band to a third column containing a Cu2+ solution.

Description

[Technical Field]

[0001] Government funding This invention was made with government support under SP8000-18-P-0007 awarded by the Defense Logistics Agency. The government has certain rights in this invention. CROSS-REFERENCE TO RELATED APPLICATIONS This patent application claims priority to co-pending U.S. Provisional Patent Application No. 62 / 982,807, filed February 28, 2020. The present novel technology relates generally to the field of chemical industry and more particularly to a method for recovering rare earth elements Nd, Pr and / or Dy from waste materials. [Background technology]

[0002] Rare earth elements (REEs) are valuable to the high-tech and clean energy industries. Three REEs, Pr, Nd, and Dy, are useful components of magnets and are widely used in the electronics, automotive, and wind energy industries, among others. REE production is highly concentrated in a few countries, putting the supply chain for REEs in general, and Pr, Nd, and Dy in particular, at risk. Once produced and used, waste magnets are almost exclusively disposed of in landfills. While this disposal of REEs in landfills represents a recoverable local source of REEs, there is no efficient recovery mechanism. Therefore, there remains a need to develop alternative supply streams for REEs in general, and for Pr, Nd, and Dy in particular. This novel technology addresses this need. Summary of the Invention

[0003] Three rare earth elements (REEs), neodymium (Nd), praseodymium (Pr), and dysprosium (Dy), are essential components of permanent magnets, which are widely used in electronics, motors, hybrid vehicles, generators, televisions, sensors, wind turbines, etc. Traditional methods for producing high-purity REEs use two-phase liquid-liquid extraction methods, which require thousands of mixer-settler units arranged in series or parallel, generating large amounts of toxic waste. This novel technology is a novel ligand-based chromatography (LBC) zoning method developed for the production of high-purity (>99%) Nd, Pr, and Dy from a crude REE mixture derived from waste magnets with high yields (>99%) and high sorbent productivity. Ligands selective for REEs can be added to the mobile phase to enable ligand-assisted displacement (LAD), and the REEs can be recovered in a ligand-assisted elution (LAE) step, or the ligands can be immobilized on the stationary phase to enable ligand-bound displacement (LBD) in sequential elution mode (LB-SMB). Ligands with affinity for one or more REEs include citric acid, aminopolycarboxylic acids (e.g., ethylenediaminetetraacetic acid (EDTA), diethylenetriaminepentaacetic acid (DTPA), nitrilotriacetic acid (NTA), etc.), bicine, etc., and other REE-selective extractants such as HDEHP, DGA, etc., and combinations thereof.

[0004] The new method introduces a two-zone ligand-assisted displacement chromatography (LAD) system that uses improved correlations to predict the minimum column length required to reach a consistent pattern state in LAD. A selectivity-weighted composition factor-based zoning method allows the two-zone design to achieve productivity two orders of magnitude higher than single-column designs. The design and simulation method is based on first principles and intrinsic (or scale-independent) engineering parameters. These can be used to design processes for a wide range of feed compositions or production scales. The total productivity of the two-zone LAD can exceed 100 kg / m3 / day of REE, which is 100 times higher than conventional extraction methods.

[0005] For LAD, sorbents include micropores, sulfonic acid, aminosulfonic acid functional groups, etc. For LBD, the sorbent IDA resin is highly selective for Cu, Ni, Co but less selective for REEs; EDTA, DTPA, and / or phosphate ligand-bound porous silica, DGA bound to PMMA, and EDTA bound to PS or polymer resins with amine functional groups. LAD and / or LBD for the purification of ternary mixtures requires only three chromatography columns, a safe extractant, EDTA, and other environmentally friendly chemicals. Most of the chemicals are recyclable, generating little waste. This method has the potential to efficiently and environmentally purify REEs from waste magnets. This method could also help transform the current linear REE economy (from ores to pure REEs to products to landfills) into a circular and sustainable REE economy.

[0006] [Table 1] [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 shows the breakdown of REEs (by mass) in various applications. [Figure 2] 1 is a graph showing Sherwood's projections of process costs for purifying REEs from magnets. [Figure 3] Figure 1 shows the REE magnet supply chain that this new technology will impact. [Figure 4] 1 is a graph showing the prices of the REEs Nd, Pr, and Dy as a function of mass fraction. [Figure 5] 1 is a graph showing the REE recovery yields comparing known methods with new technologies. [Figure 6A] Graph showing LAD loaded with free ions, with the REEs with the highest affinity eluting first. [Figure 6B] Graph showing LAD loaded with ligand-chelated REE ions, with REEs with high ligand affinity eluting first. [Figure 6C] Graph showing LBD loaded with free REE ions, with the REEs with the lowest ligand affinity eluting first. [Figure 7] Graph showing a general map of consistent patterns for LCD recovery of REEs. [Figure 8] Figure 1 shows the productivity versus yield for REE recovery for a fixed pattern LAD design to produce 99% pure product from a selected ternary complex REE mixture using a single column with a fixed breakthrough cut (θ = 0.05). [Figure 9] FIG. 1 is a schematic diagram showing an overview of the novel multi-zone constant pattern design methodology. [Figure 10] Figure 1 shows the schematic design of a two-zone LAD for separating REE crude simulants derived from waste magnets. Zone I is intended to recover high-purity Nd. In zone II, column II-A is intended to recover high-purity Dy and Nd. Column II-B is intended to recover high-purity Nd and Pr. [Figure 11]1 is a graph showing ligand elution time as a function of concentration. [Figures 12A-12F] Separation of an equimolar mixture of Nd and Pr (0.3N) using 0.09M EDTA-Na (pH 9). Experiments were designed for various X values: (a) 8.40; (b) 10.52; (c) 14.04; (d) 20.98; (e) 32.15. The mass transfer zone lengths in the experiment and simulation are compared in (f). [Figures 13A-13C] 1 is a graph showing the elution profile of the LAD test. (a) Zone I: Most of the Nd is obtained. Two mixed bands, Dy-Nd and Nd-Pr mixed bands (monochrome shades), were sent to Zone II for further separation. (b) Zone II Column A: Dy and Nd are separated from the mixed band produced in Zone I. (c) Zone II Column B: Nd and Pr are separated from the mixed band produced in Zone I. In both Zone II Column A and Zone II Column B, the mixed bands are recycled to their original feed solution (the mixed band from Zone I). [Figure 14] FIG. 1 is a schematic diagram showing the separation mechanism of LAD for rare earth elements. [Figure 15] 1 is a graph showing that the mass transfer zone length for a given system initially decreases with increasing column length, and then no longer decreases after the column length reaches a minimum column length forming a consistent pattern. [Figure 16] 1 is a graph showing a general map of a fixed pattern. [Figure 17] 1 is a flowchart of productivity optimization in constant pattern design. [Figure 18] 1 is a graph comparing calculated Nd yield (dashed curve) and productivity (solid curve) versus linear velocity u0 for fixed column lengths for separating a ternary REE mixture. [Figure 19] 1 is a graph showing elution time as a function of concentration for breakthrough cutoff θ for collecting product from column effluent in LAD. [Figure 20] FIG. 1 illustrates impurities from adjacent components. [Figures 21A-21E] FIG. 10 shows parameters from Example A. [Figures 22A-22D] FIG. 10 shows parameters from Example B. [Figures 23A-23B] FIG. 10 shows parameters from Example C. [Figures 24A-24C] FIG. 10 shows parameters from Example D. [Figure 25] 1 is a flow chart illustrating a first partitioning strategy for recovering all components of a mixture. [Figure 26] 1 is a flow chart illustrating a second partitioning strategy for recovering all components of a mixture. [Figure 27] 1 is a flow chart illustrating a partitioning strategy for recovering a group of adjacent components of a mixture. [Figure 28] 1 is a flow chart illustrating a partitioning strategy for recovering multiple groups of adjacent components of a mixture. DETAILED DESCRIPTION OF THE INVENTION

