Recovery of Fine Rare Earth Minerals from Simulated Tin Tailings by Carrier Magnetic Separation: Selective Heterogeneous Agglomeration with Coarse Magnetite Particles

Abstract

The demand for rare earth elements (REEs) is continuously increasing due to the important roles they play in low-carbon and green energy technologies. Unfortunately, the global REE reserves are limited and concentrated in only a few countries, so the reprocessing of alternative resources like tailings is of critical importance. This study investigated carrier magnetic separation using coarse magnetite particles as a carrier to recover finely ground monazite from tailings. The monazite and carrier surfaces were modified by sodium oleate (NaOL) to improve the hydrophobic interactions between them. The results of zeta potential and contact angle measurements implied the selective adsorption of NaOL onto the surfaces of the monazite and magnetite particles. Although their hydrophobicity increased, heterogenous agglomeration between them was not substantial. To improve heterogenous agglomeration, emulsified kerosene was utilized as a bridging liquid, resulting in more extensive attachment of fine monazite particles onto the surfaces of carrier particles and a dramatic improvement in monazite recovery by magnetic separation—from 0% (without carrier) to 70% (with carrier). A rougher–scavenger–cleaner carrier magnetic separation can produce REE concentrates with a total rare earth oxide (TREO) recovery of 80% and a grade of 9%, increased from 3.4%, which can be further increased to 23.2% after separating REEs and the carrier.

1. Introduction

Mining and mineral processing are essential for supplying the raw materials needed to manufacture a wide range of products used in modern society. However, these activities inevitably generate large quantities of mine wastes, such as tailings, which can significantly degrade surrounding environments [1]. For instance, tailings generated from sulfide ore deposits often contain high contents of non-valuable sulfide minerals, such as pyrite (FeS₂) and arsenopyrite (FeAsS). When exposed to oxygen and water, these minerals undergo oxidation reactions that generate acid mine drainage (AMD), a highly acidic effluent (pH < 3) enriched with heavy metals and toxic elements that can severely contaminate surrounding water bodies and ecosystems [2].

In addition to AMD, some mine tailings—such as those from tin mining in Indonesia—exhibit significant concentrations of radioactive elements (e.g., thorium (Th) and uranium (U)), which pose considerable risks to ecosystems and human health [3,4,5]. Nurtjahya et al. [4], for instance, reported that soils near former tin mining areas in the Bangka-Belitung Islands were found to emit a surface radiation dose of 0.394–0.412 µSv/h at 1 m above ground, significantly higher than undisturbed soils (0.096 µSv/h). Furthermore, agricultural products grown on these lands, such as vegetables, fruits, and tubers, have been shown to contain elevated levels of radionuclides, including radium (Ra)-226, Th-232, and potassium (K)-40, increasing the potential cancer risk for residents in these areas [3].

One of the primary reasons for the high radioactivity of Indonesian tin tailings is the inadequate recovery of radioactive minerals like monazite ((Ce,La,Nd,Th)PO₄). Beyond their environmental impact, these tailings hold potential as secondary resources for rare earth elements (REEs), which are critical for producing high-performance magnets, batteries, and other advanced technologies [6,7,8]. As the global demand for REEs continues to rise—driven by their role in low-carbon and green energy technologies like electric vehicles (EVs), wind turbines and solar panels—their strategic importance has grown considerably [6,8]. In 2024, the global production of rare earth oxides (REOs)—the majority of which were produced by China (69.2%), followed by the U.S. (11.5%), Myanmar (7.9%), and Australia (3.3%)—rose to 390,000 metric tons from 5000 metric tons before the 1950s [6,9,10]. Zglinnicki et al. [11] conducted mineralogical and chemical examinations of tailings stored on Bangka Island, Indonesia, and reported that these tailings contain significant amounts of REEs, ranging from 0.6 to 15.5%. This concentration is comparable to well-known REE deposits such as Bayan Obo deposit in China (5%–6%), Mountain Pass in the U.S. (7%), and Mountain Weld in Australia (8.8%) [10,12]. Thus, recovering REEs from tin tailings offers dual benefits: mitigating environmental hazards and unlocking a valuable resource.

Despite this potential, the effective and sustainable recovery of REEs from tin tailings remains a challenge. Tin ores are typically processed by gravity separation (e.g., jig and shaking table) to reject light gangue minerals like quartz (SiO₂), followed by electrostatic and magnetic separation to separate cassiterite (SnO₂), ilmenite (FeTiO₃), zircon (ZrSiO₄), and monazite. The primary reason for the loss of REEs to tailings is either the generation of ultrafine monazite particles due to overgrinding or the insufficient liberation of monazite from gangue minerals. To recover unliberated monazite particles from tin tailings, fine grinding is therefore essential. Yet, magnetic separation—one of the most used techniques for REE processing—becomes increasingly ineffective as particle size decreases because the magnetic force acting on a particle is proportional to the cube of its diameter. This has led to the need for alternative methods to improve the recovery of finely ground REE minerals.

