Key Factors Impacting the Decomposition Rate of REE Silicates During Sulfuric Acid Treatment

Abstract

The decomposition of silicates in sulfuric acid to extract rare earth elements (REE) is typically characterized by the formation of an amorphous silica layer surrounding the receding crystal that may act as a passivation barrier limiting the rate of mineral dissolution. In this context, sulfuric acid treatment experiments coupled with detailed characterization of the evolution of the decomposition reaction were performed on natural allanite (CaREEAl₂Fe²⁺Si₃O₁₁O[OH]), as well as synthetic neodymium disilicate (Nd₂Si₂O₇), orthosilicate (Ca₂Nd₈(SiO₄)₆O₂), and orthophosphate (NdPO₄) phases in order to investigate if there are key factors, operating on a wide range of silicates, that negatively impact REE recovery. While, as expected, the acid strength is the driver in promoting the decomposition of the orthophosphate, for the silicates investigated, no matter their crystalline structure and chemical resistance, there is a severe passivation mechanism at play in concentrated H₂SO₄. However, in all cases, this effect can be minimized by water dilution, which strongly enhances sulfate-forming cation transfer across the produced amorphous silica layer. Taking into consideration this distinct characteristic of the mode of decomposition of silicates in sulfuric acid should help in defining optimal extraction strategies.

1. Introduction

One of the major processes to extract rare earth elements (REE), in particular from the orthophosphates (e.g., monazite, xenotime) and fluorocarbonates (e.g., bastnaesite), is through a sulfuric acid (H₂SO₄) route. In an initial stage, commonly referred to as “acid baking” or “roasting”, mineral decomposition is achieved in concentrated H₂SO₄ (93–98 wt%), at temperatures typically exceeding 200 °C, leading to the formation of water-soluble REE sulfates (sulfation). After this acid treatment step, the REEs can then be recovered during a subsequent low-temperature water leach [1,2]. The use of sulfuric acid is advantageous due to its moderate cost and low vapor pressure, and although its room-temperature viscosity is ≈23.8 mPa⸱s, it drops rapidly to ≈4.1 mPa⸱s at 100 °C. The application of such concentrated sulfuric acid treatment to silicates is of interest [2,3,4] considering that they represent important REE carriers in many deposits, such as peralkaline complexes. However, during decomposition of silicates, the release of sulfate-forming cations, including the REEs, is accompanied by the formation of amorphous silica, which, depending on its nature, can lead to distinct hydrometallurgical challenges. For certain crystalline structures, in particular silicates with a low degree of SiO₄ tetrahedra linkage, the formation of a gel produced from polymerization of dissolved silica [5,6] will result in poor downstream solid–liquid separation (e.g., filtration, sedimentation). Proposed strategies to prevent silica gel formation mainly focus on limiting the water content during acid treatment [7,8], a “dry digestion” process, originally developed for the zinc industry [7], that has been successfully applied to access REEs from eudialyte [9,10,11,12,13]. On the other hand, a more upstream problem, which is the focus of the current study, is that sulfuric acid treatment of many silicate structures will lead to the development of an amorphous silica layer surrounding the residual crystals [6,8] that, under certain conditions, may negatively impact their decomposition rate. Evidence that an amorphous silica layer can act as a passivation barrier limiting metal recovery has been documented, such as, for example, nickel and magnesium from reduction roasted serpentine [14] or copper from chrysocolla [15].

There have been numerous studies on the characteristics of amorphous silica layers produced during natural weathering of silicate minerals and glasses that can be relevant to the mode of decomposition of REE silicates in sulfuric acid. The replacement is often pseudomorphic, meaning that the silica layer closely preserves the morphology and outline of the original silicate [16]. Moreover, although amorphous, at the short-range order, the produced silica phase can inherit characteristics of the parent silicate structure [16,17], especially the degree of polymerization of the SiO₄ tetrahedra (Ref. [18]). Additionally, considering that the molar volume of the amorphous silica layer will typically be lower than that of the parent mineral, it should remain permeable to solutions, as a pseudomorphic replacement will require generation of porosity to account for the volume deficit [19,20]. However, under certain conditions, subsequent pore closure, induced by densification of the amorphous silica, can result in passivation [21,22].

In a previous investigation focusing on allanite-(Ce), a mineral of the epidote group, where the REEs are accommodated through the coupled substitution [Ca²⁺][Fe³⁺] ↔ [REE³⁺][Fe²⁺] resulting in the ideal endmember formula CaREEAl₂Fe²⁺Si₃O₁₁O(OH), we identified important passivation effects after treatment using concentrated H₂SO₄ at temperatures ≥175 °C, where the extent of mineral decomposition was very low and characterized by the formation of a thin continuous amorphous silica layer that seems to act as a barrier preventing efficient exchange between the acidic solution and the crystal dissolution front [23]. In contrast, dilution of the sulfuric acid with water down to 55 wt% drastically enhanced the decomposition rate at temperatures down to 100 °C, suggesting that water addition can strongly minimize the impact of passivation. Nevertheless, the allanite-(Ce) studied showed evidence of partial amorphization due to structural damage associated with prolonged exposure to α-decay from thorium, a process known as metamictization [24]. Consequently, considering that the resulting radiation-induced crystal defects are known to lower the mineral stability [25], the extent of allanite metamictization also contributes to the observed high REE extraction efficiency in diluted sulfuric acid at low temperature (Refs. [26,27,28]).

