Aerio-metallurgical extraction of rare earth elements from NdFeB

Oct 24, 2018 - Fundamental investigations indicate that extraction proceeds by corrosion of the magnet particle's surface layer. This work demonstrate...
0 downloads 0 Views 2MB Size
Subscriber access provided by The University of Texas at El Paso (UTEP)

Article

Aerio-metallurgical extraction of rare earth elements from NdFeB magnet utilizing supercritical fluids Jiakai Zhang, John Anawati, Yuxiang Yao, and Gisele Azimi ACS Sustainable Chem. Eng., Just Accepted Manuscript • DOI: 10.1021/ acssuschemeng.8b03992 • Publication Date (Web): 24 Oct 2018 Downloaded from http://pubs.acs.org on October 25, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Aerio-metallurgical extraction of rare earth elements from NdFeB magnet utilizing supercritical fluids AUTHOR NAMES

Jiakai Zhang1†, John Anawati1†, Yuxiang Yao1†, Gisele Azimi1,2*

AUTHOR ADDRESS

*: Corresponding author: [email protected] †: These authors contributed equally to this work 1

Laboratory for Strategic Materials, Department of Chemical Engineering and Applied

Chemistry, 200 College Street, Toronto, Ontario M5S 3E5, Canada. 2

Department of Materials Science and Engineering, University of Toronto, 184 College

Street, Toronto, Ontario M5S 3E4, Canada.

ABSTRACT

ACS Paragon Plus Environment

1

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 64

There is a global need for efficient and environmentally sustainable processes to close the life cycle loop of waste electrical and electronic equipment (WEEE) through recycling. Conventional WEEE recycling processes are based upon pyrometallurgy or hydrometallurgy. The former is energy intensive, generating greenhouse gas (GHG) emissions, while the latter relies on large volumes of acids and organic solvents, thus generating hazardous wastes. Here, a novel “aerio-metallurgical” process was developed to recycle critical rare earth elements, namely neodymium (Nd), praseodymium (Pr), and dysprosium (Dy), from postconsumer NdFeB magnets utilized in wind turbines. The new process utilizes supercritical CO2 as the solvent, which is safe, inert, and abundant, along with tributyl-phosphate-nitric acid (TBP-HNO3) chelating agent and 2 wt% methanol as co-solvent. 94% Nd, 91% Pr and 98% Dy extraction was achieved with only 62% iron (Fe) co-extraction and minimal waste generation. Fundamental investigations into the extraction mechanism demonstrated that metal ion charge has an important impact on the extraction efficiency. Fundamental investigations indicate that extraction proceeds by corrosion of the magnet particle’s surface layer. This work demonstrates that supercritical

ACS Paragon Plus Environment

2

Page 3 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

fluid extraction would find widespread applicability as a cleaner, more sustainable option to recycle value metals from end-of-life products to enable the circular economy.

KEYWORDS Supercritical fluid extraction, Waste electrical and electronic equipment recycling, Tributyl-phosphate-HNO3, NdFeB magnets.

Introduction With the growing awareness to protect the urban environment and motivation for compliance with environmental regulations around the world, the sustainability of traditional linear economy (make, use, dispose) is scrutinized. With the aim of building a more sustainable future and enabling the circular economy (make, use, recover), waste valorisation and recycling of end-of-life products has been rapidly expanding. With the increasing focus on reducing greenhouse gas (GHG) emissions, the use of green energy technologies, such as wind turbines, hybrid and electric vehicles, photovoltaic cells and fluorescent lighting has been widely promoted 1. The performance of these emerging green technologies relies upon the unique physiochemical properties of their critical

ACS Paragon Plus Environment

3

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 64

building block materials, such as rare earth elements (REEs) 2. Neodymium-iron-boron (NdFeB) magnets are a class of permanent magnets widely utilized in hybrid and electric vehicles and wind turbines. The use of these magnets has been rapidly increasing all over the world in line with increasing utilization of hybrid and electric vehicles – forecasted increase from 2.3 million units in 2016 to over 10.1 million units in 2026 3 – as well as the increasing market for wind turbines, as the Global Wind Energy Council (GWEC) has reported a 12.6% cumulative capacity growth rate reaching a total of 486.6 GW in 2016 and a forecasted cumulative installed capacity of over 800 GW by 20214. Managing end-of-life NdFeB magnets in an environmentally sustainable approach could not only reduce the amount of landfilled wastes, but also enable the recovery of valuable materials they contain, including neodymium (Nd), dysprosium (Dy), and praseodymium (Pr) that are in increasingly high demand and facing supply challenge. These REEs account for more than 22% of the weight of the magnet, which is more than 10 times of the minimum industrial grade of primary ores (1.5-2.0%) 5. The current recycling rate of REEs is low – for example, in Europe, the current REE recycling rate is 6-7%, although this recycling only contributes approximately to 1% of the total European

ACS Paragon Plus Environment

4

Page 5 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

REE demand 6. Researchers across disciplines around the world are investigating techniques to extract REEs from end-of-life NdFeB magnets 7–10. Recycling of NdFeB magnet can take one of three primary routes, i.e., the direct remagnetization and reuse of end-of-life magnets; the physical reprocessing of magnet waste for direct recycling into new magnets; and the metallurgical extraction of REEs from waste magnets. Different strategies are adopted for different waste sources and magnet conditions 11. Metallurgical recycling allows the magnet to be recycled into new magnets of a different grade or even a completely different REE-containing products; whereas, direct recycling is only a viable option if the desired composition of the new magnet is the same as the old one. Because wind turbine magnets have a service life of 20-30 years11, the required specifications of the end-of-life product and those of the new product could be different. Furthermore, the magnets may become corroded or otherwise damaged during their service life, rendering them unsuitable for direct recycling. At present, pyrometallurgy and hydrometallurgy are the approaches utilized for REE recovery. Pyrometallurgy relies upon calcination of the feed with a mixture of reducing and fluxing agents and has the drawback of high energy consumption and reaction

ACS Paragon Plus Environment

5

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 64

temperature, and generation of large volumes of GHG emissions. Moreover, in most cases REEs partition into slag phases during calcination, thus an additional hydrometallurgical step is required to recover these elements

12.

Hydrometallurgy is the

preferred alternative for waste valorisation because of its adaptability to resources scale and relatively low energy consumption. However, although promising, this technique has the drawback of being reagent-intensive, often requiring the consumption of strong acids and organic solvents, and production of secondary streams of potentially hazardous wastes

12.

Considering the drawbacks of pyrometallurgical and hydrometallurgy, a

process with low energy consumption and low waste generation that yields a high recovery rate of REEs is highly desirable for the valorization of end-of-life NdFeB magnets. Supercritical fluid extraction (SCFE) is an emerging green separation technology that has been used for the recovery of various metals in recent years. This process has been developed in late 1970s to extract or produce high value products, and then expanded to less specialized applications. Major commercial SCFE processes have been focused on energy industry, including direct liquefaction of coal

13

and enhanced oil recovery from

petroleum reservoirs 14, as well as chemical industry, including food and pharmaceuticals.

ACS Paragon Plus Environment

6

Page 7 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Examples include coffee decaffeination

15,

wood and vegetable oil production

16,

extraction of organics like alcohols from aqueous solutions17, regeneration of catalysts 18 and activated carbon that is used for adsorbing organic contaminants in industrial plants 19.

Taking advantage of high diffusivity and low viscosity (gas like property) and high solvability (liquid like property) of supercritical carbon dioxide (sc-CO2), a few bench scale studies succeeded in extracting REEs from synthetic feeds (mostly oxides), a summary of which is provided elsewhere 2. Sc-CO2 offers several advantages, such as safety, inertness, abundance, and low cost. It has moderate critical points (Tc = 31.1 °C, Pc = 7.37 MPa, ρc = 469 kg/m3)20 and is easy to recycle, thus it does not generate secondary waste. Moreover, the final extraction products from this process are easily recoverable 2. Sc-CO2 can be considered a switchable solvent, i.e., near the critical point, small changes in pressure or temperature result in large changes in the physical properties of CO2, particularly density, allowing many process-critical properties of the solvent, such as the solubility, to become time-separated, i.e., the properties can be adjusted, fine-tuned, or switched at different points in time and/or space in response to triggers, in this case

ACS Paragon Plus Environment

7

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 64

depressurization. Because of these switchable properties, supercritical fluids are suitable alternatives for organic solvents in a range of industrial and laboratory processes. Because sc-CO2 is a non-polar solvent, there is a significant polarity difference between the solute and solvent when extracting metal ions and organometallic compounds. In such cases, it is therefore necessary to utilize complexing (chelating) agents to satisfy the charge neutrality and improving solvent-solute interactions. A suitable chelating agent should have significant solubility in sc-CO2 in both their complexed and noncomplexed forms

21–23.

There are several chelating agents available for REE extraction, among

which, β-diketones, dithiocarbamates, organophosphorous agents, and macrocyclic compounds have been investigated

21.