To address this issue, floc magnetic separation (FMS)—which involves the flocculation of weakly magnetic mineral fines followed by magnetic separation—has been proposed [13,14,15]. In this technique, weakly magnetic mineral fines are first mixed with flocculants to aggregate them into larger clusters called flocs, increasing the magnetic separation efficiency due to the enhanced magnetic force acting on the flocs. Meng et al. [13], for example, investigated FMS using octyl hydroxamic acid (OHA) to recover ultrafine wolframite (WO₃). Under optimal conditions, ultrafine wolframite particles (D₅₀: 4.44 µm) became flocs with a D₅₀ of 9.08 µm, and, as a result, magnetic separation efficiency increased from ~55% to ~80%. Similarly, Roy [14] investigated the recovery of fine iron oxides (hematite (Fe₂O₃) and goethite (FeO(OH))) from low-grade iron ore using FMS with sodium oleate. It was confirmed that FMS achieved a significantly higher recovery rate (79.6%) compared to conventional wet high-intensity magnetic separation (WHIMS), which yielded only 37.4%.

Although FMS has shown promising results, it is not suitable for low-grade materials such as tailings due to its reliance on high particle concentrations for effective floc formation [16]. In other words, an insufficient concentration of target mineral fines in the system hinders the formation of adequately sized flocs, leading to low recovery during the magnetic separation process. In such cases, introducing coarse magnetic particles as carriers in the flocculation process, in which valuable fines adhere to the surface of carriers, followed by magnetic separation, a method known as carrier magnetic separation (CMS), offers a viable solution, as the efficiency of CMS is unaffected by the concentration of fine particles. The utilization of carriers is known to be effective in recovering fine mineral particles even from tailings [16,17].

Therefore, this study investigates the applicability of CMS using coarse magnetite (Fe₃O₄) particles as magnetic carriers for recovering fine monazite particles from tin tailings. In order to induce the selective attachment of fine monazite particles to the carrier surface (i.e., heterogeneous agglomeration) via hydrophobic interactions, the surfaces of monazite and magnetite particles were modified using sodium oleate (CH₃(CH₂)₇CH=CH(CH₂)₇COONa) to render their surfaces hydrophobic (Figure 1). Furthermore, the effects of the surface modifications of monazite (fine particles) and magnetite (carrier) on the efficiency of CMS were evaluated to determine its potential for recovering monazite from fine-grained tailings.

2. Materials and Methods

2.1. Materials

Tailing samples used in this study were taken from Jebus subdistrict, North Bangka, Indonesia. A one-kilogram sample was initially screened using sieves with various mesh sizes (#35, #48, #65, #150, and #200), and then, each size fraction was analyzed by X-ray fluorescence spectroscopy (XRF, EDXL300, Rigaku Corporation, Tokyo, Japan). As shown in

Table 1. Elemental compositions of tin tailings for various size fractions.

Elements	Size Fractions (µm)
	+425	−425 +300	−300 +212	−212 +106	−106 +75	−75
TREO	15.7	19.0	20.3	12.8	6.2	3.7
Fe2O3	10.4	14.7	13.9	11.9	13.4	12.5
TiO2	19.0	28.1	32.6	31.4	10.1	11.4
ZrO2	9.4	3.8	6.4	15.8	35.0	32.4
SiO2	19.3	16.3	13.6	17.8	17.2	17.6
Al2O3	11.1	4.7	2.4	4.3	6.9	3.8
Others	15.2	13.3	10.9	6.0	11.3	18.6

, tin tailings contain substantial amounts of REEs, with contents ranging from 3.7% to 20.3%. The size fractions above 75 µm contain ~15% total rare earth oxides (TREOs), which can be simply upgraded as REE concentrates (40%–68% TREO) [18], and thus, this study focused on the processing of the −75 µm size fraction. Figure 2 shows the X-ray diffraction (XRD, MultiFlex, Rigaku Corporation, Tokyo, Japan) pattern of the −75 µm size fraction. As can be seen, the major REE-bearing minerals contained in the sample were monazite and xenotime (YPO₄), while the non-REE minerals that were found were ilmenite (FeTiO₃), quartz (SiO₂), and zircon (ZrSiO₄).

