Dry Magnetic Separation and the Leaching Behaviour of Aluminium, Iron, Titanium, and Selected Rare Earth Elements (REEs) from Coal Fly Ash

Abstract

Coal fly ash (CFA) is a commercially viable source of alumina comparable to traditional bauxite deposits. Due to its high silica content and alumina in the refractory mullite phase, the most suitable processing technique is the sinter-H₂SO₄ leach process. However, this process is energy-intensive, has low selectivity for Al, and generates a secondary solid waste residue. To develop a sustainable process that is economically attractive, Al can be extracted with REEs, Ti, and Fe as saleable products, while secondary solid waste is regenerated for further applications to achieve high-value and high-volume utilisation of CFA. This study focused on the potential extraction of selected REEs (Ce, La, Nd, Y, and Sc), Al, Ti, and Fe, using dry magnetic separation and the sinter-H₂SO₄ leach process. XRD analysis showed that CFA is predominantly amorphous with crystalline mullite, quartz, and magnetite/hematite. Further analysis using SEM-EDS and TIMA showed Al-Si-rich grains as the predominant phase, with discrete REE-bearing grains (phosphates and silicates) and Fe-oxide (magnetite/hematite) grains. Traces of REEs, Ti, Ca, Si, and Fe were also found in the Al-Si-rich grains. Discrete Fe-oxide was recovered using dry magnetic separation, and up to 65.9% Fe was recovered at 1.05 T as the magnetic fraction (MF). The non-magnetic fraction (non-MF) containing quartz, mullite, and amorphous phase was further processed for preliminary leaching studies. The leaching behaviour of Al, Ti, Fe, and the selected REEs was investigated using the direct H₂SO₄ and sinter-H₂SO₄ leaching processes. The maximum extraction efficiency was observed using the sinter-H₂SO₄ leach process at 6 M H₂SO₄, a 1:5 solid-to-liquid ratio, 70 °C, and a residence time of 10 h, yielding 77.9% Al, 62.1% Fe, 52.3% Ti, and 56.7% Sc extractions. The extraction efficiencies for Ce, La, Nd, and Y were relatively lower at 23.2%, 27.6%, 11.3%, and 11.2%, respectively. Overall, the results demonstrate that the extraction of REEs using the sinter-H₂SO₄ leach process is strongly influenced by the complex CFA phase composition and the possible formation of insoluble calcium sulphates. Appreciable extraction of Al, Fe, Ti, and Sc was also observed, suggesting a potential two-step leaching process for the extraction of REEs as a feasible option for the industrial recovery of multiple saleable products.

1. Introduction

South Africa operates the Hillside Aluminium Smelter, one of Africa’s largest aluminium smelters that refines alumina on a commercial scale for domestic use and exports [1]. The smelter imports alumina from Worsley Alumina in Australia and has the capacity to produce 720 kt of aluminium products per annum [2]. Despite a well-established smelter industry, there are no known exploitable alumina deposits in the country. Nonetheless, South African coal fly ash (CFA) contains 26%–31% Al₂O₃, making it the most commercially viable source after traditional bauxite deposits (30%–60% Al₂O₃) [3,4,5]. However, processing CFA using the conventional Bayer process has proven challenging due to its high silica content (40–60 wt.% SiO₂) and the presence of the refractory mullite phase (Al₆Si₂O₁₃). Due to the low solubility of silica in acidic solutions, acid leaching has been extensively studied as an ideal option for alumina processing [3,4]. Specifically, the sinter-H₂SO₄ leach process has been used to process South African CFA and has demonstrated the potential to extract over 80 wt.% of aluminium, which can be commercially processed to generate smelter-grade Al₂O₃ [6]. In this process, mullite is converted into a more soluble plagioclase/anorthite, followed by H₂SO₄ leaching, as shown in Equations (1) and (2) [6,7]. The process is energy-intensive and results in a high co-extraction of Fe and Ti, which requires further downstream processing for their removal [3,4]. Sibanda et al. [4] proposed that extracting multiple resources from CFA could add value to CFA processing and thus enhance the economic feasibility. As such, Ti and Fe, which account for 1%–2% TiO₂ and 4%–6% Fe-oxide in CFA, can also be extracted and recovered as saleable products in the form of rutile (TiO₂), a precursor for paint pigmentation, and iron ore (FeO), a feedstock for iron smelting [8].

$${3\mathrm{A}\mathrm{l}}_2{\mathrm{O}}_3·{2\mathrm{S}\mathrm{i}\mathrm{O}}_2+3\mathrm{C}\mathrm{a}\mathrm{O}+{4\mathrm{S}\mathrm{i}\mathrm{O}}_2\stackrel{1100°\mathrm{C}}{\to }3(\mathrm{C}\mathrm{a}\mathrm{O}·{\mathrm{A}\mathrm{l}}_2{\mathrm{O}}_3·{2\mathrm{S}\mathrm{i}\mathrm{O}}_2)$$

(1)

$$\mathrm{C}\mathrm{a}\mathrm{O}·{\mathrm{A}\mathrm{l}}_2{\mathrm{O}}_3·{2\mathrm{S}\mathrm{i}\mathrm{O}}_{2\left(\mathrm{s}\right)}+{4\mathrm{H}}_2{\mathrm{S}\mathrm{O}}_{4\left(\mathrm{a}\mathrm{q}\right)}\to {\mathrm{A}\mathrm{l}}_2{\left({\mathrm{S}\mathrm{O}}_4\right)}_{3\left(\mathrm{a}\mathrm{q}\right)}+{\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_{4\left(\mathrm{s}\right)}+2{\mathrm{S}\mathrm{i}\mathrm{O}}_{2\left(\mathrm{s}\right)}+{4\mathrm{H}}_2\mathrm{O}$$

(2)

Rare earth elements (REEs) have also been reported in South African CFA as phosphate minerals (e.g., YPO₄) with concentrations ranging from 400 to 600 mg/kg [9,10,11]. Therefore, extracting REEs with Al, Ti, and Fe could enhance the value of CFA processing. Additionally, the extraction of REEs from secondary resources such as CFA and phosphogypsum waste could provide a potential alternative to China, which currently dominates the REE market, accounting for approximately 40% of the global reserves [12]. However, despite the presence of REEs in South African CFA, there is limited research on their extraction reported in literature. In contrast, extensive research on REEs has been conducted and reviewed elsewhere [13]. Gupta and Krishnamurthy [13] reported that some of the challenges in processing REEs, either from primary or secondary sources, are their low concentrations and complex mineralogy. Consequently, their selective separation from major elements (i.e., Al and Fe) using hydrometallurgical processes is challenging. Another significant challenge is their poor leaching efficiency when extracted using sulphuric acid at higher acid concentrations (>3 M H₂SO₄) and leaching temperatures (>70 °C) [11,14,15]. Depending on the chemical composition of the REE resource, heavy REEs (HREEs) and light REEs (LREEs) can precipitate with calcium sulphates, resulting in the formation of insoluble compounds [16,17]. In light of these challenges with sulphuric acid leaching, hydrochloric acid has been reported as a suitable lixiviant for the extraction of REEs from CFA [18,19].

