Rare Earth Elements ( La , Ce , Pr , Nd , and Sm ) from a Carbonatite Deposit : Mineralogical Characterization and Geochemical Behavior

Geochemical characterization including mineralogical measurements and kinetic testing was completed on samples from the Montviel carbonatite deposit, located in Quebec (Canada). Three main lithological units representing both waste and ore grades were sampled from drill core. A rare earth element (REE) concentrate was produced through a combination of gravity and magnetic separation. All samples were characterized using different mineralogical techniques (i.e., quantitative evaluation of minerals by scanning electron microscopy (QEMSCAN), X-ray diffraction (XRD), and scanning electron microscopy with X-ray microanalysis (SEM-EDS)) in order to quantify modal mineralogy, liberation, REE deportment and composition of REE-bearing phases. The REE concentrate was then submitted for kinetic testing (weathering cell) in order to investigate the REE leaching potential. The mineralogical results indicate that: (i) the main REE-bearing minerals in all samples are burbankite, kukharenkoite-Ce, monazite, and apatite; (ii) the samples are dominated by REE-free carbonates (i.e., calcite, ankerite, and siderite); and (iii) LREE is more abundant than HREE. Grades of REE minerals, sulfides and oxides are richer in the concentrate than in the host lithologies. The geochemical test results show that low concentrations of light REE are leached under kinetic testing conditions (8.8–139.6 μg/L total light REE). These results are explained by a low reactivity of the REE-bearing carbonates in the kinetic testing conditions, low amounts of REE in solids, and by precipitation of secondary REE minerals.


REE Background
In recent decades, rare earth elements (REE), also called lanthanides (La to Lu, Sc, and Y), have been considered strategic metals, as they are a key component in the manufacturing of modern technologies. Their economic importance has grown as a result of their use in a wide range of industries including the manufacturing of products aimed at environmental conservation and sustainability. Applications include use in hybrid vehicles, radar systems, rechargeable batteries, mobile phones, flat screen display panels, compact fluorescent light bulbs, electronic products, etc.
Montviel alkaline proterozoic intrusion. The mining company GéoMéga Resources estimates the indicated resources to be approximately 183.9 Mt at 1.45% total rare earth oxides (TREO) and an additional 66.7 Mt at 1.46% TREO in the inferred resources category.

Sampling and Preparation
The materials used in this work were supplied by Geomega Resources Inc. Approximately 5 kg of each lithology were sampled from exploration drill core: low-grade and high-grade ferro-carbonatites (FeC-LG and FeC-HG, respectively), and calcio-carbonatite (CaC). Cores were crushed, then pulverized to mimic particle size distribution typical to the one targeted by ore processing.
In order to satisfy the characterization requirements (i.e., detection limit) and to more easily investigate the REE mineralogy and geochemical behavior during kinetic tests, REE were concentrated to increase the geochemical responses ( Figure 1). the Montviel alkaline proterozoic intrusion. The mining company GéoMéga Resources estimates the indicated resources to be approximately 183.9 Mt at 1.45% total rare earth oxides (TREO) and an additional 66.7 Mt at 1.46% TREO in the inferred resources category. The mineralized zone is mainly composed of five lithologies: silicocarbonatites, calciocarbonatites, magnesiocarbonatites, ferrocarbonatites and polygenic breccia. The mineralogical composition of these lithologies is mainly represented by carbonates (i.e., calcite, ankerite, siderite, barytocalcite, and strontianite), REE carbonates and REE-Ba-Sr mixtures [12,37].

