Environmentally friendly separation of dysprosium and neodymium by fractional precipitation of coordination polymers

Abstract

Against a backdrop of increasing demand, the recovery of neodymium (Nd) and especially dysprosium (Dy) from manufacturing scraps and used magnets has necessitated the development of Nd/Dy separation technologies. To this end, we suggest a simple and environmentally friendly separation method by fractional precipitation of coordination polymers (CPs)—extended complexes of metal ions and organic ligands. With the di(2-ethylhexyl) phosphoric acid ligand functioning as a precipitant, Dy was exclusively precipitated as a CP due to its precipitation equilibrium that is considerably different from that of Nd.

---

Dysprosium (Dy)-doped neodymium (Nd) magnets are small, lightweight, and incredibly powerful. As the demand for Nd magnets increases, it has become more difficult to secure supplies of both Nd and Dy worldwide. The recovery of these elements from manufacturing scraps and used magnets has the potential to solve the problem of their limited supply.^1,2 Because of their similar chemical properties, one of the most difficult processes is Nd/Dy separation. The solvent extraction method that is conventionally used to separate lanthanide ions (Ln³⁺), including Nd³⁺ and Dy³⁺, requires large amounts of organic solvent and energy. Thus, a precipitation method can be a powerful technique when the solubilities of the resulting solids are low and the solubility products (K_sp) are sufficiently different for partitioning. For example, the selective precipitation of lanthanide sulfates from Ln³⁺–transition metal ion mixtures (e.g., Sm³⁺ from a Sm³⁺ and Co²⁺ mixture, or Nd³⁺ from a Nd³⁺ and Fe³⁺ mixture) was suggested to recover Ln from magnet scraps.³ In the case of Nd/Dy separation by the formation of inorganic salts such as phosphate,⁴ hydroxides,⁵ and oxalate,⁶ quite similar K_sp values between Nd and Dy do not offer an efficient separation parameter.

Our research^7–10 has focused on coordination polymers (CPs) formed with di(2-ethylhexyl) phosphoric acid (PO(OH)(OCH₂CHC₂H₅C₄H₉)₂, Hdehp), hereinafter described as [Ln(dehp)₃]. CPs are solid-state extended complexes formed with organic ligands and metal ions. We previously revealed that [Ln(dehp)₃] can be formed simply by mixing LnCl₃ and Hdehp in a binary ethanol–water mixture.⁷ Powder X-ray diffraction analysis indicates that [Ln(dehp)₃] has a monoclinic structure and its lattice parameters vary only slightly with a kind of Ln³⁺.¹⁰ As the stability of the complex between Hdehp and Ln³⁺ is governed by the charge density of the ions, Hdehp prefers to form a mononuclear complex with Dy³⁺ rather than Nd³⁺ in solvent extraction systems.¹¹ Thus, it was expected that CPs based on Hdehp would impart different precipitation properties in Nd and Dy as well. In this communication, we examine the difference in precipitation behaviors of [Nd(dehp)₃] and [Dy(dehp)₃] in a binary ethanol–water mixture and its potential to separate Nd and Dy by fractional precipitation.

The precipitation equilibrium of [Ln(dehp)₃] and K_sp is given by the following equations:

where a, γ, and c represent the activity, activity coefficient, and concentration of the species, respectively. The concentration of H⁺ in the solution (c(H⁺)) affects the precipitation reaction, because dehp⁻ is generated by the dissociation of Hdehp as follows:

In order to assess the difference in the formation of [Nd(dehp)₃] and [Dy(dehp)₃], their precipitation equilibria were investigated in various acidic solutions. Water or 0.06–3.0 M aq. HCl solution (1.6 mL), 1 M LnCl₃ (Ln = Nd or Dy) in 0.25 M aq. HCl solution (0.4 mL), ethanol (14 mL), and 0.3 M Nadehp in an 80 : 20 ethanol–water mixture (4 mL) were mixed in the order shown. The Nadehp solution was prepared by mixing 14.5 M aq. NaOH and ethanolic Hdehp solutions in equimolar amounts. We previously found that a lower concentration of water results in better separation (see ESI,†); thus, an 86 vol% ethanol concentration, obtained after mixing in aqueous LnCl₃ and HCl solutions, was used. The c(H⁺) varied with HCl addition. Whitish precipitates were obtained and filtered after ca. 20 h, washed several times with 80 : 20 vol% ethanol–water, and dried. The pH value of the filtrates was measured by a pH meter (Metrohm pH meter 744) equipped with an electrode for ethanol solutions (Metrohm EtOH-trode). According to the literature,¹² the true pH values should be larger than the pH measured by 0.04–0.3 due to the difference in standard states between aqueous and binary ethanol–water solutions, thus we consider the pH in this range. The filtrates and the precipitates were decomposed in a HNO₃ solution under microwave irradiation (see ESI† for more details). Ln and P concentrations in each phase were determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Shimadzu ICPE9000).

