Extraction of Rare Earth Metal Oxide Inclusion Particles in Steel

Abstract

The conventional recovery technology is not economically viable due to the low rare metal concentration in industrial scraps. It is expected that iron- and steelmaking process can recover rare metals as a high-concentration ore substitute. For efficient enrichment of rare earth metal (REM) in this process, the establishment of the thermodynamic values related to REM deoxidation of molten iron is required. To obtain precise thermodynamic data, the concentration of elements solved in the metal sample should be determined separately from that in the inclusions, which is suspended in molten iron. Therefore, a stable extraction method for REM oxide in iron was investigated.

From the dissolution behavior of Sm₂O₃, Nd₂O₃ and Dy₂O₃ reagents in various eluents, 2 v/v% triethanolamine-1 w/v% tetramethylammonium chloride-methanol (2%TEA) is most suitable for the extraction of REM oxide. From the analysis using SEM-EDX and SEM-WDX, REM oxide in Fe-0.2 mass% REM alloy was identified as REM₂O₃ containing a small amount of MgO and FeO. From the relationships between the total O concentration and the insoluble O concentration calculated from insoluble REM concentration, the REM oxide inclusions in Fe-REM alloy could be extracted precisely by the electrolysis with 2%TEA.

1. Introduction

In recent years, soaring resource prices, rising resource nationalism, frequent political instability, increasing monopoly power of resource majors, and expanding demand in advanced industrial fields have led to securing rare metals (hereinafter abbreviated as RM). It has been positioned as a nationally important issue in the resource strategy of Japan. To stably supply RM with those risks, the Ministry of Economy, Trade and Industry stated i) securing overseas mineral resources, ii) recycling enhancement, iii) development of alternative materials, and iv) increasing stockpiling in the policy of ‘Japan’s Rare Metal Securing Strategy’.¹⁾ Among them, the recycling business is attracting attention as a new industry, and academia, industry and government are collaborating to develop a recovery system and extraction technology from materials.²⁾

In the RM recovery, to improve the production efficiency of the conventional wet and/or dry refining process, solvent extraction, adsorption, precipitation separation, halogenation volatilization, molten salt electrolysis, etc. have been developed.^3,4) Among these recovery methods, a lot of the chloride volatilization method has been investigated,^{5,6,7,8,9,10)} but the strong corrosiveness of Cl₂ gas to the reaction vessel has been a problem. In addition, the glass reduction-phase separation method¹¹⁾ and the bioprocess method^12,13,14,15) have been reported.

The conventional target of RM recovery is limited mainly to manufacturing process scraps contained in high RM concentration, and it cannot be applied to a wide variety of used small electronic devices containing low RM concentration from the viewpoint of cost.³⁾ Therefore, there are not enough reports on RM recovery from waste containing RM at low concentrations. As for the current state of industrial recycling, RM collection from manufacturing process waste, scrapped automobiles, and scrapped large electronic devices is partially carried out, but not from scrapped small electronic devices.³⁾ The reason for this is that the conventional recovery technology⁴⁾ is not economically viable due to the low RM concentration in small electronic devices. However, since a large amount of scrapped small electronic devices are abolished, a large amount of RM exists as a total amount. Therefore, it is an urgent need to develop a new recovery technology centered on efficient enrichment.

Inoue et al.¹⁶⁾ suggested that by incorporating trace amounts of V, Mn, Nb and P contained in iron ore into the by-products of steel production (steelmaking slag), it can be recovered as a high-concentration ore substitute. Their process consisted of the following four steps:

Step-1) in the smelting process, V, Mn, Nb and P are contained in the hot metal by reduction,

Step-2) in the steelmaking process, V, Mn, Nb and P are transferred to the slag melt by oxidation,

Step-3) in the cooling process of the slag melt, V, Mn, Nb and P are concentrated in the specific mineral phase that are precipitated in slag, and

Step-4) the slag is crushed and sorted into a specific mineral phase by magnetic separation, flotation and so on.

It was also expected that RM recovery can be achieved by charging the used advanced materials containing a small amount of RM and manufacturing process waste during the smelting process. This recovery technology is considered superior in that it does not require new equipment installation costs and process changes.

Regarding the concentration behavior of RM in the mineral phase in steel slag during Step-3, Kasai et al.¹⁷⁾ found that Nd, Sm and Dy, which are the rare earth metal element (hereinafter abbreviated as REM), were concentrated in the calcium phosphate (CP) phase, and Co was concentrated in the magnesiowüstite (MW) phase of the steelmaking slag. The mutual diffusion coefficient of Sm,¹⁸⁾ Nd,^19,20) and Dy¹⁷⁾ in CP and magnesiowüstite (MW) was measured, and the equilibrium hases in the REM oxide-CaO-P₂O₅ ternary phase diagram were clarified.^17,21) However, they discussed only in terms of the diffusion of REM into CP and MW phases. The consideration from the viewpoint of REM oxide activity is also important for elucidating the REM enrichment behavior in CP phase.