The high radioactivity of the tailing sample (~0.1 µSv/h) poses safety concerns, making its laboratory processing challenging. Instead of using the tailing directly, this study used simulated tailings based on the mineralogical composition of the real Indonesian tin tailings (i.e., monazite-(Ce) (Novo Horizonte, Bahia, Brazil), zircon (Sakorn Minerals Co., Ltd., Prachuap Khiri Khan, Thailand), and quartz (Wako Pure Chemical Industries, Ltd., Osaka, Japan)). Xenotime was excluded from the simulated tailings due to its similar behavior toward NaOL as that of monazite, which was used to represent rare earth minerals [19]. In addition, ilmenite was also excluded from the simulated tailings for the following reasons: (i) its high magnetic susceptibility allows for easy separation by magnetic separation, and (ii) its removal enables a clearer understanding of monazite behavior during magnetic separation. Mineral samples were roughly crushed with an agate mortar and pestle and then ground with a disc mill (RS 100, Retsch Inc., Haan, Germany). The particle size distributions of the ground products were determined using laser diffraction (Microtrac^® MT3300SX, Nikkiso Co., Ltd., Tokyo, Japan), and the median particle sizes (D₅₀) measured for quartz, zircon, and monazite were 9.5, 10.3, and 3.1 µm, respectively. The ground quartz, zircon, and monazite were mixed in the same proportion as that measured in the real Indonesian tin tailing sample. It is important to note that the use of pure minerals in the simulated tailings assumes the complete liberation of each mineral phase. This study specifically targets two types of rare earth mineral occurrences that are difficult to recover using conventional beneficiation processes: (i) liberated ultrafine rare earth minerals generated due to overgrinding and (ii) fine-grained rare earth minerals locked within gangue phases. To simulate these conditions, the mineral samples were ground to a D₅₀ of 3–10 µm, under the assumption that complete liberation of rare earth minerals can be achieved at this particle size range, even in actual tailings.

Magnetite obtained from Digkilaan, Iligan City, Mindanao Island, Philippines, was used as a carrier for monazite. It was first purified by magnetic separation using a 0.9 T hand-held magnet (Eriez Magnetics Co., Ltd., Chiba, Japan). The magnetic product was screened to collect the size fraction of 106–180 µm (D₅₀ = 120 µm), which was used as the carrier. All the mineral samples used in this study were analyzed by X-ray fluorescence spectroscopy (XRF, EDXL300, Rigaku Corporation, Tokyo, Japan), and the results are summarized in

Table 2. The chemical compositions of monazite, zircon, and magnetite used in this study.

Monazite		Zircon		Quartz		Magnetite
Elements	Contents (%)	Elements	Contents (%)	Elements	Contents (%)	Elements	Contents (%)
La	12.8	Zr	70.3	Si	99.2	Fe	79.1
Ce	33.2	Si	22.0	Al	0.3	Si	8.3
Nd	13.9	Al	4.2	P	0.4	Al	4.3
Other REEs	12.7	Mg	2.5	Others	0.1	Ti	4.5
P	8.9	Others	1.0			Others	3.8
Th	16.2
Others	2.2

.

2.2. Surface Modifications for the Attachment of Fine Monazite to Coarse Magnetite

Prior to magnetic separation, the simulated tailings were treated with sodium oleate (NaOL, C₁₇H₃₃COONa, Wako Pure Chemical Industries, Ltd., Osaka, Japan) to render monazite and magnetite surfaces hydrophobic, with the aim of attaching fine monazite particles to coarse magnetite particles via hydrophobic interactions. For this, 30 g of simulated tailings and 200 mL of solution containing 1 g/L NaOL were put into a 500 mL baffled flask and mixed at 1000 rpm for 15 min.

To further improve the hydrophobic interactions between NaOL-treated monazite and magnetite particles, the addition of emulsified kerosene (EK) stabilized by either NaOL or sodium dodecyl sulfate (SDS, CH₃(CH₂)₁₁OSO₃Na, Wako Pure Chemical Industries, Ltd., Osaka, Japan) was investigated. Emulsified kerosene was prepared according to the method of Hornn et al. [20]: (i) mixing 5 mL kerosene (Wako Pure Chemical Industries, Ltd., Osaka, Japan), 20 mL deionized (DI) water, and 50 mg NaOL or SDS, and (ii) emulsifying the mixture using an ultrasonic homogenizer (Ultra-Turrax, IKA, Königswinter, Germany) for 60 s. This emulsification was carried out immediately before agglomeration. After 15 min of treatment with NaOL, the simulated tailings were treated for another 15 min with 1.5 L/t EK.

2.3. Surface Characterization and Morphological Analysis

To check the adsorption of NaOL onto mineral surfaces, changes in the zeta potentials and contact angles of mineral samples with and without NaOL were evaluated. For the zeta potential measurements, 1 mg of pulverized mineral sample was dispersed in 10 mL deionized (DI) water containing 0 or 50 mg/L NaOL. The mixture was ultrasonicated for 1 min and stirred for 15 min. Afterward, the suspension was left to stand for 15 min, and the supernatant was measured for the zeta potential over a pH range of 2–12 using a Zetasizer Nano-series with an MPT-2 multi-purpose titration system (Malvern Corporation, Malvern, UK). The solution pH was adjusted with dilute hydrochloric acid (HCl) and sodium hydroxide (NaOH).