[0008] For the purposes of promoting an understanding of the principles of the novel technology and setting out the best presently understood modes of carrying it out, reference will now be made to the embodiments illustrated in the drawings and specific language will be used to describe the embodiments. It is understood, however, that no limitation of the scope of the novel technology is thereby intended, and that such changes and further modifications in the illustrated devices and such further applications of the principles of the novel technology illustrated herein would normally occur to one skilled in the art to which the novel technology pertains. Rare earth elements (REEs) include the 15 elements of the lanthanum series plus scandium (Sc) and yttrium (Y). They are in demand for use in magnets, metal alloys, abrasives, catalysts, ceramics, phosphors, and more, all of which are important in high-tech and clean energy applications. While the market capitalization of REEs is only $1.5 billion, the products that require them are worth more than $5 trillion.

[0009] Approximately 30% by mass of rare earth elements are used in the production of NdFeB permanent magnets, which are widely used in hard drive disks, wind turbines, and electric vehicle motors. The three REEs commonly found in NdFeB magnets are Nd, Dy, and Pr. The first two are classified as critical materials by the U.S. Department of Energy, which aims to produce 20% of its electricity from wind energy by 2030. Each direct-drive permanent generator for 1 MW of power requires 160–650 kg of permanent magnets made from Nd, Pr, and Dy. Such generators are increasingly being used in offshore and large onshore facilities due to their high energy efficiency. Rare earth permanent magnets are also used in the motors of hybrid and electric vehicles. It is estimated that each hybrid or electric vehicle requires 1.5–2.5 kg of REE-based permanent magnets. The market for REEs, which are important in magnets, is highly volatile because their production is concentrated in a few regions around the world. Currently, China controls over 80% of the world's REE supply and over 95% of the supply of rare earth alloys and magnets. This dominance could lead to over 90% of high-tech products being produced in China by 2040. REE production in other regions is limited by a lack of technical expertise for production, high capital costs, and the significant environmental impact of traditional methods.

[0010] REE production begins with the beneficiation and concentration of REE ores, such as bastnaesite, a rare earth fluorocarbonate containing approximately 7-8% rare earth oxide (REO) equivalents. After crushing and grinding, chemical vapor conditioning, flotation, and washing, the bastnaesite can be upgraded to a concentrated REE crude containing approximately 60% REO. The REE crude is further digested, refined, and smelted to pure metal. The REE purification process is the most difficult step because the REEs present in the crude product have similar physical and chemical properties. Current industrial purification procedures still use liquid-liquid extraction methods developed in the 1950s. These methods require thousands of mixer-settler units to produce high-purity REEs. Liquid-liquid extraction methods are difficult to adapt to different feedstocks or production scales. These methods are energy- and chemical-intensive, require toxic extractants, organic solvents, and concentrated extracting acids, and generate large amounts of acidic and toxic waste. Approximately 30 tons of wastewater are released into the environment to produce one ton of rare earth oxides. The extraction process also consumes large amounts of ammonia to saponify the organophosphorus extractant and hydrochloric acid to extract the extractant. On average, only 35% of the ammonia used in this process is recovered as ammonium chloride. Life cycle analysis studies have revealed that in the production of REEs from bastnaesite, the refining steps in conventional processes account for approximately one-third of the total environmental impacts in terms of global warming, carcinogenic and non-carcinogenic human toxicity, eutrophication and ecotoxicity. Furthermore, they are responsible for 70% of the ozone depletion impacts of the production process.

[0011] Recycling of waste magnets is a viable method for producing valuable REE materials. Approximately 300,000 tons of REEs are present in permanent magnets produced to date, and approximately 3,300-6,600 tons of REEs can be recovered from waste magnets annually. As solid-state disks (SSDs) begin to dominate the market, it is predicted that many HDDs containing REEs will be available for recycling. Purified REEs derived from waste magnets can be reused in other energy and defense applications.

[0012] Recycling of REEs from waste streams, e.g., scrap magnets, can reduce the environmental risks associated with mining. The production of heavy rare earth oxides from clays containing adsorbed REE ions can lead to groundwater contamination and severe vegetation and topsoil removal. The discharge of large amounts of mine tailings and wastewater into the environment can lead to permanent ecosystem damage, loss of biodiversity, and human health problems. Recovery of REEs from waste magnets is potentially profitable according to Sherwood analysis, which predicts the magnitude of price or production costs as a function of the concentration of the target product in the feedstock. Generally, the higher the concentration, the lower the production costs and, therefore, the lower the market price. Sherwood correlation lines were obtained by fitting the prices of the 10 metals and their mass fractions in the ores. Estimates of the feedstock costs and the refining costs of Nd, Pr, and Dy from waste magnets are summarized in Table 1. The Sherwood analysis indicates that waste magnets are a promising feedstock for producing Dy, Nd, and Pr, with a potential profit of approximately $5 per kg of waste magnet.

[0013] [Table 2]

[0014] Various methods have been developed to recover crude REE mixtures from waste magnets. The most commonly used method involves oxidizing metals to obtain metal oxides, followed by selective acid leaching. The waste magnets were crushed into fine particles and oxidized at high temperatures. The metal oxides were then dissolved in acid. Other metal oxides (Fe, Ni, Co, B) were also removed by selective leaching or precipitation, membrane-assisted solvent extraction, or ionic liquid extraction. REEs in waste magnets can also be converted to soluble chloride salts by various methods, including roasting with ammonium chloride under an inert atmosphere, chlorination using chlorine gas, or selective extraction using magnesium chloride and potassium chloride molten salts. Other methods for converting REEs in waste magnets to hydroxides have also been tested, including hydrothermal treatment and vacuum induction melting followed by hydrolysis. Most prior art recovery methods extract more than 90% of the REEs in waste magnets as a mixture of REE chlorides or hydroxides, with little contamination of other elements in the waste magnets.