Magnetic separation is a physical separation technique that is commercially used to reduce Fe-oxide (magnetite/hematite) during hydrometallurgical processing of REEs [13]. In acid leaching processes, REEs are co-extracted with Fe in solution, which makes the purification stage (i.e., solvent extraction) a challenge during selective stripping [13,14]. To mitigate this challenge, magnetic separation is used as the cheapest alternative for Fe-oxide recovery. CFA also contains Fe-oxide (magnetite/hematite), which can be recovered by magnetic separation as the magnetic fraction (MF) [20,21]. Phase compositions, such as mullite, quartz, and amorphous phase, are non-magnetic components in CFA and are therefore recovered as non-magnetic fraction (non-MF) [10,22]. Phase composition analysis of CFA has also demonstrated that REEs are largely associated with aluminosilicate phases. Therefore, during magnetic separation, they are recovered as non-MF [21]. Different magnetic separation techniques, such as dry and wet magnetic separation, can be used to recover Fe-oxide depending on the required final product and its intended use [20].

Rybak and Rybak [23] proposed that a multi-stage process could be an efficient method for pre-concentrating REEs from CFA using size classification, magnetic separation, and acid–alkaline treatment techniques. Due to the trace concentrations of REEs in CFA compared to major elements such as Al, Fe, and Ti, dry magnetic separation and the sinter-H₂SO₄ leach process were identified in this study as suitable processing techniques for concentrating REEs while simultaneously extracting Al, Fe, and Ti. In this paper, dry magnetic separation was used to recover Fe-oxide, thereby reducing the cost of downstream purification. The recovered MF (Fe-oxide component) was further repurposed as a seeding material for wastewater treatment elsewhere [24]. The remnant non-MF was processed in this study using the sinter-H₂SO₄ leach process due to its capacity to extract alumina worth scaling. The sintering process was also selected for possible liberation of REE-bearing phosphate grains. Both the direct H₂SO₄ and sinter-H₂SO₄ leaching processes were used to determine the leaching behaviour of Al, Fe, Ti, and selected REEs (Ce, La, Nd, and Sc). The effect of CFA phase composition on metal extraction was also evaluated. This paper represents the first stage of a broad investigation to process CFA using a two-step leaching process and an integrated purification technique to generate saleable products. In this paper, CFA is characterised for the extraction of Al, Fe, Ti, and selected REEs.

2. Materials and Methods

2.1. Materials

CFA was supplied by Eskom Power Utility in South Africa, characterised, and processed using dry magnetic separation and the direct H₂SO₄ and sinter-H₂SO₄ leaching processes, as shown in Figure 1. Coal with a fixed carbon content of 56% was sourced from Trans-alloys (Pty), Ltd. Analytical-grade sulphuric acid (98%) and hydrochloric acid (32%) used for the leaching experiments were supplied by Merck, South Africa. All solutions were prepared using deionised water.

2.2. Analyical Methods

2.2.1. Particle Size Distribution

The particle size distribution (PSD) of raw CFA was determined using a Malvern Mastersizer (3000+) and a physical screening method (Figure S1). Physical screening was conducted using test sieves (Fritsch, Idar-Oberstein Germany) of various sizes ranging from –38 μm to +212 μm, and more than 90% of the sample passed through 212 µm sieves (Figure S1a). Further analysis showed a d₅₀ of 20.05 μm (Figure S1b). Overall, the analysis indicated that the CFA used in this study is a fine-grained material with exposed surfaces that is efficient for leaching. Therefore, the material was used in the experiments without further grinding [25,26].

2.2.2. Chemical Analysis

The major chemical composition of the solid samples and aqueous solutions was analysed using X-ray fluorescence (XRF) and inductively coupled plasma–optical emission spectrometry (ICP-OES), respectively. Trace chemical analysis (i.e., REEs) was conducted using inductively coupled plasma–mass spectrometry (ICP-MS). All chemical analyses were outsourced from the ISO/IEC 17025 accredited laboratory [27].

2.2.3. Morphological Analysis

The morphologies of the CFA samples before and after leaching were determined using scanning electron microscopy–energy dispersive spectroscopy (SEM-EDS). The analysis was conducted using an MA 15 EVO SEM ZEISS instrument (Model: Sigma, Carl Zeiss, UK) at a 20 kV accelerating voltage. To prepare the samples, a double-sided adhesive carbon tape was mounted onto aluminium stubs, and a small portion of the pulverised powder sample was sprinkled and dusted with a spray to remove any contamination. The samples were coated twice with Ag/Pd to prevent surface charging during analysis.

2.2.4. Phase Composition Analysis

X-ray diffraction (XRD) was used to determine the crystalline and amorphous phases of the samples. Powdered samples were prepared using the back-loading method, and analysis was performed using an Aeris diffractometer (Malvern PANalytical, Netherland) with a PIXcel detector and fixed slits with Fe-filtered Co-Kα radiation. X’Pert Highscore plus software (version 4.9) was used to identify the phases. Further bulk composition analysis was conducted using SEM-EDS described in Section 2.2.3, as well as a TESCAN Integrated Mineral Analyzer (TIMA, 2.4.1 software) housed at the School of Geoscience (WAMLAB), University of the Witwatersrand. For TIMA analysis, elemental mapping was used to locate the trace elemental distribution. Neodymium (Nd) and cerium (Ce) were used as proxies for LREEs, while yttrium (Y) was used as a proxy for HREEs. Data analysis was performed through modal analysis at a high resolution and liberation by dot mapping. A pixel size of 1–4 μm, with a working distance of 15 mm, was selected for various sample analyses. Polished block samples with a diameter of 30 mm were used for phase composition analysis. All the samples were carbon-coated prior to analysis.