In this context, the objective of this study was to investigate if there is a common passivation mechanism at play during sulfuric acid treatment that negatively impacts decomposition of a wide range of silicate structures irrespective of their degree of reactivity, and to confirm the role of water dilution in minimizing its effect. A better understanding of the conditions that promote silicate decomposition would provide valuable information to define optimal REE extraction strategies. The approach taken was to perform experiments on natural allanite-(Ce), as well as synthetic REE silicates with distinct degrees of silica polymerization, coupled with detailed characterization of the extent of decomposition observed at the initial acid treatment stage and the resulting REE recovery after the water leach. The advantage of these simpler systems is that variables that can significantly influence the reaction mechanisms, such as metamictization, common not only in allanite but in most REE minerals, as well as the presence of elements that may change the oxidation state, such as Fe and Mn, can be eliminated.

2. Materials and Methods

2.1. Analytical Techniques

2.1.1. X-Ray Diffraction

Powder X-ray diffraction (XRD) analyses were performed to confirm the crystalline structure of the starting materials. The X-ray diffractograms were obtained on finely ground powder (<45 μm) with a Rigaku D/MAX 2500 rotating anode system (Rigaku Americas, The Woodlands, TX, USA) using monochromatic Cu Kα radiation (λ = 0.15406 nm) at 40 kV and 200 mA. Collection was done with a step scan of 0.02° in 2θ and a dwell time of 1 s per step. Phase identification was achieved using JADE version 9 interfaced with the current International Centre for Diffraction Data (ICDD) database.

2.1.2. Electron Probe X-Ray Microanalysis

To characterize the textural and compositional nature of the starting materials and residual products of the acid treatment experiments, secondary (SE) and backscattered electron (BSE) imaging as well as quantitative X-ray microanalyses by wavelength-dispersive spectrometry (WDS), in discrete and mapping modes, were performed on epoxy-impregnated polished cross-sections using a JEOL JXA 8230 Electron Probe Microanalyzer (EPMA; JEOL Canada, Saint-Hubert, QC, Canada) operated with an accelerating voltage of 20 kV and a probe current from 5 to 35 nA. Counting times for discrete analyses ranged from 20 to 50 s, with a beam defocused at 5 to 20 μm to prevent sample damage. Quantitative elemental maps were collected using a focused beam with 0.2 to 1.0 μm steps and dwell time of 35 ms. The characteristic X-ray lines and standards used for the analyses were Al Kα (orthoclase), Mg Kα (forsterite), Si Kα (orthoclase, wollastonite, zircon), Ca Kα (wollastonite), Ti Kα (ilmenite), Mn Kα (rhodocrosite), Fe Kα (hematite), Y Lα (YPO₄), La Lα (LaAlO₃), Ce Lα (Ce₂O₃), Pr Lβ (PrPO₄), Nd Lα and Lβ (NdPO₄), Sm Lβ (SmPO₄), Gd Lβ (GdPO₄), Dy Lβ (DyPO₄), and Th Mα (ThO₂). Matrix corrections were made using the CalcZAF utility [29] interfaced within the “Probe for EPMA v.14” software (Probe Software, Eugene, OR, USA).

2.1.3. Automated Mineralogy

A Tescan Integrated Mineral Analyzer (Tescan USA, Warrendale, PA, USA) equipped with four silicon drift X-ray energy-dispersive spectrometers (EDS) was used to obtain particle size distributions and modal phase abundances within the starting materials. Data were collected on the polished cross-sections using an accelerating voltage of 25 kV and a beam current of 5.5 nA with a step size 3 μm. The BSE signal and EDS spectra were acquired simultaneously at all steps and were both used to discriminate phases based on their compositional differences.

2.1.4. Solution Chemistry

The chemical composition of all recovered solutions was measured by inductively coupled plasma optical emission spectroscopy (ICP-OES) using an Agilent 5110 Synchronous Dual View instrument (Agilent Technologies Canada, Mississauga, ON, Canada) with traceable standards. Dilution to bring the concentration within the linear calibrated range was done using a 2% vol HNO₃ aqueous solution.

2.2. Materials

2.2.1. Pristine Allanite-(Ce)

Centimeter-size allanite-(Ce) crystals from the Burroughs pegmatite, Colorado, were ground to a +45 µm/−300 µm size fraction (P₈₀: 202 µm; P₂₀: 91 µm) and concentrated by elutriation to a modal abundance of 94 wt%, associated with minor chlorite, apatite, quartz, and thorite. The peaks characteristic of allanite, observed in the XRD pattern (Figure 1), are well-developed, although broad, and associated with an amorphous metamict component, consistent with concentrations of around 1.3 wt% ThO₂ (Table 1). The chemical composition, as determined by EPMA (Table 1), is consistent with a classification as allanite-(Ce) [30], while the low oxide total, likely reflecting a significant amount of incorporated H₂O, supports that the mineral is indeed partially metamict [31].

2.2.2. Heat-Treated Allanite-(Ce)

A portion of the allanite-(Ce) feed (see Section 2.2.1), loaded in a Pt boat, was heat treated at 610 °C for 2 h in a horizontal tube furnace. At these conditions, allanite, CaREEAl₂Fe²⁺Si₃O₁₁O(OH), should be partially transformed to oxyallanite with ideal endmember formula CaREEAl₂Fe³⁺Si₃O₁₁O(O), through the coupled substitution [Fe²⁺][OH⁻] ↔ [Fe³⁺][O²⁻], resulting from the iron oxidation charge balanced by dehydrogenation of the hydroxide anion [32,33]. In the XRD pattern shown in Figure 1, the shift of the diffraction peaks to higher 2θ compared to the pristine allanite-(Ce) is consistent with a unit cell contraction that would be expected from this process. However, the narrowing of the peaks and their higher intensities suggest that annealing of the thorium-induced defects, through recrystallization of the metamict allanite-(Ce), is also contributing to the transformation [31,34]. This is further supported by an increase in the oxide total of the heat-treated allanite-(Ce) chemical composition, as determined by EPMA (Table 1), that would result from the loss of excess H₂O. Although heat treatment at higher temperature may have resulted in a higher degree of annealing, a moderate temperature (610 °C) was chosen in order to minimize the potential increase in surface area that would impact reactivity. As mentioned by Reissner et al. [34], interior stress caused by heterogeneous crystallization and hydrogen loss can lead to internal and surface cracking.