For the contact angle measurements, monazite and magnetite specimens were cut using a diamond cutter to obtain small cuboid crystals (~5 mm (w) × 5 mm (d) × 10 mm (h)), fixed inside plastic holders using Technovit^® non-conductive resin (Heraeus Kulzer GmbH, Hanau, Germany), and polished on a polishing machine (SAPHIR 250 M1, ATM GmbH, Mammelzen, Germany) using silicon carbide papers (P320, P600, and P1200) and diamond suspensions (3 and 1 µm). The monazite and magnetite samples were then individually immersed into DI water with and without 1 g/L NaOL for 15 min, and their contact angles were measured using a high-magnification digital microscope (VHX-1000, Keyence Corporation, Osaka, Japan) with a built-in image analysis software. Each measurement was repeated 3 times at different spots on the mineral surface to ascertain that the differences observed were statistically significant.

To confirm whether fine monazite particle attachment to coarse magnetite particles occurred after NaOL treatment, particle size distributions of the monazite/magnetite mixture (3:7 w/w) before and after surface modifications were measured by laser diffraction and scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS, JSM-IT200, JEOL Ltd., Tokyo, Japan).

2.4. Magnetic Separation Using a Hand-Held Magnet

Magnetic separation experiments were first conducted using a 1.6T hand-held magnet (Eriez Magnetics Co., Ltd., Chiba, Japan) to check the effects of surface modification and heterogeneous agglomeration on the recovery of fine monazite. The simulated tailings (i.e., monazite, quartz, and zircon) and carrier (i.e., magnetite) were treated in 200 mL DI water containing no additives, 1 g/L NaOL, or 1 g/L NaOL and 1.5 L/t EK at 1000 rpm for 15 min. Afterward, the suspension was transferred into a plastic container, and then, magnetic minerals (e.g., monazite and magnetite) were separated from non-magnetic gangue minerals (e.g., quartz and zircon) using a 1.6 T hand-held magnet. The magnetic and non-magnetic fractions were dried in a vacuum drying oven at 40 °C for 24 h, weighed, and analyzed by XRF. The recoveries of TREO in the magnetic fraction (R_M(TREO)) and gangue minerals in the non-magnetic fraction (R_NM(G)) and the separation efficiency of TREO from gangue minerals (η) were calculated as follows:

$${\mathrm{R}}_{\mathrm{M}\left(\mathrm{T}\mathrm{R}\mathrm{E}\mathrm{O}\right)}={\displaystyle \frac{{\mathrm{M}}_{\mathrm{M}}\times {\mathrm{W}}_{\mathrm{M}\left(\mathrm{T}\mathrm{R}\mathrm{E}\mathrm{O}\right)}}{{\mathrm{M}}_{\mathrm{F}}\times {\mathrm{W}}_{\mathrm{F}\left(\mathrm{T}\mathrm{R}\mathrm{E}\mathrm{O}\right)}}}\times 100$$

(1)

$${\mathrm{R}}_{\mathrm{N}\mathrm{M}\left(\mathrm{G}\right)}={\displaystyle \frac{{\mathrm{M}}_{\mathrm{N}\mathrm{M}}\times {\mathrm{W}}_{\mathrm{N}\mathrm{M}\left(\mathrm{G}\right)}}{{\mathrm{M}}_{\mathrm{F}}\times {\mathrm{W}}_{\mathrm{F}\left(\mathrm{G}\right)}}}\times 100$$

(2)

$$\eta \left(\mathrm{\%}\right)={\mathrm{R}}_{\mathrm{M}\left(\mathrm{T}\mathrm{R}\mathrm{E}\mathrm{O}\right)}-\left(100-{\mathrm{R}}_{\mathrm{N}\mathrm{M}\left(\mathrm{G}\right)}\right)$$

(3)

where M denotes the mass (g), W is the weight present (%) of the component, and the subscripts F, M, NM, and G refer to feed, magnetic fraction, non-magnetic fraction, and gangue minerals, respectively. To ascertain that the differences observed were statistically significant, some experiments were duplicated.