2.3. Experimental Procedure

2.3.1. Dry Magnetic Separation

An induced roller magnetic separator (Model IMR. roll, Eriez, South Africa) was used to recover Fe-oxide from raw CFA. A vibrating feed hopper was fed with approximately 500 g of CFA at predetermined parameter levels. The magnetic fraction (MF) was collected from the pole piece, whereas the non-magnetic fraction (non-MF) was collected from the bottom pans for analysis. The feed rate, roller speed, and pole piece were maintained at 8 tph, 133.2 rpm, and 30°, respectively. The applied magnetic field strength was varied from 0.44 to 1.05 T. The enrichment factor (EF), which is a measure of how the concentration of an element increases after the separation process, was calculated according to Cornelius et al. [10] using Equation (3).

$$\mathrm{E}\mathrm{n}\mathrm{r}\mathrm{i}\mathrm{c}\mathrm{h}\mathrm{m}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{f}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{r}(\mathrm{E}\mathrm{F})={\displaystyle \frac{{\mathrm{C}}_1}{{\mathrm{C}}_0}}$$

(3)

where C₀ is the concentration of metals before magnetic separation, and C₁ is the concentration of metals after magnetic separation.

2.3.2. Leaching Experiments

Leaching experiments were conducted in shaking incubators, using non-MF as the starting material. The metal leaching efficiency was calculated as the ratio of the amount of metal leached into solution to the initial amount of metal in the sample. The extraction efficiency was calculated using Equation (4).

$$\mathrm{E}\mathrm{x}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{e}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y}={\displaystyle \frac{\mathrm{V}{\mathrm{C}}_2}{\mathrm{M}{\mathrm{C}}_1}}\times 100\mathrm{\%}$$

(4)

where V is the volume of a leachate in mL, and C₂ is the metal concentration in the leachate in μg/mL. M is the mass of the feed sample in g, and C₁ is the metal concentration of the feed sample in μg/g.

Direct Acid Leaching

Approximately 50 g of non-MF samples were transferred to 500.0 mL Erlenmeyer flasks. Different acidic solutions were added to each flask, and the flasks were sealed. The leaching experiments were conducted at various sulphuric acid concentrations (0.5–6 M), leaching times (6–20 h), and leaching temperatures (45–80 °C). The solid/liquid ratio and stirring rate were maintained at 1:5 and 160 rpm, respectively. After leaching, the solutions were filtered, washed, and quantitatively transferred to a 500.0 mL volumetric flask. Aliquots of each sample and dried residue were analysed for their chemical and phase compositions.

Indirect Acid Leaching

The lime-sinter process was used as a pre-treatment method prior to acid leaching [6,7]. To produce pellets strong enough for sintering, the non-MF was mixed with CaCO₃, coal (56% fixed carbon), and water (10%–30%). CaCO₃ served as a source of lime (CaO) to transform the mullite into a soluble form, while finely ground coal (similar to CFA) acted as a source of heat energy during sintering. The pellets were air-dried and then sintered in a muffle furnace at 1100 °C for 5 h [7]. The sintered pellets were crushed and ground to a powder (similar grind as raw CFA), which was then used to conduct the leaching experiments as described in the direct acid leaching procedure. The concentrations of major elements and REEs in the coal and the sintered CFA were determined.

3. Results and Discussion

3.1. Characterisation of Raw CFA

3.1.1. Chemical Composition Analysis of Raw CFA

The chemical analysis of raw CFA is presented in

Table 1. Analysis of the major chemical compositions (wt.%) and REEs (mg/Kg) in raw CFA.

Analyte	Raw CFA	* RSD
SiO2	56.32	0.07
Al2O3	28.82	0.02
CaO	5.85	0.04
FeO	0.23	0.02
Fe2O3	4.11	0.01
TiO2	1.70	0.01
MgO	1.53	0.03
MnO	0.06	0.0.
Na2O	0.26	0.01
K2O	0.84	0.01
P2O5	0.60	0.02
Carbon	2.30	0.06
Sulphur	0.16	0.02
Sc	25.7	0.04
Y	49.3	0.21
La	92	0.05
Ce	99.07	0.04
Pr	26.2	0.52
Nd	52.8	0.04
Sm	14.2	0.00
Eu	3.01	0.78
Gd	9.82	0.00
Tb	2.74	0.02
Dy	13.3	0.01
Ho	3.14	0.01
Er	4.78	0.01
Tm	1.31	0.01
Yb	5.64	0.08
Lu	1.53	0.02
ΣLREEs	297.1	--
ΣHREEs	81.74	--
ΣREEs	404.54	--

-- Not available. * Relative standard deviation, L.O. I = 2.6.

. The table shows 29.2% Al₂O₃, 4.34% Fe-oxide, and 1.70% TiO₂, which are attractive oxides for the set objectives. The ICP-MS analysis shows REE concentration of 404.54 mg/kg (0.0404%). The LREEs had higher concentrations, with 297.1 mg/kg (0.02971%), while HREEs had 81.74 mg/kg (0.008174%). Compared to other countries (i.e., China, the USA, and India), the REE concentration in South African CFA is well within the typical range for commercial consideration [28].

3.1.2. Phase Composition Analysis of Raw CFA

The phase composition of raw CFA is listed in

Table 2. The phase composition analysis of raw CFA.

Phase Composition	Chemical Formulae	Concentration, wt.%
		Phase	Al2O3
Mullite	Al6Si2O13	28.9	60.04
Quartz	SiO2	12.0	--
Magnetite	Fe3O4	1.5	--
Hematite	Fe2O3	0.3	--
Amorphous	--	57.3	39.96

-- Not available.

(also see Figure S2). The results reveal that raw CFA is predominantly amorphous, with mullite (Al₆Si₂O₁₃), quartz (SiO₂), magnetite (Fe₃O₄), and hematite (Fe₂O₃) as the crystalline components. Overall, the XRD analysis is consistent with those reported for South African CFA [7,10,29].

