The behaviour of selected rare-earth elements during the conversion of phosphogypsum to calcium sulphide and residue

The PG used in this study was collected from a fertiliser and phosphoric acid producing company in Phalaborwa, Limpopo Province, South Africa. This company uses fluoroapatite rocks as raw material to produce H₃PO₄ through wet process. The produced PG is stocked in waste stacks in an open space close to the H₃PO₄ factory. Phosphogypsum samples were collected at three different points on the waste stacks (on top, in the middle, and at the bottom of the PG stacks), mixed thoroughly and transported to the laboratory in sealed plastic bags for analyses.

The collected PG was converted to CaS following a thermal reduction reaction. The main purpose of thermal reduction was to produce a suitable starting material for CaCO₃ and S recovery as well as residue formation. In addition, this process removed both the water of crystallisation and some volatile impurities including the CO₂ contained in the PG samples [10, 27]. The PG after dewatering was mainly made of CaSO₄ (97.1 wt%) and BaSO₄ (1.2 wt%) whilst the rest of impurities were in minimal concentrations not detectable by the instrument. They might be originating from the contamination of PG with other neutralising agent. No effort was made in identifying their sources as they were in negligible amount and had no effect on the overall composition of the PG. Whilst other minor elements included REEs, Al, Th and others which were in low concentrations. Prior to the commencement of the thermal reduction process, the PG was first oven-dried for 4 h at 105 °C, thereafter it was thoroughly mixed with coal (containing no other elements except carbon (75 wt%), SiO₂ (7.3 wt%), Al₂O₃ (7.5 wt%), water (5.4 wt%) and volatiles (4.8wt%)) in a stainless steel stepped crucible in a ratio of 1-to-2.5 molar mass of PG: coal [33]. Second, the mixture was ground to finest particles possible using a mortar and pestle and sieved with a 0.57 µm sieve to obtain good particle sizes, ensure sufficient surface area for a good reactivity and good enrichment of REEs by decreasing particle size. Finally, after grinding and sieving, the mixture was then placed in a furnace at 1100 °C for 1 h to allow the reduction reaction of PG to CaS to take place. Nitrogen gas (N₂) was pumped into the furnace to remove oxygen before, throughout the reduction until the whole process was completed and the crucible removed from the furnace. Upon completion, the furnace was turned off and left to cool down to room temperature before taking the stainless steel stepped crucible out [33]. The formed CaS was thereafter kept in a desiccator for further analyses and use.

A portion of the formed CaS was processed to produce elemental sulphur, CaCO₃ and a residue following the process as described: 100 g of CaS was weighed and dissolved in 900 mL of water in a 2.5 L glass container named as reactor A. This reactor A was rubber sealed with two holes on top. The first hole was used to allow the required gas (H₂S) to flow inside the reactor A, and the second one was to allow a magnetic stirrer in which was used to stir the mixture at 500 rpm at room temperature. H₂S was continuously introduced into the reactor A at a flow rate of 450–1350 mL/min with the intent to solubilise CaS and form liquid calcium bi-sulphide (Ca(HS)₂ according to Eq. 2 [12]. This reaction was run for a maximum period of 10 min after which the formed Ca(HS)₂ was separated from the residue through filtration and transferred into the reactor B, which was also a rubber-sealed 2.5 L glass container with two holes on top as well. In this case, the first hole was for the introduction of CO₂ in excess (at a flow rate of 450–1340 mL/min) via a plastic pipe which was connected to a CO₂ cylinder in order to form CaCO₃ following Eq. 3 [33], and the second one was to allow in a plastic pipe connecting the reactor B to reactor C. The H₂S (which is one of the products formed by a reaction between Ca(HS)₂ and CO₂) released from the reaction taking place in Reactor B travelled through this pipe and was captured in Reactor C where there was a solution of iron (III) Sulphate. The H₂S was allowed to react with the ferric solution to form elemental sulphur as indicated in Eq. 5 [12]. Once enough precipitates were formed in reactors B and C, the reactions were stopped, and precipitates (CaCO₃ in reactor B and S in reactor C) were separated from the remaining solutions by filtration.

In another experiment, CO₂ was used instead of H₂S in reactor A, and the reaction was allowed to take place as shown in the Eq. 4 [10]. In this instance, both CaCO₃ and Ca(HS)₂ were formed and separated from the residue:

All the above reactions were successfully achieved at room temperature. The only parameters monitored were the concentration of the reactants and those of the products formed. A schematic presentation of the experimental procedure for the study is shown in Fig. 1

The residue, CaCO₃ and S were filtered from their respective solutions, washed with deionised water and oven-dried at 105ºC for 72 h. The dried residue and PG as well as the formed CaS were digested using both a microwave digester (CEM Corporation, MARS 240/50 Mathews, NC, USA) and traditional acid leaching for REEs identification and quantification. Digesting solvents were made up of mixing diverse concentrations and combinations of sulfuric and hydrochloric acids. This was conducted with the intent to achieve an effective release of REEs from all solid samples for their consecutive quantification [3, 34]. In an endeavour to recover more CaCO₃ and S from CaS and facilitate possible REEs recovery from the residue, the amount of some impurities was reduced.