Sampling and Preparation
The materials used in this work were supplied by Geomega Resources Inc. Approximately 5 kg of each lithology were sampled from exploration drill core: low-grade and high-grade ferrocarbonatites (FeC-LG and FeC-HG, respectively), and calcio-carbonatite (CaC). Cores were crushed, then pulverized to mimic particle size distribution typical to the one targeted by ore processing.
In order to satisfy the characterization requirements (i.e., detection limit) and to more easily investigate the REE mineralogy and geochemical behavior during kinetic tests, REE were concentrated to increase the geochemical responses ( Figure 1). Due to their high density and magnetic properties, REE-bearing minerals were enriched using gravity and magnetic separation. Ferro-carbonatite high grade (FeC-HG) was used as feed for the enrichment process. The material was dried and sieved into four different fractions (−106 µm, −315/+106 µm, −630/+315 µm, and −2 mm/+630 µm) to determine the REE distribution. The fractions were selected on the basis of the grain size distribution curve of the whole material to provide equivalent weight fractions. The heavier minerals appeared to be concentrated within the finest fraction, i.e., below 106 µm with 46,275 ppm LREE (La, 16,767 ppm; Ce, 27,504 ppm; Pr, 2009 ppm). Slurry composed of water and solids (around 40% solids) corresponding to each size fraction was prepared in the mixing tank before feeding the concentration process steps. Several stages of concentration using a Knelson concentrator and Mozley table were necessary to improve the grade of the concentrate (Figure 2). The final concentrate was created by mixing concentrates with the same initial proportions (raw sample FeC-HG). The separation efficiency was first verified through binocular loupe and then submitted to physical, chemical, and mineralogical characterization. Due to their high density and magnetic properties, REE-bearing minerals were enriched using gravity and magnetic separation. Ferro-carbonatite high grade (FeC-HG) was used as feed for the enrichment process. The material was dried and sieved into four different fractions (−106 µm, −315/+106 µm, −630/+315 µm, and −2 mm/+630 µm) to determine the REE distribution. The fractions were selected on the basis of the grain size distribution curve of the whole material to provide equivalent weight fractions. The heavier minerals appeared to be concentrated within the finest fraction, i.e., below 106 µm with 46,275 ppm LREE (La, 16,767 ppm; Ce, 27,504 ppm; Pr, 2009 ppm). Slurry composed of water and solids (around 40% solids) corresponding to each size fraction was prepared in the mixing tank before feeding the concentration process steps. Several stages of concentration using a Knelson concentrator and Mozley table were necessary to improve the grade of the concentrate ( Figure 2). The final concentrate was created by mixing concentrates with the same initial proportions (raw sample FeC-HG). The separation efficiency was first verified through binocular loupe and then submitted to physical, chemical, and mineralogical characterization.

Physical and Chemical Analysis
The specific gravity (SG) was determined with a AccuPyc 1330 Helium Pycnometer (Micromeritics, Norcross, GA, USA). The particle size distributions were analyzed using a Mastersizer S laser analyzer (Malvern Instruments, Malvern, UK). The specific surface area (SSA) was determined by using a Gemini Surface Area Analyser (Micromeritics, Norcross, GA, USA) using the BET method [38].
The whole-rock analyses were performed using an X-ray Fluorescence (Bruker, Tiger Model, Billerica, MA, USA), analyzer on a powdered aliquot of each sample to determine the major and minor elements. Chemical analyses were carried out by Inductively Coupled Plasma (ICP-AES/MS, Perkin Elmer Optima 3100-RL, Waltham, MA, USA). Over 20 elements, to a 0.001 wt % precision were analyzed after a multi-acid digestion (HCl, HNO3, and HF). The total sulfur and carbon grades were measured using an induction furnace analyzer (ELTRA CS-2000, Mettmann, Germany) with the detection limits are 0.05% of carbon and 0.009% of sulfur, respectively The physicochemical parameters of the weathering cell leachates (pH, conductivity, Eh) were determined using a pH/Eh/electrical conductivity meter. The ICP-MS was used to analyze the REE concentrations in the weathering cell leachates, while the other elements were measured using the ICP-AES on an acidified (2% HNO3) and filtered (<0.45 µm) aliquot. The acid-base accounting (AP: acid-generating potential; NP: neutralization potential; NNP = NP − AP: net neutralization potential) was determined using the percentage of sulfur (AP = 31.25 × %S) and carbon (NP = 83.3 × %C).

Physical and Chemical Analysis
The specific gravity (SG) was determined with a AccuPyc 1330 Helium Pycnometer (Micromeritics, Norcross, GA, USA). The particle size distributions were analyzed using a Mastersizer S laser analyzer (Malvern Instruments, Malvern, UK). The specific surface area (SSA) was determined by using a Gemini Surface Area Analyser (Micromeritics, Norcross, GA, USA) using the BET method [38].
The whole-rock analyses were performed using an X-ray Fluorescence (Bruker, Tiger Model, Billerica, MA, USA), analyzer on a powdered aliquot of each sample to determine the major and minor elements. Chemical analyses were carried out by Inductively Coupled Plasma (ICP-AES/MS, Perkin Elmer Optima 3100-RL, Waltham, MA, USA). Over 20 elements, to a 0.001 wt % precision were analyzed after a multi-acid digestion (HCl, HNO 3 , and HF). The total sulfur and carbon grades were measured using an induction furnace analyzer (ELTRA CS-2000, Mettmann, Germany) with the detection limits are 0.05% of carbon and 0.009% of sulfur, respectively The physicochemical parameters of the weathering cell leachates (pH, conductivity, Eh) were determined using a pH/Eh/electrical conductivity meter. The ICP-MS was used to analyze the REE concentrations in the weathering cell leachates, while the other elements were measured using the ICP-AES on an acidified (2% HNO 3 ) and filtered (<0.45 µm) aliquot. The acid-base accounting (AP: acid-generating potential; NP: neutralization potential; NNP = NP − AP: net neutralization potential) was determined using the percentage of sulfur (AP = 31.25 × %S) and carbon (NP = 83.3 × %C).