SEM-EDS was used to determine the major phase composition of raw CFA, and the results are presented in Figure 2. The analysis in Figure 2 shows that CFA is predominantly an aluminosilicate (Al-Si-rich) material with spherical and irregular shapes. Traces of REEs were also detected in the Al-Si rich grains.

Similar results indicating that Al-Si grains were the major phase component can be seen in Figure 3. These grains also contained Fe and Ca in various proportions, such as Fe-Al-Si and Ca-Al-Si, respectively. Other phases identified included the Fe-rich, Si-rich, and Ca-rich components. Using Mineral Liberation Analysis (MLA), Renani et al. [9] reported a similar Al-Si-rich component in South African CFA. The mullite phase is defined as an aluminosilicate (Al-Si-rich) component with elongated needles [30,31]. Although elongated aluminosilicate needles were not detected in this study, grains identified as mullite hosts by Dai et al. [30] and Li et al. [31] were observed, as shown in Figure 2a, grain A. In Figure 2a, grain C, a bright white grain concentrated with Fe was also observed, suggesting the presence of an Fe-oxide component. Similar observations were made by Valeev et al. [21] and Strzałkowska [32], who reported that the bright white grains were hematite/magnetite.

The BSE images of CFA obtained using TIMA, along with the normalised weight concentrations of the selected grains, are presented in Figure 4. Elemental maps were utilised to locate the trace distributions of the selected REEs, Fe, and Ti. Iron was found as discrete grains and occurred within the Al-Si-rich grains. The Fe-rich grains in Figure 4c resembled Fe-oxides and contained traces of Si, Al, Ca, and Mg. The Al-Si-rich grains (grains 3, 8, and 14) in Figure 4b–d also show traces of Fe and Ti in their chemical composition. These results are consistent with the findings by Yang et al. [33] and Dai et al. [34].

The distributions of selected REEs (Ce, Nd, and Y) are presented in Figure 5. As shown in Figure 5a, REE-phosphates were detected in the Al-Si-rich grain, while as shown in Figure 5b, they appeared as discrete phosphate grains. Palozzi et al. [35] used TIMA analysis to identify monazite (Ce, Th, La, Nd) PO₄ as a REE-bearing mineral that was somewhat bound to either silica or Al-Si-rich grains. Rerani et al. [9] used Mineral Liberation Analysis (MLA) and quantified xenotime and monazite as REE-phosphate minerals. A similar pattern was observed in the present study for Y-phosphate and Ce/Nd-phosphate. Zircon was also quantified as a REE-bearing mineral in CFA. As shown in Figure 5c,d, Al-Si-rich grains resembling zircon as a solid solution were observed. In Figure 5d, the elemental map shows the distribution of Nd and Ce in the zircon grain. However, due to the trace concentrations of REEs in this grain, colour mapping is moderate. Overall, the occurrence of REE-bearing phases as silicates and phosphates in this study is consistent with the existing literature [9,35,36].

3.2. Recovery of Fe-Oxide

In Section 3.1.2, Fe-oxide is reported as hematite/magnetite. Further analysis also showed that Fe can occur as discrete grains and inclusions in the Al-Si-rich grains. Therefore, magnetic separation (MS) was used to recover discrete Fe-containing grains.

3.2.1. Effect of Magnetic Field Intensity

The effect of magnetic field intensity on Fe-oxide recovery is shown in Figure 6. The results show that Fe-oxide recovery increased with increase in induced current, and a maximum Fe recovery of 65.9% was obtained at 1.05 T. Valeev et al. [21] investigated wet and dry MS to recover magnetite from Omsk CFA. The authors reported a maximum Fe-oxide recovery of 73.16% with 26.58% Al₂O₃ and 23.17% SiO₂ impurities at 0.11 T. The maximum Fe-oxide recovery in dry MS was approximately the same (~73%). However, the alumina and silica impurities were 5% higher [21]. Alumina and silica impurities were also quantified in this study and are further discussed in the next section.

3.2.2. Phase Composition and Chemical Analysis

The XRD patterns of the CFA magnetic fraction (MF) and non-magnetic fraction (non-MF) at 1.05 T are presented in Figure 7. The mullite, quartz, and amorphous components are reported as diamagnetic in CFA [20,21]. However, the figure shows that quartz and mullite phases were also recovered with magnetite/hematite in the magnetic fraction. The co-extraction of quartz and mullite could be due to the agglomeration of CFA particles, as reported by Shoumkova [20]. Other researchers have also reported the co-extraction of quartz and mullite in both wet and dry magnetic separation [10,20,21].

The chemical compositions of the magnetic and non-magnetic fractions are listed in

Table 3. Analysis of the major chemical compositions (wt.%) and REEs (mg/Kg) in non-MF and MF.

Analyte	Non-MF	* RSD	EF	CFA-MF	* RSD	EF
SiO2	54.9	2.01	0.97	33.0	1.32	0.59
Al2O3	29.2	2.15	1.01	19.1	1.68	0.66
CaO	5.44	1.24	0.93	2.9	1.02	0.50
FeO	0.022	0.23	0.097	0.20	0.22	0.87
Fe2O3	1.73	1.01	0.42	36.4	0.32	8.86
TiO2	1.59	0.07	0.94	1.4	0.05	0.82
MgO	1.64	0.39	1.07	0.9	0.40	0.59
MnO	0.035	0.00	0.58	0.038	0.00	0.64
Na2O	0.18	0.04	0.69	0.26	0.06	1
K2O	0.76	0.05	0.90	0.60	0.05	0,71
P2O5	0.76	0.14	1.27	--	0.12	--
Sc	31.4	1.18	1.22	14.2	0.92	0.55
Y	65.9	1.04	1.34	28.6	0.03	0.58
La	82.1	0.11	0.89	35.3	0.21	0.38
Ce	155.1	0.41	1.57	66.7	0.62	0.67
Pr	22.2	0.03	0.85	9.6	2.50	0.37
Nd	76.7	0.00	1.45	33.4	2.12	0.63
Sm	15.8	0.04	1.11	7.0	1.91	0.49
Eu	2.8	0.01	0.93	1.2	1.02	0.40
Gd	12.0	0.03	1.22	5.2	2.46	0.53
Tb	1.9	0.76	0.69	<1	--	--
Dy	13.8	0.11	1.04	6.1	1.19	0.46
Ho	2.3	0.01	0.73	1.0	0.88	0.32
Er	7.6	0.03	1.59	3.4	2.11	0.71
Tm	1.2	1.92	0.92	<1	--	--
Yb	6.7	0.00	1.19	3.0	2.23	0.53
Lu	<1	--	--	<1	--	--
ΣLREEs	354.7	--	--	153.2	--	--
ΣHREEs	111.4	--	--	47.3	--	--
ΣREEs	497.5	--	--	241.7	--	--

-- Not available. * Relative standard deviation. L.O. I = 2.7 non-MF and 0.8 MF.