[0015] Separation of crude REE mixtures derived from waste magnets into pure individual REEs is useful for the production of metal alloys or other applications. Solvent extraction using organophosphorus extractants in ionic liquids has previously been tested. However, only two REEs (Rd and Dy) were recovered in high purity, and liquid-liquid extraction methods are inefficient for REE purification because the interfacial area per unit processing volume for mass transfer is two to three orders of magnitude smaller than that of adsorption or chromatography. Functionalized silica adsorbents have been tested to separate Dy from other REEs derived from waste magnets, but these adsorbents were costly, had limited stability, and resulted in low Dy purity (approximately 70%). Because conventional low-cost adsorbents or ion exchangers are insufficiently selective for REE purification, chelating agents (or ligands) can be used in the mobile phase to substantially increase selectivity. The feasibility of ligand-assisted displacement chromatography for REE purification was first reported in the 1950s. However, because process simulation or systematic design methods were not available, LAD separation methods in the literature were designed empirically. Some experiments for the separation of three REEs took several months. Therefore, it is not feasible to use empirical methods to develop efficient LADs for large-scale separation of REEs.

[0016] It is known that constant-pattern displacement trains can form in non-ideal LAD systems (i.e., systems with diffusion or dispersive effects) if the column length is sufficiently long. A general correlation allows the design of LAD systems with a minimum column length to reach a constant-pattern state, where high-purity bands with sharp boundaries form as a result of displacement effects. Operation under a constant-pattern state can help achieve high-purity products with high yields and high sorbent productivity. A constant-pattern design method was developed for non-ideal systems based on the general correlation and an equation for the yield of the target component. The yield is a function of several important dimensionless groups that control the constant-pattern mass-transfer zone length. The design method was tested and validated using simulations and experiments for different target yields, ligand concentrations, and feed compositions. The productivity achieved using this design method on a single column was more than 800-fold higher than prior art results for a three-component separation with identical purity and similar yields.

[0017] A general correlation was derived from rate model simulations based on a constant separation factor (CSF) isotherm model applicable to ligand-containing REE feed mixtures. When the REE feed mixture does not contain ligands, the feed REEs do not separate in the loading zone because most adsorbents have negligible selectivity for different REEs. REE separation only occurs after loading by introducing the ligands. In such cases, the loading zone is shorter, and faster loading rates can be used to reduce the loading time. Here, we present an improved isotherm model and simulation to more closely simulate the separation process for ligand-free REE crudes. A modified general correlation for the conditions required to reach a constant pattern state is constructed.

[0018] Prior art techniques have focused on the separation of equimolar ternary REE mixtures. However, the REE concentrations in feedstocks derived from waste magnets differ by an order of magnitude (Table 1). Obtaining high-purity Dy (the minor component) in high yields in a single column requires a very narrow mass transfer zone between two adjacent elution bands and a low linear velocity for fixed selectivity and mass transfer coefficients. The low linear velocity results in long total elution times and consequently low productivity. To address these shortcomings, the novel technology introduces designs involving more than one column or more than one zone. An efficient partitioning strategy is developed to separate complex feed mixtures to obtain high-purity REEs in high yield and high productivity.

[0019] The new technologies shown in Figures 1-28 have the potential to convert current REE separation and purification processes into environmentally friendly and clean processes, provide the driving force for producing high-purity REEs from waste magnets, and help achieve a circular REE economy.

[0020] Constant pattern isotachic trains and general constant pattern correlations. Ligand-assisted displacement chromatography uses ligands (chelating agents) in the mobile phase to separate a mixture of different REEs into distinct REE-specific components. The separation mechanism is detailed below. The formation of an isokinetic (constant velocity) train in a long column is a distinct feature of displacement chromatography. In an ideal system, all components form rectangular solute bands and move at the same velocity. In a non-ideal system, the boundaries (or concentration waves) between adjacent solute bands broaden due to diffusion or dispersion effects. As the bands move at different velocities, the overlap area decreases as the bands begin to separate. Eventually, the wave broadening due to diffusion or dispersion effects is offset by the wave sharpening due to displacement effects, and the concentration waves reach a "constant pattern," where each mass transfer zone length has a fixed value, L MTZ,CP reaches. This novel technique presents a new general correlation (Eq. (1)) constructed using a modulated Langmuir isotherm that can more closely simulate the LAD process for ligand-free REE crudes. The detailed derivation of the mass transfer zone length and the construction of the new general correlation, Eq. (1), are provided in Supplementary Material B.

[0021]

number

number

number

[0022] Constant pattern constant velocity trains and general constant pattern correlations in LBD. In LBD systems, where the ligand is immobilized on the stationary phase, the REE mixture can also be separated into distinct displacement bands, forming a patterned isokinetic train. The ligand on the stationary phase provides selectivity for the separation. The column is presaturated with a component (Na+) that has a lower affinity than all the components in the feed, and the displacer (H+) has a higher affinity for the ligand than all the components in the feed. The feed component with the lowest affinity for the immobilized ligand elutes first. In LBD, sorbent regeneration and ligand regeneration are combined into one single step, whereas LAD requires a separate step to regenerate the ligand in the mobile phase. Therefore, LBD is expected to have lower processing costs than LAD.

[0023] Similar to the LAD, a general correlation for predicting the formation of constant-pattern isovelocity trains in the LBD, Equation (2), is also constructed.

[0024]

number

[0025] When the objective is to recover only one REE with a target yield using a single column, the design results for an REE mixture with a Nd mole fraction of 0.83, a Pr mole fraction of 0.12, and a Dy mole fraction of 0.05 are shown. The curves were generated based on a constant pattern design with a fixed breakthrough cutoff θ of 0.05. In practice, the breakthrough curves can be monitored using an online detector. A design with a fixed breakthrough cutoff is easy to implement for product recovery. However, to meet the mass balance requirements for the components under constant pattern conditions, the single-column design allows for the specification of only two of the three variables (purity, yield, and breakthrough cutoff θ). The yield, productivity, and product purity for recovering a single REE (Dy, Pr, or Nd) from an REE mixture with a fixed θ of 0.05 are shown. When the target yield is greater than 63.5%, the purity of the target component exceeds 99%. The curves are "approximate" trade-off curves because the purity of the target product is not fixed but varies slightly from 99% to over 99.9%.