Mineralogical Characterization
The mineralogical composition was identified by X-ray diffraction (XRD; Bruker D8 Advance, with a detection limit and precision of approximately 1 to 5%, operating with a copper cathode, Kα radiation) (Bruker, Billerica, MA, USA) using the DIFFRACT.EVA program, and quantified with Rietveld refinement using the Bruker's TOPAS 4.2 software [39]. In order to reconcile the final mineralogical results, chemical and mineralogical results are collected, compared, and adjusted to determine the percentages of mineral phases present within samples.
Mineralogical investigation and local X-ray microanalysis were completed on a Hitachi S-3500N scanning electron microscope (SEM, Hitachi High-Technologies Canada, Toronto, ON, Canada), equipped with an Energy-Dispersive X-ray Spectroscopy (EDS) probe (Oxford Instruments, Abingdon, UK, Oxfordshire 20 mm 2 X-Max Silicon Drift Detector (SDD)). All analyses where carried out using an accelerating voltage of 20 kV, current of 150 µA, pressure in the chamber fixed at 25 Pa and a working distance of 15 mm. The polished section observations were determined using Backscattered electrons imaging mode (BSE). Electron probe microanalysis (EPMA), which has lower detection limits than the EDS-SEM system, were also used to analyze the trace metal composition of certain accessory minerals known for their affinity to REE, such as apatite. Electron microprobe mineralogical analyses (Cameca SX 100) were performed using an accelerating voltage of 20 kV and a beam current of 20 nA.
In order to further understand the effect of mineralogy, mineral size distributions, and effect of liberation on the geochemical behavior of REE-bearing minerals, a QEMSCAN (Quantitative Evaluation of Materials by Scanning Electron Microscopy) analysis was performed [24,40]. QEMSCAN is an automated system that produces particle maps (color coded by mineral) through the collection of rapidly acquired X-rays. The distribution and corresponding data files quantify the modal mineralogy, texture (mineral association), grain size, elemental deportment and liberation of samples analyzed. The polished sections were analyzed by PMA (particle mineralogy analysis) and TMS (trace mineral search) modes. Measurement resolution was set to 0.8 m or 1 µm to capture the fine particle sizes [8,40]. The mineralogical observations were performed on polished sections prepared carefully to avoid density segregation and ensure a homogenous dispersion of particles. The mineralogical observations were mainly focused on REE carbonates.

Weathering Cells
A weathering cell test was used to evaluate the geochemistry of the REE concentrate following the procedures of Bouzahzah [41]. The weathering cells consist of bi-weekly flushes by 50 mL of deionized water on a sample weighing 67 g. The sample (particle size distribution <400 µm) is left to dry under ambient air between flushes. The bi-weekly (Mondays and Thursdays) leachates are combined and analyzed for their pH, Eh, electrical conductivity, and dissolved constituents.