Ligand chemistry and coordination geometry of

metal ions are major factors that determine the chelating agent type 21. In a recent study, Yao et al. 2 successfully utilized sc-CO2 along with tributyl-phosphatenitric acid (TBP-HNO3) chelating agent and methanol as co-solvent to recover 86% La, 86% Ce, 88% Pr, and 90% Nd from the anode material of an end-of-life NiMH battery from a Toyota Prius. Their study established the feasibility of utilizing SCFE process for extracting REEs from end-of-life NiMH batteries.

ACS Paragon Plus Environment

8

Page 9 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

In this study, a SCFE process was developed to recover REEs from end-of-life NdFeB magnets. This work was devoted to finding the optimum extraction process and elucidating the reaction mechanism with systematic experimental and theoretical approaches. We expect the findings of this study would help enable the sustainable urban mining of REEs from end-of-life NdFeB magnets using “aeriometallurgy” as a green alternative to conventional hydrometallurgical and pyrometallurgical routes, contributing to a closed-loop process for this class of products. Aeriometallurgy is a proposed terminology to differentiate supercritical fluid extraction from hydrometallurgy, pyrometallurgy, and solvometallurgy. The “aerio” prefix is from the Greek word for “air or gas”, and is chosen to invoke the idea of a gas-like solvent, which is a part of the air we breathe, rather than a specific description of the process.

Methods Chemicals and materials The following reagents were employed in this work: Tri-n-butyl phosphate (TBP, ≥ 98% – VWR), concentrated nitric acid (15.7 M, 70 wt% – VWR), concentrated hydrochloric acid (12.2 M, 37 wt% – VWR), concentrated sodium hydroxide (19.4 M, 50 wt% – VWR),

ACS Paragon Plus Environment

9

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 64

phenolphthalein indicator solution (1% in alcohol – VWR), neodymium oxide (Nd2O3, 99.9 wt% – Sigma Aldrich), iron (III) oxide (Fe2O3, 99 wt% – Sigma Aldrich), Iron (II) oxide (FeO, 99.5 wt% – Alfa Aesar), neodymium (III) nitrate hexahydrate (Nd(NO3)3∙6H2O 99.9 wt% – Alfa Aesar), carbon dioxide (CO2, grade 4.0 – Linde Canada), methanol (CH3OH, HPLC, 99.9 wt% – Fisher Chemical), hexanes (ACS grade – Fisher Chemical), and acetone (Reagent grade – Caledon Laboratory Chemicals).

Magnet Preparation The grade N52 NdFeB magnet was purchased from CMS magnetics. The magnet was heated to 350

oC

for one hour in a box furnace (CARBOLITE® HTF18/4) for

demagnetization 24. The Nickel-Copper-Nickel triple coating layer on the magnet was then removed with a Dremel 4000 rotary tool. The demagnetized uncoated magnet was crushed and ground into fine particles using a roller mill. The particle size of the magnet particles was determined to be Median Size = 17.7 μm, Mean Size = 21.6 μm, D10 = 11.1 μm, D90 = 33.8 using a laser particle size analyzer (Horiba Partica LA-950) (Supporting Information Figure S1).

ACS Paragon Plus Environment

10

Page 11 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Preparation of TBP-HNO3 complexes The TBP-HNO3 chelating agent complexes were synthesized by mixing TBP and HNO3 solutions at either 10.4 M (50 wt%), 13.05 M (60 wt%), or 15.7 M (70 wt%) – corresponding to the three levels of test factor X6 in the experimental design. The mixture was agitated manually in a separatory funnel for 5 min, then allowed to settle. The immiscible liquid phases were separated, with the upper organic phase corresponding to the TBP-HNO3 chelating agent. The nitric acid content of the complex was determined by acid-base titration with 0.1 M NaOH solution and phenolphthalein indicator (n = 3). The water content was determined by Karl Fischer titration (n = 3) with a C20 Coulometer (Mettler Toledo). Chelating agent density was determined directly by measuring the weight of a known volume of chelating agent sample. Combining the above data, the stoichiometry of the chelating agents was determined as the following: (X6 = -1: TBP(H2O)0.387(HNO3)1.224;

X6

=

0:

TBP(H2O)0.405(HNO3)1.352;

X6

=

+1:

TBP(H2O)0.560(HNO3)1.969). At the studied temperatures (35 to 55 °C for a maximum of two hours), the TBP-HNO3 complex was expected to be stable

25,26.

However, careful

safety precautions must be taken while handling this system because in a pressure vessel,

ACS Paragon Plus Environment

11

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 64

there is a risk of exothermic reactions between TBP and HNO3. Depending on the pressure, acid concentration and residence time, the onset temperature for exothermic self-accelerating oxidation processes in TBP-HNO3 mixtures is 117 °C

27

and potentially

lower depending if alcohols are present, which could result in hazardous reactions under certain reaction conditions

28,29.

As such, particular care and caution must be taken to

ensure that experiments are carried out within safe operating parameters.

Design of Experiment The experimental parameter settings for the SCFE trials were constructed using Fractional Factorial Experimental Design methodology 30, with the goal of testing a large number of process parameters and developing an empirical model that describes the effect of each operating parameter on the extraction REEs in the system, while maximizing workflow efficiency by reducing the total required number of trial runs. The following operating parameters were investigated: temperature (X1), pressure (X2), residence time (duration) (X3), the sample:chelating agent (S:CA) ratio (X4) (g/mL), agitation rate (X5), HNO3 concentration in the chelating agent (X6), and cosolvent additions (methanol in this case) (X7). The test levels were selected on the basis of the

ACS Paragon Plus Environment

12

Page 13 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

levels used in our previous study on the sc-CO2 extraction of REEs from NiMH batteries 2.

The same levels were used for both studies to allow direct comparison between the two

systems. The parameter settings (xi) were normalized and coded between -1 (low level) and +1 (high level), as the parameter coding allows for direct comparison of the relative impact of each parameter on the system response by direct comparison of the magnitude of the model coefficients. More details regarding Design of Experiment is provided in the Supporting Information (SI). A summary of the levels associated to each of the test parameters is given in Table S1 in the Supporting Information.

Empirical model building and statistical methods The results of the extraction experiments were utilized to build a set of empirical models to describe the extraction (ŷi) of the tested elements as a function of the seven experimental parameters (X1 to X7) (Equations (1-2)) yˆ i   0   1 X 1   2 X 2   3 X 3   4 X 4   5 X 5   6 X 6   7 X 7 ~ ~ ~ ~ ~ ~ ~   24 X 2 X 4   14 X 1 X 4   15 X 1 X 5   12 X 1 X 2   13 X 1 X 3   17 X 1 X 7   16 X 1 X 6

(1)

Where: (assuming 3rd order and higher interactions are negligible)

 24   24  35   67 14  14   36   57

~

~

15  15   26   47

~

~

16  16   25   34

13  13   27   46 17  17   23   45

~

~

~

12  12   37   56 (2)

ACS Paragon Plus Environment

13

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 64

The experimental data was fit to this empirical model multiple Linear Least Squares Regression (mLLSR) according to Equation (3). ˆ is the model parameter vector, containing each of the model parameters ( ˆ0 , ˆ1 , ˆ2 ,…), X is the experimental design matrix (Table S2), and Yi is the response vector, containing each of the measured experimental extraction efficiencies. More details regarding Empirical Model building is provided in the Supporting Information (SI).

ˆ  ( X T X ) 1 ( X T Yi )

(3)

SCFE process Extraction experiments were carried out in a 100 mL high-pressure reactor system manufactured by Supercritical Fluid Technology Inc., USA. A schematic diagram of the extraction system is presented in Supporting Figure S2. Magnet material or synthetic oxide, along with TBP−HNO3 was loaded into the reactor chamber prior to CO2 injection, and increasing pressure and temperature to supercritical conditions. For all tests, the volume of chelating agent was fixed at 5 mL (for example, for a test with a S:CA of 0.2 g/mL, 1 g of magnet powder was utilized). After completion of reaction, rare earth complex formed during the reaction was separated by depressurizing and collecting the liquid

ACS Paragon Plus Environment

14

Page 15 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

chelating agent. In the experiments with methanol cosolvent addition, the amount was determined on a mass/mass basis, considering total number of moles of CO2 in the reactor (Equation (4)). The density of supercritical CO2 was obtained from the National Institute of Standards and Technology (NIST) database 20.