The REM oxide activity in a specific mineral phase can be derived from thermodynamic studies on the REM-O equilibrium in molten iron. However, when the specific gravity of the deoxidizing product (REM-O inclusion) is large, those particles are easily suspended in the molten metal. Therefore, the analyzed total concentration of elements in the metal sample is considered to contain the inclusion. Since Ohta and Suito²²⁾ and Inoue et al.²³⁾ reported that ZrO₂ inclusions in iron could be stably extracted by a constant-potential electrolytic method using a non-aqueous solvent-based electrolyte. Inoue et al.²⁴⁾ quantified ZrO₂ suspended as a primary inclusion in the molten iron and derived thermodynamic values for the Zr–O equilibrium by excluding the Zr and O concentrations in the inclusions from the analysed total Zr and total O concentrations in the molten iron.

Our final purpose is to derive thermodynamic values related to REM deoxidation of molten iron required for the construction of a REM recycling system utilizing the iron- and steelmaking processes, with reference to the experimental method by Inoue et al.²⁴⁾ As the first step, we investigated a stable extraction method for REM-O inclusions (Sm₂O₃, Nd₂O₃, Dy₂O₃) in iron, which is necessary to distinguish between soluble (solved in metal) and insoluble (contained in inclusions) concentrations in steel.

2. Experimental

2.1. Dissolution Loss of REM Oxide Reagent

2.1.1. Sample

Sm₂O₃ reagent (purity 99.9 mass%, manufactured by Fujifilm Wako Pure Chemical Industries, Ltd.), Nd₂O₃ reagent (purity 99.9 mass%, manufactured by Fujifilm Wako Pure Chemical Industries, Ltd.) and Dy₂O₃ reagent (purity 99.9 mass%, manufactured by Kanto Chemical Industry Co. Ltd.) was used after heating at 473 K for 1 h in a muffle furnace for drying.

2.1.2. Extract Preparation

As the acid extract, (1 + 1)HCl:(1 + 1)HNO₃ volume ratio = 3:1 (hereinafter abbreviated as Acid) was used. Five v/v% Br₂‒methanol (hereinafter abbreviated as Br₂‒MeOH) and 14 w/v% I₂‒methanol (hereinafter abbreviated as I₂‒MeOH) were used as the halogen solvent-based extract. Non-aqueous solvent-based electrolytes include 2 v/v% triethanolamine-1w/v% tetramethylammonium chloride-methanol (hereinafter abbreviated as 2%TEA) and 10 v/v% acetylacetone-1 w/v% tetramethyl-ammonium chloride-methanol (hereinafter abbreviated as 10%AA), 4 v/v% methyl salicylate-1 w/v% tetramethylammonium chloride-methanol (hereinafter abbreviated as 4%MS), 40 w/v% maleic anhydride-1 w/v% tetramethylammonium chloride-methanol (hereinafter abbreviated as 40%MA) was supplied. All reagents used were of special grade manufactured by Fujifilm Wako Pure Chemical Industries, Ltd.

2.1.3. Procedure

Five mg of each reagent and 100 mL of each extract were placed in a brown bottle with a stopper, and the gentle stirring was kept at room temperature for 0.5 to 24 hours. The eluate was dispensed with a 10 mL whole pipette at appropriate time intervals, and was suction-filtered using a mixed cellulose ester membrane filter (open pore diameter 0.1 μm) or a polytetrafluoroethylene (PTFE) membrane filter (opening pore diameter 0.1 μm). Ten mL of (1 + 1)HCl was added to the filtrate, and the solution was heated at 373 K to evaporate the organic solvent. Then, 5 mL of (1 + 1)HNO₃ was added and heated to dryness at 393 K to completely remove the organic solvent. Again, 5 mL of (1 + 1)HNO₃ was added and heated at 353 K gently to prepare a sample solution. Each REM concentration in solution was determined by inductively coupled plasma-emission spectrophotometry (ICP-AES). To confirm the change in sample size during the dissolution loss experiment, the particle size distribution of the REM oxide reagent after immersion in Acid, I₂-MeOH, and 10%AA was obtained by a particle size measurement system with a dynamic light scattering method using 0.2 w/v% sodium hexametaphosphate aqueous solution as a dispersion solvent. For comparison, the same operation was performed for the REM oxide reagent before immersion.

2.2. Extraction of Inclusions from Fe‒0.2 mass% REM Alloy

2.2.1. Sample

The Fe-0.2 mass% REM (Sm, Nd, Dy) alloy sample, in which 0.2 mass% was not the final concentration but the charged one, was obtained by the following method.

Forty grams of electrolytic pure iron plate (purity 99.99 mass%, manufactured by Toho Zinc Co., Ltd.) was charged in dense MgO Tammann crucible (ϕ28 × ϕ23 × 50 mm) and heated at 1873 K in Ar-1 vol% H₂ flow (300 cm³/min). The MgO crucible was placed inside a graphite cylinder (ϕ42 × ϕ34 × 80 mm) to remove O₂ in atmosphere. After melting iron, 2.2 g of Fe-2 mass% REM alloy, which was previously synthesized by melting 1 g of REM metal chips (purity of Sm, Nd and Dy was 99.9, 99.5 and 99.9 mass%, respectively, supplied by the Nilaco Co.) and 50 g of electrolytic pure iron plate in an arc melting furnace, was added as a pre-deoxidation agent. The melt was stirred with a dense ZrO₂-9 mol% MgO tube for 30 s. After 1 min, 2.2 g of Fe-2 mass% REM alloy was again added, and the melt was stirred for 30 s. When the deoxidizing agent (Fe-2 mass% REM alloy) was added in two portions rather than at once, the yield of REM dissolved in metal was improved. After holding for 2 to 60 minutes, the MgO crucible was taken out from the upper part of the Al₂O₃ reaction tube and immediately quenched in water.