. The high concentration of Fe in MF concurs with magnetite/hematite as the major component, as shown in Figure 7. The results in Table 3 also show high silica and alumina recovery to the MF fraction owing to the presence of mullite, quartz, and amorphous components. In addition to the agglomerative nature of CFA [20], magnetite is also identified as a solid solution between FeO and Fe₂O₃ and therefore has a spinel structure that can accommodate atoms such as Ca, Mg, and Mn [21]. Consequently, these elements are commonly quantified in the magnetic component of Fe-oxide. The current study also identified similar impurities of Ca, Mg, and Mn in the MF, suggesting that their presence could also result from the spinel effect [21]. The high concentration of Ti in the MF can be attributed to its association with the Al-Si and Fe-oxide phases, as depicted in Figure 4. The presence of REEs in the MF is also presented in Table 3. Although REEs were not detected in the Fe-oxide grains in Section 3.1.2, they are associated with Al-Si-rich grains, which are co-extracted as impurities during MS. REE-bearing phosphates, such as monazite, are also known to be magnetic and are often separated from gangue minerals using magnetic separation techniques [13]. In Section 3.1.2, REE-bearing phosphate grains are also reported, which could result in the presence of REEs in the magnetic fraction. Abaka-Wood et al. [37] and Moghiseh et al. [38] used MS to recover REEs from iron-silica-rich tailings and thickener waste, respectively. They observed the extraction of Fe alongside REEs as the magnetic field intensity increased. Further mineralogical analyses can also be conducted on CFA to determine the effect of MS on REEs.

Magnetic separation is also reported as an alternative pre-concentration method for upgrading REEs from CFA [22,39,40]. Further analysis of the non-MF in Table 3 shows a low enrichment factor (EF) for REEs, ranging from 0.69 to 1.59. Similar findings of low EF values were reported by Cornelius et al. [10]. Rybak and Rybak [23] suggested that magnetic separation alone may not be adequate as an REE pre-concentration method. Nonetheless, the non-MF obtained from the magnetic separation stage was further processed to extract multiple products (REEs, Al, Fe, and Ti). The MF is recommended for further applications such as catalysts and coagulants for wastewater treatment [24].

3.3. Pelletisation and Sintering

In Table 2, alumina is quantified in the mullite phase; therefore, sintering was utilised as a pre-treatment method to transform mullite into soluble plagioclase/anorthite, as shown in Equation (1). Thermal pre-treatment of CFA, which includes calcination and roasting, has also been reviewed as an alternative to enhance the leaching efficiency of REEs [19,41,42]. Therefore, it is important to monitor the transformation of REE-bearing minerals and phases during sintering. In this section, the non-MF obtained in Section 3.2.2, Table 3, was further processed using sintering, and the composition of the sintered CFA was determined.

3.3.1. Phase Composition of Sintered CFA

The effect of sintering at different ratios (CFA:coal:CaCO₃) is presented in

Table 4. The phase composition analysis of the sintered CFA at different ratios.

Phase Composition	Chemical Formulae	Phase Composition, wt%
		Raw CFA	Sintered CFA
		--	4:4:2	5:4:1	5:3:2
Mullite	Al6Si2O13	29.2	0.2	6.5	--
Anorthite	CaAl6Si2O14	--	56.9	48.7	53.2
Quartz	SiO2	14.7	3.1	5.5	2.4
Calcite	CaCO3	--	0.6	0.0	0.1
Amorphous	--	54.7	33.5	35.5	37.2

-- Not available.

(also see Figure S3). The results in the table show a maximum of 77.74% mullite transformation at a ratio of 4:4:2 into an aluminosilicate phase containing 56.9% anorthite. This ratio was used in all the subsequent leaching experiments. Similar findings of the mullite transformation are reported by Shemi et al. [7], who sintered CFA residue obtained from direct acid leaching of raw CFA at a ratio of 5:4:1. In their study, 75.0% of mullite was transformed into plagioclase. The relative difference in the percentage transformation observed in this study may be due to the difference in CFA starting material used in sintering and the variation in the ratios of CFA, coal, and CaCO₃.

TIMA analysis was further conducted to observe the surface transformation of the sintered CFA, and the BSE analysis is shown in Figure 8. Based on the morphology, most of the spherical CFA grains in Figure 4 and Figure 5 were transformed into non-spherical components during the sintering process. Therefore, this observation confirmed that a reaction occurred during sintering. The possible occurrence of REE-bearing minerals as phosphates and silicates (e.g., zircon) is discussed in Section 3.1.2. According to Wu et al. [43] and Manhique et al. [44], both phases can transform into oxides and carbonates during sintering, which makes them amenable to leaching. A typical oxide transformation is illustrated in Equations (5)–(7). In Figure 8a–c, the REE-phosphate grains are identified, suggesting that during sintering, some REE-phosphate grains were liberated from the host grains (i.e., Al-Si-rich) but were not transformed. It is worth noting that coal, which was primarily used as a source of heat transformation in the sintering process in this study, has also been reviewed as an REE resource [45,46]. Therefore, a potential mineralogical contribution of REEs from coal is anticipated in sintered CFA. While a detailed understanding of the distribution of REEs in coal and during the coal combustion process is essential, the mineralogical analysis of coal was beyond the scope of this investigation.