 VCO  CO2 VCH 3OH   X 7 %  2  MWCO2 

 MWCH 3OH    CH OH 3 

(4)

Characterization of the Test Specimens Aqua Regia digestion and ICP-OES characterization For determining the composition of magnet particles before and after SCFE, they were digested in concentrated aqua regia (3 HCl:1 HNO3, approximately 20 mL per 0.25 g sample) at 200 °C (9 °C/min ramp up, 20 min dwell at 200 °C, 5 °C/min ramp down – in a MARS6 Xpress microwave digestion system), followed by Inductively Coupled Plasma Optical Emission Spectrometry (Perkin Elmer Optima 7300 DV) using the following wavelengths: Fe 238.204 nm, Nd 406.109 nm, Dy 353.170 nm, Pr 390.844 nm. Five independent digestion runs were conducted to determine the average concentration of

ACS Paragon Plus Environment

15

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 64

REEs and Fe in the magnet powder feed. Prior to ICP-OES measurement, the digested samples were diluted to 50 mL total volume with DI water, filtered with a 0.45 μm polyethersulfone syringe filter (Sarstedt), and diluted to the measurement concentration range (0.1 – 40 mg/L) with 5 wt% HNO3. The measured values corresponded to the average of concentrations obtained from the three measured dilution levels. Morphological and mineralogical analysis Morphological characterization of the unprocessed and post extraction samples was performed using scanning electron microscopy energy dispersive spectroscopy (SEMEDS, Hitachi SU8230). The mineralogical analysis of magnet was performed using X-ray Diffraction (Rigaku MiniFlex 600).

UV-Vis Complexation Investigation The Nd-TBP complexation stoichiometry was characterized by mixing a fixed amount of Nd(NO3)3, suspended in acetone, with different amounts of TBP, diluted in hexanes, which served as a model for non-polar sc-CO2 solvent. The UV-Vis absorbance of the resulting mixtures was measured with a Lambda 25 UV/Vis Spectrometer (Perkin Elmer).

ACS Paragon Plus Environment

16

Page 17 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

The stoichiometric ratio was determined as the molar ratio at which the curve of absorbance peak vs. TBP concentration reached a plateau.

Results Characterization of ground NdFeB magnet powder The chemical composition of the ground NdFeB magnet was analyzed with microwave assisted aqua regia digestion followed by inductively coupled plasma optical emission spectroscopy (ICP-OES). It was shown that the magnet comprises primarily Fe (52.4 wt%), Nd (18.5 wt%) and Pr (3.6 wt%) with a small amount of Dy (0.1 wt%), with an Fe:REE ratio of approximately 2.3:1 (Figure 1 a). This composition is slightly lower than previously reported values

9,10,31–34;

however, this quantification is based on the ground

magnet powder, which may contain contaminants, such as SiO2 from the grinding process. SEM-EDS point analysis of the individual phases present in the magnets was carried out and the results are presented in the Supporting Information Table S3. These compositions are comparable with those previously reported. Based on the characterization results using Rietveld refinement, boron concentration was determined to be below 1 wt%. Since the accuracy of Rietveld refinement technique at such low

ACS Paragon Plus Environment

17

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 64

concentrations is low, the extraction efficiency of B is not reported in this work. XRD analysis indicated that the magnet mainly consists of tetragonal Nd2Fe14B with a small amount of neodymium trihydroxide (Nd(OH)3) that is the product of the oxidation of Ndrich regions upon contact with air after crushing (Figure 1 (b)). The surface morphology and elemental mapping of the magnet material was characterized using scanning electron microscopy (Figure 1 c). Backscattered secondary electron (BSE) image and energy dispersive spectroscopy (EDS) elemental mapping of a magnet particle cross-section are presented in (Figure 1 d-i), which indicate that the Fe-rich phase forms into discrete granules with Nd- and Pr-rich phases in the intergranular space. The lighter grey region on the BSE image (Figure 1 d) corresponds to an oxygen-rich region that is likely the Nd(OH)3 phase observed in the XRD spectrum. As can be seen in the SEM image in Figure 1 c, the magnet is made of irregularly-shaped bulk particles covered with smaller particles 1 μm in size on the surface. The bulk particle corresponds to the Fe-rich granules observed in the EDS elemental mapping of the particle cross-section and the small particles on the surface correspond to the Nd-rich intergranular phase that is liberated at the surface during the grinding process. These characterization results suggest that a

ACS Paragon Plus Environment

18

Page 19 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

portion of the REE content in the magnet is readily accessible for extraction as the surface-bound Nd-rich particles, whereas the REE contained in the Fe-rich phase would require degradation of these granules for extraction.

Figure 1. Ground NdFeB magnet powder characterization results. a) ICP-OES elemental analysis, error bars represent the sample standard deviation (nFe,Nd,Dy,Pr = 5). b) X-ray diffraction diffractogram. c) Scanning electron micrograph (SEM) of the ground magnet

ACS Paragon Plus Environment

19

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 64

particle. d-i) Backscattered secondary electron (BSE) image and EDS elemental mapping of the magnet particle cross-section.

Extraction Results In this study, the operating parameters and their ranges were selected on the basis of a previous study on the SCFE of REEs from NiMH battery 2 . The relative effect of each operating parameter was assessed in a saturated 27𝐼𝐼𝐼― 4 fractional factorial design, including three replicate centre points. The resolution of this design was then increased from III to IV with a mirror-image foldover test, allowing for unaliased estimation of the primary test factors. The accuracy of the empirical model constructed from these tests was verified by a series of validation tests, which allow an independent comparison of the experimental extraction results and the model-predicted results. The experimental design matrix and the extraction results for each test are presented in Table 1.

Table 1. Overview of Experimental Design Matrix with Corresponding Processing Parameters and Extraction Efficiencies for Fe, Nd, Dy, and Pr. The saturated 𝟐𝟕𝑰𝑰𝑰― 𝟒 and

ACS Paragon Plus Environment

20

Page 21 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

mirror-image foldover results were used to construct empirical extraction models, which were verified with the validation tests. The test runs are presented in standard order but were performed in a randomized order. (S:CA =Solid:Chelating agent ratio).

Run

Extractio

Number

n

(standar d order)

Pressu Temperatu re X1 (°C)

re

Efficienc Tim

S:CA

X2

e X3 X4

(MPa)

(h)

Agitatio

HNO

n

3

(g/mL) (rpm)

X5

MeOH

X6 X7

(M)

(wt%)

y (%) Fe,

Nd,

Dy, Pr

Saturate d 2 7III4 1

48.2, 82.1, 86.8, 35

20.7

1.0

0.10

1500

15.7

0

2

79.0 60.1, 80.6, 85.8,

55

20.7

1.0

0.20

750

15.7

2

3

79.7 50.9, 70.6, 76.0,

35

31.0

1.0

0.20

1500

10.4

2

68.4

ACS Paragon Plus Environment

21

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4

Page 22 of 64

10.4

60.3, 76.4, 79.6,

55

31.0

1.0

0.10

750

5

0 10.4

75.7 61.8, 77.9, 81.5,

35

20.7

2.0

0.10

750

6

2 10.4

77.7 50.3, 67.8, 74.6,

55

20.7

2.0

0.20

1500

0

7

66.9 62.1, 76.5, 78.3,

35

31.0

2.0

0.20

750

15.7

0

8

73.7 48.3, 83.8, 88.7,

55

31.0

2.0

0.10

1500

9

15.7

2

13.0

81.3 51.5, 82.8, 85.8,

10

45

25.8

1.5

0.15

1125

45

25.8

1.5

0.15

1125

1 13.0

1

80.7 52.9, 76.2,

ACS Paragon Plus Environment

22

Page 23 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

81.3, 75.7 11

13.0

52.9, 76.5, 82.4,

45

25.8

1.5

0.15

1125

1

75.7

MirrorImage foldover 12

10.4

53.3, 69.9, 74.9,

55

31.0

2.0

0.20

750

13

2 10.4

70.7 53.3, 75.7, 78.8,

35

31.0

2.0

0.10

1500

0

14

77.1 63.6, 85.0, 87.8,

55

20.7

2.0

0.10

750

15.7

0

15

86.6 38.5, 68.5, 74.8,

16

35

20.7

2.0

0.20

1500

15.7

2

55

31.0

1.0

0.20

1500

15.7

0

69.3 50.4, 74.1,

ACS Paragon Plus Environment

23

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 64

78.9, 75.8 17

49.7, 76.7, 81.2, 35

31.0

1.0

0.10

750

18

15.7

2

10.4

76.9 31.1, 61.1, 75.7,

55

20.7

1.0

0.10

1500

19

2 10.4

56.6 49.0, 63.2, 69.4,

35

20.7

1.0

0.20

750

0

65.4

Validatio n Tests Validatio

64.4,

n1

87.7, 83.6, 55

31

2

0.2

750

15.7

0

81.4

Validatio

55.1,

n2

74.0, 55.4, 55

31

1

0.1

750

10.4

1

50.4

ACS Paragon Plus Environment

24

Page 25 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Validatio

50.7,

n3

86.6, 88.1,

Validatio

55

31

1

0.1

750

10.4

2

83.9

55

20.7

1

0.1

1500

15.7

2

67.1,

n4

93.6, 94.3, 86.9

Validatio

55

20.7

1

0.1

1500

n5

15.7

2

57.4, 94.8, 100.0, 94.2

Visual plots of the factor effect for each primary test parameter is presented in Figure S3 in the Supporting Information. It was shown that Fe extraction (53.5(± 8)%) was significantly lower (t-test, p < 0.001) than that of REEs (77.9(± 9)% on average). Among all operating parameters, increasing chelating agent (X4) and the concentration of HNO3 in the chelating agent (X6) had the largest positive effect on REE extraction. Increasing agitation rate (X5) appeared to have a strong negative effect on Fe extraction. The experimental results were used to build empirical extraction models to enable a quantitative comparison of relative effect of each tested operating parameter. In addition

ACS Paragon Plus Environment

25

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 64

to providing an estimate of the between-run variance, the centre-point runs allowed a qualitative assessment of the quadratic effects of tested parameters.