2.2.2. Procedure

The obtained metal sample was cut in the vertical direction and embedded in a resin. The cross section was mirror-polished, and then immersed in Br₂-MeOH for 20 s at room temperature. The composition of the exposed inclusion particles was determined by the Scanning Electron Microscopy-Energy Dispersive X-ray Spectroscopy (SEM-EDX) and/or the Scanning Electron Microscopy-Wavelength Dispersive X-ray Spectroscopy (SEM-WDX) using standard samples of Sm₂O₃, Sm₂O₃-5 mass% MgO-5 mass% FeO, Nd₂O₃, Nd₂O₃-5 mass% MgO-5 mass% FeO, Dy₂O₃, and Dy₂O₃-5 mass% MgO-5 mass% FeO. Those standard samples were synthesized as follows: The REM₂O₃ reagent powder was pressed to cylindrical shape (ϕ5 × 5 mm) and sintered in Pt-10 mass% Rh crucible at 1873 K for 12 h under Ar flow (300 mL/min). The mixture of REM₂O₃, MgO and FeO reagent powders was also pressed to cylindrical shape (ϕ5 × 5 mm) and sintered in MgO crucible at 1873 K for 12 h under Ar-1% H₂ flow (300 mL/min). The reason for using MgO crucible in this sintering is that Pt-10 mass% Rh crucible is easily damaged by FeO and H₂ at high temperature. During sintering, a mixture of REM₂O₃-5 mass% MgO-5 mass% FeO was placed on the inner bottom of the MgO crucible to prevent the contact between the standard sample and MgO crucible.

The metal sample was vertically cut to a square rod (5 × 5 × 15 mm). The surface of the rod was polished with an iron file, and then the rod was subjected to total oxygen analysis using a combustion in an inert gas-infrared absorption method.

A metal plate (3 × 5 × 10 mm) was polished by a pen grinder, cleaned in methanol ultrasonically, dried, and cut into metal chips (< 2 mm) with plier. The REM oxide inclusion particles in the chips, were extracted using Acid or I₂-MeOH. In the inclusion extraction with Acid (100 mL), about 0.3 g of metal chips were placed in a conical beaker and heated to 373 K. A membrane filter made by mixed cellulose ester (open pore diameter = 0.1 μm) was used for suction filtration. For the inclusion extraction with I₂-MeOH (100 mL), a metal chips of about 0.3 g was placed in an Erlenmeyer flask with a Liebig condenser, and dissolved in an ultrasonic bath maintained at 313 to 323 K. The inclusion particles were suction-filtered with a PTFE membrane filter (open pore diameter = 0.1 μm) instead of a mixed cellulose ester filter, which was broken by the reaction with methanol.

Electrolytic extraction of inclusion particles with a non-aqueous solvent-based electrolyte (10%AA, 2%TEA) was carried out by the following method. The surface of metal plate (12 to 15 × 10 × 3 mm) was polished by a pen grinder. The plate was ultrasonically cleaned in methanol, dried, and weighed. A schematic diagram of the electrolytic extraction apparatus is shown in Fig. 1. The plate was fixed to platinum tweezers, immersed in an electrolytic cell containing 150 mL of 10%AA or 2%TEA, and then subjected to galvanostatic electrolysis at a current value of 50 mA/cm². The reason for choosing this current density will be described later. After reaching 1500 coulombs (8 to 9 h), the electrolyte was suction-filtered with a PTFE membrane filter (open pore diameter = 0.1 μm). Then, the metal plate was ultrasonically cleaned in methanol, and the liquid was also suction-filtered with the same membrane filter as before. Furthermore, the inside of the electrolytic cell was thoroughly rinsed with methanol, and this liquid was also suction-filtered. The metal plate was dried and weighed. The weight of the electrolyzed sample was determined from the difference between the sample weight before and after electrolysis, which was almost 0.3 g. The membrane filter containing residue was transferred to a platinum analytical crucible, and burned off with a Bunsen burner. The residue was dissolved by 10 mL of (1 + 1)HCl to prepare a sample solution, whose concentration corresponds to insoluble REM. The filtrate was heated at 373 K to evaporate the organic solvent after addition of 10 mL of (1 + 1)HCl. Then, 5 mL of (1 + 1)HNO₃ was added and heated at 393 K to completely decompose the organic solvent. Again, 5 mL of (1 + 1)HNO₃ was added and heated at 353 K gently to prepare a sample solution, whose concentration corresponds to soluble REM. The REM concentration in the obtained two solutions was determined using ICP–AES.

Fig. 1.

A schematic diagram of the electrolytic extraction apparatus. (Online version in color.)

A higher current density during extraction is preferred to dissolve the steel sample quickly and extract inclusions in a short time. However, when the density is too high, the temperature of the electrolyte will rise and the inclusions may dissolve. Preliminary experiments confirmed that current densities of 20 to 50 mA/cm² did not affect the chemically analytical value of insoluble REM amount.

