ISIJ International, Vol. 56 (2016), No. 12, pp. 2224–2227 ISIJ International, Vol. 56 (2016), No. 12
Application of Portable Total-Reflection X-Ray Fluorescence Spectrometer to Analysis of Dysprosium in Neodymium-IronBoron Magnet Susumu IMASHUKU,1)* Jun TAKAHASHI,2) Shinsuke KUNIMURA2) and Kazuaki WAGATSUMA1) 1) Institute for Materials Research, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai, 980-8577 Japan. 2) Department of Industrial Chemistry, Faculty of Engineering, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku-ku, Tokyo, 162-8601 Japan. (Received on June 10, 2016; accepted on August 25, 2016)
Dysprosium (Dy) is expected to be recovered from end-of-life neodymium-iron-boron (NdFeB) magnets in the near future because of its worldwide shortage. Therefore, the rapid on-site elemental analysis of dysprosium in end-of-life NdFeB magnets is required. Here, we report a method of measuring the dysprosium composition in a NdFeB magnet on-site using a portable total-reflection X-ray fluorescence (TXRF) spectrometer. This method leads to drastic reduction of dysprosium loss caused by dissolving a NdFeB magnet because the portable TXRF spectrometer requires a small sample volume for measurement and its detection limit is as low as the ppb level. A NdFeB magnet was dissolved into hydrochloride acid and then iron in the solution was extracted using 4-methyl-2-pentanone (methyl isobutyl ketone: MIBK). Yttrium oxide (Y2O3) and a diluted standard solution of rubidium (Rb) were added to the solution as internal standards. We measured the X-ray intensities of dried residue of the solution using a portable TXRF spectrometer. Dy Lα line was clearly detected in the solution, whereas it overlapped with Fe Kα line in the solution before the MIBK extraction process. The dysprosium composition in the NdFeB magnet was determined from the measured intensities of Dy Lα, Y Kα, and Rb Kα, the relative sensitivities of dysprosium and yttrium to rubidium for the portable TXRF spectrometer, and the weights of the dissolved NdFeB magnet and yttrium oxide. The calculated dysprosium composition was in good agreement with that obtained by conventional ICP-AES. KEY WORDS: total-reflection X-ray fluorescence; neodymium-iron-boron magnet; dysprosium; portable analyzer; methyl isobutyl ketone; on-site analysis.
1.
sium in end-of-life NdFeB magnets is required. Furthermore, on-site analysis of dysprosium in end-of-life NdFeB magnets will contribute to an enhancement of efficiency in the recovery and recycling of dysprosium because on-site analysis can reduce time to transfer samples for the measurement from a stock site of end-of-life NdFeB magnets to a room where an analyzer is installed. Handheld X-ray fluorescence (HH-XRF) analyzers is one of the prospective candidates for on-site analysis of dysprosium in a NdFeB magnet. However, it is difficult to perform the dysprosium analysis of a NdFeB magnet in the form of solid using HHXRF because an energy-dispersive X-ray (EDX) is installed in HH-XRF: Kα line of iron (6.40 keV) overlaps Lα line of dysprosium (6.50 keV) and Kβ line of iron (7.06 keV) also overlaps Lβ line of dysprosium (7.25 keV). Wavelengthdispersive X-ray (WDX) spectrometer can separately detect the Dy Lα line and the Fe Kα line, but the detection limits of dysprosium in a NdFeB magnet become higher owing to the spectral interference of high-intensity Fe K lines.10) Even for conventional stationary XRF equipped with WDX, it is difficult to detect dysprosium in a NdFeB magnet with a concentration of dysprosium below 0.1 mass% in the form of solid. Thus, when we carry out on-site analysis
Introduction
Neodymium-iron-boron (NdFeB) magnets exhibit the highest energy product among the commercialized permanent magnets since they were developed.1) NdFeB magnets have been used in various products such as motors in hybrid and electric vehicles, electric generators in wind turbines, generators for magnetic resonance imaging, spindles for computer hard disk drives, audio speakers, and mobile phones. NdFeB magnet used in high-temperature applications, such as motors in electric vehicles and direct-drive wind turbines,2,3) have dysprosium (Dy) content of up to 10 mass% because dysprosium increases the intrinsic coercivity of NdFeB magnets at higher temperatures.4) Dysprosium is categorized as one of the most critical elements in the short and long term owing to the increasing demand for electric vehicles.3,5–9) Therefore, the recovery and recycling of dysprosium from end-of-life NdFeB magnets have attracted increasing attention. As the first step of the recovery and recycling of dysprosium, the elemental analysis of dyspro* Corresponding author: E-mail:
[email protected] DOI: http://dx.doi.org/10.2355/isijinternational.ISIJINT-2016-349
© 2016 ISIJ
2224
ISIJ International, Vol. 56 (2016), No. 12