Table 1. Mean chemical compositions, obtained by WDS-EPMA, and calculated mineral formulae of the allanite-(Ce) in its pristine form and after heat treatment at 610 °C.

	Allanite-(Ce)
	Pristine		Heat-Treated (610 °C)
	Mean (N = 30)	1 σ *	Mean (N = 25)	1 σ
Oxides (wt%)
SiO2	30.37	0.52	30.82	0.47
TiO2	0.35	0.07	0.39	0.05
Al2O3	14.64	0.93	14.73	0.23
Fe2O3 **	14.80	1.14	15.66	1.24
MgO	0.21	0.11	0.18	0.09
MnO	1.56	0.58	1.53	0.59
CaO	9.68	1.08	10.06	0.66
La2O3	4.48	0.37	4.60	0.25
Ce2O3	10.82	0.84	11.23	0.31
Pr2O3	1.24	0.11	1.27	0.05
Nd2O3	4.23	0.33	4.41	0.12
Sm2O3	0.66	0.06	0.69	0.03
Gd2O3	0.35	0.03	0.36	0.02
Dy2O3	0.13	0.03	0.14	0.03
Y2O3	0.49	0.09	0.49	0.04
ThO2	1.32	0.23	1.38	0.24
Total	95.33		97.93
apfu ***
Si	3.054		3.029
Ti	0.027		0.029
Al	1.735		1.706
Fe3+ ****	0.218		1.158
Fe2+ ****	0.902		0.000
Mg	0.032		0.026
Mn3+ ****	0.000		0.125
Mn2+ ****	0.133		0.003
Ca	1.043		1.059
La	0.166		0.167
Ce	0.398		0.404
Pr	0.046		0.045
Nd	0.152		0.155
Sm	0.023		0.023
Gd	0.012		0.012
Dy	0.004		0.004
Y	0.026		0.026
Th	0.030		0.031
Σcations ***	8.000		8.000
O ****	12.5		13.0

Note: * number of analyses (N) and standard deviation (1 σ); ** total Fe in wt% expressed as Fe₂O₃; *** atoms per formula unit (apfu) normalized on a total of 8 cations [30]; **** Fe³⁺/Fe²⁺ and Mn³⁺/Mn²⁺ in the formulae have been adjusted to balance the charge of the cations on the basis of 12.5 and 13 oxygens (O) for the pristine and heat-treated allanite-(Ce), respectively.

Table 1. Mean chemical compositions, obtained by WDS-EPMA, and calculated mineral formulae of the allanite-(Ce) in its pristine form and after heat treatment at 610 °C.

	Allanite-(Ce)
	Pristine		Heat-Treated (610 °C)
	Mean (N = 30)	1 σ *	Mean (N = 25)	1 σ
Oxides (wt%)
SiO2	30.37	0.52	30.82	0.47
TiO2	0.35	0.07	0.39	0.05
Al2O3	14.64	0.93	14.73	0.23
Fe2O3 **	14.80	1.14	15.66	1.24
MgO	0.21	0.11	0.18	0.09
MnO	1.56	0.58	1.53	0.59
CaO	9.68	1.08	10.06	0.66
La2O3	4.48	0.37	4.60	0.25
Ce2O3	10.82	0.84	11.23	0.31
Pr2O3	1.24	0.11	1.27	0.05
Nd2O3	4.23	0.33	4.41	0.12
Sm2O3	0.66	0.06	0.69	0.03
Gd2O3	0.35	0.03	0.36	0.02
Dy2O3	0.13	0.03	0.14	0.03
Y2O3	0.49	0.09	0.49	0.04
ThO2	1.32	0.23	1.38	0.24
Total	95.33		97.93
apfu ***
Si	3.054		3.029
Ti	0.027		0.029
Al	1.735		1.706
Fe3+ ****	0.218		1.158
Fe2+ ****	0.902		0.000
Mg	0.032		0.026
Mn3+ ****	0.000		0.125
Mn2+ ****	0.133		0.003
Ca	1.043		1.059
La	0.166		0.167
Ce	0.398		0.404
Pr	0.046		0.045
Nd	0.152		0.155
Sm	0.023		0.023
Gd	0.012		0.012
Dy	0.004		0.004
Y	0.026		0.026
Th	0.030		0.031
Σcations ***	8.000		8.000
O ****	12.5		13.0

Note: * number of analyses (N) and standard deviation (1 σ); ** total Fe in wt% expressed as Fe₂O₃; *** atoms per formula unit (apfu) normalized on a total of 8 cations [30]; **** Fe³⁺/Fe²⁺ and Mn³⁺/Mn²⁺ in the formulae have been adjusted to balance the charge of the cations on the basis of 12.5 and 13 oxygens (O) for the pristine and heat-treated allanite-(Ce), respectively.