The PG, CaS and the residue were characterised prior to the determination of the REEs. A portion of the PG collected from the H₃PO₄ factory, was weighed and oven-dried and the determined percentage was found to be 26.67%. All samples were dried and ground to fine particles for characterisation.

Samples excluding CaCO₃ were analysed using a PANalytical X’Pert Pro powder diffractometer in θ–θ configuration using an X’Celerator detector and variable divergence and fixed receiving slits with Fe-filtered Co–Kα radiation (λ = 1.789 Å). The sample percentage compositions were estimated using the Rietveld method in percentage form. CaCO₃ allotropic forms as well as S XRD analyses were reported in studies previously carried out by Ruto et al. [33], De Beer et al. [15] and Masukume et al., who used the same procedure and PG to form them [12]. They were all satisfied with the quality of the CaCO₃ and S obtained from this very same process. X-Ray diffraction (XRD) analyses were carried out to ascertain the crystal structure of the three sample types. The main reason of characterising the residue, which is not a product but a left over from the reaction (2), was to prove that the reaction between CaS and H₂S was not left to completion but was stopped. This is why it was necessary to determine if there were CaS crystals in it. The samples were prepared according to the standardised Panalytical back-loading system, which provides nearly random distribution of the particles. The phases were identified using X’Pert Highscore plus software.

For the morphology of the PG, CaS and residue, each sample was powdered and separately mounted onto the sample holder of a JEOL JSM − 7500F scanning electron macroscopy and energy-dispersive spectroscopy (SEM–EDS) using carbon tape. To obtain cross-sectional images of the particles, the powdered samples were embedded in an epoxy resin and polished after which they were coated with gold to ensure that they were able to conduct electricity. Transmission electron microscopy (TEM) analyses were conducted using JEOL JEM-2100 electron microscope, Hitachi, Tokyo, Japan to examine the nature and structure of all solid samples at a higher resolution to better identify the REE elemental mapping. Sample particles were dispersed onto the holey carbon film supported by the 200-mesh TEM grid and analysed as prescribed by in the user manual.

This study focussed on REEs that could be determined with acceptable level of accuracy using an inductively coupled plasma optical emission spectrometer (ICP-OES). Weighed portions of CaS, PG and residue were leached in different concentrations and combinations of H₂SO₄ and HCl traditionally, and using a microwave digester. Moreover, these acids have been widely used for the release of metallic ions from solid samples in several studies [3, 23]. The concentrations of the acids used were 0.1; 0.5; 1; 5 and 9N. The concentrations of the acids used were varied to study the effect of solvent concentration on the extraction of REEs from the samples. Precisely 5 g of each sample was accurately weighed into a 250 mL beaker and 100 mL of acid were added. The solid–liquid ratio of 1:20 used in this study was in accordance with recommendations made from optimization studies carried out by Ruto et al. [33] and Masukume et al. [12]. The digestion time was 30 min and mixing was done at 555 rpm to keep suspended solids dispersed. Thereafter digestion, the digest was filtered and the filtrate analysed for REE using an ICP-OES. The ICP-OES conditions were set as recommended by the manufacturer.

Microwave digestion of PG, CaS and residue was also carried out using two sets of acid combination to determine which one was more effective on recovering REEs from the samples. The first combination (A) was made of 6 mL of hydrochloric acid and 2 mL of nitric acid which is aqua regia that has been used in previous studies. The second combination (B) was the opposite of the first one where 6 mL of HNO₃ were mixed with 2 mL HCl. Exactly 1 g of each sample was weighed into microwave vessel, either the combination A or B of solvent was added before sealing the vessel and placing it into the microwave. The microwave was ramped up to 180 °C over a period of 20 min and held at this temperature for a time of 20 min. The microwave power during digestion was set at 50% (400 KW). After digestion, the sample vessels were left to cool down to room temperature and the digests transferred into 25 mL volumetric flasks and made up to volume with de-ionised water. All samples were filtered prior analysis.

All acids used in this study were of analytical grade unless otherwise stated. Reagent preparation was done using deionised water. Spiked recoveries were used to assess the accuracy and precision of the method used for digestion and instrument set-up [36]. The percentage recoveries obtained after spiking with 2.5 mg/L multi-element REEs standards ranged from 92 to 99.9% (Table 1). Lowest recoveries were those of Y and Pr, whereas the highest was that of Ce. The recoveries were considered satisfactory and the method was further used to determine REEs concentrations in different samples. The wavelengths used were in accordance with those applied when determining REEs in geological samples by ICP-OES [37].