of dysprosium in a NdFeB magnet with a low dysprosium concentration, such as 1 mass% and below, using portable analyzers, it is necessary to remove irons from a NdFeB magnet. In the present study, we measured the concentration of dysprosium in a solution containing a dissolved NdFeB magnet after removing iron from the solution using 4-methyl-2-pentanone (methyl isobutyl ketone: MIBK). MIBK can extract more than 99% of ferric ions (Fe3 + ) in hydrochloric acid (HCl) with a concentration of more than 5.5 M without extracting rare-earth elements.11,12) We also confirmed that more than 99% of iron in 6 M of hydrochloric acid solution containing a dissolved NdFeB magnet was extracted by MIBK without extracting the rare-earth elements in the solution such as neodymium, dysprosium, praseodymium (Pr), and terbium (Tb).13) We performed the elemental analysis of dysprosium in a NdFeB magnet using the portable total-reflection X-ray fluorescence (TXRF) spectrometer developed by one of the authors, Kunimura, and Kawai14–20) for on-site analysis because the TXRF spectrometer requires a small sample volume for measurement (approximately 10 μL), while HH-XRF and portable XRF equipped with WDX require several milli-liter. In addition, detection limit of the portable TXRF are approximately three orders of magnitude lower than HH-XRF and portable XRF equipped with WDX. Thus, we can drastically reduce dysprosium loss owing to the dissolution of a NdFeB magnet, dissolve the NdFeB magnet in a short time, and reduce waste liquid generated by the pretreatment by using the portable TXRF spectrometer. 2.
Fig. 1.
Schematic view of a portable TXRF spectrometer.
a micropipette on an optical flat whose surface was coated with diamond-like carbon.18) Intensities of Dy Lα, Y Kα, and Rb Kα lines in the dried residue of the 111-times diluted solution were measured using the portable TXRF spectrometer as shown in Fig. 1. The X-ray tube with a tungsten target (50 kV Magnum, Moxtek) was operated at 25 kV and 200 μA during the measurements. The goniometer was tilted 0.04° to the horizontal. We also measured the concentrations of iron and rareearth elements in the solutions containing the dissolved NdFeB magnet before and after MIBK extraction by inductively coupled plasma atomic emission spectroscopy (ICP–AES) (TJA Solutions, Iris Duo). When we analyzed the solution before MIBK extraction, the solution was diluted 100 times with ultrapure water and hydrochloric acid before measuring the concentrations of dysprosium, yttrium, and praseodymium or 1 000 times with ultrapure water and hydrochloric acid before measuring the concentrations of iron and neodymium so that the concentrations of hydrochloric acid in the solutions were 1 M. Before ICPAES analysis of the solution after MIBK extraction, nitric acid (HNO3) and perchloric acid (HClO4) were added to the solution, which was then heated to decompose the residual MIBK in the solution. Then, 1 M of hydrochloric acid was added to increase the volume to approximately that before heating. After that, the solution was diluted 10 times with 1 M of hydrochloric acid before measuring the concentration of iron or 100 times before measuring the concentrations of dysprosium and yttrium with 1 M of hydrochloric acid. Details of the measurement procedure were described in our previous papers.10,13)
Experimental Method
The sample analyzed in the present study was a commercially available NdFeB magnet. The NdFeB magnet (0.76762 g) was dissolved into 50 mL of 12 M of hydrochloric acid. After dissolving the NdFeB magnet, yttrium oxide (Y2O3) (0.04116 g) was dissolved into the solution as an internal standard because yttrium does not contain in the NdFeB magnet and is not extracted by MIBK.11) Then, 5 mL of hydrogen peroxide (H2O2) was added to the solution to change ferrous irons (Fe2 + ) in the solution to ferric ions (Fe3 + ). After that, the solution was diluted with ultrapure water until the total volume reached 100 mL, resulting in a hydrochloric acid concentration of 6 M. The diluted solution was mixed with MIBK using a separating funnel and shaken for at least 1 min. The solution in the aqueous layer was diluted 100 times with ultrapure water and hydrochloric acid so that the concentration of hydrochloric acid in the solution was 1 M. A diluted standard solution of rubidium (Rb) with a concentration of 10 mg L − 1 was added to the 100-times diluted solution to yield a rubidium concentration of 1 mg L − 1 ; thus the solution used for TXRF analysis was diluted 111 times from the solution comprising the aqueous layer. Rubidium was an internal standard for calculating the concentrations of dysprosium and yttrium in the solution after MIBK extraction. We chose rubidium as the internal standard because K lines of rubidium do not overlap any of the lines of yttrium and elements originating from the NdFeB magnet and components of the portable TXRF spectrometer. Ten μL of the 111-times diluted solution was added dropwise with
3.