Empirical extraction modelling The relative effect of each primary test parameter and the aliased second-order interactions were estimated by constructing empirical extraction models by multiple Linear Least Squares Regression (mLLSR). These models do not give mechanistic insight into the extraction process, but they demonstrate how system responds to changes in each parameter. Ordered charts of model parameters alongside the model accuracy are presented in Figure 2. The error bars represent the (1 – α) confidence interval for the estimated model parameter coefficients. Factors with non-zero error bars have a statistically significant effect. More details about the empirical model building are provided in the Supporting Information. As can be seen in Figure 2, the three REEs demonstrated similar extraction trends, while Fe followed a different trend, suggesting that Fe and REEs follow a different extraction mechanism. This observation is consistent with previous studies that reported in aqueous systems, the Nd-rich phases in NdFeB magnets form a galvanic couple and corrode preferentially to the ferromagnetic (Fe-rich) matrix phase 35.

ACS Paragon Plus Environment

26

Page 27 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Figure 2. Ordered chart of factor effect coefficients for the empirical extraction models. The error bars represent the (1 – α) confidence interval for each of the factors (n = 3, tdistribution). The inset panels indicate the correlation between the empirical models and the experimental data, and the coefficient of determination for the models. Each panel

ACS Paragon Plus Environment

27

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 64

represents a different analyte and its associated extraction model. a) Iron, b) Neodymium, c) Dysprosium, d) Praseodymium.

For all three REEs, the factors with significant positive impact on extraction were “HNO3 concentration in the chelating agent (X6)”, “(CA:S) ratio (X4)”, and the sum of three second-order interactions: “temperature × HNO3 concentration (X1X6)”, “pressure × agitation Rate (X2X5)”, and “time × (CA:S) ratio (X3X4)”. Higher HNO3 concentration enhances extraction because there are more acid molecules to attack the magnet particles; thus, liberating more REEs from the solid matrix, and making more NO3- counter-ions available to form sc-CO2-soluble REE-NO3-TBP complexes 36. Similarly, at higher chelating agent content, more TBP molecules and NO3ions are available to capture liberated metal ions. These results are consistent with previous studies on SCFE of Nd from Nd2O3 37. With regards to second order interactions with significant positive impact on REEs extraction, there is a synergistic effect between the two parameters. For example, in the cases of X1X6, high acid concentration at elevated temperature increases reactivity 37, and in the case X3X4, higher chelating agent

ACS Paragon Plus Environment

28

Page 29 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

content in the system requires more residence time to effectively bond to the REE ions. This observation suggests that the matrix degradation by HNO3 is the rate limiting step; thus, longer residence time is required for the added chelating agent to capture all REEs resulting from the matrix degradation by HNO3 molecules. Beyond the tested reaction conditions, previous studies suggested that the extraction behaviour of the Metal-TBPHNO3-H2O system is not only because of CA acidity, but also due to the amount of excess acid and water remaining after reacting with the metal, the TBP concentration, and the amount of metal, and there is an optimal ratio of all these components that result in maximum extraction efficiency 37,38. An exhaustive optimization of these parameters was deemed to be outside the scope of this study. For Nd and Pr, the factors with largest significant negative impact on extraction were the aliased second order interactions of “temperature × agitation rate (X1X5)”, “pressure × HNO3 concentration (X2X6)”, and “(CA:S) ratio × methanol addition (X4X7)”. These factor pairs had an antagonistic effect, meaning that when both factors were set at their high settings, extraction was partially attenuated. In the case of X4X7, simultaneous increase in the amount of chelating agent and methanol could overload the system beyond the

ACS Paragon Plus Environment

29

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 64

solvation power of sc-CO2, thus decreasing the extraction efficiency. In the case of X1X5, increasing temperature has previously been shown to slightly enhance extraction in systems without mechanical agitation;37,39 however, the elevated temperature could destabilise the structure of REE-TBP-NO3 reverse micelles, increasing their susceptibility to be disrupted by mechanical forces; thus, enhancing the negative effect of increasing the agitation rate. This negative effect of increasing agitation rate could be attributed to the unfavourable effect of shear environment on the solvation of REE-TBP-NO3 reverse micelles by sc-CO2. As was mentioned, Fe showed a fundamentally different extraction behaviour, with “HNO3 concentration in the chelating agent (X6)” and “(CA:S) ratio (X4)”, the two parameters with the largest significant positive effect on REE extraction, having no significant effect, while several parameters with no significant effect on REE extraction were shown to decrease Fe extraction. Thus, the operating parameters can be tuned to selectively enhance REE extraction while decreasing Fe extraction. Residence time (X3) and pressure (X2) had the most significant positive single-parameter effects on Fe extraction efficiency. Increasing residence time allows for longer degradation of the magnet particles, thus increasing Fe extraction, while at higher pressures, the density of

ACS Paragon Plus Environment

30

Page 31 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

sc-CO2 is higher, thus it has a higher solvation power for Fe

40,41.

These results suggest

that while Fe extraction is controlled by the parameters that increase the magnet particle solubility in sc-CO2, the extraction of REEs is controlled by the degree of complexation of these ions with the TBP-HNO3 adducts 42. Because the overall Fe extraction was significantly (α = 0.05) lower than that of the REEs, it can be concluded that the Fe-CA complex is less soluble than the REE-CA complex, thus it is more susceptible to break in high shear environments, i.e., at higher agitation rate (X5). This hypothesis is supported by previous studies that demonstrated for a given ligand, different metal ions have different mechanical stability

43.

Although

some previous studies reported an increase in metal extraction in sc-CO2 in the presence of methanol as cosolvent 2, this factor had a negative impact on Fe extraction. Methanol addition is expected to increase metal solubility in sc-CO2 by increasing the solvent polarity and by promoting rapid desorption, preventing re-adsorption, and covering the matrix active sites, or by altering the matrix 44. A hypothesis to explain this observation is that there is a difference in the Fe oxidation state in the alcohol-containing environment. Previously, it has been demonstrated that when solid Fe is exposed to dilute HNO3, it

ACS Paragon Plus Environment

31

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

oxidizes to FeIII in water environment and to FeII in alcohol environment

Page 32 of 64

45.

In the

mechanism section, it will be discussed in detail that FeII has lower extraction efficiency compared with FeIII in sc-CO2. This finding suggest that methanol can be used as a tuning agent to selectively enhance REE extraction while reducing Fe co-extraction, thus reducing downstream separation/purification costs. The complete series of fractional factorial experiments and subsequent empirical model construction enabled evaluating the relative effect of the tested operating parameters on the system response and allowing for the prediction of extraction efficiency beyond the configurations tested experimentally. For instance, the conditions that resulted in maximum REE extraction (weighted by raw magnet composition and pure oxide value 46, is given by Settings 1 in Table 2, result in a predicted extraction efficiency of 86% for the three REEs with a selectivity of 0.60 kgREE/kgFe. Alternatively, the process parameters settings could be adjusted as Settings 2 in Table 2 to maximize the selectivity of REE extraction over Fe extraction at 0.89 kgREE/kgFe with a predicted extraction efficiency of 82% for the REEs. The Settings 2 results in considerably higher REE product at more

ACS Paragon Plus Environment

32

Page 33 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

economically favourable conditions, such as low process pressure and short residence time. Table 2. Overview of optimized parameter settings and predicted extraction efficiencies, value recovery, selectivity, and value extracted in USD/kgmagnet (price calculation details are provided in the Supporting Information). (S:CA = Solid:Chelating agent ratio).

Predic

Predi

ted

cted

extrac Tempera ture X1 (°C)

Press ure X2 (MPa)

Resid ence time X3 (h)

S:CA

Agitati

HNO

X4

on X5

3

(g/mL)

(rpm)

MeOH

X6 X 7 (M) (wt%)

tion efficie ncy for Fe, Nd, Dy,

Avera Predict ge

ed

REE

Selectiv

Reco

ity

very

(kgREE/kgF

(%)

e

)

REE valu e extra cted (US D/kg

and Pr

magnet

(%)

)

Settings 1: Max REE value recovery 60.4, 55

25.8*

1.5*

0.1

750

15.7

1*

85.8,

85.8

86.3,

%

$23.