Physical and Chemical Characteristics of the Three Montviel Lithologies and the REE Concentrate
The physical and chemical properties of the studied samples are summarized in Table 1. The SG of the samples ranges from 3 to 3.60. The specific gravity of the REE concentrate (3.60) is greater than that of all the lithological units (3.00-3.36) due to the increased content of sulfide and REE-bearing minerals. Fine grinding of the FeC-HG was required to ensure a high recovery of REE-bearing minerals to the REE concentrate. The D 50 and D 90 values (the particle size passing 50% and 90% on the cumulative grain size distribution curve, respectively) also highlight the particle size differences between the waste rocks samples and REE concentrate sample. The D 50 and D 90 values of the REE concentrate are significantly lower (124-209 µm) than for the waste rock samples (D 50 : 2075-2775 µm and D 90 : 5000-5500 µm).
The chemical characterization results show that the REE concentrate sample contains higher concentrations of REE, Fe, Ca, Mg, Mn, Zn, Na, K, Ba, and Pb compared to the other samples. Indeed, the highest sum of LREE is recorded in the REE concentrate (36,226 mg/kg), while the lowest sum is found in the BreC (LG) (4400 mg/kg). The samples are rich in LREE compared to HREE. Lanthanum (La), cerium (Ce), and neodymium (Nd) are the most abundant elements in all samples, with a maximum content in the REE concentrate and a minimum content in the breccia sample.
The acid-base accounting results show that the AP values are all lower (4.1 < AP < 86.5 kg CaCO 3 /t) than the NP values (44 < NP < 602 kg CaCO 3 /t), leading to NNP values undoubtedly within the non-acid generating zone (>20 kg CaCO 3 /t). The relationship between LREE, Ba, and Sr shows a strong correlation with LREE (R 2 > 0.90), LREE vs. Ba (R 2 = 0.99) and LREE vs. Sr (R 2 = 0.98) ( Figure 3). This is confirmed by the experimental constant relative proportions of (Ce/Nd) = 2.5, (Ce/Pr) = 12.75, and (Ce/La) = 1.6 in all samples. These relationships between LREE, Sr, and Ba, verified for all samples, suggest that these elements are associated in the same mineral phases. These geochemical anomalies are confirmed in the QEMSCAN analyses section. The relationship between LREE, Ba, and Sr shows a strong correlation with LREE (R 2 > 0.90), LREE vs. Ba (R 2 = 0.99) and LREE vs. Sr (R 2 = 0.98) ( Figure 3). This is confirmed by the experimental constant relative proportions of (Ce/Nd) = 2.5, (Ce/Pr) = 12.75, and (Ce/La) = 1.6 in all samples. These relationships between LREE, Sr, and Ba, verified for all samples, suggest that these elements are associated in the same mineral phases. These geochemical anomalies are confirmed in the QEMSCAN analyses section.
These XRD results are confirmed by the other mineralogical techniques. SEM-EDS and QEMSCAN investigation show that: (i) REE are partially or totally liberated as free particles (kukharenkoite, burbankite, qaqarssukite, monazite) and/or associated with pyrochlore, baytocalcite, apatite, and Mg-Fe carbonates ( Figure 4); (ii) all analyzed REE-bearing minerals contain LREE (La, Ce, Pr, Nd, Sm); (iii) two groups of REE-bearing minerals (carbonates and phosphates) were detected with variable REE grades; and (iv) pyrite and pyrrhotite were the most abundant sulfide minerals identified in the sample.
In order to better investigate the REE mineralogy, all samples were submitted to mineralogical assessment using QEMSCAN. The main objective of the QEMSCAN work was to define the modal abundances, grain size, REE deportments and mineral liberation. The results of modal mineralogy show that REE minerals include kukharenkoite Ba 2 (La,Ce)(CO 3 ) 3 F, burbankite (Na,Ca) 3 (Sr,Ba,Ce) 3 (CO 3 ) 4 , ancylite Sr(La,Ce)(CO 3 ) 2 (OH)·(H 2 O), carbocernaite (Ca,Na)(Sr,Ce,Ba)(CO 3 ) 2 , qaqarssukite Ba(Ce,REE)(CO 3 ) 2 F, and apatite Ca 5 (PO 4 ) 3 F ( Figure 5). The QEMSCAN mineralogical data shows that carbonernaite, ancylite, barytocalcite, and apatite minerals were undetectable by XRD.  These XRD results are confirmed by the other mineralogical techniques. SEM-EDS and QEMSCAN investigation show that: (i) REE are partially or totally liberated as free particles (kukharenkoite, burbankite, qaqarssukite, monazite) and/or associated with pyrochlore, baytocalcite, apatite, and Mg-Fe carbonates ( Figure 4); (ii) all analyzed REE-bearing minerals contain LREE (La, Ce, Pr, Nd, Sm); (iii) two groups of REE-bearing minerals (carbonates and phosphates) were detected with variable REE grades; and (iv) pyrite and pyrrhotite were the most abundant sulfide minerals identified in the sample.  