Results and Discussion
Table 1 shows the concentrations of iron and rare-earth elements in the solutions containing the dissolved NdFeB magnet before and after MIBK extraction obtained by ICPAES analysis. We measured the concentrations of iron, dysprosium, and yttrium in the solution after MIBK extraction to calculate the extraction yields of iron and dysprosium for MIBK. Thus, we did not measure the concentrations of neodymium and praseodymium in the solution after MIBK extraction. The compositions of the NdFeB magnet are also listed in Table 1. The compositions were calculated from the concentrations in the solution before MIBK extraction measured by ICP-AES analysis, as shown in Table 1, and the weight of the dissolved NdFeB magnet. Rare-earth elements in the NdFeB magnet with compositions of more than 1 mass% are listed in Table 1. We calculated the extraction yields of iron and dysprosium in the NdFeB magnet for 2225
© 2016 ISIJ
ISIJ International, Vol. 56 (2016), No. 12 Table 1.
Concentrations of iron and rare-earth elements in solutions containing dissolved NdFeB magnet before and after MIBK extraction and calculated compositions of NdFeB magnet. Measurements were performed three times by ICP-AES. Errors are given as standard deviations. Fe
Nd
Dy
Pr
Y
4 747 ± 21
1 504 ± 12
522.9 ± 2.9
315.0 ± 1.6
326.3 ± 0.5
After extraction
31.90 ± 0.7
–*
508.0 ± 0.6
–
323.4 ± 1.3
Composition of NdFeB magnet (mass%)
61.84 ± 0.27
19.59 ± 0.16
6.81 ± 0.04
4.10 ± 0.02
–
Concentration in solution (mg L −1)
Before extraction
*Concentration or composition was not measured.
MIBK from the concentrations of iron, dysprosium, and yttrium in the solutions before and after MIBK extraction. In the calculations, we assumed that yttrium is not extracted by MIBK. The extraction yields of iron and dysprosium were determined to be 99.3% and 2.0%, respectively, indicating that iron in the solution was effectively extracted by MIBK without extracting dysprosium. Figure 2(a) show EDX spectra of the solution containing the dissolved NdFeB magnet before MIBK extraction, and Fig. 2(b) show the 111-times diluted solution containing the dissolved NdFeB magnet after MIBK extraction. Both EDX spectra were obtained using the portable TXRF spectrometer. Dy Lα line overlapped Fe Kα line when we analyzed the solution before MIBK extraction. On the other hand, Dy Lα line was detected and Fe Kα line disappeared when we analyzed the solution after MIBK extraction. We concluded that Fe Kα line disappeared as a result of the MIBK extraction process because the energy of the peak was shifted from 6.42 keV to 6.52 keV and the extraction yield of iron for MIBK was 99.3%. Nd L, La Lα, and Co Kα lines originated from the NdFeB magnet because they were not detected when we analyzed a mixed standard solution of dysprosium, yttrium, and rubidium using the portable TXRF spectrometer as shown in Fig. 2(c). The existence of cobalt and lanthanum in the NdFeB magnet was also confirmed by ICP-AES analysis. The concentrations of cobalt and lanthanum were approximately 1.4 and 0.02 mass%, respectively. Ni K lines came from the electroplating layer of the NdFeB magnet and the EDX detector of the portable TXRF spectrometer. Si Kα, Ar Kα, Ca Kα, and W L lines were attributed to the optical flat, air, stabilizer for polyvinyl chloride covering cables, and the target of the X-ray tube, respectively. Cl Kα line was due to the chloride ions in hydrochloric acid. No