0.60

94

85.9 Settings 2: Max REE selectivity

ACS Paragon Plus Environment

33

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 34 of 64

38.6, 55

20.7

1

0.1

1500

15.7

2

82.3,

81.5

86.3,

%

$22.

0.89

75

77.9 67.1, Experimental Results: replicate 1

93.6,

92.4

94.3,

%

$25.

0.58

78

86.9 57.4, Experimental Results: replicate 2

94.8,

94.7

100,

%

$26.

0.70

44

94.2 *: no significant effect on value recovery On the basis of these results, the optimal operating conditions were selected as Settings 2. Two replicate runs were performed under these conditions to assess the predictability of the model (Table 2). These runs showed the highest REE extraction efficiency (92– 95%) among all cases studied, even surpassing the model’s prediction; however, the REE:Fe selectivity was lower than model predictions (0.58–0.70 instead of 0.89 kgREE/kgFe). The deviation from the model can be explained by two phenomena. Because the empirical model has a resolution IV, the second order factor interaction parameters are confounded with each other, meaning that there might be interactions in the system

ACS Paragon Plus Environment

34

Page 35 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

that are not independently represented within the model that may affect the response, and/or random error and lack of model fit, since different components of the model are estimated at different confidence levels. Overall, the optimization of the operating parameters resulted a set of process conditions that yields greater than 90% overall REE extraction, which corresponds to approximately $26/kgmagnet. The fractional factorial experimental design methodology is useful for describing the system response to several variables in the most efficient number of trials; however, it has the drawback of offering limited insight into the underlying mechanism that determines the system’s behaviour. With the goal of addressing this knowledge gap, additional investigations were conducted to answer process-critical questions about the system behaviour.

Effect of surface microstructure Figure 3 (a-c) presents SEM micrographs of the magnet particles before and after SCFE. As shown, the distinctive Fe-rich granules covered in Nd-rich particles observed on magnet particles before SCFE (Figure 3 (a)) are not observed on post-SCFE samples. The lack of apparent surface Nd-rich particles is due to the preferential dissolution of the

ACS Paragon Plus Environment

35

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 64

Nd-rich and B-rich intergranular phases over the Fe-rich granules 47 because the former phases have poor corrosion resistance because of the high reactivity of Nd

48,49.

An

increase in the particle size was observed after SCFE, likely because the smaller Fe-rich particles were more readily dissolved, leaving behind the larger particles. The run showing the lowest overall REE extraction (Run 18, Figure 3 b) and the one with the highest overall REE extraction (Validation Run 5, Figure 3 c) show different morphologies. Run 18, in which only 60% of the total REEs was extracted, left residue particles with some flat regions and some rougher surfaces, indicating the regions that were attacked by the acid and chelating agent. Validation Run 5, in which 95% of the total REEs were extracted, also had some flat regions and some with more evident hierarchical roughness and porosity, indicating a more severe attack by the acid and chelating agent. It should be mentioned that the amount of remaining solid after SCFE was significantly lower in Validation Run 5 in which higher acid concentration was utilized.

ACS Paragon Plus Environment

36

Page 37 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

b)

a)

10 μm

c)

50 μm

50 μm

Figure 3. (a) SEM image of the unprocessed magnet sample; (b) SEM image of the magnet sample after extraction in the experimental run with the lowest observed overall REE extraction efficiency (64.5% – Run 18); (c) SEM image of the magnet sample after extraction in the experimental run with the highest observed overall REE extraction efficiency (96.3% – Validation Run 5).

Effect of Methanol on extraction The results of additional validation runs on the effect of methanol addition on Fe and REE extraction are presented in the Supporting Information (Figure S4). These results confirm the empirical model results that the addition of methanol has a negative effect on Fe extraction and a positive effect on REE extraction. As described in the mechanism investigation section, FeII shows a lower extraction efficiency compared with FeIII. When solid Fe is exposed to dilute HNO3, it oxidizes to FeIII in water environment and to FeII in alcohol environment

45.

Thus, in a methanol containing environment, some of the Fe in

ACS Paragon Plus Environment

37

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 38 of 64

the magnet is converted to the FeII state, resulting in the observed decrease in Fe extraction.

Validation of model and comparison with NiMH results To assess the predictability of the model, additional validation tests were conducted, which do not correspond to tests prescribed by the fractional factorial design. Furthermore, because the selected test levels were equivalent to the levels used in the previous study on NiMH batteries 2, model predictions for Nd and Pr in the NiMH study were also calculated. The results of empirical model correlation of the validation runs and those of NiMH battery runs are presented in Figure 4. For the most part, the validation runs fall within or close to the confidence intervals. The average deviation of the experimental results from the model was: Fe: 11%, Nd: 9%, Dy: 10%, and Pr: 13%, which is acceptable given that this model has a resolution of IV, and the confidence for the REE extractions are 75-95%. The Nd and Pr extraction from NiMH battery generally fell outside of the confidence intervals for the models, and the residuals appeared to follow an ordered downwards trend, indicating a difference in the extraction mechanism. Furthermore, in most cases, the observed extraction from the NiMH battery was lower than what would

ACS Paragon Plus Environment

38

Page 39 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

be expected for the NdFeB magnet under the same operating conditions. These results are consistent with the fact that the main matrix material to be broken by HNO3 is Ni for the battery and Fe for the magnet, and these two elements have different corrosion properties, as Ni is more resistant to attack by HNO350.

Figure 4. Empirical model validation with additional run results, and comparison with NiMH extraction results. The validation run experimental results are plotted against the

ACS Paragon Plus Environment

39

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 40 of 64

corresponding model predictions. Extraction predictions were generated for the NiMH and synthetic anode extraction results reported by Yao et al 2.

Exploration of mechanisms for the SCFE process To uncover the mechanism of the REE and Fe extraction from NdFeB magnet using sc-CO2, experiments were conducted on synthetic mixtures containing Nd2O3, Fe2O3, and FeO. It is known that the most common oxidation number for most REEs, including Nd, is III and that Fe exists as both FeII and FeIII. During the extraction process, HNO3 attacks ferromagnetic Nd2Fe14B and oxidizes Nd to Nd3+ and Fe to Fe2+ and Fe3+, which then react with TBP, forming metal complexes. To determine the effect of charge and elemental nature on the complexation process and hence the extraction, the SCFE of two synthetic oxide mixtures was studied, i.e., 1 Fe2+ : 1 Nd3+ (i.e., 50 wt% FeO+50 wt% Nd2O3) and 1 Fe3+ : 1 Nd3+ (i.e., Fe2O3 (50 wt%)+Nd2O3 (50 wt%)) under the test conditions of Run 16 in Table 1.

ACS Paragon Plus Environment

40

Page 41 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Figure 5. Synthetic oxide extraction and complexing trials. a) Compositions of solid extraction residues remaining after sc-CO2 extraction from synthetic mixtures of FeO, Fe2O3, and Nd2O3. The solid bars represent the starting composition of the mixed unextracted oxides and the patterned bars represent the compositions of the extraction residues. The error bars represent the sample standard deviation for the starting composition (n = 3). b) UV-Vis complexation investigation. The absorbance at 585 nm was measured for mixtures of Nd(NO3)3 and TBP in hexanes. The Nd-TBP complex stoichiometric ratio was determined as the intersection of the linear response region (i) and the plateau region (ii) linear fits.

ACS Paragon Plus Environment

41

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 42 of 64

Figure 5 (a) presents the composition of Fe and Nd in the pre- and post-SCFE solid samples. As can be seen, in the oxide mixture containing 1 Fe2+ : 1 Nd3+, the post-SCFE solid is primarily composed of Fe, with almost no Nd present, indicating that the extraction process preferentially selects Nd3+ over Fe2+. Moreover, in the oxide mixture containing 1 Fe3+ : 1 Nd3+, the post-SCFE sample contained similar amounts of Nd and Fe, indicating that there is no preferential difference between the complexation of Fe3+ and Nd3+, thus their extraction is similar. On the basis of these results, the main driver for the complexation reaction between TBP chelating agent and the metal ion is the ion charge (+3 preferred over +2) because usually higher charge is correlated with higher coordination number, which in turn results in a higher solubility in sc-CO2

51.

In the

previous study on SCFE of REEs from NiMH battery, a mechanism for the extraction process was proposed, in which bidentate nitrate anions from the TBP−HNO3 extractant chelate to the REE3+ cation, forming neutral REE nitrates REE(NO3)3 2. Because of hydrophobic interactions between aliphatic tails and dipole – dipole interactions between phosphate groups of TBP, reverse micelles are formed. Therefore, when TBP chelating agent is in contact with hydrophilic solutes like REE salts, they are attached to the polar

ACS Paragon Plus Environment

42

Page 43 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

cores. In the case of Nd, the cation is centered in a coordinate environment comprised of three bidentate nitrate anions and n TBP molecules, consistent with the mechanism suggested by Shimizu et al.