Ba(Ce,REE)(CO3)2F, and apatite Ca5(PO4)3F ( Figure 5). The QEMSCAN mineralogical data shows that carbonernaite, ancylite, barytocalcite, and apatite minerals were undetectable by XRD. Gangue minerals are dominated by carbonates that include calcite (most common in CAC), as well as ankerite, siderite, Mn siderite, and a Ca Mg Fe carbonates. Composites with higher levels of Fe have more ankerite, siderite, Ca Mg Fe carbonate compared to CAC. Silicates are dominated by biotite, chlorite with trace levels of amphiboles, pyroxene, feldspar, quartz, titanite and zircon. The QEMSCAN findings show that sulfides (mainly pyrite) and Fe-oxides are more abundant in the REE concentrate in comparison with the studied lithologies ( Figure 5). The difference between XRD and QEMSCAN analyses is perhaps due to two reasons: (i) XRD is a semi-quantitative method that has a detection limit in the 1-5% range, and/or (ii) a representativeness problem of the tested sample. Given the sample preparation procedure, the latter is unlikely.
The QEMSCAN results shown in Figure 5 demonstrate that the concentration process successfully increased the grades of all REE-bearing phases, and decreased those of the non REEbearing gangue minerals (carbonates and silicates). The degree of liberation of REE-bearing minerals is very important because it allows evaluation of mineral absolute reactivity. Liberation (based on area %) data is shown for combined REE minerals. At least, 50% of the REE-bearing minerals are liberated (more than 15% as free particles) within the samples, while the remaining particles are either included or locked within gangue minerals (e.g., silicates and carbonates) ( Figure 6). REE deportments for Ce, La, Nd, Pr and Sm have been calculated using the EPMA and QEMSCAN data. It is important to note that no deportment data is available for the REE that were not included in the EPMA analysis. Ce, La, Nd, Pr and Sm deportments are summarized in Figure 7. Ancylite, burbankite, carbonernaite, kukharenkoite, qaqarssukite, monazite, strontianite and apatite mainly contain Ce and Nd, whereas Sm and La deport mainly to monazite, ancylite, burbankite, carbonernaite, kukharenkoite, qaqarssukite. The gangue minerals are mainly represented by carbonates and silicates. The difference between XRD and QEMSCAN analyses is perhaps due to two reasons: (i) XRD is a semi-quantitative method that has a detection limit in the 1-5% range, and/or (ii) a representativeness problem of the tested sample. Given the sample preparation procedure, the latter is unlikely.
The QEMSCAN results shown in Figure 5 demonstrate that the concentration process successfully increased the grades of all REE-bearing phases, and decreased those of the non REE-bearing gangue minerals (carbonates and silicates). The degree of liberation of REE-bearing minerals is very important because it allows evaluation of mineral absolute reactivity. Liberation (based on area %) data is shown for combined REE minerals. At least, 50% of the REE-bearing minerals are liberated (more than 15% as free particles) within the samples, while the remaining particles are either included or locked within gangue minerals (e.g., silicates and carbonates) ( Figure 6). REE deportments for Ce, La, Nd, Pr and Sm have been calculated using the EPMA and QEMSCAN data. It is important to note that no deportment data is available for the REE that were not included in the EPMA analysis. Ce, La, Nd, Pr and Sm deportments are summarized in Figure 7. Ancylite, burbankite, carbonernaite, kukharenkoite, qaqarssukite, monazite, strontianite and apatite mainly contain Ce and Nd, whereas Sm and La deport mainly to monazite, ancylite, burbankite, carbonernaite, kukharenkoite, qaqarssukite. The gangue minerals are mainly represented by carbonates and silicates.
As shown in Table 2, carbonate minerals are the most abundant mineral group in the REE concentrate. As a result, the materials were proven as being non-acid generating based on quantitative evaluation of mineralogy using QEMSCAN results and the calculation of degree of liberation.