praseodymium was detected in the solution using the portable TXRF spectrometer because Pr Lα line (5.03 keV) overlaps Nd Lα line (5.23 keV). We next performed quantitative analysis of dysprosium in the NdFeB magnet using the portable TXRF spectrometer. We calculated the concentrations of dysprosium and yttrium using the relative sensitivities of dysprosium and yttrium to rubidium for the portable TXRF spectrometer. The relative sensitivities were determined by measuring the net area intensities of Dy Lα, Y Kα, and Rb Kα lines in a solution containing dysprosium, yttrium, and rubidium, each with a concentration of 1 mg L − 1. The obtained EDX spectrum is shown in Fig. 2(c). The relative sensitivities of dysprosium and yttrium to rubidium were 2.10 and 0.836, respectively. Secondary X-ray fluorescence and the absorption of X-ray fluorescence by the sample can be ignored for TXRF measurement,21,22) indicating that the net area intensity of X-ray © 2016 ISIJ
Fig. 2. (a) EDX spectrum of solution containing the dissolved NdFeB magnet before MIBK extraction obtained using the portable TXRF spectrometer. An enlarged EDX spectrum near Fe Kα peak is shown in the upper right. The measured solution was diluted 100 times without adding a standard solution of rubidium. (b) EDX spectrum of solution containing the dissolved NdFeB magnet after MIBK extraction (second measurement). An enlargement of the EDX spectrum near Dy Lα peak is shown in the upper right. The measured solution was diluted 111 times and contained 1 mg L −1 of rubidium. (c) EDX spectrum of standard solution of dysprosium, yttrium, and rubidium, each with a concentration of 1 mg L −1.
fluorescence for each element is zero when its concentration is zero. Thus, we determined the relative sensitivities from the X-ray intensities of one sample. Table 2 shows the net 2226
ISIJ International, Vol. 56 (2016), No. 12 Table 2. Net area intensities of Dy Lα , Y Kα , and Rb Kα lines in 111-times diluted solution containing dissolved NdFeB magnet after MIBK extraction. Measurements were performed three times. First
Second
Third
Dy Lα (counts)
22 547
41 392
32 538
Y Kα (counts)
4 036
5 324
4 774
Rb Kα (counts)
7 792
9 735
8 547
4.
We have demonstrated the application of a portable TXRF spectrometer as an analyzer to determine the dysprosium composition in a NdFeB magnet by measuring the EDX spectrum of the dried residue of a solution containing a dissolved NdFeB magnet and yttrium oxide as an internal standard. Prior to the measurement, iron in the solution was extracted by MIBK and then rubidium was added to the solution as an internal standard. The dysprosium composition in the NdFeB magnet obtained using the portable TXRF spectrometer was in reasonable agreement with that obtained using ICP-AES. This analytical method can detect approximately 1 mass% of dysprosium in a NdFeB magnet. By increasing the number of MIBK extraction process, we would be able to detect less than 1 mass% of dysprosium in a NdFeB magnet. The portable TXRF spectrometer will lead to reduction of dysprosium loss associated with dissolving a NdFeB magnet due to its lower detection limit and a small sample volume required for measurement. Thus, this analytical method of determining the dysprosium composition in a NdFeB magnet using the portable TXRF spectrometer will contribute to the on-site analysis of dysprosium in end-of-life NdFeB magnets for the recovery and recycling of dysprosium.