2.2.3. Synthetic REE Silicates and Phosphates

Neodymium disilicate and orthosilicate phases, with Nd₂Si₂O₇ and Ca₂Nd₈(SiO₄)₆O₂ compositions respectively (Table 2), were synthesized using a solid-state route. Appropriate stoichiometric proportions of calcium carbonate (CaCO₃), neodymium nitrate hexahydrate (Nd₂[NO₃]₃∙6H₂O), and silica glass (SiO₂), all with purity of ≥99.9%, were mixed and pressed into pellets that were loaded in Pt crucibles for an initial firing at 1375 °C for 24 h. The recovered material was finely ground with a micronizing mill, re-pelletized and fired for 72 h at 1475 °C to produce single-phase polycrystalline disks. High-contrast BSE imaging reveals small equigranular crystals (≈2 to 20 μm) for the disilicate (Figure 2a) and larger tabular crystals ranging up to ≈500 μm in size in the case of the orthosilicate (Figure 2b). The disks were then lightly ground and sieved to a +45/−500 μm size fraction leading to particle size distributions characterized by a P₈₀/P₂₀ of 287/90 μm for the disilicate and 234/103 μm for the orthosilicate. The XRD patterns shown in Figure 3 indicate that both materials are indeed single-phase with the expected structures, tetragonal P4₁ for Nd₂Si₂O₇ and hexagonal P6₃/m for Ca₂Nd₈(SiO₄)₆O₂.

Table 2. Mean chemical compositions, obtained by WDS-EPMA, and calculated mineral formulae of the synthetic silicates and phosphates.

	Nd-Disilicate		Ca-Nd-Orthosilicate		Nd-Orthophosphate
	N = 20	1 σ *	N = 20	1 σ	N = 20	1 σ
Oxides (wt%)
SiO2	26.35	0.10	19.75	0.06
CaO			6.25	0.03
Nd2O3	73.23	0.35	73.51	0.28	70.17	0.24
Gd2O3
Y2O3
P2O5					29.83	0.24
Total	99.58		99.50		100.00
apfu **
Si	2.006		6.000
Ca			2.033
Nd	1.992		7.977		0.995
Gd
Y
P					1.003
O *	7.000		26.000		4.000

Note: * number of analyses (N) and standard deviation (1 σ); ** atoms per formula unit (apfu) normalized on the number of oxygens specified as O *.

Table 2. Mean chemical compositions, obtained by WDS-EPMA, and calculated mineral formulae of the synthetic silicates and phosphates.

	Nd-Disilicate		Ca-Nd-Orthosilicate		Nd-Orthophosphate
	N = 20	1 σ *	N = 20	1 σ	N = 20	1 σ
Oxides (wt%)
SiO2	26.35	0.10	19.75	0.06
CaO			6.25	0.03
Nd2O3	73.23	0.35	73.51	0.28	70.17	0.24
Gd2O3
Y2O3
P2O5					29.83	0.24
Total	99.58		99.50		100.00
apfu **
Si	2.006		6.000
Ca			2.033
Nd	1.992		7.977		0.995
Gd
Y
P					1.003
O *	7.000		26.000		4.000

Note: * number of analyses (N) and standard deviation (1 σ); ** atoms per formula unit (apfu) normalized on the number of oxygens specified as O *.

Figure 2. High-contrast BSE images obtained on the synthesized (a) polycrystalline Nd₂Si₂O₇ disilicate, and (b) Ca₂Nd₈(SiO₄)₆O₂ orthosilicate.

Figure 2. High-contrast BSE images obtained on the synthesized (a) polycrystalline Nd₂Si₂O₇ disilicate, and (b) Ca₂Nd₈(SiO₄)₆O₂ orthosilicate.

Figure 3. Powder-XRD patterns collected on the synthesized (a) Nd₂Si₂O₇ disilicate, and (b) Ca₂Nd₈(SiO₄)₆O₂ orthosilicate starting materials. Simulated ICDD reference spectra for tetragonal Nd₂Si₂O₇ (04-015-1535) and hexagonal Ca₂Nd₈(SiO₄)₆O₂ (04-007-5968) are shown for comparison.

Figure 3. Powder-XRD patterns collected on the synthesized (a) Nd₂Si₂O₇ disilicate, and (b) Ca₂Nd₈(SiO₄)₆O₂ orthosilicate starting materials. Simulated ICDD reference spectra for tetragonal Nd₂Si₂O₇ (04-015-1535) and hexagonal Ca₂Nd₈(SiO₄)₆O₂ (04-007-5968) are shown for comparison.

Synthetic crystals of monoclinic neodymium orthophosphate (NdPO₄) ranging in size from 10s of micrometers up to 1 mm were produced by a flux method adapted from the approach used by Talla [35] and Lenz et al. [36]. A mixture of sodium trimetaphosphate (Na₃P₃O₉) and the neodymium nitrate hexahydrates (≥99.9% purity), keeping a molar Na:REE proportion of 88:12, was loaded in a Pt crucible and heated to 1170 °C. To promote crystal growth, the temperature was then slowly decreased down to 900 °C, at a rate of 1 °C/h, after which the furnace was shut off and left to cool. As the coexisting sodium phosphate phase is water-soluble, the orthophosphate crystals could be recovered through rinsing (Figure 4a). After grinding the crystals to less than 75 μm, they were pelletized and fired at 1525 °C for 72 h to obtain sintered polycrystalline disks, which were then lightly ground and sieved to a +45/−500 μm size fraction (P₈₀: 302 µm; P₂₀: 93 µm). The crystal structure as single-phase monazite analogue was confirmed by XRD analysis (Figure 4b), with a composition, determined by WDS-EPMA, consistent with NdPO₄ stoichiometry (Table 2).

