Resource Recovery of Waste Nd–Fe–B Scrap: E ﬀ ective Separation of Fe as High-Purity Hematite Nanoparticles

: Recycling rare-earth elements from Nd magnet scrap (Nd–Fe–B scrap) is a highly economical process; however, its e ﬃ ciency is low due to large portions of Fe impurity. In this study, the e ﬀ ective separation of Fe impurity from scrap was performed through an integrated nitric acid dissolution and hydrothermal route with the addition of fructose. Results showed that more than 99% of the scrap was dissolved in nitric acid, and after three dilutions that the Nd, Pr, Dy and Fe concentrations in the diluted acid were 9.01, 2.11, 0.37 and 10.53 g / L, respectively. After the acid was hydrothermally treated in the absence of fructose, only 81.8% Fe was removed as irregular hematite aggregates, whilst more than 98% rare-earth elements were retained. By adding fructose at an M fructose / M nitrate ratio of 0.2, 99.94% Fe was precipitated as hematite nanoparticles, and the loss of rare-earth elements was < 2%. In the treated acid, the residual Fe was 6.3 mg / L, whilst Nd, Pr and Dy were 8.84, 2.07 and 0.36 g / L, respectively. Such composition was conducive for further recycling of high-purity rare-earth products with low Fe impurity. The generated hematite nanoparticles contained 67.92% Fe with a rare-earth element content of < 1%. This value meets the general standard for commercial hematite active pharmaceutical ingredients. In this manner, a green process was developed for separating Fe from Nd–Fe–B scrap without producing secondary waste.


Introduction
Nd magnet scrap (Nd-Fe-B scrap) is a common material in the collection of magnetic products in the permanent magnet industry [1]. This scrap is composed of approximately 30% rare-earth elements and 50-70% Fe impurity [2]. Given that such rare-earth elements are indispensable to improving the thermal and oxidation-corrosion resistances of advanced materials in advanced manufacturing and energy industries [3], recycling rare-earth elements from scrap is a highly economical and reliable process.
Numerous approaches for recycling rare-earth elements from scrap have been developed, including selective precipitation [4], solvent extraction [5][6][7][8][9] and cationic exchange [1]. Amongst these approaches, the first step in recycling rare-earth elements is dissolving scrap in strong acids, including sulphuric, nitric and hydrochloric acids [5,6,10]. After dissolution, rare-earth elements are recycled in two ways. The first method is precipitating ferric Fe by adjusting the rare-earth element-bearing acid to above pH 3 [4]. For example, Önal et al. [4] dissolved scrap in sulphuric acid and found that ferrous Fe in

Scrap Dissolution
The scrap sample was collected from the cutting workplace of SanHe Ltd. (Jilin, China), dried at 105 • C for 3 h, ground and sifted with a sieve (1 mm mesh). The grey scrap was subjected to wet chemical analysis following the method of Sandroni and Smith [20]. The major elements of the sludge based on the dry weight was 45.5% Fe, 38.9% Nd, 9.1% Pr and 1.6% Dy, indicating that rare-earth elements and Fe predominated the scrap. The scrap was dissolved as follows: 5 g of scrap was dispersed in 100 mL 27% nitric acid, and then the acid was heated at 60 • C under constant stirring at 250 rpm. After 30 min, the acid bubbled to emit a yellowish smoke due to the redox reaction between the nitrate and Fe/rare-earth metals during the scrap's dissolution. In the nitric acid solution, nitrate had a high redox potential E NO − 3 /NO of 2.23 V [3] to oxidize the mental in the scrap to generate Fe 3+ and trivalent rare-earth elements [4]. Subsequently, a yellowish suspension was generated after 6 h and its volume was approximately 74 mL, indicating that approximately 26% of the solution evaporated. The suspension was centrifugal separation at 5500 rpm for 5 min, and then the residual solid was collected and dried at 102 • C for 4 h. The residual undissolved solid was approximately 0.08 g, suggesting that more than 99% of the scrap was dissolved. After centrifugation, the supernatant was collected and characterised via ion chromatography (IC, 881, Metrohm, Herisau, Switzerland). The results indicated that nitrate concentration was 266.1 g/L, demonstrating that approximately 25% nitrate was exhausted during the scrap's dissolution. To prevent the vessel from exploding in the subsequent hydrometallurgical experiment, the supernatant was diluted three times and characterised via inductively coupled plasma atomic emission spectroscopy (ICP-OES, Profile, Leeman, Hudson, Sustainability 2020, 12, 2624 3 of 9 NH, USA). The major elements were determined to be 88.7 g/L NO 3 − , 10.53 g/L Fe, 9.01 g/L Nd, 2.11 g/L Pr and 0.37 g/L Dy.