3. Results and Discussion

3.1. Dissolution Loss of REM Oxide (Sm₂O₃, Nd₂O₃, Dy₂O₃)

3.1.1. Sm₂O₃

The dissolution of Sm₂O₃ reagent (5 mg) in the eluent of Acid, Br₂-MeOH, I₂-MeOH, 2%TEA, 10% AA, 4%MS, and 40%MA was measured. The results are shown in Fig. 2. The dissolution of Sm₂O₃ reagent in Acid and Br₂-MeOH is remarkable, and that in 40%MA, I₂-MeOH and 10%AA occurs slightly, but that in 4%MS and 2%TEA is negligible. It is expected that the differences in the dissolution of Sm₂O₃ reagents in various solutions are due to the differences in the oxidizing power of bromine and iodine, the stability of Sm halide generated by the reaction of Sm₂O₃ with iodine and bromine, the tendency of Sm complex ion formation in the reaction between chelating agent and Sm₂O₃, and the pH values of eluents. The pH value of 40%MA and 10%AA was about 2 and 5, respectively, while that of 2%TEA was about 10.

Fig. 2.

Variation of Sm concentration in various solutions due to the dissolution of Sm₂O₃ reagent.

Figure 3 shows the particle size distribution of the Sm₂O₃ reagent before and after the dissolution experiment, where the frequency on the vertical axis represents the ratio of the number of particles in each particle size range as a percentage. The particle size of the Sm₂O₃ reagent becomes smaller after the dissolution experiment. It should be noted that almost of all Sm₂O₃ reagents were dissolved in Acid, as shown in Fig. 2. The particle size is concentrated in the range of 1.5 to 2.0 μm for I₂–MeOH and of 2.0 to 2.5 μm for 10%AA. The degree of dissolution loss is Acid >>I₂–MeOH≥10%AA, which is the same trend as the dissolution result in Fig. 2. Among various eluents, the Sm₂O₃ reagent has a particularly low dissolution loss to 4%MS and 2%TEA, so it is concluded that the electrolysis method using those eluents is suitable for the extraction of Sm₂O₃ inclusions from metal sample.

Fig. 3.

Particle size distribution of Sm₂O₃ reagent before and after dissolution experiment.

3.1.2. Nd₂O₃

Figure 4 shows the results of determining the variation of Nd concentration dissolved from 5 mg of Nd₂O₃ reagent in Acid, Br₂-MeOH, I₂-MeOH, 2%TEA, 10%AA, 4%MS, and 40%MA. Shown in this figure, the dissolution of Nd₂O₃ reagent is remarkable in Acid and Br₂-MeOH, and occurs slightly in 40%MA, 4%MS, and 10%AA, but is negligible in I₂-MeOH and 2%TEA. The particle size distribution of the Nd₂O₃ reagent before and after the dissolution experiment is shown in Fig. 5. The main particle size of the Nd₂O₃ reagent after the dissolution experiment is 0.5 to 1.0 μm in Acid, 2.0 to 2.5 μm in I₂-MeOH, and 1.5 to 2.0 μm in 10%AA, while that before the dissolution experiment is 2.0 to 2.5 μm. Therefore, the degree of dissolution loss can be estimated as Acid >>10%AA≥I₂–MeOH, which is the same as the results in Fig. 4. Of the various extracts, the extraction of Nd₂O₃ inclusions is particularly suitable for the extraction with I₂-MeOH and the electrolytic extraction using 2%TEA.

Fig. 4.

Variation of Nd concentration in various solutions due to the dissolution of Nd₂O₃ reagent.

Fig. 5.

Particle size distribution of Nd₂O₃ reagent before and after dissolution experiment.

3.1.3. Dy₂O₃

The time dependence of Dy concentration dissolved from 5 mg of Dy₂O₃ reagent in Acid, Br₂-MeOH, I₂-MeOH, 2%TEA, 10%AA, 4%MS, and 40%MA is shown in Fig. 6. The dissolution of the Dy₂O₃ reagent proceeds significantly in Acid, and slightly in 40%MA, Br₂-MeOH, I₂-MeOH, and 10% AA. On the contrary, that dissolves negligibly in 4%MS and 2%TEA. The fact that rapid dissolution in Br₂-MeOH and 40% MA is not observed at the beginning of dissolution is different from the case of Sm₂O₃ (Fig. 2) and Nd₂O₃ (Fig. 4). Figure 7 shows the particle size distribution of the Dy₂O₃ reagent before and after the dissolution experiment. The particle size distribution after the dissolution in Acid could not be measured because almost all the reagents were dissolved. According to the results in Fig. 6, the Dy₂O₃ reagent was dissolved to Acid by about 85%. Compared with the particle size distribution of the Dy₂O₃ reagent before the dissolution experiment, it becomes slightly smaller after the dissolution experiment using I₂-MeOH and 10%AA. The degree of dissolution loss is Acid >> I₂–MeOH≈10%AA, which is similar to the result in Fig. 6. It can be said that the electrolytic extraction using 2%TEA and 4%MS is suitable for the extraction of Dy₂O₃ inclusions.

Fig. 6.

Variation of Dy concentration in various solutions due to the dissolution of Dy₂O₃ reagent.

Fig. 7.

Particle size distribution of Dy₂O₃ reagent before and after dissolution experiment.

The deference in dissolution loss of Sm₂O₃, Nd₂O₃ and Dy₂O₃ in various eluates seems to depend mainly on the generation ability of REM-halide and REM complex ion rather than the oxidizing power and pH value of the eluates. This point will be discussed in a separate paper.