2.3. Sulfuric Acid Treatment and Water Leach

Details of the acid treatment conditions for all the experiments conducted in this study can be found in Table 3. ACS-grade H₂SO₄ and deionized water were used to prepare the reagent, the characteristics of which are defined by two parameters: the concentration of H₂SO₄ added relative to the feed (H₂SO₄: feed weight ratio) and the extent of dilution with water, expressed as the relative weight percent of sulfuric acid in solution (H₂SO₄^x%). Considering that the main objective was to investigate if there is a common passivation mechanism at play during sulfuric acid treatment that negatively impacts the decomposition of silicates, the concentration of H₂SO₄ was kept in significant excess of stoichiometric requirements to ensure that acid deficiency was not a variable. The solid feed and the appropriate amount of reagent were contained in a fused quartz (SiO₂) crucible covered with a lid of similar material. All experiments were conducted isothermally, where the load was directly exposed to a peak temperature of 90 or 120 °C for durations ranging from 90 to 120 min. The acid treatments at 90 °C were performed in a drying oven, where the capped crucible was fixed to a clamp held horizontally at 10 cm of a vertical stirrer rotating at 40 rpm. At 120 °C, the loaded crucible was inserted into a pre-heated oil bath over a hot plate using magnetic stirring at 250 rpm. The temperature was continuously monitored with a K-type thermocouple held at sample height in the drying oven or inserted directly in the oil bath. At the end of all acid treatments, the crucible was recovered and left to cool in a desiccator for 10 min.

For some experiments, in order to better visualize the evolution of the decomposition reaction, the cooled product was thoroughly rinsed in isopropanol through vacuum filtration (0.45 μm PTFE membrane) allowing for preservation of the water-soluble sulfates. In all other cases, after cooling, DI water was added to the residues of the acid treatment at a liquid to solid weight ratio of 10 to 1. The resulting solution was then stirred at room temperature, with the crucible covered, for durations of 60 to 90 min, after which it was vacuum-filtered using a 0.45 μm PTFE membrane. The filtrate was stabilized with 5 vol% of a 2 vol% HNO₃ solution and kept for bulk chemical analysis. The solid residues recovered from the isopropanol rinsing or the water leaching were vacuum-impregnated with epoxy resin in 2.5-cm cups, cured at room temperature, then polished on diamond lapping film using isopropanol or water as a lubricant, respectively.

Table 3. Experimental conditions for the acid treatments conducted in this study.

Feed	Experiment #	Acid Treatment				Water Leach *
		H2SO4 conc. (wt%)	H2SO4:feed (wt ratio)	T (°C)	Duration (min)	Duration (min)
Pristine allanite-(Ce)	Al-PR-55-1	55	1.5:1	90	120	90
	Al-PR-55-2	55	1.5:1	90	120	none **
	Al-PR-75-1	75	1.5:1	90	120	90
	Al-PR-97-1	97	1.5:1	90	120	90
Heat-treated allanite-(Ce)	Al-HT-55-1	55	1.5:1	90	120	90
Nd2Si2O7	NdSi-55-1	55	2.5:1	90	120	none **
	NdSi-55-2	55	2.5:1	90	120	90
	NdSi-55-3	55	2.5:1	120	90	60
	NdSi-75-1	75	2.5:1	120	90	60
	NdSi-97-1	97	2.5:1	120	90	60
Ca2Nd8(SiO4)6O2	CaNdSi-55-1	55	2.5:1	90	120	none **
	CaNdSi-55-2	55	2.5:1	90	120	90
	CaNdSi-55-3	55	2.5:1	120	90	60
	CaNdSi-75-1	75	2.5:1	120	90	60
	CaNdSi-97-1	97	2.5:1	120	90	60
	CaNdSi-97-2	97	2.5:1	120	90	none **
NdPO4	NdP-55-1	55	2.5:1	120	90	60
	NdP-75-1	75	2.5:1	120	90	60
	NdP-97-1	97	2.5:1	120	90	none **
	NdP-97-2	97	2.5:1	120	90	60

Note: # refers to the experiment number; * water leach always performed at a 10:1 liquid:solid wt ratio; ** “none” refers to experiments where the acid treatment product was rinsed and polished with isopropanol to preserve the sulfates instead of being water leached.

Table 3. Experimental conditions for the acid treatments conducted in this study.

Feed	Experiment #	Acid Treatment				Water Leach *
		H2SO4 conc. (wt%)	H2SO4:feed (wt ratio)	T (°C)	Duration (min)	Duration (min)
Pristine allanite-(Ce)	Al-PR-55-1	55	1.5:1	90	120	90
	Al-PR-55-2	55	1.5:1	90	120	none **
	Al-PR-75-1	75	1.5:1	90	120	90
	Al-PR-97-1	97	1.5:1	90	120	90
Heat-treated allanite-(Ce)	Al-HT-55-1	55	1.5:1	90	120	90
Nd2Si2O7	NdSi-55-1	55	2.5:1	90	120	none **
	NdSi-55-2	55	2.5:1	90	120	90
	NdSi-55-3	55	2.5:1	120	90	60
	NdSi-75-1	75	2.5:1	120	90	60
	NdSi-97-1	97	2.5:1	120	90	60
Ca2Nd8(SiO4)6O2	CaNdSi-55-1	55	2.5:1	90	120	none **
	CaNdSi-55-2	55	2.5:1	90	120	90
	CaNdSi-55-3	55	2.5:1	120	90	60
	CaNdSi-75-1	75	2.5:1	120	90	60
	CaNdSi-97-1	97	2.5:1	120	90	60
	CaNdSi-97-2	97	2.5:1	120	90	none **
NdPO4	NdP-55-1	55	2.5:1	120	90	60
	NdP-75-1	75	2.5:1	120	90	60
	NdP-97-1	97	2.5:1	120	90	none **
	NdP-97-2	97	2.5:1	120	90	60

Note: # refers to the experiment number; * water leach always performed at a 10:1 liquid:solid wt ratio; ** “none” refers to experiments where the acid treatment product was rinsed and polished with isopropanol to preserve the sulfates instead of being water leached.