Geochemical Behavior
The weathering cell test results are shown in Figures 8 and 9. The pH of all the cell leachates remains alkaline (7.27-8.95) during the test. The electric conductivities of the leachates stabilize rapidly between 117 and 169 µS/cm. The redox potential (Eh) varied from 228 to 553 mV. The alkalinity values stabilize between 37 and 110 mg CaCO 3 /L, while the acidity is always below 5 mg CaCO 3 /L. All these parameters stabilize after the approximately the first 60 days. The leaching concentrations of Th, Nb, Ni, Pb and Cu are not presented due to their low concentrations in the leachates from the REE concentrate. It is worth mentioning that these elemental concentrations were below the values fixed by Directive 019 (provincial legislations in Quebec). Leachate Ca concentrations vary between 0.003 and 20.6 mg/L after approximately 287 days. The Ba concentrations increase from the beginning of the test until approximately 231 days, after which they decrease until the end of the test and remain below the analytical detection limit As shown in Table 2, carbonate minerals are the most abundant mineral group in the REE concentrate. As a result, the materials were proven as being non-acid generating based on quantitative evaluation of mineralogy using QEMSCAN results and the calculation of degree of liberation.

Geochemical Behavior
The weathering cell test results are shown in Figures 8 and 9. The pH of all the cell leachates remains alkaline (7.27-8.95) during the test. The electric conductivities of the leachates stabilize rapidly between 117 and 169 µS/cm. The redox potential (Eh) varied from 228 to 553 mV. The alkalinity values stabilize between 37 and 110 mg CaCO3/L, while the acidity is always below 5 mg CaCO3/L. All these parameters stabilize after the approximately the first 60 days. The leaching concentrations of Th, Nb, Ni, Pb and Cu are not presented due to their low concentrations in the leachates from the REE concentrate. It is worth mentioning that these elemental concentrations were   The evolution of the elemental concentration of S (presumably as sulfate) Ca, Na, Ba, Mg, Fe, Sr, and REE suggests the dissolution of carbonates is a result of water-rock interaction and/or acid generated by sulfide oxidation. The LREE concentrations are higher in the leachates from the REE concentrate (1-133 µg/L) compared to those of HREE (0.03-61 µg/L). These results are consistent with those found by Edahbi et al. [24] on the Kipawa alkaline intrusion REE deposit. In this geological context the leached REE is below 15 µg/L [24]. Moreover, Purdy [42] demonstrated that the REE concentrations in the leachate using shake flask tests on REE tailings of the Nechalacho Deposit, Northwest Territories, vary between 1 µg/L and 12 µg/L, with a maximum concentration in the tests using distilled deionized water and the minimum concentration in the tests using pilot plant water. The authors also demonstrated that more than 90% of the leached REE were presented in the colloidal phase [42]. In addition, other authors showed that REE fractionation is promoted by an increase in pH [24,43,44]. Edahbi et al. showed, using thermodynamic equilibrium calculations on the humidity cell leachate data, that REE concentrations decreased significantly (from 10 −8 to 10 -16 mg/L) when pH values increase from 2.6 to 10 [45] The evolution of the elemental concentration of S (presumably as sulfate) Ca, Na, Ba, Mg, Fe, Sr, and REE suggests the dissolution of carbonates is a result of water-rock interaction and/or acid generated by sulfide oxidation. The LREE concentrations are higher in the leachates from the REE concentrate (1-133 µg/L) compared to those of HREE (0.03-61 µg/L). These results are consistent with those found by Edahbi et al. [24] on the Kipawa alkaline intrusion REE deposit. In this geological context the leached REE is below 15 µg/L [24]. Moreover, Purdy [42] demonstrated that the REE concentrations in the leachate using shake flask tests on REE tailings of the Nechalacho Deposit, Northwest Territories, vary between 1 µg/L and 12 µg/L, with a maximum concentration in the tests using distilled deionized water and the minimum concentration in the tests using pilot plant water. The authors also demonstrated that more than 90% of the leached REE were presented in the colloidal phase [42]. In addition, other authors showed that REE fractionation is promoted by an increase in pH [24,43,44]. Edahbi et al. showed, using thermodynamic equilibrium calculations on the humidity cell leachate data, that REE concentrations decreased significantly (from 10 −8 to 10 −16 mg/L) when pH values increase from 2.6 to 10 [45].
The HREE (1387 mg/kg) are in lower concentrations in the solid samples relative to the LREE (51,220 mg/kg). However, the ratio of LREE to HREE (2,5) leached during the weathering cell tests are lower than their corresponding relative ratio in the solids (37). This suggests that the LREE have a lower mobility than HREE. This could be explained by the fact that the LREE have larger ionic radii than the HREE, which favors their precipitation as secondary minerals in weathering cell conditions.

Discussion
The mineralogy of REE and the geochemical behavior of the REE-bearing phases within the Montviel carbonatites were evaluated using QEMSCAN analyses and kinetic tests (weathering cells), and observations are discussed below.

Separation Efficiency and Environmental Challenges
The chemical composition of the concentrate analyzed by XRF are shown in Table 3. In general, LREE concentrations increase with decreasing particle size. Sulfide minerals and iron oxides are also concentrated in the finest size fraction of the concentrate sample. The Nb, and Th display a similar trend ( Figure 10). The majority of these elements are present in the −106 µm fraction (41% of LREE, 36% of Nb, and 19% of Th). The concentration of LREE in the finer fractions is explained by preferential grinding of REE-bearing minerals compared to the gangue minerals (i.e., silicates, iron oxides, etc.). The HREE (1387 mg/kg) are in lower concentrations in the solid samples relative to the LREE (51,220 mg/kg). However, the ratio of LREE to HREE (2,5) leached during the weathering cell tests are lower than their corresponding relative ratio in the solids (37). This suggests that the LREE have a lower mobility than HREE. This could be explained by the fact that the LREE have larger ionic radii than the HREE, which favors their precipitation as secondary minerals in weathering cell conditions.

Discussion
The mineralogy of REE and the geochemical behavior of the REE-bearing phases within the Montviel carbonatites were evaluated using QEMSCAN analyses and kinetic tests (weathering cells), and observations are discussed below.