2.5. Wet High-Intensity Magnetic Separation

The simulated tin tailings and carrier were treated with NaOL and EK under the optimal conditions confirmed by magnetic separation using a hand-held magnet and then subject to wet high-intensity magnetic separation (WHIMS, L-4, Eriez Co., Ltd., Chiba, Japan). A total of 16 pieces of fine, expanded metal matrix (opening size: 2 mm × 5 mm) were placed in the separation cell to increase the number of high field gradient sites where magnetic particles could be collected. A magnetic induction of 1.0 T was applied to the separation cell, and then, the pretreated sample in the form of suspension (i.e., 30 g/200 mL) was fed into the WHIMS. After the slurry had completely passed through the WHIMS, the collected magnetic particles were washed with 1 L of DI water to reduce non-magnetic particles that were recovered due to entrapment. To maximize the recovery of fine monazite, the non-magnetic fraction was mixed with carrier particles, treated with NaOL and EK, and then subject to the WHIMS. The magnetic and non-magnetic fractions were dried at 40 °C for 24 h, weighed, and analyzed by XRF.

2.6. Theoretical Calculation of Particle Aggregation via EDLVO Theory

The interaction energies between fine monazite and coarse magnetite particles were theoretically calculated using the extended Derjaguin–Landau–Verwey–Overbeek (EDLVO) theory [21,22]. The total interaction energy (V_T) is given by

$${\mathrm{V}}_T={\mathrm{V}}_W+{\mathrm{V}}_E+{\mathrm{V}}_H$$

(4)

where V_W is the van der Waals interaction energy, V_E is the electrostatic double layer interaction energy, and V_H is the hydrophobic interaction energy. The sign V_T dictates the nature of the interaction between particles: when V_T > 0, repulsive forces prevail, hindering particle attachment, whereas when V_T < 0, attractive forces dominate, promoting particle attachment.

The van der Waals interaction energy (V_W) of two spherical particles is calculated as follows:

$${\mathrm{V}}_W=-{\displaystyle \frac{A{R}_f{R}_c}{6H\left(R_f+R_c\right)}}$$

(5)

where A is the Hamaker constant, R_f and R_c denote the radius of fine and coarse particles, and H represents the distance between two particles. The Hamaker constant A used in Equation (5) is calculated by the following equation:

$$A=\left(\sqrt{A_{11}}-\sqrt{A_{33}}\right)\left(\sqrt{A_{22}}-\sqrt{A_{33}}\right)$$

(6)

where A₁₁, A₂₂, and A₃₃ refer to the Hamaker constants of monazite (2.0 × 10⁻¹⁹ J), magnetite (2.4 × 10⁻¹⁹ J), and water in the vacuum (4.0 × 10⁻²⁰ J), respectively [16,23,24,25].

The electrical double layer interaction energy (V_E) between two spherical particles is expressed as follows:

$${\mathrm{V}}_E=\pi {\epsilon }_0{\epsilon }_r{\displaystyle \frac{R_f{R}_c}{R_f+R_c}}\left({\psi }_1^2+{\psi }_2^2\right)\left[{\displaystyle \frac{2{\psi }_1{\psi }_2}{{\psi }_1^2+{\psi }_2^2}}p+q\right]$$

(7)

$$\mathrm{p}=\ln \left[{\displaystyle \frac{1+\mathit{exp}\left(-\kappa H\right)}{1-\mathit{exp}\left(-\kappa H\right)}}\right]$$

(8)

$$\mathrm{q}=\ln \left[1-\exp \left(-2\kappa H\right)\right]$$

(9)

where ε₀ and ε_r refer to the absolute dielectric constant in vacuum (8.854 × 10⁻¹² F/m) and the relative dielectric constant of the dispersion medium (for water, 78.5 F/m), and ψ₁ and ψ₂ are the surface potential of monazite and magnetite, replaced by their zeta potentials. κ is the Debye constant and can be calculated by the following equation:

$$\kappa ={\left({\displaystyle \frac{2e^2{N}_{A}c{z}^2}{{\epsilon }_0{\epsilon }_{r}kT}}\right)}^{{\displaystyle \frac{1}{2}}}$$

(10)

where e is the elementary charge (1.602 × 10⁻¹⁹ C), N_A is the Avogadro number (6.023 × 10²³ mol⁻¹), c is the concentration (mol∙m⁻³), z is the valence of the ion, k is the Boltzmann constant (1.381 × 10⁻²³ J∙K⁻¹), and T is the temperature (K) [26].