3. Results and Discussion

3.1. Acid Treatment of Allanite-(Ce)

The acid treatment experiments on the pristine allanite-(Ce) feed were all performed at 90 °C using either the H₂SO₄^97%, H₂SO₄^75%, or the H₂SO₄^55% acid solutions (Al-PR-97-1, Al-PR-75-1 and Al-PR-55-1 in Table 3). BSE imaging of the solid reacted products recovered after the water leach (Figure 5) reveals a sharp contrast in the extent of decomposition. Whereas the vast majority of allanite-(Ce) grains are unreacted after treatment with concentrated H₂SO₄^97% (Al_-PR-97-1), a progressive replacement by amorphous silica is observed with increasing acid dilution, with only minor residual allanite-(Ce) cores remaining at H₂SO₄^55% (Al-PR-55-1). This strong variation in the decomposition rate evidently has a major impact on REE recovery, which ranges from ≈5 wt% for H₂SO₄^97% up to ≈80 wt% with H₂SO₄^55% (Figure 6). These results are consistent with important passivation effects limiting the decomposition of allanite-(Ce) in concentrated sulfuric acid and the role of water addition in minimizing its impact [23].

To better visualize the textural relationship between the amorphous silica layer and the water-soluble sulfates produced during mineral decomposition, quantitative elemental mapping by WDS-EPMA around residual allanite cores was performed on the product directly recovered after a H₂SO₄^55% acid treatment and preserved through rinsing and polishing with isopropanol (Al-PR-55-2). The efficient crystallization of sulfates against the outer rim of a well-developed silica layer (Figure 7) confirms that, at this level of water dilution, efficient transfer between the bulk acidic solution and the crystal dissolution front is maintained.

Figure 5. BSE images obtained on polished cross-sections of the solid products recovered after the water leach from the acid treatment experiments performed at 90 °C on the pristine and heat-treated allanite-(Ce). Detailed conditions for these experiments can be found in Table 3 as Al-PR-97-1, Al-PR-75-1, Al-PR-55-1, and Al-HT-55-1.

Figure 5. BSE images obtained on polished cross-sections of the solid products recovered after the water leach from the acid treatment experiments performed at 90 °C on the pristine and heat-treated allanite-(Ce). Detailed conditions for these experiments can be found in Table 3 as Al-PR-97-1, Al-PR-75-1, Al-PR-55-1, and Al-HT-55-1.

Figure 6. REE (Ce, Nd, Y) recoveries obtained after the water leach from the acid treatment experiments performed at 90 °C on the pristine and heat-treated allanite-(Ce). The REE recoveries in wt% were calculated following the expression: (REE_solution mass_solution)/(REE_feed mass_feed) 100.

Figure 6. REE (Ce, Nd, Y) recoveries obtained after the water leach from the acid treatment experiments performed at 90 °C on the pristine and heat-treated allanite-(Ce). The REE recoveries in wt% were calculated following the expression: (REE_solution mass_solution)/(REE_feed mass_feed) 100.

Figure 7. BSE image and quantitative WDS-based elemental maps across a partially reacted allanite-(Ce) grain in the solid residue recovered from an acid treatment experiment conducted on the pristine feed using H₂SO₄^55%, and rinsed in isopropanol to preserve the produced sulfates (Al-PR-55-2; Table 3). The elemental abundances in the quantitative maps are expressed as oxide wt% (SiO₂, SO₃, Al₂O₃, CaO, Ce₂O₃).

Figure 7. BSE image and quantitative WDS-based elemental maps across a partially reacted allanite-(Ce) grain in the solid residue recovered from an acid treatment experiment conducted on the pristine feed using H₂SO₄^55%, and rinsed in isopropanol to preserve the produced sulfates (Al-PR-55-2; Table 3). The elemental abundances in the quantitative maps are expressed as oxide wt% (SiO₂, SO₃, Al₂O₃, CaO, Ce₂O₃).

The metamict nature of the allanite-(Ce) suggests that contribution of the actinide-induced defects may also influence the extent of its decomposition [26,27,28]. Clearly, comparing the BSE images (Figure 5) of the product recovered after the water leach for H₂SO₄^55% experiments performed with the heat-treated (Al-HT-55-1) and pristine (Al-PR-55-1) allanite-(Ce) indicates that recrystallization has rendered the mineral significantly more refractory. Nevertheless, the detrimental impact of the heat treatment on REE recovery (Figure 6) when reacted at optimal acid dilution (H₂SO₄^55%) is significantly less than that of the passivation induced by the silica layer in concentrated H₂SO₄ (H₂SO₄^97%).

3.2. Acid Treatment of Neodymium Disilicate and Orthosilicate

Although having different crystal structures, which will impact their relative reactivity, the REE silicates that were synthesized present some analogy with allanite when visualized in terms of polymerization of the SiO₄ tetrahedra (Figure 8), a property that can impact the nature of the produced amorphous silica layer [6,16,18]. Monoclinic allanite is characterized by two types of tetrahedral groups, pairs of tetrahedra bridged by a single apical oxygen forming [Si₂O₇]⁻⁶ dimers, as well as isolated [SiO₄]⁻⁴ units (Figure 8a). On the other hand, the synthetic tetragonal Nd₂Si₂O₇ is a true disilicate consisting only of [Si₂O₇]⁻⁶ groups associated with Nd in higher coordination (Figure 8b). Finally, hexagonal Ca₂Nd₈(SiO₄)₆O₂, a synthetic analogue of the mineral britholite, Ca₄REE₆(SiO₄)₆(OH)₂, with the exception that the hydroxide anions are replaced by oxygen through the mass-balanced substitution: [2 Ca²⁺ + 2 OH⁻] ↔ [2 REE³⁺ + 2 O²⁻], is an orthosilicate where there is no direct linkage between each isolated [SiO₄]⁻⁴ tetrahedra (Figure 8c). Simplified decomposition reactions for these synthetic neodymium silicates in sulfuric acid can be expressed as:

Nd₂Si₂O₇ + 3 H₂SO₄ → Nd₂(SO₄)₃ + 2 SiO_2(amorphous) + 3 H₂O.