$$2{\mathrm{R}\mathrm{E}\mathrm{E}\mathrm{P}\mathrm{O}}_{4\left(\mathrm{s}\right)}+{3\mathrm{C}\mathrm{a}\mathrm{O}}_{\left(\mathrm{s}\right)}\stackrel{\mathrm{C},1000°\mathrm{C}}{\to }{\mathrm{R}\mathrm{E}}_2{\mathrm{O}}_{3\left(\mathrm{s}\right)}+{\mathrm{C}\mathrm{a}}_3{\left({\mathrm{P}\mathrm{O}}_4\right)}_{2\left(\mathrm{s}\right)}$$

(5)

$${\mathrm{Z}\mathrm{r}\mathrm{S}\mathrm{i}\mathrm{O}}_{4\left(\mathrm{s}\right)}+{\mathrm{C}\mathrm{a}\mathrm{O}}_{\left(\mathrm{s}\right)}\stackrel{\mathrm{C},1000°\mathrm{C}}{\to }\mathrm{C}\mathrm{a}\mathrm{Z}\mathrm{r}{\mathrm{O}}_{3\left(\mathrm{s}\right)}+{\mathrm{O}}_{2\left(\mathrm{g}\right)}$$

(6)

$${\mathrm{Z}\mathrm{r}\mathrm{S}\mathrm{i}\mathrm{O}}_{4\left(\mathrm{s}\right)}+{\mathrm{C}\mathrm{a}\mathrm{O}}_{\left(\mathrm{s}\right)}\stackrel{\mathrm{C},1000°\mathrm{C}}{\to }\mathrm{Z}\mathrm{r}{\mathrm{O}}_{2\left(\mathrm{s}\right)}+\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{i}{\mathrm{O}}_{3\left(\mathrm{s}\right)}$$

(7)

3.3.2. Chemical Composition of Sintered CFA

Chemical analysis was conducted to determine the presence of Al, Fe, Ti, and REEs in sintered CFA. For mass balance, the elements of interest were also determined in coal and sintered CFA, and the results are shown in

Table 5. Analysis of the major chemical compositions (wt.%) and REEs (mg/Kg) in coal and sintered CFA.

Analyte	Coal	* RSD	Sintered-CFA	* RSD
SiO2	14.3	0.01	42.33	0.14
Al2O3	1.4	0.05	22.06	0.07
CaO	1.1	0.02	19.46	0.06
FeO	--	--	4.8	0.05
Fe2O3	0.4	0.27	1.20	0.07
TiO2	--	--	1.16	0.01
MgO	0.3	0.00	0.33	0.03
MnO	0.0	0.00	0.25	0.00
Na2O	3.4	0.01	0.33	0.00
K2O	0.1	0.00	0.81	0.01
P2O5	--	--	0.63	0.01
Sc	6.46	0.44	36.5	0.31
Y	21.2	0.11	88.8	0.10
La	20.1	0.57	149	0.03
Ce	32.7	0.21	103	0.02
Pr	4.13	0.33	24.2	0.10
Nd	15.6	0.11	51.3	0.03
Sm	2.91	0.21	14.8	0.84
Eu	0.6	1.50	2.86	0.01
Gd	2.1	0.12	9.97	0.08
Tb	0.59	1.20	2.71	0.01
Dy	3.8	0.13	12.8	0.09
Ho	0.79	0.03	2.99	0.02
Er	1.26	0.05	4.54	0.07
Tm	0.37	0.71	1.21	0.01
Yb	1.57	0.51	5.37	0.07
Lu	0.45	0.07	1.29	0.01
(U)	2.56	0.31	10.9	0.01
(Th)	6.8	0.12	37.2	0.08
ΣLREEs	78.14	--	455.13	--
ΣHREEs	30.03	--	119.71	--
ΣREEs	114.63	--	611.34	--

-- Not available. * Relative standard deviation.

. The chemical composition of the non-MF is reported in Table 3. The results in Table 5 show that coal contains substantial concentrations of REEs (78.14 mg/kg). However, compared to the raw CFA in Table 2, the concentrations in coal were lower. Therefore, CFA is a significant source of REEs. The combination of coal and non-MF in the sintering process provided a total concentration of 455.13 mg/kg REEs, suggesting that both coal and non-MF could be used as resources for the extraction of REEs in sintered CFA. The major oxides quantified in non-MF were also analysed in coal and sintered CFA. This analysis also revealed the elements of interest (i.e., Al, Fe, and Ti) in coal for the set objectives; however, their concentrations were significantly lower compared to those in non-MF. Thus, for the purpose of this study, these elements are regarded as insignificant in coal. In sintered CFA, their concentrations were relatively high, suggesting that non-MF is a significant source of Al, Fe, and Ti.

3.4. Preliminary H₂SO₄ Leaching

Preliminary experiments were carried out to examine the leaching behaviour of Al, Fe, Ti, and selected REEs (Y, La, Ce, Nd, and Sc) in both non-MF and sintered CFA. The leaching behaviour was examined by evaluating the metal extraction efficiency as a function of temperature and lixiviant concentration.

3.4.1. Effect of Leaching Temperature

The effect of leaching temperature on non-MF and sintered CFA was examined, and the results are presented in Figure 9. The leaching tests were conducted using 6 M H₂SO₄ concentration, 1:4 solid-to-liquid ratio, and 10 h residence time. The effect of leaching temperature on the behaviour of Al, Fe, and Ti in non-MF is shown in Figure 9a. The figure shows an upward extraction trend for the three metals, indicating that increasing the temperature enhanced metal dissolution. However, the extraction of the three metals (Al, Fe, and Ti) in non-MF remained low (<30%). This low extraction could be attributed to the fact that most of the Al, Fe, and Ti reside in the refractory phase, which is not amenable to direct acid leaching [7,47].

The effect of leaching temperature on REEs from the non-MF is presented in Figure 9b. The figure shows an upward trend in the extractions beyond 60 °C. Nevertheless, the extraction was not higher than 15% for all the REEs. The low REE extraction could have been due to their association with the aluminosilicate (Al-Si rich) phase. REEs are also associated with phosphate and silicate phases, which are refractory and not easily leachable via direct acid leaching [13]. The effect of leaching temperature on the behaviour of Al, Fe, and Ti in sintered CFA is presented in Figure 9c. The figure shows an upward trend and a marked improvement in extractions of up to 60% for Al. The improved extractions are attributed to the transformation of the refractory mullite phase to the leachable plagioclase/anorthite phase, which is highly soluble in acidic solutions. The effect of leaching temperature on REEs from sintered CFA is also shown in Figure 9d. The figure shows a decline between 45 and 60 °C, an upward trend between 60 °C and 70 °C, and then a slight decline beyond 70 °C. In addition, the REE extraction from the sintered CFA improved to 30% for scandium. Similar to the case of the base metals, the improvement in REE extraction may also be attributed to the transformation of the aluminosilicates and the phosphates/silicates to a more leachable phase. The decline in extractions at 60 °C could be a result of the formation of insoluble calcium sulphate precipitates [48]. Similar findings showing that higher temperatures hinder REE dissolution during sulphuric acid leaching have been reported in literature [49].