[0026] When designing a single column to recover the minor component Dy with high purity (99.9%) and high yield (95%), the velocity must be very slow to sharpen the concentration wave and minimize the mass transfer zone length (the overlap region of two adjacent bands), resulting in a sorbent productivity of 0.04 kg / m of Dy. 3 In contrast, when producing major component Nd with the same purity and yield from the same feedstock, the productivity is 64.2 kg / m of Nd.3 / day, which is 1,600 times higher than the productivity of Dy. If the target yield of Nd is reduced from 95% to about 77%, the productivity of Nd is 120 kg / m of Nd, even though the purity of Nd is 99.5%. 3 / day. These results demonstrate that sorbent productivity is highly dependent on the target component for the production of high purity products (>99%) from complex feedstocks. Generally, when using a single column to recover a single component from a complex mixture, the purity of the product is controlled by the breakthrough cutoff θ and the yield of the target component. The productivity of the sorbent is determined by the selectivity-weighted composition coefficient γ, defined by Equation (2): i is controlled by

[0027]

number

number

[0028]

number

number

[0029] For complex mixtures, if the selectivity between each pair of adjacent components were the same, the constant pattern mass transfer zone length would be the same for all solute bands. i As x increases, the substitution bands become wider. The component with the highest mole fraction has the highest yield because it has the smallest overlap area relative to the total substitution band width and the smallest yield loss due to mixed bands relative to the total. i value or maximum γ i The components with the highest values ​​have the highest yields and highest productivity. Therefore, the selectivity weighted composition coefficient γ i describes the effect of composition and selectivity. i Components with a γ value can be separated from a mixture with maximum productivity using a single column. i The values ​​are listed in Table 2. In this mixture, Dy has a minimum γ i value, and therefore it can be separated from the mixture with minimum productivity, whereas maximum γ i The Nd having the desired value is separated from the mixture with maximum productivity.

[0030] [Table 3]

[0031] If a single column is used to recover all three components from a mixture in high yield and purity, the velocity or flow rate must be at least γ iThe process is limited by the yield requirement for Dy, a valuable component. If the design aims to recover Dy at 95% yield, the productivities of Nd and Pr are also small due to the low rates. As a result, the total REE productivity is only 0.7 kg / m 3 / day. However, if the separation of the three REEs is carried out in two separate zones, high purity REEs can be recovered with high yield and high productivity. A systematic separation strategy is developed. The maximum γ i The valuable component, Nd, is first recovered in zone I with high purity and high productivity. The two mixed bands in zone I, Dy / Nd and Nd / Pr, are then sent to zone II for further separation. The mixed band material in zone II is then recycled to the inlet of zone II to achieve high yields (99%) for all three components. In this way, the productivity and yield of each component are no longer limited by the trade-off curves of a single column. The total productivity of the two-zone design with high purity (>99.5%) and high yield (>99%) for all three REEs is 100-fold higher than that of a single-column design with similar product purity and 95% Dy yield. The two-zone design is described in more detail below.

[0032] Constant pattern design of two-zone LAD for separation of ternary REE mixtures of Dy, Nd and Pr Here, we present a previously reported method for designing a fixed pattern for a single column. 38 γ in the new general correlation, Eq. (1) and Eq. (2) i The multi-zone constant pattern design method was modified by incorporating a new zoning strategy based on the sorbent's specific values. The multi-zone constant pattern design method is based on advanced wave theory, a general zoning strategy, and intrinsic (or scale-independent) parameters. This method is compatible with many production scales and can handle complex feed mixtures containing multiple components and widely varying concentrations. For desired product purity and yield, when the intrinsic parameters, feed composition, and volume are known, this method can generate the zone configuration, column size, and optimal operating speed for each zone to achieve maximum sorbent productivity.

[0033] A schematic of this design is shown in the drawing for the separation of a ternary mixture of Dy (5%), Nd (83%) and Pr (12%). Zone I has the highest γ i In Zone I, Step 1, the column is designed to recover most of the Nd with Cu. 2+ In step 2, the feed mixture is pre-equilibrated with Cu 2+ In steps 3-7, the column is loaded with a ligand solution (EDTA-Na) to separate the feed mixture into three fractions: a Dy / Nd mixed band, a pure Nd band (the target product in Zone I), and a Nd / Pr mixed band. After all the REEs have been eluted from the column, the column is loaded with Na. + Then, it is again 2+ The Dy / Nd mixed band from step 5 and the Nd / Pr mixed band from step 7 are sent to columns II-A and II-B in zone II, respectively. The mixed bands from the zone II columns are collected and recycled directly back to columns II-A and II-B to further increase the yield. For simplicity, the recycle stream and column wash steps in zone II are not shown.

[0034] Kinetic model simulations were constructed for zones I and II. Batch LAD experiments were first conducted to test the kinetic model based on the modulated multicomponent Langmuir isotherm and intrinsic parameters. The validated isotherm and parameters were used to design a two-zone LAD system for separating a ternary mixture with REE compositions similar to those in waste magnets. Prior to LAD refining, the metals in the waste magnets are conveniently first converted into soluble salts. Common conversion methods include (1) hydrometallurgical, (2) pyrometallurgical, (3) hydrothermal, and (4) electrochemical methods.

[0035] In the hydrometallurgical process, waste magnets are dissolved in concentrated (up to 8M) nitric or hydrochloric acid. The REEs and other metals in the waste magnets can be completely dissolved and converted into soluble salts. Impurities, such as Fe, can be removed by adjusting the pH of the solution. Excess acid remaining in the dissolution solution must be neutralized before loading onto a chromatography column for purification. Scrap magnets can also be processed using pyrometallurgical methods. The magnets are demagnetized at 300-400°C, crushed, and ground into small particles. Alternatively, scrap magnets can be converted into small particles (100 μM) by hydrogen decrepitation. The magnet particles are then roasted at high temperatures (>800°C) to form metal oxides. Dilute acid (<2M) is used to selectively dissolve the oxides of REEs, Co, and other metals, producing a solution of metal ions for further purification.

[0036] A hydrothermal method for converting scrap magnets into soluble salts is described below. Scrap magnets (10 g) may be placed in a hydrothermal reactor (50 mL) containing water (30 g) and a small amount of salt (0.01 g or 0.03 wt. % NaCl) at 250°C and 550 psi for a period of time, e.g., about 18 hours. The Nd-rich phase reacts with water to form Nd(OH). Hydrogen is absorbed by the NdFeB phase, causing volume expansion and disintegration of the phase into fine powder. The metal alloy is oxidized to FeO and REE(OH). The metal coating is disintegrated and separated from the powder by sieving. The majority of the FeO powder can be separated using magnetic separation. The REE hydroxides typically contain less than 5% impurities and can be dissolved in dilute acid (<0.5 M) for further purification. Electrochemical methods can also be used to dissolve waste magnets. The magnets can be used as sacrificial anodes to oxidize the REEs and other metals to form metal hydroxides. Dilute acid (0.2 M) is used to selectively dissolve the REEs into an aqueous solution, while the Fe(OH)3 remains in the solid residue. The solution containing the REEs and other metal ions is typically used as a feedstock for LAD separation after filtration.

[0037] Example 1 Two packed columns were connected, resulting in a total column length of 127 cm. A ternary mixture simulating REE crude from waste magnets was prepared and separated using LAD. The detailed experimental conditions and simulation parameters are summarized in Table 9. In Zone I, the feed concentrations were 0.05 N Dy, 0.83 N Nd, and 0.12 N Pr. The column effluent was monitored with an Agilent PDA detector. The effluent was also collected in multiple fractions using an Agilent 440-LC fraction collector. The Dy concentration was analyzed by ICP-OES. Two mixed bands were produced in Zone I, one containing Cu / Dy / Nd and one containing Nd / Pr. More than 10 runs of Zone I were required to generate enough mixed-band material to test the separation in Zone II. The elution history of the mixed-band separation in Zone II was investigated using a synthetic mixture. The concentrations of REEs in the mixed band were calculated from the elution profile from Zone I. Note that the collected REE mixed band from the LAD test was a mixture of EDTA-REE complexes, rather than an aqueous solution of free REE ions.