The hydrophobic interaction energy (V_H) is given as follows:

$${\mathrm{V}}_H=2\pi {\displaystyle \frac{R_f{R}_c}{R_f+R_c}}{h}_0{V}_H^0\mathit{exp}\left({\displaystyle \frac{H_0-H}{h_0}}\right)$$

(11)

where H₀ and h₀ are the minimum equilibrium contact distance between particles (0.2 nm) and the decay length (5 nm), respectively [26]. V_H⁰ refers to the interfacial polar interaction energy constant and can be calculated by the following equation:

$$V_H^0=2\left[\sqrt{{\gamma }_3^{+}}\left(\sqrt{{\gamma }_1^{-}}+\sqrt{{\gamma }_2^{-}}-\sqrt{{\gamma }_3^{+}}\right)+\sqrt{{\gamma }_3^{-}}\left(\sqrt{{\gamma }_1^{+}}+\sqrt{{\gamma }_2^{+}}-\sqrt{{\gamma }_3^{-}}\right)-\sqrt{{\gamma }_1^{+}{\gamma }_2^{-}}-\sqrt{{\gamma }_1^{-}{\gamma }_2^{+}}\right]$$

(12)

where γ₁⁺, γ₂⁺, and γ₃⁺ represent the electron-acceptor parameters, while γ₁⁻, γ₂⁻, and γ₃⁻ denote the electron-donor parameters of the surface energy for monazite, magnetite, and water, respectively. The values γ₃⁺ and γ₃⁻ are both 25.5 × 10⁻³ J/m². The parameters γ₁⁺, γ₁⁻, γ₂⁺, and γ₂⁻ are determined using the following equation:

$$\left(1+cos\theta \right){\gamma }_L=2\left(\sqrt{{\gamma }_S^d{\gamma }_L^d}+\sqrt{{\gamma }_S^{+}{\gamma }_L^{-}}+\sqrt{{\gamma }_S^{-}{\gamma }_L^{+}}\right)$$

(13)

where θ represents the contact angle of the mineral, while the subscripts S and L denote the solid and liquid phases, respectively. The superscript d indicates the dispersive component of the surface energy. The contact angles of the minerals were measured with water, glycerol, and formaldehyde. The surface energy parameters of these liquids (γ_L, γ_L^d, γ_L⁺, and γ_L⁻) are provided in

Table 3. The surface energy parameters of liquids [27].

Liquid	γL (mJ∙m−2)	γLd (mJ∙m−2)	γL+ (mJ∙m−2)	γL− (mJ∙m−2)
Water	72.8	21.8	25.5	25.5
Glycerol	64.0	34.0	4.92	57.4
Formaldehyde	58.0	39.0	2.28	39.6

.

3. Results and Discussion

3.1. Surface Modification of Monazite and Magnetite Using Sodium Oleate

Figure 3 shows the zeta potential of monazite, magnetite, quartz, and zircon with and without NaOL. Because NaOL is an anionic surfactant, its adsorption reduced the zeta potential values of the minerals. In other words, measuring the zeta potential values of minerals with and without NaOL at various pH conditions can help determine the suitable pH at which NaOL selectively adsorbs onto the surfaces of target minerals (e.g., monazite and magnetite). Determining the appropriate pH conditions for the selective adsorption of NaOL onto monazite and magnetite surfaces is crucial, as hydrophobic interactions between these minerals are the primary force driving their agglomeration. As illustrated in Figure 3a, the isoelectric point (IEP) of monazite in DI water was around pH 5, but this shifted to about pH 3 in the presence of NaOL. The substantial change in the surface charge of monazite strongly suggests the adsorption of oleate anions. This result is to be expected because at pH values below the IEP of monazite, its surface is positively charged, making the adsorption of oleate anions via physisorption easier [28]. Aside from physisorption, oleate anion was adsorbed via chemisorption on the surface of monazite, as implied by the more negative zeta potential measured above the mineral’s IEP. Similar results were obtained for magnetite (Figure 3b); that is, the zeta potential values decreased in the presence of NaOL. Meanwhile, the changes in the zeta potentials of quartz and zircon with and without NaOL were negligible (Figure 3c,d), indicating that NaOL was barely adsorbed on their surfaces. The zeta potential measurement results revealed that NaOL could be adsorbed on monazite and magnetite surfaces under acidic (pH 2) to alkaline (pH 10) conditions. The changes in the zetapotential values of monazite and magnetite were significant between pH 6 and 8. The pH for the surface modification process was fixed at 7.

The changes in the hydrophobicity of monazite and magnetite surfaces because of oleate adsorption were evaluated by their contact angle measurements (Figure 4). The contact angles of raw monazite and magnetite were 47.6° and 43.0°, respectively, but when these minerals were immersed in oleate solution, their contact angles increased to 66.3° and 76.2°. The increase in the contact angles indicates that oleate adsorption increased the hydrophobicity of monazite and magnetite, and these results also support our earlier deductions concerning oleate adsorption on monazite and magnetite surfaces due to changes in zeta potential measurements (Figure 3a,b). Due to the more hydrophobic surfaces of monazite and magnetite, the attachment of fine monazite particles on the surfaces of coarse magnetite particles (i.e., carrier) via hydrophobic interactions is likely enhanced [16]. In the next section, the effects of NaOL treatment on carrier magnetic separation to recover fine monazite from simulated tailings will be discussed.