3.2. Extraction of Inclusion Particles from Fe‒0.2 mass%REM Alloy

3.2.1. Exposure of Sm₂O₃ Inclusion Particles by Etching with Bromine Methanol

The metal sample with a mirror-polished cross section, which was obtained after 60 min holding at 1873 K following the method described in Section 2.2.1, was immersed in Br₂-MeOH solution for 20 s. Figure 8 shows the SEM image of the metal surface. Table 1 gives the quantitative results of each inclusion particle indicated in Fig. 8, which were analyzed by SEM-EDX using two standard samples of Sm₂O₃ and Sm₂O₃-5 mass% MgO-5 mass% FeO. In this table, the O_{as Sm-oxide} content was calculated from Eq. (1) by assuming the formation of Sm₂O₃–MgO–FeO compound.   

$$\begin{array}{l}({\mathrm{O}}_{\mathrm{a}\mathrm{s}\mathrm{\hspace{0.17em} }\mathrm{S}\mathrm{m}\text{-}\mathrm{o}\mathrm{x}\mathrm{i}\mathrm{d}\mathrm{e}}\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\%)=(\mathrm{O}\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\%)-(\mathrm{M}\mathrm{g}\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\%)\cdot M_{\mathrm{O}}/M_{\mathrm{M}\mathrm{g}}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{0.5em}-(\mathrm{F}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{a}\mathrm{s}\mathrm{s}\%)\cdot M_{\mathrm{O}}/M_{\mathrm{F}\mathrm{e}}\end{array}$$(1)

where M_i denotes the atomic weight of element i.

Fig. 8.

SEM images of metal surface after etching with Br₂-MeOH. (Online version in color.)

Table 1. Chemical compositions of each inclusion particle indicated in Fig. 8.

Point in Fig. 8	Analyzed value (mass%)					mSm/mO
	O	Sm	Mg	Fe	O as Sm-oxide
1	15.0	80.4	1.7	2.8	13.1	0.65
2	14.4	80.4	1.7	3.5	12.3	0.70
3	13.9	83.1	0.6	2.4	12.8	0.69
4	14.1	83.1	0	2.7	13.3	0.66
5	14.9	82.6	0.9	1.6	13.9	0.63
6	13.6	84.5	0.5	1.4	12.9	0.69

It is predicted from Table 1 that a large number of exposed fine particles are Sm₂O₃ inclusions because the analyzed Sm/O molar ratios (m_Sm/m_O) are close to two thirds. As listed in Table 2, the average and standard deviation of m_Sm/m_O ratios were 0.69 and 0.06, respectively. Since the suspension of many REM oxide inclusion particles was observed in metal sample as shown in Fig. 8, it is important to carry out the differential quantification of soluble and insoluble REM (REM oxide inclusions) for the discussion of the REM deoxidation equilibrium of molten iron.

Table 2. Average and deviation of m_REM/m_O ratio and Fe and Mg concentrations in inclusions.

REM	Number of measured particles	mREM/mO		(mass% Mg)		(mass% Fe)
		average	Standard dev.	average	Standard dev.	average	Standard dev.
Sm	53	0.69	0.06	1.1	0.5	2.1	0.7
Nd	50	0.62	0.07	1.3	0.5	1.9	0.6
Dy	50	0.64	0.06	1.8	0.6	1.8	0.6

3.2.2. Extraction and Quantification of REM Oxide Inclusions

Since the composition, shape, size, and chemical stability of actual inclusions are different from those of reagents, it was reported that the results of extraction using reagents cannot be applied to actual inclusions.^25,26,27) Therefore, to confirm the stable extraction of Sm oxide inclusions, Fe-0.2 mass% Sm alloy was prepared following the method described in Section 2.2.1. The Sm₂O₃ inclusions were extracted from the alloy using Acid, I₂-MeOH, and the galvanostatic electrolysis with 10%AA or 2%TEA. The 2%TEA is the most difficult to dissolve Sm₂O₃ reagent in Fig. 2. The results are listed in Table 3. The concentrations of both soluble Sm (solute in metal) and insoluble Sm (contained in inclusions) decrease with the holding time. However, the insoluble Sm did not decrease significantly after 30 min, and the deviation between the analyzed soluble and insoluble Sm concentrations repeated three or four times was also reduced. The reason why the insoluble Sm concentration does not decrease after 30 min is that the difference in specific gravity between molten iron (6.94 g/cm^{3 28)}) and Sm₂O₃ (8.35 g/cm^{3 29)}) is small and Sm₂O₃ inclusions might be less to aggregate in molten iron, so Sm₂O₃ inclusion particles suspended uniformly in molten iron and were difficult to float up. It was observed by SEM that a lot of Sm₂O₃ inclusion was trapped on the boundary between metal/crucible.

Table 3. Soluble and insoluble Sm concentrations in Fe-0.2 mass% Sm alloy held for 2 to 60 min at 1873 K.