3.4.2. Effect of Acid Concentration

The effect of acid concentration was investigated from 0.5 to 6 M H₂SO₄, and the results are presented in Figure 10. Before sintering, as shown in Figure 10a, the highest extraction efficiency for the major elements (Al, Fe, and Ti) was observed at 6 M H₂SO₄, with 25% Al, 19% Fe, and 7% Ti extraction. Sedres [10] also reported low recoveries of Ti and Fe through direct H₂SO₄ leaching at 0.5–8 M H₂SO₄. Increasing the acid concentration to 15 M H₂SO₄ improved Al and Ti extractions to 54% and 58%, respectively. Working at such high acid concentrations allows for the disintegration of the Al-Si-rich phase [11]. However, Nayak and Panda [50] reported that concentrated H₂SO₄ is too viscous, hindering further leaching progress. Consequently, these extreme conditions were not considered further in this study. In the sintered CFA, the metal extraction efficiency significantly improved, as is shown in Figure 10c. The maximum recoveries were observed at 6 M H₂SO₄ with 72.3% Al, 63.4% Fe, and 40.1% Ti extraction. These results are consistent with the findings reported by Matjie et al. [6].

The extraction of REEs is shown in Figure 10b–d. In the sintered CFA shown in Figure 10d, the extraction of the selected REEs increased with increasing H₂SO₄ concentration from 0.5 to 2 M but declined at concentrations above 2 M. Kashiwakura et al. [15] reported similar findings from their work on the extraction of REEs from CFA. The authors reported that REE extraction efficiency decreased with increasing H₂SO₄ concentration. This is in agreement with other researchers who have recommended lower acid concentrations (<2 M H₂SO₄) to achieve high REE extraction efficiency [51,52]. However, operating at low acid concentrations proved challenging in this study due to the formation of silica gel, which made the filtration process difficult. Among the selected REEs investigated in this study, Sc was an exception to the decreasing trend in REE extraction efficiency with increasing acid concentration. The experimental results showed that Sc extraction increased with an increase in H₂SO₄ concentration, similar to the behaviour of the major elements (Al, Fe, and Ti). A possible explanation for the high Sc extraction could be its association with the Al-rich grain (Figure 2) which is disintegrated during sintering and thus resulting in the liberation of Sc. Nonetheless, this interesting departure in Sc behaviour was not explored further in this study. It is recommended that future research examine the impact of CFA phase composition on Sc extraction.

3.4.3. Effect of HCl Leaching

The results in the previous section indicate that sulphuric acid may not be a suitable leaching reagent for the extraction of REEs (except for Sc), as no significant improvement was observed after sintering. This may be attributed to the complex phase composition of CFA [36] the formation of insoluble calcium sulphate compounds [53,54], and silica gel formation at lower acid concentration [55]. Nonetheless, HCl has been reported as the most effective leaching reagent for REE extraction [18,19,55]. Therefore, the effect of HCl leaching was further investigated using sintered CFA to determine whether it would lead to a significant improvement in the extraction of REEs.

The tested factors were 2–6 M HCl concentration and 45–70 °C leaching temperature at 1:10 solid-to-liquid ratio and 6 h residence time. The experimental results presented in Figure 11a show that REE extraction significantly improved with increasing HCl concentration from 2 to 4 M. However, increasing the HCl concentration to 6 M did not result in significant improvement in REE extraction, except for Y, which reached ~60%. As a result, 4 M HCl was chosen as the pertinent acid concentration for all REEs, resulting in 51.3% Ce, 49.7% La, 57.8% Nd, and 34.2% Sc extraction. The improvement in REE extraction aligns with previous research, suggesting that the pre-treatment of CFA can enhance extraction efficiency and that HCl is an effective leaching reagent [19,55,56]. The effect of leaching temperature was also investigated at 45 °C, 60 °C, and 70 °C, with leaching conducted for 6 h in a 4 M HCl solution. The results are graphically reported in Figure 11b. Leaching at 70 °C yielded the highest extraction, with 57.8% Ce, 53.79% La, 51.29% Nd, and 57.01% Y. Chi et al. [53] postulated that in the extraction of REEs with HCl, increasing the leaching temperature enhanced the leaching reaction, resulting in an exothermic reaction. Furthermore, at higher temperatures (60–90 °C), HCl may have a higher diffusion rate due to increased thermal or activation energies. Although the sinter-HCl process yielded higher extraction efficiency, some REEs remained undissolved. This suggests that, in addition to the influence of HCl, the mineralogical distribution of REEs in the sintered CFA still significantly affected their extraction. As stated earlier, in raw CFA, REEs can occur as discrete minerals or be partially and fully integrated with the Al-Si-rich phase. Consequently, during the sintering process, those REEs that are fully integrated into the Al-Si-rich phase may be released from the host (i.e., Al-Si) but do not undergo transformation into a soluble form (i.e., oxides).

3.4.4. Comparison of the Sinter-HCl and Sinter-H₂SO₄ Leaching

Considering the leaching efficiency and selectivity, H₂SO₄ is a suitable lixiviant for the extraction of Al (70%, ~8.27 g/L), Fe (63%, ~1.5 g/L), and Ti (40%, ~0.8 g/L) under maximum conditions. However, the selected REEs (except Sc) showed poor extraction in the sulphate system (15.3% Y, 12.11% La, 13.29% Ce, and 12.05% Nd). A change of lixiviant from H₂SO₄ to HCl increased the extraction of the selected REEs. A graphical comparison of the selected REEs in the sulphate and chloride systems is displayed in Figure 12. In the HCl system, the extraction efficiency at maximum conditions improved from 15.3% to 65.1% Y, 12.11% to 52.2% La, 13.29% to 42.9% Ce, and 12.05% to 56.01% Nd.