Precipitation of Fe Impurity
Fe impurity in the acid was separated via a one-step hydrothermal route. Firstly, 20 mL of diluted acid was poured into a 50 mL Teflon™ vessel, followed by the addition of fructose at an M fructose /M nitrate molar ratio (hereafter referred to as "molar ratio") of 0.1. Secondly, the vessel was placed in a dry oven (DHG-9030A, Shanghai-Yiheng, China) and heated at 160 • C for 6 h. Thirdly, the deposit was collected and freeze-dried at −80 • C for 24 h. Control experiments were performed by varying the molar ratio from 0 to 0.1, 0.2 and 0.5. Optimal Fe removal occurred at a molar ratio of 0.2; thus, the time course of the removal rate of Fe was also investigated. Each experiment was repeated three times, and the standard deviation was calculated as the error bar.

Characterisation
To analyse the Fe precipitate, the obtained deposits were characterised via scanning electron microscopy (SEM, NanoSEM 450, FEI Co., Hillsboro, OR, USA) and X-ray diffractometry system (XRD, RAPID-S, Rigaku, Tokyo, Japan). Acidity before and after hydrothermal treatment was recorded using a pH meter (E-201-C, LeiCi, Shanghai, China) and organic matter in the acid solution was determined using a total organic carbon analyser (TOC, TOC-5000A, Shimadzu, Kyoto, Japan).

Fructose Dosage Optimisation
Fe impurity was efficiently separated from the rare-earth element-bearing acid, as shown in Figure 1. Without fructose, only 81.8% of the total Fe was removed, and the loss of Nd, Pr and Dy were 1.4%, 1.3% and 1.1%, respectively. With fructose, Fe removal reached 97% at a molar ratio of 0.1, nearly 100% at a molar ratio of 0.2 and only 72.5% at a molar ratio of 0.5 ( Figure 1a). Accordingly, the loss of Nd, Pr and Dy increased slightly to 1.5%, 1.8% and 1.5% at a 0.1 molar ratio and 1.7%, 1.9% and 1.6% at a 0.2 molar ratio, but dramatically to 21.9%, 27.5% and 29.5% at the 0.5 molar ratio (Figure 1b), respectively. This outcome indicated that fructose was overdosed at a molar ratio of 0.5. Thus, the molar ratio of 0.2 was optimal. Under this condition, nearly 100% Fe was removed and only 6.3 mg/L Fe was retained, but rare-earth element loss was below 2%.

Precipitation of Fe Impurity
Fe impurity in the acid was separated via a one-step hydrothermal route. Firstly, 20 mL of diluted acid was poured into a 50 mL Teflon™ vessel, followed by the addition of fructose at an Mfructose/Mnitrate molar ratio (hereafter referred to as "molar ratio") of 0.1. Secondly, the vessel was placed in a dry oven (DHG-9030A, Shanghai-Yiheng, China) and heated at 160 °C for 6 h. Thirdly, the deposit was collected and freeze-dried at −80 °C for 24 h. Control experiments were performed by varying the molar ratio from 0 to 0.1, 0.2 and 0.5. Optimal Fe removal occurred at a molar ratio of 0.2; thus, the time course of the removal rate of Fe was also investigated. Each experiment was repeated three times, and the standard deviation was calculated as the error bar.

Characterisation
To analyse the Fe precipitate, the obtained deposits were characterised via scanning electron microscopy (SEM, NanoSEM 450, FEI Co., USA) and X-ray diffractometry system (XRD, RAPID-S, Rigaku, Japan). Acidity before and after hydrothermal treatment was recorded using a pH meter (E-201-C, LeiCi, China) and organic matter in the acid solution was determined using a total organic carbon analyser (TOC, TOC-5000A, Shimadzu, Japan).