Holding time at 1873 K (min)	[sol. Sm]/[insol. Sm] (mass ppm)
	Acid	I2-MeOH	10%AA	2%TEA
2	348/1.8 383/0.2 438/0.6	68/350 83/260 102/258	66/275 71/301 91/360	63/399 76/387 93/410
10	305/0.9 328/0.1 346/0.9	64/257 76/250 82/230	60/269 70/242 78/252	58/267 63/242 70/225
30	126/1.6 137/0.9 140/1.2	37/100 40/89 48/86	35/96 37/100 41/86	30/110 34/95 36/88
60	105/0.7 117/1.9 129/1.6	31/81 35/90 41/78	30/82 33/89 34/93	19/100 20/85 25/98 28/88

On the other hand, a decrement of the soluble Sm concentration after 30 min might be due to the slow establishment of deoxidation equilibrium and/or the continuation of reaction between Sm in iron melt and MgO crucible following Eq. (2).   

$$2[\mathrm{S}\mathrm{m}]+3\mathrm{M}\mathrm{g}\mathrm{O}(\mathrm{s})=\mathrm{S}{\mathrm{m}}_2{\mathrm{O}}_3(\mathrm{s})+3[\mathrm{M}\mathrm{g}]$$(2)

By the same method, the inclusions of Nd oxide and Dy oxide were extracted from the Fe-0.2 mass% Nd alloy and the Fe-0.2 mass% Dy alloy, respectively, which were prepared by the same procedure as the Fe-0.2 mass% Sm alloy. The results are given in Table 4. As listed in Table 2, the average and standard deviation of m_Nd/m_O ratios in the inclusions are 0.62 and 0.07, respectively, and those of m_Dy/m_O ratios are 0.64 and 0.06, respectively, after SEM-EDX and SEM-WDX analysis. Therefore, the inclusions in Fe-0.2 mass% Nd and Fe-0.2 mass% Dy alloys are presumed to be REM₂O₃. Since the specific gravities of Nd₂O₃ (7.24 g/cm^{3 30)}) or Dy₂O₃ (7.81 g/cm^{3 31)}) are close to that of molten iron (6.94 g/cm^{3 28)}), the insoluble concentration of Nd and Dy in Table 4 can be seen as almost constant after the holding time of 30 min, although the analyzed values are varied. Similar to soluble Sm in Table 3, the soluble Nd and Dy concentration after 30 min are found to be decreased due to the slow achievement of deoxidation equilibrium and/or the reaction between REM in iron melt and MgO crucible. The adhesion of Nd₂O₃ and Dy₂O₃ inclusions to the boundary between metal/crucible was also observed by SEM as in the case of Sm deoxidation.

Table 4. Soluble and insoluble REM concentrations in Fe-0.2 mass% REM alloy held for 30 and 60 min at 1873 K.

REM	Holding time at 1873 K (min)	[sol. REM]/[insol. REM] (mass ppm)
		Acid	I2-MeOH	10%AA	2%TEA
Nd	30	100/6.2 108/4.3 118/5.8	40/60 46/70 50/78	37/88 41/81 51/70	36/79 41/83 43/75
	60	81/3.1 84/3.9 90/2.1	17/65 18/60 20/71	15/70 16/61 18/60	13/72 13/63 14/67 15/77
Dy	30	136/1.6 139/0.9 143/1.2	36/101 38/105 42/96	27/111 32/106 36/98	25/119 31/124 34/115
	60	113/3.7 120/1.3 128/1.3	15/108 18/106 18/100	8/113 10/99 16/106	5/113 6/119 8/99 8/108

The longer the holding time at 1873 K, the smaller the deviation of soluble and insoluble REM concentration can be expected.

Figure 9 shows the results of comparing the insoluble REM and total REM concentrations in each Fe-0.2 mass% REM alloy obtained by extraction methods, which were listed in Tables 3 and 4. Although the data variation range is large, the total REM concentration is almost the same regardless of the extraction method, while the insoluble REM concentration is the highest by the galvanostatic electrolysis method using 2%TEA. The slight difference in insoluble REM concentration in Fig. 9 is presumed to be due to the dissolution loss of the actual inclusion particles in I₂-MeOH, 10%AA and 2%TEA during extraction. It can be concluded from this figure that the electrolysis using 2%TEA is preferable for the extraction of REM oxide in iron, but the I₂-MeOH extraction method or the electrolysis using 10%AA can be used for qualitative extraction.

Fig. 9.

The concentrations of insoluble and total REM in Fe-0.2 mass% REM alloy, which was held for 60 min at 1873 K, after various extraction methods.

3.2.3. Soluble O Concentration in Metal

The determination of the soluble O concentration in iron is required to obtain the thermodynamic values related to the REM deoxidation equilibrium of molten iron. The soluble O concentration in metal was estimated by calculating the insoluble O concentration in the REM_xO_y inclusions from the analyzed insoluble REM concentration and the average insoluble Mg and Fe concentrations determined by SEM-EDX using Eq. (3) and subtracting the insoluble O concentration from the total O concentration, as indicated in Eq. (4).   

$$\begin{array}{l}[\mathrm{i}\mathrm{n}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{O}]=[\mathrm{i}\mathrm{n}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{R}\mathrm{E}\mathrm{M}]\cdot y{M}_{\mathrm{O}}/(x{M}_{\mathrm{R}\mathrm{E}\mathrm{M}})\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\left[\mathrm{i}\mathrm{n}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{M}\mathrm{g}\right]}_{\mathrm{a}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{g}\mathrm{e}}·M_{\mathrm{O}}/M_{\mathrm{M}\mathrm{g}}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+{\left[\mathrm{i}\mathrm{n}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{F}\mathrm{e}\right]}_{\mathrm{a}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{g}\mathrm{e}}\cdot M_{\mathrm{O}}/M_{\mathrm{F}\mathrm{e}}\end{array}$$(3)

  

$$[\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{O}]=[\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{\hspace{0.17em} }\mathrm{O}]-[\mathrm{i}\mathrm{n}\mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{O}]$$(4)

The total O concentration in metal was performed by the combustion in inert gas-infrared absorption method. The relationships between the total O concentration in Fe-0.2 mass% REM alloy and the holding time at 1873 K are shown in Fig. 10. The values for soluble and insoluble REM concentrations in this figure is obtained by the electrolysis using 2%TEA shown in Tables 3 and 4. From this figure, in the case of Fe-0.2 mass% Sm, both the total O and insoluble Sm concentrations decrease at the initial stage, but after 30 minutes, they decreased slightly. Since the total oxygen concentration decreases despite the decrease in the soluble REM concentration after 30 minutes, it is inferred that the REM deoxidation equilibrium is reached slowly.