38

(Reaction 1 and 4). For determining n in the case of Nd,

synthetic solutions were made, containing Nd(NO3)3 with varying composition of TBP from 0 to 6.5 in hexane, a model solution for sc-CO2, and these samples were characterized using UV-Visible spectroscopy. As shown in Figure 5 (b), the stoichiometric ratio at which Nd(NO3)3 complexes with TBP is 1 Nd: 3 TBP, indicated by the point at which absorbance peak reaches a plateau. On the basis of these results, n is determined to be 3, which is in agreement with Braatz et al.

52

and Healy and McKay53,54 proposed TBP−trisolvate

complexes of Ln(III) ions, which assumes all lanthanides have a coordination number of 9, suggesting that in the case of Nd3+ chelated with three bidentate NO3-, three TBP molecules are required to complete the coordination sphere. On the basis of these observations, we propose the complex has a formula of Nd(NO3)3(TBP)3. In recent years, a few studies have investigated the structural coordination of trivalent and tetravalent f-element ions in conventional solvent extractions using Extended X-ray absorption fine structure (EXAFS) and X-ray absorption near edge structure

ACS Paragon Plus Environment

43

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 44 of 64

(XANES)52,55,56. Some of these studies are in agreement with the proposed chemistry of TBP−trisolvate complexes of Ln(III) ions. However, there are a few that suggest a different TBP coordination number. Antonio et al.,55 reported that for the tetravalent cerium species in the tri-TBP– n-dodecane solvent extraction system, the reverse micelles consists of tetranuclear Ce(IV)-TBP solvates with (Ce4O4)8+ cores. Ellis and Antonio56 focused on the aggregation phenomena and self-assembly of reverse micelles upon lanthanide extraction. They reported the formation of cerium(III)-malonamide complexes comprising a lanthanide ion coordinated with acid molecules aggregating into higher-order structures of reverse micellar assemblies upon solvent extraction. There are also few other studies that reported a different coordination number for TBP in this complex. For example, Fox et al.51 and Zhu et al.57 reported a complex chemistry of Nd(NO3)3(TBP)4. To gain more insight into the complex chemistry of lanthanide-TBP-NO3 complex and to elucidate the chemical mechanism in the presence of sc-CO2, future study is underway that involves X-ray absorption spectroscopy (XAS). Because Fe3+ has a similar charge to Nd3+ and during the SCFE process of synthetic mixture of 1 Nd3+ : 1 Fe3+, the composition of Nd and Fe in the post-SCFE sample was

ACS Paragon Plus Environment

44

Page 45 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

the same, it was hypothesized that Fe3+ also forms a complex with three bidentate nitrate anions and three TBP molecules, i.e., Fe(NO3)3(TBP)3 (Eq. 2 and 5). However, in the case of Fe2+, it can only bind with two bidentate nitrate anions and n TBP molecules, and if one considers one TPB per bidentate anion, n is expected to be 2 (Reactions 3 and 6). It is known that complexes with a higher number of TBP molecules have higher solubility is sc-CO2 due to higher number of aliphatic tails. Thus, Fe(NO3)2(TBP)2 is expected to have a lower solubility in sc-CO2, which is consistent with the very low extraction efficiency of Fe during the SCFE process of synthetic mixture of 1 Nd3+ : 1 Fe2+. Nd + 3HNO3 → Nd3+ + 3 NO3- + 1.5 H2

(1)

Fe + 3HNO3 → Fe3+ + 3 NO3- + 1.5 H2

(2)

Fe + 2HNO3 → Fe2+ + 2 NO3- + H2

(3)

Nd3+ + 3 NO3- + 3 TBP → Nd(NO3)3(TBP)3

(4)

Fe3+ + 3 NO3- + 3 TBP → Fe(NO3)3(TBP)3

(5)

Fe2+ + 2 NO3- + 2 TBP → Fe(NO3)2(TBP)2

(6)

As demonstrated by acid-base and Karl Fisher titration, the stoichiometry of the TBPHNO3 chelating agent is complex (see Experimental section), which is consistent with the

ACS Paragon Plus Environment

45

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 46 of 64

presence of a mixture of several different distinct adduct complexes formed by strong hydrogen bonding at the P=O site of TBP58. This includes TBP-HNO3, TBP-2HNO3, TBPH2O-HNO3, and/or complicated xTBP-yH2O-zHNO3 clusters

59,60

all present in different

proportions, determined by the acid concentration utilized during chelating agent preparation. The structure of some of these adduct complexes are presented in the Supporting Information Figure S5. On the basis of the results, the following molecular mechanism was proposed for the extraction of Nd (similar for other REEs) and Fe from NdFeB magnet material using TBP− HNO3 as an extracting agent in sc-CO2 (Figure 6). First, HNO3 reacts with Nd and Fe in the Nd2Fe14B particle forming Nd3+, Fe2+, and Fe3+ (Reactions (1-3)). Then metal cations and nitrate anions react in a ratio that results in charge neutrality and complex to n TBP molecules (Reactions (4-6)). Upon complexation and reverse micelles formation, these micelles bond to one another via hydrophobic interactions of aliphatic functionalities and above a certain interaction limit, micellar condensation happens as the last step in which micellar chains are formed as a result of bonding between individual aliphatic shells, forming highly ordered macroscopic assemblies. As mentioned, future studies are

ACS Paragon Plus Environment

46

Page 47 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

underway that involve XAS to elucidate the chemical mechanism of the present extraction process.

Figure 6. Illustration of the sc-CO2 extraction model. This schematic diagram illustrates the hypothesized NdFeB magnet extraction pathway. For simplicity, Nd is the only REE represented, but Pr and Dy can be considered to behave similarly to Nd. This schematic is illustrative in nature and is not drawn to scale.

ACS Paragon Plus Environment

47

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 48 of 64

Conclusions An environmentally friendly process utilizing sc-CO2 as the solvent along with TBPHNO3 chelating agent and methanol as co-solvent was developed to extract Nd, Dy, and Pr from end-of-life NdFeB magnet. This process utilizes minimum amount of organic solvents; thus, it generates minimum amount of waste. A fractional factorial experimental design methodology was utilized to investigate the effect of seven operating parameters, resulting in 94% Nd, 94-100% Dy, and 87-94% Pr extraction with only 57-67% Fe coextraction. A preliminary economic analysis of a full-scale industrialized iteration of this process was performed. The results indicated the process can be economically viable with a break-even processing rate of 30 kg of magnet per hour, given a final product REE purity of >99 wt% and < 0.1 wt% foreign metal contamination, achieved by downstream separations, which will be developed in future work. The process was modelled as a batch process – operated continuously by employing twin extraction vessels in parallel – where one reactor is loaded while the other is operating. The largest cost drivers were capital investment for high pressure equipment and raw materials costs, particularly TBP – a

ACS Paragon Plus Environment

48

Page 49 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

TBP recycling rate of 90% was assumed (10% TBP loss was selected as a preliminary conservative estimate of the amount of the organic-soluble TBP phase, which would be lost during the stripping phase of this process). The main utility costs were the heating cost to bring CO2 to the supercritical state, and the compression cost to liquefy gaseous CO2 remaining after the depressurization stage to allow for storage. This study demonstrates that aerio-metallurgy, known as supercritical fluid extraction, is potentially a viable alternative to pyrometallurgy and hydrometallurgy for the recovery of valuable materials from secondary sources. Future work will focus on developing the REE purification and chelating agent recycling steps of the proposed process, and examining the effect of other potential contaminants, which may become entrained with the magnet waste, such as zinc and nickel from the protective coating layers. This work will have the ultimate goal of developing a sustainable and economical process for the recycling of post-consumer waste streams.

ASSOCIATED CONTENT

ACS Paragon Plus Environment

49

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 50 of 64

Supporting Information. One PDF document containing additional information on the experimental and statistical methodology and supporting results. This material is available free of charge on the ACS Publications website.

AUTHOR INFORMATION

Corresponding Author *corresponding. [email protected] Author Contributions

G.A. conceived and supervised the research. J.Z. and Y.Y. performed the experiments. J.A. produced the experimental design and conducted the data analysis and interpretation. All authors contributed to writing and revising the manuscript

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENT

ACS Paragon Plus Environment

50

Page 51 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

The authors would like to acknowledge the financial support provided by Natural Sciences and Engineering Research Council of Canada (NSERC) (No. 498382). We thank Mr. Kok Long Ng for his contribution to the SEM analysis and Dr. Raiden Acosta for his contribution to the XRD and PSA analysis. We also thank Mr. John Forster for his assistance with the grinding process. Access to the electron microscopy facility in the Canada Foundation for Innovation (CFI) funded Ontario Centre for the Characterization of Advanced Materials and the Walter Curlook Materials Characterization & Processing Laboratory is acknowledged.