Fructose Dosage Optimisation
Fe impurity was efficiently separated from the rare-earth element-bearing acid, as shown in Figure. 1. Without fructose, only 81.8% of the total Fe was removed, and the loss of Nd, Pr and Dy were 1.4%, 1.3% and 1.1%, respectively. With fructose, Fe removal reached 97% at a molar ratio of 0.1, nearly 100% at a molar ratio of 0.2 and only 72.5% at a molar ratio of 0.5 ( Figure 1a). Accordingly, the loss of Nd, Pr and Dy increased slightly to 1.5%, 1.8% and 1.5% at a 0.1 molar ratio and 1.7%, 1.9% and 1.6% at a 0.2 molar ratio, but dramatically to 21.9%, 27.5% and 29.5% at the 0.5 molar ratio ( Figure  1b), respectively. This outcome indicated that fructose was overdosed at a molar ratio of 0.5. Thus, the molar ratio of 0.2 was optimal. Under this condition, nearly 100% Fe was removed and only 6.3 mg/L Fe was retained, but rare-earth element loss was below 2%. When fructose was absent, the removed Fe precipitated in the form of irregular hematite aggregates (Figures 2a and 3a). Fructose was helpful in removing Fe. By adding fructose at a molar ratio of 0.1, Fe precipitated as hematite nanoparticles (Figures 2b and 3b) with a diameter of 400 nm. The diameter of the hematite nanoparticles was further decreased to 100 nm with an increase in molar ratio to 0.2 (Figures 2c and 3c). When the molar ratio was 0.5, the obtained deposit presented three sharp peaks at 2θ = 18.5°, 18.9° and 22.7° (Figure 2d), which belonged to humboldtine, and a broad When fructose was absent, the removed Fe precipitated in the form of irregular hematite aggregates (Figures 2a and 3a). Fructose was helpful in removing Fe. By adding fructose at a molar ratio of Sustainability 2020, 12, 2624 4 of 9 0.1, Fe precipitated as hematite nanoparticles (Figures 2b and 3b) with a diameter of 400 nm. The diameter of the hematite nanoparticles was further decreased to 100 nm with an increase in molar ratio to 0.2 (Figures 2c and 3c). When the molar ratio was 0.5, the obtained deposit presented three sharp peaks at 2θ = 18.5 • , 18.9 • and 22.7 • (Figure 2d), which belonged to humboldtine, and a broad peak at 2θ = 22.2 • , which was affiliated with the C sphere from the polymerisation of overdosed fructose [21]. Accordingly, fine spherical particles with diameters of 2-5 µm were observed (Figure 3d). This condition demonstrated that overdosed fructose was involved in the formation of C spheres. peak at 2θ = 22.2°, which was affiliated with the C sphere from the polymerisation of overdosed fructose [21]. Accordingly, fine spherical particles with diameters of 2-5 μm were observed ( Figure  3d). This condition demonstrated that overdosed fructose was involved in the formation of C spheres.

Composition of Residual Acid after Fe Removal
The residual acid solution was characterised as shown in Figure 4. Without fructose, a slight decrease in nitrate concentration from 86.3 g/L to 78.4 g/L (Figure 4a) was obeserved due to the thermal decomposition of nitrate to NO2 and/or NO (Equations (1) and (2)) [15]. Moreover, the acid's pH declined from 0.3 to 0.21 (Figure 4a) because H + was generated during the hydrolysis of ferric Fe to Fe oxyhydroxide with hematite as the final product (Equation (3)). As the hydrolysis of ferric Fe continued, H + accumulated in the residual acid, inhibiting the hydrolysis of ferric Fe (Equation (3)) and resulting in a high Fe residual (Figure 1a). peak at 2θ = 22.2°, which was affiliated with the C sphere from the polymerisation of overdosed fructose [21]. Accordingly, fine spherical particles with diameters of 2-5 μm were observed ( Figure  3d). This condition demonstrated that overdosed fructose was involved in the formation of C spheres.

Composition of Residual Acid after Fe Removal
The residual acid solution was characterised as shown in Figure 4. Without fructose, a slight decrease in nitrate concentration from 86.3 g/L to 78.4 g/L (Figure 4a) was obeserved due to the thermal decomposition of nitrate to NO2 and/or NO (Equations (1) and (2)) [15]. Moreover, the acid's pH declined from 0.3 to 0.21 (Figure 4a) because H + was generated during the hydrolysis of ferric Fe to Fe oxyhydroxide with hematite as the final product (Equation (3)). As the hydrolysis of ferric Fe continued, H + accumulated in the residual acid, inhibiting the hydrolysis of ferric Fe (Equation (3)) and resulting in a high Fe residual (Figure 1a).