3.2. Carrier Magnetic Separation to Recover Monazite from Simulated Tailings

Figure 5a shows the recovery of magnetite, monazite, and gangue minerals (i.e., quartz and zircon) by magnetic separation with and without NaOL treatment. Without NaOL treatment, the recoveries of magnetite, monazite, and gangue minerals were 98%, 9%, and 9%, respectively. However, the application of NaOL treatment did not improve the magnetic separation results; that is, a high recovery of magnetite (>95%) but low recoveries of monazite (~10%) and other minerals (~10%) were observed. Because quartz and zircon are non-magnetic minerals, they were most likely recovered via entrapment when magnetite particles were recovered. The recovery of monazite was almost the same as that of quartz/zircon, indicating that monazite was also recovered by entrapment.

Figure 5b, which gives the separation efficiency of TREO from gangue minerals, illustrates the relationship between TREO recovery in magnetic fraction and gangue recovery in non-magnetic fraction. The separation efficiency was less than 1% irrespective of NaOL treatment, which indicates that NaOL treatment alone was ineffective in improving the recovery of fine monazite from simulated tailings.

It is important to note that although NaOL treatment improved the hydrophobicity of monazite and magnetite (Figure 3 and Figure 4), this was not enough to increase the recovery of monazite (Figure 5). This indicates that hydrophobic interactions between NaOL-treated monazite and magnetite particles were not strong enough for heterogenous agglomeration to occur. As evidenced by the zeta potential results (Figure 3), both NaOL-adsorbed monazite and magnetite particles carried a negatively charge, leading to electrostatic repulsion between them (Figure 6a, left). This repulsion likely accounts for the low monazite recovery, even after hydrophobizing their surfaces. In other words, electrostatic repulsion outweighs hydrophobic attraction.

To confirm this, the interaction energies between NaOL-treated fine monazite and coarse magnetite were theoretically calculated using the EDLVO theory. As shown in Figure 6b, V_W and V_H function as attractive forces, whereas V_E acts as a repulsive force. It is important to note that the electrostatic repulsive force (V_E) is stronger than the attractive forces (V_W+V_H) until the distance between two particles becomes less than 18 nm. This indicates that under these hydrodynamic conditions, the two particles could not overcome the energy barrier required for attachment. To improve the hydrophobic interactions and increase the heterogenous agglomeration of fine monazite and coarse magnetite particles, the potential of EK as a bridging liquid during surface modification was investigated (Figure 6a, right). When EK was present, the attractive and repulsive forces between NaOL-treated fine monazite and coarse magnetite particles were balanced at an interparticle distance of 23 nm, primarily due to the increased hydrophobic interaction energy induced by EK (Figure 6c). Furthermore, the enhancement of the attractive force between particles became more pronounced as the interparticle distance decreased, suggesting stronger particle–particle interactions and the formation of stable agglomerates that are difficult to break once formed (Figure 6c).

Hornn et al. [20] reported that a kerosene/water emulsion is only stable at either a high agitation speed (~15,000 rpm) or when emulsified. As the former is more energy-intensive than the latter, this study tested two types of surfactants (e.g., NaOL and SDS) to produce a stable EK. These two surfactants were chosen because NaOL is used as a surface modifier to render monazite and magnetite surfaces hydrophobic, and SDS is already well-known as an effective surfactant in stabilizing kerosene/water emulsions [20]. As shown in Figure 7, the addition of NaOL-emulsified kerosene slightly improved the recovery of monazite from ~10% to 16%, while its recovery significantly increased to ~55% when using SDS-emulsified kerosene. The recovery of gangue minerals was almost kept constant (~10%), so the separation efficiency increased from 0.8% to 4.5% with NaOL-emulsified kerosene and to 42.4% with SDS-emulsified kerosene. This is most likely due to SDS’s better ability to stabilize EK than NaOL. When a kerosene/water emulsion becomes unstable, like in the case of NaOL-emulsified kerosene, the coalescence of oil droplets becomes inevitable, resulting in a rapid decline in dispersed oil droplets required for connecting monazite and magnetite particles [20].

Figure 8 shows the particle size distributions of the monazite/magnetite mixture with and without SDS-emulsified kerosene. Quartz and zircon particles were excluded to clearly identify the attachment of fine monazite particles onto the surface of coarse magnetite particles. After NaOL treatment, followed by the addition of SDS-emulsified kerosene, the particle size distribution of the monazite/magnetite mixture changed dramatically. As illustrated in Figure 8a, three peaks were observed: (i) un-agglomerated/unattached monazite particles, (ii) agglomerated but unattached monazite particles, and (iii) monazite particles attached to the surface of coarse magnetite particles. The attachment of fine monazite particles to the carrier surface was further confirmed by SEM-EDS analysis (Figure 9). The ratio of attached monazite particles (iii) to other monazite particles (un-agglomerated/unattached (i) and agglomerated (ii)) was 55:45 (Figure 8b), indicating that only monazite particles attached to the surface of coarse magnetite particles were recovered by magnetic separation (Figure 7).