Separation Efficiency and Environmental Challenges
The chemical composition of the concentrate analyzed by XRF are shown in Table 3. In general, LREE concentrations increase with decreasing particle size. Sulfide minerals and iron oxides are also concentrated in the finest size fraction of the concentrate sample. The Nb, and Th display a similar trend ( Figure 10). The majority of these elements are present in the −106 µm fraction (41% of LREE, 36% of Nb, and 19% of Th). The concentration of LREE in the finer fractions is explained by preferential grinding of REE-bearing minerals compared to the gangue minerals (i.e., silicates, iron oxides, etc.).  In the light of these findings, and in the framework of sustainable development, social acceptability, and environmental regulations of REE mine wastes, the finer fraction will be the most Figure 10. LREE enrichment as a function of the particle size.
In the light of these findings, and in the framework of sustainable development, social acceptability, and environmental regulations of REE mine wastes, the finer fraction will be the most problematic fraction and will potentially require a focused strategy to mitigate the risk. Fine particles are the most reactive to water-rock interactions due to its higher specific surface area and degree of liberation. Therefore, they may release metals, REE, and other contaminants in the drainage waters, which might exceed regulatory limits. As a circumvention strategy, and in the context of sound REE production, this fraction could be separated by a screening operation and reprocessed with the ore (Figure 11). The least problematic coarse fraction could be sent to waste rock piles with suitable and specific environmental management and waste disposal strategies.
REE ores are often contaminated with actinides such as thorium, uranium, and other contaminants (e.g., Ba, Sr, Mn, Ta) [11,12]. Social acceptability of REE production is an important issue for REE investors. Several publications found in the literature have reported significant environmental issues caused by REE mining and refining [13,14]. In China, the use of acids to recover REE leads to the deterioration of the water quality in the surrounding environment. In Malaysia, radioactive wastes were stored in ambient surface conditions and consequently generated deleterious impacts on the environment [15]. REE ores are generally the deposits characterized by large-tonnage, low-grade operations which could generate significant quantities of liquid and solid wastes (e.g., waste rocks, tailings, and effluents). For example, the production of one tonne of REE could be generated approximately 2000 tonnes of tailings [13]. In the case of Montviel project (Lebel-sur-Quévillon, QC, Canada) 250 million tonnes (Mt) of REE-Nb carbonatites with a REE content estimated at 1.47% of rare earth oxides (REO) will be extracted. The costs of processing, tailings, and environment of one metric tonne of is estimated to Canadian dollar 74.60 [46]. In order to implement an integrated management strategy for the REE waste rocks, the base of the waste rocks pile should be compacted and tilted to promote lateral water flow, and it should be composed of materials from the least problematic lithology. The liquid effluents, collected around the pile, should be passed through a limestone drain for passive treatment. problematic fraction and will potentially require a focused strategy to mitigate the risk. Fine particles are the most reactive to water-rock interactions due to its higher specific surface area and degree of liberation. Therefore, they may release metals, REE, and other contaminants in the drainage waters, which might exceed regulatory limits. As a circumvention strategy, and in the context of sound REE production, this fraction could be separated by a screening operation and reprocessed with the ore (Figure 11). The least problematic coarse fraction could be sent to waste rock piles with suitable and specific environmental management and waste disposal strategies. REE ores are often contaminated with actinides such as thorium, uranium, and other contaminants (e.g., Ba, Sr, Mn, Ta) [11,12]. Social acceptability of REE production is an important issue for REE investors. Several publications found in the literature have reported significant environmental issues caused by REE mining and refining [13,14]. In China, the use of acids to recover REE leads to the deterioration of the water quality in the surrounding environment. In Malaysia, radioactive wastes were stored in ambient surface conditions and consequently generated deleterious impacts on the environment [15]. REE ores are generally the deposits characterized by large-tonnage, low-grade operations which could generate significant quantities of liquid and solid wastes (e.g., waste rocks, tailings, and effluents). For example, the production of one tonne of REE could be generated approximately 2000 tonnes of tailings [13]. In the case of Montviel project (Lebel-sur-Quévillon, QC, Canada) 250 million tonnes (Mt) of REE-Nb carbonatites with a REE content estimated at 1.47% of rare earth oxides (REO) will be extracted. The costs of processing, tailings, and environment of one metric tonne of is estimated to Canadian dollar 74.60 [46]. In order to implement an integrated management strategy for the REE waste rocks, the base of the waste rocks pile should be compacted and tilted to promote lateral water flow, and it should be composed of materials from the least problematic lithology. The liquid effluents, collected around the pile, should be passed through a limestone drain for passive treatment.

Geochemical Behavior of REE
The weathering of the different lithologies within the Montviel deposit was studied in a previous study [45], where it was demonstrated that the REE-bearing carbonates from the Montviel deposit generates various elements (e.g., Ba, Sr, F, Th, and REE) in mine drainage waters. It is known that the carbonates dissolution is incomplete in near neutral conditions [47][48][49], and that the carbonate dissolution rates are lowest in alkaline and neutral conditions and increase in acid conditions [50,51]. The presence of impurities within REE-bearing minerals weaken their crystal lattice and consequently increase their dissolution rate. In the case of the Montviel materials, REE-carbonates

Geochemical Behavior of REE
The weathering of the different lithologies within the Montviel deposit was studied in a previous study [45], where it was demonstrated that the REE-bearing carbonates from the Montviel deposit generates various elements (e.g., Ba, Sr, F, Th, and REE) in mine drainage waters. It is known that the carbonates dissolution is incomplete in near neutral conditions [47][48][49], and that the carbonate dissolution rates are lowest in alkaline and neutral conditions and increase in acid conditions [50,51]. The presence of impurities within REE-bearing minerals weaken their crystal lattice and consequently increase their dissolution rate. In the case of the Montviel materials, REE-carbonates and non REE-carbonates neutralize the acidity produced by the sulfide oxidation or carried within acid rain, releasing REE and other associated elements in the leachates. However, it was demonstrated that the REE concentrations (and other associated elements) in the Montviel leachates are controlled by secondary precipitations and sorption phenomena. The behavior of the concentrate sample studied here is similar to that of the REE-bearing lithologies, although the residual REE concentrations in the leachates are higher (136 µg/L in the concentrate leachates compared to the lithologies, 9 µg/L). The use of a concentrate to mimic the behavior of REE in waste material is done in order to amplify the geochemical responses that could be obtained from mine wastes, and make measurement and interpretation more robust. This methodology was successfully applied in another context [52]. Consequently, the interpretation of the weathering cell data will not be investigated further; the interested readers are invited to consult [45].