The Ln (= Nd or Dy) and P contents of the solids was consistent with the calculated values based on the chemical formulas of [Ln(dehp)₃], indicating that Ln precipitates as a coordination polymer from the binary ethanol–water solutions (Tables S2 and S3, ESI†). From the Ln³⁺ concentration in the solutions (c(Ln³⁺)), the precipitation yields (Y/%) were calculated according to the following equation,

The Y value for [Nd(dehp)₃] and [Dy(dehp)₃] (Fig. 1(A) and (B), respectively) increases with pH, and Dy shows a more gradual change than Nd: the Y values for Dy exceed 75% within the pH range 0.7–4.0, whereas the Y values for Nd vary from 0 to ∼100% over a similar pH range. It seems that such different precipitation equilibria between the two CPs would enable Nd/Dy separation by fractional precipitation.

Fig. 1 Y values (%) of [Nd(dehp)₃] (A) and [Dy(dehp)₃] (B) from NdCl₃ and DyCl₃ solutions, respectively, as a function of pH. The initial concentrations (cⁱ) of Ln³⁺ (Ln³⁺ = Nd³⁺ or Dy³⁺), Nadehp, and H⁺ are 0.02, 0.06, and 0.05–0.24 M, respectively.

The separation of Nd and Dy was investigated under various conditions in the same manner as described above: the initial concentrations of the species are listed in Table 1. Fig. 2(A) and (B) show the Y values for Nd and Dy from Nd–Dy mixtures as a function of pH (entries a–e in Table 1), and ligand concentration (entries d and f–h in Table 1). The horizontal axes represent the values measured after equilibrium had been attained. As expected, Hdehp preferentially forms precipitates with Dy rather than Nd under any conditions except at pH 3.7. As shown in Fig. 2(A), the Y value of Nd at a pH lower than 1.0 is small enough to be separated from Dy, while both Nd and Dy give relatively high Y values when the pH is higher than 1. Since the cube of c(dehp⁻) is involved in determining K_sp (eqn (2)), c(dehp⁻) is crucial to separation efficiency. Fig. 2(B), which compares the Y values at pH = ∼0.8 as a function of c^tot(P) = c(dehp⁻) + c(Hdehp), as measured by ICP-OES analysis, indicates that Dy is exclusively precipitated regardless of c^tot(P) under conditions in which cⁱ(Nadehp) ranges from 1–2 equivalents per cⁱ(Dy³⁺) (entries e–g in Table 1). In practice, the Nd³⁺ concentration is higher than Dy³⁺ in Nd magnets, i.e., in their leached solutions. The Y values at cⁱ(Nd³⁺)/cⁱ(Dy³⁺) = 5 (h) show no substantial differences from those at cⁱ(Nd³⁺)/cⁱ(Dy³⁺) = 1 (g), offering a useful example.

Table 1 cⁱ (Ln³⁺), cⁱ(Nadehp), and cⁱ(H⁺) (cf. Fig. 2)

	c^i(Nd^3+)/M	c^i(Dy^3+)/M	c^i(Nadehp)/M	c^i(H^+)/M
a	0.010	0.010	0.060	0.0050
b	0.010	0.010	0.060	0.020
c	0.010	0.010	0.060	0.040
d	0.010	0.010	0.060	0.20
e	0.010	0.010	0.060	0.40
f	0.010	0.010	0.045	0.20
g	0.010	0.010	0.030	0.020
h	0.050	0.010	0.030	0.020

---

Fig. 2 Y values (%) of [Nd(dehp)₃] and [Dy(dehp)₃] from a NdCl₃ and DyCl₃ mixture as a function of pH (A) and c^tot(P) (B). Both maximum and minimum K^a_sp values (Table S2, ESI†) were employed to calculate Y (dotted lines in (A)).