The number of carrier particles is one of the most important parameters for carrier magnetic separation because it is directly proportional to the binding sites where fine monazite particles could attach. To further improve the recovery of monazite, the effects of the monazite/magnetite particle ratio was investigated (Figure 10). When no carrier particles were added, nothing was recovered (Figure 10a). This supports the earlier deduction that even if agglomerates of fine monazite particles are formed, they cannot be recovered (Figure 7 and Figure 8). In other words, for the recovery of fine monazite particles by magnetic separation to be effective, they should be attached to the surfaces of magnetic carrier particles. As the amount of carrier particles increased, the recovery of monazite also increased; that is, monazite recovery increased from 55% at a monazite–magnetite ratio of 3:7 to ~70% at a monazite–magnetite ratio of 2:8 (Figure 10a). Moreover, the separation efficiency increased from 42% at a monazite–magnetite ratio of 3:7 to 54% at a monazite–magnetite ratio equal to 2:8 (Figure 10b). At a higher magnetite ratio, the yield of the magnetic fraction increases, which may lead to non-selective mechanical entrapment. However, the recovery of gangue minerals increased by only 1.7%, compared to a 15% increase in monazite recovery. This indicates that the enhanced monazite recovery is primarily attributed to the increased number of attachment sites provided by the higher carrier content rather than non-selective mechanical entrapment.

Figure 11 summarizes the results of a rougher–scavenger–cleaner carrier magnetic separation using the WHIMS. After a rougher carrier magnetic separation, about 46% of the TREO was distributed in the magnetic fraction, at which point the TREO grade increased from 3.4% to 10.9%. To further improve TREO recovery, non-magnetic fraction was mixed with an additional carrier because the previously added carrier particles were all recovered by the WHIMS, treated with NaOL followed by EK for heterogeneous agglomeration, and then subject to scavenger carrier magnetic separation. As a result, of the 54% of TREO remaining in the non-magnetic rougher tailings, ~45% of the TREO could be further recovered as a magnetic product, and its grade was ~5.9%. To further improve the TREO grade, rougher and scavenger magnetic products were processed by a cleaner carrier magnetic separation, which could produce REE concentrates with a TREO recovery of 80% and grade of 9.0% from simulated tin tailings.

Although the proposed flowsheet was effective in producing REE concentrates from simulated tin tailings, there is a remaining limitation due to the use of heterogeneous carrier particles, which lowers the REE grade. To overcome this limitation, the separation of monazite and carrier from magnetic concentrates is essential. Moreover, monazite/carrier separation is important in terms of the reutilization of the carrier. As fine monazite particles are attached to coarse carrier particles via hydrophobic interaction, reducing particles’ hydrophobicity is required for them to be detached. In a previous study by the authors [29], acid treatment was proven to be effective in detaching fines attached to carriers by destroying surfactant molecules adsorbed on their surfaces. After the separation of fine monazite from carrier via acid treatment followed by sizing or low-intensity magnetic separation (LIMS), the REE grade can be increased to ~23.2%.

4. Conclusions

This study investigated the applicability of carrier magnetic separation using coarse magnetite particles as a carrier to improve the recovery of fine monazite particles from the simulated tailings based on actual tin tailings from Indonesia composed of quartz, zircon, and monazite. In order for the attachment of fine monazite particles onto the surfaces of coarse magnetite particles to occur via hydrophobic interactions, the surface modification of monazite and magnetite particles using NaOL was carried out to render their surfaces hydrophobic. This treatment was followed by the addition of EK as a bridging liquid to induce heterogenous agglomeration. After treatment using NaOL and SDS-emulsified kerosene, three types of monazite particles were observed: (i) un-agglomerated/unattached, (ii) homogeneously agglomerated, and (iii) attached monazite particles. Of these, only monazite particles attached to the surfaces of magnetite particles could be recovered by magnetic separation. Conventional magnetic separation could not recover finely ground monazite particles from the simulated tailings, while surface modification followed by rougher–scavenger–cleaner carrier magnetic separation could recover ~80% of monazite. These results indicate that carrier magnetic separation is a promising technique to recover fine rare earth minerals from tailings as well as low-grade ores that require fine grinding to achieve a sufficient degree of liberation. The monazite-rich REE concentrate recovered from tailings can be utilized as a feedstock for rare earth extraction in facilities equipped to safely manage thorium-bearing materials. This approach not only contributes to resource recovery but also offers a viable solution for mitigating the environmental impact of radioactive tailings.