[0038] To prepare the Cu / Dy / Nd mixed band, 3.143 g of Dy(NO3)3·5H2O and 3.872 g of Nd(NO3)3·6H2O were first dissolved in DI water. The mixture was loaded onto the copper-loaded column. REEs were stripped from the column using 0.09 M EDTA at pH 9. The mixed band was collected before the decrease in Cu concentration was monitored. The loaded REEs were collected in the mixed band. When this mixed band was diluted to 500 ml, the final concentrations of Dy and Nd were 0.043 N and 0.053 N. The experimental conditions and simulation parameters for the separation of Dy and Nd are summarized in Supplementary Material E, Table 10. The effluent was monitored using a PDA detector and collected in multiple fractions for Dy concentration measurement using ICP analysis. To prepare the Nd / Pr mixed band, 2.513 g of Nd(NO3)3·6H2O and 2.146 g of Pr(NO3)3·6H2O were dissolved in a 0.09 M EDTA solution (pH = 9). To prevent EDTA precipitation at low pH, approximately 1.04 g of NaOH was added to the solution. The volume of the final solution was 200 ml, and the concentrations of Nd and Pr were 0.086 N and 0.074 N, respectively. The conditions for the separation of Nd and Pr are summarized in Table 11.

[0039] Validation of velocity model simulation The results of the kinetic model simulation were compared with literature data using erbium (Er) as the presaturant and EDTA as the ligand. The simulation parameters are summarized in Table 12. The sorbent was a cation exchange resin AG 50X12 with sulfonic acid functional groups. The simulation results were in good agreement with the experimental data, indicating that the kinetic model simulation and model parameters were accurate. Er was used as the presaturant instead of Cu. The selectivity between REEs in the feed and Er was 1.1 during loading and increased to 1.95 after the ligand was introduced into the column.

[0040] Kinetic model simulations and generic maps were compared with the separation data. A binary mixture (0.3N Nd, 0.3N Pr) was prepared and fed to a 38 cm column. As previously measured, the Nd / Pr selectivity was 1.8. The mass transfer coefficient was estimated in our previous work. By varying the loading capacity, the Nd / Pr selectivity was increased at different loading rates (L f ) and the dimensionless column length (Φ) were obtained. The X value corresponding to the Φ value was calculated from a general correlation.

number

number

[0041] Zone I Design and Testing Using the improved general correlation, a three-component separation system was designed using a synthetic mixture with a composition similar to that of REE crude from waste magnets (0.05 N Dy, 0.83 N Nd, 0.12 N Pr). The actual effective selectivity between Dy and Nd should be extremely high due to the large difference in ligand affinity. The Dy / Nd selectivity in this design was set to 5, which is sufficient to generate sharp waves in this design and simulation.

[0042] The design of Zone I was aimed at maximizing Nd productivity in a single column. The yield was predicted to be 77.5% for the highest Nd productivity. Detailed experimental conditions are listed in Table 9. The pH of the ligand solution was maintained at 9. The experimental elution profile was compared with the simulation. The design target yield of Nd in Zone I and the experimental Nd yield and productivity for each test are summarized in Table 3. The target yields agreed with the experimental yields within 1%. Nd productivity exceeded 100 kg / m3 / day in Zone I.

[0043] [Table 4]

[0044] Zone II Design and Testing To increase the yield of Nd and collect high-purity Dy and Pr, the mixed bands of Dy / Nd and Nd / Pr (monochromatic region of the elution profile) in Zone I could be collected and further separated in Zone II. Simulants with similar concentrations were run to demonstrate Zone II separation. In zone I, an aqueous solution containing REE ions was directly fed into the column. Then, the mixed band collected from zone I was fed into zone II. In zone I, the REE ions were adsorbed together without separation, forming a uniform mixed band near the column inlet. This was because the selectivity was negligible. After the EDTA ligand solution was introduced into the column, the REE ions began to separate. In the theoretical model and simulation, the change in selectivity was revealed by the modulated Langmuir isotherm, as explained below.

[0045] The REE ions in the feed in zone II were bound to EDTA, and apparent displacement occurred during the loading of the feed. Because EDTA has different selectivities for different REEs, the REEs began to separate during loading. Because EDTA bound the REEs during their migration, the REE loading region was wider in zone II than in zone I. It was important to restrict the REEs to a short portion of the column length during loading, leaving sufficient column length for further separation after loading. For this reason, a safety factor of 100% additional column length was incorporated into the zone II design. Although such a large safety factor was used to demonstrate separation in zone II, in practice, the actual safety factor may have been smaller. The same design algorithm was used for both columns used in zone II. The column length in the design program was 89 cm, 50% of the actual column length in the experiment.

[0046] The elution profile is shown in the figure, and the experimental yields and productivity are summarized in Table 3. Detailed experimental conditions and simulation parameters are provided in the experimental section below. In column A of zone II, the Nd yield target in the design was 73.4% to maximize productivity, and the experimental Nd yield was 89%. In practice, Nd can be easily separated from Na by precipitating it as oxalate or carbonate, so in the experiment, the mixed band between the Nd and Na bands was defined as "pure" Nd. In column B of zone II, the Nd yield target was 74.6%, and the experimental yield was 74%. Recycling the mixed band from the two columns in zone II back to the zone II feed can further increase the yield of all components to over 99%.

[0047] Theoretical prediction of REE yields in two-zone LAD Established theoretical and kinetic model simulations for both zones I and II are in good agreement with the presented results. Experimental yields closely matched or exceeded the design target yields, with all components having purity greater than 99%. A key condition for achieving 99% yield for each component is the recirculation and refeed of the mixed band in zone II. Because experimental demonstration of mixed band recirculation in zone II is redundant, a VERSE simulation of the LAD process was performed instead. The column length was 100 cm in the design and 120 cm in the simulation, with a 20% safety factor. The Dy / Nd mixed band from a single run in zone I was used as the feed for column A in zone II, and the Nd / Pr mixed band from zone I was used as the feed for column B in zone II.

[0048] In column A of zone II, two mixed bands, Cu / Dy and Dy / Nd, were collected and mixed with the original feed to column A. The new mixture was used as the feed to column A of zone II in run 2. The loading capacity and flow rate remained the same. The feed to column A of zone II in run 3 contained the collected Cu / Dy and Dy / Nd bands from run 2 mixed with the original feed. In run 3, the column output, yield, and purity of the components remained the same as in run 2. The results showed that the system reached a cyclic steady state after run 3 (Table 4). A similar process was performed on column B of zone II for separation of the Nd / Pr mixed band. The system also reached a cyclic steady state after run 3. By continuously recycling the mixed band in the two-zone LAD, almost no REEs were lost in the mixture. The overall yield for all components was greater than 99%.