(1)

Ca₂Nd₈(SiO₄)₆O₂ + 14 H₂SO₄ → 2 CaSO₄ + 4 Nd₂(SO₄)₃ + 6 SiO_2(amorphous) + 14 H₂O.

(2)

3.2.1. Neodymium Disilicate: Nd₂Si₂O₇

In an acid treatment with the H₂SO₄^55% solution at 90 °C, the mode of decomposition of the neodymium disilicate shows similarities with that of allanite-(Ce), where a well-defined amorphous silica layer formed around the residual Nd₂Si₂O₇ particle is associated with outwards crystallization of neodymium sulfate that can be observed in the residue preserved with isopropanol (NdSi-55-1; Figure 9). However, considering the polycrystalline nature of the starting material (Figure 2a), the Nd₂Si₂O₇-silica interface is irregular, due to pseudomorphic replacement of the individual µm-sized crystals, which even preserves the original porosity (Figure 10). At these conditions, only 18 wt% of neodymium was recovered after the water leach (NdSi-55-2), increasing up to 46% at 120 °C (NdSi-55-3), indicating that the Nd₂Si₂O₇ is significantly more refractory than allanite-(Ce). Nevertheless, similar to what was observed for allanite-(Ce), there is a marked decrease in the extent of decomposition (Figure 11) and REE recovery (Figure 12a) with increasing acid concentration (H₂SO₄^55% → H₂SO₄^75% → H₂SO₄^97%) for experiments performed on the Nd₂Si₂O₇ feed at 120 °C (NdSi-55-3, NdSi-75-1, NdSi-97-1). This confirms once again the severity of silica passivation in concentrated H₂SO₄, the effect of which on recovery loss is more important than a reduction in temperature from 120 °C down to 90 °C at diluted acid concentrations (H₂SO₄^55%).

3.2.2. Neodymium Orthosilicate: Ca₂Nd₈(SiO₄)₆O₂

The complete depolymerization of the SiO₄ tetrahedra in orthosilicates is expected to significantly influence the nature or even the ability to maintain the growth of an amorphous silica layer during decomposition in acidic media [6,8]. In this context, characterization of the isopropanol-preserved product for an acid treatment performed on the synthetic neodymium orthosilicate at 90 °C with the H₂SO₄^55% solution (CaNdSi-55-1) reveals a distinct evolution of the decomposition reaction, where the Ca₂Nd₈(SiO₄)₆O₂ particles are progressively replaced by a tight intergrowth of sulfate and silica (Figure 13). The outline of this replacement layer seems to preserve the morphology of the original orthosilicate particle, against which well-developed neodymium and calcium sulfate crystals can be observed. After an acid treatment at similar conditions followed by a water leach that dissolves the sulfates (CaNdSi-55-2), an underlying amorphous silica framework representing the relic of the replacement layer can be identified (Figure 14), although the higher SiO₂ concentration suggests its densification during leaching. With such diluted acidic conditions (H₂SO₄^55%), neodymium recoveries from 54 up to 98 wt% (Figure 12a) can be achieved at temperatures of 90 and 120 °C, respectively (CaNdSi-55-2, CaNdSi-55-3). On the other hand, in progressively more concentrated acid (CaNdSi-75-1, CaNdSi-97-1) the neodymium recovery eventually drops to 21 wt% at H₂SO₄^97% (Figure 12a).

Figure 13. BSE image and quantitative WDS-based elemental maps across a partially reacted Ca₂Nd₈(SiO₄)₆O₂ grain in the solid residue recovered from an acid treatment experiment using H₂SO₄^55%, and rinsed in isopropanol to preserve the produced sulfates (CaNdSi-55-1; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (SiO₂, Nd₂O₃, SO₃). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

Figure 13. BSE image and quantitative WDS-based elemental maps across a partially reacted Ca₂Nd₈(SiO₄)₆O₂ grain in the solid residue recovered from an acid treatment experiment using H₂SO₄^55%, and rinsed in isopropanol to preserve the produced sulfates (CaNdSi-55-1; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (SiO₂, Nd₂O₃, SO₃). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

Figure 14. BSE image and quantitative WDS_-based elemental maps across a partially reacted Ca₂Nd₈(SiO₄)₆O₂ grain in the solid residue recovered from the water leach after an acid treatment experiment using H₂SO₄^55% (CaNdSi-55-2; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (Nd₂O₃, SiO₂). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

Figure 14. BSE image and quantitative WDS_-based elemental maps across a partially reacted Ca₂Nd₈(SiO₄)₆O₂ grain in the solid residue recovered from the water leach after an acid treatment experiment using H₂SO₄^55% (CaNdSi-55-2; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (Nd₂O₃, SiO₂). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

Characterization of the distribution of silicon, neodymium, and sulfur by quantitative WDS-EPMA mapping across a residual Ca₂Nd₈(SiO₄)₆O₂ particle from the isopropanol-preserved product of an acid treatment performed with the concentrated H₂SO₄^97% solution (CaNdSi-97-2) is shown in Figure 15. Only minor neodymium sulfate crystallization can be identified, suggesting that the decomposition reaction was initiated, but that the growth of a narrow continuous silica layer was sufficient to halt its progression.