Fig. 10.

Variation of total O concentration in Fe-0.2 mass% REM alloys with holding time at 1873 K.

Figure 11 shows the total O concentration in the Fe-0.2 mass% REM alloy sample held for 60 min at 1873 K plotted against the insoluble O concentration calculated by Eq. (3), where [insoluble REM] is the value obtained by the electrolysis with 2%TEA listed in Tables 3 and 4. When the experimental point is above the 1:1 line (broken line) in the figure, measurable level of soluble oxygen is considered to be absent in the metal. However, due to the large variation range of measured values, it is difficult to accurately determine the soluble oxygen concentration. As mentioned in Section 3.2.2, a longer holding time is inevitable to obtain a value with a small deviation. At least, it can be considered that the soluble oxygen concentration is extremely low.

Fig. 11.

Relationships between total O and insoluble O concentration in Fe-0.2 mass% REM alloys held for 60 min at 1873 K.

To obtain such a remarkably low oxygen concentration and oxygen activity in metal, it is effective to utilize an oxygen sensor with a ZrO₂ solid electrolyte. Inoue et al.²⁴⁾ measured O-activity in molten Fe–Zr alloy using a ZrO₂-9 mol% MgO oxygen sensor and a Mullite oxygen sensor. When obtaining the thermodynamic values concerning Zr–O equilibrium, the soluble Zr and soluble O concentrations were estimated by excluding insoluble Zr and insoluble O concentrations from the total Zr and total O concentrations, respectively. Han et al.³²⁾ measured the O activity using a ZrO₂ oxygen sensor and found the relationship between the total Nd concentration and the O activity in the low Nd concentration range of less than 1 mass ppm. It should be noted that the ZrO₂ solid electrolyte immersed in the molten metal with high REM concentration is reduced by REM and Zr solves in the metal, which strongly affects the REM-O equilibrium. On the other hand, Ono et al.³³⁾ investigated the Nd deoxidation equilibrium of molten iron using an Nd₂O₃ crucible, but they discussed using not the soluble Nd and soluble O but the total Nd and total O concentrations. The REM deoxidation equilibrium of Fe melt by determining the oxygen activity in the Fe-REM melt with an oxygen sensor using a REM oxide crucible will be reported in a separate paper.

4. Conclusion

To establish the thermodynamic values related to REM deoxidation of molten iron precisely, it is necessary to determine the concentration of solved elements in the metal sample separately from that in the inclusions, which are suspended in molten iron. Therefore, a stable extraction method for rare earth metal oxide in iron was investigated and the following conclusions were obtained.

(1) From the dissolution behavior of Sm₂O₃, Nd₂O₃ and Dy₂O₃ reagents in HCl-HNO₃-H₂O mixture, halogen-methanol solvents (I₂-MeOH and Br₂-MeOH), and non-aqueous solvent-based electrolytes (2%TEA, 4%MS, 10%AA and 40%MA), 2% TEA is most suitable for the extraction of Sm₂O₃, Nd₂O₃ and Dy₂O₃.

(2) From the analysis using SEM-EDX and SEM-WDX, REM oxide in Fe-0.2 mass% REM alloy was identified as REM₂O₃ containing a small amount of MgO and FeO.

(3) From the relationships between the total O concentration and the insoluble O concentration calculated from insoluble REM concentration, the REM oxide inclusions in Fe-REM alloy could be extracted precisely by the electrolysis with 2%TEA.