[0049] [Table 5]

[0050] The overall yields and productivities for the three components are summarized in Table 5 and compared to those of a single-column design with a fixed breakthrough cut. To achieve 99% overall REE yield and greater than 99.5% purity for each component using similar column lengths (approximately 1.2 m), the productivity of the single-column process was more than two orders of magnitude less than the productivity of the two-zone LAD process. The average sorbent productivity of the two-zone design was 0.7 kg REE / m 3 compared to the average sorbent productivity of the single-column design. 3 This is approximately 159 times the normal daily rate.

[0051] [Table 6]

[0052] Comparison of LAD with conventional liquid-liquid extraction The most widely used REE separation technique in the industry is multistage liquid-liquid extraction. In this method, an organophosphorus extractant dissolved in kerosene is used to extract the REEs from the aqueous phase. Concentrated HCl solutions (up to 6 M) are required to strip the REEs from the extractant, and concentrated aqueous ammonia is used as a saponifying agent. Most of the acid and base used cannot be recovered, resulting in large volumes of acidic wastewater containing high concentrations of ammonia salts. Furthermore, kerosene is highly flammable, and organophosphorus extractants are toxic.

[0053] In contrast, LAD uses only environmentally friendly chemicals, resulting in much improved process safety and a much lower environmental impact. The LAD process generates sodium salt by-products, which can be electrochemically converted to bases and acids. The EDTA-Cu complexes eluting from the column can be easily recovered as EDTA and copper salts. The recovery yields of EDTA and Cu are extremely high (>95%). High-purity REEs can be precipitated from the EDTA-REE complexes using oxalate, and the ligands can be recovered in high yields and reused. Therefore, little waste is generated. Table 6 summarizes the advantages of LAD over extraction processes.

[0054] [Table 7]

[0055] Rare earth elements are essential for high-tech products. Traditional mining and separation processes to produce high-purity REEs can have significant environmental impacts. Recycling REEs from waste streams such as scrap magnets is technically feasible, potentially profitable, and reduces the need for mining processes. An environmentally responsible separation method, two-zone ligand-assisted displacement chromatography, has been developed to produce high-purity REEs in high yield and productivity.

[0056] An improved model based on the modulated Langmuir isotherm has been developed and used to accurately simulate ligand-assisted displacement chromatography for ligand-free REE feedstocks. A general correlation for such feedstocks was developed to predict the minimum column length for the formation of a constant-pattern displacement train. A design method based on this correlation was developed. The kinetic model and simulations were tested and validated against literature data for the separation of seven REEs. Furthermore, the general correlation was tested and validated using new experimental data for the separation of a binary mixture of Nd and Pr.

[0057] Selectivity-weighted composition coefficient γ for two-zone LAD for separation of complex mixtures of Dy, Nd, and Pr i A precise zone division strategy based on the above method was developed and tested. In the first zone, high purity Nd was obtained with a yield of 78% and a production rate of 100 kg / m 3 Productivities higher than 111 kg / (m) of REEs / day were obtained. The second zone was designed to separate the mixed band from the first zone to produce all three REEs in high purity and yield. The total Nd yield was 95% without recycle of the mixed band in zone II. Recycle of the Nd / Pr and Dy / Nd mixed band in zone II can further increase the total yield for all three REEs to over 99%. For similar product purity and yields, the total productivity of this two-zone LAD system is 111 kg / (m) of REEs. 3 sorbent) / day, which was more than 100 times higher than that of the single-column LAD.

[0058] Two-zone LAD requires fewer columns, and its productivity is approximately 100 times higher than the average volumetric productivity of conventional liquid-liquid extraction, which requires thousands of mixer-settler units. The LAD separation process requires only environmentally friendly chemicals, most of which can be recycled and reused, and generates little waste. The multi-zone LAD method has the potential to transform the REE purification process into a cleaner, "greener" process. The results of this research may also help transform the current linear REE economy (from ore to pure REE to products to landfill) into a circular and sustainable REE economy.

[0059] Example 2 Two-zone LBD containing three columns for the separation and recovery of three high-purity REEs, Nd, Pr and Dy, from the REE mixture of Example 4 (general strategy 1) The mixture derived from waste magnets contains three REEs, Pr, Nd, and Dy. The detailed composition is listed in Table B below. To recover all three components, Pr, Nd, and Dy, a two-zone ligand-based displacement (LBD) separation involving three columns is designed. The REE-selective ligand, EDTA, is immobilized on the sorbent to provide sufficient selectivity for the separation. The monovalent metal ion Na, which has a much lower affinity than EDTA, is also used. + is selected as the presaturant. + Since Pr has the highest EDTA affinity, an acid such as HCl is used as the displacing agent. Of the three REEs to be separated, Pr has the lowest EDTA affinity and therefore the lowest sorbent selectivity, and elutes first, followed by Nd and then Dy.

[0060] [Table 8]

[0061] General resolution strategy 1 applies to this case. In this ternary mixture, Nd is γ iThe Nd / Dy concentration is highest and recovered first in Zone I. The operating conditions in Zone I are designed to achieve maximum productivity of Nd. Under these conditions, neither Pr nor Dy reach a plateau concentration. Two mixed bands, Pr / Nd and Nd / Dy, are produced from Zone I and further separated in Column A and Column B in Zone II. The binary mixed bands produced from the effluents of the two columns in Zone II are recycled to the inlets of each column, IIA and IIB, respectively, increasing the yield of all three components to over 99%. The simulated elution profiles are shown in Figures 21A-21E. The total productivity in the three-zone design was 281 kg / m 3 / day is 680 times the total productivity in a single column design.

[0062] [Table 9]

[0063] [Table 10]

[0064] [Table 11]

[0065] [Table 12]

[0066] Example 3 One-zone LBD containing one column for the separation and recovery of two products, Dy and Nd / Pr (as a mixture), from the previously disclosed REE mixture (general strategies 2 and 3) The feed mixture and sorbent in this example are the same as those in Example 5. However, the target products in this example are (1) a mixture of Pr and Nd and (2) Dy, rather than three pure products. The Nd / Pr ratio in the waste magnet is similar to that of unused magnets. The recovered Nd / Pr can be directly used to manufacture new magnets. Recovering Nd and Pr as a single product can simplify the purification process and reduce processing costs.

[0067] In this design method, Pr and Nd are grouped as one component Pr / Nd, and Dy is the other component. The mixture is treated as a binary mixture. The composition and selectivity-weighted composition coefficients for this binary mixture are shown in Table C. The γ i When calculating the values, the selectivity between the presaturant and Pr and the selectivity between Nd and Dy should be used.

[0068] [Table 13]

[0069] To recover two products from this feedstock, only a single-zone LBD containing one column is required. The grouped components Pr / Nd have a higher γ i Zone I is designed to achieve maximum productivity for producing Pr / Nd. The purity of the collected Dy meets the requirement of 99.5%, so Dy can be recovered as a pure product in Zone I. The binary mixed band generated from Zone I is recycled to the inlet of the column to prevent material loss, and the yield of both components is greater than 99%. The simulated elution profiles are shown in Figures 23A-B.