Composition of Residual Acid after Fe Removal
The residual acid solution was characterised as shown in Figure 4. Without fructose, a slight decrease in nitrate concentration from 86.3 g/L to 78.4 g/L (Figure 4a) was obeserved due to the thermal decomposition of nitrate to NO 2 and/or NO (Equations (1) and (2)) [15]. Moreover, the acid's pH declined from 0.3 to 0.21 (Figure 4a) because H + was generated during the hydrolysis of ferric Fe to Fe oxyhydroxide with hematite as the final product (Equation (3)). As the hydrolysis of ferric Fe continued, H + accumulated in the residual acid, inhibiting the hydrolysis of ferric Fe (Equation (3)) and resulting in a high Fe residual (Figure 1a).
With the addtion of fructose, a steady decrease in nitrate from 39.3 g/L to 5.8 g/L and 0.9 g/L was observed with an increase in molar ratio from 0.1 to 0.2 and 0.5 (Figure 4a), respectively. Nitrate is a strong oxidant that can react with fructose to produce ketogluconic, butanedioic and oxalic acids, and finally, CO 2 and H 2 O (Equation (4)) [22]. Figure 4b shows that the TOC of the acid apparently increased from 16.4 g/L to 78.14 g/L with an increase in molar ratio from 0.1 to 0.5 before hydrothermal reaction. After hydrothermal reaction, the residual TOC was approximately 125 mg/L within the molar ratio range of 0.2, but it remained at 26.4 g/L at a molar ratio of 0.5 (Figure 4b) in accordance with the overdosed fructose. Moreover, the redox reduction between nitrate and fructose was accompanied by the consumption of H + (Equation (4)). When the molar ratio varied from 0.1 to 0.2, the acid's pH also increased from 0.92 to 1.46 (Figure 4a), which accelerated the hydrolysis of ferric Fe to hematite particles. When fructose was overdosed, nitrate was nearly completely consumed in the redox reaction (Equation (4)), and the acid's pH sharply increased to 3.25 (Figure 4a). Thus, fructose was oxidised incompletely to accumulate the intermediate oxalic acid in the solution. Subsequently, trivalent Fe predominated in nitric acid and was reduced by fructose to form ferrous Fe. In parallel, trivalent rare-earth elements were also reduced by the overdosed fructose [23]. A portion of oxalic acid was complexed with ferrous Fe with the generation of humboldtine precipitates. Accordingly, complexation reaction between oxalic acid and rare-earth elements also occurred [24], resulting in the precipitation of these elements. During hydrothermal treatment, unoxidised fructose was polymerised into macromolecules, e.g., aromatic compounds and oligosaccharides [25], and further nucleated to C spheres via the aldol condensation and dehydration route [21]. Such macromolecules, which also have plenty of hydroxyl groups and exhibit a high affinity towards chelated ferrous Fe and rare-earth elements [26], remained in the treated acid. This condition inhibited the hydrolysation of Fe into Fe oxyhydroxide. Therefore, Fe was removed inefficiently at a molar ratio of 0.5.
Sustainability 2019, 11, x FOR PEER REVIEW 5 of 9 3 → + 2 (2) 2Fe( 3 ) 3 + 3 2 = 2 3 + 6 3 − + 6 + With the addtion of fructose, a steady decrease in nitrate from 39.3 g/L to 5.8 g/L and 0.9 g/L was observed with an increase in molar ratio from 0.1 to 0.2 and 0.5 (Figure 4a), respectively. Nitrate is a strong oxidant that can react with fructose to produce ketogluconic, butanedioic and oxalic acids, and finally, CO2 and H2O (Equation (4)) [22]. Figure 4b shows that the TOC of the acid apparently increased from 16.4 g/L to 78.14 g/L with an increase in molar ratio from 0.1 to 0.5 before hydrothermal reaction. After hydrothermal reaction, the residual TOC was approximately 125 mg/L within the molar ratio range of 0.2, but it remained at 26.4 g/L at a molar ratio of 0.5 (Figure 4b) in accordance with the overdosed fructose. Moreover, the redox reduction between nitrate and fructose was accompanied by the consumption of H + (Equation (4)). When the molar ratio varied from 0.1 to 0.2, the acid's pH also increased from 0.92 to 1.46 (Figure 4a), which accelerated the hydrolysis of ferric Fe to hematite particles. When fructose was overdosed, nitrate was nearly completely consumed in the redox reaction (Equation (4)), and the acid's pH sharply increased to 3.25 ( Figure  4a). Thus, fructose was oxidised incompletely to accumulate the intermediate oxalic acid in the solution. Subsequently, trivalent Fe predominated in nitric acid and was reduced by fructose to form ferrous Fe. In parallel, trivalent rare-earth elements were also reduced by the overdosed fructose [23]. A portion of oxalic acid was complexed with ferrous Fe with the generation of humboldtine precipitates. Accordingly, complexation reaction between oxalic acid and rare-earth elements also occurred [24], resulting in the precipitation of these elements. During hydrothermal treatment, unoxidised fructose was polymerised into macromolecules, e.g., aromatic compounds and oligosaccharides [25], and further nucleated to C spheres via the aldol condensation and dehydration route [21]. Such macromolecules, which also have plenty of hydroxyl groups and exhibit a high affinity towards chelated ferrous Fe and rare-earth elements [26], remained in the treated acid. This condition inhibited the hydrolysation of Fe into Fe oxyhydroxide. Therefore, Fe was removed inefficiently at a molar ratio of 0.5.