References

• 1)   Mining  Subgroup, Advisory Committee for Energy and Resource Policy in Ministry of Economy, Trade and Industry: Rare Metal Securing Strategy, (2009), http://www.gichokai.gr.jp/archive/member/chosabu/zengishiryo/21nen/038/038-1.pdf, (accessed 2017-04-10).
• 2)  Tohoku Leading Project Extraction Technology: High Efficiency Extraction Technology of Rare Elements, Tohoku Leading Project of Material Technology, http://tohoku-timt.net/rare-elements/, (accessed 2017-04-10).
• 3)  Ministry of the Environment: The 4th Ministry of the Environment Policy Meeting, (Recent Trends in Recycling such as Recovery of Rare Metals 2009), http://www.env.go.jp/council/seisaku_kaigi/epc004/mat04, (accessed 2017-04-10).
• 4)   Y.  Hamada and  N.  Sato: Shigen-to-Sozai, 107 (1991), 109. https://doi.org/10.2473/shigentosozai.107.109
• 5)   H.  Matsuura,  T.  Hamano and  F.  Tsukihashi: ISIJ Int., 46 (2006), 1113. https://doi.org/10.2355/isijinternational.46.1113
• 6)   R.  Nonaka,  Y.  Yamaguchi,  K.  Sugawara and  K.  Sato: Kagaku Kogaku Ronbunshu, 33 (2007), 495. https://doi.org/10.1252/kakoronbunshu.33.495
• 7)   R.  Nonaka,  H.  Funayama and  K.  Sugawara: Kagaku Kogaku Ronbunshu, 35 (2009), 403. https://doi.org/10.1252/kakoronbunshu.35.403
• 8)   Y.  Mochizuki,  H.  Mori and  K.  Sugawara: Kagaku Kogaku Ronbunshu, 37 (2011), 70. https://doi.org/10.1252/kakoronbunshu.37.70
• 9)   A.  Hosoi,  K.  Hiruta,  Y.  Takasaki and  A.  Shibayama: J. Jpn. Inst. Met., 76 (2012), 155. https://doi.org/10.2320/jinstmet.76.155
• 10)   H.  Koizumi,  S.  Iino,  Y.  Tatsuichi and  A.  Terashima: Annual Report of Tokyo Metropolitan Research Institute for Environmental Protection 2016, https://www.tokyokankyo.jp/kankyoken_contents/cmsup/pdf/report2016_11.pdf, (accessed 2017-04-10).
• 11)   H.  Kadoki,  T.  Narioka,  T.  Igura and  M.  Arita: The 24th Annual Conf. Japan Society of Material Cycles and Waste Management, Japan Society of Material Cycles and Waste Management, Tokyo, (2013), Poster 2, B1-11.
• 12)   J.  Chang,  Y.  Tani,  H.  Naitou,  N.  Miyata,  F.  Tojo and  H.  Seyama: Chem. Geol., 383 (2014), 155. https://doi.org/10.1016/j.chemgeo.2014.06.002
• 13)   T.  Nomura,  M.  Itouga and  H.  Sakakibara: Kagaku to Seibutsu, 52 (2014), 121. https://doi.org/10.1271/kagakutoseibutsu.52.121
• 14)  SIT Research Institute of Shibaura Institute of Technology: Report of Rare Metal Bio Research Center, (2016), http://www.shibaurait.ac.jp/research/sit_research_laboratories/result/mfpbut000000fxic-att/mfpbut000000fxny.pdf, (accessed 2018-03-30).
• 15)   Y.  Konishi,  M.  Yokota,  M.  Tochihara and  T.  Ogi: J. Jpn. Soc. Powder Powder Metall., 64 (2017), 222. https://doi.org/10.2497/jjspm.64.222
• 16)   R.  Inoue,  S.  Ueda and  T.  Ariyama: High Temp. Mater. Process., 29 (2010), 447. https://doi.org/10.1515/HTMP.2010.29.5-6.447
• 17)   N.  Kasai,  Y.  Takasaki,  A.  Shibayama and  R.  Inoue: Proc. 6th Int. Cong. on the Science and Technology of Steelmaking (ICS2015), (Beijing), China Machine Press, Beijing, (2015), 970.
• 18)   H.  Hikosaka: Master thesis, Akita University, (2019).
• 19)   Y.  Oikawa: Senior thesis, Akita University, (2018).
• 20)   K.  Arai: Senior thesis, Akita University, (2019).
• 21)   T.  Inoue: Senior thesis, Akita University, (2016).
• 22)   H.  Ohta and  H.  Suito: ISIJ Int., 46 (2006), 14. https://doi.org/10.2355/isijinternational.46.14
• 23)   R.  Inoue,  R.  Kimura,  S.  Ueda and  H.  Suito: ISIJ Int., 53 (2013), 1906. https://doi.org/10.2355/isijinternational.53.1906
• 24)   R.  Inoue,  T.  Ariyama and  H.  Suito: ISIJ Int., 48 (2008), 1175. https://doi.org/10.2355/isijinternational.48.1175
• 25)   K.  Kawamura,  S.  Watanabe and  T.  Uchida: Tetsu-to-Hagané, 57 (1971), 94 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.57.1_94
• 26)   S.  Wakamatsu: Tetsu-to-Hagané, 57 (1971), 656 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.57.3_656
• 27)   T.  Ishii and  M.  Ihida: Tetsu-to-Hagané, 60 (1974), 1957 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.60.13_1957
• 28)   L. D.  Lucas: Compt. Rend., 250 (1960), 1850.
• 29)   Chemical  Book: Samarium(III) oxide, 3N5, https://www.chemicalbook.com/ChemicalProductProperty_JP_CB5743142.htm, (accessed 2018-04-10).
• 30)   Chemical  Book: Neodymium(III) oxide, 3N5, https://www.chemicalbook.com/ChemicalProductProperty_JP_CB1437511.htm, (accessed 2018-04-10).
• 31)   Chemical  Book: Dysprosium(III) oxide, https://www.chemicalbook.com/ChemicalProductProperty_JP_CB6434146.htm, (accessed 2018-04-10).
• 32)   Q.  Han,  X.  Feng,  S.  Liu,  H.  Niu and  Z.  Tang: Metall. Trans. B, 21B (1990), 295. https://doi.org/10.1007/BF02664197
• 33)   H.  Ono,  Y.  Tachiiri,  K.  Yamaguchi and  T.  Usui: ISIJ Int., 49 (2009), 1656. https://doi.org/10.2355/isijinternational.49.1656