[0070] The sorbent productivity of this system (including mixed band recycle) estimated from the chromatogram shown in Figure 23B is 1,800 kg / m REE. 3 / day. Productivity is very high because of the very high selectivity between Dy and Nd (α=5) (Table 5 above). More importantly, this separation does not require the separation of the low selectivity pair Nd and Pr (α=1.8), as in Example 4, resulting in high sorbent productivity. The productivity of a single column design separating Nd, Pr, and Dy with high purity (>99.5%) and high yield (>99%) of all three REEs is only 0.4 kg / m REE. 3 Recall that the mean daily intake of 1000 mg / day is 100 mg / day (Table 8, Example 5). In general, the selectivity weighted composition coefficient γ for the target product component i i The higher the purity requirement, the lower the sorbent productivity.

[0071] Example 4 Combination of LAD (Zones I and II) and LBD (Zone III) for the separation of four REEs and Co from a crude mixture derived from waste magnets (General Splitting Strategy 1 for LAD and General Strategy 2 for LBD) Example 7 demonstrates a method for recovering three REEs (Dy, Nd, and Pr) and Co from a crude mixture derived from waste magnets. The waste magnets are first converted into soluble salts in aqueous solution. The major component Fe in the soluble salts of the waste magnets is precipitated by adjusting the pH of the solution and removed by filtration. The REE and Co ions remaining in the solution can be separated and recovered. A three-zone design is developed to recover Co and the three REEs (Figure 24A).

[0072] A cation exchange column is used as the loading column, Zone IA. Trivalent ions have a higher affinity than divalent metal ions for cation exchangers. When a mixture of REEs and Co is fed to this column, the trivalent REEs remain in the column, while the other divalent metal ions displace the trivalent ions and are eluted from the column earlier. A simulated elution profile for Zone IA is shown in the figure. The REE-loaded column, Zone IA, is then connected to the Cu-loaded column to proceed with the REE separation process using a two-zone LAD. The LAD separation design strategy is the same as in Example 4. Three pure REEs are produced. The divalent metal ions eluting early from Zone IA are collected and sent to Zone III, where Co is recovered by LBD. Zone III uses an iminodiacetic acid (IDA) column, which has high selectivity among divalent metal ions. The affinity order of the IDA column is Cu>Ni>Co>B. A displacement separation is designed to recover Co as a product. Na is used as a presaturant. + Using H as a substitute + The simulated elution profile is shown in Figure 24C.

[0073] Theoretical details A. Separation mechanism in ligand-assisted displacement (LAD) chromatography A schematic diagram of the separation mechanism of ligand-assisted displacement chromatography for a binary mixture is shown. The cation exchange resin is first pre-saturated with a pre-saturating agent (P), in this case copper ions (Cu 2+ ) pre-equilibrate with the REE ions (Nd 3+ and Pr 3+ ) is fed to the column, the trivalent REE ions are converted to divalent Cu. 2+ Since the sorbent selectivity for REE ions is very small (<1.1), no separation occurs during loading, and a homogeneous REE mixture band forms near the column inlet.

[0074] When a solution of EDTA ligand is applied to the column, EDTA competes with the sorbent to form non-adsorbed ligand-REE (L-REE) complexes. Each complex travels down the column with the mobile phase, resulting in the formation of Na. + EDTA effectively replaces REEs with Pr 3+ than Nd 3+ Since the L-Nd complex has high selectivity for Cu, the L-Nd complex migrates ahead of the L-Pr complex, resulting in separation of the Nd band from the Pr band. 2+ has the highest affinity for EDTA, so when L-Nd reaches the boundary between the Nd and Cu bands, the ligand 3+ By liberating Cu 2+ Then, the released Nd 3+ The ions are re-adsorbed onto the column. As more ligand is added to the column, the L-Pr complex reaches the boundary between the Pr and Nd bands. 3+ is released from the ligand and re-adsorbed onto the sorbent. As the adsorbed Nd ions desorb and form complexes with the ligand, the Nd band is replaced by a Pr band. If the column is long enough, this displacement process continues until two consecutive bands of Nd and Pr are formed, moving at the same velocity through the column. The separation of these two bands is driven by EDTA as an apparent displacement agent. In the ideal case, where there are no mass transfer effects, this set of bands is called a "constant velocity train." In the non-ideal case, the overlap area between adjacent bands shrinks as the bands continue to separate. Eventually, each mass transfer zone reaches a certain length, and the system reaches a "constant pattern" state.

[0075] B. Constant speed train, constant pattern constant speed train and general map Sorbent selectivity

number

number

number

number

number

[0076] For an ideal system (no broadening due to mass transfer), the minimum column length to form a constant velocity train in the LAD was first derived using h-transform theory.

[0077] For non-ideal systems, the constant pattern mass transfer zone length (L MTZ,CP )teeth,

number

number

[0078] Load factor (L f )teeth,

number

[0079] Dimensionless total mass transfer coefficient

number

number

number

number

[0080] N D is the ratio of the intraparticle diffusion rate to the convection rate

number

[0081] Constant pattern mass transfer zone length L MTZ,CP teeth,

number

number

[0082] Nonlinear distribution coefficient K d teeth,

number

number

[0083] Minimum column length L required to form a constant-velocity train with a constant pattern iso-nid was correlated with several important dimensionless groups. Using velocity model simulations, L iso-nid The dimensionless column length Φ is the ratio of the column length in a non-ideal system to the minimum column length required to form a constant velocity train in an ideal system. For a fixed column length, Φ is the ratio of the ideal loading rate to the non-ideal loading rate.

[0084] A combination of key dimensionless groups is used to reduce the multidimensional design parameter space to a two-dimensional Φ as a function of X, where X is the product of key dimensionless groups that control the mass transfer zone length.

number

number

[0085] A constant separation factor isotherm is used to simulate the LAD process:

number

number

[0086] The multicomponent Langmuir isotherm, Eq. (A13), is equivalent to the constant separation factor isotherm,

number

[0087] The sorption capacity of each component is the same and b j C p,j When the term is much larger than 1, equation (A13) becomes

number

[0088] The simplified isotherm corresponds to the constant separation coefficient isotherm, Eq. (A12), and the selectivity is

number

[0089] To simulate the process of the separation factor change after the introduction of the ligand, a modulated Langmuir isotherm is used:

number

number

[0090] b j C p,j If the term is much larger than 1, the isotherm becomes

number

number

number

number

[0091] If there is no selectivity between the two components during loading, a 0,i =a 0,j Presaturant Cu 2+ The effective selectivity between the first and earliest eluting REEs (those with the highest affinity for the ligand) was approximated as 5 in this simulation. Such a high selectivity is sufficient to predict a sharp wave between the two species. After the ligand is introduced into the column, separation begins. Since each REE has a different affinity for the ligand, as the ligand is applied to the column, the modifier concentration c m changes to 1, resulting in substantial selectivity.