Hydrothermal Time Optimisation at the Optimal Fructose Dosage
To optimise hydrothermal time, control experiments of Fe removal at the optimal molar ratio of 0.2 were performed. The result is presented in Figure 5. After hydrothermal treatment, Fe removal

Hydrothermal Time Optimisation at the Optimal Fructose Dosage
To optimise hydrothermal time, control experiments of Fe removal at the optimal molar ratio of 0.2 were performed. The result is presented in Figure 5. After hydrothermal treatment, Fe removal reached 1.35% for 0.25 h, 46.5% for 1 h, 98.5% for 3 h and nearly 100% for 6 h (Figures 5a and 1a). The loss of Nd, Pr and Dy was 0.3%, 0.2% and 0.2% after 0.25 h, apparently up to 1.2%, 0.9% and 1.1% after 1 h, and steadily increased to 1.7%, 1.7% and 1.8% after 3 h (Figure 5b), respectively, and kept nearly unchanged with an extended time course from 3 h to 6 h (Figure 1b).
Sustainability 2019, 11, x FOR PEER REVIEW 6 of 9 after 1 h, and steadily increased to 1.7%, 1.7% and 1.8% after 3 h (Figure 5b), respectively, and kept nearly unchanged with an extended time course from 3 h to 6 h (Figure 1b). Fe-bearing deposits were also characterised via XRD analysis and SEM. The generated deposit was weakly crystallised FeOOH aggregates (Figures 6a and 7a) at 1 h and was further converted into well-crystallised hematite nanoparticles with an average diameter of 100 nm after 3 h (Figures 6b and  7b), similar to that generated at 6 h (Figures 2c and 3c).   Fe-bearing deposits were also characterised via XRD analysis and SEM. The generated deposit was weakly crystallised FeOOH aggregates (Figures 6a and 7a) at 1 h and was further converted into well-crystallised hematite nanoparticles with an average diameter of 100 nm after 3 h (Figures 6b and  7b), similar to that generated at 6 h (Figures 2c and 3c).
Sustainability 2019, 11, x FOR PEER REVIEW 6 of 9 after 1 h, and steadily increased to 1.7%, 1.7% and 1.8% after 3 h (Figure 5b), respectively, and kept nearly unchanged with an extended time course from 3 h to 6 h (Figure 1b). Fe-bearing deposits were also characterised via XRD analysis and SEM. The generated deposit was weakly crystallised FeOOH aggregates (Figures 6a and 7a) at 1 h and was further converted into well-crystallised hematite nanoparticles with an average diameter of 100 nm after 3 h (Figures 6b and  7b), similar to that generated at 6 h (Figures 2c and 3c).