Table 3. Concentrations of dysprosium and yttrium in solution containing dissolved NdFeB magnet after MIBK extraction. These concentrations were calculated from the net area intensities listed in Table 2 and the relative sensitivities of dysprosium and yttrium to rubidium for the portable TXRF spectrometer. Errors are given as standard deviations. Elements −1
Dy (mg L ) −1
Y (mg L )
First
Second
Third
Average
300
410
360
360 ± 70
260
240
240
250 ± 40
Conclusions
area intensities of Dy Lα, Y Kα, and Rb Kα lines in the 111-times diluted solution containing the dissolved NdFeB magnet after MIBK extraction. The measurement was repeated three times, changing the residue of the solution after each measurement. We calculated the concentrations of dysprosium and yttrium in the solution containing the dissolved NdFeB magnet after MIBK extraction from the net area intensities of Dy Lα, Y Kα, and Rb Kα lines and the relative sensitivities of dysprosium and yttrium to rubidium. The results are summarized in Table 3. The concentrations of dysprosium and yttrium in the solution containing the dissolved NdFeB magnet were decreased by MIBK extraction because the solution was diluted with residual water inside the separating funnel and a small amount of MIBK dissolved into the aqueous layer. Thus, we calculated the dysprosium composition in the NdFeB magnet from the concentrations of dysprosium and yttrium measured by the TXRF spectrometer, the weights of the dissolved NdFeB magnet, and the yttrium oxide added as an internal standard. The dysprosium composition was determined to be 6.1 ± 1.2 mass%. This value agrees with the dysprosium composition in the NdFeB magnet calculated from the ICPAES measurement within the experimental error. Hence, we can obtain the dysprosium composition in a NdFeB magnet with reasonable accuracy by utilizing the portable TXRF spectrometer and the MIBK extraction process. Considering a detection limit of dysprosium calculated from Dy Lα peak intensity22) and the extraction yields of iron in MIBK, it is roughly estimated that this method can detect approximately 1 mass% of dysprosium in NdFeB magnet. We could detect dysprosium in NdFeB magnet with a dysprosium concentration below 1 mass% by repeating MIBK extraction process because irons in MIBK dissolved in the solution for the measurement was removed.
Acknowledgements Financial support for the present study was provided by the 23rd ISIJ Research Promotion Grant and JSPS KAKENHI Grant Number 26709056. REFERENCES 1) M. Sagawa, S. Fujimura, N. Togawa, H. Yamamoto and Y. Matsuura: J. Appl. Phys., 55 (1984), 2083. 2) K. Binnemans, P. T. Jones, B. Blanpain, T. Van Gerven, Y. Yang, A. Walton and M. Buchert: J. Clean. Prod, 51 (2013), 1. 3) S. Hoenderdaal, L. T. Espinoza, F. Marscheider-Weidemann and W. Graus: Energy, 49 (2013), 344. 4) M. Sagawa, S. Fujimura, H. Yamamoto, Y. Matsuura and K. Hiraga: IEEE Trans. Magn., 20 (1984), 1584. 5) U. S. Department of Energy: Critical Materials Strategy, Washington, DC, (2011). 6) European Commission: Critical Raw Materials for the EU, Brussels, (2010). 7) R. L. Moss, E. Tzimas, H. Kara, P. Willis and J. Kooroshy: Energy Policy, 55 (2013), 556. 8) E. Alonso, A. M. Sherman, T. J. Wallington, M. P. Everson, F. R. Field, R. Roth and R. E. Kirchain: Environ. Sci. Technol., 46 (2012), 3406. 9) J. H. Rademaker, R. Kleijn and Y. Yang: Environ. Sci. Technol., 47 (2013), 10129. 10) S. Imashuku, J. Kawai and K. Wagatsuma: Microsc. Microanal., 22 (2016), 82. 11) T. Kakita: Bunseki Kagaku, 16 (1967), 624. 12) T. Uchida, E. Tsuzuki, Y. Takahashi and U. Inoue: Bunseki Kagaku, 53 (2004), 429. 13) S. Imashuku, K. Wagatsuma and J. Kawai: Surf. Interface Anal., 48 (2016), 1153. 14) S. Kunimura and J. Kawai: Anal. Chem., 79 (2007), 2593. 15) S. Kunimura and J. Kawai: Analyst, 135 (2010), 1909. 16) S. Kunimura and J. Kawai: Adv. X-ray. Anal., 53 (2010), 180. 17) S. Kunimura and J. Kawai: Bunseki Kagaku, 58 (2009), 1041. 18) S. Kunimura, D. P. Tee and J. Kawai: Tetsu-to-Hagané, 97 (2011), 81. 19) S. Kunimura, S. Hatakeyama, N. Sasaki, T. Yamamoto and J. Kawai: AIP Conf. Proc., 1221 (2010), 24. 20) S. Kunimura and H. Ohmori: Analyst, 137 (2012), 312. 21) R. Klockenkämper and A. von Bohlen: Total-Reflection X-Ray Fluorescence Analysis and Related Methods, 2nd ed., Wiley, New Jersey, (2015), 242. 22) S. Terada: Keiko X-sen Bunseki no Jissai, ed. by I. Nakai, Asakura Shoten, Tokyo, (2005).
2227
© 2016 ISIJ
