Rare earth (RE) metals have several superior characters such as electric, magnetic, and fluorescence properties and are therefore indispensable as industrial materials throughout the world. Among the many applications of RE metals, the use of neodymium–iron–boron (Nd–Fe–B) magnets—the so-called neodymium permanent magnets—has been remarkably increasing recently. Because the properties of praseodymium (Pr) and Nd are similar, they are found in nature in the same ores and their separation is difficult. Thus, in some applications including permanent magnets, a Nd–Pr alloy called didymium (Di) is sometimes utilized. On the other hand, Nd–Fe–B magnets have the drawback of a relatively low Curie temperature of ~583 K. In order to maintain its superior coercive force even at high temperatures (above 473 K), where high-performance motors in EVs and HEVs operate, dysprosium (Dy) is necessary as an additive. One of the concerns pertaining to RE magnets is the uneven distribution of RE resources. While the present RE production capacity satisfies the demand and their price has lowered, there remains the potential for RE shortage in the future. On the basis of these supply circumstances, it can be concluded that the recycling or waste management of Nd–Fe–B magnet scraps is currently an urgent task. We have proposed a new separation and recovery process for RE metals from magnet scraps using molten salts and an alloy diaphragm as shown in the figure [1]. According to our previous studies, a specific RE element can be alloyed and de-alloyed rapidly with iron-group (IG) metals in molten salts by using electrochemical methods. In the proposed process, Nd, Dy, and Pr are separated because of differences in both the formation potentials and formation rates of the RE–IG alloys used as the diaphragm. In this study, selective formation of RE–Ni was investigated in molten NaCl–KCl–RECl3at 973 K. Based on our previous studies [2-4], the effective potential range for the separation of Dy from Pr and Nd is expected to be between 0.39 V and 0.48 V vs. Na+/Na. In this potential range, the thermodynamically stable phases are PrNi3, NdNi3, and DyNi2. Since the formation of DyNi2 is considerably faster than that of PrNi3 and NdNi3, it is expected that the formation of DyNi2 would proceed preferentially to those of NdNi3 and PrNi3. Then, the electrochemical formation of RE–Ni alloys were carried out at 0.42 V for 60 min in NaCl–KCl–0.50mol%RECl3. The thickness of the formed alloy layer was 4 µm for Nd-Ni, 2 µm for Pr-Ni, and 50 µm for Dy-Ni. These results suggest that Dy-Ni alloy is selectively formed in the melt containing both of NdCl3, DyCl3, and PrCl3. The sample was prepared by potentiostatic electrolysis at 0.42 V for 60 min in NaCl–KCl–0.50mol%NdCl3–0.50mol%DyCl3–0.50mol%PrCl3. The sample was analyzed by cross-sectional SEM and EDX. The formed alloy layer has a thickness around 60 µm with a composition of Nd:Dy:Pr = 1.3:29.8:0.7 at%. The high composition ratio, x Dy / (x Nd + x Pr), corresponding to 14.9 clearly indicates that Dy-Ni alloy can be selectively produced at this potential, and that RE-Ni alloy can be utilized as an alloy diaphragm with high separation ability for the proposed process. The results at different potentials are discussed in the presentation. References [1] T. Oishi, H. Konishi, T. Nohira, M. Tanaka, and T. Usui, Kagaku Kogaku Ronbunshu, 36, 299 (2010) (in Japanese). [2] K. Yasuda, S. Kobayashi, T. Nohira, and R. Hagiwara, Electrochim. Acta, 92, 349 (2013). [3] K. Yasuda, S. Kobayashi, T. Nohira, and R. Hagiwara, Electrochim. Acta, 106, 293 (2013). [4] K. Yasuda, K. Kondo, T. Nohira, and R. Hagiwara, J. Electroch em Soc., submitted.