Hematite Precipitation Mechanism
Fe removal was related to an increased acid pH in the oxidation of fructose by nitrate. Nitrate concentration was reduced from 86.4 g/L to 22.8 g/L and 11.6 g/L after treatment for 0.25, 1 and 3 h, respectively ( Figure 8). TOC exhibited a similar decreasing trend to nitrate, and its concentration dropped to 0.67 g/L for 1 h and to 0.23 g/L for 3 h (Figure 8) due to the oxidisation of fructose by nitrate. As the redox reaction continued, acid pH increased steadily (Figure 8), promoting the hydrolysis of ferric Fe to form the initial product FeOOH. Fe removal was related to an increased acid pH in the oxidation of fructose by nitrate. Nitrate concentration was reduced from 86.4 g/L to 22.8 g/L and 11.6 g/L after treatment for 0.25, 1 and 3 h, respectively ( Figure 8). TOC exhibited a similar decreasing trend to nitrate, and its concentration dropped to 0.67 g/L for 1 h and to 0.23 g/L for 3 h (Figure 8) due to the oxidisation of fructose by nitrate. As the redox reaction continued, acid pH increased steadily (Figure 8), promoting the hydrolysis of ferric Fe to form the initial product FeOOH. The newly generated FeOOH was weakly crystallised and had numerous hydroxyl groups for rare-earth element coordination [19,27]. However, competition adsorption occurred between H + and the rare-earth elements on the newly formed FeOOH surface in the presence of sufficient H + , releasing rare-earth elements into the acid. Therefore, the hydroxyl group was regenerated on the newly formed FeOOH surface. Subsequently, the conjunction between two adjacent hydroxyl groups on each FeOOH occurred with the release of one water molecule to generate an irreversible Fe-O-Fe bond [27,28]. As the conjunction reaction continued, rare-earth elements were released and maintained a high concentration in the acid. In parallel, the weakly crystallised FeOOH block collapsed and then polymerised to generate fine hematite nanoparticles. These hematite nanoparticles, which were generated at a hydrothermal time of 6 h, contained 67.92% Fe, 0.78% Nd, 0.34% Pr and 0.14% Dy. The composition meets the grade for the active pharmaceutical ingredients of commercial hematite powder [29], demonstrating that the precipitated hematite nanoparticles were highly purified.

Conclusion
Fe impurity in the acid solution of scrap was successfully separated via a one-step hydrothermal route with fructose as the auxiliary reagent. Through this method, fructose was oxidised by nitrate with the involvement of H + in the acid, steadily increasing the acid's pH and enhancing the hydrolysis The newly generated FeOOH was weakly crystallised and had numerous hydroxyl groups for rare-earth element coordination [19,27]. However, competition adsorption occurred between H + and the rare-earth elements on the newly formed FeOOH surface in the presence of sufficient H + , releasing rare-earth elements into the acid. Therefore, the hydroxyl group was regenerated on the newly formed FeOOH surface. Subsequently, the conjunction between two adjacent hydroxyl groups on each FeOOH occurred with the release of one water molecule to generate an irreversible Fe-O-Fe bond [27,28]. As the conjunction reaction continued, rare-earth elements were released and maintained a high concentration in the acid. In parallel, the weakly crystallised FeOOH block collapsed and then polymerised to generate fine hematite nanoparticles. These hematite nanoparticles, which were generated at a hydrothermal time of 6 h, contained 67.92% Fe, 0.78% Nd, 0.34% Pr and 0.14% Dy. The composition meets the grade for the active pharmaceutical ingredients of commercial hematite powder [29], demonstrating that the precipitated hematite nanoparticles were highly purified.

Conclusions
Fe impurity in the acid solution of scrap was successfully separated via a one-step hydrothermal route with fructose as the auxiliary reagent. Through this method, fructose was oxidised by nitrate with the involvement of H + in the acid, steadily increasing the acid's pH and enhancing the hydrolysis of ferric Fe to hematite with FeOOH as the intermediate. Nearly 100% Fe was precipitated in the form of hematite nanoparticles with Fe content of 67.92% by adding fructose at a molar ratio of 0.2. The residual Fe in the leaching acid was less than 10 mg/L, and rare-earth element concentrations were 8.8, 2.1 and 0.4 g/L for Nd, Pr and Dy, respectively. Conversely, only 81.8% of the total Fe was removed as irregular hematite aggregates without fructose, and residual Fe was approximately 1.9 g/L.
This method exhibits two advantages in the recycling of rare-earth elements from scrap. Firstly, Fe is effectively separated from the leaching acid of scrap to generate high-purity hematite nanoparticles. Secondly, low residual Fe and high concentrations of rare-earth elements are produced in the residual acid. Therefore, highly purified rare-earth products can be economically recycled due to the low Fe content caused by repeatedly using an extraction reagent.

Conflicts of Interest:
The authors declare no conflict of interest.