[0092]

number

[0093] Since the sorption capacity is still the same, the ratio between the total Langmuir "a" and "b" values ​​should be the same.

number

number

number

number

number

[0094] Simulations based on the new isotherm were used to determine the minimum column length required to achieve a consistent pattern in nonideal binary systems. In each case, two REEs (REE1 and REE2) were separated using LAD. The sorbent had negligible selectivity for these elements. REE1 has a higher affinity for the ligand, resulting in a lower effective sorbent selectivity. REE1 elutes before REE2. Various hypothetical binary systems, including 2-meter-long columns, were simulated. By varying the flow rate and the effective selectivity between the two REEs, a range of X values ​​(8 to 41) was obtained.

[0095] α e The value of was varied from 1.5 to 10. The effective selectivity between REE1 and the presaturant was set to 5, and the effective selectivity between the ligand (EDTA-Na) and REE2 was also set to 5 to simulate a sharp displacement wave. The effective capacity of these systems, the ratio of a0 to b0, was 1.45 meq / ml (bed volume). The isotherm parameters for the hypothetical binary system are shown in Table 7, A5.

[0096] The minimum column length for each binary system at which a constant pattern mass transfer zone length was reached was identified from the column profiles when the column length was increased and the simulated mass transfer zone length remained fixed. The minimum column length was then divided by the ideal column length that would form a constant velocity train in the corresponding ideal system. The dimensionless column length Φ was plotted against various X values.

[0097] A best-fit line was found through the data points and plotted. This plot is a general map that predicts whether a non-ideal system is in a constant pattern region. In the non-linear regression, which fits an exponential equation to the curve, the constant term was fixed at 1. This is because as the X value approaches infinity, the system approaches an ideal system and the minimum column length approaches that of an ideal system. Φ as a function of X min The new correlation is shown below:

number

[0098] C. Yield and productivity in constant pattern design methods for multicomponent mixtures The dimensionless yield equation is

number

number

number

[0099] Sorbent productivity P for component i in LAD systems R,i(REE production per unit time per unit volume of sorbent) was derived previously.

number

number

[0100] Design results for yield and productivity as a function of linear velocity are shown for a ternary mixture of Dy, Nd, and Pr with mole fractions of 0.05, 0.83, and 0.12, respectively. The highest productivity, Nd 120 kg / m 3 (sorbent) / day was observed with a yield of 77.5%.

[0101] D. Relationship between yield, purity and breakthrough cut θ for a given pattern design In collecting the product from the column effluent, the breakthrough cutoff θ, which controls the width of the product band, is set to the minimum concentration c bot Maximum band density c d This cut controls the yield of the product component, as it determines the amount of this component in the collected product relative to the total amount in the feed. This cut also controls the amount of impurities from adjacent bands, thus controlling the purity of the product. The collected product is the band region between the two mass transfer zones. From the elution profile, the yield of component i is given by

number

[0102] To calculate the purity of the product, the amount of impurities from two adjacent bands is first calculated. The impurities from the adjacent bands are shown enlarged on the right. It is difficult to find an analytical function that can show the change in concentration over time, but the elution time t, which corresponds to the mass transfer zone length, can be used to calculate the purity of the product. MTZ,CP is related to θ by analytical solutions, Equation (A2) and Equation (A8). Combining equation (A2) and equation (A8), we get

number

number

[0103] The area within the curved region is

number

[0104] Combining equation (A30) and equation (A31), t(θ) becomes

number

[0105] In that case, the area is

number

number

number

[0106] In that case,

number

[0107] In the limit of θ0→0,

number

number

[0108] To link yield with purity and cut, first Q f and K. d Simplify the k terms by substituting:

number

[0109] Load factor Lf teeth,

number

[0110] The purity can be calculated by combining equations (29), (A40) and (A41) to obtain the yield and breakthrough cutoff θ d Associated with:

number

[0111] E. Table of System and Simulation Parameters [Table 14]

[0112] [Table 15]

[0113] [Table 16]

[0114] [Table 17]

[0115] [Table 18]

[0116] [Table 19]

[0117] [Table 20]

[0118] While the novel technology has been illustrated and described in detail in the drawings and foregoing specification, this should be considered exemplary and not restrictive in nature. It is understood that embodiments that meet the best mode and enabling requirements have been shown and described in the foregoing specification. Those skilled in the art will appreciate that an almost infinite number of insubstantial changes and modifications to the embodiments can be readily made, and that it would be impractical to attempt to describe every variation of such embodiments in this specification. It is therefore understood that all changes and modifications that come within the spirit of the novel technology are desired to be protected.

Claims

1. 1. A method for recovering pure rare earth elements (REE) Nd, Pr and Dy from magnets, comprising: a1) dissolving a magnet to obtain a solution containing Nd, Pr and Dy; a2) Selectivity weighted composition coefficients (γ i ) b) equilibrating a first column with a presaturant to obtain a first equilibrated column; c) introducing the solution into a first equilibration column; d) introducing a ligand solution into the first equilibration column; e) For each of Nd, Pr, and Dy, γ i establishing three migration bands of different solution compositions in the column based on the values ​​of σ, wherein the three bands include a Dy / Nd mixed band, a first pure Nd band, and a Nd / Pr mixed band; f) passing the Dy / Nd mixed band into a second column containing said presaturant; g) introducing a ligand solution into the second column; h) establishing a pure Dy band, a second Dy / Nd mixed band, and a second pure Nd band in a second column; i) passing the Nd / Pr mixed band through a third column containing said presaturant; j) introducing a ligand solution into the third column; k) establishing a third pure Nd band, a second Nd / Pr mixed band, and a pure Pr band in a third column; l) eluting each pure Nd band and recovering the Nd; m) eluting the pure Dy band and recovering the Dy; n) eluting the pure Pr band and recovering Pr; The method comprising:

2. Step a) a3) converting the magnet into a mixture of soluble metal salts; a4) dissolving a mixture of soluble metal salts to obtain a solution containing Nd, Pr and Dy; The method of claim 1 further comprising:

3. The ligand is EDTA and the presaturant is Cu 2+ The method of claim 1 in a solution.

4. 10. The method of claim 1, wherein the presaturant is selected from the group consisting of sodium, copper, erbium, and combinations thereof.

5. 10. The method of claim 1, further comprising the step of introducing a displacing agent into each column after k) and before l).

6. The method of claim 5 wherein the displacing agent is EDTA.

7. 2. The method of claim 1, wherein the respective ligand is EDTA.

8. The method of claim 1 , wherein the ligands are migrating through respective columns.

9. 10. The method of claim 1, further comprising the step of eluting the second Dy / Nd mixed band from the second column and reintroducing it into the second column to produce a second pure Dy band and a fourth pure Nd band.

10. 10. The method of claim 1, further comprising the step of eluting the second Nd / Pr mixed band from the third column and reintroducing it into the third column to produce a fifth pure Nd band and a second pure Pr band.

Images