REFERENCES (1)

Jha, M. K.; Kumari, A.; Panda, R.; Rajesh Kumar, J.; Yoo, K.; Lee, J. Y. Review on Hydrometallurgical Recovery of Rare Earth Metals. Hydrometallurgy 2016, 165, 2– 26, DOI 10.1016/j.hydromet.2016.01.035.

(2)

Yao, Y.; Farac, N. F.; Azimi, G. Supercritical Fluid Extraction of Rare Earth Elements from Nickel Metal Hydride Battery. ACS Sustain. Chem. Eng. 2018, 6, 1417–1426, DOI 10.1021/acssuschemeng.7b03803.

ACS Paragon Plus Environment

51

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(3)

Page 52 of 64

Goodenough, K. M.; Wall, F.; Merriman, D. The Rare Earth Elements: Demand, Global Resources, and Challenges for Resourcing Future Generations. Nat.

Resour. Res. 2017, 27 (2), 201–216, DOI 10.1007/s11053-017-9336-5.

(4)

Global Wind Energy Council. Global Wind Report 2016 – Annual market update http://gwec.net/publications/global-wind-report-2/global-wind-report-2016/.

(5)

Tan, Q.; Deng, C.; Li, J. Innovative Application of Mechanical Activation for Rare Earth Elements Recovering: Process Optimization and Mechanism Exploration.

Sci. Rep. 2016, 6 (1), 19961, DOI 10.1038/srep19961.

(6)

European Commission. Report on Critical Raw Materials and the Circular Economy. Comm. Staff Work. Doc. 2018.

(7)

Kitagawa, J.; Uemura, R. Rare Earth Extraction from NdFeB Magnet Using a Closed-Loop Acid Process. Sci. Rep. 2017, 7 (1), 1–6, DOI 10.1038/s41598-01708629-z.

(8)

Önal, M. A. R.; Borra, C. R.; Guo, M.; Blanpain, B.; Van Gerven, T.

ACS Paragon Plus Environment

52

Page 53 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Hydrometallurgical Recycling of NdFeB Magnets: Complete Leaching, Iron Removal and Electrolysis. J. Rare Earths 2017, 35 (6), 574–584, DOI 10.1016/S1002-0721(17)60950-5.

(9)

Önal, M. A. R.; Aktan, E.; Borra, C. R.; Blanpain, B.; Van Gerven, T.; Guo, M. Recycling of NdFeB Magnets Using Nitration, Calcination and Water Leaching for REE

Recovery.

Hydrometallurgy

2017,

167,

115–123,

DOI

10.1016/j.hydromet.2016.11.006.

(10) Lee, C. H.; Chen, Y. J.; Liao, C. H.; Popuri, S. R.; Tsai, S. L.; Hung, C. E. Selective Leaching Process for Neodymium Recovery from Scrap Nd-Fe-B Magnet. Metall.

Mater. Trans. A Phys. Metall. Mater. Sci. 2013, 44 (13), 5825–5833, DOI 10.1007/s11661-013-1924-3.

(11) Peelman, S.; Venkatesan, P.; Abrahami, S.; Yang, Y. Recovery of REEs from Endof-Life Permanent Magnet Scrap Generated in WEEE Recycling Plants. In

Proceedings of the First Global Conference on Extractive Metallurgy; 2018; pp 2619–2631.

ACS Paragon Plus Environment

53

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 54 of 64

(12) Tunsu, C.; Petranikova, M.; Gergorić, M.; Ekberg, C.; Retegan, T. Reclaiming Rare Earth Elements from End-of-Life Products: A Review of the Perspectives for Urban Mining Using Hydrometallurgical Unit Operations. Hydrometallurgy 2015, 156, 239– 258, DOI 10.1016/j.hydromet.2015.06.007.

(13) Kershaw, J. R.; Jezko, J. Supercritical Gas Extraction of South African Coals. Sep.

Sci. Technol. 1982, 17 (1), 151–166, DOI 10.1080/01496398208058144.

(14) Ely, J. F.; Baker, J. K. A Review of Supercritical Fluid Extraction; U.S. Department of Commerce: Washington, 1983.

(15) Hubert P, Roselius W, V. O. Method for the Production of Caffeine-Free Coffee Extract. US3843824 A, October 1974.

(16) Yakimishen, R.; Cenkowski, S.; Muir, W. E. Oil Recoveries from Sea Buckthorn Seeds

and

Pulp.

Appl. Eng. Agric. 2005, 21 (6), 1047–1055, DOI

10.13031/2013.20021.

(17) Budich, M.; Brunner, G. Supercritical Fluid Extraction of Ethanol from Aqueous

ACS Paragon Plus Environment

54

Page 55 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Solutions.

J.

Supercrit.

Fluids

2003,

25

(1),

45–55,

DOI

https://doi.org/10.1016/S0896-8446(02)00091-8.

(18) Gumerov, F.; Sagdeev, A.; Gallyamov, R.; Galimova, A.; Sagdeev, K. Regeneration of the Catalysts by Supercritical Fluid Extraction. Int. J. Anal. Mass Spectrom.

Chromatogr. 2014, 2 (March), 1–14, DOI 10.4236/ijamsc.2014.21001.

(19) De Fillipi, R. P.; Kyukonis, V. J.; Robey, R. J.; Modell, M. Supercritical Fluid

Regeneration of Activated Carbon for Adsorption of Pesticides; 1980.

(20) National Institute of Standards and Technology. NIST Chemistry WebBook; 2017.

(21) Ding, X.; Liu, Q.; Hou, X.; Fang, T. Supercritical Fluid Extraction of Metal Chelate: A Review. Critical Reviews in Analytical Chemistry. 2016, pp 1–20, DOI 10.1080/10408347.2016.1225254.

(22) Wai, C. M.; Waller, B. Dissolution of Metal Species in Supercritical Fluids Principles and Applications. Ind. Eng. Chem. Res. 2000, 39 (12), 4837–4841, DOI 10.1021/ie0002879.

ACS Paragon Plus Environment

55

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 56 of 64

(23) Lin, Y.; Liu, C.; Wu, H.; Yak, H. K.; Wai, C. M. Supercritical Fluid Extraction of Toxic Heavy Metals and Uranium from Acidic Solutions with Sulfur-Containing Organophosphorus Reagents. Ind. Eng. Chem. Res. 2003, 42 (7), 1400–1405, DOI 10.1021/ie020804i.

(24) Schulze, R.; Buchert, M. Estimates of Global REE Recycling Potentials from NdFeB Magnet

Material.

Resour.

Conserv.

Recycl.

2016,

113,

12–27,

DOI

10.1016/j.resconrec.2016.05.004.

(25) Burger, L. L. The Chemistry of Tributyl Phosphate: A Review; 1955, DOI 10.2172/4334996.

(26) Nichols, G. S. Decomposition of the Tributyl Phosphate-Nitrate Complexes; 1960, DOI 10.2172/4117445.

(27) Nazin, E. R.; Zachinyaev, G. M.; Belova, E. V.; Tkhorzhnitskii, G. P.; Myasoedov, B. F. Exothermic Processes in Mixtures of TBP with Nitric Acid. Radiochemistry 2017, 59 (5), 512–519, DOI 10.1134/S1066362217050125.

ACS Paragon Plus Environment

56

Page 57 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

(28) Smitha, V. S.; Kumar, J. S. V.; Surianarayanan, M.; Seshadri, H.; Lakshman, N. V. Reactive Chemical Pathway of Tributyl Phosphate with Nitric Acid. Process Saf.

Environ. Prot. 2018, 116, 677–684, DOI 10.1016/j.psep.2018.03.028.

(29) J., S. V. kumar; Smitha, V. S.; Sivanesh, N. E.; Surianarayanan, M.; Seshadri, H.; Lakshman, V. Reactive Thermal Hazards of Irradiated Tributyl Phosphate with Nitric

Acid.

Thermochim.

Acta

2018,

666

(June

2017),

18–26,

DOI

10.1016/j.tca.2018.05.022.

(30) Box, G. E. P.; Hunter, J. S.; Hunter, W. G. Statistics for Experimenters: Design,

Innovation, and Discovery, 2nd ed.; Wiley, 2005.

(31) Yoon, H.-S.; Kim, C.-J.; Chung, K.; Jeon, S.; Park, I.; Yoo, K.; Jha, M. The Effect of Grinding and Roasting Conditions on the Selective Leaching of Nd and Dy from NdFeB Magnet Scraps. Metals (Basel). 2015, 5 (3), 1306–1314, DOI 10.3390/met5031306.

(32) Vander Hoogerstraete, T.; Blanpain, B.; Van Gerven, T.; Binnemans, K. From NdFeB Magnets towards the Rare-Earth Oxides: A Recycling Process Consuming

ACS Paragon Plus Environment

57

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Only

Oxalic

Acid.

RSC

Adv.

2014,

4

(109),

Page 58 of 64

64099–64111,

DOI

10.1039/C4RA13787F.