Figure 15. BSE image and quantitative WDS-based elemental maps across a partially reacted Ca₂Nd₈(SiO₄)₆O₂ grain in the solid residue recovered from an acid treatment experiment using H₂SO₄^97%, and rinsed in isopropanol to preserve the produced sulfates (CaNdSi-97-2; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (SiO₂, Nd₂O₃, SO₃). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

Figure 15. BSE image and quantitative WDS-based elemental maps across a partially reacted Ca₂Nd₈(SiO₄)₆O₂ grain in the solid residue recovered from an acid treatment experiment using H₂SO₄^97%, and rinsed in isopropanol to preserve the produced sulfates (CaNdSi-97-2; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (SiO₂, Nd₂O₃, SO₃). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

These observations strongly suggest that, even in fully depolymerized silicate structures, a silica layer can be maintained and can still induce passivation during acid treatment in concentrated H₂SO₄. Similar to the trend observed for the disilicate, the negative impact on Nd recovery resulting from an increase in H₂SO₄ concentration to 97% is significantly more pronounced than a reduction in temperature by 30 °C (120 → 90 °C) in diluted H₂SO₄^55% acidic solution. However, in the case of Ca₂Nd₈(SiO₄)₆O₂, by inheriting, at the short-range order, the depolymerized SiO₄ arrangement from the parent orthosilicate, the amorphous silica layer may grow as a more open framework where the sulfates can crystallize. Consequently, the observed pseudomorphic replacement is in the form of a tight intergrowth that can potentially act as an efficient physical barrier.

3.3. Acid Treatment of Neodymium Orthophosphate

Similar to the [SiO₄] group in the orthosilicates, the [PO₄] tetrahedra in the orthophosphate are isolated, sharing each apical oxygen with REEs (Nd) in ninefold coordination (Figure 8d). However, as P₂O₅ is known to react with sulfuric acid to produce phosphoric acid, no amorphous layer that can induce passivation should be produced during monoclinic NdPO₄ decomposition, a synthetic analogue of monazite [1,2]:

2 NdPO₄ + 3 H₂SO₄ → Nd₂(SO₄)₃ + 2 H₃PO_4(soluble).

(3)

Consequently, contrary to what was observed for the silicates, concentrated sulfuric acid (H₂SO₄^97%) promotes decomposition leading to significantly improved recovery (Figure 12b), considering that neodymium sulfate can crystallize freely around the receding NdPO₄ particle, as the dissolution front remains in direct contact with the bulk acidic solution (Figure 16).

Figure 16. BSE image and quantitative WDS-based elemental maps across a partially reacted NdPO₄ grain in the solid residue recovered from an acid treatment experiment using H₂SO₄^97%, and rinsed in isopropanol to preserve the produced sulfates (NdP-97-1; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (P₂O₅, Nd₂O₃, SO₃). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

Figure 16. BSE image and quantitative WDS-based elemental maps across a partially reacted NdPO₄ grain in the solid residue recovered from an acid treatment experiment using H₂SO₄^97%, and rinsed in isopropanol to preserve the produced sulfates (NdP-97-1; Table 3). The elemental abundances in the quantitative maps are expressed as wt% of the oxides (P₂O₅, Nd₂O₃, SO₃). To facilitate comparison, the concentration scale for each specific oxide is kept within the same range in Figure 9, Figure 13, Figure 14, Figure 15 and Figure 16.

4. Implications and Conclusions

The results obtained from this study provide a number of constraints about the mode of decomposition of silicates in sulfuric acid:

• Although the acid strength is expected to be the driver in promoting mineral decomposition, as clearly observed in NdPO4 (Section 3.3), for the silicates investigated (Section 3.1 and Section 3.2), no matter their crystalline structure and their relative chemical resistance, we observe an inverse trend with a sharp decrease in reactivity with the strongest acidic solution (H2SO497%). Consequently, there is a passivation mechanism at play that prevents efficient interaction between sulfuric acid and silicates which effect can be strongly minimized by water dilution.
• During decomposition, there is evidence of pseudomorphic replacement by a silica layer, which can even be maintained in the case of the most depolymerized silicate structure (Figure 13) where it is tightly intergrown with the produced sulfates.
• In all the H2SO455% acid treatment products preserved in isopropanol, the sulfates can be observed, often with well-defined crystalline shapes on the outer rim of the produced silica layer. This implies efficient transfer of the sulfate-forming cations across the silica layer, although their solubility limit has been exceeded in the acidic solution, even at this level of water dilution.

In this context, as proposed in our previous study on allanite-(Ce) [23], a passivation mechanism governed by the decrease in solubility of the produced sulfates with increasing acid concentration due to the “common ion effect” [37,38,39,40] may be applicable to a wide range of silicates hosting REE or any other critical elements. In essence, in concentrated H₂SO₄, as sulfate saturation is rapidly reached at the crystal dissolution front, the presence of a silica layer, even if permeable to the solution, will act as a physical barrier that prevents efficient cation transfer leading to a slow decomposition rate. In contrast, with increasing dilution, H₂SO₄ consumption at the crystal dissolution front will raise the water concentration, and hence the solubility of the sulfate-forming cations, allowing their efficient transport across the permeable silica layer until they finally crystallize in the more acidic bulk solution.

This distinct characteristic of silicates, where a compromise needs to be made between the acid contribution required to decompose the mineral and that of water to minimize passivation, can have an impact on extraction strategy. For example, in the context of an ore where the REEs are distributed between silicates and other minerals such as phosphates, a sequential approach may be considered where an initially diluted acid solution, which will ensure the efficient decomposition of the silicates, is subsequently concentrated through water evaporation using a dynamic heating path, or by simple H₂SO₄ injection. In contrast, for nonsilicate REE carriers associated with gangue silicates, the use of concentrated H₂SO₄ would optimize REE recovery, minimize acid consumption, and improve the purity of the resulting leach solution, considering that it will promote decomposition of the REE minerals; yet, at the same time, it will induce passivation of the unwanted silicates.