It has been established that at high acid concentrations and leaching temperatures, calcium reacts with sulphates to form insoluble calcium sulphate, as shown in Equation (8). The metal sulphates such as REE₂(SO₄)₃, Al₂(SO₄)₃, Fe₂(SO₄)₃, and Ti(OH)₂HSO₄ are, however, soluble in sulphuric acid. The formation of calcium sulphate will consequently precipitate with these metal sulphates, as shown in Equation (9), resulting in their low extraction efficiency [54,55].

$${\mathrm{C}\mathrm{a}\left(\mathrm{O}\mathrm{H}\right)}_2+{2\mathrm{H}}^{+}+{\mathrm{S}\mathrm{O}}_4^{2-}\to {\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4+{2\mathrm{H}}^{+}+{2\mathrm{O}\mathrm{H}}^{-}$$

(8)

$${\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4+{\mathrm{M}}_2{\left({\mathrm{S}\mathrm{O}}_4\right)}_3\to {\mathrm{C}\mathrm{a}\mathrm{S}\mathrm{O}}_4{\mathrm{M}}_2{\left({\mathrm{S}\mathrm{O}}_4\right)}_{3\left(\mathrm{s}\right)}$$

(9)

where M = Al, Fe, Ti, and REEs.

The XRD analysis shown in Figure 13 confirms that calcium sulphate precipitates out of the aqueous solution as anhydrite, as indicated in the spectra. During sulphuric acid leaching, calcium sulphate is formed as anhydrite, which is a secondary phase in both the 2 M and 6 M leached residues. Anhydrite is a form of calcium sulphate (CaSO₄) that does not contain water, distinguishing it from gypsum (CaSO₄·2H₂O), which does. Kashiwakura et al. [15] and Shemi [57] leached CFA using H₂SO₄ and identified the controlling mechanism as diffusion through the product layer. In contrast, leaching in HCl does not result in the formation of calcium sulphate due to the solubility of calcium chloride (CaCl₂). In this case, Cao et al. [18] reported that a chemical reaction at the particle surface is the controlling mechanism.

The morphologies of the sintered CFA leach residues in HCl and H₂SO₄ are also presented in Figure 14. The SEM micrographs of the HCl-leached residue (Figure 14a–b) indicate that the particles were predominantly amorphous. In the H₂SO₄ residue (Figure 14c,d), a mixture of amorphous and needle-like particles was observed. The EDS analysis of the needle-like particles in Figure 14e shows particles with calcium, silica, and sulphur as the major components, thus confirming the possible formation of calcium sulphates. In contrast, the amorphous particle in Figure 14f contained Al, Fe, Ti, Ca, and Si as the main elements. The regions containing REEs could not be quantified on the mounted stubs due to their trace concentrations. Nonetheless, the overall results indicate that the sinter-H₂SO₄ process is not very effective for REE extraction from CFA, whereas the sinter-HCl process is much more suitable. The CFA residue generated can be further studied for REE extraction. Techniques such as dry digestion and high-pressure leaching have been utilised for the treatment of silica- and calcium-rich materials such as red mud and phosphogypsum [16,56].

Purification techniques such as solvent extraction and selective precipitation can be used to separate Al, Fe, Ti, and REEs following the sinter-H₂SO₄ leach process [13,14]. In solvent extraction, acidic extractants such as primary amines (e.g., N1923 and Primene JM-T) and organophosphorus acids (e.g., D2EHPA) have been used to extract REEs from acidic sulphate solutions [58,59]. Selective separation of REEs from major elements (Al, Fe, and Ti) can also be explored.

4. Conclusions

This paper characterised CFA and evaluated the distribution of alumina, iron, titanium, and selected REEs (Y, La, Ce, Nd, and Sc). This was achieved through several analytical techniques, dry magnetic separation, and preliminary acid leaching studies. The main conclusions are as follows:

South African CFA is predominantly amorphous and contains mullite, quartz, and magnetite/hematite as the main crystalline phases. Microanalysis revealed the presence of aluminosilicate (Al-Si-rich), Si-rich, Ca-rich, and Fe-rich grains. Traces of Fe, Ti, Ca, and REEs were relatively dispersed throughout the Al-Si rich phase. The REEs were either partially associated with the Al-Si-rich grains, occluded within grains, or occurred as discrete phosphates and silicates.

Magnetic separation was successfully conducted to yield two fractions: magnetic (MF) and non-magnetic (non-MF). The recovery of discrete Fe-oxide, representing a maximum recovery of 65.9% Fe, was achieved at 1.05 T. Due to the agglomerated CFA particles, the extracted MF also contained impurities such as mullite and quartz. Further analysis showed a low EF (0.69–1.59) value for the REEs, suggesting that a multistage process could be used to pre-concentrate the REEs.

Preliminary leaching tests were conducted to examine the leaching behaviour of Al, Fe, Ti, and selected REEs in a sulphate system using direct acid leaching, which resulted in poor extraction due to the distribution of these elements in the refractory phases (i.e., Al-Si and phosphates). Sintering prior to acid leaching resulted in improved extraction of 77.9% Al, 62.1% Fe, and 52.3% Ti. However, the extraction efficiency of REEs (except Sc with 56.7%) did not improve much, resulting in 23.2% Ce, 27.6% La, 11.3% Nd, and 11.2% Y. A change in the lixiviant from H₂SO₄ to HCl resulted in significant extraction of REEs with 65.1% Y, 52.2% La, 42.9% Ce, and 56.01% Nd. It is therefore postulated that REE extraction in the sulphate system may have been adversely affected by the formation of sulphate precipitates and the REE-bearing grains, such as phosphates that did not transform. Further studies can be conducted on the residue to understand the distribution of REE-bearing grains, allowing for the development of a possible two-stage leaching process.

Although HCl is a better lixiviant for REEs, environmental concerns regarding chloride systems should be considered when evaluating the economic and environmental feasibility of this process. The sulphate system, although not as efficient as the chloride system, could still be a viable option for pre-concentration of REEs. The process will allow the selective separation of REEs from major elements such as Al, Fe, and Ti. Thus, the costs of downstream purification could be reduced.