(33) Behera, S. S.; Parhi, P. K. Leaching Kinetics Study of Neodymium from the Scrap Magnet Using Acetic Acid. Sep. Purif. Technol. 2016, 160, 59–66, DOI 10.1016/j.seppur.2016.01.014.

(34) Riaño, S.; Binnemans, K. Extraction and Separation of Neodymium and Dysprosium from Used NdFeB Magnets: An Application of Ionic Liquids in Solvent Extraction towards the Recycling of Magnets. Green Chem. 2015, 17 (5), 2931– 2942, DOI 10.1039/C5GC00230C.

(35) Prakash, V.; Sun, Z. H. I.; Sietsma, J.; Yang, Y. Electrochemical Recovery of Rare Earth Elements From Magnet Scraps ‐ a Theoretical Analysis. ERES2014 1st Eur.

Rare Earth Resour. Conf. 2014, 163–170.

(36) Baek, D. L.; Fox, R. V.; Case, M. E.; Sinclair, L. K.; Schmidt, A. B.; McIlwain, P. R.; Mincher, B. J.; Wai, C. M. Extraction of Rare Earth Oxides Using Supercritical Carbon Dioxide Modified with Tri-n-Butyl Phosphate-Nitric Acid Adducts. Ind. Eng.

ACS Paragon Plus Environment

58

Page 59 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

Chem. Res. 2016, 55 (26), 7154–7163, DOI 10.1021/acs.iecr.6b00554.

(37) Duan, W.; Cao, P.; Zhu, Y. Extraction of Rare Earth Elements from Their Oxides Using Organophosphorus Reagent Complexes with HNO3 and H2O in Supercritical

CO2.

J.

Rare

Earths

2010,

28

(2),

221–226,

DOI

https://doi.org/10.1016/S1002-0721(09)60084-3.

(38) Shimizu, R.; Sawada, K.; Enokida, Y.; Yamamoto, I. Supercritical Fluid Extraction of Rare Earth Elements from Luminescent Material in Waste Fluorescent Lamps.

J. Supercrit. Fluids 2005, 33 (3), 235–241, DOI 10.1016/j.supflu.2004.08.004.

(39) Tomioka, O.; Meguro, Y.; Enokida, Y.; Yamamoto, I.; Yoshida, Z. Dissolution Behavior of Uranium Oxides with Supercritical CO2using HNO3-TBP Complex as a

Reactant.

J. Nucl. Sci. Technol. 2001, 38 (12), 1097–1102, DOI

10.1080/18811248.2001.9715141.

(40) Zhu, L. Y.; Wang, Z. F.; He, H.; Tian, G. Competition between Lanthanides in Extraction with Tri-n-Butyl Phosphate in Supercritical CO2 from Solid Nitrates. RSC

Adv. 2016, 6 (99), 96531–96537, DOI 10.1039/C6RA20687E.

ACS Paragon Plus Environment

59

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 60 of 64

(41) Ren, Q.; Duan, D.; Zhang, H.; Su, B.; Zhang, Z.; Bao, Z.; Yang, Y. Solubility of Novel Open-Chain Crown Ether Bridged Diphosphates in Supercritical Carbon Dioxide. J. Chem. Thermodyn. 2013, 67, 40–47, DOI 10.1016/j.jct.2013.07.017.

(42) Karshigina, Z.; Abisheva, Z.; Bochevskaya, Y.; Akcil, A.; Sharipova, A.; Sargelova, E. Processing of Phosphorus Slag with Recovery of Rare Earth Metals and Obtaining Silicon Containing Cake. IOP Conf. Ser. Earth Environ. Sci. 2016, 44 (5), DOI 10.1088/1755-1315/44/5/052003.

(43) Gensler, M.; Eidamshaus, C.; Galstyan, A.; Knapp, E. W.; Reissig, H. U.; Rabe, J. P. Mechanical Rupture of Mono- and Bivalent Transition Metal Complexes in Experiment and Theory. J. Phys. Chem. C 2015, 119 (8), 4333–4343, DOI 10.1021/jp511104m.

(44) Cui, Z.; Zhang, G.; Song, W.; Song, Y. Supercritical Fluid Extraction of Metal Ions from a Solid Matrix with 8-Hydroxyquinoline and Carbon Dioxide. J. Liq.

Chromatogr. Relat. Technol. 2004, 27 (6), DOI 10.1081/JLC-120030173.

(45) Fujihara, T. Passivity of Iron by Dilute Nitric Acid. Ind. Eng. Chem. 1925, 18 (1), 62–

ACS Paragon Plus Environment

60

Page 61 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

63, DOI 10.1021/ie50193a025.

(46) Argus Consulting Services. Argus Rare Earths Monthly Outlook; 2017.

(47) Song, Y. W.; Zhang, H.; Yang, H. X.; Song, Z. L. A Comparative Study on the Corrosion Behavior of NdFeB Magnets in Different Electrolyte Solutions. Mater.

Corros. 2008, 59 (10), 794–801, DOI 10.1002/maco.200804175.

(48) Schultz, L.; El-Aziz, A. M.; Barkleit, G.; Mummert, K. Corrosion Behaviour of NdFe-B Permanent Magnetic Alloys. Mater. Sci. Eng. A 1999, 267 (2), 307–313, DOI 10.1016/S0921-5093(99)00107-0.

(49) Bala, H.; Szymura, S.; Wysłocki, J. J. Electrochemical Corrosion Resistance of FeNd-B Permanent Magnets. J. Mater. Sci. 1990, 25 (1), 571–574, DOI 10.1007/BF00714076.

(50) Cannon, R. D. The Dissolution of Iron and Nickel in Dilute Aqua Regia; U.S. Atomic Energy Commission, 1961.

(51) Fox, R. V.; Ball, R. D.; Harrington, P. D. B.; Rollins, H. W.; Jolley, J. J.; Wai, C. M.

ACS Paragon Plus Environment

61

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Praseodymium

Nitrate

and

Neodymium

Nitrate

Page 62 of 64

Complexation

with

Organophosphorus Reagents in Supercritical Carbon Dioxide Solvent. J. Supercrit.

Fluids 2004, 31 (3), 273–286, DOI 10.1016/j.supflu.2003.11.006.

(52) Braatz, A. D.; Antonio, M. R.; Nilsson, M. Structural Study of Complexes Formed by Acidic and Neutral Organophosphorus Reagents. Dalt. Trans. 2017, 46 (4), 1194–1206, DOI 10.1039/C6DT04305D.

(53) Healy, T. V.; McKay, H. A. C. The Extraction of Nitrates by Tri-n-Butyl Phosphate (TBP). Part 2.—The Nature of the TBP Phase. Trans. Faraday Soc. 1956, 52, 633– 642, DOI 10.1039/TF9565200633.

(54) Healy, T. V; Mckay, H. A. C. Complexes between Tributyl Phosphate and Inorganic Nitrates. Recl. des Trav. Chim. des Pays-Bas 2010, 75 (6), 730–736, DOI 10.1002/recl.19560750623.

(55) Antonio, M. R.; Ellis, R. J.; Estes, S. L.; Bera, M. K. Structural Insights into the Multinuclear Speciation of Tetravalent Cerium in the Tri-n-Butyl Phosphate-nDodecane Solvent Extraction System. Phys. Chem. Chem. Phys. 2017, 19 (32),

ACS Paragon Plus Environment

62

Page 63 of 64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sustainable Chemistry & Engineering

21304–21316, DOI 10.1039/c7cp03350h.

(56) Ellis, R. J.; Antonio, M. R. Coordination Structures and Supramolecular Architectures in a Cerium(III)-Malonamide Solvent Extraction System. Langmuir 2012, 28 (14), 5987–5998, DOI 10.1021/la3002916.

(57) ZHU, L.; DUAN, W.; XU, J.; ZHU, Y. Kinetics of Reactive Extraction of Nd from Nd2O3with TBP-HNO3Complex in Supercritical Carbon Dioxide. Chinese J. Chem.

Eng. 2009, 17 (2), 214–218, DOI 10.1016/S1004-9541(08)60196-2.

(58) Kinney, P. J. An Infrared Study of the Tributyl Phosphate-Nitric Acid-Water Solvent Extraction System, Iowa State University Of Science and Technology, 1963.

(59) Shuler, W. E. Nitric Acid Species in Tri-n-Butyl Phosphate Solutions - AEC

Research and Development Report; 1960.

(60) Servis, M. J.; Wu, D. T.; Braley, J. C. Network Analysis and Percolation Transition in Hydrogen Bonded Clusters: Nitric Acid and Water Extracted by Tributyl Phosphate. Phys. Chem. Chem. Phys. 2017, 19 (18), 11326–11339, DOI

ACS Paragon Plus Environment

63

ACS Sustainable Chemistry & Engineering 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 64 of 64

10.1039/c7cp01845b.

SYNOPSIS A systematic study was performed to develop a method for the sustainable recovery of rare earth elements from NdFeB magnets, using supercritical fluid extraction. TOC FIGURE

ACS Paragon Plus Environment

64