The XRD results of PG, CaS and the residue are shown in Fig. 2. These results indicated that the PG was a mixture of dehydrate (87 wt%), hemihydrate (5 wt%) and anhydrous (8 wt%) calcium sulphate. This revealed that oven drying of PG could not exhaust all crystallisation water molecules it contained. The reduction of PG to CaS was successfully (95%) achieved as the XRD results indicate CaS as the main component in the formed CaS sample (Fig. 2). Furthermore, this was confirmed as the calculated percentage CaS obtained using Eq. (5). As for the formed residue, results showed that it was mainly made of CaS (Fig. 2). This was as a result of the fact that the reaction in reactor A was not left to completion purposely as the intention was to form a residue which can be further processed in order to recover REEs. No REE mapping was detected in the samples using XRD possibly because of their low concentration.

The PG particles crystal structures were a mixture of large and small particles as indicated in Fig. 3. The morphology or shapes found in PG correlated with those reported in PG from Alberta, Canada, and this could probably be attributed to the similarities in the formation process [3]. The PG used was having a rhombohedral shape. There was an observed difference in the shapes of PG and the formed CaS as shown in the SEM images in Fig. 3. This is surely due to the reduction reaction that took place. The smaller particle size of the formed CaS indicates the thermal reduction process decreased the particle size and distorted the shape of the particles [38]. The residue did not show any precise shape probably because of the reactions it went through in reactor A. The EDS did not register any of the REEs in all samples. Elemental mapping for CaS and residue is shown in Figs. 4 and 5, respectively, and the results presented in these figures indicate the presence of major elemental components including carbon, oxygen, aluminium, sulphur and calcium.

TEM analysis was conducted to investigate the structure and nature of PG, CaS and residue at higher resolution and the results obtained are shown in Fig. 6. PG particles were seen to be a mixture of dispersed small and big rhombohedral structured particles, whilst those of CaS and residue were agglomerated and predominately very fine particles of a desultory structures. Through SEM and TEM, REEs were hardly detected because of their lower concentration, which made it hard to examine their behaviour. The behaviour of REEs was therefore monitored using the ICP-OES after complete leaching of sample followed by the subsequent analysis of liquid samples.

The concentrations of REEs in CaS were slightly higher than those in PG (Table 3). This might be attributed to the elimination of moisture and other volatile impurities as the PG was subjected to a thermal reduction for its conversion to CaS. The capacities of the acids used to extract REEs from CaS were similar to what was observed with PG. The concentrations of LREEs in the CaS samples were also higher than those of the HREEs. Worth noting is the fact that REEs were not leached from CaS by 0.1N H₂SO₄.

Since REEs seemed to be trapped in the CaS, the detection of REEs is strongly dependent on the dissolution of CaS particles. Optimum concentration of extracting REE from CaS using H₂SO₄ was 1N where the detection of more HREEs was observed compared to what was obtained for PG. Beyond this concentration, there was a decline in REE content as shown in Table 3. This is very different from what was observed with PG. Leaching CaS with different concentrations of HCl revealed a correlation between the concentration of REEs and that of the leaching agents. Rare-earth elements concentrations proportionally increased with the increase in the concentration of HCl solutions. A solution of 9N HCl extracted the highest concentration of REEs and can be considered as the optimal leaching acid concentration when using HCl. This is possibly due to the fact that more CaS were dissolved at this concentration resulting in the extraction of more REEs.

Hydrogen sulphide and CO₂ gases were separately tested by reacting them with CaS in the process of sulphur and CaCO₃ production to see if any REEs contained in the CaS were leached out at this stage of the process. Solutions were collected after reacting CaS with both H₂S and CO₂, filtered and analysed for REEs concentration. The obtained results are shown in Fig. 9. Results indicated that Ce, Pr, Nd, Sm, Gd and La were leached during this process as they were detected in the leachate of CaS. The remaining REEs could not be detected. This could be attributed to either the little effect that both gases had on them or their minimal concentrations that were below the detection limit of the equipment used.

Both gases leached out more or less the same amount for the majority of REEs, except for Ce, Nd and Gd which were more affected by CO₂ than H₂S. Thus, in view of these findings, if the residue is to be used for REEs recovery after CaS processing, it will be preferable to use H₂S in place of CO₂ gas [40].

The outcomes of this study have provided a scientific evidence to use this process as a possible way of recovering multiple products from readily available PG. In view of the presented results, it can be said with maximum conviction that the residue could constitute a better raw material for the recovery of REE than the raw PG given the amount of REEs, the ease of leaching, and reduction in the amount of impurities [41]. Maximising the recovery of REEs from readily available by-products could assist in mitigating excessive exploitation of little available natural resources and minimise both pollution and cost involved with REEs mining. The amount of PG occupying land space is concerning especially if nothing is done about it. Converting it to useful products through a viable economic treatment process would substantially assists in reducing the amount of PG waste [39].