To evaluate these results, we consider the K_sp values, and compare the calculated and experimental precipitation data. On the assumption that the values of γ(Ln³⁺), γ(dehp⁻), and γ(Hdehp) are unity, the apparent solubility product, hereinafter described as K^a_sp, is given by;

From the c^tot(P) and c(Ln³⁺) values (shown in Fig. S2, ESI†), K^a_sp(Nd) = 2.9 × 10⁻¹² (±0.8 × 10⁻¹²) mol⁴ L⁻⁴ and K^a_sp(Dy) = 1.2 × 10⁻¹⁵ = (±0.8 × 10⁻¹⁵) mol⁴ L⁻⁴ (see ESI† for more details). Their ratio is ∼2400, meaning that their concentration ratio in an equilibrated mixture of Nd³⁺ and Dy³⁺ should be ∼2400. The ratio of K_sp for the phosphate salts, for instance, is only about 6.5; the K_sp values of NdPO₄ and DyPO₄ are 1.1 × 10⁻²⁶ mol² L⁻² and 7.1 × 10⁻²⁶ mol² L⁻², respectively.⁴ By comparison, it is obvious that such a large difference in the K^a_sps of [Nd(dehp)₃] and [Dy(dehp)₃] is adequate for fractional precipitation. According to eqn (5), K^a_sp, c^tot(P), and pH allow us to calculate c(Ln³⁺), and thus Y, in the Nd–Dy mixtures. The calculated Y values are shown in Fig. 2(A) (dotted lines). For Dy, there are small differences between the experimental and calculated values, whereas for Nd, the calculated Y values are much smaller than the experimental values. The c(Nd³⁺)/c(Dy³⁺) ratio at equilibrium is as much as 100, unlike the calculated value. Hdehp may incorporate a small amount of Nd³⁺ when it forms a coordination polymer with Dy³⁺, i.e. coprecipitation results in a smaller separation performance than the calculated value. Because the coprecipitation reaction is kinetically controlled, the separation efficiency can be enhanced by changing the mixing rate and the concentrations of the metal ion and ligand. We will discuss ways to achieve a higher performance in more detail in the near future.

The precipitation behaviors of [Nd(dehp)₃] and [Dy(dehp)₃] in a binary ethanol–water solution differ considerably from each other. This allows for a simple and environmentally friendly separation of Dy and Nd from their mixture by fractional precipitation. Comparisons between the calculated and experimental data suggest that optimization of the conditions will enhance the separation efficiency.

References

1. M. Tanaka, T. Oki, K. Koyama, H. Narita and T. Oishi, in Handbook on the Physics and Chemistry of Rare Earths, Elsevier, 2013, vol. 43, ch. 255, pp. 170–178.
2. Y. Baba, F. Kubota, N. Kamiya and M. Goto, Solvent Extr. Res. Dev., Jpn., 2011, 18, 193–198.
3. N. Sato, Y. Wei, M. Nanjo and M. Tokuda, Shigen to Sozai, 1997, 113, 1082–1086.
4. F. H. Firsching and S. N. Brune, J. Chem. Eng. Data, 1991, 36, 93–95.
5. D.-Y. Chyng, E.-H. Kim, E.-H. Lee and J.-H. Yao, J. Ind. Eng. Chem., 1998, 4, 277–284.
6. I. Diakonov, K. Ragnarsdottir and B. Tagirov, Chem. Geol., 1998, 151, 327–347.
7. Y. Tasaki-Handa, K. Ooi, M. Tanaka and A. Wakisaka, Anal. Sci., 2013, 29, 685–687.
8. Y. Tasaki-Handa, Y. Abe, K. Ooi, M. Tanaka and A. Wakisaka, J. Colloid Interface Sci., 2014, 413, 65–70.
9. Y. Tasaki-Handa, Y. Abe, K. Ooi, M. Tanaka and A. Wakisaka, Dalton Trans., 2014, 43, 1791–1796.
10. K. Ooi, Y. Tasaki-Handa, Y. Abe and A. Wakisaka, Dalton Trans., 2014, 43, 4807–4812.
11. J. Stary, Talanta, 1966, 13, 421–437.
12. H. Ohtaki, Jpn. Anal., 1973, 22, 1275–1281.

---

Footnote

† Electronic supplementary information (ESI) available: Changes in precipitation yield of Nd and Dy with ethanol concentration, changes in Ln and P content in precipitates with pH, estimation of apparent solubility product (K_sp), XAFS measurement, and acid dissociation constant (K_a) of Hdehp. See DOI: 10.1039/c4ra00257a

---