Implications to the Prediction of the Geochemical Behavior and Water Quality of REE Mine Wastes
Predicting the geochemical behavior of REE-bearing mine wastes and, ultimately, the mine drainage quality of those wastes, is not significantly different than for other metals. However, prediction of contaminated neutral drainage issued from REE mine wastes is particularly challenging because of the low reactivity and complex compositions of REE-bearing minerals. Therefore, in order to better understand the geochemical behavior of REE waste rocks and tailings and predict the quality of the drainage waters, the use of detailed chemical/mineralogical characterizations such as QEMSCAN, MLA, SEM and EPMA are necessary. Figure 12 illustrates a mine water quality prediction strategy that could be applied to REE-bearing mine wastes. The purpose of this approach is to meet the following challenges ( Figure 12): (i) identify and quantify the different REE-bearing minerals (and associated potential contaminants) on representative samples of the different lithologies; (ii) evaluate the reactivity of the REE-bearing lithologies; (iii) propose an integrated management approach of the future REE-bearing waste rocks. In such an approach, representative samples of the different lithologies are submitted to percolation and leaching tests along with physical, chemical and mineralogical testing. The combination of these results helps to identify risks associated with weathering of mine waste and to move towards prediction of mine water quality. and non REE-carbonates neutralize the acidity produced by the sulfide oxidation or carried within acid rain, releasing REE and other associated elements in the leachates. However, it was demonstrated that the REE concentrations (and other associated elements) in the Montviel leachates are controlled by secondary precipitations and sorption phenomena. The behavior of the concentrate sample studied here is similar to that of the REE-bearing lithologies, although the residual REE concentrations in the leachates are higher (136 µg/L in the concentrate leachates compared to the lithologies, 9 µg/L). The use of a concentrate to mimic the behavior of REE in waste material is done in order to amplify the geochemical responses that could be obtained from mine wastes, and make measurement and interpretation more robust. This methodology was successfully applied in another context [52]. Consequently, the interpretation of the weathering cell data will not be investigated further; the interested readers are invited to consult [45].

Implications to the Prediction of the Geochemical Behavior and Water Quality of REE Mine Wastes
Predicting the geochemical behavior of REE-bearing mine wastes and, ultimately, the mine drainage quality of those wastes, is not significantly different than for other metals. However, prediction of contaminated neutral drainage issued from REE mine wastes is particularly challenging because of the low reactivity and complex compositions of REE-bearing minerals. Therefore, in order to better understand the geochemical behavior of REE waste rocks and tailings and predict the quality of the drainage waters, the use of detailed chemical/mineralogical characterizations such as QEMSCAN, MLA, SEM and EPMA are necessary. Figure 12 illustrates a mine water quality prediction strategy that could be applied to REE-bearing mine wastes. The purpose of this approach is to meet the following challenges ( Figure 12): (i) identify and quantify the different REE-bearing minerals (and associated potential contaminants) on representative samples of the different lithologies; (ii) evaluate the reactivity of the REE-bearing lithologies; (iii) propose an integrated management approach of the future REE-bearing waste rocks. In such an approach, representative samples of the different lithologies are submitted to percolation and leaching tests along with physical, chemical and mineralogical testing. The combination of these results helps to identify risks associated with weathering of mine waste and to move towards prediction of mine water quality.

Conclusions
The main objectives of this study were: (1) to characterize REE-bearing minerals and to improve the counting statistics, analytical precision, and the geochemical interpretations using a REE concentrate sample prepared by gravity and magnetic processes; and (2) to evaluate the geochemical behavior of the REE-bearing materials using kinetic tests. Three lithological units chosen from exploration dill core to represent the heterogeneity of the GéoMéga waste rocks and a REE concentrate were tested. The lithologies were chosen to represent typical residue of REE mining and refining.
The main mineralogical conclusions drawn from this study indicate that the main REE-bearing minerals in Montviel deposit are ancylite, burbankite, carbonernaite, kukharenkoite, qaqarssukite, monazite, strontianite and apatite. In addition, several accessory minerals contain REE, such as apatite and pyrochlore. At least 50% of the REE-bearing minerals are liberated within the samples, while the remaining portion is either included or locked in gangue minerals (i.e., silicates and carbonates).
Geochemical results from both kinetic tests show a stable pH around 8 and alkalinity values between 37 and 110 mg CaCO 3 /L. The main geochemical results from the kinetic tests are: (i) REE are good geochemical tracers of carbonate dissolution; and (ii) the leachability of REE from the studied carbonatite deposit is low and controlled by sorption processes in weathering cell conditions.